The Influence of Derivatives on Firm Value and Firm Risk; Evidence from Germany

(1)

1

Master Thesis Finance

The Influence of Derivatives on Firm Value and Firm Risk;

Evidence from Germany

By Dylan Kuiper

Abstract

In this research I investigate the effects of derivatives usage on firm value and firm risk for nonfinancial German firms listed on the DAX, MDAX and TecDAX between 2005 and 2013. Differences in factors that affect a firm’s decision to use derivatives, such as corporate governance systems and exposure make the dataset used in this dataset different and unique from datasets used in existing literature. I find evidence that total derivative positions have a significant negative effect on a firm’s Tobin’s q, and thus firm value. Most of this negative relation is fed by currency and commodity derivatives. When looking at firm risk, divided into systematic and idiosyncratic risk, I find that total derivative positions have no significant influence on total risk. However, I do find a significant negative relation with systematic risk and a significant positive relation for idiosyncratic risk. The two combined leave a firm’s total risk insignificantly affected. This relationship is primarily caused by currency derivatives as they have the most influence on the overall risk measures. Even after correcting for a selection bias by means of propensity score matching analysis, the effects of derivatives usage on firm value and firm risk are close to zero. This rather opposing results could be explained by the fact that German managers have less incentive for higher firm values and less firm risk, due to their lesser degree of equity portions in comparison with USA managers.

Keywords: Derivatives, interest rate, foreign currency, commodity price, firm value, Tobin’s q, firm

risk, systematic risk, idiosyncratic risk, propensity score

JEL Classification: G23, G32

Author: Dylan Frederikus Bernardus Kuiper Mail: kuiperdylan@gmail.com

Phone: +31643428118 Student number: S1882252

(2)

2 “Derivatives are financial weapons of mass destruction”

Warren Buffet, 2003

I. Introduction

Since the 2008 crisis, derivatives are seen as instruments that cause more harm than they do good. In 2003, world’s best known value investor Warren Buffet, called derivatives “financial weapons of mass destruction”. That they can be used for wrong purposes has been shown in the 2008 crisis, but initially they have a clear function for industrial companies, namely that of risk management. Bartram, Brown and Conrad (2011) state that when a firm makes use of these derivatives in a correct and non-speculating way, the downside of a firms risk exposure should and could be hedged properly.

When looking at the use of derivatives over the last decade, statistics from the Bank for International Settlements (BIS) shows that the use of derivatives has increased significantly as is shown in Figure 1. Both currency and interest rate derivatives show a great increase in nominal amounts from 1998 onwards. Especially the growth of interest rate swaps contributes to the growth in notional amounts outstanding. Commodity derivatives spiked around the 2008 crisis and decreased in notional amounts afterwards.

In 2002 the EU agreed on regulation that since January 1st_{2005 all EU listed firms have to comply} with IFRS/IAS reporting standards. With these reporting standards firms are mandatory to show their use of derivatives. With this change in regulation, much more insights in EU firm’s derivatives usage can be obtained. Both the increase in derivatives usage and the change in reporting standards give new insights in this subject and this deserves further research. Much literature is written on the effects of derivatives usage on firm value and firm risk, however most data on derivatives usage was known for IFRS complying countries, such as the USA and UK. Therefore much research is focussed on data from these two countries. This opens opportunities to do research for other regions, since all EU listed firms have to disclose their derivatives usage since the beginning of 2005. The effects of derivatives usage on firms specifically from Germany, Europe’s leading economy, have not been researched as of now. But why would German firms and their derivatives use add knowledge to the existing literature and what makes this dataset different from the ones already used in past research?

(3)

3 exposure of a firm’s earnings, the costs of using derivatives and the incentive to use derivatives. The first factor contains a firm’s exposure. Looking at the volatility of federal interest rates between the USA and Germany, little differences can be observed1_{. However, when looking at both} economies’ net export as percentage of GDP figures for 20132_{, the USA shows a NX/GDP ratio of} -3,0%3_{, whereas Germany generates a NX/GDP ratio of +3,1%, after correcting for exports/imports} to other Euro countries4_{. This implies a difference in exposure, since the economy of USA imports} more than it exports and the opposite is true for the German economy. This general difference in foreign exchange exposure could have influence on a company’s choice to employ derivatives. The second factor is about the costs of employing derivatives. Gézcy et al. (1997) state that this is primarily related to the degree of economies of scale in risk management and the use of derivatives. So large companies with large risk management departments are more likely to use derivatives than smaller companies with lesser emphasis on this subject. Later on this finding will be confirmed by the dataset. Since most data is available for listed firms and often listed firms are great in size, not much difference between datasets from the USA and Germany are expected to appear from this factor. The third factor addresses the incentive within firms to use derivatives. One of the incentives comes from the managers who direct the use of derivatives. Smith and Stulz (1985) show that managers with a large portion of company shares will experience decreasing expected utility with increasing variance of future expected profits. Therefore the manager will hedge the firm’s share price in order to increase his expected utility. Smith and Stulz therefore see a positive relationship between managerial investments in their own firm and its use of derivatives. Furthermore, Kaplan (1997) showed that ‘punishments and rewards’ tied to both short- and long-term measures appear to be the same amongst managers in the USA, Japan and Germany. However, Kaplan states that there is one important difference that prevents USA firms from overinvesting and that is that managers hold much larger equity portions than their fellow managers in Germany (and Japan). So by combining Kaplan’s findings and the theory of Smitz and Stulz, one can conclude that managers in the USA have more incentive to use derivatives, since they hold more equity portions than their peers from Germany. Therefore they want to increase their expected utility by decreasing the variance of a firm’s profits.

So it seems that there are primarily two differences between USA and German firms in factors that affect a firm’s choice in using derivatives, namely a difference in foreign currency exposure and in a firm’s incentive to use derivatives. These differences in exposure and governance systems

(4)

4 between the datasets used in existing literature and this research could give new insights on this matter and should therefore be researched.

Figure 1. – The figures below show the notional amounts of derivatives from 1998 until 2014 as

recorded by the Bank for International Settlements. Panel A shows the notional amounts for all foreign currency derivatives, Panel B shows the notional amounts for all interest rate derivatives and Panel C shows the notional amounts for all commodity price derivatives.

Panel A. – Notional amount of foreign currency derivatives recorded by the BIS from 1998-2014.

Panel B. – Notional amount of interest rate derivatives recorded by the BIS from 1998-2014.

Panel C. – Notional amount of commodity price derivatives recorded by the BIS from 1998-2014.

(5)

5

Research question

Given these differences, the research question of this paper is as follows: “What is the influence of

derivatives usage on firm value and firm risk for listed nonfinancial German firms between 2005 and 2013?”

Sub-questions are:

- What is the effect of specific types of derivatives, namely currency, interest rate and commodity derivatives, on firm value?

- What is the effect of specific types of derivatives, namely currency, interest rate and commodity derivatives, on a firm’s total risk, systematic risk and idiosyncratic risk?

This paper will start with an overview of the literature written on both subjects, namely the effect of derivatives usage on firm value and the effect of derivatives usage on firm risk. This will be followed by a section that explains the methodology used in this researched. Then the results from this research are presented and finally a conclusion will be presented. The last section contains an overview of the references used in this paper.

