Does hedging add value? Evidence from The Netherlands

1

Does hedging add value?

Evidence from The Netherlands

Master Thesis August, 2018

Abstract

This study examines the corporate use of financial derivatives and firm value. Prior research

concerning the value of hedging is mainly focused on the U.S. due to data availability. This paper

aims to find additional empirical evidence by using hand-collected hedging data of Dutch firms.

Univariate tests and multivariate regression analyses are carried out with panel data

methodology including generalized least squares and fixed effects methods. The sample includes

all public non-financial Dutch firms with non-missing data during the period of 2012 to 2017.

Unlike previous studies, weak evidence is found of the existence of a hedging premium. The

results imply that nonfinancial Dutch firms can increase their value by hedging, but the impact is

close to zero. Hence, in the case of Dutch firms, hedging does not create shareholder value.

Student:

Yirong Lo (10753974)

MSc. Finance (track: Quantitative Finance)

Supervisor:

Derya Güler, MSc.

2

Statement of Originality

This document is written by student Yirong Lo who declares to take full responsibility for

the contents of this document.

I declare that the text and the work presented in this document are 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.

Acknowledgements

To mark the end of a meaningful graduate experience, I would like to take this moment

to express my gratitude for the ample opportunities that were given to me in the past four

years at the University of Amsterdam.

A notable highlight of my academic journey was an exchange semester in Singapore,

where I encountered the challenging field of Risk Management for the first time. This

exposure has ultimately led me to pursue this specific Master track, this thesis topic, and

perhaps the direction of a professional career in the future.

I would especially like to thank everyone who was involved in the completion of my

thesis. A special word of thanks goes to my supervisor, Derya Güler, for her

encouragement since the beginning of my thesis process. Furthermore, I would like to

thank colleagues of team Risk Solutions at KAS BANK N.V. for their flexibility in allowing

me to gain practical experience next to writing this thesis. Lastly, I am most grateful for

the continuous support from my friends and family.

3

Table of Contents

Statement of Originality .............................................................................................................................. 2

Acknowledgements ....................................................................................................................................... 2

1 Introduction .......................................................................................................................................... 4

2 Literature Review ............................................................................................................................... 7

2.1 Risk Management Theory ......................................................................................................... 7

2.1.2 Hedging can reduce real costs ........................................................................................ 8

2.1.1 Hedging can address agency problems ....................................................................... 9

2.1.4 Hedging can mitigate managerial risk aversion ...................................................... 9

2.2 Hedging and Firm Value........................................................................................................... 10

2.3 Hedging in the Dutch Market ................................................................................................. 11

2.4 Hypotheses .................................................................................................................................... 12

3 Methodology ........................................................................................................................................ 13

3.1 Empirical Framework ............................................................................................................... 13

3.2 Variables ........................................................................................................................................ 14

3.3 Diagnostic Tests .......................................................................................................................... 17

4 Data .......................................................................................................................................................... 19

4.1 Sources and Collection Procedure ....................................................................................... 19

4.2 Sample ............................................................................................................................................. 20

4.3 Descriptive Statistics ................................................................................................................. 21

5 Results .................................................................................................................................................... 23

5.1 Univariate Results ...................................................................................................................... 23

5.2 Regression Results ..................................................................................................................... 25

6 Robustness & Discussion .............................................................................................................. 28

7 Conclusion ............................................................................................................................................ 29

7.1 Implications and Suggestions ................................................................................................ 29

7.2 Limitations .................................................................................................................................... 30

References ...................................................................................................................................................... 31

Appendix ......................................................................................................................................................... 33

4

1 Introduction

The corporate use of derivatives for risk management comes in the form of hedging

against adverse market movements by taking offsetting positions in an underlying asset.

Such financial instruments include forwards, futures, options, and swaps. Historical

performance shows that large amounts of shareholder value have been destroyed due to

poor hedging programs. Spectacular losses have been realized as a direct result from

hedging. Real-world examples involve Metallgesellschaft AG in 1993 being forced to close

futures contracts when spot prices fell, resulting in a historic $1.3 billion loss. Another

example includes automotive giant, Daimler-Benz, suffering DM1.2 billion for failing to

hedge its dollar receivables when the USD fell by 14% in 1995.

There is no comprehensive framework in finance theory that explains the

rationale behind risk management. The Nobel-prize winning theory of financial

irrelevance of Modigliani and Miller (M&M, 1958) state that financial policies of the firm

are irrelevant in the absence of taxes, costs of financial distress, information asymmetries,

and transaction costs; and if investors can perform the same transactions as companies.

Risk management belongs to a firm’s financial policy. This implies that risk management

is irrelevant to a firm because shareholders can manage their own risk by holding

diversified portfolios.

In recent years, however, risk management has become a growing field of

expertise both in terms of size and importance. Corporations consider risk management

as one of their most important objectives (Rawls and Smithson, 1990). This development

is further accompanied by a rapid increase in the use of derivative securities – 1700% –

in the last two decades. According to the Bank for International Settlements1, the

derivatives market recorded total notional amounts outstanding of USD$638 trillion as

of 2017/2018. This is almost ten times the global GDP. It is safe to say that the use of

derivatives will continue to have an increasingly significant role in a firm’s overall risk

management policy. As a result, academic debate has sparked since 1990 in an effort to

explain this anomalous phenomenon.

1 https://www.bis.org/

5

Mayers and Smith (1982) published the earliest work suggesting possible

explanations for corporate risk management even when shareholders are likely

diversified. Their focus was on property and liability insurance, rather than on

derivatives, as the derivatives boom had not yet occurred. Numerous papers followed in

contributing to explain hedging motives by deviating from the M&M propositions. That

is, if financial risk management affects firm value, it must do so because of its impact on

taxes, financial distress costs, agency costs, or transaction costs.

Despite various established theories explaining how hedging could potentially

create shareholder value, improving the understanding of why firms may hedge, little

empirical evidence links hedging with firm value directly. This raises the question of

whether hedging achieves reasonable economic objectives, in other words: Does hedging

add value? This question is highly relevant for shareholders and the implications of risk

management as it continues to gain an important role in business operations and finance.

Prior to 1990, companies were not required to disclose risk management policies

because it was considered to be a part of the firm’s competitive strategy. This led to

limited empirical research in this field and reliance on survey data instead. After 2000,

hedging research has been concentrated on the U.S. as hedging data of U.S. firms gradually

became available following a change in the IFRS accounting standard. Until this day,

research on hedging and firm value is limited for countries outside North America. Hence,

The Netherlands will be the geographic focus of this study due to the lack of research and

existing literature. Moreover, the recent sample period is chosen to neglect the effects of

the 2008 Financial Crisis and its aftermath for future implications. Financial firms are

excluded due to its market-making role in derivatives.

The first to address a direct relation between hedging and firm value was

Allayannis and Weston (2001). They find that hedging increases firm value by 5% for a

sample of 720 non-financial firms in the US. On the contrary, Guay & Kothari (2003)

examined the economic effects of derivatives positions for a sample of nonfinancial firms

and concluded that potential gains of hedging are small. Their interpretation is that either

the observed increase in firm value is driven by other risk management activities or that

the correlation is spurious. Jin and Jorion (2004), on the other hand, find no relation

between hedging and firm value for US oil and gas producers.

6

The sample consists of 98 firms over which gives a total of 588 observations. Data

for hedging is manually collected from over 600 annual reports. Panel data methodology

is used involving univariate and multivariate analyses. With the ambiguous empirical

findings, it is evident that further research is required to explore the value relevance of

hedging in the case of Dutch firms. This thesis aims to provide useful insights into the

value implications of hedging by examining an unexplored geographic region and a more

recent sampling period.

The remainder of this paper is outlined as follows. Hypotheses are constructed

following a review of existing theories and prior research in Section 2. The research

methodology and sample are defined in Section 3 and 4, respectively. Section 5 presents,

interprets, and discusses empirical results. Followed by robustness tests in Section 6.

Finally, Section 7 concludes.

7

2 Literature Review

This section provides background knowledge and highlights the relevance of the main

research question. Related literature is reviewed in three subsections: 1. The rationale

behind corporate hedging, 2. The effects of hedging on firm value, and 3. Hedging in the

Dutch market. Followed by the construction of the main hypotheses.

2.1 Risk Management Theory

There is no comprehensive framework in finance theory that explains the

rationale behind risk management. According to the classic Modigliani & Miller (M&M)

propositions, financial decisions are irrelevant to a firm in perfect capital markets. The

Nobel-prize-winning M&M Irrelevance Theorem suggests that financial policies do not

affect firms if investors can perform the same transactions. This can be applied to risk

management as it is part of a firm’s financial policies (MacMinn, 1987). Moreover,

shareholders are perfectly able to manage their own risk. By holding diversified

portfolios, for instance, or by implementing a hedging strategy themselves. This implies

that corporate hedging should have no effect on shareholders’ value. In stark contrast to

this principle, corporations take risk management very seriously and consider it as one

of their most important objectives (Rawls and Smithson, 1990; Froot, Scharfstein, and

Stein, 1993). The renowned M&M theory combined with the growing development of risk

management and use of derivatives products has stimulated academic debate.

Prior to 1990, several studies examined how risk management may add value

relying on theoretical models and survey data. Models were constructed mainly by

introducing some friction to the perfect market assumptions in the M&M propositions.

That merely holds in the absence of taxes, financial distress costs, information

asymmetries, and transaction costs. When markets are imperfect, risk management

creates value by reducing the volatility of the firm’s cash flows (Smith and Stulz, 1985).

Schrand and Unal (1998) divide hedging research into two broad categories: 1. Papers

that identify market imperfections that make volatility costly, and 2. Papers that examine

why one method of reducing volatility is cheaper than another. From those papers, a

theoretical framework is constructed stressing three main reasons how hedging can alter

firm value. Namely, hedging could create shareholder value by: reducing real costs,

8

addressing agency problems, and mitigating managerial risk aversion. These theories are

briefly reviewed below.

