Foreign Exchange Exposure: Risk Management During Crisis

(1)

Foreign Exchange Exposure:

Risk Management During Crisis

Evidence From the World’s Largest

Integrated Oil & Gas Companies

Advanced International Business Management & Marketing Thesis

Linsey Teuben

S173746 A98000351

l.teuben@student.rug.nl linsey.teuben@ncl.ac.uk Mobile: 06 348 74 679

Based at the University of Groningen

(2)

Abstract

Multinational corporations (MNCs) enjoy benefits of scale compared to smaller domestic corporations, but they also face a higher level of risk. One of these risks is related to higher levels of exchange rate exposure. Research into this field yielded inconclusive results as MNCs exhibit statistically insignificant sensitivity to this foreign exposure. Many

researchers have contributed this to the fact that MNCs systematically hedge their exposure risk. The extent to which firms hedge, or the exact way in which they do so have not been investigated before, because until recently MNCs were not required to release this

information. Also, research has always focused on relatively stable economic periods. This thesis attempts to fill these two gaps in the literature by addressing a period of global economic crisis in light of this topic and by examining the extent to which specific MNCs hedge their foreign exposure. A small but highly detailed sample of some of the world’s largest MNCs operating in the integrated oil and gas industry serves as the input for this research. Hedging, in theory, lowers exposure risk and thereby increases firm value. This research shows that most firms do systematically hedge a large portion of their foreign exposure. High levels of hedging do however not necessarily increase firm value. This research provides evidence for the fact that even though hedging lowers risk it also decreases potential benefits from foreign exposure, especially during crisis.

Supervisors: Rajiv Kozhikode and Chris Carter

(3)

Results ... 46 Intermediate Conclusions ... 48 Model Summary ... 49 Discussion ... 51 Conclusion ... 54 References ... 56 Appendix ... 60 SPSS Output ... 60 Multicollinearity ... 60 Conceptual Statistics ... 60 Regression Analyses 2001 - 2010 ... 65 Regression Analysis 2001 – 2006 ... 70 Regression Analyses 2005 – 2010 ... 71 Model Summary ... 73

Annual Report Outtakes ... 84

Royal Dutch Shell ... 84

BP ... 86

(5)

List of Tables

Page 1: 2: 3: 4: 5: 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20:

Regression Models Royal Dutch Shell Class A Shares Regression Models Royal Dutch Shell Class B Shares Regression Coefficients Royal Dutch Shell Class A Shares Regression Coefficients Royal Dutch Shell Class B Shares Systematic Risk Factors Royal Dutch Shell

Conceptual Statistics Royal Dutch Shell Regression Models BP

Regression Coefficients Systematic Risk Factors Conceptual Statistics BP

Regression Models ExxonMobil Regression Coefficients ExxonMobil Systematic Risk Factors ExxonMobil Conceptual Statistics ExxonMobil

Comparison of gamma coefficients and significance (2003 – 2006) Gamma coefficient ExxonMobil 2001 - 2006

Total Assets in Derivatives / Total Assets Royal Dutch Shell Total Assets in Derivatives / Total Assets BP

(6)

List of Figures and Graphs

Page 1: Change in Real GDP Global Economy (IMF, 2011) 26

1: 2: 3: 4: 5: 6: 7: 8:

Broad Dollar Index 2001 – 2010 NYSE Composite Index 2001 – 2010 Opening Prices Class A Shares Opening Prices Class B Shares Opening Prices BP Shares

Opening Prices ExxonMobil Shares Comparing Courses: Broad Dollar Index

Comparing Courses: ExxonMobil Share Price Index

(7)

Acknowledgement

First and foremost, I would like to thank my supervisors, Rajiv Kozhikode and Chris Carter. Thank you for your guidance and your extensive feedback. Sometimes you get stuck in one mindset, and you need someone to help you see things from a slightly different prospective. I would like to thank my Groningen based supervisor Rajiv especially, since he took time to meet with me several times and guide me through my literature selection and the structuring of my work. I would also like to thank the program coordinators at the universities of Groningen and Newcastle and of course the secretaries for their support.

(8)

Preface

(9)

Dedication

(10)

Introduction

Every corporation faces risk, and though multinational corporations (MNCs) enjoy benefits of scale compared to smaller domestic corporations, they also face a higher level of risk. One of these risks is related to higher levels of exchange rate exposure (Adler and Dumas, 1984), resulting from assets, liabilities and/or transactions exposed to foreign currency. Past research into exchange rate exposure and firm value (a.o. Bartram and Bodnar, 2007; Choi and Prasad, 1995; Donnelly and Sheehy, 1996; He and Ng, 1998; and Muller and Verschoor, 2006a-b, Muller and Verschoor, 2007; Chue and Cook, 2008) has lead to various conclusions, ranging from sometimes significant to mostly insignificant sensitivity to exchange rate exposure exhibited by MNCs. Bartram and Bodnar (2007) classify this mismatch between theoretical and empirical results as the Exchange Rate Exposure Puzzle. They attribute statistical insignificance of analyses to the endogenous nature of firms’ hedging policies to mitigate financial exposure risk. Given that currency fluctuations are an important source of macroeconomic instability and that companies are directly and indirectly exposed to these fluctuations, it is logical to assume that exchange rate volatility does affect firm value to some extent. Intuitively, one would expect that global financial crisis and increased levels of macroeconomic instability will equally impact exchange rate volatility and MNCs’ firm value. There are two main bodies of neglected empirical matter in this field of research. The first relates to the fact that past research has always focused on relatively stable economic conditions, neglecting the effects of global financial and economic crisis. The second relates to the fact that the extent to which companies hedge and the effect of those specific policies has not been investigated thoroughly. The results presented in this thesis will aid in filling that gap.

