The Use of Currency Derivatives by Brazilian Companies : An Empirical Investigation

This paper studies the use of foreign currency derivatives f or a sample of non-financial Brazilian companies from 1996 to 2004. The paper verifies tha t some of the hypotheses presented by the optimal hedging literature are able to expl ain the Brazilian companies’ decision to use currency derivatives and their decision on t he amount to use. Moreover, the paper shows that the macroeconomic environment and country -specific factors not analyzed in previous empirical work also play a role in determining th e companies’ risk management practices and that the use of currency derivatives has an imp act on the companies’ decision regarding their capital structure and the currency composi tion of their debt.


Introduction
The corporate finance literature has not reached a consensus on the reason companies resort to hedging activities.In a survey about the empirical studies on the determinants of companies' hedging activities, Judge (2003) concludes that the evidence is mixed with respect to the different theories discussed in the literature.
A flaw of the existent literature is that most previous papers analyze the companies' hedging activities in developed countries, especially in the U.S. Yet, the economic and political instability that characterizes emerging markets creates the exact kind of environment where risk management activities would be more useful for the companies allowing them to reduce the effects of this instability on their cash flow. 1 In addition, because financial market imperfections are more pronounced in these countries there would be more opportunity for gains of hedging.
Although extensive for developed countries, the empirical literature that analyzes the risk management practices -especially the use of derivatives by companies in emerging markets -remains scarce. 2 This paper explores this oversight by analyzing the use of foreign currency derivatives for a sample of non-financial Brazilian companies from 1996 to 2004.The fact that during the period in question Brazil suffered two main exchange rate crises3 makes the Brazilian experience an appropriate case study for analyzing the behavior of companies when subject to the high volatility of macroeconomic fundamentals. 4he paper is also related to the growing literature in international finance that analyzes the role of the interaction between the Central Banks' policy decisions and the companies' financial policies.Our dataset allows us to assess the companies' hedging activities under two different exchange rate regimes: a (quasi-)fixed exchange rate dmregime from 1996 to January 1999 and a floating exchange rate dmregime afterwards.We can therefore analyze whether the macroeconomic policy adopted by the government and the Central Bank influences the behavior of the companies.
The main results can be summarized as follows: the decision to use currency derivatives is determined by the costs of hedging, i.e., larger firms are more likely to use currency derivatives.Companies decide to use currency derivatives in order to reduce their foreign exchange exposure; exporters and companies that hold foreign currency denominated debt are more prone to use currency derivatives.Companies whose probability of incurring costs of financial distress is higher are also more likely to use currency derivatives.In addition, the results confirm that there is a relationship between the use of derivatives and the macroeconomic environment; companies have increased their hedging activities after the adoption of the floating exchange rate regime.
The results confirm that different factors drive the decision to use currency derivatives and to extend that use.The extension of the use of currency derivatives, i.e., the decision about how much to hedge, is determined positively by the company size.Firms with higher levels of foreign currency denominated debt to total debt also use currency derivatives more extensively, corroborating the idea that firms perceive foreign currency denominated debt as a risk to their balance sheets.
Unlike previous papers, this study found a negative relationship between the extension of hedging and the ratio of foreign sales to total sales and companies' foreign operations suggesting that country-specific factors have an impact on the companies' use of currency derivatives.In the case of emerging markets, companies see their foreign currency revenue as a "natural" hedge to offset negative shocks on the liability side of their balance sheets.Such is the case of Brazil, where the foreign currency denominated debt is viewed as the main risk factor for the Brazilian companies due to the country's vulnerability to negative international shocks associated to devaluations of its own currency.
Finally, the results indicate that the firms' decision on hedge and indebtness are taken simultaneously and that the use of currency derivatives allows the firms to increase their debt capacity, leading to higher levels of leverage and debt denominated in foreign currency.
The paper proceeds as follows.In section 2, it gives a brief overview of the literature that analyzes the determinants of the companies' hedging activities.Section 3 shows the data used in the analysis.Section 4 reports the main results.Section 5 concludes.

Determinants of Hedging
The relationship between corporate financial policies and the value of the firm was established by Modigliani and Miller (1958).According to this well-known theorem, in a world with no taxes, no transaction costs and a fixed investment policy, the company value would not be affected by its decision to hedge.There will be no value in hedging since investors could hedge their own portfolio by taking action themselves in financial markets.Therefore, in order to analyze the main determinants of the companies' decision to hedge, one should depart from the assumptions of Modigliani-Miller theorem by considering the effects of tax liabilities, transaction costs, or investment decisions on the companies' financial policies.