Literature review – Firm value

The classical paper of Modigliani and Miller (1958) state that in case of a perfect market, a firm with value maximizing agents and investors with equal access to capital markets, no firm value will be added by hedging activities. The theory is that investors hedge themselves by means of diversification and therefore hedging is not required for a firm. However, in imperfect markets, frictions could be costly and this is where hedging activities could be of value for a firm. Reasons for a firm to hedge anyway lay in the violation of one of the Modigliani and Miller assumptions that cause frictions in markets such as taxes, transaction costs agency problems and information asymmetries. Smith and Stulz (1985) state that for a hedging policy to affect the value of a firm, this has to do so through frictions in the market, i.e. taxes, contracting costs or the impact of hedging policy on investment decisions. Going in depth on the theorem of Smith and Stulz, Graham and Rogers (2002) state that firms have two tax incentives to hedge, namely to increase its debt capacity and interest tax deductions, and to reduce tax liabilities if the tax function appears to be convex. In their paper they test whether these two incentives influence the extent to which firms make use of derivatives for hedging purposes. They find that firms not necessarily hedge in respond to tax convexity, but primarily in response to an increase their debt capacity.

(6)

(7)

7 they find that this type of hedging increases firm value, investments and leverage. Their data consisted of 203 USA energy firms and they showed that active risk management for energy firms has direct effects for firm performance and value.

However much literature indicates that active risk management through the use of derivatives increases firm value, some papers indicate that hedging has no or modest influence on firm value. Take for instance the research of Jin and Jorion (2006). They examined the effects of hedging on firm value in the USA for 119 oil and gas producers in the period 1998-2001. They collected data on the extend of hedging and on the valuation of oil and gas reserves. They find that hedging reduces the firm’s stock price sensitivity with regard to oil and gas prices. Contradicting to previous research however, hedging does not seem to influence market values of firms in this industry. Another USA example, now from the airline industry, researches the relation between an airliner’s risk exposure, its hedging policy and firm value. Treanor et al. (2014) researched this matter and they found that the exposure of an airliner to fuel prices is higher when fuel prices are higher. Secondly they find that higher levels of exposure lead to an increase in hedging activity. Finally they find a positive hedging premium, but the interaction between exposure and hedging activity does not influence firm market value. The next paper was presented by Aziz Lookman at a European Finance Association meeting in Maastricht on 3rd_{of September 2004. In his paper,} Lookman (2004), he researched whether hedging increases firm value by looking at hedging premiums for oil and gas exploration and production firms from the USA, Canada and the Cayman Islands in the period 1999-2000. He distinguishes two firms, diversified and undiversified. For undiversified firms commodity price is a primary risk whereas it as secondary risk for diversified firms. Lookman finds a fifteen percent hedging discount for the undiversified firms and a thirty percent hedging premium for the diversified firms. Furthermore he finds that hedging primary risk is proxy for bad management/high agency costs, where hedging secondary risk proxies exactly the opposite, namely good management/low agency costs. He concludes, that his findings all together show that hedging has no or only little effect on firm value.

(8)

8

Literature review – Firm risk

(9)

9 hedging. De Jong et al. (2006) researched the exchange rate exposure of Dutch firms in the period 1994 until 1998. They find that over half of the researched Dutch firms are exposed to exchange rate risk, conforming the effects of an open economy. Furthermore, they find no significant effects on reduction of exposure by hedging with derivatives. They do find a significant reduction of exchange rate exposure via on-balance sheet hedging, i.e. foreign loans and geographical diversification.

Concluding, studies such as Guay (1999), Allayannis and Ofek (2001), Hagelin and Pramborg (2004), Bartram et al. (2011) and Nguyen and Faff (2010), all find reduction in firm risk as a result of derivatives usage. Other studies conducted by Guay and Kothari (2003), Hentschel and Kothari (2001), De Jong et al. (2006) find no reduction or hardly a reduction in firm risk by using financial derivatives.

II. Methodology

The methodology used to research the influence of derivative usage on firm value and risk is divided in two parts. The first part is concerned with the effects on firm value using regression models with instrumental and control variables. Firm value is measured by Tobin’s q, which is a ratio of market value of equity component plus all the firms liabilities to the value of its assets. It can also be interpreted as a ratio between the market value of capital and the replacement costs of capital. A higher value for q indicates that a firm’s assets are worth more than their acquisition costs, giving incentive for firms to invest.5_{The second part is concerned with the effects on firm} risk. This will be assessed using three measures of risk, namely total, systematic and idiosyncratic risk. Total risk is simply a firm’s equity return volatility. Furthermore, total risk can be divided into systematic and an idiosyncratic part. Finally a propensity score analysis will be done to correct for a selection bias.

Firm value

In order to show the effect of derivatives use on firm value, four models are being tested. With dummy variables the use of derivatives, see equation (1), and the use of specific types of derivatives, see equation (2), are being measured. The following models, based on Panaretou (2013), are being tested:

(10)

10 𝐹𝑖𝑟𝑚 𝑉𝑎𝑙𝑢𝑒_𝑖 = 𝛼 + 𝛽 × 𝐷𝐸𝑅𝑖 + 𝛴𝑗𝜈𝑗 × 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖 + 𝜀𝑖 (1)

𝐹𝑖𝑟𝑚 𝑉𝑎𝑙𝑢𝑒_𝑖 = 𝛼 + 𝛽1 × 𝐼𝑅𝐷𝑖 + 𝛽2× 𝐶𝑈𝑅𝐷𝑖+ 𝛽3× 𝐶𝑂𝑀𝐷𝑖+ 𝛴𝑗𝜈𝑗× 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖+ 𝜀𝑖

(2)

where 𝐷𝐸𝑅_𝑖 stands for the dummy variable for derivatives usage, 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟._𝑗𝑖 indicates the several control variables included in the regression and 𝐼𝑅𝐷_𝑖, 𝐶𝑈𝑅𝐷𝑖, 𝐶𝑂𝑀𝐷𝑖 all represent dummy

variables for the usage of respectively interest rate derivatives, foreign currency derivatives and commodity derivatives.

In addition, the notional value of derivatives (equation (3)) and the notional value of specific types of derivatives (equation (4)), both as a ratio to total assets are being tested. This is to see whether the degree of derivatives usage has an influence on firm value.

𝐹𝑖𝑟𝑚 𝑉𝑎𝑙𝑢𝑒𝑖 = 𝛼 + 𝛽 × 𝐷𝐸𝑅𝑖 𝑇𝐴𝑖 + 𝛴𝑗𝜈𝑗 × 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖 + 𝜀𝑖 (3) 𝐹𝑖𝑟𝑚 𝑉𝑎𝑙𝑢𝑒𝑖 = 𝛼 + 𝛽1 × 𝐼𝑅𝐷𝑖 𝑇𝐴𝑖 + 𝛽2× 𝐶𝑈𝑅𝐷𝑖 𝑇𝐴𝑖 + 𝛽3× 𝐶𝑂𝑀𝐷𝑖 𝑇𝐴𝑖 + 𝛴𝑗𝜈𝑗× 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖+ 𝜀𝑖 (4) where 𝐷𝐸𝑅𝑖

𝑇𝐴𝑖 stands for the notional amount of derivatives-to-total asset ratio, 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖

indicates the several control variables included in the regression and 𝐼𝑅𝐷𝑖

𝑇𝐴𝑖 ,

𝐶𝑈𝑅𝐷𝑖

𝑇𝐴𝑖 ,

𝐶𝑂𝑀𝐷𝑖

𝑇𝐴𝑖 all

represent notional derivative amounts-to-total asset ratios respectively for interest rate derivatives, currency derivatives and commodity derivatives.

Besides that, Panaretou and Allayannis and Weston (2001) describe a number of control variables which also could have an influence on firm value:

a) Firm size is taken in to the regression by taking the natural logarithm of total assets (Ln(TA)). Many empirical research papers, such as Peltzman (1977) prove that firm size positively influences firm efficiency.

b) Profitability is an important factor influencing Tobin’s q. The more profitable a firm, the more likely it trades at an premium relative to less profitable firms, reflected in a higher value for q than that of a less profitable firm (Allayannis and Weston (2001))Profitability is measured by the return on assets (ROA).

