Determinants and value creation from hedging in the
Netherlands
Tan Huynh (s2031736)
University of Groningen Supervisor: dr. A.A. Tsvetkov
Table of Contents
1 Introduction . . . 3
2 Literature review . . . 4
2.1 Risk management . . . 4
2.2 Modigliani and Miller . . . 4
2.3 Explanations of hedging . . . 5
2.4 Hedging and firm value . . . 6
3 Research goal and research questions . . . 7
3.1 Problem statement . . . 7 3.2 Research goal . . . 7 4 Research methods . . . 8 4.1 Sample selection . . . 8 4.2 Panel data . . . 10 4.3 Descriptive statistics . . . 11 4.4 Tobin’s Q ratio . . . 13 4.5 Control variables . . . 14 4.6 Determinants of hedging . . . 16 4.7 Overview . . . 18 5 Results . . . 19
5.1 Hedging activity over time . . . 19
5.2 Univariate test . . . 20 5.3 Multivariate test . . . 21 5.4 Robustness tests . . . 22 5.5 Diagnostic testing . . . 23 5.6 Logit regression . . . 24 6 Discussion . . . 25
7 Limitations and recommendations . . . 26
8 Conclusion . . . 27
1
Introduction
2
Literature review
In the literature review, we will discuss the main subjects that are relevant to our study. We will take a look at the Modigliani and Miller (1958) propositions that lay a foundation for our problem and look at explanations of some of the most known hedging theories that introduce frictions to these Modigliani and Miller assumptions, such that hedging activities can be justified. At last, we will take a look at what has been written about the relationship between the use of financial derivatives and firm value.
2.1 Risk management
Mason (1995) and others argue that hedging, among diversification and insurance, belongs to one of the well-known risk management activities of a firm. In case of diversification, a firm has different business segments in different industries, making it less exposed in case a specific industry experiences difficult times, because of less than perfect correlation between industries. In case of insurance, a firm can use derivative instruments such as options to protect itself from downside risk, for example by buying put options, or by having an insurance policy with an insurance company to manage risks, while paying an option premium. Tufano (1996) makes a distinction between insurance and hedging and characterizes the use of options for insurance as a nonlinear insurance strategy, whereas hedging using forward or future contracts is characterized as a linear hedging strategy: a price movement of an underlying directly results in a specific value of the contract. By buying or selling derivative contracts, the effects of risks often related to interest-rate, foreign exchange or commodities are mitigated. According to Hull (2014), hedging is a trade designed to reduces risk. Although rare, a perfect hedge is one that eliminates all risk.
2.2 Modigliani and Miller
2.3 Explanations of hedging
According to Graham and Rogers (2002), there are several explanations of why firms decide to hedge. These explanations can be categorized in tax incentives, reduction of financial distress costs, reduction of underinvestment problems, managerial risk aversion, and other incentives. Culp (2002) states that the source of value creation from risk management comes from a reduction in the firm’s cash flow volatility.
1. Tax incentives: According to Smith and Stulz (1985), using financial instruments for hedging contributes to an increase in firm value because it reduces the expected liabilities related to taxes and results in taxable income smoothing. Whereas firms that hedge have more constant cash flows, firms that do not hedge have more variable cash flows. Graham and Rogers (2002) divide the tax incentive for hedging into two parts. The first incentive is to increase debt capacity and interest tax deductions. Ross (1997) and Leland (1998) link the primary benefit of debt financing to interest deduction and show that by hedging, the debt capacity of firms can be increased. The second incentive is to reduce the expected tax liability if the tax function is convex, implying that as the taxable income increases, the tax liabilities increase as well. By hedging using derivatives, a firm can increase the predictability of the cash flows, such that income is taxed at a favorable rate1. Graham and Rogers (2002) find evidence that of two incentives mentioned above,
the main reason why firms hedge is to increase debt capacity and interest deductions and that the tax benefit from hedging increases firm value by 1.1%. They, however, find no evidence that firms hedge to reduce the expected liabilities in case a convex tax function is applicable. An explanation for this is that there are other alternatives, such as the use of specific accounting methods, to smooth taxable income.
2. Reduction of bankruptcy costs: A highly levered firm benefits more from tax advantages than a firm that is mostly financed by equity, because of the lower cost of debt. How-ever, it should be kept in mind that high leverage comes at a cost in a world where bankruptcy costs are present. High leverage puts pressure on the firm, and if it fails to fulfill its debt obligations, the firm may find itself in financial distress. The costs that arise from being in financial distress can be divided into direct and indirect costs. The direct distress costs, needed for payment to lawyers, accountants, administrative fees, in case a bankruptcy occurs, are only 1-3% of the total firm value, while they may seem huge (Warner 1977; Weiss, 1990). The firm might also face indirect distress costs, purely because stakeholders, e.g., customers or suppliers, perceive a realistic probability of default in the foreseeable future. Even the probability of a firm being bankrupt might create costs for a firm, even if the firm is not in financial distress itself. As the expected costs of financial distress increase with the volatility of cash flows and leverage, risk management helps with reducing the bankruptcy costs by decreasing the volatility of cash flows (Smith and Stulz 1985; Mayers and Smith, 1982).
1 An example is that in the Netherlands, the progressive tax rate is 20% on taxable profits up to
3. Reduction of the underinvestment problem: Mayers and Smith (1982) discuss that finan-cial distress and underinvestment are positively related to the volatility of cash flows. Froot et al. (1993) support that hedging can reduce underinvestment problems that are a result of variation in cash flow and costly access to external financing. They write that when external funding is expensive, hedging ensures the financing of positive NPV projects, making hedging activities value increasing. Especially firms that are highly levered may experience that projects with positive NPV may be foregone by rational managers because the outcome of the project mostly benefit bondholders to whom the firm has debt outstanding. In a world where information asymmetries exist, the interest of stakeholders of a company might not necessarily be aligned, and a firm may not always show optimal investment behavior. According to Aretz, Bartram and Dufey (2007) hedg-ing reduces the risk of investment projects and makes it less likely that underinvestment problems occur.