(11)

that the total revenues of the sector's top three players reached $920.6 billion in 2004,

demonstrating the highly consolidated nature of the sector. Many firms operating in this

industry exhibit a high degree of vertical integration and some of the largest players in this industry also dominate the list of largest global players (Datamonitor, 2010a-c). The Royal Dutch Shell Group accounts for 18.1% of the sector’s value, closely followed by the Exxon Mobil Corporation with 16% and BP Plc with a further 15.3%. Even though these companies are strong and powerful, with a solid financial base for their operations, the integrated oil and gas industry is highly volatile and subject to many pressures. The influence of OPEC for example is a constant hazard for MNCs operating in this industry (The Economist, 2011a) and natural disasters and other natural or macroeconomic phenomena can cause massive supply chain shocks (The Economist, 2011b). This high degree of industry volatility, combined with rapidly changing economic conditions over the past 10 years, provide an interesting setting for research into exchange rate exposure and financial risk management. The findings presented in this thesis are thus of interest to academics and practitioners alike.

(12)

(13)

Theory

Exchange Rate Exposure

When an MNC engages in foreign trade or foreign operations of any kind involving currencies other than the home currency, the corporation’s assets, liabilities and transactions become exposed to exchange rate volatility, also called currency fluctuations, given that they need to be translated at varying exchange rates and reconciled. Among others, Eiteman et al. (2006) and Muller and Verschoor (2006b) identify different types of exchange rate exposure, which can be grouped into three main categories: translation, transaction and economic exposure. Translation exposure is related to the effect of an exchange rate change on financial statement items. Transaction exposure is related to the effect of an exchange rate change on outstanding obligations, such as imports and exports. Economic exposure is related to the effect of an exchange rate change on the net present value of expected net cash flows from direct investment projects. Whereas direct exposure can mostly be managed through various forms of financial hedging, indirect exposure and its effect on competitiveness is harder to estimate (Muller and Verschoor, 2006b).

(14)

(15)

framework for exchange rate exposure suggests a significant impact of exposure on firm value, the empirical evidence is lacking. They explain this Exchange Rate Exposure Puzzle by highlighting that MNCs will mitigate financial risk through hedging of exposure in financial markets, which then minimises foreign exchange exposure. Though intuitively, greater scope of foreign operations means greater exposure, Bartram and Bodnar (2007) indicate that this greater risk results in an even higher degree of endogeneity of hedging policies in MNCs to mitigate exchange rate exposure and risk. Chue and Cook (2008) also mention that firms can hedge exposure, or offset risk through foreign-currency-denominated cash flows originating from international operations and trade. Though sometimes believed to be a result of methodological differences, Bartram and Bodnar (2007) explain that the lack of empirical evidence for significant sensitivity to exchange rate exposure is the result of firms’ strategic and often endogenous responses to changes in the exchange rates involved in their operations. Even though much of the literature on exchange rate exposure indicates that MNCs exhibit statistically insignificant sensitivity to exchange rate volatility, conceptual evidence does suggest a relationship between currency fluctuations and firm value. Past research on this topic has always focused on relatively stable historic economic periods. As indicated by Muller and Verschoor (2006b: 405): The particular effect of increased exchange rate

volatility during periods of financial turmoil on shareholder wealth deserves to be empirically assessed. It would thus be interesting to see if and to what extent the recent global economic

(16)

fluctuations, and finally on firm value, measured by the firm’s stock prices. The main research question for this proposed study is: What are the effects of global financial crisis on MNCs’ sensitivity to exchange rate volatility and consequently on firm value? More specific emergent research questions include: What happens to exchange rate sensitivity and stock returns during (global) crisis; Do the firms show different sensitivities to exchange rate exposure in the 2006 – 2010 period compared to earlier more stable economic periods (i.e. the 2001 – 2005 period); Does the exchange rate sensitivity impact firm value at all? To answer the first set of questions, and based on the existing literature as set out in the literature review, the following hypotheses can be devised and tested:

H1a: Increased exchange rate volatility during the 2006 – 2010 financial crisis has increased the sensitivity to exchange rate exposure of all firms in the sample. H1b: Increased exchange rate volatility during the 2006 – 2010 financial crisis has

had a negative effect on the stock returns of all firms in the sample.

To control for firm size and industry level effects, research is focused on the three largest MNCs in the global integrated oil and gas industry (Datamonitor, 2005): Royal Dutch Shell Plc (The Netherlands), BP Plc (United Kingdom) and the ExxonMobil Corporation (United States). In addition to answering some of these research questions through case study and event study research, many of the questions posed can be answered by means of statistical testing.

