University of Groningen
Faculty of Economics and Business
Exchange rate exposure and the impact
of the introduction of the Euro
Jelmer Veltman
Student nr: 1467514
Supervisor: A.J. Meesters and J.H. von Eije
University of Groningen
Faculty of Economics and Business
Exchange rate exposure and the impact
of the introduction of the Euro
Abstract
1. Introduction
In 1973, after the breakdown of the Bretton Woods system, exchange rates started floating. Since then, it is commonly believed by theorists as well as practitioners that exchange rate movements affect the value of the firm. For instance, the rapid growth of the foreign currency derivatives market can be seen as an indication that firms are more and more unwilling to bear exchange rate risk. However, empirical tests of exchange rate exposure have not shown a clear relationship between firm value and exchange rate movements so far.
This study will empirically test for exchange rate exposure by making use of the introduction of the Euro. To be more precise, this study will test the impact of the Euro on the exchange rate exposure of four industries in France, Germany, Spain and The Netherlands. The introduction of the Euro is an interesting event for exchange rate exposure literature, because one of the main arguments for the creation of the Euro was eliminating exchange rate exposure.
The first objective of this study is to test for exchange rate exposure of four industries in Germany, France, Spain and The Netherlands in the period from August 1986 to June 2003. The four industries are chemicals, construction, food producers and business support services. In contrast to almost all previous studies, exposures are estimated using Seemingly Unrelated Regression (SUR). Only Bodnar and Gentry (1993), who investigate exposure at the industry level for a non-Euro sample, use a similar technique. The exposures found in this study are in line with the results found by Bodnar and Gentry (1993) and Muller and Verschoor (2004a), who use a Euro sample on the firm level.
Furthermore, the influence of industry and country characteristics on exposure is tested. There is no indication that traded goods industries have different exposure than non-traded goods industries. Also, the impact of the Euro does not differ between both types of industries. In addition, the results show that The Netherlands, as an open economy, does not have more exposure than the closed economies in the sample. In contrast, industries of Spain, which had a weak currency before the introduction of the Euro, experience more reduction in market risk than industries from countries with a strong currency before the Euro.
The results of this study can be valuable in practice. As Bartram and Karolyi (2006) point out, the reduction in market risk after the introduction of the Euro implies a reduced cost of capital. As a result, firms are able to carry more business risk or sustain a higher degree of financial leverage.
The remainder of this paper is organised as follows. The next section starts with the theory of exchange rate exposure, followed by an overview of the empirical literature. Subsequently, the data are described in section three. Section four deals with the methodology, while the results are presented in section five. Finally, section six offers some concluding remarks.
2. Literature
From a theoretical point of view there is consensus that exchange rate movements affect the value of a multinational firm. According to Shapiro (1975), the exposure of a multinational depends on the distribution of the sales between domestic and export markets, the amount of import competition in its home country, and the degree of substitutability between local and imported factors of production. Bodnar, Dumas and Marston (2002) focus on the ability of a firm to pass-through any change in exchange rate to the customer. If elastic demand prevents a firm from passing-through price changes caused by exchange rate movements, then the firm has a relatively higher exposure. Alternatively, inelastic demand leads to passing-through changes in exchange rates and, as a consequence lower exposure.
changes affect the value of assets denominated in foreign currencies. These effects may differ between industries, because some industries consist of mainly exporting firms, whereas others consist of mainly importing firms. Another important industry characteristic is the difference between non-traded goods industries and traded goods industries. As Bartram and Karolyi (2006) indicate, traded goods industries are expected to be more exposed to exchange rate movements, because of the higher involvement in imports and exports.
Besides industry characteristics, exposure may differ between countries. Since the proportion of export influences the exposure, firms in countries with a relatively open economy are expected to have more exposure than firms in countries with a closed economy. Muller and Verschoor (2004a) state that the lack of exposure found in previous studies may be attributed to the dominance of the United States as the research area, because the United States is characterised as a relatively closed economy. Recent studies that investigate firms in more open economies (Domingues and Tesar, 2006; Doidge, Griffin and Williamson, 2006) seem to confirm this point.
Surprisingly, the empirical research of exchange rate exposure is relatively new. After the exchange rates started floating in 1973, it took eleven years before Adler and Dumas (1984) came up with a model to measure exchange rate exposure. Their often-cited article is the foundation for nearly all successive articles about exchange rate exposure. The model is simple and straightforward as it relates the stock return of a firm to the return on an exchange rate variable using linear regression.
In most of the studies monthly returns are used. (Jorion, 1990; Amihud, 1994; He and Ng, 1998) Recently, some studies used weekly returns (Muller and Verschoor, 2004b; Muller and Verschoor, 2004a; Dominguez and Tesar, 2006) However, Bodnar and Wong (2003) argue that it may take some time before exposure is apparent due to the complex factors influencing exposures. In addition, the noise in high-frequency data should make exposures estimates over longer horizons better. Their results show that the magnitude of the exposure increases with the return horizon, although large return horizons relative to the sample size should be avoided.
Bodnar and Gentry, 1993; Bodnar and Wong, 2003) make use of a trade-weighted index of exchange rates. Williamson (2001) points out that the use of a trade-weighted index may lack power if a firm is only exposed to a few currencies from the index. As a consequence, the results would underestimate the real exposure of the firm. A solution to this problem is to create firm-specific exchange rates. However, it is far from clear on which basis such an alternative index should be created. On the other hand, the use of one or more single exchange rates suffers from the same problem (Dominguez and Tesar, 2006). Due to the complex factors influencing exposure it is unclear which currencies have an impact on the exposure of a firm. Furthermore, the inclusion of several single currencies may lead to a correlation problem. (Miller and Reuer, 1998).