4. Managerial risk aversion: The way managers are compensated can influence the choice of engaging in hedging programs (Smith and Stulz, 1985). Compared to shareholders, who can eliminate idiosyncratic risks by holding diversified portfolios, managers are at a disadvantage to diversify and might choose to reduce this risk exposure by engag-ing in risk management through their firms. Managers that hold stock of the company have more incentives to reduce the volatility of the stock price, whereas managers that hold options are inclined to take more risk due to the upside potential of options when volatility is large. This argument is supported by Tufano (1996), who found that man-agers with stocks manage their risk more compared to manman-agers who hold options when studying the gold mining industry.
2.4 Hedging and firm value
been described by Froot, Scharfstein, and Stein (1993) and indicates that firms hedge to have sufficient amounts of cash to invest in opportunities during unfavorable shocks. Carter et al. (2006) show a positive relationship between the jet fuel price and capital expenditures indicating that airline investment is higher during periods of high prices, even if cash flows as a result of this are lower during these periods. They also show that for periods with low oil prices and volatility (1992-1996), the hedge premiums are negative and for periods with declining, increasing and high prices and volatility, hedging results in positive premiums. On the other hand, Khediri and Folus (2010) found a negative relationship between hedging and firm value for French firms between 2000 and 2002. They found in their univariate tests that non-hedgers have lower firm value, whereas in their multivariate tests, they do not find significant results, which is contradictory to the research done by Allayannis and Weston in 2001.
3
Research goal and research questions
3.1 Problem statement
The results of Allaynnis and Weston (2001) show that using foreign currency derivatives cre-ates firm value for non-financial firms in the United Stcre-ates. However, there are also sources that the use of foreign currency derivatives negative influences firm value. Markets condi-tions, as well do beliefs about hedging activity, change over time and might vary between countries. Research on this subject can be useful to see why empirical research in the past may or may not apply to specific markets and conditions. It is essential and interesting to get more insight into how empirical research applies to different geographic areas, markets and time periods.
3.2 Research goal
Now that the problem is defined, it is possible to determine the research goal, which follows from the problem statement. The research goal can be defined as follows:
The goal is to investigate value creation from hedging over time and to discover possible determinants of hedging
The main research question can be formulated as follows:
Does foreign currency hedging increase value for Dutch firms and what are possible de-terminants for firms to engage in foreign currency hedging using derivatives?
The main research question can be divided into sub-questions:
1. Has the hedging behavior of Dutch firms changed over time?
4
Research methods
4.1 Sample selection
The list of firms in the Netherland between 2012 and 2016 is constructed using the Orbis database by Bureau van Dijk, part of Moody’s Analytics Company. The Orbis database provides information on 275 million companies worldwide with the ability to search for companies with a specific profile based on specific criteria such as geographic location or ac-counting practice, making it a powerful tool to construct our list of companies. Furthermore, the database can provide financial data, extracted from financial reports, but also contains information on stock returns. The research focuses on the period between 2012 and 2016 to obtain the most recent results and because of the increased ability to collect annual reports for this period. These annual reports are required to gather information on hedging activity. The list is constructed as follows using the following criteria. Since the research will focus on companies in the Netherlands, we require the geographic location of the company to be in the Netherlands. To collect data on the use of financial derivatives, we require the accounting standard or practice of the company to be IFRS. The reason for this is that the use of derivatives must be disclosed according to IFRS 7 since 2005 in the section ”Financial Instruments: Disclosures”. Using industry classification, all financial firms with a Standard Industrial Classification (SIC) code between 6000-6999 are excluded from the sample, therefore, focusing this study on non-financial firms. These firms are excluded from the sample because they are market makers and also use financial derivatives for other purposes such as trading and speculation (Allayannis and Weston, 2001). Another criterion is that the company should have been publicly listed in the research period to retrieve the stock data.
Having selected these criteria, this resulted in a list of 96 companies, mostly2 so-called
naamloze vennootschappen (N.V.) that have their headquarters in the Netherlands. The sample consists of large industrial companies with business activities in the Netherlands, as well as companies that have their legal headquarters in the Netherlands, for example, because of the favorable corporate tax structure.
Out of 96 companies, there are 24 companies for which market value is not full unavailable for the research period. Firms that do not have complete data on market value in the examined period, for example, because they have had their initial public offering in the research period, are excluded from the sample. Having constructed the final list of 72 firms, the corresponding International Security Identification numbers (ISIN) for equities are used to retrieve financial data from the Thomson Reuters Datastream database which contains global financial and economic data. The reason for using the Datastream database is that it provides variables of interest, such as data on capital expenditures and foreign sales that were not available in the Orbis database. These variables will be described in section 4.5. Furthermore, the database provides a convenient way to export and construct the data.
2 An exception, e.g. Airbus SE, has its headquarters in the Netherlands, but is registered as a SE
Other than data from the Datastream database, information about hedging activities and foreign currency exposure will be hand-collected from annual reports. Whether a firm has foreign currency exposure and uses foreign currency derivatives is often described in the section financial risk management or risk management objectives and policies of the annual report. A firm is a foreign currency derivative user if it uses forwards, futures, swaps, or options to hedge its foreign currency risk. We also look for profits or losses as a result of using financial instruments for hedging and the notional value of hedging contracts. In the most obvious cases, a foreign currency derivative user reports the following in the Financial risk management section:
”At any point in time, Accell Group hedges 80% of its estimated foreign currency exposure in respect of forecast sales and purchases over the season (July-June). Accell Group uses forward exchange contracts to hedge its currency risk, all with a maturity of less than one year from the reporting date.” - Accell Group 2016
A company that does not use foreign currency derivatives reports the following in the obvious cases:
”Currency risks, arising mainly from purchases in dollars, are not hedged.” - Beter Bed 2016
In less obvious cases, a company reports its policy on currency risk, but does not report the use of derivative instruments in the report whatsoever:
”The currencies in which these transactions primarily are denominated are U.S. dollars and EUR. In order to hedge exposure to foreign currency risk, management attempts to balance the amount of payments in foreign currencies including debt repayments with inflows of currencies from exports sales.” - ASTARTA Holding 2016
4.2 Panel data
This research uses panel data, meaning that multiple entities are observed over time, in this case over a period of 5 years. Using panel data has several advantages such as more degrees of freedoms to improve the power of the tests, and makes it possible to control for unobserved heterogeneity, cross-sectionally and over time. Other than merely pooling all data together, there are broadly two types of panel data models available that can be used when conducting financial research, being the fixed effects model and the random fixed effects model.
