Foreign Exchange Exposure and Short-Term Cash Flow Sensitivity Laura T. Starks and Kelsey D. Wei * September 2005

Abstract We hypothesize that since exchange rate risk can affect firms’ cash flows directly, the relation between exchange rate movements and expected costs of financial distress should explain cross-sectional variation in stock return sensitivity to exchange rate movements. Controlling for market risk and interest rate risk, we find, on both the individual firm and industry level, the magnitude of exchange rate exposure is related to proxies for probabilities of financial distress, growth opportunities and product uniqueness. Further, during large exchange rate movements, firms with higher expected costs of financial distress show larger exposures as measured by their larger abnormal returns in response to exchange rate shocks. JEL classification: F31; G30 Keywords: Foreign Exchange Exposure; Financial Distress; Exchange Rate Shocks

*

Starks is from Department of Finance, McCombs School of Business, the University of Texas at Austin, Austin, TX 78712. Tel.: 512-471-5899; fax: 512-471-5073; email: [email protected]. Wei is from School of Management, Binghamton University-SUNY, Binghamton, NY 13902. Tel.: 607-777-2306; fax: 607-777-4422; email: [email protected]. The authors would like to thank Andres Almazan, Warren Bailey, Ty Callahan, Katheryn Dewenter, Li Gan, Stephen Magee, Sridhar Sundaram, Sheridan Titman, Hong Yan and participants at seminars at the University of Texas at Austin and at the FMA meetings for helpful comments. All errors are our own.

Foreign Exchange Exposure and Short-Term Cash Flow Sensitivity

Abstract We hypothesize that since exchange rate risk can affect firms’ cash flows directly, the relation between exchange rate movements and expected costs of financial distress should explain cross-sectional variation in stock return sensitivity to exchange rate movements. Controlling for market risk and interest rate risk, we find, on both the individual firm and industry level, the magnitude of exchange rate exposure is related to proxies for probabilities of financial distress, growth opportunities and product uniqueness. Further, during large exchange rate movements, firms with higher expected costs of financial distress show larger exposures as measured by their larger abnormal returns in response to exchange rate shocks.

Foreign Exchange Exposure and Short-Term Cash Flow Sensitivity

1. Introduction The increasing globalization of product and financial markets means that many firms are now exposed to changes in currency values. These changes can have dramatic effects on their operations. For example, in 1999, Sony announced that its profits fell 25% due to the effects of a yen appreciation on their foreign sales (Landers, 1999). Similarly, Xerox announced that its reduced credit rating prevented them from hedging the currency risk of their foreign operations, resulting in substantially larger losses than expected during 2001. In order to measure the effects of exchange rate changes on firm value, many studies have followed the Adler and Dumas (1984) empirical research design, which is to estimate currency exposure effects by regressing stock returns on the percentage change in exchange rates. However, this empirical approach finds low proportions of firms and industries with significant exposure, both within the U.S. and abroad. (See, for example, Jorion, 1990; Bodnar and Gentry, 1993; Amihud, 1994; and Griffin and Stulz, 2001.) Consequently, despite observed impacts of currency exposure on firms’ cash flows, the correlations between contemporaneous exchange rate movements and firms’ stock returns appear to be very low.1 Given previous theoretical studies that have demonstrated the effect of exchange rate fluctuations on firms’ cash flow volatilities (e,g., Smith and Stulz, 1985; Stulz, 1984; Froot, Sharfstein and Stein, 1993), in this paper we argue that exchange rate movements are more likely to affect a firm through direct effects on short-term cash flows. How much these direct effects on short-term cash flows matter for the firm’s value then depends on the firm’s sensitivity to short-term cash flow volatility. For example, if a firm’s liquidity is already low, then a large fluctuation in its cash flows due to an exchange rate movement can push the firm into financial distress, and as a result, lead to changes in its fundamental value. Similarly, when a firm has substantial growth 1

A stronger relationship has been found in specific industries. See, for example, Williamson (2001) on the automobile industry.

2

opportunities, exchange rate movements can have greater effects on firm value due to the firm’s larger underinvestment costs. If exchange rate changes have pronounced effects on fundamental values primarily when the resulting short-term cash flow fluctuations force the firm into financial distress or cause it to forsake positive NPV investment opportunities, the magnitude of exposures would vary cross-sectionally with the expected cost of financial distress in terms of both the probability of distress and the cost related to it. Our hypothesis implies that firms that have greater expected costs of financial distress should be more exposed to exchange rate risk. That is, the relation between exchange rate movements and financial distress costs should explain cross-sectional variation in stock return sensitivity to exchange rate movements. Recognizing this relation should also reconcile the theoretical prediction of the effects of exchange rate risk on firms’ cash flows with the empirical evidence that finds limited statistical and economic significance of that exposure. We test our hypotheses regarding the effects of exchange rate risk on firm value using monthly stock return data for a sample of U.S. manufacturing firms during 19782001. We first examine the foreign exchange exposures of these firms through a univariate analysis. At the beginning of each year, we sort firms into three groups with low, medium and high probabilities of financial distress according to a single composite measure (Ohlson’s (1980) O-Score). We then estimate each O-Score group’s average foreign exchange exposure in the subsequent 36 months controlling for the market return and the interest rate. Consistent with our hypothesis, these rolling regressions show that firms that are more likely to be financially distressed have significantly larger absolute exposures to currency risk. To control for the effects of other firm and industry characteristics, we also conduct analyses on the determinants of exposures through multivariate regressions. We find that exchange rate exposures are increasing with both the probability of financial distress and the cost of financial distress if it occurs. Specifically, exchange rate exposures are increasing with O-Score, market-to-book ratio, 3

asset growth rate and product uniqueness. In addition, we find exchange rate exposures to be increasing with foreign sales and decreasing with firm size. We also examine the effects of exchange rate changes on firm value by using what is to our knowledge a unique approach to the question. Specifically, we employ an event study methodology to examine firms’ responses to large currency shocks. Although previous studies generally focus on whether there is a statistically significant relation between exchange rate movements and stock returns over time, the economic significance of foreign exchange exposure may be better illustrated by how stock prices respond to large exchange rate shocks. Thus, we focus on the 70 individual days with the largest exchange rate changes in either direction during the sample period. For each of these events we calculate cumulative abnormal returns for a sample of firms on the CRSP/COMPUSTAT databases. We then regress the absolute value of the cumulative abnormal returns against measures of the firms’ expected costs of financial distress. The results show that firms with higher O-Scores have larger absolute abnormal returns around large exchange rate changes. We also find the cumulative abnormal returns to be greater for firms with greater growth opportunities or more unique products and smaller for larger firms. Given the problems associated with the lack of information on individual firms’ foreign business activities, we examine the foreign exchange exposure of U.S. firms at the industry level where we are able to observe monthly imports and exports. By sorting firms into industry portfolios, we find that on average U.S. manufacturing industries show significant exchange rate exposures in directions that are consistent with the nature of their foreign trade balances. Most interestingly, under a specification that allows exposure to vary both cross-sectionally and over time, we find that the overall foreign exchange exposures of U.S. industries vary with their expected costs of financial distress. This result is consistent with our findings from individual firms’ exposures. Geczy, Minton and Schrand (1997) and He and Ng (1997) argue that a higher expected cost of financial distress may also suggest a greater incentive for firms to hedge. 4

If this is the case, we should expect the effect of financial distress to be stronger on firms’ exposures to more difficult to hedge currencies. That is, currencies of developing countries. However, we do not find support for this hypothesis. In general, the expected costs of financial distress exhibit similar impacts on firms’ exposures to developed currencies and developing currencies, indicating that hedging may not necessarily lead to smaller exposures for firms with greater costs of financial distress and cannot completely insulate firms from foreign exchange rate exposure. The rest of the paper proceeds as follows. In the next section we review the previous evidence on exchange rate exposure. In Section 3 we focus on the relation between foreign exchange exposure and the expected costs of financial distress for individual firms. In Section 4 we present results from an industry level analysis. In Section 5, we explore the role of hedging on currency exposure. Finally, we conclude the paper in Section 6.