Determinants of Exchange Rate Exposure

(17)

the foreign to total sales ratio and the foreign to total assets ratio impact the degree to which MNCs have to deal with exposure. In other research, Muller and Verschoor (2006a) also find that European firms with strong short-term liquidity positions (i.e. firms with lower dividend payout ratios) have less incentive to hedge and hence have larger exchange rate exposures. This means that bigger European MNCs will tend to have larger exposure to exchange rate fluctuations. The same authors find that in the US, highly leveraged MNCs or MNCs with a lower quick ratio tend to have lower exposure to exchange rate risk (Muller and Verschoor, 2007). Nevertheless, Muller and Verschoor (2006b) state that there is no straightforward model integrating all the complexity of the effects of exchange rate shocks on firm value, or a generalised overview of the most important parameters influencing currency risk exposure.

The literature suggests that MNCs tend to be net exporters due to their size, and additionally that an appreciation of the home country currency will thus most likely have a negative effect on the home country firm’s value. Even though all case firms reconcile their annual reports in U.S. dollars and the commodities the firms trade in are directly or indirectly priced in dollars, ExxonMobil is the only American firm in the sample, which allows for the exploration of possible country of origin effects. Theoretically, an appreciation of the dollar will make ExxonMobil’s exports more expensive, reducing the firm’s competitiveness abroad. We can therefore hypothesize the following:

H2a: An appreciation of the value of the dollar will have a negative impact on the stock returns of ExxonMobil.

H2b: An appreciation of the value of the dollar will have a positive impact on the stock returns of Royal Dutch Shell and BP.

(18)

Hedging Direct Exposure

There are several ways in which MNCs can hedge their foreign exposure. Among others, Bartram and Bodnar (2007) indicate that firms may use financial and operational hedges to reduce foreign exposure risk. Financial hedging is used to reduce cash flow volatility and may take the form of foreign currency derivatives or other financial instruments. Operational hedging comprises strategies related to production, sourcing, pricing and marketing flexibility to respond to movements in exchange rates, and as such is hard to measure in a quantitative research setting.

Bartram and Bodnar (2007) have explained why there appears to be an Exposure

Puzzle. In earlier research, Dewenter et al. (2004: 124) conclude that stock prices are indeed

sensitive to contemporaneous changes in currency values and that the resulting exchange rate risk is economically relevant though they also find that the puzzle remains. According to them, the puzzle can in part be explained by the lack of specific information on hedging and detailed hedging explanations. Nydahl (1998) highlights that few companies reported their hedging positions in the past, which has made the problem almost impossible to investigate. Stulz (1999) however explains that when MNCs expand internationally and decide to participate in global capital markets, i.e. decide to list their stock on international stock exchanges, they must comply with stricter disclosure regulations. Globalization thus exerts pressure on MNCs for better performance and greater transparency. Indeed we find that large MNCs listed on one or more foreign stock exchanges are more transparent now, and (are forced to) disclose part of their hedging policies. This enables an investigation of the extent to which the exact hedging policies influence exposure and sensitivity to foreign exchange rate changes today.

(19)

related to the use of currency derivatives. This implies that firms indeed face increasing exposure as the percentage of foreign sales with regard to total sales increases. Hedging, specifically hedging using currency derivatives reduces the foreign exposure and MNC faces. Nydahl (1998) however also finds that using lagged exchange rate changes as independent variables in the regression equation, a weak impact is found. Under the assumption that changes in exchange rates may have a delayed (lagged) effect on investors and stock prices, Nydahl (1998) finds no evidence to support a mispricing hypothesis, more precisely to support that using lagged changes in the independent variables significantly changes the results from the results using more contemporaneous changes in the independent variables. Additionally, past research suggests that breaking exposure down to single currencies does not change the regression results (Nydahl, 1998) which justifies using a multilateral exchange rate in future research. It must be noted however, that the foreign exposure a firm faces is most likely different from the trade weights used to calculate the weighted exchange rate index. Dewenter et al. (2004) indicate that this will weaken the correlation between changes in a firm’s stock prices and changes in the currency index.

Dewenter et al. (2004: 142) point out that hedging incurs costs, so in practice, MNCs with sophisticated hedging policies may only hedge to the point where the marginal cost of

additional hedging equals the expected marginal benefit of reduced exposure. Dewenter et al.

(20)

start hedging they will incur a premium (i.e. an increase in firm value). Firms that stop hedging or remain unhedged will not incur this premium.

It thus appears that hedging direct exposure influences firm value to a certain extent. Recent development on the global capital markets have dictated that large multinationals who wish to participate in these markets must become more transparent. This increased transparency on firm level for the first time enables a more in-depth study of the hedging policies employed by MNCs to manage the risk they face through foreign exposure.

So another identified gap in the literature relates to the notion that previous research has been unable to identify the exact effect of hedging or hedging policies, since firms were not required to release this sensitive information. Participation in global capital markets requires of MNCs that they become more transparent and thus release more information regarding their hedging positions. Questions emerging from this new development include: How much of their foreign exposure do firms hedge; How do firms hedge their exposure; How do firms account for their hedging practices? Then, more specifically: Does hedging reduce firms’ sensitivity to exchange rate fluctuations; and Does hedging thereby indirectly increase firm value?

