Do Controlling Shareholders Expropriation Incentives Imply a Link between Corporate Governance and Firm Value? Theory and Evidence

Transcription

1 Do Controlling Shareholders Expropriation Incentives Imply a Link between Corporate Governance and Firm Value? Theory and Evidence Jae-Seung Baek, Kee-Hong Bae, Jun-Koo Kang, and Wei-Lin Liu * This version: April 2011 * Baek (jbaek@hufs.ac.kr) is from Hankuk University of Foreign Studies; Bae (kbae@schulich.yorku.ca) is from Schulich School of Business, York University; and Kang (jkkang@ntu.edu.sg) and Liu (wlliu@ntu.edu.sg) are from Nanyang Technological University. We are grateful for comments from seminar participants at the 1 st International Conference on Corporate Governance and Emerging Markets, the 2010 China International Conference in Finance, and the 2010 International Conference on Asia-Pacific Financial Markets. We also thank Jay Choi and Jordan Siegel for helpful comments. All errors are our own.

2 Do Controlling Shareholders Expropriation Incentives Imply a Link between Corporate Governance and Firm Value? Theory and Evidence Abstract We develop and test a model that investigates how controlling shareholders expropriation incentives affect firm values during crisis and subsequent recovery periods. Consistent with the prediction of our model, we find that during the 1997 Asian financial crisis, Asian firms with weaker corporate governance experience a larger drop in their share values, but during the postcrisis recovery period, such firms experience a larger rebound in their share values. We also find consistent evidence for Latin American firms during the 2001 Argentine economic crisis. Our results support the view that controlling shareholders expropriation incentives imply a link between corporate governance and firm value. JEL classification: G15; G21; G32; G33; G34 Keywords: Expropriation; Corporate governance; Firm value; Asian financial crisis; Controlling shareholder; Cash flow rights; Control rights.

3 1. Introduction Researchers have extensively examined the link between corporate governance and firm value during an economic crisis. Previous studies show that firms with weaker corporate governance suffer more during such a period. One potential explanation for this finding is that during a crisis period, controlling shareholders incentives to expropriate minority shareholders tend to go up as the expected return on investment falls (Johnson, Boone, Breach, and Friedman, 2000; Mitton, 2002; Baek, Kang, and Park, 2004). This view implies that controlling shareholders incentives to expropriate minority shareholders are the key channel through which corporate governance affects firm value during a crisis period. We term this conjecture the "expropriation hypothesis." However, the positive relation between the quality of corporate governance and the change in firm value during a crisis period is also consistent with several alternative explanations. One such explanation is an information-based argument. For example, prior to the 1997 Asian financial crisis, the relationshipbased financial system in East Asia worked well, and thus investors might have ignored the weaknesses of East Asian firms. Alternatively, perhaps investors did not have full information on whether or not their funds were being deployed appropriately, but the crisis exposed the inherent weakness in the corporate governance systems of East Asian countries, triggering greater investor awareness of the problems in the region. This increased awareness led to investors pulling out (Rajan and Zingales, 1998). This argument suggests that greater investor awareness of the weakness in corporate governance is the main driving force that links corporate governance to the change in firm value during the crisis period. According to this view, the poor performance of firms with weak corporate governance during a crisis period is not necessarily due to increased expropriation, but rather to investors paying more attention to corporate governance problems that have been hidden. Another potential explanation is that the governance measures used in previous studies are somehow closely correlated with firms sensitivity to business conditions. For example, in our sample firms, we find a strong negative correlation between firm-level governance measures and systematic risk as estimated by the market model beta. This finding suggests that the performance of poorly governed firms 1

4 is more sensitive to change in market conditions, implying that firms with weaker corporate governance suffer more when the market performs poorly. Finally, it could simply be that investors overreact to a shock, and the degree of investors overreaction is more pronounced for poorly governed firms. Like other explanations, overreaction also implies a positive relation between the change in a firm s value and the quality of its corporate governance during a crisis. While both the expropriation hypothesis and the alternative explanations above have important implications for the link between corporate governance and firm value during a crisis, previous research has largely overlooked these alternative explanations in examining that link. In this paper, we re-evaluate the validity of the expropriation hypothesis by developing and testing a model that considers the expropriation incentives of controlling shareholders not only during the crisis period, but also during the subsequent recovery period. In our tests, we explicitly consider the possibility that firms with weaker corporate governance suffer more during economic crisis periods because of the alternative reasons mentioned above. The novelty of using the post-crisis recovery period is that the prediction of the expropriation hypothesis regarding the relation between the quality of corporate governance and the change in firm value during this period is exactly opposite to that during the crisis period: If controlling shareholders increased incentive to expropriate minority shareholders during the crisis period is the main reason for the poor performance of firms with weak corporate governance, we would expect these firms to experience a larger percentage increase in value during the recovery period than do firms with good corporate governance. This prediction does not imply that poor corporate governance enhances firm value during the recovery period. Instead, our model suggests that the greater rebound in the stock prices of firms with weaker corporate governance during the recovery period is a reflection of the firms more rampant asset diversion problem during the crisis period, which severely limits their ability to take full advantage of the substantially improved investment opportunities when recovery begins. The rationale for this prediction is as follows. During the crisis period, because of the significant 2

