Financial Contagion Seminar

1

Financial Contagion: Financial Contagion: What do we Mean?What do we Mean?What do we Know?What do we Know?

Mardi DungeyAustralian National

Universityand CERF

Demosthenes TambakisPembroke College

Cambridge Universityand CERF

2

Roadmap for seminarRoadmap for seminar

1. Introduction – Definitions of Contagion2. Why does Contagion arise: Theory3. Contagion Case Studies4. Empirical Evidence5. Generalisations by Asset Market6. Contagion in Developed and Developing Markets7. Summary and Policy Implications

3

1.Introduction and background1.Introduction and background

4

Background factorsBackground factors

• Financial crises seem to occur together• Observe big shifts in financial markets

– Large changes in exchange rates: SE Asian crises 1997-1998

– Large changes in equity prices (October 1987 DJIA crash, 2000 dot.com bubble burst)

– Shifts in bond markets: Russian crisis 1998, Brazil 1999

• Policy concern is that they occur across many countries – HOW?

5

Some important recent datesSome important recent dates

• Devaluation of Mexican peso – 20 Dec 1994• Devaluation of Thai baht – 2 July 1997• Russian default – 17 August 1998• LTCM recapitalisation begins – 23 Sept 1998• Hong Kong stock market crash – 28 Oct 1998• Brazil devaluation – 13 Jan 1999• Collapse of Argentine currency board – Dec 2001• Brazil – runup to presidential election – 2003

6

And some that didn’t seem to And some that didn’t seem to attract as much attentionattract as much attention

• US and EU dot.com collapse – April 2000• Brazilian election – October 2002• Turkey banking and currency crises – 2000

• What will be the final outcome for the current Argentine problems – seems no contagion

7

Thai baht against US dollar

0

10

20

30

40

50

60

Jan-97 Apr-97 Jul-97 Oct-97 Jan-98 Apr-98 Jul-98

8

Indonesian rupiah against US dollar

0

2000

4000

6000

8000

10000

12000

14000

16000

18000


9

Korean won against US dollar

0

500

1000

1500

2000

2500


10

East Asian currencies against USD

0

100

200

300

400

500

600

700

800

Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03

Indonesia rupiah

Thai baht

Korean won

index: Jan 1 1997 =100

11

East Asian equity indices 1990-2003

0

50

100

150

200

250

Jan-90 Jan-92 Jan-94 Jan-96 Jan-98 Jan-00 Jan-02

Thailand

Hong Kong

Indonesia

Index: Jan 1997=100

12

Returns: Hong Kong equity index 1996-1999

-8

-6

-4

-2

0

2

4

6

8

Jan-96 Jul-96 Jan-97 Jul-97 Jan-98 Jul-98 Jan-99 Jul-99

%

13

Returns: Indonesian equity index 1996-1999

-8

-6

-4

-2

0

2

4

6

8


14

Returns: Thai equity index 1996-1999

-8

-6

-4

-2

0

2

4

6

8


%

15

Bond spread: Russian sovereign - US Treasury

0

1000

2000

3000

4000

5000

6000

7000

8000

Feb-98 Apr-98 Jun-98 Aug-98 Oct-98 Dec-98

basis points

16

Bond spread: Bulgarian sovereign - US Treasury

0

500

1000

1500

2000

2500


basis points

17

Bond spread: Argentinian sovereign - US Treasury

0

200

400

600

800

1000

1200


basis points

18

Policy makers’ and market Policy makers’ and market participants’ viewsparticipants’ views

• “Malaysia is concerned that the risks of contagion from the Asian crisis have increased….” Mustapa Mohamed

Malaysian Finance Minister 4/10/98

• “It’s like there are two businesses here. The old business, which works fine under normal conditions, and this stand-by business, when the world goes mad.” Eric Rosenfeld of Long-Term Capital,

New York Times Magazine, January 24, 1999[1]

19

Why is contagion a ‘problem’?Why is contagion a ‘problem’?

• Contagion is seen as a feature of financial crises.

• Internationally diversified portfolios to protect against country risk.

• In times of financial crisis the relationships used to diversify break down through unanticipated shocks = CONTAGION

• How do we cope with this?

20

Defining contagionDefining contagion

• Myriad of definitions• Problems across theory and empirical work• Attempt to draw this together using the

World Bank’s definitions• First, taxonomies of transmission paths

during crises

21

Financial linkages

Flight to Safety

Cash in effects

Portfolio adjustment

Herd behavior

Shifts in Investor

sentiment

Wake up callDemonstration Effects

Heightened Awareness

Spillovers

Competitive Dynamics

Demand Effects

Competition Effects

Economic Linkages

IMF 99Goldstein 98Perry & Lederman 98

Lowell et al 98

22

Broad definition of contagion Broad definition of contagion (World Bank)(World Bank)

• “Contagion is the cross-country transmission of shocks or the general cross-country spillover effects”

• This is very broad. Includes fundamentals linkages such as due to trade, terms of trade effects, things which we can name.

