Did the use of currency derivatives
create a premium on firm value for
public manufacturing companies
in the Eurozone for the years 2013
and 2014?
Karen Guit
10542949
Economics and Business
Specialization: Finance and Organization
Field: Finance
In combination with an internship at Tata Steel
June 2016
This document is written by Student Karen Guit who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Using a sample of public manufacturing firms from the Eurozone, the effect of currency derivative usage on firm value is examined. Cross-sectional time series regressions are used. These are: pooled ordinary least squares, fixed effects, treatment effects and instrumental variables regression. Furthermore, a sensitivity analysis for firm value is performed. Three different measures for firm value are used. A premium between 15.1 percent and 19.8 percent is found for foreign currency derivative usage when using the ordinary least square and fixed effects regression. However, this could be due to the selection and causality bias. The difference in premiums for 2013 and 2014 was not significant. The results of the sensitivity analysis differed; the model was not robust to different measurements of firm value. Furthermore, a discount is found for the usage of interest rate derivatives and no effect is found for the usage of commodity price derivatives.
1. Introduction
According to the classic Modigliani and Miller paradigm, risk management is irrelevant to the value of a firm. Shareholders can practice risk management on their own, for example, by holding well-diversified portfolios. However, recent theories suggest that hedging with currency derivatives adds value.
Most empirical studies on the derivatives topic focus on the relation between corporate hedging and firm characteristics. They research whether the behavior of firms that hedge is consistent with existing theories. Most of these studies provide only indirect evidence that hedging increases firm value. For example, Géczy et al. (1997) found that research and development expenses divided by net sales influence the decision to use currency derivatives. Other studies found that research and development expenses increase market value (Johnson & Pazderka, 1993). Recent studies directly test the relation between firm value and the usage of derivatives; firm value used as a dependent variable and the usage of derivatives as the main explanatory variable.
Allayannis and Weston (2001) and Allayannis et al. (2012) researched the impact of the usage of currency derivatives on firm value. They found a premium on firm value if a company used derivatives. On the other hand, Guay and Kothari (2003) argue that the effect of derivative usage on firm value should be modest. However, holding currency derivatives is likely to be rewarded by investors with higher valuation in the marketplace, due to the underinvestment problem, the cost of financial distress and tax incentives. This will be elaborated in section two, literature review.
In this thesis the effect of currency derivatives on firm value in the Eurozone for the years 2013 and 2014 is researched. Cross-sectional time-series regressions are used. Moreover, a sensitivity analysis for firm value is performed. The expectation is that there is a premium on firm value when a company uses currency derivatives. Furthermore, the effect on firm value of the usage of interest rate and commodity price derivatives are shortly reviewed.
This research is most similar to the research of Allayannis et al. (2012). However, in this thesis additional control variables are added: Altman Z score, quick ratio and commodity price derivative dummy. Furthermore, the relevance of this research is enlarged due to the sensitivity analysis of the measurement of firm value. Three different proxies are used for firm value. It is believed that this thesis is the first to test the valuation effect of derivative usage in the Eurozone. Moreover, this research controls for endogeneity issues while a lot of other researches, for example Allayannis and Weston (2001), do not account for this.
To control for financial condition, liquidity and distress, which affect firm value, the Altman Z score and quick ratio are used (Bartram et al., 2011). According to Hull (2012) firms can also use commodity price derivatives. Like the use of currency rate derivatives, commodity price derivatives will probably affect firm value as well, because hedging with commodity price derivatives decreases the price risk. It is believed that this control variable has never been used in previous researches.
The years 2013 and 2014 are chosen, because these are recent years and in 2013 the Euro appreciated and in 2014 the Euro depreciated against the largest trading partners. This thesis is combined with an internship at the Treasury department of Tata Steel Europe, the second largest steel maker in Europe. Due to the internship, manufacturing companies are researched. The sample is reduced from Europe to the Eurozone, because in the Eurozone there is only one currency. More information about Tata Steel, the Treasury department and my experiences can be found in the internship report.
2. Literature review
Basic economic theory implies that corporate risk management does not contribute to firm value. Based on Modigliani and Miller capital risk management can be seen as a financing policy, it cannot contribute to firm value in a perfect capital market. Hence, one or more assumptions of Modigliani and Miller must be questioned. Capital market imperfections prevent shareholders from the ability to replicate perfect capital risk management.
There are two kinds of explanations why managers undertake risk management activities: based on shareholder value maximization and based on diversification motives for owners or personal utility maximization for managers (Jin & Jorion, 2006). The shareholder value maximization theory is in line with the theory about capital market imperfections.
First, the theory about shareholder value maximization. Firms choose to hedge to reduce the various costs involved with highly volatile cash flows. This includes: the reduction of costs of financial distress, relieving the problem of underinvestment and tax incentives. According to Smith and Stulz (1985) bankruptcy costs create incentives for stakeholder to support optimal hedging. By reducing the variance of the cash flows of a firm, hedging decreases the probability and therefore the expected costs, of financial distress.
The potential underinvestment costs provide incentives for hedging. Firms which do not hedge are more likely to pursue suboptimal investment projects (Myers, 1977). External financing is more expensive than internally generated funds, hedging is useful to secure the availability of those internal funds (Froot et al., 1993). Furthermore, Nance, Smith and Smithson
(1993) state that firms which use derivatives have more growth options in their investment opportunity set.
There are two tax incentives for corporations to hedge: to increase debt capacity and interest tax deductions (Leland, 1998), and to reduce expected tax liability if the tax function is convex (Smith and Stulz, 1985). The first is supported by a more recent research of Graham and Rogers (2002). They indicate that firms hedge to increase debt capacity resulting in increased tax benefits averaging 1.1 percent of firm value.
The theory based on diversification motives for owners or personal utility maximization for managers, does not affect market value. According to Stulz (1984) and Smith and Stulz (1985), risk-averse managers choose to hedge if their wealth and human capital are concentrated in the firm and if the cost to hedge at firm level is lower than the cost of hedging on their own account. Furthermore, engaging in hedging increases the informativeness of corporate earnings as a signal of management ability, it projects quality by eliminating extraneous noise (DeMarzo & Duffie, 1995).
Most of the theories why managers undertake risk management activities are supported by Géczy et al. (1997), they investigate why firms use currency derivatives. Firms with greater growth opportunities and tighter financial constraints are more likely to use currency derivatives. Furthermore, firms with extensive foreign exchange rate exposure and large economies of scale are also more likely to use currency derivatives.