2. Overview of Exchange Rate Exposure For firms domiciled in the United States, current empirical studies have documented only weak correlations between exchange rate movements and firm value. For example, Jorion (1990) examines 287 U.S. multinational firms and finds that only 5% of them exhibit significant exposures. Although the evidence for firms domiciled in other countries is somewhat stronger, it is still relatively weak. For example, He and Ng (1998) and Glaum, Brunner and Himmel (2001) investigate Japanese and German firms, respectively, and find a greater relation between stock returns and exchange rate movements. But even in these countries, where presumably the large firms have relatively more foreign trade than do their U.S. counterparts, the percentage of firms with significant return exposures is still less than would be expected. The puzzle of why U.S. firms show such little apparent exchange rate exposure has not been explained. Exchange rate exposure certainly has the potential to be a significant risk factor for firms. As pointed out by Jorion (1990), the volatility of 5

exchange rates is substantially larger than that of interest rates or inflation. Several possible explanations have been offered as to why researchers have documented such small exposures for U.S. firms. First, the small observed exposures may be due to the offsetting nature of currency exposures. Since researchers generally lack complete data on individual firms’ imports, exports and business competitors, they cannot identify which firms are exposed to a given currency. For example, in Brown’s (2001) study of the hedging practices of one U.S. firm, he finds that the firm hedges 24 different currencies due to both extensive foreign sales and the importation of a major portion of their manufacturing inputs. As a result, some studies have chosen to examine exchange rate exposure at the industry level where it is more appropriate to proxy for exchange rate movements with changes of a trade-weighted index. Second, the small observed exposures may be due to the complexity of the firms’ foreign exchange exposures since exchange rate risk can vary over time as well as crosssectionally. For example, it can vary with the level of a firm’s foreign trade, the demand elasticity of the firm’s product, or the competitive reactions of other firms in the same industry. Allayannis (1997), Bodnar, Dumas, and Marston (2002), Allayannis and Ihrig (2001), and Francis, Hasan and Hunter (2005) examine time-varying exposure at the industry level. They provide evidence that exchange rate exposures increase with the level of foreign trade and decrease with firms’ ability to mark up prices and pass through the impact of exchange rate movements to customers. These studies indicate that it is important to measure exposure in a specification that allows it to vary both crosssectionally and over time. Third, with the rapid development of hedging instruments since the 1980s, more firms are actively involved in the management of foreign exchange risk. However, a survey of derivative usage by Bodnar, Hayt and Marston (1998) indicates that although many firms engage in currency hedging, they hedge selectively. Further, Guay and Kothari (2003) find that the potential effects of hedging with derivatives are small compared to firm size. Hentschel and Kothari (2001) find no differences in risk for firms 6

that hedge with derivatives versus those that do not. Given these evidences, we conclude that it is unlikely that hedging can completely insulate firms from currency risk.

3. Foreign Exchange Exposure and Financial Distress of Individual Firms A firm’s exchange rate exposure can come from foreign businesses, foreign competitors, or a business relationship with a party that is exposed to exchange rate risk. Given that firms often have multiple sources of direct and indirect exposure to exchange rate changes, instead of restricting our analysis to firms with foreign businesses, we examine all manufacturing firms with at least 30 months of stock return data available from CRSP and financial information available from Compustat. These firms are from 27 industry groups according to the Fama-French 48 industry classification. To reduce the effect of outliers, for each year of the sample period we only include firms with assets greater than $10 million dollars and which have non-negative book values of equity. Since previous studies have shown that a firm’s foreign exchange exposure is strongly impacted by its foreign sales, we obtain firms’ foreign sales as a percentage of total sales from Worldscope. Worldscope’s coverage on U.S. firms varies significantly overtime. Much fewer firms are covered in the 1980s than in the 1990s. Our sample would have been reduced by more than 65% as the result of lack of foreign sales data. Therefore, we instead measure each firm’s foreign sales as the average foreign sales of the 4-digit industry group to which they belong. By measuring foreign sales at the industry level, we hope to capture variation in foreign business that is related to industry structures and competitive environment. We measure foreign exchange movement as the percentage change in the real, trade-weighted USD exchange rate index against major currencies as published by the Federal Reserve. We check the sensitivity of our tests with respect to a number of alternative exchange rate indices and find almost no differences in the results. Also, as indicated in Jorion (1990), the choice of a nominal or real index is not crucial to the

7

results since inflation is only a small component of the total exchange rate variation over the sample period. Previous studies have measured the foreign exchange exposure of stock returns by regressing stock return on percentage change of foreign exchange controlling for the market return (see, for example, Allanyannis and Ofek, 2001; Bodnar and Gentry, 1993; Bodnar and Wong, 2003; Choi and Prasad, 1995; Glaum, Brunner and Himmel, 2001; Jorion, 1990; Williamson, 2001). Since our main interest is in the relationship between the expected cost of financial distress and foreign exchange exposure, controlling for firms’ exposure to macroeconomic risk is particularly important because firms with high expected costs of financial distress may also be more sensitive to other macroeconomic types of risk. Although the market return can be used to control for the effect of macroeconomic factor, we include real interest rate as an additional control variable to control for the possibility that the sensitivity of return to exchange rate is actually due to interest rate risk. Specifically, we calculate exposure by regressing monthly stock returns on the percentage change of the foreign exchange rate index, returns on the equallyweighted CRSP index, and the return on the 1-month T-bill rate minus the monthly inflation rate: R j ,t = α j + β j Rt fx +γ j EWRETt + δ j INTERESTt + ε j ,t

for i=1...N

(1)

where Ri ,t is the stock return for firm j in month t, Rt fx is the percentage change in the exchange rate index for month t, EWRETt is the equally-weighted CRSP index return and

INTERETt is the real interest rate. The regression coefficient, β j , can then be interpreted as the elasticity of firm j’s value to the exchange rate, controlling for other macroeconomic risks through EWRETt and INTERESTt .2

2

We also estimate foreign exchange exposure using value-weighted CRSP index returns as the proxy for macroeconomic effects. The results are not materially different. See Bodnar and Wong (2003) for discussions of different control variables for macroeconomic risk.