(11)

11 dividend, so less access to financial markets, it has a higher value for q. This will be tested with a dividend dummy (DIV), 1 if a firm pays a dividend and 0 if not.

d) Leverage is measured as the ratio of a firm’s long-term debt to its market value of equity (DE) and capital structure can have an impact on the value of a firm, due to distress costs, debt tax shield, etc.

e) Investment growth is one of the most important metrics in determining firm value. According to Koller et al. (2010), investment growth and return on (newly) invested capital drive firm value. Therefore there is high possibility it influences Tobin’s q. Investment growth is measured as a ratio of capital expenditures to total sales (CAPEX). This ratio is also used by Panaretou (2013). Allayannis and Weston also propose two other measures, namely R&D expenses and advertising expenses as ratio of total sales. They include R&D expenses, since it proxies besides a firm’s investment opportunities, also the amount of intangible assets of a firm. The same reasoning, but more devoted to a firm’s goodwill, goes for advertising expenses.

f) Industrial diversification seems to reduce the value of firm, and thus negatively related to Tobin’s q. Allayannis and Weston and Panaretou list some empirical arguments that industrial diversification, causes a valuation discount on average6_{. Diversification is} measured through an industry dummy (INDD), the dummy value is 1 if a firm is active in one or more industries and 0 if it is not.

g) Industry effect is included to see whether a certain industry influences the level of Tobin’s q. Allayannis and Weston (2001) state the following concerning derivative users’ Tobin’s q and industry effects: “If hedgers are concentrated in high-q industries, then hedgers will have higher values, not because of their use of derivatives, but because of the industry they belong to.” Industry dummies are used to check for differences between industries (IDD). h) Geographical diversification is measured through foreign sales as ratio of total sales

(FORSAL). Allayannis and Weston (2001) and Panaretou (2014) both describe number of papers that prove the positive relationship between geographical diversification, or multinationality, and Tobin’s q, and thus firm value7_.

Firm value in this model is approximated by Tobin’s q, a widely used proxy for firm value. Chung and Pruitt (1994) present a simple model to approximate Tobin’s q as presented in equation (5).

6_{See Lewellen (1971), Denis, Denis and Yost (2002), Berger and Ofek (1995), Lang and Stulz (1994) and, Serveas}

(1996).

(12)

12

𝐴𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑒 𝑇𝑜𝑏𝑖𝑛′_{𝑠 𝑞 = (𝑀𝑉𝐸 + 𝑃𝑆 + 𝐷𝐸𝐵𝑇) 𝑇𝐴}_⁄ ₍₅₎

In this notation MVE equals the market value of equity, PS is the liquidation value of preferred stock of a firm, DEBT equals the net of a firm’s short-term liabilities and assets plus its long-term debt and TA is book value of a firm’s total assets. As can be noted immediately, this model assumes that book values equal market values. Therefore this model can be inaccurate for below investment grade firms. The approximation of Tobin’s q as proxy for firm value is, according to Wernerfelt and Montgomery (1988), is risk-adjusted, forward-looking and accounting standard changes have little influence on the approximation.

Firm risk

Besides the effects of derivatives usage on firm value, also the effects on firm risk are going to be tested. Risk is a broad concept that can be measured using several models and theories. One can define risk through equity volatility, default probabilities and many other methods explore various other types of risk. In this research, we will follow the methodical framework of Nguyen and Faff (2010). They use daily stock returns and their volatilities to distinguish between two types of risk, namely systematic risk and idiosyncratic risk. Both combined one obtains the standard deviation of the daily stock returns of a certain firm. This framework is also known from the Capital Asset Pricing Model and is shown in equation (6):

𝑅_𝑖 = 𝛽0+ 𝛽𝑚𝑅𝑚+ 𝜀𝑖 (6)

with 𝑅_𝑖 as daily stock return of firm i, 𝑅_𝑚 is the daily return of the corresponding index (in this research either DAX, MDAX or TecDAX) and 𝜀_𝑖 as the error term. In this framework, systematic risk is represented by 𝛽_𝑚 and the idiosyncratic risk by 𝜀_𝑖. Combined they comprise a firm’s total risk, simply the standard deviation of returns. To derive systematic risk from total risk one must regress the returns of firm i with the returns of the corresponding market/index. The 𝑅2_{of the}

regression tells which degree of the deviation of a firm’s return can be explained by the deviations of the market returns. By multiplying the 𝑅2_{with the standard deviation, one obtains the}

systematic risk of a firm. The part of total risk that stays unexplained by the market deviations is the idiosyncratic risk of a firm and therefore the degree of idiosyncratic risk is 1 − 𝑅2_.

(13)

13 derivative usage reduces a firm’s stock price volatility, it is also expected to reduce its cash flow volatility. In order to show the effect of derivatives use on a firm’s total risk, systematic risk and idiosyncratic risk, equation (7) to (12) are being tested. Based on Nguyen and Faff, equations (7) and (8) test both total derivative notional amounts and specific type of notional derivative amount, and their effect on total firm risk. The same goes for equations (9) and (10), except now the effect on systematic risk is measures. Equations (11) and (12) test for the influence on idiosyncratic risk of a firm.

𝑇𝑜𝑡𝑎𝑙 𝑅𝑖𝑠𝑘𝑖 = 𝛼 + 𝛽 × 𝐷𝐸𝑅𝑖 𝑇𝐴𝑖 + 𝛴𝑗𝜈𝑗 × 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖 + 𝜀𝑖 (7) 𝑇𝑜𝑡𝑎𝑙 𝑅𝑖𝑠𝑘𝑖 = 𝛼 + 𝛽1 × 𝐼𝑅𝐷_𝑖 𝑇𝐴_𝑖 + 𝛽2× 𝐶𝑈𝑅𝐷_𝑖 𝑇𝐴_𝑖 + 𝛽3× 𝐶𝑂𝑀𝐷_𝑖 𝑇𝐴_𝑖 + 𝛴𝑗𝜈𝑗× 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖+ 𝜀𝑖 (8) 𝑆𝑦𝑠𝑡𝑒𝑚𝑎𝑡𝑖𝑐 𝑅𝑖𝑠𝑘𝑖 = 𝛼 + 𝛽 × 𝐷𝐸𝑅_𝑖 𝑇𝐴_𝑖 + 𝛴𝑗𝜈𝑗 × 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖 + 𝜀𝑖 (9) 𝑆𝑦𝑠𝑡𝑒𝑚𝑎𝑡𝑖𝑐 𝑅𝑖𝑠𝑘𝑖= 𝛼 + 𝛽1 × 𝐼𝑅𝐷𝑖 𝑇𝐴𝑖 + 𝛽2× 𝐶𝑈𝑅𝐷𝑖 𝑇𝐴𝑖 + 𝛽3× 𝐶𝑂𝑀𝐷𝑖 𝑇𝐴𝑖 + 𝛴𝑗𝜈𝑗× 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖+ 𝜀𝑖 (10) 𝐼𝑑𝑖𝑜𝑠𝑦𝑛𝑐𝑟𝑎𝑡𝑖𝑐 𝑅𝑖𝑠𝑘𝑖 = 𝛼 + 𝛽 × 𝐷𝐸𝑅𝑖 𝑇𝐴𝑖 + 𝛴𝑗𝜈𝑗 × 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖 + 𝜀𝑖 (11) 𝐼𝑑𝑖𝑜𝑦𝑠𝑛𝑐𝑟𝑎𝑡𝑖𝑐 𝑅𝑖𝑠𝑘𝑖 = 𝛼 + 𝛽1× 𝐼𝑅𝐷𝑖 𝑇𝐴𝑖 + 𝛽2× 𝐶𝑈𝑅𝐷𝑖 𝑇𝐴𝑖 + 𝛽3× 𝐶𝑂𝑀𝐷𝑖 𝑇𝐴𝑖 + 𝛴𝑗𝜈𝑗× 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖+ 𝜀𝑖 (12)

The same as with the analysis for the impact of derivative usage on firm value, Nguyen and Faff use a number of control variables to isolate other factors that might contribute to the overall risk of a firm. In this research the control variables are similar and they are listed below:

a) Size and its influence on firm risk is defined as an ambiguous relationship. Firms that are larger in size are more likely to engage in risk managing behaviour with the help of derivative instruments. On the other hand, smaller firms tend to be more risky (Guay (1999), Hentschel and Kothari (2001)) and should have more incentive to hedge risks and use derivative instruments. Therefore it is unknown whether size has a positive or negative impact on firm risk. Firm size is measured by the natural logarithm of total asset value.

b) Market-to-book value (MVBV) is a ratio that displays the options for growth for a certain firm. Growth firms tend to be more risk, due to increased risk of underinvestment. Froot et al. (1993) that underinvestment represents the failure to maximize shareholder value and therefore it is presumed to be costly.