Dominguez and Tesar (2006) compare the exposures using different exchange rate variables. The trade weighted index seems to give a slightly higher exposure than the US dollar or the currency of the main trading partner. However, the percentage of firms that is exposed to the dollar, but not to the trade weighted index is relatively high, suggesting that the trade weighted index may not be the perfect indicator of exposure.
The coefficient of the exchange rate variable is often called the total exposure of a firm. However, from a statistical point of view it is only the part of the stock return variance that is correlated with the exchange rate variable. Jorion (1990) argues that macroeconomic variables may have an impact on the stock returns and the exchange rate simultaneously. Therefore, the exposure coefficient does not only measure the total exposure of a firm, but also the impact of macroeconomic events. Examples of these macroeconomic variables are the risk-free rate, the market risk premium and investor sentiment (Bodnar and Wong, 2003). Hence, Jorion (1990) included a market return to the original model.
have exchange rate exposure. In that case, the exposure would be equal to the exposure of the market.
According to Bodnar and Wong (2003), the form of the macroeconomic control variable can have a substantial impact on the exposure. Most studies make use of a value-weighted market return (e.g. Jorion, 1990). The value-value-weighted market return gives more importance to large firms, whereas in an equally-weighted market return the small firms are relatively more important. Generally, large firms or multinationals are export oriented, while small firms are import oriented. As a result, the exposure of the market differs between both the value- and the equally-weighted market return. In addition, the residual exposure, measured as the difference between the total exposure and the market exposure, differs as well. The results of Bodnar and Wong (2003) confirm this point.
Besides the aforementioned ways to measure exchange rate exposure, there have been alternative attempts. For example, Bartram (2007) relates the exchange rate variable to corporate cash-flows. The results are similar to the stock return approach. However, his idea is called into question by an earlier work of Muller and Verschoor (2006), who state that the cash-flow approach is past-oriented. They prefer the forward-looking approach of the stock return.
Doidge, Griffin and Williamson (2006) measure exposure by forming portfolios. The portfolio with high international sales outperforms the portfolio with no international sales in times of large depreciations. On the other hand, during large appreciations, the portfolio without international sales performs better. The advantage of this approach is that it gives economic significance to the exposure theory. However, they also conclude that the exchange rate does not explain a large portion of the variance in individual firm stock returns.
The empirical results of measuring exchange rate exposure are disappointing. Jorion (1990) finds that only 15 of the 287 US multinationals exhibit exposure between 1971 and 1987. Additionally, Amihud (1993) finds almost no exposures in his sample. Evidence from more open economies shows higher exposures. For example, He and Ng (1998) find that 25% of the Japanese firms have significantly positive exposure. An extended overview of the empirical results can be found in table 1.
changes in international conditions at low costs, then the exchange rate exposure can be reduced. However, there are difficulties investigating this relationship, because of the lack of data about hedging activities of firms.
Empirical results from exchange rate exposure at the industry level are more scarce. Bodnar and Gentry (1993) relate the return on industry portfolios in the US, Canada and Japan to the return on a trade-weighted index. 28% of the industries in the US are exposed, in Canada and Japan this percentage is respectively 21 and 35. The variance in exposures is higher for Canada and Japan than for the US, suggesting that exchange rate movements have more impact in more open economies. Furthermore, industry characteristics help determine exposure at the industry level.
Griffin and Stulz (2001) conclude that exchange rate shocks have a negligible impact on the value of industries across the world. In addition, they find no evidence that an industry in one country performs better at the cost of the same industry in another country. Muller and Verschoor (2006) point out that the weak results of Griffin and Stulz (2001) are caused by the aggregation of individual exposures. They believe that the aggregation of exposures is questionable, because the individual exposures may have an opposite sign. The result of the aggregation process then shows insignificant results, whereas the firms individually are exposed to exchange rate movements. This problem is severe for aggregation of a diverse set of firms. Nevertheless, aggregation on the industry level is justified, if industry characteristics determine for a large part the exposure. In that case, firms in the same industry would have the same direction of exposure.
Table 1: overview of the empirical literature on exchange rate exposure on a firm-level.
Authors Sample
period
Sample area Sample size(firms)
Data frequency
Market variable Exchange rate
variable
Exposure
Jorion (1990) 1971-1987 US 287 monthly Value-weighted TW-index 5%
Amihud (1993) 1979-1988 US 32 monthly Equally-weighted EW-index None
Choi & Prasad (1995) 1978-1989 US 409 monthly Value-weighted TW-index 15%
Miller & Reuer (1998) 1987-1992 US 126 monthly Value-weighted Multiple currencies 13 to 17%
He & Ng (1998) 1979-1993 Japan 171 monthly Value-weighted TW-index 25%
Bodnar & Wong (2003) 1977-1996 US 910 monthly Equally-weighted
Value-weighted
TW-index 23%
22%
Muller & Verschoor (2004a) 1988-2002 Europe 817 weekly Value-weighted Multiple currencies 13-22%
Muller & Verschoor (2004b) 1993-2003 Asia 3634 weekly Value-weighted Multiple currencies 22-25%
Muller & Verschoor (2004c) 1990-2001 US 935 weekly Value-weighted TW-index 29%
Dominguez & Tesar (2006) 1980-1999 Chili
The reduction in exchange rate exposure following the introduction of the euro is investigated by Nguyen, Faff and Marshall (2007) and Bartram and Karolyi (2006). Nguyen et al. (2007) prove, for a French sample, that the number of firms that has exposure after the introduction of the euro is reduced as well as the size of the exposure. Additionally, French firms tend to use less foreign currency derivatives as a result of the reduction in exposure.