8

3.1 Exchange Rate Exposure and Financial Distress To test our hypothesis that the more a firm is financially distressed, the greater will be the effect of exchange rate movements on the firm’s stock price, we need measures of financial distress. Since financial distress can be directly related short-term liquidity and short and long term leverage. we expected firms’ quick ratio, coverage ratio, total dividend payout and debt ratio to be related to the probability of financial distress. Since some of these variables may potentially be highly correlated with each other, we employ a composite measure for the probability of financial distress, the Ohlson’s (1980) O-Score measure, which provides an estimate of bankruptcy risk for industrial firms.3 The higher is a firm’s O-Score, the more likely it is in financial distress. The model for estimating the O-Score was introduced in 1980, but Begley, Ming and Watts (1996) show that its predictive power for Compustat firms also remained strong in the 1980s. Further, other studies (e.g., Dichev, 1998; Griffin and Lemmon, 2002) have employed the model through the 1990s.4 The O-Score has the advantage of being a summary measure and the weights of the ratios are of less importance than this advantage. More importantly, compared with other variables like total dividend payout or quick ratio, O-Score is less of a choice variable that can be managed through hedging. Previous studies have shown that a firm’s foreign exchange exposure depends on a number of firm and industry characteristics, including the amount and the nature of its 3

The O-score is defined as:

O-score= -1.32-0.407 log (total assets) + 6.03 ⎛⎜ total liability ⎞⎟ -1.43 ⎛⎜ working capital ⎞⎟ + 0.076 ⎜ ⎟ ⎟ ⎜ ⎝ total assets ⎠

⎝

total assets

⎛ current liability ⎞ -1.72 (1 if total liabilities> total assets, o if other wise)-2.37 ⎟⎟ ⎜⎜ ⎝ current assets ⎠

⎠

⎛ net Income ⎞ ⎜⎜ ⎟⎟ − 1.83 ⎝ total assets ⎠

⎛ funds from operation ⎞ +0.285 ( 1 if a net loss for the last two years, 0 otherwise) –0.521 ⎟⎟ ⎜⎜ total liability ⎠ ⎝ ⎛ net incomet − net incomet −1 ⎞ ⎜⎜ ⎟⎟ ⎝ | net incomet | − | net incomet −1 | ⎠

9

foreign operations, the market structure, the competitive reactions of other firms in the same industry and the use of hedging tools. Because determinants such as these change over time, we would expect the firm’s exposure to be time varying. Williamson (2001) finds time variation in the foreign exchange exposure of the auto industry. Bodnar et al (2002) and Allayannis and Ihrig (2001) also examine the effect the time variation of industry structure has on exposure. Given previous evidences of time variation in exchange rate exposure, we first measure exposures using rolling regressions. Each year, we classify firms into three groups of equal number of firms according to their previous year end O-Scores. For each O-Score group, we estimate each firm’s foreign exchange exposure using its monthly stock returns in the subsequent 36 months according to equation (1). This process is repeated for every year from 1978 to 2001. Since previous studies (see, for example, Allayannis and Ihrig, 2001; Bodnar, Dumas, and Marston, 2002; Francis, Hasan and Hunter, 2005) have shown that exchange rate exposures exhibit industry variation due to differences in competitive structures and levels of foreign businesses, we divide firms into O-Score groups within each of the 27 manufacturing industry according to the Fama-French 48 industry classification.5 These rolling regressions will help us understand whether the current level of financial distress leads to foreign exchange exposure in the future. Since we only have information about firms’ foreign sales, we cannot determine whether a firm is a net importer or a net exporter. Thus, we focus on the magnitude rather than the sign of the exposure as estimated using equation (1). Table 1 presents the timeseries average of the absolute value of the foreign exchange exposures for low, medium and high O-Score groups in each of the 27 industries. The results provide strong support 4

We check our results with Altman’s (1968) Z-score. The results are similar. The Spearman rank correlation of O-score and Z-score is about –0.66 for our sample.

10

for our hypothesis that firms that are more financially distressed are more affected by foreign exchange movement. In every industry, the high O-score group has greater exchange rate exposure than the low O-Score group. We also conduct t-tests to test the significance of the differences in exposure between the high and low O-Score groups. Given the overlapping of data in the rolling regressions, the test statistics are calculated based upon Newey-West autocorrelation and heteroskedasticity consistent standard errors. The null hypothesis of equal exposure between low and high O-score groups is significantly rejected for 22 out of the 27 industries. We have also examined the median value of absolute exposures for each O-Score group to make sure that the result is not driven by outliers. The comparison of median values is very much similar to that with the mean. 3.2 Determinants of Foreign Exchange Exposures Although Table 1 establishes the relationship between financial distress and foreign exchange rate exposure, a firm’s foreign exchange exposure is likely to be a function of many other firm and industry characteristics, in addition to their probability of financial distress. For example, cash flow volatility caused by exchange rate movement can be more costly for firms with more investment opportunities due to their greater underinvestment costs. As a result, stock price of these firms should be more sensitive to currency risk. Therefore, although growth opportunities do not directly measure probability of financial distress, they can proxy for costs of financial distress (if it occurs). In this paper, we employ two alternative measures of growth opportunities: the market-to-book ratio (M/B) and the asset growth rate.6

5

We have also examined all results of the paper using 2-digit SIC code to define the industry groups and find similar results. 6 We measure market-to-book ratio as the ratio of the market capitalization to the book value of equity. We calculate book equity as total assets less total liabilities and preferred stock plus deferred taxes and convertible debt. Asset growth rate is computed as the percentage growth of total assets.

11

The cost of financial distress may also depend on the degree to which the firm has specialized products and intangible assets. As discussed in Titman and Wessels (1988), financial distress can be more costly for firms with specialized products because the consequence of disrupted long-run relationships with customers and business partners is more severe. In addition, firms that have a lot of intangible assets are more likely to face liquidity problem when their cash flow is volatile. Thus, we examine the impact of R&D expenses on exchange rate exposure. Since this variable is often missing or equal to zero for many firms in Compustat, we instead use a binary variable that takes the value of one when a firm’s R/D over sales is above the sample average, and zero otherwise. Jorion (1990), Bodnar and Wong (2003), Doidge, Griffin and Williamson (2002) provide evidence that firms’ foreign sales significantly affect the sensitivity of stock returns to exchange rate movement. We therefore hypothesize that the higher the level of foreign sales, the more negative is the firm’s exchange rate exposure. We include a control variable for firm size (total market capitalization). Firm size is included as a control variable rather than a proxy for the probability of financial distress because of its ambiguous relation with the firms’ exchange rate exposure. On the one hand, large firms generally have more access to external and internal financing and therefore have a lower probability of financial distress. On the other hand, large firms are also more likely to conduct multinational business and thus be more affected by exchange rate movements through multiple potential channels. Table 2 presents the correlation matrix of the average firm characteristics. As expected, firms with higher O-Score are smaller on average. They also tend to have fewer investment opportunities and lower levels of foreign sales. In general, since some of the firm characteristics proxy for the probability of financial distress while others are more related to the cost of financial distress if it occurs, their correlations are not very high. 12

We next examine the determinants of foreign exchange exposures through a twostep approach. First, we estimate the sensitivity of firms’ returns to exchange rate exposure through equation (1). The estimated coefficient is then regressed on the average value of firm characteristics controlling for firm size. We also include industry dummies for the 27 manufacturing industries in the cross-sectional analysis to control for factors related to industry structures, such as the markup and pass-through effects, that are important determinants of foreign exchange exposure. Using equation (1), we find that 419 firms out of the 3717 firms in the sample period show significant exposure at 10% significance level. This low proportion of firms with significant exposure is broadly consistent with evidence reported in previous studies (see, for example, Jorion, 1990; Bodnar and Wong, 2003). Similar to Table 1, we first focus on the magnitude of the estimated exchange rate exposures. That is, we estimate the following regression: abs( β jFX ) = a + a1O − Score j + a2 M / B j + a3 Asset Growth j + a4 High R & D j 26

