## Exchange Rate Exposure of Real Sector Firms in an Emerging Economy

**Turan Erol**^{1}, **Ayhan Algüner**^{2}, **Güray Küçükkocaoğlu**^{2,}

^{1}Prime Minister of Turkey, Ankara, Turkey

^{2}Management Department, Başkent University, Ankara, Turkey

### Abstract

Firms in emerging economies face a potentially higher degree of transaction and economic exposure compared to those in advanced economies. This is due to fact that most of hedging instruments available in advanced financial markets are not available for firms of emerging markets. Another peculiarity of the exposure of firms in emerging economies derive from the fact that these firms are usually price-takers in international trade and have little power to pass through the changes in exchange rates to foreign buyers. This study is one of the few attempts to directly estimate the exposure of real sector firms from an emerging economy. The paper focus on the determinants of exposure as revealed by the estimates from the two-factor CAPM model. Five main determinants of exposure that are thought to be the most relevant in the particular case of Turkey. These are the level of foreign sales, industry competition, net foreign non-financial liabilities, net foreign currency debt and finally domestic risk free assets. The estimates based on a sample of Turkish manufacturing firms reveal a positive but lagged exposure to real exchange rate changes. A result from the estimations of the two-factor model is the insignificant coefficients on the current exchange rate under all categories. This is a strong evidence against the Efficient Market Hypothesis (EMH) and an evidence for mispricing or a form of market inefficiency. The foreign sales ratio is found to be the most important positive determinant of exposure. However, foreign debt and other foreign non-financial liabilities make negative contribution to exposure. This indicates that measurable exposures are not easily hedged because of the missing derivative markets. Finally, competition is found to make a marginal contribution to the exposure.

**Keywords:** exchange rate exposure, firm value, currency derivatives, industry structure

*Journal of Finance and Accounting*, 2013 1 (1),
pp 1-12.

DOI: 10.12691/jfa-1-1-1

Received January 02, 2013; Revised January 21, 2013; Accepted February 28, 2013

**Copyright:**© 2013 Science and Education Publishing. All Rights Reserved.

### Cite this article:

- Erol, Turan, Ayhan Algüner, and Güray Küçükkocaoğlu. "Exchange Rate Exposure of Real Sector Firms in an Emerging Economy."
*Journal of Finance and Accounting*1.1 (2013): 1-12.

- Erol, T. , Algüner, A. , & Küçükkocaoğlu, G. (2013). Exchange Rate Exposure of Real Sector Firms in an Emerging Economy.
*Journal of Finance and Accounting*,*1*(1), 1-12.

- Erol, Turan, Ayhan Algüner, and Güray Küçükkocaoğlu. "Exchange Rate Exposure of Real Sector Firms in an Emerging Economy."
*Journal of Finance and Accounting*1, no. 1 (2013): 1-12.

Import into BibTeX | Import into EndNote | Import into RefMan | Import into RefWorks |

### 1. Introduction

Exchange rate exposure is broadly defined as the sensitivity of firm value to exchange rate changes^{1}. The exposure occurs as the unexpected exchange rate changes alter the firm’s market value by changing the expected cash flows in its home currency. Changes in the expected cash flows can in turn be traced back to two basic sources: those from the value of net monetary assets held in foreign currency and those from the value of real assets located both in domestic economy or abroad. The former class includes not only the pure foreign monetary assets with fixed nominal returns (e.g., bonds, shares, etc.) but also nominal contracts fixed in foreign currency (e.g., receivables and payables)^{1}. The exposures related to this class of assets are respectively referred to as translation and transaction exposures ^{[4]}. The translation and transaction exposures may actively be managed through various covering instruments known as hedging ^{[5]} and ^{[6]}. The latter class includes the real assets of all types of firms from purely local producers (e.g., utilities) to import competing and to sole exporting producers (e.g., multinationals). All types of firms are potentially exposed because their real assets are influenced by the resulting changes in fundamentals such as (domestic and foreign) demand, cost of imported inputs and market competition. The exposure related to the changes in these variables is referred to as economic exposure, and it may not be covered through hedging activities ^{[7]} and ^{[8]}.