A more complete overview of the impact of the euro on exchange rate exposure is given by Bartram and Karolyi (2006). Their study includes three stages. In the first stage they show that the introduction of the Euro was accompanied with an increase in stock market return volatility. However, this increase in volatility is lower for firms that have more exposure to the Euro. In the second part they demonstrate a reduction in market risk following the introduction of the Euro, suggesting that exchange rate risk is part of systematic risk. The change in market beta is not caused by a change in financial leverage. The last stage shows that in general the incremental exchange rate exposure is reduced, although the results are not very significant. In addition, firm-characteristics, like total sales and the proportion of sales in Europe, and industry characteristics, like competition and trade, are important determinants of exposure. The influence of country characteristics is less clear. The difference between strong EMU countries and weak EMU countries is not significant. An important point that the authors make is that the reduction of exposures is expected to be bigger for firms that have more ex-ante exposure.
3. Data
The sample consists of industries from Germany, France, Spain and The Netherlands. The countries are selected on the basis of data availability as well as country characteristics. For example, The Netherlands is selected, because it has a relatively open economy. According to data from the Worldbank, The Netherlands had in 2000 an index of openness1 of 61, whereas Germany, France and Spain had an index of openness of respectively, 33, 29 and 30. Spain is selected because it had a weak currency before the Euro. As in Bartram and Karolyi (2006) a weak currency is defined as a currency which has suffered a crisis in the years before the Euro and has been the target of speculative attacks.
1
The dataset includes, for each country in the sample, the domestic market return, the trade weighted currency index and four industry returns. All return measures are on a monthly basis covering the period from August 1986 to June 2003. The measurement point is the first day of the month and returns are continuously compounded. The data is obtained from Datastream.
The market return for Germany, France, Spain and The Netherlands is respectively, the DAX 30, CAC 40, IBEX 35 and the AEX-index. These four indices are all value-weighted indices. The sample period includes several extreme events, which influenced the market return heavily. The minimum return of all four market returns is on November 1987. This is a result of the global stock market crash of October 19th 1987. In the late nineties the stock markets were influenced by the bubble of the internet stocks, while the low returns in October 2001 can be explained by the stock market decline following September 11th. A full overview of the descriptive statistics of the market returns is in table 2A.
This study makes use of JP Morgan’s trade-weighted currency index. The index measures the strength of a currency relative to a basket of 18 other OECD currencies. The currencies are selected on the basis of trading amount with the home currency. The basket-weights reflect the pattern of bilateral trade between the countries. The index is defined as the price, measured in the home currency, for a foreign currency. Consequently, an increase of the index indicates a relative depreciation of the home currency. Accordingly, a negative return denotes a relative appreciation of the home currency.
The industries included in the sample are chemicals, construction, food producers and business support service. They are selected on the basis of trading characteristic (traded goods or non-traded goods) and data availability. Bodnar and Gentry (1993) classify a non-traded goods industry as an industry where high transportation costs prevent international trade. Accordingly, chemicals and food producers are considered traded goods industries, whereas construction and business support service are considered non-traded goods industries.
Industry portfolio returns are created as the average of the continuously compounded return of individual firms in the industry. Firms are selected from the aforementioned four industries according to the industry classification of Datastream. Furthermore, firms should have stock returns prior to and after the introduction of the Euro. More specifically, firms are excluded from the sample if they have less than two years of stock returns in one of the two periods. A full list of the firms can be found in appendix A. More details of the industry returns are presented in table 3.
Table 2A: Descriptive statistics of the market return
The returns are continuously compounded on a monthly basis. The sample period, August 1986 to June 2003, includes 202 observations. Moreover, the Jarque-Bera values reject the normality hypothesis at the 1% level.
minimum maximum median mean standard
deviation Jarque-Bera Germany -0,2669 0,1830 0,0128 0,0043 0,0678 49,5 France -0,2633 0,1954 0,0087 0,0038 0,0660 21,4 Spain -0,2724 0,2342 0,0073 0,0047 0,0709 24,7 The Netherlands -0,3534 0,1400 0,0121 0,0043 0,0642 300,0
Table 2B: Descriptive statistics of the return on the trade weighted index
The returns are continuously compounded on a monthly basis. The sample period, August 1986 to June 2003, includes 202 observations. Moreover, the Jarque-Bera values of France and Spain reject the normality hypothesis at the 1% level.
minimum maximum median mean standard
Table 3: Descriptive statistics of the industry portfolio returns.
The returns are continuously compounded on a monthly basis. The sample period, August 1986 to June 2003, includes 202 observations, however, some industries have less observations because of data availability. The number of industry observations is displayed in the third column: “N”. Furthermore, the second column “F” shows the number of firms from which an average industry portfolio return is drawn.
F N minimum maximum median mean standard
deviation Germany
Chemicals 13 202 -0,232 0,184 -0,001 0,000 0,051
Construction 21 202 -0,216 0,159 -0,002 -0,001 0,056
Food producers 9 202 -0,121 0,136 0,001 0,001 0,045
Bus. sup. service 5 176 -0,579 0,320 -0,017 -0,018 0,107
France
Chemicals 8 190 -0,247 0,174 0,000 0,001 0,062
Construction 20 190 -0,299 0,156 0,005 0,003 0,057
Food producers 18 190 -0,158 0,156 0,004 0,003 0,046
Bus. sup. service 10 188 -0,604 0,299 0,005 0,006 0,085
Spain
Chemicals 1 195 -0,707 0,657 -0,025 -0,013 0,172
Construction 9 195 -0,552 0,295 0,005 0,000 0,109
Food producers 5 195 -0,459 0,242 0,003 0,005 0,086
Bus. sup. service 2 169 -0,515 0,903 -0,011 -0,004 0,133
The Netherlands
Chemicals 3 202 -0,413 0,162 0,007 0,002 0,069
Construction 7 202 -0,252 0,204 0,005 0,003 0,058
Food producers 5 202 -0,178 0,108 0,005 0,001 0,044
4. Methodology.
Following many authors (e.g. Miller and Reuer, 1998; He and Ng, 1998), this study uses the market model, introduced by Jorion (1990), to measure exchange rate exposure. Although the original model was used for measuring exposures at the firm level, Bodnar and Gentry (1993) show that equation 1 is also applicable for measuring exposures at the industry level.