+ a5 Foreign Sales j + a6 Size j + ∑ d n Ind jn + ε j

(2)

n =1

where O-Score, M/B, Asset Growth and size are firms’ average O-Score, market-to-book ratio, asset growth rate and market value during the sample period. High R&D is a dummy variable that takes the value of 1 if a firm’s average R&D/Sales is above the sample median and 0 otherwise. Foreign Sales are average foreign sales over total sales for each firm’s affiliated 4-digit SIC code industry group. Indjn are industry dummies for every industry except for the boxes industry. Table 3 shows the result of the crosssectional regression of the absolute value of the exchange rate exposures on these firm and industry characteristics. Consistent with our hypothesis on the relationship between financial distress and foreign exchange exposure, we find that firms’ exchange rate exposures are significantly positively related to their O-Scores, indicating that firms that 13

are more financially distressed are more affected by exchange rate movements. In addition, Table 3 also shows that firms with higher market-to-book ratios, higher asset growth rates, or higher R&D expenses have significantly larger foreign exchange rate exposures. Since a higher level of foreign sales leads more negative exposure and thus greater absolute value of the exposure, we expect the coefficient on foreign sales to be positive. This is indeed what we find in Table 3. A 1% rise in foreign sales on average increases the sensitivity of stock price to exchange rate movements by 0.56%. Lastly, larger firms seems to have smaller exchange rate exposures. It is possible that they are less likely to be financially distressed, or they simply have more resources to hedge against currency risk, either financially or operationally. Since both firms’ foreign exchange rate exposures and their financial conditions change over time, Table 3 also presents the result of the sub-period analysis. Given the sample period of 24 years and the monthly frequency of the return series, we estimate firms’ foreign exchange exposures in two sub-periods: 1978-1989 and 1990-2001. This allows for sufficient number of time-series observations in each sub-period to ensure the accuracy of our exposure estimates. In each of the two sub-periods, a firm must have at least 30 monthly return observations to be included in the analysis. The sub-period analysis result is broadly consistent with that for the whole sample period although the effects of growth opportunities and uniqueness of products seem to be stronger in the second period. In general, the absolute values of estimated exposures are significantly increasing with proxies for financial distress and the cost of financial distress in both periods. To see whether these measures of expected cost of financial distress used in Table 3 have similar effects on net importers versus net exporters, in Table 4 we allow them to have different coefficients on firms with positive and negative exposures. Particularly, we 14

interaction O-Score, M/B, asset growth rate, and the high R&D dummy with the dummy variables indicating positive and negative estimated exposures, respectively. We expect higher expected cost of financial distress to cause positive exposures to be more positive and negative exposures to be more negative. That is, we estimate the following crosssectional regression:

β jFX = a + a1 D + * O − Score j + a2 D + * M / B j + a3 D + * Asset Growth j + a4 D + * High R & D j + b1 D − * O − Score j + b2 D − * M / B j + b3 D − * Asset Growth j + b4 D − * High R & D j 26

+ cForeign Sales j + dSize j + ∑ d n Ind jn +ε j

(3)

n =1

where D + ( D − ) is a dummy variable that takes the value of 1 if firm j has positive (negative) exposure and zero otherwise. This analysis is done for the whole sample period and for each of the two sub-periods examined in Table 3. As expected, Table 4 shows that although measures of the probability of financial distress and the cost of financial distress have slightly different impacts on firms with positive versus negative exposures, they tend to exuberate the exchange rate exposure for both groups of firms. That is, among firms with positive exposures, those with higher expected cost of financial distress have significantly larger exposures. The same effect is observed among firms with negative exposures. This finding contrasts with that in He and Ng (1998) as they find that the explanatory power of the determinants of exposure is mainly driven by exporting firms. This is probably due to the fact that we have relatively balanced numbers of firms that are net importers versus net exporters in our sample, while the sample of He and Ng (1998) is dominated by exporting firms. Similar to Table 3, the sub-period result is not significantly different from that of the whole period.

15

3.3. Stock Price Reaction to Exchange Rate Shocks So far, we have provided some evidences on the relationship between exchange rate exposures and firms’ average expected cost of financial distress under two assumptions. First, the sensitivity of stock price to currency movement is constant. Second, firm characteristics are relatively stable over time. However, previous studies have shown that a firm’s exposure depends on the level of foreign trade (see, for example, He and Ng, 1998; Doidge et al, 2002), the competitive structure of the market in which it operates (see, for example, Allayannis and Ihrig, 2001; Williamson, 2001; Bodnar, Dumas, and Marston, 2002), and the size of the foreign exchange movements (see Doidge et al, 2002), all of which may be time-varying. Even though the potential misspecification problem is alleviated through rolling regressions and sub-period analyses, the approach of imposing a linear relation between exchange rate risk and return is debatable. Doidge, Griffin and Williamson (2002) provide evidence that firms with large foreign sales underperform firms with no foreign sales during periods of large currency appreciation and overperform them during periods of large currency depreciation. This suggests that the correlation between stock prices and exchange rates may be stronger during large currency shocks. Further, our results in previous sections suggest that although the correlation between stock price and exchange rate risk is weak as a whole, the effects of exchange rate movements become more pronounced under certain conditions. If exchange rate movements have the potential to cause severe liquidity problems or affect the fundamental value of a firm, the most likely effects will be observed during large currency movements. Consequently, in this section, we employ an alternative approach to examine whether stock price reactions to large exchange rate shocks vary systematically with firm’s short-term cash flow sensitivity, the event study. An advantage 16

of the event study methodology is that we avoid imposing any specific structure on the relation between currency risk and firm value. This approach also allows us to focus on the cross-sectional variation of foreign exchange exposure at one point in time without worrying about the specification problem across time. We take a series of dates with large currency movements and examine the stock price reaction for a large sample of firms and how that reaction is related to our proxy for the firm’s expected cost of financial distress. We note the possibility that stock prices react to the currency movements through the influence of very broad macroeconomic factors that widely impact all firms, in additional to through their direct exposure to foreign exchange fluctuations. Thus, we are interested in examining the cross-sectional variation in stock price reactions around the large movements rather than simply the level of the reactions. We expect that if firm value is affected by exchange rate shocks, then the reaction of stock prices should be more prominent for those firms that have large expected financial distress costs. To identify large currency movements, we calculate the standard deviation of daily percentage changes in exchange rates for the 1978-2001 sample period.7 We define large currency movements as those daily exchange rate changes that are more than three standard deviations, in absolute value, from the sample mean daily movement.8 A total of 74 days fall into this category. For each of the 74 days, we calculate each firm’s cumulative abnormal returns (CAR) during a three-day window of (t-1, t+1) and regress the absolute value of abnormal returns on the firm’s expected cost of financial distress. Since we use a three-day window to calculate the firms’ cumulative abnormal returns, we drop four large currency movement days whose event windows overlap with those of 7

The sample mean daily percentage exchange rate change is 0.0025%. The standard deviation is 0.4284%.