More recently, researches have been examining the direct relation between firm value and hedging. Graham and Rogers (1999) confirm the tax incentives to hedge. By hedging, the debt capacity and firm value increase which results in a higher tax shield. They state that currency hedging increases firm value by 2.1 percent. Interest rate hedging increases firm value by 1.4 percent. Consequently, Batram et al. (2011) found that the use of financial derivatives for nonfinancial companies resulted in a higher firm value. Moreover, there was strong evidence that both total risk and systematic risk diminish if a firms uses derivatives.
On the other hand, magnitude derivatives positions and the cash flows generated by hedging are from an economical point of view insignificant in relation to typical risk exposures of firms (Guay & Kothari, 2003). Firms use derivatives to fine-tune an overall risk-management program that likely includes other means of hedging, for example, operational hedging.
Jin and Jorion (2006) established that hedging reduces the stock price sensitivity of the firm to oil and gas prices, but find no evidence of an effect on the market value of the firm. Investors may easily identify and hedge commodity risk exposure for oil and gas producers. The type of risk that the firm is exposed to, influences the existence of a hedging premium. Based on
the research of Allannis and Weston (2001) and Allayannis et al. (2012), it could be stated that foreign exchange risk cannot be easily hedged by investors on their own.
Allayannis and Weston (2001) researched the effect of foreign exchange derivatives on firm value. They investigated 720 nonfinancial firms in the United States and found that firms that face currency risk and use currency derivatives have a 4.87 percent hedging premium. Firms that begin to hedge experience an increase in value compared to those firms that choose to remain unhedged. The opposite is true for firms that stop hedging.
This is consistent with the results of Allayannis et al. (2012): the premium on firm value is significant for foreign exchange hedgers. They researched a cross-listed sample of 39 countries and found a hedging premium of 14.5 percent for firms with foreign currency exposure.
A lot of researches who investigate the premium on firm value do not control for endogeneity. However, Bartram et al. (2011) control for endogeneity by matching users and nonuser on the basis of their propensity to use derivatives. Allayannis et al. (2012) also use this, furthermore the instrumental variable approach is used by them. To control for endogeneity issues is very important. This is based on previous researches, variables which increase firm value could also influence the decision to use derivatives. That is the reason why the instrumental variable and treatment regression are used in this thesis, this will be elaborated in the methodology.
3. Sample selection
The sample consists of Eurozone public manufacturing companies. The data is gathered from the Compustat database, Bureau van Dijk database and the annual reports of the companies of the sample. The usage of derivatives must be reported according to the International Accounting Standards, which is mandatory to public companies in the EU (Ernst & Young Accountants, 2014). They will be hand-collected from the annual reports.
To account for any time trends two years are researched. The years 2013 and 2014 have been chosen, because these are the most recent years with all the data available. Another reason for this choice is that in 2013 the Euro appreciated and in 2014 the Euro depreciated compared to the currencies of the twelve largest non-Eurozone trading partners (European Central Bank, 2016a). This is based on the nominal effective exchange rate, see appendix 1.
In 2013 and 2014 the companies probably had a net long position in foreign currency, because the value of import of intermediate goods was lower than the export of manufactured products for the Eurozone for the years 2013 and 2014 (European Central Bank, 2016b). Appreciation and depreciation of the Euro affects the firm value in combination with foreign
currency hedging. For example, with a depreciation of the Euro hedging is relatively less beneficial as firms that are not hedging experience a windfall, relating to the hedging firm.
Manufacturing companies are defined as companies with a Standard Industrial Classification code of 2000 up to 3999. Companies that miss data in the Compustat or Bureau van Dijk database, companies with a fiscal year ending other than 31 December, companies which are not listed in Bureau van Dijk Amadeus Stock Annual database, companies with a reporting currency other than Euros and companies that do not have data available for the years 2013 and 2014 are excluded. The companies with a fiscal year ending different to 31 December are deleted, because the effect of the depreciation and appreciation of the Euro cannot be researched correctly.
A random sample was drawn of 150 companies from the database, for the years 2013 and 2014 this results in 300 observations. The sample was reduced, because hand-collecting the foreign currency derivative dummy, commodity price derivative dummy and interest rate derivative dummy is very time-consuming. Some consolidated financial statements were missing or were in a different language than English; in total there are 228 valid observations.
It is possible that derivatives can be used for speculative purposes. However, the sample of firms is all nonfinancial and in the annual reports was never stated that the usage of derivatives was for speculative purposes. This is supported by Guay (1999), he investigated the effect of derivative usage on risk exposures of firms. Firm risk declines following derivatives use, so this is consistent with firms using derivatives to hedge. Consequently, Allayannis & Ofek (2001) found evidence that firms use currency derivatives for hedging, while their use reduces exchange rate exposure for firms.
4. Data
A cross-sectional time-series regression will be used with robust standard errors (𝜀): 𝐿𝑛 𝑇𝑜𝑏𝑖𝑛𝑠𝑄 ! = 𝛽!+ 𝛽!𝐹𝐶𝐷!+ 𝛽!𝐶𝑜𝑛𝑡𝑟𝑜𝑙1!+ 𝛽!𝐶𝑜𝑛𝑡𝑟𝑜𝑙2!+ ⋯ + 𝜀!
The dependent variable (𝐿𝑛 𝑇𝑜𝑏𝑖𝑛𝑠𝑄 ) is the natural log of Tobin’s Q as a proxy for firm value. Tobin’s Q will be calculated by subtracting the book value of equity and adding the market value of equity to the total assets and divide this number by the book value of assets (Allayannis et al., 2012). For the sensitivity analysis the market to sales and the market to book ratio will be used as proxies for firm value.
The foreign currency derivative dummy is the main explanatory variable. This is a dichotomous measurement, because the magnitude of off-balance sheet activities is either inconsistent or missing. Moreover, the notional amounts, as a measure of exposure is not
favorable; since the annual report disclosures are noisy, often because of aggregation and netting. Other variables will be obtained from Compustat and Bureau van Dijk.
A number of control variables are used to control for other determinants that influence firm value. Previous literature is used to determine the control variables in this research (Lang and Stulz (1994), Allayanis and Weston (2001), Jin and Jorion (2006), Allyannis et al. (2012)). The firm size affects the value of the firm (Lang & Stulz, 1994), therefore the natural log of total assets is used as an control variable.