Theoretical literature
The theoretical literature gives some guidance about the determinants of the companies' hedging activities.Smith and Stulz (1985) assert that given a convex corporate tax system, by reducing the variability of the cash flow, hedging would reduce the companies' tax liability, leading to an increase in the post-tax value of the firms.Therefore, it would be optimal for firms to hedge.They also show that hedging would reduce the expected bankruptcy costs, increasing the expected payoff to the firms' claimholders; thus, hedging would raise the companies' value by reducing the variability of their future cash flow.
The managerial risk aversion might also be a determinant of the companies' hedging activities.Smith and Stulz (1985) analyze the possibility that risk-averse managers prefer to hedge.The study explains this preference by suggesting that reducing the variability of a firm's cash flow would increase managers' expected utility, since it is dependent on the firm's payoffs.DeMarzo and Duffie (1995) show that hedging would allow the market to draw better inferences on managers' ability; therefore, managers would like to signal their ability by hedging.Froot et al. (1993) show that if capital markets are imperfect, hedging may increase the firms' value by insuring they have sufficient internal funds.A variable cash flow would lead to more variability either in the amount raised externally or in the investment.Hence, firms with higher growth opportunities would prefer to hedge in order to mitigate their underinvestment problem.
Firms might hedge to reduce their exposure to fluctuations of the exchange rate.Firms whose cash flow is more sensitive to exchange rate fluctuations would derive great benefits from hedging; this would be the case of exporters, firms with foreign subsidiaries or firms that hold foreign currency denominated debt.Foreign currency denominated debt per se might be used as a manner to hedge exchange rate fluctuations.Firms with revenue in foreign currency might issue debt denominated in foreign currency in order to avoid mismatches in their balance sheets.This alternative seems unreasonable in emerging markets, where foreign currency debt is the main concern with respect to exchange rate exposure. 5he international finance literature indicates the possible existence of a relationship between the companies' hedging activities and the macroeconomic policy, especially the monetary policy.Burnside et al. (2001) build a model in which implicit guarantees given by the government induce firms and financial intermediaries to borrow from abroad, but do not completely hedge against exchange rate risk.According to these authors, a bank has no incentive to hedge since the expected value of this strategy is null.In the event of no devaluation, buying forward to hedge would generate losses to the bank, and, in the event of devaluation, the government would seize profits derived from such hedging activities.Moreover, they show that apart from the government guarantees, it would be optimal for the firms to hedge their exchange rate risk completely. 6chneider and Tornell ( 2004) emphasize the role of government guarantees and asymmetries in sectoral behavior.They highlight the dichotomy between tradables and non-tradables.In their model, given the presence of bailout guarantees and the inability of the non-tradable sector to make a clear commitment to the repayment of its debt, currency mismatches arise endogenously, since foreign creditors would extend credit to the non-tradable sector.This currency mismatch would lead to a self-fulfilling crisis.Again, if there were no guarantees, managers would have no incentive to create currency mismatches.In the presence of bankruptcy costs, they would prefer to hedge the exchange rate risk.These authors show that, under the fixed regime, firms in the non-tradable sector grow faster by relaxing their borrowing constraints; however, in the event of depreciation, these companies would suffer heavily from balance sheet problems; therefore, the existence of government guarantees related to the choice of a fixed exchange rate regime imposes temporal restrictions on the companies' hedging activities.The existence of government guarantees implies that, under the fixed exchange rate regime, companies would not fully internalize the risk of exchange rate fluctuations, incurring currency mismatches on their balance sheets.On the contrary, the floating exchange rate regime would encourage companies to take the exchange rate fluctuations seriously, leading them to improve their risk management activities by matching the currency composition of their assets and liabilities.

Empirical literature
There is a vast empirical literature that attempts to discriminate among different theories about determinants of hedging7 (Wysocki, 1995, Mian, 1996, Geczy et al., 1997, Graham and Rogers, 2002, Allayannis and Ofek, 2001, Carter et al., 2003).Judge (2003) summarizes the results of fifteen studies on the topic.In general, he finds little support for tax reasons and managerial risk aversion for hedging. 8In addition, almost none of the papers studied corroborate the financial distress hypothesis.Yet, the evidence with respect to capital market imperfections is mixed; half of the papers in his study confirm that there is an interaction between growth opportunities and hedging.In addition, he finds strong support for the existence of economies of scale in hedging.
He also establishes a strong relationship between foreign currency cash flow volatility and hedging.For example, Allayannis and Ofek (2001) discover from a sample of 500 U.S. companies that the choice and the extension of the companies' foreign borrowing can be explained in light of foreign exposure.These authors see this fact as evidence that American companies use foreign debt to hedge their foreign exposure.Similarly, Kedia and Mozumbar (2003) find evidence that U.S. companies use foreign debt as a way to hedge their foreign exposure.
Although extensive, none of the previous papers analyze hedging practices in emerging markets, where exchange rate crises create a natural experiment in risk management practices.Exceptions are Allayannis et al. (2003), Kim and Sung (2005) and Saito and Schiozer (forthcoming).Allayannis et al. (2003) study the use of foreign currency derivatives from a sample of East-Asian companies right before the financial crisis in 1997.Kim and Sung (2005) analyze hedging practices for a sample of Korean companies.Saito and Schiozer (forthcoming) analyze the risk management practices for a sample of Latin-American firms.9They find that firms in these countries mainly hedge foreign currency denominated debt and confirm that consistent with theory, economies of scale, financial distress costs, informational asymmetry and growth opportunities are important to the use of currency derivatives.Unlike this paper, none of the previous papers study the relationship between the use of derivatives and the macroeconomic environment.They find that firms in these countries mainly hedge foreign currency denominated debt and confirm that consistent with theory, economies of scale, financial distress costs, informational asymmetry and growth opportunities are important to the use of currency derivatives.Unlike this paper, none of the previous papers study the relationship between the use of derivatives and the macroeconomic environment.