17

other event days. This leaves us with 70 days for the analysis. Specifically, we estimate the following market model, using a 200-day window that ends 40 days before each event,

Ri ,t = α i + β i Rmt + ε i ,t

t=(-240, -41)

(4)

where Rmt is the return on the CRSP equally-weighted market index. Then we calculate the abnormal returns for each day in the 3 days around the large currency movements through the following equation, ^

^

ARi ,t = Ri ,t − α i − β i Rmt

t=(-1, 1)

(5)

After calculating the sample firms’ cumulative abnormal returns (CAR) over the three-day window around the large exchange rate movements, for each of the 70 events we examine whether the abnormal returns vary systematically across the sample by regressing the absolute value of CAR on proxies for the firm’s expected costs of financial distress. We focus on the absolute value of CAR because exchange rate movements should have opposite impacts on net importers and net exporters. To control for potential heteroskedasticity across firms, the dependent and independent variables are scaled by the standard deviation of residuals from equation (4). The time series mean estimates and Fama-MacBeth t-statistics from these weighted least squares analyses are presented in Table 5. The independent variables in column one of Table 5 are O-Score, market-tobook ratio, asset growth rate, and the high R&D dummy. Foreign sales of the firm’s affiliated four-digit SIC industry group and firm size are also included. The results indicate that in general firms’ stock price reactions to large exchange rate changes are related to their financial condition. Abnormal returns around the large exchange rate movements are significantly higher for firms with high O-Scores, supporting the hypothesis that the more financially distressed is a firm, the more its value is affected by

8

We choose to examine currency movements that are out of three standard deviations because there are very few days where daily changes are beyond four (14 days) standard deviations and too many days (328) that are beyond two standard deviations.

18

exchange rate movements. The results also suggest that firms with higher potential costs of financial distress if the distress occurs (those with more growth opportunities and greater R&D expenses) are more affected by the large exchange rate movements, as predicted by our hypothesis. The coefficient on the size variable indicates that the larger the firm, the less effect a currency movement has. As a robustness check, we also report in column 2 the CAR during a 11-day window of (t-5, t+5) to account for the possibility that information about a potential exchange rate movement is well anticipated by the public and is factored in by the market before the actual change. Similar to the results from column 1, we again find abnormal returns to be positively related to O-Score, growth opportunities, and product uniqueness when we examine the abnormal returns in a longer event window. In summary, by focusing on the impact of exchange rate risk on firm value at particular points of time during large currency movements, we show that stocks of firms with larger expected costs of financial distress experience greater abnormal returns associated with such exchange rate shocks.

4. Industry Level Analysis In the analyses above, we have controlled for the effect of foreign sales on firms’ exchange rate exposures. However, the level of foreign sales itself may not be adequate to determine the nature of a firm’s currency exposure. For example, a large firm may appear to have very high level of foreign sales relative to its total sales. However, it is possible that it also imports substantially from foreign countries so that its net exposure may be minimal or even positive. Although we do not have detailed firm level data on foreign operations, monthly U.S. foreign trade data are readily available at the industry level. Therefore, industry-level analysis will give us more power to detect any exchange rate exposure. An additional advantage of the industry approach is that the tradeweighted index serves as a better proxy of exchange rates at the more aggregated level. That is, since a firm can face multiple currency exposures through direct and indirect 19

channels, this study, as well as most previous empirical studies, uses a trade-weighted index to test the effect of exchange rate changes on firm value. By using this index, we are implicitly assuming that a firm is exposed to all currencies in the basket in the same magnitude as the composition of the basket. If firms’ exposures vary from this basket of exposures, the empirical tests may fail to capture the true underlying exposures. We extract monthly values of worldwide imports and exports of U.S. manufacturing industries from the Trade DataWeb of U.S. International Trade Commission. Because this data source provides detailed data on industry import and export shares only after 1989, we conduct the industry level analysis for the 13 year period, 1989-2001. The imports data are defined as U.S. general imports based upon general custom values. The total exports data are based upon FAS values.9 To construct the monthly foreign trade shares, we also obtain total industry productions as proxied by manufacturing industry shipments from the Bureau of Economic Analysis. This data is only available on an annual basis at the industry level; thus, we calculate the monthly import and export shares by dividing the trade data by one-twelfth of the annual industry shipments. For each year from 1989 to 2001 we form value-weighted industry portfolios among all manufacturing firms according to the Fama-French 48 industry classification. Firms’ previous year-end market capitalization is used as the weight. We estimate an industry panel regression using monthly industry portfolio returns. This specification has several advantages. First, the panel regression can capture the fact that both foreign exchange exposures and distress costs are varying simultaneously over time and crosssectionally. Second, given the shorter time-series of this sample of industry portfolios, a panel regression has more statistical power to detect any exchange rate exposure than does a rolling regression.

9

The total FAS (free alongside ship) value is the value of exports at the U.S. port, based on the transaction price, including inland freight, insurance, and other charges incurred. The value excludes the cost of loading the merchandise and also excludes any further costs.

20

During 1989-2001, we regress monthly industry portfolio returns on CRSP equalweighted index returns, the percentage change of exchange rate, real interest rate, and the interaction terms between exchange rate and average (median) industry O-Score, marketto-book ratio, asset growth rate, R&D expenses as a percentage of total sales. Since size and book-to-market ratio have been found to have explanatory power for cross-sectional variation of stock returns, we include the logarithm of the market value and the marketto-book ratio separately in case the effects of their interaction terms with exchange rate changes are merely driven by the size and book-to-market effects. According to our hypothesis, we expect exposure to be higher for industries with higher expected cost of financial distress. In addition, it should be increasing with their import shares and decreasing with their export shares. Therefore, we interaction exchange rate with the monthly imports and exports shares of each industry portfolio. We also include the interaction between exchange rate change and average firm size because larger firms’ may have greater ability to engage in operational or financial hedging, although they may also be more likely to conduct multinational businesses. Therefore, the effect of size is ambiguous. To control for differences in exposure due to industry structures, we include industry effect in the unbalanced panel regression. t-statistics of the coefficients are calculated with panel-corrected standard errors (PCSE) that adjust for the autocorrelation within each industry as well as heteroskedasticity across industries (Beck and Katz, 1995). Table 6 presents results of the industry analysis. We first run the regression without including proxies for expected costs of financial distress. Model 1 shows that on average U.S. manufacturing firm have significantly positive exposures, indicating that they are net importers. This is consistent with the imports and exports data of the U.S. manufacturing industries that show a trade deficit for the majority of the industries during this period. Although larger export shares lead to more negative exposures and larger import shares lead more positive exposures to currency risk, the coefficient on the

21

interaction term between import shares and exchange rate changes fails to be statistically significant. Model 2 includes the interaction terms of exchange rate changes with median industry O-Score, market-to-book ratio, asset growth rate and R&D expenses. After controlling for these firm characteristics, on average the foreign exchange rate exposure of U.S. manufacturing industries is still significantly positive. More importantly, the significantly positive interaction term between median O-Scores and exchange rate changes provides a strong support to our hypothesis that industries that are more likely to be financially distressed exhibit greater currency exposures. We also find exchange rate exposure to be increasing with the industry’s median R&D expenses, consistent with the effect of cost of financial distress relating to product uniqueness. However, the coefficients on market-to-book ratio and asset growth rates are found to be negative. One possible reason may be that industries with higher growth opportunities are more likely to be high-tech firm that do not require large amount of import from foreign countries. As a result, they are net exporters that have negative exposures. As a whole, under a specification that allows both time-series and cross-sectional variation of exposure, the industry analysis provides support for our hypothesis that exchange rate risk is more likely to affect the value of firms with larger expected cost of financial distress.

