Pacific-Basin Finance Journal 11 (2003) 121 138 www.elsevier.

com/locate/econbase

The Asian exposure of U.S. firms: Operational and risk management strategies
Jongmoo Jay Choi a,*, Yong-Cheol Kim b,1
a

Department of Finance, Fox School of Business and Management, Temple University, Philadelphia, PA 19122, USA b University of Wisconsin-Milwaukee, P.O. Box 742, Milwaukee, WI 53201, USA Received 10 November 2001; accepted 27 August 2002

Abstract This paper examines the Asian currency exposure of U.S. firms with regard to their international operational and risk management strategies. We find that contemporaneous and lagged changes in real exchange rates have significant impacts on firm value for about 30% of the U.S. firms with Asian operations. The effects of a strong dollar are heterogeneous, with both significantly positive and significantly negative coefficients. The exchange exposure coefficients are then estimated as a function of international operational and risk management variables. A strong dollar has an adverse effect on firm value when the firm has a negative initial exposure position, and is related to exports and local sales activities of the firms. However, asset deployment in Asia raises the exposure in absolute terms regardless of initial exposure condition. Variables for hedging incentives explain exposure in both positive and negative exposure cases. Finally, a disaggregate study by country shows significant intra-regional differences, indicating the different ways in which the U.S. firms used their Asian subsidiaries operationally. D 2002 Elsevier Science B.V. All rights reserved.
JEL classification: G3; F23; F31 Keywords: Exchange exposure; International operational strategy; Corporate risk management; Asian currencies; American firms in Asia

The wide currency swings experienced during the Asian financial crisis heightened interest in the potential vulnerability of U.S. firms to exchange risk in general, and to the

* Corresponding author. Tel.: +1-215-204-5084; fax: +1-215-204-1697. E-mail addresses: jjchoi@temple.edu (J.J. Choi), ykim@uwm.edu (Y.-C. Kim). 1 Tel.: +1-414-229-4997; fax: +1-414-229-5999. 0927-538X/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0927-538X(02)00109-9

122

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

Asian exposure in particular.2 The Asian financial crisis occurred against the backdrop of decades of unprecedented high economic growth and accelerating volumes of international trade and investment in the region. The firms accustomed to seeing only the opportunity side of the ledger were suddenly awakened to the risk that their Asian operations might entail. In this paper, we assess and investigate the Asian currency exposure of U.S. multinational firms with regard to their international operational strategies. We estimate the exposure of U.S. firms to Asian currencies and relate the resulting exposure coefficients to the profile of international operational and risk management strategies of U.S. firms during 1992 1997, a period of economic growth as well as a prelude to the financial crisis in Asia. The present paper is also motivated by the literature on exchange exposure. It is commonly believed, among practitioners as well as academics alike, that exchange rates are an important source of macroeconomic uncertainty that influence the performance and the value of the firm in an international context. As such, the measurement and management of exchange risk exposure is a central issue in international finance. However, empirical evidence on currency exposure is mixed at best. Jorion (1990), Bodnar and Gentry (1993), Choi and Prasad (1995), and Dominguez and Tesar (2001) report varying significance of contemporaneous exchange rate changes on stock returns. Bartov and Bodnar (1994) and Amihud (1994) document the significant influence of lagged effects of exchange rate changes for U.S. export firms. He and Ng (1998), however, show that contemporaneous changes in exchange rates, not the lagged changes, are significant in explaining stock returns of Japanese firms. These seemingly conflicting results suggest further study is needed regarding currency exposure, especially in light of the concern about the Asian exposure of U.S. firms. Upon further reflection, the fact that exchange exposures are different for different firms is not surprising. The estimated exposure coefficients may simply reflect the different cash flow profiles of international operational and asset utilization (as well as different risk management strategies) of the firms. In principle, the cash flow impact of exchange rate changes depends on the nature of the business in which the firm is engaged. For instance, U.S. export firms generally do well with a weak dollar while import firms show improved performance with a strong dollar. On the other hand, for U.S. multinational firms with foreign production facilities, the effects of exchange rate changes depend on the market demand conditions and sourcing strategy. If the firm produces abroad for local sales, it may suffer from a strong U.S. dollar because of the conversion of a given amount of transactional local-currency sales revenue into a lower dollar amount. On the other hand, if the local subsidiary is used primarily as an outsourcing base to supply the parent, a strong dollar would lower the cost basis of the parent U.S. firm and thereby raise the value of the firm. Furthermore, if the products are exported to a third country (and are invoiced in the U.S. dollar), the effect of a strong dollar may also be positive for the firm. The net impacts of currency exposure depend on the combination of operational and conversion effects.3
For practitioner interest on the impact of the Asian currency crisis on U.S. firms, see Wall Street Journal, (1997a,b). 3 Here we are focusing on exchange rate impacts on foreign production. Exchange rate changes can also influence the value of the firms domestic operation because of the competitive changes caused by changes in real exchange rates.
2

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

123

The measured exchange exposure coefficients also reflect the outcomes of risk management. A firm that engages in aggressive hedging would show lower or insignificant measured exchange exposure coefficients compared to the original exposure position it faced. A firm that is less active or less efficient in hedging, however, would show an exposure that is little changed from its initial position. In addition, operational strategies such as international production or marketingregardless of whether they were undertaken on their own merit or out of risk management considerationmay alter the exposure profile of the firm. This interdependence of currency exposure and operational strategies of the firm has not received much attention in the literature.4 In this paper, we estimate and analyze the Asian currency exposure of U.S. manufacturing firms. The results show that contemporaneous and lagged exchange exposures of U.S. firms are significant for about 30% of the sample, but the significant coefficients are divided between positive and negative effects of a strong U.S. dollar. We then attempt to identify the firm-specific determinants of exchange exposure coefficients. The results confirm the dynamic interdependence of the measured exposure of U.S. firms on one hand and their operational and risk management strategies on the other. A disaggregate study of U.S. multinational firms by country, however, indicates noted national differences in exposure coefficients, suggesting important intra-regional differences in terms of the ways in which the U.S. firms utilized their Asian subsidiaries operationally. We believe that this is the first study of its kind that formally investigates the operational contents of the Asian exposure of U.S. firms. We also believe that the basic issues and results raised in the present paper are potentially of relevance and need to be empirically validated in other regions as well. Section 1 describes the data and characteristics of sample firms. Section 2 investigates the exposure measurement and discusses the exposure of U.S. firms to the trade-weighted real exchange index of the Asian currencies. Section 3 explains the exposure coefficient by various firm-specific factors related to operational profile and risk management strategies. Section 4 reports national differences in the exposure coefficients for each of the nine Asian currencies. The conclusion follows in Section 5.

