International Stock Market Liquidity

INTERNATIONAL STOCK MARKET LIQUIDITY DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Christof W. Stahel, M.A. ***** The Ohio State University 2004

Dissertation Committee:

Approved by

Ren´e M. Stulz, Adviser Kewei Hou Ingrid M. Werner

Adviser Graduate Program in Business Administration

ABSTRACT

This dissertation contributes to the international asset pricing literature. The research it presents in its two essays is related to papers that investigate commonalities in individual stock liquidity in the domestic US setting, to research that estimates risk premia related to liquidity risk in the US, and to articles that explore properties and determinants of market-wide liquidity in the US, while expanding the scope to an international setting. The first essay shows that individual liquidity exhibits commonalities in monthly measures of individual stock liquidity within and across countries for a sample from Japan, the UK, and the US from 1980 to 2001. An asset pricing analysis suggests that expected stock returns are cross-sectionally related to the sensitivity of returns to shocks in global liquidity in this sample and that global liquidity is a priced state variable in an international framework at the portfolio as well as at the individual stock level. The second essay analyzes cross-regional and time-series properties of weekly market-wide liquidity measures from 1990 to 2002 for five regional aggregates: developed Asia, North America, Europe, emerging Asia, and emerging America. The aggregates are calculated from a sample that contains 39 developed and emerging countries. The results suggest that liquidity shocks are contemporaneously correlated and dynamically spread across regions. However, there is only week evidence ii

that liquidity affects returns in this sample. An investigation of determinants of liquidity indicates that market-wide returns, market-wide averages of individual stock volatilities, and world net bond flows are fundamental drivers of market-wide liquidity. There is little evidence that equity fund flows and interest rates consistently affect liquidity in the sample. Even though changes in liquidity can to some extent be explained by returns and other determinants, shocks to liquidity continue to be contemporaneously correlated across markets. But the empirical results from an application of extreme value theory offers evidence that extreme shocks to liquidity are asymmetrically correlated in the tail of the distribution. In particular, it is mostly negative extreme liquidity shocks that are correlated between North America, Europe, and emerging America. The overall conclusions from this dissertation are twofold. First, changes in global liquidity constitutes an international risk factor, and financial assets with returns that are more sensitive to this factor reward investors with higher expected returns. However, the contribution of liquidity risk to expected returns seems to be more relevant for developed markets. Second, market-wide liquidity is contemporaneously and dynamically related across regions. Furthermore, these relationships do not simply reflect other variables that are related across markets but constitute a phenomenon by themselves.

iii

This dissertation is dedicated to the ones I miss and love.

iv

ACKNOWLEDGMENTS

I would like to thank my advisor Ren´e Stulz and my committee members Kewei Hou and Ingrid Werner for their guidance, insightful feedback, and encouragement that helped make this dissertation possible. I also thank Tom Bates, Terry Campbell, Jeff Harris, Jean Helwege, David Hirshleifer, Roberto Ragozzino, and seminar participants at the University of Delaware, Drexel University, George Mason University, HEC Montreal, The Ohio State University, Queen’s University, and Texas Tech University for helpful comments and suggestions, and Laurie Pomerson for her help with too many versions of the manuscript.

v

VITA

June 25, 1964 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Born – Z¨ urich, Switzerland 1995 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . lic.oec.publ. – University of Z¨ urich, Switzerland 1997 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M.A. Economics – The Ohio State University, USA

PUBLICATIONS Research Publications Michel Peytrignet and Christof W. Stahel, Stability of money demand in Switzerland: A comparison of the M2 and M3 cases, Empirical Economics, 23:437–454, 1998.

FIELDS OF STUDY Major Field: Business Administration Concentration: Finance

vi

TABLE OF CONTENTS

Page Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ii

Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

iv

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

Vita . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vi

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

Chapters: 1.

2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

1.1 1.2 1.3 1.4

. . . .

2 4 5 7

Is there a Global Liquidity Factor? . . . . . . . . . . . . . . . . . . . . .

11

2.1 2.2

19 24 24 27 32 33 38 39

2.3

International Asset Pricing Models . . . Market Integration . . . . . . . . . . . . Capital Flows, Spillovers and Contagion Essay Summary . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

Sample and Liquidity Measures . . . . . . . . . Commonalities in Liquidity . . . . . . . . . . . 2.2.1 Contemporaneous Variation . . . . . . . 2.2.2 Sources of Commonalities . . . . . . . . Asset Pricing Implications . . . . . . . . . . . . 2.3.1 Liquidity Risk . . . . . . . . . . . . . . 2.3.2 Expected Returns and Trading Cost . . 2.3.3 Stock Level Fama-MacBeth Regressions vii

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . .

. . . . . . . .

. . . . . . . .

2.4 2.5

Alternative Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

41 41

Are liquidity Shocks correlated across Equity Markets? . . . . . . . . . .

43

3.1

. . . . . .

48 51 52 56 61 65

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

67

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

70

3.

3.2 3.3 3.4 3.5 4.

Sample and Liquidity Measure . . . . . . 3.1.1 Sample Properties . . . . . . . . . Dynamic Transmission of Liquidity Shocks Liquidity and Returns . . . . . . . . . . . Correlation of Extreme Liquidity Shocks . Conclusion . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

Appendices: A.

Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

B.

Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

viii

77

LIST OF TABLES

Table

Page

A.1 Liquidity Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . .

78

A.2 Commonality in Liquidity . . . . . . . . . . . . . . . . . . . . . . . .

80

A.3 Decomposition of Commonality . . . . . . . . . . . . . . . . . . . . .

81

A.4 Variance Decomposition . . . . . . . . . . . . . . . . . . . . . . . . .

82

A.5 Liquidity Risk Premium - country and industry portfolios . . . . . . .

84

A.6 Liquidity Level and Return . . . . . . . . . . . . . . . . . . . . . . . .

85

A.7 Liquidity Risk Premium - size portfolios . . . . . . . . . . . . . . . .

86

A.8 Liquidity Risk Premium - country, industry, and size portfolios . . . .

87

A.9 Test of constant Risk Premiums . . . . . . . . . . . . . . . . . . . . .

88

A.10 Risk Premium and Liquidity as an Asset Characteristic . . . . . . . .

89

A.11 Fama-MacBeth Risk Premiums . . . . . . . . . . . . . . . . . . . . .

90

A.12 Commonalities and Decompositions - IPO . . . . . . . . . . . . . . .

91

A.13 Variance Decomposition - IPO . . . . . . . . . . . . . . . . . . . . . .

92

A.14 Liquidity Risk Premium - IPO . . . . . . . . . . . . . . . . . . . . . .

93

A.15 Sample Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

94

ix

A.16 Dynamic Transmission of Liquidity Shocks . . . . . . . . . . . . . . .

96

A.17 Granger-Causality Tests . . . . . . . . . . . . . . . . . . . . . . . . .

97

A.18 Correlation of Liquidity Shocks . . . . . . . . . . . . . . . . . . . . .

98

A.19 Liquidity Spillovers . . . . . . . . . . . . . . . . . . . . . . . . . . . .

99

A.20 Liquidity and Returns . . . . . . . . . . . . . . . . . . . . . . . . . . 101 A.21 Correlation of Liquidity Shocks and Return Shocks . . . . . . . . . . 106 A.22 Bivariate Peak Over Threshold Tail Correlations . . . . . . . . . . . . 107

x

LIST OF FIGURES

Figure

Page

B.1 Changes in Market-wide Liquidity . . . . . . . . . . . . . . . . . . . . 109 B.2 Bivariate Peak Over Threshold Tail Correlations . . . . . . . . . . . . 110

xi

CHAPTER 1

INTRODUCTION

International capital flows steadily increased over the last couple of decades as many financial markets opened their borders to foreign investors. For example, international portfolio investments have gradually grown from cumulative net flows into Japan, the UK, and the US of about one billion US dollars in the 1980s to about two billion US dollars in the 1990s. While most European and G-7 countries liberalized their stock markets in the early 1970s, many emerging countries opened their markets to foreign investors in the late 1980s and early 1990s.1 This liberalization process allowed investors to extend their investment opportunity set to include multiple markets. Markets are said to be integrated if assets are traded at the same price regardless of where they are traded, and markets are said to be segmented if an asset’s price depends on where it is traded. If markets are integrated, all investors face the same investment opportunity set with no barriers to international investment. In such world of integrated markets, expected returns on risky assets are related to risk factors that are common to all markets. 1

See, for example, Kaminsky and Schmukler (2001).

1

The following sections survey the field of international asset pricing and offer a bird’s-eye view of where this dissertation is embedded.2 The survey is followed by a brief outline and overview of the results of the two essays presented in chapter 2 and chapter 3, and the final chapter offers some conclusions.

1.1

International Asset Pricing Models

The modeling of asset prices in a world of integrated financial markets depends on the assumption of whether or not consumption opportunity sets of investors can differ across countries. If consumption opportunity sets are the same for each investor, markets are called perfect and the specific location of an investor does not matter when maximizing life-time utility via an optimal investment portfolio choice. The law of one price applied to international finance states that the domestic and foreign price of a consumption good is proportional with the proportionality factor being the price of foreign currency. If the law of one price holds across goods and countries, consumption opportunity sets are equal across countries up to this common proportionality factor, and the purchasing power parity of international finance holds. In such a world, the location of an investor does not matter when he is translating future wealth into consumption as the exchange rate adjusts accordingly to equate prices across countries. Moreover, the returns on assets are related to risk factors which affect all markets, and these factors can be viewed as common, global counterparts to factors in domestic asset pricing models. In such a setting, the numerous models of asset pricing, which have implicitly been developed for single markets, can simply be translated into an international setting by viewing the extended investment opportunity set to 2

See the survey of the field of international finance by Karolyi and Stulz (2003b) and the compilation of the most important work in the field by Karolyi and Stulz (2003a).

2

represent one global market. For example, in the simplest case of only one common global risk factor, financial assets could be priced using a world version of the capital asset pricing model (CAPM), where expected excess returns are proportional to the expected excess return of the world market portfolio. The number and nature of the risk factors is ultimately an empirical question, since theory does offer little guidance in this respect. There is a vast empirical literature investigating the performance of the world CAPM and its multifactor extensions in unconditional and conditional forms for portfolios and individual stocks. For example, tests of unconditional versions of the models have been performed by Stehle (1977), Mark (1988), Harvey (1991), Chan, Karolyi, and Stulz (1992), Fama and French (1998); and tests of conditional versions have been carried out by Ferson and Harvey (1993, 1994, 1995, 1997) and Harvey (1995). Overall, the evidence shows that portfolio risk premia can be explained by the covariance of returns with the world market portfolio and that individual stock risk premia can be explained by multifactor models with, for example, additional global size and value-growth factors which resemble the domestic US factors proposed by Fama and French (1993).3 All of the above discussed models depend on purchasing power parity, but there is a vast literature in international economics that casts strong doubts on the purchasing power parity’s validity. This in turn implies that an investor’s evaluation of the riskiness of an asset must depend on the specific location of the investor. 3

See also Griffin (2002), who decomposes the global factors into domestic and international components and shows that the domestic component of the global value-growth factor is most relevant.

3

Consider, for example, a riskless asset with a return denominated in a particular currency. This asset is riskless for a domestic investor, but not for a foreign investor. In such a world where consumption opportunity sets differ across countries, whenever a domestic investor holds a foreign asset, the return in domestic currency depends on the exchange rate and the investor faces exchange rate risk. Solnik (1974), Sercu (1980), Stulz (1981a), Adler and Dumas (1983), and Solnik (1983) derive asset pricing models in which foreign exchange rate risk is priced. The research which empirically tests these pricing models, such as Jorion (1991), Dumas and Solnik (1995), Santis and G´erard (1998), Dahlquist and Sallstrom (2002) finds that foreign exchange rate risk is priced in international portfolios.

1.2

Market Integration

The integration of financial markets implies that investors who think that one market offers higher expected returns over another market can move their funds to that market. The enormous growth in cross-border net capital flow over the past three decades offers testimony to the liberalization and integration of capital markets.4 All of the international asset pricing models discussed so far rely on the assumption that stock markets are integrated and investors can freely move their funds from one country to another. However, there might exist a large number of direct and indirect costs and barriers, such as foreign investment restrictions, legal constraints, discriminatory taxes and higher investment cost, as well as psychological, cultural, and behavioral aspects, that could segment markets and impede investors to take 4

For information on capital flows, see, for example, the Treasury International Capital Reporting System from the US Treasury, and the World Economic and Financial Surveys from the International Monetary Fund.

4

full advantage of international investment opportunities.5 For example, Black (1974), Stulz (1981b), Errunza and Losq (1985), and Eun and Janakiramanan (1986) directly model the portfolio choice under the assumption of different levels of investment barriers and restrictions on foreign ownership. Most studies that investigate market integration work in an asset pricing framework. They generally assume under the null hypothesis that expected returns are generated by an international asset pricing model and under the alternative hypothesis that expected returns are explained by pricing models related to segmented markets. These studies regularly cannot reject the assumption of market integration for developed markets for the 1980s and 1990s. Even though explicit barriers mostly have been removed for emerging markets, the extent to which the markets are integrated into the world market depends ultimately on whether foreign investors actually invest in these markets. There exists some empirical evidence in Bekaert and Harvey (1995) and Foerster and Karolyi (1999), for example, which suggests that markets are implicitly integrated, but with a degree that varies over time.

1.3

Capital Flows, Spillovers and Contagion

Given the evidence that most markets are integrated and risk premia are determined globally, it is natural to observe that stock prices around the world tend to exhibit some comovement. There are a number of studies investigating the contemporaneous correlation among equity markets and lead and lag patterns in returns and volatilities. While Longin and Solnik (1995) and Bekaert and Harvey (2000) 5

See Kaminsky and Schmukler (2001) for a comprehensive list of liberalization dates.

5

document general patterns and time-variation in international stock market correlations, Eun and Shim (1989), Engle, Ito, and Lin (1990), and Hamao, Masulis, and Ng (1990) document lead and lag effects across markets in returns as well as spillovers in volatilities. However, linking the correlation patterns and spillovers to market and economics factors has so far resulted in little significant evidence. There is a growing literature in international finance that investigates the joint dynamics of capital flows and asset returns. The basic question here is whether flows reflect additional information of foreigners about changes in expected returns, or whether flows impact returns themselves. As in Bohn and Tesar (1996), Seasholes (2000), and Froot, O’Connell, and Seasholes (2001), for example, most of the results suggest that foreign investors are feed-back traders – buying following positive returns and selling following negative returns – and are worse informed than domestic investors. However, Choe, Kho, and Stulz (1999) offer some evidence that this behavior of feedback trading weakened during the Korean crisis in the late 1990s, and, hence, foreign investors are not to blame for enforcing the destabilization of markets during crises periods. One concern about comovements in returns and volatility spillovers is that such cross-market relationships might at times be unwarranted by fundamentals in each market. For example, Kaminsky, Lyons, and Schmukler (2000) demonstrate that mutual funds engage in what they call contagion trading, that is selling in one market when returns in another market are poor, and Edison and Warnock (2001) show that portfolio flows to emerging markets depend little on the fundamentals of these markets, but are related to US interest rates. The literature defines so called contagion to be an increase in correlations among asset returns during periods of crises 6

beyond what fundamentals would suggest.6 For example, Calvo and Reinhart (1996), Frankel and Schmukler (1998), and Bailey, Chan, and Chung (2000) present evidence that the correlation increased beyond fundamental determinants in the aftermath of the Mexican crisis in 1994. Yet, Forbes and Rigobon (2002) caution that increased volatilities around crises periods could falsely lead investigators to conclude that the correlation increased as it might simply be the consequence of strong transmission mechanisms that exist during more stable periods.

1.4

Essay Summary

Given the integration of financial markets and the large scale international portfolio flows, it is a natural question to investigate the existence, nature, and impact of market-wide liquidity in an international framework. The dissertation provides empirical evidence in two related essays. The market microstructure literature has extensively investigated how trading activity, inventory costs, and market conditions affect liquidity – roughly speaking the price concession an investor has to make to trade now versus tomorrow – and the price formation process of individual securities. One central argument, originally put forth by Amihud and Mendelson (1986), states that empirically observed asset prices reflect liquidity costs, and assets that are less liquid are traded at a discount. More recently, a number of studies have investigated whether liquidity constitutes a risk factor in the pricing of financial assets (see, among others, Pastor and Stambaugh 2003). They found for the US that unexpected market-wide changes in liquidity constitutes a domestically priced risk factor. 6

See Claessens and Forbes (2002) and Karolyi (2003).

7

The first dissertation essay is presented in chapter 2. It is related to the above discussed articles by employing three different international asset pricing models. In particular, after identifying country, industry, and global commonalities in liquidity of individual stocks, it analyzes the implication of such a global liquidity factor for the pricing of financial assets in an international framework for a sample from the United States, the United Kingdom, and Japan covering the period from 1980 to 2001. The results for three different monthly liquidity measures — based on daily return and trading volume data — suggest that individual stock liquidity exhibits commonalities within countries and industries and co-moves globally. Furthermore, global and country-specific commonalities dominate industry effects as the source of common variation in liquidity. The asset pricing analysis suggests that expected stock returns are cross-sectionally related to the sensitivity of returns to shocks in global liquidity and that global liquidity is a priced risk factor on the portfolio and the individual stock level. The hypotheses that the liquidity risk premiums are equal across countries and industries cannot be rejected. Moreover, the results are neither driven by time-varying levels of asset-specific liquidity, nor by observations from recently listed firms, for which liquidity and return processes are likely different. The second dissertation essay is presented in chapter 3. It investigates the properties of changes in market-wide stock liquidity for a sample of weekly information from 1990 to 2002 for five regional aggregates: Asia, Europe, North America, emerging Asia, and emerging America. The analysis is based on a measure that does not rely on trading volume information, but only on return data. This is an important

8

difference with the measures used in chapter 2, since thin trading activities, which might occur especially in emerging markets, biases the measures used in chapter 2. The analysis of market-wide aggregates is important to better understand liquidity. For example, to what extent can the same cross-market patterns discovered in returns also be found in liquidity? Even though it is clear that changes in liquidity affect the return of financial assets in developed markets, it is unclear how liquidity is related across markets and returns, and what drives market-wide liquidity. Is the cross-market relationship of liquidity simply reflecting a common underlying variable or is it a phenomenon by itself? The analysis offers evidence of significant cross-market relationships: liquidity shocks are contemporaneously correlated; shocks from Asia dynamically spread into North America and vice versa; shocks from Europe spread into Asia and vice versa; shocks from North America spread into all regions; and shocks from emerging markets are only transmitted between themselves. The investigation of volatility spillovers in liquidity reveals that liquidity shocks in any market increase volatility of liquidity in all other markets. An analysis of the joint dynamic relationship of liquidity and market-wide returns offers, however, only weak evidence that liquidity determines returns. But the results suggest that changes in market valuations do affect liquidity, and market-wide averages of individual stock volatilities and world net bond flows are further fundamental drivers of market-wide liquidity. Beyond these results, there is little evidence that equity flows and interest rates affect liquidity consistently across markets. Even though changes in liquidity can to some extent be explained by returns and other determinants, shocks to liquidity continue to be contemporaneously correlated 9

across markets. However, the empirical results from an application of extreme value theory offers evidence that extreme shocks to liquidity are asymmetrically correlated in the tail of the distribution. In particular, mostly negative extreme liquidity shocks are correlated between North America, Europe, and emerging America.

10

CHAPTER 2

IS THERE A GLOBAL LIQUIDITY FACTOR?

In this chapter, I address the question whether there exist commonalities in liquidity in an international context, and whether they are country- or industry-specific or of a global nature. Moreover, I analyze whether global liquidity is a priced risk factor in an international framework. The analyses are based on a sample containing daily observations from 1980 to 2001 for all stocks from the US, UK, and Japan. The results from the investigation of commonalities suggest that individual stock liquidity co-moves within countries and industries, as well as with global liquidity. Furthermore, two separate analyses that are based on decompositions of the aggregate measures show that a global and independent country factors dominate industry effects as the source of common variation in liquidity. The asset pricing analysis suggests that average stock returns are cross-sectionally related to the sensitivity of returns to shocks in global liquidity, and that global liquidity risk is priced internationally at the portfolio and the individual stock level. Moreover, the hypothesis that the liquidity risk premiums are equal across countries and industries cannot be rejected. Additional results suggest that the cross-sectional relationship between average returns and shocks to liquidity is not driven by time-varying expected liquidity,

11

and the traditional view where assets with lower expected liquidity level command higher expected returns finds some support. The integration of capital markets over the last quarter of a century allowed investors to substantially improve the trade-off between risk and return. However, the possibility to diversify beyond the domestic investment opportunity set also gave rise to new and additional factors that determine the cross-sectional distribution of returns. In integrated markets, assets are no longer priced in a domestic context but rather relative to international risk factors. Cross-border capital flows and coordinated monetary policies, both determinants of liquidity from a market microstructure perspective, raise the question whether global liquidity constitutes such a risk factor. To the extent that individual stock liquidity is driven by a common underlying factor, shocks to this factor generate market-wide effects, and if asset returns and marketwide liquidity are correlated, the source of common liquidity effects could constitute a non-diversifiable risk factor for which investors might demand a risk premium for bearing this risk. International commonalities in liquidity could arise from several sources. Grossman and Miller (1988) point out that market liquidity of an individual asset is the result of the interaction of a market making sector that balances the expected net return from offering immediacy for an asset and the demand of investors to trade now versus tomorrow. The market microstructure literature has extensively investigated how order flow and trading volume affect liquidity and the price formation process of individual securities focusing on two paradigms: the level of inventory cost for immediacy suppliers; and the degree of asymmetric information among market participants. Therefore, common factors that determine inventory cost and level across 12

many assets may induce international commonalities in liquidity through correlations in the supply of immediacy. For example, a dealer’s inventory level is directly related to trading volume which in turn is, at least partially, determined by international portfolio flows. Moreover, global factors that influence price volatilities and interest rates determine the cost of maintaining a market. Demand for immediacy, on the other hand, is related to portfolio decisions. If, for example, investors reallocate portfolios after a common shock to asset prices or interest rates, international liquidity effects could arise. Or, if asset values in one market experienced a negative shock and investors are required to satisfy margin calls they might rather liquidate assets with non-depressed values in another markets and hence create cross-border liquidity effects. However, there are not only rational explanations for time-varying liquidity. For example, Baker and Stein (2002) argue that irrational investors underreacting to information contained in order flow, induce a lower price impact and thereby boost liquidity in general. This implies, under short-sales constraints, that higher levels of liquidity can be associated with positive investor sentiments on the individual stock, or on the aggregate level. Whatever the underlying sources are, if individual stock liquidity is correlated across assets, commonalities may constitute a non-diversifiable and priced risk factor. The traditional view of liquidity as an asset specific issue, put forth by Amihud and Mendelson (1986), is that the risk adjusted expected return is related to the expected level of liquidity. For example, in order to induce an investor to hold an asset that is on average more difficult to resell, the seller has to offer a price concession, and such expected transaction costs are factored into the prices when assets are traded. Thus, the observed relative price changes are simply gross returns. But what about 13

shocks to liquidity? These shocks generate unexpected changes in asset prices and returns. As long as such changes cancel each other out across assets, investors are able to diversify away these idiosyncratic liquidity shocks. But if individual stock liquidity is driven by a common underlying factor, individual liquidity is correlated across stock and values can simultaneously be depressed for many assets. However, this implies that such liquidity shocks are non-diversifiable, that investors demand compensation for holding assets with values that are more sensitive to unexpected changes in liquidity and are willing to accept lower expected returns from assets that are less sensitive. In this paper, I address whether there exist commonalities in liquidity in an international context, and whether they are country- or industry-specific or of a global nature. Moreover, I analyze whether global liquidity is a priced risk factor in an international framework. The analyses are based on a sample containing daily observations from 1980 to 2001 for all stocks from the US, UK, and Japan.7 For each asset in the sample, I calculate several monthly liquidity measures based on daily data and aggregate them to form country, industry, and global liquidity measures. The time series of these monthly individual and aggregate liquidity measures constitute my working sample. This paper is related to the two strands of research on commonalities in liquidity and on the effects of liquidity on the cross-section of expected stock returns in domestic contexts.8 7

The time frame and country selection allow me to investigate the liquidity relation among the three largest financial markets each representing a different continent. By restricting the analyses to these developed countries, I can abstract from issues related to market liberalization and the degree of financial integration. 8

See, for example, Fleming (2001), Brandt, Edelen, and Kavajecz (2001), Schultz (2001), and Chordia, Sarkar, and Subrahmanyam (2003) for research on bond liquidity.