5. Can Hedging Explain the Exchange Rate Exposure? Although we have established the connection between firm characteristics that proxy for expected costs of financial distress and currency risk at both the firm and the industry levels, these same characteristics also suggest strong incentives for firms to hedge foreign exchange risk. For example, Geczy, Minton and Schrand (1997) provide evidence that firms with greater growth opportunities and tighter financial constraints are more likely to use currency derivatives. Further, using a sample of Japanese multinational firms, He and Ng (1998) show that the extent to which a firm is exposed to exchange rate 22

risk can be explained by variables such as the level of financial leverage and the firm’s short-term liquidity position. By assuming that these variables explain hedging incentives, they find that less hedged firms are more exposed to exchange rate risk. If hedging explains the low correlation between stock return and currency risk, we should expect firms to be more exposed to difficult-to-hedge currencies. Francis, Hasan and Hunter (2005) find significant mean exposure to both the foreign exchange index that includes the currencies of developed countries and the one that includes developing countries. Therefore, it is unlikely that hedging can insulate firms from exchange rate risk. In this paper, we implicitly assume that hedging cannot fully insulate firms from currency risk. This assumption is consistent with evidence provided in current literature that most U.S. firms hedge selectively rather than completely (e.g., Brown, 2001; Bodnar, Hayt and Marston, 1998). To the extent that firms that have larger costs of financial distress are also more likely to hedge their currency exposure, the empirical tests will be biased against our hypotheses. Furthermore, if hedging incentive leads to a negative relationship between exposure and the expected cost of financial distress, as argued by He and Ng (1997), we should see the positive effect of financial distress being stronger in earlier years since hedging tools have become more popular and widely available over time. However, this is not what we find in Tables 3 and 4 when we conduct sub-period analyses. So far, the exchange rate index we use is the trade-weighted average index of a basket of major currencies since exposures to these currencies are most relevant to firm value economically. If higher expected costs of financial distress lead firms to hedge more, we should expect to see a stronger relationship between costs of financial distress and firms’ exposures to more difficult to hedge currencies. To further examine the effect of potential hedging by firms, in Table 7 we explore their exposures to more difficult to hedge currencies. Following Francis, Hasan and Hunter (2005), we use a trade-weighted index of the currencies of “other important trading partners” (OITP) of 19 developing 23

economies as defined by the Board of Governors of the Federal Reserve Bank to proxy for more difficult to hedge currencies. Using the same two-step procedure used in Tables 3 and 4, we regress firms’ exposures to OITP on their expected costs of financial distress. Panel A of Table 7 presents the estimates from equation (2) when the absolute value of estimated exposures is regressed on firm characteristics. In general, there is still a positive relationship between absolute exposure to OITP and the cost of financial distress. However, although the point estimate of the coefficient on O-Score is slightly higher than that in Table 3, coefficients on M/B, asset growth and the high R&D dummy are slightly smaller than those in Table 3. Similarly, when we allow the coefficients on these proxies of expected costs of financial distress to be different for net exporters and net importers, the finding is again fairly close to that in Table 4 when we examine the determinants of firms’ exposures to currencies of major trade partners. Taking these evidences together, it seems that hedging is not likely to lead to smaller currency exposures for firms with greater expected costs of financial distress.

6. Conclusion In this paper we examine the exchange rate exposure of U.S. manufacturing firms from a new perspective. By noting that exchange rate fluctuations can affect a firm’s value through their direct impact on cash flows, we propose that whether or how much a firm’s fundamental value is exposed to currency risk depends on how sensitive its value is to changes in its short-term cash flows. We further argue that firms that are more likely to be financially distressed, have greater growth opportunities or more unique products are more likely to have values with higher sensitivities to short-term cash flows volatility. Controlling for macroeconomic risks, we find evidence that firms’ foreign exchange exposures increase with their expected costs of financial distress. That is, firms with higher O-Scores, greater growth opportunities and more unique products exhibit larger foreign exchange exposures. Further examining exchange rate exposure through an event study methodology, we find that during large, unexpected movements of the dollar, 24

firms with higher expected costs of financial distress show larger exposures as measured by their larger abnormal returns in response to exchange rate shocks. We also provide additional support to our hypothesis by showing that at the aggregate level, U.S. manufacturing industries’ exchange rate exposures also exhibit significant correlation with the stock price’s sensitivity to short-term cash flow. Finally, by comparing firms’ exposures to currencies of developed and developing countries, we do not find evidence that hedging can explain the weak correlation between exchange rate risk and stock return and hedging leads to smaller currency exposures for firms with greater expected costs of financial distress.

25

References Adler, Michael and Bernard Dumas, 1984, Exposure of currency risk: Definition and measurement, Financial Management 13, 41-50 Allayannis, George, 1997, The time-variation of the exchange rate exposure: An industry analysis, Working paper, Darden Graduate School of Business, University of Virginia Allayannis, George and Jane Ihrig, 2001, Exposures and markups, Review of Financial Studies 14, 805-835 Allayannis, George and Eli Ofek, 2001, Exchange rate exposure, hedging, and the use of foreign currency derivatives, Journal of International Money and Finance 20, 273-296 Allayannis, George and James P. Weston, 2001, The Use of Foreign Currency Derivatives and Firm Market Value, Review of Financial Studies 14, 243-276 Altman, E.I., 1968, Financial ratios, discriminant analysis and the prediction of corporate bankruptcy, Journal of Finance 23, 589-609 Amihud, Yakov, 1994, Exchange rates and the valuation of equity shares, in Yakov Amihud and Richard M. Levich, ed.: Exchange Rates and Corporate Performance (Irwin, New York). Bailey, Warren and Y. Peter Chung, 1995, Exchange rate fluctuations, political risk and stock returns: some evidence from emerging markets, Journal of Financial and Quantitative Analysis 30, 541-561 Begley, J., J. Ming, and S. Watts, 1996, Bankruptcy classification errors in the 1980s: An empirical analysis of Altman and Ohlson’s models, Review of Accounting Studies 1(4), 267-284 Bodnar, Gordon M., Bernard Dumas, and Richard Martson, 2002, Pass-through and exposure, Journal of Finance 57, 199-231 Bodnar, Gordon M., Greg Hayt, and Richard Marston, 1998, 1998 Survey of risk management by U.S. non-financial firms, Financial Management 27, 70-91 Bodnar, Gordon M. and William M. Gentry, 1993, Exchange rate exposure and industry characteristics: Evidence from Canada, Japan, and the USA, Journal of International Money and Finance 12, 29-45 Bodnar, Gordon M. and M. H. Wong, 2003, Estimating exchange rate exposures: Issues in Model Structure, Financial Management 32, 35-67 Brown, Gregory, 2001, Managing foreign exchange risk with derivatives, Journal of Financial Economics 60, 401-448.