Case study research will provide answers to the questions posed in relation to how exactly and to what extent Shell, BP and ExxonMobil hedge their foreign exposure. Since the literature suggests that hedging will decrease foreign exposure risk, we can hypothesize the following moderating effect of hedging on sensitivity:

H3a: The firms remaining unhedged against foreign exposure will be more sensitive to exchange rate volatility than the firms that choose to hedge their foreign exposure.

(21)

Method

Variables

Two types of price data have been gathered to compose the dependent and independent variables in the regression models and for the conceptual analysis: The opening prices of stocks and the price indices. Since the NYSE Composite and the Broad Dollar are presented as indices, the regression analyses are performed using the share price indices of the case firms Royal Dutch Shell, BP and ExxonMobil. This data has been gathered from the DataStream database at the University of Groningen. Changes in these indices are recorded on a daily basis, leading to a highly detailed sample over a period of 10 years with thousands of entries. For the purpose of linear regression analysis, variables were comprised for the three different firms, the NYSE Composite Index and the Broad Dollar Index by calculating the relative difference in the values of the data gathered since the absolute values cannot be used for comparison purposes. These changes, as mentioned, are measured on a daily basis for each of the dependent (Royal Dutch Shell, BP and ExxonMobil stock price indices) and the independent variables (Broad Dollar and NYSE Composite indices). Regression analyses are repeated using one single dependent variable that comprised all the case firms, for the purpose of control testing and to allow for improved model building. These tests also lead to the composition of a model summary.

(22)

no evidence to suggest multicollinearity between these variables. So based on multicollinearity tests performed in SPSS prior to the regression analyses, there are no apparent multicollinearity problems. As a control test, the regression analyses performed to devise the model summary each include a set of multicollinearity statistics, known as Tolerance and VIF, to ensure that no multicollinearity problems have occurred.

Finally, the absolute values of the stock returns are used for the analyses that are not based on multivariate linear regression. The opening prices of the shares of Royal Dutch Shell, BP and ExxonMobil are used in a more conceptual analysis of price fluctuations, stocks returns and consequently firm value in this research.

The Regression Models

The Ordinary Least Squares (OLS) regression method is applied in this research. The OLS regression method is parametric given that it is defined by a finite number of estimated parameters based on the data entered. The goal of the OLS method is to produce predictor parameters that may be used to estimate values in the dependent variable based on the characteristics of (and changes in) the independent variables in the model. Since that is exactly the aim of this research, the OLS is an appropriate regression method. As mentioned, the regression models are based on the following regression equation: Rιt = αι + βιRMt + γ

δiRFXt + ειt , where Rιt represents stock returns on firm level, as a measurement for firm value.

RMt represents the average returns of the NYSE stock market (included to improve relevance

of the regression model (Bartram and Bodnar, 2007)), and δiRFXt represents changes in the

(23)

a fixed and identical number of variables is used in every regression. Alternatives to using the OLS regression method include using Tobin’s Q as a measurement for changes in firm value, but regression analysis is more appropriate for this research given that we are trying to find out if changes in the dependent variables can be explained by changes in the independent variable(s) at the measurement levels described above.

Practical Constraints

Practical constraints are most applicable and limited to the case study research, given that stock market data and exchange rate data will be more readily available than detailed information on foreign currency exposure, or information on hedging in financial markets and using foreign currency derivatives to mitigate exposure risk on an individual MNC basis. This case study research is imperative to control for external variables that may have influenced the regression models tested in the earlier stages of the research. Coverage in news article will also be limited given that for competitive reasons, most firms will not reveal much information on their risk management practices.

Ethical Implications

(24)

Approach

The conducted research consists of both quantitative research and case study research. To test the hypotheses outlined in the previous section a wide range of data needs to be gathered and transformed before statistical tests can be performed using statistical analysis software. Justification of the choice of case firms lays in the fact that all firms have a significant market share in the integrated oil and gas industry (Datamonitor, 2005) and are all (cross)listed on the New York Stock Exchange (NYSE) to allow for comparison. These companies reconcile their annual reports in U.S. dollars, further facilitating the comparison. Additionally, most of the markets in which the case firms operate are directly or indirectly priced in U.S. dollars (Royal Dutch Shell, 2011; BP; 2011; ExxonMobil, 2011).

The regression equation for this research is based on past research on exchange rate volatility and firm value, which has been performed using the following regression equation: Rιt = αι + βιRMt + γ δiRFXt + ειt (Bartram and Bodnar, 2007: 644) where Rιt represents stock

returns on firm level, RMt represents the average returns of the stock market (included to

improve relevance of the regression model), and δiRFXt represents changes in the exchange

rate. The sensitivity coefficients (β and γ) and systematic risk factor (α) are produced by the regression analyses for further analysis and comparison.