14

In order for liquidity to be a risk factor, shocks to liquidity need to be correlated across assets. Chordia, Roll, and Subrahmanyam (2000), Hasbrouck and Seppi (2001), and Huberman and Halka (2001) investigate whether liquidity of individual assets are related. Using principal component and canonical correlation analyses, Hasbrouck and Seppi (2001) analyze commonalities in order flows and returns for the 30 stocks in the Dow Jones Index using high-frequency data for 1994. They find one to two common factors in order flows and show that these factors explain roughly two-thirds of the commonality in returns. They also find, however, evidence that idiosyncratic liquidity strongly dominates the common liquidity factor in explaining returns. They argue that despite the existence of commonalities, liquidity risk seems unlikely to be a priced risk factor but rather an asset specific characteristic. Chordia, Roll, and Subrahmanyam (2000) and Huberman and Halka (2001) analyze the change in cross-sectional averages of daily liquidity measures derived from intra-day data for two different sets of NYSE stocks for 1992 and 1996, respectively. The first authors find that the daily relative changes in individual asset liquidity are strongly related to changes in market and industry aggregates. The second authors show that timeseries model innovations in average liquidity for mutually exclusive groups of stocks are correlated, which they interpret as evidence of the presence of a common liquidity factor. Given the inter-market focus, this work is also related to a recent paper by Chordia, Sarkar, and Subrahmanyam (2003). They investigate the relationship between liquidity in the US treasury notes market and an aggregated liquidity measure for commons stocks on the NYSE. Based on intra-day data from 1991 to 1998, they find similar liquidity commonalities in both markets, and show that the primitive factors 15

driving liquidity are monetary conditions and mutual fund flows. Their results are therefore related to this work. To the degree that monetary conditions and fund flows are country specific, they point to the existence of factors that drive liquidity within countries. And integration of financial markets, interdependence of monetary policy and cross-border capital flows suggest that commonalities in an international context may exist as well. The question of whether liquidity determines expected returns has been investigated in a large body of literature. Earlier papers analyze how asset-specific trading costs and liquidity impact the cross-section of expected returns. After controlling for risk, disparities in expected returns are related to differences in trading cost (for example see Amihud and Mendelson 1986, Brennan and Subrahmanyam 1996, Brennan, Chordia, and Subrahmanyam 1998, Datar, Naik, and Radcliffe 1998, Lesmond 2002). This research mainly analyzes liquidity as an asset characteristic that determines the cross-section of expected returns. Using a variety of liquidity measures based on either intra-day or daily data, these studies find that less liquid stocks have higher average returns as expected. Easley, Hvidkjaer, and O’Hara (2002) argue that not only liquidity, but also the probability of information-based trading commands a premium. They show for a large cross-section of NYSE-listed stocks that higher rates of returns are positively related to a measure for the probability of information-based trading. Taking liquidity one step closer to representing a priced risk factor, Amihud (2002) and Jones (2002) focus on the time-series aspects of aggregate liquidity. They document for the US a time-series relationship between liquidity and expected return on the market level. Acharya and Pedersen (2003) investigate asset returns net of stochastic liquidity cost in an overlapping generations model. Their results imply that 16

the cross-section of expected returns depends not only on the asset’s sensitivity to the market return but also on individual and market liquidity and the two covariances between these three factors. In particular, the model implies that assets which have depressed values when overall market liquidity is low need in equilibrium to compensate investors for holding these assets. Moreover, their model implies that market liquidity commands a positive risk premium. Using a liquidity measure proposed by Amihud (2002), they find for a sample of all common stocks on CRSP from 1962 through 1999 that liquidity risk is priced and that the covariance between individual liquidity and market returns is particularly important. Two more recent papers by Bekaert, Harvey, and Lundblad (2003) and Pastor and Stambaugh (2003) offers the conceptually very different view of liquidity as a common risk factor: correlated liquidity shocks constitute a non-diversifiable risk and investors demand a premium for bearing that risk in equilibrium. In particular, Bekaert, Harvey, and Lundblad (2003) investigate two different monthly liquidity measures for emerging markets for two different samples starting in the late 1980s and running through 2001. The first measure is the equity market turnover based on the IFC market indexes, and the second measure is the monthly average of the daily proportions of all firms in a country with a zero return day. They find that the second measure is a better predictor of future returns and that the level of liquidity increases with equity market liberalization. Moreover, they analyze the implications for pricing financial assets with liberalization taken into account and find no evidence that global liquidity is priced. However, they find that country-level liquidity risk matters. Market liquidity as a state variable in an asset pricing framework has also

17

been investigated by Pastor and Stambaugh (2003). Along the line of previous studies that document commonality in liquidity, they argue that changes in aggregate liquidity could be non-diversifiable and, hence, a priced risk factor. Based on daily observation for all NYSE and AMEX stocks from 1962 through 1999, they construct an aggregate monthly liquidity measure and show that monthly portfolio returns command a positive risk premium for the changes in this measure even after controlling for other systematic risk factors.9 This paper extends the first strand of research of commonality in liquidity in two dimensions. First, by investigating the relationship of monthly relative changes in individual asset liquidity to changes in a global aggregate, and second, by analyzing the sources of commonalities employing two approaches from the asset pricing literature. The paper extends the asset pricing strand of literature by investigating whether shocks to aggregate liquidity are a globally priced risk factor or whether only the level of liquidity is priced. The remainder of the paper is organized as follows: The next section discussions of the sample and liquidity measures used in the analyses. Section 2.2 introduces the methodology employed in analyzing commonalities in liquidity and discusses the results. Section 2.3 investigates the asset pricing implications of commonalities in liquidity. Section 2.4 contains the results for an alternative sample that excludes observations for firms within four years of their first listing, and in section 2.5 I offer some conclusions. 9

Their asset specific liquidity measure is based on the model by Campbell, Grossman, and Wang (1993) and is the first-order autocorrelation measure in returns conditional on signed volume.

18

2.1

Sample and Liquidity Measures

The sample includes daily observations from January 1, 1980 to December 31, 2001 for all stocks from the US, UK, and Japan that are traded on NYSE, AMEX, Nasdaq, the London Stock Exchange, and the Tokyo Stock Exchange. The sample of the US firms is from the CRSP database, the UK sample from Datastream, and the Japan sample from the PACAP database from 1980 to 1996 and from Datastream from 1997 to 2001. The sample includes all live and dead ordinary shares from the three data sources, but excludes Investment Trusts and Funds since their trading characteristics might differ from ordinary shares. Depositary Receipts are included if they are associated with an underlying asset traded in any of the three countries. For all remaining stocks the sample contains the company’s country of origin and industry classification, the daily return, the number of shares traded, and the number of shares outstanding.10 The industry classification is based on the seven sector FT Actuaries/Goldman Sachs International Equity Indexes as reported in Roll (1992): Financial Industry, Energy Industry, Utility Companies, Transport Industry, Consumer Goods, Capital Goods, Basic Industry. To construct monthly liquidity measures from daily observations, the available information for each stock during a given month must meet minimal requirements for the measure to reflect an average tendency not distorted by too few observations or outliers. A stock is excluded from the sample for any given month if it did not trade during at least 10 days of that month, or if the number of shares outstanding is smaller than 1,000,000. Furthermore, to avoid contamination from tick-size considerations, a stock is deleted from the sample for any given month if on the last day of the 10

DRs are assigned to the country of the underlying stock.

19

previous month the domestic currency price falls outside of the respective U5 and U10,000, the £1 and £2,500, or the $5 and $1,000 range.11 To remove the influence of merger and acquisition activities and stock net issuances, a stock is deleted from the sample for any given month if on any day of the given month its trading volume exceeds the number of shares outstanding, or there is a change in the number of shares outstanding. Based on daily observations, I construct for each stock i and month t several monthly liquidity measures, Li,t , following Amihud (2002) and the market microstructure literature. While these measures are less precise than those that are based on trade-by-trade data, they benefit from the availability for most assets and over longer time periods — a fact that has been stressed by Amihud (2002). The three measures, based on equally-weighted averages of daily observations, are the stock’s turnover, the normalized absolute price change, and the normalized absolute return. While the first measure has been used extensively to proxy for transaction cost and liquidity, the second and the third follow Amihud (2002). They are related to Kyle’s (1985) price impact measure λ reflecting the price impact that accompanies a certain trading volume. The more liquid a stock is the lower is the price impact given a certain trading volume. Hasbrouck (2003) compares several trading cost measures based on daily data to estimates from high-frequency data and finds that the measure suggested by Amihud (2002) has the highest correlation with TAQbased price impact measures. To be more precise, let Pi,dt , V Oi,dt , N Oi,dt , and Di,t be the stock price on day d in month t, the number of shares traded, the number of 11

Basing the exclusion on the final price of the previous month avoids introducing a survivorship bias in the data.

20

shares outstanding , and the number of daily observations in month t, respectively.12 The stock’s turnover, T Oi,t , is calculated as T Oi,t

Di,t 1 X V Oi,dt = , Di,t d=1 N Oi,dt

(2.1)

g the normalized absolute price change, P V i,t , as

Di,t 1 1 X g , |Pi,dt − Pi,dt −1 | P V i,t = Di,t d=1 Pi,dt V Oi,dt

(2.2)

g i,t , as and the normalized absolute return, RV g i,t RV

Di,t 1 X |Pi,dt − Pi,dt −1 | 1 = . Di,t d=1 Pi,dt −1 Pi,dt V Oi,dt

(2.3)

In order to facilitate the interpretation I transform the latter two measures to quantify liquidity instead of illiquidity by taking the natural log of the inverse plus one.13 The transformations ensure that the measures are bounded below by zero, are linearly increasing in liquidity and are not artificially skewed by taking the inverse. In particular, the transformed measures are −1 g P Vi,t = log(1 + P V i,t ),

(2.4)

and g −1 RVi,t = log(1 + RV i,t ).

(2.5)

To analyze commonalities in liquidity, I aggregate the individual liquidity measures to obtain country, industry, and global liquidity measures. To remove the 12

Trading volume on Nasdaq is overstate relative to other markets since it is a dealers market. Atkins and Dyl (1997) show that trading volume falls by about 35% when a stock moves from Nasdaq to NYSE. All the results in the paper are very similar if I scale trading volume for all stocks on Nasdaq by 2/3. 13

The results in the subsequent analyses are qualitatively the same using the illiquidity measures in lieu of the transforms.

21

influence of outliers and measurement errors, I truncate the sample of individual liquidity measures at the 5% and 95% percentile. Following Chordia, Roll, and Subrahmanyam (2000) the aggregate measures La,t , a ∈ {c, f, w} are calculated as the equally-weighted average of the individual monthly liquidity measures La,t

Nt 1 X = Li,t , Nt i=1

(2.6)

where c, f, w refers to the country, industry, and global aggregate, respectively. The analyses below are, for the most part, based on relative changes in individual and aggregate liquidity which are calculated as DLi,t =

Li,t − Li,t−1 Li,t−1

and

DLa,t =

La,t − La,t−1 , La,t−1

(2.7)

respectively. Panel A in table A.1 offers an overview of the sample, panel B and C report summary statistics for the level of individual and aggregate liquidity measures and their proportional monthly change, respectively, and panel D shows the pairwise correlation measures between returns, relative changes in liquidity, and innovations in liquidity as calculated in chapter 2.3.1. Since the magnitude of the measures is arbitrary as they depend on the units in which the underlying variables are measured, the results are scaled for expositional purpose. Panel A shows that the US and the consumer, capital, financial, and basic industries have more observations than other countries and industries, with the average number of monthly observations following a similar distribution. The number of stocks in the different country-industry intersection ranges from 39 to 4,632 with an average number of monthly observations between 1,148 and 66,057. The aggregates, which are equally-weighted averages of the individual measures, are based on 460 up 22

to 19,724 stocks and cover all 264 months in the sample. Panel B reports the mean and standard deviation of the individual liquidity measures, as well as the mean and standard deviation of the various aggregates. While the dispersion in most of the individual stock liquidity levels is too large to render any mean significantly different from zero, the levels of the aggregates are statistically larger than zero. Panel C reports summary statistics of the relative changes in liquidity, which is the main focus of the paper. For each country and industry intersection, the average relative change in individual liquidity suggests an increase in liquidity for all liquidity measures. However, the large standard deviations indicate that none of those means are statistically different from zero but rather that their distributions have large crosssectional or time-series dispersion. The relative changes in the aggregates offer mixed results with different implications for different measures. However, as with individual liquidity, their large dispersions imply that no mean is statistically different from zero. Panel D reports the pairwise correlation measures between returns, changes in liquidity, and innovations in liquidity. While changes and innovations in liquidity of the first two measures, T O and P V , are positively and significantly correlated with returns, RV is significantly related only with the return of the US market. The correlation between the liquidity measures and the innovation measures are positive between T O and P V , as well as between P V and RV , but the correlation is negative for the pair T O and RV . This implies that while T O and P V as well as P V and RV measure similar aspects of liquidity, RV and the component of P V that is correlated with RV must capture an effect that is not measured by turnover T O.

23

2.2

Commonalities in Liquidity

In this section I investigate international commonality by calculating simple measures of covariation of individual stock liquidity with aggregated liquidity measures. Moreover, I investigate the sources of commonality in liquidity via a decomposition of commonality into a common global component and into country and industry factors. While the first set of tests establishes the existence of commonalities in an international setting along the line of Chordia, Roll, and Subrahmanyam (2000), the second set explores the sources of commonalities using approaches that have been advocated in the asset pricing literature by Heston and Rouwenhorst (1994) and Griffin (2002). In particular, I analyze if there is global commonality, because if individual stock liquidity does not co-move worldwide, there can be no global liquidity factor relevant for asset pricing.

2.2.1

Contemporaneous Variation

I run simple time-series regressions for each stock by relating the monthly relative changes of individual liquidity to the monthly relative changes of different aggregate liquidity measures DLi,t = αi + βi DLia,t + εi,t ,

t = 1, . . . , Ti ,

(2.8)

where DLi,t is the relative change from month t − 1 to t in any of the liquidity variables Li,t (e.g. T Oi,t ) outlined above, and DLia,t is the concurrent change in the respective cross-sectional country, industry, or global aggregate of the same variable as described in equation (2.6), but excludes stock i so that the explanatory variable is slightly different for each stock’s time-series regression. This effectively removes 24

the mechanical constraint that the average coefficient βi across all stocks is exactly unity. As Chordia, Roll, and Subrahmanyam (2000) point out, the exclusion of stock i from the calculation of the aggregate liquidity measure is irrelevant in large samples and makes only a small difference in the coefficient of any individual regression, but those small differences can accumulate to a material total when averaged across equations. I analyze relative changes because I am interested in whether liquidity comoves across assets and to avoid possible spurious results if individual and aggregate liquidity are trending variables. Following Chordia, Roll, and Subrahmanyam (2000), each regression includes several additional variables intended to remove spurious dependencies: the contemporaneous and two lags of the world portfolio return, and the contemporaneous change in the stock’s return volatility. If there are commonalities in liquidity and individual liquidity moves in the same direction for a majority of stocks, I expect that the cross-sectional average of the individual βi coefficients is positive. Moreover, I expect that the variation in the changes of aggregate liquidity has explanatory power in regression (2.8). Note that if individual stock liquidity is related to a common factor, but with opposite sign on the loadings, liquidity effects cancel each other out in the aggregate and investors can diversify away this liquidity risk. Table A.2 reports the cross-sectional, equally-weighted averages of the estimated βi , the percentage of coefficients that are positive, the percentage of coefficients that are significantly positive at the 5% level of a one-sided test that the coefficient is smaller or equal to zero, and the equally-weighted averages of the adjusted Ri2 from regression (2.8), using the three different aggregate liquidity measures individually as the right-hand side variable. The standard deviations of the average are calculated 25

under the assumption that the estimation errors in βi are independent across regressions and are based on the Lindeberg version of the Central Limit Theorem. Each average coefficient is statistically positive with between 76% and 59% of them having the correct sign.14 Moreover, the number of significantly positive coefficients is much larger than 5%, the size of the test, for all liquidity measures and aggregates. For comparison, the number of significantly negative coefficients is only between 2% and 6% and, hence, well within the range of the size of the one-sided test. The equallyweighted adjusted Ri2 ranges from 12% to 8%. These results are broadly comparable to the numbers found in Chordia, Roll, and Subrahmanyam (2000) for domestic data. The table also reports the cross-sectional, equally-weighted averages of the estimated βi , the percentage of coefficients that are positive, the percentage of coefficients that are significantly positive at the critical 5% level for each size quintile, where size is measured as the average US dollar market value of the company over its life in accordance with the liquidity measure. The degree of co-movement seems to be stronger for the intermediate quintiles for all aggregates and liquidity measures with the exception of the industry and global turnover aggregates. The regression results in this subsection offer strong support for the existence of commonalities among individual stock liquidity.

Thus far, whether these co-

movements in liquidity are driven by country-, industry-, or global factors is unclear. 14

Another way to gauge the significance of the results is by ignoring the size of the coefficient and to assume that each coefficient is independent and randomly chose to have a positive or a negative sign with probability π. Under this assumption, the number of positive coefficients is binomially distributed with mean nπ and variance nπ(1 − π). From this it follows that the probability of observing any of the fractions of positive coefficients in table A.2, which is a point estimate for π, is practically zero under the null hypothesis π = 1/2.

26

2.2.2

Sources of Commonalities

In order to investigate the sources of commonality, I follow the asset pricing literature and employ two different approaches. The first approach follows the idea in Griffin (2002), who investigates whether asset pricing factors are global or country specific. The second approach borrows from a decomposition of stock returns into country and industry effects proposed by Heston and Rouwenhorst (1994). For the first approach, I decompose for each industry f and country c the global liquidity measure into four mutually exclusive aggregates, which sum up to global aggregate. The four aggregates are the joint country and industry component, Lf c,t ; the pure country component, Lnc,t ; the pure industry component, Lf n,t ; and the remainder, Lnn,t . They are for a particular combination f and c defined via P Lw,t = N1 N Li,t P P P N P N L L L = Nf c F ∩C Ni,t + NNnn N \(F ∪C) + Nf n F \C Nfi,tn + NNnc C\F Ni,t nc fc = wf c Lf c,t + wf n Lf n,t + wnc Lnc,t + wnn Lnn,t ,

Li,t Nnn

(2.9) where F and C are the sets containing all stocks associated with industry f and country c, respectively, and N is the set of all assets. Since the sample consists of 3 countries and 7 industries, I obtain a total of 21 sets with each containing a different set of four aggregates. This decomposition has the advantage of circumventing the problem of high correlation among country, industry, and global aggregates that mechanically arise from averaging over non-disjoint sets.

27

Analogously, the relative change in global liquidity, DLw,t , can be decomposed into a weighted average of changes in the mutually exclusive aggregates15 DLw,t = vf c DLf c,t + vf n DLf n,t + vnc DLnc,t + vnn DLnn,t .

(2.10)

I replace global liquidity in equation (2.8) with the decomposition in (2.10) and analyze the cross-section of the least-square estimates of the β.,i coefficients in DLi,t = αi + βf c (vf c DLf c,t ) + βf n,i (vf n DLf n,t ) + βnc,i (vnc DLnc,t ) + βnn,i (vnn DLnn,t ) + εi,t .

(2.11)

The results of the analysis depends on the relative importance of commonalities within industries and countries and on the extent to which they are global. There are several cases that can be distinguished. If individual stock liquidity is related within each industry (country), I expect the cross-sectional average of the individual βf n,i (βnc,i ) coefficients to be on average larger than zero. If individual stock liquidity is related only to a global factor with no country- or industry-specific effects, I expect the cross-sectional average of the individual coefficients to be positive and equal. The case where individual liquidity is driven by a global factor and independent industry (country) factors implies that the averages of βnn,i and βf n,i (βnc,i ) should be positive. Finally, the case where individual stock liquidity is related within industries and within countries as well as to a global liquidity factor implies that all cross-sectional averages are larger than zero. Table A.3 reports the equally-weighted averages of the estimated βf n,i , βnc,i , βnn,i and the adjusted Ri2 from regression (2.11). The averages are again positive for all components. The number of positive and significantly positive coefficients is only 15

If the weights, w, in equation (2.9) are time-varying, then this is only a first order approximation. Furthermore, the first difference of each component is divided by the global liquidity measure to ensure consistency.

28

slightly lower for the global aggregate compared to the individual analyses above, but dropped dramatically for the country and industry aggregates. These results suggest that the global aggregate is the most important source of commonalities in individual stock liquidity. The average adjusted R2 are between 12% to 10%. The results for the size quintiles in table A.3 offer additional insight. Only 8 out of 15 coefficient averages for the industry aggregate are significantly different from zero, which is also reflected in the low number of positive and significantly positive individual coefficients. Moreover, the hump-shaped pattern across size quintiles discovered in the individual analysis above disappeared for two of the three liquidity measures at the country level. For the global aggregate on the other hand, the results and pattern from the individual analysis above continue to hold. The second approach investigates the importance of country and industry effects in commonalities. The methodology follows Heston and Rouwenhorst (1994), who analyze whether country or industry effects dominate the cross-sectional difference in stock return volatility (see also Heston and Rouwenhorst (1995) and Griffin and Karolyi (1998)). In particular, I postulate the following model for the relative change in liquidity of the ith security that belongs to industry f and country c DLi,t = αt + βf t + γct + εi,t ,

(2.12)

where αt is a common, base level change in liquidity in month t, βf t is the industry effect, γct is the country effect, and εi,t is a stock specific disturbance. Defining an industry dummy If,i that is equal to one if the stock i belongs to industry f and zero otherwise, and a country dummy Cc,i that is equal to one if security i belongs to

29

country c and zero otherwise, equation (2.12) can be rewritten for each month t as DLi = α + β1 I1,i + β2 I2,i + β3 I3,i + β4 I4,i + β5 I5,i + β6 I6,i + β7 I7,i +γ1 C1,i + γ2 C2,i + γ3 C3,i + εi .

(2.13)

As Heston and Rouwenhorst (1994) point out, it is not possible to estimate (2.13) by cross-sectional regressions because of perfect multicollinearity between the regressors. Since each stock belongs to one country and one industry, only cross-sectional differences between industries and cross-sectional differences between countries can be measured. Rather than choosing an arbitrary benchmark, it is more natural to measure the industry and country effects relative to the equally-weighted relative change in stock liquidity, which I interpret as a common, global change in liquidity for all individual stocks in a given month. To implement this definition, I follow Heston and Rouwenhorst (1994) and impose the following restrictions for each cross-sectional regression in month t

P7 n β = 0, P3f =1 f f c=1 mc γc = 0,

(2.14)

where nf and mc denote the number of assets in industry f and country c, respectively. The estimation results allow me to decompose DLc , the equally-weighted average of the relative change liquidity for all stocks of country c, into a global effect common to all countries α ˆ , the average industry effect of securities in that country, and a country-specific component γˆc 7 1 XX ˆ βf If,i + γˆc , DLc = α ˆ+ mc i f =1

(2.15)

where the i-summation is taken over the stocks in country c. Similarly, each equallyweighted average of the relative change in liquidity of all stocks from industry f , DLf , can be decomposed into a component that is common to all industries, α ˆ , the 30

weighted average of several country components, and an industry-specific component, βˆf DLf = α ˆ + βˆf +

3 1 XX γˆc Cc,i , nf i c=1

(2.16)

where the i-summation is taken over stocks in industry f . Note that an equallyweighted average of the relative change in liquidity can be calculated over any set of stocks, but the specific choice of country and industry indexes has the appeal that the pure effects can directly be compared to the global component and the contribution from the particular industry and country composition. Moreover, the logic follows the decomposition employed in the first approach. The regression (2.13) produces estimates of the global component and country and industry effects for one particular month. By running the cross-sectional regression for every month, I obtain a time series of the global component and the country and industry effects and, hence, of the decomposition of country and industry averages. Table A.4 presents the variances and variance ratios from the decomposition. The first column contains the variance of the global component, α ˆ , and its ratio to the variances of the country and industry indexes. The second and third column report the variances of the country and industry components and their ratios to the variances of the country and industry indexes. Panel A shows that between 40% and 47% of the variation in the equally-weighted country indexes is explained by the global factor, while the pure country effect accounts for about 60% of the variation. The industry composition of the countries explains basically nothing in the variation of the indexes. Panel B reports the result for the industry averages. Between 75% and 84% of the variation in the seven industry aggregates is explained by the global component and roughly another 30% of the variation is related to pure industry 31

effects. The fraction the weighted average of the country components explains is between 5% and 7%, which is far more than the combined industry effect explains in the country averages. The results have two implications. First, the variation in country and industry liquidity is strongly related to a global factor, and second, country effects are much more pervasive than industry effects. Following Heston and Rouwenhorst (1994), I interpret these results to show that individual liquidity is dominated by a global factor and independent country effects. The evidence in this section suggests that each stock’s liquidity is driven by at least two dominant sources, a global factor and a country factor. The results are consistent with a world where liquidity co-moves within countries and where liquidity is related to a common global factor. The next section explores the implication of these findings for the cross-section of expected returns.

2.3

Asset Pricing Implications

The previous section offered evidence supportive of the existence of commonalities in individual stock liquidity within countries and across industries. The fact that individual liquidity co-moves globally implies that investors are unable to diversify away this risk and, hence, that shocks to aggregate liquidity could command a positive risk premium. I take a standard international asset pricing model for globally integrated markets as my benchmark (see, for example, Karolyi and Stulz 2003b) and investigate the implication for the pricing of portfolios of stocks. First, I explore the nature of liquidity as a priced state variable in Merton’s (1973) intertemporal CAPM by relating the expected returns to shocks in aggregate global liquidity. Second, I relate the cross-section of expected returns to the level of expected liquidity and address the

32

question whether it is rather the traditional view of liquidity, as put forth by Amihud and Mendelson (1986), that explains the cross-sectional differences in returns.