We can estimate the fixed model either by creating dummies for each cross section and period, also known as the least squares dummy variable (LSDV) approach. Given the number of cross sections, it would be highly impractical to create dummy variables for each observed cross-section, in case we want to estimate a cross-section fixed effects model. EViews software package provides a tool for estimating fixed effects models without having to create dummies, using the within transformation. When estimating a cross section fixed model, the within transformation subtracts the time-mean of each entity away from the values of the control variables, which results in the facts that we cannot estimate variables that are time-invariant (Wooldridge, 2002). This is important to keep in mind because it can be expected that hedging activity over time does not change for most firms in our sample, which will be discussed later on.
This research will use the integrated analysis tool by EViews, but the LSDV estimator has been estimated to check our work. Since both estimators are mathematically identical, they should yield the same result. We create year dummies to control for time-fixed effects to confirm our findings. However, there is a difference between using both estimation methods. Whereas in the LSDV approach we retrieve a different intercept for each cross-section, or period in case of period fixed-effects, the integrated method of EViews presents a single constant, the mean value or overall intercept of the fixed effects. To test whether fixed effects are present, a redundant fixed effects test will be performed.
4.3 Descriptive statistics
The final sample exists of 72 firms over five years, resulting in 360 firm-year observations and contains several firms that do not have foreign sales. These firms, however, might be subject to foreign currency exposure (FCE), arising from foreign currency denominated assets and liabilities. A firm may be exposed to currency risk on sales, purchases, and borrowings that are denominated in a currency other than the respective functional currency. Instead of dividing the sample into two groups based on foreign sales, like Allayannis and Weston did in 2001, the groups are divided based on foreign currency exposure, based on information in the annual report. Whether a firm is exposed to foreign currency risk, is reported in the Financial risk management section. The reason for this is that a company with foreign sales does not necessarily have foreign currency exposure if sales are denominated in euros. Using this method, we do not run into the problem that we do not know whether a firm with foreign sales has exposure to exchange movements or not. Descriptive statistics of all samples can be found in table 1.
Almost all firms in the sample are exposed to foreign currency risk, not necessarily arising from foreign sales. Out of 72 firms, nine firms have no exposure to foreign currency risk, and it is not surprising that of that sample none of them uses foreign currency derivatives. This results in 63 firms or 88% of the Dutch firms analyzed having exposure to foreign currency risk. Out of these 63 remaining firms with foreign currency exposure, 43 firms are identified as foreign currency derivative users, making the percentage of FCD users in the sample of firms with foreign currency exposure 65.1%. This percentage is similar compared to the 60% that Allayannis and Weston (2001) found for firms with foreign sales in the United States between 1990 and 1995. It can be observed that the mean (1.65) of Tobin’s Q differs from the median (1.38) and that the distribution is positively skewed. From all firms in the sample, 43% of the firms are industrially diversified.
Table 1. Descriptive statistics
N Mean Std. Dev. Median Min. Max. Panel A: All firms
Total assets (millions) 360 6786 17081 571 0.1 103576
Total sales (millions) 360 5975 15405 644 0.0 111018
Market value of equity (millions) 360 5460 14017 359 0.6 113852
FCD Dummy 360 0.59 0.49 1.00 0.00 1.00 Tobin’s Q 360 1.65 1.01 1.38 0.64 9.85 Controls Size 360 13.24 2.67 13.25 4.39 18.46 Dividend dummy 360 0.63 0.48 1.00 0.00 1.00 Leverage 342 0.59 1.03 0.28 0.00 9.69 Profitability 359 0.04 0.21 0.06 -1.37 1.97 Investment growth 347 0.06 0.10 0.03 0.00 0.66
Industrial diversification dummy 360 0.43 0.50 0.00 0.00 1.00
Geographical diversification 313 0.56 0.37 0.67 0.00 1.00
Panel B: Firms with FCE
Total assets (millions) 315 7730 18064 1006 0.6 103576
Total sales (millions) 315 6792 16305 942 0.0 111018
Market value of equity (millions) 315 6229 14826 776 2.1 113852
FCD Dummy 315 0.68 0.47 1.00 0.00 1.00
Tobin’s Q 315 1.62 0.82 1.40 0.64 8.02
Panel C: Firms without FCE
Total assets (millions) 45 175 204 72 0.1 778
Total sales (millions) 45 249 309 90 0.0 1098
Market value of equity (millions) 45 82 91 72 0.6 422
FCD Dummy 45 0 0 0 0 0
Tobin’s Q 45 1.86 1.83 1.24 0.65 9.85
4.4 Tobin’s Q ratio
Tobin’s Q ratio will be used to measure firm value and is defined by the ratio of the market value of a company to the replacement costs of the assets. A high Q ratio would imply that the replacement costs of the assets are lower than the market value of equity making the company overvalued and vice versa. A simplified approximation of Tobin’s Q will be used like Allayannis and Weston did in 2012. We calculate Tobin’s Q as:
Q = T A + M V E − BV E
T A (1)
where
T A = the book value of the firm’s total assets in thousand euros, M V E = the market value of equity in thousand euros,
BV E = the book value of equity in thousand euros,
4.5 Control variables
We want to see whether firms that use financial currency derivatives have higher firm value than firms that do not use financial currency derivatives. In the univariate test, we compare the means of Tobin’s Q’s to see whether there is a significant difference in firm value between hedgers and non-hedgers.