(14)

14 reluctant to issue debt than less risky firms. This implies that when a firm issues debt, their cash flows are presumed to be more stable, hence, less volatile. Leverage is measured as the long term debt-to-common equity ratio.

d) Liquidity as a control variable comes back to the costs associated with underinvestment, shown by Froot et al. (1993). Highly liquid firms can employ more easily internal funds, rather than expensive external funds and therefore the risk of underinvestment is relatively low. This low risk is associated with low overall firm risk. Extreme liquid firms however, have to cope with agency costs. Jensen and Meckling (1976) and Schleifer and Vishny (1989) show that extreme levels of liquidity within a firm lead to irrational investments and management entrenchment. This implies additional volatility to the cash flows of a firm. This means that liquidity has a non-linear relationship with firm risk and that both too low and extreme levels of liquidity add risk to cash flow volatility. Liquidity is measured with the current ratio (CR) and the cash-to-total assets ratio (CATA).

e) Managerial discretion is again involved with agency theory. Managerial behaviour has great impact on future firm cash flows and therefore their incentives should be aligned with firm incentives to maximize shareholder wealth. Nguyen and Faff use two indicators to measure the degree of managerial discretion, namely the portion of executive share-to-total ordinary shares and executive options-to-share-to-total ordinary shares. In this research however we use a measure created by Thomson Reuters Datastream and is called “Board of directors/Compensation policy”. This category measures a company's management commitment and effectiveness towards following best practice corporate governance principles related to competitive and proportionate management compensation. It reflects a company's capacity to attract and retain executives and board members with the necessary skills by linking their compensation to individual or company-wide financial or extra-financial targets. The measure is a score between 0 and 100 by comparing a firms management ability to an universe of 4000 firms. In the rest of this paper this variable is called “Compensation Policy Score” (CPS).

f) Industry effects can have substantial effects on overall firm risk. For example, Shin and Soenen (1999) show that metal electrical equipment industries tend to be more exposed to foreign currency risk than other industries. Therefore different industries can have different influences on a firm’s risk. Dummy variables to check for industry effects are included as a control variable.

Propensity score matching analysis

(15)

15 larger firms are more likely to make use of derivatives than smaller firms. This creates a selection bias as described in Bartram et al. (2011). They present a measure to check for selection bias in the firm characteristics and it is given by equation (13):

𝐵𝐼𝐴𝑆 = |(𝜇1− 𝜇0)

√(𝜎₁2_{+ 𝜎} 02)/2

⁄ | (13)

where 𝜇₁ and 𝜇₀ are mean values of the control variables for derivative users and non-users respectively and 𝜎₁2_{together with 𝜎}

02 are the squared standard deviations of the control variables

of users and non-users respectively.

The best way to see the impact of derivatives usage on firm value and firm risk is to compare two identical firms with one not using derivatives and the other using derivatives. A method to correct for this selection bias is to match users and non-users based on propensity scores. The idea is to compare individuals who based on observables have a very similar probability of receiving treatment, but one of them received treatment and the other didn’t. Then the assumption is that all the difference in the dependent variable is due to the treatment.8_{In this case treatment is the} use of derivatives. This way the effect of derivatives usage on firm value and firm risk can be shown without other lurking variables influencing this effect. In this analysis one estimates the probability for an individual receiving treatment, in this case a firm using derivatives, based on the firm characteristics and control variables. By estimating probit model (14) one obtains coefficients, noted as

𝛽

_𝑖

,

for the control variables that have significant influence on a firms choice to use derivatives.

𝐼

_𝐷𝐸𝑅

= 𝛽

₀

+ 𝛽

₁

𝑇𝐴 + 𝛽

₂

𝐷𝐸 + 𝛽

₃

𝐹𝑂𝑅𝑆𝐴𝐿 + 𝛽

₄

𝐼𝑁𝐷𝐷 + 𝛽

₅

𝐶𝑅 + 𝛽

₆

𝐶𝑃𝑆 + 𝛽

₇

𝐼𝐷𝐷

(14)

where

𝐼

_𝐷𝐸𝑅 is the dummy variable, indicating derivatives usage with a value of 1 and 0 otherwise. The other variables are given and explained above. Then the propensity score is the predicted value 𝐼̂_𝐷𝐸𝑅 that you get from multiplying the value of

𝛽

_𝑖with the value of the corresponding control variable. Now you have two groups, derivative users and non-users, with different mean propensity scores. Since the data consists of much more derivative users than non-users, the mean value for non-users is used as a benchmark to select a group of users that come to a similar, or at least as similar as possible, mean propensity score. This way we obtain two approximately similar groups of firms accept for the fact that one group contains of users and the other of non-users.

(16)

16 Then we can test for differences between these two groups in values for Tobin’s q ,total risk, systematic risk and idiosyncratic risk and see if the difference in derivatives usage has an effect on these measures. Table 1 Panel A shows the regression coefficients resulting from estimating probit model (14). Panel B shows the selection of users and non-users and the descriptive statistics regarding their propensity scores.

Table 1. –Summary of the propensity score matching analysis. Panel A shows the probit model used

to estimate the coefficients of the control variables. These coefficients are then used to make a propensity score for each firm. Panel B shows the descriptive statistics for the firms selected on basis of their propensity score.

Panel A: Probit regression coefficients for propensity scores

𝛽1 𝛽2 𝛽3 𝛽4 𝛽5 𝛽6 𝛽7

0,4053 0,0055 0,0174 -1,1684 -0,1819 -0,0001 -0,1587

Panel B: Descriptive statistics on the propensity scores

N Mean Median St. Dev

Non-users 114 6,7209 6,6673 1,4166

Users 123 6,7212 6,7250 0,4761

Difference 0,0003 - -

T-test 0,9981 - -

III. Data

The data consists of all listed German nonfinancial firms on three German indices, namely the DAX, MDAX and TecDAX from the period between 2005 and 2013. This period is chosen because it is recent and derivative disclosure is compulsory since 2005 for all listed firms in the EU. The initial panel dataset consisted of 990 observations and after correcting for financial firms, the dataset shrank to an unbalanced panel dataset with 918 observations. Amongst these 918 observations a distinction can be made between users and non-users of derivatives. Data on the control variables are gathered through Thomson Reuters Datastream. This is also the case for the dependent variables. The components of Tobin’s q are just as all the other control variables derived from yearly data. Total, systematic and idiosyncratic risk is derived from daily stock returns. The notional derivative amounts are not available in the Datastream database so this information is collected by scanning all annual reports for the given firm selection and time period.