Data
I gathered data from two main sources: Economática and the companies' annual reports.Economática gives stock market returns and accounting data for all publicly traded companies in Brazil.Data were also gathered directly from the companies' annual reports, in case some information was not available, or to confirm the quality of data.I used data from a sample of 212 Brazilian non-financial companies from 1996 to 2004, which represents more than two thirds of all publicly traded companies and more than three quarters of market capitalization..The description of all variables used throughout the text is shown in the appendix.The choice of the period 1996-2004 was due to the fact that the requirement to report the use of derivatives came into effect only after 1995. 10here is no systematic information about foreign sales.Sometimes the report comes together with gross sales, sometimes through comments made by managers to shareholders, or in the explanatory notes.Some companies name themselves exporters but they do not report the amount of sales; in this case, the companies were contacted directly through electronic mail.In the end, seven companies mentioned as exporters had to be discarded because none reported the amount of foreign sales or answered the emails.
The use of currency derivatives and foreign currency denominated debt variables are available in the annual reports under the explanatory notes.The amount of foreign debt is located under the item loans and financing.The use of currency derivatives is registered under the item financial instruments.
I use the total gross notional value of currency derivatives as my proxy for the extension of the use of currency derivatives.Graham and Rogers (2002) argue that ideally in order to identify a more precise picture of the companies' risk management practices one should compute the net position of the companies in the derivatives markets.However, as this information is only made available to a small number of firms, I opted to use the total gross notional value that is available to a larger number of companies. 11 group foreign assets as any asset the company holds that earns the variation in the nominal exchange rate plus a premium during the period.These can be Treasury bonds (NTN-E), Central Bank bonds (NBC-E), assets invested in foreign banks, and cash in foreign currency. 12igure 1 shows the evolution of nominal exchange rate R$/US$ during the period. 13From 1995 to 1998, Brazil adopted a crawling-peg exchange rate regime,14 , and suffered several speculative attacks, especially during the Asian and Russian crises.The Central Bank reacted promptly to such attacks by raising interest rates in order to maintain the regime, clearly demonstrating its commitment to the exchange rate regime even at the cost of maintaining high interest rates, increasing the public debt, and causing economic recession.This first period was characterized by a low volatility of the nominal exchange rate.After a speculative attack in January 1999, currency was allowed to float, and an inflation-target regime was adopted.After tightening monetary and fiscal policies, Brazil succeeded in stabilizing inflation and the economy quickly recovered from the crisis.In 2002, due to the possibility that a new president against current policies would be elected, a reversal of capital flows took place and the exchange rate depreciated more than 50% during the year with a consequent rise in inflation.After 2003, home currency started to appreciate because the new government opted to reinforce the orthodox macroeconomic policy and a positive external shock represented by an increase in the price of the main exported commodities struck the country.

Sample characteristics
Table 1 shows a summary of the statistics of the companies in the sample.The results in Table 1 indicate that there is a substantial variability with respect to the size of the companies in the sample.Regardless of the proxy used for the size of the companies, the sample encompasses small and large companies whose sizes vary from R$ 50.3 million to R$142,000 million when the book value of total assets is used as proxy for size, and from R$1.44 million to R$150,000 million when the total sales is adopted as proxy for size.Therefore, the results are not biased by the size of the companies in the sample.Another important result presented in Table 1 is that most firms in the sample are exposed to fluctuations of the exchange rate.Nine hundred and thirty-one companies out of 1664 (55.9% of the sample) are classified as exporters.On average, these companies have 26.9% of their revenues expressed in foreign currency.In addition, results in Table 1 show that 82.1% of the companies in the sample hold debt denominated in foreign currency.
Table 2 reports that the use of currency derivatives and foreign assets varies considerably from 1996 to 2004 and shows that the number of users of derivatives and those that hold foreign assets increased steadily from 1996 to 2002.These facts contradict Eichengreen and Hausmann (1999) when they suggest the possibility that an increase in the volatility of the exchange rate would lead to higher costs of hedging; therefore, one could observe less rather than more hedging when exchange rates are less stable.In addition, Table 2 shows that after the appreciation of home currency in the period 2003-2004, there was a reduction in the use of currency derivatives and foreign assets.That fact signals an indication that although the BIS data on derivative usage show an increase trend on the use of currency derivatives, the macroeconomic environment has an impact on the use of these instruments.Finally, Table 2 also reveals that companies prefer to use currency derivatives rather than foreign assets to hedge their exposure.During the whole period, more than half of the hedgers preferred currency derivatives to foreign assets.
As in studies conducted by Saito and Schiozer (forthcoming), Table 3 reports that currency swaps are the most preferred among all possible currency derivatives.This can be viewed as evidence that the use of currency derivatives by Brazilian companies is linked to the attempt to reduce their foreign currency exposure and not to speculative purposes, since swaps are usually preferred when the sources of exposure extend for multiple periods, but are predetermined.This is the case when liabilities are denominated in foreign currency.In contrast, forward contracts are preferred when the main source of exposure is related to short-term transactions that are characterized by uncertainty.This is the case of foreign revenues derived from exports.

