M

Ra

b

a

ARRAA

KFI

1

sap2k

gitoh

itovm

mobsu

8

1h

Journal of Financial Stability 12 (2014) 16–25

Contents lists available at ScienceDirect

Journal of Financial Stability

journal homepage: www.elsevier.com/locate/jfstabil

easuring the costs of short-termism

ichard Daviesa, Andrew G. Haldaneb, Mette Nielsenb, Silvia Pezzinib,∗

The Economist, United KingdomBank of England, United Kingdom

r t i c l e i n f o

rticle history:eceived 5 October 2012

a b s t r a c t

A potential cost of modern capital markets is short-termism, with agents in the financial intermediationchain weighing near-term outcomes too heavily at the expense of longer-term opportunities and thus

eceived in revised form 9 July 2013ccepted 9 July 2013vailable online 1 August 2013

eywords:inancial economics

forgoing valuable investment projects and potential output. This paper sets out an analytical frameworkand empirical estimates of the potential costs of short-termism arising from distortions to the cost ofcapital and investment intentions.

© 2013 Bank of England.. Published by Elsevier B.V. All rights reserved.

dhret

irabmaftott

waomcausing a violation of the dividend irrelevance hypothesis. Sec-

nvestment

. Introduction

Modern capital markets come with costs. As recent events havehown, the most visible and violent of those costs are experiencedt times of financial crisis. These costs, for example in foregone out-ut, have been extensively studied (for example, Hoggarth et al.,002). But there is a second potential cost of modern capital mar-ets – the costs of short-termism.

Although it has no off-the-shelf definition, short-termism isenerally taken to refer to the tendency of agents in the financialntermediation chain to weight too heavily near-term outcomes athe expense of longer-term opportunities (Haldane, 2011). This haspportunity costs, for example in foregone investment projects andence future output.

Unlike crises these opportunity costs are neither violent nor vis-ble. Rather, they are silent and invisible. Perhaps for that reason,here have been very few attempts to capture the potential costsf short-termism in quantitative terms. Nevertheless, existing sur-ey evidence is strongly suggestive of short-termist tendencies inodern capital markets.For example, a 2004 MORI survey of members of the Invest-

ent Managers Association (IMA) and the National Associationf Pension Funds (NAPF) found a third and two-thirds of mem-

ers respectively believed their investment mandates encouragedhort-termism. Poterba and Summers (1995) surveyed Chief Exec-tive Officers (CEOs) at Fortune-1000 firms and found that the

∗ Corresponding author at: Bank of England, Threadneedle Street, London EC2RAH, United Kingdom. Tel.: +44 2076014663.

E-mail address: silvia.pezzini@bankofengland.co.uk (S. Pezzini).

oe

F

W

572-3089/$ – see front matter © 2013 Bank of England.. Published by Elsevier B.V. All rittp://dx.doi.org/10.1016/j.jfs.2013.07.002

iscount rates applied to future cash-flows were around 12%, muchigher than either equity holders’ average rate of return or theeturn on debt. Based on a survey of over 400 executives, Grahamt al. (2005) found over 75% would give up a NPV-positive projecto smooth earnings.

Perhaps reflecting that, short-termism has a rising public pol-cy profile. In the UK, a government review of UK equity marketsecently found short-termism in equity markets caused by mis-ligned incentives in the investment chain.1 In America, bothusiness groups and think-tanks are concerned about investoryopia.2 And the European Commission, Financial Stability Board

nd Group of Thirty have all recently expressed concerns aboutactors hindering long-term investment strategies, including short-ermism.3 Given its rising public policy profile, the relative dearthf quantitative evidence on the scale and importance of short-ermism is an important gap. This paper aims to help fill some ofhat gap.

We make three specific contributions to the literature. First,e show that if investors discount future returns excessively,

manager seeking to maximise the value of the firm will pri-ritise near-term cash-flows over distant ones. Specifically, theanager will prioritise dividend distributions over reinvestment,

nd, we provide evidence that investors discount future returnsxcessively. Our estimates of investor discount rates in the US

1 See Kay (2012); industry responses to the report varied considerably seeinancial Times (2012).2 See Business Roundtable (2006) and Aspen Institute (2009, 2010).3 See European Commission (2011), Financial Stability Board (2013) and G30orking Group (2013).

ghts reserved.

nancia

abfitpsmof

tsfsmen

2

gveetiW

tbpiFpTsfp

igiitti

optttri

iaiotc

e

daiitcisf

(orcorpi

aoaaaa

3

wdusp

3

aotflv

N

Tipai

I

Tchoice variable is the timing of dividends. Dividend payments toinvestors can either be paid out as earnings become available at theend of each period, or cash can be held within the firm for payout

R. Davies et al. / Journal of Fi

nd UK suggest significant evidence of myopia, which appears toe increasing over time. Third, we show that ownership of therm matters: private firms (who would be unaffected by short-ermism on our definition) tend to invest more than equivalentublicly owned firms. Together, these findings suggest that thesehort-termist distortions can affect materially the rates of invest-ent by companies and the stock of capital – whether physical

r human. This would have important implications for countries’uture growth rates.

The paper proceeds as follows. Section 2 surveys the litera-ure on formal tests for short-termism and on the link betweenhort-termism and investment. Section 3 sets out an analyticalramework for capturing the potentially adverse consequences ofhort-termism, in particular for the cost of capital and for invest-ent intentions. Section 4 presents some empirical evidence on

ach of these short-termism channels. Section 5 concludes withext steps for policy and research on this topic.

. The short-termism literature

Formal, quantitative evidence on short-termism is thin on theround. An exception would be Miles (1993). Using an augmentedersion of a basic asset pricing framework, he finds evidence ofxcessive discounting of future cash-flows using company-levelquity price data from the UK. Similar approaches, applied to longerime-series across a range of countries, have reached broadly sim-lar conclusions (Cuthbertson et al., 1997; Black and Fraser, 2002).

e follow a similar approach to Miles (1993) in Section 3.There is also relatively little empirical evidence linking short-

ermism and investment. Asker et al. (2011, 2013) provide a testased on a panel of US companies. They find that firms whose sharerice (and by implication, investors) are very sensitive to earn-

ngs announcements tend to forgo good investment opportunities.irms that are held privately invest significantly more than similarublic firms and are more responsive to investment opportunities.ests of periods when firms move from private to public owner-hip confirm these results. The inference is that private firms do notace the same earnings-driven pressure to scrimp on investment asublically quoted firms.

Bushee (1998) identifies firms that fall short of last year’s earn-ngs, so that a slight boost to earnings would deliver earningsrowth. These firms might face strong incentives to reduce R&Dn order to achieve this. The author also identifies institutionalnvestors that are likely to be myopic, measuring this by momen-um trading and portfolio turnover. The finding confirms those inhe other empirical papers, with ownership of shares by myopicnstitutional investors increasing the prevalence of R&D cuts.