1. The characteristics of sample firms The firms in the sample are U.S. firms that reported Asian sales, income and assets in the COMPUSTAT geographic segment data for the period from January 1992 to December 1997. The sample period ends in 1997 to exclude the period of structural turmoil faced by U.S. multinationals following the Asian financial crisis.5 To calculate the exposure coefficient, monthly individual stock return and value-weighted market return (including
4 As an example of some limited existing work, Kim (1997) examines the stock price reactions to international investment and divestiture announcements of the U.S. firms. Miller and Reuer (1998) similarly report that foreign direct investments reduce the firms exchange exposure. 5 The Asian financial crisis began in 1997 when the involved countries received assistance from the IMF, which ranged from July 1997 for Thailand to December 1997 for South Korea. However, the bulk of the negative fallout for the U.S. firms occurred in 1998 and beyond. Also note that the cross-sectional estimations are done for each year of the sample period in Section 3.

124

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

Table 1 Descriptive statistics for firms with Asian operations Variable Total Assets (million US$) Export Sales/Net Sales Return on Asset (Asia) Return on Asset (Foreign) Return on Asset (All) Asian Sales/Net Sales Foreign Sales/Net Sales Asian Assets/Total Assets Foreign Assets/Total Assets Tobins Q Long-term Debt/Total Assets Tax Amount/Total Assets Cash Flow/Total Assets R&D Expenses/Net Sales Dividend Payout Ratio Earnings per Share N 1052 413 1052 1052 1052 1052 1052 1052 1052 1052 1052 1052 1052 802 814 814 Mean 5477.58 15.86 12.72 8.70 9.28 14.89 39.31 13.66 34.65 2.08 13.49 3.03 9.66 7.98 0.97 0.00 Lower quartile 122.94 5.55 2.32 3.39 4.90 5.13 24.88 4.89 20.60 1.26 1.16 0.63 6.34 2.30 0.00 0.01 Median 421.61 10.26 10.19 10.29 10.39 10.60 37.43 9.49 32.60 1.69 9.79 2.69 11.24 5.77 0.00 0.04 Upper quartile 1766.17 19.40 20.61 18.65 15.63 17.84 52.10 16.11 43.60 2.54 20.31 5.10 15.29 10.81 1.70 0.06

Firms are included in the sample if they report sales, assets, and income from Asian operations in the geographic segment file and other financial information from the COMPUTAT database. The sample covers the period of January 1992 to December 1997. Total assets are in millions of U.S. dollars; all other variables are defined as a ratio in percentage terms, except for the Tobins Q and the earnings per share ratios. Variable definition: Return on Asset (Asia) is the operating income from Asia over assets in Asia. Return on Asset (Foreign) is the operating income from all foreign operations over foreign assets. Return on Asset (All) is the operating income from all assets over total assets. Asian Sales/Net Sales is the ratio of sales from Asia to net sales. Foreign Sales/Net Sales is the ratio of sales from all foreign operations to net sales. Tobins Q is defined as (Total Assets Book Value of Equity + Market value of Equity)/Total Assets. Cash Flow/Total Assets is the cash flow from all operations over total assets.

dividends) data were collected from the CRSP data files. Regional corporate datasales, export revenues, operating income and assets from operations in Asia and elsewhere were obtained from the COMPUSTAT geographic segment data. Other financial information was taken from the COMPUSTAT industry files. The sample includes manufacturing firms exclusive of regulated firms in utilities and telecommunications or those in financial service industries. Firms listed in the geographic segment with no information on foreign operations were dropped from the sample. A total of 1052 firm-years is identified for U.S. firms that have Asian operations during the period. The distribution of firm-years for each year is 121, 150, 171, 184, 188, and 238, respectively, from 1992 to 1997. Of these, only 413 report exports. However, 802 report research and development expenses, while all 1052 firm-years report sales, income and asset deployment in Asia.6
6 The use of firm-years may contain the possibility of multiple accounting if the same firm appears multiple times during the sample period. However, since firm-specific variables vary during the period, it is not likely that such method would introduce a systematic bias regarding the number of significant exposure cases as a percentage of total data points, and is common in empirical corporate finance literature. Inspection of current data set, in fact, reveals significant variability in firm-specific variables during the period, which indicates dynamic changes in U.S. corporate strategies in Asia as well as the Asian economies.

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

125

The return on asset figures in Table 1 indicate that the Asian operations were generally more profitable for U.S. firms than either domestic operations at home or operations in other international regions. However, it is interesting that the higher Asian profitability is observed in the upper quartiles and the mean, but not in the lower quartile. It is also interesting that the bigger U.S. firms generally achieved higher profitability than smaller ones in Asia. A similar disparity is seen in earnings per share (EPS). Ten percent of the sample firm-years report negative EPS, the magnitudes of which are sufficient to offset positive EPS firm-years and make the overall mean EPS zero. More than half of the firmyears did not show dividend payments, opting instead for retention and growth. Overall, a firsthand reading of these numbers suggests that the U.S. firms were increasingly adopting a local norm of emphasizing growth and market share in Asia in the midst of the looming financial crisis.