5 decline in firms profit prospects and poorer investment opportunities, controlling shareholders have stronger incentives to divert firm resources for their own benefits. Consequently, firms with weaker corporate governance experience more asset diversion and larger decline in firm value than those with better corporate governance. However, as the economy recovers, firms profit outlook and investment opportunities improve substantially. Since controlling shareholders can benefit more from profitable firm investments than from expropriation during this recovery period, the improved economic conditions alleviate controlling shareholders incentives to expropriate minority shareholders. Because firms with weaker corporate governance had more extant asset diversion before the recovery period, during the recovery period they have limited resources for undertaking all the profitable investments available and are thus forced to undertake only the most profitable ones. As a result, during the recovery period, on a per dollar basis, firms with weaker corporate governance realize higher returns on investments than those with better corporate governance, and thus experience greater percentage increases in firm value. The case of Samsung Fine Chemical (SFC) illustrates how firms with weak corporate governance experience large declines in firm value during the crisis period but have strong rebounds as the economy recovers. In 1996, a year before the onset of the Asian financial crisis, the controlling shareholders of Samsung Group directly and indirectly (through firms affiliated with the business group) own 0.07% and 8.49% of the outstanding shares of SFC, and the affiliated firms additionally hold 32.47% of its outstanding shares. This ownership structure (i.e., divergence between ownership and control) allows controlling shareholders to exercise full control over SFC despite holding a relatively small portion of its cash flow rights. During the crisis period, the stock price of SFC plummeted from 29,500 won in July 1997 by almost 74% to just above 7,700 won by the end of September However, as the economy recovers, the stock price of SFC bounced back strongly, reaching 26,600 won (an increase of 245%) by the end of December In comparison, the increase in the Korean stock market index during the same period was 152%. To test the model predictions, we first analyze a sample of 608 Korean firms listed on the Korean Stock Exchange during the financial crisis and the post-crisis recovery periods. As 3

6 a further test, we then examine 598 firms listed on seven other East Asian stock markets during the same periods and 302 firms listed on four Latin American stock markets during the Argentine crisis and the post-crisis recovery periods. We require sample firms to have data available for both the crisis and recovery periods, since otherwise our results could be driven by different firms being in the two periods. We find strong support for the expropriation hypothesis: compared to well-governed firms, poorly governed firms drop more in stock price during the crisis period, but experience significantly more increase in stock prices during the recovery period. Our finding that poorly governed firms experience a greater rebound in stock prices during the recovery period is inconsistent with the information-based explanation. The information-based explanation suggests that greater investor awareness of corporate governance problems revealed during a crisis period is the main reason for the poor performance of firms with weak corporate governance. Since the economic recovery does not change investor awareness of these firms corporate governance, this explanation also suggests that during the recovery period, the rebound in the stock prices of poorly governed firms should be modest or at least no greater than that of well-governed firms. Our evidence shows the opposite results: during the recovery period, the stock return performance of firms with weaker corporate governance is better than that of firms with better corporate governance. We also find that for Korean sample firms, those with weaker corporate governance suffer larger loss of accounting profits during the crisis period, but experience a greater rebound in accounting profits during the recovery period. This result provides further evidence that is consistent with our model. To control for the alternative explanations (i.e., risk and overreaction) in addressing the link between corporate governance and firm value, we include beta and past holding period returns as explanatory variables in the regressions of stock returns during both the crisis and recovery periods. If poorly governed firms have a higher sensitivity to changes in market conditions, they are likely to suffer more when the market performs poorly but perform better when the market recovers. Similarly, firms whose stock prices drop more during the crisis period may perform better during the recovery period because of the contrarian effects in stock returns that have been emphasized in the asset-pricing literature. Consistent 4

7 with these arguments, we find that the coefficients on beta and past holding period returns have the predicted signs in both crisis and recovery period regressions. However, controlling for beta and overreaction effects does not attenuate the impact of governance variables on the stock return. To provide further support to the expropriation hypothesis, we provide ancillary evidence showing that the bad news announcements about the deteriorating economic environment in Korea (e.g., a downgrade of the sovereign rating of Korea by Moody s or other rating agencies, nationalization of the commercial bank by the Korean government, chaebol bankruptcy, etc.) affect firms with weak corporate governance more negatively than they do those firms with good corporate governance. In contrast, the good news announcements that are associated with brighter prospects for future investment opportunities (e.g., an upgrade of the sovereign rating of Korea by Moody s or other rating agencies, settlement of the negotiation on foreign debt payments, abolition of restrictions on foreign direct equity investment, etc.) affect firms with weak corporate governance more positively than they do those firms with good corporate governance. To the extent that these news events are largely unexpected and thus represent relatively exogenous shocks that significantly affect the expected return on investment, our results further confirm the expropriation hypothesis in explaining the link between corporate governance and firm value. We perform several robustness checks on the data. We use various measures of the quality of corporate governance and find evidence that is consistent with the expropriation hypothesis. We also experiment with alternative time periods for the shock and recovery periods and find robust results. Our findings have an important implication for the growing literature on law and finance. The existing literature has demonstrated the importance of investor protection in the various aspects of financial markets. For example, La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998), hereafter referred to as LLSV, show that countries with better investor protection have larger and deeper capital markets. Markets with better investor protection also have higher valuation of listed firms relative to their assets (LLSV, 2002; Claessens, Djankov, and Lang, 2000), a larger number of listed firms (LLSV, 1997), and higher quality of accounting information (Hung, 2001; Ball, Robin, and Wu, 2002; Fan and Wong, 2002; Leuz, Nanda, and Wysocki, 2003). Furthermore, better investor protection enables firms to make 5