0, 1, 2,
ijt ij ij jt ij jt ijt
R
=
β
+
β
Rm
+
β
Rfx
+
ε
(1)Where Rijt is the stock return of industry i in country j at time t. Rmtjt is the return on the
market index and Rfxjt is the return on the trade-weighted index. All return measures are
continuously compounded. The choice for a trade-weighted index, instead of single currencies, is related to the industry level. The main argument against a trade-weighted index is that some firms may only be exposed to a few currencies from the index. However, this problem is less likely for an aggregation of firms. Therefore a trade-weighted index is included as the exchange rate variable.
In equation 1, β2,ij indicates the average residual exposure of industry i in country j
over the estimation period. An important implication of equation 1 is that the coefficients are assumed to be constant over time. This assumption is called into question by, among others, Koutmos and Martin (2007), who argue that exchange rate exposures are time-varying.
The idea of time-varying exposures is an important element in this study. However, instead of continuous time-varying exposures, only the change in exposure around the introduction of the Euro is of interest. Therefore, a dummy variable is used instead of a more complex econometric model. The impact of the Euro on the exchange rate exposure of industries is measured by equation 2:
0, 1, 2, 3, 4, 5,
ijt ij ij jt ij jt ij t ij t jt ij t jt ijt
R
=
β
+
β
Rm
+
β
Rfx
+
β
D
+
β
D Rm
+
β
D Rfx
+
ε
(2)Where Rijt, Rmjt and Rfxjt are defined as in equation 1. Dt is a dummy variable, which takes
in market risk, while β5,ij measures the change in residual exposure following the
introduction of the Euro.
Equation 2 is almost similar to the model used by Bartram and Karolyi (2006) to test for the impact of the Euro on the exchange rate exposure of firms. The only difference is the inclusion of β3,ijDt in equation 2. β3,ij measures the difference in the constant, β0,ij , after
the introduction of the Euro. Although the term is not directly related to one of the main research questions, from a statistical point of view, this term should be included. Brambor, Clark and Golder (2005) show that constitutive terms should always be included in multiplicative interaction models except for very rare circumstances. Omitting β3,ijDt
from the model is the same as assuming that β3,ij is zero. But even if there are good
reasons to assume that β3,ijDt does not influence the industry return, the term should still be included, because it influences the estimation of the coefficients of the other variables.
The sample consists of four countries and four industries. Hence, equation 1 and 2 are estimated sixteen times. These sixteen equations could be estimated separately by ordinary least squares. However, the individual estimated relationships may contain subtle interactions with each other. For example, the relationship between the return on the trade weighted index of Germany and the industry return of Germany’s chemicals may be somewhere related with the relationship between the trade weighted index of Germany and the industry return of Germany’s construction. Moreover, Srivastava and Giles (1987) argue that if the equations contain interactions with each other, then treating the model as separate equations is suboptimal. Therefore, following Bodnar and Gentry (1993), this study estimates equation 1 and 2 using SUR.
The SUR approach is a joint estimation of the individual equations, based on the idea that the errors of the individual equations are correlated with each other. The assumption of correlated errors adds extra information to the system of equations. Hence, estimating the equations jointly should improve the estimates.
of each industry i are estimated jointly. This leads to four different models of equations, one model per industry.
In addition a SUR will be performed for the full sample. The rationale behind one model of sixteen equations is that the domestic market return does not fully represent the macroeconomic variables that have a simultaneous impact on the industry return and the trade weighted exchange rate. This effect in combination with the industry effect leads to a joint estimation of all sixteen equations.
The third stage of this study deals with the influence of industry and country characteristics on the exposure and on the impact of the Euro on the exposure. From the previous discussions in the literature and data section, three testable hypotheses are drawn. 1. The reduction of exposure is bigger for traded goods industries than for
non-traded goods industries.
2. The reduction of exposure is bigger for industries in open economies than for industries in closed economies
3. The reduction of exposure is bigger for industries in a country with a weak former currency than for industries in a country with a strong former currency. For the first two hypotheses the difference in the extent of the exposure is also tested. This is done, since traded goods industries and open economies are expected to have more reduction in exposure, because they have more ex-ante exposure. Coefficients from equation 1 and 2 are used for answering the hypotheses. The coefficients of interest are compared by means of a Wald test. The null hypothesis of the Wald test is that both coefficients are equal to each other. However, comparing positive exposures with negative exposures gives only information about the difference in exposure and not about the difference in the extent of the exposure. Therefore, all coefficients that represent positive exposures have been given a minus sign in the Wald test. In this way the difference in the extent of the exposure is tested. A similar procedure is executed for the change in the exposure following the introduction of the Euro. An important assumption of this procedure is that exposure is symmetric, that is, positive exposures behave in the same way as negative exposures.
5. Results
Table 4 presents estimates for regression equation 1. This equation test for exchange rate exposure on the industry level. The estimates are obtained with the SUR-approach. Table 4 shows the results of the joint estimation of all sixteen equations. The results of a
Table 4. Exchange rate exposures on the industry level.
The table shows the estimates of Rijt=β0,ij+β1,ijRmjt+β2,ijRfxjt+εijt over the full period: 1986-2003. The
equations are estimated jointly with the SUR-approach. The standard deviations are displayed between the brackets. The significance levels are ***.** and * for respectively, 1%,5% and 10%.
