حساسیت بازده سهام به شوک بحران: شواهدی از بازارهای توسعه یافته و نوظهور

We consider three “crisis shocks” related to key features of the 2007–2008 crisis, for emerging and developed economies: (1) the collapse of global trade, (2) the contraction of credit supply, and (3) selling pressure on firms’ equity. Using an international cross-section of firms, we find that returns’ sensitivities to these shocks imply large and statistically significant influences on residual equity returns during the crisis period (after controlling for normal risk factors that are associated with expected returns). Similar analysis for several placebo periods shows that these effects are generally less severe or absent in non-crisis periods. Relative to developed economies, emerging markets are more responsive to global trade conditions (in crisis and in placebo periods), but less responsive to selling pressures. An analysis of portfolios of firms during various placebo periods indicates that investors are not compensated for the risks associated with the crisis shocks. Finally, a month-by-month analysis of returns during the crisis period shows that the time variation of the importance of each of the sensitivities to shocks tracks related changes in the global economic environment.

The financial crisis of 2007–2008, which started in the US mortgage market, was characterized by three types of global shocks: a sharp contraction in the supply of credit, distressed sales of risky assets as banks and investors scrambled to shore up their liquidity and capital ratios, and a significant contraction in global trade. In this paper, we examine the extent to which the sensitivities of firms to these shocks explain the behavior of firm-level stock returns during the crisis. Stock returns are a unique measure of performance that is comparable across firms and countries, forward-looking, comprehensive in scope, and insensitive to differences in accounting rules. In normal times, a firm’s stock returns reflect a combination of expected returns (its loadings on risk factors) and residual returns that are associated with firm-specific news. At times of significant economy-wide shocks, however, the cross-section of residual returns can be understood as reflecting the exposure or sensitivity of firms to unexpected shocks. Our strategy is to construct measures of firm-level sensitivity to each of the three categories of “crisis shocks” described above and then identify their relative contribution to the observed declines in equity returns. As a measure of sensitivity to global product demand shocks, we employ a measure of global trade exposure. The sensitivity to selling pressure is captured by the amount of trading in each stock prior to the crisis. We measure firms’ sensitivity to credit supply shocks through a combination of variables relating to the capital structure (leverage ratio), its dividend behavior (dividend to sales ratio), and the ability of the firm to cover its debt obligations (interest coverage). We collect data on over 16,000 firms in 44 countries around the world to study whether cross-sectional stock returns over the period of August 2007 to December 2008 can be explained by firms’ sensitivities to the “crisis shocks” described above.1 We use a methodology similar to Tong and Wei (2011) which employs a cross-sectional model of stock returns and captures expected returns with a standard set of control variables.2 In this framework, our sensitivities to shocks capture unexpected influences of crisis-related shocks on residual stock returns. Empirically, we use values from 2006 to construct our measures of sensitivities, which are based on firm characteristics observed prior to the crisis. We then compare our results for the crisis period with a similarly structured model of the “placebo” period that runs from August 2005 to December 2006 as well as with two longer placebo periods spanning 5 and 10 years each, going back as far as 1997. To complement the regression analysis, we also build portfolios of “crisis-shock exposed” and “crisis-shock robust” firms (which are defined later in the paper) and test whether returns before and during the 2007–2008 crisis, as well as the different placebo periods, have differed across these two types of firms. To preview our results, we find that firms sensitive to credit supply shocks, global demand shocks, and selling pressures in the equity market had lower returns during the crisis. On the other hand, sensitivity to these shocks had generally less severe effects during the placebo periods. Our results are robust to different measures of beta, momentum, and weighting. Because the crisis originated in the developed countries (largely in the US and UK), and later spread to emerging markets, we also investigate whether there are major differences in the impact of these sensitivities in the developed countries and emerging markets samples. We find meaningful differences. Global demand sensitivity is higher in the emerging markets sample, likely because trade is more important for firms in emerging economies. On the other hand, the sensitivity to selling pressures is higher in the sample of developed countries, reflecting the fact that stock markets in developed countries tend to be more liquid than in emerging markets. Both developed and emerging markets display similar sensitivity of returns to credit supply shocks, but the magnitudes differ. Confirming the conclusions of our placebo period analysis, our portfolio analysis during the pre-crisis period (1997–2006) reveals that the influences we identify do not appear to be “priced.” Mean returns are higher for firms with low leverage, high interest coverage, and high dividends. This is the opposite of what we would expect if the market was generally compensating investors for risk associated with higher leverage. This pattern conforms with the negative effects of financial distress on returns identified in prior work (Campbell et al., 2008). At the same time, we find no difference in mean returns for firms with high and low foreign sales or liquidity, which indicates that the market is not compensating investors for the risks associated with holding liquid stocks and stocks from firms with high foreign sales. Finally, consistent with the regression results for the crisis and placebo periods, we find that during the crisis period stock returns are lower for “crisis-shock exposed” firms, namely those with high leverage, low dividend to sales, low interest coverage, high foreign sales, and high liquidity. A month-by-month analysis of returns’ sensitivities shows that the time variation of the importance of each of the sensitivities to “crisis shocks” tracks related changes in the global economic environment. The magnitude of the negative coefficient associated with the sensitivity to a global demand shock rises during times of greatest decline in exports. Time variation in the coefficients associated with the sensitivity to a credit supply shock is similar to that observed in credit risk spreads that reflect the timing of credit supply shocks. The variation over time in the coefficient that measures sensitivity to a stock market selling pressure shock closely tracks the variation in the returns to the stock market. Developed countries and emerging markets have similar patterns across time, with two exceptions: the credit supply