University of Amsterdam
Faculty of Economics and Business
June 2014
Bachelor Thesis
Determinants of hedging activity of large European energy companies in the years around the 2008 financial crisis.
Michał Dmoch dhr. dr. J.E. Ligterink
Abstract
This study expands on the previously investigated topic of determinants of derivative hedging in corporations. The literature mostly analyzes data on U.S. companies before the recent recession of 2008. This study contributes to that stock by extending the research into a new environment – the energy sector in Europe in years 2007-2012. It investigates whether the factors most commonly found to influence derivative hedging policies in firms are also valid in such an setting. Thus, it tests the validity of other studies’ conclusions over a new sample. Additionally, it tests the hypothesis that the state of economy as connected to the stage of the Great Recession influenced the use of derivatives for corporate hedging purposes. The results are surprising – none of the commonly investigated determinants of derivative hedging is valid in this sample. Moreover, the estimates of the time variables indicate that the stage-of-the-crisis hypothesis does not hold either. Instead, firm-specific effects and idiosyncrasies seem to dominate the issue. Hence, the conclusion that can be drawn is that the variables that determine the use of derivatives for corporate hedging purposes, at least for this sample, are different than the ones most commonly found in other studies. Hence, more research in this field is necessary to discover and describe those.
1. Introduction
Derivatives, despite having been around for a long time, have only become popular in recent years. Ever since the 1970’s, when the theoretical groundwork and economic circumstances created favorable conditions, derivative markets experienced rapid growth and expansion (Stulz, 2004, p. 10). However, since they were developed and became popular only recently, there is still a lot of unknowns in that field of study – such as the reasons why companies use them. That is, it is well recognized that derivatives can serve either for speculative or hedging purposes, however it is not clear what determines the volume of derivatives that a particular company uses. And, while speculative trading is motivated by bringing in new revenues, hedging aims at securing the revenues already brought in. It attempts to diminish the uncertainty of cash flows inherent in a company’s operations.
Such uncertainty is particularly pronounced in electricity markets due to the high degree of price volatility, which results from the fact that power cannot be economically stored. Additionally, the numerous drivers of energy prices, as well as those of other related commodities, often change radically and independently between markets (Liu & Wu, 2007, p. 690). Apart from the basic supply and demand interaction, they are determined by factors like weather, regulations or international agreements, whose effects are difficult to predict. Hence, many of the large electric utilities take great care to manage their risks, including setting up derivative hedging programs (Deng & Oren, 2006, p. 941).
Much has changed in the world economy since 2007 and, given the shock that the recent crisis was, it could be expected that much has changed in the perceptions of risk as well. Thus, throughout the crisis exposures could have been treated differently, resulting in different hedging policies. For example, in the deeper stages of the crisis, firms could have hedged more as a result of higher financial vulnerability. Therefore, the way companies use derivatives could have changed as well. This would affect not only the U.S., where the crisis began, but also the rest of the world, including Europe.
This paper explores the determinants of derivative hedging activity among electricity firms in recent years. By considering a timeframe encompassing the recent crisis, it pays special attention to the question of how they were affected by the state of economy and the stage of the crisis it was in at a particular time. That is, it investigates how they changed over time, with respect to different stages of the crisis. Additionally, it narrows down its sample to large European electricity companies, to see what conclusions can be drawn about an industry that is particularly under-researched in this field. Thus, the purpose of this paper boils down to
one question: What were the determinants of derivative hedging activities of large European energy companies in the years 2007-2012?
This study contributes to the current stock of academic literature by extending the research that has been done previously into a new environment. Most of the studies conducted on the determinants of hedging concern large U.S. companies in the period 1980-20001. This study not only shifts the focus towards European companies to check for international robustness of the previous results, but also explores the validity of those results in a period of financial and economic instability. In other words, it verifies their conclusions in a different setting, to see whether they still apply. Additionally, it investigates whether the progress of the recent 2008 crisis had a significant effect on those conclusions. The outcomes are surprising – none out of five popularly investigated determinants of corporate hedging are significant for this sample. Further, the insignificance of the time variables indicates that nothing can be said about how the stage of the crisis influenced the use of derivatives. Instead, these activities appear to be mostly driven by firm-specific factors, although even that is not certain for some companies. Thus, this paper shows that derivative hedging activity of large European electricity companies in recent times is not determined by the factors typically important for the U.S. market, but by idiosyncratic ones. Whether this can be generalized to the whole European market is another question and a matter to be further investigated in other studies.
This work progresses as follows: Section 2 gives some background information on the electricity market and derivative instruments, as well as their relation to each other and to the financial crisis. Section 3 gives an overview of the existing literature related to the topic, specifies the variables to be explored and the hypotheses. Section 4 outlines the investigation method used, describes the choice of the sample and gives some basic statistical characteristics to set it in context, as well as develops the proxies that are later used to test the hypotheses. Section 5 gives the empirical results that emerge after processing the data and analyzes them in comparison to Section 3 to see what they imply. Section 6 gives a summary of the findings and a conclusion.
1
See for example: Nance, Smith & Smithson (1993), Géczy, Minton & Schrand (1997), Hentschel & Kothari (2001), Howton & Perfect (1998).
2. Background
This section provides context to the paper’s investigation. It outlines some particulars of the electricity market, gives some information on derivative instruments and hedging, as well as justifies the timeframe chosen by relating the above to the recent financial crisis.