Bernstein (2012) considers patents as a measure of innovationutput. The author compares US firms which went from private toublic by listing on NASDAQ with similar ones which had startedhe process but did not complete it (instrumenting for the poten-ial bias inherent in withdrawing from listing). The author findshat, after going public, firms do not reduce the number of patentsegistered, but they do tend to reduce considerably a measure ofnnovation novelty (patent citations).

At a theoretical level, the possibility of short-termism amongnvestors is related to a broader literature on behavioural biasesnd hyperbolic discounting (Laibson, 1997). Hyperbolic discount-ng refers to the tendency to choose a ‘smaller and sooner’ rewardver a ‘larger and later’ reward, in a way that is not consistent over

ime. This provides one explanation for the excess sensitivity ofonsumption to shocks to current income.

Some theoretical papers link short-termism and investmentxplicitly. The literature relies on informational problems which

tc

l Stability 12 (2014) 16–25 17

ividends can help solve, but at the expense of investment. In Millernd Rock (1985), managers know the current state of earnings butnvestors do not. Dividends provide a signal about earnings thatnvestors can observe. This means the manager has the incentiveo surprise the market with high dividends, even if this meansutting investment. Investors understand this, and discount thesenflated dividend signals accordingly. In equilibrium there are nourprises, but dividends are higher and investment lower than withull information.

A different type of information asymmetry appears in Stein1989). In this model, investors base their valuation of the firmn expected future earnings. Future earnings are known to be cor-elated with current earnings. The manager understands this anduts investment to boost current earnings. This lifts expectationsf future earnings, increasing the firm’s share price.4 In equilib-ium, the manager’s signal has no effect on share prices, but arisoners’ dilemma means that this behaviour continues to reduce

nvestment.Investors might also be uncertain about the quality of the man-

ger, as in Narayanan (1985). In this model, shareholders cannotbserve the manager’s ability or the project that is selected. Profitsre observable and boost the investor’s perception of managerialbility, which translates into higher wages. Knowing this, the man-ger may select a project that yields short-term profits, even if therere better long-term projects available.

. Theory

In this section we use a forward-looking asset pricing frame-ork to show how our definition of short-termism – excessiscounting on the part of investors – might affect project val-ation and selection. This framework also shows how investorhort-termism may cause a firm’s manager, acting rationally, torioritise dividends over investment.

.1. The trade-off between investing and distributing dividends

We start with a simple model in which two agents – an investornd a manager – face investment decisions. An investor’s valuationf a project (either a new firm or an expansion of an existing firm)hat operates for n periods is equal to the present value of the cashows or dividends in each period (Di), plus the discounted terminalalue (Pn) minus up-front investment costs (C).5

PV =n∑

i=1

Di

(1 + r)i+ Pn

(1 + r)n − C (1)

he investor’s decision is, in this simple model, all-or-nothing. Thenvestor follows a simple rule, choosing to invest in all positive NPVrojects and rejecting all negative NPV projects. The investor’s NPVssessment will therefore determine whether he or she invests,ncurring cost C, or walks away from the investment.

={

C, NPV > 0

0, NPV ≤ 0(2)

he manager of the project seeks to maximise its NPV. The only

4 The model used is one of “signal-jamming” – i.e. firms engage in costly behaviouro prevent information from appearing, rather than investing in a costly signal thatonveys information.

5 For more detail see, for example, Brealey et al. (2010).

1 nancia

wt(

i“oTt(ewnDo

trff

N

Tpiat(ino

Sddi

3t

bdimdztd

N

ta

(i

Wti

iA

N

Ihanpmp

3

t

N

Ai

Td(ativalado

ptrmrmprti

8 R. Davies et al. / Journal of Fi

hen the firm is wound up in year n. We assume cash held withinhe firm funds its research programme, which yields a return of1 + r), the risk-free rate.6

Under certain conditions, the timing of dividends will have nonfluence on the value of the firm (Modigliani and Miller, 1958). Thisdividend irrelevance” will occur when any change in Di is perfectlyffset by a counterbalancing change in Pn (in present value terms).o take a simple example of dividend irrelevance, consider a firmhat cuts its year 1 dividend from the maximum possible amountD1) to zero. The investor loses D1 in year 1. The cash funds research,arning the risk-free rate (1 + r), and is paid out when the firm isound up in year n, so that the investor gains D1(1 + r)n−1 in year

. In present value terms, the NPV loss is D1/(1 + r), the NPV gain is1(1 + r)n−1/(1 + r)n = D1/(1 + r) and the overall change in the valuef the firm is zero.

More generally, the firm chooses to pay a fraction �i ∈ [0, 1] ofhe maximum possible dividend, Di, to the investor in period i. Theemainder (1 − �i) is held within the firm, earning the risk-free rate,or payout when the firm is wound up in period n. This yields theollowing payoffs:

PV =n∑

i=1

Di�i

(1 + r)i+

n∑i=1

Di(1 − �i)(1 + r)n−i

(1 + r)n + Pn

(1 + r)n –C (3)

he first sum in Eq. (3) is the present value of dividends that areaid out immediately as cash becomes available. The second sum

s the present value of the cash that is not paid out as it becomesvailable, but is reinvested in the firm, used to fund research, andhen paid out at the wind-up date n. The third and fourth termsthe terminal value of the firm and the initial investment cost) arenvariant to the manager’s dividend timing decisions, �i. We canow assess how the manager’s decision to alter �i affects the valuef the firm by computing the derivative.

∂NPV

∂�i= Di

(1 + r)i− Di(1 + r)n−i

(1 + r)n = 0 (4)

ince this applies in all periods (i.e. ∀i ∈ 1, . . ., n), dividend timingoes not alter the value of the firm. So the managers’ reinvestmentecision does not influence either firm value or the investor’s initial

nvestment decision.

.2. How excess discounting can alter the dividend-investmentrade-off

Dividend irrelevance may not hold in reality. One reason coulde faulty discounting or short-termism. Since there are variousefinitions of short-termism, we begin by using an example to

llustrate our definition using the model above.7 Consider an invest-ent project costing $60. This investment is riskless and pays a $10

ividend at the end of each of 10 years. Its terminal value, Pn, isero. Plugging these values into Eq. (1) above, we obtain an equa-ion for the present value of the project as the sum of the discounted

ividends.

PV = $10(1 + r)

+ $10

(1 + r)2+ · · · + $10

(1 + r)10− $60 (5)

6 In reality a firm would want its R&D programme to beat the risk-free rate, andhe returns from research would be uncertain. Adding these extensions does notlter the conclusions here.7 Specifically, we are not assuming any managerial myopia, and unlike Stein

1989) and Miller and Rock (1985), there is no asymmetric information or signallingn this model.

cica

ii

m

l Stability 12 (2014) 16–25

ith a discount rate of 9%, the project’s cash-flows are worth $65oday, so its NPV is $5. In other words, this is a project worth invest-ng in.