2. Measurement of exchange rate exposure Economically, the exchange risk exposure of a firm is quantified by measuring the degree of change in firm value in response to changes in exchange rates. Jorion (1990) and Bodnar and Gentry (1993) report that the exposure of U.S. firms to contemporaneous changes in exchange rates is limited. The studies focused on multinational firmsChoi and Prasad (1995) for U.S. multinational corporations and He and Ng (1998) for Japanese multinationalsare somewhat more positive. At the same time, Amihud (1994) and Bartov and Bodnar (1994) maintain that firms are exposed to lagged as well as contemporaneous changes in exchange rates. Following existing work, we use a two-factor model where stock return of an individual firm is regressed on exchange rate and market risk factors. However, we make several methodological innovations specific to Asian countries. First, unlike most existing work that uses nominal exchange rates, we use trade-weighted real exchange rates of Asian currencies. While the random walk and market efficiency hypotheses might make both nominal and real exchange rates acceptable for advanced industrialized countries, real exchange rates make more sense for Asian emerging countries where the same degree of market efficiency cannot be expected. The real exchange rate of each currency relative to the U.S. dollar is calculated by adjusting the nominal exchange rate by the relative inflation rates. We then construct a real multilateral Asian currency index based on a weighted average of nine Asian currenciesthe Chinese yuan, the Hong Kong dollar, the Indian rupee, the Japanese yen, the South Korean won, the Malaysian ringgit, the Singapore dollar, the Thai baht, and the Indonesian rupiah. The weight of each currency is determined by the value of trade of goods and services of the country concerned with the U.S., and is updated monthly based on the U.S. Commerce Department trade database. Nominal exchange rates and consumer inflation rates were obtained from the International Financial Statistics database. We use changes in the real multilateral Asian currency index as the exchange risk variable in the regression. An increase in the index represents an increase in the value of the U.S. dollar against the trade-weighted basket of Asian currencies in real terms. We also utilize disaggregate bilateral real exchange rates (the local currency per U.S. dollar) in individual country estimations.

126

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

The second methodological point concerns pooling of observations. Since we are interested in the differential exposure of firms based on their operational profiles, the pooling of cross-section and time series data (e.g., Bartov and Bodnar, 1994) would obscure different characteristics of firms. In addition, there is potential econometric bias from pooling. To capture the intertemporal as well as firm-specific effects of exposure, we use the Fama and McBeth (1974) method and estimate the return of individual firms as a function of market return and the change in exchange rate for 60 months on a rolling basis. For example, 60 monthly returns for each firm from the beginning of 1987 to the end of 1991 are used in the cross-sectional estimation for 1992.7 Third, we estimate the exchange exposure by using contemporaneous exchange rate changes only, and also by including both contemporaneous and lagged exchange rates jointly. In addition, since the latter variables are both significant, they are combined in the distributed lag model to produce a single measure of exchange exposure necessary in cross-sectional estimations. The distributed lag model (Green, 1993) uses the weighted average of contemporaneous and lagged coefficients as a measure of the change in exchange rates in the two-factor model:8 Ri;t a0 bm Rm;t bx Lwt Xt ei;t where Ri,t is stock return of firm i at time t, Rm is the market return, Xt is the change in the real multilateral Asian currency index value of the U.S. dollar (which varies for each

7 The rolling method enables us to estimate the firm-specific determinants of exposure each year. An alternative method is to regress a firms single measure of exchange exposure on the 1992 1997 average of firmspecific variables. This would mean a loss of information regarding the effects of time variability of these variables. Given the dynamic changes of the Asian economies and U.S. firms in the 1990s, this loss of information can be significant. 8 The estimation of exposure is based on the weighted average of contemporaneous and lagged exchange rate changes. First, we run a regression using both contemporaneous and lagged changes in exchange rates in Eq. (1):

Ri;t a0 bm Rmi;t bx;0 Xi;t bx;1 Xi;t1 et;i :

Second, the estimated coefficients of contemporaneous and lagged variables are added, which is then used as the denominator in the weight calculation: b bx;0 bx;1 ; wi bx;i =b; where i 0; 1: Third, we create the weighted average of the contemporaneous and lagged exchange rate changes: Lwt Xt w0 Xi;t w1 Xi;t1 : The currency exposure is the estimated distributed lag coefficient in Eq. (3): Ri;t a0 bm Rm bx Lwt Xt ei;t : 3 2

Note that with appropriate assumptions on the weights, Eq. (3) collapses to the case of contemporaneous exposure only, or to the case where both contemporaneous and lagged exchange rate changes are included with equal weight.

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

127

Asian country based on trade weights), L is a distributed lag operator, and wt is the weight of contemporaneous and lagged exchange rates.9 To relate to existing work, we first estimate the equation with contemporaneous exchange rate changes only, and also with both contemporaneous and lagged exchange rate changes, without the use of the distributed lag method. When we measure the exposure by the contemporaneous change only in a two-factor model, 13% of the sample shows significance at the 10% level for the entire sample period (Table 2, panel A). When both contemporaneous and lagged changes are used in panel B, 13.4% is exposed to contemporaneous changes in exchange rates and 9.0% to lagged exchange rate coefficients. These results show the relative importance of contemporaneous and lagged exchange rate changes. The results of distributed lag estimation with the weighted average of contemporaneous and lagged exchange rates are presented in Table 2, panel C. For the period of 1992 1997, the number of firms with significant exchange exposure is 29% at the 10% level of significance. The peak is seen in 1995 when 32% of all observations show significant exchange exposure. These results on the Asian operations of U.S. firms are more pronounced than those reported by Choi and Prasad (1995) for general international operations of U.S. multinationals.10 Of the firms that show significant exposure, about half of them have significant positive exposures. The positive exposure coefficient indicates that an appreciation of the U.S. dollar against the Asian currencies has a positive impact on stock returns of U.S. firms. It is also interesting to see the changes in the coefficients over time. While 50% of the firms with significant exposure coefficients show positive coefficients at the 10% level for the entire sample period, there is a significant change of positions over time. For example, 58% of the firms in the significant category show a positive exchange exposure in 1995, but the percentage of positive exposure was reduced to 42% in 1997. One interpretation of this intertemporal change is the worsening Asian economic condition in the 1990s. In the latter part of the sample period, the percentage of the U.S. firms that benefited from a weak Asian currency (or a strong U.S. dollar) had decreased dramatically because declining Asian currency values in real terms portended the deteriorating Asian business opportunity and performance for U.S. firms in the region.