26

Brown, Gregory, P. Crabb and D. Haulshater, 2002, Are firms successful at selective hedging? Working paper, University of North Carolina at Chapel Hill Choi, J. J. and A. M. Prasad, 1995, Exchange risk sensitivity and its determinants: A firm and industry analysis of US Multinationals, Financial Management 24, 77-88 Dewenter, K., R. Higgins, and T. Simin, 2003, Can Event Study Methods Solve the Currency Exposure Puzzle? Working paper, University of Washington Dichev, I.D., 1998, Is the risk of bankruptcy a systematic risk? Journal of Finance 53, 1131-1147 Doidge, Craig, J. M. Griffin and R. Williamson, 2002, Does exchange rate exposure matter? Working paper, Ohio State University Francis, Hasan and Hunter, 2005, The role of currency risk in industry cost of capital, Working paper, University of South Florida Froot, K. and K. Rogoff, Perspectives on PPP and long-run real exchange rates, Handbook of International Economics, Elsevier Science, Amsterdam, 1647-1688. Froot, K., D. Scharfsten, and J. Stein, 1993, Risk management: coordinating investment and financing policies, Journal of Finance 48, 1629-1658 Geczy, Christopher, B. A. Minton and C. Schrand, 1997, Why firms use currency derivatives, Journal of Finance 52, 1323-1354 Glaum, Martin, Marko Brunner, and Holger Himmel, 2001, The DAX and the dollar: The economic exchange-rate exposure of German corporations, Journal of International Business Studies, 715-724 Griffin, John M. and Michael L. Lemmon, 2002, Book-to-market equity, distress risk, and stock returns, Journal of Finance 57, 2317-2336. Griffin, John M. and Rene M. Stulz, 2001, International competition and exchange rate shocks: A cross-county industry analysis of stock returns, Review of Financial Studies 14, 215-241 Guay, Wayne and S.P. Kothari, 2003, How much do firms hedge with derivatives? Journal of Financial Economics 70, 423-461 He, Jia and Lilian K. Ng, 1998, The foreign exchange exposure of Japanese multinational corporations, Journal of Finance 53, 733-753 Hentschel, Ludgar and S.P. Kothari, 2001, Are corporations reducing or taking risks with derivatives? Journal of Financial and Quantitative Analysis 36, 93-118 Jorion, Philippe, 1990, The exchange-rate exposure of U.S. multinationals. Journal of Business 63, 331-345 27

Landers, Peter, 1999, Sony 's net falls 25%, Underlining strong yen 's impact on exports, The Wall Street Journal (10/28/1999), A21 Nance, Deana R., Clifford W. Smith Jr., and Charles W. Smithson, 1993, On the determinants of corporate hedging, Journal of Finance 48, 267-284 Ohlson, James, 1980, Financial ratios and the probabilistic prediction of bankruptcy, Journal of Accounting Research 18, 109-131. Shumway, T., 1997, The delisting bias in CRSP data, Journal of Finance 52, 327-340 Smith, Clifford W., and Rene Stulz, 1985, The determinants of firms’ hedging policies, Journal of Financial and Quantitative Analysis 20, 391-405 Stulz, R.M., 1984, Optimal Hedging Policies, Journal of Financial and Quantitative Analysis 19, 127-140 Titman, Sheridan, and Roberto Wessels, 1988, The determination of capital structure choice, Journal of Finance 43, 1-19 Williamson, Rohan, 2001, Exchange rate exposure and competition: evidence from the world automotive industry, Journal of Financial Economics 59, 441-475

28

Table 1 The Relation between absolute exchange rate exposures and expected costs of financial distress by industry using rolling regressions We group firms by their Fama-French 48 industry sectors. At the beginning of each year, we classify the firms within each industry sector into three groups with equal number of firms according to their O-Scores. Within each O-Score group, we estimate each firm’s foreign exchange exposure using its monthly stock returns in the subsequent 36 months according to equation (1). This process is repeated for every year from 1978 to 1998. The table presents the time-series average of the absolute value of foreign exchange exposure for the low, medium and high O-score groups for each of the 27 manufacturing industries. tstatistics calculated with Newey-West robust standard errors for the difference in exchange exposure coefficients between the low and high groups are reported in parentheses. a, b and c denote significance at the 1%, 5% and 10% levels, respectively.

Industry

Low

Medium

High

Low-High (t-statistics)

Food Products

0.6920

0.6519

0.9730

(2.36)b

Candy & Soda

0.7325

0.7304

0.7457

(0.00)

Beer & Liquor

0.6421

0.7624

0.9763

(2.04)b

Tobacco Products

0.5305

0.6288

0.7051

(0.83)

Recreation

0.9182

1.2562

1.3733

(2.33)b

Printing and Publishing

0.6133

0.6971

0.9625

(2.90)a

Consumer Goods

0.7490

0.8941

1.1535

(3.23)a

Apparel

0.7537

0.9354

1.3204

(5.23)a

Medical Equipment

1.0023

1.0509

1.2503

(1.77)c

Pharmaceutical Products

1.0130

1.0376

1.4056

(1.58)

Chemicals

0.6340

0.7348

1.0616

(5.42)a

Rubber and Plastic Products

0.7206

0.8693

1.0615

(4.22)a

Textiles

0.7390

0.8606

0.9822

(1.94)

Construction Materials

0.7709

0.8128

1.1160

(3.42)a

Steel Works Etc

0.7432

0.8271

0.9536

(2.46)b

Fabricated Products

0.7445

0.9271

1.3530

(2.85)a

29

Table 1.--Continued

Industry

Low

Medium

High

Low-High (t-statistics)

Machinery

0.8637

0.9451

1.1883

(2.34)b

Electrical Equipment

0.7429

0.7817

1.1331

(3.39)a

Automobiles and Trucks

0.7566

0.7528

0.9869

(2.02)b

Aircraft

0.8169

0.7967

1.0999

(2.06)b

Shipbuilding, Railroad Equipment

0.8885

0.8131

1.7982

(3.10)a

Defense

0.8122

0.7788

1.1747

(1.64)

Computers

1.1577

1.2295

1.4724

(1.70)c

Electronic Equipment

1.0670

1.1262

1.4633

(2.15)b

Measuring and Control Equipment

0.9826

0.9549

1.2364

(1.31)

Business Supplies

0.6654

0.7104

1.0784

(2.95)a

Shipping Containers

0.7000

0.8005

1.5103

(3.71)a

30

Table 2 Correlation Matrix Between Firm Characteristics This table reports the correlation matrix between average firm characteristics over the sample period of 1978 to 2001. “O-Score” and “M/B” stand for the Ohlson’s (1980) O-Score and market-to-book ratio, respectively. “Asset growth” is the percentage change of total assets. “High R&D” is a binary variable that takes the value of 1 if a firm’s R&D/Sales is above the sample median level. “Foreign Sales” is measured as the average percentage of foreign sales over total sales for firms’ affiliated industry groups. “Size” is measured as the logarithm of the firm market value. a, b and c denote significance at the 1%, 5% and 10% levels, respectively.

O-Score M/B Asset Growth High R&D Foreign Sales Size

O-Score 1 -0.1029 -0.1513 -0.2314 -0.0749 -0.1608

M/B

Asset Growth

High R*D

Foreign Sales

Size

1 0.2887 0.4419 0.1033 0.1464

1 0.1328 0.0488 0.0602

1 0.1801 0.0619

1 0.0719

1

31

Table 3 Determinants of Foreign Exchange Exposures Over the sample period, firms’ monthly stock returns are regressed on the percentage changes of the foreign exchange rate index, returns on the equally-weighed CRSP index, and the return on the1-month Tbill rate minus the monthly inflation rate. The absolute value of the estimated exposure is then regressed on a firm’s average O-Score, market-to-book ratio, asset growth rate, a dummy variable indicating high R&D expense firms, foreign sales as the average percentage of foreign sales over total sales for its affiliated industry groups, the logarithm of firm market value and industry dummy variables. t-statistics are reported in parentheses. R-square is reported at the bottom. a, b and c denote significance at the 1%, 5% and 10% levels, respectively. 1978-2001 1.5466 (6.42)a

1978-1989 1.0456 (5.69)a

1990-2001 2.1514 (7.53)a

O-Score

0.0578 (4.22)a

0.0647 (6.62)a

0.0544 (3.10)a

M/B

0.1670 (10.32)a

0.0950 (6.11)a

0.1086 (5.85)a

Asset Growth

0.4070 (7.74)a

0.2073 (2.96)a

0.4179 (7.06)a

High R&D

0.2308 (4.44)a

0.0414 (1.06)