Our definition of short-termism translates into excess discount-ng on the part of the investor and is captured by the parameter x.mending Eq. (5) to include our short-termism measure

PV = $10x

(1 + r)+ $10x2

(1 + r)2+ · · · + $10x10

(1 + r)10− $60 (6)

f x is less that unity, the project’s dividends are discounted tooeavily. For example, if x = 0.95 so that one-period-ahead dividendsre undervalued by 5%, discounted dividends are now worth $52,ot $65, so that the NPV of the project falls from $5 to −$8.8 Theroject is no longer investable. In other words, modest degrees ofyopia can flip the NPV, and the initial investment decision, from

ositive to negative.

.3. Adding a short-termism measure to the valuation model

We now generalise this idea by adding the myopia parameter xo Eq. (3). The equation for the NPV of the firm becomes:

PV =n∑

i=1

Di�ixi

(1 + r)i+

n∑i=1

Di(1 + r)n−i(1 − �i)xn

(1 + r)n + Pnxn

(1 + r)n –C (7)

nd the payoff from altering the fraction of cash paid out in period becomes:

∂NPV

∂�i= Dix

i

(1 + r)i− Di(1 + r)n−ixn

(1 + r)n = Di

(1 + r)i(xi − xn) (8)

here are three possible outcomes, depending on the value of x. Ifiscounting is rational then x = 1. This means that (xi − xn) = 0 and∂NPV)/(∂�i) = 0. In this case the investor is indifferent to the man-ger’s dividend decision and so dividend policy has no impact onhe value of the firm. The second case is more interesting. If thenvestor is myopic, x < 1 so that (xi − xn) > 0 and (∂NPV)/(∂�i) > 0. Thealue of the firm is then boosted by distributing more dividends,lthough the firm would then be investing at a sub-optimally lowevel. Third, if the investor is long-sighted, then x > 1 so (xi − xn) < 0nd (∂NPV)/(∂�i) < 0. In this case, the investor values future divi-ends more highly and the firm should delay dividend payment inrder to maximise its value.

In the myopic case (x < 1), the investor prefers dividends to beaid out today, rather than allowing the firm’s manager to holdhem for payout tomorrow. This means increasing � in any periodaises the firm’s value. It has another implication too. Consider aanager that is considering raising � by cutting the firms’ current

esearch budget to pay a higher dividend in just one period. Theanager considers two periods j and k, where j < k. If the maximum

ossible dividend payment is constant over time (Dj = Dk), then theatio of the benefits from cutting early compared to those from cut-ing late is given by (xj − xn)/(xk − xn) > 1. That is, the benefits fromncreasing � diminish with time, so the trade-off between holdingash within the firm (to support research, say) versus paying outs most acute in early periods. That is, early stage research is mostostly in terms of lost NPV and the manager should pay dividendss soon as possible in order to maximise the firm’s value.

This simple model explains how we define short-termism andts relationship to discounting. It also shows how discounting cannfluence a firm’s decision between paying dividends and keeping

8 A $10 return received at the end of year 5 should be worth $6.65 today. Withyopia, it is worth $5.14.

nancia

cgitm

3

ftwtwp

E

Ta

R

Aeei

P

Sdat(

P

Tsu

E

E

Espfcc

P

SecE

Ten

4

titrt

tteftf

4

Uscas

ipist

bb

�

wfiiTofi

cRohf

i

R. Davies et al. / Journal of Fi

ash within the firm to finance investment. The simple model sug-ests that if investors are myopic, dividend timing becomes verymportant. A manager acting entirely rationally may commit rela-ively less cash (and time and effort) to investment and relatively

ore to maintaining a constant stream of dividends.

.4. A more formal model with testable implications

In order to test whether investors are myopic, we use a moreormal model of firm valuation. Finance theory typically assumeshat investors care about both the level and uncertainty of theirealth and are risk averse.9 In this world, agents require a premium

o invest in a company. More formally, the expected return can beritten as the sum of the risk-free rate and a company-specific riskremium for company j10:

t(Rjt) = Rft + �jt (9)

he actual return on an investment is the sum of the capital gainnd the dividend yield:

jt = Pjt+1 − Pjt

Pjt+ Djt+1

Pjt(10)

ssuming an efficient market, actual returns differ only fromxpected returns due to a forecast error which is uncorrelated withxpected returns.11 Using this assumption, we can substitute (9)nto (10) to give an equation for the equity price.

jt = Et(Pjt+1 + Djt+1)Rft + �jt + 1

(11)

o the price of the security is simply the expected price and divi-end in the next period, discounted by the sum of the risk-free ratend the company-specific risk premium. By repeated substitution,his asset pricing equation can be written as a generalised form of1):

jt =N∑

i=1

Et(Djt+i)

(1 + rt1,t+i + �jt)i+ Et(Pjt+N)

(1 + rt1,t+N + �jt)N

(12)

he current share price is a function of future discounted dividendtreams and a discounted terminal share price, where we havesed:

t(�jt+k) = �jt, ∀k (13)

t(Rft+k) = rt1,t+k, ∀k (14)

q. (13) is very close to Eq. (9). It says that the expected company-pecific risk premium is constant and pre-determined based oneriod t information. Eq. (14) says that expectations of future risk-ree rates are defined by the path of the risk-free forward rateurve observed at time t. Eq. (12) can be modified with a myopiaoefficient, x.

jt =N∑

i=1

Et(Djt+i)xi

(1 + rt1,t+i + �jt)i+ Et(Pjt+N)xN

(1 + rt1,t+N + �jt)N

(15)

9 For background on the mean-variance approach, see Hillier et al. (2008).10 This is the case with the capital asset pricing model (CAPM) (Lintner, 1965;harpe, 1964) and arbitrage pricing theory (Ross, 1976). Under the CAPM, forxample, the company specific risk premium is equal to the company spe-ific beta multiplied by the market risk premium �jt = ˇjt(Rmt − Rjt). This meanst(Rjt) = Rjt + ˇjt(Rmt − Rjt).11 That is, we assume that Rjt = Et(Rjt) + εjt .

befia

d

tg

l Stability 12 (2014) 16–25 19

he null hypothesis – no short-termism – implies x = 1. Drawing onvidence across time and industrial sectors, it is this restriction weow test.

. Empirical evidence of short-termism

A testable hypothesis from the models presented is that short-ermism may have dented the willingness of firms to engage innvestment spending. There are two related but distinct channelshrough which this might occur. The first is that short-termismaises to sub-optimally high levels the marginal cost of new capitalo finance projects. This is the channel explored in Section 4.1.

The second channel is that short-termism induces firms to dis-ribute a sub-optimally high share of their revenues and profitso shareholders in the form of dividends. This would come at thexpense of ploughing back profits into the business to financeuture investment growth opportunities. Section 4.2 provides illus-rative estimates of the potential cost of short-termism in terms oforgone investment.