(Qi,t) = α + β1Hedgei,t+ ui,t (2)
In equation 2, Hedgei,t represents a dummy hedge variable that is 1 when firm hedges
and 0 when there is no hedge using foreign currency derivatives. Since we know that other factors influence firm value, we need to control for other variables that may affect Tobin’s Q. Allayannis and Weston (2001) describe several control variables that might influence the value of the firm, other than the use of foreign currency derivatives. We will first discuss the variables and our expectations (hypotheses) of the effect on firm value.
1. Size: There are sources that size of a firm leads to higher profitability and efficiency (Peltzman, 1977). However, there are sources that large firms have lower firm value, because they are more diversified, and this diversification leads to lower firm value (Lang and Stulz, 1994). Large firms are less likely to be financially distressed and more likely to engage in hedging activities compared to small firms, partly due to the set-up costs of hedging (Mian, 1996). This argument is supported by Booth, Smith, and Stolz (1984) that argue that effective hedging is associated with economies of scales, making it more likely for larger firms to use derivative instruments for hedging. Empirical evidence is inconclusive about the effect of size on firm value, and therefore the effect is ambiguous. Using the book value of total assets, which is also used in the calculation of Tobin’s Q, we control for size. Size, S, is calculated as
S = ln(T A) (3)
where TA is the book value of total assets in thousand euros.
3. Leverage: Leverage might influence the value of the firm in the sense that the cost of debt is lower than the cost of equity. Since interest payments on debt are tax-deductible in the Netherlands, leverage is expected to increase the value of the firm. Graham and Rogers (2002) showed that leverage is positively related with firm value, but it has to be kept in mind that high leverage also comes with distress costs which influence the market value of a firm. To control for this, the leverage control variable is defined as the long-term debt divided by the book value of equity. Leverage, L, is calculated as
L = LT D
BV E (4)
where LTD is the long-term debt and BVE is the book value of equity in thousand euros. 4. Profitability: If firms that are more profitable are more attractive than firms that are less profitable, this could result in a higher valuation by the market. Using the return on assets, the ratio of net income to total assets, the profitability of we control for the profitability of the firm. We expect that the relationship between profitability and firm value is positive. Profitability, P, is calculated as
P = N I
T A (5)
where NI is net income and TA is the book value of total assets in thousand euros. 5. Investment growth (IG): Froot, Scharfstein, and Stein (1993) argue that hedgers are more
likely to have more considerable investment opportunities. Earlier research by Smith and Watts (1992) indicated that firm value is also dependent on future investment opportu-nities. They show that investment opportunities positively affects firm performance. To control for investment growth, either capital expenditures or R&D can be used. Because of the lack of data on R&D, the ratio of capital expenditures to sales is used to control for investment opportunities. Investment growth, IG, is calculated as
IG = CE
T S (6)
where CE is capital expenditures and TS is total sales in thousand euros.
6. Industrial diversification (ID): Several sources indicate that industrial diversification leads to an increase in firm value (Lewellen, 1971). However, there are also sources that it leads to a discount in firm value as shown by Servaes in 1996. This is supported by Jensen (1986) that states that industrial diversification is a result of agency problems between shareholders and management and therefore it decreases firm value. Using the four-digit SIC code, a firm is industrially diversified if it has operations in more than one segment, using the four-digit SIC code. In case a company operates in more than one segment, a dummy variable of 1 will be used and 0 otherwise.
On the other hand, there are sources that multinationality positively affects firm value because of foreign direct investments (Dunning, 1973). To control for this variable, the ratio of foreign sales to total sales is used. Geographic diversification, GD, is calculated as
GD = F S
T S (7)
where FS is foreign sales and TS is total sales in thousand euros.
Therefore, the following multivariate test is used when we estimate a pooled regression:
ln (Qi,t) = α + β1Hedgei,t+ β2Sizei,t+ β3AT F Mi,t+ β4Leveragei,t+
+β5P rof itabilityi,t+ β6IGi,t+ β7IDi,t+ β8GDi,t+ ei,t
(8)
In our multivariate test we use the log of Tobin’s Q to control for the skewness in the distribution, whereas in the univariate test, we compare the mean values of Tobin’s Q. In addition, we want to control for unobserved heterogeneity in our data. Therefore we investigate the fixed effects model, which allows each cross-section or time-period to have a different intercept:
ln (Qi,t) = α + β1Hedgei,t+ β2Sizei,t+ β3AT F Mi,t+ β4Leveragei,t+
+β5P rof itabilityi,t+ β6IGi,t+ β7IDi,t+ β8GDi,t+ µi+ vit
(9)
Note that in the fixed effects model, the disturbance ei,t in equation 8 has been decomposed
into a time or cross-section specific effect µi and the remainder disturbance vit.
4.6 Determinants of hedging
To investigate possible determinants of hedging, a logit regression model will be used with the foreign currency dummy as the dependent variable, and our control variables in previous regressions are used as indicators, except for geographic diversification, which we do not regard as a possible determinant of FCD usage. We try to connect each of the variables to the explanations of hedging to find possible determinants of hedging. We will first discuss our variables and explain how they might affect the usage of foreign currency derivatives.
in case taxable profits do not exceed 200.000 euros, which increases the incentives to hedge.
2. Access to financial markets (ATFM): Nance et al. (1993) show that low dividend yields have a negative relationship with derivative usage since it enables firms to preserve liquidity by imposing constraints on dividend payments. It provides an alternative or substitute for hedging and makes corporate hedging ineffective. On the other hand, there are theories that liquid firms have fewer incentives to engage in hedging programs (Tufano, 1996). We expect the relationship with FCD usage to be ambiguous.