Table 3
The choice of currency derivatives These practices are completely different from those found in previous studies for developed countries.Geczy et al. (1997) show for a sample of U.S. companies that forward contracts, or a combination between forward and option contracts, were the most frequently preferred instruments.Judge (2006) presents similar results for a sample of British companies.He finds that forward was the most frequently used instrument, followed by swaps and options.The preference for swaps is stable across periods and, therefore, is independent of the exchange rate regime.It might indicate that the main concern of the Brazilian hedgers was the possibility that fluctuations of the exchange rate could affect their liabilities.This indication will be explored in next section.
Table 4 reports summary statistics for the comparison between users and nonusers of currency derivatives.Although Table 4 does not show any causal relationship, it helps to clarify differences between foreign currency derivative users and non-users.
The Use of Currency Derivatives by Brazilian Companies: An Empirical Investigation The corporate finance literature states that the relationship between the use of derivatives and the size of the company is ambiguous.If fixed costs of using derivatives are important, one would expect large companies to use more currency derivatives than small firms.In opposition, if small firms are more financially constrained, i.e., more dependent on their internal funds, they would use more in order to avoid fluctuations in their cash flow.Table 4 shows that users of currency derivatives are larger than non-users -a fact that strongly indicates the existence of fixed costs in the use of derivatives.
Table 4 also supports the idea that companies use currency derivatives to reduce their foreign exposure.Companies with higher ratios of foreign sales to total sales, firms that maintain foreign operations, and those with higher levels of foreign debt to total debt are more likely to use currency derivatives. 15If firms want to hedge in order to mitigate the underinvestment problem, theory says that firms with higher growth opportunities would use more currency derivatives.Table 4 shows that this pattern appears in the data.There is a positive relationship between investment opportunities measured by companies' market-to-book ratio and the use of derivatives.
The results reported in Table 4 confirm that firms use currency derivatives in order to reduce their expected bankruptcy costs.Derivative users have higher ratio of debt to assets.Table 4 also shows that there is no evidence that firms use currency derivatives due to taxation.There is no clear relationship between the ratio of tax loss carry-forward to total assets and the use of currency derivatives.Nance et al. (1993) argue that firms that are more profitable make less intense use of currency derivatives, since they will be more able to offset variations in their cash flow.Yet, data show no statistical difference between users and non-users of currency derivatives with respect to their profitability.In the same line, liquidity represented by the current ratio would also be a substitute to the use of currency derivatives.Table 4 reports that there is no difference between users and non-users with respect to this variable.Companies can use foreign assets as substitutes for or complements to the use of derivatives.Table 4 suggests that Brazilian companies see foreign assets as a complement to the use of derivatives; derivatives users have higher ratios of foreign assets to total assets than non-users do.
Interesting that, as mentioned by Mello and Parsons (2000), the fact that liquidity is a complement to the use of currency derivatives might reflect that more financially constrained firms have more difficulty in implementing hedging policies due to their credit risk.Therefore, the relationship between the use of currency derivatives and liquidity should be seen with caution since it might characterize the impact of the companies' financial constraints on their financial policies. 16t is important to discuss the reason behind the choice of the proxies in the paper.The ratio of market value to book value of the equity was used as a proxy for growth opportunities.Elliot et al. (2007) show that this variable can be split into two components: one representing the firms' growth options, and the other, market mispricing; therefore, it would not be an optimal proxy for growth.Several studies use the ratio of R&D investment to sales as a proxy for the companies' growth opportunities.Unfortunately, this variable is not available to most of the companies.Because most empirical papers in the optimal hedging literature use the ratio market-to-book as a proxy for growth opportunities, we decided to use this ratio as our proxy.Graham and Smith (1999) argue that the use of variables like tax loss carryforward is too simple to capture the tax incentives to hedge.They show that depending on whether firms expect to have losses or not, carryforwards might provide disincentives rather than incentives to hedging.They also show that carryforward variables are uncorrelated with the tax function concavity constructed by them.One alternative that could circumvent this problem would be the construction of a variable indicative of the companies' tax function convexity as in Graham and Smith (1999).Unfortunately, for most of the companies in our dataset, data on the companies' taxable income is not available before 1996, or if it is available, the time span is too short to estimate the volatility of the companies' taxable income.Therefore, the use of this variable as a proxy for the convexity of tax system would reduce substantially the size of the sample which would reduce the gains of using such methodology.Facing this trade-off, the ratio of the companies' tax loss carry-forward to total book assets was maintained as a proxy for tax reasons for hedging. 17able 5 displays the correlations among the variables used throughout the paper.Some interesting results come up from the analysis of Table 5.First, the explanatory variables used in the paper are not highly correlated; most of them have correlation lower than 10% and these correlations are not statistically significant.Second, decisions to use currency derivatives, the currency composition of the debt and leverage seem to be taken simultaneously by the firms.Results in Table 5 indicate the existence of a statistically significant correlation among these decisions.This result will be further explored in the multivariate analysis.Finally, defying expectations, and despite not being statistically significant, results in Table 5 show a negative relationship between the ratio of total notional value of derivates to total assets, and the companies' foreign revenue and foreign operations.These facts indicate that rather than a risk, foreign revenue might be perceived as a way to hedge fluctuations on the firms' cash flow.This will also be analyzed later in the text.