Another peculiarity of the exposure of firms in emerging economies derive from the fact that these firms are usually price-takers in international trade and have little power to pass through the changes in exchange rates to foreign buyers. That is, the limited role of these firms in international markets and limited pass-through capacity may amplify their exposure. Firms facing a high foreign competition (and demand elasticity) in local markets are thus expected to have a high exposure ^{[8]} and ^{[13]}. Price-taking position does not however mean that domestic currency prices of goods sold in domestic and foreign markets are the same. In contrast, a large differential between the two prices can arise depending on the direction of deviations from purchasing power parity (PPP) and intensity of competitions in foreign versus domestic markets. For instance, in times of real appreciations exporters can be more aggressive and eager for price cuts in foreign markets. Similarly, exporter can compensate revenues losses in foreign markets by increasing margins in domestic markets whereby competition is relatively softer (due to protection or market structure). More important, volatile real exchange rates (owing to volatile inflation and/or nominal exchange rates) in emerging countries like Turkey may themselves impinge on exposure by changing the present and future cash flows in domestic currency ^{[12]}.

The empirical analysis is based on a self-constructed data set that consists of 52 Turkish manufacturing firms from the three major industries of textile, machinery and food. Categorical differences based on criteria such as foreign market involvement and industry competition are also investigated. The data set is in quarterly (highest disclosure) frequency and covers the 1998-2001 period only for which the relevant footnote information is available**.**

The exposure estimates based on a two-factor capital asset pricing model yields basically the followings. First, the exposure observed is not contemporaneous but lagged and positive in real depreciation. Second, the foreign sales ratio is the largest positive determinant and foreign currency debt is the largest negative determinant of exposure. Foreign non-financial liabilities ratio is the second largest negative determinant. The competitiveness and informal hedging make statistically marginal contribution to exposure.

### 2. A Selective Review

Shapiro** **presents the first formal analysis wherein foreign exchange exposure is determined as a function of three factors: the degrees of export sales, domestic competition and substitution between domestic and imported inputs ^{[14]}. Ongoing research in this area adds the type of competition ^{[15]} and ^{[8]} and operational and financial hedging activities ^{[16]}, ^{[17]} and ^{[5]} as the determinants of the exposure. To stress role of the degrees of allocation of production and financial hedging the term ‘net’ exposure is adhered to ^{[18]} and ^{[19]}.

Most of the previous exposure models studying the effect of exchange rate shocks on firm value are based on the assumption of monopolistic competition. The basic implications in these models are that the net foreign revenue is the primary determinant of the exposure and that the elasticity of the firm’s product is irrelevant. The monopolistic firm is also expected to display a high ability to pass cost increases through to customers and thus a small and probably undetectable exposure^{2}. Moreover, given the export ratio, the exposure of monopolistic firm can easily be predicted and managed through financial instruments.

The exposure models that are based on a more competitive oligopolistic setting find the elasticity as a second important determinant of the exposure^{3} ^{[4]} and ^{[7]}. The elasticity in turn is determined by the substitution between the products by the domestic (exporting) firm and foreign (importing) firm. As a result, oligopolistic firms are expected to display low pass-through and thus greater exposure than monopolistic firms.

The exposure is expected to be a negative function of the costs denominated in foreign currency because the latter is a part of the net foreign currency revenue. The exposure of an exporting firm deceases with its ratio of foreign costs to revenue, given the standard assumption that foreign costs are smaller than foreign revenues.

A review of the theory thus identifies a set of real and financial operations as the potential determinants of exposure: foreign sales, competitive structure, distribution of costs and production, and foreign currency positions (both financial and non-financial) and hedging possibilities ^{[21, 22, 23, 24]}. Firms from every category, including large multinationals, small exporters and import competitors, can be exposed as their expected cash flows and therefore values are altered through any of the determining factors.

Foreign currency debt is generally considered as a natural hedge and firms are expected to use this financing parallel with their foreign market involvement and thus to reduce the currency risk ^{[25]}. However, owing to frequent cycles of overvaluations and currency crises whereby foreign financing is initially motivated, foreign debt may be increasing the foreign exchange risk rather than serving as a natural hedge in emerging markets. Apriori, exchange rate exposure is expected to be positively related to the foreign sale ratio and available hedging instruments, negatively related to the elasticity and transaction position, but ambiguously related to foreign currency debt.