The energy market’s inherent volatility has always been its defining characteristic. The fact that electricity cannot be economically stored, aside from sparking intense technological research in that direction, has been the reason for setting up the complex mechanisms that ensure the balancing of market supply and demand at any given moment. Whenever a trading company in the market has an unbalanced position in electricity for a given moment (purchases are unequal to sales) they will need to pay a penalty, which is classified as balancing cost (Holttinen, 2005, pp. 2052-2053). This cost is one of the risks that the participants in the electricity market must face. However, it can be mitigated using derivatives – if the cost of implementing hedging structures is lower than that of the fees, it can be composed so that the entity has some flexibility with respect to the amount of electricity bought/sold.
As is the case in any market, there is always a risk associated with contracting, the so-called counterparty default risk. However, the particulars of the electricity market make it more pronounced, since a counterparty’s failure to deliver may be more difficult to forecast. This is the case with producers of electricity utilizing renewable sources, whose output is heavily dependent on the weather conditions at a given moment. Since it is virtually impossible to predict the wind speed, water level or sunlight intensity over a long time horizon, the exact amount of electricity received from such producers at a particular point in time is uncertain (Holttinen, 2005, p. 2053). Thus, electricity purchase agreements with such producers may be at risk from a wholesale firm’s point of view when technical difficulties arise – such as the counterparty’s failure to deliver the contracted amount of electricity. In these cases, the buyer’s position diverges from its forecasts and if the market prices move unfavorably, profitability issues may arise. Thus, a company buying and selling energy is often risking costs if its position is unbalanced and it is forced to buy/sell electricity at an unfavorable price to compensate for the counterparty’s default.
Thus, the essence of both the above risk sources is the risk connected with market prices, since a firm is always able to buy/sell any amount of electricity at current market prices to mitigate balancing or counterparty issues. Given the large number of their drivers – coal prices, gas prices, emission allowance prices, renewable certificate prices, tax rates,
quotas, legal requirements or even the weather – these electricity market prices are extremely difficult to forecast in the long-term. However, since some agreements in the electricity market are made for long periods (10-20 years), the wholesale/distribution firms present in that market need some way of insuring themselves against unfavorable price shifts that would endanger their profits – whether because they wouldn’t be able to sell contracted power at a profit, or because they would have to purchase it for balancing purposes, or simply because they wouldn’t be able to conduct trading activities profitably.
This is where risk management comes in. Companies set up entire programs and divisions that are designated to monitor and minimize expected costs resulting from risk exposures (Statkraft, 2012, p. 16). And even though operational hedging is common and often effective, financial hedging is an integral part of many of those programs. Hence, for the past few years, many major power exchanges introduced some kind of electricity derivatives and saw a rise in their popularity. Derivative instruments on other commodities (like coal or gas) have also been present (Weron, 2000, pp. 128-129). Apart from that, firms often resort to interest rate (IR) and foreign exchange (FX) derivatives. And even though derivatives cannot fully hedge the price exposure mentioned above due to volume uncertainty, they can eliminate that risk at least partly. Partial hedge can be achieved by analyzing the expected volume of balancing needed and using forward contracts to ensure that volume will be bought/sold. Then, the deficit/excess energy can be hedged otherwise or sold on the balancing market. Either way, even though derivatives can’t ensure a perfect hedge, they can certainly contribute to mitigating the risks involved.
A derivative is a financial instrument (a security), whose price and payoffs depend on the values of other variables (‘the underlying’) (Hull, 2012, p. 1). Most common underlyings are assets, such as stocks, commodities, bonds or currency. Most common types of derivatives are futures, forwards and options (Hull, 2012, p. 5-8). They can all be used for hedging by overlaying a particular derivative strategy on a portfolio already possessed in order to achieve a particular payoff profile, which in case of hedging activity means minimizing risks and stabilizing cash flows (Hull, 2012, p. 10-12). By adjusting a portfolio appropriately, respective risk exposures can be adjusted using relevant derivatives.
However, the approach to hedging, and to exposures in general, may have changed in recent times. That is, since the crisis was an intense experience to many people around the world, it could have altered the way they have perceived risk. In such a case, firms would change their approach to exposures and alter their hedging policies to fit the former, bearing in mind that hedging aims at minimizing risk. Hence, in deeper stages of the crisis, companies
could be more concerned about their financial well-being and try to offset their vulnerability with lower risk-taking and higher hedging, which would increase the use of derivatives at those times. Thus, arguably, the perceptions of risk may have been changing in recent years: from a more light-hearted approach before the crisis (less hedging), to a more reserved one during the crisis (more hedging), back to a rise in boldness during the recovery phase (less hedging). If people’s views on risk were changing over time, they should have been adjusting their exposures accordingly. This could have resulted in differing profiles of derivative usage in recent years.
Hence, as can be seen, the motivations for risk management are numerous, especially in a market as volatile and uncertain as that of energy. There, the benefits of diminishing risk exposures are particularly clear. These could have been useful to companies especially at times of increased financial vulnerability, like the recent crisis. However, apart from the stage of the crisis, there must be other factors determining the amount of derivative hedging by a particular company. This is the topic of the next section.
3. Determinants of derivative hedging
There have been several prominent studies in the field of derivative hedging determinants (Bartram, Brown & Fehle, 2009, p. 187). This section summarizes some of their main conclusions. It synthesizes the most commonly discussed factors, explains them and hypothesizes on the relation between them and the volume of derivative hedging.