(17)

17 and then decrease back to around the level of 2007. This is similar to the pattern shown in Figure 1, the total notional derivative amounts recorded by the BIS. The degree of hedging can be grasped by the notional derivative amount-to-total asset ratio. Panel C tells us that, looking at the median for the entire dataset, notional amounts of derivatives used sum up to 15% of a firm’s total assets. In this total, foreign currency derivatives make up for the highest notional amounts, followed by interest rate derivatives. Respectively, they make up for 5,5% and 2,5% of total assets. Interesting is to see that the mean values for interest, currency and commodity derivatives approximately sum up to the mean value for derivatives. However this is not the case for the median values indicating that total derivative positions not always evenly consist of interest rate, currency and commodity derivatives. Firms choose for a specific type of derivative, depending on their exposure and incentives, rather than hedging by means of all three derivative types.

Table 2. – Number of derivative users and notional derivative amounts by year. Panel A consist of

data on usage of derivatives, Panel B shows total notional derivative values per year and finally the notional value of derivative positions divided by the firms total assets is shown in Panel C. The amount of firms (N), mean, median, standard deviation (St. Dev.) the first quartile (Q1) and third quartile (Q3) values give insight in the dataset used in this research. Note that specific types of derivatives not necessarily need to add up to the total derivative positions because other type of derivatives, such as credit default swaps, are not included in this dataset.

Panel A: by year

N Users Non-users % users % non-users

2005 82 63 19 76,83 23,17 2006 84 71 13 84,52 15,48 2007 90 79 11 87,78 12,22 2008 91 82 11 90,11 9,89 2009 93 81 12 87,10 12,90 2010 94 84 10 89,36 10,64 2011 96 87 9 90,63 9,37 2012 96 87 9 90,63 9,37 2013 102 91 11 89,22 10,78

Panel B: Notional derivative values per year in million €

N Mean Median St. Dev. Q1 Q3

2005 82 12.137,73 249,66 51.380,67 - 1.940,95 2006 84 11.979,71 209,70 59.134,54 2,70 1.738,36 2007 90 13.080,50 358,85 66.703,02 19,13 2.061,41 2008 91 6.439,33 525,60 21.952,31 49,60 2.155,11 2009 93 6.111,76 514,53 21.692,92 56,35 2.057,72 2010 94 7.925,02 578,80 28.248,66 48,55 1.878,35 2011 96 8.634,61 438,53 33.200,57 55,82 1.948,80 2012 96 8.470,78 420,41 31.185,12 30,00 2.302,90 2013 102 7.723,25 377,88 27.810,60 30,00 2.154,00

Panel C: Notional value of derivative-to-total assets ratio

(18)

18 The value of a firm is measured by Tobin’s q, as explained in the previous section. Firm risk is measured through three variables, namely total risk, systematic risk and idiosyncratic risk. Table 3 gives an overview of Tobin’s q and total, systematic and idiosyncratic risk for all firms, firms that report they do not make use of derivatives and firms that report they do. As can be seen from Table 3, the median of Tobin’s q for all firms in the dataset is 0,6352. This value is somewhat similar for users of derivatives and a little higher for non-users with a value of 0,6560. When looking at the risk parameters one observes that total risk is a little lower for non-users versus users. Non-users experience roughly half the systematic risk in comparison with derivative users, but on the other hand non-users exhibit more idiosyncratic risk than derivative users.

Table 3. – Descriptive statistics on Tobin’s q, total risk, systematic risk and idiosyncratic risk amongst

all firms, non-users of derivatives and users of derivatives. The amount of firms (N), mean, median, standard deviation (St. Dev.) the first quartile (Q1) and third quartile (Q3) are shown in all three panels.

N Mean Median St. Dev. Q1 Q3

Panel A: Tobin’s q

All firms 918 0,6021 0,6352 0,2060 0,4849 0,7489

Non-users 123 0,6234 0,6563 0,2631 0,5273 0,7910

Users 795 0,5988 0,6318 0,1953 0,4802 0,7409

Panel B: Total risk

All firms 918 0,0197 0,0189 0,0106 0,0142 0,0251

Non-users 123 0,0190 0,0193 0,0121 0,0125 0,0269

Users 795 0,0198 0,0189 0,0103 0,0143 0,0248

Panel C: Systematic risk

All firms 918 0,0060 0,0047 0,0054 0,0021 0,0084

Non-users 123 0,0038 0,0025 0,0040 0,0005 0,0054

Users 795 0,0063 0,0050 0,0054 0,0025 0,0089

Panel D: Idiosyncratic risk

All firms 918 0,0137 0,0129 0,0082 0,0092 0,0175

Non-users 123 0,0150 0,0150 0,0112 0,0065 0,0221

Users 795 0,0135 0,0128 0,0076 0,0093 0,0170

(19)

19 that derivative users are more financially levered than non-users of derivatives. This makes sense, since highly levered firms have more incentive to hedge interest rate risk and control their interest expenses than firms with lower leverage levels. Another confirming observation would be the difference in foreign sales-to-total sales ratio between users and non-users of derivatives. One would expect a difference, as with the debt-to-equity ratio, since firms with a higher foreign sales ratio have higher probabilities of cash flows occurring in different currencies. Firms with a higher foreign sales ratio would have more incentive to use foreign currency derivatives. When looking at Table 3 one observes a median foreign sales ratio of 54% for non-users and a ratio of 64% for derivative users. This confirms that firms with foreign currency exposure have higher incentive to immunize that risk through derivatives than firms with lower foreign currency exposure.

Table 4. – Descriptive statistics on the several control variables included in the regressions. The table

contains data amongst all firms, non-users of derivatives and users of derivatives. Panel A shows descriptive statistics on all firms in the dataset, Panel B shows only the firms that report no usage of derivatives and Panel C shows the firms that report derivative usage. The number of firms (N), mean, median, standard deviation, first quartile and third quartile are shown in all three panels.

N Mean Median St. Dev. Q1 Q3

Panel A: All firms

Total assets (k EUR) 887 18.478.798 2.972.650 40.639.817 801.799 11.541.600 Ln (Total assets) 887 14,97 14,91 2,00 13,59 16,26 Return on assets 870 5,98 5,55 7,68 2,95 8,79 Dividend per share 918 0,62 0,40 0,82 - 0,90 Dividend dummy 918 0,69 1,00 0,46 - 1,00 Market value equity (k EUR) 887 4.538.026 1.047.618 9.273.531 289.351 3.791.642 Preferred stock (k EUR) 883 666,57 - 6.261,57 - - Total debt (k EUR) 887 5.093.146 547.874 13.687.455 84.968 3.302.000 Long term debt-to-common equity ratio 882 72,24 42,44 208,68 8,96 83,68 Capital Expenditures-to-total sales ratio 886 6,93 3,68 18,43 2,14 6,36 Foreign sales-to-total sales ratio 824 55,64 61,23 28,67 36,56 78,94 Diversification dummy 856 317.811 33.806 886.714 - 103.925

Market-to-book value 822 2,32 1,91 1,96 1,23 2,84

Current ratio 833 1,77 1,43 1,23 1,09 1,97

Cash-to-total assets ratio 887 0,15 0,11 0,17 0,05 0,19

Compensation policy score 802 26,43 19,83 26,97 0,00 48,65

Panel B: Non-users

Total assets (k EUR) 263 5.861.686 1.097.096 14.983.337 284.235 3.992.100 Ln (Total assets) 263 13,98 13,91 1,81 12,56 15,20 Return on assets 253 4,32 4,90 10,65 1,77 8,61

Dividend per share 291 0,14 - 0,54 - -

Dividend dummy 291 0,09 - 0,29 - -

Market value equity (k EUR) 263 1.984.233 321.094 5.332.792 145.886 1.576.554 Preferred stock (k EUR) 263 - - - - - Total debt (k EUR) 263 1.591.524 214.584 4.894.191 16.404 1.258.914 Long term debt-to-common equity ratio 258 75,20 27,50 299,64 2,38 75,96 Capital Expenditures-to-total sales ratio 262 7,07 3,77 22,79 1,90 6,91 Foreign sales-to-total sales ratio 231 47,78 54,27 34,15 11,03 78,18 Diversification dummy 291 0,28 - 0,45 - 1,00