Determinants of the Use of Currency Derivatives
In this section, the main determinants of the use of currency derivatives during the period 1996 -2004 are empirically analyzed.Assuming that the firms proceed with the decision to use currency derivatives in two steps, the first one would be when the firms choose whether to use currency derivatives or not; in case the firm opts to use currency derivatives, the second one would be the decision on the amount of currency derivatives to use.This procedure would allow us to analyze whether the determinants of the decision to use currency derivatives are different from the determinants of the extension of hedging.This procedure is similar to that adopted by Allayannis and Ofek (2001) and Saito and Schiozer (forthcoming).
For all estimations, the following empirical specification is used: where dependent i,t stands for the dependent variables; α i and α t represent, respectively, firm and time specific effects and X i,t is the set of explanatory variables.
For the first step, a binary variable that assumes the value of 1 if the firm uses currency derivatives, and 0 otherwise is used as our dependent variable.Therefore, a logistic panel estimation is performed.
For the second step, using data only for the firms that decide to use currency derivatives, the main determinants of the extension of the use of currency derivatives is estimated.In this case, the ratio of total gross notional amount of currency derivatives to total assets18 is used as the dependent variable.Given the nature of this variable that is limited in the interval between 0 and 1, a Tobit panel regression is performed.
An advantage of this specification is that unlike other papers, in this study we can control not only differences across firms that are not captured by the explanatory variables, but also the effect of the change on the macroeconomic environment that took place during the period.

Results
The results for the determinants of the companies' decision to use currency derivatives are presented in Table 6.Three different specifications are estimated.In the first specification, only the variables proxies for the main hypothesis levied by the optimal hedging literature are used.In order to test different hypotheses and to verify the robustness of the results other variables are added to the estimation in specifications 2 and 3.The results in Table 6 show that there is a positive and statistically significant relationship between the size represented by the logarithm of the companies' total sales and the decision to use currency derivatives.Consistent with the existence of fixed costs of hedging, larger firms are more likely to use currency derivatives.This result is also consistent with previous results for more developed countries.
Table 6 confirms that the companies' desire to reduce their foreign exchange exposure is one of the main determinants of their decision to use currency derivatives.Table 6 shows that firms with higher ratio of foreign sales to total sales and firms with foreign subsidiaries are more likely to use currency derivatives, although this last result is not robust across the different specifications.
Table 6 reports that the ratio of foreign debt to total debt is a significant determinant of the companies' decision to use currency derivatives.In light of the additional fact that the swap is the most used currency derivative, it is possible to conclude that Brazilian companies use currency derivatives in order to reduce the exposure on the liability side of their balance sheets to fluctuations of the exchange rate. 19The results in Table 6 give no clear evidence that there is a relationship between the company's decision to use currency derivatives and growth opportuni-ties.Although the coefficient of the market-to-book ratio is statistically significant in two specifications, the coefficient loses its significance once we add all control variables.Moreover, the proxy for the interaction between growth opportunities and financial distress is not statistically significant. 20The results indicate that the possibility of incurring in costs of financial distress is a determinant of the use of currency derivatives.All three specifications confirm that there is a positive relationship between the ratio of debt to assets and the use of currency derivatives.
In none of the estimations presented on Table 6 is the ratio of total loss carriedforward to total assets statistically significant, an indication that tax reasons fail to drive the companies' decision to use currency derivatives.
Table 6 shows no evidence that companies see profitability as a substitute for the use of currency derivatives.The coefficient of the gross margin is not statistically significant.Although the results in Table 6 indicate that liquidity represented by the current ratio is viewed as a complement to the use of currency derivatives, represented by the positive relationship between the current ratio and the use of currency derivatives, this result is not statistically significant.Similar result was found with respect to the ratio of foreign assets to total assets.
Finally, Table 6 confirms not only that cross-sectional variables determine the use of currency derivatives by Brazilian companies, but also that the macroeconomic environment impacts on the companies' decision about hedging.All time dummies are statistically significant.Moreover, the results confirm the hypothesis that companies will hedge less under fixed exchange rate regimes.The opposite would also happen under a flexible exchange rate regime, corroborating the idea that a fixed exchange rate regime leads companies to disregard their exchange rate risk and the floating regime would induce them to take their exchange rate exposure seriously.
Consistent with Allayannis and Ofek (2001), the results confirm that there are differences in the determinants of the companies' decision to use currency derivatives and their decision on the amount to use.That is an indication that our two-step procedure is more suitable for analyzing the companies' hedging activities.
Results in Table 7 show that larger firms make a more extensive use of currency derivatives, confirming that there are economies of scale in using currency derivatives.Table 7 also shows that an interesting pattern happens with respect to the relationship between the use of currency derivatives and the companies' foreign exchange risk.Companies with higher ratios of foreign debt to total debt make a more extensive use of foreign currency derivatives.That is expected since foreign currency liabilities represent the main risk factor for the companies' exchange rate exposure.However, defying expectations, there is a negative relationship between the extent of hedging, the ratio of foreign sales to total sales 21 and the companies' foreign operations.Brazilian exporters might see their foreign sales and foreign activities as a "natural" hedge to the exposure that derives from their foreign currency liabilities.Moreover, exporters consider there is little likelihood of the appreciation of the domestic currency; therefore, they do not expect a loss of revenues due to fluctuations in the exchange rate.Given the costs of hedging, the exporters prefer to hedge less, focusing more on the liability side of their balance sheets.Saito and Schiozer (forthcoming) found similar results for a sample of Latin-American companies.