8 greater use of external finance (LLSV, 1998) and larger investments from external funds (Rajan and Zingales, 1998; Demirgüç-Kunt and Maksimovic, 1998). In contrast, poor investor protection increases liquidity costs (Brockman and Chung, 2003), and impedes informed arbitrage that capitalizes on firmspecific information, thereby resulting in less efficient stock prices (Morck, Yeung, and Yu, 2000). Finally, Djankov, La Porta, Lopez-de-Silanes, and Shleifer (2007) develop the anti-self-dealing index, and show that it predicts a variety of stock market outcomes. As LLSV (2000) argue, the fundamental premise of this literature is the importance of investor protection in preventing the expropriation of minority investors. Our evidence shows that investor expropriation is indeed the main channel through which corporate governance affects firm value. Although our results for the recovery period generally support the expropriation hypothesis, they are also consistent with controlling shareholders propping behavior ( negative tunneling in which controlling shareholders use their private cash to temporarily prop up troubled group affiliates (Friedman, Johnson, and Mitton (2003)). During the recovery period, controlling shareholders in business groups may use their private funds to prop up affiliated firms so as to bring in more capital quickly or to build a reputation that can be used to raise outside capital. Such propping activities improve firm value and lead to large rebounds in the stock prices, thereby benefitting a firm s outside shareholders. Even though propping reduces controlling shareholders current wealth, in the long run it can produce a net benefit to controlling shareholders by enhancing the values of their options to expropriate firms future profits (Friedman, Johnson, and Mitton (2003)). While we do not explicitly model this propping-based explanation as an alternative to our expropriation hypothesis, our results during the recovery period are consistent with both the expropriation and propping arguments. 1 The rest of the paper proceeds as follows. In Section 2, we develop a simple model of managerial expropriation. In Section 3, we discuss the data, variables, and sample characteristics. It is followed by the discussion of the main results for the cross-sectional determinants of firm performance in Korea during the crisis and post-crisis periods. Section 5 presents the results of robustness tests and shows 1 See Bae, Cheon, and Kang (2008) for evidence on propping in Korean markets. 6

9 further evidence supporting the expropriation hypothesis. In Section 6, we report the results from the outof-sample tests using other Asian and Latin American firms. Finally, we present summary and concluding remarks in Section A simple model of managerial expropriation In this section, we develop a simple model of managerial expropriation during crisis and recovery periods. To simplify the analysis, we consider a one-period model in which a firm has an investment project and initial capital endowment of w init. The payoff from the project is π I, which depends on both the investment amount, I, and investment profitability, π. The net present value of the project is π I I. We assume that the profitability of the firm s project is independent of the level of managerial expropriation, but depends only on whether the firm is in the crisis period or the recovery period. In practice, a firm s profitability is likely to be affected by both macroeconomic conditions and its managerial decisions. However, given that the economic conditions significantly affect a firm s investment opportunities and their profit prospects, it appears reasonable to assume that macro-factors play a key role in determining its profitability. Thus, in our model, the firm s manager cannot affect the profitability of its investment, but he can only alter the extent of expropriation and thus the level of investment. Furthermore, although the firm can obtain w init from internal operations as well as external financing, given the difficulties in obtaining external financing during a crisis, we assume that the firm obtains the entire w init from its previous operations. 2 Finally, since we intend to apply our model to both the crisis and the recovery periods, ideally we need a two-period dynamic model, which begins with the crisis period and ends with the recovery period. 2 We find that during our sample period of , our sample median Korean firm could not issue any bonds and had difficulty extending its long-term bank debt. Moreover, although the median Korean firm managed to issue a small amount of equity during the recovery period of (the number of shares issued as a percentage of total shares outstanding was only a little over 10% of the figure in the pre-crisis period of ), it could not issue any equity during the crisis period of