Most of the main theories of hedging determination revolve around the variability of cash flows and liquidity problems (Bartram, Brown & Fehle, 2009, p. 185). That is, when cash flows vary between periods, the firm’s liquidity and its ability to meet immediate financial obligations varies as well. Periods of low liquidity can impose costs on the firm. Minimizing them implies managing and stabilizing the cash flows, and one way of doing that is with derivative instruments. The three main hypotheses discussed below are: underinvestment, cost of financial distress and hedging substitutes. Other than liquidity problems, the hedging determinants that are often mentioned include tax incentives and management compensation (Bartram, Brown & Fehle, 2009, p. 185). These are not discussed in the study for reasons described later in this section.
One of the main determinants of hedging activity investigated in previous research has been the problem of underinvestment. That is, when sufficient internal funds are not available at a certain time, a firm may refrain from investing in potentially profitable projects. There could be several reasons for this, including liquidity concerns or a high cost of external financing (Géczy, Minton & Schrand, 1997, p. 1328). The use of derivatives can help manage the accessibility of internal funding. Thus, it can also reduce the disadvantages of underinvestment by reducing the volatility of firm value. Additionally, underinvestment can be a more severe problem in highly leveraged firms, where a majority of the potential profits would go to the debt-holders and thus could result in projects not being undertaken (Nance, Smith & Smithson, 1993, p. 270). Also, previous research shows that this problem is more pronounced for firms with high growth opportunities, because they have more options to choose from (Bartram, Brown & Fehle, 2009, p. 187). The underinvestment hypothesis is the first of the three main ones investigated in this paper. Since both leverage and growth opportunities should make it more severe, the effects of both these factors are explored.
Another major factor that, theoretically, should influence the use of derivatives is the expected cost of financial distress. Apart from the costs imposed on the firm by periods of low liquidity, like the fees for external financing, that cost is influenced by another factor that should be considered. In a balance-of-probability calculation, the likelihood of bankruptcy
affects the expected cost of bankruptcy (Géczy, Minton & Schrand, 1997, pp. 1327-1328). These costs are something a firm wants to minimize and hence manage the variability of its cash flows. By smoothing them, liquidity can be stabilized and the probability of bankruptcy can be lowered, thus diminishing the expected costs of financial distress. This hypothesis is the second one investigated in this study. According to previous research, the costs of bankruptcy increase less than proportionally with firm size (Bartram, Brown & Fehle, 2009, p. 187). Also, a high ratio of intangible assets means the firm is not able to liquidate them at book value, which can aggravate the costs of distress (Howton & Perfect, 1998, p. 112). Thus, these two factors are taken into account. Additionally, higher leverage implies more debt in the firm’s capital structure, which makes the financial distress problem more pronounced (Bartram, Brown & Fehle, 2009, p. 187). Even though leverage is already included due to its relation to the underinvestment problem, it also plays a part in this hypothesis, giving it two channels to influence the use of derivatives.
A third main conclusion often reached is that the use of derivatives is affected by the availability of hedging substitutes. That is, since cash flow stabilization is the primary purpose of hedging, if a firm has better ways of ensuring that, it is less likely to implement an intensive derivative-hedging program (Nance, Smith & Smithson, 1993, p. 270). Since the main reason for the latter is liquidity management, it has been found that sufficient liquidity acts as a substitute for derivative hedging, because the firm already has the ability to meet its financial obligations (Howton & Perfect, 1998, pp. 112-113). That is, since maintaining liquidity is the goal of derivative usage, having high liquidity to begin with diminishes the need for hedging. The hedging substitute hypothesis is the last of the three main ones investigated in this study. As such, a firm’s liquidity’s impact on its use of derivatives is analyzed.
The factors outlined above are the most prominent ones in the previously conducted studies on the determinants of corporate derivative hedging. All three are investigated in this paper. However, there are also other factors that are named by individual authors, whose effect on hedging differs between studies or is difficult to measure. These most prominently include tax incentives in case of convex tax schedules and management incentives in case of manager stock and stock-option holdings (Bartram, Brown & Fehle, 2009, p. 185). The tax hypothesis states that, since convex tax schedules imply lower tax liabilities in case of stable than variable income, it should stimulate smoothing of cash flows (Nance, Smith & Smithson, 1993, p. 268). The management incentive hypothesis says that when managers own a piece of the firm, they want to minimize the risk resulting from the volatility of its value, thus
implementing hedging programs (Géczy, Minton & Schrand, 1997, pp. 1326). These two hypotheses are not investigated in this paper due to the ambiguity of their effects and lack of widespread support among authors (Bartram, Brown & Fehle, 2009, p. 190). Arguably, excluding potentially influential factors from the study may result in omitted variable bias and misestimation of the coefficients included. However, the bias only occurs if the omitted variables are correlated with both the dependent and one or more independent variables. Therefore, it is debatable whether the bias has an effect in this study, since the results of other authors on both these requirements are mixed at best, as shown by Nance, Smith & Smithson (1993, pp. 280-281) and Bartram, Brown & Fehle (2009, p. 196). Hence, it is difficult to say whether excluding these variables indeed results in a bias.
There is one additional factor that is taken into account in this study. It is the aforementioned effect of the state of economy, as affected by the stage of the crisis it was in, which is assumed to be closely related to risk perception. Since it has been hypothesized that the way risk has been seen and handled differed in various stages of the recent crisis, it could have had an influence on the use of derivatives. That is, since risk perceptions have been tied to the (broadly defined) economic prosperity, the use of derivatives should be too. Since this prosperity was changing over time as the crisis developed and then receded, the use of derivatives could have varied over time due to that factor as well, to adjust the hedging policy to the changing economic conditions.