2.1.2 Hedging can reduce real costs

There is an absence of asymmetric information in the M&M environment. That is,

all market participants are able to assess costs correctly. Such a world allows for

“optimal” transactions because principal-agent problems do not exist when all parties

have the same information. Managers aim to maximise value for shareholders and

shareholders are able to diversify the risks of the corporations on their own. Therefore,

the only way that hedging can add value is if it can reduce real costs. Existing literature

has identified two such gains: 1) Hedging can reduce expected tax liabilities; 2) Hedging

can reduce financial distress costs.

Smith and Stulz (1985) assume that bankruptcy involves some exogenous

transactions costs; a violation of the M&M assumption. They found a higher probability

of financial distress if a firm does not hedge as well as higher costs incurred if it does

encounter financial distress. Thus, hedging adds value by reducing the probability of not

being able to repay debt. If financial distress costs are high, hedging may be used to

increase debt capacity implying that corporate hedging increases with the probability of

distress (Smith and Stulz, 1985).

In addition, Smith and Stulz (1985) argue that, if taxes are a convex function of

earnings, it is generally optimal for firms to hedge. Convexity implies that volatile

earnings lead to higher expected taxes, which is plausible for some firms. In particular,

those who face a significant chance of negative earnings and are unable to carry tax losses

to subsequent periods. Hence, hedging increases firm value by reducing the present value

of future tax liabilities. Graham and Rogers (2002) find evidence that firms hedge to

increase debt capacity and interest deductions. More specifically, they estimate that

hedging adds 1.1% to firm value through tax benefits.

Apart from hedging incentives resulting from convexity, there is a significant

indirect tax incentive to hedge (Froot et al., 1993). Hedging can reduce taxes by providing

the opportunity to increase leverage. Trade-off between tax benefits and bankruptcy

costs produce an optimal capital structure (Ross, 1996). In his study, Ross derives an

optimal hedge portfolio and concludes that a firm that hedges its risk increases its optimal

9

amount of debt, therefore, realizes more tax benefits from leverage. He was the first to

assign numerical values relating hedging with firm value saying that hedging is worth an

extra 10% - 15% for current shareholders. This claim is later confirmed by empirical

studies in the US2.

2.1.1 Hedging can address agency problems

Agency problem is a common topic in Corporate Finance that arise when there is

a conflict of interest between shareholders and other stakeholders, i.e. bondholders. One

such problem is when managers pass on valuable investment opportunities because

debtholders would capture a portion of the benefits, leaving insufficient returns to

shareholders. This is also known as the underinvestment problem put forward by Myers

(1977). Another example is the risk-shifting problem, where high-risk projects are

preferred over low-risk projects – at the expense of debtholders – to increase returns for

shareholders (Jensen and Meckling, 1976). Either case shows that agency conflicts exist

when a company is (partially) debt financed.

Froot, Scharfstein, and Stein (1993) developed a general framework for analysing

corporate risk management policies. The authors observe that there is a benefit to

hedging if external sources of financing are more costly than internally generated funds.

In this case, hedging adds value due to explicit and implicit cost savings by ensuring that

a firm has sufficient internal funds available to take on profitable investment

opportunities. Therefore, hedging reduces the probability that a corporation will have to

engage in costly external financing (Froot et al., 1993). Consequently, hedging also

reduces the probability that a firm would forego profitable investments due to a lack of

internal funds. Leland (1998), in addition, examines risk management and find that

hedging permits greater leverage. His evidence indicates that hedging benefits are higher

when agency costs are low. In other words, hedging can create shareholder value by

reducing the need for debt financing.

2.1.4 Hedging can mitigate managerial risk aversion

The previous theories assumed that managers act in the best interest of shareholders. In

addition to reducing potential financial distress costs and tax expenses, Stulz (1985)

suggests that risk aversion of managers drives corporate hedging. While outside

2 See Dolde (1995); Géczy, Minton, and Schrand (1997); Hentschel and Kothari (1995); Tufano (1996).

10

stockholders are able to diversify their portfolio risk, making them indifferent to a firm’s

hedging activity, the same cannot be said for managers who often hold a large amount of

their wealth in the firm’s stock. Stulz (1985) argues that managers are strictly better off,

without costing outside stockholders anything, by reducing the variance of total firm

value.

Tufano (1996) obtains little empirical support for the predictive power of theories

that view risk management as a means to maximize shareholder value. He does find

evidence that firms whose managers hold more stock hedge more gold price risk. Thus,

suggesting that risk management may be affected by managerial risk aversion. He further

observes a negative relation between the tenure of CFO’s and risk management, which

perhaps reflects managerial interests, skills, or preferences.

Another managerial theory of hedging, based on asymmetric information, is

explored by Breeden and Viswanathan (1990) and DeMarzo and Duffie (1992). Their

models suggest that the labour market revises the ability of managers based on firm

performance. This can result in some managers undertaking hedges in an attempt to

influence the labour market’s perception.

2.2 Hedging and Firm Value

Despite the range of theories explaining the rationale of hedging, there is little

empirical evidence that directly relates hedging with firm value. Early studies on risk

management practices were restricted due to data availability. Companies were not

required to disclose risk mitigation policies as risk mitigation was considered to be a

firm’s strategy competitive. The rapid rise of derivatives usage has changed accounting

standards requiring firms to expose their risk management policies. This change has

allowed a new generation of research with derivatives use as the explanatory variable.

Among the earliest studies on a direct relation between hedging and firm value is

by Allayannis & Weston (2001). They found that firm value, measured in Tobin’s Q ratio,

is 5% higher for firms that hedge currency risk in a sample of 720 large US firms between

1990 and 1995. With a median market value of $4 billion, this translates into an average

value added of almost $200 million for firms using foreign currency derivatives. On the

contrary, Guay & Kothari (2003) examined the economic effects of derivatives positions

11

for a sample of nonfinancial firms and concluded that potential gains of hedging are small.

Their interpretation is that either the observed increase in firm value is driven by other

risk management activities or that the correlation is spurious. Additionally, Jin & Jorion

(2006) verified that hedging reduces the firm’s stock price sensitivity but does not affect

firm value for a sample of 119 US oil and gas producers. A more recent paper by

Panaretou (2014) investigated 1372 large non-financial UK firms. He finds that the

hedging premium is statistically and economically significant for foreign currency

derivative users. Consistent with this, Bua et al. (2015) find an average hedging premium

of 1.53% with respect to company value (Tobin’s Q) in a sample of 400 Spanish firms.

Overall, the empirical evidence regarding the effect of hedging on firm value remains

inconclusive. A full overview of the existing research can be found in Appendix 1 (p.33).

2.3 Hedging in the Dutch Market

To the best of our knowledge, hedging and firm value have not yet been

investigated concerning the Dutch market mainly due to data availability. As of 2007,

Dutch firms are required to disclose information about the significance of financial

instruments to an entity, and the nature and extent of risks arising from those financial

instruments, both in qualitative and quantitative terms (IFRS 7, 2007). Unlike the US

GAAP, the IFRS does not require companies to disclose notional amounts of derivatives.

To this date, this has undoubtedly restricted the extent of hedging research in European

markets. This paper aims to contribute to the lack of empirical evidence in Europe by

investigating firms in The Netherlands.

Bodnar, de Jong, and Macrae (2003) compared derivatives usage of U.S. and Dutch

firms and found institutional differences. Their results indicate that Dutch firms hedge

more financial risk than U.S. firms which can be explained by the greater openness of The

Netherlands. Dutch companies experience far more foreign exchange exposure (Bodnar

et al., 2003). Hence, Dutch firms hedge more currency risk. Whereas US firms are more

careful with the use of derivatives due to stricter disclosure requirements in the US.

Moreover, Bodnar et al. (2003) argue that U.S. firms focus more on accounting earnings,

which may be attributable to the importance of shareholder value in the US versus the

stakeholder value in The Netherlands. Another institutional difference is that Dutch firms

rely more on over-the-counter transactions. While US firms use exchange-traded

12

derivatives more. Therefore, U.S. firms require a higher credit rating for derivatives

transactions. These findings prove the existence of institutional differences in the

financial environments between the U.S. and The Netherlands. Therefore, the effect of

risk management practices and derivatives use may be significantly different between

U.S. and Dutch firms.

2.4 Hypotheses

The objective of this paper is to investigate whether hedging is a value-adding activity for

Dutch firms. The diverse results in Section 2.2 lead to the following formulation of the

hypotheses. If hedging adds value, hedgers should have higher firm value on average.

Hence, the first hypothesis tests the difference in firm value between hedgers and non-

hedgers.

Hypothesis 1: Hedgers and non-hedgers are equally valued (μHedgers = μNon-Hedgers)

The second hypothesis identifies a causal relationship between hedging and firm value

by testing whether hedging increases firm value. In other words, the effect of hedging

should not equal zero if hedging adds value.

Hypothesis 2: Hedging does not increase firm value (β1 = 0, θ1 = 0 )

These hypotheses is tested following the approach of Allayannis & Weston (2001) and

Magee (2009) explained in the next section. This research aims to contribute to existing

literature by providing new evidence on a different geographic region and time period

employing a unique dataset.

13

3 Methodology

This section describes the empirical approach of the research method. The framework is

based on various prior studies explaining the impact of hedging on firm value. Followed

by diagnostic tests to confirm fundamental model assumptions. Furthermore,

explanations of the choice and construction of variables are provided.

3.1 Empirical Framework

3.1.1 Univariate Analysis

A straightforward way to test whether hedgers have higher firm value than non-

hedgers is to compare the two groups. By splitting the sample into hedgers and non-

hedgers, the difference in mean values can be tested using simple t-tests. In economic

terms, the existence of a difference in firm value is a hedging premium (or discount).