10 However, mainly for ease of exposition, we focus on a one-period model and assume that w init for the crisis period is exogenously given. However, we link the crisis period to the recovery period by assuming the firm s initial investable capital for the recovery period to be equal to the capital produced from its investment during the crisis period. That is, w init for the recovery period is equal to the cash flow generated from the firm s investment in the crisis period. 3 Following LLSV (2002), we assume that the firm has one controlling shareholder who is also the firm s manager. The manager owns α fraction of the firm and can divert the firm s capital to benefit himself at the expense of minority shareholders. 4 Diversion of firm resources generally requires costly transactions (Burkart, Gromb, and Panunzi, 1998; Johnson et al., 2000; LLSV, 2002). However, as the manager commands greater expropriation power, he incurs less cost and derives more benefits from diverting capital. We assume that the manager receives a benefit of μ(w init w ) from expropriating w init w amount of capital, where μ is the manager s expropriation power. The manager first decides how much of the firm s initial capital w init to expropriate and then how much of the remaining capital to invest in the project. Note that when the manager expropriates w init w amount of capital, the remaining capital available for investment is w. Since w init is unaffected by the manager s decisions in the current period, choosing w init w is equivalent to choosing w. Thus, the manager s optimal choices are determined by the following problem: Max I,w α w + π I I + μ(w init w ) (1) Subject to I w, (2) w init w 0 (3) 3 To the extent that firms are unlikely to be able to forecast how long the unanticipated economic crisis will last at its onset, it will be very challenging for firms to plan ahead for recovery. Furthermore, intuitions following our analysis below suggest that a full-fledged dynamic model is not likely to change the key features of our predictions based on the one-period model. 4 Managerial expropriation can take many forms. For example, it can take subtle legal forms, such as dilutive share issues that discriminate against minority shareholders and mergers between affiliated firms to siphon resources out of the bidder or the target. It can also take the form of outright theft or fraud. 8

11 In equation (1), α w + π I I is the manager s share of the firm s end-of-period value, and μ(w init w ) is his gain from expropriation. Condition (2) states that investment I should be smaller than the amount of capital available for investment after managerial expropriation, w. Condition (3) is the capital constraint that requires that w be smaller than the firm s initial investable capital. To solve for the manager s optimal decisions, consider first his choice of w. From the manager s objective function in equation (1), a dollar increase in w reduces his gains from expropriation by μ but increases his gains from managerial equity holding by α. If μ α, the manager has no incentive to expropriate minority shareholders. In the rest of the analysis, we follow Johnson et al. (2000), by assuming that α is small enough so that μ > α. Under this assumption, the amount of expropriation is always non-negative (i.e., w w init ). When μ > α, the manager chooses to minimize w so that constraint (2) is binding: w = I. (4) Condition (4) says that the manager expropriates all of the firm s capital except for the minimum amount needed for investment. Given (4), the firm s end-of-period value, w + π I I, will be π w, so the manager s objective function in (1) becomes απ w + μ(w init w ). (5) Maximizing the payoff in equation (5) subject to constraint (3) shows that the optimal level of w, denoted as w, is given by w = Min{ απ 2μ 2, w init }. (6) Equation (6) identifies the key factors that affect the manager s incentive to divert the firm s capital. First, when the firm s initial capital w init is large relative to απ 2μ 2, then w = απ 2μ 2. Thus, as managerial equity ownership α increases, the profitability of the firm s project π improves, or as managerial expropriation power μ declines, the firm s investment w increases while managerial expropriation w init w decreases. The intuition for this result is straightforward. As α or π increases, or 9

12 as μ declines, the manager s benefits from the firm s investment, the first term in (5), increase relative to his gain from expropriation, the second term in (5), resulting in greater managerial incentive to invest in the project and weaker incentive to expropriate capital. Second, when the firm has limited initial capital such that w init < απ 2μ 2, then w = w init. In this case, there is no managerial expropriation. To separately examine the implications of the model for the crisis and recovery periods, let s denote 1) the profitability for the crisis and recovery periods as π c and π r, respectively, 2) the investable capital at c the beginning of the crisis and recovery periods as w init r and w init, respectively, and 3) the optimal investment in the crisis and recovery periods as w c and w r, respectively. To solve for the manager s optimal decisions during the crisis period, we assume that at the beginning of the crisis period the firm is endowed with a large amount of initial investable capital. c Assumption 1: For the crisis period, w init > απc 2μ 2. Assumption 1 is motivated by the fact that the external shock at the beginning of the crisis substantially reduces the profit prospect of the firm s investment, causing π c c to be very small. Thus, w init, which corresponds to the capital generated from the firm s operations during the pre-crisis period, is likely to exceed the capital needed for investment during the crisis period. It follows from equation (6) and Assumption 1 that the manager s optimal choice of investment during the crisis period is w c = απc 2μ 2. The resulting managerial expropriation is c w init w c c = w init απc 2μ 2. (7) The amount of capital produced from the investment is π c w c = α(πc ) 2 2μ. (8) Note that π c w c is also the value of the firm at the end of the crisis period. r When the recovery period begins, the firm s initial capital w init is equal to the capital produced during the crisis period, π c w c in (8), that is, 10