Current ratio 103 2,38 1,82 1,78 1,32 3,05

(20)

20

Panel C: Users

Total assets (k EUR) 624 23.796.587 4.747.000 46.451.096 1.232.176 18.231.250 Ln (Total assets) 624 15,39 15,37 1,92 14,02 16,72

Return on assets 617 6,67 5,76 5,92 3,48 8,83

Dividend per share 627 0,84 0,60 0,83 0,32 1,14

Dividend dummy 627 0,97 1,00 0,17 1,00 1,00

Market value equity (k EUR) 624 5.614.385 1.669.579 10.312.661 479.933 5.487.500 Preferred stock (k EUR) 620 949 - 7.455 - - Total debt (k EUR) 624 6.568.989 774.686 15.775.505 174.550 4.480.075 Long term debt-to-common equity ratio 624 71,01 46,01 156,28 14,79 86,79 Capital Expenditures-to-total sales ratio 624 6,87 3,65 16,26 2,23 6,21 Foreign sales-to-total sales ratio 593 58,70 64,14 25,59 41,18 79,05

Diversification dummy 627 0,57 1,00 0,4947 - 1,00

Current ratio 730 1,69 1,39 1,11 1,08 1,93

Compensation policy score 698 28,83 24,10 27,31 0,00 51,23

IV. Results

First the results for the several tests for the influence of derivatives usage on firm value, represented by Tobin’s q are being treated. Later on, the test results for the influence of derivatives usage on firm risk, represented by total risk and divided into a systematic part and an idiosyncratic part, are being covered.

Firm value

The first two models that will be tested are models (1) and (2). Model (1) includes dummy variables on whether a firm makes use of derivatives and model (2) makes a distinction between three types of derivatives, namely interest rate, currency and commodity derivatives. All three will be tested with dummy variables in model (2). Model (3) and (4) tests whether the notional amount of derivatives influences Tobin’s q, by taking the ratio of a firm’s notional derivative amounts to its total assets. Model (3) tests the total notional amount of derivatives used and model (4) again makes a distinction between interest rate, currency and commodity derivatives, using the notional amounts.

Table 5 shows the results of the regressions of models (1)9_{and (2)}10_{. Both models make use of} dummy variables to measure the impact of derivatives usage on a firm’s value, represented by Tobin’s q. As one can see from the results, almost all variables in model (1) prove to be significant. Only the foreign sales-to-total sales ratio has no significant effect on Tobin’s q, the dependent variable. The variable most relevant in this model is the derivative dummy variable. It is

9_{(1); 𝐹𝑖𝑟𝑚 𝑉𝑎𝑙𝑢𝑒}

𝑖 = 𝛼 + 𝛽 × 𝐷𝐸𝑅𝑖 + 𝛴𝑗𝜈𝑗 × 𝐶𝑜𝑛𝑡𝑟. 𝑉𝑎𝑟.𝑗𝑖 + 𝜀𝑖 10_{(2); 𝐹𝑖𝑟𝑚 𝑉𝑎𝑙𝑢𝑒}

(21)

21 significant at a 1% level and has a negative effect on Tobin’s q. A coefficient of -0,074 indicates that on an average q for all firms of 0,6, derivatives usage has a severe negative influence on Tobin’s q. In model (2) there is been made a distinction between specific types of derivatives. In this model nearly all variables prove to have significant influence on Tobin’s q at a 5% confidence level, except for the dividend dummy and the dummy for the use of interest rate derivatives. Currency and commodity derivatives however do significantly influence Tobin’s q, both negatively with coefficients of respectively -0,07 and -0,05.

Table 5. – The effect of derivatives usage, both total and specific types, on firm value. Firm value in

the form of Tobin’s is the dependent variable and derivatives usage is measured by dummy variables. Furthermore, several control variables are included to decrease the number of lurking variables.

Dependent Variable: Model (1): Tobin’s q Model (2): Tobin’s q

Variable Coefficient t-Statistic Coefficient t-Statistic

C 0,8166 24,0388*** 0,7693 23,3865***

Derivative dummy -0,0743 -3,8168***

Interest rate derivative dummy 0,0041 0,2873

Currency derivative dummy -0,0705 -4,4307***

Commodity derivative dummy -0,0506 -3,5533***

Ln (Total Assets) -0,0061 -4,0243*** -0,0042 -2,6992***

Return on Assets 0,0028 3,4720*** 0,0024 3,0178***

Long term debt-to-common equity ratio 0,0004 3,4180*** 0,0003 2,2252**

Dividend dummy -0,0304 -2,0319** -0,0239 -1,6433

Foreign sales-to-total sales ratio 0,0004 1,5996 0,0005 2,0541**

Diversification dummy -0,0822 -6,5780*** -0,0771 -6,3586***

Capital Expenditures-to-total sales ratio 0,0020 4,8392*** 0,0018 4,4135***

Total panel (unbalanced) observations: 762 762

R-squared 0,1874 0,2097

*** Significance at 1% level with a two-tailed test. ** Significance at 5% level with a two-tailed test. * Significance at 10% level with a two-tailed test.

Table 6 shows the results of models (3)11_{and (4)}12_{. In model (3), almost all variables prove to} have significant influence on Tobin’s q. Again the foreign sales-to-total sales ratio shows no significant effect on Tobin’s q. This coincides with Table 4, where there is little difference in mean values for this control variable between users and non-users. The variable most relevant in this model is the notional amount of derivatives-to-total assets ratio variable. It is significant at a 1% level and has a negative effect on Tobin’s q. In model (4) there has again been made a distinction between specific types of derivatives. In this model nearly all variables prove to have significant

(22)

22 influence on Tobin’s q at a 5% confidence level, except for the foreign sales-to-total sales ratio, the interest rate derivative and the commodity derivative variables. Currency derivatives however do significantly influence Tobin’s q, and it does so negatively.

Table 6. – The effect of derivatives usage, both total and specific types, on firm value. Firm value in

the form of Tobin’s is the dependent variable and derivatives usage is measured by the ratio of their notional amounts-to-total assets. Furthermore, several control variables are included to decrease the number of lurking variables.

Dependent Variable: Model (3): Tobin's q Model (4): Tobin's q

C 0,7991 20,8738*** 0,8686 21,0354***

Notional amount total derivatives / Total

assets -0,0836 -3,2551***

Notional amount interest rate derivatives /

Total assets 0,0290 0,6705

Notional amount currency derivatives /

Total assets -0,1297 -3,1800***

Notional amount commodity derivatives /

Ln(Total Assets) -0,0063 -3,5438*** -0,0096 -4,9749***

Return on Assets 0,0029 2,9584*** 0,0027 2,6967***

Long term debt-to-common equity ratio 0,0004 2,9613*** 0,0003 2,1608**

Dividend dummy -0,0433 -2,4421** -0,0404 -2,2795**

Foreign sales-to-total sales ratio 0,0001 0,2483 0,0001 0,2510

Diversification dummy -0,0604 -4,2842*** -0,0498 -3,4394***

Capital Expenditures-to-total sales ratio 0,0018 4,1021*** 0,0016 3,6557***

R-squared 0,1853 0,1963

*** Significance at 1% level with a two-tailed test. ** Significance at 5% level with a two-tailed test. * Significance at 10% level with a two-tailed test. Firm risk

(23)

23 Table 7 shows the results of the regressions of models (7)13_{and (8)}14_{. Both models make use} notional derivative amounts-to-total asset ratios to measure the impact of derivatives usage on a firm risk. In these two models the impact on a firm’s total risk is being tested. As can be seen from Table 7 model (7), all control variables are significant within a 1% confidence interval, except for the current ratio, which is significant at a 10% level and the industry dummy which checks for industry effects shows no significant influence on total risk. The same goes for the notional amount total derivatives-to-total assets ratio, which shows no significant influence on a firm’s total risk. Model (8) shows similar results for all variables. This means that not one of the specific type of derivatives usage show significant influence on total risk. So firms using derivatives do not experience a lower level of equity return volatility, hence total risk. The effects of derivatives on the systematic and idiosyncratic part are mentioned below.