Although univariate analysis is performed in numerous related articles3, it cannot test

cause and effect relationships because only one variable is analyzed at a time. For this

reason, the remainder of the methodology consists multivariate regression analyses.

3.1.2 Regression Analysis

This study aims to investigate whether hedging with financial derivatives is a value-

adding activity. In order to investigate the value implications of hedging, other factors

affecting firm value must be controlled for. Similar to previous studies3, I make use of

multiple linear regressions with panel data to test the relationship between hedging and

firm value. There are various benefits of using panel data. It is more informative than

cross-sectional or time-series data as it greatly increases the number of observations for

this research. Moreover, panel data considers heterogeneity and allows for more

variation as well as increases degrees of freedom. It also shows less collinearity and

makes the results more generalizable. The regression equations look as follows:

ln(FVi,t) = α + β1𝐻𝑒𝑑𝑔𝑖𝑛𝑔 𝐷𝑢𝑚𝑚𝑦i,t + 𝛾X′i,t + εi,t

ln(FVi,t) = δ + θ1𝐴𝑚𝑜𝑢𝑛𝑡 𝑜𝑓 𝐻𝑒𝑑𝑔𝑖𝑛𝑔i,t + 𝜆X′i,t + ωi,t

3 Allayannis and Weston (2001); Allayannis et al. (2012); Panaretou (2013); Ayturk et al. (2016)

(1)

(2)

14

Where FVi,t is a measure of firm value for firm i in year t. Hedging, Hi,t, is the main

explanatory variable and represented by a hedger variable (dummy) or, alternatively, the

notional amount of hedging (continuous variable). X’i,t is a vector of all control variables

which are explained in the next section. The intercepts are α and δ. The error terms are

εi,t and ωi,t, which represent the residual part of firm value that is not explained by the

independent variables. In other words, β1 is the change in the log of firm value as a result

of whether the firm hedges. Whereas θ is the change as a result of a one-unit change in

hedging.

3.2 Variables

3.2.1 Dependent Variable

The dependent variable is firm value, which in theory and practice can be

measured in numerous ways. The primary objective of a firm is to maximize value for its

shareholders and/or stakeholders. In the case of public firms, this can be calculated from

the share price or book values on the balance sheet. Furthermore, the efficient market

hypothesis suggests that share prices fully reflect all available information. Hence, the

market value of equity should reflect shareholder value as it is forward-looking, risk-

adjusted and less susceptible to changes in accounting practices (Wernerfelt and

Montgomery 1988). Following Panaretou, we make use of the Chung and Pruitt (1994)

approximation of Tobin’s Q as they have shown that there is a high degree of correlation

between this simple construction of Tobin’s Q and more rigorous approximations as a

proxy for firm value. Hence, the market value of the firm is calculated as the market value

of equity plus the liquidation value of firm’s outstanding preferred stock and total debt.

FV1 = ln (𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐸𝑞𝑢𝑖𝑡𝑦 + 𝑃𝑟𝑒𝑓𝑒𝑟𝑟𝑒𝑑 𝑆𝑡𝑜𝑐𝑘 + 𝑇𝑜𝑡𝑎𝑙 𝐷𝑒𝑏𝑡

𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠)

To check the robustness of our results, we employ two alternative measures for firm

value. Another approximation for firm value, following Allayannis and Weston (2001), is

calculated as the ratio of the book value of total assets minus the book value of equity plus

the market value of equity to the book value of total assets (FV2).

FV2 = ln (𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 + 𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐸𝑞𝑢𝑖𝑡𝑦 − 𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐸𝑞𝑢𝑖𝑡𝑦

𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠)

15

Additionally, an industry-adjusted firm value measure is constructed. This is relevant

given the large percentage of industrially diversified firms in the sample. Following

Panaretou (2013), industry-adjusted firm value is calculated as the difference between

the individual firm value and the industry median, both log-transformed.

𝐹𝑉3 = 𝑙𝑛(𝐹𝑉1𝑖) − 𝐼𝑛(𝐹𝑉1𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦−𝑚𝑒𝑑𝑖𝑎𝑛)

3.2.2 Main Independent Variable

The main explanatory variable is hedging measured by the use of financial

derivatives. Dutch firms follow the accounting framework of the International Financial

Reporting Standards (IFRS) as adopted by the EU. The IFRS “requires disclosure of

information about the significance of financial instruments to an entity, and the nature

and extent of risks arising from those financial instruments, both in qualitative and

quantitative terms” (IFRS 7, 2007). More specifically, effective as of 2005, it is required

that all derivatives are marked-to-market with changes in the mark-to-market being

taken to the profit and loss account (IAS39, 2003). Thus, in addition to reporting risk

exposures and mitigation strategies, the IFRS requires firms to disclose derivative

positions in terms of fair value. The fair value of a financial instrument is the mark-to-

market price at which it is traded on the day of valuation. Hence, it greatly fluctuates

depending on the underlying asset and maturity of the contract. Since financial

statements merely document values at one specific moment in time, it does not reflect the

annual amount of derivatives usage. Previous research on European countries under IFRS

have all used hedging dummies for this reason. However, I have found quite a large

number of Dutch firms that report the principal notional amounts of outstanding

derivative contracts voluntarily. The notional value is the total amount of a security’s

underlying asset at its spot price at the time the contract was entered into. This is a

comparable measure of units that is constant over time. Therefore, allowing for the

creation of a continuous variable for hedging in monetary terms which is a more efficient

and accurate measure. Hence, in this research, hedging is measured in two ways:

1. A dummy variable that equals 1 if a firm hedges risk with interest rate, currency,

and/or commodity derivatives; 0 otherwise.

2. The principal notional amount of the derivatives position.

16

3.2.3 Control Variables

Firm Size – Intuitively, larger firms are expected to have higher firm value simply for

having more assets and/or revenue. Empirical evidence regarding the effect of firm size

on firm value, however, remains inconclusive. It is still important to control for size

because larger firms might have more resources for hedging than smaller firms.

Following Allayanis and Weston (2001), firm size is controlled for by taking the natural

logarithm of total assets.

𝑆𝑖𝑧𝑒 = 𝑙𝑛 (𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠)

Profitability – Shareholders naturally value profitability because it generates return.

Hence, more profitable firms are expected to have a higher firm value. Profitability can

be measured by net income scaled by total assets, also known as ROA.

𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 = 𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒

𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠

Leverage – Capital structure may be related to firm value in accordance with the trade-off

theory of leverage (Kraus and Litzenberger, 1973) . This is because firms can choose how

much debt and equity to use for financing. Hence, firms can optimize its value by

balancing cost of capital and tax benefits of debt. A straightforward proxy for leverage is

the debt to assets ratio:

𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 =𝑇𝑜𝑡𝑎𝑙 𝐷𝑒𝑏𝑡

𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠

Growth Opportunities – Similar to profitability, shareholders value a firm’s growth

potential for its future returns. Firm value is expected to be higher for firms with more

investment opportunities. A proxy for future investment opportunities is capital and R&D

expenditure divided by total assets.

𝐺𝑟𝑜𝑤𝑡ℎ 𝑜𝑝𝑝𝑜𝑟𝑡𝑢𝑛𝑖𝑡𝑖𝑒𝑠 =𝐶𝑎𝑝𝑒𝑥 + 𝑅&𝐷 𝐸𝑥𝑝𝑒𝑛𝑑𝑖𝑡𝑢𝑟𝑒

𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠

International Diversification – Firms operating in multiple countries are diversified

against geographic factors, agency problems. controlled for by the percentage of foreign

sales out of total sales.

𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛 =𝐹𝑜𝑟𝑒𝑖𝑔𝑛 𝑆𝑎𝑙𝑒𝑠

𝑇𝑜𝑡𝑎𝑙 𝑆𝑎𝑙𝑒𝑠

17

Liquidity – Firms that are more liquid are better able to repay debt, hence, less likely to

go bankrupt. Shareholders are expected to value this aspect positively. Liquidity is

measured as the ratio of cash and cash equivalents to current liabilities.

𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 =𝐶𝑎𝑠ℎ 𝑎𝑛𝑑 𝐶𝑎𝑠ℎ 𝐸𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡𝑠

𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠

Capital Constraint – Access to financial markets is argued to affect a firm’s investment

decisions. If a firm is capital constrained, it can only take on projects with the highest

NPV’s. A dividend dummy that equals 1 if the firm issued dividends in the current year is

used to proxy a firm’s capital constraint. Dividends can be viewed as a positive signal from

management. Hence, the sign of the coefficient may be negative or positive.

Time effects – Year dummies are used in all regressions to control for time effects.

Furthermore, industry effects are controlled for by the firm value variable adjusted for

industry. Unlike Allayannis and Weston (2001), business diversification and credit rating

controls are not included in the model. The former due to presence of multicollinearity

with other dummies. The latter because adding seven credit dummies may reduce the

power of the test, given the relatively small sample size in the multivariate regression

analysis.

3.3 Diagnostic Tests

As described in the methodology above, this research makes use of econometric models

involving linear regressions. It is therefore important to test certain primary assumptions

of linear regression. So that adjustments can be made accordingly when assumptions are

violated.

3.3.1 Linearity

One of the key assumptions of a linear regression is that the relationships between

the dependent and independent variables are linear. To detect non-linearity, joint Wald

tests are performed for the parameters of each variable (see Appendix 3).

3.3.2 Homoscedasticity

OLS estimation assumes that the variance of the error term is constant.

Homoscedasticity can be checked by plotting the residuals of the regression model

against its fitted values. The graph shows a clear clustered pattern and indicates

18

heteroscedasticity (see Appendix 4). Hence, robust standard errors are used in all

regressions hereafter to allow for correlation among the residuals.

3.3.3 No Autocorrelation

The Wooldridge test for autocorrelation in panel data is performed (see Appendix

5). Serial correlation in the residuals is found. Hence, to allow for correlation across

panels, generalized least squares method is used instead of ordinary least squares.