.1. The impact of short-termism on investors’ discount rates

The data set we use comprises a panel of 624 firms listed on theK FTSE and US S&P indices over the period 1980–2009.12 These

pan a broad range of industrial sectors, as shown in Table 1. Theore inputs to the analysis are firm-level measures of dividendsnd equity prices. The average dividend-price ratio in each industryegment is shown in Table 2.

The mean dividend-price ratio across the panel is 2.6%. But theres a fairly significant degree of cross-sectoral and time-series dis-ersion. For example, dividend-price ratios are almost twice as high

n the energy and utilities sector as the health and pharmaceuticalsector. And mean dividend-price ratios were two-thirds higher inhe 1990s compared to the 1980s.

Following Miles (1993), the company risk premium is modelledased on firm-specific characteristics, in particular the companyeta and the level of gearing Z:

jt = ˛1ˇjt + ˛2Zjt (16)

here Z = D/E.13 Betas are estimated using daily return data forrms listed on the S&P 500 and FTSE, together with daily data for the

ndices themselves.14 Mean estimated betas are shown in Table 3.hese average below one for both UK and US firms. The distributionf betas is fairly wide, with over a third of US firms and almost afth of UK firms having a beta in excess of unity.

The second component of the firm-specific discount factor isompany gearing. This was constructed using annual Thomsoneuters Datastream data for book value per share, the numberf shares outstanding and debt outstanding. Other things equal,igher gearing would suggest a higher company-specific discount

actor.15

The final element in the firm-level discount factor calculations the risk-free rate. The yield on government securities was used,ased on data from the Federal Reserve and Bank of England. Having

stimated (15) using pooled US and UK regressions over 20 years, arm-specific risk premium can be calculated. The average premiumcross the sample is 5.9%.

12 Data are from Thomson Reuters Datastream.13 Because a firm’s beta ought also to be a function of its business and financingecisions, we also estimate a restricted version of (16): �jt = ˛1ˇjt .14 We exclude 7 firms where the estimated beta is greater than 5 in absolute value.15 The gearing variable is of poorer quality than others in the data, leading to nega-ive gearing observations for some firms. Any firm-year observations with negativeearing are excluded from the analysis (a total of 37 observations).

20 R. Davies et al. / Journal of Financial Stability 12 (2014) 16–25

Table 1Number of firms in each industry segment.

Index Consumer Energy and utilities Financials Health IT Industrials Materials Total

S&P 117 65 78 47 73 47 23 450FTSE 52 14 42 5 34 18 9 174

Table 2Mean dividend-price ratio for firms in each industry segment.

Index Consumer Energy and utilities Financials Health IT Industrials Materials

S&P 1.94 2.82 3.19 1.87 1.69 2.55 2.22FTSE 4.12 3.96 2.63 1.38 2.92 3.81 3.16

Table 3Estimated betas.

Index Number of firms Number of observations Mean Median S.D

dvs

P

Faeb

E

Ap

P

Ccspa

aaus

tIy8T(afi

ts

m

fn

atsls

mtmhifac

t0ymscddayTUUefpuI

f

S&P 401 10,140

FTSE 168 3765

The right-hand-side of Eq. (15) defines the stream of future divi-ends and terminal prices. To generate these, consider a simplifiedersion of (15) which abstracts from discount rates, company-pecific risk premia and dividends:

jt = Et(Pjt+N)xN (17)

ollowing Wickens (1982), the rational expectation t + N periodshead are formed on the basis of information available at time t. Forach company j, these expectations differ from the realised valuesy a forecast error (Ujt+N) unpredictable at time t:

t(Pjt+N) = Pjt+N − Ujt+N (18)

dding and subtracting the average forecast error across all com-anies (Ujt+N) gives:

jt = Et(Pjt+N)xN = Pjt+NxN − Ujt+NxN − (Ujt+N − Ujt+N)xN (19)

onsistent estimation of (18) is possible using a set of instrumentsorrelated with Pjt+N but which are independent of the company-pecific excess forecasting errors. In the estimation, lagged sharerices, lagged dividends per share and lagged earnings per sharere used as instruments for future dividends and equity prices.

Given estimates of company beta and gearing, risk-free ratesnd the instrumented variables, Eq. (15) can be estimated to gener-te estimates of the short-termism parameter, x. This was achievedsing non-linear least squares on a set of cross-sectional regres-ions for each of the years 1985–2004.16

Table 4 presents the results for each of the years. Our estimate ofhe x parameter differs significantly from unity in 11 of the 20 years.n 13 of the 20 years, x is lower than 1, significantly so in 9 of theseears. Moreover, there is more evidence of myopia in later years:

of the 9 myopic years occur in the final decade of the sample.able 5 shows point estimates of x over two decadal sub-samples1985–1994 and 1995–2004) and over the full sample. Estimatesre significantly below unity in the second sub-sample, but not therst. The point estimate of x over the second sub-sample is 0.94.17

There are some interesting patterns in the degree of short-ermism across sectors. Table 6 shows estimates of x over theame three time periods (1985–1994, 1995–2004, full sample) on

16 Because the estimation uses expected values up to five periods ahead, the esti-ates only run to 2004.

17 Various robustness checks were conducted. These included dropping gearingrom the estimation of the risk premium and varying the effects of taxes. These didot alter significantly the empirical estimates.

nes2

t

0.91 0.86 0.490.63 0.62 0.45

sectoral basis. There is statistically significant evidence of short-ermism in the second half of the sample for all seven industrialectors. And in all of these sectors except health and materials, x isower in the second half of the sample than the first – in those twoectors x is below unity throughout the sample.

There are various reasons why short-termism may have becomeore important in equity markets in recent years. In particular,

he proportion of assets held by long-term institutional investorsay have fallen relative to say, hedge funds and, more recently,

igh-frequency trading firms. In addition, various commentators,ncluding investment professionals, have observed an increasedocus on quarterly earnings and a rise in portfolio turnover evenmong institutional investors.18 Both of these portfolio trends areonsistent with our findings.

The estimates of short-termism are economically as well as sta-istically significant. The estimates for x often lie between 0.9 and.95, suggesting excess discounting of between 5% and 10% perear. To illustrate the impact of this degree of myopia on invest-ent choice, Chart 1 shows the present value of those income

treams under three counter-factual assumptions: rational dis-ounting; myopic discounting – lower bound (5%); and myopiciscounting – upper bound (10%). The cumulative impact is fairlyramatic. Ten-year ahead cash-flows under rational discountingre valued similarly to between six-year (lower bound) and four-ear (upper bound) ahead cash-flows under myopic discounting.his is illustrated even more clearly if we consider payback periods.nder rational discounting, payback occurs in 9 years (Chart 2).nder upper bound myopic discounting, the investor today wouldrroneously assume that payback would never be made. These dif-erences have the potential to alter radically project choice. The netresent value of this project evaluated over 50 years falls from $56nder rational discounting to a loss of $11 under extreme myopia.

n other words, a NPV-positive project would be rejected.Put differently, consider how quickly the present value of a

uture cash-flow decays to a level of 1% of its face value, under ratio-al and myopic discounting. Under rational discounting, cash-flowsven 50 years ahead retain more than 1% of their face value. Under

trong myopic discounting, this residual threshold is reached after5 years. Virtually zero weight – less than 1000th of the face value

18 On portfolio turnover see for example Haldane and Davies (2011) and on quar-erly earnings see Aspen Institute (2009).

R. Davies et al. / Journal of Financial Stability 12 (2014) 16–25 21

Table 4Results, main specification.