3. The determinants of exchange exposure Theoretically, the effects of exchange rates on the firm are ambiguous depending on the potential conflict between output and input effects, between home and foreign market
9 We use unadjusted market return and exchange risk factors. The multicollinarity problem is not significant because the correlation coefficient of market return and the change in exchange rate is 0.124 with p-value of 0.16. The use of unadjusted factors is consistent with the random walk hypothesis assumed by Jorion (1990) and others. However, we accept the possibility that the value of the beta coefficients can be sensitive to whether unexpected or orthogonalized factors are used and how such adjustment is achieved. 10 Choi and Prasad (1995) use 409 U.S. multinational firms and trade-weighted nominal and real exchange rate changes of the U.S. dollar for 1978 1989 in a nonoverlapping Fama MacBeth estimation of the two-factor model. Fifteen percent of the firms show significant exchange exposure coefficients.

128
Table 2

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

The Asian exposure of sample firms

Panel A: The exposure coefficients are estimated by regressing monthly stock returns on value-weighted market returns and contemporaneous ACU. 1992 1993 1994 1995 27 14.67 16 59.26 11 40.74 1996 24 12.77 11 45.83 13 54.17 1997 34 14.29 10 29.41 24 70.59 All 137 13.02 76 55.47 61 44.53

Number of firm-years with significant sig N 11 PctN 9.09 Signs of significant exposures + N 9 PctN 81.82 N 2 PctN 18.18

exchange exposures 19 22 12.67 12.87 14 73.68 5 26.32 16 72.73 6 27.27

Panel B: The exposure coefficients are estimated by regressing monthly stock returns on value-weighted market returns and both contemporaneous and lagged ACU values. 1992 1993 1994 1995 1996 1997 All

(1) Significance of contemporaneous ACU Number of firm-years with significant exchange exposures sig N 12 19 23 PctN 9.92 12.67 13.45 Signs of significant exposures + N 9 14 17 PctN 75.00 73.68 73.91 N 3 5 6 PctN 25.00 26.32 26.09 (2) Significance of lagged ACU Number of firm-years with significant exchange exposures sig N 10 11 15 PctN 8.26 7.33 8.77 Signs of significant exposures + N 6 5 5 PctN 60.00 45.45 33.33 N 4 6 10 PctN 40.00 54.55 66.67

27 14.67 16 59.26

23 12.23 9 39.13

37 15.55 12 32.43

141 13.40 77 54.61

11
40.74

14
60.87

25
67.57

64
45.39

17 9.24 10 58.82 7 41.18

15 7.98 7 46.67 8 53.33

27 11.34 16 59.26 11 40.74

95 9.03 49 51.58 46 48.42

Panel C: The exposure coefficients are estimated by regressing monthly stock returns of firms on market returns and distributed lag of ACU, i.e. the weighted averages of contemporaneous and lagged exchange rate changes. 1992 Total number 121 Significant exchange exposures N 29 PctN 23.97 Signs of significant exposures + N 16 PctN 55.17 N 13 PctN 44.83 1993 150 39 26.00 22 56.41 17 43.59 1994 171 47 27.49 24 51.06 23 48.94 1995 184 59 32.07 34 57.63 25 42.37 1996 188 51 27.13 22 43.14 29 56.86 1997 238 76 31.93 32 42.11 44 57.89 All 1052 301 28.61 150 49.83 151 50.17

ACU is the change in the trade weighted real exchange rate of nine Asian countries, where the real exchange rate is the nominal rate adjusted for the inflation rate of each country relative to the U.S. inflation rate. ACU is defined as the change in Asian currency per 1 U.S. dollar. Included in the study are nine Asian currencies: the Chinese yuan, the Hong Kong dollar, the Indian rupee, the Japanese yen, the South Korean won, the Malaysian ringgit, the Singapore dollar, the Thai baht, and the Indonesian rupiah. The significance of exposure coefficients is evaluated at the 10% level. The number of firms (N) and the percentage of firms (PctN) with significant exposure are shown. The sign + ( ) indicates a positive (or negative) sign of the significant coefficient.