3. Leverage: We discussed that hedging reduces the volatility of a firm’s value which may reduce the costs of financial distress. Firms with high leverage are more likely subject to the underinvestment problem. Therefore, we expect that firms with high leverage have strong incentives to hedge their foreign currency exposure using financial derivatives. 4. Profitability: If firms are more profitable, it might be the case that they are less likely to
be financially constrained. Profitable firms are less likely to fail their debt obligations. Hence we would expect that firms that are profitable have fewer incentives to hedge. 5. Investment growth (IG): We expect that relationship between investment growth and the
use of foreign currency derivatives is positive since hedging can alleviate underinvestment problems as described by Froot et al. in 1993. Nance et al. (1993) state that firms with more growth options are more likely to engage in hedging programs to reduce the volatility of the firm value.
6. Geographic diversification (GD): Firms that have foreign sales are more likely to have foreign currency exposure. Therefore we expect the relationship to be positive.
Using this framework, we model the probability of an event, whether a firm engages in foreign currency hedging or not, depending on other independent variables (determinants). In our logit regression, we pool the data from all years together to estimate our equation. Having defined our explanatory variables, we estimate the following logistic function:
P ri = F (zi) =
exp (zi)
1 + exp (zi)
(10)
where P ri is a value between 0 and 1 and a function of random variable zi defined as:
zi= α + β1Sizei+ β2AT F Mi+ β3Leveragei
+β4P rof itabilityi+ β5IGi+ β6GDi+ ei
4.7 Overview
Now that we have defined our control variables and relationships to hedging and foreign currency usage, table 2 summarizes our hypotheses for both regressions. Formally, in our regression analysis we test H0 that the coefficients are zero, but based on the literature, we
expect the following relationships.
Table 2. Summary of variables and hypotheses
Variable Description Tobin’s Q FCD
FCD Dummy The value is 1 if a firm uses FCD
during a year, and 0 otherwise.
+/-Size The natural logarithm of the
book value of total assets. +/-
+/-Access to financial markets
The value is 1 if a firm payed
dividends during a year, and 0 otherwise. +/-
+/-Leverage The ratio of long term debt
divided to the book value of equity. + +
Profitability The ratio of net income to book
value of total assets. +
-Investment growth The ratio of capital expenditures
to total sales. + +
Industrial diversification
The value 1 if a firm has operations
in more than one segment, and 0 otherwise. +/-Geographic
diversification The ratio of foreign sales to total sales. +/- +
5
Results
This section will discuss the hedging activity over time and the results of the univariate, multivariate and logit regressions. We will first look at the use of foreign currency derivatives by firms over five years and see whether firms have changed their hedging activity. In the univariate test, we will see whether there is a difference in Tobin’s Q values between hedgers and non-hedgers and in the multivariate test we investigate the relationship between hedging and firm value while controlling for variables that may influence firm value. At last, we will discuss results from the logit regression to find possible determinants of hedging.
5.1 Hedging activity over time
Using annual reports, the hedging activity of firms between 2012-2016 has been analyzed for each year separately. Out of 63 firms that have foreign currency exposure three firms in the period 2012-2016 have shown to change their foreign currency hedging activities either by starting or stopping hedging their foreign currency exposure by using financial derivatives. In two cases a firm has begun to report the use of FCD’s, and in one case a firm ceases to do so. The rest of the firms have not changed their foreign currency hedging activity. Table 3 presents the number and percentage of firms that hedge using foreign currency derivatives. The number of firms that hedge has remained reasonably stable, but is not surprising given the relatively small sample compared to the research of Allayanis and Weston in 2001. They found that the percentage of firms using FCDs increases by 8% from 232 to 291 firms during 1990-1995. This change of hedging activity may also be a result of a difference in research period. Out of 63 firms, that all have foreign currency exposure, approximately 67% uses FCDs, and 33% did not use foreign currency derivatives.
Table 3. Hedging activity of firm’s over time
2012 2013 2014 2015 2016
Number of firms using derivatives 42 43 42 43 43
Firms without FCD 21 20 21 20 20
Percentage of sample with FCD 66.7% 68.3% 66.7% 68.3% 68.3%
5.2 Univariate test
In the univariate test, we compare the Tobin’s Q ratio’s and see whether firms that hedge have higher firm value than firms that do not hedge. Table 4 presents the mean Qs for all firms that have exposure to foreign currency risk, for both hedgers (column 1) and non-hedgers (column 2). Column 3 presents the difference in Qs between both hedgers and non-hedgers. For the period 2012-2016, the mean Q for hedgers is 1.48, whereas for nonhedgers the mean value is 1.94. This premium is statistically significant at a 5% significance level and results in a discount of 0.46 as shown in column 4 and 5. The results are in contrast to the results of Allayannis and Weston (2001) that found a positive hedging premium. Naturally, there are more factors that affect firm value, and we need to control for these variables in a multivariate setting.
Table 4. Comparison of Tobin’s Q ratios
Year Hedgers Nonhedgers Difference
(1) (2) (3)=(1)-(2) t-statistic p-value
All years Mean 1.48 1.94 -0.46 -1.974 0.049
Std. Dev. 0.53 1.17 N 213 102 2012 Mean 1.32 1.58 -0.27 -1.775 0.081 Std. Dev. 0.48 0.67 N 42 21 2013 Mean 1.46 1.87 -0.42 -1.109 0.027 Std. Dev. 0.51 1.03 N 43 20 2014 Mean 1.48 1.98 -0.51 -0.038 0.970 Std. Dev. 0.54 1.57 N 42 21 2015 Mean 1.56 2.00 -0.44 -0.792 0.431 Std. Dev. 0.54 1.06 N 43 20 2016 Mean 1.57 2.14 -0.58 -1.375 0.174 Std. Dev. 0.55 1.16 N 43 20
5.3 Multivariate test
We have seen earlier that the number of firms that change hedging activity is small, making the foreign currency hedging dummy time-invariant. Due to the properties of the within estimation that demeans the variables over time when estimating a cross-sectional fixed model, we cannot estimate our variable of interest. Therefore, we estimate a pooled OLS model and time-fixed model to control for time-specific effects. In the time-fixed model, we allow each period to have a different intercept. In table 5, the results of both regressions are presented.