2.3.1

Liquidity Risk

Extending Pastor and Stambaugh (2003), I investigate whether liquidity is a priced risk factor with a constant premium across countries as well as industries in a GMM framework. Standard asset pricing models suggest that if liquidity is a priced risk factor and financial markets are integrated, the risk premium on different test assets should be equal. The test assets are the equally-weighted monthly US dollar country and industry portfolio returns. The pricing models that I use to test the implications are: a World CAPM, which uses the world portfolio return, and the innovation in global liquidity as the only two factors; a standard International CAPM with the world portfolio return, the change in the log trade-weighted US dollar exchange rate index, which accounts for exchange rate risk, and the innovation in liquidity; and the standard International CAPM augmented with the returns from the US SMB and HML portfolios introduced by Fama and French (1993). To construct innovations in liquidity I estimate the regression DLw,t = a + bDLw,t−1 + cLw,t−1 + εt

(2.17)

over the full sample and use the residuals as the innovations in liquidity U Lt := εt . This approach is similar to the methods employed by Pastor and Stambaugh (2003) and Acharya and Pedersen (2003). Let xt be a 10 × 1 vector containing for month t the equally-weighted country and industry portfolio US dollar returns in excess of the three month US Treasury rate, F t a 1 × k vector with the realized returns on k traded factor portfolios, and U Lt the 33

innovation in liquidity. I investigate whether liquidity risk is a priced state variable in the following multifactor pricing model E[xt ] = BλF + bλU L ,

(2.18)

where λF and λU L are the risk premia associated with the traded factor portfolios and the innovation in aggregate liquidity, respectively. The pricing equation (2.18) is based on the assumption that the returns are generated from the following multivariate factor model xt = b0 + BF t + bU Lt + εt ,

(2.19)

where B and b contain the corresponding factor sensitivities. Equation (2.19) with the restriction b0 = λU L − E[U Lt ] implied by (2.18) can be estimated using GMM, an approach that produces robust estimates under general error term structures.16 In particular, letting θ contain the unknown parameters B, b, λF , λU L , and E[U Lt ] equations (2.19) and (2.18) imply the orthogonality conditions E[ht (θ)] = E

·

et ⊗ Z t U Lt − E[U Lt ]

¸

= 0,

(2.20)

where Z t = [1 F ′t U Lt ]′ et = xt − b[λU L − E[U Lt ]] − BF t − bU Lt . Table A.5 reports for the three different factor model specifications the liquidity risk premium estimates and model test statistics.17 The model specification is not 16

See, for example, Pastor and Stambaugh (2003), or Hansen (1982) for details about the estimation procedure and the test of overidentifying restrictions. 17

The model test statistics are the χ2 –tests for overidentifying restrictions.

34

rejected at the 1% critical level for any liquidity measure, and the estimated liquidity risk premium is significant and fairly constant across all specifications, except for the augmented ICAPM with the normalized absolute return liquidity measure RV . The estimated premium λU L is positive, which implies that assets which have lower returns when global liquidity is low have to offer higher expected returns for investors to hold them. This result is consistent with the prediction of the theoretical model of Acharya and Pedersen (2003), in which investors demand a positive risk premium for stocks that exhibit a positive (negative) correlation between returns and market-wide liquidity (transaction cost). The magnitude of the estimated risk premiums depends on the arbitrary scaling of the liquidity measures. But, as Pastor and Stambaugh (2003) point out, the scaling does not affect t-statistics or the product of b and λU L , which is the contribution of liquidity risk to expected returns. Table A.5 reports the estimate of the contribution [ of liquidity risk to expected returns of the test assets, bλ U L , along with their standard deviations. The results show that the estimated contribution for the different test portfolios are fairly constant across liquidity measures and model specifications, with significantly positive contributions to the expected return of almost all industry portfolios, and the Japanese and US country portfolios. The contribution to the expected return of the UK portfolio is, albeit positive for all measures and model specification, only significantly positive for the normalized absolute price change using the world CAPM and the International CAPM. This result is consistent with the evidence in table A.4 that the variation in the equally-weighted average of the relative change in liquidity for all stocks from the UK is dominated by a pure country effect.

35

The country and industry portfolios used in the asset pricing analysis so far offer no answer to the concern that the results could be driven by a small stock effect. Panel A in table A.6 reports the average level of liquidity for the intersection of independent sorts of size and liquidity, and shows that liquidity increases within each size group on average about fourteenfold from the lowest to the highest quintile. On the other hand liquidity increases on average only by 18% from the smallest to the largest stocks from the same liquidity quintile. One reason for not finding any distinct pattern could be that firms migrate across size quintiles during their life and, therefore, obscure the results. In order to address this problem, I replace the test assets with the equallyweighted returns of 10 size portfolios that are formed each month t based on the the US dollar market capitalization in month t−1. The estimation results for the liquidity premium along with the contribution to the expected returns are reported in table A.7. All the estimated risk premiums are lower than in table A.5, but significant for all specifications and measures except, again, for the augmented ICAPM with the normalized absolute return liquidity measure RV . The pattern of the contribution to expected returns across all liquidity measures and specifications is fairly constant and hump-shaped. The fact that the intermediate size firms command the largest contribution confirms the pattern of different degrees of commonalities across the size quintiles reported earlier in tables A.2 and A.3 for the global aggregate. In order to increase the number of test assets, I reestimate (2.20) with the ICAPM model specification using portfolios based on the intersection of the 3 countries and the 7 industries, and portfolios based on the intersection of the 3 countries and 7 size percentiles for two sets of 21 test assets. The results reported in table A.8 are very similar to the results for the individual analyses and the liquidity risk premium 36

estimates are strongly significant for all liquidity measures. The contribution to the expected returns of the portfolios is significant for most test assets and liquidity measures, with the exception of most portfolios from the UK. The hump-shaped pattern is no longer present in the size portfolios for each country. But now small Japanese firms exhibit a larger contribution to expected returns from liquidity risk than large firms. The US size percentiles show an opposite pattern with small US firms receiving a larger contribution to expected returns from liquidity risk than large firms. One important feature of integrated capital markets is that a risk factor must command a premium that is the same across markets. Since the three markets in my sample — Japan, the UK, and the US — are financially integrated markets, global liquidity risk should be priced equally across those markets. In order to test this assertion, I reformulate the model to allow the liquidity risk premiums to differ across assets E[xt ] = BλF + B U L λU L ,

(2.21)

where B U L is a 10 × 10 diagonal matrix with bii , i = {JP, UK, US, Finance, Energy, Utilities, Transport, Consumer, Capital, Basic} on the diagonal and zeros on the off diagonal, and λU L is a 10×1 vector containing the asset specific global liquidity shock risk premiums. I reestimate (2.21) and investigate the following three hypotheses: are the three country risk premiums equal; are the risk premiums the same across the seven industries; and are the risk premiums equal across countries and industries. Table A.9 reports the respective Wald test statistics and their associated asymptotic probabilities based on the International CAPM. The hypotheses of equal liquidity risk premiums across countries, equal liquidity risk premiums across industries, 37

and constant premiums across countries and industries cannot be rejected for any liquidity measure at the 10% level.18

2.3.2

Expected Returns and Trading Cost

The market microstructure research along the line of Amihud and Mendelson (1986) suggests that the cross-sectional variation in expected returns, after controlling for common risk factors, is related to asset specific variation in expected liquidity. Panel B in table A.6 reports the equally-weighted average return for all stocks in the intersection of independent sorts of size and the level of liquidity, and shows that on average more liquid stocks have either higher returns or exhibit no clear pattern. This surprising result could be related to time-variation in expected liquidity if stocks migrate across liquidity quintiles. In order to investigate whether time-varying expected level of liquidity supports the traditional view put forth by Amihud and Mendelson (1986), I include the two-month-lag of the six-month average measure of asset spe¯ t−2 , in the pricing equation (2.18) and estimate the following pricing cific liquidity, L model19 ¯ t−2 + BλF + bλU L . E[xt ] = αL

(2.22)

Table A.10 reports the estimates of the liquidity risk premium along with the contributions of liquidity risk to expected returns, and the liquidity premium α based on the International CAPM specification. The imposed overidentifying restrictions imply that the model specification cannot be rejected for any of the liquidity measures. ˆ U L are statistically different from zero for all The estimated liquidity risk premiums λ 18

Note that given the results that liquidity risk contributes to the expected return of only some country and industry portfolios a test of equality might be difficult to reject. 19

I use a two month lag to avoid any correlation with the innovation in liquidity.

38

measures. Moreover, the introduction of the liquidity level variable increases the risk premium estimates for all measures. The estimated compensation for the level of liquidity α ˆ is negative for all measures, but only significant for the turnover T O. The negative signs imply that assets which are expected to be more liquid have lower expected returns, which is consistent with Amihud and Mendelson (1986).

2.3.3

Stock Level Fama-MacBeth Regressions

The evidence that global liquidity is a priced risk factor has been established so far at the portfolio level with a limited number of test assets. One way to increase the number of test assets is to conduct the analysis at the individual stock level. However, there is a drawback since individual stocks change character over time. For example, they increase or decrease in size, vary the exposure to risk factors, alter leverage, or even change the nature of the business. In order to address these issues, I test whether liquidity is priced at the individual stock level by performing a standard Fama-MacBeth analysis (for example, see Cochrane 2001) which accounts for such issues. In June each year from 1985 to 2001 I first sort all stocks into 10 deciles based on the firms’ market capitalization. I then subdivide each decile independently into 3 times 10 portfolios based on pre-ranking βs of the market, the foreign exchange, and the liquidity risk factor. The βs are jointly estimated for each stock over the previous 24 to 60 months by regressing the US dollar excess return on the three factors. After assigning the firms to the size-β portfolios in June, I calculate the equally-weighted return on the portfolios for the next 12 month from July to June. In the end, I have monthly returns from July 1985 to December 2001 on 3 times 100 portfolios formed

39

on size and each of the pre-ranking βs. I then estimate for each liquidity measure 3 sets of post-ranking βs for the full sample using the above multi-factor pricing model and assign the respective βs back to the stocks in each portfolio. In order to account for non-synchronous trading, the pre-and post-ranking βs are estimated as the sum of the slopes in the regressions of the returns on the current value of the factors, as well as one lead and one lag. Table A.11 reports the time-series averages of the slopes from the month-to-month Fama-MacBeth regressions of the cross-section of individual stock returns on the three post-ranking βs f l m + εi,t , i = 1, . . . , N + λf βi,t + λl βi,t xi,t = λm βi,t

(2.23)

f m with βi,t the post-ranking β-estimate for the market risk, βi,t the post-ranking βl estimate for the foreign exchange risk, and βi,t the post-ranking β-estimate for the

liquidity risk. Note that the βs may change over time for a particular stock if the stock switches from one pre-ranking decile portfolio to another over the sample period. The standard deviations are the time-series standard errors of the average slope coefficients. The estimated liquidity risk premium is positive and significant for all three liquidity measures and ranges from from 0.020 to 0.038. However, the estimated market and the foreign exchange risk premiums are negative, albeit not significant for most liquidity measures. In summary, the analyses of the asset pricing implications of liquidity offer support for the view that global liquidity is a priced state variable in an international framework. Moreover, there is supportive evidence that the premium is constant across the test assets, and the results suggest that it is not the time-varying level of 40

expected liquidity that drives the results. The results show also evidence that global liquidity is priced at the individual stock level.

2.4

Alternative Sample

Table A.1 shows that the US is dominating the sample, and it is natural to question whether the results are driven by the relative large number of young firms just past their IPO event. Since it is likely that liquidity and return processes are very different for this subset, it could be the post IPO performance together with the relative low turnover for such stocks that dominates the sample and results. In order to test this assertion, I exclude all observations for each firm within the first 4 years of the first record in the original sample. Tables A.12 through A.14 report the results for the commonality and asset pricing analyses based on this subsample along with the original estimates. The results for the investigation of commonality reveal that the exclusion of the recent IPO firms strengthens the evidence that there exist commonalities, and that there are at least two dominant sources, a global factor and a country factor. The results of the asset pricing analysis, the risk premiums and the contributions to expected returns of the test assets are also stronger than in the full sample. No model specification can be rejected, and the contribution to the expected return for the UK are now significantly positive as well. Moreover, the hump-shaped contribution to the size-sorted test assets prevails.

2.5

Conclusion

Individual liquidity appears to exhibit commonalities within the sample of this study and the findings are robust with respect to different liquidity measures employed

41

in this chapter. Analyzing the sources of commonalities reveals that individual stock liquidity is related to a common global factor and co-moves within countries and industries. Furthermore, the results suggest that the global liquidity factors and country effects dominate the commonality. The asset pricing analysis suggests that expected stock returns are cross-sectionally related to the sensitivity of returns to shocks in global liquidity, and that global liquidity is a priced state variable in an international framework at the portfolio as well as at the individual stock level. The hypotheses that the liquidity risk premiums are equal across countries and industries cannot be rejected for all liquidity measures. Moreover, time-varying levels of expected liquidity are not driving the results, and there exist some evidence for a cross-sectional relationship between expected returns and the level of expected liquidity. The results also hold for a sample that excludes observations for firms within the first 4 years after their IPO, for which liquidity and return processes are likely very different. I draw two main conclusions from the results in this paper. First, there exist independent country and global commonalities in liquidity. And second, liquidity risk is a priced state variable in a global context.

42

CHAPTER 3

ARE LIQUIDITY SHOCKS CORRELATED ACROSS EQUITY MARKETS?

The market microstructure literature has extensively investigated how trading activity and inventory cost affect liquidity and the price formation process of individual securities. One central argument, originally put forth by Amihud and Mendelson (1986), states that empirically observed asset prices reflect liquidity costs, and assets that are less liquid are traded at a discount. More recently, a number of studies have investigated whether liquidity constitutes a risk factor in the pricing of financial assets (see, among others, Pastor and Stambaugh (2003) and chapter 2 of this dissertation). They found that unexpected changes in market-wide liquidity constitutes a domestically and internationally priced risk factor. The cross-market dependence of market-wide liquidity among a broad sample of international equity markets and its implication for stock returns has not been investigated so far. This paper fills this void by investigating properties of market-wide liquidity, by identifying determinants in an international context, and by exploring dynamic relationships between liquidity and returns. The integration of financial markets, interdependence of monetary policy, and cross-border capital flows suggest that such relationships across markets may exist in an international context. 43

It is important to understand cross-market dependence in equity liquidity for several reasons. First, if an increase in liquidity in one market affects liquidity in another market, trading and portfolio allocation strategies that incorporate this correlation can lower transaction costs and increase returns. Hence, this is a relevant issue in light of Amihud and Mendelson’s (1986) argument. Second, the evidence presented in the international asset pricing literature that returns are correlated across stock markets raises the question whether the cross-market relationship of liquidity simply reflects other common variables. Hence, understanding what primitive factors drive liquidity sheds some light on the question whether the cross-market dependence of liquidity is a phenomenon by itself. And third, if liquidity shocks are positively correlated across markets around the world, unexpected changes in liquidity may not be diversifiable and global liquidity might constitute a risk factor for which investors demand a premium. Hence, understanding to what extent liquidity shocks and asset returns returns are related offers insight into the question question whether liquidity can constitute a priced risk factor in this international setting. This paper is naturally related to the existing literature that explores comovement in asset liquidity and that investigates whether liquidity is a risk factor. The question whether there exist commonalities in individual stock liquidity within the US has been investigated and documented, for example, by Chordia, Roll, and Subrahmanyam (2000), Hasbrouck and Seppi (2001), Huberman and Halka (2001). Using principal component and canonical correlation analyses, Hasbrouck and Seppi (2001) analyze commonalities in order flows and returns for the 30 stocks in the Dow Jones Index using high-frequency data for 1994 and find one to two common factors in order flows. Chordia, Roll, and Subrahmanyam (2000) and Huberman and Halka (2001) analyze 44

the change in cross-sectional averages of daily liquidity measures derived from intraday data for two different sets of NYSE stocks for 1992 and 1996, respectively. The first authors find that daily relative changes in individual asset liquidity are strongly related to changes in market and industry aggregates. The second authors show that time-series model innovations in average liquidity for mutually exclusive groups of stocks are correlated, which they interpret as evidence of the presence of a common liquidity factor. Commonalities in individual stock liquidity in an international context have been investigated in chapter 2 of this dissertation. There I show that relative changes in individual asset liquidity are strongly related to changes in country and global aggregates for monthly measures based on daily data from 1980 to 2001 for the US, the UK, and the Japanese equity markets. The general result from the studies of commonalities in individual stock liquidity is that liquidity is correlated across assets within markets and across markets. Cross-market movements in market-wide liquidity between the US equity market and the US Treasury notes market have been explored and documented by Chordia, Sarkar, and Subrahmanyam (2003). They investigate the relationship between liquidity in the US treasury notes market and an aggregated liquidity measure for common stocks on the NYSE based on intra-day data from 1991 to 1998. They find similar liquidity commonalities in both markets and show that the primitive factors driving liquidity are monetary conditions and mutual fund flows. Determinants of market-wide liquidity in the US have also been investigated by Chordia, Roll, and Subrahmanyam (2001). Based on high-frequency data for a comprehensive sample of stocks from the NYSE and different bid–ask spread measures, 45

they find that the change in a daily market-wide liquidity measure is significantly related to the contemporaneous market return, the absolute market return, a short-term interest rate, and the US term premium. Financial markets need to be integrated in order for any liquidity shock to be transmitted across markets. The empirical literature on market integration provides evidence that risk premia are determined globally and many markets are integrated.20 In such a world it is natural to observe that stock prices around the world tend to exhibit some comovement, and a number of studies investigate the contemporaneous correlation among equity markets and lead-lag pattern in returns and volatilities. For example, Longin and Solnik (1995) and Bekaert and Harvey (2000) document general patterns and time-variation in international stock market correlations, and Eun and Shim (1989), Engle, Ito, and Lin (1990), and Hamao, Masulis, and Ng (1990) document lead and lag effects in returns across markets as well as spillovers in volatilities. There is a large literature that investigates the relationship between liquidity and returns.21 For example, Amihud (2002) focuses on the time-series aspects of aggregate liquidity. He documents for the US a time-series relationship between liquidity and expected return on the market level. In particular, he shows for a sample of NYSE stocks from 1964 to 1997 that changes in market-wide liquidity have a significantly negative effect on expected returns. Bekaert, Harvey, and Lundblad (2003) investigate a liquidity measure based on the proportion of zero daily firm returns averaged over 20

See Kaminsky and Schmukler (2001) for a comprehensive list of liberalization dates, and Bekaert and Harvey (1995) and Foerster and Karolyi (1999) for evidence of time-varying degrees of market integration for emerging markets. 21

See chapter 2 of this dissertation and, for example, Amihud and Mendelson (1986), Brennan and Subrahmanyam (1996), Brennan, Chordia, and Subrahmanyam (1998), Datar, Naik, and Radcliffe (1998), Lesmond (2002), Acharya and Pedersen (2003), and Pastor and Stambaugh (2003).

46

the month following Lesmond, Ogden, and Trzcinka (1999) and Lesmond (2002). In their emerging markets sample starting in the late 1980s and running through 2001, they find that liquidity is a predictor of future returns. However, when they analyze the implications for the pricing of financial assets they find no evidence that global liquidity is priced. However, they find that country-level liquidity risk matters. In this paper I analyze cross-regional and time-series properties of market-wide liquidity for five regional aggregates. The measures are calculated from individual stock information for a sample from 39 developed and emerging countries aggregated across all stocks in developed Asia, North America, Europe, emerging Asia, and emerging America. In particular, I attempt to answer the question whether similar cross-market patterns in returns as documented in the international asset pricing literature can be found in liquidity. Furthermore, I analyze in an international context the relationship between changes in liquidity and regional average returns and investigate determinants of liquidity. The results suggest that liquidity shocks are contemporaneously correlated and that shocks from Asia dynamically spread into North America and vice versa, that shocks from Europe spread into Asia and vice versa, that shocks from North America spread into all regions, and that shocks from emerging markets are only transmitted between themselves. An investigation of volatility spillovers in liquidity reveals that a liquidity shock in any market increases volatility of liquidity in all other markets. An exploration of the dynamic relationship between market-wide liquidity and returns offer weak evidence that changes in market-wide liquidity predicts market-wide returns. On the other hand, the results suggest that liquidity responds asymmetrically to market returns, and that the average of individual stock volatilities and world 47

net bond flows are fundamental drivers of market-wide liquidity. Furthermore, even though changes in liquidity can to some extent be explained by determinant variables, liquidity shocks continue to be contemporaneously correlated across markets. In light of the systemic risk of extreme liquidity shocks that occur simultaneously and irrespective of fundamentals, I analyze the correlation of shocks in liquidity in the tail of the distribution. The empirical results from an application of extreme value theory offers evidence that extreme shocks to liquidity are asymmetrically correlated in the tail of the distribution for some markets even after controlling for fundamental determinants. In particular, negative extreme liquidity shocks are correlated between North America, Europe, and emerging America. The paper is organized as follows: The second section discusses the sample and liquidity measure used in the analyses. Section 3.2 investigates dynamic and contemporaneous aspects of liquidity shocks. Section 3.3 explores the dynamic relationship between liquidity and returns as well as determinants of liquidity. Section 3.4 investigates the relationship across markets of extreme shocks in liquidity. Section 3.5 offers some conclusions.

3.1

Sample and Liquidity Measure

The sample includes daily observations for all stocks from 39 developed and emerging countries grouped into five regions: aggregate Asia includes Australia and Japan; Europe includes Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the UK; North America includes Canada and the US; emerging Asia includes China, Hong Kong, India, Indonesia, South Korea, Malaysia, Pakistan, the

48

Philippines, Singapore, Thailand and Taiwan; and emerging America includes Argentina, Brazil, Columbia, Chile, Mexico, Peru, and Venezuela. The sample covers the period from January 1, 1990 to January 2, 2002, and stocks are incorporated as soon as their information becomes available. The information for the US is from the CRSP database and the data for the Japanese stocks for the years from 1990 to 1996 is from the PACAP database. The data for all other stocks and dates is from Datastream. The sample includes all live and dead ordinary shares, but excludes Investment Trusts and Funds since their trading characteristics might differ from ordinary shares. Based on daily observations, I construct for each stock i and week t a weekly liquidity measure, Li,t , following Lesmond, Ogden, and Trzcinka (1999), Lesmond (2002), and Bekaert, Harvey, and Lundblad (2003) as the fraction of zero return trading days over the week. While this measure is less precise than others based on trade-by-trade data, it has several major advantages. First, the measure is available for a stock as long as there is return information available. Second, there is no need for positive trading volume on a given day in the week to include that day in the weekly average. This is important for less liquid markets and, hence, in particular for emerging markets where trading usually is thin. Measures that are functions of trading volume as, for example, the one proposed by Amihud (2002), will overestimate liquidity for such markets as such measures naturally only include information for days where trading volume is non-zero. And third, even if volume information was available, volume data is very challenging as it is plagued by trends and outliers and by trading and reporting systems that differ across markets.

49

The measure is related to implicit and explicit trading cost and is based on the argument that the marginal, informed investor will rationally trade only if the value of the accumulated information exceeds transaction costs. The larger the trading cost, the less likely information is incorporated into prices and volume and return is zero for that stock on that given day. Lesmond, Ogden, and Trzcinka (1999) show that the annual average of zero daily returns and the average NYSE/AMEX specialist spreads are strongly related, and Lesmond (2002) documents for a broad cross-section of emerging markets that the annual average of zero daily returns and hand-collected bid-ask spreads exhibit a correlation of about 72%. For the few markets where spread information is available on Datastream Bekaert, Harvey, and Lundblad (2003) show that the correlation of the monthly average of zero daily returns with bid-ask spreads is on average 67%. I calculate the measure as the number of zero return days over a week from Wednesday to Tuesday to circumvent any day-of-the-week problem. To be more precise, let Ri,dt and Di,t be the stock return on day d in week t and the number of daily observations in week t, respectively. The weekly individual stock liquidity measure is calculated as Li,t

Di,t 1 X 1{|Ri,dt |=0} = Di,t d=1

(3.1)

where 1{A} is an indicator function equal to 1 if the condition A is true and otherwise zero. To analyze market-wide liquidity, I aggregate the individual liquidity measures to obtain regional measures, Lr,t , calculated as the equally-weighted average of the individual weekly liquidity measures Lr,t

Nt 1 X Li,t , = Nt i=1

50

(3.2)

where r ∈ {Asia, Europe, North America, emerging Asia, emerging America}. In order to eliminate problems of spurious results due to non-stationarity, I follow Chordia, Roll, and Subrahmanyam (2001) and analyze throughout the paper changes in market-wide liquidity ∆Lr,t = −(Lr,t − Lr,t−1 ) with the negative taken only to facilitate the interpretation. Without reporting the results of KPSS stationarity tests, all market-wide liquidity measures contain a unitroot in levels but are stationary in the first differences.22

3.1.1

Sample Properties

Table A.15 reports summary statistics and pairwise correlation measures for the change in liquidity for the full sample and three subsamples covering financial calm and crises periods, respectively, and figure B.1 offers a graphical representation.23 The means for each region over the full sample are small relative to their standard deviations. The averages are all positive over the non-crises periods, negative over the Mexican crisis and mixed over the Asian/Russian crises. This suggests that liquidity generally increased in all markets during the 1990’s but declined during periods of financial turmoil. The large standard deviations indicate that there is substantial time variation in the measures, a fact that can easily be seen from figure B.1. However, there is no clear pattern in the standard deviation between non-crises and crises 22

The KPSS test, proposed by Kwiatkowski, Philips, Schmidt, and Shin (1990), tests a series by assuming stationarity under the null. The results from Augmented Dickey-Fuller tests, which assume non-stationarity as the null, are similar. 23

The crises periods cover the Mexican crisis from 7/1/1994 through 6/3/1995 and the Asian/Russian crises from 1/1/1997 through 3/31/1999.

51

periods. The table also reports skewness and excess kurtosis. All distributions show fat tails over the full sample as well as over the three subsamples. The pairwise correlation measures of raw changes in liquidity are reported in panel B. They indicate that liquidity on average moves in the same direction across all regions, and the correlation measures for the three subsamples show mostly an increase in correlations during the financial crisis periods.