To control for financing constraints, the dividend dummy and the KZ index are used. The determinants of financing constraints are used, because limited access to financial markets influences firm value, while firms that are constrained only take on projects with the highest net present value so that their firm value remains high (Allayannis & Weston, 2001). The KZ index is an index which measures financing constraints (Kaplan & Zingales, 1995). This index is executed based on the research of Lamont et al. (2001): they also used Compustat for the calculation of the index. Furthermore, the dividend dummy is used, this equals one if a firm paid dividend in the current year and zero otherwise. The expectation is that the coefficient is negative, because when a firm paid dividend it is less likely to be capital constrained and may thus have a lower firm value (Allayannis & Weston, 2001).
The Altman Z-score and quick ratio are used to control for financial condition, liquidity and distress. A negative relation between Altman Z-score and firm value and a positive relation between quick ratio and firm value is expected. The capital structure could also influence firm value, to account for this the debt to equity ratio is used. As a proxy for profitability the return on assets is included. A profitable firm is expected to trade at a premium relatively to a less profitable one.
Firm value could be influenced by the future investment growth opportunities. Therefore, capital expenditures to net sales and research and development to net sales are used. A positive relation between those values and firm value is expected. The global industry Q is used as a proxy for growth opportunities (Doidge et al., 2002). Furthermore, year dummies are used to control for time effects. Industry dummies and country dummies were initially used to control for industry effects and cross country variation. However, due to the relative small sample these variables are eliminated, the adjusted R squares are higher for the regressions without industry and country dummies and the results did not change significantly when adding these control variables.
The interest rate derivative and commodity price derivative dummy are included to ensure that the effect of foreign currency derivative usage on firm value is due to the usage of
foreign currency derivatives and not due to the usage of other kinds of derivatives. Furthermore, the effect on firm value of the usage of interest rate and commodity price derivatives can be reviewed. Previous literature found a premium on firm value if a firm used interest rate derivatives or commodity price derivatives (e.g. Graham and Rogers (1999) and Batram et al. (2011)).
In respect of Allayannis and Weston (2001) and Allayannis et al. (2012) the diversification dummy, credit rating variable and advertising to net sales variable is excluded. Since, those variables were not available or not of decent quality, they would bias the results. Moreover, these variables were not significant according to these previous researches.
The definition of the variables can be found in the second appendix. The descriptive statistics without winsorizing can be found in appendix three and with winsorizing in appendix four. There are relatively a lot of foreign currency users compared to the research of Allayannis et al. (2012). Allayannis et al. (2012) is used as a reference, because in a lot of other top journal researches only the United States was researched. Allayannis et al. (2012) had 31.4 percent non-users, this research has 14.9 percent non-users. However, Allayannis et al. (2012) researched countries around the world, although not in all countries of their research a derivative market existed. In this thesis all the researched countries had a derivative market. Moreover, in recent years the turnover of foreign exchange contracts in Europe increased exponentially (Bank for International Settlement, 2016). Therefore, the distribution of the foreign currency derivative dummy is not a concern.
5. Econometric methods
In this part the econometric methods to test the hypotheses will be described. Different tests and models are used to optimize the research. In total four models are used to test the hypotheses: pooled ordinary least squares, fixed effects, treatment effects and instrumental variables. Moreover, the models will also be performed for 2013 and 2014 separately. It could be the case that in one year there is a premium and in one year there is no premium. It is decided to do separate regressions, because in this way there is a clear view of the value of the premiums for both years. To test if there is a significance difference between the years, the significance of the year dummy is tested for all four the regressions. Furthermore, there is a sensitivity analysis for firm value performed. Three different proxies for firm value are used: Tobin’s Q, market to book ratio and market to sales ratio.
Winsorizing is applied to avoid the problem of large outliers. All values which are above the 95th percentile or below the 5th percentile are put at the point of the threshold. Hence, the
extreme variables are replaced by the trimmed minimum and maximum. By winsorizing instead of deleting outliers, the number of observations is not reduced. In appendix three and four the descriptive statistics with and without winsorizing can be found. Tobin’s Q, Global Q, Altman Z score, capital expenditures divided by net sales, debt to equity, KZ index, Quick ratio, Research and development divided by net sales, net income divided by assets and total assets had some extreme outliers.
Furthermore, the model is tested for high degrees of multicollinearity. If there is severe multicollinearity, the coefficients on at least one individual regressor will be imprecisely estimated (Stock & Watson, 2015).
Some variables were skewed. The ladder command of Stata was used to decide whether to use a log transformation. Therefore, the natural logs of Tobin’s Q, Global Q, capital expenditures divided by net sales, debt to equity, research and development divided by net sales and quick ratio are used. In this way the results can still be interpreted and the skewness is reduced. The natural log of total assets was already determined before using the ladder command, while this is a proxy for firm size.
A main assumption of the ordinary leas squares regression is the homogeneity of variance of the residuals. The Breusch-Pagan/Cook-Weisberg test for heteroskedasticity is used. A p-value of zero rejects the null hypothesis that the variance of the fitted values of natural log of Tobin’s Q is constant (appendix 5). Therefore, robust standard errors are used in all the models.
The pooled ordinary least squares model is used to test the hypotheses. This model is executed based on the research of Allayanis and Weston (2001). The fixed effects model is used to control for unobservable country characteristics that may affect value. In the fixed effects model each country is assigned a unique intercept (Allayannis et al., 2012).
To mitigate any potential endogeneity concerns in the analyses between firm value and foreign currency derivative usage, the instrumental variables and treatment effect model will be used (Allayannis et al., 2012). The instrumental variable regression can indicate errors in the foreign currency derivative dummy, simultaneous causality bias and omitted variable bias. One variable will be used as potential instrument to estimate the likelihood of using currency derivatives. This will be elaborated further in this part.
There is a possibility that firms with high value tend to use foreign currency dummies for reasons unrelated to risk management and that the controls for firm characteristics do not capture this information. A relation between firm value and foreign currency derivatives could be drawn while no relation exists. To avoid this potential self-selection bias, a treatment effects
model with full maximum likelihood estimation will be used (Allyannis et al., 2012). With the treatment effect model users and nonusers are matched based on their propensity to use derivatives. The instrument used for this is the same as employed in the instrumental variable approach.
There are two conditions for the independent treatment variable. These are the same as the conditions for the instrument of the instrumental variables model, because there is data on the treatment and on the initial random assignment. The conditions are the relevance condition and the exclusion condition. The instrument is correlated with the usage of foreign currency derivatives and is uncorrelated with firm value.
The validity of the instrument is tested by doing a regression of the foreign currency derivative dummy on the control variables and the instrumental variable. Robust standard errors are used. With the F test the null hypothesis that the instrumental variable is zero, is tested. The rule of thumb of ten is used to indicate if the instrument is valid. This is called the Cragg-Donald test. To test the instrument exogeneity the J-test is normally used. Nonetheless, this cannot be used because there is only one instrument. The exogeneity of the instrument must be done by rational thinking.