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

129

effects, and between conversion and economic effects (Choi, 1986). However, as a first approximation, firms with international business will have more reason to be concerned about exchange risk, depending on the nature of the firms international operations and the extent to which the firm engages in hedging. U.S. exporters will be hurt by a strong U.S. dollar under the reasonable elasticity assumptions on demand and supply for manufactured goods. Importers, on the other hand, will generally be helped by a strong dollar. For U.S. firms with foreign production, the effect of a strong dollar depends on whether it sells its product in the foreign local market or exports to the world market. The firm that sells in the foreign local market will be adversely affected because the local currency cash inflows will be converted into the lower U.S. dollar amount. However, if the sales are exported to the parent or to a third country with dollar denomination, then the weak local currency (or strong U.S. dollar) will increase the firms value. The operational effects of exchange rates also depend on the extent to which the firm depends on local inputs when the local input prices rise with the weak local currency. The measured value of exchange exposure coefficients also reflects the firms attempt to reduce its exchange risk exposure through financial as well as operational means. It is generally accepted that the usual financial hedging through derivatives can reduce the variability of the cash flow. Hedging can also positively impact the value of the firm by mitigating the firms underinvestment problems, by reducing agency costs of financial distress, or by alleviating information asymmetry.11 However, financial hedging is less effective in incomplete markets where the firm faces a long-term unpredictable operating cash flow. Operational strategies involving production, marketing, or organization may be more appropriate in such situations. We perform two separate estimations: the firms that have significant positive exposure coefficients and the firms that have significant negative coefficients. For both groups, we include international operational variables and risk management variables as potential determinants of exchange exposure. Mixing the two groups of firms would confound the interpretation. If the firm has a positive exposure initially, a positive sign of an operational variable in the exposure determinant model would mean that the variable would raise the firms exposure even more. On the other hand, for firms with a negative exposure, a negative sign would make the exposure more significant. 3.1. Effects of international operation Table 3 shows the results of regressions of the Asian currency exposures of U.S. firms on firm-specific variables, using the exchange exposure measures estimated by a distributed weighted lag of contemporaneous and lagged exchange rate changes (panel A) or by contemporaneous exchange rate changes only (panel B). Panel A indicates that the ratio of export to sales in general or the one specific to Asiathe ratio of Asian sales to total salesreduces the firms exchange exposure. This means that with a strong U.S.

11 DeMarzo and Duffie (1991) argue that hedging enhances corporate valuation because of the benefit of reducing information asymmetry. See Stulz (1984) and Tufano (1996) for additional references on corporate hedging policy.

130

Table 3 Determinants of the Asian currency exposures Panel A: The exposure coefficients are estimated by a distributed lag of contemporaneous and lagged exchange rate changes. J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138 POS 1 Intercept R&D Expenses/ Net Sales Dividend Payout Ratio Export Sales/ Net Sales Log (Total Assets) Long-term Debt/ Total Assets Q Return on Asset (Asia) Asian Sales/ Net Sales Asian Assets/ Total Assets Tax Amount/ Total Assets Number of observations Adjusted R squared F statistics 1.97 (8.85) POS 2 2.42 (14.21) POS 3 2.27 (11.41) POS 4 2.34 (12.03) 2.08 (3.36) POS 5 1.04 (7.45) NEG 6 1.08 ( 2.24) NEG 7 1.69 ( 8.18) NEG 8 1.56 ( 8.01) NEG 9 1.73 ( 7.24) 1.31 ( 1.28) NEG 10 0.89 ( 5.54)

9.00 ( 4.73) 0.57 ( 1.71) 0.16 (6.96) 1.49 ( 3.30) 0.14 (1.93) 0.66 ( 1.80) 0.13 (2.42) 0.02 (0.85) 0.18 ( 0.42) 0.96 (2.43) 8.34 ( 4.28) 531 0.05 7.05

13.28 (4.39)

0.23 (10.78)

0.19 ( 7.73)

0.21 ( 9.00)

0.14 (6.03)

0.10 (4.01)

0.16 (4.53) 0.62 (1.90) 0.06 ( 0.76) 0.07 (1.08) 1.45 ( 3.04) 1.11 ( 3.67) 2.60 (0.87) 521 0.04 5.21

0.12 (1.78)

0.14 (2.78)

0.12 (1.94)

0.10 ( 0.63)

0.07 ( 0.80)

0.00 (0.03)

0.50 ( 1.27)

0.33 ( 0.84)

0.13 ( 0.25)

2.58 ( 2.48)

0.73 ( 2.51)

0.90 ( 3.17)

4.94 ( 2.21) 213 0.15 8.64

6.25 ( 3.92) 531 0.20 33.36

5.18 ( 2.99) 417 0.19 20.20

2.91 ( 2.11) 410 0.22 29.07

0.08 (0.01) 200 0.13 7.10

1.02 (0.36) 521 0.09 14.21

0.11 (0.04) 397 0.13 12.47

0.15 ( 0.05) 392 0.12 14.59

Panel B: The exposure coefficients are estimated by contemporaneous exchange rate changes. POS 1 Intercept R&D Expenses/ Net Sales Dividend Payout Ratio Export Sales/ Net Sales Log (Total Assets) Long-term Debt/ Total Assets Q Return on Asset (Asia) Asian Sales/ Net Sales Asian Assets/ Total Assets Tax Amount/ Total Assets Number of observations Adjusted R squared F statistics 1.80 (7.94) POS 2 1.88 (14.80) POS 3 1.76 (12.13) POS 4 1.91 (12.97) 1.22 (2.84) POS 5 0.97 (8.87) NEG 6 1.71 ( 3.73) NEG 7 1.86 ( 8.58) NEG 8 1.72 ( 7.17) NEG 9 1.70 ( 7.23) 1.99 ( 2.84) NEG 10 0.69 (5.65) J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

8.92 ( 4.25) 0.58 ( 1.56) 0.14 ( 6.08) 0.37 ( 1.09) 0.19 (2.97) 0.30 ( 1.08) 0.09 (2.24) 0.03 ( 1.81) 0.03 ( 0.10) 0.45 (1.83) 6.05 ( 4.43) 693 0.04 6.43

2.95 (2.27)

0.15 ( 10.06)

0.11 ( 6.60)

0.15 ( 8.95)

0.18 (6.36)

0.15 (4.69)

0.18 (5.38) 0.27 (0.69) 0.04 (0.72) 0.42 (1.69) 0.33 ( 0.65) 0.86 (2.35) 1.53 (0.67) 359 0.06 5.17

0.07 (1.52)

0.08 (2.35)

0.07 (1.64)

0.00 (0.06)

0.03 (0.64)

0.02 (0.23)

0.08 (0.21)

0.06 ( 0.26)

0.10 ( 0.39)

1.08 ( 1.38)

0.01 ( 0.04)

0.28 ( 0.76)