Some notable findings from previous empirical research, which exclusively focus on advanced countries, are as follows. First, the lagged exposure can be stronger than the contemporaneous exposure ^{[3]} and ^{[26]} and exposure is more detectable in the longer-run data ^{[27]} and ^{[28]}. Second, the foreign sales ratio is a common determinant of exposure when it is found to be significant ^{[29]} but firms with no material foreign assets, revenues or debt may well be exposed ^{[2]}. Third, intra-industry competition is an important determinant ^{[18]}, ^{[30]} and ^{[29]} and financial and operational hedging reduces exposure ^{[16]} and ^{[31]}.

### 3. Statistical Analysis

**3.1. Construction of Data Set**

In parallel with the current trend, we constructed the data set at quarterly horizon that is believed to be sufficient if not optimal. Another compelling reason that motivated the construction of the data set in quarterly frequency has to do with the estimation methodology adopted. As will be explained, there are two alternative procedures to ultimately estimate the determinants of exposures. These are the one-step (direct) and two-step (indirect) procedures that have quite different statistical properties. Although widely adopted, the two-step procedure, which runs a regression of exposures betas from the first stage estimation on a set of determinants, is usually problematic because of the potential correlations across betas and thus across the error terms. The alternative direct procedure does not involve this statistical problem because it directly incorporates the set of determinants as interaction terms in the first stage of estimation. This off course requires a data set that has sufficient observations but a unique frequency. To implement this second procedure we have chosen the quarterly return horizon that is the highest disclosure frequency for balance sheets.

The choice of the proxy for market return in empirical tests is another concern. Previous empirical works overwhelmingly use country specific market returns although a global portfolio return might be more appropriate in a world of highly integrated capital markets. Another concern is the weighting of the specific proxy chosen for market return, be it value-weighted or equally-weighted. The value-weighting might underestimate the exposure coefficient as it removes the negative cash flow effects of larger firms that dominate the market portfolio ^{[32]}. Smaller local or import-competing firms might easily avoid the negative cash flows. However, ^{[33]} present evidence based on an international data set that conditioning on the value-weight versus the equally-weighted market portfolio has no discernable effect on exposure coefficient. We principally use the largest published market index, namely, the capitalization-weighted national 100 firms index.

A final point discussion relates to the choice of the exchange rate index. A variety of indexes from the end of month bilateral to the average trade-weighted indexes are used in empirical exposure studies. We do not embark on these methodological discussions, but take a practical approach in deciding on the type of indexes to be used. We prefer to basically use the published capitalization-weighted share prices and end of month bilateral (US dollar) exchange rate indexes. This preference is based on the cross-country evidence in ^{[26]}, ^{[28]} and ^{[33]} that the exposure estimates are not seriously affected by the weighting of market portfolio and that the trade-weighted exchange rate index leads to underestimation. However, we will also experiment with the equally-weighted share prices and trade-weighted exchange rate indexes to see if the results are sensitive to choice of alternative indexes.

We now present some relevant statistics for each industry and for the manufacturing panel. The textile industry has the highest foreign sales and foreign debt ratios, followed by the food and machinery industries. Similarly, the ratios of the contract in foreign currencies (receivables and payables) are the highest in the textile industry, but now followed by the machinery and food industries. Moreover, the textile industry is the most suffered one from the drastic decline in the real stock returns. The real returns on the exchange rate stayed positive and much more stable (with the standard deviations of 6 % versus 32 %) over the sample period. The final observation from Table 1 is about the preliminary exposure statistics. Parallel with the other figures of foreign involvements, the textile industry has again the highest coefficient.

**3.2. Estimation**

A prior decision is whether the variables should be expressed in nominal or real terms. In the high inflationary environment of Turkey, the inflation is volatile and it needs to be treated as random. This requires using the real returns on exchange rates and asset prices in regressions, and thus measuring the exposure in real terms. The wholesale price index is used to convert the nominal returns into real returns.

The basic estimating equation is a two-factor CAPM modeled as

(1) |

Three different coefficients of exposure from equation (1) are obtained by including different specifications of the exchange rate variable. First, in line with the literature, contemporaneous and lagged exposure coefficients are separately estimated. Second, an average exposure coefficient based on a weighted average of exchange rate variable is estimated. The average exposure coefficient, besides some statistical advantages over alternative dynamic specifications as explained below, combines both the contemporaneous and lagged effects and gives a net measure of exposure. Moreover, each of the three coefficients of exposure is obtained for four different categories of samples. The distinguishing criteria are foreign sales, competition, industry classification, which are widely discussed in the theoretical and empirical literature. Table 2 presents the estimates of three different parameters of exposure under four categories of samples from A to D, as well as the numbers of significantly exposed firms in each categories.