The decision to use foreign currency derivatives will be modeled as a function of a firm and industry specific variable. Allayannis et al. (2012) used the following four variables: dummy variable indicating whether the firm resides in a country with a floating currency regime; the percent in the industry which uses derivatives; dummy indicating whether a derivatives market exists in the home country; a variable indicating the number of foreign segments a firm has. They expected a positive influence of these variables on the decision to use derivatives.
Not all variables described above can be used in this research. The whole Eurozone has a flexible exchange rate regime. In all Euro countries there exists a derivative market. The variable indicating the number of foreign segments is not available in Compustat or Bureau van Dijk. Foreign net income divided by net income can be calculated as a proxy for it, but it is not a strong instrument based on the Cragg-Donald test. Hence, foreign net income divided by net income will not be used.
The percentage in the industry that uses derivatives can be calculated. However, the percentage per country that uses derivatives is a stronger instrument, based on the Cragg-Donald test. Therefore, the country percentage is used instead of the industry percentage. It is likely that the average usage of foreign currency derivatives per country is exogenous, because previous researches did not found any link between this variable and firm value. Furthermore,
the correlation between firm value and the average usage of foreign currency derivatives per country in this research is negligible.
Other variables that could be correlated with the usage foreign currency derivatives and not correlated with firm value were considered, like the over-the-counter foreign exchange derivatives per country and the corporate tax rates per country. According to Graham and Rogers (1999) firms are more likely to hedge due to the tax incentives. If the corporate tax rate is high, there are more tax incentives to hedge than when the corporate tax rate is low. Nonetheless, these instruments were not strong enough according to the Cragg-Donald test.
It is considered to do separate regressions for percentiles of the dependent variables, while Tobin’s Q, market to book ratio and market to sales ratio are not perfectly normally distributed even when the log of these variables is used. However, due to the relative small sample the results of the separate regression could be biased. Therefore, this is not implemented in this thesis. Furthermore, it is considered to do an instrumental variable and a treatment effect regression with instruments for interest rate derivative and commodity price derivative dummy. Nonetheless, this would be very time-consuming while a lot of theoretical research must be done to compute the right instruments and the data gathering would also cost a lot of time. Therefore, it is beyond the scope of this research.
6. Statistical hypotheses
I. H0: The usage of foreign currency derivatives has no effect on firm value: the beta of the foreign currency derivative dummy is 0
II. H0: The usage of interest rate derivatives has no affect on firm value: the beta of the interest rat derivative dummy is 0
III. H0: The use of commodity price derivatives has no affect on firm value: the beta of the commodity price derivative dummy is 0
IV. H0: The effect of the use of foreign currency derivatives is equal for the years in which the Euro appreciates and in which it depreciates against the currencies of the twelve largest trading partners of the Eurozone: the beta the year dummy is 0
V. H0: The use of foreign currency derivatives has no affect on firm value in the year 2013: the beta of the foreign currency derivative dummy is 0 for the 2013 regression
VI. H0: The use of foreign currency derivatives has no affect on firm value in the year 2014: the beta of the foreign currency derivative dummy is 0 for the 2014 regression
market to sales: the beta of the foreign currency derivative dummy is 0
VIII. H0: The use of foreign currency derivatives has no affect on firm value measured by market to book ratio: the beta of the foreign currency derivative dummy is 0
7. Results
In this part the results of the regressions are described. The main results can be found in the table below.
Table
Results regressions
This table displays the impact of foreign currency derivatives on firm value for the ordinary least squares (OLS), fixed effects (FE), treatment effects (TE) and instrumental variable regression (IV) . The dependent variable is ln(TobinsQ). The variables are all winsorized at the 5% level. The standard errors are robust. There are 228 observations. The absorbed variable in the fixed effects regression is the country variable (the country variable has a specific number for each country). The endogenous treatment in the treatment effects regression is FCDdummy. The treatment independent variable is average usage of foreign currency derivatives in the country, this is winsorized and transformed based on the ladder command of Stata. The estimator of the treatment regression is computed with the maximum likelihood estimation. The instrumental variable for FCDdummy used in the instrumental variable regression is also the average usage of foreign currency derivatives in the country and is also winsorized and transformed based on the ladder command of Stata. With the Cragg-Donald Wald F statistic the validity of the instrumental variable is tested. The stars ***, ** and * indicate significance at the 1-percent, 5-percent and 10-percent level, respectively.
OLS
FE
TE
IV
Variable Coef. t
Coef. t
Coef. z
Coef. t
FCDdummy 0.1512*** 2.8 0.1636*** 3.0 -0.0088 -0.1 -0.0063 0.0 IRDdummy -0.1247*** -2.6 -0.1225*** -2.8 -0.1307*** -2.8 -0.0934 -1.3 CODdummy 0.0209 0.7 0.0357 1.1 0.0289 1.0 0.0215 0.7 logGlobalQ 0.6638*** 12.7 0.6644*** 11.8 0.6632*** 12.7 0.6834*** 11.5 AltmanZscore -0.0087 -0.7 -0.0065 -0.5 -0.0088 -0.7 -0.0051 -0.4 logCapexsales 0.0222 0.8 0.0098 0.3 0.0195 0.7 0.0169 0.5 logdebttoequity -0.0719** -2.2 -0.0668** -2.2 -0.0734** -2.4 -0.0699** -2.1 Kzindex -0.0049** -2.4 -0.0045** -2.1 -0.0049** -2.5 -0.0045** -2.0 logQuickratio -0.0897* -1.8 -0.1021** -2.1 -0.0929* -1.9 -0.0943* -1.9 logRDsales 0.0395*** 2.8 0.0324** 2.0 0.0396*** 2.9 0.0377*** 2.8 Netincomeassets -0.0570 -0.2 -0.1250 -0.5 -0.0661 -0.3 0.0206 0.1 logTotalassets 0.0074 0.7 -0.0027 -0.3 0.0065 0.7 0.0172 0.9 Yeardummy -0.0182 -0.6 -0.0184 -0.6 -0.0181 -0.6 -0.0200 -0.6 Dividendsdummy -0.0219 -0.6 0.0088 0.2 -0.0190 -0.6 -0.0193 -0.5 Intercept 0.2362* 2.0
0.2067 1.6
0.3720** 2.4
0.2398* 1.9 F-test 46.65*** 38.46*** 40.5*** R-squared 0.6925 0.7102 0.6798
Adj. R-squared 0.6723 0.6775 0.6587
Cragg-Donald 13.25
According to the ordinary least squares regression there is a premium of fifteen percent on firm value if a firm uses derivatives, see the table above. Except for the interest rate derivative dummy, all control variables are in line with the expectation. Some control variables did not meet the expectation. For instance, a positive sign was expected for the variable net income divided by assets, but appeared to be negative for three out of four regressions. Although this is the case, these coefficients are not significant.