Table 5. Results pooled and fixed-effects regression
Pooled regression Fixed effects
Dependent variable: ln (Tobin’s Q) (1) (2)
Observations 270 270 R2 0.26 0.30 Intercept 0.281 0.312 1.072 1.197 FCD dummy -0.181 -0.178 2.006** 1.974** Size -0.003 -0.006 -0.137 -0.305 Dividend dummy 0.210 0.218 2.564** 2.629*** Leverage 0.032 0.037 1.086 1.248 Profitability 0.949 0.937 1.982** 1.912* Investment growth 0.237 0.307 0.909 1.214 Geographical diversification -0.112 -0.115 -0.876 -0.889
Industrial diversification dummy 0.226 0.223
2.607*** 2.550**
This table represents the pooled and time-fixed effects regression of firms with foreign currency exposure. Tobin’s Q is calculated as the ratio of the book value of total assets and market value of equity minus the book value of equity to the book value of total assets. The FCD dummy equals 1 if the company reports the use of forward currency derivatives, such as forwards, futures, options or swaps. Size is calculated as the natural logarithm of total assets. The dividend dummy is set to 1 in case a company pays dividends during that year. Leverage is the ratio of long-term debt to shareholder’s equity. Profitability is the return on assets and is calculated using net income. Investment growth is calculated as the ratio of capital expenditures to sales. Geographical diversification is calculated by the ratio of foreign sales to total sales. The industrial diversification dummy is set to 1 if a firm is active in one more segment at the four-digit SIC code. P-values smaller than 0.01, 0.05 and 0.10 are indicated by ***, ** and*, respectively. T-statistics are based on White’s (1980) standard errors.
biased estimators, since it ignores these effects. The time-fixed effects model has been es-timated by using both within transformation and least dummy square variable methods. As expected, both methods yield the same coefficients and standard errors. Note that the intercept represents the overall intercept of the time-fixed effects.
It can be observed that the foreign currency derivative is significant at a 5% level for both pooled and fixed-effects regressions with a negative coefficient of -0.181 and -0.178, respectively, implying a discount in the valuation for foreign currency derivative users. We find that next to the foreign currency dummy, dividend dummy, industrial diversification dummy, and profitability are all statistically significant and positive. The sign of the dividend dummy is positive, which is in contrast with Servaes (1996) that writes that firms that pay dividends are less likely to be financially constrained and have more exposure to negative NPV projects. However, it supports the findings of Asquit and Mullins (1983) that dividend payments increase shareholder wealth. The sign of the industrial diversification dummy supports the finding of Lewellen (1971) that industrial diversification leads to increase in firm value. We expected the effect of size on firm value to be ambiguous, and results show that the coefficient is both economically as statistically highly insignificant. It can be observed that leverage positively affects firm value which is in line with Gram and Rogers (2002); however, the variable is statistically insignificant. Investment growth is positive which is in line with the findings of Smith and Watts (1992), but the variable is not statistically significant. At last, we find no statistically significant effect for geographical diversification, but the sign of the coefficient is in line with Jensen (1986) that states that geographical diversification leads to a discount in firm value, as result of agency problems. At last, we try to estimate our model using a random effects model. We reject the Hausman test that errors are uncorrelated with the explanatory variables and will not consider this model.
5.4 Robustness tests
Earlier, we calculated Tobin’s Q ratio as the ratio of the book value of total assets and market value of equity minus the book value of equity to the book value of total assets. We will perform a robustness test using a different approach to Tobin’s Q, defined as the market value of equity to total sales. This alternative approximation of the log of this Tobin’s Q has a correlation of 77% with the log Q we used before. We notice that this approximation foreign currency dummy is still negative. The coefficient for the foreign currency dummy is -0.385, but now significant at a 10% level with a t-statistic of 1.695, due to a high standard error.
5.5 Diagnostic testing
To avoid possible violations of the OLS assumptions, a series of diagnostic testing has been performed, the first one being testing for multicollinearity. Multicollinearity occurs when the explanatory variables are highly correlated with each other and results in high standard errors and wide confidence intervals which may affect statistical inference. To detect forms of multicollinearity, the correlations between independent variables have been analyzed. It can be observed that the maximum correlation between regressors is 0.49, indicating that there are no signs of multicollinearity.
Table 6. Correlation matrix
FCD SIZE ATFM LEV. PROFIT IG GD ID
FCD 1 SIZE 0.49 1 ATFM 0.36 0.29 1 LEVERAGE 0.03 0.24 -0.17 1 PROFIT 0.06 0.08 0.17 -0.23 1 IG -0.20 0.05 -0.23 0.16 0.09 1 GD 0.13 0.21 0.22 0.04 -0.09 -0.19 1 ID 0.21 0.30 0.23 -0.07 0.09 -0.09 0.24 1
This table presents the correlation matrix of the regressors in order to detect cases of multicollinear-ity. The values are calculated based on pooled data.
The classical linear regression model assumes linearity in the parameters. To formally test this assumption and to detect misspecification of the model a Ramsey’s (1969) RESET test has been performed. For all tests, the Ramsey RESET test turns out to be statistically insignificant, indicating that there is limited evidence for non-linearity in the regression model.
We use the Jarque-Bera test to test whether the residuals are normally distributed. The distribution of the residuals appears to have a skewness of 0.76 and kurtosis of 4.38, whereas a normal distribution is not skewed and has a kurtosis of 3. Based on the Jarque-Bera test statistic of 47.08 and p-value of 0.0000, we strongly reject the null hypothesis that the residuals are normally distributed. We conclude that the non-normality appears to be caused by a small number of larger positive residuals. For sample sizes that are sufficiently large, violating the normality assumption is virtual without consequences (Brooks, 2014). Given the sample size, there are fewer concerns compared to the case where the sample size would be small.