Using this regression equation means the following data had to be gathered:

(25)

example of such an index is the Broad Dollar Index. The American Federal Reserve Bank calculates this index based on the relative weight of America’s trading partners and their currencies. Twenty-six economies are taken into account to calculate the Broad Dollar Index (American Federal Reserve Bank, 2005). Historical index levels can be accessed through online resources of the American Federal Reserve Bank. The method of gathering all the data mentioned above has resulted in the composition of a unique database.

(26)

Results

The Effects of Global Crisis

Results

The World Economic Outlook, published semi-annually by the International Monetary Fund (IMF) gives an overview of the economic state of the global economy. Data presented in the September 2011 World Economic Outlook (IMF, 2011) clearly depicts the economic turmoil of the past years.

Figure 1: Change in Real GDP Global Economy (IMF, 2011)

(27)

Graph 1: Broad Dollar Index 2001 – 2010 In fact, the range of the index between 2007 and 2010 is 20.3 points, with a minimum of 94.8 and a maximum of 115 points. During the same amount of time preceding the crisis the index ranged from 106.1 and 124.7 points. This range of 18.6 points is considerably smaller than the range between 2007 and 2010. What can be concluded is that indeed the global crisis of the past years, accompanied by plummeting GDP growth rates in both developed and developing nations across the world has impacted the stability of the value of the dollar and, since their value is incorporated in the calculation of the Broad Dollar Index, other major world currencies.

In addition to increased volatility in the Broad Dollar Index, increased volatility is also observed in the NYSE Composite Index.

Graph 2: NYSE Composite Index 2001 – 2010 What can be observed is that between 2007 and 2010, the Broad Dollar Index shows a dip, followed by a peak, whilst the NYSE Composite Index shows a dip a bit later. The movements are practically mirror images of each other. It is interesting to investigate whether

(28)

or not this increase in exchange rate volatility has influenced the stock returns of Shell, BP and ExxonMobil. Regression analysis will show whether or not the stock prices of these firms show sensitivity to changes in the market portfolio or, more importantly, to changes in the weighted average value of the U.S. dollar.

Royal Dutch Shell

It is important to note that Royal Dutch Shell actually has two separate listings on all major stock exchanges. These are designated with the letters A and B. The difference between these two is related to taxation (Shell Investor Centre, 2011). Class A shares have a Dutch source for tax purposes and are subject to the Dutch tax system. Dividends paid on Class B shares are not subject to Dutch or UK withholding tax, and some shareholders (not including U.S. based shareholders) are even entitled to a UK tax credit. Also, their stocks are traded on the NYSE as American Depository Receipts (ADR), since Royal Dutch Shell is not an American company. For completeness, the analyses comprise both classes of shares.

Multiple regression analyses yielded the following results, where significance is determined by a P<0.05, i.e. a 95% confidence interval:

Class A Shares Explained Variance

Significant? F-Statistic Significant?

Full Period 62.0% Yes 1701.631 Yes

2003 - 2006 40.7% Yes 357.120 Yes

2007 - 2010 67.8% Yes 1095.449 Yes

Table 1: Regression Models Royal Dutch Shell Class A Shares

Class B Shares Explained Variance

2003 - 2006 37.3% Yes 308.727 Yes

2007 - 2010 63.7% Yes 913.665 Yes

(29)

First and foremost, all six models are statistically significant, which means that a statistically significant percentage of the variance observed in the dependent variables, here Class A and B shares, can to some extent be explained by movement in one or more of the independent variables. What can be seen is that the models for Class A shares explain more variance that occurs in the dataset than the models for Class B shares. The F-statistics, as general measures of the quality of the regression model are quite high, which renders the models even more significant. It is striking that the models for the period before the global crisis are weaker than the models for the crisis period, and that the crisis period models are even stronger than the model for the full time period. Now that the models have been identified as significant, the sensitivity coefficients and systematic risk factors may be examined.

The regression analyses produced the following sensitivity coefficients:

Class A β Significant? γ Significant? Full Period 1.030 Yes 0.058 No

2003 - 2006 1.009 Yes 0.121 No

2007 - 2010 1.035 Yes 0.025 No

Table 3: Regression Coefficients Royal Dutch Shell Class A Shares

Class B Shares β Significant? γ Significant? Full Period 1.019 Yes 0.051 No

2003 - 2006 1.007 Yes 0.061 No

2007 - 2010 1.022 Yes 0.044 No

(30)

statistical significance, we see that the coefficient is not larger during the second period, but in fact smaller. Explanations for this observed phenomenon include possible high levels of endogeneity of hedging practices or even an increased amount of hedging on firm level to counteract the instable economic conditions.

In addition to the figures above, the following systematic risk factors were produced:

Class A α Significant? Class B α Significant? Full Period 0.00003956 No Full Period 0.00003741 No 2003 – 2006 -0.00006994 No 2003 – 2006 -0.00009057 No 2007 – 2010 0 No 2007 – 2010 0 No

Table 5: Systematic Risk Factors Royal Dutch Shell From these figures, it can be concluded that there is very little systematic risk in the markets according to these regression models. These factors that serve as the constant in the regression equation above, are calculated at close to zero. Additionally, the systematic risk factors are found to be statistically insignificant.