3.3.4 Normality

Linear regressions require all variables to be normally distributed. Histograms of

all variables are provided in Appendix 6. As expected, firm value and total assets are not

normal, hence log-transformed.

3.3.5 No Multicollinearity

Multicollinearity is present when there is a strong correlation between the

independent variables. A Pearson correlation matrix including all the independent and

control variables is presented in Appendix 7. A commonly used rule of thumb is that

correlation coefficient between two explanatory variables greater than 0.8 indicates a

strong linear association. The correlations between our variables are generally small.

Hence, there is no indication of multicollinearity. None of the variables have a correlation

above 0.8.

3.3.6 Fixed Effects versus Random Effects

Panel data analysis requires assumptions of its model parameters, generally

Pooled OLS, Random Effects, or Fixed Effects. Prior studies have run regressions with the

Fixed Effects model to control for unobservable firm characteristics that may affect value.

To confirm this model assumption, I test Fixed Effects against Random Effects with a

Hausman test (see Appendix 8). The result of the Hausman test suggests that Fixed Effects

is indeed preferred. Hence, fixed firm effects are used in all regressions.

19

4 Data

Panel data is obtained from Compustat Global, Datastream, and annual reports. The

dataset includes all publically-listed Dutch non-financial firms with non-missing values

between 2012 and 2017. The final sample consists of 98 individual firms covering 6 fiscal

years, which results in 588 firm-year observations.

Prior research on the relationship between hedging and firm value was restricted

mainly due to data availability. Companies were not required to disclose risk mitigation

strategies before 1990. This led to a reliance on survey data instead. Moreover, The

Netherlands is the geographic focus of this study due to the lack of research and existing

literature. The recent sample period is chosen to neglect the effects of the 2008 Financial

Crisis and its aftermath for future implications. Financial firms are excluded due to its

market-making role in derivatives.

4.1 Sources and Collection Procedure

4.1.1 Compustat and DataStream

First, the ISO country code of incorporation (FIC) is set equal to The Netherlands.

This classifies a firm as Dutch when they are legally registered in the country, without

requiring to have operations in The Netherlands or be listed on a Dutch stock exchange.

Secondly, non-missing data over the fiscal years of 2012 to 2017 is required for a

balanced panel. This means that firms with missing values are dropped, i.e., due to late

IPO, early bankruptcy, or being acquired between 2012-2017. Thirdly, financial

institutions are excluded by dropping banks and insurers, rather than dropping the entire

financial sector. This would otherwise exclude real estate firms and other funds that are

not market makers in derivatives but do use it for hedging. Similar to Ayturk et al. (2016),

no firm size threshold is included since it would significantly decrease the sample size. A

full list of the 98 Dutch firms can be found in Appendix 2, p. 36.

The following standard annual accounting variables are obtained from Compustat:

total assets, revenue, net income, stockholders equity, long-term debt, short-term debt,

current liabilities, R&D expenses, capital expenditure, and dividends. Additional data, not

available in Compustat, are retrieved from DataStream including market value of equity

20

(=market capitalization), foreign sales/total sales ratio, and multiple SIC codes per

company.

4.1.2 Hedging Data

Data for hedging is hand-collected from annual reports obtained from company

websites. Over 600 documents are manually scanned with search terms such as ‘hedg’,

‘notional’, ‘futures’, ‘forward contract’, and ‘swap’. When such terms are found, the text is

carefully read to confirm that the derivatives were used for hedging purposes. Four

dummy variables and, when available, principal notional amounts are collected. The four

hedging dummy variables are created as follows:

- Interest rate hedger = 1 if the firm hedged interest rate risk with interest rate

derivatives in that year; 0 otherwise.

- Currency hedger = 1 if the firm hedged currency risk with currency derivatives in

that year; 0 otherwise.

- Commodity price hedger = 1 if the firm hedged commodity price risk with

commodity derivatives in that year; 0 otherwise.

- Hedger = 1 if the sum of the three aforementioned dummies is greater than 0; 0

otherwise.

4.2 Sample

Merging the datasets from Compustat, DataStream, and annual reports resulted in

98 unique firms with 6-year non-missing values, thus 588 firm-year observations. An

overview of the sample is provided in Table 1. Roughly 60% of the observations use

financial derivatives for hedging purposes. These hedgers are further segmented into

currency hedgers (80%), interest rate hedgers (68%), and commodity price hedgers

(27%). The distribution of hedging over the years can be found in Appendix 3, p.37.

In line with previous findings, firms use currency derivatives the most. Currency

risk being the most hedged could be explained by the fact that The Netherlands has an

open economy and depend heavily on foreign trade with a current account surplus. This

is consistent with the findings of Bodnar et al. (2003). The largest industries in The

Netherlands by the number of observations are Manufacturing (276) and Services (150).

The smallest number of observations are found in Mining (18) and Wholesale Trade (18).

Moreover, firms in the Utilities segment are all hedgers.

21

Table 1. Sample Overview The sample includes all publicly-listed non-financial Dutch firms with non-missing data during the years 2012 to 2017. A firm is classified as hedger if it reported hedging interest rate, currency, and/or commodity price risk with derivatives in that year. Non-hedgers are firms that do not make use of any financial derivatives for hedging purposes in that year. Industries are classified according to 2-digit SIC codes.

Hedgers Non-Hedgers N Currency (%) Interest rate (%) Commodity price (%) (%) By industry:

Mining 18 66.67 55.56 11.11 11.11 Construction 30 43.33 73.33 20.00 23.33 Manufacturing 276 60.87 44.57 29.71 29.35 Utilities 30 93.33 80.00 23.33 0.00 Wholesale Trade 18 100.00 72.22 0.00 11.11 Retail Trade 42 57.14 54.76 0.00 30.95 Real Estate 24 8.33 8.33 0.00 87.50 Services 150 16.00 19.33 0.00 74.00

Total 588 80.34 68.26 27.25 40.31

Note that multiple types of risk can be hedged simultaneously. An example to interpret the table: 40.31% of the total sample are non-hedgers. This means that 59.69% are hedgers. From the total amount of hedgers, 80.34% hedge currency risk. Thus, approximately 48% of the total sample uses currency derivatives for hedging purposes (0.5969×0.8034=0.4795).

4.3 Descriptive Statistics

Summary statistics of relevant variables are shown in Table 2. Firms in the sample

have a mean (median) market cap of €4,264 (€404) million and mean (median) value of

assets of €9,983 (€625) million. This is similar to the samples in previous studies focusing

on the US, Spanish, and French markets, except that small firms have not been excluded

which is similar to Ayturk who studied the Turkish market. The average D/E ratio is 68%,

34% issue dividends, average profit margin of -0.35%. This is also similar to the sample

in previous studies. Additionally, it can be seen that median values are smaller than mean

values. For example, the relatively large difference in mean and median of total assets

indicates that there are more smaller firms in the sample than larger firms. The same is

true for total sales and market cap. This skewness is corrected for by log-transformation

of the variables.

Out of the 588 observations, 351 are classified as hedgers. Total notional

derivatives amounts are recorded for 434 observations. Thus, the sample size is smaller

for the regression with the continuous hedging variable. On average, a Dutch firm hedges

€1.97 billion per year in total notional amounts. Hedgers that did not report notional

22

values are accepted as missing values. The low hedging medians are due to non-hedgers

(40%) having a notional derivatives value of 0, in line with the hedging dummies.

Table 2. Descriptive Statistics This table presents summary statistics that describe the sample, including all variables. The sample includes all public non-financial Dutch firms with non-missing data from 2012-2017. Firm characteristics are directly obtained from Compustat and DataStream. The dependent variable is firm value, measured in four ways as described on p.14. Detailed construction of the control variables are explained on p.15&16. Data of hedging variables are manually collected from annual reports. N Mean Median St. Dev. Min Max Firm characteristics Total assets (€ mln) 588 11,944.26 663.19 44815 0.007 565258 Total sales (€ mln) 588 12,816.72 793.81 81593.39 0 1295008 Market Cap (€ mln) 588 4419.06 405.85 9904.83 0.95 86096.44 Business Diversification (dummy) 588 0.6531 1 0.4764 0 1 Dependent variable FV1 588 0.1433 0.0629 0.9113 -1.8072 4.7885 FV2 588 0.4589 0.3303 0.7648 -1.5238 4.7958 FV3 588 0.1092 7.45e-9 0.9002 -1.9232 4.7861 Hedging variables Hedger dummy 588 0.5969 1 0.4909 0 1 Interest rate hedger dummy 588 0.4184 0 0.4937 0 1 Currency hedger dummy 588 0.4915 0 0.5004 0 1 Commodity price hedger dummy 588 0.1650 0 0.3715 0 1 Total Hedging (€ mln) 434 343.4272 0 10544.84 0 107909 Interest Rate Hedging (€ mln) 510 343.43 0 1232.93 0 12400 Currency Hedging (€ mln) 508 1,509.2 0 8696.50 0 99606 Commodity price hedging (€ mln) 540 12.333 0 77.62 0 765 Control variables Firm Size 588 6.3844 6.4971 2.9126 -4.9618 13.245 Profitability 588 -0.1289 0.0229 1.0232 -7.9061 1.7180 Leverage 588 0.3011 0.2172 0.6505 0 9.1660 Growth Opportunities 588 0.0796 0.0422 0.2023 0 3.9239 International Diversification 588 0.4604 0.5255 0.4107 0 1 Liquidity 588 0.6422 0.2198 1.4206 -0.6512 9.5911 Dividend dummy 588 0.3384 0 0.4736 0 1

The 1% extreme outliers of the dependent variables is winsorized by replacement.

23

5 Results

This section presents empirical results from the analyses that test the main hypothesis.

That is, whether hedging adds value to shareholders. Two approaches are used: 1.

univariate tests to compare hedgers with non-hedgers, and 2. Multivariate regression

analysis to test the effect of hedging on firm value.