β0,ij β1,ij β2,ij Adj. R2
Germany Chemicals -0,001 (0,002) 0,548*** (0,035) -0,253 (0,270) 0,543 Construction -0,004 (0,003) 0,574*** (0,043) 0,288 (0,328) 0,444 Food producers -0,000 (0,003) 0,251*** (0,043) -0,261 (0,329) 0,130
Bus. sup. service -0,018**
(0,008) 0,460*** (0,118) -0,141 (0,867) 0,061 France Chemicals -0,002 (0,003) 0,653*** (0,049) 0,391 (0,367) 0,449 Construction 0,001 (0,003) 0,609*** (0,041) 0,107 (0,298) 0,486 Food producers 0,001 (0,003) 0,415*** (0,040) 0,022 (0,300) 0,328
Bus. sup. service 0,002
(0,005) 0,853*** (0,068) 0,904* (0,505) 0,415 Spain Chemicals -0,017* (0,010) 1,347*** (0,140) 2,269*** (0,786) 0,299 Construction -0,004 (0,006) 1,003*** (0,078) 0,390 (0,428) 0,422 Food producers 0,002 (0,005) 0,761*** (0,063) -0,044 (0,341) 0,388
Bus. sup. service -0,005
(0,009) 0,923*** (0,128) 0,365 (0,677) 0,189 The Netherlands Chemicals -0,001 (0,003) 0,795*** (0,046) -0,982** (0,442) 0,624 Construction 0,001 (0,003) 0,503*** (0,050) -0,826* (0,484) 0,376 Food producers -0,001 (0,002) 0,461*** (0,037) 0,000 (0,356) 0,456
Bus. sup. service -0,002
joint estimation on the industry level differ only marginally and are therefore included appendix B.
The market variable is significantly positive for all sixteen industries. This result should not come as a surprise, as it is generally known that stock returns are positively correlated with the market return. The average value for the market coefficient is 0,458; 0,633; 1,009 and 0,630 for respectively, Germany, France, Spain and The Netherlands.
Furthermore, four of the sixteen industries document significant residual exchange rate exposure at the ten percent significance level. Compared to previous evidence of exchange rate exposure, the 25 percent is about equal to exposures found on the firm level for the same region. Muller and Verschoor (2004a) find that 13 to 22 percent of the firms in the Euro-area have exposure, although there are more exposures in Germany, France, Spain and The Netherlands than in other Euro countries. Also, it has to be said that the industry with the biggest residual exposure, the chemicals industry of Spain, is the only industry which is made out of the return of a single firm.
A positive coefficient of exposure means that the return of an industry increases if the return on the trade-weighted index increases. In other words, industries that have a positive exposure benefit from a depreciation of the home currency. At the same time, they lose from an appreciation of the home currency. A depreciation leads to relatively decrease in the price of goods of the home currency. Consequently, industries with positive exposures can be seen as net-exporters. The two Dutch industries that have a significant negative residual exposure gain from an appreciation of the home currency and therefore can be seen as net-importers.
Table 5 shows the impact of the Euro on the exchange rate exposure. The coefficients are estimated using equation 2. The standard deviation of the coefficients is displayed between the brackets. The individual equations are estimated jointly with SUR. Once more, the joint estimation on the industry level can be found in the appendix (C), because the results differ only marginally with the results of the joint estimation of all sixteen equations.
The change in the residual exchange rate exposure is measured by β5,ij. Only one of
the sixteen industries has a significant value for this coefficient. The sign of Germany’s business support service industry is negative, whereas the residual exposure of that industry, β2,ij is positive. This means that there is a reduction of exchange rate risk.
Interestingly, the initial sign of the exposure, β2,ij in table 4, was negative. Taken as a
whole, about half of the industries in the sample report residual exposures of the same sign as the residual exposure multiplied by the dummy variable. Consequently, there is no evidence that the residual exposure is reduced following the introduction of the Euro.
As opposed to finding significant values for the residual exposures, significant values for the market variable are found. In fifteen of the sixteen cases, there is a reduction in market risk, while in 80 percent of the cases this reduction is significant. This surprising result is also found by Bartram and Karolyi (2006). Since they prove that this reduction is not caused by a change in financial leverage, it leads them to suggest that exchange rate exposure is part of systematic risk. A reduction in market risk in combination with an insignificant change in residual exchange rate exposure can have important implications for the exchange rate exposure literature. Since Jorion (1990), nearly all studies have used a market return as a control variable. However, the market return captures a big part of the exchange rate exposure, and consequently the measure for exchange rate exposure loses its value. This result is similar to the statement of Bodnar and Wong (2003), that the exposure coefficient measures the residual exchange rate exposure of a firm or industry. Consequently, the market variable is not the ideal control variable in controlling for macroeconomic variables that simultaneously affect the exchange rate variable and the stock return.
Table 5. The impact of the Euro on the exchange rate exposure on the industry level.
The coefficients are estimated over the full period 1986-2003 with equation 2:
0, 1, 2, 3, 4, 5,
ijt ij ij jt ij jt ij t ij t jt ij t jt ijt
R =β +β Rm +β Rfx +β D +β D Rm +β D Rfx +ε . The standard deviation is displayed between the brackets, while the last column shows the adjusted r-squared. All equations are estimated jointly, using the SUR approach. The significance levels are ***.** and * for respectively, 1%,5% and 10%.
Β0,ij Β1,ij Β2,ij β3,ij Β4,ij β5,ij Adj.
The third objective investigates whether industry and country characteristics influence exposures and/or the impact of the Euro on exposures. Given the results on the impact of the Euro, the change in market variable is included in the analysis, instead of the change in the residual exposure. It is impossible to make distinctions in the reduction of the residual exposures, because several industries show an increase of residual exposure. On the other hand, the reduction in the market variable can not be seen as a reduction of exposure. Consequently, the hypotheses about reductions in exposure cannot be correctly tested. However, because the reductions in market risk are significant, it is interesting to investigate if this reduction is influenced by industry and country characteristics.