Class A Mean Minimum Maximum Range Full Period 59.89 37.98 88.09 50.11

2003 - 2006 56.02 37.98 72.20 34.22

2007 - 2010 63.71 40.00 88.09 48.09

Class B Mean Minimum Maximum Range Full Period 59.41 38.46 87.75 40.29

2003 - 2006 49.62 38.46 62.06 23.61

2007 - 2010 69.09 42.83 87.75 44.92

(31)

Graph 3: Opening Prices Class A Shares

Graph 4: Opening Prices Class B Shares In line with expectations, the stock returns of Shell are far more volatile during the 2007 – 2010 period with regard to the preceding period. The observed movements are very similar to the movements observed in the NYSE Composite Index over the same period of time. This observation is in line with what the regression analyses revealed. It is striking that overall, and regardless of the economic crisis, the share prices rose. The dip observed between 2008 and 2009 is in stark contrast with the steady increase in prices in the preceding period. The period from mid-2009, 2010 onwards illustrates the beginning of economic recovery for the global economy and many firms, Royal Dutch Shell included.

BP

Due to its multidomestic structure Royal Dutch Shell gives out two classes of shares. BP does not share this multidomestic structure, and lists one class of shares on the global

(32)

capital markets. Similar to Royal Dutch Shell, BP being a British multinational, BP stocks are traded on the NYSE as American Depository Receipts. The stocks are traded as equity on the European capital markets and the London Stock Exchange (LSE).

Multiple regression analyses yielded the following results:

Explained Variance

2003 - 2006 33.5 Yes 262.493 Yes

2007 - 2010 51.9% Yes 561.410 Yes

Table 7: Regression Models BP Similar to the models for Shell, the regression models for the different time periods for BP are also all statistically significant. The F-statistics are slightly lower, as are the percentages of explained variance. The R2 which indicates the percentage of variance that can be explained by the model and the F-statistics are still quite high. The regression models for BP constitute a further investigation into the regression coefficients produced by the analysis.

β Significant? γ Significant? Full Period 0.963 Yes 0.049 No

2003 - 2006 0.914 Yes 0.013 No

2007 - 2010 0.970 Yes 0.063 No

(33)

than the coefficients for the full period and for the period preceding the crisis. This is mild evidence to support the hypothesis that increased exchange rate volatility during the global economic crisis of 2007 - 2010 may indeed to some extent impact the stock returns of BP. The lack of statistical significance may, as previously explained, be due to the presumed endogeneity of hedging practices or due to increased hedging during the crisis episode.

Class A α Significant?

Full Period -0.00007911 No

2003 – 2006 -0.000002949 No

2007 – 2010 0 No

Table 9: Systematic Risk Factors BP From these figures, it can be concluded that there is very little systematic risk in the markets according to these regression models. These factors that serve as the constant in the regression equation that was identified earlier, are calculated at close to or actually at zero. Additionally, the systematic risk factors are found to be statistically insignificant. This means that there is no relevant constant factor in the regression equation.

BP Mean Minimum Maximum Range

Full Period 56.61 36.65 79.75 43.10

2003 - 2006 47.92 36.65 61.97 25.32

2007 - 2010 65.20 38.60 79.75 41.15

Table 10: Conceptual Statistics BP

(34)

Very similar to the results for Royal Dutch Shell, and thus in line with expectations, BP stock returns are far more volatile during the 2007 – 2010 period compared to the preceding period. Here too, the observed movements are very similar to the movements observed in the NYSE Composite Index over the same period of time. This observation is in line with what the regression analyses revealed. It is striking that overall, and regardless of the economic crisis, the share prices rose. The main difference between BP and the other firms, is that where for example the Royal Dutch Shell shares starts to show signs of recovery, the BP share prices plummet even further between 2009 and 2010. This is due to the oil spill in the Gulf of Mexico in 2010 (BP, 2010), for which BP was held responsible.

The ExxonMobil Corporation

The ExxonMobil Corporation is the only American firm in the sample. Its stocks are traded as equity on the U.S. based capital markets, including the NYSE. The regression analyses for ExxonMobil yield slightly different results than the analyses for the previous firms.

Multiple regression analyses yielded the following results:

Explained Variance

2003 - 2006 38.6% Yes 327.177 Yes

2007 - 2010 67.8% Yes 1096.572 Yes

(35)

global economic crisis, is the highest of the percentages observed in all three models. The regression models for ExxonMobil definitely constitute a further investigation into the regression coefficients produced by the analysis.

β Significant? γ Significant? Full Period 0.925 Yes 0.227 Yes

2003 - 2006 0.983 Yes -0.159 No

2007 - 2010 0.913 Yes 0.499 Yes

Table 12: Regression Coefficients ExxonMobil What can be observed from these results is that it appears that BP’s stock returns are, similarly to earlier results, more sensitive to changes in the NYSE Composite Index than to changes in the Broad Dollar Index. The beta coefficients are slightly smaller than one, which means that any change in the NYSE Composite Index may be observed in approximately the same magnitude in ExxonMobil’s share price index. ExxonMobil’s stocks are however the only out of the three sets of stocks that appear to be significantly sensitive to changes in the value of the dollar. Between the period preceding crisis and the occurrence of the economic crisis, the sign of the coefficient has changed from negative to positive. Overall the sign of the coefficient is positive. The reason for the observed statistical significance may have country of origin effects, since ExxonMobil is the only U.S. firm in the sample and the U.S. dollar is not only its reconciliation currency but also its primary currency. It may also be due to valuation differences in stocks traded as equity (ExxonMobil) and stocks traded as American Depository Receipts (Royal Dutch Shell and BP) on the NYSE, though this is less likely.

