Now we’retalking

Voice computingcomes of age

JANUARY 7TH–13TH 2017

Turkey torn apart

Theresa Maybe, Britain’s indecisive PM

Nestlé goes on a health kick

Meet China’s Shakespeare

EconCow.com

Latest video (Jan 2017)

Were Richer Voters More Likely to Vote Trump?

Sorry for this ad. I’m just a random guy on the internet. I’m also a

new YouTuber who makes educational videos (mostly on

economics). Please help me by checking them out and let me know

how I can improve! Thank you!

The Economist January 7th 2017 3

Daily analysis and opinion tosupplement the print edition, plusaudio and video, and a daily chartEconomist.com

E-mail: newsletters andmobile editionEconomist.com/email

Print edition: available online by7pm London time each ThursdayEconomist.com/print

Audio edition: available online to download each FridayEconomist.com/audioedition

The Economist online

Volume 422 Number 9022

Published since September 1843to take part in "a severe contest between intelligence, which presses forward, and an unworthy, timid ignorance obstructing our progress."

Editorial offices in London and also:Atlanta, Beijing, Berlin, Brussels, Cairo, Chicago,Lima, Mexico City, Moscow, Mumbai, Nairobi, New Delhi, New York, Paris, San Francisco, São Paulo, Seoul, Shanghai, Singapore, Tokyo,Washington DC

Contents continues overleaf

Contents

1

Theresa Maybe It is stillunclear what Britain’s newprime minister standsfor—perhaps even to her:leader, page 10. The makingand meaning of a primeminister, pages 14-16. Thesudden departure of Britain’sman in Brussels lays bare thelack of Brexit plans, page 41.The “WTO option” for Britain isfar from straightforward: Freeexchange, page 55. The firstcrop of Brexit books includesentries rich in detail andanalysis, page 59

On the coverVoice technology is makingcomputers less daunting andmore accessible: leader, page7. Computers have got muchbetter at translation, voicerecognition and speechsynthesis. But they stilldon’t understand whatlanguage means: TechnologyQuarterly, after page 34

6 The world this week

Leaders7 Voice technology

Now we’re talking8 Japan’s economy

The second divine wind8 Trumponomics

Men of steel, houses ofcards

9 Fixing failed statesFirst peace, then law

10 British politicsTheresa Maybe

Letters11 On China, management,

elections, nuclear power,Japan, the elderly,economists

Briefing14 Theresa May

Steering the course

Asia17 Ageing in Japan

Cities vie for the young18 Japan’s elderly workers

Keep on toiling18 Alcohol in Indonesia

Calls for a ban19 New Zealand’s national

parksLord of the ker-chings

20 BanyanSelling Malaysians downthe river

China21 Selection year

A reshuffle looms22 Literature

Meet China’s Shakespeare

United States23 Inequality

Fat tails24 Congressional ethics

Old bog, new tricks25 Recruiting police officers

The force is weak25 Gun laws

Still standing

26 CharlestonCobblestones and bones

26 Markets for ticketsBattling bots

27 LexingtonLearning to love Trumpism

The Americas28 Brazil’s prisons

Horror in the jungle29 Bolivia

Run, Evo, run

Middle East and Africa30 South Africa’s schools

Bottom of the class31 Astronomers v sheep

farmers in South AfricaStars and baas

31 Zimbabwe’s sex tradeLess stigma, morecompetition

32 Iraq’s long warA goody and Abadi

33 America and IsraelUnsettled by Trump

34 Israel’s divisionsConvicting a soldier

Technology Quarterly:LanguageFinding a voiceAfter page 34

Europe36 Terror in Turkey

From celebration tocarnage

37 Obama sanctions RussiaPutin gets the last laugh

37 Bavaria’s angry driversTaking their toll

38 Spain and CataloniaCatalexit?

39 CharlemagneMartin Luther’s Germany

Turkey torn apartThe murderous Islamic Stateattack on a nightclub widensthe secular-religious divide,page 36

Failed states How to savenations from collapse: leader,page 9. The lessons fromAfghanistan and South Sudan,page 43. Why South Africa hasone of the world’s worsteducation systems, page 30

© 2017 The Economist Newspaper Limited. All rights reserved. Neither this publication nor any part of it may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording orotherwise, without the prior permission of The Economist Newspaper Limited. Published every week, except for a year-end double issue, by The Economist Newspaper Limited. The Economist is a registered trademark of The Economist Newspaper Limited.Publisher: The Economist. Printed by Times Printers (in Singapore).

M.C.I. (P) No.030/09/2016 PPS 677/11/2012(022861)

PEFC certified

This copy of The Economistis printed on paper sourcedfrom sustainably managedforests, recycled and controlledsources certified by PEFCwww.pefc.orgPEFC/01-31-162

Principal commercial offices:25 St James’s Street, London sw1a 1hgTel: +44 20 7830 7000

Rue de l’Athénée 321206 Geneva, SwitzerlandTel: +41 22 566 2470

750 3rd Avenue, 5th Floor, New York, NY 10017 Tel: +1 212 541 0500

1301 Cityplaza Four,12 Taikoo Wan Road, Taikoo Shing, Hong KongTel: +852 2585 3888

Other commercial offices:Chicago, Dubai, Frankfurt, Los Angeles,Paris, San Francisco and Singapore

4 Contents The Economist January 7th 2017

Subscription for 1 year (51 issues)Print onlyAustralia A$425China CNY 2,300Hong Kong & Macau HK$2,300India INR 7,500Japan Yen 41,000Korea KRW 344,000Malaysia RM 780New Zealand NZ$460Singapore & Brunei S$425Taiwan NT$8,625Thailand US$288Other countries Contact us as above

Subscription serviceFor our full range of subscription offers, including digital only or print and digital combined visitEconomist.com/offersYou can subscribe or renew your subscription by mail, telephone or fax at the details below:Telephone: +65 6534 5166 Facsimile: +65 6534 5066Web: Economist.com/offersE-mail: Asia@subscriptions.economist.comPost: The Economist Subscription Centre, Tanjong Pagar Post Office PO Box 671 Singapore 910817

Ageing Japan An olderpopulation is changingsuburbia, page 17. Japan’sworkforce is ageing, too, page18. Toshiba, an enfeebledJapanese giant, faces amulti-billion-dollar write-down, page 47. The strongdollar has given Abenomicsanother chance. Nowcorporate Japan must do itsbit: leader, page 8

Nestlé’s health kick As rivalsencroach and consumers fretabout their waistlines, theincoming boss of Switzerland’sfood multinational must find anew formula for growth, page 46

The third regime First therewas Bretton Woods; thencapital controls ended andregulations were slashed; nowcomes the third post-warfinancial regime. But whatdoes it entail? Buttonwood,page 52

China’s Shakespeare Officialshave been using the 400thanniversary of Shakespeare’sdeath to promote a Chinesebard who they claim standsshoulder to shoulder with theSwan of Avon, page 22

Britain40 Crime

How low can it go?41 Brexit preparations

Rogers and out41 Foreign aid

A stingy new year42 Bagehot

Pierogi and integration

International43 Fixing fragile nations

Conquering chaos

Business46 Nestlé

A life less sweet47 Toshiba

Losing count48 Donald Trump and Ford

Wheel spin49 Schumpeter

The three Rs of banking

Finance and economics50 Indian economics

Many rupee returns51 Impact investing

Coming of age51 Bank capital

Polishing the floor52 Buttonwood

The new global regime53 Sub-national currencies

Local difficulties53 Futures and options

Out of the pits54 Anthony Atkinson

For poorer, for richer54 Insuring talent

Death Star55 Free exchange

Brexit and the WTO option

Science and technology56 Medicine and computing

The shoulders of gAInts57 Olfactory medicine

Whiff of danger58 Atmospheric physics

The storm before the calm58 Palaeontology

Cracking a puzzle

Books and arts59 Britain and the EU

Why Brexit won60 Johnson

Word of the year61 Chinese economics

Western takeaway61 Fiction

Crazy city62 Car-park architecture

Pile ‘em in style

64 Economic and financialindicatorsStatistics on 42 economies,plus a closer look at GDPforecasts

Obituary66 Vera Rubin

Astronomy’s dark star

6 The Economist January 7th 2017

BarackObama expelled 35Russian diplomats and im-posed new economic sanc-tions in retaliation againstRussian hackers’ interferencein America’s election. Ameri-can intelligence agencies saythat Russia released stolene-mails ofDemocratic Partyofficers in order to aid thecampaign ofDonald Trump.Vladimir Putin declined tostrike back, winning praisefrom Mr Trump.

A gunman attacked a night-club in Istanbul during NewYear’s Day festivities, killing atleast 39 people. Islamic Stateclaimed responsibility. Turk-ish religious authorities whohad criticised new year’scelebrations as un-Islamiccondemned the attack. It cametwo weeks after a policemanshouting “Don’t forget Alep-po!” fatally shot the Russianambassador to Turkey.

Relations between Israel andAmerica became strainedwhen John Kerry, the soon-to-retire secretary ofstate, saidthat the Israeli governmentwas undermining the pros-pects for a “two-state solution”with the Palestinians. Hiscomments came soon afterAmerica abstained in the UNSecurity Council vote thatcriticised Israel’s constructionofsettlements.

Politicians in the DemocraticRepublic ofCongo struckadeal in which elections will beorganised in 2017 and Presi-dent Joseph Kabila will stepdown by the end of the year.Mr Kabila himselfhas notsigned the deal, however.

Argentina’s president, Maur-icio Macri, dismissed thefinance and treasury minister,Alfonso Prat-Gay. He left ap-parently because ofdisagree-ments over the structure of theeconomic team. Mr Macri splitthe finance ministry into two.Luis Caputo, the new financeminister, will be responsiblefor borrowing. A new treasuryminister, Nicolás Dujovne, willoversee tax and spending.

A battle between gangs at aprison in the Brazilian state of

Amazonas left 56 inmatesdead. Some were decapitated;severed limbs were stacked bythe entrance.

Odebrecht, a Brazilian con-struction company, and Bras-kem, a petrochemical firm inwhich it owns a stake, pleadedguilty to bribing officials andpolitical parties to win con-tracts in Latin American andAfrican countries. The compa-nies agreed to pay a penalty ofat least $3.5bn, the largestsettlement ever in a globalbribery case.

Stockmarkets had a good2016. The S&P 500 rose by10%over the 12 months and theDow Jones by13%. The FTSE100 recovered from its Brexitwobbles to end 14% up; Rus-sia’s RTS index soared after theelection ofMr Trump to finish52% higher; and Brazil’s Bo-vespa rose by 39%, despite, orbecause of, the defenestrationof the president. But Italy’smain index fell by10%, andChina’s Shanghai Compositenever fully recovered from itsturbulent start to 2016, endingthe year12% lower.

Donald Trump picked JayClayton, a legal expert onmergers and acquisitions, to bethe next head of the Securitiesand Exchange Commission.

The European Central Bankraised its estimate of the capi-tal shortfall at Monte deiPaschi di Siena to €8.8bn($9.1bn). The troubled Italianbankhas requested a bail-outfrom the government afterrunning out of time to raisenew capital privately.

Shortly before Christmas,Deutsche Bank agreed to pay$7.2bn to settle with America’sDepartment of Justice formis-selling subprime mortgagesecurities, about half theamount the regulator hadinitially sought. Credit Suisseagreed to pay $5.2bn to resolveclaims. But Barclays rejected asettlement, prompting thedepartment to file a lawsuit.

Ford made a U-turn when itscrapped plans for a newfactory in Mexico to buildcompact cars, and divertedsome of the investment to aplant near Detroit to produceelectric vehicles. Ford stressedthat this was a commercialdecision. Donald Trump hadcriticised the proposed Mex-ican factory when he cam-paigned on the theme ofsav-ing American jobs.

Meanwhile, Paul Ryan, themost senior Republican in theHouse ofRepresentatives, saidthat Congress was not going to

raise tariffs, portending whatmay be one ofhis biggest fightswith Mr Trump.

Luis Videgaray was rehabilitat-ed in Mexico’s government bybeing appointed foreign min-ister. Mr Videgaray resigned asfinance minister after suggest-ing that Donald Trump visitMexico last year, a hugelyunpopular move at the time.

Japan’s prime minister, ShinzoAbe, paid his first visit to theAmerican naval base at PearlHarbour. He expressed “sin-cere and everlasting condo-lences” to those who died inJapan’s attackon it 75 yearsago. Soon after, however, hisdefence minister, TomomiInada, paid a visit to YasukuniShrine in Tokyo where Japa-nese war criminals are hon-oured among the war dead.

The British government ap-pointed Sir Tim Barrow, aformer ambassador to Russia,as its new ambassador to theEU, three months before it isdue to trigger negotiationsover Brexit. This followed theearly exit ofSir Ivan Rogersfrom the job. His resignationnote decried “muddled think-ing” by ministers.

The world this week

Other economic data and newscan be found on pages 64-65

The Economist January 7th 2017 7

ANY sufficiently advancedtechnology, noted Arthur C.

Clarke, a British science-fictionwriter, is indistinguishable frommagic. The fast-emerging tech-nology of voice computingproves his point. Using it is justlike casting a spell: say a few

words into the air, and a nearby device can grant your wish. The Amazon Echo, a voice-driven cylindrical computer that

sits on a table top and answers to the name Alexa, can call upmusic tracks and radio stations, tell jokes, answer trivia ques-tions and control smart appliances; even before Christmas itwas already resident in about 4% of American households.Voice assistants are proliferating in smartphones, too: Apple’sSiri handles over 2bn commands a week, and 20% of Googlesearches on Android-powered handsets in America are inputby voice. Dictating e-mails and text messages now works reli-ably enough to be useful. Why type when you can talk?

This is a huge shift. Simple though it may seem, voice hasthe power to transform computing, by providing a naturalmeans of interaction. Windows, icons and menus, and thentouchscreens, were welcomed as more intuitive ways to dealwith computers than entering complex keyboard commands.But being able to talk to computers abolishes the need for theabstraction of a “user interface” at all. Just as mobile phoneswere more than existing phones without wires, and cars weremore than carriages without horses, so computers withoutscreens and keyboards have the potential to be more useful,powerful and ubiquitous than people can imagine today.

Voice will not wholly replace other forms of input and out-put. Sometimes it will remain more convenient to conversewith a machine by typing rather than talking (Amazon is saidto be working on an Echo device with a built-in screen). Butvoice is destined to account for a growing share of people’s in-teractions with the technology around them, from washingmachines that tell you how much of the cycle they have left tovirtual assistants in corporate call-centres. However, to reachits full potential, the technology requires further break-throughs—and a resolution of the tricky questions it raisesaround the trade-offbetween convenience and privacy.

Alexa, what is deep learning?Computer-dictation systems have been around for years. Butthey were unreliable and required lengthy training to learn aspecific user’s voice. Computers’ new ability to recognise al-most anyone’s speech dependably without training is the lat-est manifestation ofthe powerof“deep learning”, an artificial-intelligence technique in which a software system is trainedusing millions of examples, usually culled from the internet.Thanks to deep learning, machines now nearly equal humansin transcription accuracy, computerised translation systemsare improving rapidly and text-to-speech systems are becom-ing less robotic and more natural-sounding. Computers are, inshort, getting much better at handling natural language in allits forms (see Technology Quarterly).

Although deep learning means that machines can recog-nise speech more reliably and talkin a less stilted manner, theystill don’t understand the meaning of language. That is themost difficult aspect of the problem and, if voice-driven com-puting is truly to flourish, one thatmustbe overcome. Comput-ers must be able to understand context in order to maintain acoherent conversation about something, rather than just re-sponding to simple, one-off voice commands, as they mostlydo today (“Hey, Siri, set a timer for ten minutes”). Researchersin universities and at companies large and small are workingon this very problem, building “bots” that can hold more elab-orate conversationsaboutmore complextasks, from retrievinginformation to advising on mortgages to making travel ar-rangements. (Amazon is offering a $1m prize for a bot that canconverse “coherently and engagingly” for 20 minutes.)

When spells replace spellingConsumersand regulatorsalso have a role to play in determin-ing how voice computing develops. Even in its current, rela-tively primitive form, the technology poses a dilemma: voice-driven systems are most useful when they are personalised,and are granted wide access to sources of data such as calen-dars, e-mails and other sensitive information. That raises pri-vacy and security concerns.

To further complicate matters, many voice-driven devicesare always listening, waiting to be activated. Some people arealready concerned about the implications of internet-connect-ed microphones listening in every room and from everysmartphone. Notall audio is sent to the cloud—devices wait fora trigger phrase (“Alexa”, “OK, Google”, “Hey, Cortana”, or“Hey, Siri”) before they start relaying the user’s voice to theservers that actually handle the requests—but when it comesto storing audio, it is unclear who keeps what and when.

Police investigating a murder in Arkansas, which may havebeen overheard by an Amazon Echo, have asked the companyfor access to any audio that might have been captured. Ama-zon has refused to co-operate, arguing (with the backing ofpri-vacyadvocates) that the legal statusofsuch requests isunclear.The situation is analogous to Apple’s refusal in 2016 to help FBIinvestigators unlock a terrorist’s iPhone; both cases highlightthe need for rules that specify when and what intrusions intopersonal privacy are justified in the interests ofsecurity.

Consumers will adopt voice computing even if such issuesremain unresolved. In manysituationsvoice is farmore conve-nient and natural than any other means of communication.Uniquely, it can also be used while doingsomethingelse (driv-ing, working out or walking down the street). It can extend thepower of computing to people unable, for one reason or an-other, to use screens and keyboards. And it could have a dra-matic impactnot juston computing, buton the use of languageitself. Computerised simultaneous translation could renderthe need to speak a foreign language irrelevant for many peo-ple; and in a world where machines can talk, minor languagesmay be more likely to survive. The arrival of the touchscreenwas the last big shift in the way humans interact with comput-ers. The leap to speech matters more. 7

Now we’re talking

Voice technology is making computers less daunting and more accessible

Leaders

8 Leaders The Economist January 7th 2017

1

JAPAN’S prime minister,Shinzo Abe, was the first for-eign leader to meet Donald

Trump after his improbableelection victory. The photo-graphs show him smiling al-most as broadly as the new pres-ident-elect. But not even Mr Abe

could have guessed how much he would have to smile about.The prospect of stronger spending in America, which has

raised bond yields and strengthened the dollar against theyen, has rekindled some optimism about Abenomics, MrAbe’s campaign to lift the economy out of its decades-longstagnation. At the Bank of Japan’s most recent meeting, onepolicymaker said that the prospects for growth and reflationstand at a “critical juncture”. They likened conditions to thoseof2013 and early2014, when the currencywascheap, the stock-marketwasbuoyantand inflation wasrising. Thatmomentumwas not sustained. On its fourth anniversary, Abenomics hasfound a second wind. But this time Mr Abe must tackle theweak link in his programme: corporate Japan.

The golden hoardThe ability of Abenomics to lower borrowing costs, weakenthe yen and lift share prices was nevermuch in doubt. The pro-blem is that these gifts to Japanese industry have generateddisappointingly meagre increases in domestic investment,wages and consumption. Many firms would rather hold cashor securities than make big capital outlays (although countingR&D as investment, as Japan’s new statistics do, improves thepicture). They have also been happier paying one-off bonusesor hiring temporary workers than increasing the base pay ofcore workers, which would be harder to reverse. Abenomicshas run into a bottleneckofcorporate timidity.

Business leaders argue that Japan is an uninviting place toinvest, not least because it already has a large stock of capital,paired with a dwindlingpopulation (see page 17). But if the Jap-anese are an increasingly scarce and precious commodity, cor-porate Japan has a funny way of showing it. Despite low un-employment, real wages have declined under Abenomics.Last month the boss of Dentsu, Japan’s biggest advertisingagency, said he would resign after an investigation concludedthat overwork drove an employee to suicide. Japan’s coreworkers cannot easily be fired, but nor can they easily quit, be-cause their skills and status in a firm are not seamlessly trans-ferable elsewhere. That limits their bargaining power.

There are signsofchange. The investigation and resignationat Dentsu—like the huge losses unveiled by Toshiba, a troubledconglomerate (see page 47)—may be a paradoxical sign of pro-gress, of problems long hidden now coming to light. The com-position of Japan’s workforce is slowly changing, with greaternumbers of workers, especially women, on more flexible con-tracts that are more exposed to market forces, for better orworse. The government’s next budget will help by raising theamount that second earners, usually women, can make beforetheir spouses lose a generous tax exemption.

But a bigger shove is needed. The government ought to re-tain a tax exemption for all couples, regardless of how muchthe second earner makes. It should redesign corporate taxes todiscourage the hoardingofprofits. Ifannual wage negotiationsin the spring yield disappointing results, blunter options, likebig rises in the minimum wage, exist.

Abenomics has succeeded in stemming deflation during adifficult few years when many other big economies looked indanger of succumbing to it. If the reflationary trend of recentmonths persists, the global economy may become more sup-portive of Abenomics. But for Japan to prosper, Japan’s firmsmust swap caution for courage. 7

Japan’s economy

The second divine wind

Yen per dollarInverted scale

J A S O N D2016

J2017

120

110

100

130

TRUMP WINS ELECTION

The strong dollarhas given Abenomics anotherchance. Nowcorporate Japan must do its bit

IT MUST seem to DonaldTrump that reversing globali-

sation is easy-peasy. With a cou-ple of weeks still to go before heis even inaugurated, contritefirms are queuing up to invest inAmerica. This week Ford can-celled a $1.6 billion newplant for

small cars in Mexico and pledged to create 700 new jobs build-ing electric and hybrid cars at Flat Rock in Michigan—whilepraising Mr Trump for improving the business climate inAmerica. Other manufacturers, such as Carrier, have changedtheir plans, too. All it has taken is some harsh words, the odd

tax handout and a few casual threats.Mr Trump has consistently argued that globalisation gives

America a poor deal. He reportedly wants to impose a tariff of5% or more on all imports. To help him, he has assembled ad-visers with experience in the steel industry, which has a richhistory of trade battles. Robert Lighthizer, his proposed tradenegotiator, has spent much of his career as a lawyer protectingAmerican steelmakers from foreign competition. Wilbur Ross,would-be commerce secretary, bought loss-making Americansteel mills just before George W. Bush increased tariffs on im-ported steel. Daniel DiMicco, an adviser, used to run Nucor,America’s biggest steel firm. Peter Navarro, an economist, au-thorofbookssuch as “Death byChina” and nowan adviseron

Trumponomics

Men of steel, houses of cards

The president elect’s team needs to realise that America’s economyis not like a steel mill

The Economist January 7th 2017 Leaders 9

2

EIGHT years ago Ashraf Gha-ni and Clare Lockhart wrote

a book called “Fixing FailedStates”. Now Mr Ghani is in aposition to follow his own ad-vice. He is the president of Af-ghanistan, a state that failed inthe 1990s and could fail again. 

State failure causes untold misery (see page 43). Broadly de-fined, it is the main reason poor countries are poor. Its chiefcause is not geography, climate or culture, but politics. Somecountries build benign, efficient institutions that foster eco-nomicgrowth; othersbuild predatoryones that retard it. SouthSudan is an extreme example of predation. Its politics consistof warlords fighting over oil money. The warlords also stir uptribal animosity as a tool to recruit more militiamen. The statemakes Big Men rich while ordinary folksubsist on food aid. 

Ashes to assetsAfghanistan must overcome several hurdles to avoid the samefate. Since Barack Obama pulled out most of the NATO troopssupporting the government, the Taliban, an Islamist militia,has recaptured parts of the country. In the past year it has beenfought to a stalemate. But were Donald Trump to withdrawthe remaining American forces, the jihadists would probablytake over again. The last time they were in power they bannedfemale education, crushed gay people with bulldozers andhosted Osama bin Laden, so the stakes are high.

As a first step, Mr Trump should maintain at least the cur-rent level of air support, training and funding for the Afghanarmy. He should also ramp up pressure on Pakistan to stop let-ting the Taliban use its territory as a rear base. (Pakistan insistsit is doing all it can; no one believes it.)

Foreign military support can buy time for a fragile state tobuild the right kind ofinstitutions. This worked in Sierra Leone

and Liberia, two war-scorched African nations where UNpeacekeepers gave new governments breathing-space to startafresh. It worked in Colombia, too, where American supporthelped the government drive back the drug-dealing leftist in-surgents of the FARC and force them to the negotiating table,producing a historic peace deal in 2016. However—and this isthe lesson of Iraq—good government cannot be imposed fromoutside. National leaders have to want it and work for it, over-coming stiff resistance from the militia bosses and budget-bur-gling ministers who benefit from its absence.

Mr Ghani has the right priorities. First, establish a degree ofphysical security. Next, try to entrench the rule of law. Both arehard in a nation where suicide-bombers kill judges and war-lords grow rich from the poppy trade. Yet he has made pro-gress. The Afghan army is becoming more capable. Tax collec-tion has improved, despite the economic shock of theAmerican troop drawdown. Corruption, though still vast, isbeing curbed in some areas.

This is not a side issue. If ordinary Afghans see the state aspredatory, they will not defend it against the Taliban. Rightnow the jury is out: most Afghans are terrified of the Taliban,but trust in the government is low, too. Mr Ghani needs time toimplement his reforms; donors must be patient.

After a civil war ends somewhere, Western donors oftenpour in more money than the damaged state can absorb, andpull back when results disappoint. NGOs parachute in, poachthe best staffwith higher wages and form a costly parallel statethat will one day packup and go. This undermines national in-stitutions. It would be better if donors scaled up their largessegradually, channelled it through national coffers where possi-ble and stuckaround for the long run.

None of this will succeed if a country’s leaders do not wantit to. In South Sudan neither of the two main warlords is inter-ested in nation-building, so donors have no one to work with.But in Kabul they do. They should not cut and run. 7

Fixing failed states

First peace, then law

Howto save nations from collapse

trade, sees the decline of America’s steel industry as emblem-atic ofhow unfair competition from China has hurt America.

But the steel business is not a model for trade policy in gen-eral and companies are capable ofbeing tricksy, too. MrTrumpmay simply be looking for good headlines, but if he wantsmore, his plans threaten to be an expensive failure.

The miller’s taleOne reason is that Paul Ryan, the Speaker of the House of Rep-resentatives, said this weekthat Congress would not be raisingtariffs. Executive orders are bad politics and can get Mr Trumponly so far. Another is that Ford’s plans are not as simple asthey look. It will still build its new small car in Mexico—at anexisting plant (see page 48). But above all, Mr Trump gravelyunderestimates the complexity ofmessing with tariffs.

The men of steel are right to complain about China. Its gov-ernment has indeed subsidised its steelmakers, leading to aglut that was dumped on the world market. Successive Ameri-can governments have put up tariffs to protect domestic pro-

ducers (in 2016 the Obama administration placed a tariff of522% on cold-rolled Chinese steel), as has the European Union.

Yet this way of thinking fails to deal with the question ofwhether an ample supply of cheap steel courtesy of a foreigngovernment is really so terrible: it benefits American firms thatconsume steel—and they earn bigger profits and employ morepeople as a result. Moreover, trade in most goods and servicesis not like steel. America’s biggest import from China is electri-cal machinery. China’s government does not subsidise theoverproduction of iPhones which are then dumped on themarket, causing iPhone-makers in America to be laid-off. In-stead, a smartphone might be designed and engineered in Cal-ifornia and assembled in China, using components made ordesigned in half a dozen Asian and European countries, usingmetals from Africa. Likewise, every dollar of Mexican exportscontains around 40 cents of American output embeddedwithin it. For producers ofsuch goods, tariffs would be a costlydisaster. American steelmakers might seek out governmentprotection. Apple and its kind will not. 7

10 Leaders The Economist January 7th 2017

1

WITHIN hours of the Brexitreferendum last summer

David Cameron had resigned,and within three weeks TheresaMay had succeeded him asprime minister. The speed of herascent to power, on July 13th2016, without a general election

or a full-blown Tory leadership contest, meant that Mayismwas never spelt out in any manifesto or endorsed by the elec-torate. Yet the new prime minister soon made clear the scale ofher ambitions for Britain. Not only would she make a successof Brexit, she would also set in motion a sea-change in socialmobility to correct the “burning injustices” faced by the down-trodden, and reshape “the forces of liberalism and globalisa-tion which have held sway...across the Western world.” Her al-lies talked of an epochal moment, comparable to MargaretThatcher’s break with the past in 1979. The feeble condition ofthe Labour opposition gave Mrs May control of a one-partystate. As for her mandate, she cited the referendum: a “quietrevolution” by people “not prepared to be ignored any more”.

Yet after half a year in office there is strikingly little to showfor this May revolution (see pages14-16). The strategy for Brexit,which is due to be triggered in less than three months, remainsundefined in any but the vaguest terms, and seems increasing-ly chaotic. At home, the grand talk about transforming societyand tamingcapitalism hasyielded only timid proposals, manyof which have already been scaled down or withdrawn. Thegrowing suspicion is that the Sphinx-like prime minister isguarded about her plans chiefly because she is still strugglingto draw them up.

The emperor’s new trousersMrs May built a reputation for dogged competence during sixyears at the Home Office, a tricky beat that has wrecked manypolitical careers. She skilfully survived the Brexit referendumdespite backing the losing side. In the abbreviated Conserva-tive leadership race she stood out as the only grown-up; fewTories regret plumpingforherover the unprepared and unseri-ous other contenders. In negotiating Brexit, the hardest taskforany prime minister since the second world war, she faces apowerful drain on political capital and governmental capacity.Half the country is against the idea and the rest may sour onceits drawbacks materialise. Most of the civil servants imple-mentingBrexit thinkit a mistake. IfBritain’s next few years willbe about avoiding traps, then the wary tenacity of Mrs Maycould be just what the country needs.

Yet caution has started to look like indecision. Her most se-nior official in Brussels has just resigned, saying that the gov-ernment does not have a clear Brexit plan (see page 41). Aftersixmonths it ishard to name a single signature policy, and easyto cite U-turns. Some are welcome: a silly promise to put work-ers on company boards, for instance, was abandoned; a dread-ful plan to make firms list their foreign employees lasted lessthan a week; and hints at curbing the Bank of England’s inde-pendence were quietly forgotten. Selective “grammar”

schools will be resurrected—but only on a small scale, and per-haps not at all, given how many Tory MPs oppose the idea.Other reversals smack of dithering. The construction of a newnuclear plant at Hinkley Point was put in doubt, then given thego-ahead; a new runway at Heathrow airport was all butagreed on, then deferred until a parliamentary vote next year.“Just-about-managing” households were the prime minister’slodestar for a week or so, then dropped. So were suggestionsthat Britain would seek a transitional deal with the EU afterBrexit—until they were recirculated a few weeks later whenMrs May apparently changed her mind once again.

The cause of this disarray could be that Mayism itself ismuddled. While vowing to make Britain “the strongest globaladvocate for free markets”, the prime minister has also talkedof reviving a “proper industrial strategy”. This is not about“propping up failing industries or picking winners”, she in-sists. Yet unspecified “support and assurances” to Nissan topersuade the carmaker to stay in Sunderland after Brexitamount to more or less that. Herenthusiasm for trade often sitsuncomfortably with her scepticism ofmigration. Consider therecent trip to India, where her unwillingness to give way onimmigration blocked progress on a free-trade agreement.

A citizen ofnowhereThere is one lesson in the overdone comparison ofMrs May toThatcher. The woman who really did transform Britain had ashambolic first term; privatisation and union reform, withwhich she is now associated, did not really get going until after1983. Angela Merkel also made a shaky start as Germany’schancellor. MrsMaycould yetfind herfeet—and given the stateof Labour, she will have time to do so, if Brexit does not pro-vide her own party with a reason to oust her.

Yet Mrs May could turn out to resemble another, less obvi-ous predecessor: Gordon Brown. He, too, was thin-skinned.Like her, he moved into DowningStreetwithoutan election, in2007. He also started with a fearsome reputation and bigpromises. And when it became clear he had little idea what todo with the job he had so coveted, he flopped. The financialcrisis paralysed his government because ofhis desire to micro-manage every decision.

There is more than a little of this in Mrs May. One personcan just about run the Home Office single-handed. But beingprime minister requires delegation—especially when Brexitlooms so large. Care for the elderly is fraying. The NationalHealth Service is running out of money. A housing shortage isworsening. Scotland and Northern Ireland are raising awk-ward constitutional questions. As long as every proposal hasto be pored over by the prime minister, radical decisions of thesort needed to solve these problems will not be taken. To get agrip on Britain, Mrs May must learn to loosen hers.

For this, she must decide what the grand promises of hergovernment actually amount to. The need for every policy tobe agonised over in Downing Street, the secrecy over Brexitand the silence on the government’s broader plans for Britainall point to the same problem: Theresa Maybe does not reallyknow what she wants. 7

British politics

Theresa Maybe

Aftersixmonths, what Britain’s newprime ministerstands for is still unclear—perhaps even to her

The Economist January 7th 2017 11

Letters are welcome and should beaddressed to the Editor at The Economist, 25 St James’s Street,London sw1A 1hgE-mail: letters@economist.comMore letters are available at:Economist.com/letters

Snooping on shoppers

You shone a light on theharrowing implications of theChinese Communist Party’s“social-credit system” (“Creat-ing a digital totalitarian state”,December17th). But privateindustries, too, have imple-mented a social-credit system.Through Alibaba’s financearm, for example, SesameCredit scores people based ontheir consumption habits anddigital behaviour. The scorecan affect one’s ability to takeout a loan, buy movie tickets oreven find a significant other(dating firms often requirecourters to display their creditscores).

The technology poweringsuch systems has significantbenefits for Chinese consum-ers and the businesses servingthem. Alibaba and other Chi-nese firms make this tech-nology available to marketresearchers, who use it toassess where a likely customerlives, where they typicallyshop and how much it costs toget them to a store. This hasspurred significant investmentin China from multinationalsthat want a slice of its retail pie,and has also helped Chinabecome the largest retaile-commerce market in theworld. JOE NORAMarketing directorExport Now Digital SolutionsShanghai

Ideas management

One may easily take issue withSchumpeter, who believes thatmanagement theorists havegone astray by subscribing tothe dead ideas of increasingcompetition, widespreadenterprise, the growing speedofbusiness operations andglobalisation (December17th).For anecdotal evidencedisproving Schumpeter, justglance at the article thatpreceded his column. It wasabout the competitive successofZara, a highly entrepreneur-ial company with a globalfootprint, and the edge it hasattained by adjusting its cloth-ing lines in lightning speed tothe most current fashion(“Behind the maskofZara”,

December17th).Ranging more widely, the

bone-breaking changes in suchindustries as retail and mediaare now reaching finance, withfintech. Ford will become aninformation-technologycompany competing at thatindustry’s speed, using itsautonomous cars and theservices enabled by theinternet of things. The cheapglobal connectivity of theinternet, combined with thelarge and increasing share ofinformation and knowledge inproducts, will obviate anypolitical moves towardsautarky. VLADIMIR ZWASSEditor-in-chiefJournal of Management Information SystemsSaddle River, New Jersey

Election advice for Italy

Why do you recommendfirst-past-the-post elections inItaly (“Salvaging the wreck-age”, December10th)? It is aninherently undemocraticvoting system. Take the mostrecent British general election.In 2015 the Conservative Partywon 330 seats with only 37% ofthe total vote, giving it a major-ity government without anactual electoral majority. TheUK Independence Party gotjust one seat with 13% of thevote, whereas the ScottishNationalists secured 56 seatswith 5% of the vote.

The single-transferablevote, used in Ireland andMalta, is a better system,because it reflects the will ofthe electorate and keepspoliticians more in tune withtheir constituents. MICHAEL RYANDublin

Nuclear v solar

I doubt that the El RomeroSolar Plant in the Chileandesert would power a city of amillion people (Bello, Decem-ber10th). In fact, it wouldpower120,000 Chilean house-holds today, and far fewer inthe future, if the forecasts ofrapid growth in demand mate-rialise. Globally, electricityconsumption far outpaces newsolar and wind power. Car-

bon-free electricity generationas a percentage ofoverallgeneration has fallen. This isexplained by both the declineofnuclear power and thefailure of renewables to makeup the difference.

In the United States alone,five nuclear plants have closedover the past several years.Together they generated asmuch electricity as all ofAmer-ica’s solar plants and resi-dential installations put to-gether. Many more nuclearplants are at riskofclosing inthe Western hemisphere with-out any replacement in sight.

Clean electricity is likely tocontinue declining for years tocome. Policymakers have beenslow to realise that the man-dated purchases ofheavilysubsidised renewables havedepressed electricity prices.Even with very low fossil-fuelprices, ageing nuclear plants,which often have remaininglifetimes longer than new solarand wind facilities, are at adisadvantage. Yet they also donot pollute.CESAR PENAFIELNew York

Japan’s broadside

Lexington mentioned thatJapan’s new destroyer isnamed the Izumo (December10th). The original Izumo wasan armoured cruiser thatserved as the Japanese navy’sflagship in China in the 1930sand 1940s. She saw battle inboth the 1932 and 1937Sino-Japanese wars, shellingChinese positions from themiddle of the Huangpu river inShanghai. She also sankthelast British gunboat andcaptured the last Americangunboat in Shanghai in 1941.

By giving the new Izumoher name, and, indeed, nam-

ing the entire class ofships theIzumo class, Japan is sending aclear message to China.DOUG CLARKHong Kong

With winterhere…

The British government’sresponse to the crisis in care forthe elderly is, as you say, “Toolittle, too late” (December17th).You are also right that fundingservices for old people throughlocal-government taxes oftenleaves the councils that need itthe most with the least cash.But there is an even greaterdefect in the system.

Responsibility for care ofthe elderly is divided betweenthe National Health Serviceand local councils, and theirinterests are usually diamet-rically opposed. Every elderlyperson who has to remain inhospital because there is nospace in a care home is afinancial gain for the councilbut a considerable cost for thehospital (as well as denying abed to someone who needs it).The only way to resolve thisconflict of interest is to putsocial care in the communityunder the control of the NHS.This is perfectly logical as it is anational “health” service not anational “hospital” service.DAVID TERRYDroitwich, Worcestershire

Grouping economists

Professor Ben-Gad answeredyour call for a collective nounfor economists with the admi-rable suggestion of“aggregate”(Letters, December17th). Butgiven the befuddling diversityofeconomic mantras andeconomists, that suggestionrisks mixing apples andoranges.

There must be at least twoother collective nouns foreconomists: an inefficiencyand a disutility.DONALD NORBERGSturminster Newton, Dorset 7

Letters

14 The Economist January 7th 2017

1

AS A student at Oxford, Theresa Maylooked like a typical ambitious young

Tory. The daughter of a vicar, she had beenstuffing envelopes for her local Conserva-tive association for years. She was a mem-ber of the Oxford University ConservativeAssociation; it was at one of its discos thatBenazir Bhutto, later the prime minister ofPakistan, introduced her to the man shewould marry. She also joined the OxfordUnion, a debating society where politi-cians in embryo learn to speechify, ingrati-ate themselves and stab each other in theback. She told a tutorial partner that shewanted to be prime minister.

Yet various things distinguished herfrom the classic Tory hack. For one, she didnot read philosophy, politics and econom-ics (PPE), the course designed to train futureelites. She read geography. For David Wil-letts, who was minister for universities inthe 2010-15 coalition government in whichMrs May was home secretary, this distinc-tion is more than incidental.

He notes thatPPEists (like David Camer-on, Mrs May’s predecessor, and indeedLord Willetts) tend to concentrate on Brit-ain’s sectoral strengths—its booming ser-vice industries, its great universities, theCity—whose success might trickle down topoorer areas, or into whose orbit residentsof poorer areas might be persuaded to

move. By contrast Mrs May cares aboutplaces, their preservation and people’s at-tachment to them, an attitude whichmakes her particularly concerned withdown-and-out areas that need help pick-ing themselves up.

In this she is well-suited to her times.Britain’s vote for Brexit (the responsibilityfor whose realisation she inherited fromMr Cameron) was partly a cry ofprotest byparts of the country that felt left behind, ex-cluded from its successes, oroverwhelmedbyrapid change. It showed howmuch peo-ple’s sense ofbelonging in the place wherethey live mattered to them, and the valuethey placed on stability and order. Theprime minister’s talk of reviving manufac-turing, reducing immigration and tacklingcorporate excess plays well to such feel-ings. The public likes her considerably bet-ter than it did Mr Cameron two years intothe previous parliament, and much betterthan the lamentably led Labour Party (seechart on next page). In a YouGov poll pub-lished on January 3rd, every region, everysocial class and every age group said shewould be a better prime minister than Je-remy Corbyn, the Labour leader.

The outlook, education and characterof a leader always matter; but with MrsMay they matter more than usual. Mostprime ministers travel on tracks of tradi-

tion, convention and precedent. The legal,political, economic and diplomatic com-plexities of Brexit have put paid to that. Acostly and possibly bitter divorce must benegotiated. Trade deals with the remain-der of the EU, and possibly the rest of theworld, must be struck. A new immigrationregime must be established, economicshockscontained, partners reassured, Scot-land held in the union, peace in NorthernIreland preserved and painful fractures inBritish society closed. There are no prece-dents. It is forMrs May to create her own; tomake choices that dwarfmost ofthose thatconfronted her predecessors.

A prime minister who had won a gen-eral election, or even a contested partyleadership campaign, would have had togive some sense of how she would makesuch choices. But Mrs May has done nei-ther of those things. Thus for an idea ofhow she reads the lay of the unknownland ahead, and how adept she will proveat navigating it, it pays to look closely atwho she is and where she came from.

Onward Christian soldiersMrs May was born in 1956 to the ReverendHubert Brasier and his wife Zaidee. Whenshe was a girl her father became vicar of StKenelm’s in Church Enstone, a cinemati-cally idyllic huddle ofgolden stone housesamid the drystone walls and rolling fieldsof the Cotswolds. Her ecclesiastical up-bringing has prompted comparisons toAngela Merkel (whose father was a Luther-an pastor in East Germany) and GordonBrown, Tony Blair’s successor as Labourprime minister (whose fatherwasa Presby-terian minister in Fife, near Edinburgh). Allthree grew up in households dominated

Steering the course

CHURCH ENSTONE AND MAIDENHEAD

The making and meaning ofa prime minister

Briefing Theresa May

The Economist January 7th 2017 Briefing Theresa May 15

1

2 by the moral and practical duties imposedby the life of the church; all were therebyfurnished with an unflashy, serious andcautious character.

Her vicarage childhood lives on in MrsMay’s very English traits. She drinks EarlGrey tea, reads Jane Austen, watches JamesBond films, regularly attends church in herconstituency (Maidenhead, a posh town inthe Thames valley) and adores cricket. Ech-oes of this can be seen in her leadership.Anglicanism often combines stormy, king-dom-of-God language with a restrainedconservative culture: hymns about crusad-ers and the devil belted out before tea andbiscuits. In herfirst months as prime minis-ter Mrs May, too, has been bolder in herrhetoric than in her actions—big ideas havereceived little follow-through, or beendropped altogether. There is a touch of hercricketing hero, Geoffrey Boycott, abouther too. It is hard not to detect her admira-tion for the stolid style of the Yorkshirebatsman in her matter-of-fact demeanour.When her aides say “She just gets on withthe job” it is the sort of praise their bosswould like.

A social reformism rooted in her Angli-can upbringingand practice (“partofwho Iam and therefore how I approach things”,she has said) has been a constant of her ca-reer. When the voters of Maidenhead firstsent her to Westminster in 1997 she was, inthis respect, to the left of her party. In 2002she warned her colleagues and their sup-porters that they had become known as“the nasty party”. The following year, asshadow transport minister, she argued formore state intervention in the economy, amore nuanced relationship with trade un-ions and limits on fat-cat excesses.

All of this lives on in her premiership.When, having lost the Brexit referendum,Mr Cameron resigned, Mrs May enumerat-ed the inequities of modern Britain as shelaunched her campaign to succeed him:boys born poor die nine years earlier thanothers; children educated in state schoolsare less likely to reach the top professionsthan those educated privately; manywomen earn less than men.

When she became prime minister sherepeated some of these “burning injus-tices” on the steps of Downing Street. Shehas talked up a newgeneration ofstate-rungrammar schools (schools, like the one sheattended, that are allowed to select theirpupils through competitive exams) to giveclever children from poor backgrounds aleg up. She has hinted at worker represen-tation on companyboards; she has lament-ed the effect of the Bank of England’s lowinterest rates on savers.

Mrs May patently stands apart frommany of her colleagues in ways that go be-yond this reformism; there is a social dis-tance, too. Some say it has to do with theisolating shock of losing both of her par-ents when she was relatively young. Oth-

ers cite her experience of diabetes—theprime minister must inject herself with in-sulin several times a day. But the best ex-planation is her career as a woman educat-ed at a provincial grammar-school (thegranddaughter of domestic servants, noless) in a party dominated by public-school boys given to cavalier confidenceand clever-clever plans. When her alliespraise Mrs May’s methodical style and herdisdain for chummy, informal “sofa gov-ernment”, they are channelling her long-held exasperation with the know-it-allposh boys—particularly Mr Cameron andGeorge Osborne, his chancellor.