The three main hypotheses investigated in this study are the underinvestment problem, the expected cost of financial distress and the hedging substitute. They are all connected to the problem of cash flow variability and this makes liquidity stabilization the primary role of derivatives in concern of this study. Other hypotheses, like tax schedule convexity or management incentives, have also been brought forward, but they are not focused on here. One additional hypothesis taken into account in this study is the derivative hedging activity’s relation to the progress of the recent financial crisis, making time another factor to be taken into account.
4. Model and hypotheses
This section outlines the method used for testing the hypotheses. Then it develops and explains the proxies that are later used and states the hypothesized relationship between the relevant variables. It also describes the choice of the sample and gives some of its basic characteristics to set it in context.
The method used to analyze the data is a regression model with multiple independent variables, incorporating the aforementioned hypotheses. The proxies for these variables are outlined below. By estimating the model in that way, the direction and magnitude of the effect of each of the proxies on the amount of derivatives used can be found. It can also be tested on significance. By estimating a coefficient for each of the proxies, the hypotheses stated can be evaluated by assessing the impact of the related variables on the use of derivatives. The software used to run the regression is the free program ‘R’. The formula for the model is presented below and a link to the software’s website can be found in the Appendix.
Model equation: 5 4 3 2 1 29 30 31 32 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 10 9 8 7 6 5 4 3 2 1 Y Y Y Y Y S R Q P O N M L K J I H G F E D C B A QR TAR LEV TA MTB DERREL
where DERREL is the value of derivatives used by particular company in a particular year scaled by the value of total assets, α is the intercept, MTB is the respective Market-to-book ratio, TA is the value of Total Assets, LEV is the Leverage, TAR is the Tangible Assets Ratio, QR is the Quick Ratio, A to S are firm-specific dummies and Y1 to Y5 are time dummies.
As can be seen in the equation of the model, for each of the hypotheses and variables investigated, a proxy is developed (formulas can be found in the Appendix). Since, within the underinvestment problem, it is hypothesized that both leverage and growth opportunities aggravate the underinvestment problem, both these variables should have a significant effect on the amount of derivatives a firm uses. That is, high leverage makes the problem more pronounced, because the majority of potential profits from the available projects would go to the debt holders, thus discouraging the firm from undertaking them. Additionally, more growth opportunities emphasize the underinvestment problem, because the firm has more options to choose from and it increases the chance of suboptimal projects being chosen. Thus, both these variables are included in the estimation. Following studies by authors such as Bartram, Brown & Fehle (2009, p. 189), Géczy, Minton & Schrand (1997, p. 1329) or Nance, Smith & Smithson (1993, p. 274), growth opportunities are proxied by the Market-to-book ratio. It shows how the market perceives a firm’s growth opportunities by measuring whether it over- or under-values the firm’s assets. If the ratio is higher than 1, then the current value of
assets is too low to explain the market value of the firm – the extra value comes from the intangible market expectation that the company will grow and generate profits in the future (Barclay, Smith & Watts, 1997, pp. 10-11). Thus, when the assets are undervalued, the firm has good growth prospects. Hence it is more exposed to the underinvestment problem and should be more prone to hedge. Therefore, the higher the market-to-book ratio, the higher the use of derivatives should be, due to higher expected growth and more pronounced underinvestment problems.
For the financial distress cost hypothesis, three variables are of substance: firm size, leverage and relative amount of tangible assets. The size of the company is simply proxied by the value of its total assets. Due to the aforementioned fact that bankruptcy costs increase less than proportionately with firm size, a higher value of total assets should imply relatively lower derivative use. The leverage is already included and has been discussed above – the distress cost effect only magnifies the contribution of leverage to the final amount of derivatives used, because leverage multiplies the costs of financial distress if they occur due to the firm’s obligations towards debt holders. The amount of tangible assets is estimated by calculating the Tangible Assets Ratio, which relates the book value of tangible assets to the book value of total assets. A higher relative amount of tangible assets implies that the firm, if needed, is able to liquidate them quickly to obtain funds and mitigate the costs of financial distress. On the other hand, intangible assets are more difficult to liquidate, making them less effective in countering funding issues. Thus, a lower Tangible Assets Ratio increases the expected costs of financial distress, increasing the use of derivatives as well.
Lastly, the hedging substitute hypothesis focuses mainly on the firm’s short-term liquidity as a means of handling cash flow issues. This variable can be proxied by the Quick Ratio, or acid test, which relates the amount of current assets (excluding inventory, which is not always easily liquefiable) to current liabilities. It shows whether the firm is able to meet its short-term financial obligations and how much effort does it cost to do so. A high Quick Ratio implies a high relative amount of assets that can be used to repay the liabilities at hand and means that cash flows are easily manageable. Since high liquidity is assumed to be a substitute to derivative hedging, a high Quick Ratio means that not much hedging is needed to maintain stability of cash flows. Therefore, the Quick Ratio and the use of derivatives should be negatively related.