A major result from the estimations of the two-factor model in Table 2 is the insignificant coefficients on the current exchange rate under all categories. This is a strong evidence against the Efficient Market Hypothesis (EMH) and an evidence for mispricing or a form of market inefficiency. The market inefficiency is double checked by the strong coefficients on the lagged exchange rate under all categories. All lagged exposure coefficients are significant (mostly at 1 % level) and have positive sign, a positive but lagged exposure. That is, the exchange rate risk is priced (requiring a positive premium) with some delay in stock prices.

Two categories of firms, those of the textile and high foreign sales firms, have the largest exposure based on the lagged or average coefficients, while those of the low foreign sales and food industry categories have the smallest exposure coefficients. Moreover, the largest intra-category divergences are according to the foreign sale and industry division. That is, the most divergent exposure is between the high and low foreign sales categories, followed by the one between the textile and food industries. A division according to the competitiveness does not lead to a notable divergence in exposure coefficients. This may be an indication that domestic industry setup of these internationally price-taking firms has a limited influence on their exposures.

Finally, parallel with the signs of panel coefficients, the significant individual exposure coefficients are predominantly positive and account about one third of the total, compared to the individually significant negative coefficients that account only one tenth of the total.

The second stage estimations for the determinants of exposure are based on the following extended model that involves interactions,

(2) |

where ^{j}*X*_{it} the *j*th determinant fir firm *i*, and *j*= 1,...,5 and *i*= 1,…,52. *R*_{m} and *R*_{s} are respectively the real rate of returns on the market portfolio and the exchange rate.

(3) |

where β_{2ji }is the coefficient on the *j*th determinant (^{j}*X*_{i}) interacted with the exchange rate for firm *i*.

The explicit forms of these five interactive terms are as follows:

for* j =1*, the interaction term is *S*_{f}**R*_{s}_{,} where *S*_{f}* *is the ratio of foreign to total sales;

for *j = 2*, the interaction term is *(-1/OPCM)*R*_{s}_{,} where *OPCM* is the operational price-cost margin;

for *j = 3*, the interaction term is *D*_{f*}*R*_{s}, where *D*_{f} is the ratio net foreign currency debt to total assets;

for *f = 4*, the interaction term is *L*_{f*}*R*_{s}, where *L*_{f}* *is the ratio of net foreign non-financial liabilities (payables minus receivables) to total sales;

for *j = 5*, the interaction term is *R*_{rf}**R*_{s}, where R_{rf} is the real return on repos and reverse repos in government papers up to two weeks of maturity.

As in the first stage estimations, equation (3) is estimated through the fixed-effects GLS method based on the cross-sectional weights. The estimation results, whose details are relegated to Appendix 2, are given in Table 3. Eight alternative parameters for each interaction term are estimated based on the eight categorical division of the sample, a critical issue in the exposure literature. The panel of 52 firms are divided into seven subsamples of: high versus low foreign sales firms, more versus less competitive firms, and three subsamples of industrial classification (textile, machinery and food). Finally, each interaction term is parameterized for the complete sample of 52 firms.

As expected, the average coefficients are extremely stable as the offsetting effects of different periods (from t to t-3) are netted out, representing a statistically reliable average effect. A much clearer picture emerges from the average coefficients. Firstly, the interaction terms on foreign sales have the largest and positive coefficients. The single category in which the foreign sales and exchange rate interaction term has no significant coefficient is the less competitive category. Secondly, the second largest but negative interaction coefficients are related to the foreign currency debt. The negative exposure from the (net) foreign currency debt is confirmed for all categories except the textile firms. Similarly, all significant interaction coefficients on foreign non-financial liabilities are negative, reducing the firm value as the real exchange rate depreciates. Thirdly, firms with high foreign sales ratio gain enormously from informal hedging through government papers as they have an exceptionally large and positive interaction term. The firms in others categories have no significant interaction coefficient, while the complete sample has a positive coefficient at only a questionable level of significance. Finally, the food, more competitive and high foreign sales categories are the ones that benefit from competition and increase their values in times of real depreciation, while less competitive firms lose from the real depreciation.