The prime minister has little time forthe parliamentary village, avoiding its barsand tea rooms, decliningdinner-party invi-tations in London—let alone in Brussels, orWashington, DC. She is the opposite of cos-mopolitan. “If you believe you’re a citizenof the world, you’re a citizen of nowhere,”she told her party conference in October.She struggles with the small talk that oilsdiplomatic (and cabinet) wheels. The Euro-pean Council summit on December 16thsaw the prime minister fiddling awkward-ly with her cuffs as fellow leaders air-kissed behind her. She is far more at homein her constituency on the banks of theThames. Her house in the village of Son-ning sits by what Jerome K. Jerome, a Vic-torian humorist, described as “the mostfairy-like little nook on the whole river”.Here, in her natural habitat, she is by all ac-counts witty, relaxed and gregarious.

Ordering theirestateMrsMay’s time running the Home Office, adepartment institutionally obsessed withorder and control, earned her a reputationfor inscrutability, formality and obsessionwith detail (“she was always asking formore papers in her red box,” says one lieu-tenant). She worked well with people withwhom she had things in common, likeLynne Featherstone, the Liberal Democratminister whose commitment to introduc-

ing gay marriage she shared. But she ex-cluded and ignored those—like JeremyBrowne and Norman Baker, Ms Feather-stone’s two successors in the department—with whom she did not.

She clashed with Michael Gove, thenthe education secretary, over measures todeal with extremism in schools and withMr Osborne over immigration—she want-ed to tighten up Britain’s student visa re-gime. She was typicallyone ofthe last min-isters to agree on her department’s budgetin the annual financial round. She also hada run-in with Boris Johnson, then mayor ofLondon, over three water cannon hebought without seeking the Home Office’snecessary—and, in the event, withheld—approval. The incident serves her inner cir-cle as a house parable showing the perfidyof civil servants (who talked Mr Johnsoninto the idea), the folly of ill-scrutinised de-cisions, the danger of informal structuresand the comeuppance ofthose who do notdo things Mrs May’s way.

In Downing Street Mrs May has im-posed the centralised, formal workingpractices that she honed at the Home Of-fice. The day is governed by the 8.30ammeeting, a shoeless free-for-all under MrCameron that now has a strict invitationlist. Blue-sky thinking and speculationabout the headlines that evening are out;firm instructions to staffers are in. In theprime minister’s office a table and chairs(and vases of hydrangeas) have replacedthe sofa. Ministers and staffers must sub-mitpapersearlier than underMrCameron,to allow her to work through them late inthe evening (he would do them the nextday). The whole machine is run by a small,powerful team centred on her two chiefs-of-staff, Fiona Hill and NickTimothy.

Cabinet and sub-cabinet meetings arevenues for serious discussion, not Potem-kin forums with pre-decided outcomes.Having for the most part distributed minis-terial portfolios evenly between Leaversand Remainers, Mrs May appointed three

The one-party state

Sources: ICM; Ipsos MORI; Opinium; YouGov

Britain, 2016

10

30

50Theresa May

Theresa May becomes prime minister

Mrs May backs newrunway at Heathrow

Mrs May retreatsfrom workers-on-boards pledge

Jeremy Corbyn

Labour leadership contest

Brexit vote

Article 50Supreme

Court case

EuropeanCouncilSummit

21 Labourfrontbenchersresign

Conservative party conference.Pledge to trigger Article 50by April 2017

July August September October November December0

20

40

60

treers-dge

June

Who would do better as prime minister?% replying:

VOTING INTENTION% replying:

High Court rulesParliamentshould voteon Article 50

CONSERVATIVE

LABOUR

Tory leadershipcontest

16 Briefing Theresa May The Economist January 7th 2017

2 people who, unlike her, campaigned forBrexit to the departments most concernedwith bringing it about—Mr Johnson to theForeign Office, Liam Fox to a new Depart-ment for International Trade and DavidDavis to a new Department for Exiting theEU. Giving the Brexit-related jobs topaid-up Brexiteers insulates her from criti-cisms of not supporting the policy. It alsocannily reduces the chance ofa single Brex-iteeremergingas a rival if the process’s out-come disappoints the diehard Leavers.

One minister says that, whereas thecabinetsofMrBlairand MrBrown were fu-rious power struggles, and Mr Cameron’scabinets mostly shams, Mrs May’s cabinetfeatures open discussions in which theprime minister really listens. Anotherclaims that she is more interested in evi-dence than her predecessor was andpraises the fluency with which she shiftsbetween subjects. Acolytes insist that themighty chiefs-of-staff produce decisionsthat have been properly tested (not so un-der Mr Cameron) without prime ministeri-al overload (not so under Mr Brown).

Most ofall, though, these arrangementsgive the prime ministerwhat she most cov-ets: control. Even close allies call Mrs May acontrol freak—and as is often the case, thefreakery comes at the expense of trust andefficiency. The “Nick and Fi” filter on poli-cies creates a bottleneck delaying urgentmeasures (new funding to soothe the so-cial-care crisis was unveiled almost amonth later than planned). Apparent pri-orities—like those grammar schools—havefailed to turn into flagship policies. Thesuggestions of workers on boards, govern-ment meddling in monetary policy andobligations on firms to list their foreignworkers have all come to nothing. More re-grettably, so have hints of big new infra-structure investments and house-buildingschemes. Westminster feels dead.

Comments by ministers have been dis-owned, the Treasury feels sidelined, dip-lomats believe they are ignored. When aconsultant’s memo to the Cabinet Officecriticising Mrs May’s leadership styleleaked, the prime minister reportedly de-manded that Deloitte, the firm in question,be “punished”. It has since withdraw froma series of bids for government contracts,and ministers’ e-mails and phone recordsare to be seized to prevent further leaks.Even the queen has reportedly grumbledabout Mrs May’s slogan-heavy furtivenessabout how Britain will leave the EU.

Indeed, six months after coming topower all the prime minister can say onthat subject is that “Brexit means Brexit”and that it will be “red, white and blue” (iepatriotic, rather than Caucasian, bloodiedand bruised). Her fear of losing control ex-plains why, instead of holding a simpleparliamentary vote on triggering Article 50of the EU Treaty (the process by which Brit-ain will leave the union), she stubbornly

plunged into a legal bunfight to prevent it.As the Deloitte memo put it, she seems tohave no coherent plan for Brexit, her gov-ernment is “struggling” and still she isprone to “drawing in decisions and detailsto settle matters herself”.

Some confirmation of this came on Jan-uary 3rd when Sir Ivan Rogers, Britain’sambassador to the EU, left his job tenmonths early. In a leaked e-mail he tookaim at “muddled thinking” on Brexit (seepage 41). He is not the first senior civil ser-vant to leave early; Helen Bower, the re-spected chief spokeswoman at 10 Down-ing Street, went first. A senior minister inthe upper house, Jim O’Neill, has alsowalked out.

All of which is a reminder that, al-though the Labour Party’s disarray makesMrs May look unassailable, her position isnot entirely safe. She has a very small par-liamentary majority and the ConservativeParty has a knackfor regicide. It looks quitelikely that the Brexit talks will founder; MrsMay insists that she wants to maintain cer-tain economic benefits of EU membershipbut end free movement of labour, a dealdeemed unthinkable in Brussels. Thatcould lead to economic chaos and exposeher to a challenge from MrOsborne, who isremaking himself as the backbench stan-dard-bearer for liberal Toryism. Alterna-tively, a final deal could involve trade-offsunpalatable to her most keenly BrexiteerMPs, who would then cut up rough.

When things start to go south the defen-sive and needlessly belligerent toneshown in her tenure to date will serve herill. For most of her end-of-term grilling bythe liaison committee—a panel of MPswhich scrutinises the government—shewore an aquiline scowl, quibbling withthe questions and, when pushed, cleavingto evasive platitudes: “I gave the answer I

gave.” MrBoycott, one feels, mightapprovesuch dogged defensiveness; but fewwouldlook to him for lessons on team building.

On coming to power it was not enoughfor Mrs May to fire Mr Osborne and MrGove: she capriciously gave each a dress-ing down in the process. Close observerssay she is allergic to cutting deals and thatin cabinet she sees eye-to-eye only withministers who, like Philip Hammond, herchancellor, and Damian Green, her wel-fare secretary (and the husband of her Ox-ford tutorial partner), she has known fordecades. Her sporadic attempts to lightenup are hit-and-miss: her frequent publicmockery of Mr Johnson is making an ene-my of him—and feels weird coming fromthe woman who gave him his powerfuljob in the first place.

Many a conflict, many a doubtThere may be lessons as to Mrs May’s pos-sible longevityand success from herfellowchildren of the cloth, Mr Brown and MrsMerkel. Mr Brown, whose brief premier-ship was dominated by the global finan-cial crisis, never unified his party and wasup against a strong opposition led by MrCameron. Mrs Merkel has faced crises,too—but for more than a decade has grownthrough them, outwitting or co-opting heropposition, maintaining unquestioned su-premacy in her party.

Like Mrs Merkel, Mrs May has seen offrivals through canny manoeuvring; shebides her time, knowing when to speak upand when (as in the referendum cam-paign) to stay quiet. Like Mr Brown, she isprone to overblown rhetoric, irritabilityand indecisiveness. The biggest worry,though, is that she may also share his in-ability to adapt—the key difference be-tween Mr Brown and Mrs Merkel.

Mrs May shows few signs of the abilityto assimilate the new that has made MrsMerkel so successful. Her vision of leader-ship, it seems, is focused on giving state-ments, installing processes, gathering upinformation and control—and little else.This makes it worryingly easy to imaginethe Britain of 2018 or 2019 in disarray: herparty in revolt, her ministers and partnersalienated, her government sclerotic, Brexittalks breaking down, the economy tankingand Number10 in bunker mode.

For there is more to leadership than MrsMay’s procedures. There is also what PeterHennessy, a contemporary historian, calls“the emotional geography” of power. Thismeans adapting to events and institutions,building networks and—yes—being judi-ciously informal sometimes: a dose of in-stinct, a snap decision, a deal cut, a risk tak-en on a wingand a prayer. Itmeanssharinginformation, accepting dissent, seeking al-ternative opinions, staking out a positionand persuading people of it. It is this emo-tional landscape that Britain’s geographerprime minister must master, if she can. 7Very well, alone

The Economist January 7th 2017 17

For daily analysis and debate on Asia, visit

Economist.com/asia

1

MIEKO TERADA moved to Tama in1976, at about the same time as every-

one else there. Back then, the fast-growingcity in Tokyo’s suburban fringe was busywith young married couples and children.These days, however, the strip of shopswhere Ms Terada runs a café is deathly qui-et, herclientele elderly. The people ofTamaand their apartments are all growing oldand decrepit at the same time, she says.

In the mid-1990s Japan had a smallerproportion of over-65s than Britain or Ger-many. Thanks to an ultra-low birth rate, ad-mirable longevity and a stingy immigra-tion policy, it is now by far the oldestcountry in the OECD. And senescence isspreading to new areas. Many rural Japa-nese villages have been old for years, be-cause young people have left them for cit-ies. Now the suburbs are greying, too.

Between 2010 and 2040 the number ofpeople aged 65 or over in metropolitan To-kyo, of which Tama is part, is expected torise from 2.7m to 4.1m, at which point one-third of Tokyo residents will be old. InTama, ageing will be even swifter. Thenumber of children has already droppedsharply: its city hall occupies a formerschool. Statisticians think the share of peo-ple over 65 in Tama will rise from 21% to38% in the three decades to 2040. The num-ber ofover-75s will more than double.

The city’s inhabitants have alreadybeen spooked by an increasing number of

square-metre apartments in the old blockswere sufficient for the post-wargeneration,modern Japanese families demand morespace. Tama’s authorities intend to trans-form other districts in a similar way.

This is smart policy, but there is a pro-blem with it. The number of20- to 29-year-olds in Japan has crashed from 18.3m to12.8m since 2000, according to the WorldBank. By 2040 there might be only10.5m ofthem. Cities like Tama are therefore play-ing not a zero-sum game but a negative-sum game, frantically chasing an ever-di-minishing number of young adults andchildren. And some of their rivals have ex-tremely sharp elbows.

Follow the Tama river upstream, intothe mountains, and you eventually reach atiny town called Okutama. What Tama istrying to avoid has already happenedthere. Okutama’spopulation peaked in the1950s, as construction workers flocked tothe town to build a large reservoir that sup-plies water to Tokyo in emergencies. It hasgrown smaller and older ever since.

Today 47% of people in the Okutamaadministrative area—the town and sur-rounding villages—are 65 or older, and 26%are at least 75. Children have become soscarce that the large primary school is onlyabout one-quarter full. Residents in their70s outnumber children under ten bymore than five to one (see chart, next page).

And Okutama’s residents are as stub-born as theyare long-lived. Some ofits out-lying villages have become so minusculethat providing them with services is diffi-cult, says Hiroki Morita, head of the plan-ning and finance department. It would bebetter for their residents, and certainly bet-ter for the local government, if they con-solidated into larger villages. But old peo-ple refuse to leave their shrunken hamletseven during heavy snowstorms, and are

confused old people wandering around.By 2025, officials in Tama predict, almostone in four elderly residents will be bed-ridden and one in seven will suffer fromdementia. And the city is hardly ideal forold people. It is built on steep hills, and thefive-storey apartment blocks where manyof the residents live do not have lifts.

For Tama, though, the most worryingeffectsofageingare fiscal. Two-thirds ofthecity’s budget goes on social welfare, whichold people require lots of. They do not con-tribute much to the city’s coffers in return.Although Japan’s central government re-distributes money between municipal-ities, much of what local governmentsspend comes from local residency taxes,which fall only lightly on pensioners. Inshort, says Shigeo Ito, the head of commu-nity health in Tama, it pays for a place toavoid growing too old.

Tama’s enticementsSo, as well as providing more in-home careand laying on aerobics classes to keep peo-ple fit enough to climb all those stairs,Tama is once again trying to lure youngfamilies. With a developer, Brillia, it has al-ready razed 23 five-storey apartmentblocks and put up seven towers in theirplace. The number of flats in the redevel-oped area has almost doubled, and manyare larger than before. That has attractednew residents: although the poky 40-

Demography in Japan

A negative-sum game

TAMA

The ageing of Japan’s population is changing its cities. The first of two stories looksat the impact on suburbia

AsiaAlso in this section

18 Japan’s elderly workers

18 Indonesia debates a booze ban

19 New Zealand’s national parks

20 Banyan: Malaysia’s political puzzle

18 Asia The Economist January 7th 2017

1

2 unlikely to move permanently just tomake a bureaucrat’s life easier. The inter-net and home delivery help them cling on,points out Mr Morita.

Okutama has tried to promote agricul-ture: wasabi, a spicy vegetable that isground up and eaten with sushi, growswell there. It hopes to appeal to families byoffering free vaccinations, free schoollunches and free transport. None of thathasstaved offageingand decline. So nowitis touting free housing. Mr Morita esti-mates that the town has about 450 emptyhomes. He wants the owners to give theirhomes to the town government, whichthey might do in order to avoid propertytaxes. The government will then rent thehomes to young couples, the more fecundthe better. If they stay for15 years their rentwill be refunded.

Although its setting, amid steep hills, isspectacular, Okutama is not a pretty town.Itshousesare neitherold enough to be con-sidered beautiful nor modern enough tobe comfortable. Some feature post-warwheezes like plastic siding. Still, the pros-pect of free accommodation some twohours’ journey from central Tokyo mighttempt some young families. And in themeantime, Okutama has another plan.

A building once occupied by a juniorhigh school, which closed for lack of pu-pils, is becoming a language college. Jelly-fish, an education firm with tentacles inseveral countries, will use it to teach Japa-nese to young graduates from East andSouth-East Asia. It hopes to enroll 120 stu-dents, plus staff, which ought to make a no-table difference in a districtwhere there arenow fewer than 350 people in their 20s.Some of those students might even decidethey like the place, and settle down. Whis-per it, but this sounds a little like a more lib-eral immigration policy. 7

Where are the kids?

Sources: Okutama town records; Statistics Japan

Population of Okutama and JapanFive-year age groups, 2015, % of total

0 3 6 936912

95+90-9485-8980-8475-7970-7465-6960-6455-5950-5445-4940-4435-3930-3425-2920-2415-1910-145-90-4

OKUTAMA, TOKYO PREFECTURE JAPAN

LIKE many firms in Aichi prefecture, Ja-pan’s manufacturing heartland, Nishiji-

max, a maker of machine tools for the carindustry, is struggling to find workers. Itssolution in a country with a drum-tight la-bour market is one that is increasinglycommon in Japan: raising the age of retire-ment. More than 30 of the company’s 140employees are over60; the oldest is 82. Put-ting qualified people out to pasture early isa waste, says Hiroshi Nishijima, a manag-er; “If they want to work, they should.”

Since peaking at over 67m in the late1990s, Japan’s workforce has shrunk byabout 2m. The government says it couldcollapse to 42m by mid-century as the pop-ulation ages and shrinks. The number offoreigners inched up in 2015 to a recordhigh of 2.2m, but that is far from enough tofill the labour gap. Instead of opening itsdoors wider to immigrants, Japan is tryingto make more use of its own people whoare capable ofworking.

Large companies in Japan mostly set amandatory retirement age of 60—mainlyas a way of reducing payroll costs in a sys-tem that rewards seniority. But other busi-nesses are less stringent. About12.6m Japa-nese aged 60 or older now opt to keepworking, up from 8.7m in 2000. Two-thirdsof Japan’s over-65s say they want to staygainfully employed, according to a govern-ment survey. The age of actual retirementformen in Japan isnowclose to 70, says theOECD, a rich-country think-tank. In mostcountries people typically stop workingbefore the age at which they qualify for astate pension. Japan, where the state pen-sion kicks in at 61 (it is due to rise to 65 by2025), is a rare exception.

The greying of Japan’s workforce isclearly visible. Elderly people are increas-ingly seen driving taxis, serving in super-markets and even guarding banks. Bossesare getting older, too. Mikio Sasaki, thechairman of Mitsubishi Corporation, atrading company, is 79. Masamoto Yashiro,the chairman and CEO of Shinsei Bank, is87. Tsuneo Watanabe, editor-in-chiefof theworld’s biggest-circulation newspaper, theYomiuri Shimbun, is a sprightly 90.

It is inevitable that people will stay inthe workforce longer, says Ken Ogata, thepresident of Koreisha, an agency that pro-vides temporary jobs exclusively to peopleover 60. He notes that the country has littleappetite for importing workers, so it willhave to make more use of pensioners,women and robots. Many of those who

find work through Koreisha were once em-ployees of Tokyo Gas, Japan’s largest sup-plier of natural gas to homes. They do thesame kind of work now—reading metersand explaining the use of appliances tohomeowners. “They have so much experi-ence and knowledge that can be put togood use,” says Mr Ogata.

They can also be cheaper. Companiesoften hire back retirees on non-permanentcontracts offering poorer terms than theirprevious ones. Takashimaya, a depart-ment-store chain, has introduced a perfor-mance-based system for such employeesaged 60-65 (at no extra cost to the company,it says).

Japan’s labour crunch has created achronic shortage ofnursing care for elderlypeople who are no longer fit enough towork. McKinsey, a consultancy, says Japanshould encourage able-bodied elderlypeople to help. If 10% of them were to takeup such work, the country would have anadditional 700,000 carers by 2025, it reck-ons. One way of encouraging this wouldbe to give priority to those who haveworked as carers when allocating places innursing homes, says McKinsey. It does nothelp, however, that the state pension sys-tem discourages some elderly people fromworking by cutting their benefits if theyearn more than a certain amount.

At Nishijimax, managers clearly wantelderly workers to stay. The company’swork routine is tailored to their needs. So,too, are the canteen’s offerings—rightdown to the reduced-salt miso soup. 7

Japan’s elderly workers

Silver lining

TOKYO

As Japan ages, so too does its workforce

ONE of Indonesia’s newest brands ofbeer, Prost, traces its ancestry back to

1948 when Chandra Djojonegoro, a busi-nessman, started selling a “health tonic”,known as Anggur Orang Tua, from theback of a bright-blue lorry at night marketsin the coastal city ofSemarang. A troupe ofdancing dwarves would pull in the punt-ers, while Djojonegoro peddled shots ofwhat was, in essence, a fortified herbalwine to fishermen. It kept them warm dur-ing the chilly nights in the Java Sea.

The tonic is still sold in bottles with dis-tinctive labels depicting an old Chineseman with a thick white beard. The com-pany that makes it now produces a vastrange of consumer goods, and Prost beer isthe latest addition to its range. It is made ina $50m brewery that opened in August2015, filled with shiny stainless-steel ma-

Alcohol in Indonesia

Dry talk

SEMARANG

Debating a ban on booze

The Economist January 7th 2017 Asia 19

2

NEWZEALAND’Schiefconservation of-ficer, LouSanson, caused a stir in Octo-

ber by suggesting that it might be time tostart charging tourists for using the coun-try’s wilderness trails. New Zealanders arekeen fans of their national parks. Manywould be outraged at having to pay. Butmany also worry about a huge influx offoreigners who have been seeking thesame delights.

In 2016 New Zealand hosted 3.5m tour-ists from overseas; by 2022 more than 4.5mare expected every year—about the sameas the country’s resident population. Tou-rism has overtaken dairy produce as thebiggest export, helped by a surge in thenumberofvisitors from China. The nation-al parks, which make up about one-thirdof the territory, are a huge draw. About halfof the foreign tourists visit one. They arekeen to experience the natural beautypromised by the country’s “100% PureNew Zealand” advertising campaign (andshown off in the film adaptations of “TheLord of the Rings” and “The Hobbit”,

which were shot in New Zealand’s breath-taking wilderness).

But for every happy Chinese couplesnuggling up for a selfie next to a tuatarathere is a grumpy New Zealander who re-members the way things used to be—whenyou could walk the tracks without runninginto crowds at every clearing. Many localsnow wonderwhy their taxes, as they see it,are paying for someone else’s holiday. MrSanson would seem to agree. Entry feescould be used to upgrade facilities such ascabins, car parks and trails. A varying levycould also help reduce numbers at some ofthe popular locations by making it cheaperto use lesser-known, but no less beautiful,trails farther afield.

Some are not so sure it would work.Hugh Logan, a former chief of conserva-tion for the government who now runs amountaineering club, worries it wouldcost too much to employ staff to take mon-ey from hikers at entrances. It would alsobe difficult to prevent tourists from sneak-ing around the toll booths.

Some argue that it would be easier tocharge visitors a “conservation tax” whenthey enter the country. The Green Party, thethird-largest in parliament, says that add-ing around NZ$18 ($12.50) to existing bor-der taxes would still make the totalamount levied less than visitors to arch-ri-val Australia have to pay. But some travelcompanies oppose the idea. They note thattourists already contribute aroundNZ$1.1bn through the country’s 15% salestax. Better, such firms say, to use foreigntourists’ contribution to this tax for themaintenance of the parks.

Amongthe fiercest criticsofa charge arethose who point out that unfettered accessto wilderness areas is an important princi-ple for New Zealanders. It is enshrined in aNational Parks Act which inspires almostconstitution-like devotion among thecountry’s nature-lovers. Mr Sanson has arocky path ahead. 7

New Zealand’s national parks

Lord of theker-chings QUEENSTOWN

A proposal to taxusers ofnational parkshas aroused fierce argument

Welcome to Orcland

chinery from Germany. Thomas Dosy,chief executive of the subsidiary that pro-duces Prost, says that given Orang Tua’shistory in the booze business itwasnaturalfor the company to move into Indonesia’s$1bn-a-year beer market.

It will not be straightforward. Conser-vative Muslim groups have become moreassertive. Onlymonthsbefore the breweryopened, the government slapped a ban onthe sale of beer at the small shops wheremost people buy their groceries. It led to a13% slump in sales, according to Euromon-itor, a research firm. The government min-ister who issued the decree has since beensacked, but his ban remains in place. AndMuslim parties in parliament are still notsatisfied. They are pushing legislation thatwould ban the production, distributionand consumption of all alcoholic bever-ages. Drinkers could face two years in jail.

The law is unlikely to pass. Muslim par-ties control less than one-third of the legis-lature’s seats. The government is propos-ing a far more limited law aimed at curbingthe production of toxic home-brews,known as oplosan, which are responsiblefor nearly all alcohol-related deaths in In-donesia. Turning Indonesia dry would beseen by many people as an affront to thecultural diversity of the sprawling archi-pelago, which has large Buddhist, Chris-tian and Hinduminorities, aswell as manyMuslims who are partial to a cool one.

Brewers argue that alcohol is not an im-port from the decadent West, as the puri-tans often claim, but has been producedand consumed in Indonesia forat least 700years. “It is part of the culture of Indone-sia,” says Michael Chin, chief executive ofMulti Bintang, the country’s biggest brew-er. Indonesiansconsume less than one litreof alcohol per head a year, belying Muslimgroups’ claims that booze is creating ahealth crisis. Still, even without a nationalprohibition, Islamists will push for localbans—such as the one in force in Aceh since2005 and adopted elsewhere.

Beyond booze, the state-backed councilof clerics, the Indonesian Ulema Council(MUI), has in recent years passed edictscondemningeverythingfrom homosexualpartnerships to the wearing of Santa hats.Although these have no legal force underIndonesia’s secularconstitution, vigilanteshave sometimes used the edicts to targetrevellers as well as religious and sexual mi-norities. Partly at the MUI’s urging, parlia-ment has passed sweeping anti-pornogra-phy laws, which some Indonesians see asa threat to artistic and cultural liberties.Muslim groups are petitioning the courtsto interpret the law in a way that wouldcriminalise extramarital sex. They are alsomaking more use of laws against blasphe-my—notably in the trial against the gover-nor of Jakarta, Basuki Tjahaja Purnama, aChristian ofChinese descent.

Still, for a country with the world’s larg-

est Muslim population, Indonesia is re-markably permissive. Night spots in Jakar-ta, the capital, and tourist magnets such asthe island of Bali have their raunchy sides.In Semarang, Mr Dosy predicts steadygrowth in domestic sales of 8-9% per year,buoyed by a growing number of middle-class tipplers. Most Indonesians, proud oftheir tradition of tolerance, will be hopingthat he is right. 7

20 Asia The Economist January 7th 2017

AROUND of applause, ladies and gentlemen. Any typicalleader of a typical democracy, when found with nearly

$700m of ill-explained money from an unnamed foreign donorin his accounts, would experience a swift and fatal fall. Yet, nearlytwo years after news first broke that Najib Razak’s bank balancehad been thus plumped up, his high-wire act continues.

You could even argue that the Malaysian prime minister, whodenies any wrongdoing, is at the top of his game. Mr Najib ap-pears to command the unstinting loyalty of the party, the UnitedMalays National Organisation (UMNO), which leads the co-alition that has ruled the country since independence in 1957. Hehas undermined a fractious opposition, not least by peeling an Is-lamist party away from it. And as investigations proceed in sever-al other countries into the alleged bilking of colossal sums from1MDB, an indebted state investment-fund whose advisory boardMr Najib once chaired, the prime minister himself remains un-touched. Staying in powerhelpsstave offanyriskhe might face ofinternational prosecution. A general election is due by late Au-gust 2018, but perhaps Mr Najib will call a snap poll in the nextfew months to give himself several more years’ rule.

The question is how, despite the mysteries surrounding 1MDBand his personal accounts, the prime minister appears to be con-solidating his power. Patronage is a big part of it. Though his wifehas a lusty appetite for Hermès Birkin bags, and the wedding ofhis daughter to a nephew ofthe Kazakhstani president was an oc-casion of such bling that the Malaysian media were discouragedfrom publishing photographs, Mr Najib may be essentially rightwhen he says the cash in his accounts was not for personal gain.An UMNO leaderneedsmoney to buy loyalty from powerful pol-iticians. It is also handy for spreading largesse among ordinaryMalays—including helping devout Muslims make the haj.

Threatsare as importantasmoney. Anwar Ibrahim, the charis-matic leader of the informal opposition coalition which won thepopular vote in an election in 2013 (though not, thanks to gerry-mandering, a majority of seats), has been in prison since 2015 ontrumped-up charges of sodomy. In November the leader of ananti-corruption rally in Kuala Lumpur was arrested and held un-der tough new security laws. Newspapers and bloggers havebeen hounded. The number of activists and politicians charged

with sedition hasshotup. As for1MDB, the onlyconviction in Ma-laysia related to it has been of a whistle-blowing legislator whohighlighted alleged wrongdoing by the fund’s managers.

Now perhaps Mr Najib feels that the chief risks from 1MDB arebehind him. Bear in mind that among the most assiduous investi-gations to date have been those by America’s Department of Jus-tice, which claims $3.5bn is missing from the fund. Yet the nextAmerican president, Donald Trump, speaksadmiringlyofMr Na-jib, a golfing buddy. It might be hard for the department to pursuea full-throttle investigation ifMr Trump expressed displeasure.

At any rate, the prime minister is at work covering his domes-tic bases, including wooing the Islamist party, the Pan-MalaysianIslamic Party (PAS). Some analysts mock the PAS leader, AbdulHadi Awang, as having ayatollah-like aspirations: the party haslongurged for sharia punishments to apply much more widely tothe Malay Muslims who make up nearly two-thirds of the popu-lation. Such a proposal is not only morally but also constitution-ally iffy. Undaunted, Mr Najib took the extraordinary step lastyear of backing Mr Hadi’s private member’s bill, which aims toincrease the powerofIslamic courts. With little discussion in cab-inet or with the other12 coalition members, the government sub-mitted it to Parliament.

Mr Najib’s strategy is clear. Although his image has not hither-to been one of ostentatious piety, he is rebranding himself as aMuslim devout. And the message to Malays is also clear: eitheryou are with him, or, as Jayum Anak Jawan of Ohio Universityputs it in New Mandala, a website on South-East Asia, “your Ma-lay-ness or Muslim-ness are brought into question.”

If that seems a masterstroke, looks may deceive. Mr Najib’speople insist that the issue isno businessofthe (non-Muslim) eth-nic Chinese, who make up a quarter of the population, or withethnic Indians, who make up a tenth. Yet these largely urban andprosperous groups worry that Mr Najib is playing a potentiallyexplosive game of racial politics, targeted at them. That could gal-vanise the opposition.

Friendless in high placesAs it is, Mr Najib is counting on the squabbling opposition not toget its act together—in particular, on its failing to acknowledge thefutility of Mr Anwar leading the opposition from jail. Yet there isample scope for surprises. Agrowingnumberofopposition sym-pathisers say that a heavyweight with political experience isneeded to take on UMNO. The obvious candidate is MuhyiddinYassin, a former UMNO deputy prime minister who fell out withMr Najib over 1MDB. Last year he and Mahathir Mohamad, whoran the country for 22 years, founded an ethnic-Malay party op-posed to Mr Najib. The opposition would need to swallow a lotofpride and some principles to askMrMuhyiddin to be its leader.But he shares some of its reformist agenda, and it would trans-form the opposition’s chances ofvictory.

Lastly, MrNajib is runningnotonlyagainst the opposition, butagainst the economy. Since April the currencyhas fallen bynearlya quarter, reflecting the weak price of oil, a crucial export, andconcern about cronyism under Mr Najib (Malaysia ranks second,after Russia, in The Economist’s crony-capitalism index). China’shelp in bailing out 1MDB may have bought Mr Najib time, butbudgets are strapped as economic growth starts to slow. If hecan’t keep the money flowing, his seemingly loyal allies wouldabandon him in a jiffy. So if anything keeps the prime ministerawake at night, it may well be a future without friends. 7

Selling Malaysians down the river

An authoritarian prime minister looks more secure than ever. Looks can deceive

Banyan

The Economist January 7th 2017 21

1

For daily analysis and debate on China, visit

Economist.com/china

EVERY four years the United States holdsan election that can change national

policy and unseat many decision-makers.Every five years China holds a selectionprocess that can do the same thing. Com-munist Party officials tout it as evidence ofa well-ordered rhythm in their country’spolitics. This year it may turn out as unpre-dictable as America’s election in 2016.

The people up for re-selection are the350-odd members of the party’s CentralCommittee, the political elite, along withits decision-taking subsets: the Politburo,the Politburo’s Standing Committee (a sortof inner cabinet) and the army’s rulingcouncil. The choice of new leaders will bemade at a party congress—the 19th sincethe founding one in 1921—which is expect-ed to be held in Beijing in October or No-vember, and at a meeting of the newly se-lected Central Committee which will beheld directly afterwards.

Party congresses, which are attendedby more than 2,000 hand-picked dele-gates, and the Central Committee meet-ings that follow them, are little more thanrubber-stamp affairs. But they are of hugesymbolic importance to Chinese leaders.They matter for three reasons. First, theyendorse a sweeping reshuffle of the leader-ship that is decided in advance during se-cretive horsetrading among the elite. Thecomingcongresswill be MrXi’sfirstoppor-tunity to pack the Central Committee withhis own allies; the outgoing one was

der way. They involve a massive operationfor the selection of congress delegates. Onpaper, this is a bottom-up exercise. Partycommittees down to village level arechoosing people who will then chooseother representatives who, by mid-sum-mer, will make the final pick. Thousands ofparty members are also scrutinising theparty’s charter, looking for bits that mightneed changing.

It may sound like a vast exercise indemocratic consultation, but Mr Xi is leav-ing little to chance. Provincial party bossesare required to make sure that all goes to(his) plan. Over the past year, Mr Xi has ap-pointed several new provincial leaders, allallies, who will doubtless comply.

Hands up who likes XiThose chosen to attend the congress willfollow orders, too, especially when itcomes to casting their votes for membersof the new Central Committee. And thenewly selected committee will stick evencloser to script. The processes that lead toits selection of the party’s and army’s mostsenior leaders are obscure—a bit like thepicking of cardinals in the Vatican. But anaccount in the official media of what hap-pened in 2007 suggests that at some pointin the summer, Mr Xi will convene a secretmeeting of the current Central Committeeand other grandees for a straw poll to rankabout 200 potential members of the newPolitburo (which now has 25 members).This is called “democratic recommenda-tion”, although those taking part will bemindful ofwho Mr Xi’s favourites are.

Candidates for the Politburo must fulfilcertain criteria, such as holding ministerialrank. For the coming reshuffle, Mr Xi hasadded a new stipulation: faithful imple-mentation ofhis policies. For all his power,Mr Xi has struggled with widespread pas-sive resistance to his economic reforms. To

picked in 2012, when he took over, not byhim but by the people then running thecountry, including his two predecessors.After previous congresses held five yearsinto a leader’s normally ten-year term—that is, those convened in 2007 and 1997—itbecame clear who that leader’s successorwas likely to be. If the coming meetings arelike those earlier ones—a big if—they willgive a strong clue to Mr Xi’s choice of suc-cessor and start the transition from onegeneration of leaders to another.

Second, congresses can amend theparty’s constitution. China’s leaders likethe document to give credit to their favour-ite ideological themes (and Mr Xi is partic-ularly keen on ideology). When Jiang Ze-min stepped down as party chief in 2002his buzzwords were duly incorporated; sotoo were those of his successor, Hu Jintao,five years later. Mr Xi’s contribution toparty-thought—such as on the need topurge it of corruption while strengtheningits grip—is likely to gain similar recognition.

Third, congresses are the setting for akind of state-of-the-union speech by theparty leader, reflecting an elite consensushammered out during the circulation ofnumerous drafts. In the coming months,Mr Xi will be devoting most of his politicalenergy to ensuring that his will prevails inall three of these aspects. His authority inthe coming years will hugely depend onthe degree to which he succeeds.

Preparations for the gatherings are un-

The party congress

Selection year

BEIJING

For20 years China’s political transitions have been predictable. Not in 2017

ChinaAlso in this section

22 China promotes its own Shakespeare

22 China The Economist January 7th 2017

2 ram home the importance of obedience,Mr Xi recently held what he called a“democratic life session” at which Polit-buro members read out Mao-era-style self-criticisms as well as professions of loyaltyto Mr Xi as the “core” leader (as the partydecided last October to call him).

By August, when Mr Xi and his col-leagues hold an annual retreat at a beachresort near Beijing, the initial lists of lead-ers will be ready. Probably in October, theCentral Committee will hold its last meet-ingbefore the congress to approve its docu-ments. The “19th Big” will start soon after,and will last for about a week. The firstmeeting of the new Central Committeewill take place the next day, followed im-mediatelybythe unveilingbefore the pressof Mr Xi’s new lineup (no questions al-lowed, ifofficials stick to precedent).

The process is cumbersome and elabo-rate, but over the past 20 years it has pro-duced remarkably stable transfers of pow-er for a party previously prone to turbulentones. This has been helped by the intro-duction of unwritten rules: a limit of twoterms for the post of general secretary, andcompulsory retirement for Politburo mem-bers if they are 68 or over at the time of acongress. Mr Xi, however, is widely be-lieved to be impatient with these restric-tions. He has ignored the party’s hallowednotion of “collective leadership”, by accru-ing more power to himself than his post-Mao predecessors did.

If precedent is adhered to, five of theseven members ofthe Politburo’s StandingCommittee, six of its other members andfour of the 11 members of the party’s Cen-tral Military Commission (as the armycouncil is known) will all start drawingtheir pensions. In addition, roughly halfthe 200-odd full members of the CentralCommittee (its other members, known asalternates, do not have voting rights) willretire, or will have been arrested during MrXi’s anti-corruption campaign. This wouldmake the political turnover at this year’sgatherings the biggest for decades, akin tochanging half the members of the Houseof Representatives and three-quarters ofthe cabinet.

Until late in 2016 there was little to sug-gest any deviation from the informal rules.But in October Deng Maosheng, a directorof the party’s Central Policy Research Of-fice, dropped a bombshell by calling theparty’s system of retirement ages “folk-lore”—a custom, not a regulation.

The deliberate raising of doubts aboutretirement ages has triggered a round ofru-mour and concern in Beijing that Mr Ximay be considering going further. Themain focus is his own role. Mr Xi is in themiddle ofhisassumed-to-be ten-year term.By institutional tradition, any party leadermust have served at least five years in theStanding Committee before getting the topjob. So if Mr Xi is to abide by the ten-year

rule, his successor will be someone whojoins the Standing Committee right afterthe coming congress.

But there is widespread speculationthat Mr Xi might seekto stay on in some ca-pacity when his term ends in 2022. Hemight, for instance, retire as state president(for which post there is a clear two-termlimit) but continue as party general-secre-tary. He faces a trade-off. The more hebreaks with precedent, the longer he willretain power—but the more personalisedand therefore more unstable the politicalsystem itselfmaybecome. Trying to squarethat circle will be Mr Xi’s biggest challengein the politicking of the year ahead. 7

LIKE many countries, China had a busyschedule of Shakespeare-themed cele-

brations in 2016, 400 years after his death.There were plays, lectures and even plansannounced for the rebuilding of his home-town, Stratford-upon-Avon, at Sanweng-upon-Min in Jiangxi province. But as manyorganisers saw it, Shakespeare was just anexcuse. Their main aim was to use the Eng-lish bard to promote one oftheirown: TangXianzu. Whatever the West can do, theirmessage was, China can do at least as well.

Tang is well known in China, thougheven in his home country he does not en-joy anything like the literary status of hisEnglish counterpart—he wrote far fewerworks (four plays, compared with Shake-

speare’s 37), and is not as quotable. But nomatter. The timing was perfect. Tang diedin 1616, the same year as Shashibiya, asShakespeare is called in Chinese. PresidentXi Jinping described Tang as the “Shake-speare of the East” during a state visit toBritain in 2015. The Ministry of Culture lat-er organised a Tang-themed exhibition,comparing his life and works to those ofShakespeare. It has shown this in morethan 20 countries, from Mexico to France.

The two playwrights would not haveheard ofeach other: contacts between Chi-na and Europe were rare at the time. Butthat has not deterred China’s cultural com-missars from trying to weave a commonnarrative. A Chinese opera companycreated “Coriolanus and Du Liniang”, inwhich Shakespeare’s Roman general en-counters an aristocratic lady from Tang’sbest-known play, “The Peony Pavilion”.The musical debuted in London, then trav-elled to Paris and Frankfurt. Last monthXinhua, an official news agency, releasedan animated music-video, “When Shake-speare meets Tang Xianzu”. Its lines, set bi-zarrely to a rap tune, include: “You tell lovewith English letters, I use Chinese ink to de-pict Eastern romance.”

The anniversary of Shakespeare’sdeath is now over, but officially inspiredadulation of Tang carries on (a musicalabout him premiered in September in Fu-zhou, his birthplace—see picture). Chinesemedia say that a recent hit song, “The NewPeony Pavilion”, is likely to be performedat the end of this month on state televi-sion’s annual gala which is broadcast onthe eve of the lunar new year. It is often de-scribed as the world’s most-watched tele-vision programme. Officials want to culti-vate pride in Chinese literature, and boostforeign awareness of it. It is part of whatthey like to call China’s “soft power”.

Shakespeare’s works only began to takeroot in China after Britain defeated theQing empire in the first Opium War of1839-42. They were slow to spread. Afterthe dynasty’s collapse in the early 20thcentury, Chinese reformers viewed thelack of a complete translation of his worksas humiliating. Mao was less keen on him.During his rule, Shakespeare’s works werebanned as “capitalist poisonous weeds”.Since then, however, his popularity hassurged in tandem with the country’s grow-ing engagement with the West.

Cong Cong, co-director of a recentlyopened Shakespeare Centre at NanjingUniversity, worries that without a push bythe government, Tang might slip back intorelative obscurity. But Ms Cong says the“Shakespeare ofthe East” label does Tang adisservice by implying that Shakespeare isthe gold standard for literature. Tangworked in a very different cultural envi-ronment. Thatmakes itdifficult to comparethe two directly, she says. Officials, how-ever, will surely keep trying. 7

Literature

There is flattery infriendshipSHANGHAI

Officials are using Shakespeare topromote a bard ofChina’s own

The balcony scene of the East

The Economist January 7th 2017 23

For daily analysis and debate on America, visit

Economist.com/unitedstatesEconomist.com/blogs/democracyinamerica

1

APUZZLE exists where America’s eco-nomics meet its politics. Income in-

equality is higher than in other rich coun-tries, and the recent election wasinterpreted by many as the revenge of theleft-behind, who found their champion inDonald Trump. Yet the candidate whomade income inequality a campaigntheme, wanted higher taxes on the richand promised more financial regulationlost. Since the election, Mr Trump hasnominated a cabinet with a combined networth of over $6bn, by one estimate. Hehas invited the bosses of big corporationsto advise him on economic policy. And hehas filled key White House posts withGoldman Sachs alumni. The riches of top

the middle mayseem to explain a lotpoliti-cally, but it is not true. Much federal policybenefits middle earners more than thepoor. One example is the tax-deduction formortgage-interest payments. This handoutcurrently costs slightly more than theearned income tax-credit (EITC), the flag-ship anti-poverty programme that tops uppoor workers’ earnings. Yet it benefits onlythose who can afford to own their home(the bigger the mortgage, the more gener-ous the deduction). Another example isthe tax exemption for employer-providedhealth insurance. Unlike the mortgage-in-terest deduction, this does help many poorworkers. But it benefits the middle more,and this disparity has become sharper inrecent decades as insurance has becomemore expensive (see chart 2).

Handouts to the relatively well-off donot end with tax exemptions. Ignoringpublic pensions, America’s biggest federalredistribution programme is Medicare,which offers free health insurance toover-65s of any income. Much Medicarespending, which totalled $589bn (around3% of GDP) in 2016, benefits the middle

earners do not seem to bother voters near-ly as much as many on the left would likethem to.

In fact, some argue that a focus on in-equality actually harmed Democrats’chances. Most ofthe rise in inequality hap-pened over a decade ago (see chart 1). Pollsusually suggest that Americans care lessabout inequality than they do about eco-nomic opportunity. And voters have rea-son to worry about stagnation in the mid-dle-classes. Median weekly earnings,adjusted for inflation, were the same in2014 as they were in 2000. Health-insur-ance premiums have soared. A recent pa-per by Raj Chetty of Stanford Universityand colleagues documents the “fadingAmerican dream”. In 1970 more than ninein ten 30-year-olds earned more, in infla-tion-adjusted terms, than their parents didat the same age. In 2014 only halfdid.

Democrats, the logic goes, focus toomuch on helping the poor and taxing therich, ignoringjustified feelingsofabandon-ment in the middle. But there is anotherhalf to the political argument: the potentcharge thatgovernment redistribution alsopicks the pockets of the hard-working mid-dle, offering welfare to the feckless poor.This suspicion of redistribution explainshow Mr Trump could run simultaneouslyas populist insurgent and as champion ofhuge tax-cuts for the highest earners.

The idea thatgovernmenthasexploited

Inequality

Fat tails

WASHINGTON, DC

America’s government spends a lot on middle earners and little on the poor

United StatesAlso in this section

24 Congressional ethics

25 Recruiting police officers

25 Stand-your-ground laws, measured

26 Charleston’s new museum

26 Markets for tickets

27 Lexington: Learning to loveTrumpism

1For richer

Sources: CBO; Piketty,Saez and Zucman (2016)

United States

*0=perfect equality,1=perfect inequality

8

10

12

14

16

18

0

0.1

0.2

0.3

0.4

0.5

0.6

1980 85 90 95 2000 05 10 14

Richest 1%, share ofafter-tax income, %Gini index*

2A healthy chunk

Source: CBO

United States, employers’ health-insurancecontributions, as % of market income

01234567

1stPoorest

2nd 3rdINCOME QUINTILE

4th 5thRichest

1990 2013

24 United States The Economist January 7th 2017

2 class, notes Gabriel Zucman of the Univer-sity of California, Berkeley. With ThomasPiketty and Emmanuel Saez, two othereconomists, MrZucman recentlyproducednew estimates which harness GDP data toimprove the familiar figures from surveysand tax returns. They find that the incomesof those in the 50th to 90th income percen-tiles have grown by 40% since 1980, morethan previously thought, thanks to grow-ing tax exemptions. The poorer half ofAmericans pay roughly as much in taxes asthey receive in cash redistribution, in spiteof the EITC.