Additionally a set of time variables is introduced. Since the hypothesis is that derivatives were less popular before the crisis (2007-2008), their use rose during the crisis (2008-2010) and fell back down during the recovery period (2011-2012), a set of 6
year-specific dummies are added to the equation. This is because the hypothesized path of the use of derivatives implies an downward-open parabolic relationship: starting off with low values, rising towards a maximum in the middle and decreasing by the end of the set of arguments. Hence, the coefficients of the dummies should be negative for years 2007-2008 (Y1-Y2), positive for 2009-2010 (Y3-Y4) and negative again for years 2011-2012 (Y5-Y6).
Additionally, since some of the companies in the sample are state-owned, the data on their market capitalizations is not available, since they are not publicly traded. This means that the Market-to-book ratios cannot be calculated, as using the nominal share capital would be misleading. Hence, the market value of these firms is estimated indirectly using average industry PE ratios and the firms’ annual earnings. The average PE ratios are calculated manually for each year, using the PE ratios of the other sixteen companies and weighing them by their market capitalization. The average annual PE ratios obtained in this way are then multiplied by the state-owned firms’ annual earnings to get the estimated market values need to calculated the Market-to-book ratios. Additionally, extra dummies are introduced (one per firm), so firm-specific effects (like tradition, experience, culture, management attitude, etc.) can be isolated and quantified. However, one firm dummy (T for Edison) has to be eliminated, due to singularities in the equation, as the intercept α is present. Similarly, one time dummy (Y6 for 2012) has to be taken out of the equation. This is because the intercept acts as the default describing Edison in year 2012 and all the other coefficients are estimated in relation to that.
All the above proxies are related to the amount of derivatives used by each company. However, since each company reports its derivatives differently than the other, and this reporting often differs even between a single company’s individual reports, a good amount of reason had to be applied to standardize the data. Sometimes companies distinguish derivatives by the underlying assets (commodity, interest rate, currency); sometimes they distinguish them by their maturity (current, non-current); or by their value (positive/assets, negative/liabilities). Sometimes they use combinations of those. However, most of the investigated companies include information on how much derivatives were used for each purpose (cash flow hedging, fair value hedging, trading, other). Hence, the data used as an input here is a selection that attempts to approximate the aggregate net amount of hedging derivatives as closely as possible – if given a choice, all underlying assets (commodity/IR/FX) and time horizons (current/non-current) are taken into account, as well as all types of hedging (cash flow/fair value). Derivatives labeled as ‘other’ or not including the word ‘hedge’ in the label are excluded. The amount of derivatives used for hedging is considered in the form of
net assets – liabilities (negative values) are subtracted from assets (positive values). Thus, the total quantity can possibly be very small, or even negative, for some companies. Further, the amount of derivatives is scaled by firm size (total assets) to eliminate absolute values in favor of relative ones. This helps standardize the differences between firms to achieve a homogenous picture of the results.
The sample has been combined by taking twenty of the largest electric utility firms in Europe.2 These are: EDF (France), GDF Suez (France), E.ON (Germany), EDP (Portugal), Iberdrola (Spain), SSE (UK), Centrica (UK), Enel (Italy), Fortum (Finland), EnBW (Germany), CEZ (Czech Republic), Vattenfall (Sweden), Axpo (Switzerland), Dong (Denmark), Statkraft (Norway), A2A (Italy), Alpiq (Switzerland), BKW (Switzerland), Drax (United Kingdom) and Edison (Italy). This choice was dictated primarily by the availability and transparency of the necessary data, as well as the fact that larger companies have proportionally larger balance sheets, making random errors proportionately less significant. On the other hand, this may create a bias in the validity of the final conclusions, in the sense that they would only apply to large companies. The actual numbers were obtained by analyzing the end-of-year balance sheets and the respective notes to the financial statements, all included in the annual reports available online. The timeframe chosen was the period from 2007 to 2012, giving six sets of data per company. That timeframe is partly justified by data availability again (quality of annual reports declines back in time), but mostly by the need for the study to focus on the effects of the recent financial crisis, as discussed above.
The average characteristics of the sample calculated over the period investigated are shown below. The absolute values expressed originally in currencies other than the Euro were converted into the Euro for comparability using official ECB exchange rates from the end of the period investigated (31/12/2012)3. Other values are relative and immune to currency differences. This, overall, allows to compare the average values between the firms investigated, despite their operating using different currencies.