5.1 Univariate Results

The main hypothesis that firms using derivatives for hedging are rewarded by

investors with higher valuation is tested here. Hedgers and non-hedgers are split using

the hedger dummy variable using four measures of firm value. A firm is classified as

hedger if it reported in that year to have used derivatives for hedging purposes. Firm

value is compared between hedgers and non-hedgers using two-sample t-tests of mean

values, as shown in Table 3.

Table 3. Comparison of Firm Value between Hedgers and Non-hedgers This table presents a univariate comparison of firm value (FV) between hedgers and non-hedgers with three different measures. The construction of the three measures of firm value are explained in detail on p.14. A firm is classified as hedger for a given year if it reports to have managed risk with the use of derivatives. The sample consists of 588 observations and includes all public non-financial Dutch firms in the period 2012-2017. Hedgers Non-Hedgers Difference t-value General hedging FV1 0.0422 0.2930 -0.2508*** -3.3010 FV2 0.3714 0.5886 -0.2172*** -3.4086 FV3 0.0496 0.3020 -0.2524*** -3.9798

Currency hedging FV1 0.0650 0.2190 -0.1540** -2.0537 FV2 0.3885 0.5270 -0.1385** -2.2021 FV3 0.0499 0.2494 -0.1995*** -3.1918

Interest rate hedging FV1 0.0418 0.2163 -0.1745** -2.2990 FV2 0.3946 0.5052 -0.1106* -1.7326 FV3 0.0798 0.2028 -0.1230* -1.9310

Commodity price hedging FV1 -0.0731 0.1860 -0.2591** -2.5715 FV2 0.3161 0.4871 -0.1710** -2.0175 FV3 -0.0334 0.1878 -0.2213*** -2.6208

Difference in the means are compared using t-tests. The 1% outliers of Firm Value are winsorized by replacement. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

24

Table 3 reports differences in mean firm values and related t statistics obtained from

twelve t-tests. The hypothesis is tested separately for general hedging, currency hedging,

interest rate hedging, and commodity price hedging. Four types of hedging are examined,

each with three measures of firm value. General hedging represents firms that reported

to have used currency, interest rate, and/or commodity derivatives for hedging purposes

in a given year. Whereas currency hedging, interest rate hedging, and commodity price

hedging represent the corresponding type of hedging separately.

The univariate results suggest that hedgers have lower firm value in comparison

to non-hedgers. This is true for all three different measures of firm value and indicates

some robustness. The difference in firm value between firms that use either type of

hedging (General hedging) as opposed to firms that do not hedge at all is -0.25, -0.22, and

-0.25 for all three measures respectively. In economic terms4, this translates into value

differences of -22.18%, -19.52%, and -22.31% resulting in an average hedging discount

of -21.34%. This finding rejects the hypothesis at a 1% significance level and suggests

that hedgers are valued lower than non-hedgers. More specifically, using the same

calculation method, the average hedging discounts are -15.09% for currency hedging, -

12.68% for interest rate hedging, and -19.47% commodity hedging.

From the three types of hedging, commodity price hedging seems to be the most

negative. Commodity derivatives are also the least used for hedging, as shown in the

sample overview (Table 1). A plausible reason for this may be that commodities are often

part of a firm’s core business. Therefore, fluctuations in commodity prices would impact

the firm’s core business more than currency or interest rate changes would in general.

Another reason why the results of commodity price hedging deviate from currency and

interest rate hedging could be due to the small sample size. The sample records only 16%

of commodity hedgers. While roughly 50% and 40% is reported for currency and interest

hedgers.

Overall, the univariate findings are vastly inconsistent with numerous studies that

infer a valuation premium imposed by the use of derivatives. An explanation may be due

to the existence of large differences in structural characteristics as well as market

valuation of Dutch and U.S. firms. Instead, evidence of a hedging discount is in line with

4 Calculation of hedging discount e.g. (e-0.2508-1×100%) = -22.18%

25

Khediri (2010) and, to an extent, supports Jin & Jorion (2006) and Ayturk et al. (2016).

Khediri (2010) argues that a hedging discount exists due to investors’ perception of

corporate hedging decisions which may be linked to insider’s motives and their risk

aversion. Hence, investors value the derivatives use at a discount.

To further analyze the differences between hedgers and non-hedgers, an

additional comparison between hedgers and non-hedgers are drawn from other

variables (see Appendix 9, p.38). The results in A3 show that the characteristics of

hedgers and non-hedgers are significantly different within the sample. More specifically,

hedgers are larger in size; more profitable and diversified; and more likely to issue

dividends. Whereas non-hedgers are more leveraged, liquid, and have more growth

opportunities. Differences in these variables could also explain why hedgers are valued

lower by shareholders. So it is important to further investigate with regression analyses

in order to identify a causal relationship.

5.2 Regression Results

The findings in the univariate tests suggest that hedgers are valued lower by

investors than non-hedgers. In this paragraph, linear regressions are performed to

control for other variables that may have impact on firm value. To estimate the

coefficients of Equations 1 & 2, two panel regression models are used. Model 1 regresses

on the Hedging Dummy and estimates with generalized least squares method. Whereas

Model 2 regresses the Amount of Hedging with fixed firm effects method. Following

previous studies, the model includes various control variables and year dummies to fix

for time-specific effects as well.

For a clear appearance, one measure of firm value (FV1) as dependent variable is

presented here. The other two measures are presented in the robustness section later.

The number of observations for notional amounts of hedging (434) is lower than that of

hedging dummies (588). This is due to some firms that do not report notional values of

derivatives usage although they do report hedging activity. In such cases, the notional

value of derivative instruments are accepted as missing data. Additionally, for some

observation years, multiple types of hedging are executed, but not all notional values are

reported. Therefore, the number of observations in Model (2) for General Hedging is

lower while it represents the total sample. Regression results are indicated in Table 4.

26

Table 4. Effect of Hedging on Firm Value This table presents panel regression results. Model 1 is estimated with Generalized Least Squares method on a hedging dummy that equals 1 if the firm has reported the use of derivatives for hedging in a given year. While Model 2 regresses on the total notional amount of hedging reported in a given year with Fixed Effects method. Firm Value is calculated as the natural log of [(MV Equity + Preferred Stock + Total Debt)/BV Assets]. Construction of control variables are described on p. 16&17. Year dummies are included to control for time effects. General Hedging Currency Hedging Interest Rate Hedging Commodity Price Hedging

Dependent variable: Firm Value (1) (2) (1) (2) (1) (2) (1) (2) Hedging (Dummy) 0.0063 0.5171 0.0949 -0.0960 (0.01) (0.85) (0.23) (-0.14) Amount of Hedging (Notional) 7.87e-6*** 7.51e-6** 3.13e-5** -0.0006*** (2.74) (2.41) (2.37) (-3.16) Firm Size -0.3590*** -0.5667*** -0.3879*** -0.5610*** -0.3512*** -0.5569*** -0.3571*** -0.5509*** (-4.24) (-12.22) (-4.63) (-12.66) (-3.95) (2.37) (-4.07) (-12.57) Profitability -0.5982* -0.0804 -0.6289** -0.0948 -0.6188* -0.1011 -0.5973* -0.1126 (-1.77) (-0.38) (-2.20) (-0.45) (-1.92) (-0.47) (-1.87) (-0.53) Leverage 1.5263 0.1899*** 0.0879 0.1921*** 1.4899 0.1891*** 1.4696 0.1914*** (1.46) (8.53) (0.16) (8.45) (1.51) (7.74) (1.47) (7.88) Growth Opportunities -1.3369 0.2272 -1.1952 0.2710* -1.3934 0.2524 -1.3618 0.2025 (-1.38) (1.51) (-1.10) (1.78) (-1.51) (1.60) (-1.48) (1.30) International Diversification 0.9411* 0.3235** 1.4401*** 0.2967** 0.9763** 0.1811 0.9903** 0.1643 (1.89) (2.08) (4.15) (2.01) (2.31) (1.21) (2.12) (1.22) Liquidity 0.4009* -0.0331** 0.2706 -0.0312* 0.4128* -0.0259 0.4023* -0.0265 (1.81) (-2.11) (1.12) (-1.95) (1.94) (-1.51) (1.86) (-1.55) Dividend Dummy 0.9854*** 0.1315 0.7681 0.1288 0.9854*** 0.1303* 0.9783*** 0.1323* (3.62) (1.47) (2.07) (1.59) (3.56) (1.76) (3.13) (1.79) Year Dummies Yes Yes Yes Yes Yes Yes Yes Yes Fixed Effects No Yes No Yes No Yes No Yes Intercept 0.4840 3.4469*** 1.5615*** 3.5097*** 0.4253 3.5506 0.4922 3.4947 (0.76) (12.24) (3.34) (12.52) (0.71) (12.30) (0.95) (12.79) R2 0.2539 0.2329 0.2256 0.2154 N 588 434 588 508 588 510 588 540 In parentheses are robust t-statistics. All standard errors are clustered at the firm level. The 1% outliers of Firm Value and Profitability is winsorized by replacement. R2 values are the model’s Overall-R2 (weighted average of the Within-R2 and Between-R2). R2 from GLS estimation does not have the same interpretation. Hence, R2 of model (1) are not reported. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

27

From the regression analysis, weak evidence is found of the existence of a hedging

premium for Dutch firms during the sample period. Overall, the coefficients are positive

for both the Hedging Dummy and the Amount of Hedging except when considering

commodity hedging separately.

The coefficients of the Hedging Dummy variable represent the change in firm value

when a firm hedges compared to a firm that does not hedge. Firm value increases by

0.63%5 when a firm hedges either currency, interest rate, and/or commodity price risk.

As for currency, interest rate, and commodity hedging, the effects are 67.72%, 9.95%, and

-9.15%, respectively. These results, however, are not significant and, thus, provide no

evidence that hedging increases firm value. Moreover, results from the Hedging Dummy

can be misleading as it does not measure the exact level of hedging. To solve this issue, a

continuous variable is included measuring the notional amounts of outstanding

derivatives.