The unobservable country characteristics used in the fixed effects regression, which are processed in this regression, do not affect the results a lot. The premium is a bit higher with 1.24 percent in comparison with the results of the ordinary least squares regression. The control variables do not deviate a lot.
The treatment effects regression shows a different effect than the two regressions before: there is a discount of 0.9 percent. This indicates that firms select their derivative use due to the country characteristics. There is possibly a self-selection bias. However, the coefficient of the foreign currency derivative dummy is not significant. Consequently, the result can differ from the discount if for example a larger sample is used.
A discount of 0.6 percent is shown by the instrumental variables regression. This indicates there could be an omitted variable bias, that the foreign currency derivative dummy could be measured with error or there could be is a simultaneous causality bias. An omitted variable bias is probably not the case, because a lot of previous literature is studied for determining the control variables and in previous literature there is no acknowledgement about omitted variable bias. Measurement with error is also not likely.
Nonetheless, a simultaneous causality bias is possible. To test this, the four regressions (including all the control variables) are performed with interaction variables. Firms could select their currency derivative usage to avoid potential underinvestment costs (Myers, 1977). Therefore, the interaction variable foreign currency derivative dummy times research and development expenses divided by net sales is introduced. Besides, the interaction variable foreign currency derivative dummy multiplied by quick ratio is used, because companies could choose to use derivatives so that there are more internal funds available if a firm hedges (Froot et al., 1993).
Furthermore, costs associated with implementing a derivative strategy affect the decision to use currency derivatives. Firms could decide to hedge with currency derivatives based on economies of scale. It is relatively cheap to implant a derivative strategy if the firm already uses
another kind of derivative strategy. For example, a firm is more likely to hedge with currency derivatives if it already uses an interest rate derivative strategy (Géczy et al., 1997). Furthermore, firm size is a proxy for economies of scale. Large firms are more likely to hedge than small firms (Géczy et al., 1997). Therefore, the commodity price derivative dummy, interest rate derivative dummy and the firm size are used for the interaction variables.
All these interactive variables are in line with the determinants of the usage of currency derivatives based on the research of Géczy et al. (1997). The Research and development expenses divided by net sales, quick ratio, firm size and the usage of other derivatives affect the firm value and are significant determinants of the usage of currency derivatives at the one percent level. Hence, these are good interaction variables to indicate if there is a simultaneous causality bias.
The foreign currency derivative dummy is no longer significant in all of the regressions (appendix 6). This could indicate that there is a simultaneous causality bias. Besides, the interaction variable with commodity price derivative dummy and the interaction variable with the interest rate derivative dummy are significant in some of the regressions. This indicates that firms hedge due to the economies of scale described above.
In all the regressions the coefficient of interest rate derivative dummy is negative. For three out of four regressions the coefficient is also significant. Faulkender (2005) states that the final interest rate exposure is largely driven by the slope of the yield curve when the debt is issued. This suggests that interest rate risk management practices are primarily driven by speculation or myopia, not hedging considerations. Therefore, the discount could be due to bad speculation of the firms. However, it could also be due to the small sample size. The discounts of the treatment effect and instrumental variable regression must be interpreted with care while for the interest rate derivative no instrument is used.
The premiums of the usage of commodity price derivatives are low and not significant. Jin and Jorion (2006) did not find a premium on firm value. They suggest that there is no premium while the nature of the commodity risk exposure is easy to identify and investors can do it on their own. Likewise, this could be the case in this thesis while there are low and not significant premiums. On the other hand, it could also be due to the small sample size.
In both years there is a premium when using the ordinary least squares and fixed effects regression. When measuring with the treatment effects and instrumental variables regression, there was a discount in 2013 and a premium in 2014 (appendix 7). This contradicts with the hypothesis. There could be a (larger) premium in 2014, because of a higher import than export value. However, this could also be due to macro-economic events, which are not accounted for.
Furthermore, the difference of firm value for the years 2013 and 2014 is not significant while the year dummies were not significant for all four of the regressions, see the table above.
A sensitivity analysis of Tobin’s Q is performed. The results can be found in appendix eight. With the market to book ratio the premium is larger than the Tobin’s Q measurement for the firm value. Furthermore, the premium still exists with treatment effect and instrumental variable regression. Tough, the premiums are not significant for the treatment effect and instrumental variable regression. The results with market to sales as dependent variable differ across the four regressions. This could be due to the fact that market to sales is a volatile measurement for firm value for this sample, while the value of net sales is very different per firm. It could be stated that the model is not very robust to different measurements of firm value, because of the small sample size.
8. Discussion
There are three recommendations for further research. The first recommendation is to enlarge the amount of observations. The foreign currency dummy, the interest rate dummy and the commodity price derivative dummy were hand-collected from the annual reports. Since this was very time-consuming, a random sample of 150 companies was used instead of the whole dataset. When enlarging the sample size, the significance of the results would be enlarged. This would contribute to more reliable results. In addition, the robustness of the model would probably be enlarged too. Because of the small sample size, the results of the sensitivity analysis differed a lot and the standard errors were large. Therefore, it is recommended to enlarge the amount of observations.
The second recommendation is to control for endogeneity for the commodity price derivative dummy and the interest rate derivative dummy. It is likely that these variables also have some endogeneity issues. However, the search of good instruments for the treatment and instrumental variable regression was to time-consuming. It took a while to find an instrument that adhered to the relevance and exclusion condition for the foreign currency derivative dummy. The last recommendation is related to the 2013 and 2014 analysis. The differences in premiums could be due to macro-economic events. This research does not take into account these macro-economic events. The gathering of good indicators of macro-economic events would have been to time-consuming. Therefore, it is chosen to focus on other parts of the research, like the endogeneity issues with the foreign currency derivative dummy, while the differences in the premiums for 2013 and 2014 were not significant.