First of all, the differences between traded goods and non-traded goods industries are presented in table 6. The traded goods industries, chemicals and food producers, are in the columns, whereas the non-traded goods industries, construction and business support services, are displayed in the rows. The numbers in the table indicate the difference in the extent of, the exposure, and the impact of the Euro on the market variable. To measure the difference in the magnitude, and not the difference between positive and negative exposures, all coefficients of interest from equation 1 and 2 have been given a minus sign. Subsequently, the value for the non-traded goods industry is deducted from the traded goods industry. For example, the value of 0,03 for Germany’s chemical industry and the construction industry is calculated as -0,253 minus -0,288. As a result, the values in the first two columns of table 6 which have a minus sign, signify that the extent of the exposure is bigger for the traded goods industry than for the non-traded goods industry. The same yields for the values for the change in market variable. The values between the brackets in table 6 show the probability that the difference in magnitude is equal to zero. It is the probability that results from the Wald test.
Table 6. Traded goods versus non-traded goods industries.
The reported values are the difference in magnitude between traded goods and non-traded goods
industries. The coefficient of interest for exposure is β2,ij of equation 1, and for the market variable β4,ij
of equation 2. The probability of equality is shown between the brackets. This probability is calculated
with the Wald test for equal coefficients.
Exposure Reduction in Market variable
Germany Chemicals Food Prod. Chemicals Food Prod.
Construction 0,03 (0,94) 0,03 (0,96) 0,11 (0,21) -0,12 (0,23)
Bus. Sup. service -0,11 (0,90) -0,12 (0,89) 0,35 (0,13) 0,12 (0,60)
France
Construction -0,28 (0,45) 0,09 (0,81) 0,15 (0,17) 0,07 (0,49)
Bus. Sup. service 0,51 (0,36) 0,88 (0,11) -0,24 (0,13) -0,32 (0,04)
Spain
Construction -1,88 (0,02) 0,35 (0,57) 0,02 (0,94) 0,49 (0,01)
Bus. Sup. service -1,90 (0,03) 0,32 (0,68) -0,53 (0,15) -0,06 (0,83)
The Netherlands
Construction -0,16 (0,81) 0,83 (0,13) -0,14 (0,32) -0,07 (0,54)
Bus. Sup. service -0,94 (0,11) 0,05 (0,93) -0,33 (0,02) -0,26 (0,04)
Also, the impact of the Euro on the market variable does not differ between traded goods and non-traded goods industries. A negative value means that the reduction in the market variable is bigger for the traded goods industry than for the non-traded goods industry. However, positive values are just as likely as negative values. Only The Netherlands confirms the expectations set out in the hypothesis; all four combinations are negative. Overall, the results differ from Bartram and Karolyi (2006), who come to the conclusion that the difference between traded goods and non-traded goods industries influences the exposure and the reduction in exposure.
countries. There are less negative values than positive values. So, the industries of The Netherlands do not seem to have more exposure than industries in closed economies. However, there are some differences between the industries. The Netherlands has more exposure in the construction industry, whereas the food producing industry of The Netherlands has less exposure. This suggests that other industry characteristics, as an alternative to traded goods and non-traded goods industries, may play a role in influencing exposure.
Table 7B shows that the reduction in market risk is not bigger for The Netherlands than for a relatively closed economy. This result is a logical consequence of the findings in table 7A, which denote that the industries of The Netherlands do not have more ex-ante exposure than industries in closed economies. In contrast to the results of the exposures, the reduction in market risk does not seem to be influenced by other industry characteristics.
Table 7. Open versus closed economies.
The reported values are the difference in magnitude between The Netherlands, as an open economy, and “closed” countries.
Table 7A. Difference in the extent of the exposure between open and closed economies.
The coefficient of interest for exposure is β2,ij of equation 1. The probability of equality is shown between the
brackets. This probability is calculated with the Wald test for equal coefficients.
Chemicals Construction Food Producers Bus. Sup. service
Germany -0,73 (0,14) -0,54 (0,38) 0,26 (0,55) 0,09 (0,92)
France -0,59 (0,33) -0,72 (0,24) 0,02 (0,96) 0,86 (0,25)
Spain 1,29 (0,16) -0,44 (0,50) 0,04 (0,93) 0,32 (0,69)
Table 7B. Difference in the reduction in market risk between open and closed economies.
The coefficient of interest for the market variable is β4,ij of equation 2. The probability of equality is shown
between the brackets. This probability is calculated with the Wald test for equal coefficients.
Chemicals Construction Food Producers Bus. Sup. service
Germany -0,35 (0,00) -0,10 (0,45) -0,05 (0,63) 0,33 (0,20)
France -0,13 (0,31) 0,16 (0,19) 0,02 (0,87) -0,05 (0,82)
Table 8 shows the difference in reduction between the industries of Spain and the other three countries. The industries of Spain are expected to have a greater reduction in exposure, because the former currency of Spain is considered weak. As a consequence, the introduction of the Euro would have a bigger stabilization effect than in other countries. The values in table 8 are obtained in the same way as in table 6 and 7. Ten of the twelve industry comparisons show a bigger reduction in market risk for Spanish industries. This suggests that the strength of the former currency plays a role in the reduction of the market variable. Although part of this result can be explained by higher initial values for the market variable.
There is also a difference in significance between industries. The reduction in market risk of the construction industry of Spain is significantly greater than its counterparties in the other three countries. In contrast, the reduction in market risk of the food producing industry of Spain is about equal to the food producing industries of Germany, France and The Netherlands.
6. Conclusion
This study empirically tests for exchange rate exposure and the impact of the Euro on the exchange rate exposure of four industries in four European countries. In addition, it investigates if the impact differs between industries and countries. In case of the industries, there is a distinction between traded goods and non-traded goods industries. The countries are categorized as an open economy or a closed economy. Moreover, countries are compared on the basis of the strength of the former currency.