Before the financial crisis, governmentredistribution kept median incomes risingeven as wages stagnated (see chart 3). Sincethen it has kept incomes flat as wages havefallen. By 2013 median household incomebefore taxes was 1.6% lower than it was in1999. But after taking off taxes and addingin government transfers, it was fully 13.7%higher. More recent data suggest that evenpre-tax incomes are now growing again:they were up by 5.2% in 2015.

The economic safety net for the poor-est, however, remainsperilously thin by in-ternational standards. A typical joblessmarried couple with two children can ex-pect a welfare income, including the valueof food stamps, worth 23% of median pay.The average in the OECD, a club of mostlyrich countries, is 40%. Partly as a result, rel-ative poverty is higher than every othermember of the club bar Israel. This lookseven worse as the lower-paid have bornethe brunt of rising inequality. Messrs Pi-ketty, Saez and Zucman find that the trendsince 1980 can be summarised as a shift of8% of national income from the bottomhalf of earners to the top 1%, with no effecton those in between.

That all still leaves those whose earn-ings place them between the middle andthe poor. Median household income in2015 was nearly $57,000. Exit polls suggestthat Mr Trump lost among voters with in-comes beneath $50,000, as a Republicanpresidential candidate would be expectedto. But he did much better with such votersthan Mitt Romney did in 2012. The positive

swingmight be thought ofas a revolt of thelower middle. However, it was largestamong those with incomes beneath$30,000. Most of these voters are probablyin the poorest fifth of households, thoughsome may previously have held more lu-crative jobs. The Pew Research Centre esti-mates that the middle class, defined asthose with incomes between two-thirdsand twice the median, shrank from 55% ofthe population in 2000 to 51% by 2014.

Reaganite orkryptonite?Inequality will rise if Mr Trump succeedsin slashing taxes for the highest earners, asit did after Ronald Reagan’s tax cuts in the1980s. Then, the labour market was aboutto bifurcate into winners and losers fromglobalisation and technological change.Today, rising inequality in wealth, ratherthan in wages, might be a bigger concern.Mr Zucman and his co-authors find that aboom in investment income at the top has

driven inequality since 2000. A recentcompendium published by the RussellSage Foundation warns of growing differ-ences in wealth even amongthose who arenot rich. Mr Trump’s plan to reduce taxeson capital returns and abolish them on in-heritance could exacerbate these trends,much as the Reagan income-tax cuts coin-cided with growing disparity in wages.

The effectofMrTrump’seconomicpoli-cies on median incomes will depend onwhether they encourage firms to invest,boosting workers’ productivity. Historicalevidence is not encouraging: median earn-ings barely grew in the 1980s. But if wagescontinue their recent recovery, MrTrump issure to claim the credit. And, unlike hisparty, Mr Trump has shown little appetiteto curb spending on the middle class. Avery rich elite, high poverty and plentifulgovernment spendingon the middle couldmake Mr Trump look like a continuity can-didate after all. 7

3Easing the pain

Source: CBO *Adjusted for household size

United States, median real household income*1979=100

90

100

110

120

130

140

150

1979 85 90 95 2000 05 10 13

Market income:plus transfers plus transfers less taxes

Congressional ethics

Old bog, new tricks

AS THEIR first major initiative of thenew year, Republican congressmen

announced a scheme so crassly self-interested as to suggest they had learnednothing from the old one. Denizens of areviled institution, and a party railroadedby Donald Trump’s populist insurgency,they planned to gut the Office ofCongres-sional Ethics (OCE), an independentinvestigative body designed to root outcorruption. Less than 24 hours later, aftera hail ofcondemnation, they turned tail;even so their bungling was damning.

The OCE was founded by the Demo-crats in 2008 after a run ofscandals—including a big one concerning the Re-publican lobbyist JackAbramoff—high-lighted the impunity with which somelawmakers were abusing their office inexchange for campaign contributions.The office is empowered and equipped toinvestigate allegations of impropriety. Itmay then report its findings to the HouseEthics Committee and, even if that bodydecides to take no further action, pub-licise them. Anti-corruption campaignersconsider it a bulwarkagainst officialcorruption. Many congressmen considerit unjust and wasteful.

Several who have been subject to theoffice’s inquiries were involved in theeffort, at a closed-door meeting of Repub-lican congressmen, to nobble it. Theyincluded Blake Farenthold ofTexas, whowas investigated and exonerated by theoffice over an allegation ofsexual harass-ment. The plan was to put the OCE under

congressional control and limit the scopeof its investigations and its ability topublicise its work. Paul Ryan, the Repub-lican Speaker of the House ofRepre-sentatives, warned against this; the planwas nonetheless approved, by a vote of119 Republican congressmen to 74.

Mr Trump, no doubt aware ofhowbadly this was playing, offered a mea-sured criticism of the congressmen’sinitiative. In a tweet, he called the OCE“unfair” but suggested his Republicancolleagues had bigger things to be gettingon with. A deluge ofnegative commentsreceived at their district offices made thatpoint more forcefully. So did SenatorLindsey Graham ofSouth Carolina, whocalled the attempted takedown of theOCE “the dumbest frickin thing I’ve everheard”. On January 3rd, the opening dayof the new Congress, the plotters hastilyagreed to leave the OCE alone after all.

This delivered an easy triumph to MrTrump, whose tweet was credited bymany headline writers with havingpersuaded the congressmen to changecourse, albeit without much evidence. Italso showed those lawmakers to lackself-awareness to an amazing degree. Ifthe OCE is not working well, they shouldstart a debate—in and with the public—about how better to investigate andprevent their abuses. To avoid an unnec-essary partisan fight, they also plainlyneed Democratic support. This is basicpolitics. Republican congressmen shouldreally learn how to do it.

WASHINGTON, DC

Howto lose votes and irritate people

The Economist January 7th 2017 United States 25

STAR WARS can be used to sell almostanything, from Lego to a career in polic-

ing. Fort Worth’s police department re-leased a recruitment video on its Facebookpage in December featuring an officer attarget practice with a stormtrooper. Thewhite-clad soldiers are notoriously poorshots, and the video shows the galactic GImissing every attempt he makes until hecreeps so far forward that his goggles arevery nearly touching his target. When theexasperated officer asks “who referred youto us?” Darth Vader peeks out from theback of the room, shaking his helmetedhead in disgust. The scrolling text at theend of the video, which has garnered 17mviews thus far, urges: “Join ourForce! Ifyouhave what it takes to be a Fort Worth PoliceOfficer and are a better aim than a Storm-trooper.” The advert underscores a seriousproblem affecting police forces nation-wide. Economic and social changes havemade it harder for police departments tokeep their forces fully staffed, and lead toincreasingly desperate recruitment.

The Los Angeles Police Departmentwas short of nearly 100 officers as of mid-December—only 1% of its total workforce,but still enough to be felt on the ground,says Captain Alan Hamilton, who runs re-cruitment for the department. Philadel-phia had 350 vacancies, largely due to aspate of retirements. Last spring, Dallascancelled two academy classes for lack ofapplicants; its preliminary applicationsdropped by over 30% between 2010 and2015. In 2012, the ratio of police officers topopulation hit its lowest level since 1997,according to Uniform Crime ReportingProgramme data published by the FBI.

The dynamics underpinning the short-ages vary by department, but there are na-tional trends making it harder for policeforces to attract applicants. The first is astrong economy. Nelson Lim, a researcherat the RAND Corporation, a think-tank,says this is nothing new. When plenty ofjobs are available, people are usually lessmotivated to enter dangerous professions.Police forces as well as the armed forcestend to field less interest in boom times.

The second is the perception of in-creased danger associated with policing:135 officers were killed in the line of dutybetween January 1st 2016 and December28th 2016—a 10% increase from 2015 butfewer than the 192 killed in 2007. Shootingdeaths increased from 41 to 64. Several ofthem were high profile and gruesome,

such as the assassination of five Dallas po-lice officers in July 2016. “When you lookaround the nation and you see the acts ofviolence directed at police officers—itmakes people reluctant to join. Many peo-ple join the profession when they’re 22 or23 when parents still have a heavy influ-ence,” says Scott Walton, deputy chief in

Dallas, though sympathycan also boost re-cruitment. Dallas has seen an uptick in ap-plications since its officers were attacked.

The last is the image of policing. Thedeaths of several unarmed black men atthe hands of police officers and the ensu-ing backlash seem to have made policework less appealing. “We have a situationwhere law enforcement is being scruti-nised more heavily,” says Mr Hamilton ofthe LAPD. According to Gallup, a polling or-ganisation, trust in law enforcement gener-ally has remained fairly stable since it be-gan surveying the topic in 1993. Butaccording to data collected by Harris, an-other polling group, the share of bothwhites and blacks who believe that Afri-can Americans are discriminated againstby the police has risen markedly between1969 and 2014.

Baltimore Police Department’s officershortage led it to Puerto Rico in search offresh faces. The department also mulled re-laxing its stance on past marijuana use.Chicago has cut its minimum age require-ment for its police academy from 25 to 21.Several departments have lowered educa-tional requirements for recruits. If Presi-dent-elect Trump follows through on hispromises to beef up military and infra-structure spending, the plight of police de-partments might worsen, worries Mr Lim.The armed and police forces tend to com-pete for applicants. If more jobs becomeavailable in industry and construction,puttingon a badge might become even lessappealing to young workers. 7

Recruiting police officers

The force is weak

LOS ANGELES

Police departments struggle to stayfullystaffed

No college required

Source: “Evaluating the impact of Florida’s ‘stand-your-ground’ self-defence law on homicide and suicide by firearm” by D.K. Humphreys, A. Gasparrini and D.J. Wiebe, January 2017

Florida, United States, homicides per 100,000 population, victims aged 20-34 years

0

0.5

1.0

1.5

2.0

1999 2000 01 02 03 04 05 06 07 08 09 10 11 12 13 14

ALL HOMICIDES

Of which: HOMICIDES BY FIREARM

“STAND-YOUR-GROUND”LAW COMES INTO EFFECT

With the stroke of his pen in 2005, Jeb Bush, then governor of Florida, ignited enthusi-asm for “stand-your-ground” laws. Citizens who “reasonably believed” their lives to bethreatened were given the right to “meet force with force, including deadly force”—even in public places and, critically, without the duty to try and retreat first. More than20 states have passed similar laws since then. Critics warned that, rather than protect-ing self-defence rights as intended, the bill would result in unnecessary deaths. Re-search published in the Journal of the American Medical Association appears tovindicate those fears. Soon after the law took effect in Florida, there was a sudden andsustained 24% jump in the monthly homicide rate. The rate of homicides involvingfirearms increased by 32%. The authors found that in states without a stand-your-ground law over the same time period those rates remained flat, suggesting that anationwide crime wave was not to blame for the abrupt increase.

Gun laws

26 United States The Economist January 7th 2017

JUST inside the gates of the UnitarianChurch in Charleston sits a slab of sal-vaged bricks. Affixed to the front is a met-

al bird looking backwards—a West Africansymbol, a plaque explains, which means“learning from the past in order to moveforward”. An inscription dedicates themonument to “the enslaved workers whomade these bricks and helped build ourchurch.” The church is off most tourists’trails, so many miss the memorial. But anoverdue museum aims to spread its frankmessage more widely.

Like much of the South, for a long timethe city glossed or downplayed the abom-ination that made it rich and left it beauti-ful. Visitors to its grand townhouses, or tothe sumptuous plantation mansions near-by, might be shown suspiciously well-ap-pointed “servants’ quarters”. The conflictknown as the “war between the states”was not, repeat not, fought over slavery.There ismore honesty these days; the tradein human beings is documented in a smallexhibition in an old slave mart. But suchacknowledgments are not commensuratewith the role the institution once played inCharleston, and Charleston in it. Rose-tint-ing continues. As Michael BoulwareMoore, boss of the planned InternationalAfrican American Museum (IAAM), says,formany the plantations are less “places ofhorrific inhumanity” than picturesquebackdrops for weddings.

The IAAM’s progenitor was Joseph Ril-ey, Charleston’smayorforfourdecadesun-til last year. He announced the idea in2000, aiming to remedy an amnesia he de-scribesas“a societal defect in America”. (Inthe 1970s his outreach to the blackcommu-nity earned him the sobriquet L’il BlackJoe: “an honour”, he now says.) Ser-endipitously, he recalls, the city was able toacquire “one of the most sacred sites of Af-rican-American history in the Westernhemisphere”, the location of Gadsden’sWharf, where perhaps 40% of the slavesimported to America first set foot on themainland. Overall Charleston’s wharfs ac-counted for around half of those arrivals.As MrRiley says, nowhere else in the coun-try was as important to slavery, and “noplace has more of a duty” to remember it.He is helping to raise the $20m needed forthe project to meet its target of$75m.

When the museum opens in 2019—the400th anniversary of the first slave ship toland in the colonies—its “greatest artefact”,says Mr Moore, “will be the ground.” The

building will duly be raised on pedestals,the waterfront windows affording viewsof the Cooper River and out towards FortSumter, where the civil war began in 1861.That setting isone ofthe featuresMr Mooresays will differentiate it from the new Afri-

can-American museum in Washington(the two share an exhibit designer). So, hehopes, will its emphasis on genealogicalresearch, a bid to fill some of the gapsscoured by enslavement, plus its interest inAfrica itself. Some locals, he says, fear that,in a rapidly gentrifying environment, theIAAM might “pimp black history”. On thecontrary, says Mr Riley, it will tell the “un-varnished, harsh story” of the country’s“original sin”, including the roughly 700people who froze to death in a warehousenear the wharf in the winter of1807-08.

Yet along with the horror, promises MrMoore, the museum will commemoratethe skills and accomplishments of slavesand their descendants, reassuring blackyoungsters that “there are heroeswho lookjust like them”. One such is his own great-great-grandfather, Robert Smalls (pic-tured), who in 1862 won his freedom bycommandeering a Confederate steamshipand delivering it to the federal fleet. Laterhe was elected to Congress and bought hisformer master’s house. A century and ahalf after his escapade, two markers wereerected in Charleston in his honour. Onewas promptly vandalised. 7

Charleston’s new museum

Cobblestones andbones

CHARLESTON

Filling in the gaps in America’s history

Smalls by name, big in daring

Markets for tickets

Battling bots

“HAMILTON”, the hip-hop Broad-way musical about one ofAmeri-

ca’s founding fathers, has broken all sortsofbox-office records. Demand is highbecause it is exceptionally good. But thisis not the only reason tickets are so scare.Every time the show’s producers releasea new block to sell, they immediately getsnapped up by “ticket bots”, high-speedticket-buying software. The bots cut thevirtual queue, manipulating and paralys-ing sites like Ticketmaster before realpeople can get a look-see. Jeffrey Seller,“Hamilton’s” producer, has called bots“computerised cheaters”.

Despite Mr Seller working with Tick-etmaster—which runs software in aneffort to stop bots—to get tickets into thehands of real fans, too many tickets endup on secondary market websites forsubstantially inflated prices. According toTicketmaster, about 60% of the hottesttickets are bought by bots. A single bro-ker, using a bot, purchased 1,012 to a 2014U2 concert in under a minute, despite thevenue limiting sales to four tickets percustomer. By the end of the day, thatsame broker had purchased 15,000 tick-ets. Even tickets for free events, like PopeFrancis’s visit to New York in 2015, weregobbled up by bots and sold for thou-sands ofdollars on secondary sites.

Last month President Obama signedlegislation which aims to eliminate botsand intends to slap hackers with heftyfines. Under the new law, the BetterOnline Ticket Sales Act, the federal gov-ernment can also intervene and file suiton behalfofpeople shut out ofbuyingtickets because ofbots. The law hassupport from across the industry, in-cluding Stubhub, a secondary market-place. Stubhub does not sell anythingdirectly, but takes a transaction fee. Je-remy Liegl, a lawyer at Ticketfly, whichsells and promotes music events, saidduring a congressional hearing that botsharm everyone in the music industryexcept for the bots’ operators. The ticketmarkups end up in the pockets of the botoperator, not the promoter, not the venueand not the performers.

But the secondary market, worth asmuch as $8bn worldwide, may be toovaluable for virtual scalpers to give up.Federal law enforcement may be unableto hunt down bot-operators based out-side America. And the scale of the racketis daunting. Last year bots made 5bnattempts to buy tickets on Ticketmaster,at a rate of roughly10,000 a minute. Acold-hearted economist would propose asimple solution: make the tickets muchmore expensive in the first place.

NEW YORK

Whyyou still can’t get a ticket for“Hamilton”

The Economist January 7th 2017 United States 27

ATRUE politician is someone who, upon spying a torch-wield-ing mob marching on his legislature, declares: “Oh good, a

parade—I must lead it.” A striking number of conservatives aretaking that approach ahead of Donald Trump’s inauguration onJanuary 20th. The president-elect’s followers include many whodistrustboth main partiesand, ifhanded a pitchfork, might skew-er half the Republicans in Congress. Undaunted, party bigwigsand intellectuals have begun making the case that, for all itsrough edges, “Trumpism” is a recognisably conservative way ofviewing the world, with the potential to rescue swathes ofAmer-ica from feelings of abandonment and despair, securing major-ities for Republicans for years to come.

Republican leaders who clashed with Mr Trump during theelection campaign now urge colleagues to see his victory as a les-son in humility. After the new Congress was sworn in on January3rd Paul Ryan, the Speaker of the House of Representatives, toldmembers that for too long leaders in Washington had treatedcomplaints about closed factories with “condescension”. NowAmericans had let out a “great roar”, Mr Ryan continued: theyhave given Republicanscontrol ofCongressand the White Housenot as an act ofgenerosity but as a demand for “results”.

Embracing the results soughtbyTrumpism will notbe easy forconvinced free-marketeers such as MrRyan. On the day that Con-gress returned to work, Ford announced that it is cancelling plansfor a $1.6bn plant in Mexico and will create 700 jobs in Michiganbuilding electric vehicles. This follows months of public brow-beatingbyMrTrump, including threatsofpunitive tariffs on firmsmaking things abroad—though Ford’s chief executive cast the de-cision to invest in America as a vote of confidence in “pro-growth” policies outlined by the president-elect. Ford joins Carri-er, Lockheed Martin and Boeing as companies that have changedinvestment or pricing decisions after Trumpian arm-twisting(hailing Carrier’s climb-down, eased by tax breaks from the stateof Indiana, Mr Trump declared that the free market had failedAmerican workers “every time”). As Mitt Romney’s vice-presi-dential running-mate in 2012, Mr Ryan scorned the idea of gov-ernments picking “winners and losers”. Today the Speaker talksup the prospects of tax reforms and deregulation giving all com-panies good cause to stay in America.

Other conservative grandees wonder if a dose of economicnationalism is the price of solidifying the coalition that carriedMr Trump to power, including blue-collar voters in the Midwestwho abandoned the Democrats in droves. They praise Mr Trumpas a patriot-pragmatist in the spirit ofLincoln or Theodore Roose-velt. They are slower to note more recent models for Trumpism,starting with populist-nationalist movements sweeping Europe.Parallels with Nicolas Sarkozy, France’s centre-right presidentfrom 2007-12 and a hyperactive corporatist, are startling. Mr Sar-kozy denounced French carmakers for producing cars in easternEurope (“not justifiable”, he growled), and rushed to a steelworksto promise workers he would save their jobs (a pledge he couldnot keep).

Hugh Hewitt, a conservative talk-radio host, this month willpublish “The Fourth Way”, a book-length guide to how Trum-pism mightadvance bitsofthe Reagan agenda, bypromoting con-servative judges, stronger armed forces (Mr Hewitt likes MrTrump’s talk of a 350-ship navy) and free enterprise (above allrolling back “the vast and growing regulatory state”). To that hewould add a “repatriation window” for corporate profits heldabroad, and a new, voter-pleasingwave ofinfrastructure projects,including reopened shipyards, modernised airports, local sportsfacilities and other visible signs of Trumpian largesse. Billionswould be disbursed by temporary, county-level commissions ap-pointed by Congress and the White House (“the patronage!”sighs Mr Hewitt) and as classic pork-barrel spending by membersofCongress. To breakthe partisan stalemate over immigration MrHewitt would have MrTrump lay out detailed plans fora double-row border fence along the southern border, which when half- orthree-quarters built would trigger a legalisation programme formost of the 11m immigrants in America without the right papers.Get all this right, Mr Hewitt says, and Mr Trump can realign na-tional politics. Get itwrongand MrTrump could face “catastroph-ic” midterm elections in 2018, a primary challenger in 2020 or, ifembroiled in scandals, impeachment.

Seeking MrTrump’s innerReaganIn the spring Newt Gingrich, a former Speaker and adviser to MrTrump, will publish “Understanding Trump.” He calls the presi-dent-elect the third attempt, after Reagan’s election in 1980 andhis own Contract with America in 1994, to breakfree from the big-government mindset of Franklin Roosevelt’s New Deal. Tosquare that claim with Mr Trump’s free-spending, distinctly stat-ist campaign promises, Mr Gingrich portrays the businessman asa disruptive innovator, using social media and a genius for pub-licity to win a presidential election on the cheap. That thriftiness,also displayed in Mr Trump’s business life, tells Mr Gingrich thatPresident Trump will run a lean federal bureaucracy, root outwaste and generally “kick over the table”. Pondering Mr Trump’sdesire for better relations with Russia, Mr Gingrich has called itspresident, Vladimir Putin, “a thug” but at the same time scoldedRepublicans who treat Russia as if it were still the Soviet Union,rather than a competitor that needs dealing with “as it is”.

Some conservative enthusiasm for Trumpism reflects a sin-cere desire to grapple with voterangst. Some isborn of opportun-ism and fear that the next president will set his voters on Repub-licans who defy him. Yet Mr Trump will need allies in Congress,too, if he is to rack up achievements to impress supporters.Trumpism, born as a populist revolt, must become a programmefor government. That’s harder than leading a parade. 7

Learning to love Trumpism

Conservatives are working hard to reconcile theirbeliefs with the next president’s agenda

Lexington

28 The Economist January 7th 2017

1

THE rampage lasted 17 hours. By the endof it, 56 inmates of the Anísio Jobim pri-

son complex in Manaus, a city set amid theAmazon rainforest, were dead. Many hadbeen decapitated; severed arms and legswere stacked by the entrance to the jail,known to mostasCompaj, a contraction ofits full name. Luís Carlos Valois, a judgewho negotiated an end to the violence onJanuary 2nd, called the hellish scene “Dan-tesque”. It was Brazil’s bloodiest prison riotin a quarter-century.

Only the death toll makes the carnageat Compaj stand out. Brazil’s prisons eruptoften. Last year 18 inmates died in clashesbetween gangs at prisons in the northernstates of Roraima and Rondônia. In Per-nambuco, a north-eastern state whose pri-sonsare overstuffed even byBrazilian stan-dards, violent deaths are a frequentoccurrence. In January 2016, 93 prisonersbroke out of two of the state’s jails.

Prisons are both hellholes and head-quarters for Brazil’s most powerful crimi-nal gangs. The country’sprison populationof 622,000, the world’s fourth-largest, iscrammed into jails built to hold 372,000 in-mates. Compaj houses 2,200, nearly fourtimes its capacity. Guards often do littlemore than patrol the perimeters, leavinggangs free to manage far-flung criminal op-erations via mobile phones.

The riot at Compaj suggests that prisonviolence—and the behaviour of the gangsbehind it—is entering a new phase. Offi-cials in the state of Amazonas say mem-

alliance with the Comando Vermelho (RedCommand, or CV). But the paulistas usedtheir growing might to force their partnerinto a subordinate position, which pro-voked a rupture. The PCC has since teamedup with the CV’s main rival, the Amigosdos Amigos (Friends of Friends). Prosecu-tors say the arrangement has allowed theSão Paulo group to take control ofRocinha,a favela in Rio de Janeiro, thought to be thecity’s most profitable drug market.

The CV has responded by forming alli-ances with other crime groups threatenedby the PCC’s expansion. Among them arethe FDN, Brazil’s third-biggest gang, whichcontrols drug-smuggling routes in the Am-azon. The clashes in Roraima and Rondô-nia were a harbingerofthe Compaj massa-cre. Most of the dead were members of theFDN and the CV, targeted by the PCC in re-venge for attacks mounted by the FDN theyear before.

Payback timeOfficials now wonder where and whenthe PCC will retaliate. The retribution willcome from calculation, not rage, says Gua-racy Mingardi, a criminologist. But come itwill. The PCC “cannot remain quiet, asthey will lose prestige, and prestige in thelong term represents money”.

There is little prospect that govern-ments will do much to end the cycle of vio-lence. Alexandre de Moraes, Brazil’s justiceminister, said the ringleaders of the Com-paj massacre will be transferred to federalprisons. The federal government promisedat the end of 2016 to spend an extra 1.2bnreais ($370m) quickly to build andmodernise state prisons. But that will notbe enough to improve conditions that aprevious justice ministerdescribed as “me-dieval”. The cash-strapped federal govern-ment will have a hard time finding more.Historically, it has preferred to let state gov-ernments, which house nearly all of Bra-

bers of Família do Norte (Family of theNorth, or FDN), which controls drug traf-ficking in the Amazon region, organisedthe Compaj massacre. Having gained con-trol over much of the prison, the gangsought to wipe out opposition from Pri-meiro Comando da Capital (First Com-mand of the Capital, or PCC), a larger rivalbased in São Paulo, a south-eastern state.

The assault on the PCC seems to be a re-action to its growing strength. Formed in1993 by inmates in São Paulo after policemassacred more than 100 prisoners at thenotorious Carandiru jail, it has branchedout into drugrunning, extortion and prosti-tution, often with the tacit consent of pri-son authorities. The PCC killed the last bigrival drug trafficker in Paraguay in 2016.That gave it dominance over smugglingalong the borders with Paraguay and Bo-livia, and thus over the supply of cocaineand marijuana to the south-east, Brazil’srichest region. It used that advantage to be-come the country’sbiggestand mostprofit-able organised-crime group. Exploiting itsgrowing control of the main entry pointsfor drugs, the PCC moved beyond its homeregion and now has a nationwide pres-ence. The battle at Compaj is “principally areaction to the growing power of the PCCacross Brazil in the distribution of drugs”,says Bruno Paes Manso, a criminologist atthe University ofSão Paulo.

At first, the PCC co-operated with thedominant forces in other states. In Rio deJaneiro it formed a narcotics-distribution

Brazil’s prisons

Horror in the jungle

SÃO PAULO

A massacre in Manaus shows that competition among gangs is increasing

The AmericasAlso in this section

29 Evo Morales for ever?

Bello is away

The Economist January 7th 2017 The Americas 29

2 zil’s prisoners, bear the burden of manag-ing and paying for the system.

They, in turn, have neither the moneynor the ideas needed to improve condi-tions. Politiciansand judgesare more eagerto lock up criminals, especially if they arepoor and black, than they are to reduceovercrowding. About two-fifths of Brazil’sprisoners are awaiting trial rather thanserving sentences; university graduates,priests and others are entitled to wait incomfier conditions.

Governments also fear that a crack-down on violence in prisons will causetrouble outside them. An attempt by SãoPaulo’sgovernment in 2006 to curb the pri-son-based operations of the PCC set off acampaign ofviolence by the gang’sconfed-erates across the state. Hundreds died overten days in attacks on policemen and thereprisals they provoked. Politicians preferto keep the violence within prison walls. 7

AS 2016 drew to a close, a resignation let-ter from Bolivia’s president, Evo Mo-

rales, lit up social media. It was a hoax, per-petrated on día de los inocentes (“day of theinnocents”), the Latin American version ofApril Fools’ Day, which falls on December28th. The unamused communicationsminister, Marianela Paco, denounced it asan “attack on the people’s right to reliableand truthful news”.

In fact, MrMorales’s alliesare schemingto keep him in office indefinitely, eventhough, in a referendum last February, Bo-livians voted to deny him the right to runfora fourth term in 2019. On December17thhis party, the Movement to Socialism(MAS), named him as its presidential can-didate for the next election. “If the peopledecide it, Evo will continue,” Mr Moralespromised his supporters.

The clamour to keep him is not a sign ofrecent success. Bolivia is suffering from asevere drought, whose effects are madeworse by the government’s failure to planand invest. Economic growth, which sus-tained Mr Morales’s popularity for most ofhis 11 years in office, has lost momentum.Scandals, strikes and clashes betweenprotesters and police have turned someBolivians against the government. Yet MrMorales, Bolivia’s first president of indige-nous origin, dwarfs his rivals. Nearly halfof voters still approve of his performance.No one in the MAS has the stature to suc-ceed him. The opposition is fragmented.

Bolivia’s woes may cut Mr Moralesdown to size. Despite the onset of the rainyseason in December, manydistrictsare stillrationing water. The state-owned watercompany that supplies La Paz, the seat ofgovernment, and El Alto, a populous cityperched on a cliffabove it, ran out of waterin November. The water level in the Inca-chaca reservoir, which serves parts of LaPaz, was farbelow normal in early January.Residents of the city queue for hours to getdeliveries by lorry. Farmers and ranchersare reporting large and growing losses.

This is contributing to the slowdown ofthe economy, which depends largely ongas exports. Their price is linked to theprice ofoil, which has halved since 2014. In2017 Bolivia is expected to earn $2.1bn fromgas sales, just a third ofwhat it made whenprices were high.

As a result, GDP will grow 3.9% in 2017, abit more than this year but far below thepeak of 6.8% in 2013, forecasts the IMF (seechart). Bolivia’s overvalued currency ishurting producers of goods besides rawmaterials, warned the IMF last month.Alarmingly, the current-account and bud-get deficits were around 8% ofGDP in 2016.

By the profligate standards of LatinAmerica’s left-wing leaders, Mr Moraleshas been a fairly responsible economicmanager. He invested Bolivia’s gas wind-fall in roads, bridges, hospitalsand schools.Until 2014 he kept a lid on budget deficits.But the oil-price slump and wackyweatherare exposing the government’s failures.High taxes on the production ofoil and gasand the absence ofan independent regula-tor have discouraged investors from pros-pecting for new reserves, says Hugo delGranado, a former energy official. Just twoforeign energy companies, Russia’s Gaz-prom and Venezuela’s PDVSA, have cometo Bolivia in the past decade. Debt-riddenPDVSA has only a token presence; Gaz-prom spentyearsbattlingbureaucracyandended up scaling back its ambitions. Rath-er than invest on its own, it formed a part-nership with France’s Total, which alreadyhad operations in Bolivia.

The government’s infrastructure-spending binge did not extend to water. It

nationalised utilities, put party hacks incharge and failed to invest in them. La Paz’swater reserves reached dangerously lowlevels even before the drought tookhold.

Such setbacks have damaged the gov-ernment’s prestige. In October it had tocancel a popular bonus of an extramonth’s wages paid to all workers in theformal sector, but only in years when GDPgrowth is more than 4.5%. Mr Morales’s re-lations with trade unions and social move-ments, which once gave him unstintingsupport, have been hurt by disputes overinfrastructure projects and benefits for dis-abled people. A conflict over regulation ofmining by co-operatives led to the deathsof four miners and the murder of a vice-minister. The Central Obrera Boliviana,the main trade-union federation, has fallenout with the government. The MAS lostcontrol of El Alto, Mr Morales’s politicalstronghold, in regional elections in 2015.

None of that deters his allies from plot-ting to keep him in power. Manylookto theexample of the late Hugo Chávez, Vene-zuela’s left-wing leader, who lost a referen-dum to end term limits in 2007 only to holdanother one 14 months later, which hewon. Under Bolivia’s constitution, a peti-tion signed by a fifth of the electorate couldtrigger a re-run of the referendum to lift theterm limit facing Mr Morales.

Another option, suggested by the vice-president, Álvaro García, is that Mr Mo-rales resign for real, turning the hoax intoreality. If this were coupled with a reformof the constitution, Mr Morales could ar-gue thathis currentmandate, served underan outdated constitution, should not counttowards one of the two terms he is al-lowed. He used this manoeuvre once be-fore, in 2014, to run for re-election.

The president says that what he reallywants is to return to his career as a growerof coca, a traditional stimulant that is alsothe raw material for cocaine. In fact, he wasmore an organiser of farmers than a culti-vator himself. Only inocentes believe it ishis ambition to become one. 7

Bolivia

For Evo, for ever

LA PAZ

Evo Morales’s supporters are lookingforways around term limits

Evo’s economy

Source: IMF *Estimate †Forecast

Bolivia

9

6

3

0

3

6

9

+

–

2006 08 10 12 14 16*17†

GDP, % increase on a year earlier

General-government budget balance, % of GDP

Why stop in 2025?

30 The Economist January 7th 2017

For daily analysis and debate on the Middle Eastand Africa, visit

Economist.com/world/middle-east-africa

1

AFTER halfan hourofpencil-chewing Li-zeka Rantsan’s class lines up at her

desk to hand in its maths tests. The teacherat Oranjekloof primary school in CapeTown thanks the 11- and 12-year-olds andflicks through the papers. Ms Rantsansighs, unimpressed. Pullingone sheet ofer-rant scribbles from the pile she asks: “Howare we supposed to help these children?”

It is a question that South Africa is fail-ing to answer. In a league table of educa-tion systems drawn up in 2015 by the OECDclub of mainly rich countries, South Africaranks 75th out of76. In November the latestTrends in International Mathematics andScience Study (TIMSS), a quadrennial testsat by 580,000 pupils in 57 countries, hadSouth Africa at or near the bottom of its va-rious rankings (see chart), though its scoreshad improved since 2011. Its children arebehind those in poorer parts of the conti-nent. A shocking 27% of pupils who haveattended school for six years cannot read,compared with 4% in Tanzania and 19% inZimbabwe. Afterfive years ofschool abouthalf cannot work out that 24 divided bythree is eight. Only 37% of children startingschool go on to pass the matriculationexam; just 4% earn a degree.

South Africa has the most unequalschool system in the world, says Nic Spaullof the University of Stellenbosch. The gapin test scores between the top 20% ofschoolsand the rest iswider than in almostevery other country. Of 200 black pupils

After Nelson Mandela became presi-dent in 1994 his government expanded ac-cess to schooling. It also replaced a schoolsystem segregated by race with one divid-ed by wealth. Schools in poorer areas re-ceive more state funding. But schools inricher areas can charge fees on top.

In theory these schools must admit pu-pils even if their parents cannot afford thefees. In practice they are fortresses of privi-lege. There are still about 500 schools builtfrom mud, mainly in the Eastern Cape. TheWestern Cape has some of the largest cam-puses in the southern hemisphere, withcricket pitches as smooth as croquet lawns.

And yet money is not the reason for themalaise. Few countries spend as much toso little effect. In South Africa publicspending on education is 6.4% of GDP; theaverage share in EU countries is 4.8%. Moreimportant than money are a lack of ac-countability and the abysmal quality ofmost teachers. Central to both failures isthe South African Democratic TeachersUnion (SADTU), which is allied to the rul-ing African National Congress (ANC).

The role of SADTU was laid bare in a re-port published in May 2016 by a team ledby John Volmink, an academic. It found“widespread” corruption and abuse. Thisincluded teachers paying union officialsfor plum jobs, and female teachers beingtold they would be given jobs only in ex-change for sex. The government has donelittle in response. Perhaps this is unsurpris-ing; all six of the senior civil servants run-ning education are SADTU members.

The union’s influence within govern-ment belies its claim that officials are toblame for woeful schools. Last year it suc-cessfully lobbied for the cancellation ofstandardised tests. It has ensured that in-spectors must give schools a year’s noticebefore showing up (less than 24 hours isthe norm in England). And although par-

who start school just one can expect to dowell enough to study engineering. Tenwhite kids can expect the same result.

Many of the problems have their rootsin apartheid. The Bantu Education Act of1953 set out to ensure that whites received abetter education than blacks, who were,according to Hendrik Verwoerd, the futureprime minister then in charge of educa-tion, to be educated only enough to be“hewers of wood and drawers of water”.Black pupils received about a fifth of thefunding of white peers. They were taughtalmost no maths or science. Most indepen-dent church-run schools that provided agood education in blackareas were shut.

South Africa’s schools

Bottom of the class

CAPE TOWN

WhySouth Africa has one of the world’s worst education systems

Middle East and AfricaAlso in this section

31 Science amid the sheep farmers

31 Zimbabwe’s sex trade

32 The battle for Mosul continues

33 Trump and the West Bank

34 Israel’s divisions

Far behind

Source: Trends in InternationalMathematics and Science Study *500=Average

Selected TIMSS maths scores*, 2015, % of pupils

0 20 40 60 80 100

Singapore

South Korea

Japan

Australia

Sweden

Italy

Botswana

Morocco

South Africa

Below 400 400-475 475-550550-625 At or above 625

The Economist January 7th 2017 Middle East and Africa 31

1

2 ent-led school governing bodies are meantto hold teachers to account, they are moreoften controlled by the union, or in somecases by gangs.

But even if there were better oversightmost teachers would struggle to shape up.In one study in 2007 maths teachers of 11-and 12-year-olds sat tests similar to thosetaken by their class; questions includedsimple calculations of fractions and ratios.A scandalous 79% of teachers scored be-low the level expected of the pupils. Theaverage 14-year-old in Singapore and SouthKorea performs much better.

It does not have to be this way. SparkSchool Bramley in Johannesburg is a low-cost private school, spending roughly asmuch per pupil as the average state school.

And it is everything state schools are not.Its 360 pupils begin learning at 7.30am andend around 3pm-4pm; most state schoolsclose at 1.30pm. At the start of the day pu-pils gather for mindfulness exercises,maths questions, pledges to work hard—and a blood-pumping rendition of KatyPerry’s “Firework”. “We have an emotionalcurriculum as well as an academic one,”says Bailey Thomson, a Sparkdirector.

Pupils attend maths lessons based onSingapore’s curriculum; literacy classesdraw on how England teaches phonics.Crucially, teachers are not members ofSADTU. But they receive 250 hours of pro-fessional development per year, about asmuch as the average state-school teachergets in a decade.

Early results show that its pupils are onaverage a year ahead of their peers. Sparkruns eight schools and plans to have 20 by2019. Other operators, such as Future Na-tion, co-founded by Sizwe Nxasana, a for-mer banker, are also expanding. “We arenever going to have a larger footprint than[the] government but we can influence it,”hopes Stacey Brewer, Spark’s founder.

Another promising scheme is the “col-laboration schools” pilot in the WesternCape, based on academies in England andcharter schools in America. The five col-laboration schools are funded by the statebut run by independent operators. In whatHelen Zille, the premier of the WesternCape, calls “a seminal moment”, the par-ents of Oranjekloof pupils petitioned tokeep the school in the collaboration pro-gramme when unions tried to oppose it.Ms Zille wants to open a “critical mass” ofcollaboration schools to inject competitioninto the public system.

Spark and the collaboration schoolssuggest that South African education neednot be doomed. But together they accountfor a tiny fraction of the country’s morethan 25,000 schools. Widespread im-provement will require loosening the gripof SADTU. In local polls in August the rul-ingparty saw its worst results since the endof apartheid. This may force it to reviewvested interests. More likely it will contin-ue to fail children. “The desire to learn hasbeen eroded,” saysAngusDuffett, the headof Silikamva High, a collaboration school.“That is the deeper sickness.” 7

Astronomers v sheep farmers in South Africa

Stars and baas

THERE is a haunting beauty to theKaroo, a vast swathe ofsemi-desert

that seems empty save for the stars over-head and sheep grazing below. Economicopportunities here are few. Scrubby,sprawling farms support sheep, ostrich,springbokand little else. (To be fair, Karoolamb is delicious.)

But the Karoo’s clear skies also drawsome of the world’s best scientists. Aradio telescope project called the SquareKilometre Array (SKA) is under construc-tion, with the latest cluster of64 giantantennae due to be completed late nextyear. South Africa won the right to hosthalfof the $2bn international project in2012. When finished it will be the biggestradio telescope in the world and should

allow scientists to peer into the origins ofthe universe.

Still, some sheep farmers are grum-bling. Because of the sensitivity of thetelescope, the surrounding area must bekept free from radio interference causedby everything from mobile phones tomicrowave ovens and some car engines.The SKA is buying up more farms thanoriginally expected to ensure radio si-lence over an area ofsome 130,000 hect-ares. There will be no mobile phonesignals allowed, except in the few townsin the area. Residents will instead begiven an alternative radio communica-tion system. Save the Karoo, an advocacygroup, isn’t swayed by the prospect ofgroundbreaking astronomical discover-ies. Its members fear the restrictions willmake the Karoo “a cut-offand backwardregion”, and warn that abattoirs andwindmill repairmen serving farms nearthe SKA site could face financial ruin. “Icouldn’t give a damn about a blackholesitting somewhere out in space,” says EricTorr, an organiser with the group. “It doesnot put food on the table.”

Sky-high expectations in this down-at-heel area are also a problem. An SKAofficial grumbles that the locals expectthe telescope to solve all their woes.Some jobs have been created, but fewlocals have the skills to decipher thesecrets ofdistant galaxies. Until recentlythe high school in Carnarvon, a nearbytown, didn’t even have a maths andscience teacher. The SKA organisationhired one, and is also offering bursariesto college students. Perhaps if the nextgeneration’s horizons are raised, theywill be able to take advantage of the radiotelescopes in their own backyard.

THE KAROO

A telescope in the desert meets NIMBYism

Where lambs once frolicked

IT IS midnight at a shopping centre in Ho-pley, a poor suburb east of Harare. Mas-

celine, an orphaned 15-year-old, stands inthe darkness. Her aim is to find a client. Butshe also wants to avoid getting beaten upby older prostitutes who resent the compe-tition she represents.

Zimbabwe’s Constitutional Courtruled in 2015 that the police could not arrestwomen for prostitution. At first, sex work-ers cheered. No more would they bedragged to police stations and shakendown for bribes to avoid six-month jailterms. It was a relief, too, for the many oth-er women whom the cops used to arrestjust for walking alone at night or drinkingin a pub.

Yet there was a catch. Tambudzai Miko-rasi, a 40-year-old sexworker, saysshe mayhave celebrated too early. Nowthat the old-

Decriminalising Zimbabwe’s sex trade

Less stigma, morecompetitionHARARE

Why some prostitutes wish their jobwas still banned

32 Middle East and Africa The Economist January 7th 2017

1

2 est profession has been decriminalised, aflood of young women and girls are join-ing it and driving down prices. (Jobs ofanykind are scarce thanks to Robert Mugabe’sdisastrous management of the economy.)

Some veterans have responded by hir-ing thugs to protect their turf. In return, themen can have sex whenever they want.“We have no option,” says Ms Mikorasi,her muscles flexing as she pulls down herminiskirt. “It has never been this bad.These little girls are pushing us out of busi-ness.” Malaika Chatyoka, a 37-year-old,complains that the fee she can charge hasslumped from $10 for 30 minutes to as littleas $2 on some nights. The price for a fullnight is down to $10, one fifth of what sheused to charge.

Police raids used to scare away compet-itors. “Only the brave remained on thestreets. Now it’s free-for-all,” says 25-year-old Sazini Ngwenya from Bulawayo. Sheadds that without the police paying atten-tion there has been an increase in robber-ies and rape.

Faced with a slumping economy Zim-babweans are so desperate that, for many,even cut-price sex work seems like theleast-bad option. “With no education,faced with the responsibility to fend fortheir siblings and/or their own children,many girls and young women are beingforced to sell sex for survival,” laments Tal-ent Jumo, a director of the Katswe Sista-hood, a charity. With no sign of an eco-nomic recovery, many more girls could beforced onto the streets, she says.

For Masceline, the need to put food onthe table forher family means she will con-tinue renting out her body. “Everything Iface here is better than going hungry orwatching my baby sister drop out ofschool,” she says hurriedly, as a potentialclient approaches. 7

Supply and demeaned

THEY said it would be over by Christ-mas. NowHaideral-Abadi, Iraq’sprime

minister, is suggesting that the battle forMosul could last until Easter. For almost amonth his forces had stalled on what wassupposed to be the easier eastern bank ofIraq’s second city. And the costs have beengruelling. A fifth of Iraq’s elite force has re-portedly fallen in the assault. With the sup-port of more American special forces, MrAbadi has launched a second phase, takingthe city’s industrial zone. Progress is beingmade. But what Iraqi soldiers clear by day,Islamic State (IS) fighters often regain bynight, thanks to a warren of tunnels underthe front lines.

Not only are IS fighters holding the lineagainst Iraqi soldiers and their Americanbackers after ten weeks of fighting in Mo-sul, but they are also fending off Turkishtroops 515km (320 miles) to the west,around the town ofal-Bab in northern Syr-ia. They have also recaptured Palmyrafrom the Syrian regime.