3.2

Dynamic Transmission of Liquidity Shocks

In this section I investigate the dynamic transmission of liquidity shocks across regions in order to answer the question whether a shock to liquidity in one market is propagated into other markets. At this point I am only documenting the cross-market lead-lag response in liquidity and make no attempt to distinguish whether a shock is directly transmitted or whether the response is the result of an additional underlying variable that is driving liquidity across markets. I will investigate in section 3.3 whether variables, which are known to be correlated across markets, are the source of such cross-market correlation in liquidity. I address the question in a standard vector autoregressive framework ∆Lt = c +

p X

Ai ∆Lt−i + εt

(3.3)

i=1

where ∆L is a 5 × 1 vector containing the changes in liquidity for the five regions, c is a constant vector, and Ai are square coefficient matrices. I set the lag length to p = 3 by minimizing the Schwarz-Bayesian Information Criterion, BIC. Table A.16 reports the estimation results. The adjusted R2 with values ranging from 26% to 32% indicate that changes in liquidity are correlated across time and regions. All the main diagonal elements of Ai are statistically negative, but decrease in absolute 52

values, which implies that liquidity is mean reverting and shocks die out. The results also shed some light on the dynamic transmission of liquidity shocks across markets. Positive shocks in North America are positively transmitted into all other markets at lag one and into emerging Asia up to lag three. Furthermore, the results show some evidence of tradeoff effects in liquidity between Asia and the two regions of North America and Europe, as well as tradeoff effects between emerging America and the two regions of Europe and emerging Asia. In particular, a negative shock to Asia induces a subsequent positive response in liquidity in North America and Europe, and a negative shock in emerging America and Europe leads to increased liquidity in emerging Asia. In order to summarize the results from the vector autoregressive system, I explore whether observed shocks in one market contain information about future changes of liquidity in other markets. Table A.17 reports standard Granger-Causality test statistics along with their p-values. While North America Granger-causes all other markets, there is some evidence of feedback or bi-directional causality between North America and Asia, between Europe and Asia, and between emerging Asia and emerging America. The analysis so far investigated the dynamic transmission of liquidity shocks but did not answer the question whether liquidity shocks might be systemic in nature. The contemporaneous correlations between the disturbances of the vector autoregressive system, ε, summarize the cross-market linkage in liquidity shocks after taking the dynamic dependance into account. The estimated pairwise correlation matrix, i.e. the normalized covariance matrix Σ = E[εε′ ], is reported in table A.18. The correlations among market-wide liquidity shocks turn out to be generally larger compared 53

to the estimates in table A.15, and shocks to Europe and North America exhibit the strongest correlation. Europe generally exhibits the strongest correlations with other markets while Asia is the least correlated. With an average correlation of 29.8% the results indicate that market-wide liquidity shocks are correlated across markets and, hence, support the notion that liquidity shocks can be systemic and could command a risk premium. Moreover, the evidence that the correlation between emerging and developed markets is significant even at moderate levels of changes in liquidity indicates that portfolio diversification into emerging markets introduces liquidity risk. The analysis so far focused on the cross-market dependence of liquidity shocks in the mean. It is of considerable interest whether a liquidity shock in a particular market also increases the variance of liquidity shocks in other markets, since an increase in liquidity volatility might increase inventory risk for example for market makers. The question of the existence of meteor showers, that is whether shocks in one market affect volatility in other markets, has been well investigated for returns on assets that are simultaneously traded around the world (see, for example, Engle, Ito, and Lin 1990). I employ a similar framework using a multivariate GARCH model, where the results will also shed some light on whether the reported excess kurtosis in table A.15 panel A is due to time-varying variances. Engle and Kroner (1995) proposed a multivariate GARCH model specification known as BEKK, that guarantees symmetry and positive semi-definiteness of the covariance matrix while offering rich dynamics in the second moment. I estimate a BEKK(1,1) model with the mean equation given as in (3.3) and time-varying second moment of the following form Et−1 [εt ε′t ] = Σt = CC ′ + A(εt−1 ε′t−1 )A′ + BΣt−1 B,

54

(3.4)

where C a 5 × 5 lower triangular matrix, and A and B unrestricted 5 × 5 square matrices. This general specification allows for past shocks in one market to affect the variances of other markets as well. Let ε(i),t be the shock to liquidity in market i and ε(i,j),t = ε(i),t ε(j),t . For example, the dependance of the variance of market 1 on a shock to market 2 is given by v(1,1),t = a(1,2) a(1,2) ε(2,2),t−1 + 2a(1,2) a(1,1) ε(2,1),t−1 + 2a(1,2) a(1,3) ε(2,3),t−1 + 2a(1,2) a(1,4) ε(2,4),t−1 + 2a(1,2) a(1,5) ε(2,5),t−1

(3.5)

and the hypothesis that there exist no meteor showers from market 2 into market 1 can be assessed via a Likelihood-Ratio test of the hypothesis a(1,2) = 0. Without tabulating the results, all Ljung-Box tests for zero autocorrelation in the squared standardized residuals up to lag 12 indicate that the BEKK(1,1) specification is successful at modeling the time variation in the covariance matrix and the estimated coefficients of the mean equation are roughly comparable to the results in table A.16. Table A.19 reports the estimated A and B coefficient matrices of the variance equation. The significantly positive diagonal elements of A and B suggest the presence of ARCH and GARCH effects for most markets in the sense that their own liquidity shocks somewhat persistently increas future volatility. Panel C in table A.19 reports Likelihood Ratio Tests for the hypotheses, a(i,j) = 0, that a past liquidity shock in market j does not increase volatility in market i at time t. All Likelihood Ratio Test values clearly exceed the 5% level of significance, which indicate that liquidity shocks in every market increase liquidity volatility in all other markets.

55

3.3

Liquidity and Returns

Overall, the results suggest that shocks are correlated across markets and time, but the connection between liquidity and returns has not been explored so far. Furthermore, the cause for the contemporaneous and the lead-lag relationships in liquidity among the five regions is unclear. It is therefore uncertain whether the cross-market relationships of liquidity simply reflect common underlying variables or whether they are a phenomenon by themselves. This section first analyzes to what extent shocks in liquidity and returns are dynamically transmitted between each other. And second, it investigates to what extent the cross-market relationship documented in the previous sections is subsumed by the influence of regional market returns, return volatilities and other candidate determinants of liquidity. There is substantial evidence for a linkage between liquidity and asset prices. For example, Amihud and Mendelson (1986) point out that less liquidity stocks sell at a discount in order to reward investors for holding these stocks, and Pastor and Stambaugh (2003) and chapter 2 of this dissertation show that asset returns are related to unexpected changes in market-wide liquidity domestically and internationally. Also, there are a number of studies that investigate the contemporaneous correlation among equity market and lead-lag pattern in return levels and volatilities. While Longin and Solnik (1995) and Bekaert and Harvey (2000) document general patterns and timevariation in international stock market correlations, Eun and Shim (1989), Engle, Ito, and Lin (1990), and Hamao, Masulis, and Ng (1990) document lead and lag effects across markets in returns as well as spillovers in return volatilities. Hence, it is a

56

natural question to ask whether the observed correlations and spillovers in liquidity simply reflect the relationship of asset returns across markets and whether stock market liquidity constitutes a risk factor and affects returns. Unfortunately, given the relative short sample it is difficult to employ the asset pricing models used in chapter 2 to directly estimate the liquidity risk premium. However, if liquidity is priced, it must be the case that negative shocks to liquidity are positively correlated with contemporaneous return shocks and at the same time increase expected returns (See, for example, Bekaert, Harvey, and Lundblad 2003). There are several other variables besides returns and return volatilities that influence market-wide liquidity and are correlated across markets. For example, Chordia, Roll, and Subrahmanyam (2001) and Chordia, Sarkar, and Subrahmanyam (2003) present evidence that fund flows and monetary policies affect liquidity in domestic US equity and bonds markets.24 To account for flow effects, I include the regional net equity portfolio flows and global net bond portfolio flow from the TIC system of the US Treasury department as additional explanatory variables.25 Even though a positive equity flow presumably increases the demand for immediacy, the response of liquidity is unclear. In equilibrium, liquidity also depends on the degree of information asymmetry among market participants and the larger the degree of information asymmetry the lower is liquidity (see, among others, O’Hara 1995, Grossman and 24

See also Griffin, Nardari, and Stulz (2004) for a setting where equity flows are positively related to local and foreign stock returns, and Brennan and Cao (1997) for a model where equity flows are related to public information contained in local market returns. 25

TIC data is available on a monthly bases only. I calculate weekly net portfolio flows by assigning a equal proportion to each week in a given month such that they add up to the monthly figure reported in TIC. The net equity flow to the US is the negative of the sum of all flows from the US reported in the TIC. The global net bond portfolio flow is the sum of the net bond portfolio flows to each region. All equity flows are normalized by the contemporaneous regional equity market capitalization, and the global bond flows are normalized by Lehman Brother’s global bond index capitalization.

57

Miller 1988). Based on the evidence presented in Bohn and Tesar (1996), Seasholes (2000), and Froot, O’Connell, and Seasholes (2001) that foreign investors are feedback traders and less informed than domestic investors, the resulting effect of cross-border equity flows on liquidity is an empirical question. To account for monetary policy effects, I include changes in the 3-month US Treasury Bill rate and changes in the US term spread.

26

The market microstructure literature shows that not only information asymmetry determines stock liquidity, but also the cost of holding inventory. In particular, an increase in stock volatility renders market-making more risky and, hence, reduces individual stock liquidity.27 To control for such market-making risk on the aggregate level, I include regional averages of weekly measures of individual stock return volatility. I analyze whether stock market returns and changes in liquidity are dynamically related and whether changes in liquidity are related to determinants by extending the VAR framework introduced in section 3.2. Following Bekaert, Harvey, and Lundblad (2003), I estimate the following system of equations ∆Lt = cL + Rt

= cR +

P

P

ALi ∆Lt−i + AR i Rt−i +

P

P

bLi ⊙ |Rt−i | +

P

dLi ⊙ Rt−i + F X t + εLt (3.6)

R bR i ⊙ ∆Lt−i + εt

where ∆L is a 5 × 1 vector containing the changes in market-wide liquidity of all five regions, R is a 5 × 1 vector containing the equally-weighted average stock market 26

Using the Eurodollar interest rate instead of the Treasury Bill rate does not change the results qualitatively. See also Edison and Warnock (2001), who show that portfolio flows to emerging markets depend little on the fundamentals of these markets but are related to US interest rates. 27

See, for example, O’Hara (1995). See also Chordia, Shivakumar, and Subrahmanyam (2001) for an interpretation of absolute returns as a proxy for information flow.

58

returns, X contains the candidates of liquidity determinants, and ⊙ denotes elementby-element multiplication. I use the same number of lags for all the endogenous variables as in the VAR specification of section 3.2 to keep the system parsimonious and the estimates comparable. There are two testable hypotheses if liquidity is priced. First, shocks to liquidity must be positively correlated with contemporaneous return shocks, E[εR εL ] > 0, and second, negative shocks to liquidity must increase expected returns, bR < 0. I include the absolute return and the signed return in the liquidity equations, because positive and negative changes in the stock market may reduce liquidity but the response to shocks in the market valuation might be asymmetric. Hence, this is an extension of Bekaert, Harvey, and Lundblad’s (2003) setting which only includes returns. The parameter estimates are reported in table A.20 along with the adjusted R2 for each regression. The extension of the model improves the explanatory power for changes in liquidity to values between 34% and 42%, which corresponds to an increase of roughly 10% for each equation. The extension of the system, however, does not change the autoregressive coefficients of liquidity in any significant way. The main diagonal elements of ALi are still negative and significant, and the off-diagonal values are comparable to the numbers in table A.16. All coefficients on the absolute market returns are negative with nine out of fourteen significant at the 5% level. Only North America’s market liquidity seems not to be affected by valuation shocks. This result suggests that shocks to market valuation generally reduce liquidity. However, the positive coefficients on the return variables indicate that this response is asymmetric for some markets where positive shocks reduce liquidity far less than negative shocks.

59

The only two determinant variables that significantly affect liquidity are average stock volatilities and total net world bond flows. The coefficients on the average stock volatilities are significant and positive suggesting that an increase in average volatilities seems to proxy for an increase in average investor disagreement which in turn leads to an increase in liquidity rather than measuring an inventory cost component. The significant negative signs on the net world bond flows for Asia, emerging Asia, and emerging America suggests that stock market liquidity is adversely affected when investors shift capital into bond markets. Since neither North America nor Europe are affected, this could indicate a general flight to quality behavior. The results of the other determinants are less consistent and insignificant across markets. The explanatory power of the equations for market-wide returns, shown in table A.20, are lower with values between 2% and 16%. All markets show positive return autocorrelation up to at least lag two, and there is evidence of some cross-market relationships. The cross-equation coefficients of changes in liquidity are mostly negative implying that negative shocks to liquidity increase future expected returns as anticipated. However, only the coefficients for North America are significant. Table A.21 reports the pairwise correlations between the unexplained changes in liquidity and unexplained portion of returns. The correlation is positive and significant only for Asia and emerging America. However, the results for the pairwise correlation of shocks to liquidity indicates that, even after controlling for determinants, liquidity shocks are correlated across all regions with the highest correlation between Europe and North America and Europe and emerging America.

60

To summarize, while liquidity shocks are clearly correlated across regions even after controlling for other factors, there is only weak evidence for liquidity to be priced.

3.4

Correlation of Extreme Liquidity Shocks

So far, the results indicate that liquidity shocks are correlated across markets and time. Furthermore, the previous section offered evidence that changes in liquidity can be predicted with stock market returns and that changes in liquidity are determined by average volatilities and net world bond flows. However, the analysis has not been able to address the question whether there is a tendency for extreme shocks in liquidity to be correlated beyond what determinants suggest since standard correlation measures speak to the overall distributions only. There is an emerging body of literature that investigates whether there exist contagion among asset returns in different markets during crises periods, where contagion is defined as an increase in correlations beyond fundamental relationships (see, for example, Forbes and Rigobon 2002). However, as Karolyi (2003) points out, one of the limitation of this literature is that it focuses on correlation measures. Some recent studies have investigated the coincidence of extreme changes in returns across different markets without relying on linear correlation measures. For example, Longin and Solnik (2001) apply a multivariate extreme value analysis to the joint distribution of monthly stock returns and find that the correlation among large negative returns is much higher than what would be expected from a multivariate normal distribution. And Bae, Karolyi, and Stulz (2003) employ a multinomial logistic regression model to events of simultaneous extreme returns called coexceedances.

61

They show that coexceedances in one region can help explain coexceedances in other regions even after controlling for fundamental factors. Contagion in liquidity is an important issue by itself since liquidity can simultaneously dry up in several markets and possibly lead to a widespread financial crisis. In order to answer the question whether extreme shocks to liquidity are correlated beyond what fundamentals suggest, I follow the literature that investigates extreme correlations in returns (see, for example, Longin and Solnik 2001). In particular, I estimate bivariate peak over threshold models for each pair of residuals from the liquidity equations of the model (3.6) in section 3.3, which effectively filters out fundamental determinants of liquidity. The advantage of using this branch of extreme value theory is that no assumptions have to be made about the underlying distribution of liquidity since the approach relies on asymptotic arguments. Let ∆L be the unexplained change in liquidity for a single market as estimated in (3.6), F∆L the cumulative distribution function of ∆L, and define an extreme change in liquidity if ∆L exceeds a large threshold denoted by θ. The probability that ∆L > θ is given by p = 1 − F∆L (θ), and the cumulative tail distribution of θ-exceedance is defined by θ F∆L (x) =

F∆L (x) − F∆L (θ) 1 − F∆L (θ)

(3.7)

Extreme value theory shows that as θ tends to the upper endpoint of the distribuθ tion’s support, the only nondegenerate distribution that approximates F∆L (x) is the

generalized Pareto distribution 1/ξ

Gθ∆L (x) = 1 − [1 + ξ(x − θ)/σ]+

62

(3.8)

where the tail index ξ is intrinsic to the distribution and characterizes its tail, the dispersion parameter σ depends on θ, and the + operator gives the positive part of the expression in brackets. The extension to the bivariate case relies on the property of multivariate extreme value distributions that the dependence structure is preserved under monotone transformations of the marginal distributions (Smith 1994). This feature allows to separately model aspects of the tail behavior of each individual series and the dependence structure among the series.28 Letting ∆L = (∆L1 , ∆L2 ) and θ = (θ1 , θ2 ), the bivariate nondegenerate distribution that approximates the tail distribution of ∆L is of the following form Gθ∆L(x1 , x2 ) = exp[−D∆L(y1 , y2 )]

(3.9)

where D∆L is an unknown dependence function, and y1 and y2 are monotone transforms of the marginal distributions. A model commonly used in the literature (see, among others, Longin and Solnik 2001, Qi 2001) connecting corresponding marginal generalized Pareto distributions is the logistic function proposed by Gumbel (1961) with a dependence function given by −1/α

D∆L(y1 , y2 ) = (y1

−1/α α

+ y2

)

(3.10)

i (xi ). The parameter α controls the level of dependence beand yi = −1/ log Gθ∆L i

tween extreme shocks to liquidity and is related to the correlation coefficient ρ of extremes by ρ = 1 − α with the special cases of α = 1 and α = 0 corresponding to asymptotic independence and total dependence, respectively. It is important to note 28

The model for the dependence structure is commonly known as a copula function in the EVT literature (see, for example, Embrechts, McNeil, and Straumann 1999, Dehauvels 1994).

63

that asymptotic independence is not a property of only a few particular distributions. For example, two variables from a bivariate normal distribution are asymptotically independent as long as their correlation is different from ±1 (see, for example, Longin and Solnik 2001). Theoretically, the above results hold when the threshold θ approaches the respective upper endpoints of the joint distribution. In practice, since any sample contains only a finite number of observations, the threshold naturally needs to be finite as well. However, its choice is critical since a high value of θ leads to few observations and inefficient parameter estimates with large standard errors, and a low value introduces a bias in the estimation, since observations are included that do not belong to the tail of the distribution. Following Longin and Solnik (2001) and Qi (2001), I consider predetermined threshold levels corresponding to 200, 150, 100, and 50 positive and negative joint exceedances. Figure B.2 offers a graphical representation of the estimated tail correlations ρ, and table A.22 reports the results for the estimated tail correlations associated with the largest three threshold levels θ on the first line for each market pair. If the variables ∆L were bivariate normal with correlations different from ±1, the estimated tail correlations would naturally decrease with an increase in the threshold. Since the above theoretical results rely on large sample properties, I formally test whether the estimated tail correlations are within the 95% range of tail correlations estimated from 100 Monte Carlo simulations. In particular, I simulate 100 samples of 624 observations drawn from a bivariate normal distribution with matched means and covariances, report the average estimated ρsim on the second line in table A.22, and indicate whether ρ is above or below the 95% range of the simulation estimates. 64

The estimated ρ coefficients exhibit a general pattern of decreasing tail correlation for all but three market pairs as the threshold levels increase. For these three market pairs, Europe – North America, Europe – emerging America, North America – emerging America, the tail correlations for negative exceedances is above the 95% level of the Monte Carlo based estimates for the largest negative threshold level. Furthermore, judging by the plots in graph B.2 there is evidence that the tail correlation estimates are not only not falling off but rather increase with larger negative threshold levels. The peak over threshold analysis also allows to addresses the question whether the correlation of extreme shocks to liquidity differs for negative and positive shocks, since for symmetric distributions the tail correlations for positive and negative exceedance levels are naturally equal. The results in table A.22 and in graph B.2 clearly indicate that liquidity shocks between the three market pairs, Europe – North America, Europe – emerging America, North America – emerging America, are mostly correlated for negative shocks. The strong correlation in liquidity in the overall distribution between North America, Europe, and emerging America reported in table A.21 seems to be mostly driven by common large negative shocks.

3.5

Conclusion

In this paper, I analyze cross-regional and time-series properties of weekly marketwide liquidity measures from January 1, 1990 to January 2, 2002 for five regional aggregates. The aggregates, consisting of developed Asia, North America, Europe,

65

emerging Asia, and emerging America, are calculated from a sample covering 39 developed and emerging countries. Furthermore, I study the joint dynamics of liquidity and returns and investigate determinants of liquidity in an international context. The results suggest that liquidity shocks are contemporaneously correlated and that shocks from Asia dynamically spread into North America and vice versa, that shocks from Europe spread into Asia and vice versa, that shocks from North America spread into all regions, and that shocks from emerging markets are only transmitted between themselves. An investigation of volatility spillovers in liquidity reveals that a liquidity shock in any market increases volatility of liquidity in all other markets. The analysis of the relationship between liquidity and market-wide returns offers only weak evidence that liquidity determines returns in the sample. On the other hand, the results suggest that changes in market valuations does affect liquidity, and market-wide averages of individual stock volatilities and world net bond flows are further fundamental drivers of market-wide liquidity. There is little evidence that equity flows and interest rates affect liquidity consistently across markets. Even though changes in liquidity can to some extent be explained by returns and other determinants, shocks to liquidity continue to be contemporaneously correlated across markets. The empirical results from an application of extreme value theory offers evidence that extreme shocks to liquidity are asymmetrically correlated in the tail of the distribution. In particular, mostly negative extreme liquidity shocks are correlated between North America, Europe, and emerging America.

66

CHAPTER 4

CONCLUSIONS

This dissertation contributes to the finance literature in international asset pricing. It is related to papers that investigate commonalities in individual stock liquidity in the domestic US setting (see, for example, Chordia, Roll, and Subrahmanyam 2000), to research that estimates risk premia related to liquidity risk in the US (see, for example, Pastor and Stambaugh 2003), and to articles that explore properties and determinants of market-wide liquidity in the US (see, for example, Chordia, Roll, and Subrahmanyam 2001). However, it expands the scope of these papers in two related essays to an international setting. The first essay shows that monthly measures of individual stock liquidity exhibit commonalities for a sample from Japan, the UK, and the US from 1980 to 2001. Analyzing the sources of commonalities reveals that individual stock liquidity is related to a common global factor and co-moves within countries and industries. Furthermore, the results suggest that the global liquidity factor and country effects dominate the commonalities. An asset pricing analysis suggests that expected stock returns are cross-sectionally related to the sensitivity of returns to shocks in global liquidity, and that global liquidity is a priced state variable in an international framework at the portfolio, as 67

well as at the individual stock level. Hypotheses that the liquidity risk premiums are equal across countries and industries cannot be rejected for any liquidity measure. Moreover, time-varying levels of expected liquidity are not driving the results, and there exists some evidence for a cross-sectional relationship between expected returns and the level of expected liquidity. The results also hold for a sample that excludes observations for firms within the first four years after their IPO for which liquidity and return processes are likely very different. The second essay analyzes cross-regional and time-series properties of weekly market-wide liquidity measures from January 1, 1990 to January 2, 2002 for five regional aggregates. The aggregates, consisting of developed Asia, North America, Europe, emerging Asia, and emerging America, are calculated from a sample of daily return measures covering 39 developed and emerging countries. Furthermore, I investigate determinants of liquidity in an international context. The results suggest that liquidity shocks are contemporaneously correlated and that shocks from Asia dynamically spread into North America and vice versa, that shocks from Europe spread into Asia and vice versa, that shocks from North America spread into all regions, and that shocks from emerging markets are only transmitted between them. An investigation of volatility spillovers in liquidity reveals that a liquidity shock in any market increases volatility of liquidity in all other markets. An analysis of the relationship between liquidity and market-wide returns offers only weak evidence that liquidity determines returns. On the other hand, the results suggest that changes in market valuations does affect liquidity, and market-wide averages of individual stock volatilities and world net bond flows are further fundamental

68

drivers of market-wide liquidity. There is little evidence that equity flows and interest rates affect liquidity consistently across markets. Even though changes in liquidity can to some extent be explained by returns and other determinants, shocks to liquidity continue to be contemporaneously correlated across markets. The empirical results from an application of extreme value theory offers evidence that extreme shocks to liquidity are asymmetrically correlated in the tail of the distribution. In particular, mostly negative extreme liquidity shocks are correlated between North America, Europe, and emerging America. The overall conclusions from this dissertation are twofold. First, changes in global liquidity constitutes an international risk factor, and financial assets with returns that are more sensitive to this factor reward investors with higher expected returns. However, the contribution of liquidity risk to expected returns seems to be more relevant for developed markets. Second, market-wide liquidity is contemporaneously and dynamically related across regions. Furthermore, these relationships do not simply reflect other variables that are related across markets but constitute a phenomenon by themselves.