3.41 ( 2.27) 289 0.09 6.61

4.22 ( 3.61) 693 0.15 30.95

3.78 ( 2.98) 565 0.15 20.50

1.71 ( 1.90) 542 0.16 26.02

0.32 ( 0.10) 124 0.13 4.82

1.79 (0.88) 359 0.15 17.05

1.91 (0.75) 249 0.15 9.56

1.15 (0.51) 260 0.21 17.82

Regression is run separately for the groups of positive and negative exposure coefficients. The dependent variable is the estimated currency exposure from regressing firm stock return to the market return and the trade-weighted real exchange rate of nine Asian countries. The t value is based on White (1980) adjusted standard errors, and is reported in parenthesis. POS or NEG denotes the group of firms with positive or negative exposure coefficients, respectively. 131

132

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

dollar, a higher export ratio or a higher Asian sales ratio adversely affects the U.S. firms, confirming the theory that a strong dollar makes exports less competitive. Its statistical significance, however, is found only in the group of firms with a negative exposure. Another international operational variable, the ratio of Asian assets over total assets, however, significantly influences the exposure coefficients for both groupsa positive sign for firms with a positive currency exposure coefficient and a negative sign for those with a negative exposure. Higher Asian assets raise the firms exposure in absolute terms, regardless of whether the firm had a positive or negative exchange exposure to begin with. These results on asset deployment may indicate the nature of Asian business for most U.S. firms: they may simultaneously sell to local markets and export to the U.S. and to other countries leading to negative or positive effects from a strong U.S. dollar, but further increases in asset deployment in the region would increase the exposure position in absolute value for both cases. To further investigate the effects of international operation, we compare the exposure coefficients by the quintile of variables that summarize the nature of international business activities in Table 4. Since exports here originate from parent firms in the U.S., the number of firms that have significant positive (negative) exposures is expected to decrease (increase) as the ratio of exports to net sales increases. If international operations generate sales and income mostly from the foreign local market, economic theory predicts that the number of firms that are positively (negatively) exposed decreases (increases) with a strong U.S. dollar or a weak local currency. For our sample, that is the case only for firms with a negative exposure. The group of firms that have a positive exchange exposure is more heterogeneous, including firms that engage in export sales of foreign production to the U.S. or other countries where the U.S. dollar is used as the currency of invoicing or pricing. The results of asset ratios provide another insight into the effects of international operation on exchange exposure. In the positive exposure category shown by quintile in Table 4, the number of firms increases as the percentage of Asian assets increases. In contrast to foreign sales and income ratios which indicate the global source of revenue, the asset ratios more directly show the pattern of asset deployment for production, as larger asset commitment is generally required for foreign production than for foreign sales. If the U.S. parent firm uses a foreign subsidiary as a production base for, say, parent input sourcing, then a higher ratio of Asian assets may be beneficial to the parent by reducing the U.S. dollar cost of foreign sourced input. 3.2. Effects of risk management Froot et al. (1993) suggest that firms with more growth opportunity have more incentive to hedge. Firms with growth opportunities require additional funds when attractive investment opportunities occur. The need for external financing thus increases. However, external financing may be constrained for firms with high growth opportunities either because their assets are not good collateral or because asymmetric information makes their valuation difficult. Hedging would be useful in such a case because reduced volatility with hedging would make external financing easier. This is supported in Table 3. The two variables used as a proxy for growth opportunitiesthe Tobins Q ratio and the ratio of research and development expenses to total salesshow significant positive

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

Table 4 Significance and the sign of exposure coefficient by quintile Export sig N All quintile 1 2 3 4 5 P_ PCHI 104 19 24 20 18 23 0.86 PctN 25.18 22.89 28.92 24.39 21.69 28.05 + N 57 13 16 10 14 4 0.11 N 47 6 8 10 4 19 0.01 Incasia sig N 301 48 55 54 64 80 0.04 PctN 28.61 22.75 26.19 25.59 30.48 38.10 + N 150 26 32 28 34 30 0.86 N 151 22 23 26 30 50 0.00 Salasia sig N 301 52 56 61 58 74 0.32 PctN 28.61 24.64 26.67 28.91 27.62 35.24 + N 150 34 29 35 23 29 0.55 N 151 18 27 26 35 45 0.01 Taasia sig N 301 47 54 57 67 76 0.07 PctN 28.61 22.27 25.71 27.01 31.90 36.19 + N 150 31 21 24 41 33 0.08 N 151 16 33 33 26 43 0.01

The table shows the number and the percentage of sample with significant Asian exposure at the 10% level of significance. The quintiles are based on four variables that indicate the extent of Asian operations: (a) exportthe ratio of export sales to total sales; (b) incasiaincome from Asian operations divided by total income; (c) salasia sales revenue from Asian operations divided by total sales; and (d) taasiaassets from Asian operations divided by total assets. Quintile 1 is the lowest quintile. N is the number of observations; PctN is the percentage of significant exposures in each category compared to total. Total number of observations in export is 413, and 1092 observations for the other three categories. + ( ) indicates positive (negative) significance. P_ PCHI is the p-value of the significance of the chi-square statistic testing that each quintile is uniformly distributed.