When the Amount of Hedging is considered, the coefficients indicate a slight

increase in firm value following an increase in the extent of hedging in terms of notional

amounts. In other words, the greater the hedging position, the higher the firm value. The

coefficients are significant at a 1-5% level. However, the effects seem to be very small. For

example, hedging an additional €1 million of currency exposure increases firm value by

0.00075%. The results imply that nonfinancial Dutch firms can increase their value by

hedging, but the impact is extremely small.

Furthermore, some control variable have an effect on firm value as well.

Significant results are found for firm size, leverage, international diversification, liquidity,

and dividend dummy. The signs of these coefficients seem to be logical and as expected.

Leverage, International Diversification, and Dividend Dummy are positively related to

firm value, whereas, Firm Size and Profitability are negatively related to firm value. The

negative signs for size is in line with the findings of Chun et al. (1985). They explained

that smaller firms are riskier and captured the size effect in the risk premium that justifies

higher returns for smaller firms. As expected, dividends are positively associated with

firm value. Negative signs for profitability are counter-intuitive.

5 The coefficient of General Hedging in Model (1) is 0.0063. Thus, the effect on firm value is (e0.0063-1)x100%=0.63%.

28

6 Robustness & Discussion

Several additional tests are performed to examine the robustness of our results. We test

whether the results obtained from the estimation of Models (1) and (2) are robust to the

alternative measures of firm value (FV2 and FV3). Results of the robustness tests are

presented in Appendix 10, 11, and 12. The inference of our results does not change after

running the regression models another sixteen times with different measures. Thus, the

robustness checks support the earlier results that hedging does not increase firm value.

Overall, the results are largely inconsistent with earlier studies. Allayannis and

Weston (2001), Panaretou (2014), and Bua et al. (2015) found a statistically and

economically significant hedging premium of 5%, 14%-16%, and 1.53%, respectively.

The discrepancy in the results could be explained by the different country focus as well

as the different sampling period. On the contrary, our results are most in line with Guay

& Kothari (2003) who concluded that potential gains of hedging are positive but small.

Their interpretation is that either the observed increase in firm value is driven by other

risk management activities or that the correlation is spurious. In addition, the

insignificance supports the findings of Jin & Jorion (2006) who verified that hedging

reduces the firm’s stock price sensitivity but does not affect firm value for a sample of 119

US oil and gas producers. Another reasoning suggested by Tufano (1996) is that hedging

reflects managerial risk aversion and may actually harm firm value if risk management is

costly.

Our evidence raises doubts about the conclusions of existing literature. The small

increase in firm value documented in our research could indicate that the relationship is

indeed either driven by other risk management activities or that the results are spurious.

However, given the small number of firms in our sample, the lack of significant results

could be due to the relatively low power of the empirical tests.

29

7 Conclusion

This study examines the corporate use of financial derivatives and firm value in The

Netherlands. Univariate tests and multivariate regression analyses are carried out with

panel data methodology including generalized least squares and fixed effects methods.

The sample consists of 98 public non-financial Dutch firms during the period of 2012 to

2017. Total firm-year observations of 588 are segmented into currency, interest rate, and

commodity hedgers. Similar to U.S. and U.K. studies, the use of currency and interest rate

derivatives is more common than the use of commodity derivatives. More than half of the

sample – roughly 60% – reported the use of derivatives for hedging purposes. At the

same time, a Dutch firm hedges €1.97 billion per year in total notional amounts on

average. This supports the view that Dutch firms are widely exposed to currency and

interest rate risk. Therefore, Dutch firms have high incentives to hedge. The empirical

results, however, indicate that hedging does not add value for Dutch firms. On the one

hand, univariate results indicate an average hedging discount of -21.34% for any type of

hedger compared to a non-hedger. More specifically, the average hedging discounts are -

15.09% for currency hedging, -12.68% for interest rate hedging, and -19.47% commodity

hedging. On the other hand, regression results suggest a very small hedging premium of

0.0000787% significant at a 1% level for hedging in general. This translates into a mere

0.0000787% increase of firm value following an additional €1 million notional amount

in hedging instruments. The results are furthermore robust to different measures of firm

value. Conclusively, hedgers are valued significantly lower than non-hedgers and the

value-adding component of hedging is close to zero. Hedging, therefore, does not add

value in the case of Dutch firms.

7.1 Implications and Suggestions

The above empirical results are an extension of existing research. Since prior

studies are mostly concentrated on the U.S., the main contribution of this research is the

use of a unique dataset for The Netherlands that is not readily available to investigate the

value implications of hedging. The discrepancy between this research and numerous

existing studies suggests that there are major differences between Dutch and U.S. firms.

For instance, corporate governance may indirectly affect the valuation of derivatives use.

The research topic is relevant for risk management practices as well as shareholders.

Further research is therefore required to determine whether corporate hedging should

30

be considered an important component of a firm’s risk management policy and/or

whether corporate hedging is a value-enhancing activity. Finally, while this study

provides useful insight in the hedging activity of Dutch firms, the implications are sample

specific. Future research should focus on more generalizable out-of-sample analysis in

order to assess the value implications of corporate hedging.

7.2 Limitations

Shortcomings and limitations in this research are briefly discussed in this paragraph.

First, the greatest issue in empirical Corporate Finance research is endogeneity

(Roberts & Whited, 2012). Endogeneity is present when there is a correlation between

the independent variables and the error term in a regression. This can lead to biased and

inconsistent parameter estimates making the results unreliable. Moreover, Guay and

Kothari (2003), Jin and Jorian (2004), Bua et al. (2015), and Magee (2009) raised serious

endogeneity concerns regarding firm value and hedging. Linear regressions used in this

research assume that all variables in the model are strictly exogenous. An example of

endogeneity is reverse causality. For instance, we tested the effect of hedging on firm

value. But firm value may also affect the decision to hedge because higher valued firms

have more resources and perhaps have more incentivized to use derivatives.

Secondly, research regarding the use of derivative instruments of European firms

is restricted due to data availability. Despite the time and effort put in obtaining hedging

data, the data is prone to errors from manual collection. This affects the accuracy of the

results. Moreover, due to small number of observations, no size threshold has been

implemented in the selection of the sample firms for this research. The majority of related

papers, however, excluded small firms from their sample due to lower risk exposure and

less need for hedging. However, direct costs of insolvency are widely independent on firm

size and imply that small firms should have more incentives to hedge. Dolde (1993) finds

evidence that large firms use significantly more hedging instruments, but small firms

hedge to a greater extent. Although firm size is controlled for in the regression models,

including small firms might have driven contaminated the results.

31

References

Allayannis, G., and Weston, J.P. (2001). “The Use of Foreign Currency Derivatives and Firm

Market Value”, The Review of Financial Studies 14 (1), pp. 243-276.

Allayannis, G., Lel, U., and Miller, D.P. (2012). “The use of foreign currency derivatives,

corporate governance, and firm value around.” Journal of International Economics

87(1), pp. 65-79.

Ayturk, Y., Gurbuz, A.O., and Serhat, Y. (2016). “Corporate derivatives use and firm value:

Evidence from Turkey”, Borsa Istanbul Review 16 (2), pp. 108-120.

Bartram, S.M., Brown, G.W., and Conrad, J. (2011). “The effects of derivatives on firm risk

and value”, Journal of Financial and Quantitative Analysis 46 (2), pp. 967-999.

Belghitar, S.M., Clark, E., and Mefteh, S. (2013). “Foreign currency derivative use and

shareholder value”, International Review of Financial Analysis, pp. 283-293.

Bodnar, G.M., de Jong, A., and Macrae, V. (2003). “The Impact of Institutional Differences

on Derivatives Usage: A Comparative Study of U.S. and Dutch Firms.” European

Financial Management 9, pp. 271–297.

Búa, M.V., González, L.O., López, S.F., and Santomil, P.D. (2015). “Is value creation

consistent with currency hedging?” The European Journal of Finance 21, pp. 912-

945.

Carter, D., Rogers, D., and Simkins, B. (2006). “Does Fuel Hedging Make Economic Sense?

The Case of the U.S. Airline Industry.” Financial Management 35 (1), pp. 53–86.

Froot, K.A., Scharfstein, D.S., and Stein, J.C. (1993). “Risk Management: Coordinating

Corporate Investment and Financing Policies.” The Journal of Finance 48 (5), pp.

1629–1658.

Graham, J. R., and Rogers, D. A., (2002). “Do firms hedge in response to tax incentives?”

The Journal of Finance 57 (2), pp. 815-839.

Guay, W., and Kothari, S. P., (2003). “How much do firms hedge with derivatives?” Journal

of Financial Economics 70, pp. 423-461.

Jensen, M., and Meckling, W. (1976). “Theory of the firm: Managerial behavior, agency

costs, and ownership structure”, Journal of Financial Economics 4, pp. 305–360.

Jin, Y., and Jorion, P. (2006). “Firm value and hedging: evidence from U.S. Oil and Gas

Producers”, The Journal of Finance 61(2), pp. 893-919.

32

Khediri, K. B. (2010). “Do investors really value derivative use? Empirical evidence from

France”, The Journal of Risk Finance 11(1), pp. 62-74.

Kraus, A., and Litzenberger, R. H. (1973). “A state-preference model of optimal financial

leverage.”, The Journal of Finance 28(4), pp. 911-922.

Leland, H. E. (1998). “Agency Costs, Risk Management, and Capital Structure”, The Journal

of Finance 53(4), pp. 1213-1243.

Lookman, A. (2004). “Does hedging increase firm value? Evidence from oil and gas

producing firms”, EFA 2004 Maastricht Meetings, Working Paper (no. 5174).

Modigliani, F., and Miller, M.H. (1958). “The Cost of Capital, Corporation Finance, and the

Theory of Investment.” American Economic Review 48, pp. 261–297.