69

BIBLIOGRAPHY

Acharya, Viral V., and Lasse H. Pedersen, 2003, Asset Pricing with Liquidity Risk, Working paper, NYU Working Paper. Adler, Michael, and Bernard Dumas, 1983, International Portfolio Choice and Corporate Finance: A Synthesis, Journal of Finance 38. Amihud, Yakov, 2002, Illiquidity and Stock Returns: Cross-Section and Time-Series Effects, Journal of Financial Markets 5, 31–56. Amihud, Y., and H. Mendelson, 1986, Asset Pricing and the Bid-Ask Spread, Journal of Financial Economics 17, 223–249. Atkins, Allen B., and Edward A. Dyl, 1997, Market Structure and Reported Trading Volume: Nasdaq versus the NYSE, The Journal of Financial Research 10. Bae, Kee-Hong, Andrew Karolyi, and Rene Stulz, 2003, A New Approach to Financial Contagion, Review of Financial Studies 16. Bailey, W., K. Chan, and Y.P. Chung, 2000, Depositary Receipts, Country Funds, and the Peso Crash: The Intraday Evidence, Journal of Finance 55. Baker, Malcom, and Jeremy C. Stein, 2002, Market Liquidity as a Sentiment Indicator, Working paper, 8816 NBER. Bekaert, Geert, Campbell Harvey, and Christian Lundblad, 2003, Liquidity and Expected Returns: Lessons from Emerging Markets, Working paper, Duke University. Bekaert, Geert, and Campbell R. Harvey, 1995, Time-Varying World Market Integration, Journal of Finance 50. Bekaert, Geert, and Campbell R. Harvey, 2000, Capital Flows and the Behavior of Emerging Market Equity Returns, in S. Edwards, eds.: Capital Flows and the Emerging Economies (University of Chicago Press, Chicago, IL, USA ). Black, Fisher, 1974, International Capital Market Equilibrium with Investment Barriers, Journal of Financial Economics 1. 70

Bohn, H., and L. Tesar, 1996, U.S. Equity Investment in Foreign Markets: Portfolio Rebalancing or Return Chasing?, American Economic Review 86. Brandt, M., R. Edelen, and K. Kavajecz, 2001, Liquidity in the U.S. Treasury Market: Asymmetric Information, Inventory, and Supply Shifts, Working paper, University of Pennsylvania. Brennan, Michael, Tarun Chordia, and Avandihar Subrahmanyam, 1998, Alternative Factor Specifications, Security Characteristics, and the Cross-Section of Expected Stock Returns, Journal of Financial Economics 49, 354–373. Brennan, Michael, and Avanidhar Subrahmanyam, 1996, Market Microstrucure and Asset Pricing: On the Compensation for Illiquidity in Stock Returns, Journal of Financial Economics 41, 441–464. Brennan, M. J., and H. Cao, 1997, International Portfolio Investment Flows, Journal of Finance 52. Calvo, G., and C.M. Reinhart, 1996, Capital Flows to Latin America: Is There Evidence of a Contagion Effect?, in G. Calvo, M. Goldstein, and E. Hochreiter, eds.: Private Capital Flows to Emerging Markets After the Mexican Crisis (Institute for International Economics, Washington, DC ). Campbell, J.Y., S.J. Grossman, and J. Wang, 1993, Trading Volume and the Serial Correlation in Stock Returns, Quarterly Journal of Economics. Chan, K.C., Andrew Karolyi, and Ren´e Stulz, 1992, Global Financial Markets and the Risk Premium on U.S. Equity, Journal of Financial Economics 32. Choe, H., B. Kho, and R.M. Stulz, 1999, Do Foreign Investors Destabilize Stock Markets? The Korean Experience in 1997, Journal of Financial Economics 54. Chordia, Tarun, Richard Roll, and Avanidhar Subrahmanyam, 2000, Commonality in Liquidity, Journal of Financial Economics 56, 3–28. Chordia, Tarun, Richard Roll, and Avanidhar Subrahmanyam, 2001, Market Liquidity and Trading Activity, Journal of Finance 56. Chordia, Tarun, Asani Sarkar, and Avanidhar Subrahmanyam, 2003, An Empirical Analysis of Stock and Bond Market Liquidity, Working paper, Emory University, Federal Reserve Bank of New York, and UCLA. Chordia, Tarun, Lakshmanan Shivakumar, and Avanidhar Subrahmanyam, 2001, The Cross-Section of Daily Variation in Liquidity, Working paper, Goizuata Business School, LBS, and Anderson School of Business. 71

Claessens, S., and K. Forbes (eds.), 2002, International Financial Contagion. (Kluwer Academic Publisher New York). Cochrane, John, 2001, Asset Pricing. (Princeton University Press Princeton, New Jersey). Dahlquist, Magnus, and Torbjorn Sallstrom, 2002, An Evaluation of International Asset Pricing Models, Working paper, DP3145 CEPR. Datar, Vinay, Narayan Naik, and Robert Radcliffe, 1998, Liquidity and Stock Returns: An Alternative Test, Journal of Financial Markets 1, 203–219. Dehauvels, Paul, 1994, Probabilistic Aspects of Multivariate Extremes, in J. Tiago de Oliveira, eds.: Statistical Extremes and Applications (NATO ASI Series, Dordrecht ). Dumas, Bernard, and Bruno Solnik, 1995, The World Price of Foreign Exchange Risk, Journal of Finance 50. Easley, David, Soeren Hvidkjaer, and Maureen O’Hara, 2002, Is Information Risk a Determinant of Asset Returns?, Journal of Finance 57. Edison, Hali J., and Francis E. Warnock, 2001, A Simple Measure of the Intensity of Capital Controls, International finance discussion paper Board of Governors of the Federal Reserve System. Embrechts, Paul, Alexander McNeil, and Daniel Straumann, 1999, Correlation and Dependence in Risk Management: Properties and Pitfalls, Working paper, ETH Z¨ urich. Engle, R.F., T. Ito, and W. Lin, 1990, Meteor Showers or Heat Waves? Heteroskedastic Intra-Daily Volatility in the Foreign Exchange Market, Econometrica 58. Engle, R.F., and K.F. Kroner, 1995, Multivariate Simultaneous Generalized ARCH, Econometric Theory 11. Errunza, Vihang, and Etienne Losq, 1985, International Asset Pricing under Mild Segementation: Theory and Test, Journal of Finance 40. Eun, Cheol, and S. Janakiramanan, 1986, A Model of International Asset Pricing with a Constraint on Foreign Equity Ownership, Journal of Finance 41. Eun, Cheol, and Sangdal Shim, 1989, International Transmission of Stock Market Movements, Journal of Financial and Quantitative Studies 24. Fama, Eugene, and Kenneth French, 1993, Common Risk Factors in the Returns on Stocks and Bonds, Journal of Financial Economics 33. 72

Fama, Eugene, and Ken French, 1998, Value versus Growth: The International Evidence, Journal of Finance 53. Ferson, Wayne, and Campbell Harvey, 1993, The Risk and Predictability of International Equity Returns, Review of Financial Studies 6. Ferson, Wayne, and Campbell Harvey, 1994, The Sources of Risk and Expected Returns in Global Equity Markets, Journal of Banking and Finance 18. Ferson, Wayne, and Campbell Harvey, 1995, Predictability and Time-Varying Risk in World Equity Markets, Research in Finance 13. Ferson, Wayne, and Campbell Harvey, 1997, Fundamental Determinants of National Equity Market Returns: A Perspective on Conditional Asset Pricing, Journal of Banking and Finance 21. Fleming, M., 2001, Measuring Treasury Market Liquidity, Staff Report 33 The Federal Reserve Bank of New York. Foerster, Steven, and Andrew Karolyi, 1999, The Effects of Market Segmentation and Investor Recognition on Asset Prices: Evidence from Foreign Stock Listings in the United States, Journal of Finance 54. Forbes, Kristin, and Roberto Rigobon, 2002, No Contagion, Only Interdependence: Measuring Stock Market Comovements, Journal of Finance 57. Frankel, J., and S. Schmukler, 1998, Crises, Contagion, and Coutry Funds: Effects on East Asia and Latin America, in R. Glick, eds.: Management of Capital Flows and Exchange Rates: Lessons from the Pacific Rim (Cambridge University Press, Cambridge, MA ). Froot, K.A., P.G. O’Connell, and M.S. Seasholes, 2001, The portfolio flows of international investors, Journal of Financial Economics 59. Griffin, John, 2002, Are the Fama and French Factors Global or Country Specific?, Review of Financial Studies 15. Griffin, John, Frederico Nardari, and Rene Stulz, 2004, Daily Cross-Border Flows: Pushed or Pulled?, Review of Economics and Statistics forthcoming. Griffin, John M., and G. Andrew Karolyi, 1998, Another Look at the Role of the Industrial Structure of Markets for International Diversification Strategies, Journal of Financial Economics 50. Grossman, S.J., and M.H. Miller, 1988, Liquidity and Market Structure, Journal of Finance 43. 73

Gumbel, E.J., 1961, Multivariate Extremal Distributions, Bulletin de L’Institut International de Statistique 33. Hamao, Yasushi, Ronald Masulis, and Victor Ng, 1990, Correlation of Price Changes and Volatilities Across International Stock Markets, Review of Financial Studies 3. Hansen, Lars P., 1982, Large Sample Properties of Generalized Method of Moment Estimators, Econometrica 50, 1029–1054. Harvey, Campbell, 1991, The World Price of Covariance Risk, Journal of Finance 46. Harvey, C.R., 1995, Predictable Risk and Returns in Emerging Markets, Review of Financial Studies 8. Hasbrouck, Joel, 2003, Trading Costs and Returns for US Equities: The Evidence from Daily Data, Working paper, New York University. Hasbrouck, Joel, and Duane J. Seppi, 2001, Common Factors in Prices, Order Flows, and Liquidity, Journal of Financial Economics 59. Heston, Steven, and Geert Rouwenhorst, 1994, Does Industrial Structure Explain the Benefits of International Diversification?, Journal of Financial Economics 36. Heston, Steven, and Geert Rouwenhorst, 1995, Industry and Country Effects in International Stock Returns, Journal of Portfolio Management Spring. Huberman, Gur, and Dominika Halka, 2001, Systematic Liquidity, Journal of Financial Research. Jones, Charles M., 2002, A Century of Stock Market Liquidity and Trading Cost, Working paper, GSB, Columbia University. Jorion, Philippe, 1991, The Price of Exchange Rate Risk in the Stock Market, Journal of Financial and Quantitative Studies 26. Kaminsky, Graciela, Richard Lyons, and Sergio Schmukler, 2000, Managers, Investors, and Crises: Mutual Fund Strategies in Emerging Markets, Working paper, 7855 NBER. Kaminsky, Graciela, and Sergio Schmukler, 2001, On Booms and Crashes: Financial Liberalization and Stock Market Cycles, Working paper, World Bank. Karolyi, Andrew, 2003, Does International Financial Contagion Really Exist?, International Finance 6.

74

Karolyi, Andrew, and Ren´e Stulz (eds.), 2003a, International Capital Markets. (Edward Elgar Publishing, Inc. Northampton, MA, USA). Karolyi, G.A., and R.M. Stulz, 2003b, Are Financial Assets Priced Locally or Globally?, in G.M. Constantinides, M. Harris, and R.M. Stulz, eds.: Handbook of the Economics of Finance (North-Holland, New York ). Kwiatkowski, D., P.C.B. Philips, P. Schmidt, and Y. Shin, 1990, Testing the Null of Stationarity Against the Alternative of a Unit Root: How Sure are We That Economic Time Series Have a Unit Root?, Journal of Econometrics 54. Kyle, A., 1985, Continous Auctions and Insider Trading, Econometrica 53. Lesmond, David, 2002, Liquidity of Emerging Markets, Working paper, Tulane University. Lesmond, David, Joseph Ogden, and Charles Trzcinka, 1999, A New Estimate of Transaction Cost, Review of Financial Studies 12. Longin, Fran¸cois, and Bruno Solnik, 1995, Is the Correlation in International Equity Returns Constant: 1960-1990?, Journal of International Money and Finance 14. Longin, Francois, and Bruno Solnik, 2001, Extreme Correlation of International Equity Markets, Journal of Finance 56. Mark, Nelson, 1988, Time-Varying Betas and Risk Premia in the Pricing of Forward Foreign Exchange Contracts, Journal of Financial Economics 22. Merton, Robert C., 1973, An Intertemporal Capital Asset Pricing Model, Econometrica 41. O’Hara, Maureen, 1995, Market Microstructure Theory. (Blackwell Publisher Cambridge, MA, USA). Pastor, Lubos, and Roberto F. Stambaugh, 2003, Liquidity Risk and Expected Stock Returns, Journal of Political Economy 111. Qi, Rong, 2001, Return-Volume Relation in the Tail: Evidence from Six Emerging Markets, Working paper, Columbia Business School. Roll, Richard, 1992, Industrial Structure and The Comparative Behavior of International Stock Market Indices, Journal of Finance 47. Santis, Giorgio De, and Bruno G´erard, 1998, How Big is the Permium for Currency Risk?, Journal of Financial Economics 49.

75

Schultz, P., 2001, Corporate Bonds: A Peek Behind the Veil, Journal of Finance 56, 677–698. Seasholes, M. S., 2000, Smart Foreign Traders in Emerging Markets, Working paper, Harvard University. Sercu, Piet, 1980, A Generalization of the International Asset Pricing Model, Revue de l’Association Fran¸caise de Finance 1. Smith, Richard L., 1994, Multivariate Threshold Methods, in Janos Galambos, James Lechner, and Emil Simiu, eds.: Extreme Value Theory and Applications (Kluwer Academic Publishers, The Netherlands ). Solnik, Bruno, 1974, An Equilibrium Model of the International Capital Market, Journal of Economic Theory 8. Solnik, Bruno, 1983, International Arbitrage Pricing Theory, Journal of Finance 38. Stehle, Richard, 1977, An Empirical Test of the Alternative Hypothesis of National and International Pricing of Risky Assets, Journal of Finance 32. Stulz, Ren´e, 1981a, A Model of International Asset Pricing, Journal of Financial Economics 9. Stulz, Ren´e, 1981b, On the Effects of Barriers on International Investment, Journal of Finance 36.

76

APPENDIX A

TABLES

77

Panel A: Sample Total Number of Stocks Utilities Transport Consumer Capital

Finance

Energy

JP UK US

361 330 2,615

39 84 337

50 135 740

180 84 239

1,160 1,590 4,632

Aggregate

3,306

460

925

503

7,382

Average Number of Monthly Observations Utilities Transport Consumer Capital

Basic

Aggregate

1,148 1,338 5,403

1,321 1,504 16,444

4,590 1,318 4,557

27,680 20,986 66,057

32,767 6,650 46,476

31,388 5,599 32,135

264 264 264

264

264

264

264

264

264

264

264

Basic

Aggregate

Finance

Energy

1,129 495 2,679

1,051 336 1,458

3,970 3,054 12,700

9,715 4,278 45,814

4,303

2,845

19,724

Finance

Energy

Average Utilities Transport Consumer

Capital

Basic

Aggregate

Finance

Energy

Standard Deviation Utilities Transport Consumer

Capital

Basic

Aggregate

JP

TO PV RV

0.9960 3.1717 4.7743

1.3846 3.4069 4.5411

0.8605 3.5380 4.4982

1.2850 2.8146 3.8883

1.4427 2.8046 4.3367

1.5873 2.8071 4.2540

1.5680 3.0163 4.2977

1.5313 2.9417 4.4080

1.3435 1.3821 1.3371

1.6533 1.4966 1.6783

1.2288 1.5861 1.7543

1.6424 1.5396 1.7396

1.6928 1.3897 1.5641

1.8303 1.3639 1.5887

1.8159 1.4349 1.6190

0.4643 0.2421 0.5043

UK

TO PV RV

2.5902 3.6191 3.5297

2.7800 3.6296 3.3618

3.6771 4.0035 4.0875

2.7852 3.6391 4.0571

2.6591 3.6184 3.6194

2.6629 3.7066 3.7268

2.5945 3.9407 3.9311

2.7008 3.2948 3.0611

2.3569 1.4526 1.9107

2.4257 1.4877 1.9287

2.5001 1.5499 2.3173

2.2766 1.4303 1.9749

2.4000 1.4300 1.8363

2.3448 1.4037 1.8360

2.1124 1.3819 1.8307

1.0992 0.9947 1.4128

US

TO PV RV

2.5246 2.0329 1.1741

3.2112 2.6790 1.5444

2.5630 3.2611 2.2760

3.7689 2.6513 1.5239

3.6188 2.5687 1.5188

3.6266 2.5596 1.5386

2.8236 2.5801 1.6520

2.9957 2.4400 1.4640

2.2805 1.5350 1.4502

2.5963 1.5091 1.4474

2.3309 1.6001 1.6492

2.8718 1.5717 1.4526

2.8923 1.5269 1.4841

2.8484 1.5195 1.5497

2.3682 1.5581 1.6275

0.6675 0.3264 0.3206

Aggregate

TO PV RV

2.2672 2.3586 1.9865

2.7384 2.8660 2.1747

2.4138 3.2695 2.4043

2.5805 2.7752 2.8167

2.7368 2.6573 2.4400

2.6429 2.6816 2.7353

2.1969 2.8489 2.9976

2.5093 2.6992 2.5557

0.4136 0.2132 0.3319

0.5668 0.2565 0.3653

0.7266 0.3333 0.4104

0.5487 0.2593 0.3689

0.5805 0.3306 0.3612

0.5041 0.2398 0.3003

0.4135 0.2113 0.3076

0.4551 0.2168 0.2673

Finance

Energy

Average Utilities Transport Consumer

Capital

Basic

Aggregate

Finance

Energy

Standard Deviation Utilities Transport Consumer

Capital

Basic

Aggregate

JP

TO PV RV

0.1021 0.0285 0.0194

0.1066 0.0196 0.0169

0.0859 0.0227 0.0224

0.1069 0.0257 0.0251

0.1102 0.0263 0.0197

0.1186 0.0253 0.0192

0.1136 0.0236 0.0192

0.0223 0.0026 0.0012

0.5199 0.2292 0.1980

0.5488 0.2032 0.2007

0.4755 0.1949 0.2178

0.5453 0.2392 0.2387

0.5526 0.2345 0.2112

0.5717 0.2352 0.2125

0.5649 0.2265 0.2102

0.1910 0.0519 0.0372

UK

TO PV RV

0.1292 0.0313 0.0416

0.1109 0.0347 0.0448

0.0862 0.0168 0.0256

0.1342 0.0253 0.0402

0.1296 0.0319 0.0424

0.1378 0.0312 0.0445

0.1194 0.0288 0.0390

0.0404 0.0189 0.0207

0.5963 0.2503 0.3137

0.5665 0.2505 0.3140

0.5001 0.1997 0.2612

0.5974 0.2440 0.2876

0.6055 0.2505 0.3087

0.6119 0.2473 0.3077

0.5823 0.2385 0.2896

0.3602 0.2239 0.2352

US

TO PV RV

0.1192 0.0315 0.0579

0.1062 0.0229 0.0416

0.1020 0.0196 0.0359

0.1159 0.0250 0.0464

0.1078 0.0259 0.0473

0.1073 0.0258 0.0472

0.1068 0.0243 0.0437

-0.0058 -0.0043 0.0042

0.5717 0.2535 0.3576

0.5427 0.2207 0.3215

0.5143 0.1855 0.2634

0.5583 0.2261 0.3284

0.5531 0.2298 0.3325

0.5482 0.2270 0.3309

0.5395 0.2233 0.3174

0.1097 0.0885 0.1616

Aggregate

Panel B: Level

TO PV RV

-0.0028 0.0087 0.0388

-0.0015 0.0002 0.0215

-0.0040 0.0011 0.0146

0.0011 0.0040 0.0229

-0.0177 0.0030 0.0328

-0.0082 0.0032 0.0288

0.0022 0.0026 0.0161

-0.0138 0.0049 0.0320

0.1307 0.0734 0.1531

0.1743 0.0758 0.1424

0.1222 0.0684 0.0901

0.2136 0.0589 0.1108

0.1097 0.0460 0.0820

0.1603 0.0500 0.1114

0.1402 0.0466 0.0956

0.1114 0.0428 0.1004

Panel C: Relative Change

continued

Table A.1: The table reports sample size information, equally-weighted averages of the monthly stock liquidity measures in the respective intersection, as well as pairwise correlation measures with average market returns. The liquidity measures TO, PV, and RV are for each stock monthly averages using daily observations: TO represents turnover; PV absolute price change divided by trading volume in local currency; and RV absolute return divided by trading volume in local currency (see text for the details on how they are constructed). The relative changes are the monthly percentage change from month t-1 to month t for the individual stocks and the aggregates. JP, UK, and US refer to Japan, the United Kingdom, and the United States; the industry classification is from FT Actuaries/Goldman Sachs International Equity Index as reported in Roll (1992). The sample for levels and relative changes is truncated at the 5% and 95% range.

78

Table A.1 continued

Return

JP JP

.

UK

0.3639 [0.00] 0.1633 [0.01] 0.1794 [0.00] 0.2938 [0.00] 0.0712 [0.25] 0.1759 [0.00] 0.2588 [0.00] 0.0345 [0.58]

US

Liquidity

TO PV RV

Innovation

TO PV RV

Return UK 0.3639 [0.00] . 0.4737 [0.00] 0.1205 [0.05] 0.2085 [0.00] 0.0720 [0.25] 0.1395 [0.02] 0.1938 [0.00] 0.0416 [0.50]

Panel D: Pairwise Correlations Liquidity US TO PV 0.1633 [0.01] 0.4737 [0.00] . 0.1880 [0.00] 0.4111 [0.00] 0.1360 [0.03] 0.1768 [0.00] 0.3964 [0.00] 0.1292 [0.04]

0.1794 [0.00] 0.1205 [0.05] 0.1880 [0.00] . 0.3086 [0.00] -0.4238 [0.00] 0.9595 [0.00] 0.2864 [0.00] -0.3691 [0.00]

79

0.2938 [0.00] 0.2085 [0.00] 0.4111 [0.00] 0.3086 [0.00] . 0.4512 [0.00] 0.3142 [0.00] 0.9860 [0.00] 0.4493 [0.00]

RV

TO

0.0712 [0.25] 0.0720 [0.25] 0.1360 [0.03] -0.4238 [0.00] 0.4512 [0.00] .

0.1759 [0.00] 0.1395 [0.02] 0.1768 [0.00] 0.9595 [0.00] 0.3142 [0.00] -0.4045 [0.00] .

-0.4045 [0.00] 0.4510 [0.00] 0.9213 [0.00]

0.3188 [0.00] -0.3408 [0.00]

Innovations PV 0.2588 [0.00] 0.1938 [0.00] 0.3964 [0.00] 0.2864 [0.00] 0.9860 [0.00] 0.4510 [0.00] 0.3188 [0.00] . 0.4752 [0.00]

RV 0.0345 [0.58] 0.0416 [0.50] 0.1292 [0.04] -0.3691 [0.00] 0.4493 [0.00] 0.9213 [0.00] -0.3408 [0.00] 0.4752 [0.00] .

Global

Industry

Country

All beta std %+ sign +

0.544 0.010 75% 29%

adj R2

0.11

beta std %+ sign +

0.213 0.008 61% 15%

adj R2

0.10

beta std %+ sign +

0.195 0.009 59% 14%

adj R2

0.10

1

TO Size Quintile 2 3 4

5

All

0.467 0.028 69% 18%

0.567 0.025 72% 25%

0.585 0.022 78% 31%

0.550 0.018 79% 36%

0.529 0.021 77% 33%

0.526 0.007 76% 31%

0.266 0.022 62% 13%

0.246 0.020 60% 12%

0.228 0.017 61% 15%

0.179 0.015 61% 16%

0.153 0.015 61% 18%

0.509 0.010 73% 24%

0.251 0.023 61% 12%

0.220 0.022 58% 11%

0.200 0.019 59% 13%

0.164 0.016 58% 15%

0.147 0.016 60% 18%

0.541 0.011 72% 23%

PV Size Quintile 2 3 4

1

5

All

0.398 0.026 65% 18%

0.572 0.018 72% 27%

0.639 0.013 78% 36%

0.507 0.011 81% 37%

0.428 0.017 79% 29%

0.556 0.006 76% 32%

0.467 0.039 63% 15%

0.656 0.025 73% 22%

0.625 0.018 78% 27%

0.482 0.014 78% 29%

0.251 0.020 65% 19%

0.186 0.007 66% 16%

0.513 0.046 64% 15%

0.701 0.029 73% 21%

0.691 0.020 77% 28%

0.532 0.016 76% 29%

0.189 0.020 64% 18%

0.167 0.007 64% 15%

0.10

1

RV Size Quintile 2 3 4

5

0.538 0.026 66% 20%

0.641 0.015 72% 29%

0.677 0.011 79% 38%

0.477 0.010 81% 37%

0.418 0.016 79% 28%

0.152 0.026 60% 11%

0.251 0.020 66% 16%

0.242 0.014 70% 18%

0.188 0.010 70% 20%

0.058 0.010 59% 12%

0.111 0.028 58% 10%

0.234 0.021 65% 14%

0.229 0.015 68% 17%

0.179 0.010 68% 18%

0.029 0.009 56% 11%

0.12

0.09

0.10

0.08

0.10

Table A.2: Table reports the average coefficient from time-series regressions of the monthly proportional change in individual liquidity on the proportional change in the equally-weighted average liquidity measures. ’% +’ and ’sign +’ report the percentage of the coefficients that are positive and significantly positive at the 5% critical level in a one-sided test, respectively. In each individual regression, the right hand side average excludes the dependent variable. The contemporaneous and two lags of the world portfolio return and the proportional monthly change in the individual firm squared return are additional right-hand side regressors; coefficients are not reported. See table 2 and the text for an explanation of the liquidity variables. ’adj R2’ is the mean of the individual time-series regressions’ adjusted R-squared. Cross sectional standard deviations are calculated under the assumption that the estimation error for the coefficients are independent across regressions. Bold and italic numbers indicates significance at the 5% and 10% level, respectively. The size sort is based on the average market value in US dollar.

80

<99




:; 5+6 78+-+9  



<99




34 5+6 78+-+9  



<99




24 5+6 78+-+9  



G EFD C B?

  '( *+,- (

    ' '

  !  ' )'

  !  ' )'

  !  ' '

   # " ' '

 $   " ' '

  %%  ' '

    & "" ' '

   #  )' '

  #  
' '

    ' '

 
   " " ' '

  %#  ' '

    % " 
' '

  !  ' )'

  %  ' '

  $  '
'

    $$ " '
'

G EFJ IC D H

  '( *+,- (

  $ &  " ' )'

$& ""  #/  ' )'

   $/  ' )'

   / /  ' )'

  / /  ' )'

   $!  ' )'



  """ ' '


   /!  ' )'

   %  ' '



  $%$  ' '



  # " ' '

  % " ' '

  % " " ' )'

 

  ##  
' )'

    %%!  ' '


  $!  " ' )'

  $  ' '

. $!&   #! " ' )'

>A @ >? =

  '( *+,- ( 01 2


  %  )'
'

   " ' )'

  !&  ' '

  &$  )' '

  #!  )'
'

  /#  ' '

  $$  ' '

   !!  '
'



  #%  ' '



  /$  )' '

  %  ' )'

  $&  ' '

   $$  )'
'

  !#  ' '

  #%  )' '

   //  
' '

  %  ' '

    $&  ' '




  




Table A.3: Table reports the average coefficient from time-series regressions of the monthly proportional change in individual liquidity on a country-industry-global decomposition. See table 1 and the text for an explanation of the variables and how the decomposition is constructed. ’% +’ and ’sign +’ report the percentage of the coefficients that are positive and significantly positive at the 5% critical level in a one-sided test, respectively. The contemporaneous and two lags of the world portfolio return and the proportional monthly change in the individual firm squared return are additional right-hand side regressors; these coefficients are not reported. ’adj R2’ is the mean of the individual time-series regressions’ adjusted R-squared. Cross sectional standard deviations are calculated under the assumption that the estimation error for the coefficients are independent across regressions. Bold and italic numbers indicates significance at the 5% and 10% level, respectively. The size sort is based on the average market value in US dollar.