• Most of the literature distinguishes ‘fundamental’ linkages from contagion.

• Eg Lowell, Neu and Tong (1998), Reside & Cochoco-Bastista (1999), Calvo and Reinhart (1996) ‘fundamentals-based’ contagion, Kaminskyand Reinhart (2000)

23

Restrictive Definition Restrictive Definition (World Bank)(World Bank)

• “Contagion is the transmission of shocks to other countries or the cross-country correlation, beyond any fundamental link among the countries and beyond common shocks.”

• Excludes herding behavior and so forth.• Fundamental links include:

– Financial - Real (- Political)more on these later

Eg. Eichengreen, Rose and Wyplosz (1995,1996)

24

Very Restrictive Definition Very Restrictive Definition (World Bank)(World Bank)

• “Contagion occurs when cross-country correlations increase during ‘crisis times’ relative to correlations during ‘tranquil times.’”

• This needs to control for general volatility rising during financial crises (Forbes and Rigobon (2002))

• The fundamental linkages are again not acknowledged• Only increases in correlation are recognized as contagion

25

Our preferred version Our preferred version –– akin to akin to the restrictive definitionthe restrictive definition

• The transmission of shocks beyond the fundamental linkages

• Other terms:‘unwarranted contagion’, ‘pure contagion’ • Closest in the empirical literature is Eichengreen, Wyplosz

and Rose (1995,96) and Pesaran and Pick (2003) who want to control for a large variety of fundamentals first.

• Measured contagion may be relative to the particular fundamentals chosen

26

2. Why does contagion arise? 2. Why does contagion arise? Theoretical modelsTheoretical models

27

MicroMicro--foundations of contagionfoundations of contagion

• The cross-section dimension as opposed to the time domain.

• Investors’ actions do not reveal their private information.

• Herds arise when Information gets trapped: underlying signals driving investment decisions are not revealed.– For example, when traded asset prices are not market-

determined.• Investors can then rationally decide to mimic the

behavior of others.

28

Rational herding Rational herding behaviorbehavior by by international investorsinternational investors

• Herding arises when there is incomplete information about a country’s fundamentals and investors are free to choose when they move.

• Different classes of investors may change positions at the same point in time:– Banks, corporates, multinationals, hedge funds

• The potential for destabilizing collective action by herding investors.

29

How to prevent herding?How to prevent herding?

• In a bad equilibrium, bank runs or speculative attacks on a currency can be unrelated to fundamentals.

• Therefore unpredictable! • The importance of enlarging the amount of public

information available.• Need to enhance the transparency of institutions,

objectives and governance (see also policy implications I).

30

International illiquidity International illiquidity and ‘sudden stops’and ‘sudden stops’

• Capital account reversals have become more severe for developing economies.

• The availability of a rescue package (country bailout option) can make the problem worse because of moral hazard.

• Reliance on short-term financing can lead to sharp real slowdown if capital inflows stop.

• International creditors covering losses in other markets can lead to contagion (portfolio links).

31

The international debate on the speed The international debate on the speed of capital account liberalization: Iof capital account liberalization: I

• Comparing the costs and benefits of capital market integration.

• Against the costs, arguments for less capital controls include: – Increase the overall availability of funds for financing

socially valuable projects– Promote transparency and accountability– Reduce moral hazard and liquidity problems– Improve the functioning of the financial system (though

not necessarily deepening it!)

32

The international debate on the speed The international debate on the speed of capital account liberalization: IIof capital account liberalization: II

• The conventional view: the benefits outweigh the costs (Rogoff (1999)).

• The recent microstructure view: more market interconnectedness is bad because it leads to cross-market hedging and contagion (Kodres and Pritsker (2002)).

• The middle way: phasing in of opening up capital short-term flows

33

The role of fundamentalsThe role of fundamentals

• Control variables – spillovers.• Fundamentals-based contagion – Calvo and

Reinhart (1996), Kaminsky and Reinhart (2000).

• Usually there is a unique equilibrium for each possible set of fundamentals.

34

The role of beliefsThe role of beliefs

• Control variables – expectations.• Hard to measure and even harder to manipulate.• Beliefs-based contagion – Calvo and Reinhart

(1996), Kaminsky and Reinhart (2000).• There can be multiple equilibria even with

complete and symmetric information if investors are sufficiently forward-looking (Jeanne and Masson (2000)).

35

3. Contagion Case Studies3. Contagion Case Studies

36

The European Monetary System’s ERM The European Monetary System’s ERM crises: 1992crises: 1992--9393

• Germany’s problems at the ‘center’ affected ‘periphery’ countries: UK, Italy, Spain, Portugal, Sweden, Finland, France.