13 r. w init = α(πc ) 2 2μ (9) To solve for the manager s optimal decisions during the recovery period, we assume that at the beginning of the recovery period, the firm has limited capital available for investment. r Assumption 2: For the recovery period, w init < απr 2μ 2. This assumption is motivated by the fact that the economic recovery substantially improves the firm s investment prospect, so the profitability of the firm s project during the recovery period, π r, is expected to be significantly larger than that during the crisis period, π c. It follows from equation (6) and Assumption 2 that during the recovery period the manager has few incentives to expropriate and the optimal investment is w r r = w init (10) Therefore, firm value at the end of the recovery period is π r r w init. We can summarize the above discussions of managerial expropriation during the crisis and recovery periods as follows: Proposition 1: During the crisis period, the managerial incentive to expropriate minority shareholders increases with the manager s expropriation power and decreases with his equity ownership in the firm. However, during the recovery period, the manager has few incentives to expropriate minority shareholders. Proposition 1 is similar to Jensen s (1986) free cash flow argument in that excess cash and low growth opportunities result in agency problems. During the crisis period, the firm tends to have a relatively large amount of initial investable capital, but few profitable investment opportunities. This opens the door for agency conflicts, such as managerial expropriation and investment in negative net present value projects. Expropriation is possible due to the manager s significant controlling power, but 11

14 his equity ownership alleviates the extent of expropriation by aligning his interests with those of minority shareholders. On the other hand, coming off the crisis and entering the recovery period, the firm is likely to have better investment prospects but with limited investable capital due to the extant managerial expropriation during the crisis period. Thus, during the recovery period, the manager has few incentives to engage in expropriation. The above analysis provides two testable predictions on the relations between change in firm value and the manager s expropriation power and equity ownership. For the crisis period, the firm is initially c worth w init this period is and is worth π c w c at the end of the period. Thus, the percentage change in firm value over. (π c w c c w init ) (11) c w init We know from equation (8) that π c w c is decreasing in the manager s expropriation power but increasing in his equity ownership, so the change in firm value should also be decreasing in his expropriation power but increasing in his equity ownership. The following prediction summarizes the dependence of the change in firm value during the crisis period on the manager s expropriation power and equity ownership. Prediction 1: The change in firm value during the crisis is negatively related to the manager s expropriation power but is positively related to his equity ownership in the firm. r At the beginning of the recovery period, the firm value is w init, and at the end of the recovery period, it is π r r w init. Thus, during the recovery period, the change in firm value is (π r r w init r w init ) r w init = ( πr r ) 1. (12) w init r Recall from equation (9) that w init is decreasing in the manager s expropriation power but increasing 12

15 in his equity ownership. Thus, equation (12) leads to the following testable prediction: Prediction 2: The change in firm value during the recovery period is positively related to the manager s expropriation power but is negatively related to his equity ownership in the firm. As we pointed out previously, Prediction 2 does not mean either that during the recovery period managerial expropriation contributes to firm value or that managerial equity ownership reduces firm value. As shown above, during the recovery period the manager has few incentives to expropriate. The role of managerial expropriation power and managerial equity ownership in explaining the change in firm value during the recovery period comes indirectly from their influence on the level of initial investable capital, which is determined by the capital produced from investment during the crisis period. During the crisis period, as the manager s expropriation power increases and/or equity ownership declines, his incentives to expropriate minority shareholders increase and thus he allocates less capital to investments, resulting in a smaller amount of investable capital available when the recovery arrives. As illustrated in Figure 1, because of the concavity in the firm s production technology, there is decreasing return to investment. Thus, for firms with high managerial expropriation power and/or low managerial equity ownership, the limited amounts of initial investable capital produce, on a per dollar investment basis, a large increase in firm value. 3. Data, key variables, and sample characteristics 3.1. Data Our initial sample consists of 608 non-financial firms listed on the Korean Stock Exchange (KSE) during , which covers both the crisis and recovery periods. To check the robustness of our results, we perform an out-of-sample test using a sample of 598 other East Asian firms from Hong Kong, Indonesia, Malaysia, Philippines, and Singapore, Taiwan, and Thailand during the same periods and a 13

16 sample of 302 Latin Ameriacan firms from Argentina, Brazil, Chile, and Mexico during the 2001 Argentine economic crisis and recovery periods. We discuss the results of using these alternative samples in Section 5. Since focusing on a single nation (i.e., Korea) allows us to examine corporate governance and firm-specific measures at a level of detail that would be hard to aggregate across countries, our main analysis is based on the single country of Korea. Figure 2 shows the changes in the Korean Composite Stock Price Index (KOSPI) from January 1996 to December The KOSPI is a market-capitalization-weighted price index of all firms listed on the KSE and is the index that is most widely used to evaluate market performance. At the end of January 1996, the KOSPI was 879. It went through periods of ups and downs, and reached its highest point at the end of June 1997, when it stood at 944. The first event that appears to trigger the Asian financial crisis is the announcement on July 2, 1997 regarding the Thai baht s devaluation (Chowdhry and Goyal, 2000). After that, the KOSPI started to fall until September 1998, at which point the index level had dropped to 406. Thus, as in the previous study by Baek, Kang, and Park (2004), we set the period from July 1997 to September 1998 as the crisis period. In October 1998, the Korean stock market recovered its previous losses. In particular, during 1999, investor optimism gained momentum and the stock market performance was spectacular. At the end of 1999, the KOSPI reached a new high of This movement in the KOSPI suggests that investors would have realized a holding period return of about 300% if they had invested in the market portfolio from October 1998 to December Thus, we set the period from October 1998 to December 1999 as the recovery period Key variables We measure the change in firm value during the crisis (recovery) period by the holding period stock return from the beginning of the crisis (recovery) to the end of crisis (recovery) and use it as the dependent variable. The explanatory variables include the following. 14