81

Y SN V m i n K L M N OPQ i k k P V k S OV o pq t mL u Sm U L r v L N VN O

Y M P V U L M N O P Q V s sV U O

f M r L s ] T N l M oO P Q V s sV U O o

K L M N OPQ

R S PTSN U V

W S OT L

R S PTSN U V

W S OT L

R S PTSN U V

W S OT L

XY

Z [Z Z \ \ \

]]^

Z [Z Z _ _ `

_ \^

Z [Z Z Z Z \

`^

ab

Z [Z Z \ \ \

`c^

Z [Z d Z d _

e `^

Z [Z Z Z d `

`^

af

Z [Z Z \ \ \

_ e^

Z [Z ` g h h

\ `^

Z [Z Z Z Z _

Z^

i j VPSk V

Z [Z Z \ \ \

dZ^

Z [Z ` h h d

c e^

Z [Z Z Z ` \

`^

YR t mL u Sm U L r v L N VN O

Y M P V U L M N O P Q V s sV U O

f M r L s ] T N l M oO P Q V s sV U O o

K L M N OPQ

R S PTSN U V

W S OT L

R S PTSN U V

W S OT L

R S PTSN U V

W S OT L

XY

Z [Z Z ` g e

]\^

Z [Z Z Z c h

g_^

Z [Z Z Z Z `

Z^

ab

Z [Z Z ` g e

`d ^

Z [Z Z h \ d

he^

Z [Z Z Z Z \

`^

af

Z [Z Z ` g e

g e^

Z [Z Z ` \ \

d ]^

Z [Z Z Z Z Z

Z^

i j VPSk V

Z [Z Z ` g e

dg^

Z [Z Z g \ g

c \^

Z [Z Z Z Z _

Z^

WR t mL u Sm U L r v L N VN O

Y M P V U L M N O P Q V s sV U O

f M r L s ] T N l M oO P Q V s sV U O o

K L M N OPQ

R S PTSN U V

W S OT L

R S PTSN U V

W S OT L

R S PTSN U V

W S OT L

XY

Z [Z Z _ c ]

cZ^

Z [Z Z ` Z Z

_Z^

Z [Z Z Z Z `

Z^

ab

Z [Z Z _ c ]

`g ^

Z [Z ` ] \ g

h h^

Z [Z Z Z ` d

`^

af

Z [Z Z _ c ]

]e^

Z [Z Z _ d g

]d ^

Z [Z Z Z Z `

Z^

i j VPSk V

Z [Z Z _ c ]

d ]^

Z [Z Z ] Z _

\ `^

Z [Z Z Z Z c

Z^

K L N OT N M V l

Table A.4: Table reports the variances of the global component and the country and industry specific components of equally-weighted country and industry averages of individual stock liquidity. Each country average in panel A is decomposed into a global component and a pure country effect and a sum of seven industry effects. Each industry average in panel B is decomposed into a global component and the sum of three country effects and a pure industry effect. See text and Heston and Rouwenhorst (1994) for the methodology and an explanation of the decomposition. ’Ratio’ gives the ratios of the variance of the global component, the country and industry effects to the variance of the weighted average. ’Average’ is the equally weighted average across the three measures.

82

w x y z { | }~  €  ‚ƒ  „ { …

˜ x  { z Ÿ   †  … „ ‡‚ ˆ ‰ | — — ˆ { — x ‚ { ‡ wž š z€ y xz  € › œ €  { ‚

 „ › € ™ ”  € „  ‚ ˆ ‰ { ™ ™{  ‚

†  … „ ‡‚ ˆ ‰

Š x ˆƒx  {

‹ x ‚ƒ €

Š x ˆƒx  {

˜ „ ˆ { ƒ  … „ ‡‚ ˆ ‰ { ™ ™ {  ‚

‹ x ‚ƒ €

Š x ˆƒx  {

‹ x ‚ƒ €

Œ

 }  Ž Ž Ž

Œ  

 }   ~ ‘



 }  Œ  ~

‘ ’

‘

 }  Ž Ž Ž

~~

 }   “ Œ

”

 }  • • ’

ŽŽ

”

 }  Ž Ž Ž



 }  Œ Œ ~

Œ” 

 }  ~ Ž ”

“~ 

~

 }  Ž Ž Ž

“‘ 

 }   • ”



 }  ~ “ ‘

”Ž

“

 }  Ž Ž Ž

Œ ~ 

 }   Œ ’

”

 }   ~ 



Ž

 }  Ž Ž Ž



 }   “ ‘

“

 }  Œ Œ 

Œ‘ 



 }  Ž Ž Ž

Ž ’

 }   • 

Œ 

 }   ’ •

•

| – {ˆx— {

 }  Ž Ž Ž

“

 }   Ž 



 }  ” ” ”

”

˜Š š z€ y xz  € › œ €  { ‚

 „ › € ™ ”  € „  ‚ ˆ ‰ { ™ ™{  ‚

˜ „ ˆ { ƒ  … „ ‡‚ ˆ ‰ { ™ ™ {  ‚

†  … „ ‡‚ ˆ ‰

Š x ˆƒx  {

‹ x ‚ƒ €

Š x ˆƒx  {

‹ x ‚ƒ €

Š x ˆƒx  {

‹ x ‚ƒ €

Œ

 }  Œ ” •

Œ • 

 }   Œ ‘

•

 }   ~ 

” Œ

‘

 }  Œ ” •

“ ’

 }   Œ 



 }  Œ ~ •

Ž‘ 

”

 }  Œ ” •

Œ ’

 }   ” ‘

‘ “

 }   “ “

~”

~

 }  Œ ” •

Ž“

 }    “

‘

 }    ’

”Ž

“

 }  Œ ” •

•

 }    “

~

 }    ’

Ž

Ž

 }  Œ ” •

”

 }    Œ

Œ

 }   Œ Ž

’



 }  Œ ” •

 ’

 }    “

”

 }   Œ ‘

Ž

| – {ˆx— {

 }  Œ ” •

’~ 

 }   Œ Œ



 }   “ Œ

‘ ’

‹Š š z€ y xz  € › œ €  { ‚ †  … „ ‡‚ ˆ ‰

Š x ˆƒx  {

 „ › € ™ ”  € „  ‚ ˆ ‰ { ™ ™{  ‚

‹ x ‚ƒ €

Š x ˆƒx  {

˜ „ ˆ { ƒ  … „ ‡‚ ˆ ‰ { ™ ™ {  ‚

‹ x ‚ƒ €

Š x ˆƒx  {

‹ x ‚ƒ € ‘ 

Œ

 }  ‘ “ 

•‘ 

 }   Œ 

Ž

 }    ~

‘

 }  ‘ “ 

~~

 }   ‘ Ž

~

 }  ” Ž 

Ž” 

”

 }  ‘ “ 

’‘ 

 }   “ ”

Œ

 }  Œ ~ ‘

~ Ž

~

 }  ‘ “ 

Ž

 }    •

”

 }  Œ  ’

”‘

“

 }  ‘ “ 

•

 }    Ž

‘

 }   Œ ’

Ž

Ž

 }  ‘ “ 

~ 

 }    ‘



 }   ‘ 

’



 }  ‘ “ 

•Ž

 }    ’

”

 }   Œ “

Ž

| – {ˆx— {

 }  ‘ “ 

 ’

 }   Œ 

“

 }  Œ  

‘ 

83

ÕÞ ßÊ ºá £ÏͨÊب

¦§¨©ª§ º»ª

ä æé åí ð ï åí î í ì ë ê ã å ä ã çå é è çæ åä ã â

Á£ º»ª ÆÇ º»ª ÆÈ º»ª ÉÊ˧ËÌÍ º»ª ÎËÍÏÐÑ º»ª ƻʦʻÊͺ º»ª Õϧ˺Ö×Ï» º»ª ¡×˺بÍÏ º»ª ¡§ÖÊ»§¦ º»ª Ù§ºÊÌ º»ª

¤×ªÍ¦ Õͺ»

Ú¿ ÖÜݧ¦ØÍ

¡¢£¤

¥¡¢£¤

¢¥¡¢£¤

¡¢£¤

£à ¥¡¢£¤

¢¥¡¢£¤

¡¢£¤

ßà ¥¡¢£¤

¢¥¡¢£¤

« ­®¯° ¬ ¶ ¶¹¼ · « «­¯´ ¬ ¶ ¶¶¸ · ¶ ¶¶¼¾ · ¶ ¶¶¸ · « ««´° ¬ ¶ ¶¶½ · « ««®­ ¬ ¶ ¶¶½ · ¶ ¶¶¹¿ · ¶ ¶¶¸ · « ««´­ ¬ ¶ ¶¶½ · « «­­² ¬ ¶ ¶¶½ · « ««³® ¬ ¶ ¶¶½ · « «­«­ ¬ ¶ ¶¶½ · « ««´² ¬ ¶ ¶¶½ · ¼¹ ¾¸À · ¶ ¶À ·

« ­±­² ¬ ¶ ¶¸½ · « «­µ­ ¬ ¶ ¶¶¸ · ¶ ¶¶¿¸ · ¶ ¶¶¸ · « ««´´ ¬ ¶ ¶¶½ · « ««®µ ¬ ¶ ¶¶½ ·  ÂÂÒÓ Ã ¶ ¶¶¸ · « ««®® ¬ ¶ ¶¶½ · « «­­® ¬ ¶ ¶¶½ · « ««®¯ ¬ ¶ ¶¶½ · « «­«® ¬ ¶ ¶¶½ · « «­«¯ ¬ ¶ ¶¶½ · ¼À ½¶¼ · ¶ ¶¸ ·

« ­³«´ ¬ ¶ ¶¸¸ · « «­µ¯ ¬ ¶ ¶¶¸ · ¶ ¶¶¿¾ · ¶ ¶¶¸ · « ««´® ¬ ¶ ¶¶½ · « ««®² ¬ ¶ ¶¶½ ·  ÂÂÒ à ¶ ¶¶½ · « ««´­ ¬ ¶ ¶¶½ · « «­­´ ¬ ¶ ¶¶½ · « ««®¯ ¬ ¶ ¶¶½ · « «­«° ¬ ¶ ¶¶½ · « ««´´ ¬ ¶ ¶¶½ · ¼¹ ÛÛ¾ · ¶ ¶À ·

« «µ±° ¬ ¶ ¶¶¾ · « «­µµ ¬ ¶ ¶¶¸ · « ««±­ ¬ ¶ ¶¶½ · « «­°³ ¬ ¶ ¶¶½ · « «­¯« ¬ ¶ ¶¶¿ · « ««´« ¬ ¶ ¶¶½ · « «­¯² ¬ ¶ ¶¶¿ · « «­µ± ¬ ¶ ¶¶½ · « «­¯® ¬ ¶ ¶¶¿ · « «­°° ¬ ¶ ¶¶½ · « «­µ« ¬ ¶ ¶¶½ · Û ÛÔ¶ · ¶ ½¹ ·

« «µ²° ¬ ¶ ¶¶¾ · « «­µ³ ¬ ¶ ¶¶¸ · « ««±² ¬ ¶ ¶¶½ · « «­°° ¬ ¶ ¶¶½ · « «­­´ ¬ ¶ ¶¶¿ · « ««´® ¬ ¶ ¶¶½ · « «­¯¯ ¬ ¶ ¶¶¿ · « «­µ± ¬ ¶ ¶¶½ · « «­¯® ¬ ¶ ¶¶¿ · « «­°µ ¬ ¶ ¶¶½ · « «­µ¯ ¬ ¶ ¶¶½ · Û ¹¿Û · ¶ ½Û ·

« «°²± ¬ ¶ ¶¼¿ · « «­°­ ¬ ¶ ¶¶¸ · ¶ ¶¶¸¸ · ¶ ¶¶½ · « «­¯² ¬ ¶ ¶¶¿ · « «­¯« ¬ ¶ ¶¶¿ · « ««®¯ ¬ ¶ ¶¶½ · « «­¯­ ¬ ¶ ¶¶¿ · « «­µ² ¬ ¶ ¶¶½ · « «­«® ¬ ¶ ¶¶¿ · « «­­® ¬ ¶ ¶¶½ · « «­¯° ¬ ¶ ¶¶½ · ¼¿ ÔÀ½ · ¶ ¼¾ ·

« ¯³±³ ¬ ¶ ¼½¸ ·  ÂÂÄÅ Ã ¶ ¶¶¸ · ¶ ¶¶¶¸ · ¶ ¶¶¸ · « «­­³ ¬ ¶ ¶¶½ · « ««´­ ¬ ¶ ¶¶¿ · ¶ ¶¶¸¶ · ¶ ¶¶¸ · « «­¯­ ¬ ¶ ¶¶¿ · « ««´± ¬ ¶ ¶¶½ · « ««®³ ¬ ¶ ¶¶½ · « ««´® ¬ ¶ ¶¶½ · « ««®­ ¬ ¶ ¶¶½ · ¿ Ô¿½ · ¶ Û¾ ·

« ¯±´´ ¬ ¶ ¼½¼ · « ««®² ¬ ¶ ¶¶¸ · ¶ ¶¶¶¹ · ¶ ¶¶¸ · « «­­° ¬ ¶ ¶¶½ · « ««®® ¬ ¶ ¶¶½ · ¶ ¶¶¸Ô · ¶ ¶¶¸ · « «­¯« ¬ ¶ ¶¶¿ · « ««´± ¬ ¶ ¶¶½ · « ««®° ¬ ¶ ¶¶½ · « ««´± ¬ ¶ ¶¶½ · « ««®¯ ¬ ¶ ¶¶½ · ¿ ¹¹¼ · ¶ Û¾ ·

¶ ¸¹¶¹ · ¶ ¿¾À · « ««´­ ¬ ¶ ¶¶¸ · ¶ ¶¶¶¾ · ¶ ¶¶¸ · « «­«³ ¬ ¶ ¶¶¿ · « ««®µ ¬ ¶ ¶¶¿ · ¶ ¶¶½Ô · ¶ ¶¶¸ · « «­¯° ¬ ¶ ¶¶¿ · « ««´² ¬ ¶ ¶¶½ · « ««³® ¬ ¶ ¶¶¿ · « ««®® ¬ ¶ ¶¶½ · « ««®« ¬ ¶ ¶¶½ · ½ ÛÛÀ · ¶ Û¼ ·

Table A.5: Table reports the estimated liquidity risk premium associated with the innovation in liquidity, as well as the contribution of liquidity risk to the expected return of the test assets. The test assets are three country and seven industry equallyweighted portfolio returns. Liquidity is the unexpected relative change in the equallyweighted global liquidity aggregate. The pricing models are a World CAPM, an International CAPM with the change in the trade-weighted US dollar exchange rate as an additional factor, and an augmented International CAPM with the change in the trade-weighted US dollar exchange rate and the returns on the US SMB and HML spread portfolios as the other two factors. ’Contribution’ is the estimate of the product of the portfolio’s factor sensitivity and the risk premium. ’Model Test’ refers to the GMM test of overidentification with the test statistic and its associated probability denoted as ’X2’ and ’p-value’, respectively. Bold and italic numbers represent significance at the critical 5% and 10% level, respectively.

84

 ÿ    ýþ

  

ñ ò ó ô õ

ñ ö÷ôøòö ñ÷ööûò ñ÷ùúõö ó÷ñòúù ø÷ùøñõ

    

ò ö÷ôõöù ö÷úúñù ñ÷ùúôû ó÷ñóúû ø÷úñøñ

ó ö÷ôøóö ñ÷ööôñ ñ÷ùûñô ó÷ñòóñ ø÷úòòñ

ô ö÷ôùöú ñ÷öòöú ñ÷ùûôõ ó÷ñöóô ø÷ûñòö

ÿ

õ ö÷ôøûø ñ÷öõöû ñ÷ûòñù ó÷öøóú ø÷óøóù

ñ ö÷øûøõ ñ÷øøöú ò÷õóôõ ó÷ôöûù ô÷õùóù

    

ò ö÷ùøúú ñ÷ùñûö ò÷øóûû ó÷õôøø ô÷øöûö

ó ö÷ûòñô ñ÷ùóûú ò÷øùôò ó÷øõûø ô÷ùõóô

ô ö÷ûôùò ñ÷ùòúô ò÷øõûó ó÷ùöñó ô÷úøøô

õ ö÷ûøñø ñ÷ûñöò ò÷ùòóö ó÷øøøû õ÷ñóúù

ñ ö÷ñõøø ö÷úñóù ò÷óñøô ó÷óûñó ô÷ùôöô

 
  ýþ

   

ñ ò ó ô õ

ñ üö÷öñûó üö÷öñóö üö÷ööõö ö÷ööóû ö÷öòòû

    

ò üö÷ööúô üö÷ööòû ö÷ööôñ ö÷öñóö ö÷öóñõ

ó üö÷ööúú üö÷ööóñ ö÷ööøò ö÷öñøó ö÷öóöò

ô üö÷ööúù üö÷ööóú ö÷ööõø ö÷öñûñ ö÷öóöù

ÿ

õ üö÷ööùø ö÷öööñ ö÷ööûø ö÷öñõñ ö÷öòõõ

ñ üö÷ööñõ ö÷ööøô ö÷öñôñ ö÷öñôô üö÷öñòø

    

ò ö÷ööóõ ö÷ööùö ö÷öñóñ ö÷öòñô ö÷öòûø

ó ö÷ööòô ö÷ööóô ö÷öñöö ö÷öñøô ö÷öòôô

ô üö÷ööõñ üö÷öööû ö÷ööõù ö÷öññú ö÷öñøù

õ üö÷öôñù üö÷öñûö üö÷ööòø ö÷ööúõ ö÷öñôõ

ñ ö÷öööú ö÷ööûò ö÷ööóó ö÷öòõû ö÷öûöû


    

ò ö÷òñûô ö÷ûôûú ò÷óòöù ó÷ùõóñ õ÷öóóú

ó ö÷òùöñ ö÷úûõö ò÷ñóóñ ó÷úñøô õ÷ôõöù

ô ö÷óòóø ñ÷öøöú ò÷òøöõ ó÷ùòöò õ÷úòúû

õ ö÷óùúò ñ÷òóùø ò÷óùõô ó÷ûññú õ÷ùôõô


    

ò ö÷ööôô ö÷öñóó ö÷ööúò ö÷ööôö ö÷öôñô

ó ô õ üö÷öööô üö÷öòûó üö÷öûúõ ö÷ööúú ö÷öööõ üö÷öóöù ö÷öñùñ ö÷öñòõ üö÷ööñö ö÷ööóñ ö÷öñùó ö÷öññô ö÷öñõù ö÷ööøò ö÷öñõû

Table A.6: Panel A and B report equally-weighted average levels of liquidity and returns based on independent liquidity and size sorts, where size is measured as the average market value in US dollar. See table A.1 and A.2 for more details.

85

Y;5Z [=";I"

NL MQ XU WU MV TU RS MKL OMKQ PO MNL JK

\$= ]=56

!"#$! 56$ :;<= 4 56$ :;<= 9 56$ :;<= 1 56$ :;<= 7 56$ :;<= 8 56$ :;<= ? 56$ :;<= 3 56$ :;<= @ 56$ :;<= 2 56$ :;<= 4/ 56$ E9 FGH! I=

 %&%''( /0/13 %&%%-* /0//1 %&%%'> /0//1 %&%%'+ /0//1 %&%(%> /0//1 %&%(%+ /0//1 %&%(%+ /0//1 %&%(%( /0//1 %&%%'/0//1 %&%%'( /0//1 %&%%.+ /0//1 ?0421 /039

]^  %&%')' /0/11 %&%%-+ /0//1 %&%%'( /0//1 %&%%') /0//1 %&%(%( /0//1 %&%(%, /0//1 %&%(%) /0//1 %&%%'* /0//1 %&%%'> /0//1 %&%%'% /0//1 %&%%.+ /0//1 70@43 /0@8

 %&()*+ /0/78 %&%%.% /0//1 %&%%'* /0//1 %&%(%% /0//1 %&%(%' /0//1 %&%((% /0//1 %&%((% /0//1 %&%(%. /0//1 %&%(%) /0//1 %&%%'/0//1 %&%%.' /0//1 30@4/ /088

 %&%),( /0//3 %&%%.. /0//9 %&%%'+ /0//1 %&%(%) /0//1 %&%((% /0//1 %&%((+ /0//1 %&%(%. /0//1 %&%(%. /0//1 %&%%'/0//1 %&%%*+ /0//9 %&%%.% /0//9 70818 /0@3

_   %&%),( %&%,() /0//3 /0/4/ %&%%.+ %&%%.( /0//9 /0//9 %&%%'> %&%%'% /0//1 /0//9 %&%(%) %&%%'. /0//1 /0//9 %&%((( %&%(%. /0//1 /0//1 %&%((%&%((, /0//1 /0//1 %&%(%* %&%(%> /0//1 /0//1 %&%(%* %&%((% /0//1 /0//1 %&%%'. %&%(%) /0//1 /0//1 %&%%*%&%%') /0//9 /0//1 %&%%.( %&%%.. /0//9 /0//9 707/@ 308@7 /0@@ /08@

 %&%+-* /0/98 %&%%,/0//9 %&%%>' /0//9 %&%%++ /0//9 %&%%-% /0//9 %&%%+. /0//9 %&%%+% /0//9 %&%%+% /0//9 %&%%,' /0//9 /0//9? /0//9 /0//93 /0//9 @09@8 /084

Y_  %&%.() /0/1/ %&%%>) /0//9 %&%%+* /0//9 %&%%-/0//9 %&%%.% /0//9 %&%%-. /0//9 %&%%-% /0//9 %&%%+' /0//9 %&%%>* /0//9 /0//19 /0//9 ABAACD /0//9 30384 /08?

 /01234 /0138 %&%%-> /0//9 %&%%'> /0//9 %&%(%+ /0//9 %&%(%/0//1 %&%(%+ /0//1 %&%(%% /0//1 %&%%'' /0//1 %&%%'> /0//1 %&%%*+ /0//1 %&%%./0//1 @01?3 /08/

Table A.7: Table reports the estimated liquidity risk premium associated with the innovation in liquidity, as well as the contribution of liquidity risk to the expected return of the test assets. The test assets are 10 size sorted equally-weighted portfolio returns. Liquidity is the relative unexpected change in the equally-weighted global liquidity aggregate. The pricing models are a World CAPM, an International CAPM with the change in the trade-weighted US dollar exchange rate as an additional factor, and an augmented International CAPM with the change in the trade-weighted US dollar exchange rate and the returns on the US SMB and HML spread portfolios as the other two factors. ’Contribution’ is the estimate of the product of the portfolio’s factor sensitivity and the risk premium. ’Model Test’ refers to the GMM test of overidentification with the test statistic and its associated probability denoted as ’X2’ and ’p-value’, respectively. Bold and italic numbers represent significance at the critical 5% and 10% level, respectively.

86

€a†}` ¤Ÿ ™˜|†n‘ z ¥†d| mn‘ €˜n ˜`…˜m £… m¦ €}b…|b §˜d}` —}mn ¬ª «³¯ ¶ µ³ «´ ²³ °± «©ª ­«©¯ ®­ «¬ª ¨©

£… m¦ €}b…|b §˜d}` —}mn ¬ª «³¯ ¶ µ³ «´ ²³ °± «©ª ­«©¯ ®­ «¬ª ¨©

`abcda mnd uq yz{a`|} „…†a†‡} mnd Ž†}‘ mnd n…`…n…}m mnd —a† my˜n mnd ™˜† m|b} mnd ™ay…na` mnd šam…‡ mnd

€ ˆ‰ˆˆŠ‹ opoov ˆ‰ˆˆ’Š opoov ˆ‰ˆˆŠ” opoov efegge opoov efeekl opoov efegeg opoov efeeŒ• opoov

`abcda mnd uq yz{a`|} ƒ…ž} r mnd ƒ…ž} q mnd ƒ…ž} x mnd ƒ…ž} v mnd ƒ…ž} ~ mnd ƒ…ž} t mnd ƒ…ž} œ mnd

€ efegŒg opoov efegil opoov efeg“j opoov efeekŒ opoov efeeŒi opoov efee•g opoox efeelh opoox

—¡

€¢

£¢

efgghi opoqr

efejek opoos

efiggl oprrt

qvpovv opqv

rtpvwo optw

rwpxvt op~o

‚ opoovv opoox opooos opoov efeeŒl opoox opoox~ opoox opooxo opoox opooxo opoox ˆ‰ˆˆ”› opoox

ƒ efeekŒ opoox efeeŒŒ opoov efeeŒg opoox efeeŒk opoov efeek• opoov efeg“h opoov efege“ opoox

€ ‚ ƒ efeggk efeell efeghe opoox opooq opooq efeg“h opoorr efeggh opoov opoox opoox efeg“• efegei efegjh opoov opooq opooq efegil efeehe efegjh opoov opooq opoox efeg“Œ efeejg efeg•k opoov opooq opoox efeg“k efeejh efegŒk opoov opooq opoox efeg“k efeehŒ efeglg opoov opooq opoox €a†}` šŸ ™˜|†n‘ z ƒ…ž} €˜n ˜`…˜m

€ ˆ‰ˆˆŠ opoov opoots opoo~ ˆ‰ˆˆ–ˆ opoo~ efeeŒl opoov ˆ‰ˆˆ– opoov ˆ‰ˆˆ opoov opooœo opoov

‚ opooxw opoox zopooxr opoox efeeŒ• opoox opooxq opoox opoorr opoox opooo~ opoox opooxo opoox

—¡

€¢

£¢

efgj“e oporw

efejig opooœ

ˆ‰”’Š opqsw

qqpvrr opxq

rwpœvt opvœ

qœpwtr oprr

‚ ˆ‰ˆˆ”Š opoox opoooq opoox opoooq opoox zopooow opoox opooot opoox opooox opoox opooo~ opoox

ƒ ˆ‰ˆˆ”Š opoox efeeŒj opoox efegeg opoov efeekk opoov efege“ opoov efegeg opoox efegg• opoox

€ efegk“ opoov efeghk opoov efeghg opoov efeg“k opoov efeggl opoov efege“ opoox efeek• opoox

‚ efeelk opooq opooqo opooq opooqx opooq opooqo opooq ˆ‰ˆˆ›– opooq efeejl opooq efeej• opooq

ƒ efegei opooq efegjj opoox efeg•j opoox efegŒe opoox efegŒj opoox efegŒg opoox efeglj opooq

€ efeglŒ opoo~ efeg“Œ opoo~ efeggj opoo~ ˆ‰ˆˆ’’ opoo~ opootw opoov opoott opoov ˆ‰ˆˆŠ opoov

‚ efeele opoox opooox opoox opoooo opoox zopoooœ opoox zopooor opoox opooov opoox opooov opoox

ƒ efegei opoox opoovw opoov efegi“ opooq efee•l opoov efeg“• opoox efeg“k opoox efegek opoox

ƒ ˆ‰ˆˆ”ˆ opoox efeekg opoox efeg“e opoox efeg“Œ opoox efeg“k opoox efegii opoox efegig opoox

Table A.8: Table reports the estimated liquidity risk premium associated with the innovation in liquidity, as well as the contribution of liquidity risk to the expected return of the portfolios formed from the intersection of the countries and industries as well as countries and size. The results are from the International CAPM (ICAPM). See tables A.5 and A.7 and the text for an explanation of the liquidity variables and the approach.