Summarizing, the largest significant interaction coefficient is a positive one and is related to the foreign sales ratio. The second largest interaction coefficient is a negative one and is related to the foreign currency debt ratio, followed by the foreign non-financial liabilities ratio. The firms most active in foreign markets are the only beneficiaries of informal hedging through domestic risk-free assets. High competition makes moderate positive contribution to exposure in most categories but it is destructive for the less competitive firms.

Besides the separate estimates for each category just discussed (seven subsamples and complete sample), we have also estimated the model for the complete panel with the competition and foreign sales dummies, two critical determinants of exposure. That is, rather than breaking into more (and less) competitive and high (and low) foreign sales categories, we assigned dummies for these categories and run the regression for the entire sample. The relevant equation is now

(4) |

*D*_{c} and* D*_{f} are respectively the dummies of competition and foreign sales.

### 4. Conclusion

All types of firms from pure local producers to pure exporters can be exposed to real exchange rate changes as the value of their real assets are influenced by the ensuing changes in demand, cost and other fundamentals. This type of exposure, known as the economic exposure, is not easily measured and covered even in financially mature economies. Moreover, measurable exposures from fixed foreign currency contracts, known as the translation and transaction exposures, may not easily be hedged in developing countries because the markets for currency derivatives are generally not functioning.

Firms from emerging economies may thus face a higher degree of exposure compared to those in advanced economies. For instance, many of the formal foreign exchange derivatives are not operational in Turkey, whose currency is not traded and not a part of these hedging activities, except the few special contracts between the central bank and commercial banks and between large holdings and their banks. Similarly, foreign currency debt, generally seen as a natural hedge, may be value-reducing if short positions are soon to be considered unsustainable after a period of optimism that and over-borrowing. The price-taking position and thus limited pass-through capacity can be another factor amplify the exposure of emerging market firms. Finally, volatile real exchange rates (owing to volatile inflation and/or nominal exchange rates) in emerging countries like Turkey may themselves impinge on exposure by changing the present and future cash flows in domestic currency.

The present empirical analysis is based on a sample of firms with varying foreign market involvement measured by their foreign sales ratios, which on average ranges between 2 and 91 %. That is, the sample includes both primarily exporting firms with the foreign sales ratio close to hundred percent, primarily import competing firms with the foreign sales ratio close to zero, and firms in between. A categorization based on the foreign sales ratio is critical because it allows contrasting exposures across different categories.

Two unusually additional determinants of exposure are considered in estimations. These are the net foreign currency payables (net transaction position) and an informal instrument of foreign currency hedging. Short transaction positions in foreign currency may not be fully covered because formal foreign currency derivatives for hedging are not available in Turkey as in many emerging markets. This fact leaves no choice but find informal instruments of hedging. Potential instruments are the foreign currency holdings, foreign currency debt and other indirect (domestic currency) instruments such governments bonds and papers. Indirect instruments provide coverage for foreign currency risks through the risk-free (and mostly higher) real interest rate returns that can be easily converted into foreign currency. Investments in government papers (for short-term hedging) and bonds (for longer-term hedging) are the indirect instruments considered.

A major result from the estimations of the two-factor model is the insignificant coefficients on the current exchange rate under all categories. This is a strong evidence against the Efficient Market Hypothesis (EMH) and an evidence for mispricing or a form of market inefficiency. The market inefficiency is double checked by the strong coefficients on the lagged exchange rate under all categories. The exchange rate risk is priced but with some delay in stock prices.

The second striking result follows from the average exposure estimates. Two categories of firms, those of the textile and high foreign sales firms, are the most (positively) exposed, while those of the low foreign sales and food industry categories are the least exposed. Moreover, the largest intra-category divergences are according to the foreign sale and industry division. That is, the most divergent exposure is between the high and low foreign sales categories, followed by the one between the textile and food industries. A division according to the competitiveness does not lead to a notable divergence in exposure.

The third set of results concern the determinants of exposure. The largest significant determinant of exposure is a positive one and is related to the foreign sales. The second largest determinant is a negative one and is related to the foreign currency debt, followed by the foreign non-financial liabilities. The firms most active in foreign markets are the only beneficiaries of informal hedging through domestic risk-free assets. High competition makes moderate positive contribution to exposure in most categories but it is destructive for the less competitive firms.