Across Iraq the insurgency has a newlease of life. The sickening rhythm of sui-cide bombs in Shia suburbs of Baghdadand southern Iraq is quickening again. InAnbar and Salahuddin, provinces longsince reclaimed by the government, IS isalso flexing its muscles. On January 2nd itwon control of a police station in Samarraforseveral hoursand it cutbriefly the Bagh-dad to Mosul road. It is putting out lights inDiyala. “It is not an organisation that isclose to collapse,” says an analyst in touchwith people in Mosul.

The prolonged campaign carries politi-cal costs for Mr Abadi, who had sought toturn himself from a bumbling office-hold-

er into a victorious commander by don-ning military fatigues. Should there be fur-ther mishaps, Mr Abadi’s rivals in Baghdadwill be waiting to pounce. Among them ishis predecessor, Nouri al-Maliki, and hiscohorts in Iraq’s assortment of predomi-nantly Shia militias, which are collectivelyknown as the Hashd al-Shaabi, or popularmobilisation forces (PMF). As the govern-ment-led advance on Mosul slows, theyare calling for the deployment of Iranian-backed brigades. So short ofmen are Iraq’sarmy and police that even some Americancommanders now welcome the use ofthese auxiliaries.

To date the Iraqi government’s assault(with air support from the American-ledcoalition) has managed to minimise civil-ian casualties in Mosul. But if Shia militiasare unleashed another bout of sectariankillings might ensue. It could push thecity’s remaining 900,000 Sunnis towardsIS. Moreover an enhanced role for the mili-tias would weaken the state by boostingthe PMF. Its brigades already display fac-tional flags, run several secret prisons andraise money by extorting bribes at gun-point at checkpoints.

Mr Maliki, for his part, seems to be in-tent on building a sectarian pasdaran, orrevolutionary guard, much like the onethat wields real power in neighbouringIran. As prime minister from 2006 to 2014Mr Maliki built up networks that still givehim influence. From the judiciary to thecivil service to parliament, many in Iraq’supper echelons owe him their positions.One of his last acts as prime minister wasto form the PMF. Having tried for monthsto chip awayatMrAbadi’sauthorityby im-peaching his senior ministers, Mr Malikihas recently switched tack in favour of bol-stering the PMF. On November 26th MPsvoted to create an autonomous force com-prising110,000 PMF militiamen paid by thegovernment and overseen by parliament.

In response Mr Abadi is using the newPMF law to woo the Sunni militias that MrMaliki had ostracised when he was primeminister. He has brought thousands ofSunni officers and soldiers from SaddamHussein’s former army onto the govern-ment payroll under the command of a for-mer governor of Mosul, Atheel al-Nujaifi.Charges of terrorism against Khamis al-Khanjar, a Sunni politician financed by theGulf, have been quietlydropped. And withMr Abadi’s blessing, Ammar al-Hakim,who heads the largest Shia parliamentary

Iraq’s long war

A goody and Abadi

BAGHDAD

Behind the battle forMosul lies a struggle forpowerin Baghdad

A N B A R

S Y R I AI R A Q

IRAN

JORDAN

S A U D IA R A B I A

T U R K E Y

Damascus

Baghdad

Mosul

Samarra

AleppoRaqqa

Palmyra

al-Bab

DIYALA

SALAHUDDIN

Tigris Euphrates

200 km

Areas of IslamicState controlDecember 2016

Source: Institute forthe Study of War

Iraqi distributionof population

KurdishSunni ArabShia Arab

Sparsely populatedSource: CIA World Factbook

The Economist January 7th 2017 Middle East and Africa 33

1

2 bloc, has gone to Amman and Beirut to ne-gotiate terms for a national reconciliation,including an amnesty, with Sunni exileswho had long since despaired of a dealwith Baghdad’s Shia masters.

The hope is that if they are promised afuture inside Iraq, former Baathists andother Sunni Islamists might join forces torid Iraq of IS. “Some Shia are starting to re-alise that if they can’t absorb Sunnis andKurds, what remains of Iraq risks becom-inganotherwilaya [province] ofIran,” saysa diplomat in Baghdad.

Mr Abadi’s main failure has been politi-cal: he has not broken the hold that sectari-an parties, including his own, have onIraq’s coffers. Political parties in his cabinetcontinue to take handsome cuts from gov-ernment contracts and collect the pay ofghost workers in defunct factories. Ratherthan tackling political corruption, he hassquandered the backing of protesters andIraq’s leading ayatollahs by slashing thepay ofcivil servants and raising taxes.

Yet for all his setbacks, Mr Abadi has re-gained most of the territory that Mr Malikihad lost. He has secured renewed Ameri-can military support and overseen a four-fold increase in America’s troop deploy-ment over the past year. With Americanhelp, the panic-stricken army that fled Mo-sul in June 2014 has been rearmed andnow sports an air force and a division ofspecial-forces soldiers who are proving ca-pable fighters. His men are also operatingalongside Kurdish fighters for the first timein a decade.

Socially, too, Baghdad is regaining asemblance of normality. The concrete bar-ricades encasingpublic buildings like lugu-brious tombstones have slowly comedown, and checkpoints have thinned.Iraq’s economy is now weathering a crisiscaused by a slump in the oil price and asurge in war-related spending. Oil produc-tion reached record levels in 2016 and theIMF has extended a $5.3bn loan, whichpromises to attract an additional $11bn ininternational creditand exportguarantees.The bits of the economy not related to oil,which slumped 5% in 2016 and 14% the yearbefore, will probably expand by 5% thisyear, the IMF reckons. American compa-nies that had previously fled as Iraqslipped into mayhem, such as GeneralElectric, are now tiptoeing back.

Even in parliament unusual alliancesare forming across sectarian lines. Kurds,Sunni Arabs and Shias no longer vote asunited blocs. Selim Jabbouri, the Sunniparliamentary speaker who hails from theMuslim Brotherhood, speaks of establish-ing a cross-confessional party in the 2018elections and canvasses in Shia as well asSunni parts of Iraq. For all its faults, Iraq isstill the Arab world’s most boisterousmultiparty democracy. Perhaps it mayafter all convince a majority of its peoplethat they have a future together. 7

WITH its shopping malls, sports cen-tres and new residential blocks,

Maale Adumim (pictured above) looksand feels like any other Israeli dormitorysuburb. Less than ten minutes’ drive east-ward from Jerusalem, the town on theedge of the Judaean Desert, which is alsothe third-largest Jewish settlement in theWest Bank, may be about to become thefirst test case ofAmerica’s Middle East poli-cy under Donald Trump.

Elements of Binyamin Netanyahu’sright-wing coalition are agitating to annexparts of the West Bank, starting with MaaleAdumim, in the belief that the Trump ad-ministration will reverse a long-standingAmerican policy opposing Israeli settle-ments in territory it occupied in 1967.

The mayor, Benny Kashriel, says that heis not concerned about international poli-tics so much as local laws: “I wantmytownto have the same rights as any other townin Israel.” He complains that, because theWest Bank (called Judea and Samaria bythe Israeli government) is under militaryrule, Maale Adumim’s residents need toapply to the army iftheywish, for instance,to close a veranda: “There is no reason thatthe Israeli law shouldn’t apply here. Netan-yahu tells us he is in favour but that wehave to wait for better timing.” Emphasis-ing that the time may indeed be approach-ing, this week Maale Adumim welcomedthe former presidential candidate and go-vernor of Arkansas, Mike Huckabee, to in-augurate a new building.

Naftali Bennett, the leader of the hard-

right Jewish Home party, says he will pro-pose a bill to extend Israeli sovereignty toMaale Adumim, arguing that the newAmerican administration offers “a uniquewindow of opportunity” to redraw themap. “For the first time in 50 yearswe Israe-lis have to decide what we want—a Pales-tinian state in Judea and Samaria, or Israelilaw replacing military law where Israelicitizens live.” Mr Bennett believes that heand his colleagues must press the primeminister to persuade Mr Trump to recog-nise the settlements as permanent at hisfirst meeting.

After eight years of friction with BarackObama’s administration, Mr Netanyahuhas good reason to feel optimistic. He is ex-pected to be the first foreign leader to meetthe new president after his inaugurationon January 20th. Although Mr Netanyahumaynotgeta promise to scrap the Iran nuc-lear deal, in every other way he will bepushing on an open door when it comes toenlisting Mr Trump’s help to counter Irani-an influence in the region.

Then there is the pre-election promiseby Mr Trump to relocate the American em-bassy from Tel Aviv to Jerusalem, with allthe symbolism that would involve. MrTrump’s pick of David Friedman to beAmerica’s ambassador to Israel has alsosent a strong message. A bankruptcy law-yer without diplomatic experience, MrFriedman believes that Israel is legally enti-tled to annex the West Bank and supportsthe building of new settlements there—steps that would rule out any possibility of

America and Israel

Unsettled

MAALE ADUMIM

President Trump may not be all good news forBinyamin Netanyahu

34 Middle East and Africa The Economist January 7th 2017

2 Israel’s divisions

Uniform justice

ONE of the most contentious cases inIsrael’s military history reached its

verdict on January 4th when three mil-itary judges found a serving soldier,Sergeant Elor Azaria, guilty ofman-slaughter for killing a Palestinian.

The public controversy was not overthe facts. Both the prosecution and de-fence agreed that on March 24th 2016, inthe West Bankcity ofHebron, Mr Azariahad fired point-blankat Abdel-Fattahal-Sharif, a Palestinian man lying griev-ously wounded after he had been shotwhile stabbing an Israeli soldier.

Nor was it over the court’s dismissalofMr Azaria’s claim to have been actingin self-defence; the judges reached aunanimous decision that he had acted“calmly, without urgency and in a calcu-lated manner” and that, as he said on thescene to a comrade, he thought Mr al-Sharif “deserve[d] to die”.

Instead the controversy relates to thefact that a large section of the Israeli

public seems to believe that Mr Azariawas right to have shot a wounded prison-er who no longer posed a danger. A polllast August by the Israeli DemocracyInstitute and Tel Aviv University indicat-ed that 65% of Israel’s Jewish majoritysupported his actions. A seemingly con-tradictory finding in the same poll putpublic support of the Israel DefenceForces (IDF), the same organisation thatput Sergeant Azaria on trial, at 87%.

One explanation for this discrepancyis that anger is high over the sporadiccampaign ofstabbings by Palestinianssince 2015. Another is that admiration forthe army does not necessarily extend tothe generals, who rushed to condemnhim when video of the shootingemerged. This mixed message is comingfrom the politicians, too. The primeminister, Binyamin Netanyahu, called MrAzaria’s parents to express solidarityshortly after his arrest. Since the convic-tion he joined several ministers in callingfor Mr Azaria to be pardoned by thepresident, Reuven Rivlin. The only seniormember of the ruling Likud party tocondemn the shooting unequivocallywas Moshe Yaalon, then the defenceminister. Shortly afterwards he waspushed out ofoffice and replaced byAvigdor Lieberman, a hardliner.

Manslaughter convictions of IDFsoldiers on duty are very rare. The lastwas11years ago when one was foundguilty ofkilling Tom Hurndall, a Britishpro-Palestinian activist. Human-rightsorganisations say other similar shootingshave gone unpunished.

Still, the generals can take comfortfrom the judges’ firm line: the IDF’s claimto be a highly moral army requires it toact against cases ofblatant indiscipline.And Mr Netanyahu, though bending topopulist sentiment at home, will doubt-less hold out the verdict as proof to theworld that his country is a democracywhere the rule of law prevails, in a regionwhere such virtues are rare.

TEL AVIV

The conviction ofa soldierdivides Israel

Rallying for the right to shoot prisoners

a peace deal based on the establishment ofa Palestinian state next to Israel.

The gulf between the outgoing and in-coming administrations was laid bare onDecember 23rd. Mr Obama’s decision notto veto UN Security Council Resolution2334 (reiterating that the settlements are il-legal and expressing concern that pros-pects for the two-state solution were beingsabotaged by both sides) was a well-aimedparting shot. Mr Trump’s administrationwill not be able to overturn a resolutionthat may embolden the InternationalCriminal Court to take action over settle-ments; Israeli officials will have to considerthe risks ofbuilding more.

Mr Netanyahu’s fury must at least havebeen soothed by Mr Trump’s response onTwitter: “Stay strong Israel, January 20th isfast approaching!” The Israeli prime minis-ter tweeted back: “President-elect Trump,thank you for your warm friendship andyour clear-cut support for Israel!”

A few hours later America’s secretaryofstate, John Kerry, in a speech highly criti-cal of Israel’s government, explained thecontext of the UN abstention. He warnedthat right-wing ideologues within Mr Net-anyahu’s coalition were leading Israel in-exorably towards abandoning even thepretence of interest in a two-state solution,with profound consequences for both thecountry’s future security and its status as ademocratic Jewish state. There was little inMr Kerry’s speech, which also forthrightlycondemned Palestinian glorification ofterrorism, that deviated from Americanpolicy towards Israel that goes back morethan 40 years. But his disappointment andfrustration were clear.

In practice, Mr Netanyahu may not findMr Trump’s uncritical friendship an unal-loyed blessing. It has often suited him toplay off hardliners, such as Mr Bennett,against Washington. It has served his pur-pose to keep the idea of the two-state sol-ution alive while doing nothing to helpmake it a reality. Typically, he has yet to re-spond to the latest demands for annexingMaale Adumim. One of Mr Netanyahu’sallies, the regional co-operation minister,Tzachi Hanegbi, warned this week that “itwould be bad for Israel to unilaterally an-nex Judea and Samaria.” However, manymembers of the ruling Likud party also fa-vour the Maale Adumim law.

The right wing has scented an opportu-nity in legal problems facing Mr Netanya-hu. The prime minister was questionedthis week by police over fraud and graft al-legations. This is one of a number of cor-ruption probes into his financial affairsand the prime minister is suddenly vulner-able. Although he insists that there is noth-ing of substance in the allegations, he willnot have forgotten how his predecessor,Ehud Olmert, was forced by his cabinetcolleagues eight years ago to resign over al-legations of bribe-taking, and was later

jailed. Mr Netanyahu may have to concedesomething to the hardliners to keep hisown job.

“The Maale Adumim lawis the first signthat the Netanyahu government is using itsnewfound power in the Trump era to makeunilateral moves,” says Danny Seide-mann, a director of Terrestrial Jerusalem,an NGO that monitors building works inand around Jerusalem. “Annexing MaaleAdumim would virtually cut the West

Bank in two, making it almost impossibleto establish in the future a viable Palestin-ian state.” He doubts that Mr Netanyahuwants to go that far, which means he mayhave to throw his right-wingers othermeat, such as more settlement-building ineast Jerusalem and the West Bank. Delay-ing tactics and ambiguity have been thetwin hallmarks of Mr Netanyahu’s pre-miership. Both may become harder withthe advent ofMr Trump. 7

TECHNOLOGY QUARTERLY January 7th 2017 LANGUAGE

Finding a voice

The Economist January 7th 2017 3

1

I’M SORRY, Dave. I’m afraid I can’t do that.”With chilling calm, HAL 9000, the on-board computer in “2001: A Space Odys-sey”, refuses to open the doors to DaveBowman, an astronaut who had venturedoutside the ship. HAL’s decision to turn on

his human companion reflected a wave of fear aboutintelligent computers.

When the film came out in 1968, computers thatcould have proper conversations with humansseemed nearly as far away as manned flight to Jupiter.Since then, humankind has progressed quite a lot far-therwith buildingmachines that it can talkto, and thatcan respond with something resembling naturalspeech. Even so, communication remains difficult. If“2001” had been made to reflect the state of today’slanguage technology, the conversation might havegone something like this: “Open the pod bay doors,Hal.” “I’m sorry, Dave. I didn’t understand the ques-tion.” “Open the pod bay doors, Hal.” “I have a list ofeBay results about pod doors, Dave.”

Creative and truly conversational computers able

to handle the unexpected are still far off. Artificial-in-telligence (AI) researchers can only laugh when askedabout the prospectofan intelligentHAL, TerminatororRosie (the sassy robot housekeeper in “The Jetsons”).Yet although language technologies are nowhere nearready to replace human beings, except in a few highlyroutine tasks, they are at last about to become goodenough to be taken seriously. They can help peoplespend more time doing interesting things that only hu-mans can do. After six decades of work, much of itwith disappointing outcomes, the past few years haveproduced results much closer to what early pioneershad hoped for.

Speech recognition has made remarkable ad-vances. Machine translation, too, has gone from terri-ble to usable for getting the gist ofa text, and may soonbe good enough to require only modest editing by hu-mans. Computerised personal assistants, such as Ap-ple’s Siri, Amazon’s Alexa, Google Now and Micro-soft’s Cortana, can now take a wide variety ofquestions, structured in many different ways, and re-turn accurate and useful answers in a natural-sound-

ALSO IN THIS TQ

SPEECH RECOGNITIONI hear you

SYNTHETIC SPEECHHasta la vista,robotvoice

MACHINE TRANSLATIONBeyond Babel

MEANING AND MACHINEINTELLIGENCEWhat are you talkingabout?

BRAIN SCANTerry Winograd

LOOKING AHEADFor my next trick

TECHNOLOGY QUARTERLY Language

“

Finding a voice

Computers have got much better at translation, voice recognition and speech synthesis, says LaneGreene. But they still don’t understand the meaning of language

2 ing voice. Alexa can even respond to a request to “tell me a joke”,but only by calling upon a database of corny quips. Computerslacka sense ofhumour.

When Apple introduced Siri in 2011 it was frustrating to use, somany people gave up. Only around a third ofsmartphone ownersuse their personal assistants regularly, even though 95% have triedthem at some point, according to Creative Strategies, a consultan-cy. Many of those discouraged users may not realise how muchthey have improved.

In 1966 John Pierce was working at Bell Labs, the research armofAmerica’s telephone monopoly. Havingoverseen the team thathad built the first transistor and the first communications satellite,he enjoyed a sterling reputation, so he was asked to take charge ofa report on the state of automatic language processing for the Na-tional Academy ofSciences. In the period leadingup to this, schol-ars had been promising automatic translation between languageswithin a few years.

But the report was scathing. Reviewing almost a decade ofworkon machine translation and automatic speech recognition, itconcluded that the time had come to spend money “hard-head-edly toward important, realisticand relativelyshort-range goals”—another way of saying that language-technology research hadoverpromised and underdelivered. In 1969 Pierce wrote that boththe funders and eager researchers had often fooled themselves,and that “no simple, clear, sure knowledge is gained.” After that,America’s government largely closed the money tap, and researchon language technology went into hibernation for two decades.

The story of how it emerged from that hibernation is both sal-utary and surprisingly workaday, says Mark Liberman. As profes-sorof linguistics at the University ofPennsylvania and head oftheLinguistic Data Consortium, a huge trove of texts and recordingsof human language, he knows a thing or two about the history oflanguage technology. In the bad old days researchers kept theirmethods in the dark and described their results in ways that werehard to evaluate. But beginning in the 1980s, Charles Wayne, thenatAmerica’sDefence Advanced Research ProjectsAgency, encour-aged them to try another approach: the “common task”.

Step by stepResearchers would agree on a common set of practices, whetherthey were trying to teach computers speech recognition, speakeridentification, sentiment analysis of texts, grammatical break-down, language identification, handwriting recognition or any-thing else. They would set out the metrics they were aiming to im-prove on, share the data sets used to train their software and allowtheir results to be tested by neutral outsiders. That made the pro-cess far more transparent. Funding started up again and languagetechnologies began to improve, though very slowly.

Many early approaches to languagetechnology—and particularly translation—got stuck in a conceptual cul-de-sac: therules-based approach. In translation, thismeant trying to write rules to analyse thetext ofa sentence in the language of origin,breaking it down into a sort of abstract“interlanguage” and rebuilding it accord-ing to the rules of the target language.These approaches showed early promise.But language is riddled with ambiguitiesand exceptions, so such systems werehugely complicated and easily brokedown when tested on sentences beyondthe simple set they had been designed for.

Nearly all language technologies began to get a lot better withthe application of statistical methods, often called a “brute force”approach. This relies on software scouring vast amounts of data,looking for patterns and learning from precedent. For example, inparsing language (breaking it down into its grammatical compo-nents), the software learns from large bodies of text that have al-ready been parsed by humans. It uses what it has learned to makeits best guess about a previously unseen text. In machine transla-tion, the software scans millions of words already translated byhumans, again looking for patterns. In speech recognition, thesoftware learns from a body of recordings and the transcriptionsmade by humans.

Thanks to the growing power of processors, falling prices fordata storage and, most crucially, the explosion in available data,this approach eventually bore fruit. Mathematical techniques thathad been known fordecadescame into theirown, and big compa-nies with access to enormous amounts of data were poised tobenefit. People who had been put off by the hilariously inappro-priate translations offered by online tools like BabelFish began tohave more faith in Google Translate. Apple persuaded millions ofiPhone users to talknot only on their phones but to them.

The final advance, which began only about five years ago,came with the advent of deep learning through digital neural net-works (DNNs). These are often touted as having qualities similarto those ofthe human brain: “neurons” are connected in software,and connections can become stronger or weaker in the process oflearning. But Nils Lenke, head of research for Nuance, a language-technology company, explains matter-of-factly that “DNNs arejust another kind ofmathematical model,” the basis ofwhich hadbeen well understood for decades. What changed was the hard-ware being used.

Almostbychance, DNN researchersdiscovered that the graphi-cal processing units (GPUs) used to render graphics fluidly in ap-

plications like video games were also bril-liant at handling neural networks. Incomputer graphics, basic small shapesmove according to fairly simple rules, butthere are lots of shapes and many rules, re-quiring vast numbers of simple calcula-tions. The same GPUs are used to fine-tunethe weightsassigned to “neurons” in DNNsas they scour data to learn. The techniquehas already produced big leaps in qualityfor all kinds ofdeep learning, including de-ciphering handwriting, recognising facesand classifying images. Now they are help-ing to improve all manner of languagetechnologies, often bringing enhance-ments of up to 30%. That has shifted lan-guage technology from usable at a pinch toreally rather good. But so far no one hasquite worked out what will move it onfrom merely good to reliably great. 7

1954 60 1965 70 75 80 85 90 95 2000 05 10 16

Source: The Economist

Scientists from IBM and Georgetown demonstratea limited machine-translation system

John Pierce’s highly critical report on language technologiespublished. Funding languishes for decades

Dawn of “common task” method. Researchers share data, agree on common methods of evaluation Google releases neural-net machine

translation for eight language pairs

Now I understandA history of language technologies

“2001: A Space Odyssey”released

Statistics-based version ofGoogle Translate launched

Siri debuts on iPhone

“Hey Siri”

Microsoft speech-recognitionsystem reaches human parity

No US governmentresearch fundingfor machinetranslationor speechrecognition

Many earlyapproachesto languagetechnologygot stuck ina conceptualcul-de-sac

4 The Economist January 7th 2017

TECHNOLOGY QUARTERLY Language

The Economist January 7th 2017 5

TECHNOLOGY QUARTERLY Language

1

WHEN a person speaks, air is forced out through thelungs, making the vocal chords vibrate, whichsends out characteristic wave patterns through theair. The features of the sounds depend on the ar-rangement of the vocal organs, especially the ton-gue and the lips, and the characteristicnature ofthe

sounds comes from peaks of energy in certain frequencies. Thevowels have frequencies called “formants”, two ofwhich are usu-ally enough to differentiate one vowel from another. For example,the vowel in the English word “fleece” has its first two formants ataround 300Hz and 3,000Hz. Consonants have their own charac-teristic features.

In principle, it should be easy to turn this stream of sound intotranscribed speech. As in other language technologies, machinesthat recognise speech are trained on data gathered earlier. In thisinstance, the training data are sound recordings transcribed to textby humans, so that the software has both a sound and a text input.All it has to do is match the two. It gets better and better at workingout how to transcribe a given chunk of sound in the same way ashumans did in the training data. The traditional matching ap-proach was a statistical technique called a hidden Markov model(HMM), making guesses based on what was done before. More re-cently speech recognition has also gained from deep learning.

English has about 44 “phonemes”, the units that make up thesound system of a language. P and b are different phonemes, be-cause theydistinguish words like patand bat. But in English p witha puff of air, as in “party”, and p without a puff of air, as in “spin”,are not different phonemes, though they are in other languages. Ifa computerhears the phonemes s, p, i and n backto back, it shouldbe able to recognise the word “spin”.

But the nature of live speech makes this difficult for machines.Sounds are not pronounced individually, one phoneme after theother; they mostly come in a constant stream, and finding theboundaries is not easy. Phonemes also differ according to the con-text. (Compare the l sound at the beginning of “light” with that atthe end of “full”.) Speakers differ in timbre and pitch of voice, andin accent. Conversation is far less clear than careful dictation. Peo-ple stop and restart much more often than they realise.

All the same, technology has gradually mitigated many ofthese problems, so error rates in speech-recognition software havefallen steadily over the years—and then sharply with the introduc-tion of deep learning. Microphones have got better and cheaper.With ubiquitouswireless internet, speech recordingscan easilybebeamed to computers in the cloud for analysis, and even smart-phones now often have computers powerful enough to carry outthis task.

Bear arms orbare arms?Perhaps the most important feature of a speech-recognition sys-tem is its setofexpectationsaboutwhatsomeone is likely to say, orits “language model”. Like other training data, the language mod-els are based on large amounts of real human speech, transcribedinto text. When a speech-recognition system “hears” a stream ofsound, it makes a number of guesses about what has been said,then calculates the odds that it has found the right one, based onthe kinds of words, phrases and clauses it has seen earlier in thetraining text.

At the level of phonemes, each language has strings that arepermitted (in English, a word may begin with str-, for example) orbanned (an English word cannot start with tsr-). The same goes forwords. Some strings of words are more common than others. Forexample, “the” is farmore likely to be followed by a noun oran ad-jective than by a verb or an adverb. In making guesses about ho-mophones, the computer will have remembered that in its train-ing data the phrase “the right to bear arms” came up much moreoften than “the right to bare arms”, and will thus have made theright guess.

Training on a specific speaker greatly cuts down on the soft-ware’s guesswork. Just a few minutes of reading training text intosoftware like Dragon Dictate, made by Nuance, produces a bigjump in accuracy. For those willing to train the software for longer,the improvement continues to something close to 99% accuracy(meaning that ofeach hundred words oftext, not more than one iswrongly added, omitted or changed). A good microphone and aquiet room help.

Advance knowledge ofwhat kinds of things the speaker mightbe talking about also increases accuracy. Words like “phlebitis”and “gastrointestinal” are not common in general discourse, anduncommon words are ranked lower in the probability tables thesoftware uses to guess what it has heard. But these words are com-mon in medicine, so creating software trained to look out for suchwords considerably improves the result. This can be done by feed-ing the system a large number ofdocuments written by the speak-er whose voice is to be recognised; common words and phrasescan be extracted to improve the system’s guesses.

As with all other areas of language technology, deep learninghas sharply brought down error rates. In October Microsoft an-

Speech recognition

I hear you

Computers have made huge strides in recognising humanspeech

6 The Economist January 7th 2017

TECHNOLOGY QUARTERLY Language

2 nounced that its latest speech-recognition system had achievedparity with human transcribers in recognising the speech in theSwitchboard Corpus, a collection of thousands of recorded con-versations in which participants are talking with a stranger abouta randomly chosen subject.

Error rates on the Switchboard Corpus are a widely usedbenchmark, so claimsofquality improvementscan be easily com-pared. Fifteen years ago quality had stalled, with word-error ratesof 20-30%. Microsoft’s latest system, which has six neural net-works running in parallel, has reached 5.9% (see chart), the same asa human transcriber’s. Xuedong Huang, Microsoft’s chief speechscientist, says that he expected it to take two or three years to reachparity with humans. It got there in less than one.

The improvements in the lab are now being applied to pro-ducts in the real world. More and more cars are being fitted withvoice-activated controls of various kinds; the vocabulary in-volved is limited (there are only so many things you might want tosay to yourcar), which ensureshigh accuracy. Microphones—or of-ten arrays of microphones with narrow fields of pick-up—are get-ting better at identifying the relevant speaker among a group.

Some problems remain. Children and elderly speakers, as wellas people moving around in a room, are harder to understand.Background noise remains a big concern; if it is different from thatin the training data, the software finds it harder to generalise fromwhat it has learned. So Microsoft, for example, offers businesses aproduct called CRIS that lets users customise speech-recognition

systems for the background noise, special vocabulary and otheridiosyncrasies theywill encounter in thatparticularenvironment.That could be useful anywhere from a noisy factory floor to a carehome for the elderly.

But fora computer to knowwhata human hassaid isonly a be-ginning. Proper interaction between the two, of the kind thatcomes up in almost every science-fiction story, calls for machinesthat can speakback (see box below). 7

Loud and clearSpeech-recognition word-error rate, selected benchmarks, %

The Switchboard corpus is a collection of recordedtelephone conversations widely used to train andtest speech-recognition systems

1993

Sources: Microsoft; research papers

96 98 2000 02 04 06 08 10 12 14 16

Broadcast speech

Meeting speech

Microsoft, Switchboard

IBM, Switchboard

Switchboard Switchboard cellular

5.9%

100Log scale

10

1

Hasta la vista, robot voice

Machines are starting to sound more like humans“I’LL be back.” “Hasta la vista, baby.”Arnold Schwarzenegger’s Teutonic dronein the “Terminator” films is world-fam-ous. But in this instance film-makerslooking into the future were overly pessi-mistic. Some applications do still featurea monotonous “robot voice”, but that ischanging fast.

Creating speech is roughly the inverseofunderstanding it. Again, it requires abasic model of the structure ofspeech.What are the sounds in a language, andhow do they combine? What words doesit have, and how do they combine insentences? These are well-understoodquestions, and most systems can nowgenerate sound waves that are a fairapproximation ofhuman speech, at leastin short bursts.

Heteronyms require special care. Howshould a computer pronounce a wordlike “lead”, which can be a present-tenseverb or a noun for a heavy metal, pro-nounced quite differently? Once again alanguage model can make accurateguesses: “Lead us not into temptation”can be parsed for its syntax, and once thesoftware has worked out that the firstword is almost certainly a verb, it cancause it to be pronounced to rhyme with“reed”, not “red”.

Traditionally, text-to-speech models

have been “concatenative”, consisting ofvery short segments recorded by a hu-man and then strung together as in theacoustic model described above. Morerecently, “parametric” models have beengenerating raw audio without the need torecord a human voice, which makesthese systems more flexible but lessnatural-sounding.

DeepMind, an artificial-intelligencecompany bought by Google in 2014, hasannounced a new way ofsynthesisingspeech, again using deep neural net-works. The network is trained on record-ings ofpeople talking, and on the textsthat match what they say. Given a text toreproduce as speech, it churns out a farmore fluent and natural-sounding voicethan the best concatenative and paramet-ric approaches.

The last step in generating speech isgiving it prosody—generally, the mod-ulation ofspeed, pitch and volume toconvey an extra (and critical) channel ofmeaning. In English, “a German teacher”,with the stress on “teacher”, can teachanything but must be German. But “aGerman teacher” with the emphasis on“German” is usually a teacher ofGerman(and need not be German). Words likeprepositions and conjunctions are notusually stressed. Getting machines to put

the stresses in the correct places is about50% solved, says MarkLiberman of theUniversity ofPennsylvania.

Many applications do not requireperfect prosody. A satellite-navigationsystem giving instructions on where toturn uses just a small number ofsentencepatterns, and prosody is not important.The same goes for most single-sentenceresponses given by a virtual assistant on asmartphone.

But prosody matters when someone istelling a story. Pitch, speed and volumecan be used to pass quickly over thingsthat are already known, or to build interestand tension for new information. Myriadtiny clues communicate the speaker’sattitude to his subject. The phrase “a Ger-man teacher”, with stress on the word“German”, may, in the context ofa story,not be a teacher ofGerman, but a teacherbeing explicitly contrasted with a teacherwho happens to be French or British.

Text-to-speech engines are not muchgood at using context to provide suchaccentuation, and where they do, it rarelyextends beyond a single sentence. WhenAlexa, the assistant in Amazon’s Echodevice, reads a news story, her prosody isjarringly un-humanlike. Talking comput-ers have yet to learn how to make humanswant to listen.

The Economist January 7th 2017 7

TECHNOLOGY QUARTERLY Language

1

IN “STAR TREK” it was a hand-held Universal Translator; in“The Hitchhiker’s Guide to the Galaxy” it was the BabelFish popped conveniently into the ear. In science fiction,the meetingofdistantcivilisationsgenerally requires somekind ofdevice to allow them to talk. High-quality automat-ed translation seems even more magical than other kinds

of language technology because many humans struggle to speakmore than one language, let alone translate from one to another.

The idea has been around since the 1950s, and computerisedtranslation is still known by the quaint moniker “machine transla-tion” (MT). It goes back to the early days of the cold war, whenAmerican scientists were trying to get computers to translate fromRussian. They were inspired by the code-breaking successes of thesecond world war, which had led to the development of comput-ers in the first place. To them, a scramble of Cyrillic letters on apage of Russian text was just a coded version of English, and turn-ing it into English was just a question ofbreaking the code.

Scientists at IBM and Georgetown University were amongthose who thought that the problem would be cracked quickly.Having programmed just six rules and a vocabulary of 250 wordsinto a computer, they gave a demonstration in New York on Janu-ary 7th 1954 and proudly produced 60 automated translations, in-cluding that of “Mi pyeryedayem mislyi posryedstvom ryechyi,”which came out correctly as “We transmit thoughts by means ofspeech.” Leon Dostert of Georgetown, the lead scientist, breezilypredicted that fully realised MT would be “an accomplished fact”in three to five years.

Instead, after more than a decade ofwork, the report in 1966 bya committee chaired by John Pierce, mentioned in the introduc-tion to this report, recorded bitter disappointment with the resultsand urged researchers to focus on narrow, achievable goals suchas automated dictionaries. Government-sponsored work on MTwent into near-hibernation for two decades. What little was donewas carried out by private companies. The most notable of themwas Systran, which provided rough translations, mostly to Ameri-ca’s armed forces.

La plume de mon ordinateurThe scientists got bogged down by their rules-based approach.Having done relatively well with their six-rule system, they cameto believe that if they programmed in more rules, the systemwould become more sophisticated and subtle. Instead, it becamemore likely to produce nonsense. Adding extra rules, in the mod-ern parlance ofsoftware developers, did not “scale”.

Besides the difficulty of programming grammar’s many rulesand exceptions, some early observers noted a conceptual pro-blem. The meaning of a word often depends not just on its dictio-nary definition and the grammatical context but the meaning ofthe restofthe sentence. Yehoshua Bar-Hillel, an Israeli MT pioneer,realised that “the pen is in the box” and “the box is in the pen”would require different translations for“pen”: any pen bigenoughto hold a box would have to be an animal enclosure, not a writinginstrument.

How could machines be taught enough rules to make this kindofdistinction? Theywould have to be provided with some knowl-edge of the real world, a taskfar beyond the machines or their pro-grammers at the time. Two decades later, IBM stumbled on an ap-

proach that would revive optimism about MT. Its Candide systemwas the first serious attempt to use statistical probabilities ratherthan rules devised by humans for translation. Statistical, “phrase-based” machine translation, like speech recognition, neededtraining data to learn from. Candide used Canada’s Hansard,which publishes that country’s parliamentary debates in Frenchand English, providing a huge amount of data for that time. Thephrase-based approach would ensure that the translation of aword would take the surrounding words properly into account.

Butqualitydid not take a leap until Google, which had set itselfthe goal of indexing the entire internet, decided to use those datato train its translation engines; in 2007 it switched from a rules-based engine (provided by Systran) to its own statistics-based sys-tem. To build it, Google trawled abouta trillion web pages, lookingfor any text that seemed to be a translation of another—for exam-ple, pages designed identically but with different words, and per-haps a hint such as the address of one page ending in /en and theother ending in /fr. According to Macduff Hughes, chief engineeron Google Translate, a simple approach using vast amounts ofdata seemed more promising than a clever one with fewer data.

Training on parallel texts (which linguists call corpora, the plu-ral of corpus) creates a “translation model” that generates not onebuta seriesofpossible translations in the target language. The nextstep is running these possibilities through a monolingual lan-guage model in the target language. This is, in effect, a set ofexpec-tations about what a well-formed and typical sentence in the tar-get language is likely to be. Single-language models are not toohard to build. (Parallel human-translated corpora are hard to comeby; large amounts of monolingual training data are not.) As withthe translation model, the language model uses a brute-force sta-tistical approach to learn from the training data, then ranks theoutputs from the translation model in order ofplausibility.

Statistical machine translation rekindled optimism in the field.Internet users quickly discovered that Google Translate was farbetter than the rules-based online engines they had used before,such as BabelFish. Such systems still make mistakes—sometimesminor, sometimes hilarious, sometimes so serious or so many asto make nonsense of the result. And language pairs like Chinese-English, which are unrelated and structurally quite different,make accurate translation harder than pairs of related languages

Machine translation

Beyond Babel

Computer translations have got strikingly better, but stillneed human input

For the former secretary of state, this is to forget a month of bungling and convince the audience that Mr

Trump has not the makings of a president

The former secretary of state has to put behind her a month of setbacks

and convince the audience that Mr Trump

does not have what it takes to be a president

Speak easy Human scorers’ rating* of Google Translate and human translation

3 4 5 Perfect translation=6

Chinese

Source: Google *0=completely nonsense translation, 6=perfect translation †Machine translation

Spanish

Pour l’ancienne secrétaire d’Etat, il s’agit de faire oublier un mois de cafouillages et de convaincre l’auditoire que M. Trump n’a pas l’étoffe

d’un président

For the former secretary of state, it is a question of forgetting a month of muddles and convincing

the audience that Mr Trump does not have the

stuff of a president

Neural-network† Human

Input sentence

French

Spanish

French

Chinese

English

English

English

English

Phrase-based†

(2007)Neural-network†

(2016)HumanTranslation method

Phrase-based†

8 The Economist January 7th 2017

TECHNOLOGY QUARTERLY Language

2

1

like English and German. But more often than not, Google Trans-late and its free online competitors, such as Microsoft’s Bing Trans-lator, offer a usable approximation.

Such systems are set to get better, again with the help of deeplearning from digital neural networks. The Association for Com-putational Linguistics has been holding workshops on MT everysummer since 2006. One of the events is a competition betweenMT engines turned loose on a collection of news text. In August2016, in Berlin, neural-net-based MT systems were the top per-formers (out of102), a first.

Now Google has released its own neural-net-based engine foreight language pairs, closing much of the quality gap between itsold system and a human translator. This is especially true forclosely related languages (like the big European ones) with lots ofavailable trainingdata. The resultsare still distinctly imperfect, butfar smoother and more accurate than before. Translations be-tween English and (say) Chinese and Korean are not as good yet,but the neural system has brought a clear improvement here too.

The Coca-Cola factorNeural-network-based translation actually uses two networks.One is an encoder. Each word of an input sentence is convertedinto a multidimensional vector (a series ofnumerical values), andthe encoding of each new word takes into account what has hap-pened earlier in the sentence. Marcello Federico of Italy’s Fonda-zione Bruno Kessler, a private research organisation, uses an in-triguing analogy to compare neural-net translation with thephrase-based kind. The latter, he says, is like describing Coca-Colain terms of sugar, water, caffeine and other ingredients. By con-trast, the former encodes features such as liquidness, darkness,sweetness and fizziness.

Once the source sentence is encoded, a decoder network gen-erates a word-for-word translation, once again taking account ofthe immediately preceding word. This can cause problems whenthe meaning of words such as pronouns depends on words men-tioned much earlier in a long sentence. This problem is mitigatedby an “attention model”, which helps maintain focus on otherwords in the sentence outside the immediate context.

Neural-network translation requires heavy-duty computingpower, both for the original training of the system and in use. Theheart of such a system can be the GPUs that made the deep-learn-ing revolution possible, or specialised hardware like Google’s Ten-sorProcessingUnits (TPUs). Smaller translation companiesand re-searchers usually rent this kind of processing power in the cloud.But the data sets used in neural-networktrainingdo not need to beas extensive as those forphrase-based systems, which should givesmaller outfits a chance to compete with giants like Google.

Fully automated, high-quality machine translation is still along way off. For now, several problems remain. All current mach-ine translations proceed sentence by sentence. If the translation ofsuch a sentence depends on the meaning ofearlier ones, automat-ed systems will make mistakes. Long sentences, despite tricks likethe attention model, can be hard to translate. And neural-net-based systems in particular struggle with rare words.

Training data, too, are scarce for many language pairs. They areplentiful between European languages, since the European Un-ion’s institutions churn out vast amounts ofmaterial translated byhumans between the EU’s 24 official languages. But for smallerlanguages such resources are thin on the ground. For example,there are few Greek-Urdu parallel texts available on which to traina translation engine. So a system that claims to offer such transla-tion is in fact usually running it through a bridging language, near-ly always English. That involves two translations rather than one,multiplying the chance oferrors.

Even if machine translation is not yet perfect, technology canalready help humans translate much more quickly and accurately.“Translation memories”, software that stores already translatedwords and segments, first came into use as early as the 1980s. Forsomeone who frequently translates the same kind of material(such as instruction manuals), they serve up the bits that have al-ready been translated, saving lots ofduplication and time.

A similar trick is to train MT engines on text dealing with a nar-row real-world domain, such as medicine or the law. As softwaretechniques are refined and computers get faster, training becomeseasier and quicker. Free software such as Moses, developed withthe support of the EU and used by some of its in-house translators,

can be trained by anyone with parallel cor-pora to hand. A specialist in medical trans-lation, for instance, can train the system onmedical translations only, which makesthem far more accurate.

At the other end of linguistic sophistica-tion, an MT engine can be optimised forthe shorter and simpler language peopleuse in speech to spew out rough but near-instantaneous speech-to-speech transla-tions. This is what Microsoft’s Skype Trans-lator does. Its quality is improved by beingtrained on speech (things like film subtitlesand common spoken phrases) rather thanthe kind of parallel text produced by theEuropean Parliament.

Translation management has also ben-efited from innovation, with clever soft-ware allowing companies quickly to com-bine the best of MT, translation memory,customisation by the individual translatorand so on. Translation-management soft-ware aims to cut out the agencies that havebeen actingas middlemen between clientsand an army of freelance translators. JackWelde, the founder of Smartling, an indus-try favourite, says that in future translationcustomers will choose how much humanintervention is needed for a translation. Aquick automated one will do for low-

The Economist January 7th 2017 9

TECHNOLOGY QUARTERLY Language

2

1

stakes content with a short life, but themost important content will still require afully hand-crafted and edited version. Not-ing that MT has both determined boostersand committed detractors, Mr Welde sayshe is neither: “If you take a dogmaticstance, you’re not optimised for the needsof the customer.”

Translation software will go on gettingbetter. Not only will engineers keep tweak-ing their statistical models and neural net-works, but users themselves will make im-provements to their own systems. Forexample, a small but much-admired startup, Lilt, uses phrase-based MT as the basis fora translation, but an easy-to-use interfaceallows the translator to correct and improve the MT system’s out-put. Every time this is done, the corrections are fed back into thetranslation engine, which learns and improves in real time. Userscan build several different memories—a medical one, a financialone and so on—which will help with future translations in thatspecialist field.

TAUS, an industrygroup, recently issued a reporton the state ofthe translation industry saying that “in the past few years thetranslation industry has burst with new tools, platforms and sol-utions.” Earlier this year Jaap van derMeer, TAUS’s founderand di-rector, wrote a provocative blogpostentitled “The Future DoesNotNeed Translators”, arguing that the quality of MT will keep im-proving, and that for many applications less-than-perfect transla-tion will be good enough.

The “translator” of the future is likely to be more like a quality-control expert, decidingwhich texts need the most attention to de-tail and editing the output of MT software. That may be necessarybecause computers, no matter how sophisticated they have be-come, cannot yet truly grasp what a text means. 7

Computertranslationis stillknown as“machinetranslation”

IN “BLACK MIRROR”, a British science-fiction satire seriesset in a dystopian near future, a young woman loses herboyfriend in a car accident. A friend offers to help her dealwith hergrief. The dead man was a keen social-media user,and his archived accounts can be used to recreate his per-sonality. Before long she is messaging with a facsimile,

then speaking to one. As the system learns to mimic him ever bet-ter, he becomes increasingly real.

This is not quite as bizarre as it sounds. Computers today canalready produce an eerie echo of human language if fed with theappropriate material. What theycannotyetdo ishave true conver-sations. Truly robust interaction between man and machinewould require a broad understanding of the world. In the absenceofthat, computers are not able to talkabout a wide range of topics,follow long conversations or handle surprises.

Machines trained to do a narrow range of tasks, though, canperform surprisingly well. The most obvious examples are thedigital assistants created by the technology giants. Users can askthem questions in a variety ofnatural ways: “What’s the tempera-ture in London?” “How’s the weather outside?” “Is it going to be

cold today?” The assistants know a few things about users, such aswhere they live and who their family are, so they can be personal,too: “How’s my commute looking?” “Text my wife I’ll be home in15 minutes.”

And they get better with time. Apple’s Siri receives 2bn re-quests per week, which (after being anonymised) are used for fur-ther teaching. For example, Apple says Siri knows every possibleway that users ask about a sports score. She also has a delightfulanswer for children who ask about Father Christmas. Microsoftlearned from some ofits previous natural-language platforms thatabout 10% of human interactions were “chitchat”, from “tell me ajoke” to “who’s your daddy?”, and used such chat to teach its digi-tal assistant, Cortana.