5.6 Logit regression
The results of the logit regression can be found in table 7. First, a model has been estimated using all control variables from which we expect to increase the likelihood of FCD usage. It can be seen that out of six variables, three variables seem to be significant. When looking at size, we observe that the sign of size is positive and significant and a 1% level, which indicates that larger firms are more likely to use foreign currency derivatives. When we look at the dividend dummy, we see that the sign is positive and significant at the 1% level as well. Firms that pay dividends are more likely to use foreign currency derivatives. In our hypothesis development, the effect of size and dividend payments on the likelihood of using foreign currency derivatives was ambiguous. However, when we look at investment growth, we see that the variable is statistically significant at the 1% level, but the sign is negative. We cannot conclude that Dutch firms with high investment growth, are more likely users of foreign currency derivatives, because they hedge to alleviate the underinvestment problem. For the rest of the variables, which are statistically insignificant, we cannot conclude that they increase or decrease the likelihood of using foreign currency derivatives. In our second model, we estimate a model with our significant predictors. We notice that our explanatory variables remain significant at the 1% level, except for the dividend dummy which is now significant at a 5% level.
Table 7. Variables explaining the use of foreign currency derivatives
Dependent variable: FCD dummy Model 1 Model 2
Observations 270 305 Intercept -7.146 -6.773 -6.094 -6.712 Size 0.596 0.579 6.375*** 7.077*** Dividend dummy 1.0223 0.662 2.71*** 2.006** Leverage -0.004 -0.023 Profitability -0.189 -0.144 Investment growth -6.835 -7.142 -2.685*** -3.478*** Geographical diversification -0.232 -0.475
6
Discussion
We analyzed the annual reports for our sample of Dutch firms and found that a very few select companies changed their hedging activity using foreign currency derivatives. Given the relatively small sample, this is not surprising since most firms do not adjust their hedging activity over a few years. We have seen that the hedging premium is negative for our sample of firms, which is in contrast with the research of Allayannis and Weston for U.S. firms (2001). However, the results are consistent with Khediri and Folus (2010), who found a negative hedge premium for French firms in their univariate test. We will discuss several possible explanations for these differences.
Firms that do not use foreign currency derivatives do not necessarily have more exposure to foreign currency risks, as there are different methods available to hedge the currency risk, other than the use of derivatives. When analyzing the annual reports, we notice that firms that do not use financial derivatives, try to mitigate their foreign currency exposure in different ways, for example by matching the foreign currency inflows and outflows or by holding foreign assets or by using foreign debt. This may result in the fact that currency risks are partly mitigated, possible decreasing value creation from the use of derivatives. Also, given the costs related to engage in hedging programs, it has to be taken into account that a firm should have enough exposure to foreign currency risks when assessing the value creation from using financial derivatives (Muller and Verschoor, 2005).
Differences in results between geographic areas can be explained by a difference in char-acteristics between U.S. firms and European or Dutch firms. Khediri and Folus (2010) sug-gest that there may be differences between countries in information asymmetries, making investors unable to make a correct judgment about the use of derivative instruments for hedging purposes. It might be the case that foreign currency derivatives are partly used for speculative purposes since not all companies explicitly state that this is not the case. Hence it might be the case that investors cannot make a clear distinction whether the use of derivatives is for hedging or speculative purposes. Allayannis et al. (2004) state that corpo-rate governance might play a role in how the use of FCD affects firm value. Further research on these mechanics is required to understand the difference from value creation between countries.
derivatives. When a firm pays dividends, this substitute for hedging is not present, which should make hedging using derivative more effective (Nance et al., 1993). The negative coefficient on investment growth, proxied by the ratio capital expenditures to total sales is in contrast with what we expected since we would expect firms hedge to alleviate the underinvestment problem. We see that the firms with the growth opportunities are the firms that do not use financial derivatives.
We looked at the performance of our estimated model using the coefficients from the regression and calculated the probabilities of a firm using foreign currency derivatives, based on the three significant indicators or determinants. When we look at the group that uses foreign currency derivatives, the model predicts an average probability of 81% that the firm uses derivatives, whereas for the sample without foreign currency derivatives the average probability is 46%. Hence we see that there might be more determinants that we did not include in our model since the model predicts a relatively high probability for the sample of firms that do not use foreign currency derivatives.
7
Limitations and recommendations
Our sample was relatively small compared to other studies. This however also resulted in the fact that almost all our companies either remained unhedged or hedged. This may be a result of a difference in sample size and research period when we compare the number of firms that changed their hedging activity. We were not able to estimate a fixed effects model in which the cross-section was fixed, due to the time-invariant variables such as industrial diversification and our variable of interest, the foreign currency hedging dummy. When we remove these variables from the equation and run our model using both cross-section and time-fixed model, we reject the redundant fixed effects test, indicating that there might be unobserved heterogeneity across firms as well. To estimate time-invariant variables, so-called hybrid models or between within models could be considered, as described by Schunck (2003). The small sample size may also result in the fact that the research is subject to selection bias, making the results sensitive to changes in the data set or specifications.
For the same reason as previously discussed, this study lacks a reverse causality test. Since almost all companies did not change their hedging activity, we cannot compare the Q ratios after a change in hedging activity. Therefore, we cannot formally test if it is the case that firms that hedge have lower firm value or that firms hedge because they have lower firm value.
is value enhancing for large firms that have significant exposures and value-destroying for small firms with smaller exposures since we do not make a distinction between these firms. An interesting follow-up study would also be to investigate the determinants that influence the level of foreign currency derivatives used.
We looked at the logit regression to discover possible determinants of hedging. We used main variables to proxy for hedging incentives or hedging substitutes and could relate most of our control variables in our logit regression to the underinvestment problem or the costs of financial distress. However, we also discussed tax incentives and managerial decisions, which our control variables were less related to. For further research, we could extend our research focusing on these theories, for example by looking at variables related to management com-pensation or amount of shares held by management in a company, or specific tax-related variables.