Myers, S. (1977). “Determinants of Corporate Borrowing,” Journal of Financial Economics

5, pp. 147-175.

Nelson, J., Moffitt, J., and Affleck-Graves, J. (2005). “The impact of hedging on the market

value of equity.” Journal of Corporate Finance 11(5), pp. 851-881.

Panaretou, A. (2013). “Corporate risk management and firm value: evidence from the UK

market.” The European Journal of Finance 20(2), pp. 1161-1186.

Perez-Gonzalez, F., and Yun, H. (2013). “Risk management and firm value: evidence from

weather derivatives.” The Journal of Finance 68(5), pp. 2143-2176.

Rawls, S.W., and Smithson, C.W. (1990). “Strategic risk management.” Journal of Applied

Corporate Finance 1, pp. 6-18.

Tufano, P. (1996). “Who Manages Risk? An Empirical Examination of Risk Management

Practices in the Gold Mining Industry.” Journal of Finance 51(4), pp. 1097-137.

33

Appendix

A1 Literature Overview

Type of hedging

Authors Sample period Country focus Currency Interest rate Commodity Results

Allayannis et al.

(2001)1990-1995 US X

Positive relation between firm value and use of currency derivatives. Hedging

premium is 4.87% of firm value. Some evidence of causality.

Panaretou (2014) 2003-2010 UK X X

Hedging premium is significant for currency hedging, but weak for interest rate

hedging. The extent of hedging and the horizon have an impact on the hedging

premium.

Jin et al. (2006) 1998-2001 US XHedging reduces the firm's stock price sensitivity to oil and gas prices. However,

hedging does not seem to affect market values.

Búa et al. (2015) 2004-2007 Spain XHedging generated a premium of 1.53% wrt company value. The contribution of

currency hedging firm value fluctuates according to volume of hedging.

Allayannis et al.

(2012)1990-1999 39 countries X

Strong evidence that the use of currency derivatives is associated with a

significant value premium (9-20%).

Bartram et al.

(2011)1998-2003 47 countries X X X

The effect of derivative use on firm value is positive, but sensitive to endogeneity

and omitted variable concerns.

Belghitar et al.

(2008)1995 UK X X Strong relationship between derivatives hedging and firm value.

Gleason et al. () 1998 US XFinancial hedging adds value to high-tech firms (5.6-5.8%), while operational

hedging does not.

Kim et al. (2006) 1996-2000 US X Hedging increases firm value.

Ayturk et al.

(2016)2007-2013 Turkey X X X Use of financial derivatives does not affect firm value in Turkish market.

Carter et al.

(2006)1993-2003 US X

Airlines that hedge their fuel costs have Tobin's Q of 5-10% higher than those

that do not hedge.

Khediri (2010) 2000-2002 France X XDecision to use derivatives has no effect on firm valuation. However, the extent

of derivatives use is associated with lower firm value.

Lookman (2004) 2000 US & Canada X Hedging does not increase firm value of oil and gas producers.

Nelson et al.

(2005)1995-1999 US X X X

No abnormal returns for firms that hedge interest rates or commodities. Some

evidence of abnormal returns for firms that hedge currencies.

Pérez-González

et al. (2010)1997 US Weather derivatives lead to higher market valuations in the utilities sector.

34

A2 Sample Firms

# Company Name # Company Name 1 AALBERTS INDUSTRIES NV 57 KONINKLIJKE PHILIPS NV 2 ACCELL GROUP NV 58 LASTMINUTE.COM NV 3 AD PEPPER MEDIA INTL NV 59 MILKILAND NV 4 AFC AJAX NV 60 MKB NEDSENSE NV 5 AIRBUS SE 61 NAVIGATOR EQUITY SOLUTIONS 6 AKZO NOBEL NV 62 NEDAP NV 7 ALTICE NV 63 NEWAYS ELECTRONICS INTERNTL 8 ALUMEXX NV 64 NORD GOLD SE 9 AMG ADVANCED METALLURGICAL 65 NOVISOURCE NV

10 AMREST HOLDINGS SE 66 OCI NV 11 AMSTERDAM COMMODITIES NV 67 ORANJEWOUD NV 12 AND INTL PUBLISHER NV 68 ORDINA NV 13 ARCADIS NV 69 OVOSTAR UNION NV 14 ARGEN-X SE 70 PEIXIN INTERNATIONAL GROUP 15 ASM INTERNATIONAL NV 71 PHARMING GROUP NV 16 ASML HOLDING NV 72 PHOTON ENERGY NV 17 ASTARTA HOLDING NV 73 PORCELEYNE FLES (NV KONINK) 18 AVANTIUM N.V. 74 POSTNL NV 19 BATENBURG TECHNIEK NV 75 QIAGEN NV 20 BESI-BE SEMICONDUCTOR INDS 76 RANDSTAD NV 21 BETER BED HOLDING NV 77 REFRESCO GROUP NV 22 BOSKALIS WESTMINSTER NV 78 RELX NV 23 BRUNEL INTERNATIONAL NV 79 ROODMICROTEC NV 24 CATALIS SE 80 SBM OFFSHORE NV 25 CORBION NV 81 SIF HOLDING NV 26 CORE LABORATORIES NV 82 SLIGRO FOOD GROUP NV 27 CTAC NV 83 SNOWWORLD NV 28 CURETIS NV 84 STEINHOFF INV HOLDINGS NV 29 DGB GROUP N V 85 STERN GROEP NV 30 DPA GROUP NV 86 STMICROELECTRONICS NV 31 EASE2PAY NV 87 TIE KINETIX NV 32 EUROPEAN ASSETS TRUST NV 88 TKH GROUP NV 33 FERRARI NV 89 TOMTOM NV 34 FIAT CHRYSLER AUTOMOBILES NV 90 TRADER MEDIA EAST LTD 35 FNG NV 91 UNILEVER NV 36 FORFARMERS NV 92 UNIQURE NV 37 FORTUNA ENTERTAIN GRP NV 93 VOPAK (KONINKLIJKE) NV 38 FUGRO NV 94 WESSANEN NV 39 FUNCOM N.V. 95 WOLTERS KLUWER NV 40 FYBER N.V. 96 X5 RETAIL GROUP NV 41 GEMALTO 97 YANDEX N.V. 42 GRANDVISION NV 98 ZIGGO NV 43 HEIJMANS NV 44 HEINEKEN HOLDING NV 45 HEINEKEN NV 46 HYDRATEC INDUSTRIES NV 47 ICT GROUP NV 48 IMCD NV 49 IT COMPETENCE GROUP NV 50 KARDAN NV 51 KENDRION NV 52 KONINKLIJKE AHOLD DELHAIZE 53 KONINKLIJKE BAM GROEP NV 54 KONINKLIJKE BRILL NV 55 KONINKLIJKE DSM NV 56 KONINKLIJKE KPN NV

35

A3 Hedging distribution over years

Total Hedgers Non-Hedgers N Currency Interest rate Commodity price Years:

2012 98 52 44 14 38 2013 98 48 41 15 40 2014 98 47 43 16 39 2015 98 47 41 17 39 2016 98 48 40 18 40 2017 98 47 37 17 41

Total 588 289 246 97 237

36

A4 Diagnostic Test: Linearity

Tests for non-linearity are conducted using Stata’s -nlcheck- command after running the

regression model. All independent variables are tested individually, except dummy

variables since they are linear by definition. Overall, no non-linearity is found. Below is a

visual presentation of the linear relationship between the dependent variable and a main

independent variable.

A5 Diagnostic Test: Homoscedasticity

Heteroscedasticity is found when plotting the residuals against the fitted values. A clear

inconsistent pattern can be seen in the scatterplot below, which indicates that the

residuals in the model are heteroscedastic. To correct for this, robust standard errors are

used.

-4-2

02

4A

ug

me

nte

d c

om

po

ne

nt p

lus r

esid

ua

l

0 20000 40000 60000 80000 100000Derivatives Use (notional amount)

-6-4

-20

24

Re

sid

uals

3 4 5 6 7 8Linear prediction

37

A6 Diagnostic Test: No Autocorrelation

The Woolridge test rejects the hypothesis that there is no first-order autocorrelation in

the error term. To correct for serial correlation: I estimate parameters with the

generalized least squares method –xtgls- to fit the model with first-order autoregressive

error term.

A7 Diagnostic Test: Normality

Prob > F = 0.0000

F( 1, 97) = 43.213

H0: no first-order autocorrelation

Wooldridge test for autocorrelation in panel data

. xtserial FV1_w hedger size roa lev growth div int_div y1 y2 y3 y4 y5 y6

Prob > F = 0.0000

F( 1, 73) = 38.641

H0: no first-order autocorrelation

Wooldridge test for autocorrelation in panel data

. xtserial FV1_w total size roa lev growth div int_div y1 y2 y3 y4 y5 y6

0.2

.4.6

.8D

en

sity

-2 0 2 4 6FV1

0.2

.4.6

.8D

en

sity

-2 0 2 4 6FV2

0.1

.2.3

.4.5

De

nsity

-4 -2 0 2 4FV3

0.2

.4.6

.8D

en

sity

-2 0 2 4 6FV1_adj

38

A8 Diagnostic Test: No Multicollinearity Pearson correlation coefficients among independent variables. Firm Size Profitability Leverage Growth Opp. Diversif. Liquidity Dividend Firm Size 1.0000 Profitability 0.2656 1.0000 Leverage -0.0633 0.2263 1.0000 Growth Opp. -0.1053 -0.5931 0.0966 1.0000 Diversification 0.2163 0.0824 -0.070 -0.1240 1.0000 Liquidity -0.1721 -0.2705 0.0373 0.3803 -0.0407 1.0000 Dividend 0.2163 0.0824 -0.0730 -0.1025 -0.0144 -0.1526 1.0000 Hedger dummy 0.5375 0.0359 -0.0116 -0.0042 0.2156 -0.1027 0.2805 Curr. Hedger dummy 0.5716 0.0171 -0.0365 -0.0198 0.3173 -0.1182 0.2314 IR Hedger dummy 0.4631 0.0075 0.0409 0.0015 0.0435 -0.0870 0.1806 Comm Hedger dummy 0.3998 0.3998 -0.0144 -0.0402 0.0971 -0.0936 0.0659 Total Notional Hedging 0.3213 0.0194 -0.0203 -0.0178 0.1005 -0.0510 -0.0747 Curr. Notional Hedging 0.3016 0.0178 -0.0244 -0.0156 0.0971 -0.0477 -0.0770 IR Notional Hedging 0.3901 0.0271 0.0157 -0.0323 0.0895 -0.0661 -0.0361 Comm. Notional Hedging 0.3112 0.0179 -0.0266 0.1491 0.1581 -0.0421 -0.0810