• A case study for the self-fulfilling crisis view? (Contagion unrelated to fundamentals)

• Not really: in most crisis countries, high unemployment and interest rates were very undesirable, including politically. Also systemic banking sector problems in Scandinavia

• Need to distinguish the credibility of policies from the credibility of policymakers.

37

The Mexican peso crisis 1994The Mexican peso crisis 1994

1. Driven by fundamentals- role of weak banking, weak reserves

2. Evidence of contagion - coined the term ‘Tequila effect’- tests provide mixed results

3. Was it regional- largely confined to Latin America

4. Which asset markets were affected- currency and equities

38

The East Asian crises: 1997The East Asian crises: 1997--9898

1. Driven by fundamentals- terms of trade effects due to export competing nations

2. Evidence of contagion- mixed evidence from formal testing- much commentary says that Indonesia particularly was contagion

3. Was it regional- largely- relatively little spillover to developed markets

4. Which asset markets were affected?- questions as to whether the crisis started in the equity rather than currency market as commonly presumed

39

The Russian and LTCM crises: 1998The Russian and LTCM crises: 1998

1. Driven by fundamentals- liqudity crisis and credit crisis- promulgated by hedging

2. Evidence of contagion- were these crises connected by contagion

3. Was it regional- very widespread- Russia affects developing markets, LTCM affects developed- did Russian crisis prompt the Brazilian crisis

4. Which Asset Markets were affected- bonds, equities

40

Recent Latin American financial crisesRecent Latin American financial crises

1. Driven by fundamentals– Brazil 1999– Argentina 2001

2. “Twin crises”: spillover from currency to banking and vice versa

3. Little evidence of international contagion 4. Was it regional?5. Which financial markets were affected?

- primarily currency and bonds

41

4. Empirical Evidence on 4. Empirical Evidence on Financial ContagionFinancial Contagion

42

Methods of testing for contagionMethods of testing for contagion

A taxonomy loosely based on the World Bank’s classification:

• Unexpected shocks or news– Dungey et al (2002,2003), Favero and Giavazzi (2003)

• Correlation tests– Forbes and Rigobon (2002), Baig and Goldfajn (1999)

• Probability tests– Eichengreen, Rose and Wyplosz (1995), Kaminsky and Reinhart (2000)

• Extreme returns tests– Bae, Karolyi and Stultz (2003), Baur and Schulze (2002)

• Other tests– Glick and Rose (1999), Lowell, Neu and Tong (1998)

43

Empirical Tests for ContagionEmpirical Tests for Contagion

1. Contagion as ‘unexpected shocks’ or news– contagion arises because transmission arises over and

above the anticipated links– the reaction is beyond what could have been expected

beforehand– Sometimes links are so complex so as to behave as if

there is contagion (Kiyotaki and Moore (2002))

Egs. Dungey et al (2002,2003), Favero and Giavazzi (2002)

44

Empirical tests for contagionEmpirical tests for contagion

2. Correlation TestsContagion as a significant increase in the correlation between assets during a period of crisis, compared with a period of calm

Eg. Forbes and Rigobon (2002)

- consistent with World Bank’s ‘very restrictive’ definition

45


3. Probability Tests- if the probability of a domestic crisis is affected by the occurrence of a foreign crisis this is consistent with contagion

Eg. Eichengreen, Rose and Wyplosz (1995,1996)

46


4. Extreme Returns Tests- the transmission between asset markets is different in times of extreme returns (crisis times) from that of normal times

Eg. Bae, Karolyi and Stulz (2003)

47


5. Other Tests- encompassing spillovers (fundamental linkages)

Glick and Rose (1999) tradevan Rickjem and Weder (2001) financial links

- other thingsLowell, Neu and Tong (1998) fundamentals

48

Some key practical issuesSome key practical issues

• How to define the crisis sample period?– Practically either ad hoc or data driven

• How to define the threshold at which a crisis occurs?– Sample dependence

• How to deal with different time zones?• How to deal with missing observations?

49

5. Generalisations on contagion 5. Generalisations on contagion by Asset Marketsby Asset Markets

50

Foreign exchange marketsForeign exchange markets

1. Fundamentals- often in conjunction with a banking system crisis; exchange rate pressure often leads banking problems- a large devaluation is often a trigger – used as a critical date, eg float of Thai baht, devaluation of Mexican peso

2. Evidence of contagion3. Regional

- crises seem to spread across wide range of currencies, both spillovers and contagion

51

Equity marketsEquity markets

1. Fundamentals2. Evidence of Contagion

- Forbes and Rigobon result that ‘no contagion, only interdependence’- Other methods wide ranging evidence of contagion

3. Regional nature of crises and contagion- That developed markets act as a conduit for crises between developing regions