Year Constant X Beta Gearing R2 Log L N x < 1 x > 1

1985 −1.125 (7.648) 1.194 (0.168) 0.121 (0.187) 0.134 (0.498) 0.448 −390 831986 5.953 (2.317) 1.207 (0.062) 0.18 (0.087) 1.376 (0.277) 0.791 −390 89 *

1987 −3.201 (2.532) 1.094 (0.032) 0.065 (0.043) 0.133 (0.083) 0.904 −427 100 *

1988 0.822 (1.777) 1.019 (0.024) 0.067 (0.034) 0.105 (0.065) 0.887 −473 1081989 −1.192 (3.996) 1.048 (0.035) −0.007 (0.045) 0.193 (0.092) 0.908 −501 1141990 4.862 (4.539) 0.976 (0.035) −0.046 (0.006) 0.101 (0.127) 0.793 −572 1161991 5.293 (2.297) 0.967 (0.016) 0.082 (0.022) 0.07 (0.082) 0.89 −551 123 *

1992 2.224 (6.412) 1.097 (0.062) 0.083 (0.077) 0.155 (0.166) 0.902 −566 1251993 −2.71 (5.564) 1.044 (0.028) 0.105 (0.04) 0.056 (0.101) 0.956 −588 1331994 8.943 (7.728) 0.981 (0.03) 0.066 (0.056) −0.15 (0.107) 0.877 −709 1371995 4.839 (7.547) 0.969 (0.031) −0.005 (0.056) 0.205 (0.227) 0.942 −681 1451996 14.264 (7.244) 0.821 (0.033) −0.028 (0.067) −0.341 (0.132) 0.844 −794 152 *

1997 19.582 (7.922) 0.86 (0.013) −0.062 (0.028) −0.025 (0.08) 0.923 −1602 292 *

1998 9.317 (7.639) 0.948 (0.016) −0.052 (0.018) 0.192 (0.053) 0.902 −1931 340 *

1999 11.296 (6.848) 0.914 (0.012) −0.08 (0.014) 0.186 (0.041) 0.888 −2070 354 *

2000 18.479 (11.467) 0.9 (0.013) −0.105 (0.01) 0.042 (0.048) 0.789 −2323 362 *

2001 7.753 (12.51) 0.997 (0.016) 0.023 (0.015) 0.225 (0.044) 0.897 −2267 3712002 20.674 (8.957) 0.937 (0.016) 0.022 (0.019) 0.162 (0.038) 0.885 −2273 380 *

2003 2.92 (8.094) 0.927 (0.015) 0.012 (0.016) 0.079 (0.04) 0.847 −2318 391 *

2004 21.877 (8.088) 0.97 (0.015) −0.007 (0.017) 0.119 (0.041) 0.802 −2445 394 *

Notes. Results shown are for the full specification, with a separate parameter for gearingassumed to enter the beta (i.e. the parameter on gearing is restricted to zero) yields simil

* x is significantly different from 1, using a 95% confidence interval.

Table 5Short-termism estimates for the US and UK.

Year X Standard error Evidence ofshort-termism?

Full sample (1985–2004) 0.937 0.004 Yes1985–1994 1.001 0.008 No1995–2004 0.938 0.005 Yes

Notes. The significance column refers to a test of whether x is significantly differentfrom 1 at the 5% confidence level.

ob

4

hfstreams, at the expense of retaining, or ploughing back, profits intothe business to finance future growth opportunities.

TS

i

able 6ectoral results.

Year x

Consumer1985–1994 325 1.007

1995–2004 814 0.94

All years 1139 0.939

Energy and utilities1985–1994 127 1.05

1995–2004 429 0.934

All years 556 0.939

Financials1985–1994 213 1.087

1995–2004 673 0.962

All years 886 0.963

Health1985–1994 69 0.857

1995–2004 230 0.932

All years 299 0.932

IT1985–1994 183 0.957

1995–2004 465 0.892

All years 648 0.902

Industrials1985–1994 135 1.018

1995–2004 287 0.936

All years 473 0.937

Materials1985–1994 76 0.871

1995–2004 183 0.86

All years 259 0.892

* x is significantly different from 1, using a 95% confidence interval.

included. Standard errors in parentheses. A restricted model, in which gearing isar results.

f the cash-flow – is placed on projects with income streams mucheyond 35 years.

.2. The impact of short-termism on investment

Short-termism may induce firms to distribute a sub-optimallyigh share of their revenues and profits to shareholders in the

orm of dividends, to meet their demands for near-term income

There is anecdotal evidence that firms may increasingly be seek-ng to actively manage the quantum and timing of their dividends.

s.e x < 1 x > 1

0.0170.010 *

0.008 *

0.022 *

0.013 *

0.011 *

0.0440.008 *

0.007 *

0.0840.027 *

0.233 *

0.0240.019 *

0.015 *

0.0510.025 *

0.021 *

0.045 *

0.048 *

0.040 *

22 R. Davies et al. / Journal of Financial Stability 12 (2014) 16–25

Not es: The ch art assu mes $10 is pai d a t the en d of e ach year. The risk- free disc oun t rate use d is 1. 085.

$0

$1

$2

$3

$4

$5

$6

$7

$8

$9

$10

0 1 2 3 4 5 6 7 8 9 10

x=1.00 x=0.95 x=0.90

Years

Chart 1. Present value of future cash-flows.

Notes: The cumulati ve NPV o f $10 c ash-f lows rises to $61 in year 9 under rational discounting. With mild myopia (x=0.95) it only pass es $60 at year 15. With severe myopia (x=0.90)

$0

$20

$40

$60

$80

$100

$120

$140

0 5 10 15 20 25 30 35 40 45 50

x = 1.00 x = 0.95

x = 0.90 Investment cost

Years

Ha“dddcwt

mb

B

0

20

40

60

80

100

120

140

Fixed assets by profits Fixed assets by sales turno ver

Quoted Pr ivateRa�o

Cs

iiKBiRflwd

mfifipffli

dtaat

otaquoted firms. Similarly, if we compare stocks of fixed assets tosales turnover, private firms have stocks of fixed assets 127 timeslarger than their sales turnover, while for quoted firms they are

the i nvestor cal cula tes t hat payback is n ot ac hieved.