9. Conclusion
In this thesis the effect of the usage of foreign currency derivatives on firm value is researched. Four different kinds of regressions are used to determine the effect. A premium is found between 15.1 percent and 19.8 percent for the ordinary least square and the fixed effects regression. Both were significant at the one-percent level. However, a small discount is found for the treatment effects and instrumental variable regression. This could be due to the self-selection bias and simultaneous causality bias. The conclusion of the self-selection bias was implied by the results of the treatment effect model. The simultaneous causality bias was based on the results of the instrumental variable regression and was tested by introducing interaction variables to the models. It is important to note that the results of the treatment effect and the instrumental variable regressions were not found to be significant.
The interest rate derivative dummy had a negative and significant coefficient. This could be due to bad speculation of the firms in this sample. The premiums of commodity price derivative dummy were low and not significant. The reason for this could be that investors can easily identify the commodity risk exposure in this sample and can hedge on their own. Nonetheless, it could also be due to small sample size.
The differences in premiums of foreign currency derivative usage were not significantly different for each year. Furthermore, the model is not very robust to different measurements of firm value. This could be because of the small sample size.
Recommendations for further research are to enlarge the sample size, to control for endogeneity for the commodity price derivative and interest rate derivative dummy, and to include variables that account for macro-economic events in the year-to-year analysis. These shortfalls were behind the scope of this thesis, due to the limited time.
The relevance of this thesis was enlarged by adding additional control variables compared to previous researches. Furthermore, the relevance of this research was enlarged due to the sensitivity analysis of the measurement of firm value. Moreover, this thesis controlled for endogeneity issues by using the treatment effect and instrumental variable regression. An instrument was used for the foreign currency derivative dummy.
References
Allayannis, G., Lel, U., & Miller, D. P. (2012). The use of foreign currency derivatives, corporate governance, and firm value around the world. Journal of International Economics, 87, 65- 79.
Allayannis, G., & Weston, J. P. (2001). The Use of Foreign Currency Derivatives and Firm Market Value, The Review of Financial Studies, 14(1), 243-276.
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.
Bank for International Settlements (2016). Exchange-traded futures and option, by location of
exchange [A, U, B, T, TO1, 8C]. Retrieved from
http://stats.bis.org/statx/srs/tseries/XTD_DERIV/A.U.B.T.TO1.8C?t=d1&c=&p=2009&i=16 .4
Bartram, S. M., Brown, G. W., & Conrad, J. (2011). The Effects of Derivatives on Firm Risk and Value. Journal of Financial and Quantitative analysis, 46(4), 967-999. DeMarzo, M., & Duffie, D. (1995). Corporate Incentives for Hedging and Hedge Accounting. The
Review of Financial Studies 8(3), 743-771.
Doidge, C., Karolyi, A., & Stulz, R.M. (2004). Why are foreign firms listed in the U.S. worth more? Journal of Financial Economics 71, 205-238.
Ernst & Young Accountants, Directoraat Vaktechniek (2014). Handboek jaarrekening,
Toepassing van de Nederlandse wet -en regelgeving en IFRS. Deventer, The
Netherlands: Kluwer.
European Central Bank (2016a). Eurostat External Trade Statistics [M, X, INT, MAN]. Retrieved from http://sdw.ecb.europa.eu/browseChart.do?TRD_PRODUCT=INT&TRD_PRODUCT =MAN&COUNT_AREA=I8&TRD_FLOW=M&TRD_FLOW=X&node=2018791&q=&TRD_f SUFFIX=VAL&DATASET=0&advFil=y
European Central Bank (2016b). Effective exchange rates [D, E0]. Retrieved form http://sdw.ecb.europa.eu/browseChart.do?node=2018795&FREQ=D&CURRENCY=E0& DATASET=0&advFil=y
Fauver, L., & Naranjo, A. (2010). Derivative usage and firm value: The influence of agency costs
and monitoring problems 16, 719-735.
Faulkender, M. (2005). Hedging or Market Timing? Selecting the Interest Rate Exposure of Corporate Debt. The Journal of Finance, 60 (2), 931-962.
Froot, K., Scharfstein, D., & Stein, J. (1993). Risk management: Coordinating investment and financing policies. The Journal of Finance 48 (5), 1629-1658.
Géczy, C., Minton, B.A., & Schrand, C. (1997). Why Firms Use Currency Derivatives. Journal of
finance, 52, (4), 1323-1354.
Graham, J.R., & Rogers, D.A. (2002). Do Firms Hedge in Response to Tax Incentives. The
Journal of Finance 57(2), 815-839.
Graham, J.R., & Rogers, D.A. (1999). Is corporate hedging consistent with value maximization? An empirical analysis. Journal of Finance 2, 815-840.
Guay, W., & Kothari, S.P. (2003). How much do firms hedge with derivatives. Journal of
Financial Economics, 70, 423-461.
Guay, W.R. (1999). The impact of derivatives on firm risk: An empirical examination of new derivative users. Journal of Accounting and Economics 26, 219-251.
Hull, J.C. (2012). Options, Futures, And Other Derivatives, Global Edition. Essex, England: Pearson Education Limited.
Jin, Y., & Jorion, P. (2006). Firm Value and Hedging: Evidence from U.S. Oil and Gas Producers. The Journal of Finance, 61 (2), 893-919.
Johnson, L.D., & Pazderka, B. (1993). Firm Value and Investment in R&D. Managerial and
Decision Economics 14, 15-24.
Kaplan, S.N., & Zingales, L. (1995). Do financing constraints explain why investment is
correlated with cash flow? (NBER Working Paper no. 5267). Retrieved from National
Bureau of Economic Research website:
http://www.nber.org.proxy.uba.uva.nl:2048/papers/w5267.pdf
Lamont, O., Polk, C., & Saá-Requejo, J. (2001). Financial Constraints and Stock Returns. The
Review of Financial Studies 12(2), 529-554.
Lang, L. H. P., & Stulz, R. M. (1994). Tobin’s q, Corporate Diversification, and Firm Performance. Journal of Political Economy, 102(6), 1248-1280.
Leland, H.E. (1998). Agency Costs, Risk Management, and Capital Structure. Journal of Finance
53(4), 1213-1242.
Mackay, P., Moeller, Sara B., 2007. The value of corporate risk management. Journal of
Finance 62, 1349–1419.
Myers, S.C. (1977). Determinants of corporate borrowing. Journal of financial economics 5, 147-175.
Nance, D.R., Smith, C.W., & Smithson Jr., C.W. (1993). On the determinants of corporate hedging. The Journal of Finance 48, 267-284.
Smith C.W., & Stulz, R.M. (1985). The Determinants of Firms’ Hedging Policies. Journal of
Stock, J.H., & Watson, M.W. (2015). Introduction to Econometrics. Essex, England: Pearson Education Limited.