8
Conclusion
9
References
[1] Asquit, P., Mullins, D.W., 1983. The Impact of Initiating Dividend Payments on Share-holders Wealth. Journal of Business 56, 77-96.
[2] Allayannis, G., Lel, U., Miller, D., 2004. Corporate governance and the hedging pre-mium around the world. Darden Business School Working Paper, University of Virginia, Charlottesville.
[3] Allayannis G., Lel, U., Miller D.P. 2012. The use of foreign currency derivatives, corpo-rate governance, and firm value around the world. Journal of International Economics 87, 65-79.
[4] Allayanis, G., Weston, J.P., 2001. The use of foreign currency derivatives and firm market value. The review of financial studies 14, 243-276.
[5] Aretz, K., Bartram, S.M., Dufey, G., 2007. Why Hedge? Rationales for Corporate Hedging and Value Implications, The Journal of Risk Finance, 8, 434-449.
[6] Booth, J.R., Smith, R.L., Stolz, R.W., 1984. The use of interest futures by financial Institutions. Journal of Bank Research 15, 15-20.
[7] Brooks, C., 2014. Introductory econometrics for finance. Cambridge University Press, Cambridge.
[8] Carter, D., Rogers, D., Simkins, B., 2006. Does fuel hedging make economic sense? Evidence from US Airline Industry. Financial Management 35, 53-68.
[9] Chung, K.H., Pruitt, S.W., 1994. A simple approximation of Tobins Q. Financial Man-agement 23, 70-74.
[10] Culp, C., 2002. The ART of Risk Management: Alternative Risk Transfer, Capital Struc-ture, and the Convergence of insurance and Capital Markets. John Wiley Sons, New York.
[11] Dunning, J., 1973. The Determinants of International Product. Oxford Economic Papers 25, 289-336.
[12] Froot, K.A., Scharfstein D.S., Stein, J.C., 1993. Risk Management: Coordinating Cor-porate Investment and Financing Policies. The Journal of Finance 48, 1624-1658.
[13] Graham, J.R., Rogers, D.A. 2002. Do firms hedge in response to tax incentives? The Journal of Finance 57, 815-839.
[15] Jensen, M., 1986. Agency Costs of Free Cash Flow, Corporate Finance and Takeovers. American Economic Review 76, 323-329.
[16] Jin, Y., Jorion, P., 2006. Firm Value and Hedging, Evidence from U.S. Oil and Gas Producers. The Journal of Finance 61, 893-919.
[17] Khediri, K., Folus, D., 2010. Does Hedging Increase Firm Value? Evidence from French Firms. Applied Economics Letters 17, 995-998.
[18] Koch, P. D., Shenoy, C., 1999. The Information Content of Dividend and Capital Struc-ture Policies. Financial Management 28, 16-16.
[19] Lang, L. H. P., Stulz, R. M., 1994. Tobin’s Q, Corporate Diversification, and Firm Performance. Journal of Political Economy 102, 1248-1248.
[20] Lautier, T.O., 2007. Corporate Risk Management for Value Creation: A Guide to Real-life Applications. Risk Books.
[21] Leland, H.E., 1998. Agency costs, risk management, and capital structure. The Journal of Finance 53, 1213-1243.
[22] Lewellen, W., 1971. A Pure Financial Rationale for the Conglomerate Merger. The Journal of Finance 26, 521-537.
[23] Mason, S.P., 1995. The Allocation of Risk, The Global Financial System: A Functional Perspective. Harvard Business School, Boston.
[24] Mayers, D., Smith, C.,1982. On the Corporate Demand for Insurance. The Journal of Business 55, 281-296.
[25 ]Mian, S., 1996. Evidence on Corporate Hedging Policy. Journal of Financial and Quan-titative Analysis 31, 419-439.
[26] Miller, M.H., Modigliani, F., 1961. Dividend policy, growth, and the valuation of shares. Journal of Business 34, 411-433.
[27] Modigliani F, M.H. Miller., 1958. The cost of capital, corporation finance and the theory of investment. American Economic Review 48, 261-297.
[28] Muller, A., W.F.C. Verschoor., 2005. The Impact of Corporate Derivative Usage on Foreign Exchange Risk Exposure. Working paper, HEC Management School University of Lige and Erasmus University Rotterdam.
[30 ]Peltzman, S., 1977. The Gains and Losses from Industrial Concentration. Journal of Law and Economics 20, 229-263.
[31] Ramsey, J. B., 1969. Tests for Specification Errors in Classical Linear Least-squares Regression Analysis. Journal of the Royal Statistical Society B 31, 350-71.
[32] Ross, M.P., 1997. Corporate hedging: What, why and how? Working paper. University of California, Berkeley.
[33] Schunck, R., 2013. Within and between estimates in random-effects models: Advantages and drawbacks of correlated random effects and hybrid models. The Stata Journal 13, 65-76.
[34] Servaes, H., 1996. The value of diversification during the conglomerate merger wave. The Journal of Finance 51, 1201-1225.
[35] Smith, C., Stulz, R., 1985. The Determinants of Firms Hedging Policies. The Journal of Financial and Quantitative Analysis, 20, 391-405.
[36] Smith, C., Watts, R., 1992. The Investment Opportunity Set and Corporate Financing, Dividend, and Compensation Policies. Journal of Financial Economics 32, 263-292.
[37] Tufano, P., 1996. Who Manages Risk? An Empirical Examination of Risk Management Practices in the Gold Mining Industry. The Journal of Finance 51, 1097-1137.
[38] Warner, J.B., 1977. Bankruptcy costs: some evidence. The Journal of Finance 32, 337-47.
[39] Weiss, L.A., 1990. Bankruptcy resolution: direct costs and violation of priority claims. Journal of Financial Economics, 27, 285-314
[40] White, H., 1980. A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. Econometrica 48, 817-838.