2
According to a combination of sources: see Appendix 3
(absolute values marked *, in millions EUR) EDF GDF Suez E.ON EDP Iberdrola SSE Centrica Total intangible assets* 14,951.0 33,782.0 22,491.3 7,808.3 18,305.4 1,362.2 3,646.8 Current assets* 61,735.2 50,781.3 43,734.2 6,300.0 15,086.1 8,104.7 8,186.3 Inventories* 11,847.0 3,621.7 4,454.8 318.9 2,048.3 336.1 483.2 Total Assets* 224,851.8 170,220.9 148,855.3 38,652.9 87,985.7 20,520.2 22,554.4 Current liabilities* 53,824.2 46,748.4 40,618.7 7,869.9 15,679.3 8,321.3 7,888.9 Total Equity* 31,432.0 62,566.7 43,593.3 9,904.7 30,254.2 4,382.0 5,997.8 Market cap* 68,109.8 56,188.8 50,716.7 10,631.4 32,241.5 14,367.4 18,208.6 State-owned? NO NO NO NO NO NO NO Market-to-book ratio= 32.3% 36.9% 34.4% 28.5% 38.1% 76.8% 81.9% Total Assets*= 224,851.8 170,220.9 148,855.3 38,652.9 87,985.7 20,520.2 22,554.4 Leverage= 7.2 2.8 3.5 3.9 2.9 4.8 3.8
Tangible Assets ratio= 93.5% 80.2% 84.9% 80.1% 79.1% 93.4% 84.1%
Quick ratio= 93.1% 100.8% 97.7% 76.1% 84.7% 92.4% 96.8%
Absolute derivatives use* 340.33 544.80 214.50 9.46 139.81 -422.40 -43.09 Derivatives scaled by total assets= 0.167% 0.368% 0.134% 0.019% 0.210% -1.991% -0.190%
(absolute values marked *, in millions EUR) Enel Fortum EnBW CEZ Vattenfall Axpo DONG Total intangible assets* 33,973.7 361.2 1,870.1 734.4 4,808.6 1,207.0 368.9 Current assets* 31,444.8 2,732.8 9,916.7 4,433.0 15,429.5 5,806.3 5,702.2 Inventories* 2,616.2 419.8 962.1 225.8 1,783.1 202.1 461.6 Total Assets* 156,083.0 21,230.5 33,949.7 20,890.1 57,883.6 14,363.2 17,141.4 Current liabilities* 33,146.3 2,409.5 8,161.0 4,584.0 12,496.4 3,427.2 4,163.2 Total Equity* 43,364.2 9,212.8 6,486.9 8,546.9 16,219.8 6,470.1 6,528.9 Market cap* 34,845.5 17,487.3 10,226.8 18,678.5 29,018.3 8,198.1 3,262.9
State-owned? NO NO NO NO YES YES YES
Market-to-book ratio= 22.9% 85.3% 30.9% 96.4% 56.8% 61.5% 19.2% Total Assets*= 156,083.0 21,230.5 33,949.7 20,890.1 57,883.6 14,363.2 17,141.4
Leverage= 3.8 2.3 5.3 2.4 3.6 2.2 2.6
Tangible Assets ratio= 78.4% 98.3% 94.5% 96.4% 92.4% 91.6% 97.7%
Quick ratio= 86.8% 99.5% 109.4% 91.9% 108.8% 165.0% 126.3%
Absolute derivatives use* -1,219.33 44.83 64.83 -214.44 220.27 -5.95 55.57 Derivatives scaled by total assets= -0.766% 0.163% 0.190% -0.996% 0.157% -0.036% 0.603%
(absolute values marked *, in millions EUR) Statkraft A2A Alpiq BKW Drax Edison Total intangible assets* 355.0 1,154.2 1,650.3 167.7 16.7 3,512.2 Current assets* 3,553.8 2,786.8 3,944.3 1,625.2 918.4 3,902.5 Inventories* 131.9 252.2 91.7 21.8 184.6 305.8 Total Assets* 19,125.8 11,701.6 14,662.4 5,451.0 2,445.9 15,582.7 Current liabilities* 3,449.7 2,316.2 3,277.8 612.4 653.0 3,721.5 Total Equity* 8,732.9 4,290.4 5,575.7 2,443.9 1,188.1 7,824.7 Market cap* 13,271.5 3,837.4 6,141.8 3,249.0 2,236.1 6,124.8 State-owned? YES NO NO NO NO NO Market-to-book ratio= 72.5% 33.1% 40.4% 62.6% 94.3% 39.4% Total Assets*= 19,125.8 11,701.6 14,662.4 5,451.0 2,445.9 15,582.7 Leverage= 2.2 2.8 2.7 2.3 2.4 2.0
Tangible Assets ratio= 98.2% 90.1% 89.0% 97.0% 99.4% 77.4%
Quick ratio= 99.5% 109.1% 117.5% 263.8% 115.1% 95.5%
Absolute derivatives use* -79.77 1.38 -21.54 6.97 -32.45 -6.67 Derivatives scaled by total assets= -0.555% 0.007% -0.143% 0.132% -2.202% -0.061%
The summary shows some basic descriptive characteristics of the twenty firms investigated. It can be seen that the largest of them all (by the value of total assets) is EDF, the only one with assets worth over €200bln. However, GDF Suez, E.ON and Enel are also very large, with assets worth around €150bln. The smallest firms are Drax and BKW, with assets worth €2bln and €5bln respectively. This shows how much variation in size there is between the firms investigated, which adds versatility to the sample. On the other hand, the scaled use of derivatives varies between -2.2% and 0.6%, which is a much smaller range. The range of the values of the Market-to-book ratio is quite wide, going from 19.2% in the case of Dong, up to 96.4% in the case of CEZ. However, since none of the values is higher than 100%, it implies that all of the sample firms’ assets are overvalued on the market. Further, the values of leverage range from 2.0 in the case of Edison to 7.2 at EDF. Even though the bracket is wide, the majority of values oscillate around the value of 2.5. Similarly, the tangible assets ratio does not bring many surprises: obviously, it cannot be higher than 100%, but most of the values revolve around 90%, ranging from 77.4% to 99.4%. The quick ratio values range from 76.1% (low liquidity) to 263.8% (high liquidity), giving a wide spectrum of liquidities to test the hypotheses over.
The above paragraph shows how diverse the chosen firms are, despite being active in similar geographic and business areas. This yields some gravity to this study, as it makes its results applicable to a wider range of entities.