Table 7. – The effect of derivatives usage, both total and specific types, on firm risk. In these two

models a firm’s total risk is the dependent variable. Derivatives usage is measured by the ratio of their notional amounts-to-total assets. Furthermore, several control variables are included to decrease the number of lurking variables.

Dependent Variable: Model (7): Total risk Model (8): Total risk

C 0,0337 17,5525*** 0,0341 15,6027***

assets -0,0001 -0,0635

Total assets -0,0021 -1,1416

Total assets -0,0006 -0,1762

Ln(Total Assets) -0,0005 -6,7424*** -0,0005 -5,9728***

Market-to-book value -0,0004 -2,7119*** -0,0004 -2,6781***

Long term debt-to-common equity ratio 0,0000 4,1755*** 0,0000 4,6266*** Cash-to-total asset ratio 0,0113 5,4609*** 0,0113 5,3808***

Current ratio -0,0005 -1,8693* -0,0006 -1,9363*

Board of directors/compensation policy -0,0000 -5,5959*** -0,0000 -5,2361***

Industry dummy 0,0001 1,1855 0,0001 0,8623

R-squared 0,5597 0,5606

(24)

24 The results for models (9)15_{and (10)}16_{, which control for a firm’s use of derivatives and their} influence on systematic risk, are shown in Table 8. The control variables of both model (9) and (10) show little to no influence on the systematic part of a firm’s risk. This is no surprise, since firm specific ratios and characteristics can be expected to have more influence on the idiosyncratic part of total risk, rather than the systematic part. When looking at total derivative amounts in model (9), a significant relation at a 10% level show a positive influence of derivatives on the systematic part of a firm’s risk. Model (10) tests whether specific types of derivatives have influence on systematic risk. Interest rate derivatives show a negative significant relation and currency derivatives show a strong positive significant influence on systematic risk. Concluding, commodity derivatives show no significant influence on systematic risk.

models a firm’s systematic risk is the dependent variable. Derivatives usage is measured by the ratio of their notional amounts-to-total assets. Furthermore, several control variables are included to decrease the number of lurking variables.

Dependent Variable: Model (9): Systematic risk Model (10): Systematic risk

C 0,0044 3,2214*** 0,0046 2,9883***

assets 0,0014 1,8230*

Total assets -0,0026 -1,9430**

Total assets 0,0047 4,2464***

Ln(Total Assets) 0,0000 0,4914 0,0000 0,4214 Market-to-book value 0,0000 0,4665 0,0001 1,0258 Long term debt-to-common equity ratio -0,0000 -0,1126 0,0000 0,8001 Cash-to-total asset ratio 0,0009 0,6120 0,0004 0,2926 Current ratio -0,0002 -0,8762 -0,0002 -1,0679 Board of directors/compensation policy 0,0000 3,0971*** -0,0000 -2,3177**

Industry dummy 0,0001 2,1729** 0,0001 1,1669

R-squared 0,3997 0,4089

(25)

25 Finally, models (11)17_{and (12)}18_{control for a firm’s use of derivatives and their influence on} idiosyncratic risk. The results are shown in Table 9. Nearly all control variables in both models, as expected, have strong significant influence on the idiosyncratic risk of a firm. However, the current ratio and the dummy variable that checks for industry effects show no significant relation. In model (11), total notional amounts of derivatives show a significant negative influence on idiosyncratic risk. For specific types of derivatives, foreign currency derivatives show a significant negative relation with idiosyncratic risk, where interest rate and commodity derivatives show no significant influence on a firm’s idiosyncratic risk.

models a firm’s idiosyncratic risk is the dependent variable. Derivatives usage is measured by the ratio of their notional amounts-to-total assets. Furthermore, several control variables are included to decrease the number of lurking variables.

Dependent Variable: _{Idiosyncratic risk}Model (11): _{Idiosyncratic risk}Model (12):

Variable Coefficient t-Statistic Coefficient t-Statistic

C 0,0288 16,4014*** 0,0290 14,4589***

Notional amount total derivatives / Total assets -0,0019 -1,9690** Notional amount interest rate derivatives / Total

assets -0,0006 -0,3330

Notional amount currency derivatives / Total

assets -0,0035 -2,4509**

Notional amount commodity derivatives / Total

assets -0,0006 -0,1912

Ln(Total Assets) -0,0005 -7,5602*** -0,0005 -6,6737***

Market-to-book value -0,0004 -3,0774*** -0,0004 -3,4524***

Long term debt-to-common equity ratio 0,0000 4,6642*** 0,0000 4,5419***

Cash-to-total asset ratio 0,0104 5,5133*** 0,0109 5,6568***

Current ratio -0,0003 -1,2666 -0,0003 -1,2086

Board of directors/compensation policy 0,0000 -8,1702*** 0,0000 -7,1606***

Industry dummy 0,0000 -0,4045 0,0000 0,0232

R-squared 0,4765 0,4763

(26)

26

Propensity score matching analysis

As mentioned in the methodology section, the models and results above are subject to a selection bias. A propensity score matching analysis is used to correct for this bias. Bias indicators, as measured by equation (13), give high bias values for firm characteristics such as the log of total assets, foreign sales-to-total sales ratio, cash-to-total asset ratio and CPS with respectively 147%, 114%, 90% and 104%. This means that, between users and non-users, there is a big difference in the mean value of these firm characteristics while the variances of the two groups do not differ much. By matching propensity scores of non-users with those of users, a comparison can be made between more or less identical firms. This selection of users and non-users based on propensity scores reduced bias measures significantly for all control variables. Most variables now have bias measures below 10% and only one variable with a maximum bias of 27%, meaning that firm characteristics between users and non-users have more similar mean and variance values. Table 10 summarizes the propensity score matching analysis with both Panel A and B giving the comparison between Tobin’s q, total risk, systematic risk and idiosyncratic risk. Panel A shows the test results performed with a parametric Student T-test. After testing for normality, all four risk and value measures show little prove that the data is normally distributed. This observation makes a T-test less valid and therefore a non-parametric test, the Wilcoxon Rank Sum Test, is added to test for differences between users and non-users for median risk and value measures. Therefore, Panel B shows the test results performed with a non-parametric Wilcoxon Rank Sum test. When looking at the differences between the firm value (Tobin’s q) and firm risk measures (Total, systematic and idiosyncratic risk), no significant differences can be observed using the parametric Student’s T-test. When testing with the non-parametric Wilcoxon Rank Sum Test, all four risk and value parameters show a significant difference at a 1% level. For all four risk and value parameters a negative difference in median values is can be observed for non-users in comparison with users of derivatives. From this specific analysis is shown that somewhat identical firms that make use of derivatives exhibit higher values for Tobin’s q and higher risk parameters than non-users. Although this difference is significant, it is very modest and close to zero.

(27)

27 Table 10. –Summary of the propensity score analysis. Panel A shows the mean values of users and

non-users for Tobin’s q, total risk, systematic risk and idiosyncratic risk. Student’s T-test is used as parametric test to test whether the two series are significantly different. Finally, Panel B shows the median values of users and non-users for Tobin’s q, total risk, systematic risk and idiosyncratic risk. Wilcoxon’s rank sum test is used as a non-parametric test to test whether the two series are significantly different. Note that the differences are absolute values.