The writing team forCortana includes two playwrights, a poet,a screenwriter and a novelist. Google hired writers from Pixar, ananimated-film studio, and The Onion, a satirical newspaper, tomake its new Google Assistant funnier. No wonder people oftenthank their digital helpers for a job well done. The assistants’ re-plies range from “My pleasure, as always” to “You don’t need tothankme.”

Good at grammarHow do natural-language platforms know what people want?They not only recognise the words a person uses, but break downspeech for both grammar and meaning. Grammar parsing is rela-tively advanced; it is the domain of the well-established field of“natural-language processing”. But meaning comes under theheadingof“natural-language understanding”, which is farharder.

First, parsing. Most people are not very good at analysing thesyntax of sentences, but computers have become quite adept at it,even though most sentences are ambiguous in ways humans arerarely aware of. Take a sign on a public fountain that says, “This isnotdrinkingwater.” Humansunderstand it to mean that the water(“this”) is not a certain kind ofwater (“drinkingwater”). But a com-puter might just as easily parse it to say that “this” (the fountain) isnot at present doing something (“drinking water”).

As sentences get longer, the number ofgrammatically possiblebut nonsensical options multiplies exponentially. How can amachine parserknowwhich is the rightone? Ithelps for it to knowthat some combinations of words are more common than others:the phrase “drinking water” is widely used, so parsers trained onlarge volumes of English will rate those two words as likely to bejoined in a noun phrase. And some structures are more commonthan others: “noun verb noun noun” maybe much more commonthan “noun noun verb noun”. A machine parser can compute theoverall probability ofall combinations and pick the likeliest.

A “lexicalised” parser might do even better. Take the GrouchoMarx joke, “One morning I shot an elephant in my pyjamas. Howhe got in my pyjamas, I’ll neverknow.” The first sentence is ambig-uous (which makes the joke)—grammatically both “I” and “an ele-phant” can attach to the prepositional phrase “in my pyjamas”.But a lexicalised parser would recognise that “I [verb phrase] inmypyjamas” is farmore common than “elephant in mypyjamas”,and so assign that parse a higher probability.

But meaning is harder to pin down than syntax. “The boykicked the ball” and “The ball was kicked by the boy” have thesame meaning but a different structure. “Time flies like an arrow”can mean either that time flies in the way that an arrow flies, orthat insects called “time flies” are fond ofan arrow.

“Who plays Thor in ‘Thor’?” Your correspondent could not re-member the beefy Australian who played the eponymous Norsegod in the Marvel superhero film. But when he asked his iPhone,Siri came up with an unexpected reply: “I don’t see any moviesmatching ‘Thor’ playing in Thor, IA, US, today.” Thor, Iowa, with apopulation of 184, was thousands of miles away, and “Thor”, thefilm, has been out of cinemas for years. Siri parsed the questionperfectly properly, but the reply was absurd, violating the rules ofwhat linguists call pragmatics: the shared knowledge and under-

Meaning and machine intelligence

What are you talkingabout?Machines still cannot conduct proper conversations withhumans because they do not understand the world

10 The Economist January 7th 2017

TECHNOLOGY QUARTERLY Language

2

1

Terry Winograd

THE Turing Test was conceived as a wayto judge whether true artificial intelli-gence has been achieved. Ifa computercan fool humans into thinking it is hu-man, there is no reason, say its fans, tosay the machine is not truly intelligent.

Few giants in computing stand withTuring in fame, but one has given hisname to a similar challenge: Terry Wi-nograd, a computer scientist at Stanford.In his doctoral dissertation Mr Winogradposed a riddle for computers: “The citycouncilmen refused the demonstrators apermit because they feared violence.Who feared violence?”

It is a perfect illustration ofa well-recognised point: many things that areeasy for humans are crushingly difficultfor computers. Mr Winograd went intoAI research in the 1960s and 1970s anddeveloped an early natural-languageprogram called SHRDLU that could takecommands and answer questions abouta group ofshapes it could manipulate:“Find a blockwhich is taller than the oneyou are holding and put it into the box.”This workbrought a jolt ofoptimism tothe AI crowd, but Mr Winograd later fellout with them, devoting himselfnot tomaking machines intelligent but to mak-ing them better at helping human beings.(These camps are sharply divided byphilosophy and academic pride.) Hetaught Larry Page at Stanford, and after

Mr Page went on to co-found Google, MrWinograd became a guest researcher atthe company, helping to build Gmail.

In 2011Hector Levesque of the Univer-sity ofToronto became annoyed bysystems that “passed” the Turing Test byjoking and avoiding direct answers. Helater asked to borrow Mr Winograd’sname and the format ofhis dissertation’spuzzle to pose a more genuine test ofmachine “understanding”: the WinogradSchema. The answers to its battery of

questions were obvious to humans butwould require computers to have somereasoning ability and some knowledge ofthe real world. The first official WinogradSchema Challenge was held this year,with a $25,000 prize offered by Nuance,the language-software company, for aprogram that could answer more than90% of the questions correctly. The best ofthem got just 58% right.

Though officially retired, Mr Winogradcontinues writing and researching. Oneofhis students is working on an applica-tion for Google Glass, a computer with adisplay mounted on eyeglasses. The appwould help people with autism by read-ing the facial expressions ofconversationpartners and giving the wearer infor-mation about their emotional state. Itwould allow him to integrate linguisticand non-linguistic information in a waythat people with autism find difficult, asdo computers.

Asked to tricksome of the latest digitalassistants, like Siri and Alexa, he asks themthings like “Where can I find a nightclubmy Methodist uncle would like?”, whichrequires knowledge about both nightclubs(which such systems have) and Methodistuncles (which they don’t). When he tried“Where did I leave my glasses?”, one ofthem came up with a link to a bookof thatname. None offered the obvious answer:“How would I know?”

Brain scan

The Winograd Schema tests computers’ “understanding” of the real world

standing that people use to make sense of the often messy humanlanguage they hear. “Can you reach the salt?” is not a request forinformation but for salt. Natural-language systems have to bemanually programmed to handle such requests as humans expectthem, and not literally.

Multiple choiceShared information is also built up over the course of a conversa-tion, which is why digital assistants can struggle with twists andturns in conversations. Tell an assistant, “I’d like to go to an Italianrestaurant with my wife,” and it might suggest a restaurant. Butthen ask, “is it close to her office?”, and the assistant must grasp themeanings of“it” (the restaurant) and “her” (the wife), which it willfind surprisingly tricky. Nuance, the language-technology firm,which provides natural-language platforms to many other com-panies, is working on a “concierge” that can handle this type ofchallenge, but it is still a prototype.

Such a concierge must also offer only restaurants that are open.Linking requests to common sense (knowing that no one wants tobe sent to a closed restaurant), as well as a knowledge of the realworld (knowing which restaurants are closed), is one of the mostdifficult challenges for language technologies.

Common sense, an old observation goes, is uncommonenough in humans. Programming it into computers is harder still.

Fernando Pereira of Google points out why. Automated speechrecognition and machine translation have something in common:there are huge stores ofdata (recordings and transcripts for speechrecognition, parallel corpora for translation) that can be used totrain machines. But there are no training data for common sense.

Knowledge of the real world is another matter. AI has helpeddata-rich companies such as America’s West-Coast tech giants or-ganise much of the world’s information into interactive databasessuch as Google’s Knowledge Graph. Some of the content of thatappears in a box to the right ofa Google page ofsearch results for afamous figure or thing. It knows that Jacob Bernoulli studied at theUniversity of Basel (as did other people, linked to Bernoullithrough this node in the Graph) and wrote “On the Law of LargeNumbers” (which it knows is a book).

Organising information this way is not difficult for a companywith lots ofdata and good AI capabilities, but linking informationto language is hard. Google touts its assistant’s ability to answerquestions like “Who was president when the Rangers won theWorld Series?” But Mr Pereira concedes that this was the result ofexplicit training. Another such complex query—“What was thepopulation of London when Samuel Johnson wrote his dictio-nary?”—would flummox the assistant, even though the Graphknows about things like the historical population of London andthe date of Johnson’s dictionary. IBM’s Watson system, which in

The Economist January 7th 2017 11

2

1

2011 beat two human champions at thequiz show “Jeopardy!”, succeeded mainlyby calculating huge numbers of potentialanswers based on key words by probabili-ty, not by a human-like understanding ofthe question.

Making real-world information com-putable is challenging, but it has inspiredsome creative approaches. Cortical.io, a Vienna-based startup, took hundreds ofWikipedia articles, cut them into thou-sandsofsmall snippetsofinformation andran an “unsupervised” machine-learningalgorithm over it that required the comput-er not to lookfor anything in particular butto find patterns. These patterns were thenrepresented as a visual “semantic finger-print” on a grid of 128x128 pixels. Clumpsof pixels in similar places represented se-mantic similarity. This method can be usedto disambiguate words with multiplemeanings: the fingerprintof“organ” sharesfeatures with both “liver” and “piano” (be-cause the word occurs with both in different parts of the trainingdata). This might allow a natural-language system to distinguishbetween pianos and church organs on one hand, and livers andother internal organs on the other.

Proper conversation between humans and machines can beseen as a series of linked challenges: speech recognition, speechsynthesis, syntactic analysis, semantic analysis, pragmatic under-standing, dialogue, common sense and real-world knowledge. Be-cause all the technologies have to work together, the chain as awhole isonlyasstrongas itsweakest link, and the first fewoftheseare far better developed than the last few.

The hardest part is linking them together. Scientists do notknow how the human brain draws on so many different kinds ofknowledge at the same time. Programming a machine to replicatethat feat is very much a work in progress. 7

IN “WALL-E”, an animated children’s film set in the future,all humankind lives on a spaceship after the Earth’s envi-ronment has been trashed. The humans are whiskedaround in intelligent hovering chairs; machines take careof their every need, so they are all morbidly obese. Eventhe ship’s captain is not really in charge; the actual pilot is

an intelligent and malevolent talking robot, Auto, and like somany talking machines in science fiction, he eventually makes agrab for power.

Speech is quintessentially human, so it is hard to imagine ma-chines that can truly speak conversationally as humans do with-out also imagining them to be superintelligent. And if they aresuperintelligent, with none ofhumans’ flaws, it is hard to imaginethem not wanting to take over, not only for their good but for thatof humanity. Even in a fairly benevolent future like “WALL-E’s”,where the machines are doingall the work, it is easy to see that the

lackofanything challenging to do would be harmful to people.Fortunately, the tasks that talking machines can take off hu-

mans’ to-do lists are the sort that many would happily give up.Machines are increasingly able to handle difficult but well-de-fined jobs. Soon all that their users will have to do is pipe up andaskthem, usinga naturally phrased voice command. Once upon atime, just one tinkerer in a given family knew how to work thecomputer or the video recorder. Then graphical interfaces (iconsand a mouse) and touchscreens made such technology accessibleto everyone. FrankChen ofAndreessen Horowitz, a venture-capi-tal firm, sees natural-language interfaces between humans andmachines as just another step in making information and servicesavailable to all. Silicon Valley, he says, is enjoying a golden age ofAI technologies. Just as in the early 1990s companies were pilingonline and building websites without quite knowing why, noweveryone is going for natural language. Yet, he adds, “we’re in 1994for voice.”

1995 will soon come. This does not mean that people will com-municate with their computers exclusively by talking to them.Websites did not make the telephone obsolete, and mobile de-vices did not make desktop computers obsolete. In the same way,people will continue to have a choice between voice and textwhen interacting with their machines.

Not all will choose voice. For example, in Japan yammeringinto a phone isnotdone in public, whether the interlocutor isa hu-man or a digital assistant, so usage of Siri is low during businesshours but high in the evening and at the weekend. For others,voice-enabled technology is an obvious boon. It allows dyslexicpeople to write without typing, and the very elderly may find iteasier to talk than to type on a tiny keyboard. The very young,some of whom today learn to type before they can write, maysoon learn to talk to machines before they can type.

Those with injuries or disabilities that make it hard for them towrite will also benefit. Microsoft is justifiably proud of a new de-vice that will allow people with amyotrophic lateral sclerosis(ALS), which immobilises nearly all of the body but leaves themind working, to speak by using their eyes to pick letters on ascreen. The critical part is predictive text, which improves as it getsused to a particular individual. An experienced user will be ableto “speak” at around 15 words per minute.

People may even turn to machines for company. Microsoft’sXiaoice, a chatbot launched in China, learns to come up with theresponses that will keep a conversation going longest. Nobodywould think it was human, but it does make users open up in sur-prising ways. Jibo, a new “social robot”, is intended to tell childrenstories, help far-flung relatives stay in touch and the like.

Another group that may benefit from technology is smallerlanguage communities. Networked computers can encourage awinner-take-all effect: if there is a lot ofgood software and contentin English and Chinese, smaller languages become less valuableonline. If they are really tiny, their very survival may be at stake.But Ross Perlin of the Endangered Languages Alliance notes thatnew software allows researchers to document small languagesmore quickly than ever. With enough data comes the possibilityof developing resources—from speech recognition to interfaceswith software—for smallerand smaller languages. The Silicon Val-ley giants already localise their services in dozens of languages;neural networksand othersoftware allownewversions to be gen-erated faster and more efficiently than ever.

There are two big downsides to the rise in natural-languagetechnologies: the implications for privacy, and the disruption itwill bring to many jobs.

Increasingly, devices are always listening. Digital assistants likeAlexa, Cortana, Siri and Google Assistant are programmed to waitfor a prompt, such as “Hey, Siri” or “OK, Google”, to activate them.But allowing always-on microphones into people’s pockets andhomes amounts to a further erosion of traditional expectations ofprivacy. The same mightbe said forall the ways in which language

Looking ahead

For my next trick

Talking machines are the new must-haves

Whatmachinescannot yetdo is havetrue con-versations

TECHNOLOGY QUARTERLY Language

officer, likes to think that this will free chief financial officers fromhaving to write up the same old routine analyses for the board,giving them time to develop more creative approaches.

Carl Benedikt Frey, an economist at Oxford University, has re-searched the likely effect of artificial intelligence on the labourmarket and concluded that the jobs most likely to remain immuneinclude those requiring creativity and skill at complex social inter-actions. But not every human has those traits. Call centres mayneed fewerpeople as more routine workis handled by automatedsystems, but the trickier inquiries will still go to humans.

Much of this seems familiar. When Google search first becameavailable, it turned up documents in seconds that would have tak-en a human operator hours, days or years to find. This removedmuch of the drudgery from being a researcher, librarian or jour-nalist. More recently, young lawyers and paralegals have taken tousing e-discovery. These innovations have not destroyed the pro-fessions concerned but merely reshaped them.

Machines that relieve drudgery and allow people to do moreinteresting jobs are a fine thing. In net terms they may even createextra jobs. But any big adjustment is most painful for those leastable to adapt. Upheavals brought about by social changes—likethe emancipation of women or the globalisation of labour mar-kets—are already hard for somepeople to bear. When thosechanges are wrought by ma-chines, theybecome even harder,and all the more so when thosemachines seem to behave moreand more like humans. People al-ready treat inanimate objects asif they were alive: who has nevershouted at a computer in frustra-tion? The more that machinestalk, and the more that they seemto understand people, the moretheir users will be tempted to at-tribute human traits to them.

That raises questions aboutwhat it means to be human. Lan-guage is widely seen as human-kind’s most distinguishing trait.AI researchers insist that theirmachines do not think like peo-ple, but if they can listen and talklike humans, what does thatmake them? As humans teachever more capable machines touse language, the once-obviousline between them will blur. 7

12 The Economist January 7th 2017

TECHNOLOGY QUARTERLY Language

OFFER TO READERSReprints of Technology Quarterly are available from the Rights and Syndication Department. A minimum orderof five copies is required.

CORPORATE OFFERCustomisation options on corporate orders of 100 or moreare available. Please contact us to discuss your requirements.For more information on how toorder special reports, reprints or any queries you may have pleasecontact:

The Rights and Syndication DepartmentThe Economist20 Cabot SquareLondon E14 4QW

Tel +44 (0)20 7576 8148Fax +44 (0)20 7576 8492e-mail: rights@economist.comwww.economist.com/rights

software improvesby trainingon a single user’svoice, vocabulary,written documents and habits.

All the big companies’ location-based services—even the accel-erometers in phones that detect small movements—are makingever-improving guesses about users’ wants and needs. The mo-ment when a digital assistant surprises a userwith “The chemist isnearby—do you want to buy more haemorrhoid cream, Steve?”could be when many may choose to reassess the trade-off be-tween amazing new services and old-fashioned privacy. The techcompanies can help by giving users more choice; the latest iPhonewill not be activated when it is laid face down on a table. But hack-ers will inevitably find ways to get at some of these data.

Hey, Siri, find me a jobThe other big concern is for jobs. To the extent that they are rou-tine, theyface beingautomated away. Agood example is customersupport. When people contact a company for help, the initial en-counter is usually highly scripted. A company employee will veri-fy a customer’s identity and follow a decision-tree. Language tech-nology is now mature enough to take on many of these tasks.

For a long transition period humans will still be needed, butthe work they do will become less routine. Nuance, which sellslots of automated online and phone-based help systems, is bull-ish on voice biometrics (customers identifying themselves by say-ing “my voice is my password”). Using around 200 parameters foridentifyinga speaker, it is probably more secure than a fingerprint,says Brett Beranek, a senior manager at the company. It will alsoeliminate the tedium, for both customers and support workers, ofgoing through multi-step identification procedures with PINs,passwords and security questions. When Barclays, a British bank,offered it to frequent users of customer-support services, 84%signed up within five months.

Digital assistants on personal smartphones can get away withmistakes, but for some business applications the tolerance for er-ror is close to zero, notes Nikita Ivanov. His company, Datalingvo, aSilicon Valley startup, answers questions phrased in natural lan-guage about a company’s business data. If a user wants to knowwhich online ads resulted in the most sales in California lastmonth, the software automatically translates his typed questioninto a database query. But behind the scenes a human working forDatalingvo vets the query to make sure it is correct. This is becausethe stakes are high: the technology is bound to make mistakes inits early days, and users could make decisions based on bad data.

This process can work the other way round, too: rather thannatural-language input producing data, data can produce lan-guage. Arria, a company based in London, makes software intowhich a spreadsheet full of data can be dragged and dropped, tobe turned automatically into a written description ofthe contents,complete with trends. Matt Gould, the company’s chief strategy

2

36 The Economist January 7th 2017

For daily analysis and debate on Europe, visit

Economist.com/europe

AFTER a year of terrorist attacks and aviolent coup attempt, Istanbul resi-

dents are getting used to the sound of ex-plosions. When blasts rang out near thecity’s best-known nightclub just after 1amon January 1st, some thought they werenew-year pyrotechnics. Yet the skies abovethem were empty. A massacre was unfold-ingbelow. By the time it was overat least 39people, mostly foreigners, were dead, anddozens more wounded. Autopsies suggest-ed that many had been shot at close range.Some saved themselvesby leaping into theBosporus. As The Economist went to pressthe attacker, a suspected follower of Islam-ic State (IS), had not been caught.

IS has carried out at least eight big at-tacks in Turkey, including the deadliest inthe country’s history, a suicide-bombingthat killed more than 100 people in Octo-ber 2015. The nightclub attack is the first ithas undisputedly claimed. In an onlinestatement the group praised the shootingas an attack on an “apostate” celebrationand revenge for a Turkish offensive againstit in Syria. Turkey’s army cleared IS fromstrongholds overlooking the border in ear-ly September, and fighting continues nearal-Bab, a town north-east ofAleppo.

Under pressure in Syria, IS has struckback by destabilising Turkey. The group’searliest attacks in 2015 helped to reignite awar between Kurdish militants and Tur-key’s armed forces. A second wave scaredaway tourists and fanned resentment of

an alliance of Western powers. An MPfrom the governing party blamed—whoelse?—the CIA.

The shooting also raised questions ofaccountability. More than 400 lives havebeen lost in big terrorist attacks since thesummer of 2015, yet not one minister hasresigned. Just over a weekbefore the night-club attack, Russia’s ambassador was fatal-ly shot by an off-duty Turkish policeman.The government says it foiled 339 attackslast year. But it has also used the war on ter-rorism as an excuse to silence critics. In De-cember authorities detained a Wall StreetJournal reporter for three nights, allegedlyfor retweeting an image from an IS murdervideo. Days later they arrested an investi-gative reporter, Ahmet Sik, on farcical ter-rorism charges. Since the coup, more than100 journalists have been locked up.

Largely because of the state’s controlover religious debate, support for IS amongTurks is minimal. Yet the group is deter-mined to pit Turkey’s traditionally tolerantbrand of Islam against an emboldenedfundamentalist fringe. IS wants to galva-nise those Islamists who condemn secularways of life, says Rusen Cakir, a journalist.“They want to transform Turkey into a bat-tlefield,” he says.

The New Year’s Day attack could serveas a wake-up call. The ruling Justice andDevelopment party is realising that polar-isation can win elections “but that it makesthe country ungovernable,” says Ozgur Hi-sarcikli, head of the Ankara office of theGerman Marshall Fund, a think-tank. TheDiyanet has declared that an attack on anightclub is as reprehensible as an attackon a mosque. Mr Erdogan himself haswarned against allowing the fault lines inTurkish society to widen, which is exactlywhat IS wants. Alas, Mr Erdogan’s populistauthoritarianism, jingoism and repressionare only wedging them further apart. 7

the 2.8m Syrian refugees living in Turkey.The latest, which hit a venue where celeb-rities dance and drinkalongside foreignersand the monied elite, threatens to inflametensions between Islamists and secularTurks, many of whom blame the pro-Is-lamist government for the spread of ex-tremism. “Islamic State reads Turkish soci-ety very well and it knows to strike at thekey pressure points,” says Hilmi Demir, anexpert on Muslim sects and radicalisation.

Those pressure points are multiplying.Instead of healing his divided countryafter the coup in July, Turkey’s president,Recep Tayyip Erdogan, cracked down onhis opponents, including Kurdish activists,leftists and secularists. Official discourse isincreasingly conservative. In Decemberthe country’s religious-affairs directorate,the Diyanet, joined Islamist groups in pro-claiming that new-year festivities were“alien” to Turkish values. Meanwhile, agroup of young ultranationalists staged aprotest at which they pretended to holdSanta Claus—that unwelcome Western in-truder—at gunpoint.

Losing the plotMany Turkish conservatives refuse to ad-mit that innocents, including Muslims, arebeing murdered by a group acting in thename of Islam. They prefer conspiracy the-ories. A pro-government newspaperclaimed the attack on New Year’s Day wasthe workofa “mastermind”, shorthand for

Terror in Turkey

From celebration to carnage

ISTANBUL

An IslamicState attackon a nightclub widens the secular-religious divide

EuropeAlso in this section

37 Putin outmanoeuvres Obama

37 Bavaria’s angry drivers

38 Spain and Catalonia

39 Charlemagne: How Martin Luthernailed it

The Economist January 7th 2017 Europe 37

Motorway charges

Another European crisis

EVERYwinter, northern Europeansbound for ski holidays zip insouciant-

ly through the Netherlands, Belgium andGermany on motorways that are free ofcharge. But near the borders ofAustria orSwitzerland they must pull over to buystickers so that they can drive on theAlpine motorways—even as Austrian andSwiss cars zoom in the opposite directiononto Germany’s free Autobahn. To theperceived injustices in the EuropeanUnion (EU), add another: the nuisancesofa quilt of road-tolls.

Bavarians are particularly cranky. Ifyou live in Munich, say, workand playextend naturally across the border. Hencethe grousing about paying on Austrianroads while Austrians “free-ride” onBavarian ones. In 2013 the CSU, a regionalparty that governs Bavaria, made fixing“this unfair situation” a condition forjoining the coalition of the chancellor,Angela Merkel. The CSU’s AlexanderDobrindt, who became transport min-ister, got to work.

He knew that his biggest hurdlewould be the EU, which forbids dis-criminating against the citizens of othermember states. So he came up with twonominally separate laws. In one, every-one, German or foreign, would becharged a new road toll, like Austria’s. Inthe other, Germans would get a cut intheir vehicle tax that miraculously equalsthe price of the new toll. In effect, onlyforeigners would have paid more to usethe Autobahn. That was a bit too cheekyfor the European Commission, which in2015 flashed a stop sign in front of MrDobrindt. But now he has strucka com-promise with Brussels. The proposed toll

will be cheaper, and the tax relief forGermans better disguised.

That still has other Europeans fumingabout Bavarian harassment aimed atthem. Austria and the Netherlands, pos-sibly joined by Denmarkand Belgium,may sue Germany before the EuropeanCourt of Justice. A Bavarian pet peeve hasthus escalated to crisis diplomacy. “We inAustria are very unhappy about this,”Christian Kern, the Austrian chancellor,said in December. “This is a stress test forgood German-Austrian relations.” MrDobrindt retorted: “I have little sympathyfor this toll-whingeing, especially when itcomes from Austria.” It seems that the EUwill always find new ways to puncture itsown tyres.

KIEFERSFELDEN

One continent, divided by lots of toll booths

IT WAS Vladimir Putin as we have cometo know him: unpredictable, cynical and

skilful at trumping real events with propa-ganda. On December 29th Barack Obamaexpelled 35 Russian diplomats involved inintelligence work (along with their fam-ilies), ordered two Russian diplomaticcompounds in America closed, and im-posed fresh sanctions against Russian se-curity agencies and a list of individuals. MrObama was retaliating forRussian interfer-ence in America’s elections, which includ-ed hacking the computers of high-levelDemocratic Party officials and leaking theembarrassing contents to the press. TheWhite House expected Russia to eject anequal number of American diplomats. In-stead, MrPutin responded asymmetrically,parrying the American action and mock-ing Mr Obama as a bitter loser.

Mr Putin’s performance was carefullychoreographed. In the first move SergeiLavrov, the foreign minister, appeared onstate television to declare that Russiawould respond to America’s actions inkind. The foreign ministry and other agen-cies, he said, had proposed to Mr Putin that31 diplomats from the American embassyand four diplomats from the consulate inSt Petersburg be declared personae nongratae. MrLavrovalso recommended shut-ting down the American embassy’s dachain a wooded Moscow park.

In the second move, Mr Putin publiclyoverruled his foreign minister. WhileAmerica’s actions were “unfriendly” and“provocative”, and merited the toughestresponse, Mr Putin said, he would not “re-sort to irresponsible ‘kitchen’ diplomacy”but would instead plan steps for improvedrelations with the incoming president, Do-nald Trump. Moreover, he would not pun-ish the children ofAmerican diplomats forthe tensions between the two countries.Instead, he invited them to a Christmasand New Year’s Day show at the Kremlin.He then wished Mr Obama and Mr Trumpa happy new year.

Russian state television, which hasbeen pumping out anti-American propa-ganda for years, quickly seized on a newnarrative. “The provocation hasfailed,” an-nounced a news anchor on Channel One.Mr Obama had found himself “in a pud-dle”, while Mr Putin had displayed dip-lomatic “mastery at the world level”. Theinvocation of Christmas and familyseemed a backhanded jab at Mr Obama’spacific reputation and his public displays

of affection for children. The media stuntearned an unctuous tweet from Mr Trump:“Great move on delay (by V. Putin)—I al-ways knew he was very smart!”

Mr Trump’s tweet added to the mysteryof his apparent infatuation with Mr Putinand fuelled anxiety about Russia’s abilityto undermine American democracy. Yetfor pro-Western Russian liberals, the pan-icked attitude of some of America’s main-stream media was equally discomfiting. Itseemed a mirror image of Russia’s ownhysteria about the role of America in sow-ing chaos and staging colour revolutions inRussia’s backyard.

“In the eyes of the West, Russia appearsto be the source ofmost uncomfortable so-cial changes,” wrote Maxim Trudolyubovin a column in Vedomosti, an independent

daily. “As a Russian, it is amusing to watchthis. The West now identifies all its pro-blems with ‘Russia’, just as Russia identi-fies all its problems with ‘the West’.”

Americans’ treatment of Russia as a bo-geyman fills Mr Putin’s supporters withpride. They see it at a sign of Russia’s re-newed great-power status. But while theKremlin may be benefiting from fears of itsinfluence in the short term, it is unclearhow it plans to turn those fears to its lon-ger-term advantage. Mr Putin has long de-pended on fear of America as a mighty en-emy to reinforce his hold on power.Kicking the outgoing Mr Obama may be apoor substitute. Paradoxically, Mr Trump’sdismissal of Russian influence could bemore harmful to the Kremlin’s narrativethan fears of its interference. 7

America’s new Russia sanctions

Feint praise

MOSCOW

As Putin outmanoeuvres Obama,Trump applauds

38 Europe The Economist January 7th 2017

IFYOUlookup from the bustle ofthe win-ter tourists thronging the streets of Barce-

lona, you will see some balconies drapedwith the estelada, a blend of the Catalanand Cuban flags that has become the ban-ner of those who want their land to be-come independent. There are fewer thanthere once were, but still enough to inspirethe Catalan regional government’s pledgeto hold a binding referendum on indepen-dence in 2017. Since the Spanish govern-ment refuses to contemplate such a vote, aconfrontation seems inevitable.

Indeed, it has already begun. Some 300Catalan officials face court cases for flout-ing the law, in acts ranging from a previousunilateral effort in 2014 to organise an inde-pendence vote to petty protests, such asflying the estelada from town halls. CarlesPuigdemont, the president of the General-itat (the Catalan government), promises topush through “laws of disconnection” thissummer, such as one setting up its own taxagency, prior to holding a referendum,probably in September. His pro-indepen-dence coalition has a majority in the Cata-lan parliament. On December14th Spain’sConstitutional Tribunal warned the Gen-eralitat that the referendum would be ille-gal. Spain could face unprecedented de-fiance of its democratic constitution.

How has it come to this? Spain’s consti-tution of 1978 gave Catalonia, one of thecountry’s most prosperous regions, moreself-government than almost any otherpart of Europe. The Generalitat controlsnot just schools and hospitals but policeand prisons. It has made Catalan the mainlanguage of teaching. Under Jordi Pujol,the skilful moderate nationalist cacique(political boss) who headed the Generali-tat from 1980 to 2003, Catalonia was con-tent with this settlement, using its votes inthe Madrid parliament to extract incre-ments to it powers and revenues.

Two things upset matters. The first waswhen the Constitutional Tribunal in 2010watered down a new autonomy statute,which recognised Catalonia’s sense of na-tionhood and granted additional legalpowers to the Generalitat. It had been ap-proved by referendum in Catalonia and bythe Spanish parliament. The second factorwas the economic crisis after the burstingofSpain’s property bubble in 2008.

In 2012 demonstrators against austeritybegan to put the blame on Madrid, ratherthan Artur Mas, Mr Pujol’s heir. Supportfor independence surged from less than

25% to more than 45%. “Society moved to-wards more radical positions,” thinks JoanCulla, a historian. Others see this as at leastin part induced by the Generalitat, with itsmoney and powerful communicationsmachine. It allowed the nationalists tokeep power, despite budget cuts and reve-lations that for decades they had takenrake-offs on public contracts.

Catalan society remains split. “Therearen’t the numbers to advance [to indepen-dence] but there’s enough to make a lot ofnoise,” says Jordi Alberich of the Cercled’Economia, a business group.

Best ofenemiesThis stand-off has been politically profit-able not just for the Catalan nationalistsbut also for Mariano Rajoy, Spain’s primeminister, and his conservative People’sParty. His staunch defence of his country’sterritorial unity is popular in most placesoutside Catalonia. For years Mr Rajoy didnothing to respond to Catalan grievances,some of which are justified. Cataloniapays more into the central kitty than it getsback, but its transport systems have beenneglected while Madrid has spiffy metrolines and a surfeit ofmotorways.

Yet weariness with the deadlock hastaken hold, in both Barcelona and Madrid.Last month Mr Rajoy put his deputy, So-raya Sáenz de Santamaría, in charge of theCatalan question. She is putting feelers outto the Generalitat. Mr Puigdemont haspublished a list of 46 points to negotiate. It

starts with the “binding referendum”.It is not hard to divine the contours of a

deal. Mr Rajoy could offer concessions onfinancing and infrastructure. More contro-versially, he could propose recognising theCatalan language or that Catalonia is a na-tion within Spain.

The toughest issue is the referendum.This is no moment to contemplate any sortof plebiscite with equanimity. Catalan na-tionalists claim to be exemplary pro-Euro-peans. But there are many echoes of Brexitin Catalonia. Instead of Brussels, it is Ma-drid the nationalists accuse ofstealing Cat-alans’ money. They argue that indepen-dence would be quick and easy. “The greatgrowth in support for independence from2012 was the first manifestation of popu-lism in Spain,” says Javier Cercas, a writerwho lives in Barcelona.

MrPuigdemont insists thatblocking thereferendum “would be bad news for de-mocracy”. He is prepared to negotiate itstiming. But he adds: “We won’t easily re-nounce it. I think we’ve earned the right tobe heard.” Some in Barcelona believe theGeneralitat’s leaders are searching for adignified way to back down. Mr Puigde-mont talks also of “constituent” electionsto found a new state. But his party, cloudedby corruption, may suffer. The Catalan var-iant of Spain’s left-wing Podemos, whichalready runs Barcelona’s city governmentand which is forminga new, broader, party,is likely to gain ground. It wants Cataloniato form part of a “plurinational” Spain, acleverly vague formula.

“Is being part of Spain a problem in thedaily life of Catalans?” asks Inés Arrima-das of Ciudadanos, an anti-nationalistparty that leads the opposition in the Cata-lan parliament. “For us the problems ofCatalonia are unemployment, povertyand corruption.” The longer the deadlocklasts, the harderMrPuigdemontmayfind itto persuade Catalans otherwise. 7

Spain and Catalonia

Catalexit?

BARCELONA

A search fora face-saving deal to avoid anotherunsettling referendum

Let my pueblo go

The Economist January 7th 2017 Europe 39

SET foot in Germany this year and you are likely to encounterthe jowly, dourportraitofMartin Luther. With more than 1,000

events in 100 locations, the whole nation is celebrating the 500thanniversary of the monkissuing his 95 theses and (perhaps apoc-ryphally) pinning them to the church door at Wittenberg. He setin motion a split in Christianity that would forever change notjust Germany, but the world.

At home, Luther’s significance is no longer primarily theologi-cal. After generations of secularisation, not to mention decadesof official atheism in the formerly communist east (which in-cludes Wittenberg), Germans are not particularly religious. Butthe Reformation was not just about God. It shaped the Germanlanguage, mentality and way of life. For centuries the countrywas riven by bloody confessional strife; today Protestants andCatholics are each about 30% of the population. But after Ger-man unification in the 19th century, Lutheranism won the culturewars. “Much of what used to be typically Protestant we todayperceive as typically German,” says Christine Eichel, author of“Deutschland, Lutherland”, a bookabout Luther’s influence.

Start with aesthetics. For Luther this was, like everything else,a serious matter. He believed that Christians were guaranteedsalvation through Jesus but had a duty to live in such a way as todeserve it. Ostentation was thusa disgraceful distraction from theasceticism required to examine one’s own conscience. The tracesof this severity live on in Germany’s early 20th-century Bauhausarchitecture, and even in the furniture stylesat IKEA (from Luther-an Sweden). They can be seen in the modest dress, office decorand eating habits of Angela Merkel, the daughter of a Lutheranpastor, and of Joachim Gauck, Germany’s president and a formerpastor himself. Both may partake of the glitz of the French presi-dency while visiting Paris, but it would never pass in Berlin.

Luthershared hisdistaste forvisual ornamentwith other Prot-estant reformers. But he differed in the role he saw for music. TheSwiss Protestants John Calvin and Huldrych Zwingli viewed mu-sic as sensual temptation and frowned on it. But to Luther musicwas a divinely inspired weapon against the devil. He wanted be-lievers to sing together—in German, in church and at home, andwith instruments accompanying them. Today Germany has 130publicly financed orchestras, more than any other country. And

concerts are still attended like sermons, sombrely and seriously.Luther’s inheritance can also be seen in the fact that Germany,

the world’s 17th-most populous country, has the second-largestbook market after America’s. After he translated the Bible intoGerman, Luther wanted everyone, male or female, rich or poor,to read it. At first Protestants became more literate than Catholics;ultimately all Germans became bookish.

Finally, a familiar thesis links Luther to German attitudes to-wards money. In this view Catholics, used to confessing and be-ing absolved after each round ofsins, tend to run up debts (Schul-den, from the same root as Schuld, or “guilt”), whereas Protestantssee saving as a moral imperative. This argument, valid or not, hasa familiar ring in southern Europe’s mainly Catholic and Ortho-dox countries, which have spent the euro crisis enduring lectureson austerity from Wolfgang Schäuble, Germany’s devoutly Lu-theran finance minister.

Yet on money, too, Luther differed from other reformers.When Max Weber wrote of the Protestant work ethic in 1904, hehad in mind Calvinism and its relatives, such as American Puri-tanism. Calvin viewed an individual’s ability to get rich as a signthat God had predestined him to be saved. To Luther, Christianswere already saved, so wealth was suspect. Instead of amassingit, Christians should work for their community, not themselves.Work (Beruf) thus became a calling (Berufung). Not profit but re-distribution was the goal. According to Gerhard Wegner, a profes-sor of theology, this “Lutheran socialism” finds secular expres-sion in the welfare states ofScandinavia and Germany.

Luther’s “subcutaneous” legacy keeps popping up in surpris-ing places, says Mrs Eichel. Germans, and especially Lutherans,buy more life insurance but fewer shares than others (Lutherdidn’t believe in making money without working for it). Andeverywhere they insist on conscientious observance ofprincipleand order. Theyreligiouslyseparate their rubbish by the colourofglass and are world champions at recycling (65% ofall waste), eas-ily beating the second-place South Koreans.

Holier than thouLuther also shares blame for some negative qualities ascribed toGermans. He was deeply anti-Semitic, a prejudice his country-men have shed at great cost (he blamed evil stares from Jews forthe illness that eventually killed him). Germans’ legendary obe-dience to authority is attributed to Luther’s insistence on separat-ing spiritual and worldly authorities (which princes in his dayfound useful in suppressing a peasants’ revolt). And althoughpersonally fond of boisterous jokes, he was among the foundingfigures ofGermany’s rather humourless and preachy tradition ofpublic discourse. Germans today are the first to bemoan their na-tional habit ofdelivering finger-wagging lectures.

Such rigid moralism can make Germans hard to deal with, es-pecially in Brussels, where the EU’s problems demand a willing-ness to let misdemeanours slide. But there are worse traits thanexcessive morality. Besides, 500 years on, Lutheran Germany isbeing transformed by globalisation. Germany today has not onlydevout ascetics but everything from consumerist hipsters to Om-chanting yogis. A growing Muslim population is pushing thecountry towards a new kind of religious pluralism. Mrs Eichelherself finds German churches “too serious”; she attends oneheaded by an African-American gospel preacher. If the downsideof Germans’ Lutheran heritage is a difficulty in lightening up oraccepting alternative lifestyles, they seem to be getting over it. 7

Nailed it

HowLuther’s ideas have shaped Germany forhalfa millennium

Charlemagne

40 The Economist January 7th 2017

For daily analysis and debate on Britain, visit

Economist.com/britain

1

MARY ANN COTTON was one of thegreat Victorian poisoners. She proba-

bly killed three ofher four husbands, a lov-er, her mother and 11 of her 13 children. Ar-senic was her weapon of choice. As thedeaths mounted, the authorities becamesuspicious. Only when they tested theempty medicine bottles that littered herhouse and found evidence of arsenic wasCotton caught. She was tried and hangedin 1873. Today murder by poison is rare—dangerous substances are more tightlycontrolled and the accuracy of autopsiesmakes the crime harder to pull off. In theyear to March 2015 only 11 people werekilled in this way.

Poisoning is not the only offence almostto have disappeared. Since the mid-1990sBritain has seen a steady and dramatic de-cline in lawbreaking: the numberofcrimeshasmore than halved, according to the offi-cial Crime Survey for England and Wales.Vehicle theft has fallen by 86% and bur-glary by 71% since 1995. Violent crime hasdropped by two-thirds and robberies bymore than half. Even with the onset of thefinancial crisis in 2007 and the ensuingcuts to welfare and public services, includ-ing the police, Britain has grown ever safer.

The explanations range from the fallingvalue of items once stolen, such as televi-sions, to clever policing and improved se-curity. Ram-raiding, a once-common crimein which criminals crash cars or vansthrough the front of shops or banks in or-

the switch it dropped to 4%.Yet the fall in crime seems to have

slowed. The overall number of offencesdropped by just 1% in the year to June 2016,according to the Crime Survey. Comparethat with the decline of13% over the previ-ous two years. And a few crimes are risingagain. Car theft edged up by 1%. The kindsof violent crimes that do not cause physi-cal injury (such as stalking, harassmentand death threats) climbed by 18%. Pick-pocketing, which had been going up evenwhile instances of other offences fell, con-tinued to rise. Have Britain’s crime ratesreached their nadir?

The recent levellingoffmay be the long-expected impact of the financial crisis, sug-gests Tim Newburn, a criminologist at theLondon School of Economics. In times ofhardship people steal more. Attempts tobreakinto homes have risen by 5% over thepast year. Domestic violence goes up withthe anxiety of poverty and appears to beincreasing. The Crime Survey stops count-ing offences after five incidents involvingthe same victim. But remove that cap andviolence against women has been risingsince 2008. That suggests that even if thenumber of women being abused at homehas not risen, victims are being attackedmore often.

Still, further declines in crime are possi-ble. Pinchingcars is one example ofa “gate-way crime”—the first rung on the ladder tomore serious lawbreaking. The drastic re-duction in car theft in the past couple ofde-cades has thus meant fewer entrants intothe pool of criminals. Young people makeup a shrinking share of the prison popula-tion. In June 2011, men aged 18-24 account-ed for 26% of those locked up. They nowrepresent just17%. And whereas youths aregrowing less likely to reoffend, among old-er cons recidivism is on the increase.

Older, experienced crooks lie behind

der to loot them, is largely a thing of thepast, says Nick Tilley, a criminologist atUniversity College London, because of in-novations such as the introduction of bol-lards in front ofsuch premises and securityshutters to protect shopfronts.

Other criminal enterprises have be-come less rewarding. Phone theft in-creased as smartphones took off, says Gra-ham Farrell, a criminologist at theUniversity of Leeds, but it has since ebbedas owners have gained the power to trackand disable their stolen devices. The pro-portion of owners reporting their phonenicked fell by half between 2009 and 2016.Nor is it any longerworth robbingbus driv-ers, because card payments and cash-dropboxes mean they no longer carry muchmoney. Data from Transport for Londonshow a fall of 56% in bus robberies be-tween 2013 and 2015, which coincides witha reduction in the number of cash pay-ments for fares.

In some cases, the harm has been re-duced even as the crime has persisted. En-suring that pubs and clubs give drinkersventuring outside receptacles made ofplastic or toughened glass, which breaksinto blunt little cubes rather than jaggedshards, has cut the number of severe inju-ries, particularly to the face, incurred bydrunken brawlers. Before the drinks indus-try switched to toughened glass in 1997, 13%of violence between strangers involvedthe use of glasses or bottles. The year after

Crime

How low can it go?

Adecades-long fall in offending seems to be tapering off. Yet crime rates could dropeven further in future

BritainAlso in this section

41 Brexit preparations

41 International development

42 Bagehot: Pierogi and integration

The Economist January 7th 2017 Britain 41

2 the recent rises in certain crimes. Althoughcar theft in general has been falling, sophis-ticated thefts of expensive cars by skilledcriminals have increased. Swiping poshvehicles for resale and export is more diffi-cult than nicking them offthe street for joy-riding. Thieves are pinching car keys ratherthan simply breaking into vehicles, or un-locking them remotely by hacking intotheir security systems.

That such professionals are responsiblehints at why crime rates may have furtherto fall. Studies in America, where crime hasalso been declining for a long time, suggestthat men over 40 today offend at a muchhigher rate than men of that age did a cou-ple of decades ago. Today’s middle-agedcrooks learned their trade in the 1980swhen crime was relatively easy, and havecarried on offending, says Mr Farrell. Intime this light-fingered generation will “re-tire”, or die. With fewer novices takingtheir place, crime may dip lower still. 7

CIVIL servants mostly operate behindthe scenes and off camera. Sir Ivan

Rogers, Britain’s permanent representativeto the European Union since November2013, fits the bill nicely. Yet his announce-ment on January 3rd that he was leavingearly became big news because of what itshowed about the government’s unreadi-ness for Brexit negotiations.

Sir Ivan, a former Treasury and EU offi-cial, knows everything there is to knowabout how Brussels works. He was DavidCameron’s chief adviser on Europe. He gotto know Theresa May when, as home sec-retary, she engaged in tortuous talks overBritain’sopt-out from EU policieson justiceand home affairs. But he has long beenfaulted by Eurosceptics as too gloomy overBrexit. In December he was pilloried for re-porting that it could take ten years to nego-tiate and ratify a trade deal with the EU. Yetas Sir Ivan put it this week, free trade “doesnot just happen when it is not thwarted byauthorities.” Experience shows that suchdeals can indeed take years to agree.