### References

### Data Appendix

**A.1. Detailed Estimations**

**A.1.1. First stage Estimations**

The first set of first stage estimates involves the contemporaneous exposure coefficients based on equation (1), which is specified in the fixed-effects form as

(1') |

and is estimated by a GLS based on the cross-section weights.

The detailed test statistics along with the exposure coefficients for different categories or sub-samples are presented in Table 2A.

The set of first stage estimations involve the lagged exposure coefficients from equation (1), which is specified in the fixed-effects form as

(1'') |

where, k = 1, ,4, is the first significant lag encountered on the exchange rate variable. That is, only the lagged exposure coefficients but not the current ones take place. The detailed test statistics along with the exposure coefficients for different categories or sub-samples are presented in Table 2B. A caution about the lagged coefficients is need: only the first significant lag is included and this is most time the third lag. However, some other lags, which are insignificant when individually included, tend to be significant when included together with other lags (especially with the third lag that is significant in all cases). Moreover, these other lags sometimes take opposite (negative) signs. All these imply that the exposure coefficients based on the lagged exchange rate variable in Table 2B cannot be taken as a final measure of exposure (average or total).

The next set of estimates of exposure is based on a weighted exchange rate series (arithmetically weighted from periods t to t-3). Alternative average series based on higher dimension (from periods t to t-4) and geometric weighting are also tried, but their results are not presented as the differences are small. As in the previous two estimations the fixed-effects form equation (1)

(1''') |

where ^{a}*R*_{s} is the arithmetically weighted average exchange rate series.

The last set of first stage estimates are the cross-section specific exposure coefficients based on the average exchange rate series (the last two columns in Table 2). Given the large number of coefficients involved (number of exposure coefficients equals the number of cross-section units for each category), these estimates, available upon request, are not presented.

**A.1.2 Second Stage Estimations**

The second stage of estimations that involve five interaction terms to parameterize the determinants of exposure are based on,

(3) |

or, more explicitly,

(3') |

where β_{2j,}* j = 1, 2, 3, 4, 5* are five parameters on the interaction terms (between the real exchange rate change and the determinants of exposure, ^{j}*X R*_{s})., and

^{1}*X*: the ratio of foreign to total sales,

^{2}*X*: *(-1/OPCM), OPCM* is the operational price-cost margin,

^{3}*X*: the ratio of net foreign currency debt (debt minus assets) to total assets,

^{4}*X*: the ratio of net foreign currency payables (payables minus receivables) to total sales,

^{5}*X*: real (quarterly) return on repos and reverse repos in government papers up to fourteen days of maturity.

As in the first stage estimations, three different sets of (interactive) exposure coefficients, namely, the contemporaneous, lagged and average, are obtained. However, only two sets of coefficients, contemporaneous and average, are presented to simplify the exposition. The difference between the two sets of estimates, like the first stage estimates, lies in the definitions of the exchange rate variables entering the interaction term ^{j}*X R*_{s}. That is, the contemporaneous estimates are based on the current exchange rate *R*_{s} while the average estimates are based on the weighted exchange rate ^{a}*R*_{s}. The statistical details of the second stage estimations presented in Table 3 are given in Tables 3A (for contemporaneous coefficients) and 3B (for average coefficients).