0.3394 (5.44)a

Foreign Sales

0.5561 (3.52)a

-0.0274 (-0.24)

0.5279 (2.61)a

Size

-0.1273 (-9.49)a

-0.0798 (-7.49)a

-0.1653 (-10.25)a

Yes

Yes

Yes

0.1263

0.1354

0.1142

Intercept

Industry Dummy R2

32

Table 4 Determinants of Positive and Negative Foreign Exchange Exposures Over the sample period, firms’ monthly stock returns are regressed on the percentage change in the foreign exchange rate index, returns on the equally-weighed CRSP index, and the return on the1-month T-bill rate minus the monthly inflation rate. The estimated exposures is then regressed on interaction terms between dummy variables indicating positive and negative exposures (denoted as D+ and D-, respectively), and average O-Score, market-to-book ratio, asset growth rate and a dummy variable indicating high R&D expense firms. The control variables include foreign sales as the average percentage of foreign sales over total sales for firms’ affiliated industry groups, the logarithm of the firm market value and industry dummy variables. t-statistics are reported in parentheses. R-square is reported at the bottom. a, b and c denote significance at the 1%, 5% and 10% levels, respectively. 1978-2001 0.6359 (2.51)b

1978-1989 0.2228 (1.09)

1990-2001 0.4066 (1.31)

D+ * O-Score

0.0822 (4.41)a

0.0798 (5.64)a

0.0885 (3.51)a

D- * O-Score

-0.0344 -(1.79)c

-0.0265 (-1.88)c

-0.0046 (-0.18)

D+ * M/B

0.2828 (13.85)a

0.2836 (12.90)a

0.2845 (11.73)a

D- * M/B

-0.2799 (-13.67)a

-0.2271 (-10.68)a

-0.2565 (-10.98)a

D+ * Asset Growth

0.3351 (5.01)a

0.1273 (1.02)

0.3386 (4.34)a

D- * Asset Growth

-0.4834 (-4.97)a

-0.4170 (-4.25)a

-0.4978 (-4.36)a

D+ * High R&D

0.4248 (6.02)a

0.1222 (2.21)b

0.6455 (7.10)a

D- * High R&D

-0.3206 (-4.58)a

-0.1222 (-2.21)b

-0.4528 (-5.20)a

Foreign Sales

0.1880 (1.13)

0.0149 (0.12)

0.2674 (1.21)

Size

-0.0075 (-0.53)

-0.0218 (-1.84)c

-0.0012 (-0.07)

Yes

Yes

Yes

0.4184

0.4387

0.4184

Intercept

Industry Dummy R2

33

Table 5 Reaction of Stock Prices to Large Exchange Rate Shocks For each of 70 days over the 1978-2001 sample period with daily percentage exchange rate changes at least three standard deviations from the sample mean, the absolute value of the 5-day cumulative abnormal return is regressed against previous year-end firm characteristics using weighted least squares regressions. The average coefficients from these regressions are reported along with Fama-MacBeth t-statistics in parenthesis. The average value of R-square is also reported at the bottom. a, b and c denote significance at the 1%, 5% and 10% levels, respectively.

Intercept

O-Score

(t-1, t+1) 3.5748

(t-5, t+5) 7.1183

(21.17)a

(25.83)a

0.1447

0.3066

(11.80)

a

(12.06)a

0.0974

0.1948

a

(5.91)a

0.2195

1.0228

a

(7.38)a

High R&D

0.4147 (7.60)a

0.7972 (8.01)a

Foreign Sales

-0.1568

-0.1006

(-1.36)

(-0.49)

-0.1469 (-7.12)a

-0.2696 (-6.80)a

0.060

0.077

M/B

(6.03) Asset Growth

(3.25)

Log(size)

R2

34

Table 6 Industry Exchange Rate Exposures During 1989-2001, industry portfolios are formed based on Fama-French 48 industry classification. The monthly returns on industry portfolios are regressed on CRSP equal-weighted index returns, the percentage change of exchange rate, real interest rate, size as measured by the logarithm of median market value, median market-to-book ratio, and interaction terms between exchange rate and monthly industry export and import shares, median industry O-score, market-to-book ratio, asset growth rate, R&D expenses as a percentage of total sales, and average size. t-statistics (in parentheses) of the coefficients are calculated with panel-corrected standard errors that adjust for the autocorrelation within each industry as well as heteroskedasticity across industries. a, b and c denote significance at the 1%, 5% and 10% levels, respectively. Model 1 -0.0013 (-0.32)

Model 2 0.0344 (2.87)a

EWRET

0.5763 (33.00)a

0.5809 (33.48)a

Interest Rate

4.1388 (9.91)a

4.1850 (10.13)a

FX

0.8320 (2.26)b

0.8198 (2.23)b

FX*Export

-1.2349 (-2.18)b

-0.8723 (-1.48)

FX*Import

0.2127 (0.59)

0.5083 (1.39)

Intercept

Size

-0.0079 (-3.08)a

M/B

0.0024 (0.79)

FX * O-Score

0.3346 (2.07)b

FX * M/B

-0.3749 (-2.53)b

FX * Asset Growth

-1.5652 (-1.09)

FX * R&D

2.1669 (2.06)b

FX*Size

-0.1735 (-2.50)b

35

0.0070 (0.09)

Table 7 Determinants of Foreign Exchange Exposures to OITP Over the sample period, firms’ monthly stock returns are regressed on the percentage changes of the foreign exchange rate index of the currencies of “other important trading partners” (OITP), returns on the equally-weighed CRSP index, and the return on the1-month T-bill rate minus the monthly inflation rate. In Panel A the absolute value of the estimated exposure is regressed on a firm’s average O-Score, market-tobook ratio, asset growth rate, a dummy variable indicating high R&D expense firms, foreign sales as the average percentage of foreign sales over total sales for its affiliated industry groups, the logarithm of firm market value and industry dummy variables. In Panel B we regress the estimated exposures on the same set of variables but allowing firms with positive and negative exposures to have different coefficients on OScore, market-to-book ratio, asset growth rate, and the dummy variable indicating high R&D expense firms. t-statistics are reported in parentheses. R-square is reported at the bottom. a, b and c denote significance at the 1%, 5% and 10% levels, respectively.

Panel A: Estimates according to equation (2) 1978-2001 1.9088 (5.29)a

Intercept

O-Score

0.1015 (4.96)a

M/B

0.1217 (5.02)a

Asset Growth

0.1995 (2.53)b

High R&D

0.2125 (2.73)a

Foreign Sales

0.7115 (3.00)a

Size

-0.1797 (-8.94)a

Industry Dummy R2

Yes 0.0718

36

Panel B: Estimates according to equation (3) 1978-2001 0.6265 (1.61)

Intercept

D+ * O-Score

0.0637 (2.05)b

D- * O-Score

-0.0892 (-3.29)a

D+ * M/B

0.4162 (12.52)a

D- * M/B

-0.3173 (-10.69)a

D+ * Asset Growth

0.1556 (1.17)

D- * Asset Growth

-0.2015 (-1.84)c

D+ * High R&D

0.3410 (2.97)a

D- * High R&D

-0.5513 (-5.40)a

Foreign Sales

0.2680 (1.05)

Size

-0.0114 (-0.53)

Industry Dummy R2

Yes 0.2998

37