This section explained the regression method used to build the model. It also outlined the proxies chosen and justified them. Additionally, it described the hypothesized relationships between the proxies and derivatives use, which can be summarized as follows: - Market-to-book ratio: higher ratio means assets are undervalued and the underinvestment problem will be more severe, so more derivatives are needed to manage it: positive relation; - Total assets: more assets mean larger firm, so the bankruptcy costs grow slower, so relatively less hedging is needed: negative relation;
- Leverage: higher leverage means more risk and higher expected costs of financial distress: positive relation;
- Tangible Assets ratio: higher ratio means that more assets can be easily liquidated and the expected costs of financial distress are lower: negative relation;
- Quick ratio: higher ratio means the firm can meet its short-term liabilities without incurring high costs, so less hedging is needed: negative relation.
The above relationships are summarized in the table below.
Proxy variable Name in model Coefficient Market-to-book ratio MTB +
Total assets TA -
Leverage LEV +
Tangible assets ratio TAR -
Quick ratio QR -
5. Results and analysis
This section outlines the results of the previously mentioned regression. It presents the estimated coefficients and discusses these results in relation to the set hypotheses.
The regression is run as specified previously. The resulting estimates of the model coefficients are presented below (significance level key can be found in the Appendix):
Coefficient Variable Estimate Std. Error t value Pr(>|t|) Significance (Intercept) α 1.15E-01 8.76E-02 1.315 0.19209
MTB Market-to-book ratio -1.31E-02 8.89E-03 -1.468 0.14573 TA Total Assets -2.09E-08 8.27E-08 -0.253 0.80098 LEV Leverage -2.19E-02 5.03E-03 -4.357 0.00004 **** TAR Tangible Assets Ratio -1.06E-01 1.10E-01 -0.960 0.33993 QR Quick Ratio 1.41E-02 1.29E-02 1.091 0.27856
A EDF 1.37E-01 3.76E-02 3.646 0.00046 ****
B GDF Suez 2.63E-02 1.87E-02 1.402 0.16451
C E.ON 4.42E-02 2.04E-02 2.170 0.03280 **
D EDP 4.33E-02 1.62E-02 2.664 0.00924 ***
E Iberdrola 2.74E-02 1.49E-02 1.841 0.06908 *
F SSE 6.34E-02 2.63E-02 2.411 0.01807 **
G Centrica 5.06E-02 1.82E-02 2.777 0.00675 ***
H Enel 3.63E-02 1.88E-02 1.931 0.05680 *
I Fortum 3.67E-02 2.70E-02 1.358 0.17800
J EnBW 8.95E-02 2.81E-02 3.184 0.00203 ***
K CEZ 3.88E-02 5.26E-02 0.738 0.46238
L Vattenfall 6.26E-02 4.73E-02 1.324 0.18913
M Axpo 1.32E-02 2.23E-02 0.590 0.55697
N DONG 3.57E-02 2.93E-02 1.216 0.22729
O Statkraft 2.80E-02 2.99E-02 0.937 0.35138
P A2A 2.68E-02 1.99E-02 1.345 0.18217
Q Alpiq 2.40E-02 2.02E-02 1.185 0.23946
R BKW 7.23E-03 3.33E-02 0.217 0.82859
S Drax 1.40E-02 2.81E-02 0.500 0.61863
Y1 2007 -4.59E-03 1.04E-02 -0.442 0.65985 Y2 2008 1.13E-02 8.07E-03 1.395 0.16657 Y3 2009 5.22E-03 7.16E-03 0.728 0.46836 Y4 2010 -4.94E-03 7.14E-03 -0.692 0.49102 Y5 2011 3.29E-03 7.08E-03 0.466 0.64276 Residuals Min -0.094372 1Q -0.008087 Median -0.000002 3Q 0.006836 Max 0.119397
Residual standard error: 0.02225
Degrees of freedom 85
Multiple R-squared: 0.3358
Adjusted R-squared: 0.1092
F-statistic: 1.482
It can be seen that some of the firm-specific dummies are relatively significant, which means that there is a large amount of peculiarities about these firms that determine their use of derivatives. When it comes to the hypothesized proxies, the results are weak. The only variable with a significant coefficient is Leverage. However, the coefficient’s direction is opposite to the hypothesized – negative. This would imply that lower leverage stimulates the use of derivatives. The rest of the proxies, despite being statistically insignificant at the commonly accepted levels of confidence (with the Total Assets coefficient having a wildly high p-value), show variety in the correctness of their coefficients’ directions – while Total Assets and the Tangible Assets Ratio are correct, the Market-to-book Ratio and Quick Ratio are not. Additionally, the time dummies display a fluctuating pattern of derivative use, rather than a parabolic one as hypothesized. However, due to the high p-values of those coefficients, it is difficult to say whether this bears any significance. Thus, most of the coefficients are statistically insignificant using the regular confidence levels, as shown above. This is a problem, because it means it is rather difficult to state that these variables have any effect on the use of derivatives.