### Notes

^{1}However, cross currency hedging (i.e., between two foreign currencies) is always possible but this provides no coverage for domestic firms concerned.^{2}This amounts to assuming a unique profit margin in domestic and foreign markets.^{3}More precisely, most of these instruments are formally available but the markets are not functioning because the Lira is not accepted as an international currency, that is, the Lira side of the market is missing.^{4}Only recently, limited amount of government bonds denominated in foreign currency (foreign currency indexed bonds) are available but these are mostly held by banks rather than the nonfinancial sector.^{5}The data set, constructed from the scratch since databases such as the Compustat in the USA or Exstat in the UK are unavailable, includes both the market and balance sheets variables.^{6}Note that the exposure is expected to be independent of the time horizon in theoretical models that assume market efficiency and complete information. That is, in all horizons, the impact of (observed and expected) exchange rate changes on the current and future cash flows are incorporated in the current stock prices. See the above papers for more discussions and references.^{7}The random-effects approach, subjecting the residuals from the fixed-effects to a randomness check through some arbitrary weighting, may sometimes be very restrictive and therefore not preferred. See e.g. [34].^{8}The tests, not presented but can be obtained upon request, are based on the statistic,,whereβ_{fe}_{}andβ_{fe}are the vectors of respectively the fixed and random-effects parameters. The null is the random-effects model. Under the null hypothesis, both the random-effects and fixed-effects estimators are consistent and random-effects model is efficient. Therefore, a large Wald measure withχ^{2}distribution weighs against the null (random-effects) in favor of the alternative (fixed-effects) model. We calculated the Hausman test statistics for the complete samples in the first and second stage estimations (option D in Table 2 and the all-firms category in Table 3).^{9}It is interesting to note that the lagged exposure coefficient, as well as the average exposure coefficient, is at the margin of significance in the food industry whereby more than the half of the firms have low foreign sales ratio.^{10}The results are qualitatively the same if the constructed series is extended from period t to t-4 or if the weights are altered.^{11}As noted before, an insignificant average coefficient, against a significant individual (third) lag, is to be explained by the presence of other significant but offsetting lags that dominates the former.^{12}Some other potential determinants such as foreign operations, domestic versus foreign currency production and costs are not considered because of the data restriction.^{13}As discussed before, we prefer the direct estimation method for parameterization of determinants. The connection with the alternative two-stage method is easily be established through the relation. See e.g. [3] for further discussion^{14}Of course, the lags that yield significant coefficients differ, and sometimes, there are multiple significant lags.^{15}This restriction on the exchange rate series and thus on the interaction terms eliminates the potential multicollinearity problem when otherwise multiple lags of each interaction variables are used.^{16}The calculated series is an arithmetically weighted average of the periods from t to t-3. However, the results are qualitatively same when it is extended to period t-4 or when a geometric weighting is adopted.^{17}Firms with margins between 0.07 and 0.29 (or elasticity between 14.4 and 3.44) are classified as more competitive and firms with profit margins between 0.30 and 0.43 (or elasticity between 3.33 and 2.32) are classified as less competitive.^{18}Foreign sales ratios below 0.30 are classified as low and high otherwise.

[1] | Adler, M., Dumas, B., 1984. Exposure to currency risk: definition and measurement. Financial Management 13, 41-50. | ||

In article | CrossRef | ||

[2] | Booth, L., Rotenberg, W., 1990. Assessing foreign exchange exposure: theory and application using Canadian firms. Journal of International Financial Management and Accounting 2, 1-22. | ||

In article | CrossRef | ||

[3] | Jorion, P., 1990. The exchange rate exposure of US multinationals. Journal of Business 63, 331-341. | ||

In article | CrossRef | ||

[4] | Marshall, A.P., 2000. Foreign exchange risk management in UK, USA and Asia Pacific multinational companies. Journal of Multinational Financial Management 10, 185-211. | ||

In article | CrossRef | ||

[5] | Bartram, S.M., 2004. Linear and nonlinear foreign exchange rate exposures of German nonfinancial corporations. Journal of International Money and Finance 23, 673-699. | ||

In article | CrossRef | ||

[6] | Crabb, P.R., 2002. Multinational corporations and hedging exchange rate exposure. International Review of Economics and Finance 11, 299-314. | ||

In article | CrossRef | ||

[7] | Lee, W.Y., Solt, M.E., 2001. Economic exposure and hysteresis Evidence from German, Japanese, and U.S. stock returns. Global Finance Journal 12, 217-235. | ||

In article | CrossRef | ||

[8] | Krishnamoorthy, A., 2001. Industrial structure and the exchange-rate exposure of industry portfolio returns. Global Finance Journal 12, 285-297. | ||

In article | CrossRef | ||

[9] | Elliott, W.B., Huffman, S.P., Makar S.D., 2003. Foreign-denominated debt and foreign currency derivatives: complements or substitutes in hedging foreign currency risk? Journal of Multinational Financial Management 13, 123-139. | ||

In article | CrossRef | ||

[10] | Lin, C., 2011. Exchange rate exposure in the Asian emerging markets. J. of Multi. Fin. Manag. 21, 224-238. | ||

In article | CrossRef | ||

[11] | Bartram, S.M., Bodnar, G.M., 2008. Crossing the lines: The conditional relation between exchange rate exposure and stock returns in emerging and developed markets. Journal of International Money and Finance 31, 766-792. | ||