Class A α Significant?

Full Period -0.00007911 No

2003 – 2006 -0.000002949 No

2007 – 2010 0 No

(36)

From the above figures, it can be concluded that there is very little systematic risk in the markets according to these regression models. These factors that serve as the constant in the regression equation that was identified earlier, are calculated at (close to) zero. Additionally, the systematic risk factors are found to be statistically insignificant. This, in turn, means that there is no influential constant factor in the regression equation for ExxonMobil either.

ExxonMobil Mean Minimum Maximum Range Full Period 56.29 30.70 95.08 64.38

2003 - 2006 40.08 30.70 51.80 21.10

2007 - 2010 72.32 49.30 95.08 45.78

Table 14: Conceptual Statistics ExxonMobil

Graph 6: Opening Prices ExxonMobil Shares Compared to the other two firms, ExxonMobil seems to have come out of the crisis slightly less damaged. Regardless of the dip in the opening share prices that can be observed, the decline appears to be more gradual between 2008 and 2010 than as seen in the other firms. Increased volatility between 2007 and 2010 compared to the preceding period is nonetheless obvious, both from the statistics as from the graph that illustrates the movements in share prices.

Intermediate Conclusions

Coming back to the first set of hypotheses, several conclusions can be drawn from the research results above. First of all, the statistical significance and high F-statistics of the

(37)

regression models indicate that the results are of high quality and can be used for deduction of tentative conclusions.

H1a: Increased exchange rate volatility during the 2006 – 2010 financial crisis has increased the sensitivity to exchange rate exposure of Royal Dutch Shell, BP and ExxonMobil.

In all the models, the alpha, or systematic risk factors are close to zero, from which it can be concluded that there is very little systematic risk in the markets according to these regression models. Additionally, the systematic risk factors are found to be statistically insignificant. The research has shown that even though exchange rate volatility increased, the gamma coefficients, or sensitivity to exchange rate fluctuations, has not always increased. In all case firms, it appears as though it appears that the stock returns are more sensitive to changes in the NYSE Composite Index than to changes in the Broad Dollar Index. In all regression models, the beta coefficients are statistically significant and approximately equal to one, which means that any change in the NYSE Composite Index may be observed in approximately the same magnitude in the firm’s share price indices.

In the models for Royal Dutch Shell, exchange rate sensitivity was statistically insignificant and, ignoring statistical significance, weaker during the crisis than during the preceding period. This may be explained by high levels of endogeneity of hedging practices or an increased amount of hedging on firm level to counteract the instable economic conditions. Overall, the findings mean that that for Royal Dutch Shell, H1a is rejected.

(38)

insignificance may be due to the reasons listed above for Royal Dutch Shell. Overall, we find some evidence to support H1a.

Finally, the models for ExxonMobil show that ExxonMobil’s stock prices are also more sensitive to changes in the NYSE Composite Index than to changes in the Broad Dollar Index. However, unlike the previous models, the models for ExxonMobil actually reveal that ExxonMobil’s stocks are statistically significantly sensitive to changes in the value of the dollar. Though the sign of the gamma coefficients has changed from the first to the second period, overall the coefficient is significant and larger in magnitude during the second period. This also implies that there may be evidence to support H1a.

Overall, we only find weak evidence to support H1a. Bartram and Bodnar (2007) warned that statistical significance may be very hard to find when investigating exchange rate exposure due to a wide range of reasons including methodological issues and the fact that hedging policies are more often than not endogenous to the firms’ risk management policies. Though H1a is not fully rejected, it can only tentatively be accepted.

H1b: Increased exchange rate volatility during the 2006 – 2010 financial crisis has had a negative effect on the stock returns of Royal Dutch Shell, BP and ExxonMobil.

(39)

prices drastically changed the overall positive trend observed in the period preceding the global economic crisis. With evidence to support increased share price volatility and evidence to support the negative effect it had on the stock returns of Royal Dutch Shell, BP and ExxonMobil, there is sufficient evidence to support H1b.

Country of Origin Effects

Results

(40)

A direct comparison of the graphs depicting the course of the value of the dollar and the opening share prices of ExxonMobil shows that indeed, the course of the dollar mirrors the course of ExxonMobil’s share prices.

Graph 7: Comparing Courses: U.S. Dollar

Graph 8: Comparing Courses: ExxonMobil Share Prices

It is however more likely that these mirror images can be attributed to changes in the NYSE Composite Index, given that all case firms exhibited significant sensitivity to changes in the NYSE Composite Index with a coefficient of approximately one. Also, the graphs for ExxonMobil, BP and Royal Dutch Shell are too similar to be able to state that the effect observed above is unique to ExxonMobil, even though the mirror image to the dollar fluctuations is slightly more visible in ExxonMobil. An examination of the produced sensitivity coefficients of all case firms provides more conclusive evidence.