A9 Diagnostic Test: Fixed Effects vs Random Effects (Hausman Test)

Prob>chi2 = 0.0000

= 165.70

chi2(8) = (b-B)'[(V_b-V_B)^(-1)](b-B)

Test: Ho: difference in coefficients not systematic

B = inconsistent under Ha, efficient under Ho; obtained from xtreg

b = consistent under Ho and Ha; obtained from xtreg

y5 -.1959221 -.1439524 -.0519697 .0045523

y4 -.0905756 -.0463021 -.0442734 .0045431

y3 -.1132793 -.0567675 -.0565118 .0071788

y2 -.3074691 -.1828513 -.1246177 .0106958

y1 -.3640398 -.2539525 -.1100873 .0104416

dividend .143761 .2183241 -.0745631 .0469698

liquidity -.0205842 -.0003652 -.020219 .0036709

internatio~v .0017024 .0029484 -.001246 .0005078

growth .0102226 .0215932 -.0113707 .0036533

leverage -.0693197 -.1092891 .0399694 .004248

roa .2843394 -.2059141 .4902535 .054055

size .2108194 .5141773 -.3033579 .0242115

hedger -.0042123 .183748 -.1879603 .0665942

fixed random Difference S.E.

(b) (B) (b-B) sqrt(diag(V_b-V_B))

Coefficients

39

A10 Univariate Results

Comparison of Hedgers and Non-hedgers This table contains t-test results comparing the mean values of hedgers and non-hedgers for each control variable. Constructions of all variables are explained in detail on p. 16-17.

Hedgers Non-Hedgers Difference Firm Size 7.59 4.51 3.08*** (14.77) Leverage 1.13 1.46 -0.33 (0.96) Profitability 0.03 -0.05 0.08*** (5.24) Growth opportunities 0.05 0.76 -0.71*** (-3.24) International Diversification 51.45 36.79 14.68*** (4.33) Liquidity 0.54 1.07 -0.53*** (3.12) Dividend dummy 0.46 0.16 0.30*** (7.96) Business segment diversification 0.75 0.54 0.21*** (5.47) N 356 238 In parentheses are t-statistics of the difference in mean values. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

40

A11 Robustness Check: Different measures (FV2) This table presents panel regression results. Model 1 is estimated with Generalized Least Squares method on a hedging dummy that equals 1 if the firm has reported the use of derivatives for hedging in a given year. While Model 2 regresses on the total notional amount of hedging reported in a given year with Fixed Effects method. Firm Value is calculated as the natural log of [(BV Assets + MV Equity – BV Equity)/BV Assets]. Construction of control variables are described on p. 16&17, Year dummies are included to control for time effects. General Hedging Currency Hedging Interest Rate Hedging Commodity Price Hedging

Dependent variable: Firm Value (1) (2) (1) (2) (1) (2) (1) (2) Hedger (Dummy) 0.2568 0.6084 0.6539 -0.4167 (0.65) (1.27) (1.44) (-0.51) Hedging (Notional) 6.34e-6** 6.15e-6** 2.81e-5** -0.0002 (2.38) (2.04) (2.31) (-1.46) Firm Size -0.3772*** -0.4893*** -0.3890*** -0.4818*** -0.3905*** -0.4812*** -0.3879*** -0.4746*** (-4.31) (-12.03) (-4.64) (-12.09) (-4.56) (-11.53) (-4.25) (-11.63) Profitability -0.5110* -0.0858 -0.5130* -0.1025 -0.5326* -0.1040 -0.4291 -0.1178 (-1.84) (-0.47) (-1.89) (-0.57) (-1.95) (-0.57) (-1.54) (-0.64) Leverage 0.5736 0.1977*** 0.5504 0.1996*** 0.5771 0.1957*** 0.3782 0.1982*** (1.18) (7.85) (1.14) (7.63) (1.20) (7.04) (0.74) (7.09) Growth Opportunities 0.4131 0.2381* 0.2245 0.2695** 0.2956 0.2566* -0.9365 0.2182* (0.52) (1.94) (0.28) (2.16) (0.38) (1.97) (-0.85) (1.69) International Diversification 1.1131*** 0.2869** 1.1281*** 0.2606** 1.2205*** 0.1617 1.1770*** 0.1487 (3.27) (2.28) (3.52) (2.18) (3.60) (1.35) (3.14) (1.38) Liquidity -0.0362 -0.0336** -0.0178 -0.0319* -0.0357 -0.0279* 0.2705 -0.0281* (-0.20) (-2.11) (-0.10) (-1.96) (-0.20) (-1.67) (1.07) (-1.67) Dividend Dummy 0.1214* 0.1144* 0.1116* 0.4120 0.1152** (1.71) (1.76) (1.92) (0.74) (2.00) Year Dummies Yes Yes Yes Yes Yes Yes Yes Yes Fixed Effects No Yes No Yes No Yes No Yes Intercept 1.7074*** 3.2784*** 1.8346*** 3.3227*** 1.7829*** 3.3680*** 1.8386*** 3.3218 (4.23) (13.01) (4.80) (12.98) (4.63) (12.70) (3.60) (12.85) R2 0.2757 0.2555 0.2444 0.2318 N 588 434 588 508 588 510 588 540 In parentheses are robust t-statistics. All standard errors are clustered at the firm level. The 1% outliers of Firm Value and Profitability is winsorized by replacement. R2 values are the model’s Overall-R2 (weighted average of the Within-R2 and Between-R2). R2 from GLS estimation does not have the same interpretation. Hence, R2 of model (1) are not reported. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Dividend dummy variable in Model (1) is omitted due to collinearity.

41

A12 Robustness Check: Different measures (FV3) This table presents panel regression results. Model 1 is estimated with Generalized Least Squares method on a hedging dummy that equals 1 if the firm has reported the use of derivatives for hedging in a given year. While Model 2 regresses on the total notional amount of hedging reported in a given year with Fixed Effects method. Firm Value is calculated as the natural log of firm value as calculated in FV1 minus the industry median firm value. Construction of control variables are described on p. 16&17, Year dummies are included to control for time effects. General Hedging Currency Hedging Interest Rate Hedging Commodity Price Hedging

Dependent variable: Industry-adjusted Firm Value (1) (2) (1) (2) (1) (2) (1) (2) Hedger (Dummy) -0.0659 -0.4208 0.6539 -0.4167 (-0.14) (-0.69) (1.44) (-0.51) Hedging (Notional) 8.17e-6*** 7.86e-6** 2.81e-5** -0.0002 (2.75) (2.42) (2.31) (-1.46) Firm Size -0.3531*** -0.5736*** -0.3626*** -0.4818*** -0.3905*** -0.4812*** -0.3879*** -0.4746*** (-4.14) (-12.42) (-4.35) (-12.09) (-4.56) (-11.53) (-4.25) (-11.63) Profitability -0.5863* -0.0736 -0.6408** -0.1025 -0.5326* -0.1040 -0.4291 -0.1178 (-1.72) (-0.35) (-2.08) (-0.57) (-1.95) (-0.57) (-1.54) (-0.64) Leverage 1.5440 0.1935*** 1.4196 0.1996*** 0.5771 0.1957*** 0.3782 0.1982*** (1.42) (8.11) (1.45) (7.63) (1.20) (7.04) (0.74) (7.09) Growth Opportunities -1.0370 0.2228 -0.9906 0.2695** 0.2956 0.2566* -0.9365 0.2182* (-1.33) (1.49) (-0.95) (2.16) (0.38) (1.97) (-0.85) (1.69) International Diversification 0.9057* 0.3304** 0.9370** 0.2606** 1.2205*** 0.1617 1.1770*** 0.1487 (1.73) (2.13) (2.32) (2.18) (3.60) (1.35) (3.14) (1.38) Liquidity 0.4126* -0.0331 0.3322 -0.0319* -0.0357 -0.0279* 0.2705 -0.0281* (1.81) (-2.12) (1.35) (-1.96) (-0.20) (-1.67) (1.07) (-1.67) Dividend Dummy 1.0653*** 0.1347 0.9328*** 0.1144* 0.1116* 0.4120 0.1152** (3.68) (1.50) (2.94) (1.76) (1.92) (0.74) (2.00) Year Dummies Yes Yes Yes Yes Yes Yes Yes Yes Fixed Effects No Yes No Yes No Yes No Yes Intercept 0.3517 3.4505 0.4407 3.3227*** 1.7829*** 3.3680*** 1.8386*** 3.3218 (0.52) (-3.72) (0.77) (12.98) (4.63) (12.70) (3.60) (12.85) R2 0.2653 0.2555 0.2444 0.2318 N 588 434 588 508 588 510 588 540 In parentheses are robust t-statistics. All standard errors are clustered at the firm level. The 1% outliers of Firm Value and Profitability is winsorized by replacement. R2 values are the model’s Overall-R2 (weighted average of the Within-R2 and Between-R2). R2 from GLS estimation does not have the same interpretation. Hence, R2 of model (1) are not reported.

***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

42