133

134

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

impact on the firms exposure.12 However, for both variables, the significant positive impacts are shown only for firms with a positive exchange exposure; similar, but statistically insignificant, effects are shown in absolute terms for firms with a negative exposure. In an international context, an important instrumental variable is wealth as a means of financing growth opportunities abroad. A strong U.S. dollar would increase the funds available for foreign projects, reducing the need for external financing for overseas operations. In that sense, a strong dollar can be viewed as a substitute for external financing. Given a positive growth opportunity in Asia, a strong dollar will make it easier for the firm to expand in Asia. While U.S. firms with greater growth opportunities have greater incentive to hedge, with strong dollar they would be less concerned with volatility and underinvestment cost (and hence would have a lower desire to hedge). Thus, the relationship between the coefficient of positive exchange exposure and growth opportunity is positive. Firm size, measured by the logarithm of total assets, shows the significant impact on the Asian currency exposure for both groups of firms. However, the signs are opposite. The effect of firm size is negative for positive-exposure firms, and positive for negativeexposure firms. This suggests an economy of scale in hedging activities: bigger firms are more efficient in managing currency risk than smaller ones regardless of whether exposures are positive or negative initially. The effects of profitability, measured by the return on assets in Asia on the Asian currency exposure of U.S. firms, however, are insignificant. Nance et al. (1993) argue that firms with greater liquidity could reduce the agency costs and financial distress associated with long-term debt. Also, a higher dividend payout ratio, as a proxy for short-term liquidity, would make hedging more cost effective. Similarly, Smith and Stulz (1985) predict that hedging can reduce the expected costs of financial distress. Therefore, we use the ratio of long-term debt to total assets and dividend yield as our empirical hedging variables. The results in Table 3 show that both variables are significant as a determinant of exposure. However, dividend yield and long-term debt ratios are both negatively significant in the group of firms with positive Asian currency exposures, and are positively significant for firms with negative Asian exposures. The results indicate that firms with a higher dividend payout ratio engage in hedging more, and thereby reduce the exposures in absolute terms for both groups. Firms with a higher debt ratio face higher expected costs of bankruptcy, and have more incentive to hedge. The results of the long-term debt ratio support this argument. Smith and Stulz (1985) indicate that the concavity of a progressive corporate tax schedule makes hedging more likely by increasing the expected claim by stockholders. The concavity of the tax schedule implies an asymmetric nature of the tax incentive for hedging. Geczy et al. (1997) document that, ceteris paribus, a higher taxable income makes a firm hedge more when the firms value increases than when it declines. Applied to multinational firms, this implies that firms with higher taxes are more likely to hedge when a strong U.S. dollar has a positive effect on firm value than a negative one. The results
12 In international business literature, the research and development expenses often measure the effect of internalization of the firm.

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

135

regarding the effect of taxes as a percentage of total assets in Table 3 support this argument. Results of panel Bwhich uses contemporaneous exchange rate changes rather than distributed lag measures of exposuresconfirm the findings in panel A, except for the following. First, the effects of Asian sales ratios become insignificant (but still of the same signs) for the case of negative exposures, while they were significant in panel A. Second, again for negative exposure cases, the level of significance of the research and development expense variable has increased somewhat in panel B compared with panel A, while that of export sales has decreased. This indicates that, with respect to these two variables, the operational content of negative exchange exposure cases is somewhat less solid than that of positive exposure cases. All other discussions above remain intact, including the effects of all explanatory variables for positive exposure cases, regardless of whether distributed lags or contemporaneous exchange rate changes are used as the exposure variable. It should also be noted that the present results on the effects of risk management are based on the variables identified in existing work influencing the firms demand for hedging. No attempt has been made in this paper to directly measure the extent of the use of derivatives by firms.13 However, since we use both operational and risk management variables simultaneously, the results reported in this section are consistent with the notion that there exists an interaction between operational and financial strategies.

4. Country exposure To further investigate the nature of the Asian exposure of U.S. firms, we disaggregate the Asian currency index by individual currencies in Table 5. For simplicity, we use contemporaneous real exchange rates (local currency per U.S. dollar) in Table 5. The number of firms significantly exposed to exchange risk contemporaneously varies from 11% to 15% at the 10% level. Three currenciesthe Hong Kong dollar, the Korean won, and the Indian rupeeshow the largest proportion (more than 65%) of all significant exposure coefficients to be negative, with another currency (the Malaysian ringgit) showing 57% of all significant exposure to be negative. The negative coefficients indicate that the U.S. firms, on balance, lose from a strong U.S. dollar in terms of their business operations in these countries. This is the case if U.S. firms are net exporters to these countries and/or if they set up operations there for the purpose of local sales. In contrast, a majority of U.S. firms (more than 60%) are positively exposed to the Chinese yuan, the Singapore dollar, and the Indonesian rupiah. A strong U.S. dollar against these three currencies benefits U.S. firms more than it hurts them. Operationally, this is consistent with a notion that U.S. firms are net importers from these countries, and/or have operations in these countries for export to the world market. U.S. firms have more or less balanced exposure positions with respect to two remaining currencies, the Japanese yen and the Thai baht, implying both patterns of operational characteristics.

13

See Allayannis and Ofek (2001) for an analysis of actual data concerning corporate use of derivatives.

136 Table 5 Country exposure CNY Significant at 10% 10.93 Sign of exposure + 65.22 34.78

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

HKD 15.21

INR 11.50

JPY 12.17

KRW 12.64

MYR 11.22

SGD 11.21

THB 9.98

IDR 11.22

28.13 71.88

33.06 66.94

47.66 52.34

31.58 68.42

43.22 56.78

66.10 33.90

51.43 48.57

61.02 38.98

This table reports the currency exposure of U.S. firms to each Asian country. The exposure coefficients are estimated by regressing firm returns on market returns and each of nine Asian currencies. The numbers for the signs of exposures indicate the percentage of firms with a positive or a negative sign from the group of firms that have been found to be significant for their exchange exposure coefficients at the 10% level. The nine currencies are the Chinese yuan (CNY), the Hong Kong dollar (HKD), the Indian rupee (INR), the Japanese yen (JPY), the South Korean won (KRW), the Malaysian ringgit (MYR), the Singapore dollar (SGD), the Thai baht (THB) and the Indonesian rupiah (IDR).

It should be pointed out that these results in Table 5 indicate the degree of sensitivity of U.S. firms to individual currencies rather than countries as such. Therefore, obtaining a statistically significant exposure for a currency does not necessarily indicate that the firm has actual operations in that countryit might merely suggest that the firms products face competition from imports of similar Asian products into the U.S. and, therefore, are sensitive to exchange rates through operating exposure.