In article | CrossRef | ||

[12] | Timothy, K. Chue, T.K., Cook, D., 2008. Emerging market exchange rate exposure. Journal of Banking & Finance 32, 1349-1362. | ||

In article | CrossRef | ||

[13] | Doidge, C., Griffin, J., Williamson R., 2006. Measuring the economic importance of Exchange rate exposure. Journal of Empirical Finance 13, 550-576. | ||

In article | CrossRef | ||

[14] | Shapiro, A.C., 1975. Exchange rate changes, inflation, and the value of the multinational corporation. Journal of Finance 30, 485-502. | ||

In article | CrossRef | ||

[15] | Marston, R.C., 2001. The effects of industry structure on economic exposure. Journal of International Money and Finance 20, 149-164. | ||

In article | CrossRef | ||

[16] | Allayannis, G., Ofek, E., 2001. Exchange rate exposure, hedging, and the use of foreign currency derivatives. Journal of International Money and Finance 20, 273-296. | ||

In article | CrossRef | ||

[17] | Allayannis, G., Brown, G.W., Klapper, L.F., 2001. Exchange rate risk management: evidence from East Asia. Unpublished. | ||

In article | |||

[18] | Williamson, R., 2001. Exchange rate exposure and competition: evidence from the automotive industry. Journal of Financial Economics 59, 441-475. | ||

In article | CrossRef | ||

[19] | Chow, E.H., Chen, H., 1998. The determinants of foreign exchange rate exposure: Evidence on Japanese firms. Pacific-Basin Finance Journal 6, 153-174. | ||

In article | CrossRef | ||

[20] | Bodnar, G.M., Dumas, B., Marston, R.C., 2000. Pass-through and exposure. Unpublished. | ||

In article | |||

[21] | Dominguez, K.M.E, Tesar, L.L., 2006. Exchange rate exposure. Journal of International Economics 68, 188-218. | ||

In article | CrossRef | ||

[22] | Bartram, S.M., 2004. Linear and nonlinear foreign exchange rate exposures of German nonfinancial corporations. Journal of International Money and Finance 23, 673-699. | ||

In article | CrossRef | ||

[23] | Koutmos, G., Martin, A.D., 2003. Asymmetric exchange rate exposure: theory and Evidence. Journal of International Money and Finance 22, 365-383. | ||

In article | CrossRef | ||

[24] | Huffman, S.P., Makar, S.D., Beyer, S.B., 2010. A three-factor model investigation of foreign exchange-rate exposure. Global Finance Journal 21, 1-12. | ||

In article | CrossRef | ||

[25] | Goswami, G., Shrikhande, M.M., 2001. Economic exposure and debt financing choice. Journal of Multinational Management 11, 39-58. | ||

In article | CrossRef | ||

[26] | Bartov, E., Bodnar, G.M., 1994. Firm valuation, earnings expectations, and the exchange- rate exposure effect. Journal of Finance 49, 1755-1785. | ||

In article | CrossRef | ||

[27] | Chow, E.H., Lee, W.Y., Solt, M.E., 1997. The economic exposure of US multinationals. Journal of Finnacial Research 20, 191-210. | ||

In article | |||

[28] | Dominguez, M.E.K., Tesar, L.L., 2001b, Exchange rate exposure, NBER Working Paper 8453. | ||

In article | |||

[29] | Bodnar, G.M., Gentry, W., 1993. Exchange rate exposure and industry characteristics: evidence from Canada, Japan, and the USA. Journal of International Money and Finance 12, 29-45. | ||

In article | CrossRef | ||

[30] | Allayannis, G., Ihrig, J., 2000. Exposure and markups. Unpublished Manuscript. | ||

In article | |||

[31] | Gao, T., 2000. Exchnage rate movements and the profitability of US multinationals. Journal of International Money and Finance 19, 117-134. | ||

In article | CrossRef | ||

[32] | Bodnar, G.M., Wong, M.H. F., 2000. Estimating exchange rate exposures: some weighty issues. NBER Working Paper 7497. | ||

In article | |||

[33] | Dominguez, M.E.K., Tesar, L. L., 2001a. A re-examination of exchange rate exposure. NBER Working Paper 8128. | ||

In article | |||

[34] | Hsiao, C., 1986. Analysis of panel data. Cambridge University Press. | ||

In article | |||