17 Expropriation variables: As a direct empirical measure of the manager s expropriation power (variable μ in our model), we use the disparity between cash flow rights and voting rights of the controlling shareholders. The controlling shareholders with high ownership disparity may have strong incentives to siphon resources out of the firm to benefit themselves. We measure the cash flow rights of the controlling shareholders (defined as owner-managers and their family members) as the sum of the direct equity ownership and the product of the ownership stakes obtained indirectly. We obtain the information on these stakes by tracing up to two layers along the control chain in a pyramid structure. We measure voting rights as the sum of direct equity ownership and the minimum value of ownership in the chain of voting rights. We use both total and block ownership to estimate voting rights. By obtaining data on total holdings, we can accurately measure the controlling shareholders voting rights. We compute the divergence between cash flow rights and control rights as the logarithm of the ratio of voting rights to cash flow rights (disparity variable). We use equity ownership by controlling shareholders as the measure for managerial equity ownership (variable α in our model). Following Mitton (2002) and Baek, Kang, and Park (2004), we use two additional variables related to block holdings as alternative measures of equity ownership. The first measure is the sum of block holdings by all shareholders who own 5% or more of issued shares. The second measure is the block holdings by the largest shareholder owning 5% or more of issued shares. We further separate the block holdings by the largest shareholder into two components, the largest managerial block ownership and the largest nonmanagerial block ownership. As an alternative measure for controlling shareholders expropriation power μ, we use the extent of a firm s diversification. The degree of expropriation may be more severe for highly diversified firms, as these firms tend to suffer more from information asymmetry and agency costs (Mitton, 2002; Lins and Servaes, 1999). To measure diversification, we use a dummy variable that takes a value of zero if 90% or more of a firm s sales come from one three-digit SIC and a value of one otherwise. We then interact this 15

18 diversification dummy with the diversity of investment opportunities. To measure the diversity of investment opportunities, we follow Rajan, Servaes, and Zingales (2000). We first identify competing undiversified firms for each segment of diversified firms at a four-digit SIC level and then measure the standard deviation of the Tobin s q (book value of debt plus market value of equity/total assets) of matched undiversified firms for all segments of a diversified firm. We define diversified firms with standard deviations above (below) the median for all diversified firms as having a high (low) variation of investment opportunities. Risk and overreaction variables: We use beta as the key measure of firm risk. We estimate beta by the slope of the market model regression. To estimate beta before the crisis (recovery) period, we use twoyear daily stock returns preceding the crisis (recovery) period. We also include as a risk variable firmspecific risk measured as the standard deviation of residual errors from the market model regression. A contrarian (overreaction) effect in stock returns predicts that better performing firms before the crisis period drop more during the crisis period and worse performing firms during the crisis period perform better during the recovery period. To control for such a contrarian effect, we use past holding period returns (HPRs) as our key measure. We compute the HPR of firm i between t 1 and t 2 as HPRi (t 1, t 2 ) = (1 + Ri,t 1 )(1 + Ri,t 1+1 )(1 + Ri,t 1+2 )(1 + Ri,t 1+3 ) (1 + Ri,t 2 ) 1 where Ri,t is the daily return of firm i at time t. As past holding period returns for the shock and recovery periods, we use the HPR during year 1996 and the HPR during the shock period, respectively. Control variables: As in Baek, Kang, and Park (2004), we control for several other variables that may affect firm performance during the crisis and recovery periods. First, we control for firm size (logarithm of total assets). Larger firms tend to have easy access to external finance and suffer less from information asymmetry. Thus, we expect larger firms to be less vulnerable to external shock. To control for the 16