Chart 2. Cumulative present value of future cash-flows.

aldane (2011) presents evidence on firms seeking to maintainnd smooth dividends. Others have observed that firms engage inequity recycling” – issuing new equity at the same time as payingividends, even though it would be less costly to simply cancel theividend.19 Over recent years, some firms continued to pay divi-ends, even though they were running at a loss. This is in stark

ontrast to dividend behaviour in the 19th and early 20th century,hen US and UK firms were as likely to reduce dividends as raise

hem.20

19 Weld (2007) finds that 40% of common equity issuance is by firms that areaking regular positive dividend payments, and that 20% of this raised equity could

e avoided if firms eliminated dividends.20 See Haldane (2011) for a fuller discussion and Goetzmann et al. (2001) andraggion and Moore (2008) for the US and UK history.

In

viAp

hart 3. Stocks of fixed assets of private and quoted firms scaled by profits andales.

There is also evidence that investment in long-term researchn the US and UK, two countries where capital market financings most developed, is lagging behind other countries. Japan, Southorea and China have increased their R&D to GDP ratio since 1980.y comparison, the US has dropped back, maintaining its R&D

ntensity at a roughly stable level over the past 30 years. The UK&D record is worse still, with the ratio of R&D to GDP having

allen secularly. These aggregate numbers are borne out in firm-evel data. In 1993, 94 of the top 200 firms by R&D expenditure

ere American or British. By 2009, just 77 were.21 Could this beue to short-termism?

One way to answer this question is to look at whether invest-ent patterns are different for quoted firms relative to private

rms. To test this, we use a cross-section of around 140,000 UKrms in 2010 from Bureau van Dijk, comprising both quoted andrivate firms.22 Summary statistics are available in Appendix A. Weocus on measures of the capital stock, rather than the investmentow in a particular year, to pick up the long-term or accumulated

nfluence of short-termism on investment choice.Specifically, we look at corporates’ stock of fixed assets. We

efine the variable of interest as the ratio of the stock of fixed assetso the flow of profits or turnover of a company. Profits and turnoverre used as a normalising measure of the flow of resources avail-ble for investment in a given year, as well as helping to control forhe different sizes of (public versus private) firms.23

On average, private firms tend to have materially larger stocksf fixed assets relative to their incoming resources (profits or salesurnover) than public firms (Chart 3). The average ratio of fixedssets to profits is over 100 for private firms but only 24 for

21 Source: the Department for Business Innovation and Skills “R&D Scoreboard”.n the 1993 survey, 81 of the top 200 firms were US firms and 13 were UK firms; theumbers in 2009 were 68 and 9 respectively.22 Quoted firms are 1.1% of all UK firms in this sample.23 Other papers (e.g. Bernstein, 2012) use the number of patents as dependentariable. As described in the literature review, we consider this as an outcome ofnvestment, while in this paper we focus on the choice of investment expenditure.lso, not all corporate investment results in patents, so such a study investigates aarticular subsample of firms.

R. Davies et al. / Journal of Financial Stability 12 (2014) 16–25 23

Table 7Effect of private versus public ownership on companies’ stocks of fixed assets.

Dependent variable: fixed assets scaled by profits (1) (2) (3) (4) (5) (6) (7)

Private 77.09* 76.60* 125.98* 82.58* 144.84* 231.33* 79.14*

(1.85) (1.85) (2.54) (1.76) (1.96) (2.69) (1.97)Sales turnover −4.79e−10 −1.65e−9 −5.65e−10 3.59e−10 3.35e−11 2.6e−10

(0.52) (1.30) (−0.54) (0.37) (0.02) (0.69)ROA 0.20*** 0.19*** 0.21* 0.20***

(4.12) (3.30) (2.69) (3.07)ROA × private −0.74** −0.86** −1.30** −0.98**

(−2.34) (−2.34) (−3.32) (−2.58)AGE of the company −0.01 −0.05 0.08

(−0.04) (−0.23) (0.43)AGE of the company × private −3.36* −5.08* −2.77

(−1.87) (−2.32) (−1.58)Constant 23.91*** 24.41*** 30.13*** 25.27*** 24.49** 31.07**

(3.55) (3.65) (4.23) (3.52) (2.91) (3.62)Sectoral fixed effects YesSectoral fixed effects × private Yesı 4.22 4.14 5.18 4.27 6.91 8.45 see Table 8R2 0.00 0.00 0.00 0.00 0.00 0.00 0.00Observations 143,445 143,445 92,843 140,419 140,413 90,797 140,413Sectors All All Top 5 All All Top 5 All

Notes. t-Statistics in parentheses. Standard errors are robust and clustered at the sector level. “Top 5 sectors” are the five largest sectors by number of companies in thesample; they are agriculture, manufacturing, wholesale/retail trade, financial intermediation and real estate and cover 91% of total corporate assets in the sample.

2fifipfi

awntcp

F

Taapm

fi(mfi(

grolsro

i

F

Tsrmllm

(mcover over 90% of total corporate assets in the sample. Private firmsin these five sectors appear to have built even larger (and signif-icant) stocks of fixed assets relative to their profits than quotedfirms - the investment multiplier from being a private firm is 5.2.

Table 8Investment multipliers (ı) by sector.

Sector ı

1 Agriculture, hunting and forestry −34.2

* Effect statistically significant from zero at 10% level of significance.** Effect statistically significant from zero at 5% level of significance.

*** Effect statistically significant from zero at 1% level of significance.

5 times larger. In other words, investment stocks are four orve times larger in our sample for private firms than for quotedrms. This is consistent with private firms re-investing a largerroportion of their profits or sales in fixed assets than quotedrms.

To test this formally, we regress companies’ stocks of fixedssets normalised by profits (FIXEDASSETS) on an indicator forhether the company is private (PRIVATE), controlling for compa-ies’ sales turnover (SALES), return on assets (ROA) and the age ofhe company (AGE).24 ROA and AGE are interacted with whether aompany is private to allow for differential effects for private andublic companies. The main specification is as follows.

IXEDASSETSit = + ˇ1PRIVATEit + ˇ2SALESit + ˇ3ROAit + ˇ4ROAit

× PRIVATEi + ˇ5AGEit + ˇ6AGEit × PRIVATEi + εit

(20)

he constant (˛) yields the baseline ratio for quoted firmsnd ˇ1 measures the extra boost to fixed assets from being

private firm. The difference in investment from being arivate firm is: ı = ( + ˇ1)/˛. We call this the “investmentultiplier”.In further specifications we check the robustness of the results,

rst, by restricting the sample to the five largest productive sectorsagriculture, manufacturing, wholesale/retail trade, financial inter-

ediation and real estate); and second, by controlling for sectoralxed effects � i (also interacted with the PRIVATE indicator) (Eq.21)). When sectoral fixed effects are included, the sector-specific

24 For comparison, Asker et al. (2011) include as controls: assets; sales growth;earing; cash/assets ratio; retained earnings/assets ratio; age of the company; andeturn on assets. We decided not to include assets as a control because the orderf magnitude is very different between private and public firms, even after takingogs, and it would have forced doubtful comparisons. We have experimented addingales growth and gearing as controls but this did not produce easily interpretableesults. The other two ratios (cash/assets and retained earnings/assets) are not inur dataset.