His other purported sin came duringMr Cameron’s attempted renegotiation ofBritain’s membership terms before the ref-erendum. At one point the prime ministerwanted to demand a unilateral emergencybrake on free movement of EU citizens toBritain. Sir Ivan advised him that other EUleaders, including Germany’s Angela Mer-kel, would reject this out of hand. So Mr

Cameron settled instead for a four-yearfreeze on in-work benefits for EU migrants.Brexiteers claim that, without Sir Ivan’s ex-cessive caution, Britain could have got a lotmore. Yet the EU’s attachment to freemovement is genuine and deep—even thebenefits change that Mr Cameron wontook48 hours ofhard pounding to secure.

Uncertainty clouds Brexit, even afterthe speedy replacement of Sir Ivan by SirTim Barrow, previously ambassador toRussia. Sir Ivan’s letter makes clear that thegovernment has no detailed exit strategyand that its negotiating team is not evenfully in place. Mrs May insists she will trig-ger Article 50, the legal way to leave the EU,by the end of March, earlier than Sir Ivanadvised. That will set a two-year deadlinefor Brexit. A chunk of 2017 will be taken upby Dutch, French, German and probablyItalian elections. Sir Ivan pointedly notesthat serious multilateral negotiating expe-rience is in short supply in Whitehall. Inthe EU institutions in Brussels, it is not.

His resignation supports the idea thatMrs May and her ministers mistrust advis-ers tainted by time in Europe. Lord Mac-pherson, a former permanent secretary tothe Treasury, tweeted that, with other de-partures, it was a “wilful & total destruc-tion of EU expertise”. Anyone with experi-ence of Brussels knows it is a place inwhich knowledge ofEU customs, laws andprocedures is valuable, especially aftermidnight. Mrs May could be repeatedlyambushed if she is bereft ofadvisers ready,as Sir Ivan puts it, to “challenge ill-foundedarguments and muddled thinking.”

OtherEU countrieshave longbeen frus-trated by the British, who favour a transac-tional approach to the project over dreamsof ever closer union. But they have alsocome to admire the talent and dedicationof British diplomats and officials. It wouldbe gravely damaging to Mrs May if shewere to lose those advantages at a timewhen they are needed more than ever. 7

Brexit preparations

Rogers and out

The departure ofBritain’s man inBrussels lays bare a lackofBrexit plans

Ivan to break free

International development

A stingy new year

IN 2015 Britain gave away £12.1bn($18.5bn) in foreign aid, more than any

country bar America. It was one of justsix countries to meet the UN’s target ofspending 0.7% ofGDP on internationalassistance. Yet although the leaders ofallBritain’s main political parties supportthis generosity, grumbles that the moneyshould stay at home are growing louder.

For the past few months newspapershave been digging up examples ofexor-bitant aid-industry salaries and allegedmis-spending. According to the DailyMail, £5.2m ofBritish cash went to anEthiopian pop group (defenders pointout that the band was part ofa project tochange attitudes about women’s roles).Some backbench Conservatives havecalled for aid to be redirected to pay forsocial care for elderly Britons. The UKIndependence Party wants to spend it onhomeless veterans instead.

The appointment ofPriti Patel ashead of the Department for Internation-al Development in July raised hawks’hopes, since she had previously calledfor the department to be abolished. Sofar, though, Ms Patel has done little tolive up to her ferocious reputation be-sides talking tough to internationalagencies. Few aid-watchers expect bigchanges to government policy. The 0.7%target was enshrined in law by the Con-servative-Liberal Democrat coalition in2015; maintaining it is a Tory manifestocommitment.

Some charity bosses whisper that thefuss is not all bad. A few executives areindeed paid too much, and consul-tancies sometimes overcharge, they say.Pointing this out is worthwhile, especial-ly since the industry can be too defen-sive about instances ofgenuine incom-petence. Yet much of the criticism ishollow. Some newspapers complainedabout bureaucracy, then whinged aboutprogrammes in which cash handoutswere given directly to the poor.

The government could tweakaidpolicy to use more of the funds to pro-mote British commercial interests, orhand out more money bilaterally ratherthan through intermediaries. Neitherwould make aid spending more effec-tive, says Owen Barder of the Centre forGlobal Development, a think-tank. Norwould they insulate the governmentfrom criticism. With public servicesfacing cuts, the generous aid budget willcome under continued fire however wellit is spent.

The juicy aid budget sparks jealousy

42 Britain The Economist January 7th 2017

THE new year finds Britain tense. Brexit looms: Theresa Maywill soon start the two-year countdown. There is still the tan-

gle of divisions that contributed to last June’s referendum vote.The gaps between liberal, Remain-voting places and conserva-tive, Leave-voting ones will widen as the trade-offs of Brexit be-come clear. Cultural grievances concerning rapid inflows of for-eigners, and of existing immigrants perceived to have integratedpoorly, remain unresolved. Racist slogans have appeared onwalls and hate crimes are up. Still Mrs May refuses to guaranteeEuropean Union citizens the right to stay put. “A disunited King-dom”, bellow headlines.

How to reunite it? Debates rage over Louise Casey’s reviewinto integration, published last month. Commissioned by DavidCameron, the civil servant’s report paints a grim picture of a landcleft by segregation, where citizens live parallel lives. It recom-mends that schools teach British values and that immigrants takean oath of loyalty (an idea endorsed by Sajid Javid, the communi-ties secretary). Much of the opposition agrees, Stephen Kinnockurging his fellow Labour MPs to “move away from multicultural-ism and towards assimilation”. On January 5th a new Parliamen-tary group on integration advocated a middle way between thetwo approaches. Expect more in this vein as the year unfolds.

Much of it will be warranted. Segregation scars parts of Brit-ain, some immigrant groups remain poorly integrated and mi-norities within them are hostile to liberal values. But the gloomlacks a sense of the bigger picture. Accompany Bagehot to Ely, acathedral city sprouting from the prairies of eastern England,where thousands of central and eastern Europeans have movedto pick vegetables for low wages—and where many have settled.Your columnist visited on December 23rd to witness the segrega-tion. What better litmus test than Christmas, with its many na-tional variations?

Sure enough, the local Poles’ traditions were widely evident.In a shop named “White & Red”, green-grey carp glistened in iceboxes; shelves groaned with pierogi (dumplings) and bottles ofbison-grass vodka; piles of sachets variously containing hay andcommunion wafers teetered by the till. Gosia Bates and JoannaBialas, two locals, explained that each ingredient features in theWigilia, or Christmas Eve, meal. This involves hay scattered on

the floor (to evoke the nativity), wafers broken before the mealand 12 dishes including carp, herring, pierogi, mushrooms, beet-root soup and poppy-seed cake. No meat or alcohol is taken, sothe vodka comes out at midnight. The steady traffic of local Polesin the shop spoke to the strength of this foreign culture. “Everyyearmy uncle sends me this from Silesia, forgood financial luck,”said Mrs Bates, producing a shiny carp scale from her purse.

Below the surface, however, something else is happening.Britons also shop at “White & Red”, lured by the garlicky sausageand crusty bread. And Poles are picking up British habits like eat-ing turkey and watching the queen’s Christmas speech. Thosewho, like Mrs Bates, have British partners are leading the fusion:her Anglo-Polish son receives British chocolate from the Polish StNicholas on December 6th; her Christmas tree is decorated to-wards the startofthe month (the British way) butwill stayup wellinto January (the Polish way); her son receives halfof his presentson December 24th (Polish) and half on the 25th (British). Sheserves turkey on Christmas Day, as is typical in Britain, but alsoleaves a chair empty—a Polish tradition respecting strangers.When relatives visit she cooks an English breakfast, which theylove (apart from the baked beans). She enjoyscrackersbut feels “abit silly with a paper hat on my head”. Ms Bialas plans to create asimilar mix ofcultures for her baby, due in 2017.

Without oaths, integration classes or other forms of state do-goodery, central European cultures in Britain are melding with lo-cal ones. Children are leading the way. Right after the Brexit voteteachers in Ely had to sooth not just upset Polish pupils but alsoBritish ones who fretted about losing their pals. Ms Bialas de-scribes school pick-up time, when Polish and British parents tendto stick to their own, but children pour out in a multinationalmuddle. Ask the Polish ones which football teams they support,she says, and they often name two: one Polish and one British.Some have become so British that they now struggle in their na-tive tongue, gettingA* grades in maths but Ds in Polish written ex-ams. This even extends to the liturgy. Mariusz Urbanowski, a lo-cal Polish priest, says he mixes the two languages in his festiveservices, to cater for different generations ofAnglo-Poles.

Szczesliwego New Year!Such is the Britain forgotten by the gloomsters. Fully82% ofits citi-zens socialise at least monthly with people from different ethnicor religious backgrounds; from 2003 to 2016 the proportion call-ing their vicinity “cohesive” rose from 80% to 89%; over half offirst-generation migrants have friends of a different ethnicity(among their kids the proportion is nearer three-quarters); num-bers of inter-ethnic marriages and households are rising; educa-tional and employment gaps are shrinking. The proportion ofBritish-Pakistani households using English as their main lan-guage rose from 15% to 45% in the 13 years to 2010.

The story of British life in 2017 is that new immigrants are en-riching and combining with this mongrel culture as loyally astheir predecessors once did. In pubs and churches, gyms andschools, Britishness is being made and remade not by politicaldiktat but by an organic process of mixing and mingling. Britaincontains few French-style banlieues or American-style ghettos.London’s mayor is a liberal Muslim. Sikhs in turbans protectBuckingham Palace. Tikka masala (Indian-ish) and fish and chips(Jewish-ish) are the country’s national dishes. Polish-ish pierogiwill surely join them soon. In a troubled age, let this diverse coun-try take more pride in all that. 7

Pierogi and the British genius

Britain is a more cohesive society than the doom-mongers claim

Bagehot

The Economist January 7th 2017 43

1

IN THE middle of 2016 a suicide-bomberblew up a minibus full of judicial staff in

Kabul. The injured were rushed to theEmergency Hospital in the Afghan capital.One was married to a nurse there, whowas on duty when he arrived. He is nowparaplegic. She is “coping”, says a col-league: “She’s one tough woman.”

It is striking how many of the hospital’spatients were targeted for upholding thelaw. Amir Muhammad, a policeman withshrapnel wounds, says the Taliban at-tacked his post and killed seven of his 14fellow officers. “They had heavier weap-ons than us,” he explains.

The Taliban are as shrewd as they arebrutal. Afghanistan is close to becoming afailed state again. To avert that catastrophe,the government must provide adequatesecurity and establish something resem-bling the rule of law. But it is tricky to set upa functioning legal system when judgesand police officers keep getting murdered.Moreover, the government can hardlyclaim to be keeping people safe when theyfear being blown up on their way to work.

Since Barack Obama drastically re-duced the number of American troops inAfghanistan, the Taliban have madealarming gains. NATO forces there fell froma peak of 132,000 in 2011 to around 13,000today. Only about 60% of Afghans live in

American troops should leave Afghani-stan, and that they should “probably” stay.Afghans are nervous. “We hope this newAmerican administration will be suppor-tive too,” says AshrafGhani, the president.

Few things matter more than fixingfailed states. Broadly defined, state failureprovides “a general explanation for whypoorcountries are poor”, argue Daron Ace-moglu of the Massachusetts Institute ofTechnology and James Robinson of theUniversity of Chicago in “Why NationsFail”. Life in a failed or failing state is shortand harsh. Life expectancy in the bottom 16countries on the Fragile States Index com-piled by the Fund for Peace, a think-tank(see map on next page), is 85% of the globalaverage. Measured at purchasing-powerparity, income per head is a miserable 21%.

There goes the neighbourhoodLawless regions, such as the badlands ofPakistan and Yemen, act as havens for ter-rorists. And civil wars tend to spill acrossborders. The Rwandan genocide of 1994,forexample, sparked an even deadlier con-flagration in Congo.

In the most extreme form of state fail-ure, in places like Somalia, the central gov-ernment does not even control the capitalcity. In milder forms, as in Nigeria, the stateis far from collapse but highly dysfunction-al and unable to control all of its territory.Or, as in North Korea today or China underMao Zedong, it controls all of its territorybut governs in a way that makes everyonebut a tiny elite much worse off.

This article will lookat two main exam-ples: an unambiguously failed state, SouthSudan, and a state tottering on the brink,Afghanistan. It will argue that, as Mr Ace-moglu and Mr Robinson put it, the key to

areas controlled by the government. Oth-ers live under Taliban control (about 10%)or in areas that are violently disputed.

Wherever they can, the Taliban replacethe government’s justice with their ownswifter, harsher (and, some say, less cor-rupt) variety. If two peasants quarrel overapiece of land, a Taliban official will hearboth sides and make a ruling. Such rulingsoften stick, for no one doubts that the Tali-ban will enforce them.

The pull-out offoreign troops has madeAfghanistan not only more dangerous, butpoorer, too. Byone estimate, the NATO mis-sion cost almost $1trn between 2001 and2014—more than six times as much as Af-ghanistan’sGDPoverthatperiod. ManyAf-ghans sold stuff to and built things for theforeigners. Now that boom is over. Eco-nomic growth plunged from 14% in 2012 to0.8% in 2015.

“Day by day we are losing our busi-ness,” says Ashad Wali Safi, who runs anelectronics store in Kabul. The aid agenciesthat used to buy printers from him aregone. Security is “very bad”. (The previousday, a suicide-bomb had killed at least 30people in a nearby mosque.) “Even in day-time, we don’t feel safe. At night? Forget it.”

Adding to the uncertainty, no oneknows what Donald Trump’s Afghan poli-cy will be. In the past he has said both that

Fixing fragile nations

Conquering chaos

JUBA AND KABUL

Whystates fail and howtheycan be rebuilt: lessons from Afghanistan and South Sudan

International

44 International The Economist January 7th 2017

1

2 understanding state failure is “institutions,institutions, institutions”. The world’snewest country, South Sudan has receivedbillions of dollars of aid and the advice ofswarms ofconsultants since seceding fromSudan in 2011, buthas failed to build any in-stitutions worthy of the name. Afghani-stan faces a terrifying insurgency but has apresident doing his best to restore order.

States are not wretched and unstablebecause of geography—if so, how to ex-plain the success of landlocked Botswana?Nor is culture the main culprit: if so, SouthKoreans would not be more than 20 timesricher than North Koreans. Some societieshave “inclusive institutions that foster eco-nomic growth”; others have “extractive in-stitutions that hamper [it]”. South Sudan isan extreme example of the latter.

Never lookback“Everybody I know is getting out,” saysJoyce Mandi, as she mixes maize porridgefor her six children at a bus stop in SouthSudan. Around her, youngmen heave bagsand mattresses onto the roof of a minibus.Ms Mandi is fleeing her village and head-ing for Kakuma, a refugee camp in neigh-bouring Kenya. Her husband has gone intothe bush, she says, to fight the government.

The South Sudanese, who are mostlyblack African and non-Muslim, fought forhalf a century to secede from Sudan. ArabMuslims from the north used to oppressand enslave them. Perhaps 2m southern-ers died in the war of secession. But fewthink life has got better since then.

Those now in charge are former guerril-las from the Sudan People’s LiberationArmy (SPLA), a group of tribal militias un-ited only by hatred of the north. The firstpresident, Salva Kiir, tried to hold theSPLA’s factions togetherbypayingthem offwith petrodollars (oil is almost the onlything South Sudan exports). Alex de Waalof Tufts University estimates that in 2013the government paid salaries for 320,000soldiers, police and militiamen—morethan a tenth of all men aged 15-54. Many ofthese soldiers did not exist: their pay waspocketed by the warlordssupposedlycom-manding them.

But the state’s largesse did not buy loy-alty. Instead, it encouraged the men withguns to demand more. Then the moneyran out, thanks to collapsing oil prices anda suicidal game of chicken, in which thegovernment stopped production to try tosqueeze better terms from Sudan (whichcontrols the pipeline through which SouthSudanese oil is exported).

As his coffers emptied, Mr Kiir startedflagrantly to favour his own Dinka tribe,South Sudan’s biggest, to stay in power. Hisvice-president, Riek Machar, who is fromthe Nuer tribe, the second-biggest, wasforced out of government in July 2013 andwent back to war later that year. A peacedeal in 2015 quickly broke down. Some 3m

South Sudanese—a quarter of the popula-tion—have fled their homes. Were it not forfood aid, often dropped out of planes ontoremote villages, hundreds of thousandswould starve.

South Sudan failed to build institutionsthat transcended tribal loyalties or curbedthe power of warlords. Torit, where MsMandi boarded that minibus to Kenya, is agood place to observe the hollowness ofthe country’s government. Though it iscapital of one of South Sudan’s 28 states, itfeels like a military outpost. Troops in“technicals”—pick-up trucks with mount-ed machine-guns—patrol the streets.

There are plenty of government build-ings, including state ministries of educa-tion, culture and health. But none of themdoes much. Teachers were last paid in Sep-tember, says Jacob Atari, the local educa-tion minister. Inflation of over 800%means their monthly salary ofaround 300South Sudanese pounds is now worth lessthan $4. Over 70% of children are out ofschool, says Mr Atari.

Nowhere in South Sudan does the statedo what it is supposed to. Only 27% ofadults can read, according to the UN. Pre-ventable diseases such as cholera, measlesand malaria are rampant. The rule of law isa distant dream.

The country’s political system is in the-ory decentralised, but in reality the moneyflows through Juba, the national capital.And instead ofbeingdistributed to states, itis typically stolen or spent on weapons.Politics is a euphemism for armed battlesover plunder. The warlord who wins cansteal the oil and pay his troops. (Or, he cansimply let them rob civilians.)

The fighting becomes tribal becausewarlords recruit by stirring up ethnic ten-sion so that their kinsmen will rally tothem. This creates a vicious circle. Lackingprotection from other institutions, peopleseekit from theirown tribe. Rather than de-mand evenhanded government, they backtribal leaders, knowing that they will stealand hoping they will share the spoils withtheir kin.

The splintering of South Sudan can beglimpsed in the “protection of civilians”camps maintained by the UN. One in Jubaholds almost 40,000 people. At night, gun-shots are common and aid workers refuseto venture inside. Most of the residents areNuers, like Mr Machar, who have beenstranded here since the civil war began.They are sure who is to blame. “Our tribewas killed by the government and so wecame here,” says Kikany Kuol Wuol, a com-munity chairman in the camp. “We cannotleave, we have nowhere to go. If our wom-en just go outside to look for firewood theyare raped.” When fighting broke out inJuba in July between Mr Machar’s forcesand the government, it spread into thecamp as UN peacekeepers withdrew. Theproblem, says Mr Kuol, is that: “This is agovernment only for Dinkas. The rest of usthey want to starve to death.” Everyone inthe camp supports Mr Machar, he says.

MrAcemoglu and MrRobinson are pes-simistic about failed nations’ chances ofturning around. Extractive institutionstypically have historical roots. For exam-ple, the authors trace the failure of today’sDemocratic Republic of Congo partly tothe pre-colonial Kingdom of Kongo, wheretaxes were arbitrary (one was leviedwhenever the king’s beret fell off) and theelite sold their subjects to European slav-ers. Peasants therefore lived deep in the for-est, to hide from slavers and tax-collectors.They did not adopt new technology, suchas the plough, even when they heard of it.Why bother, when any surplus was sub-ject to seizure? Modern Congolese farmersmake similar complaints.

“Why Nations Fail” argues that “thepolitics of the vast majority of societiesthroughout history has led, and still leadstoday, to extractive institutions.” Thesetend to last because they give rulers the re-sources to pay armies, bribe judges and rigelections to stay in power. These rulersadopt bad policies not because they are ig-norant of good ones but on purpose. Let-ting your relatives embezzle is bad for thenation but great for your family finances.

1880.7

20

1.5

3.9

AMERICASAFRICA

WESTERNMED.

SOUTH-EASTASIA

WESTERNPACIFIC

Fragile StatesIndex score2016

<60.0

60.0-89.9

90.0-99.9

100.0+Source: The Fund For Peace

Afghanistan 107.9

S. Sudan 113.8

Congo 110.0

Yemen 111.5

Nigeria 103.5

Syria 110.8

Rwanda 91.3

WORST: Somalia 114.0

UnitedStates 34.0

Canada 23.8

Brazil 65.3

Britain 32.4

BEST: Finland 18.8

S. Korea 36.1

N. Korea 93.9

Libya 96.4

Liberia 95.5

Sierra Leone 91.0China 74.9

Indonesia 74.9

Botswana 63.5

Mexico 70.4

The Economist January 7th 2017 International 45

2 But failed states are not doomed to staythat way. Between 2007 and 2016, accord-ing to the Fragile States Index, 91 countriesgrew more stable and 70 grew shakier.Among those improving were giants suchas China, Indonesia, Mexico and Brazil.The worst performers were mostly small-er, such as Libya and Syria.

Even states that have collapsed com-pletelycan be rebuilt. Liberia and Sierra Le-one were stalked by drug-addled child sol-diers a decade and a halfago; now both arereasonably calm. The key is nearly alwaysbetter leadership: think of how Chinachanged after Mao died. Many bad rulerscontinue deliberately to adopt bad poli-cies, but they can be—and often are—re-placed with better ones.

Instructions includedAfghanistan’s president since 2014, MrGhani is a former academic and author ofa book called “Fixing Failed States”. HisTED talk on fixing broken states has beenviewed 750,000 times. Now he is trying toput his own theories into practice.

Yet he admits that rebuilding Afghani-stan is more complex than he expected.The insurgents draw support from severalsources: local grievances, tribal animos-ities, global Islamist networks, organisedcrime (Afghanistan is the world’s largestproducer ofopium) and the Pakistani secu-rity services. In 2015 Mr Ghani accusedPakistan of being in an “undeclared stateof hostility” towards his country. Now hegoes further. “In October it was almost adeclared state of hostility,” he says. The Ta-liban enjoy havens in Pakistan’s lawless ar-eas and, analysts suspect, direct help fromPakistani spooks, some of whom wouldrather have Afghanistan in chaos than seeIndia gain a foothold there. Recent suicide-bombs in Kabul appear to have containedmilitary-grade explosives, which Afghansassume came from over the border.

Mr Ghani has a clear idea of the state’sbasic functions. First, it must uphold therule of law. Second, it must secure a mo-nopoly on the use of violence. The two arelinked. As Sarah Chayes points out in“Thieves of State”, when people see thestate as predatory, they are more likely tosupport insurgents. She cites the exampleof an Afghan who was shaken down ninetimes by police on a single journey, andvowed not to warn them ifhe saw the Tali-ban planting a bomb to kill them.

Mr Ghani justly takes credit for the factthat the Taliban did not overthrow thestate after Mr Obama’s pull-out. “In 2015we were in danger, because the global andregional consensus was that we would notbe able to hold,” he says. Now, says Gen-eral John Nicholson, the commander ofthe NATO forces in Afghanistan, the Tali-ban have been fought to a stalemate. Theyseized a big city, Kunduz, in 2015, but weredriven out and have taken no more since.

NATO air power combined with Ameri-can-trained Afghan special forces pack “anoffensive punch”, says General Nicholson.The Taliban cannot mass troops for fear ofNATO bombs. However, they have “safetyoutside the country”.

Mr Ghani is less tolerant of corruptionthan was his predecessor, Hamid Karzai,and appears to have cleaned up customsand government procurement a fair bit. Hehas improved tax collection and promotedinfrastructure projects, such as rail linksand powerplants, in the hope that Afghan-istan will become a central Asian hub. (Henotes with satisfaction that the Talibanhave said they will not attack suchschemes.) He promotes education forwomen, which was banned when the Tali-ban ruled Afghanistan in 1996-2001. Toconservative Afghans who think thiswould lead to illicit mixing with men, hehas a convincing response. “In the remoteprovinces, they are asking for women doc-tors,” he points out. How can they have fe-male doctors if they do not allow theirdaughters to go to school?

Nonetheless, a surveybythe Asia Foun-dation finds that only 29% of Afghans be-lieve the country is moving in the right di-rection. This is largely because 70% fear fortheir safety—the highest level in over a de-cade. However, a slim majority (54%) saythe army is gettingbetterat providing secu-rity, while only 20% say it is getting worse.

Public perceptions of corruption havebarely budged since Mr Ghani came topower, with 89% of Afghans saying it is aproblem in their daily life. More encourag-ingly, the share of those who had dealtwith police and reported sometimes hav-ing to pay bribes is falling somewhat: from53% in 2015 to 48% in 2016.

Foreign donors warm to Mr Ghani. In

October he convinced them to pledge$15bn over the next four years. Yet he isleery of how aid is dispensed. The flood offoreign cash that followed the American-led toppling of the Taliban government in2001 often undermined the state or waswasted. Aid agenciespaid salaries20 timeshigher than the Afghan civil service,prompting the best officials to quit to workas drivers and interpreters. Mr Ghani haslong argued that aid should flow throughthe national government, rather than sup-port a parallel state that can packup and gowhen donor fashions change. He may begetting his way: roughly half of aid nowpasses through the national budget, ashare that is expected to rise.

Even with a leader determined to makegood choices, building an honest state ishard. MrGhani complainsofinaccurate in-formation. “There were three databases inthe MinistryofEducation: one for teachers,one for salaries, one for schools…theyweren’t talking to each other.” Faulty re-cords make it easier for money to vanish.Digital payments should help, he says—thepolice thought they had received a pay risewhen the first experiments with mobilepayments began, because commanderscould no longerskim theirwages. But thereis a long way to go.

“A decade ago, if you went into a minis-ter’s office, you’d see dust on the desk, nocomputer and the minister picking his toe-nails,” says a Western official in Kabul.“Now you have competent ministers andlots of young professional staff who keepin touch via WhatsApp and speak English.The bad news is that Ghani is still learninghow to be a politician. Karzai would get onthe phone with tribal leaders and chatabout their fathers’ health [before talkingbusiness]. Ghani tries to book them for aten-minute meeting, and hustles them outof the door before the tea is cold.”

This is a common criticism. Mr Ghani isgood at retail politics (he won the disputedelection in 2014 partly because he hadspent so much time sitting in villages ask-ing ordinary Afghans what they wanted).But he is a technocrat among warlords,some of whom have been made billion-aires by the drugs trade. He rules in uneasycoalition with a “chief executive” with ill-defined powers: Abdullah Abdullah, theman he beat in 2014. His vice-president is ablood-spattered warlord. The presidentwill struggle to build a clean state when somany bigwigs prefer it dirty, critics say.

Mr Ghani dismisses the charge. “If poli-tics becomes all tactics, where would youproduce change?” he asks. He insists thathe bends over backwards to be respectfulof tribal leaders, “but it cannot be at the ex-pense of building institutions.” This is acrucial point. Countries whose stabilitydepends on an individual strongman arebrittle. Those that create inclusive institu-tions need never fail again. 7

Politics by other means in South Sudan

46 The Economist January 7th 2017

For daily coverage of business, visit

Economist.com/business-finance

1

LARGE food companies have long beenamong the world’s most solid, with re-

assuringly consistent returns even in hardtimes. None would seem steadier thanNestlé, based in the Swiss town of Vevey,on a lake near snowy peaks. For its 150thanniversary in 2016 it opened a new muse-um filled with corporate heirlooms: thefirst written notes about a new productcalled milk chocolate, laid out in black cur-sive; an old tin ofNescafé, used by soldiersas a stimulant in the second world war;and an early can of Henri Nestlé’s infantformula, which in 1867 saved the life of apremature baby.

It has come a long way since then. Itsold goods worth nearly $90bn in 189countries in 2015. Ofthe 30,000 cupsofcof-fee sipped around the world each second,Nestlé estimates, one-fifth are cups of Nes-café. But the industry it presides over is inupheaval. On January1st a new chief exec-utive, Ulf Mark Schneider (pictured), tookover. He is the first outsider to get the topjob since 1922, and his background—run-ning a health-care firm, not selling choco-late bars or frozen pizza—suggests the mainsource ofworry for the business.

More and more consumers are snub-bingpackaged food’s sugar, salt and unpro-nounceable preservatives. Meanwhile,swarms of smaller firms, emboldened bythe ease ofpeddling goods online, are tout-ing supposedly healthier options. From2011 to 2015 big sellers of consumer-pack-

cuts wreck firms’ growth prospects evenfurther, or whether, in fact, they are best offaccepting that robust expansion is a thingof the past and wringing out profits.

Nestlé is not immune to such pressures.In recent years it has often missed its goalof 5-6% sales growth. Excluding acquisi-tions, its numbers have not met investors’expectations in 11 of the past 17 quarters. Inthe most recent quarter, the firm registeredorganic sales growth of3.2%.

Changing consumer tastes explainsome of these shortfalls. So does a shiftingretail landscape. Managing a giant portfo-lio ofbrands, from KitKat and Nespresso toDiGiorno pizza and Purina dog food, hasbecome harder. Mr Schneider will have tomaster online ways to market and deliverits well-known brands. The firm needs tocoax customers to pay more for premiumproducts as ordinary ones get commodi-tised, and discounted by firms such as Ger-many’s Lidl and Aldi.

The firm can still boast impressive stay-ing power—its global market share acrossits entire range of products has remainednear 20% for the past decade. François-Xa-vier Roger, Nestlé’s chief financial officer,points out that the group’s sales growth inthe first nine months of 2016 was amongthe fastest of the top ten biggest food anddrink companies. Yet a detailed examina-tion of its position by Sanford C. Bernstein,a research firm, shows that when growthfrom acquisitions is excluded, it lost sharein all but three of its top 20 product catego-ries between 2007 and 2015. Some of itscore offerings, such as bottled water andsingle-serve coffee, fared the worst. (Keu-rig, Nestlé’s arch-rival in coffee pods,slurped share in America.)

Such results are likely to attract particu-lar censure from investors because of Nes-tlé’s past heavy emphasis on growth andmarket share, which sometimes came at

aged goods, mainly food and drinkcompa-nies, lost three percentage points of marketshare in America—a lot in the industry’scontext—according to a study by the Bos-ton Consulting Group, a consultancy, andIRI, a data provider.

As super-sized companies swat at suchtiny attackers, another foe is gainingground. 3G, a Brazilian private-equity firm,likes to buy big, slow-growing food anddrinks companies and slash their costs.Targets have included Kraft and Heinz, twogiants which 3G helped merge into onegroup in 2015, as well as several of theworld’s biggest brewers. Other food com-panies are scrambling to make cuts of theirown, lest theybecome 3G’snextmeal. Thathas prompted a debate over whether such

Nestlé

A life less sweet

VEVEY

As rivals nibble at its business, Nestlé’s new boss must find a formula forgrowth

BusinessAlso in this section

47 Toshiba’s latest write-down

48 Ford and Donald Trump

49 Schumpeter: The fat-cow years

The latest recipe

Sources: Bloomberg; company reports *Year to October 2016

2015 fiscal year

0 10 20 30 40 50

Return on equity%

Operating costAs % of sales

Operating profitAs % of sales

Gross profit$bn

AB InBev NestléKraft-Heinz*

The Economist January 7th 2017 Business 47

1

2 the expense of the firm’s profits. In 2015 itsoperating-profit margin was 15%, betterthan the 13% at Danone, a French competi-tor, but far below the 21% at Kraft-Heinz.Shareholders in the firm are waiting to seewhether Mr Schneider will shake thingsup. Some want him to sell off businessesthat seem most at riskof long-term decline,such as frozen food, as shoppers look forfresher fare.

Food for life?For now, Nestlé is defiant. “We started 150years ago having a product that actually—there’s symbolism there—saved the life ofachild,” says Paul Bulcke, the outgoing chiefexecutive. He and his colleagues say thatinvestment in health and related innova-tion will produce strong growth at thecompany for years to come. Mr Schneider,who used to run Fresenius, a big Germanfirm that offers kidney-dialysis productsand other medical services, will certainlyemphasise that message. Nestlé differen-tiates itself from 3G, with its keen focus oncuts. Mr Roger says he respects what 3Gdoes, but that “they have a strategy whichis very different from ours.”

Still, few observers would call Nestlé ahealth company. Many of its products areperfectly healthy, including bottled waterand coffee. Many are not—milk chocolateand ice cream, to name but two. And fornow, the purest forms of Nestlé’s focus onhealth contribute relatively little to itssales. A business unit called Nestlé HealthScience, for example, sells nutritional pro-ducts for medical needs, such as vitamin-packed drinks for the elderly and for can-cer patients. It contributes less than 5% ofrevenue.

The firm has a research institute de-voted to studying food’s role in the man-agementand prevention ofdisease—for ex-ample, better understanding nutrition’sability to promote brain health. It maybring growth but probably only in the longterm. Nestlé has also partnered withyoung drugs firms, including one that istesting a treatment for ulcerative colitis.

More immediately rewarding may beits efforts to make best-selling but un-healthy foods a bit more wholesome. InNovember the companysaid ithad createdhollow sugar crystals that taste sweet butcontain fewer calories than the usual stuff.It will begin to put the new ingredient in itschocolate in 2018.

It is also proud of changes to the mil-lions of frozen dinners it sells every weekin America. Shoppers had been avoidingthe frozen-food aisle. Nestlé first tried dis-counts, and then in 2015 introduced newversions of its Lean Cuisine products, strip-ping out unpalatable ingredients and re-placing them with organic ones. AtStouffer’s, another frozen brand, Nestlé de-cided to target men with easy, protein-packed meals that are more nutritionally

valuable. It worked—its frozen-food salesin America grew faster. In November 2015they were 6% above what they had been ayear earlier. But Bernstein’s Andrew Woodpoints out that the revival of frozen foodnow looks wobbly again.

Nor is Nestlé ignoring 3G’s strategy en-tirely: it is trying to trim expenses. “We arevery much in an investment position, notin a cost-cutting exercise,” says Mr Roger,“but that doesn’t mean that we don’t wantto be cost-efficient in what we do.” One ef-fort, which includes trimming waste at fac-tories, is credited with saving aboutSFr1.5bn ($1.5bn) a year. Last year Nestlé an-nounced organisational changes, such as

consolidating procurement, which willsave about SFr2bn each year from 2020.

Whatever else Mr Schneider has on themenu for Nestlé, radical changes may besomewhat limited by the fact that so manyof those who built the company into whatit is now are sticking around. Mr Bulcke isexpected to become its chairman. The out-going chairman, Peter Brabeck-Letmathe, aformer Nestlé chief executive, may be-come honorary chairman. Mr Bulcke, forone, seems sure that the company shouldmaintain its strong emphasis on the longterm. He taps his hand on the table, rattlingsome Nespresso cups, as he insists thatgrowth is still the key. 7

THE probe in 2015 into one of Japan’slargest-ever accounting scandals, at

Toshiba, an electronics and nuclear-powerconglomerate that has been the epitome ofthe country’s engineering prowess, con-cluded that number-fiddling at the firmwas “systemic”. It was found to have pad-ded profits by ¥152bn ($1.3bn) between2008 and 2014. Its boss, and half of theboard’s 16 members, resigned; regulatorsimposed upon it a record fine of$60m.

Now its deal-making nous is in doubttoo. In December 2015—the very samemonth that it forecast hundreds of billionsof yen in losses for the financial year thenunder way, as it struggled to recover fromthe scandal—Toshiba’s American arm,

Westinghouse Electric, bought a nuclear-construction firm, CB&I Stone & Webster.One year on, on December 27th, Toshibaannounced that cost overruns at that newunit could lead to several billionsof dollarsin charges against profits.

Its shares fell by 42% in a three-daystretch as investors dumped them, fearinga write-down that could wipe out its share-holders’ equity, which in late Septemberstood at $3.1bn. Moody’s and S&P, two rat-ings agencies, announced credit down-grades and threatened more. Toshiba’s ex-planation for how it got the numbers sowrong on a smallish purchase is woolly.But it is clear that missing constructiondeadlines on nuclear-power plants cansend costs skyrocketing. Its projects inAmerica, and in China, are years behindschedule. Mycle Schneider, a nuclear ex-pert, says that in America, as elsewhere,engineeringproblemsare compounded bya shortage of skilled manpower. Fewplants have been built there recently.

Part of the $229m that Westinghousepaid for CB&I Stone & Webster included$87m of goodwill (a premium over thefirm’s book value based on its physical as-sets). It is that initial estimate that is nowbeing recalculated.

Toshiba had looked to be bouncingbackfrom its accounting nightmare. Beforethe latest plunge it had made the second-biggest gains on the Nikkei 225 index in2016, where its shares were up by 77%. InApril it wrote off $2.3bn on the goodwillvalue of Westinghouse, purchased for$5.4bn in 2006—a write-down that it hadlong avoided. In August it announced itsfirst profit in six quarters. It forecast a netprofit of ¥145bn for the financial year of

Toshiba

Losing count

TOKYO

Japan’s enfeebled giant faces a multi-billion-dollarwrite-down

Ritual contrition

48 Business The Economist January 7th 2017

2 2016-17, a clearreversal from its¥460bn lossof the previous year. Part of that wasthanks to a bold turnaround plan: firing14,000 staff, as well as selling lossmakingparts of its manufacturing empire, like TVs,and one of its star units, a medical-equip-ment maker, for $6bn.

That left it free to focus on its semicon-ductor arm, which has been buoyed by de-mand from Chinese smartphone makers,and its nuclear unit, which accounts for athird of its revenue. The latest write-downcould dampen future investment in both.Toshiba has limited ways left to raise cash.It has been barred from doing so on thestockmarket ever since it was put on alertafter the accounting fiasco—one step shortofa delisting.

Observers reckon that Toshiba hassome room to manoeuvre, and that it willnot ditch its nuclear business. It could raiseas much as $4bn from the sale of somepart-owned subsidiaries, including Nu-Flare, a spinoff of its semiconductor unit,says Seth Fischer of Oasis Management inHong Kong, a hedge fund, and a share-holder in Toshiba’s power-station affiliate.It could even choose to sell its lucrativechip business altogether (Toshiba is theworld’s second-biggest maker of NANDchips after Samsung Electronics of SouthKorea), as well as some of its remainingconsumer-electronics ones.

Toshiba’s central part in a plan by thegovernment ofShinzo Abe, the prime min-ister, to pep up growth by exporting nuc-lear-power technology to emerging coun-tries may help. In June Westinghouseclinched a deal in India to build six new-generation AP1000 reactors, Toshiba’s firstorder since the triple meltdown at the Fu-kushima Dai-ichi nuclear plant in 2011.Toshiba is also involved in that site’s costlyand complex clean-up. Some think thatJapanese banks, known for keeping zom-bie firms on life support, will stand behindit, come what may. Shares in Toshiba’s twomain lenders, Sumitomo and Mizuho, slidlast week after the profit warning. Inves-tors expect more big bank loans or a debt-for-equity swap, which allows a bank toturn bad loans into shares.

The consensus on Toshiba’s latestscrew-up is that a long-standing culture ofpoor management is to blame. Toshiba’saudit committee, for example, was until2015 headed by its formerchieffinancial of-ficer; such bodies should be fully indepen-dent, says Nicholas Benes of the Board Di-rector Training Institute of Japan. It is notclearwhetherornot the firm hasfully over-hauled its culture as part of its response to

the scandal laid bare in 2015. Satoshi Tsuna-kawa, who was installed as the company’snew boss in June 2016, said last week thathe had only become aware of the problemwith CB&I Stone & Webster in December. Itwas in 2015 that Mr Abe introduced Japan’sfirst detailed rules on how companiesshould run themselves. The spectacle ofToshiba’s apparently endless crisis sug-gests more needs to be done. 7

Westinghouse woes

Source: Thomson Reuters

Toshiba, share price, yen

2015 16 170

100

200

300

400

500

600ACCOUNTING SCANDAL

CB&I STONE &WEBSTER

WRITE-DOWN

RECORD LOSS $4.6BN

WESTINGHOUSEWRITE-DOWN

CEO RESIGNS

Donald Trump and business

Wheel spin

IT WAS in the spring of2016 that DonaldTrump singled out Ford Motors, calling

its plans to build a plant in Mexico an“absolute disgrace” and promising itwould not happen on his watch. Backthen, it seemed remarkable that thecandidate thought he could boss arounda firm ofFord’s stature. On January 3rdFord cancelled its $1.6bn project in theMexican state ofSan Luis Potosí and saidit would instead invest $700m into anexisting plant in Flat Rock, Michigan, tobuild electric and autonomous cars.

Ford’s manoeuvre seems more wheel-spin than U-turn. Mr Trump’s strong-arming ofcorporate America is realenough, and the carmaker will havegained much favour with the president-elect. But its decision can be explainedlargely in operational terms. The originalplan was for the new Mexican plant tobuild chiefly Focus cars—small passengervehicles for which demand has fallen,thanks to America’s love affair withSUVs, crossovers and pick-up trucks andto low petrol prices. The decision to scrapthe new plant looks far more like Fordreducing its exposure to the small-cargame in North America than reducing itsfootprint in Mexico, says George Galliersat Evercore, an investment bank.

The firm will still move production ofthe Focus away from its plant in Wayne,Michigan to an existing plant in Hermosi-llo, Mexico. As for the upgrade of the Flat

Rockfacility, where Ford this week trum-peted 700 new jobs to come, the firm hadalready announced back in December2015 that it would invest in electrificationand in 13 new electric vehicles. Linkingone location for that (Flat Rock) with theMexican plant cancellation looked likeyet more accomplished spin.

Things would undoubtedly be diffi-cult for global carmakers ifMr Trumptried to follow through on a campaignpromise to slap a 35% tariffon cars ex-ported from Mexico to America. In 2015the country exported 2.7m vehicles, overfour-fifths ofwhich went to North Ameri-ca. By appearing to kowtow to the newboss-in-chief, Ford’s chiefexecutive,MarkFields, may hope to keep this threatat bay—and to extract other favourableconcessions, such as softer rules on emis-sions standards. “We have a president-elect who has said very clearly that oneofhis first priorities is to grow the econ-omy,” enthused Mr Fields. “That shouldbe music to our ears.”

Next in the line offire is General Mo-tors, America’s biggest carmaker, whichsaid in 2013 that it would invest $5bn inMexico over six years. This weekMrTrump admonished it for making itsChevy Cruze, another compact car, most-ly over the border. “Make in U.S.A. or paybig border tax!” he tweeted. The com-pany may find it hard to match Ford’sskilful road-handling.

Ford Motors cancels a new plant in Mexico

The Economist January 7th 2017 Business 49

IN THE Bible, seven years of feast were followed by seven yearsof famine. For banks there have been ten lean years. Subprime-

loan defaults started to rise in February 2007, causing a near-col-lapse of the industry in America and Europe. Next came bail-outsfrom governments, then years of grovelling before regulators,mass firings of staff and quarter after quarter of poor results thatleft banks’ shareholders disappointed. Now, a decade later, themoneylenders are quietly wondering if 2017 is the year in whichtheir industry turns a corner.

Over the past six months the FTSE index ofglobal bankshareshas leapt by 24%. American banks have led the way, with the val-ue of Bank of America rising by 67%, and that of JPMorgan Chaseby 39%. In Europe BNP Paribas’ market value has risen by 52%. InJapan shares in the lumbering Mitsubishi UFJ Financial Group—the rich world’s biggest bank by assets—have behaved like thoseof a frisky internet startup; they are up by 57%. Predictions aboutglobal banks’ future returns on equity have stopped falling, noteanalysts at UBS, a Swiss bank. Some of the biggest casualties ofthe financial crisis are even expanding. On December 20thLloyds, bailed out by British taxpayers in 2009 at a cost of $33bn,said it would buy MBNA, a credit-card firm, for $2bn.

The excitement can be explained by three Rs: rates, regulationand returns. Consider interest rates first. The slump in rates hasbeen terrible forbanks. Between 2010 and 2015, the net interest in-come of the rich world’s 100 biggest banks fell by $100bn, orabout halfof2010 profits. When rates across the economy rise, bycontrast, banks can expand margins by chargingborrowers more,while passing on only some of the benefit of higher rates to de-positors. So bankers have been watching the bond market withbarely concealed joy. Ten-year government yields have risen byone percentage point in America, and by 0.30-0.64 points in thebig euro-zone economies and Japan over the last six months. In-vestors are talking about a Trump-inspired “reflation”: the presi-dent-elect promises to embark on a public-spending boom. InGermany inflation is at a three-year high of1.7%.

Banks’ CEOs are also chipper because they think that regula-tion has peaked. In America the new administration is likely ei-ther to repeal the Dodd-Frank act, an 848-page law from 2010, orto prod regulators to enforce it less zealously. Bank-bashing fa-

tigue seems to have set in among the public. True, when firmsmisbehave, there is still a firestorm of outrage. John Stumpf, theboss ofWells Fargo, quit in October after his bankadmitted creat-ing fake accounts. But many people can see that power has mi-grated from banking to the technology elite in California. Thebrew of high pay, monopolistic tendencies and huge profits thatattracts populist resentment is now more to be found in SiliconValley than in Wall Street or the City ofLondon.

Global supervisors are still cooking up new rules, known as“Basel 4” (see page 51), but are unlikely to demand a big rise in thesafety buffer the industry holds in the form of capital. The stron-gest banks are signalling that they will lay out more in dividendsand buy-backs, rather than hoard even more capital (today, thetop 100 rich-world banks pay out about 40% of their profits).