52

Fixed income marketsFixed income markets

1. Fundamentals2. Evidence of contagion

- Much more limited evidence, lack of data

3. Regional effects - seem less pronounced- Less evidence that developed markets act as

conduits

53

CrossCross--market studiesmarket studies

• No clear causation from one market to another

• Most work concentrated on geographical separation (Bayoumi et al (2003))

• Evidence not yet systematic enough to be sure

• Growing area of research, and certainly important for the policy agenda

54

Mexican and Argentine peso against Mexican and Argentine peso against the USD: 1994the USD: 1994--19951995

0

1

2

3

4

5

6

7

8

9

Jan-94

Mar-94

May-94

Jul-94

Sep-94

Nov-94

Jan-95

Mar-95

May-95

Jul-95

Sep-95

Nov-95

Mexican peso

Argentine peso

55

The equity indices of Mexico and The equity indices of Mexico and Argentina during the 1994Argentina during the 1994--1995 crisis1995 crisis

20

40

60

80

100

120

140

Jan-94

Mar-94

May-94

Jul-94

Sep-94

Nov-94

Jan-95

Mar-95

May-95

Jul-95

Sep-95

Nov-95

Index: 1994:1=100

Mexican equity

Argentinian equity index

56

-6

-4

-2

0

2

4

6

Jan-94

Mar-94

May-94

Jul-94

Sep-94

Nov-94

Jan-95

Mar-95

May-95

Jul-95

Sep-95

Nov-95

Mexican equity index - blueArgentinian equity index- red

Equity index returnsEquity index returns

57

Country 1

Equity market

Currency market

58

Country 1

Equity market

Currency market

59

60

61

62

63

64

• Bonds, equities and currencies

65

6. Contagion in Developing and 6. Contagion in Developing and Developed Financial MarketsDeveloped Financial Markets

66

Relating the stage of financial Relating the stage of financial market development and market development and

contagioncontagion• Developed markets seem less affected• Developing markets have largest contagion

effects• Regional nature of contagion and crises

usually involves regions of developing –and opening - financial markets (Latin America, Eastern Europe, East Asia)

67

The Russian and LTCM crisesThe Russian and LTCM crises

• Did – Russian crises mainly affect developing

markets and – LTCM mainly affect developed marketsAs claimed by BIS (1999) ??

68

Volatility Decomposition: Bond spreads in Volatility Decomposition: Bond spreads in Russian and LTCM crisesRussian and LTCM crises

80%

85%

90%

95%

100%Arg

ent

Brazil

Mexico Indo

Korea

Thail

and

Bulgari

aPola

ndRus

siaNeth

er UK US

World Idiosyncratic Regional Contagion from Russia Contagion from LTCM

69

0

20

40

60

80

100

120

140

160

180

200

US UKNeth

erlan

ds

Poland

Indon

esia

South

Korea

Thaila

nd

Mexico

Argenti

na

Contagion from U.S. Contagion from Russia

Contagion in basis points- the smaller contributions

70

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

Argentina Brazil Russia Bulgaria

Contagion from U.S. Contagion from Russia

Contagion in basis points - the larger contributions

71

7. Summary and Policy 7. Summary and Policy ImplicationsImplications

72

Summary of questions and Summary of questions and evidenceevidence

1. Contagion is an important problem• Statistically significant contagion occurs• It is not usually the dominant cause of volatility (cost-benefit

trade off required)2. Contagion is a regional issue

• Varies across crises (and asset markets)• Some evidence that developed markets operate as a conduit

between regions3. Asymmetry: developing countries are more

affected by contagion than developed countries• True in terms of the levels effect• Not clear in terms of proportionate effect on volatility

73

Policy Implications IPolicy Implications I

If contagion is statistically significant:

- Implications of responding – moral hazard- How can we improve the outcomes

- improved transparency (information)- improved fundamentals (policy formation)- improved public institutions (infrastructure,

bankruptcy laws)

74

Policy Implications IIPolicy Implications II

If contagion is regional:

- Argument for greater regional cooperation in terms of shared information - concern that the current focus of the international institutions does not adequately reflect regional concerns- disadvantage could be parochialism

75

Policy Implications IIIPolicy Implications III

If developed markets transmit crises between developing regions:

- Do developed markets then have some responsibility to the developing regions in helping to cope with this effect- Possibly transmitted through portfolio rebalancing effects- Repeated prisoner’s dilemma game: better outcome for all participants if they cooperate

76

Policy Implications IVPolicy Implications IV

If developed and developing markets are proportionally both affected:

- What is more important, the proportion or level of the effect?

If developing markets see the larger levels effects:

- developing markets provide profitable capital opportunities for capital, settling for lower global capital allocations ultimately means lower global growth

77

More resources:More resources:

• http://www.cerf.cam.ac.uk/links/index.php

Mardi Dungey Demosthenes [email protected] [email protected]

Documents

Financial Contagion Seminar