87

Å Ä¿ Æ ÇÀ» Ï ÐÑ Ä¹ Ç

Å Ä¿ Æ ÇÀ» Ï ÐÑ Ä¹ Ç

Å Ä¿ Æ ÇÀ» Ï ÐÑ Ä¹ Ç

·¸¹º»¼½

ÆÙ ¾º ¿¹ À»¼½

Á¸à ÄÂ

È ÉÊËÊ Ê ÉËÍ

È ÉÊÌÊ Ê ÉÌ Ò

ÍÉÎÍÊ Ê ÉÍÓ

·¸¹º»¼½

Ø× ¾º ¿¹ À»¼½

Á¸à ÄÂ

È ÉÎËÊ Ê ÉÈ Ó

ÌÉÌÊÊ Ê ÉÒË

ÒË ÉÌÌÊ Ê ÉÒÈ

·¸¹º»¼½

Ö× ¾º ¿¹ À»¼½

Á¸à ÄÂ

ÒÉÒÌÊ Ê ÉÎÎ

Ë ÉÈÊÊ Ê ÉÕÓ

Ë ÉÔ ÎÊ Ê ÉÌÔ

Table A.9: Table reports Wald tests of constant liquidity premiums across countries and industries. The estimation allows the liquidity risk premiums to differ for are all country and industry portfolios. ’Country’ refers to the test that the premiums are equal across countries; ’Industry’ refers to the test that the premiums are constant across industries; and ’Global’ refers to the test that the premiums are equal across countries and industries. All estimation results are based on the International CAPM and three country and seven industry portfolio returns (see table A.5 and the text for an explanation of the factors). ’Wald Test’ and ’p-value’ report the test statistics and the corresponding probabilities based on asymptotic distribution of the test statistics.

88

ÚÛ Þßàá Ü òóä ßôä

âãä å æã àõæ

ø ßùôßæßõú Ü òóä ßôä

ãâûü ã àõæ Ü àõæ





" ! $" & % $# " ! 

æóâ Ú óàõ





 

ì íìîïð ç ñèñ ë

ç ñèê é ñ ç÷ ñè ö













ñèññê

ñèññê

ì íìï  












çç÷



 


 

ñèñ











ì íïïþ 

ì íìï  ð

ì íìïïì

ì íì  

ì íìïîð

ñèññëë

ì íïììð

ì íìïþ ÿ
















ì íìÿ  ì

ì íìï  ï

ì íìï  

ì íìï  þ
















ì íìï  ì
















ì íìï  
















ì íì   ì

ì íìï  ì




 














 

  


  

û ãâô ó





 


 



, * " + * ) ( '

ì íì   ð

ÞÝ

ì íìÿÿï

àõæ ! # & " * -

ç èéê ëé ñèëö ñ ý ì íìþðÿ ç ñèñ ö

ÜÝ

àõæ ß  ã  ó  àõæ   óò  ú àõæ

õßâßõßóà àõæ Ú òã  àû  òõ àõæ    àôä óò àõæ  ãûßõãâ àõæ  ãàß  àõæ














 



 



 

ì íïì  ï



ì íïìðþ

 

ì íïì  î





 































Table A.10: Table reports the estimation results of the liquidity risk premium and the liquidity level compensation for the three country and seven industry portfolio returns based on the International CAPM. The liquidity risk factor is the contemporaneous innovation in global liquidity, and the liquidity level is the a two month lagged six month average level of the test assets’ equally-weighted average liquidity. ’lambda’ and ’alpha’ are the risk premium and the liquidity premium, respectively, associated with the innovation in liquidity and the level of liquidity. ’Contribution’ is the product of the liquidity risk premium and the assets’ sensitivity. ’X2’ is the GMM test of overidentification with its ’p-value’ below. Bold and italic numbers represent significance at the critical 5% and 10% level, respectively.

89

SZ . / 01 2 3 4 05 6 7 8 9 :5

; / 0 69< = > ? @A 4= < 6 8 9 :5

B 1 / C 41 D 9 E F 9 2 97G 8 9 :5

H 0 6I 9F I

JK LK K M N

J K LK N NK

K LK NM M

:7 2

K LK K O

K LK K P

K LK N N

Q / F = 70G

R= 2 F :70G

B 1 / C 41

. 41 2 S 6 :7

JK LK T NU

J K LK N NV

K LK T K P

:7 2

K LK N

K LK N

K LK N

Q / F = 70G

R= 2 F :70G

B 1 / C 41

. 41 2 S 6 :7

JK LK NP P

J K LK N NK

K LK W X K

:7 2

K LK N

K LK N

K LK T

HY

8Y

Table A.11: Table reports the time-series averages of the slopes of monthly crosssectional regressions of individual stock excess returns on post-ranking betas from January 1985 to December 2001. ’std’ is the time-series standard error of the average slopes. See text for details.

90

^c

€

a \ [ ˜ ™ š › › š c\ b a € ~ ~ }

§_

— – • ” “ ’ ‘

— – • ž “  ” œ

 ¢ ¡ ’  Ÿ

^

[ \\

]^ _

[ \\

]^ _

[ \\

]^ _

` abc

d ef g g

d eh g d

d ef i h

d ef f f

d ef f h

d eh j k opo o r

l mn

o p o qo

opo q q

opo o r

opo o s

opo o t

u v

wxu

yz u

w{u

w|u

w{u

yz u

}~  € v

 |u

‚ƒu

‚ „u

‚ƒu

‚u

‚ yu

c…† ‡ 

z ˆ„ „

z ˆ „

z ˆ„z

z ˆ „

z ˆ „

z ˆ„ x

` abc

d ei k j

d ek ‰ i

d ef d ‰

d ej ‰ f

d ek Š h

d ek ‰ d

l mn

opo o s

opo o‹

o p o qo

o p o qo

opo o r

opo o r

u v

{ „u

{ u

w‚u

w „u

{{u

{|u

}~  € v

„xu

„xu

ƒu

‚u

„{u

 „u

c…† ‡ 

z ˆ„z

z ˆ„z

z ˆz |

z ˆ„z

z ˆ„z

z ˆ„‚

` abc

d ek ‰ f

d ek Š Š

d ef g k

d eg k Œ

d ek h Œ

d ek f h

l mn

opo o‹

o p o qo

opo q q

opo q q

opo o r

opo o r

u v

x|u

x|u

w u

w u

{ƒ u

{wu

}~  € v

„ƒ u

„ƒ u

‚u

‚u

„xu

„|u

c…† ‡ 

z ˆ„z

z ˆ„z

z ˆz y

z ˆ„z

z ˆ„z

z ˆ „

]^ _

[ \\

^c

€ a \ £ ˜ ¤ a ¥ š › ¦ š }~ b ~ š €

§_ [ \\

— – • ” “ ’ ‘

— – • ž “  ” œ

 ¢ ¡ ’  Ÿ

‡

^ [ \\

]^ _

‡ ]^ _

` abc

d eg f Œ

d eg g Š

d ek ‰ k

d ei Š j

d ek ‰ f

d ek k h

l mn

o po r

o po 

opoŽ Ž

opoŽ Ž

opoŽ t

opoŽ

u v

x{u

x{u

{zu

{ „u

xxu

x{u

}~  € v

„z u

„z u

„‚ u

„ƒ u

„ „u

„ „u

` abc

d e‰ Œ ‰

d eŒ i Š

d eΠd k

d ef d i

d ej i j

o p qs

l mn

o p qs

o p q‹ Ž

op qqq

opo‹ r

o p qŽ q

opo‹ r

u v

xƒ u

xxu

xwu

xwu

xƒ u

x‚u

}~  € v

|u

|u

„ „u

„ „u

|u

|u

` abc

d eg Š d

d ej Š i

d ef j g

d ef j j

d ej d ‰

d ej Œ Š

l mn

opoŽ r

o p o ‹

o p o 

o p o o

o p o 

o p o q‹

u v

x|u

{zu

{ƒ u

{|u

x|u

{‚u

}~  € v

„ u

„ u

„{u

zu

„ u

„xu

c…† ‡ 

z ˆ „

z ˆ „ƒ

z ˆ„z

z ˆ„‚

z ˆ „

z ˆ„ x

Table A.12: Panel A reports the average coefficient for the original sample, and the alternative sample, IPO, that excludes the first four years of data for each stock. Panel B reports the results for regressions on the country, industry, and global decomposition. See table A.2, table A.3 and the text for an explanation of the liquidity variables and the approach. 91

Á ©± ­ ¬ ¶ Î Æ µ Ç ± ´¯È ¶ Ï Ï ¯ ­ Ï © ´­ Ê Ð Â Ì ¬µ Í ©¬ ² µ ª « µ ± ­± ´

Á Ç ¯ ­ ² µ Ç ± ´¯È ­ Ë Ë­ ² ´

¨ Ç ª µ Ë Ã °± É Ç Ê´ ¯È ­ Ë Ë­ ² ´ Ê

¨ ©ª « ¬­

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

¶ ¬¬

· ¸· · ¹ ¹ ¹

º·»

· ¸· ¼ ½ ½ º

¾ ¿»

· ¸· · · ¼ ¹

¼»

ÀÁ Â

· ¸· · Ã Ä º

¾·»

· ¸· ¼ ¾ ¼ ¾

¹Å »

· ¸· · · ¼ Å

¼»

Á® Ì ¬µ Í ©¬ ² µ ª « µ ± ­± ´

Á Ç ¯ ­ ² µ Ç ± ´¯È ­ Ë Ë­ ² ´

¨ Ç ª µ Ë Ã °± É Ç Ê´ ¯È ­ Ë Ë­ ² ´ Ê

Æ µ Ç ± ´¯È

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

¶ ¬¬

· ¸· · ¼ Ä ¿

ºÄ»

· ¸· · Ä ¹ Ä

¾ ¹»

· ¸· · · · Å

·»

ÀÁ Â

· ¸· · ¼ Å ¿

¹Å »

· ¸· · ¼ ½ Ã

¹ ¹»

· ¸· · · · ¼

·»

³® Ì ¬µ Í ©¬ ² µ ª « µ ± ­± ´

Á Ç ¯ ­ ² µ Ç ± ´¯È ­ Ë Ë­ ² ´

¨ Ç ª µ Ë Ã °± É Ç Ê´ ¯È ­ Ë Ë­ ² ´ Ê

Æ µ Ç ± ´¯È

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

¶ ¬¬

· ¸· · Å ¾ Ã

º û

· ¸· · Ã · Å

¹ ¼»

· ¸· · · · ¾

·»

ÀÁ Â

· ¸· · ¼ Ã Å

¾·»

· ¸· · Å ¹ ¿

¾ ¹»

· ¸· · · · Å

·»

Á © ± ­ ¬ Ñ Î À ± É Ç Ê´ ¯È ¶ Ï Ï ¯ ­ Ï © ´ ­ Ê Ð Â Ì ¬µ Í ©¬ ² µ ª « µ ± ­± ´

¨ Ç ª µ Ë Ä ² µ Ç ± ´¯È ­ Ë Ë­ ² ´

Á Ç ¯ ­ ° ± É Ç Ê´¯È ­ Ë Ë­ ² ´

À ± É Ç Ê´¯ È

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

¶ ¬¬

· ¸· · ¹ ¹ ¹

þ»

· ¸· · · ¹ Ã

û

· ¸· · Ä Ä Ä

Ä·»

ÀÁ Â

· ¸· · Ã Ä º

þ»

· ¸· · · Ã ¿

¿»

· ¸· · Ä º ¿

Ä ¼»

Á® Ì ¬µ Í ©¬ ² µ ª « µ ± ­± ´

¨ Ç ª µ Ë Ä ² µ Ç ± ´¯È ­ Ë Ë­ ² ´

Á Ç ¯ ­ ° ± É Ç Ê´¯È ­ Ë Ë­ ² ´

À ± É Ç Ê´¯ È

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

¶ ¬¬

· ¸· · ¼ Ä ¿

½º »

· ¸· · · ¼ ¼

û

· ¸· · · ¾ ¼

Å ½»

ÀÁ Â

· ¸· · ¼ Å ¿

¿ ½»

· ¸· · · ¼ Å

¼Å »

· ¸· · · º º

Ä·»

³® Ì ¬µ Í ©¬ ² µ ª « µ ± ­± ´

¨ Ç ª µ Ë Ä ² µ Ç ± ´¯È ­ Ë Ë­ ² ´

Á Ç ¯ ­ ° ± É Ç Ê´¯È ­ Ë Ë­ ² ´

À ± É Ç Ê´¯ È

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

® © ¯°©± ² ­

³ © ´° µ

¶ ¬¬

· ¸· · Å ¾ Ã

à ½»

· ¸· · · ¼ Ã

¾»

· ¸· · ¼ · Ã

Šû

ÀÁ Â

· ¸· · ¼ Ã Å

½¾»

· ¸· · · ¼ ¹

¿»

· ¸· · · ¹ Ã

Ä·»

Table A.13: Table reports the variances of the global component and the country and industry specific components of equally-weighted country and industry averages of individual stock liquidity for the original sample and the ’IPO’ sample. 92

Õ×ÿùÓ Ò ÿç ×ÿÚ ÔÿÚæç Õ ç Ó÷ æ

Ö ÷ æ ÕùØ÷Ø

Ó×ØÙÚ× æçÚ

  " % $ " # " !          

òÕ æçÚ ûü æçÚ ûö æçÚ þ÷ÿ×ÿ ù æçÚ
ÿù æçÚ ûç÷Ó÷ç÷ùæ æçÚ ×ÿæ ç æçÚ

ÿæØù æçÚ

×÷ç×Ó æçÚ ×æ÷ æçÚ

& ÚùÓ ùæç

ì ×Óù

Õ×ÿùÓ  ö÷øù Õ ç Ó÷ æ

Ö



Õ

ÒÓÓ

ÔÕÖ

ÒÓÓ

ÔÕÖ

ÒÓÓ

ÔÕÖ

Û ÝÞÝß Ü è èêë é Û ÛÝáÝ Ü è èèê é è èèìê é è èèê é Û ÛÛàà Ü è èèë é Û ÛÛäá Ü è èèë é ó óó ô è èèê é Û ÛÛää Ü è èèë é Û ÛÝÝä Ü è èèë é Û ÛÛäã Ü è èèë é Û ÛÝÛä Ü è èèë é Û ÛÝÛã Ü è èèë é ïñ ëèï é è èê é

Û ÛàÝÞ Ü è èìê é Û ÛÝåÞ Ü è èèú é Û ÛÛàÝ Ü è èèê é Û ÛÛåà Ü è èèê é Û ÛÝÛÝ Ü è èèë é Û ÛÛäà Ü è èèê é Û ÛÛßß Ü è èèì é Û ÛÝÞÛ Ü è èèê é Û ÛÝÛä Ü è èèë é Û ÛÝáå Ü è èèê é Û ÛÝáä Ü è èèê é ïï ñèú é è ñï é

Û Ûáßâ Ü è èèí é Û ÛÝáå Ü è èèê é Û ÛÛÞß Ü è èèë é Û ÛÝââ Ü è èèë é Û ÛÝÝà Ü è èèì é Û ÛÛàä Ü è èèë é Û ÛÝãã Ü è èèì é Û ÛÝáÞ Ü è èèë é Û ÛÝãä Ü è èèì é Û ÛÝâá Ü è èèë é Û ÛÝáã Ü è èèë é ð úìð é è ëð é

Û Ûãäå Ü è èèî é Û ÛÝåÝ Ü è èèê é Û ÛÝÛÝ Ü è èèë é Û ÛÝáâ Ü è èèì é Û ÛÝáÞ Ü è èèë é Û ÛÝáâ Ü è èèë é Û ÛÛàá Ü è èèì é Û ÛÝÞå Ü è èèê é Û ÛÝâã Ü è èèë é Û ÛÝßß Ü è èèë é Û ÛÝâä Ü è èèë é ð ñíí é è êí é

Û ãÞàà Ü è ïëï é Û ÛÛäß Ü è èèê é è èèèú é è èèê é Û ÛÝÝâ Ü è èèë é Û ÛÛää Ü è èèë é è èèêî é è èèê é Û ÛÝãÛ Ü è èèì é Û ÛÛàÞ Ü è èèë é Û ÛÛäâ Ü è èèë é Û ÛÛàÞ Ü è èèë é Û ÛÛäã Ü è èèë é ì úúï é è ðí é

Û ÝåÞã Ü è èîð é ó óóõó ô è èèú é ó óóýó ô è èèê é Û ÛÝãâ Ü è èèë é Û ÛÝÝß Ü è èèë é Û ÛÝÛä Ü è èèê é Û ÛÝÛÝ Ü è èèì é Û ÛÝãÛ Ü è èèê é Û ÛÛàß Ü è èèë é Û ÛÛàÛ Ü è èèë é Û ÛÛäâ Ü è èèê é í èíð é è îê é

Ó×ØÙÚ× æçÚ ö÷øù ï æçÚ ö÷øù ì æçÚ ö÷øù ë æçÚ ö÷øù ê æçÚ ö÷øù ú æçÚ ö÷øù î æçÚ ö÷øù ñ æçÚ ö÷øù í æçÚ ö÷øù ð æçÚ ö÷øù ïè æçÚ

ì ×Óù



Õ

ÒÓÓ

ÔÕÖ

ÒÓÓ

ÔÕÖ

ÒÓÓ

ÔÕÖ

Û Ûàãà Ü è èëë é Û ÛÛÞß Ü è èèë é Û ÛÛàÝ Ü è èèë é Û ÛÛàã Ü è èèë é Û ÛÝÛÝ Ü è èèë é Û ÛÝÛá Ü è èèë é Û ÛÝÛã Ü è èèë é Û ÛÛàä Ü è èèë é Û ÛÛàâ Ü è èèë é Û ÛÛàÛ Ü è èèë é Û ÛÛåß Ü è èèë é ê íïñ é è íú é

Û Ûåäß Ü è èìñ é Û ÛÛàÞ Ü è èèê é Û ÛÝÝå Ü è èèê é Û ÛÝãÞ Ü è èèê é Û ÛÝãà Ü è èèê é Û ÛÝãß Ü è èèê é Û ÛÝãã Ü è èèê é Û ÛÝÝÞ Ü è èèê é Û ÛÝÛä Ü è èèë é Û ÛÛàâ Ü è èèë é Û ÛÛåå Ü è èèë é î ìúí é è îî é

Û ÛãáÝ Ü è èèñ é Û ÛÛåß Ü è èèì é Û ÛÛàâ Ü è èèë é Û ÛÝÛã Ü è èèë é Û ÛÝÝÝ Ü è èèë é Û ÛÝÝÞ Ü è èèë é Û ÛÝÛä Ü è èèë é Û ÛÝÛä Ü è èèë é Û ÛÛàå Ü è èèë é Û ÛÛäÞ Ü è èèì é Û ÛÛåÝ Ü è èèì é ê êèí é è íí é

Û Ûááß Ü è èèí é Û ÛÝãã Ü è èèë é Û ÛÝâà Ü è èèë é Û ÛÝßÞ Ü è èèë é Û ÛÝÞâ Ü è èèë é Û ÛÝâä Ü è èèë é Û ÛÝßã Ü è èèë é Û ÛÝâÝ Ü è èèë é Û ÛÝáÛ Ü è èèë é Û ÛÝÝÞ Ü è èèì é Û ÛÝÛÛ Ü è èèì é ïï ïïì é è úë é

Û ÛåÝã Ü è èëè é Û ÛÛâã Ü è èèì é Û ÛÛßä Ü è èèì é Û ÛÛÞÞ Ü è èèì é Û ÛÛåÛ Ü è èèì é Û ÛÛÞå Ü è èèì é Û ÛÛÞÛ Ü è èèì é Û ÛÛßà Ü è èèì é Û ÛÛâä Ü è èèì é è èèëì é è èèì é ó óó  ô è èèì é ñ ñúï é è úî é

Û Ýåãã Ü è èñë é Û ÛÝÛâ Ü è èèë é Û ÛÝáâ Ü è èèë é Û ÛÝãß Ü è èèë é Û ÛÝãã Ü è èèë é Û ÛÛàä Ü è èèë é Û ÛÝÛß Ü è èèë é Û ÛÛàÞ Ü è èèë é Û ÛÛäß Ü è èèë é Û ÛÛåå Ü è èèë é Û ÛÛåã Ü è èèë é ïï ïìú é è íì é

Table A.14: Table reports the estimated liquidity risk premium associated with the innovation in liquidity, as well as the contribution of liquidity risk to the expected return of the country and industry portfolios in panel A and the size portfolios in panel B for the original sample and the ’IPO’ sample. See tables A.5 and A.7 and the text for an explanation of the liquidity variables and the approach.

93

Asia Mean Standard Deviation Skewness Excess Kurtosis

0.00132 0.10943 0.14348 4.57163

Mean Standard Deviation Skewness Excess Kurtosis

0.00158 0.10618 0.36623 3.76353

Mean Standard Deviation Skewness Excess Kurtosis

-0.00186 0.11243 0.09178 5.23600

Mean Standard Deviation Skewness Excess Kurtosis

0.00161 0.12117 -0.43793 6.43115

Panel A: Summary Statistics Europe North America Emerging Asia Full Sample 0.00019 -0.00007 0.00087 0.05898 0.06993 0.09844 -0.51051 -0.27113 0.31999 8.51827 4.02697 4.06564 No Crises 0.00017 0.00011 0.00127 0.05555 0.06762 0.09958 -0.47533 -0.15041 0.44095 7.47366 3.65569 4.45294 Mexican Crisis [7/1/94–6/3/95] -0.00235 -0.00060 -0.00120 0.04725 0.07042 0.08911 -0.57841 -0.23539 -0.24107 6.55641 4.48643 4.11422 Asian/Russian Crises [1/1/97–3/31/99] 0.00130 -0.00057 0.00014 0.07480 0.07875 0.09833 -0.55437 -0.58145 -0.01876 8.24517 4.69663 2.60579

Emerging America 0.00069 0.05358 0.05931 2.04338 0.00106 0.05730 0.00651 1.77207 -0.00236 0.04294 0.49288 3.70771 0.00050 0.04127 0.32377 1.34946

continued

Table A.15: Panel A reports summary statistics for the full sample of weekly observations of market-wide changes in liquidity from 1/9/1990 to 1/2/2002 and for three subsamples of crises and non-crises periods. The crises periods cover the Mexican crisis in 1994/1995 and the Asian/Russian crisis in 1997/1999. The regional aggregates are equallyweighted averages of individual stock liquidity, where the measure of individual stock liquidity is the average number of days with zero returns calculated from the previous Wednesday to Tuesday. The aggregate Asia includes Australia and Japan; Europe includes Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the UK; North America includes Canada and the US; Emerging Asia includes China, Hong Kong, India, Indonesia, South Korea, Malaysia, Pakistan, the Philippines, Singapore, Thailand and Taiwan; and Emerging America includes Argentina, Brazil, Columbia, Chile, Mexico, Peru, and Venezuela. Panel B reports pairwise correlations over the full sample as well as over the three subsamples.

94

Table A.15 continued

Asia Asia Europe North America Emerging Asia Emerging America

. 0.1739 0.0500 0.2202 0.2141

Asia Europe North America Emerging Asia Emerging America

. 0.1535 0.0429 0.2078 0.2016

Asia Europe North America Emerging Asia Emerging America

. 0.2607 0.2269 0.2065 0.1957

Asia Europe North America Emerging Asia Emerging America

. 0.2082 0.0115 0.2707 0.3025

Panel B: Pairwise Correlations Europe North America Emerging Asia Full Sample 0.1739 0.0500 0.2202 . 0.5568 0.2834 0.5568 . 0.1090 0.2834 0.1090 . 0.5113 0.3197 0.2780 No Crises 0.1535 0.0429 0.2078 . 0.5476 0.2536 0.5476 . 0.0859 0.2536 0.0859 . 0.4947 0.2872 0.2480 Mexican Crisis [7/1/94–6/3/95] 0.2607 0.2269 0.2065 . 0.5567 0.2116 0.5567 . -0.0016 0.2116 -0.0016 . 0.4636 0.3127 0.2610 Asian/Russian Crises [1/1/97–3/31/99] 0.2082 0.0115 0.2707 . 0.5878 0.3990 0.5878 . 0.2262 0.3990 0.2262 . 0.6732 0.5098 0.4633

95

Emerging America 0.2141 0.5113 0.3197 0.2780 . 0.2016 0.4947 0.2872 0.2480 . 0.1957 0.4636 0.3127 0.2610 . 0.3025 0.6732 0.5098 0.4633 .