However, these results might be distorted by the strongly significant firm-specific dummies. Thus, a second regression is run using only the five original proxies to see if the results are different. They are shown below:
Coefficient Variable Estimate Std. Error t value Pr(>|t|) Significance (Intercept) α 2.64E-02 2.59E-02 1.019 0.31060
MTB Market-to-book ratio -1.15E-02 5.51E-03 -2.081 0.03980 ** TA Total Assets 1.10E-08 1.45E-08 0.757 0.45100 LEV Leverage -2.33E-03 1.73E-03 -1.347 0.18080 TAR Tangible Assets Ratio -2.81E-02 3.16E-02 -0.888 0.37670 QR Quick Ratio 7.67E-03 5.67E-03 1.351 0.17940
Residuals Min -0.147576 1Q -0.006183 Median -0.000329 3Q 0.008653 Max 0.114662
Residual standard error: 0.02302
Degrees of freedom 109
Multiple R-squared: 0.08767
Adjusted R-squared: 0.04582
F-statistic: 2.095
Table 4. Results of regression no. 2
Now the results are even worse – none of the coefficients is significant, except for the Market-to-book ratio. However, the direction of the effect is contrary to the one predicted – it is negative. Additionally, the directions of the other coefficients are opposite to the
hypothesized ones, apart from the Tangible Assets Ratio’s. Also, the R2 coefficient is much lower in the second model (0.0877 compared to 0.3358), indicating that the first model better explains the determinants of derivative hedging.
The conclusion that can be reached here is that the amount of derivative hedging a particular firm implements to a large extent depends on variables different than the ones here presented. Whether these are quantifiable and measurable, or abstract and qualitative, is impossible to say on the basis of this study. It is, however, concerning, that the commonly investigated factors do not apply in this case. It raises a question of whether the electricity market differs from other sectors to such an extent or if the reason for this discrepancy is different. Either way, it shows how much more research this field needs and points in a potential direction for future studies.
6. Summary and conclusion
This study has focused on the determinants of derivative hedging in large European energy firms in the recent years. Utilizing a multiple regression model, it investigates the validity of the three most common hypotheses in this field of research over a sample picked from an under-researched sector. The three factors that are often said to determine the amount of corporate derivative hedging are: underinvestment, costs of financial distress and availability of hedging substitutes. Additionally, it attempts to investigate how much of it is also determined by unique firm-specific factors and the state of the economy, as determined by the stage of the recent crisis.
The main conclusion is that these popular hypotheses should not be accepted for this sample, assuming the usual levels of confidence. Some of the proxies showed slight significance, but most of them, despite being vaguely correct in the direction of their influence, were rather weakly estimated. On the other hand, the firm-specific effects appear to be quite strong in some cases. The above results imply that in recent years the traditionally accepted determinants of corporate hedging do not apply in the European energy market – more relevant are firm-specific effects that are more difficult to measure.
This study demonstrates how much more research in this field needs to be done before hedging can be fully understood. Especially in the face of all the changes that the recent financial crisis has stimulated, it is important to extend the research to more sectors, more markets and a wider timeframe. This could yield results that would be universally true, instead of being only valid for a particular type of sample.
Bibliography
Bartram, S. M., Brown, G. W., & Fehle, F. R. (2009). International evidence on financial derivatives usage. Financial management, 38(1), 185-206.
Barclay, M. J., Smith, C. W., & Watts, R. L. (1997). The determinants of corporate leverage and dividend policies. Journal of Financial Education, 1-15.
Deng, S. J., & Oren, S. S. (2006). Electricity derivatives and risk management. Energy, 31(6), 940-953.
Géczy, C., Minton, B. A., & Schrand, C. (1997). Why firms use currency derivatives. The Journal of Finance, 52(4), 1323-1354.
Hentschel, L., & Kothari, S. P. (2001). Are corporations reducing or taking risks with derivatives?. Journal of Financial and Quantitative Analysis, 36(01), 93-118.
Holttinen, H. (2005). Optimal electricity market for wind power. Energy Policy, 33(16), 2052-2063.
Howton, S. D., & Perfect, S. B. (1998). Currency and interest-rate derivatives use in US firms. Financial Management, 27(4), 111-121.
Hull, J. C. (2012). Fundamentals of futures and options markets (8th ed.). Boston, MA: Pearson Education Inc.
Liu, M., & Wu, F. F. (2007). Risk management in a competitive electricity market. International Journal of Electrical Power & Energy Systems, 29(9), 690-697.
Nance, D. R., Smith, C. W., & Smithson, C. W. (1993). On the determinants of corporate hedging. The Journal of Finance, 48(1), 267-284.
Statkraft Markets GmbH. (2012). Annual Report 2012. Retrieved from: http://www.statkraft.com/Statkraft/Documents/en/Reports%20and%20presentations/2 012/Statkraft%20Markets%20GmbH_Annual%20Report%202012%20engl._WEB_tc m9-27519.pdf on 01.05.2014.
Stulz, R. M. (2004). Should we fear derivatives? (No. w10574). National Bureau of Economic Research.
Weron, R. (2000). Energy price risk management. Physica A: Statistical Mechanics and its Applications, 285(1), 127-134.
Appendix 1. Proxies: Market-to-book ratio = Assets Total Cap Market ratio P/B Equity Total Assets Total Leverage
Tangible Assets ratio =
Assets Total Assets Tangible Quick ratio = s Liabilitie Current s Inventorie -Assets Current test Acid
2. Link to the software website: http://www.r-project.org/
3. Sources for sample company selection:
http://www.statista.com/statistics/263424/the-largest-energy-utility-companies-worldwide-based-on-market-value/
http://en.wikipedia.org/wiki/List_of_European_power_companies_by_carbon_intensity http://en.wikipedia.org/wiki/List_of_largest_European_companies_by_revenue
http://www.iie.org.mx/promocio/notisel/re924a-brochure.pdf
4. Significance level key
Mark Significance level
* 0.1 10%
** 0.05 5%
*** 0.01 1%