Table 7
Results for determinants of the extension of the use of currency derivatives -tobit estimation Table 7 shows that, although not robust across different specifications, the underinvestment hypothesis is important to explain cross-sectional differences among users of currency derivatives.Once we add all control variables, the results indicate that there is a positive relationship between the amount of derivatives used and the companies' growth opportunities, confirming that firms with higher growth opportunities use currency derivatives in order to avoid the underinvestment problem as argued by Froot et al. (1993).The results also confirm the prediction of avoided complaints about the valuation of Brazilian Real if they had considered that a floating exchange rate dmregime does not mean a movement towards a higher devaluation of the currency.As proved in recent times, Brazilian Real can value with respect to US Dollar... Exporters could have avoided losses caused by the volatility of the exchange rate by hedging their exposures, but they didn't do it, because they expected Brazilian Real to depreciate even more, and by hedging they would have limited the value of their revenues".Sonia Racy, O Estado de São Paulo, 08/12/2003.
the underinvestment theory about optimal hedging that the hedging activity is the result of the interaction between the companies' growth opportunities and costly external finance.The cross variable between the ratio market-to-book and debt-toassets is statistically significant with the positive sign predicted by the theory.
Results in Table 7 indicate the probability of financial distress impacting the companies' decision about the extension of the use of currency derivatives only through its interaction with growth opportunities.Once we add the control variables, the ratio of debt to assets loses its statistical significance, implying that there is no robust relationship between financial distress and the extension of hedging.Finally, other variables and the macroeconomic environment seem not to be important to the decision about the amount of currency derivatives to use. 22

Robustness tests
Previous estimations considered that the decision to use currency derivatives and the decision on the amount to use are taken separately by the firms.An alternative approach would be to estimate a one-step procedure including all firms in the sample using as the dependent variable the ratio of total gross notional value of the use of currency derivatives to total assets.Behind this estimation lies the assumption that the decisions of the firms are taken one-step and that the explanatory variables have the same impact on both decisions of the firm.
As discussed by Allayannis and Ofek (2001), this is an empirical question to be analyzed.I also estimated a one-step simultaneous decision equation.The results presented in the first column of Table 8 are in line with the previous results discussed in the text. 23Larger firms -firms with higher ratio of foreign currency denominated debt to total debt and firms with higher ratio of debt to assets -use currency derivatives more extensively.In addition, the results confirm that firms perceive their foreign revenue as a "natural" hedge to offset the exposure of their liabilities, indicated by the negative relationship between the ratio of foreign sales to total sales and the extension of the use of currency derivatives.Moreover, the results confirm that after 1999, Brazilian firms make more extensive use of foreign currency derivatives, implied by the positive value and statistical significance of time dummies. 22Exception for the year 2002 that has a positive impact on the companies' use of currency derivatives.
23 Judge (2003) argues that this distinction in the firms' decision is important since most of the theoretical hypotheses establish a relationship between the extent of hedging and firm characteristics.According to Allayannis and Ofek (2001), another weakness of the one-step procedure is that one constrains the coefficient of the variables so as to be the same in two different moments of decision.Our results confirm that this is a strong assumption.