Asia

Europe

North America

Emerging Asia

Emerging America

Constant

0.0003 (0.09)

0.0003 (0.15)

0.0002 (0.10)

0.0008 (0.25)

0.0005 (0.29)

Asia

-0.5620 (-13.84)

-0.0217 (-0.97)

A Lag 1 -0.0633 (-2.49)

0.0129 (0.36)

-0.0206 (-1.03)

Europe

-0.0040 (-0.04)

-0.6200 (-11.30)

-0.0247 (-0.40)

-0.0494 (-0.57)

-0.0443 (-0.90)

North America

0.1789 (2.21)

0.1039 (2.33)

-0.6550 (-12.96)

0.3734 (5.29)

0.0984 (2.47)

Emerging Asia

0.0705 (1.50)

0.0289 (1.11)

0.0563 (1.92)

-0.5621 (-13.67)

0.0327 (1.41)

Emerging America

-0.0979 (-1.01)

-0.0207 (-0.39)

-0.0340 (-0.56)

-0.2326 (-2.75)

-0.6107 (-12.77)

Asia

-0.3892 (-8.87)

-0.0540 (-2.23)

A Lag 2 0.0259 (0.94)

-0.0579 (-1.51)

-0.0296 (-1.37)

Europe

0.2445 (2.23)

-0.3629 (-5.98)

-0.0267 (-0.39)

-0.1983 (-2.07)

0.0133 (0.25)

North America

0.0343 (0.36)

-0.0374 (-0.71)

-0.4206 (-7.05)

0.2412 (2.89)

0.0271 (0.57)

Emerging Asia

0.0890 (1.77)

0.0091 (0.33)

0.0376 (1.20)

-0.4366 (-9.96)

0.0061 (0.25)

Emerging America

-0.0268 (-0.25)

0.0215 (0.37)

-0.0483 (-0.73)

-0.0142 (-0.15)

-0.3511 (-6.70)

Asia

-0.1982 (-4.92)

-0.0425 (-1.91)

A Lag 3 0.0273 (1.08)

-0.0454 (-1.29)

0.0063 (0.32)

Europe

0.1099 (1.11)

-0.2143 (-3.93)

-0.0435 (-0.70)

-0.1046 (-1.21)

0.0413 (0.85)

North America

0.1013 (1.26)

-0.0024 (-0.05)

-0.1953 (-3.88)

0.1416 (2.01)

-0.0506 (-1.27)

Emerging Asia

0.0704 (1.53)

0.0146 (0.58)

0.0141 (0.49)

-0.2899 (-7.23)

-0.0321 (-1.42)

Emerging America

0.1219 (1.27)

0.0496 (0.93)

-0.0149 (-0.25)

0.0439 (0.52)

-0.0991 (-2.09)

Adj. R-squared

26.3%

25.9%

32.4%

32.0%

27.1%

Table A.16: Table reports estimation results of a vector autoregressive system for the full sample of 624 weekly observations from 1/9/1990 to 1/2/2002. The lag length p = 3 minimizes the Schwarz-Bayesian information criterion, BIC. The estimated coefficients are in the first row with the corresponding t-statistics in parentheses below. Bold and italic bold letters report significance at the 5% and the 10% level, respectively. 96

Response

Asia

Impulse

Europe North America Emerging Asia Emerging America

Asia

Europe

.

6.0681 [0.11] .

6.6020 [0.09] 8.2570 [0.04] 4.2823 [0.23] 3.8891 [0.27]

12.0232 [0.01] 1.4027 [0.70] 1.4032 [0.70]

North America

Emerging Asia

Emerging America

12.5448 [0.01] 0.5522 [0.91] .

4.2590 [0.23] 4.5695 [0.21] 28.5114 [0.00] .

3.1808 [0.36] 2.1442 [0.54] 10.2496 [0.02] 5.8121 [0.12] .

3.8696 [0.28] 0.6021 [0.90]

10.3262 [0.02]

Table A.17: Table reports Granger-Causality tests based on the VAR(3) model reported in table A.16. The first row reports the Wald-test statistic and the second row reports the corresponding p-value.

97

Asia Europe North America Emerging Asia Emerging America

Asia

Europe

.

0.1731 [0.00] .

0.1731 [0.00] 0.1009 [0.01] 0.2333 [0.00] 0.1853 [0.00]

0.6148 [0.00] 0.2912 [0.00] 0.5146 [0.00]

North America

Emerging Asia

Emerging America

0.1009 [0.01] 0.6148 [0.00] .

0.2333 [0.00] 0.2912 [0.00] 0.2614 [0.00] .

0.1853 [0.00] 0.5146 [0.00] 0.3699 [0.00] 0.2432 [0.00] .

0.2614 [0.00] 0.3699 [0.00]

0.2432 [0.00]

Table A.18: Table reports estimated correlation matrix of residuals from the VAR(3) model reported in table A.16. The corresponding p-values for the hypothesis that the correlation is zero are in the brackets below.

98

Panel A: Mean Equations Asia

Europe

North America

Emerging Asia

Emerging America

Constant

0.0014 (0.24)

0.0003 (0.07)

-0.0011 (-0.22)

-0.0024 (-0.40)

0.0003 (0.10)

Asia

-0.5329 (-8.22)

-0.0194 (-0.46)

A Lag 1 -0.0640 (-1.39)

0.0470 (0.80)

-0.0036 (-0.10)

Europe

-0.0858 (-0.57)

-0.5925 (-7.15)

0.0047 (0.05)

-0.0618 (-0.48)

-0.0564 (-0.81)

North America

0.1480 (1.32)

0.0669 (0.88)

-0.6610 (-7.80)

0.3627 (3.72)

0.0685 (1.20)

Emerging Asia

0.0846 (1.33)

0.0176 (0.36)

0.0464 (0.95)

-0.5708 (-8.47)

0.0491 (1.26)

Emerging America

-0.0234 (-0.19)

-0.0333 (-0.37)

-0.0639 (-0.63)

-0.1851 (-1.56)

-0.6093 (-8.41)

Asia

-0.3487 (-5.44)

-0.0484 (-1.29)

A Lag 2 0.0215 (0.55)

-0.0145 (-0.22)

-0.0166 (-0.46)

Europe

0.1446 (1.04)

-0.3558 (-3.70)

-0.0159 (-0.14)

-0.2414 (-1.84)

-0.0089 (-0.11)

North America

0.0175 (0.14)

-0.0617 (-0.75)

-0.4210 (-4.30)

0.2486 (2.28)

-0.0021 (-0.03)

Emerging Asia

0.0776 (1.14)

-0.0027 (-0.05)

0.0244 (0.45)

-0.4263 (-5.67)

0.0163 (0.42)

Emerging America

0.0605 (0.42)

0.0458 (0.49)

-0.0190 (-0.18)

0.0455 (0.34)

-0.3141 (-3.98)

Asia

-0.1528 (-2.28)

-0.0441 (-1.41)

A Lag 3 0.0300 (0.88)

-0.0241 (-0.43)

0.0076 (0.24)

Europe

0.0671 (0.52)

-0.2086 (-2.12)

-0.0362 (-0.33)

-0.1141 (-0.87)

0.0451 (0.61)

North America

0.0840 (0.80)

-0.0128 (-0.18)

-0.1787 (-1.97)

0.1524 (1.57)

-0.0704 (-1.21)

Emerging Asia

0.0448 (0.60)

0.0129 (0.31)

0.0017 (0.03)

-0.2898 (-4.56)

-0.0234 (-0.79)

Emerging America

0.1229 (1.08)

0.0327 (0.36)

-0.0076 (-0.07)

0.0504 (0.38)

-0.0974 (-1.37)

Continued

Table A.19: Table reports estimation results for a multivariate GARCH - BEKK(1,1) model for the sample of 624 weekly observations from 1/9/1990 to 1/2/2002. The estimated coefficients are in the first row and the corresponding t-statistics in parentheses below. Panel A and B report the estimation results for the mean and variance equation, respectively, and Panel C reports Likelihood Ratio tests for the existence of meteor showers. See text for further details. 99

Table A.19 continued

Panel B: Variance Equations ARCH Term A Asia

0.2505 (3.75)

0.0107 (0.23)

-0.0012 (-0.02)

0.0225 (0.29)

0.0333 (0.81)

Europe

0.0257 (0.16)

0.1725 (1.58)

-0.0547 (-0.48)

-0.0561 (-0.35)

-0.0039 (-0.04)

North America

-0.0740 (-0.53)

-0.0514 (-0.54)

0.1566 (1.72)

0.0250 (0.17)

-0.0448 (-0.55)

Emerging Asia

-0.0364 (-0.39)

0.0142 (0.21)

-0.0040 (-0.06)

0.3345 (2.74)

-0.0244 (-0.52)

Emerging America

-0.0080 (-0.05)

0.0343 (0.31)

0.0895 (0.90)

-0.0320 (-0.16)

0.2751 (3.15)

Asia

0.8829 (12.63)

0.0273 (0.60)

0.0190 (0.45)

0.0238 (0.21)

-0.0097 (-0.26)

Europe

-0.0849 (-0.55)

0.9074 (11.89)

-0.0453 (-0.71)

-0.0453 (-0.23)

-0.0524 (-0.76)

North America

0.0014 (0.01)

0.1170 (1.11)

0.9721 (11.10)

0.1495 (0.67)

0.0706 (0.83)

Emerging Asia

0.0363 (0.41)

-0.0313 (-0.54)

0.0117 (0.23)

0.8035 (5.87)

0.0074 (0.16)

Emerging America

0.0639 (0.53)

-0.0316 (-0.41)

-0.0357 (-0.67)

-0.0005 (-0.00)

0.9277 (16.75)

GARCH Term B

Panel C: Likelihood Ratio Tests .

112.7925 [0.00]

127.8499 [0.00]

124.0348 [0.00]

118.1525 [0.00]

Europe

115.9540 [0.00]

.

137.9742 [0.00]

115.3177 [0.00]

100.4178 [0.00]

North America

116.9866 [0.00]

123.7880 [0.00]

.

118.5501 [0.00]

124.9365 [0.00]

Emerging Asia

118.3014 [0.00]

115.5729 [0.00]

118.5349 [0.00]

.

116.7439 [0.00]

Emerging America

148.2822 [0.00]

114.2453 [0.00]

85.2325 [0.00]

109.7952 [0.00]

.

Asia

100

Table A.20: Table reports the estimation results for a dynamic simultaneous equations model for the full sample from 1/9/1990 to 1/2/2002 using an SUR framework. The model contains five liquidity and five return equations and includes three lags of the endogenous variables. The liquidity equations also contain the absolute return. Additional determinants are the average individual stock volatility in the respective market, the net equity flow to the market, the world net bond flow, the change in the US TBill rate, and the change in the US Term spread. The coefficients are in the first row and the corresponding t-statistic in parentheses below. Bold and italic bold coefficients indicate significance at the 5% and the 10% level, respectively.

101

Table A.20 continued

∆ Market Liquidity Equations Asia

Europe

North America

Emerging Asia

Emerging America

Intercept

-0.1338 (-10.01)

-0.0456 (-7.22)

-0.0599 (-5.36)

-0.1170 (-10.04)

Asia

-0.5628 (-14.84)

∆ Market Liquidity – Lag 1 -0.0230 -0.0527 0.0249 (-1.09) (-2.15) (0.76)

-0.0468 (-7.70)

Europe

-0.0767 (-0.84)

-0.6553 (-12.36)

-0.0548 (-0.92)

-0.1322 (-1.64)

-0.0292 (-0.65)

North America

0.1415 (1.92)

0.1058 (2.50)

-0.6591 (-13.64)

0.3310 (5.04)

0.0779 (2.13)

Emerging Asia

0.0476 (1.11)

0.0230 (0.93)

0.0467 (1.65)

-0.5921 (-15.41)

0.0286 (1.34)

Emerging America

-0.0774 (-0.87)

-0.0006 (-0.01)

-0.0170 (-0.29)

-0.1468 (-1.85)

Asia

-0.3625 (-8.83)

∆ Market Liquidity – Lag 2 -0.0480 0.0364 -0.0286 (-2.10) (1.38) (-0.81)

-0.6186 (-13.42)

Europe

0.1625 (1.61)

-0.3798 (-6.46)

-0.0584 (-0.88)

-0.2268 (-2.55)

0.0197 (0.39)

North America

0.0234 (0.27)

-0.0174 (-0.35)

-0.3982 (-6.96)

0.2330 (3.02)

0.0293 (0.68)

Emerging Asia

0.0462 (1.01)

0.0037 (0.14)

0.0307 (1.02)

-0.4361 (-10.68)

0.0050 (0.22)

Emerging America

-0.0041 (-0.04)

0.0147 (0.26)

-0.0473 (-0.74)

0.0257 (0.30)

Asia

-0.1732 (-4.62)

∆ Market Liquidity – Lag 3 -0.0367 0.0335 -0.0187 (-1.74) (1.38) (-0.57)

-0.3819 (-7.63)

Europe

0.0428 (0.47)

-0.1847 (-3.55)

-0.0426 (-0.71)

-0.1207 (-1.51)

0.0326 (0.73)

North America

0.0998 (1.35)

-0.0009 (-0.02)

-0.1825 (-3.78)

0.1483 (2.26)

-0.0264 (-0.72)

Emerging Asia

0.0407 (0.97)

0.0198 (0.83)

0.0159 (0.58)

-0.2638 (-7.08)

-0.0273 (-1.31)

Emerging America

0.1644 (1.88)

0.0177 (0.35)

-0.0221 (-0.38)

0.0578 (0.74)

-0.1287 (-2.89)

-0.0183 (-1.00)

-0.0240 (-1.21)

0.0164 (0.90)

continued

102

Table A.20 continued

∆ Market Liquidity – Cross Equation and Exogenous Variables Asia

Europe

North America

Emerging Asia

Emerging America

|Market Return| t-1

-0.3967 (-1.96)

-0.1019 (-0.52)

-0.0233 (-0.14)

-0.2309 (-1.35)

-0.7441 (-5.59)

|Market Return| t-2

-0.7086 (-3.70)

-0.3920 (-1.98)

-0.1257 (-0.75)

-0.4383 (-2.68)

-0.3775 (-2.88)

|Market Return| t-3

-0.7077 (-3.72)

-0.8026 (-4.25)

-0.0308 (-0.19)

-0.5151 (-3.18)

-0.1979 (-1.51)

Market Return t-1

0.3897 (2.96)

0.3074 (2.22)

0.0961 (0.81)

0.1173 (1.06)

0.3298 (3.59)

Market Return t-2

0.3208 (2.45)

0.1300 (0.90)

0.1927 (1.59)

0.1217 (1.10)

-0.0428 (-0.47)

Market Return t-3

-0.1288 (-1.01)

0.0463 (0.33)

0.1726 (1.41)

0.0082 (0.07)

0.0737 (0.82)

avg Volatility

0.7426 (12.04)

0.3440 (9.08)

0.1620 (6.12)

0.5148 (11.66)

0.3737 (10.63)

∆ TBill Rate

4.9053 (1.42)

-0.7808 (-0.39)

3.0751 (1.35)

1.7248 (0.56)

-2.4559 (-1.43)

∆ US Term spread

1.0432 (0.38)

-1.1233 (-0.72)

-1.8582 (-1.04)

1.2827 (0.53)

-1.0661 (-0.79)

Net Equity Flow

-0.0277 (-0.20)

-0.0001 (-0.02)

0.0899 (1.30)

0.0054 (0.34)

0.0021 (0.63)

Total Net World Bond Flow

-0.4940 (-4.12)

-0.1079 (-1.55)

0.0861 (1.07)

-0.2326 (-2.17)

-0.1130 (-1.89)

Adj. R-squared

38.9%

34.0%

38.1%

42.3%

36.8%

continued

103

Table A.20 continued

Market Return Equations Asia

Europe

North America

Emerging Asia

Emerging America

-0.0015 (-1.30)

-0.0003 (-0.52)

-0.0017 (-2.20)

-0.0017 (-1.45)

0.0019 (2.58)

Asia

0.1271 (2.89)

Market Return – Lag 1 -0.0127 -0.0492 0.0334 (-0.63) (-1.67) (0.75)

0.0190 (0.68)

Europe

-0.0554 (-0.47)

0.3678 (6.82)

0.1164 (1.48)

0.2343 (1.99)

0.1875 (2.53)

North America

0.0969 (1.24)

-0.0025 (-0.07)

0.2156 (4.12)

0.0233 (0.30)

-0.0735 (-1.49)

Emerging Asia

-0.0010 (-0.02)

-0.0359 (-1.81)

-0.0084 (-0.29)

0.0004 (0.01)

-0.0040 (-0.15)

Emerging America

-0.0897 (-1.32)

-0.0086 (-0.27)

0.0292 (0.63)

0.0182 (0.26)

0.1648 (3.70)

Asia

0.0795 (1.79)

Market Return – Lag 2 0.0258 0.0162 0.0358 (1.26) (0.54) (0.80)

-0.0111 (-0.40)

Europe

0.0259 (0.21)

0.0834 (1.48)

-0.0251 (-0.31)

-0.0976 (-0.79)

0.0975 (1.26)

North America

-0.0039 (-0.05)

0.0195 (0.53)

0.1336 (2.50)

0.0975 (1.21)

-0.0288 (-0.57)

Emerging Asia

-0.0143 (-0.33)

0.0121 (0.61)

0.0393 (1.36)

0.0216 (0.50)

0.0256 (0.94)

Emerging America

-0.0338 (-0.49)

-0.0019 (-0.06)

-0.0314 (-0.68)

0.0289 (0.42)

0.1315 (2.94)

Asia

0.0281 (0.64)

Market Return – Lag 3 0.0114 0.0250 -0.0609 (0.56) (0.85) (-1.38)

-0.0306 (-1.11)

Europe

0.1680 (1.42)

0.0470 (0.86)

0.1350 (1.70)

-0.3552 (-2.97)

0.0187 (0.25)

North America

-0.0518 (-0.66)

0.0075 (0.21)

0.0194 (0.37)

0.1354 (1.72)

0.0624 (1.27)

Emerging Asia

-0.0382 (-0.88)

-0.0116 (-0.59)

-0.0137 (-0.47)

0.0806 (1.85)

0.0194 (0.71)

Emerging America

0.0720 (1.07)

0.0213 (0.68)

0.0032 (0.07)

0.1299 (1.90)

-0.0045 (-0.10)

Intercept

continued

104

Table A.20 continued

Market Return – Cross Equation Variables Asia

Europe

North America

Emerging Asia

Emerging America

∆ Market Liquidity

t-1

0.0131 (1.26)

-0.0026 (-0.34)

-0.0056 (-0.57)

-0.0006 (-0.05)

-0.0010 (-0.07)

∆ Market Liquidity

t-2

-0.0049 (-0.44)

-0.0039 (-0.47)

-0.0255 (-2.27)

-0.0035 (-0.29)

-0.0015 (-0.10)

∆ Market Liquidity

t-3

0.0028 (0.28)

-0.0018 (-0.23)

-0.0177 (-1.80)

0.0005 (0.05)

-0.0124 (-0.89)

1.6%

15.7%

12.4%

3.0%

9.0%

Adj. R-squared

105

Asia

Europe

North America

Emerging Emerging Asia America

Asia

Europe North America Emerging Asia Emerging America Asia Europe North America Emerging Asia Emerging America

Market Return Equations

Asia

∆ Market Liquidity Equations

∆ Market Liquidity Equations . 0.1766 [0.00] 0.1005 [0.01] 0.2792 [0.00] 0.1742 [0.00] 0.0754 [0.06] 0.0798 [0.05] 0.0450 [0.26] 0.0806 [0.04] 0.0021 [0.96]

0.1766 [0.00] . 0.6015 [0.00] 0.2882 [0.00] 0.5220 [0.00] -0.0581 [0.15] -0.0672 [0.09] 0.0101 [0.80] -0.0115 [0.77] -0.0447 [0.27]

0.1005 [0.01] 0.6015 [0.00] . 0.2477 [0.00] 0.3680 [0.00] 0.0424 [0.29] -0.0269 [0.50] -0.0041 [0.92] -0.0028 [0.94] -0.0575 [0.15]

0.2792 [0.00] 0.2882 [0.00] 0.2477 [0.00] . 0.2453 [0.00] -0.0080 [0.84] -0.0369 [0.36] 0.0042 [0.92] 0.0252 [0.53] 0.0238 [0.55]

Europe

North America

Emerging Emerging Asia America

Market Return Equations 0.1742 [0.00] 0.5220 [0.00] 0.3680 [0.00] 0.2453 [0.00] . 0.0543 [0.18] 0.0925 [0.02] 0.1375 [0.00] 0.1098 [0.01] 0.1903 [0.00]

0.0754 [0.06] -0.0581 [0.15] 0.0424 [0.29] -0.0080 [0.84] 0.0543 [0.18] . 0.3599 [0.00] 0.2803 [0.00] 0.2960 [0.00] 0.1725 [0.00]

0.0798 [0.05] -0.0672 [0.09] -0.0269 [0.50] -0.0369 [0.36] 0.0925 [0.02] 0.3599 [0.00] . 0.6153 [0.00] 0.3496 [0.00] 0.3561 [0.00]

0.0450 [0.26] 0.0101 [0.80] -0.0041 [0.92] 0.0042 [0.92] 0.1375 [0.00] 0.2803 [0.00] 0.6153 [0.00] . 0.3284 [0.00] 0.3683 [0.00]

0.0806 [0.04] -0.0115 [0.77] -0.0028 [0.94] 0.0252 [0.53] 0.1098 [0.01] 0.2960 [0.00] 0.3496 [0.00] 0.3284 [0.00] . 0.1899 [0.00]

0.0021 [0.96] -0.0447 [0.27] -0.0575 [0.15] 0.0238 [0.55] 0.1903 [0.00] 0.1725 [0.00] 0.3561 [0.00] 0.3683 [0.00] 0.1899 [0.00] .

Table A.21: Table reports the estimated correlation matrix of residuals from the model reported in table A.20. The corresponding p-values for the hypothesis that the correlation is zero are in the brackets below.

106

-50 ex

[DA]

Europe [DE] North America [DN] Emerging Asia [EA] Emerging America [ES]

[DE]

North America [DN] Emerging Asia [EA]

[EA]

[DN]

Emerging America [ES]

Emerging Asia [EA] Emerging America [ES]

Emerging America [ES]

0.2077 0.1793 0.1587 0.1358 0.2749 0.2439 0.2310 0.1710 0.6846 0.4515 0.3225 0.2338 0.6386 0.3803

-100 ex

** ** ** **

** + **

0.3185 0.2934 0.2883 0.2585 0.4089 0.3453 0.3374 0.2918

** ** ** **

100 ex 0.3298 0.2934 0.2337 0.2585 0.3329 0.3453 0.2847 0.2918

** ** ** **

** **

50 ex 0.2774 0.1793 0.1583 0.1358 0.2633 0.2439 0.2723 0.1710

** ** ** **

** +

0.5831 ** 0.5575 0.3480 ** 0.3479 0.5773 ** 0.4942

0.4597 0.5575 0.3177 0.3479 0.4273 0.4942

0.3073 ** 0.2216 0.5415 ** 0.2804 +

0.3103 ** 0.3277 0.4914 ** 0.3901 +

0.3164 ** 0.3277 0.3595 ** 0.3901

0.2081 ** 0.2216 0.2549 ** 0.2804

0.2310 ** 0.1710

0.3782 ** 0.3335

0.3537 ** 0.3335

0.2179 ** 0.2154

**

0.4128 ** 0.4515 0.2110 ** 0.2338 0.3568 ** 0.3803

Table A.22: Table reports estimated tail correlations for threshold values with 50 and 100 positive and negative exceedences. The first row reports the tail correlation and the second row reports the average tail correlation from a Monte Carlo simulation of 100 simulated samples of bivariate normally distributed values with matching means and covariances. ** and * indicate significance at the 5% and the 10% level for the estimated tail correlation, respectively. + and - indicate that the estimated tail correlation is above or below the 95% range of the simulated estimates, respectively.

107

APPENDIX B

FIGURES

108

0.015 Asia 0.01 Europe

Weekly Change in Liquidity

0.005

0

-0.005 North America

emerging Asia -0.01

emerging America

-0.015 October-93

October-94

October-95

October-96

October-97

October-98

October-99

October-00

October-01

Figure B.1: Figure presents a 50-week moving average of market-wide changes in five regional aggregates over the full sample of liquidity from 12/18/1990 to 1/2/2002. See table A.15 and the text for more details.

109

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0 -50

-100

-150

-200

200

150

100

DA - DE

DA - DN

DA - EA

DA - ES

DE - DN

DE - EA

DE - ES

DN - EA

DN - ES

EA - ES

50

Figure B.2: The figure reports the estimated tail correlations ρ = 1−α2 from bivariate peak over threshold models, where α is the dependence parameter from a logistic dependence function D which maps the marginal extreme value distributions Gi of each variables into the one-dimensional space D(DL1 , DL2 ) = (G1 (DL1 )−1/α + G2 (DL2 )−1/α )α The results are calculated for threshold values with exceedences of positive and negative 200, 150, 100, and 50 values. See text for a discription of the estimation method.

110