19 nonlinear effect of firm size on HPR, we include the square of the logarithm of total assets as an additional control variable. In addition, we control for leverage (total debt / total assets). Since highly leveraged firms have difficulty obtaining external financing during a crisis, we expect such firms to experience a larger drop in equity value. In Korea, a large business group is often referred to as a "chaebol." The Korea Fair Trade Commission (KFTC) ranks business groups according to the size of their total assets and identifies the 30 largest business groups. Chaebol firms that belong to the 30 largest business groups operate in many different industries, are bound together by a nexus of explicit and implicit contracts, maintain substantial business ties with other firms in their group, and are mostly family controlled. 5 They are also characterized by an extensive arrangement of pyramidal or multilayered shareholding agreements and the existence of cross-debt guarantees among member firms. 6 If the chaebol affiliation allows risk-sharing among chaebol firms, we expect chaebol firms to suffer less during a crisis. We use a dummy variable that equals one if a firm belongs to one of the 30 largest business groups and zero otherwise. We expect firms with high foreign ownership and those having a listed ADR to suffer less during a crisis. To control for these effects, we use equity ownership by foreign investors and a dummy variable that takes the value of one if the firm has an ADR listed in the United States. We control for inefficiency in resource allocation among member firms within the same chaebol by including the ratio of financial securities invested in affiliated (non-affiliated) firms to the firm s total assets. Finally, we control for other determinants of firms stock returns, including future investment opportunity (Tobin s q), liquidity (ratio of cash flow (operating income plus depreciation) to total assets), 5 Fogel (2006) shows that greater oligarchic family control over large corporations, which is the case for chaebol firms, is associated with worse social economic outcomes. 6 Chang (2003) examines the simultaneous relationship between ownership and firm performance for chaebol affiliates. He shows that using insider information, chaebol families and their affiliates take higher ownership stakes in more profitable and more promising firms and transfer profits to affiliates through intragroup trade. 17

20 and industry effects (industry dummies) Sample characteristics Table 1 provides summary statistics of our sample firms as of the end of fiscal years 1996 and 1997, respectively. We obtain the daily stock return data from the Stock Database of the Korea Securities Research Institute, which includes all firms listed on the KSE, and financial data from the Listed Company Database of the Korean Listed Companies Association. The average ratio of voting rights to cash flow rights is 3.22 in 1996 and 1.91 in When we closely examine the ratio of voting rights to cash flow rights, we see that outlier problems are much more severe in 1996 than in When we delete observations in which the ratio of voting rights to cash flow rights is larger than the 99 th percentile of the sample, the average ratio in 1996 becomes 2.31, but the average ratio in 1997 is In 1996, the average equity ownership by controlling shareholders and the average equity ownership by affiliated firms are 20.51% and 8.79%, respectively. The corresponding values in 1997 are 19.69% and 10.91%. Tests of differences in mean and median equity ownership by controlling shareholders between 1996 and 1997 do not reject the null hypothesis that they are equal, but those in mean and median equity ownership by affiliated firms between 1996 and 1997 reject the null hypothesis that they are equal. These results suggest that during the crisis period, rather than using their own wealth, controlling shareholders use the resources of member firms to increase their control power in the chaebol. The sum of block ownership by all shareholders averages 30.96% in 1996 and 31.15% in The average of largest blockholder concentration, as measured by the block holdings of the largest shareholder owning 5% or more of shares, is 19.29% in 1996 and 20.97% in The average largest managerial blockholder concentration, i.e., block holdings by owners involved with management, is 11.3% in 1996 and 12.43% in The corresponding values for the largest non-managerial blockholder concentration, i.e., block holdings by largest shareholders who are not associated with management, are 7.99% and 18

21 8.54%, respectively. The median equity ownership by foreign investors is 1.62% in 1996 and 1.07% in 1997, suggesting that during the crisis, the median foreign investor reduces her previous holdings by almost 34%. All of these differences in block ownership and foreign ownership between 1996 and 1997 are statistically significant at the 0.01 level. Not surprisingly, betas show an increase during the crisis period. The mean betas are 0.85 in 1996 and 0.9 in However, firm-specific risk shows little changes. The past HPR shows that firm value is significantly lower in 1997 than in The average of past HPRs in 1996 is 0.17, while the corresponding number in 1997 is The average size of firms measured by total assets is 621 billion won in 1996 and 713 billion won in 1997, and the average leverage ratio is 70.1% in 1996 and 73.3% in In 1996 and 1997, the ratios of cash flow (operating income plus depreciation) to total assets average 5.5% and 4.7%, respectively, and the means of Tobin s q are 1.1 and The average ratio of financial investment in affiliated (unaffiliated) firms to total assets is 7.9% (4.7%) in 1996 and 6.9% (4.0%) in We note a slight increase in the proportion of sample firms whose ADR is listed in the U.S., from 2.65% in 1996 to 2.82% in The proportion of diversified firms decreased from 32.28% in 1996 to 29.91% in Determinants of firm performance during the crisis and recovery periods This section presents the main results of the paper. First, using univariate tests, we examine the relation between expropriation variables and firm performance. Then, using multivariate regression analyses, we show that the most important determinants of firm performance during the crisis and recovery periods are expropriation variables Univariate test We partition our sample firms by the sample medians of key expropriation variables at the end of fiscal years 1996 and 1997, respectively, and then compare the HPRs during the shock (July 1997 to 19