A third reason for optimism in bank boardrooms is returns.Global banking’s return on equity (ROE) has crept back towards arespectable 10%. The worst ofthe fines imposed by American reg-ulators are over. So far, “fintech” startups that use technology tocompete with rich-world banks have not won much marketshare; banks have used technology to boost efficiency. They havealso got better at working out which of their activities create val-ue after adjusting for risk and the capital they tie up. Barclays,once known for cutting corners, says it can calculate the ROE gen-erated by each of its trading clients. It is ditching 7,000 of them.

Given the giddy mood, the big danger starts with a C, for com-placency. Regulators believe that banks now pose less of a threatto taxpayers. American lenders have $1.2trn of core capital, morethan twice what they held in 2007. Citigroup, the most systemi-cally important bank to be bailed out, now has three times morecapital than its cumulative losses in 2008-10. European banks’capital buffers have risen by 50% since 2007, to $1.5trn.

Yet there are still plenty of weak firms that could cause may-hem. Deutsche Bank, several Italian lenders and America’s twostate-run mortgage monsters, Fannie Mae and Freddie Mac, areexamples. Mega-banks may simply be too big for any mortal tocontrol. For every dollar of assets that General Electric’s Jeff Im-melt manages, Jamie Dimon at JPMorgan Chase looks after $5.

Once bittenAnd banks still lack a post-crisis plan beyond cost-cutting. De-spite their surging shares, most are valued at around the levelthey would fetch if their assets were liquidated, which hardly in-dicates optimism about their prospects. Before the crisis, they in-flated their profits by expanding in unhealthy ways. They cap-tured rents from state guarantees, created evermore layers ofdebtrelative to GDP, and grewtheirbalance-sheetsbymeansof heavyover-borrowing. They have reversed much of this expansionover the past decade but that strategy cannot go on for ever.

In 2017 banks will need to articulate a new growth missionand show that they can expand profits without prompting publicoutrage or a regulatory backlash. One area ofpromise is the driveto raise rich-world productivity. That would boost economiesbroadly, and their own profits. There is plenty that banks coulddo: get more credit to young firms, improve payments systems sothat a higher proportion of midsized firms can engage in cross-border e-commerce, and harness technology to make banking ascheap and easy to use as a smartphone app. Forward-thinkingbank bosses are already emphasising such goals. If they couldachieve them over the next decade, they might even realise afourth R—redemption. 7

The fat-cow years

Banks in the rich world are getting theirappetites back. Don’t be too scared

Schumpeter

50 The Economist January 7th 2017

For daily analysis and debate on economics, visit

Economist.com/economics

1

MOSTeconomistsmighthazard a guessthat voiding the bulk of a country’s

currency overnight would dent its imme-diate growth prospects. On November 8thIndia took this abstruse thought experi-ment into the real world, scrapping twobanknotes which made up 86% of all ru-pees in circulation. Predictably, the econ-omy appears indeed to have been hobbledby the sudden “demonetisation”. Evidenceof the measure’s costs is mounting, whilethe benefits lookever more uncertain.

At least the newyearhasbrought a sem-blance of monetary normality. For sevenweeks queues had snaked around banks,the main way for Indians to exchange theirold notes for new ones or deposit them intheiraccounts. That is over, largely becausethe window to exchange money closed onDecember30th. The numberoffresh notesthat can be withdrawn from ATMs or bankcounters is still curtailed, but the acute cashshortage is abating, at least in big cities.

As data trickle through, so is evidenceofthe economic price paid fordemonetisa-tion. Consumers, companies and investorsall wobbled in late 2016. Fast-moving con-sumergoods, usuallya reliable growth sec-tor, retrenched by 1-1.5% in November, ac-cording to Nielsen, a research group.Bigger-ticket items seem to have been hitharder. Year-on-year sales at Hero Moto-corp, the biggest purveyor of two-wheel-ers, slid by more than a third in December.

A survey of purchasing managers in

tors, such as the rise in the oil price and thesurge in the value of the dollar after theelection ofDonald Trump, are also at play.

Whether the costs of the exercise justifythe benefits depends, of course, on whatthose benefits are. In his speech announc-ingthe measure, Narendra Modi, the primeminister, highlighted combating corrup-tion and untaxed wealth. Gangsters andprofiteers with suitcases full of moneywould be left stranded. But reports suggestthat nearly 15trn rupees of the 15.4trn ru-pees taken out of circulation are now ac-counted for. So either the rich weren’thoarding as much “black money” as wassupposed, or they have proved adept atlaundering it. The Indian press is full oftales ofhousehold staffpaid months in ad-vance in old notes, or of bankers agreeingto exchange vast sums illegally.

Fans of demonetisation point to threebeneficial outcomes. First, banks, ladenwith fresh deposits, will lend this moneyout and so boost the economy. Big bankscut lending rates this week (quite possiblynudged by government, the largest share-holder of most of them). But their lendingrecently has not been constrained by a lackof deposits, so much as by insufficientshareholder capital to absorb potentiallosses, and by the over-borrowed balance-sheets ofmany industrial customers.

Second, Indians will move from livingcash in hand into the taxed formal econ-omy. Mr Modi has recently promoted theidea of a cashless, or “less-cash”, India (notsomething mentioned at the outset), asone reason for demonetisation. Progresstowards getting Indians to pay for thingselectronically is indeed being made, butfrom an abysmally low base.

The third upshot is the most controver-sial. Now that the demonetised banknotesare worthless, the government is intent onin effect appropriating the proceeds. The

manufacturing plunged from relative opti-mism throughout 2016 to the expectationofmild contraction. Firms’ investment pro-posals fell from an average of2.4trn rupees($35bn) a quarter to just 1.25trn rupees inthe one just ended, according to Centre forMonitoring Indian Economy, a data pro-vider. As a result, corporate-credit growth,already anaemic, has reached its lowestrate in at least 30 years (see chart).

All this amounts to “a significant butnot catastrophic” impact, says Shilan Shahof Capital Economics, a consultancy. An-nual GDP growth forecasts for the fiscalyear ending in March have slipped byaround half a percentage point, to under7%, from an actual rate of7.3% in the last fullquarter before demonetisation. Other fac-

Indian economics

Many rupee returns

MUMBAI

The impact of India’s radical monetaryreform is becoming clearer

Finance and economicsAlso in this section

51 Impact investing

51 Bank capital requirements

52 Buttonwood: The new global regime

53 Sub-national currencies

53 Futures and options trading

54 Sir Anthony Atkinson, RIP

54 Deaths of the rich and famous

55 Free exchange: Brexit and the fallacyof the “WTO option”

Less cash, less credit

Source: Thomson Reuters

India, bank credit to the commercial sector% increase on a year earlier

2000 02 04 06 08 10 12 14 160

5

10

15

20

25

30

The Economist January 7th 2017 Finance and economics 51

1

2 procedure requires trampling on the credi-bilityofthe Reserve BankofIndia (RBI), thecentral bank, which must first agree to dis-honour the promise, on all banknotes, to“pay the bearer” the value. If it does so, “ex-tinguishing” the notes and its liability forthem, it can transfer an equivalent amountto the government budget.

With so much cash handed in at banks,the amount remitted to government by theRBI might amount to perhaps 0.2-0.3% ofGDP. Proceeds from a tax-amnesty schemefor cash-hoarders may swell the figure.Even so, it will not be enough to justify thecosts ofdemonetisation—oreven, perhaps,the damage to the reputation of the RBI,which is already facing questions about itsindependence. But having imposed thecosts, Mr Modi will be keen to trumpetwhatever benefits he can find. 7

WHEN investors gathered in Amster-dam in late 2016 for perhaps the larg-

est annual conference on “impact invest-ing”, the mood was upbeat. The concept ofinvesting in assets that offer measurablesocial or environmental benefits as well asfinancial returnshascome a longway fromits modest roots in the early 2000s. Panel-lists at the conference included, amongothers, representatives of two of theworld’s largest pension funds, TIAA ofAmerica and PGGM of the Netherlands,and of the asset-management arm of AXA,a French insurance behemoth. Aniche pro-duct is inching into the mainstream.

In the past two years BlackRock, theworld’s biggest asset manager, launched anew division called “Impact”; GoldmanSachs, an investment bank, acquired animpact-investment firm, Imprint Capital;and two American private-equity firms,Bain Capital and TPG, launched impactfunds. The main driver ofall this activity isinvestor demand. Deborah Winshel, bossof BlackRock Impact, points to the transferofwealth to women and the young, whoseinvestment goals, she says, transcend merefinancial returns. Among institutions,sources of demand have moved beyondcharitable foundations to hard-bitten pen-sion funds and insurers.

The sector has also been boosted by in-creased attention from policymakers andthe development of industry standards. In-ternational organisations—such as the UN,and a global task force founded under theaegis of the G8—have promoted impact in-

vestment. Bodies such as the council of in-vestors and borrowers that sets the GreenBond Principles, guidelines for bonds ear-marked for environmental projects, havehelped set common standards.

Definitional squabbles still plague theimpact community. For sticklers, invest-ment only deserves “impact” status if it de-livers both near-market level returns andstrict measurement of the non-financialimpact: eg, of the carbon emissions savedby a renewable-energy project; or of thenumber of poor people who borrow froma microcredit institution. Others, however,include philanthropic investment, wherefinancial returns are sacrificed for greatersocial benefits; or less rigorous types ofdo-good investments.

Such disagreements make it hard togauge the true extent of impact invest-ment. For instance, BlackRock Impact andGoldman both also offer two looser invest-ment categories: “negative screening” (ie,not investing in “bad” sectors—say, tobaccoor oil); and “integrated” investments thattake environmental, social or governance(ESG) considerations into account (eg, byselecting for firms with, say, good workingconditions). Neither firm, however, pro-vides a complete breakdown of these cate-gories by assets under management.

The industry is also held back by a re-stricted choice of asset classes, and by thelimited scale of investment opportunities.According to a surveyby the Global ImpactInvesting Network, which organised theconference in Amsterdam, investors weremanaging $36bn in impact investments in2015. But the median size of investment re-mained just $12m. Urban Angehrn, chiefinvestment officer of Zurich Insurance,says the Swiss firm has had trouble fulfill-ing its pledge to commit 10% of its private-equity allocation to impact investments.

Cynics may still dismiss impact invest-ing as faddish window-dressing. Of Zu-rich’s $250bn-plus in assets under manage-ment, only $7bn-worth are classified asimpact investments. At Goldman’s asset-management arm, impact and ESG-inte-grated investments combined only makeup $6.7bn out ofa total $1.35trn in assets un-

der management. But that is to ignore the scale and pro-

gress that large institutional investors havebrought to impact investing. Although$7bn is a tiny slice of Goldman’s portfolio,it is huge compared with the investmentsof even well-established impact special-ists, such as LeapFrog, whose commit-ments total around $1bn. And the entry ofhard-nosed financial giants sends an im-portant message about impact investing:that they see it as profitable for themselvesand their clients. It is not enough to makeinvestors feel good about themselves; theyalso want to make money. 7

Impact investing

Coming of age

Investing to do good as well as to makemoneyis catching on

Doing good doing better

Source: Global ImpactInvesting Network *61 respondents worldwide

Impact investingAssets under management, by sector*, $bn

0

10

20

30

40

2013 14 15

Microfinance

Other financial services

EnergyOther

SOME banks find existing capital require-ments too taxing. To no one’s surprise,

on December 23rd Monte dei Paschi di Si-ena, at present Italy’s fourth-biggest bank,asked the Italian state for help, havingfailed to raise from the private sector €5bn($5.2bn) in capital demanded by the Euro-pean Central Bank before the year’s end.Three days later Monte dei Paschi said thatthe ECB had redone its sums—and conclud-ed that the stricken lender faced an evenbigger shortfall, of€8.8bn.

Plenty of other European banks—in farbetter nick than poor old Monte dei Paschi,which is overloaded with bad loans—aregrumbling that they too may eventuallyhave to find more capital. They have spentyears plumpingup cushions that the finan-cial crisis showed to be worryingly thin,but fear that proposed adjustments to Ba-sel 3, the latest global standards, will re-quire more. The Basel Committee on Bank-ing Supervision, which draws up thestandards, had hoped to agree on the revi-sions by the end of 2016. It’s not there yet:on January 3rd an imminent meeting ofcentral-bankgovernorsand supervisors, toapprove the changes, was postponed.

The amendments are intended to re-duce the variation in banks’ own calcula-tions of risk-weighted assets (RWAs), large-ly by restricting their use of in-housemodels. Under Basel rules, the ratio of abank’sequity to itsRWAsare a keygauge ofits strength: if lenders are too sanguineabout risk, their estimated RWAs will betoo low and their reported capital ratiosmisleadingly high.

The main obstacle to an agreement isthe committee’s proposal of an “outputfloor”—a lower bound for banks’ RWAs—calculated as a percentage of the figure

Bank capital

Polishing the floor

Supervisors put offfinalising reforms tothe Basel rules

52 Finance and economics The Economist January 7th 2017

2 churned out by a “standardised” method.The higher the percentage, the tighter thestandard: a first version of the proposalssuggested 60-90%; a failed compromiselast month proposed gradually raising it to75% over four years, starting in 2021.

American officials like the floor, believ-ing that it limits banks’ ability to playgames with the rules. European banks andofficials don’t. Both the Association of Ger-man Banks and the Bundesbank, forexam-ple, want no floor at all. They argue that in-ternal models make capital calculationsmore, not less, sensitive to risk.

America’s banks would be little affect-

ed; several European lenders could bestung. That is partly because America hasalready installed floors in its domesticrules—and, Americans would add, itsbanks shaped up faster after the crisis.Europeans retort that it also reflects trans-atlantic differences in business models.European lenders tend to keep more resi-dential mortgages on their books thanAmerican banks, which often sell them on;they also lend more to companies and forproject finance. All this may carry heavierrisk-weights under the revised rules.

Officials are still aiming for agreementin the first quarter of 2017. That probably

means fixing a floor, but how high? OmarKeenan and Kinner Lakhani, of DeutscheBank, estimate that a 75% floor would in-crease the RWAs (and hence reduce thecapital ratios) of 26 of the 34 listed Euro-pean banks they cover; at 60%, the numberdrops to ten, mainly in the Netherlandsand Nordic countries.

Phasing in the rules would give bankstime to adapt. Under the timetable envis-aged by the committee, they would haveuntil 2025—almost two decades after theworld’s financial system started to crack. Ifthe stand-off continues, the repairs willtake even longer to complete. 7

THANKS to Brexit and the election ofDonald Trump, 2016 is widely viewed

as a political turning-point. But it mayalso come to be seen asan economic turn-ing-point, marking the third big change ofdirection since the second world war.

The post-war period from 1945 to 1973was the era of the Bretton Woods systemof fixed exchange rates and capital con-trols. It was a time of rapid economicgrowth in the rich world as countries re-built themselves after the war and as thetechnological innovations of the first halfof the 20th century—cars, televisions, andso on—came into widespread use. Hightaxes reduced inequality; fiscal policywas used to control the economic cycle. Itall came crashing down in the early 1970sas the fixed-currency system collapsed,and an oil embargo imposed by Arab pro-ducers ushered in stagflation (ie, high un-employment combined with inflation).

By the early 1980s, a new system hademerged. Currencies floated, capital con-trols were abolished, the financial sectorwas liberalised, industry was privatisedand tax rates on higher incomes were cut.In this system inequality widened again(although economists still debate how toparcel out the blame between technologi-cal change and globalisation, as Chinaand other countries took a full part intrade). Growth was slower than in theBretton Woods era but inflation wasreined in. Monetary measures replacedfiscal ones as the main policy tool. Thisera suffered its defining crisis in 2007-08and has come to an end.

The final years of both periods weremarked by a degree of monetary experi-mentation. In the late 1970s many policy-makers were converted to the doctrine ofmonetarism—the idea thatbysetting a tar-get for the growth of the money supplygovernments could control inflation (and

that controlling inflation should be themain aim of their policies). But moneta-rism proved harder to implement than itsproponents thought; the monetary targetsbehaved unpredictably. By the mid-1980s,monetarism had been quietly dropped.

Since the 2008 crisis, monetary policyhas had to be rethought again, with centralbanks grappling with the “zero bound” forinterest rates. Their first move was to adoptquantitative easing, the purchase of assetsto drive down longer-term borrowingcosts. Some have since followed this upwith negative rates on bankreserves.

Financial-market trends have playedout against the backdrop of these two poli-cy eras. Equities did very well for 20 yearsunder the Bretton Woods regime, but start-ed to falter in the mid-1960s, well beforethe system’s collapse. Perhaps investors al-ready took fright at signs of inflation; bondyields had been trending upwards sincethe end of the second world war.

In the era ofglobalisation a great equitybull market began in 1982 but declined in2000-02 with the bursting of the dotcom

bubble. That was a portent of the biggercrisis of 2007-08. Both showed how in-vestors could be prey to “irrational exu-berance” and push asset prices to absurdlevels. Just as rising bond yields in the1960s presaged the inflationary battles ofthe 1970s, so falling bond yields in the1990s and 2000s foreshadowed today’sstruggles with deflation and slow growth.

Financial markets seem to expect thatpolitical turmoil will indeed lead to an-other change of economic regime. Sincethe American election the MSCI Worldequity index has rallied and the DowJones Industrial Average has hit recordhighs. Valuations reflect this optimism. Inthe early 1980s price-earnings ratios werein single digits. In contrast, the S&P 500now trades on an historic price-earningsratio of 25. Another contrast with the1980s is that, back then, short-term inter-est rates were at double-digit levels andequity valuations were able to climb asrates fell. That cannot happen now.

So what kind of economic regime areinvestors expecting? They seem to becherry-picking the best bits from the pre-vious two regimes—the tax cuts and de-regulation of the 1980s with an expecta-tion that (as under Bretton Woods) fiscal,rather than monetary, policy will be usedto smooth the upsand downsofthe cycle.

But the populist revolt is, in large part,a reaction against the free movement ofcapital and labour that has made so manyfinanciers rich. A much bleaker outcomeis possible, whereby rising nationalismleads to trade wars and an ageing work-force makes it impossible for the richworld to regain the growth rates of pastdecades. Change is coming. But ratherthan resembling the 1980s, the new re-gime could lookmore like the 1930s.

The third regime

Seismic shifts

Sources: Robert Shiller; The Economist

1945 60 70 80 90 2000 10 160

5

10

15

20

1945 60 70 80 90 2000 10 16

Real S&P composite-price indexLog scale

United States, long-term interest rates, %

10

50100

5001,000

5,000

BRETTON WOODSSYSTEM

GLOBALISEDSYSTEM

Buttonwood

The world is changing and investors maybe too optimistic about the results

Economist.com/blogs/buttonwood

The Economist January 7th 2017 Finance and economics 53

TUCKED away in a corner of Brixton, insouth London, a rainbow-coloured

ATM dispenses cash, looking for all theworld like any other. But the notes it spewsout are not pounds sterling. They are Brix-ton pounds (B£). Not to be mistaken for sil-ly Monopoly money, the Brixton poundcan actually be spent, legally: the currency,which has a fixed one-for-one exchangerate with sterling, is accepted at over 150 lo-cal shops and businesses. It can even beused to pay local taxes.

Launched in 2009, this is one of manysuch initiatives. Local currencies havebeen adopted in other towns and cities inBritain, such as Bristol, Exeter and Totnes.Elsewhere, examples include the eusko,used in the French Basques; BerkShares,used in western Massachusetts; and theIthaca Hour, in Ithaca, New York. Barcelo-na plans an experiment in 2017.

Such schemes aim to boost spending atlocal retailers and suppliers, by encourag-ing the recirculation of money within acommunity. Because the currency isworthless outside its defined geographicarea, holders spend it in the neighbour-hood, thus creating a “local multiplier ef-fect”. Backers of the schemes also claim en-vironmental benefits: stronger localbusinesses cut transport distances and car-bon emissions.

But local currencies have a poor record.Of over 80 launched in America since 1991,only a handful survive. Elsewhere, theGuardiagrele simec in Italy, the Torontodollar, the Stroud pound and others arelanguishing or are already defunct. Even

the Ithaca Hour, the most hyped “success”,has seen its circulation fall precipitouslyfrom two decades ago, says its founder,Paul Glover.

Local currencies face three hurdles.First, they are relatively illiquid, being ac-cepted only at willing local businesses.They are, in effect, a form of self-imposedeconomic sanction, narrowing the rangeof choice for consumers and businesses.Second, local-currency schemes sufferfrom a trust deficit: they are not backed bythe central bank, so holders do not want torisk having too much. Finally, having todeal with two parallel currencies imposestransaction costs—and those wanting tobacklocal businesses can easily use the na-tional currency.

All of which helps explain why local-currency circulation in most of theseplaces is very low. Just B£100,000($123,000) circulates, for example, in anarea of 300,000 people. That is too little tohave much of an economic impact oneway or another. The odd-looking notes,however, do make good souvenirs. 7

Sub-national currencies

Local difficulties

From Brixton to NewYork, localcurrencies struggle to survive

Five Bowies make a Winston

AS A new trading year began this week inthe art-deco tower that houses the Chi-

cago Board of Trade, big men were clus-tered around pitsdealing in futuresand op-tions tied to various commodities. Theirapproach dates back to the building’sopening in 1930, and was once familiar incities throughout America. But after de-cades of attrition, on December 30th theCME Group (named after the Chicago Mer-cantile Exchange) closed the “open outcry”trading pits that it operated in New York. InAmerica, Chicago’s hue and cry has be-come unique.

Even this exchange is a shadow of itsformer self. There are now nine pits, downfrom 32 in 2007. A once teeming tradingfloor was closed in 2015. Most activity inthe contracts still traded in the pits is elec-tronic. No one in the surviving CME pits inChicago seems worried by the New Yorkclosures. But they have a symbolic impact.The markets have long been a colourful,fractious component of America’s finan-cial architecture.

They have always lured the ambitious.Two alumni of New York’s commoditymarkets have joined the Trump adminis-tration. Gary Cohn parlayed a cab ride intoa job as a silver trader, into a position atGoldman Sachs, and, eventually, that

bank’s presidency. His new post is as MrTrump’s chief economic adviser. VincentViola swapped a job at an exchange for Vir-tu Financial, the electronic-trading firm hefounded that made him a fortune. He willbe nominated as army secretary.

Tales of failure as well as of successabound. A scandalous default in the pota-to market in the 1970s wiped out severalfirms. A failure to corner the silver marketin the 1980s led to the spectacular bank-ruptcy ofone ofAmerica’s richest families.The destruction of the World Trade Centrein 2001obliterated the floorused by four ofNew York’s commodity exchanges buteven before the flames were extinguishedthey were back to business, some in smalltemporary facilities like technological junkshops, knit together by familiar cries.

In the end it was not scandal or terro-rism that undermined open outcry; it wasefficiency. Computers turned out to bequicker, cheaperand more precise than hu-mans. Almost all the important contractsended up in the hands of the CME Group,which was first to realise that the most dy-namic business was not in traditional com-modities but in interest rates, stock indicesand currencies. The strong volume theseproducts provided enabled the CME tocreate economies of scale in clearing andtrading systems, and to scoop up other ex-changes as they faltered.

Bit by bit, the exchange has shed its real-estate assets. The Board of Trade buildingwas sold in 2012 and the equivalent NewYork facility in 2013. This contraction, how-ever, is far from reflecting the health ofoverall business—which is booming. In2016 Brexit, the American election and In-dia’s monetary experimentation, to citejust three examples, each created demandfor futures and options tied to interestrates, precious metals and currency. Tran-saction volume on the CME grew by12% toreach a new record. The markets are moreimportant than ever, even if, increasingly,they can be neither seen nor heard. 7

Futures and options trading

Out of the pits

CHICAGO

A long era of trading closes as theunderlying business gathers steam

Nothing to outcry about

54 Finance and economics The Economist January 7th 2017

“TIME is of the essence,” wrote Sir An-thony Atkinson, a British economist,

in a report on measuring global poverty,published in July 2016. His sense of urgen-cy may have been influenced by anotherconstraint. In 2014 Sir Anthony had beendiagnosed with incurable cancer. Somemight have paused; he sped up. He chairedthe World Bankcommission thatproducedthe poverty report, and wrote a book, “In-equality: WhatCan Be Done?”, in just threemonths. On January1st, his time ran out.

In his lifetime, he was tipped for a No-bel prize. On his death, fellow economistsrushed to describe him as “one of the all-time greats” and emphasised his extraordi-nary “decency, humanity and integrity”.The two were linked. For him, economicswas about improving people’s lives.

A six-month stint volunteering as anurse in a hospital in deprived inner-cityHamburg was an early influence. He sawpoverty, and went on to spend his life com-bating it. He fought his battles gently—shy-ing away from the adversarial style he ex-perienced as a student at Cambridge—butwith rigorous precision and an unfailingsense ofsocial justice.

As economists fell in love with marketsin the 1980s and 1990s, he wrote the besttextbook on their failures, with Joseph Sti-glitz, another economist. (Mr Stiglitz’sscrawl was some comfort to Sir Anthony,as evidence of handwriting even worsethan his own.) Faced with an imperfectworld, he showed how to achieve a sec-ond-best compromise.

The theoretical pontificating of 18th-and 19th-century political economists onwelfare and inequality had rather fallenout of fashion. Sir Anthony quickly identi-fied a big obstacle to getting the messageacross: a lack of good data. He poredthrough historical sources to unearth pasttrends in income inequality. He createddata sets on the highest incomes, findingsfrom which would support the slogans onprotesters’ placards. Sir Anthony was amentor and collaborator of Thomas Pi-ketty, famous for his book, “Capital in theTwenty-FirstCentury”. MrPikettysays thatall work on trends in income and wealthinequality stems from Sir Anthony’s.

In the course of his career, Sir Anthonycontributed to an average of nearly a booka year and sat on numerous governmentcommissions. The legacy of his most citedpaper, published in 1970, is an inequalityindex that bears his name. Existing mea-

sures, he showed, might seem like neutralindicators of the spread of incomes in acountry. In fact they contained implicit val-ue judgments. Some were more sensitiveto sagging incomes for the poorest; otherswould respond more to soaring incomes atthe top. Always constructive, he thencreated a new class of inequality mea-sures, making explicit what had been im-plicit. Today they are used by the US Cen-sus Bureau.

He went beyond analysing the world totrying to fix it—in ways that many rejected.

His faith in the power of government toright the world’s ills led to radical propos-als. His final bookon inequality argued fora participation income (a payment for allwho contribute to society) and a tax onwealth to finance an inheritance for every-one on reaching the age of 18. He pushedback against pressure to cut taxes and prio-ritise containing inflation over reducingunemployment. To the end, he was bat-tling lifelong challenges: inadequate data;how to harness government for good; andclosed minds. 7

Anthony Atkinson

For poorer, forricher

AnthonyAtkinson, a great Britisheconomist, died on January1st, aged 72

Insuring talent

Death Star

THE death ofCarrie Fisher, a much-loved actor in the “Star Wars” movies,

left a hole in the force for fans. It may alsoburn a hole in the pockets ofunderwrit-ers, syndicated under Lloyds ofLondon.They may have to forkout as much as$50m to meet Disney’s claim for its loss.The studio, which owns the sci-fi saga,had wisely taken out so-called contractu-al-protection insurance (CPI) in casedeath thwarted a contractual obligation:in Ms Fisher’s case to film and promotefuture “Star Wars” episodes.

Contrary to the headlines, 2016 wasnot an especially lethal year to be a celeb-rity. Like the rest ofus, they do die. Butunlike most ofus, their employers can beleft with astronomic bills. When PaulWalker, an actor in “The Fast and theFurious”, a series ofaction movies, diedin 2013 while filming the seventh in-stalment, Universal Pictures had to spendconsiderable effort (and dollars) to make

his on-screen persona live on. This in-cluded hiring body-doubles and digitallyinserting Mr Walker into the movie withhundreds ofcomputer-generated images.

Most workers are easier to replace.Employers can take out simple life insur-ance that pays a fixed lump sum. But thevalue ofa film star to a studio, or a strikerto a football club, is harder to calculate inadvance. It depends on all sorts of things,especially timing. This is where contin-gency insurance, such as CPI, comes in.Unlike a life policy, how much of the$50m Disney receives depends on how itnow calculates and justifies the lossescaused by Ms Fisher’s death. This couldinclude, for example, her role in boostingsales ofstorm-trooper figurines.

Insuring talent is becoming popularoutside Hollywood. The aptly namedExceptional RiskAdvisors, a companybased in New Jersey that reportedlybrokered the Fisher policy, also helpsinsure against the deaths ofhedge-fundmanagers, company executives andsports teams’ star players. Publishershave taken out CPI in case bestsellingauthors die with books half-written.

Jonathan Thomas, from Munich Re,who has written contingency policies forover 30 years, says they are “exactly whatLloyd’s is good at”. The greatest changehe has seen is in the sums involved. Butsome worry that underwriters are drop-ping their standards and taking on toomuch risk. This could well become aproblem if contingency insurance growsmuch larger. But today it is still tiny com-pared with life insurance.

With rockstars remaining on stageinto their dotage and long-running se-quels one of the surest ways to makemoney in Tinseltown, the risks of losing a“key human” (or on occasion animal) aregrowing. That creates business opportu-nities for insurers, so long as they remainprudent and don’t become star-struck.

When the famous die, it is increasingly costly for insurers

Carrie trade

The Economist January 7th 2017 Finance and economics 55

THE two sides of the Brexit debate do not agree on much, butthey agree on this: if Britain fails to reach a trade deal with the

EU it will have to revert to the “WTO option”. This involves trad-ing only under rules set by the World Trade Organisation. TheLeave camp is happy with this idea; Remainers less so. But theawkward truth is that the WTO option is not much of a fallback.Becoming an independent WTO member will be tortuous.

It is puzzling that Brexiteers, whose campaign was summedup as “Vote Leave, take back control”, seem happy with the WTOoption. The WTO is truly global, with only a handful ofcountriesoutside it (zealous as they are about sovereignty, Brexiteers do notwant to join the ranks ofTurkmenistan and Nauru). But forsakingone unelected, unaccountable bureaucracy in Brussels for anoth-er housed in a leafy district of Geneva seems perverse. WTOmembers are at the mercy of its “dispute-settlement” regime,which allows other countries to enforce penalties.

Inconsistency has its upside. Membership of the WTO ap-pears to be good for trade. Mosteconomists believe Britain’sover-all trade will suffer if Britain leaves the single market. But Brexi-teersargue that, outofthe EU’s clutches, Britain will be the WTO’sstar pupil, striking trade deals across the world. China’s explosiveexport growth after joining in 2001testifies to its potency.

However, there is a snag. Britain is already a member of theWTO, but operates through the EU. To become a fully indepen-dent member, Britain needs to have its own “schedules”, WTO-speakfor the lists of tariffs and quotas that it would apply to othercountries’ products. Alan Winters, of the UK Trade Policy Obser-vatory at the University of Sussex, says that, in theory, it wouldnot be too hard for Britain to acquire its own schedules. Anychange would require the acquiescence of other members. But,using a “rectification” procedure, the government would simplycut “EU” at the top of the page and paste in “UK” instead. Biggerchanges—say, raising tariffs on certain goods—might require amore ambitious“modification” and more thorough negotiations.

The most simple course, then, would seem to be for Britain tokeep its schedules as they are under the EU, including the “com-mon external tariff”, applied uniformly by EU members to im-ports from third countries. The government has recently hintedasmuch. Thisavoidsdiplomaticwrangling. But simply to readopt

EU-approved commitments hardly looks like “taking back con-trol”. It would also lead to other problems.

WTO trade agreements assume that the EU as it currentlystands is a coherent economic bloc. Trade in goods between the28 member states is pretty free. Multinationals, which need tomove components back and forth frequently between differentmember states, have set up supply chains accordingly. Brexitcomplicates this arrangement. If Britain kept the common exter-nal tariff in place, then it might also apply to a company movingcomponents between the EU and Britain. Such a firm could incurtariff charges each time a border is crossed. A WTO membermight kick up a fuss if, say, one of its car companies with produc-tion facilities in both Britain and the EU suddenly found it moreexpensive to assemble a model.

A related problem concerns the WTO’s “tariff-rate quotas”(TRQs). These allow a certain amount of a good to enter at acheaper tariff rate. The EU has almost100 of them. Peter Ungpha-korn, formerly of the WTO secretariat, uses the example of the“Hilton” beef quota (named after a hotel where the agreementwas reached) to illustrate how gnarly Brexit could be.

The EU’s current official quota on beef imports is about40,000 tonnes, charged 20% import duty, he reckons. Above thequota, the duty is much higher. Britain and the EU will need to di-vide those 40,000 tonnes. The EU might push Britain to take a bigshare, appeasing European beef producers. British farmerswould howl as low-tariff beef flooded in. The quotas might needto be increased because Britain-EU trade would now come underthem. Expect to hear more about TRQs in 2017. According to LuisGonzález García of Matrix Chambers, a legal-services firm, theyare likely to become “the most contentious issue” in Britain’s re-establishment of its status as an independent WTO member.

Least-favoured nationThe WTO will even shape the Brexit negotiations themselves. Inrecent weeks, the government has appeared keen to ensure that,even after Brexit, Britain’s big exporters will be able to sell freelyto the single market. It has mooted paying into the EU budget toguarantee access for the City of London’s financiers. It has as-sured Nissan, a carmaker, that it will not lose from Brexit. It hasstudiously refused to spell out the terms of this guarantee, ru-moured to entail as-yet-unspent regional-development funds.

WTO rules, however, make such industry-specific deals hard.If Britain were to agree bilaterally with the EU not to apply tariffson cars, the WTO’s “most-favoured nation” principle would forceit to offer tariff-free access to othercountries’ too, saysMrUngpha-korn. And free-trade deals are not supposed to cover just one ortwo goods, but “substantially all the trade” between the coun-tries involved. Meanwhile, channelling government money toboost exports is frowned on in Geneva.

Some of these problems are surmountable. The WTO is not aslegalistic as you might think, says Mr Winters; countries that stayin others’ good books find things easier. But so far, British politi-cians are also struggling on that front. Boris Johnson, the foreignsecretary, has irritated his counterparts with clownish com-ments. “We are pro-secco but by no means anti-pasto,” he recent-ly told the Sun, a newspaper, alluding to food imports from theEU. When the reality of Brexit dawns, Mr Johnson and his fellowBrexiteers will find no trade deal to be especially appetising. 7

The fallacy of the fallback

The “WTO option” forBrexit is farfrom straightforward

Free exchange

Economist.com/blogs/freeexchange

56 The Economist January 7th 2017

For daily analysis and debate on science andtechnology, visit

Economist.com/science

1

IN A former leatherworks just off EustonRoad in London, a hopeful firm is starting

up. BenevolentAI’s main room is large andopen-plan. In it, scientists and coders sitbusily on benches, plying their varioustrades. The firm’s star, though, has a priv-ate, temperature-controlled office. Thatstar is a powerful computer that runs thesoftware which sits at the heart of Bene-volentAI’s business. This software is an ar-tificial-intelligence system.

AI, as it is known for short, comes inseveral guises. But BenevolentAI’s versionof it is a form of machine learning that candraw inferences about what it has learned.In particular, it can process natural lan-guage and formulate new ideas from whatit reads. Its job is to sift through vast chemi-cal libraries, medical databases and con-ventionally presented scientific papers,looking for potential drug molecules.

Nor is BenevolentAI a one-off. Moreand more people and firms believe that AIis well placed to help unpick biology andadvance human health. Indeed, as ChrisBishop of Microsoft Research, in Cam-bridge, England, observes, one way ofthinkingabout livingorganisms is to recog-nise that they are, in essence, complex sys-tems which process information using acombination ofhardware and software.

That thought has consequences.Whether it is the new Chan Zuckerberg Ini-

turned quickly into understanding. Scientific output doubles every nine

years. And data are, increasingly, salami-sliced for publication, to lengthen re-searchers’ personal bibliographies. Thatmakes information hard to synthesise. Acentury ago someone could still, with ef-fort, be an expert in most fields of medi-cine. Today, as Niven Narain of BERGHealth, an AI and biotechnology firm inFramingham, Massachusetts, points out, itis not humanly possible to comprehend allthe various types ofdata.

This is where AI comes in. Not only canit “ingest” everything from papers to mo-lecular structures to genomic sequences toimages, it can also learn, make connectionsand form hypotheses. It can, in weeks, elu-cidate salient linksand offernewideas thatwould take lifetimes of human endeavourto come up with. It can also weigh up theevidence for its hypotheses in an even-handed manner. In this it is unlike humanbeings, who become unreasonably at-tached to their own theories and pursuethem doggedly. Such wasted effort besetsthe best ofpharmaceutical firms.

For example, Richard Mead, a neurosci-entist at the University of Sheffield, in Eng-land, says BenevolentAI has given himtwo ideas for drugs for ALS, a neurodegen-

tiative (CZI), from the founder of Facebookand his wife, or the biological subsidiariesbeing set up by firms such as Alphabet(Google’s parent company), IBM and Mi-crosoft, the new Big Idea in Silicon Valley isthat in the squidgy worlds of biology anddisease there are problems its software en-gineers can solve.

Drug moneyThe discovery of new drugs is an early testofthe beliefthatAI hasmuch to offerbiolo-gy and medicine. Pharmaceutical compa-nies are finding it increasingly difficult tomake headway in their search for novelproducts. The conventional approach is toscreen large numbers of molecules forsigns of pertinent biological effect, andthen winnow away the dross in a series ofmore and more expensive tests and trials,in the hope of coming up with a goldennugget at the end. This way ofdoing thingsis, however, declining in productivity andrising in cost.

One explanation suggested for whydrug discovery has become so hard is thatmost of the obvious useful molecules havebeen found. That leaves the obscure ones,which leads to long development periodsand high failure rates. In theory, growingknowledge of the basic science involvedought to help. The trouble is that too muchnew information is being produced to be

Medicine and computing

The shoulders of gAInts

Artificial intelligence mayhelp unpickthe complexity ofbiology

Science and technologyAlso in this section

57 Diagnosing illness by smell

58 A hurricane paradox

58 A tale of dinosaur eggs

The Richard Casement internship

We invite applications for the 2017 Richard Casementinternship. We are looking for a would-be journalist tospend three months of the summer working on thenewspaper in London, writing about science andtechnology. Applicants should write a letter introducingthemselves and an article of about 600 words that theythink would be suitable for publication in the Scienceand Technology section. They should be prepared tocome for an interview in London or New York. A stipendof £2,000 a month will be paid to the successfulcandidate. Applications must reach us by January 27th.These should be sent to: casement2017@economist.com

The Economist January 7th 2017 Science and technology 57

2 erative disease that he works on. Both mol-ecules remain confidential while their util-ity is being assessed. One is bang in themiddle of what he and his team are doingalready. To him, this confirmsthat the artifi-cial intelligence in question is generatinggood ideas. The other, though, is compli-cated and not obvious, but mechanistical-ly interesting. Without the AI to promptthem, it is something his team might haveignored—and that, he admits, might in turnbe a result of their bias.

For now, BenevolentAI is a small actorin the theatre ofbiology and artificial intel-ligence. But much larger firms are also in-volved. Watson, a computer system builtby IBM, is being applied in similar ways. Inparticular, IBM has gone into partnershipwith Pfizer, an American pharma com-pany, with the intention of acceleratingdrug discovery in immuno-oncology—apromising area of cancer therapy that en-courages the body’s own immune systemto fight tumours.

Artificial intelligence will also moveinto clinical care. Antonio Criminsi, who,like DrBishop, worksatMicrosoftResearchin Cambridge, observes that today the pro-cess of delineating the edges of tumours inimagesgenerated byMRI machinesand CTscans is done by hand. This is tedious andlong-winded (it can take up to four hours).AI can reduce the time taken to minutes, oreven seconds—and the results are com-pletely consistent, unlike those arrived atby human doctors.

Another example of AI’s move into theclinic is described in a recent paper in JAMA,an American medical journal. This papershowed that it ispossible to use AI to detectdiabetic retinopathyand macularoedema,two causes of blindness, in pictures of theretina. Enlitic, a new firm based in SanFrancisco, is using AI to make commercialsoftware that can assist clinical decisions,including a system that will screen chest X-rays for signs of disease. Your.MD, a firmbased in London, is using AI, via an app, tooffer diagnoses based on patients’ queriesabout symptoms. IBM is also, via Watson,involved in clinical work. It is able to sug-gest treatment plans for a number ofdiffer-ent cancers. All this has the potential totransform doctors’ abilities to screen forand diagnose disease.

The powerofnetworkingAnother important biological hurdle thatAI can help people surmount is complex-ity. Experimental science progresses byholding steady one variable at a time, anapproach that is not always easy whendealing with networks of genes, proteinsor other molecules. AI can handle thismore easily than human beings.

At BERG Health, the firm’s AI systemstarts by analysing tissue samples, geno-mics and other clinical data relevant to aparticular disease. It then tries to model

from this information the network of pro-tein interactions that underlie that disease.At that point human researchers interveneto test the model’s predictions in a real bio-logical system. One of the potential drugsBERG Health has discovered this way—fortopical squamous-cell carcinoma, a formof skin cancer—passed early trials for safe-ty and efficacy, and now awaits full-scaletesting. The company says it has others indevelopment.

For all the grand aspirations of the AIfolk, though, there are reasons for caution.Dr Mead warns: “I don’t think we are in astate to model even a single cell. The modelwe have is incomplete.” Actually, that in-completeness applies even to models ofsingle proteins, meaning that science is notyet good at predicting whether a particularmodification will make a molecule intend-ed to interact with a given protein a better

drug or not. Most known protein struc-tures have been worked out from crystal-lised versions of the molecule, held tightby networks of chemical bonds. In reality,proteins are flexible, but that is much hard-er to deal with.

More work at the molecular level istherefore needed before AI will be able tocrack open the inner workings of a cell.One of CZI’s first projects is generating justsuch basic data. That, in itself, is a massiveundertaking—but it isone which collabora-tion with artificial intelligence will alsospeed up. AI will nudge people to generatenew data and run particular experiments.Those people will then askthe AI to sift theresults and make connections. As IsaacNewton put it, “If I have seen further, it isby standing on the shoulders of giants.” Ifthe brains of those giants happen to bemade ofsilicon chips, so be it. 7

Olfactory medicine

Whiff of danger

ONE ofa doctor’s most valuable toolsis his nose. Since ancient times,

medics have relied on their sense ofsmell to help them workout what iswrong with their patients. Fruity odourson the breath, for example, let themmonitor the condition ofdiabetics. Foulones assist the diagnosis of respiratory-tract infections.

But doctors can, as it were, smell onlywhat they can smell—and many com-pounds characteristic ofdisease areodourless. To deal with this limitationHossam Haick, a chemical engineer atthe Technion Israel Institute ofTech-nology, in Haifa, has developed a devicewhich, he claims, can do work that thehuman nose cannot.

The idea behind Dr Haick’s inventionis not new. Many diagnostic “breathalys-ers” already exist, and sniffer dogs, too,can be trained to detect illnesses such ascancer. Most of these approaches,though, are disease-specific. Dr Haickwanted to generalise the process.

As he describes in ACS Nano, he andhis colleagues created an array ofelec-trodes made ofcarbon nanotubes (hol-low, cylindrical sheets ofcarbon atoms)and tiny particles ofgold. Each of thesehad one of20 organic films laid over it.Each film was sensitive to one ofa scoreofcompounds known to be found on thebreath ofpatients suffering from a rangeof17 illnesses, including Parkinson’sdisease, multiple sclerosis, bladder can-cer, pulmonary hypertension andCrohn’s disease. When a film reacted, itselectrical resistance changed in a predict-

able manner. The combined changesgenerated an electrical fingerprint that,the researchers hoped, would be diag-nostic of the disease a patient was suf-fering from.

To test their invention, Dr Haick andhis colleagues collected 2,808 breathsamples from 1,404 patients who weresuffering from at least one of the diseasesthey were looking at. Its success varied. Itcould distinguish between samples frompatients suffering from gastric cancer andbladder cancer only 64% of the time. Atdistinguishing lung cancer from head andneckcancer it was, though, 100% success-ful. Overall, it got things right 86% of thetime. Not perfect, then, but a useful aid toa doctor planning to conduct furtherinvestigations. And this is only a proto-type. Tweaked, its success rate would beexpected to improve.

A prototype device to detect the smell ofdisease

The nose knows

58 Science and technology The Economist January 7th 2017

Palaeontology

Cracking a puzzle

DID dinosaur eggs hatch quickly, likethose ofbirds (which are dinosaurs’

direct descendants), or slowly, like thoseofmodern reptiles (which are dinosaurs’collateral cousins)? That is the questionaddressed by Gregory Erickson ofFloridaState University and his colleagues in apaper just published in the Proceedings ofthe National Academy of Sciences. It ispertinent because it touches on the widermatter of just how “reptilian” the dino-saurs actually were. Researchers alreadyknow that many were warm-blooded,and that some had insulation in the formoffeathers, even though they could notfly. Fast-developing embryos would drivea further wedge between them and theirtruly reptilian kin.

To investigate, Dr Erickson looked attwo sets of fossilised dinosaur eggs. Thefirst, from a Mongolian nest (pictured),

was laid by Protoceratops andrewsi, asheep-sized creature that lived 70m yearsago. The second, from Canada, was laidby Hypacrosaurus stebingeri—a speciescontemporary with P. andrewsi that grewto something between the weights of arhinoceros and an elephant.