shock more significantly affects developed countries in the periods of March 2008 (Bear Stearns collapse) and the summer and fall of 2008 (Lehman Brother’s collapse); the liquidity shock is larger and more variable in developed countries than it is in the emerging markets sample. While our regression methodology builds on Tong and Wei (2011), our focus is different. Tong and Wei explore the role of country level exposure to financial globalization, specifically through the composition of capital flows. They also find an important firm-specific factor in cross-sectional returns related to financial dependence (specifically, working capital financing needs). Our focus is entirely on firm-specific sensitivities to shocks, which arise as a result of an unexpected crisis event. We abstract from the effect of country characteristics by using country fixed effects. Didier et al. (forthcoming) provides a detailed analysis of country-specific factors in aggregate equity returns during the crisis. In considering the significance of firm-specific variables, we explore a broader range of firm characteristics, both relating to financing structure and other characteristics of firms than did Tong and Wei. This paper is related to the growing literature on the origin and consequences of the crisis. Most of the existing papers have focused on the causes and consequences of the crisis and, thus, have mostly analyzed its epicenter, the United States.3 A few others have studied the global transmission of this crisis. For instance, Fratzscher (2009) and Obstfeld et al. (2009) focus on the transmission via exchange rates. Dooley and Hutchison (2009) provide evidence of transmission to credit default swap spreads in emerging markets. Rose and Spiegel, 2010 and Rose and Spiegel, 2012) conduct an analysis of the international propagation of the crisis based on a measure of crisis incidence and severity which combines changes in real GDP, stock markets, credit ratings, and exchange rates. However, these papers use macro data to analyze the incidence and determinants of the propagation of the crisis. Ehrmann et al. (2009) study the transmission of the US 2007–2008 crisis to stock markets around the world by focusing on the performance of about 450 industry-equity portfolios across 64 countries. That paper primarily emphasizes the role of macro factors on the performance of industry portfolios rather than the role of the micro sensitivities to crisis shocks we consider here. The rest of our paper is organized as follows. Section 2 explains our approach to identifying firms’ sensitivities to crisis shocks. Section 3 describes the data and the empirical model used in our regression analysis. Section 4 presents our main empirical results for the global cross-section of stock returns during the crisis, and shows that our identified sensitivities to crisis shocks played a uniquely important role in explaining equity returns during the crisis, as compared with several pre-crisis “placebo” periods. Section 5 presents our results separately for developed countries and emerging markets. Section 6 presents an alternative approach to examine the behavior of firms’ stock returns during crisis and non-crisis periods using portfolio analysis. Section 7 examines the cross-section of returns during the crisis period in more detail, performing a month-by-month analysis of the changing importance of firms’ sensitivities to crisis shocks over time. Section 8 concludes. 2. Identifying firms’ sensitivities to crisis shocks 2.1. Global demand shock The financial crisis was associated with a remarkable decline in global trade. World exports fell by 9 percent between July 2007 and December 2008. That decline reflected a variety of potential influences, including the sensitivity of export financing to credit supply contraction (Amiti and Weinstein, 2011 and Chor and Manova, forthcoming). Our interest is not in explaining export decline, but rather examining firms’ differing sensitivity to the decline in global demand during the crisis. In particular, we want to assess whether firms that had positioned themselves prior to the crisis to be more dependent on trade were relatively more vulnerable to global demand shocks during the financial crisis. We, therefore, measure global demand shock sensitivity using a firm-specific measure that captures the exposure of a firm to global trade. Our measure is the firm’s pre-crisis proportion of sales outside the company’s home country (i.e., the ratio of foreign to total sales). 2.2. Stock market selling pressure shock There have been numerous studies of the effects of the crisis and the role of credit contraction and illiquidity crisis-induced selling on the redemptions of money market debts and the widening of bond spreads. These studies identify important effects of correlated selling pressure traceable to illiquidity problems in generating the contraction of quantities and the declines in prices in different debt markets.4 In publicly traded equity markets, crisis-related shocks could have even greater effects than in debt markets, given the consequences of the crisis for firms’ immediate and future incomes and their debt financing options. Just as in debt markets, problems of “funding illiquidity” for investors in publicly traded firms (due to declines in investor equity, rising market volatility, and the decline in available credit), could have been transformed into “market illiquidity” as owners of publicly traded shares were forced to liquidate their shares. Billio et al. (2010) examine correlations in returns across different equity investors and document apparent crisis-specific linkages in returns that they argue reflect this selling pressure.5 Additionally, publicly traded firms’ expected performance was itself affected by declining expected sales and by contraction in the supply of credit. Equity selling pressure, therefore, could have magnified declines in share prices that reflected the influences of declining demand and tight credit in reducing the discounted expected future cash flows of firms. We measure the sensitivity of a firm’s equity to selling pressures in the stock market using pre-crisis stock turnover (the volume of trading relative to outstanding market value of equity).6 This measure is intended to capture the relative liquidity of a stock prior to the crisis.7 In theory, the effect of stock liquidity on returns is ambiguous. On the one hand, greater liquidity may be associated with steeper declines in equity prices, as investors select their most liquid risky assets to sell during a liquidity squeeze. On the other hand, liquidity becomes more valuable during a crisis, implying that relatively illiquid stocks may experience relative price declines. The interpretation of any observed liquidity effects on returns is also controversial. For example, if liquid stocks decline more during the crisis, one could argue that relatively illiquid stocks also experienced similar or even larger “shadow” declines in value during the crisis that were masked by the lack of sales of these illiquid stocks. In other words, had someone tried to sell a large amount of an illiquid stock, its price would have been much lower. Selectivity bias related to endogenous decisions to sell, therefore, complicates the interpretation of the meaning of the effects of liquidity on stock returns during the crisis.