First, four of the sixteen industries document exchange rate exposure at the ten percent significance level. This result may look disappointing given the fact that there is a clear Table 8. Difference in the reduction in market risk between weak and strong currencies.
The coefficient of interest for the market variable is β4,ij of equation 2. The probability of equality is shown
between the brackets. This probability is calculated with the Wald test for equal coefficients.
Chemicals Construction Food Producers Bus. Sup. service
Germany -0,76 (0,02) -0,67 (0,00) -0,07 (0,69) 0,12 (0,74)
France -0,55 (0,10) -0,42 (0,02) 0,00 (0,99) -0,26 (0,42)
theoretical relationship between exchange rate movements and firm value. However, the results are completely in line with previous evidence of exchange rate exposure.
The results show that only one industry experiences a significant reduction in exposure following the introduction of the Euro. However, twelve of the sixteen industries display a significant decrease in market risk. This result suggests that exchange rate risk is captured by the market variable instead of the exchange rate variable. It is this result that can have important implications for the future literature about exchange rate exposure. The market variable, which is included as a control variable, seems to do more than only controlling. Moreover, it may explain the low exposures found in empirical tests of exchange rate exposure.
Furthermore, this study shows that the difference between traded goods and non-traded goods industries does not influence the exposures nor the impact of the Euro on the reduction in market risk. Also the openness of a country is not an important determinant of exposure. On the other hand the strength of the former currency seems to play a role, as in general the industries of Spain experience a bigger reduction in market risk than industries of Germany, France and The Netherlands.
References
Adler, Michael, and Bernard Dumas, 1984, Exposure to currency risk: definition and measurement, Financial Management Summer, 41-50.
Amihud, Yakov, 1994, Evidence on exchange rates and valuation of equity shares, in Y. Amihud and R.M. Levich, eds.: Exchange Rates and Corporate Performance (Irwin Professional Publishing, New York).
Bartov, and Gordon M. Bodnar, 1994, Firm valuation, earning expectations, and the exchange-rate exposure effect, Journal of Finance 49, 1755-1785.
Bartram, Sohnke M., and G. Andrew Karolyi, 2006, The impact of the introduction of the Euro on foreign exchange rate risk exposures, Journal of Empirical Finance 13, 519-549. Bartram, Sohnke M., 2007, Corporate cash flow and stock price exposures to foreign
exchange rate risk, Journal of Corporate Finance 13, 981-994.
Bodnar, Gordon M., and William 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., Bernard Dumas, and Richard C. Marston, 2002, Pass-through and exposure, Journal of Finance 62, 199-231.
Bodnar, Gordon M. and M.H. Franco Wong, 2003, Estimating exchange rate exposures: Issues in model structure, Financial Management 32 (1), 35-67.
Brambor, Thomas, William Roberts Clark, and Matt Golder, 2005, Understanding interaction models: Improving empirical analyses, Political Analysis 14, 63-82.
Bun, Maurice J.G., and Franc J.G.M. Klaassen, 2007, The Euro effect on trade is not as large as commonly thought, Oxford Bulletin of Economics and Statistics 69, 473-496.
Choi, Jongmoo Jay, and Anita Mehra Prasad, 1995, Exchange risk sensitivity and its determinants: a firm and industry analysis of U.S. Multinationals, Financial Management
24 (3), 77-88
Doidge, Craig, John Griffin, and Rohan Williamson, 2006, Measuring the economic importance of exchange rate exposure, Journal of Empirical Finance 13, 550-576.
Dominguez, Kathryn M.E., and Linda L. Tesar, 2006, Exchange rate exposure, Journal of
International Economics 68, 188-218.
He, Jia, and Lilian K. Ng, 1998, The foreign exchange exposure of Japanese multinational corporations, Journal of Finance 53, 733-753.
Hill, R. Carter, William E. Griffiths, and George G. Judge, 2001, Undergraduate Econometrics, (John Wiley & Sons, Inc., New Jersey).
Jorion, Phillipe, 1990, The exchange rate exposure of U.S. Multinationals, Journal of
Business 63 (3), 331-345.
Griffin, John, and René M. Stulz, 2001, International competition and exchange rate shocks: A cross-country industry analysis of stock returns, Review of financial studies 14 (1), 215-241
Koutmos, Gregory, and Anna D. Martin, 2007, Modelling time variation and asymmetry in foreign exchange exposure, Journal of Multinational Financial Management 17, 61-74. Lopez, Claude and David H. Papell, 2007, Convergence to purchasing power parity at the
commencement of the Euro, Review of International Economics 15 (1), 1-16
Micco, Alejandro, Ernesto Stein, and Guillermo Ordonez, 2003, The currency union effect on trade: early evidence form EMU, Economic Policy 18, 315-356.
Miller, Kent D., and Jeffrey J. Reuer, 1998, Firm strategy and economic exposure to foreign exchange rate movements, Journal of International Business Studies 29, 493-514.
Muller, Aline, and Willem Verschoor, 2004a, European foreign exchange risk exposure,
European Financial Management 12 (2), 195-220.
Muller, Aline, and Willem Verschoor, 2004b, Asian foreign exchange risk exposure, Journal
of Japanese International Economies 21, 16-37.
Muller, Aline, and Willem Verschoor, 2004c, Asymmetric foreign exchange risk exposure: Evidence from U.S. multinational firms, Journal of Empirical Finance 13, 495-518.
Muller, Aline, and Willem Verschoor, 2006, Foreign exchange risk exposure: Survey and suggestions, Journal of Multinational Financial Management 16, 385-410.
Nguyen, Hoa, Robert Faff, and Andrew Marshall, 2007, Exchange rate exposure, foreign currency derivatives and the introduction of the Euro: French evidence, International
Review of Economics and Finance 16, 563-577.