Table 8
Robustness tests Table 8 reports the results of some robustness tests in the estimation of the determinants of the use of currency derivatives.The dependent variable is the ratio of total gross notional amount of currency derivatives to total assets.One-step decision stands for the estimation of equation ( 1) for all companies in the sample assuming that decisions of using and the extension of hedging are taken simultaneously.Panel random effects estimation is performed.Simultaneity stands for the estimation of equation ( 1) considering that the decisions of hedging and indebtness are taken simultaneously.FE-2SLS stands for Balestra and Krishnakumar (1987) generalized two-stage least squares.EC2SLS stands for the error component two-stage least squares method developed by Baltagi (1981) Decisions of hedging and borrowing might be taken simultaneously by the firms; Stulz (1996), Ross (1997) and Leland (1998) suggest that by reducing the probability of financial distress through hedging activities, firms would be increasing their debt capacity provoking a potential increase in their leverage.This problem also appears with respect to the decision of the currency denomination of the debt.By hedging, a firm might reduce the proportion of the debt denominated in foreign currency.This reduction might lead to more space for borrowing in foreign currency considering that the hedge reduced the probability of financial distress.Therefore, in the case of the estimation of (1), both, the ratio of debt to assets and the ratio of foreign currency denominated debt to total debt might be endogenous variables affected by hedging activities, biasing the results.
In order to control this problem, equation ( 1) is estimated using two different methods: Balestra and Krishnakumar (1987) generalized two-stage least squares -FE2SLS and the error component two-stage least squares method developed by Baltagi (1981) -EC2SLS.
The main concern about this simultaneous equation framework is that the theoretical literature does not provide guidance with a model explaining the simultaneous determination of the companies' hedging activities, capital structure and currency composition of the debt.Following Geczy et al. (1997) and Graham and Rogers (2002), the debt-equity decision is specified by a model whose variables are suggested by the capital structure literature.The ratio of tangible assets to total assets used as a proxy for the importance of collateral, the ratio of investment to total sales for investment opportunities, and the gross margin for profitability are used as a set of instruments.
In addition, with respect to the currency composition of the debt, we use a dummy variable if the firm has issued ADR (American Depository Receipts) and a dummy regardless of whether the firm has a foreign origin.The former is expected to have a positive impact on the ratio of foreign debt to total debt since, according to Martinez and Werner (2002), these firms have a lower cost of monitoring, and therefore easier access to international financial markets, the same reason foreign firms might have higher ratios. 24able 8 shows the results of the estimation of the determinants of the use of currency derivatives correcting the possible endogeneity problem.Results in Table 8 confirm the importance of size.Independently of the method of estimation, size is a statistically significant determinant of the companies' use of currency derivatives, confirming that fixed costs of hedging are an important factor on the companies' risk management decisions.
The results also confirm that companies that hold higher levels of foreign currency denominated debt use foreign currency derivatives more intensively.The ratio of foreign currency denominated debt to total debt is significant in both regressions.This fact ratifies that Brazilian companies see foreign currency denominated debt as the main risk they face, and use currency derivatives to reduce the impact of fluctuations of the exchange rate on the liability side of their balance sheets.
In addition, the results in Table 8 indicate the existence of a negative relationship between the use of currency derivatives and the ratio of foreign sales to total sales.It confirms that, given the costs of hedging, the Brazilian firms expect their foreign revenue to act as a "natural hedge" to offset the impact of the fluctuation of the exchange rate on their liabilities.That fact leads exporters and firms with foreign subsidiaries to use currency derivatives less extensively.
The effect of the macroeconomic environment is robust with respect to control for endogeneity.All time dummies from 2001 to 2004 continue to be positive and statistically significant, confirming that under a more volatile macroeconomic environment, firms use currency derivatives more intensively.None of the other theories levied by the optimal hedging literature are statistically significant; they are not robust to the control of the simultaneity issues.
At the bottom of Table 8, a Hausman test is performed with respect to the choice of the fixed effects 2SLS or the random component EC2SLS procedure.The test fails to reject the null hypothesis, indicating that the random effects estimation would be more appropriate, although except for the foreign operations dummy, this result does not change the main conclusions.
One concern about the estimation of a simultaneous equation framework is the quality of the instruments.The first stage is also run to check for weak instruments.For both regressions, F-statistics are higher than 10, an evidence that the instruments used are valid.
Another contribution of the paper is that we can analyze the mechanism through which firms decide their hedging activities, leverage, and currency composition of the debt.Results in Table 9 show that the use of currency derivatives has impact on the companies' decision on leverage and currency composition.The positive and statistically significant impact of the companies' hedging activities on their leverage decisions indicates that the use of derivatives opens more space for the firms to increase their indebtness.Two different reasons might drive this result.Ross (1997) and Leland (1998) argue that if the main benefit of debt financing is the tax deductibility of interest, by hedging firms can increase their debt capacity and through this mechanism increase the value of the firm.Departing from tax reasons, Graham and Rogers (2002) argue that firms might hedge to increase debt capacity and use these funds to invest in profitable projects.Why firms would wish to increase their debt capacity goes beyond the objective of this paper and is a suggestion for future research.
Results in Table 9 also confirm that firms with more tangible assets are more levered, indicating the importance of collateral in the determination of the companies' capital structure.The results indicate that there is a negative relationship, although not robust across different specifications, between investment opportunities and the companies' leverage.Confirming the pecking order theory, the results in Table 9 show a negative relationship between profitability and capital structure decisions.Defying expectations, size seems to have a negative impact on the companies' leverage.
The Use of Currency Derivatives by Brazilian Companies: An Empirical Investigation Table 9 Robustness tests Table 9 reports the results of the determinants of the companies' capital structure and the currency composition of the debt.The dependent variables are, respectively, the ratio of total debt to total assets and the ratio of foreign currency denominated debt to total debt.The estimation considers that the decisions of hedging and indebtness are taken simultaneously.FE-2SLS stands for Balestra and Krishnakumar (1987) generalized two-stage least squares.EC2SLS stands for the error component two-stage least squares method developed by Baltagi (1981) Hausman (FE x EC2SLS) χ 2 (12) = 14.62 -Prob > χ 2 =χ 2 χ 2 (15) = 5.32 -Prob χ 2 =0.9890 Revista Brasileira de Financ ¸as 2007 Vol. 5, No. 2 Results in Table 9 confirm that the use of currency derivatives increases the companies' capacity to borrow in foreign currency.There is a positive relationship between the ratio of foreign debt to total debt and the use of currency derivatives.Therefore, a more extensive use of currency derivatives opens more space for the firms to increase their foreign currency borrowing.
Results in Table 9 indicate that larger firms, firms with higher ratio of foreign sales to total sales, firms with foreign subsidiaries, firms that issued ADRs, and foreign firms are able to hold higher levels of foreign currency denominated debt.
It is interesting to note that the results in Table 9 show, after 2002, a positive and statistically significant impact of the macroeconomic environment on the companies' leverage, and a negative impact on the ratio of foreign currency debt to total debt.It might be an indication that the firms are more aware of the risks of holding debt denominated in foreign currency, which, together with the reduction in the domestic cost of capital, led firms to reduce their debt denominated in foreign currency and to increase their debt in domestic currency to fulfill their investment needs.