Stulz, R.M. (1984). Optimal Hedging Policies. The Journal of Financial and Quantitative Analysis
Appendix 1
Nominal effective exchange rate 2013-2014
Appendix 2 Definition of variables
Name Meaning
TobinsQ (total assets+market value of equity-book value of equity)/total assets FCDdummy
dummy is equal to 1 if the firm used foreign currency derivatives, equal to 0 if a firm did not used it
IRDdummy
dummy is equal to 1 if the firm used interest rate derivatives, equal to 0 if the firm did not used it
CODdummy dummy is equal to 1 if the firm used commodity price derivatives, equal to 0 if the firm did not used it GlobalQ based on four-digit SIC code, the average tobin's Q is calculated
AltmanZscore
(6.56*(working capital/total assets)+3.26*(retained earnings/total assets)+6.72*(earnings before interest and taxes/total assets)+1.05*(stockholders equity+preferred/preference stock)/(total liabilities)
Capexsales capital expenditures/(net sales) debttoequity total liabilities/stockholders equity
KZindex
−1.001909((earnings before interest and taxes+depreciation and amortization)/property, plant and equipment)+ 0.2826389(total assets+market capitalisation-common/ordinary equity-deferred taxes/total assets) + 3.139193(long-term debt+debt in current liabilities/long-term debt+debt in current liabilities+stockholders equity)−39.3678(dividends
common/ordinary+dividends preferred/preference/property, plant and
equipment)−1.314759(cash and short-term investments/property, plant and equipment) Quickratio (current assets-inventories)/current liabilities
RDsales research and developments expense/(net sales) Netincomeassets net income (loss) consolidated/total assets Totalassets total assets
Yeardummy dummy is equal to 1 for the year 2014 and 0 for the year 2013
Dividendsdummy dummy is equal to 1 if the firm paid dividend, equal to 0 if a firm did not paid dividend ICFCDyd interaction variable: FCDdummy multiplied by Yeardummy
102 103 104 105 106 107 108 109 110 jan -‐1 3 fe b-‐13 m rt-‐ 13 ap r-‐ 13 mei -‐13 jun-‐1 3 jul-‐1 3 au g-‐ 13 se p-‐13 ok t-‐ 13 nov -‐1 3 de c-‐1 3 jan -‐1 4 fe b-‐14 m rt-‐ 14 ap r-‐ 14 mei -‐14 jun-‐1 4 jul-‐1 4 au g-‐ 14 se p-‐14 ok t-‐ 14 nov -‐1 4 de c-‐1 4
ICFCDIRD interaction variable: FCDdummy multiplied by IRDdummy ICFCDCOD interaction variable: FCDdummy multiplied by CODdummy ICFCDfirmsize interaction variable: FCDdummy multiplied by Totalassets ICFCDquickratio interaction variable: FCDdummy multiplied by Quickratio ICFCDrd interaction variable: FCDdummy multiplied by RDsales Markettobook market value of equity/book value of equity
Markettosales market value/(net sales)
Appendix 3
Descriptive statistics without winsorizing
This table provides descriptive statistics for the sample used throughout the regressions. It displays the mean and the standard deviation for firms who use foreign currency derivatives, interest rate derivatives and commodity price derivatives and firms who do not. The definition of the variables can be found in appendix 2. There are in total 228 observations for each variable. The data is not winsorized. The Euro figures are measured in million EUR and reported in absolute terms, although some of them are used in logs throughout the regressions.
Variable Mean Std. Dev. Mean Std. Dev.
FCD non-users (34 obs.) FCD users (194 obs.)
TobinsQ 2.44 2.69 1.58 0.95 FCDdummy 0.00 0.00 1.00 0.00 IRDdummy 0.35 0.49 0.84 0.37 CODdummy 0.06 0.24 0.34 0.48 GlobalQ 2.23 1.64 1.62 0.77 AltmanZscore -0.44 8.88 2.51 3.46 Capexsales 0.85 2.47 0.05 0.03 debttoequity -18.74 114.75 1.78 1.67 KZindex -3.68 13.09 -4.94 16.61 Quickratio 2.24 2.88 1.16 0.63 RDsales 38.02 189.76 0.06 0.15 Netincomeassets -0.17 0.33 0.02 0.15 Totalassets 121.00 121.03 8611.13 21783.23 Yeardummy 0.50 0.51 0.50 0.50 Dividendsdummy 0.18 0.39 0.46 0.50
IRD non-users (53 obs.) IRD users (175 obs.)
TobinsQ 2.62 2.48 1.43 0.57 FCDdummy 0.58 0.50 0.93 0.25 IRDdummy 0.00 0.00 1.00 0.00 CODdummy 0.17 0.38 0.34 0.47 GlobalQ 2.06 1.26 1.60 0.84 AltmanZscore 1.44 7.94 2.26 3.27 Capexsales 0.54 2.01 0.05 0.08
debttoequity -11.87 91.89 1.92 1.66 KZindex -4.98 27.93 -4.69 10.28 Quickratio 2.01 2.43 1.11 0.51 RDsales 23.82 152.34 0.24 1.80 Netincomeassets -0.11 0.36 0.02 0.08 Totalassets 2352.39 7968.04 8857.12 22570.88 Yeardummy 0.49 0.50 0.50 0.50 Dividendsdummy 0.26 0.45 0.46 0.50
COD non-users (160 obs.) COD users (68 obs.)
TobinsQ 1.84 1.60 1.40 0.48 FCDdummy 0.80 0.40 0.97 0.17 IRDdummy 0.73 0.45 0.87 0.34 CODdummy 0.00 0.00 1.00 0.00 GlobalQ 1.82 1.11 1.43 0.40 AltmanZscore 1.91 5.61 2.43 1.55 Capexsales 0.22 1.17 0.05 0.03 debttoequity -2.66 52.96 1.97 1.94 KZindex -4.29 16.63 -5.86 14.90 Quickratio 1.44 1.52 1.03 0.35 RDsales 8.14 87.84 0.03 0.02 Netincomeassets -0.03 0.23 0.03 0.08 Totalassets 2504.57 6992.74 18734.45 33080.97 Yeardummy 0.49 0.50 0.51 0.50 Dividendsdummy 0.36 0.48 0.56 0.50 Appendix 4
Descriptive statistics with winsorizing
This table provides descriptive statistics for the sample used throughout the regressions. It displays the mean and the standard deviation for firms who use foreign currency derivatives, interest rate derivatives and commodity price derivatives and firms who do not. The definition of the variables can be found in appendix 2. There are in total 228 observations for each variable. The data is winsorized at a five percent level, so the threshold is set at the 5th and 95th percentile. The Euro figures are measured in million EUR and reported in absolute terms, although some of them are used in logs throughout the regressions.