Shapiro, Alan C., 1975, Exchange rate changes, inflation, and the value of the multinational corporation, Journal of Finance 30, 485-502.
Srivastava, Virendra K., and David E.A. Giles, 1987, Seemingly unrelated regression equation models: Estimation and inference, (CRC Press).
Appendix A – list of firms from which an industry return is drawn
Germany - Chemicals Germany – Bussiness Support Services
Altana AIS
Basf Agor
Bayer Kauffing
Fuchs Petrolub Pinguin
H&R Wasag Sero Entsorgung
Hench Thermoplast France – Chemicals
L&G Farbenindustrie Air Liquid
K + S Dynaction
Kalichemie Explos
Linde Exxon Mobil Chemical
New York Hamburger Gummi Orgasynth
Pongs & Zahn PCAS
Simona Plast Val loire
Germany – Construction Robertet
Bilfinger Berger France – Construction
Creaton A & CO
Didier Werke Bouygues
DT Steinzug Ciments Francais
Dyckerhoff CNIM
Heidelbergcement Colas
Hochtief Debuschere
Innotech Eiffage
Keramag Grosse Leon
Lindner Imerys
MDB Industr. Finc
Muehl Installux
Norddeutsche Steingut Lafarge
Pfleiderer Poujoulat
Pilkington Rougier
Sto Saint Gobain
Strabag Samse
Villeroy & Boch Selcodis
Walter Bau Vicat
Weru Vinci
Westag VM Materiaux
Germany – Food Producers France – Food Producers
A. Moksel Agricole de Crau
ADM Biscuits Gardeil
Baywa Bongrain
Frosta Cofigeo
KWS Saat Danone
Sachsenmilch Evialis
Schwaleben Molkerei Fala
Suedzucker Fromageries
LDC Boskalis
Naturex Gouda Vuurvast
Pltns T-rge hold Grontmij
SIPH Heijmans
Sucriere Pithiviers Ballast
Tipiak The Netherlands – Food Producers
Unibel Alanheri
Vermandoise Amsterdam commodities
Vilmorin CSM
France – Business Support Services Unilever
Altran Wessanen
Aurea The Netherlands – Business Support Services
Derichebourg Arcadis
Fimalac Corp Express
GAI DNC
Manutan Hagemeyer
Prodef Imtech
Securidev Randstad
Synergie Royal Reesink
Thermador RSDB
Spain – Chemicals USG People
Ercros
Spain - Construction Abbengoa
Acciona ACS
Cementos port Valder. Cleop
Formento constr. Obrascon Huarte Sacyr Vallehermoso Uralita
Spain – Food Producers Campofrio
Ebro Puleva Pescanova Rusticas Viscofan
Spain – Business Support Services Prosegur
Service Point Solutions The Netherlands – Chemicals
AKZO Nobel DSM
Holland Colours
The Netherlands – Construction BAM
Appendix B – The results of SUR with a joint estimation on the industry level.
Table 9. Exchange rate exposures on the industry level.
The table shows the estimates of Rijt=β0,ij+β1,ijRmjt+β2,ijRfxjt+εijt over the full period 1986-2003. Four
systems of equations are created, one for each industry. Subsequently, the equations in the systems are estimated jointly with the SUR-approach. The standard deviations of the coefficients are displayed between the brackets. The significance levels are ***.** and * for respectively, 1%,5% and 10%.
β0,ij β1,ij β2,ij Adj. R2
Germany Chemicals -0,001 (0,002) 0,541*** (0,036) -0,354 (0,279) 0,544 Construction -0,004 (0,003) 0,570*** (0,044) 0,318 (0,333) 0,443 Food producers -0,000 (0,003) 0,235*** (0,044) -0,435 (0,338) 0,132
Bus. sup. service -0,018**
(0,008) 0,394*** (0,121) -0,596 (0,896) 0,063 France Chemicals -0,001 (0,003) 0,648*** (0,051) 0,321 (0,382) 0,449 Construction 0,000 (0,003) 0,607*** (0,043) 0,199 (0,317) 0,485 Food producers 0,001 (0,003) 0,405*** (0,041) -0,048 (0,309) 0,328
Bus. sup. service 0,002
(0,005) 0,838*** (0,070) 0,829 (0,527) 0,415 Spain Chemicals -0,017* (0,010) 1,296*** (0,143) 2,286*** (0,796) 0,300 Construction -0,004 (0,006) 0,990*** (0,080) 0,450 (0,440) 0,422 Food producers 0,001 (0,005) 0,768*** (0,066) -0,097 (0,369) 0,388
Bus. sup. service -0,006
(0,009) 0,846*** (0,132) 0,207 (0,698) 0,190 The Netherlands Chemicals -0,001 (0,003) 0,789*** (0,048) -1,166** (0,459) 0,624 Construction 0,001 (0,003) 0,499*** (0,052) -0,850* (0,496) 0,376 Food producers -0,001 (0,002) 0,466*** (0,037) -0,072 (0,361) 0,456
Bus. sup. service -0,002
Appendix C
Table 10. The impact of the Euro on the exchange rate exposure on the industry level.
The coefficients are estimated over the full period, 1986-2003, with equation 2:
0, 1, 2, 3, 4, 5,
ijt ij ij jt ij jt ij t ij t jt ij t jt ijt
R =β +β Rm +β Rfx +β D +β D Rm +β D Rfx +ε . Four systems of equations are created, one for each industry. Subsequently, the equations in the systems are estimated jointly with the SUR-approach. The standard deviation is displayed between the brackets, while the last column shows the adjusted r-squared. The significance levels are ***.** and * for respectively, 1%,5% and 10%.
Β0,ij Β1,ij Β2,ij β3,ij Β4,ij β5,ij Adj.