Conclusion
This paper studies the use of foreign currency derivatives for a sample of nonfinancial Brazilian companies from 1996 to 2004.The paper shows that there are differences between the determinants of the decision to use currency derivatives and the decision on the amount to use.The explanatory variables have a different impact on each decision, implying that the use of a two-step decision is more appropriate to analyze the companies' use of currency derivatives.
Larger companies, with higher foreign currency exposure and higher probability of incurring costs of financial distress are more likely to use foreign currency derivatives.Moreover, this paper gives evidence that cross-sectional differences do affect the decision to use currency derivatives, as well as the macroeconomic environment that causes an impact on the companies' hedging policies.
Given that the firm chooses to use currency derivatives, the results show that larger companies, companies with higher growth opportunities and with higher levels of the ratio of foreign currency denominated debt to total debt, make more extensive use of currency derivatives.Unlike the results for developed countries, those of exporters and companies with foreign subsidiaries reflect the use of less currency derivatives.It happens because companies see their revenue in foreign currency as a "natural" hedge to the exposure on the liability side of their balance sheets.
Another contribution of the paper is that it reports that the firms take decisions on hedging and indebtness simultaneously.The use of currency derivatives increases debt capacity, leading to higher levels of leverage and debt denominated in foreign currency.
The paper corroborates the idea of Judge (2003) that the study of hedging policies outside the developed world might show the importance of country-specific factors not encountered in previous studies.This paper shows that the fact that developing countries are more susceptible to negative external shocks, usually associated with huge devaluations of home currency, and the companies hold high levels of foreign currency denominated debt change the nature of the companies' hedging activities.

Figure 1
Figure 1Evolution of nominal exchange rate Figure1shows the evolution of nominal exchange rate R$/US$ from January 1996 to December 2004.Source: Central Bank of Brazil

Table 1
Summary statisticsTable1reports the main statistics of the sample.Total Assets represent companies' book value of the assets.Total Sales stands for companies total gross sales.Foreign Sales stands for companies' revenue denominated in foreign currency.Foreign Debt stands for companies' debt denominated in foreign currency.

Table 2
Summary statistics for companies' hedging activitiesTable2reports the firms' choice to hedging from 1996 to 2004 to all firms in the sample.Foreign Assets includes government bonds and investment abroad.Currency Derivatives includes the use of swaps, futures, and options.Both indicate the number of firms that use currency derivatives and hold foreign assets.

Table 3
shows the choice of currency derivatives among Brazilian companies reported in their annual reports from 1996 to 2004.

Table 4
Comparison between users and non-users of currency derivativesTable4reports the comparison between the mean of variables used in the analysis for users and non-users of currency derivatives.Asterisks (*,**) denote statistical significance at 5%, and 10% level of significance for a two-tailed Wilcoxon two-sample test between users and non-users of foreign currency derivatives.Asterisks are placed next to the value that is significantly larger.

Table 5
Correlation among explanatory variablesTable5provides information about the correlation among all variables used in the analysis.*, ** represent, respectively, statistical significance at 5 and 10% level of significance.

Table 6
Results for determinants of the decision to use currency derivatives -logit estimation Table6reports the results for determinants of the decision to use Currency Derivatives from 1996 to 2004.The dependent variable is a dummy that assumes the value of 1 if the firm used currency derivatives and 0 otherwise.1996to2004aredummies for the different years.Results in column (1) are from a random effects logistic estimation.Results in columns (2) and (3) are from a conditional fixed-effects logistic regression.Asterisks (*,**) denote 5% and 10% level of significance.Standard errors are in parenthesis.Hausman shows the results of the Hausman test between the fixed and random effects estimation.

Table 7
reports the results for determinants of the extension of the use of Currency Derivatives from 1996 to 2004.Only data for users of currency derivatives are used in the estimation.The dependent variable is the ratio of total gross notional amount of currency derivatives to total assets.Results are from a panel Tobit regression.Asterisks (*,**) denote 5% and 10% level of significance.Standard errors are in parenthesis.