Panel A: Mean values for value and risk indicators, tested with Student’s T-test

Tobin’s Q Total risk Systematic risk Idiosyncratic risk

Non-users 0,5911 0,0205 0,0041 0,0162

Users 0,6241 0,0220 0,0046 0,0172

Difference 0,0330 0,0015 0,0005 0,0010

T-test 0,1943 0,2915 0,3795 0,4105

Panel B: Median values for value and risk indicators, tested with Wilcoxon rank sum test

Tobin’s Q Total risk Systematic risk Idiosyncratic risk

Non-users 0,6949 0,0202 0,0029 0,0155

Users 0,7198 0,0214 0,0033 0,0166

Difference 0,0249 0,0012 0,0004 0,0011

Wilcoxon -7,5571*** -10,5810*** -10,0996*** -10,5905***

V. Discussion and limitations

(28)

28 Although these results and findings may suggest that derivatives should not be used, they should be interpret with caution. Although a propensity score matching analysis was done to correct for a selection bias, all firms in the dataset are listed. Another limitation of this research could be the definition of risk. In this research firm risk is defined as a firm’s equity return volatility. This is assumed to be a proxy for cash flow volatility, as stated by Guay (1999). However, stock markets can be subject to inefficiency, irrationality and many other factors that influence prices and volatilities. In order to make more solid conclusions on this matter, other measures of risk should be included, such as the Altman Z-score analysis and Merton’s credit risk model. The same applies to the measure for firm value, Tobin’s q. As mentioned earlier, Wernerfelt and Montgomery (1988) state that Tobin’s q is risk-adjusted, forward-looking and accounting standard changes have little influence on the approximation. On the other hand critics state that Tobin’s q is outdated and no longer applicable to today’s firms. This is, because over- and undervaluation can hold for a long time and the fact that in the previous century, markets were dominated by industrial firms. These days markets are much more dominated by firms that make use of intangible assets. Determining the replacement costs of a machine is much easier and straightforward than that of an intangible asset, such as intellectual property and brands. This way, Tobin’s q can be misleading in some cases.

VI. Conclusion

This research addresses two main problems, namely what is the influence of derivatives usage on firm value and what is the influence of derivatives usage on firm risk. Firm value is measured by the concept of Tobin’s q and firm risk is assessed through daily stock return volatility. This study on derivatives and their influence on firm value and risk is of interest, because much literature have researched either one of the two subjects, but most of the data comes from the USA and UK. Differences in foreign currency exposure and corporate governance systems between USA/UK and Germany could give new insights in how derivatives are being used, and what effects they have on firms. New EU regulation on IFRS complied reporting since 2005 made new data available for in this case listed German firms. As much research has shown and as described in the introduction, one could expect that derivatives used for hedging purposes add value to a firm. For firm risk, derivative hedging is expected to reduce the volatility of future cash flows, and thereby lowering equity volatility.

(29)

29 have no effect on total firm risk, since the increase in systematic risk is offset by a reduction in idiosyncratic risk. Specific derivative types show that currency derivatives significantly increase systematic risk, but decrease idiosyncratic risk. Even after correcting for a selection bias by means of propensity score matching analysis, the effects of derivatives usage on firm value and firm risk are close to zero. This is rather an opposing conclusion in comparison with the literature on this subject. They mostly find a positive relation or no significant relation at all for both firm value and firm risk.

(30)

30

VII. References

Allayannis, G., and Weston, J. P., 2001, “The use of foreign currency derivatives and firm market value”, The Review of Financial Studies 14, 243-276.

Allayannis, G. and Ofek, E., 2001, “Exchange rate exposure, hedging, and the use of foreign currency derivatives”, Journal of International Money and Finance 20, 273–96.

Allayannis, G., Lel, U., and Miller, D. P., 2012, “The use of foreign currency derivatives, corporate governance, and firm value around the world”, Journal of International Economics 87, 65-79.

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, 967-999.

Berger, p., Ofek, E., 1995, “Diversification’s effect on firm value”, Journal of Financial Economics

37, 39-65.

Bodnar, G., Tang, C., Weintrop, J., 1997, “Both sides of corporate diversification: the value impact of geographical and industrial development”, Working paper 6224, NBER

Brainard, W. C., Tobin, J., 1968, "Pitfalls in Financial Model Building", American Economic Review

58-2, 99–122.

Chiarella, C., Pham, T., Sim, A. and Tan, M., 1991, “Determinants of corporate capital structure: Australian evidence”, Unpublished Working Paper, University of New South Wales, Sydney. Chung, K. H., Pruitt, S. W., 1994, “A simple approximation of Tobin’s q”, Financial Management

23-3, 70-74

Denis, D. J., Denis, D. K., Yost, K., 2002, “Global diversification, industrial diversification, and firm value”, Journal of Finance 57-5, 1951-1979.

Doukas, J., Travlos, N., 1988, “The effect of corporate multinationalism on shareholders’ wealth: Evidence from international acquisitions”, Journal of Finance 43-5, 1161-1175.

Fischer, B., Scholes, M., 1973, “The Pricing of Options and Corporate Liabilities”, Journal of

Political Economy 81-3, 637-655.

Froot, K. A., Scharfstein, D. S. and Stein, J. C., 1993, “Risk management: coordinating corporate investment and financing policies”, Journal of Finance 48, 1629–58.

Géczy, C., Minton, B. A. and Schrand, C., 1997, “Why Firms Use Currency Derivatives”, The Journal

(31)

31 Gómez-González, J. E., Rincón, C. E. L., Rodríguez, K. J. L., 2012, “Does the Use of Foreign Currency Derivatives Affect Firms’ Market Value? Evidence from Colombia”, Emerging Markets

Finance & Trade 48-4, 50-66.

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

of Finance 61-2, 893-919.

Guay, W., 1999, “The impact of derivatives on firm risk: An empirical examination of new derivative users”, Journal of Accounting and Economics 26, 319-351.

Guay, W., Kothari, S. P., 2003, “How much do firms hedge with derivatives?”, Journal of Financial

Economics 70, 423-461.

Hagelin, N. and Pramborg, B., 2004, “Hedging foreign exchange exposure: risk reduction from transaction and translation hedging”, Journal of International Financial Management and

Accounting 15, 1–20.

Hentschel, L., Kothari, S. P., 2001, “Are corporations reducing or taking risks with derivatives?”,

Journal of Financial and Quantitative Analysis 36-1, 93-118.

Jensen, M. C. and Meckling, W. H., 1976, “Theory of the firm: managerial behaviour, agency costs and ownership structure”, Journal of Financial Economics 3, 305–60.

Jin, Y., and Jorion, P., 2006, “Firm value and hedging: Evidence from U.S. oil and gas producers”,

Journal of Finance 61-2, 893-919.

Jong, de, A., Ligterink, J., Macrae, V., 2006, “A firm-specific analysis of the exchange-rate exposure of Dutch firms”, Journal of International Financial Management and Accounting 17-1, 1-28.

Kaplan S., 1997, “Corporate governance and corporate performance: comparison of Germany, Japan and the U.S.”, Journal of Applied Corporate Finance 9, 86–93.

Koller, T., Goedhart, M., Wessels, D., 2010, “Valuation: Measuring and managing the value of companies”, Wile, 5th_edition.

Lang, L., Stulz, R., 1994, “Tobin’s Q, corporate diversification and firm performance”, Journal of

Political Economy 102, 1248-1280.

Lewellen, W., 1971, “A pure financial rationale for the conglomerate merger”, Journal of Finance

26, 521-537

Lookman, A. A., 2004, “Does Hedging Increase Firm Value? Evidence from Oil and Gas Producing Firms”, Paper presented at the European Finance Association meetings, Maastricht,