Variable Mean Std. Dev. Mean Std. Dev.
FCD non-users (34 obs.) FCD users (194 obs.)
TobinsQ 1.80 1.07 1.52 0.68 FCDdummy 0.00 0.00 1.00 0.00 IRDdummy 0.35 0.49 0.84 0.37 CODdummy 0.06 0.24 0.34 0.48 GlobalQ 2.26 1.56 1.62 0.77 AltmanZscore 1.48 3.46 2.67 2.04
Capexsales 0.06 0.05 0.05 0.03 debttoequity 1.04 0.78 1.74 1.19 KZindex -4.81 7.43 -4.69 5.98 Quickratio 1.66 0.86 1.14 0.50 RDsales 0.17 0.22 0.05 0.09 Netincomeassets -0.09 0.16 0.02 0.08 Totalassets 124.45 117.94 6292.80 12101.72 Yeardummy 0.50 0.51 0.50 0.50 Dividendsdummy 0.18 0.39 0.46 0.50
IRD non-users (53 obs.) IRD users (175 obs.)
TobinsQ 1.99 1.09 1.44 0.57 FCDdummy 0.58 0.50 0.93 0.25 IRDdummy 0.00 0.00 1.00 0.00 CODdummy 0.17 0.38 0.34 0.47 GlobalQ 2.07 1.23 1.60 0.83 AltmanZscore 2.57 3.18 2.47 2.02 Capexsales 0.06 0.04 0.05 0.03 debttoequity 0.93 0.68 1.85 1.20 KZindex -7.35 8.56 -3.91 5.05 Quickratio 1.57 0.82 1.11 0.47 RDsales 0.14 0.19 0.05 0.09 Netincomeassets -0.04 0.16 0.02 0.07 Totalassets 2355.18 7967.21 6286.91 12090.83 Yeardummy 0.49 0.50 0.50 0.50 Dividendsdummy 0.26 0.45 0.46 0.50
COD non-users (160 obs.) COD users (68 obs.)
TobinsQ 1.63 0.84 1.40 0.48 FCDdummy 0.80 0.40 0.97 0.17 IRDdummy 0.73 0.45 0.87 0.34 CODdummy 0.00 0.00 1.00 0.00 GlobalQ 1.83 1.09 1.44 0.39 AltmanZscore 2.52 2.61 2.43 1.55 Capexsales 0.05 0.03 0.05 0.03 debttoequity 1.53 1.18 1.89 1.10 KZindex -4.98 6.57 -4.05 5.21 Quickratio 1.29 0.67 1.04 0.34 RDsales 0.09 0.14 0.03 0.02 Netincomeassets 0.00 0.12 0.03 0.05 Totalassets 2414.88 6306.99 12333.14 16548.17 Yeardummy 0.49 0.50 0.51 0.50 Dividendsdummy 0.36 0.48 0.56 0.50
Appendix 5
Breusch-Pagan / Cook-Weisberg test for heteroskedasticity
The null hypothesis is tested that the variance of the residuals is constant. The p-value is very small, so the null hypothesis is rejected.
Ho: Constant variance
Variables: fitted values of logxTobinsQ
chi2(1) 31.57
Prob > chi2 0.000
Appendix 6
The FCDdummy and interaction variables coefficients when adding interaction variables
This table displays the impact of foreign currency derivatives on Tobin’s Q when interaction variables are added. The dependent variable is ln(TobinsQ). The variables are all winsorized at the 5% level. The standard errors are robust. There are 228 observations. The control, treatment and instrumental variables are similar to table displayed in part seven, results. OLS=Ordinary Least Square regression, FE=Fixed effects regression, TE=Treatment effects model and IV=Instrumental variables regression. The stars ***, ** and * indicate significance at the 1-percent, 5-percent and 10-percent level, respectively.
OLS FE TE IV
Variable Coef. t Coef. t Coef. z Coef. t
FCDdummy 0.0105 0.0 0.1156 0.39 -0.1130 -0.39 -0.6979 -0.54 ICFCDCOD 0.2412*** 3.11 0.2409** 2.52 0.181** 1.96 0.2037* 1.71 ICFCDfirmsize -0.0454 -0.46 -0.0731 -0.64 -0.0563 -0.6 0.0064 0.04 ICFCDIRD 0.1804* 1.77 0.1506 1.43 0.1816* 1.88 0.1268 0.9 ICFCDquickratio 0.0187 0.35 0.0126 0.21 0.0143 0.28 0.1560 0.62 ICFCDrd 0.0124 0.35 0.0265 0.7 0.0150 0.43 -0.0039 -0.08 Table 7
Premiums and discounts for 2013 and 2014
This table displays the impact of foreign currency derivatives on Tobin’s Q for the years 2013 and 2014. The regressions are performed for each year. There are 114 observations for the regressions for 2013 and for 2014. The dependent variable is ln(TobinsQ). The variables are all winsorized at the 5% level. The standard errors are robust. The control, treatment and instrumental variables are similar to variables used in the table in part seven, results. OLS=Ordinary Least Square regression, FE=Fixed effects regression, TE=Treatment effects model and IV=Instrumental variables regression. The stars ***, ** and * indicate significance at the 1-percent, 5-percent and 10-percent level, respectively.
OLS FE TE IV
2013 0.1252* 0.1339* -0.0118 -0.0405
Appendix 8
The FCDdummy coefficients when using alternative measures of Tobin’s Q
This table displays the impact of foreign currency derivatives on Tobin’s Q and two alternative measures of Tobin’s Q. The control, treatment, instrumental and independent variables are similar to the table displayed in part seven, results. The dependent variables are: TobinsQ=ln(TobinsQ) (the same as in previous regressions), Markettobook=ln(market value/book value) and Markettosales=ln(market value/net sales). The variables are all winsorized at the 5% level. OLS=Ordinary Least Square regression, FE=Fixed effects regression, TE=Treatment effects model and IV=Instrumental variables regression. Robust standard errors are used. There are 228 observations. The stars ***, ** and * indicate significance at the 1-percent, 5-percent and 10-percent level, respectively.
OLS FE TE IV
TobinsQ 0.1512*** 0.1636*** -0.0088 -0.0063
Markettobook 0.2197* 0.2679** 0.1795 0.1778