Nt

nvestment multiplier becomes: ı = (ˇ1 + ˇ7 + ˇ8)/ˇ7.

IXEDASSETSit = ˇ1PRIVATEit + ˇ2SALESit + ˇ3ROAit + ˇ4ROAit

× PRIVATEi + ˇ5AGEit + ˇ6AGEit × PRIVATEi

+ ˇ7�i + ˇ8�i × PRIVATEi + εit (21)

able 7 presents estimates of the potential impact of public owner-hip on corporate investment. Column 1 presents the baselineegression without other firm-level characteristics. The investmentultiplier from being a private firm is 4.2. In column 2, control-

ing for sales turnover, private firms appear to hold significantlyarger stocks of fixed assets relative to their profits – the investment

ultiplier from being a private firm is 4.1.In column 3, we narrow the sample to the five largest sectors

agriculture, manufacturing, wholesale/retail trade, financial inter-ediation and real estate), which comprise over 92,000 firms and

2 Fishing −5.33 Mining and quarrying −13.24 Manufacturing 10.75 Electricity, gas and water supply −6.16 Construction 0.87 Wholesale and retail trade 8.58 Hotels and restaurants −5.79 Transport, storage and communication 3.510 Financial intermediation 15.711 Real estate, renting and business activities 6.612 Public administration and defence −1.913 Education −6.614 Health and social work 4.6

ote. Benchmark sector: wholesale and retail trade (for median ratio of fixed assetso profits).

24 R. Davies et al. / Journal of Financial Stability 12 (2014) 16–25

Table 9Summary statistics.

Variable Obs Mean Std. Dev. Min Max

Private firms (98.9% of the sample)Fixed assets (GBP) 141,920 32,000,000 1,240,000,000 1 322,000,000,000Fixed assets scaled by profits 141,920 101 8858 −609809 1927855Fixed assets (% of assets) 141,920 36 34 0 100Total assets (GBP) 141,920 73,000,000 2,990,000,000 1 481,000,000,000Profit after tax (GBP) 141,920 1,663,835 73,900,000 −5,210,000,000 12,500,000,000Sales turnover (GBP) 141,920 25,400,000 516,000,000 −23,500,000 120,000,000,000Return on assets 138,905 16 53 −223 328Age of the company (years) 141,914 17 18 1 154

Quoted firms (1.1% of the sample)Fixed assets (GBP) 1525 1,520,000,000 14,500,000,000 386 319,000,000,000Fixed assets scaled by profits 1525 24 431 −5479 11795.26Fixed assets (% of assets) 1525 59 28 0 100Total assets (GBP) 1525 2,290,000,000 18,200,000,000 90,866 370,000,000,000Profit after tax (GBP) 1525 79,100,000 601,000,000 −2120,000,000 13,100,000,000Sales turnover (GBP) 1525 1,040,000,000 8,740,000,000 −246,000 242,000,000,000Return on assets 1514 −3 27 −217 103Age of the company (years) 1525 26 31 2 153

Table 10Summary statistics by sector.

Obs Fixed assets (% total assets) Fixed assets scaled by profits

Private firms (98.9% of the sample)1 Agriculture, hunting and forestry 1421 50 22 Fishing 116 49 43 Mining and quarrying 655 41 174 Manufacturing 13,451 28 15 Electricity, gas and water supply 486 55 −86 Construction 10,310 26 −547 Wholesale and retail trade; repairs 15,845 25 148 Hotels and restaurants 4063 63 379 Transport, storage and communication 5439 37 010 Financial intermediation 5690 33 57411 Real estate, renting and business activities 53,180 38 19312 Public administration and defence 374 25 1113 Education 3649 44 1414 Health and social work 7275 42 615 Other community, social and personal services 16,945 40 19

Quoted firms (1.1% of the sample)1 Agriculture, hunting and forestry 12 59 22 Fishing 03 Mining and quarrying 90 70 −74 Manufacturing 309 52 105 Electricity, gas and water supply 15 72 −36 Construction 37 33 157 Wholesale and retail trade; repair of mo 105 45 128 Hotels and restaurants 24 79 −139 Transport, storage and communication 58 62 1910 Financial intermediation 342 75 4211 Real estate, renting and business activities 442 54 3712 Public administration and defence 0

4

2

4

prpowmbtCtws

dtsoiservices exhibit positive multipliers on their investment relative toquoted firms, with multipliers ranging from 4 to 15.

13 Education

14 Health and social work 115 Other community, social and personal services 6

Column 4 adds return on assets, alone and interacted for being arivate firm. In general, firms appear to invest more, the larger theireturn on assets, but less so for private firms. The investment multi-lier remains significant and above 4. Column 5 adds the longevityf the company, with age of the company both alone and interactedith the ‘private’ dummy. Controlling for age raises the invest-ent multiplier to over 6. There is a non-significant relationship

etween investment and the age of the company; but private firmsend to invest more, the younger they are, as might be expected.

olumn 6 restricts the latter specification to the five largest sec-ors. The direction and significance of all effects is the same, onlyith stronger magnitudes. The investment multiplier of these five

ectors is estimated at over 8. m

58 −860 1165 7

Finally, column 7 adds fixed effects for each sector. Again, theirection, magnitude and significance of the effects are robust tohis specification. Table 8 presents the investment multipliers byector. Relative to the median sector,25 private firms in the sectorsf manufacturing; transport, storage and communication; financialntermediation; real estate; health and social work and community

25 The benchmark category is the wholesale and retail trade sector, with theedian ratio of fixed assets to profits in the sample.

nancia

pb

5

ctdddt

etipwAsctois

diiotpa

utaiii(rlFrk

A

R

A

A

A

A

B

B

B

B

B

B

B

C

E

F

FG

G

G

H

H

H

H

K

L

L

M

M

M

MN

P

R

S

Stein, J., 1989. Efficient capital markets, inefficient firms: a model of myopic corpo-

R. Davies et al. / Journal of Fi

These results suggest, overall, that UK private firms tend tolough back between 4 and 8 times more of their profits into theirusiness over time than publicly held firms.

. Conclusions and next steps

On the basis of the evidence presented here, short-termism inapital markets appears to come with a potentially significant priceag, albeit one which is often hidden from view. The evidence israwn from asset prices in capital markets and from investmentecisions by companies. Both point to a quantitatively significantegree of short-termism in capital markets, whether measured byhe cost of capital or investment intentions.