In each case the researchers used anX-ray scanner to examine the teeth ofembryos found inside the eggs. In cross-section, dinosaur teeth display growthrings, called von Ebner lines, that arereminiscent of the annual growth ringsofa tree trunk. In all living species whichhave von Ebner lines those lines repre-sent a day’s growth. It therefore seemsreasonable to believe that this was truefor dinosaurs as well.

Assuming also, as Dr Erickson and hiscolleagues did, that dinosaurs’ teethbegan to grow about halfway throughembryonic development (which is whena crocodile’s embryonic teeth first ap-pear), they conclude that the P. andrewsieggs they looked at were about 83 daysold, making that the lower bound of theirincubation period. This compares withthe 42 days an ostrich egg takes to in-cubate and the 200-plus days required bya Komodo dragon egg—both of theseanimals being, when adult, ofcompara-ble size to P. andrewsi.

The bigger eggs ofH. stebingeri need-ed, according to Dr Erickson’s calcula-tions, a minimum of171days incubation.Sadly, no egg-laying animal of its size isaround today for comparison. But projec-tions based on size and incubation-period data from modern birds andreptiles suggest171days is substantiallymore than would be expected if the eggsofH. stebingeri were developing in abirdlike way.

The truth, then, is that in this as inother matters, dinosaurs are less reptilianthan was once thought, but not as avianas some revisionists would like to be-lieve. A messy answer, perhaps. But, innature, things are not always clear-cut.

Howreptilian were dinosaureggs?

A dinosaur’s nest

IN 2015, a bit over two years after Hurri-cane Sandy hit his city, Bill de Blasio,

New York’s mayor, announced the cre-ation of a $3 billion restoration fund. Partof the money is intended to pay for seawalls that will help protect the place fromfuture storms.

Building such walls may be an evenmore timely move than Mr de Blasiothought when he made his announce-ment. As a paper just published in Natureexplains, for the past two decadesa naturalform of protection may have been shield-ing America’s Atlantic coast, stopping bigstormsarriving. Such protection, though, isunlikely to last forever. Mr de Blasio is thustaking the prudent course of mending theroofwhile the sun is shining.

The study in question was conductedby James Kossin of America’s NationalOceanic and Atmospheric Administration,using wind and ocean-temperature datacollected since 1947. In it, Dr Kossin showsthat the intensity of hurricanes whichmake landfall in the United States tends tobe lowest when the Atlantic’s storm-gener-ation system is at its most active.

In Dr Kossin’s view, the cause of this ap-parent paradox is that, when conditions inthe deep Atlantic conspire to produce themost hurricanes, precisely the oppositeconditions obtain along the Americancoast. Thatcreatesa bufferzone which low-ers the intensity of incoming storms beforethey make landfall. The agent responsiblefor this lowering of intensity is verticalwind shear—in other words, wind speedsand directions that vary greatly with alti-tude. Vertical wind shear removes energyfrom hurricanes by pulling heat and mois-ture out of a storm’s centre. When the At-lantic is in its hurricane-producing phase,with low wind shearand high surface tem-peratures in its central region, the partalong the American coast behaves in theopposite manner, with high wind shearand low surface temperatures that sapstorms’ energy.

The obverse is also true. When windshear and sea-surface temperatures keepthe Atlantic’s hurricane-generating regionquiet, as they did between 1970 and 1992,those storms which do appear are two tothree times more likely to intensify rapidly(defined as gaining 15 knots of wind speedin six hours) when they are near the coastthan is the case during active periods.

Not everyone agrees with Dr Kossin’sproposed mechanism. James Elsner, a ge-

ographer at Florida State University, sug-gests that the correlations between stormgeneration and storm strength at landfallwhich Dr Kossin observes could be ex-plained another way. The biggest stormstend to start out far from land rather thannear it, and during periods of high activityhurricanes are generated farther out in theAtlantic than happens during lulls. Thesedistant storms thus have more time to veernorth—pushed that way by the interaction

between Earth’s rotation and their own, aphenomenon called the Coriolis effect—and therefore avoid landfall altogether.

Whatever its physical explanation,though, the correlation looks secure. And,with the current period ofactive hurricaneformation now 24 years old, a lull, with ac-companyingsuperstorms, maynotbe longin coming. Time, perhaps, forothermayorsalongAmerica’sAtlanticcoast to follow Mrde Blasio’s example. 7

Atmospheric physics

The storm beforethe calm

Something is damping down cyclonesbefore theyreach the American coast

The Economist January 7th 2017 59

For daily analysis and debate on books, arts andculture, visit

Economist.com/culture

1

ONE explanation of Britain’s vote toquit the European Union last June is

that Eurosceptics worked towards it for de-cades. A young Daniel Hannan joinedtheir number in the early 1990s, first as astudent, later as a journalist and Tory MEP.In hisnewbookMrHannan dulyslams theEU’s erosion of national sovereignty andsupposed antipathy to free markets. His vi-sion isofa more liberal, open and less regu-lated Britain, trading freely around theglobe and no longerheld backby a bureau-cratic and stagnant EU.

Yet this differs sharply from the ideas ofother Brexiteers, such as Nigel Farage, theformer leader of the UK IndependenceParty. Because Mr Hannan has economicnous, he likes a Norwegian-style “softBrexit”, at least as a transition. Norway isoutside the EU but in its single market, so itaccepts most of its rules, freely admits EUmigrants and pays into its budget. Mr Far-age will have none of this: anything lessthan a “hard Brexit” that takes Britain outof the single market would betray voters.

This tension between hard and softBrexit is one reason why Theresa May’sTory government has remained so opaqueabout its goals. It was also evident duringthe campaign, as Owen Bennett’s bookshows. Indeed, the rival Brexiteers hatedeach other even more than they did theiropponents—or the EU. On one side stoodMr Farage and his millionaire backer, Ar-ron Banks (whose diary of the campaign is

draftofhistory. Itwill notgive Mr Cameronmuch satisfaction.

Partly because they expected to wineasily, as Harold Wilson did in 1975, MrCameron and the Remainers made tacticalmistakes. These included accepting a pre-vote period of official government “pur-dah”, constraining what it could publish;allowing cabinet ministers to back Leavewithout resigning; and avoiding direct“blue-on-blue” attacks on fellow Tories. MrCameron’s renegotiation ofBritain’s mem-bership in February was also successfullyportrayed by Leavers as trivial.

In the campaign itself, Mr Cameron’steam relied heavilyon whatbecame tarredas“ProjectFear”. Modelled on the defeat ofthe Scottish independence referendum inSeptember 2014, it stressed Brexit’s risks tothe economy. George Osborne, the chan-cellor, issued gloomy forecasts of lost in-come, output and jobs. Many domesticand international bodies were wheeledout to support such warnings, culminatingwith Barack Obama saying that Britainwould be at “the back of the queue” fortrade deals. There was little effort to put outa positive message about the EU or to de-fend immigration, Leavers’ key weapon.

The main Vote Leave campaign led byMatthew Elliott and Dominic Cummingswas more vigorous and more aggressivethan the Stronger In team led by Will Strawand Craig Oliver from 10 Downing Street.Downing Street also misjudged the moodof Tory MPs. It hoped gratitude for the 2015Tory election victory and respect for MrCameron’s leadership would reduce rebelnumbers to 50-60. But careful canvassingby Steve Baker, a Eurosceptic backbencher,pushed them up to over 140, including thecritical duo of Mr Gove and Mr Johnson.Letting the Remain campaign seem largelyTory-run was another error.

The Leavers made mistakes, too. They

tellingly called “The Bad Boys of Brexit”),bent on talking about immigration and lit-tle else. On the other, with Mr Hannan,were leading Tory MPs like Michael Goveand Boris Johnson, backed by UKIP’s onlyMP, Douglas Carswell, who played downimmigration and talked up global trade lib-eralisation instead.

Mr Bennett is good on the internecinewarfare among Brexiteers, but his booklacks the detailed reporting that is in TimShipman’s “All Out War”. MrShipman, po-litical editorof the Sunday Times, has inter-viewed almost everyone involved in thereferendum (though apparently not MrsMay’s predecessor, David Cameron, whois writing his own memoir). He has in a re-markably short time produced a story thatis thorough, comprehensive and utterlygripping. It is hard to imagine a better first

Britain and the European Union

Why Brexit won

The first crop ofBrexit books includes entries rich with detail and analysis

Books and artsAlso in this section

60 Johnson: Word of the year

61 Chinese economics

61 Nigerian fiction

62 High-end car parks

What Next: How to Get the Best fromBrexit. By Daniel Hannan. Head of Zeus;298 pages; £9.99. To be published inAmerica in February

The Brexit Club. By Owen Bennett.Biteback; 340 pages; £12.99

The Bad Boys of Brexit. By Arron Banks.Biteback; 338 pages; £18.99

All Out War. By Tim Shipman. WilliamCollins; 630 pages; £25

60 Books and arts The Economist January 7th 2017

2 failed to answer the economic argument,being reduced to Mr Gove’s notorious at-tackon “experts”. They did not set out clearalternatives to membership. Their internalsplits and focus on immigration oftenmade them seem nasty, a big worry whena Labour MP, Jo Cox, was brutally mur-dered in mid-June, justbefore the vote, byaman linked to the far right. By then manyLeavers thought they would lose.

That they won is down to three causesdeeper than Remainers’ tactical errors.One was the Labour Party leadership. Thearrival of Jeremy Corbyn, a far-left anti-EUfigure, in late 2015 made winning the refer-

endum harder. Although he nominallybacked Remain, he and his team often sab-otaged the Labour In campaign, for exam-ple refusing to use the word “united” to de-scribe Labour’s position or to share aplatform with former party leaders.

A second was the rising anti-elite, anti-London and anti-globalisation mood ofmany voters, especially in the Midlandsand north. Those who feel they were leftbehind after the financial crisis haveturned to populists in many countries (in-cluding to Donald Trump in America). Inthe Brexit referendum they voted in unex-pectedly large numbers, a big reason why

many pollsters got the result wrong.The third goes back to Mr Hannan and

his friends. For three decades British gov-ernments of both parties, egged on by ashrilly Eurosceptic press, did little but carpat Brussels. Mr Cameron’s delusion wasthat, having himself hinted that he mightcampaign to Leave, he could turn senti-ment round completely in just threemonths. Instead, his past stance made himseem unconvincingwhen he portrayed EUmembership as vital for Britain’s economyand security. This same legacy could nowmake it trickier for Mrs May to persuadevoters to accept a soft Brexit. 7

CHOOSING the “word of the year” canbe an unenlightening exercise. The

last several years have seen language ma-vens and dictionary publishers pick anemoji (the one meaning “crying withjoy”), “because” as a preposition (becauseteenagers), and “hashtag” (as in “I’m sohappy, hashtag irony,” to signal a hashtagin speech). Most are probably passingfads; a “word of the year” should ideallyboth summarise the feel of the 12 monthsand have a chance ofsurviving.

If recent years have offered slim pick-ings, that is certainly not the case of 2016.Last year gave the English language anunusually big crop. Take “adulting”, anunlikely verb used by younger millenni-als to describe the joys of paying rent andmaking it to work on time and sober.Memes circulate online with the likes ofapicture ofa puppy lyingpassed out on thefloor under the text “I Can’t Adult Today.Please Don’t Make Me Adult”. With slangrising and falling faster than ever before,though, it is anyone’s guess whetheradulting will survive as long as it takes forits users to become seasoned grown-ups.

The same short shelf-life might be re-served for “hygge”, a venerable Danishword for a kind of relaxed happiness, anda phenomenon that hit Britain’s publish-ing industry like the hammer of Thor in2016. No fewer than nine books on hyggewere released or planned. Danes areamused that Britons think its joys can befound in a book, as it has a lot more to dowith good company than things like thesocks and mulled wine touted on thesebooks’ covers. It is hard to imagine non-Danes still going on about hygge in 2026.

Many words do look more likely tosurvive. The Chinese do not actuallycurse you with “may you live in interest-ing times,” but 2016 certainly has been abit too interesting, its politics making a

markon the lexicon.First came “Brexit”, a strong runner for

word of the year. It isn’t the first portman-teau word with a country name and“-exit”—that was Greece’s possible exitfrom the euro, or “Grexit”—but it’s the onethat has actually happened, and its conse-quences will be around for a long time.Britain’s vote to leave the European Unionhas others talking of a potential “Frexit”,should Marine Le Pen become president ofFrance, or “Italeave”, if Italy should beforced out of the euro. The portmanteauthat spawned a thousand others, Brexithas also resulted in “Remoaners”, thosewho voted for Britain to remain in the EU,and who are still grousing about the result.

It was America’s turn to embrace leap-into-the-unknown populism with the elec-tion of Donald Trump in November. The“alt-right”, another newly prominentgroup, played a role in Mr Trump’s victory.After firing two more conventional cam-

paign managers, the candidate hiredSteve Bannon to run his election bid. MrBannon had been the chairman of Breit-bart, a website devoted to the worldviewof maverick conservatives who some-times call themselves “race realists”,while others call them “white national-ists”. Most reject labels like “white su-premacist” or the dreaded “racist”: whitenationalists merely say that whites, likeother peoples, should have their owncountries, for everyone’s good.

Many people voted for Mr Trump notbecause they thought he was a racist, butbecause they could believe anything theyliked about him and his opponent, Hilla-ry Clinton. It was the year of “fake news”,“viral” stories in that word’s original in-fectious-plague sense, convincing manyvoters thatMrsClinton had sold weaponsto Islamic State, or that Pope Francis hadendorsed Donald Trump.

Truly fake stories were relatively rare,though. The more worrying phenome-non was a general disappearance of theexpectation that politicians should evenbe expected to stick to the facts. So John-son’s word ofthe year is “post-truth”. Poli-ticians have always strayed from thetruth, but shame kept them in the generalpostcode. But in 2016 Pro-Brexit cam-paigners said falsely that Britain sent theEU £350m a week, successfully goadingthe Remain camp into debating the figureendlessly—and so keeping the topic in thepublic’s mind. Mr Trump, after a series ofmisogynist comments, said that nobodyin the world respected women more thanhe does. In 2016 the only rule was “any-thing goes, so long as it gets attention,”and the most audacious at following itwere the winners. Other campaignershave been watching and taking note, afrightening sign that “post-truth” may bearound for some time to come.

Word of the yearJohnson

The past12 months sawmany words enjoy a breakthrough. Unfortunatelymost of them are grim

The Economist January 7th 2017 Books and arts 61

1

IN 1985 James Tobin, a Nobel laureate ineconomics, delivered a talk at a confer-

ence in China. Mao had died less than a de-cade earlier and modern economic con-cepts, shorn of socialism, were stillunfamiliar to many in the country, includ-ing the interpreter on this occasion. Strug-gling to find the right words, she burst intotears. Two conference participants steppedaside after Tobin spoke and, on the spot,devised the Chinese term for “macroeco-nomic management”. Future interpreterswould have it easier.

Chinese officials and academics, espe-cially those with a reformist bent, wereacutely aware of their tenuous grasp oneconomics at the time. Five years earlier,Deng Xiaoping, the country’s paramountleader, had put it bluntly when meetingRobert McNamara, president of the WorldBank: “We have lost touch with the world.”

With China’s economic rise now intoits fourth decade, it is easy to forget howshaky its footing was at the start of its as-cent. It began not just in poverty, but besetby basic uncertainty about how to devel-op. There was even disagreement overwhether development, in so far as it en-tailed market forces, was the right goal.

The oft-told story is that the Commu-nist Party forged ahead with policy experi-ments—“crossing the riverbyfeeling for thestones”, as the Chinese reformers’ sayinggoes—and, little by little, found the ingredi-ents for growth. There is much truth to this.But the role ofWestern economists in help-ing shape that journey is missing. “Unlike-ly Partners” by Julian Gewirtz, a doctoralcandidate in Chinese history at Oxford(and an occasional reviewer for thesepages), fills that gap. It vividly brings to lifeChina’s economic debates from Mao’sdeath in 1976 until 1993, by which time thecountry’s direction was clearer.

The claim is not that Westerners wereresponsible for China’s development. Alarge constellation of Chinese reformersdeserves the credit for that. Indeed, one ofthe book’s virtues is that it puts the spot-light on Zhao Ziyang, the Communist Partychief who wound up under house arrestafter the 1989 Tiananmen protests. MrZhaohas been written out of official histories,but his consistent support for bold think-ing was critical to China’s success.

Nevertheless, to understand how Chi-

na found its way, it is also necessary to re-cognise the influence of foreign ideas. Insome cases the impact was immediate.The concept of special economic zones,which enabled coastal regions to flourish,began with a Chinese vice-premier’s trip towestern Europe in 1978, where he saw ex-port-processing zones.

More often, the impact was diffuse. Ac-ademics trained in Marxist economicslapped up translated versions of Westerntextbooks. American professors came forweeksata time to teach econometrics. Chi-nese institutions invited a succession ofWestern economists to give talks and thensifted through their ideas for those that

were actually relevant to China.The Chinese were most receptive to

economists who themselves hailed fromplanned economies and understood theirflaws but also knew that sudden changeswere impractical. Ota Sik, from Czechoslo-vakia, inspired a phased-in pricingstrategyin the early 1980s, whereby China gave en-terprises ever more control over settingprices. The biggest star was Janos Kornai, aHungarian economist who moved to Har-vard after writing a seminal book in whichhe identified shortage as the chronic pro-blem of socialism. What came to be called“Kornai fever” gripped the study of eco-nomics in China in the late 1980s, and his

Chinese economics

Western takeaway

Unlikely Partners: Chinese Reformers,Western Economists, and the Making ofGlobal China. By Julian Gewirtz. HarvardUniversity Press; 389 pages; $39.95. To bepublished in Britain on January 31st

Fiction

Crazy city

AT LEAST in their conception of theworld, there are two broad categories

ofNigerians—or for that matter Kenyans,Pakistanis, Chinese or anyone from thepoor world. The vast majority are thosefor whom national boundaries representinsurmountable barriers, for whom evena bus ride to the city seems an other-worldly journey. And then there arethose who flit between African andEuropean cities as easily as if they wereriding the Victoria line from Brixton toGreen Park. They are the lucky ones withconnections at embassies or stores ofcapital certified and triple-stamped bybankofficials, or, best ofall, the burgundypassports of the European Union.

Lagos, a sprawling shambles ofsome21m souls, has its fair share ofboth cate-gories, and they come crashing togetherin Chibundu Onuzo’s second novel,“Welcome to Lagos”.

Some welcome. It is hard to imagine amegacity less hospitable to newcomers.At every turn lurkscammers, thieves,crooked cops and rent-extracting gang-lords. Into this metropolis come Chikeand Yemi, two soldiers deserting theirposts in the Niger Delta after one toomany orders to shoot civilians and burndown villages. Along the way they meet,and become fellow travellers with, amotley crew ofrunaways: Fineboy, amilitant fleeing from the very same army;Isoken, a young girl near-raped by thosemilitants; and Oma, a housewife escap-ing her abusive husband. Clueless, practi-cally penniless and unaccustomed to thenasty ways of the big city, they find refugeunder a bridge until, one day, Fineboyfinds an abandoned flat to squat in.

From another world come Ahmed,the pampered, British-passport-holdingcrusader who returns to Lagos to start amuckraking newspaper after a decade asa banker in London, and ChiefSandayo,the minister ofeducation, on the runfrom Abuja with $10m in his suitcase.These worlds—rich and poor, urban andrural, privileged and powerless, Muslimand Christian, Igbo and Yoruba—collideto spectacular effect as their paths crossand power shifts hands in surprising andunexpected ways, and then does soagain, and again. It is an unlikely plot, butMs Onuzo pulls it off, revealing the faultlines in her country’s society—or indeedthose ofany half-formed democracy.Though drenched in Lagosian atmo-sphere, the bookwears its Nigerian set-ting lightly: it is clearly the work ofapan-African and an internationalist—andis all the better for it.

Welcome to Lagos. By Chibundu Onuzo.Faber and Faber; 358 pages; £12.99

Where paths cross

62 Books and arts The Economist January 7th 2017

2 booksold more than 100,000 copies.The World Bank also had a big hand in

China’s take-off. The bank has a taintedreputation from that era, when it was seenas pushing a “Washington consensus”agenda of liberalisation that harmed LatinAmerica. Much less attention is paid to itssubtler positions in China in the 1980s. Itcarried out two major studies of the econ-omy (the first of their kind), became Chi-na’s largest source of foreign capital and,responding to Chinese requests, providedreams ofuseful policy advice.

Mr Gewirtz’s book does not attempt toprovide a definitive account of China’seconomic rise. It dwells in the world ofideas, tracing the arc of debates. Little at-tention is paid to what was actually hap-

pening on factory floors or in farm fields.But it is still a gripping read, highlightingwhat was little short ofa revolution in Chi-na’s economic thought.

Reading the booktoday, it is tempting toconclude that China is ignoring a basic les-son from its success: that being open to for-eign ideas served it so well. Under Xi Jin-ping, officials rail against “Western values”.Yet there is also a less gloomy conclusion.China’s path has never been linear: re-formists and conservatives have constant-ly jostled for the upper hand. But voices foropenness have ultimately prevailed. Andthe gains that China has made in its under-standing of economics and, more funda-mentally, in the lives of its people will notbe easily undone. 7

CAR parks are rarely well-designed.Even more rarely do they amount to

“design”: something to enjoy on a purelyaesthetic level. However, in Miami Beach,Florida, the car park has become not just abuilding type that is visually pleasing, butsomething else entirely: a set piece that of-fers architects the chance to show off.

Perhaps because the city has expandedrapidly as a travel destination, its new ho-tels are invariably disappointing dumb cit-adels of glass and steel that dominate thecity’s charmingold art-deco lookapparent-ly on purpose. Most galleries and muse-ums are soulless, too, the glamorous ve-randa of the Pérez Art Museum not-withstanding. Miami is largely built onsand or swamp and has a high water-table,making subterranean parking expensive;building above ground is a better option.

The first building to turn this inconve-nience into a design opportunity was theevocatively titled Ballet Valet. Arquitecton-ica, a local firm, had established itself inthe 1980s with a series of brash, colourfulapartment blocks that were immediatelysnapped up as sets for “Miami Vice”, a tele-vision series, and “Scarface”, a Cubangangster epic. Asked in the following de-cade to create a car park that would addsomething to a block of boutique shops,Arquitectonica adapted its garish palate tothe more sensitive 1990s by wrapping thebuilding in a fibreglassmesh with an irriga-tion system, and filling it with indigenousclusias and sea lettuce, which ran riot.

Ballet Valet might have remained a one-off were it not for the arrival of Art Basel inMiami Beach. When one of the largest art

fairs in Europe was seeking to expand intoAmerica it made an inspired choice. Artwas popular there, both among the Ameri-can celebrity set, who had taken to MiamiBeach as a place to party, and among thewealthy Latin Americans who saw the cityas both theirhome and theirfinancial basein America. There were only a handful ofgalleries, however. Entering into the art-ledregeneration ofMiami Beach, the car parksare in many ways monuments to the suc-cess of that relationship, creating spacesthat enable commerce and art to exist sideby side.

Car parks put developers at the centreofupcomingareas. Herzogand de Meuron,

a Basel firm that also specialises in muse-ums, completed 1111Lincoln Road in 2010. Aziggurat of bare concrete linked by precipi-tous ramps, it provides accommodationfor a series of art-crowd-friendly shops onthe ground floor and a home for the devel-oper, Robert Wennett, on the roof. This gid-dy stack of concrete cards set a benchmarkfor audacity, its upper deckproviding stun-ning views and one of the most sought-after party spaces during Art Basel MiamiBeach. From this example, the high-end carparkbecame firmly established.

In November, as part ofa new six-blockdevelopment in the mid-Beach area, AlanFaena, an Argentine developer, revealedhis parking garage (pictured). It boasts aglazed side elevation that exposes the ro-botic car elevator, which installs and re-trieves cars from closely stacked shelves: apreparation for the dance performances inthe Faena Forum arts centre to which it isappended. The car park actually only pro-vides room for around 100 cars (thoughthere are 300 subterranean parking placesbeneath the development). Yet still Mr Fae-na felt that the development needed anabove-ground car park, to be “a state-ment”. He had his designed, like the adja-cent arts centre, by OMA, the fashionablefirm founded by Rem Koolhaas.

Soon the designer car park will breachthe borders of the Beach into the widermetropolitan area. Later this year in down-town Miami, Terence Riley, a former cura-tor at New York’s Museum of Modern Art,will open an 800-car garage that will beclad in a crazy collage of different façadesdesigned by five of the world’s trendiestpractices. Although Miami has no morecars per person than the rest of America, itis still hugely car-dependent. The competi-tion among developers to build the mostextravagantormost striking take on an oth-erwise dull building is typical of Miami’speculiarly intimate glamour. 7

Architecture

Pile ‘em in style

The most exciting architecture in Miami Beach is carparks

Starchitects and their car parkitecture

Statistics on 42 economies, plus acloser look at GDP forecasts

Economicdata

Economic data% change on year ago Budget Interest

Industrial Current-account balance balance rates, %Gross domestic product production Consumer prices Unemployment latest 12 % of GDP % of GDP 10-year gov't Currency units, per $latest qtr* 2016† latest latest 2016† rate, % months, $bn 2016† 2016† bonds, latest Jan 4th year ago

United States +1.7 Q3 +3.5 +1.6 -0.6 Nov +1.7 Nov +1.3 4.6 Nov -476.5 Q3 -2.6 -3.2 2.47 - -China +6.7 Q3 +7.4 +6.7 +6.2 Nov +2.3 Nov +2.0 4.0 Q3§ +264.6 Q3 +2.5 -3.8 2.93§§ 6.95 6.52Japan +1.1 Q3 +1.3 +0.7 +4.6 Nov +0.5 Nov -0.2 3.1 Nov +184.2 Oct +3.7 -5.6 0.04 117 119Britain +2.2 Q3 +2.3 +2.0 -1.2 Oct +1.2 Nov +0.6 4.8 Sep†† -138.1 Q3 -5.7 -3.7 1.27 0.81 0.68Canada +1.3 Q3 +3.5 +1.2 +1.6 Oct +1.2 Nov +1.5 6.8 Nov -53.6 Q3 -3.5 -2.5 1.71 1.33 1.39Euro area +1.7 Q3 +1.4 +1.6 +0.6 Oct +1.1 Dec +0.2 9.8 Oct +380.4 Oct +3.2 -1.8 0.27 0.95 0.93Austria +1.2 Q3 +2.4 +1.5 +0.2 Oct +1.3 Nov +1.1 5.9 Oct +8.0 Q3 +2.1 -1.4 0.50 0.95 0.93Belgium +1.3 Q3 +0.7 +1.2 +2.8 Oct +2.0 Dec +1.9 7.9 Oct +3.4 Sep +0.7 -2.8 0.47 0.95 0.93France +1.0 Q3 +1.0 +1.2 -1.8 Oct +0.6 Dec +0.3 9.7 Oct -40.0 Oct‡ -1.1 -3.3 0.78 0.95 0.93Germany +1.7 Q3 +0.8 +1.8 +1.2 Oct +1.7 Dec +0.4 6.0 Dec +296.2 Oct +8.8 +1.0 0.27 0.95 0.93Greece +1.6 Q3 +3.1 +0.4 +6.8 Oct -0.9 Nov nil 23.1 Sep -1.0 Oct -0.2 -5.6 6.72 0.95 0.93Italy +1.0 Q3 +1.0 +0.8 +1.3 Oct +0.5 Dec -0.1 11.6 Oct +49.5 Oct +2.4 -2.6 1.88 0.95 0.93Netherlands +2.4 Q3 +3.1 +2.0 +0.6 Oct +0.6 Nov +0.2 6.6 Nov +57.1 Q3 +8.5 -1.1 0.43 0.95 0.93Spain +3.2 Q3 +2.9 +3.2 -2.1 Oct +1.6 Dec -0.4 19.2 Oct +23.0 Oct +1.6 -4.6 1.43 0.95 0.93Czech Republic +1.6 Q3 +0.9 +2.4 -1.7 Oct +1.5 Nov +0.6 4.9 Nov§ +3.7 Q3 +1.5 nil 0.50 25.8 25.0Denmark +1.1 Q3 +1.5 +0.9 -0.3 Oct +0.4 Nov +0.3 4.2 Oct +23.2 Oct +5.9 -1.0 0.40 7.10 6.91Norway -0.9 Q3 -1.9 +0.6 nil Oct +3.5 Nov +3.5 4.8 Oct‡‡ +18.0 Q3 +4.4 +3.5 1.67 8.60 8.91Poland +2.0 Q3 +0.8 +2.6 +3.3 Nov +0.8 Dec -0.7 8.2 Nov§ -2.4 Oct -0.5 -2.7 3.71 4.17 3.98Russia -0.4 Q3 na -0.5 +2.6 Nov +5.4 Dec +7.0 5.4 Nov§ +29.0 Q3 +2.4 -3.7 8.45 60.6 73.0Sweden +2.8 Q3 +2.0 +3.1 -0.5 Oct +1.4 Nov +0.9 6.2 Nov§ +22.2 Q3 +5.0 -0.3 0.60 9.11 8.50Switzerland +1.3 Q3 +0.2 +1.4 +0.4 Q3 -0.3 Nov -0.4 3.3 Nov +68.2 Q3 +9.4 +0.2 -0.15 1.02 1.01Turkey -1.8 Q3 na +2.9 +0.2 Oct +8.5 Dec +7.8 11.3 Sep§ -33.8 Oct -4.8 -1.8 11.32 3.57 2.97Australia +1.8 Q3 -1.9 +2.9 -0.2 Q3 +1.3 Q3 +1.3 5.7 Nov -47.9 Q3 -3.5 -2.1 2.79 1.37 1.40Hong Kong +1.9 Q3 +2.5 +1.6 -0.1 Q3 +1.3 Nov +2.8 3.3 Nov‡‡ +13.3 Q3 +2.6 +0.6 1.89 7.76 7.75India +7.3 Q3 +8.3 +7.2 -1.9 Oct +3.6 Nov +4.9 5.0 2015 -11.1 Q3 -0.9 -3.8 6.37 68.1 66.6Indonesia +5.0 Q3 na +5.0 -2.7 Oct +3.0 Dec +3.5 5.6 Q3§ -19.2 Q3 -2.1 -2.6 7.85 13,440 13,918Malaysia +4.3 Q3 na +4.3 +4.2 Oct +1.8 Nov +1.9 3.5 Oct§ +5.6 Q3 +1.8 -3.4 4.26 4.50 4.35Pakistan +5.7 2016** na +5.7 +2.3 Oct +3.7 Dec +3.8 5.9 2015 -4.1 Q3 -0.9 -4.6 8.03††† 105 105Philippines +7.1 Q3 +4.9 +6.9 +8.3 Oct +2.6 Dec +1.8 4.7 Q4§ +3.1 Sep +0.9 -1.0 5.08 49.7 47.1Singapore +1.1 Q3 +9.1 +1.3 +11.9 Nov nil Nov -0.6 2.1 Q3 +63.0 Q3 +21.5 +0.7 2.56 1.44 1.43South Korea +2.6 Q3 +2.5 +2.7 +4.8 Nov +1.3 Dec +0.9 3.1 Nov§ +99.0 Nov +7.2 -1.3 2.10 1,206 1,188Taiwan +2.0 Q3 +3.9 +1.0 +8.8 Nov +1.7 Dec +1.3 3.8 Nov +74.7 Q3 +14.4 -0.5 1.13 32.2 33.0Thailand +3.2 Q3 +2.2 +3.2 +3.8 Nov +1.1 Dec +0.2 1.0 Nov§ +47.9 Q3 +11.8 -2.3 2.68 35.8 36.2Argentina -3.8 Q3 -0.9 -2.0 -2.5 Oct — *** — 8.5 Q3§ -15.7 Q3 -2.5 -5.3 na 16.1 13.1Brazil -2.9 Q3 -3.3 -3.4 -7.3 Oct +7.0 Nov +8.3 11.9 Nov§ -20.3 Nov -1.1 -6.3 11.31 3.22 4.04Chile +1.6 Q3 +2.5 +1.8 -1.4 Nov +2.9 Nov +3.7 6.2 Nov§‡‡ -4.8 Q3 -1.9 -2.7 4.26 672 717Colombia +1.2 Q3 +1.3 +1.8 +0.4 Oct +6.0 Nov +7.5 7.5 Nov§ -13.7 Q3 -5.1 -3.7 7.05 2,963 3,212Mexico +2.0 Q3 +4.0 +2.1 -1.4 Oct +3.3 Nov +2.8 3.6 Nov -30.6 Q3 -2.8 -3.0 7.65 21.4 17.4Venezuela -8.8 Q4~ -6.2 -13.7 na na +424 7.3 Apr§ -17.8 Q3~ -2.8 -24.3 10.43 10.0 6.31Egypt +4.5 Q2 na +4.3 -4.9 Oct +19.4 Nov +13.1 12.6 Q3§ -20.8 Q3 -7.0 -12.4 na 18.2 7.83Israel +5.1 Q3 +3.4 +3.3 -0.8 Oct -0.3 Nov -0.5 4.5 Nov +13.3 Q3 +2.8 -2.4 2.06 3.85 3.93Saudi Arabia +1.4 2016 na +1.1 na +2.3 Nov +3.8 5.6 2015 -46.8 Q3 -5.6 -11.7 na 3.75 3.75South Africa +0.7 Q3 +0.2 +0.4 -1.3 Oct +6.6 Nov +6.3 27.1 Q3§ -12.3 Q3 -4.0 -3.4 8.93 13.6 15.6Source: Haver Analytics. *% change on previous quarter, annual rate. †The Economist poll or Economist Intelligence Unit estimate/forecast. §Not seasonally adjusted. ‡New series. ~2014 **Year ending June. ††Latest 3 months. ‡‡3-month moving average. §§5-year yield. ***Official number not yet proved to be reliable; The State Street PriceStats Inflation Index, Nov 35.38%; year ago 25.30% †††Dollar-denominated bonds.

64 The Economist January 7th 2017Economic and financial indicators

The Economist January 7th 2017 Economic and financial indicators 65

Indicators for more countries and additionalseries, go to: Economist.com/indicators

Othermarkets

Other markets% change on

Dec 31st 2015Index one in local in $

Jan 4th week currency termsUnited States (S&P 500) 2,270.8 +0.9 +11.1 +11.1United States (NAScomp) 5,477.0 +0.7 +9.4 +9.4China (SSEB, $ terms) 344.8 +1.1 -13.5 -19.1Japan (Topix) 1,554.5 +1.2 +0.5 +3.1Europe (FTSEurofirst 300) 1,443.8 +1.0 +0.4 -3.1World, dev'd (MSCI) 1,774.0 +1.3 +6.7 +6.7Emerging markets (MSCI) 871.5 +2.4 +9.7 +9.7World, all (MSCI) 427.2 +1.4 +7.0 +7.0World bonds (Citigroup) 877.6 +0.2 +0.9 +0.9EMBI+ (JPMorgan) 776.0 +0.7 +10.2 +10.2Hedge funds (HFRX) 1,203.2§ -0.1 +2.5 +2.5Volatility, US (VIX) 11.9 +13.0 +18.2 (levels)CDSs, Eur (iTRAXX)† 67.9 -5.5 -12.0 -15.1CDSs, N Am (CDX)† 63.4 -6.6 -28.2 -28.2Carbon trading (EU ETS) € 5.7 -10.0 -31.6 -34.1Sources: Markit; Thomson Reuters. *Total return index. †Credit-default-swap spreads, basis points. §Dec 29th.

The Economist commodity-price index

The Economist commodity-price index2005=100 % change on Dec 20th Dec 27th Jan 3rd one one

2016 2016 2017* month year

Dollar indexAll items 142.0 140.8 141.9 -1.7 +13.8

Food 154.9 152.6 154.8 -1.2 +6.4

Industrials

All 128.6 128.5 128.5 -2.3 +24.8

Nfa† 136.5 136.8 138.1 +2.4 +27.9

Metals 125.2 124.9 124.4 -4.3 +23.4

Sterling indexAll items 209.0 208.9 210.8 +2.1 +36.2

Euro indexAll items 170.1 167.4 169.9 +1.6 +17.5

Gold$ per oz 1,131.6 1,133.5 1,156.1 -1.3 +7.3

West Texas Intermediate$ per barrel 51.9 52.0 52.3 +2.7 +45.9Sources: Bloomberg; CME Group; Cotlook; Darmenn & Curl; FT; ICCO;ICO; ISO; Live Rice Index; LME; NZ Wool Services; Thompson Lloyd & Ewart; Thomson Reuters; Urner Barry; WSJ. *Provisional †Non-food agriculturals.

Markets

Markets% change on

Dec 31st 2015Index one in local in $

Jan 4th week currency termsUnited States (DJIA) 19,942.2 +0.5 +14.4 +14.4China (SSEA) 3,307.5 +1.8 -10.7 -16.6Japan (Nikkei 225) 19,594.2 +1.0 +2.9 +5.6Britain (FTSE 100) 7,189.7 +1.2 +15.2 -3.9Canada (S&P TSX) 15,516.8 +1.0 +19.3 +24.6Euro area (FTSE Euro 100) 1,121.9 +1.1 +2.5 -1.1Euro area (EURO STOXX 50) 3,317.5 +1.2 +1.5 -2.1Austria (ATX) 2,682.6 +1.7 +11.9 +7.9Belgium (Bel 20) 3,665.7 +1.6 -0.9 -4.5France (CAC 40) 4,899.4 +1.1 +5.7 +1.9Germany (DAX)* 11,584.3 +1.0 +7.8 +4.0Greece (Athex Comp) 657.5 +3.4 +4.1 +0.4Italy (FTSE/MIB) 19,626.6 +2.0 -8.4 -11.6Netherlands (AEX) 487.6 +0.7 +10.4 +6.4Spain (Madrid SE) 956.1 +1.5 -0.9 -4.5Czech Republic (PX) 934.2 +1.2 -2.3 -5.8Denmark (OMXCB) 805.4 +1.0 -11.2 -14.0Hungary (BUX) 32,649.0 +1.9 +36.5 +35.0Norway (OSEAX) 772.6 +0.8 +19.1 +22.5Poland (WIG) 52,753.8 +2.8 +13.5 +7.5Russia (RTS, $ terms) 1,176.7 +3.4 +28.9 +55.4Sweden (OMXS30) 1,530.9 +0.2 +5.8 -2.0Switzerland (SMI) 8,354.8 +1.2 -5.3 -7.4Turkey (BIST) 76,143.6 -1.8 +6.2 -13.3Australia (All Ord.) 5,788.2 +1.0 +8.3 +8.4Hong Kong (Hang Seng) 22,134.5 +1.7 +1.0 +0.9India (BSE) 26,633.1 +1.6 +2.0 -0.9Indonesia (JSX) 5,301.2 +1.8 +15.4 +18.4Malaysia (KLSE) 1,647.5 +1.1 -2.7 -7.1Pakistan (KSE) 48,705.0 +2.7 +48.4 +48.3Singapore (STI) 2,921.3 +0.8 +1.3 -0.2South Korea (KOSPI) 2,045.6 +1.0 +4.3 +1.4Taiwan (TWI) 9,287.0 +0.9 +11.4 +13.5Thailand (SET) 1,563.6 +2.6 +21.4 +22.0Argentina (MERV) 18,143.1 +9.9 +55.4 +24.9Brazil (BVSP) 61,589.1 +3.0 +42.1 +74.4Chile (IGPA) 20,809.5 +1.4 +14.6 +20.9Colombia (IGBC) 10,288.4 +1.6 +20.4 +28.9Mexico (IPC) 46,587.7 +2.2 +8.4 -12.5Venezuela (IBC) 31,839.2 +4.7 +118 naEgypt (EGX 30) 12,608.4 +2.8 +80.0 -23.3Israel (TA-100) 1,287.8 -0.1 -2.1 -1.1Saudi Arabia (Tadawul) 7,198.1 -0.6 +4.1 +4.2South Africa (JSE AS) 50,760.2 +0.9 +0.1 +14.1

Indicators for more countries and additionalseries, go to: Economist.com/indicators

GDP forecasts

Source: Economist Intelligence Unit

2017, % change on a year earlier

WorstBest0 2 4 6 8 10

Myanmar

Ivory Coast

Bhutan

Laos

Cambodia

Tanzania

Ghana

India

Djibouti

Vietnam

6 4 2 0 2+–

Trinidad & Tobago

Ecuador

Azerbaijan

Chad

Syria

Timor-Leste

Puerto Rico

Equatorial Guinea

Libya

Venezuela

66 The Economist January 7th 2017

WHEN in 1965 Vera Rubin arrived for afour-day stint at “the monastery”, as

the Palomar Observatory, home of theworld’s largest telescope, was dubbed,there were no women’s lavatories. No fe-male astronomer had ever worked therebefore. How could they, when it wouldmean walking home late at night?

It had been the same thinking at highschool. When she told her revered scienceteacher of her scholarship to Vassar hesaid: “You should do OK as longas you stayaway from science.” She was the only as-tronomy major to graduate there in heryear. When in 1947 she requested a gradu-ate-school catalogue from Princeton, thedean told her not to bother: women werenot accepted for physics and astronomy.George Gamow, later her doctoral adviser,said she could not attend his lecture at theJohns Hopkins Applied Physics Lab “be-cause wives were not allowed”.

She was indeed a wife. She married—aged 19—Robert Rubin, a physicist whomshe followed to Cornell, sacrificing herplace at Harvard. He was, she said, hergreatest ally. Later, when she attendednight classes at Georgetown University, hedrove her there, eating his dinner in the caruntil he could drive her home, while herparents baby-sat. Still, she found raisingfour children “almost overwhelming”.

When she halted her academic career—theworst six months of her life—she wept ev-ery time the Astrophysical Journal arrivedin the house. But, working part-time, shemade sure to be home when the kids re-turned from school. She never inspectedtheir rooms, she said, and they grew upfine, all with PhDs in science or maths.

Her master’s thesis was, her Cornell su-pervisor said, worthy of being presentedto the American Astronomical Society. Butshe was about to give birth, so, he suggest-ed, he would present it—but in his name.

She refused. Her parents drove up fromWashington and took their 22-year-olddaughter, nursing her newborn, on a gruel-ling snowy trip from upstate New York toPhiladelphia . She addressed the roomfulof strangers for ten minutes about galaxyrotation, soaked up some patronising criti-cism and a smidgen ofpraise—and left.

Though rows were unpleasant, defeatwas worse. “Protest every all-male meet-ing, every all-male department, every all-male platform,” she advised. At Palomar,she made a ladies’ room by sticking ahandmade skirt sign on a men’s room door(she returned a year later: it was gone).

She’d never anticipated such problems.Her father encouraged her childhood habitof watching meteor showers, leaning outof her bedroom window and memorising

theirgeometry in order to lookthem up lat-er. He even helped her make her first tele-scope, from a cardboard tube; she had al-ready made her own kaleidoscope. Shehadn’t evermetan astronomer, but it neveroccurred to her that she couldn’t be one.

But her early research was largely ig-nored. In otherwork, male astronomers el-bowed her aside. Fed up, she looked for aproblem “that people would be interestedin, but not so interested in that anyonewould bother me before I was done.”

She found it. In the 1930s Fritz Zwicky,an idiosyncratic Swiss astrophysicist, hadsuggested that the brightly shining starsrepresented only a part of the cosmicwhole. There must also be “dark matter”,unseen but revealed indirectly by the ef-fects of its gravity. That conjecture lan-guished on the margins until Ms Rubin,working with her colleague Kent Ford, ex-amined the puzzle of galactic rotation. Spi-ral galaxies such as Andromeda, sheproved, were spinning so fast that theirouter stars should be flying away into thenever-never. They weren’t. So either Ein-stein was wrong about gravity, or gravita-tional pull from vast amounts of some-thing invisible—dark matter—was holdingthe stars together.

The discovery reshaped cosmology,though initially her colleagues embraced itunenthusiastically. Astronomers hadthought they were studying the whole uni-verse, not just a small luminous fraction ofit. New theories developed on what thematter might be—but its fugitive particlesescaped all direct detection.

Some are worried by the absence. MsRubin was unbothered. Astronomy, shereckoned, was “out of kindergarten, butonly in about the third grade”. Many of theuniverse’s deep mysteries remained to bediscovered by eye and brain, with all thejoy that involved.

Shining a lightThere were other scientific feats, too: in1992 she discovered NGC 4550, a galaxy inwhich half the stars orbit in one direction,mingled with half that head the other way.She won medals aplenty: the Gold Medalof Britain’s Royal Astronomical Society(last awarded to a woman in 1828) andAmerica’s National Medal of Science.Princeton, which had once shunned her,wasamongthe manyuniversities to awardher an honorary doctorate. She gave nota-ble commencement speeches.

The plaudits were pleasant, but num-bers mattered more: the greatest compli-ment would be ifastronomers years hencestill used her data, she insisted. She was aperennial favourite for a Nobel prize inphysics—only ever awarded to two wom-en. That call never came: like dark matter,her fans lamented, she was vitally impor-tant, but easy to overlook. 7

Dark star

Vera Rubin, an American astronomerwho established the existence ofdarkmatter,died on December25th, aged 88

Obituary Vera Rubin