5. Conclusion The recent Asian financial crisis in the aftermath of a long economic malaise in Japan heightened the interest of U.S. management concerning the issue of exposure to Asia. The paper examines the exposure of U.S. multinational firms to Asian currencies using the cross-section and time series sample of 1052 firm-years during 1992 1997. We find both contemporaneous and lagged changes in real exchange rates are significant in explaining stock returns for about 30% of the sample. However, the signs of exposure coefficients are both positive and negative, which reflect the heterogeneous nature of exchange rate effects. We attempt to estimate the determinants of exposure coefficients formally on the basis of international operational and risk management variables. Estimation results show that higher export and Asian sales affect U.S. firms adversely when the U.S. dollar appreciates. Higher asset deployment in Asia, however, has a positive effect on firm value with a strong dollar. This suggests that the Asian operations are used as an export base to third countries. The value of exchange exposure also depends on the risk management strategies of the firm. Firms with higher growth opportunities and lower liquidity have more incentive to hedge, and thereby reduce the firms currency exposure. These results are consistent with the notion that interactions exist between operational and financial strategies of the firm, and that these interactions are reflected in the firms exchange exposure coefficient.

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

137

An analysis of individual country exposure also shows the heterogeneous nature of the Asian currency exposure of U.S. firms. We suggest that these results capture the different operational patterns of U.S. firms: net exporter versus importer from these countries, and production for local sales versus export to the world market. The basic conclusion is that the measured exchange exposure coefficients reflect the outcomes of operational and risk management strategies undertaken by the firm. We believe that this is the first study of its kind that investigates the nature of the Asian exposure of U.S. firms. However, the present work needs to be corroborated with country-specific financial and operational information, as well as validated in other regions.

6. For further reading The following works are also of interest to the reader: Adler and Dumas (1984), Choi (1989), Hodder (1982).

Acknowledgements The authors would like to thank Lilian Ng, the referee, and participants in the 2001 PACAP session for their helpful comments.

References
Adler, M., Dumas, B., 1984. Exposure to currency risk: definition and measurement. Financial Management, Summer, 41 50. Allayannis, G., Ofek, E., 2001. Exchange rate exposure, hedging, and the use of foreign currency derivatives. Journal of International Money and Finance 20 (2), 273 296. Amihud, Y., 1994. Evidence on exchange rates and valuation of equity shares. In: Amihud, Y., Levich, R.M. (Eds.), Exchange Rates and Corporate Performance. Irwin Professional Publishing, New York. Bartov, E., Bodnar, G.M., 1994. Firm valuation, earnings expectations, and the exchange-rate exposure effect. Journal of Finance 49, 1755 1785. Bodnar, G.M., Gentry, W.M., 1993. Exchange rate exposure and industry characteristics: Evidence from Canada, Japan, and the USA. Journal of International Money and Finance 12, 29 45. Choi, J.J., 1986. A model of firm valuation with exchange risk. Journal of International Business Studies 17 (2), 153 160 (Summer). Choi, J.J., 1989. Diversification, exchange risk and corporate international investment. Journal of International Business Studies 20 (1), 145 155 (Spring). Choi, J.J., Prasad, A.M., 1995. Exchange risk sensitivity and its determinants: A firm and industry analysis of U.S. multinationals. Financial Management 24 (3), 77 88 (Autumn). DeMarzo, P., Duffie, D., 1991. Corporate financial hedging with proprietary information. Journal of Economic Theory 53, 261 286. Dominguez, K.M.E., Tesar, L., 2001. A reexamination of exchange-rate exposure. American Economic Review, Papers and Proceedings, May 91 (2), 396 399. Fama, E.F., MacBeth, J.D., 1974. Tests of the multi-period two-factor models. Journal of Financial Economics 1, 43 66. Froot, K., Scharfstein, D., Stein, J., 1993. Risk management: Coordinating corporate investment and financing policies. Journal of Finance 48, 1629 1658.

138

J.J. Choi, Y.-C. Kim / Pacific-Basin Finance Journal 11 (2003) 121138

Geczy, C., Minton, B.A., Schrand, C., 1997. Why firms use currency derivatives? Journal of Finance 52-4, 1323 1354. Green, W.H., 1993. Econometric Analysis, 2nd ed. Macmillan, New York. He, J., Ng, L., 1998. Foreign exchange exposure of Japanese corporations. Journal of Finance 53 (2), 733 753. Hodder, J., 1982. Exposure to exchange rate movements. Journal of International Economics 13, 375 386. Jorion, P., 1990. The exchange rate exposure of U.S. multinationals. Journal of Business 63, 331 345. Kim, Y.-C., 1997. Stock price reactions to international investment and divestiture and management of currency operating exposure. Journal of Economics and Business 49, 419 437. Miller, K.D., Reuer, J.J., 1998. Firm strategy and economic exposure to foreign exchange rate movements. Journal of International Business Studies 29, 493 513 (Third Quarter). Nance, D.R., Smith Jr., C.W., Smithson, C.W., 1993. On the determinants of corporate hedging. Journal of Finance 48, 391 405. Smith, C.W., Stulz, R., 1985. The determinants of firms hedging policies. Journal of Financial and Quantitative Analysis 20, 391 405. Stulz, R., 1984. Optimal hedging policies. Journal of Financial and Quantitative Analysis 19, 127 140. Tufano, P., 1996. Who manages risk? An empirical examination of risk management practices in the gold mining industry. Journal of Finance 51, 1097 1137. Wall Street Journal, 1997a. Strong dollar creates winners and losers. July 6. Wall Street Journal, 1997b. How U.S. firms cope with Asian crisis. December 26. White, H., 1980. A heteroscedasticity-consistent covariance matrix estimator and a direct test for heteroscedasticity. Econometrica 48, 817 838.