To assess the broader implications of these estimates for theconomy, consider a simple ready-reckoner. Assume the short-ermism problem can be approximated by the difference in thenvestment performance of public and private firms. Given theublic/private split of firms, the stock of capital at quoted firmsould then be several times larger in the absence of short-termism.ssume a standard Cobb-Douglas production function with con-tant returns to scale and an elasticity of output with respect toapital of 0.3. A simple back of the envelope exercise suggests thathe elimination of short-termism would then result in a level ofutput around 20% higher than would otherwise be the case. Evenf this is an upper bound, it suggests the gains are potentially sub-tantial.

From a research perspective, further work could usefully beone to better calibrate these gains. Using alternative identify-

ng restrictions to gauge the potential impact of short-termism onnvestment choice is one potential avenue – for example, the effectsf firms switching from private to public or vice versa. Comparinghe experience of different countries, with different sets of com-any law and different patterns of capital market financing, maylso be revealing.

From a policy perspective, there are a variety of potential meas-res that could be considered which lean against short-termistendencies by investors (see Kay, 2012). Among the more obviousre alterations to the tax code to provide companies with strongerncentives to retain (rather than distribute) profits and to providenvestors with stronger incentives to hold (rather than churn)nvestment assets – for example, by issuing “loyalty bonuses”Bolton and Samama, 2010). Another policy option would be toebalance the distribution of voting rights within a firm towardsonger-duration investors – for example, as is the case for somerench companies. All of these options would benefit from greateresearch, given the quantitatively significant degree of capital mar-et myopia estimated in this paper.

ppendix A.

See Tables 9 and 10

eferences

sker, J., Farre-Mensa, J., Ljungqvist, A., 2011. Comparing the Investment Behaviorof Public and Private Firms, NBER Working Paper 17394.

sker, J., Farre-Mensa, J., Ljungqvist, A., 2013. Corporate Investment andStock Market Listing: A Puzzle?, Working Paper http://papers.ssrn.com/sol3/papers.cfm?abstract id=1603484

WW

l Stability 12 (2014) 16–25 25

spen Institute, 2010. Short-Termism and US Capital Markets, A Compelling Casefor Change. Aspen Institute, USA.

spen Institute, 2009. Overcoming short-termism: a call for a more responsibleapproach to investment and business management. In: The Aspen InstituteBusiness and Society Program, September 9.

ernstein, S., 2012. Does Going Public Affect Innovation? Stanford Graduate Schoolof Business, Working Paper.

lack, A., Fraser, P., 2002. Stock market short-termism—an international perspective.Journal of Multinational Financial Management 12, 135–158.

olton, P., Samama, F., 2010. L-Shares: Rewarding Long-term Investors. Committeeon Global Thought, Columbia University, mimeo.

raggion, F., Moore, L., 2008. Dividend Policies in an Unregulated Market: TheLondon Stock Exchange 1895–1905. Tilburg University, Center for EconomicResearch, Discussion Paper 2008-83.

realey, R.A., Myers, S.C., Allen, F., 2010. Principles of Corporate Finance. McGraw-Hill, UK.

ushee, B.J., 1998. The influence of institutional investors on myopic R&D invest-ment behavior. The Accounting Review 73 (July (3)), 305–333.

usiness Roundtable, Krehmeyer, D., Orsagh, M., Schacht, K.N., 2006. Breaking theShort-Term Cycle. Business Roundtable Institute for Corporate Ethics, 24 July.

uthbertson, K., Hayes, S., Nitzsche, D., 1997. The behaviour of UK stock pricesand returns: is the market efficient? The Economic Journal 107 (443),986–1008.

uropean Commission, 2011. The EU Corporate Governance Framework. In:COM(2011), 164 final.

inancial Stability Board, 2013. Financial Regulatory Factors Affecting the Avail-ability of Long-term Investment Finance, Report to G20 Finance Ministers andCentral Bank Governors.

inancial Times, 2012. Industry shows mixed response to Kay Review, 29th June.oetzmann, W.N., Ibbotson, R.G., Peng, L., 2001. A new historical database for the

NYSE 1815 to 1925: performance and predictability. Journal of Financial Markets4 (1), 1–32.

raham, J.R., Harvey, C.R., Rajgopal, S., 2005. The economic implications of corporatefinancial reporting. Journal of Accounting and Economics 40, 3–73.

30 Working Group, 2013. Long-Term Finance and Economic Growth, SpecialReport.

aldane, A.G., 2011. Capital discipline, remarks based on a speech given atthe American Economic Association, Denver. http://www.bankofengland.co.uk/publications/speeches/2011/speech484.pdf

aldane, A.G., Davies, R., 2011. The short long. In: Speech given at the 29th SociétéUniversitaire Européenne de Recherches Financières Colloquium, Brusselshttp://www.bankofengland.co.uk/publications/Documents/speeches/2011/speech495.pdf

illier, D., Grinblatt, M., Titman, S., 2008. Financial Markets and Corporate Strategy.McGraw-Hill, UK.

oggarth, G., Reis, R., Saporta, V., 2002. Costs of banking system instabil-ity: some empirical evidence. Journal of Banking and Finance 26 (5),825–855.

ay, 2012. The Kay Review of UK Equity Markets and Long-Term Decision Making.Department for Investment, Business and Skills, United Kingdom, July.

aibson, D., 1997. Golden eggs and hyperbolic discounting. Quarterly Journal ofEconomics 112 (2), 443–477.

intner, J., 1965. The valuation of risk assets and the selection of risky investmentsin stock portfolios and capital budgets. Review of Economics and Statistics 47(1), 13–37.

iles, D., 1993. Testing for short termism in the UK stock market. The EconomicJournal 103 (421), 1379–1396.

iller, M.H., Rock, K., 1985. Dividend policy under asymmetric information. TheJournal of Finance 40 (September (4)), 1031–1051.

odigliani, F., Miller, M., 1958. The cost of capital, corporation finance and the theoryof investment. American Economic Review 48 (3), 261–297.

ORI, 2004. NAPF/IMA Short-termism Study Report.arayanan, M.P., 1985. Managerial incentives for short-term results. Journal of

Finance 40 (December (5)), 1469–1484.oterba, J., Summers, L.H., 1995. A CEO survey of U.S. companies’ time horizons and

hurdle rates. Sloan Management Review, 43–53, Fall.oss, S.A., 1976. The arbitrage theory of capital asset pricing. Journal of Economic

Theory 13, 341–360.harpe, W.F., 1964. Capital asset prices: a theory of market equilibrium under con-

ditions of Risk. Journal of Finance 19 (3), 425–442.

rate behavior. Quarterly Journal of Economics 104 (4), 655–669.eld, W., 2007. Recycled Equity. Cornell University, Working Paper.ickens, R.M., 1982. The efficient estimation of econometric models with rational

expectations. Review of Economic Studies 49 (1), 55–67.