22 September 1998) and recovery (October 1998 to December 1999) periods. In Panel A of Table 2, we use the ratio of voting rights to cash flow rights as a key expropriation variable. During the shock period, firms with high disparity realize an average (median) HPR of -73.5% (- 79.7%), while firms with low disparity realize an average (median) HPR of -68.2% (-73.9%). The differences in mean and median HPRs between these two subsamples are significant at the 0.05 level. These results are consistent with Prediction 1 of our model. During the recovery period, firms with high disparity realize an average (median) HPR of 158.7% (96.4%), while firms with low disparity realize an average (median) HPR of 123.6% (50.7%). The differences in mean and median HPRs between these two subsamples are significant at the 0.01 level. These results, together with those during the shock period, suggest that poorly governed firms suffer more during the shock period but perform better during the recovery period. This evidence of reversal in stock returns supports Prediction 2 of our model. In Panel B of Table 2, we use equity ownership by controlling shareholders as our key expropriation variable. We find that firms with lower equity ownership by controlling shareholders suffer more during the shock period but bounce back more during the recovery period, again supporting Predictions 1 and 2 of our model. However, the greater reversal of stock prices for firms with weak governance could be due to factors that are not related to ownership structure. It could be that ownership variables are correlated with other aspects of firm characteristics, such as firm risk. For example, firms with high betas perform worse during the shock period and better during the recovery period. While not reported in the paper, we split the sample firms according to the sample median beta and examine whether stock return performance differs according to the beta. During the shock period, high-beta firms experience a mean (median) loss of about 71.4% (76.9%), while the corresponding loss is 67.5% (74.5%) for low-beta firms. However, during the recovery period, this pattern is reversed. High-beta firms experience a mean (median) gain of about 151.1% (63.9%), while the corresponding gain is 110.3% (58.9%) for low beta firms. These findings underscore the importance of controlling for firm risk in detecting the relation between firm performance and 20

23 governance variables. In Panel C of Table 2, we use sum of block ownership by all shareholders as our key expropriation variable. We find that firms with lower block ownership by all shareholders realize more loss during the shock period but gain more during the recovery period, supporting the expropriation hypothesis. We further break down block ownership into managerial block ownership and non-managerial block ownership and examine their relation to firm performance. Panel D of Table 2 shows that firms with lower managerial block ownership have lower returns during the crisis period and higher returns during the recovery period. Non-managerial block ownership, however, shows no relation to firm performance (not reported) Cross-sectional variation in HPRs In this section we estimate cross-sectional regression of HPRs on firm-specific variables that capture expropriation while controlling for other variables that may affect firm value such as risk and overreaction. Table 3 shows the results from the cross-sectional regression. As the dependent variable, regressions (1) through (3) use the HPRs during the shock period, and regressions (4) through (6) use the HPRs during the recovery period. Regression (1) includes the logarithm of the ratio of voting rights to cash flow rights as a key explanatory variable. 7 The coefficient estimate on this variable is negative and significant, showing that firms with high disparity between voting rights and cash flow rights perform worse during the crisis period. In regression (2), we include beta, firm-specific risk, and past HPRs as explanatory variables, in addition to ownership disparity. We find that the coefficient estimates on beta, previous HPRs, and ownership disparity are negative and significant, showing the importance of including risk and contrarian 7 In untabulated tests, we also experiment with the difference between control rights and cash flow rights instead of their ratio as an alternative measure of the expropriation variable. We obtain similar results as those reported in the paper. 21

24 variables in the regression. In regression (3), we control for all other variables. We add the logarithm of total assets, squared logarithm of total assets, total debt over total assets, Top 30 chaebol dummy, equity ownership by foreign investors, ADR dummy, the ratio of financial investments in affiliated firms to total assets, the ratio of financial investments in unaffiliated firms to total assets, Tobin s q, and the ratio of cash flow to total assets. We also include four industry dummy variables (construction, manufacturing, wholesale and retail, and transportation and services) to control for a possible industry effect. To conserve space, we do not report coefficients on these industry dummies. Controlling for these variables increases the adjusted R 2 of the regression to 11%, suggesting that ownership disparity, beta, firm-specific risk, and previous HPRs, together with other control variables, are able to capture successfully the variation of the HPRs during the crisis. Adding control variables does not affect the significance of the disparity variable. This result is consistent with Prediction 1 of our model. Among the control variables, previous HPR, logarithm of total assets, and leverage variables are negatively and significantly related to HPR during the shock period. In contrast, squared logarithm of total assets, chaebol dummy, and ADR dummy are positively and significantly related to HPR. Our regression analyses for the recovery period begin with regression (4), in which we regress HPRs during the recovery period on the disparity variable. Supporting Prediction 2 of our model, the result shows that the coefficient estimate on the disparity variable is positive and significant at the 0.01 level. In regression (5), we add beta, firm-specific risk, and previous HPRs as explanatory variables. We find that the coefficient estimate on beta is significantly positive, whereas the coefficient estimate on previous HPR is significantly negative. In regression (6), we control for other variables. Adding the control variables does not change the magnitude and significance levels of our key variables. In Table 4, we use equity ownership by controlling shareholders as the key expropriation measure. In regression (1), we replace the disparity variable in regression (1) of Table 3 with equity ownership by controlling shareholders. Consistent with Prediction 1 of our model, we find that the coefficient estimate 22