(41)

γ Coefficients Full Period 2003 - 2006 2007 - 2010 Royal Dutch Shell A 0.058 0.121 0.025

Significance 0.426 0.244 0.812

Royal Dutch Shell B 0.051 0.061 0.044

BP 0.049 0.013 0.063

ExxonMobil 0.227 -0.159 0.499

Table 15: Comparison of gamma coefficients and significance (2003 – 2006)

ExxonMobil 2001 - 2006 γ coefficient -0.161

Significance 0.224

Table 16: Gamma coefficient ExxonMobil 2001 – 2006 As the figures in the table above show, ExxonMobil is the only firm in the sample which has a negative gamma coefficient over a certain period of time. A regression analysis for ExxonMobil over the 2001 – 2006 period even reveals a gamma coefficient of -0.161, though statistical significance lacks. If we accept a lower confidence interval, say 85% as opposed to the more usual 95%, we may even say that the coefficient of -0.159 is statistically significant. The fact that this negative coefficient is reversed over the 2007 – 2010 period may be related to the increased and larger fluctuations in both the value of the U.S. dollar and the NYSE Composite Index during that period but we find no conclusive evidence to explain this change in sign.

Intermediate Conclusions

H2a: An appreciation of the value of the dollar will have a negative impact on the stock returns of ExxonMobil.

(42)

Additionally, ExxonMobil is the only firm with a negative gamma coefficient over a certain period. During this period, accepting a slightly lower confidence interval than usual, we can thus say that indeed an appreciation of the value of the dollar will have a negative impact on ExxonMobil’s stock returns.

H2b: An appreciation of the value of the dollar will have a positive impact on the stock returns of Royal Dutch Shell and BP.

Even though mild evidence is given to support H2a, H2b is more difficult to prove since all gamma coefficients produced by regression analyses for BP and Royal Dutch Shell over varying periods of time are always statistically insignificant. Their sign is nevertheless positive, which means that movements in the Broad Dollar Index may also be seen in the course of BP’s and Royal Dutch Shell’s stock prices. The hypothesis however states that an appreciation of the value of the dollar will have a positive impact on the stock returns of Royal Dutch Shell and BP, so we must reject H2b based on the statistical insignificance found in the analyses. Some evidence is found to support H2b, but this evidence is not conclusive.

Hedging Explored

Results

To assess the effectiveness of specific approaches to hedging, these specific approaches need to be identified first. Though some characteristics are shared between the case firms, ExxonMobil once again differs from BP and Royal Dutch Shell in many ways.

Royal Dutch Shell

(43)

enable Royal Dutch Shell to acquire or supply commodities at future dates. These contracts are accounted for at fair value, so the value of the hedged assets and liabilities will be approximately equal to each other, given that the contracts are held to mitigate exposure risk. More specifically, the derivative contracts Royal Dutch Shell holds are specified as either interest rate swaps, forward foreign exchange contracts, currency swaps, commodity swaps, options, futures, forwards, or others (Royal Dutch Shell, 2010). In addition to reporting regular revenue, income and profit, Royal Dutch Shell also reports their annual comprehensive income. This comprehensive income comprises, among several others, any gains and losses incurred due to foreign exchange exposure and hedging activities. Royal Dutch Shell only started reporting their comprehensive income from 2008, possibly to comply with stricter regulations imposed by varying capital markets. Before 2008, the items listed in comprehensive income were listed under ‘other reserves’. Derivative contracts have been included in the annual reports from 2006 onwards. Hedging policies are approached in a top-down manner, and subsidiaries usually are not allowed to engage in financial derivatives on their own (Royal Dutch Shell, 2007). To assist these subsidiaries in managing exposure, Royal Dutch Shell has regional treasury centres in place. In 2007, Royal Dutch Shell explicitly stated that foreign forward exchange contracts were the most common instruments to mitigate exposure risk.

(44)

derivatives compared to total (regular) assets is 14.4% in 2008, where this percentage usually fluctuates in the range of 6.5% to 8.6%.

2006 2007 2008 2009 2010 Total Assets 235276 269470 282401 292181 322560

Total Assets in Derivatives 20264 20005 40623 20456 21160

Relative Percentage 8.61 7.42 14.38 7.00 6.56

Table 17: Total Assets in Derivatives / Total Assets Royal Dutch Shell Overall we can say that Royal Dutch Shell engages in relatively elaborate hedging practices to mitigate their exposure risk. Based on the information presented, we can tentatively conclude that Royal Dutch Shell has even increased its hedging efforts when the economic crisis hit. Royal Dutch Shell most likely decided to increase their risk management efforts due to a combination of decreasing profit margins and increased macro-economic stability.

BP

First and foremost, it is important to recognize that the 2010 oil spill in the Gulf of Mexico has had a significant impact on BP’s performance during that year (BP, 2010). Similar to Royal Dutch Shell, BP reconciles all cash flows in U.S. dollars for reporting

reasons and because most of the markets in which these firms operate are directly or indirectly priced in dollars. Due to the global scale of operations, BP is also exposed to a range of

(45)

strategy than the one we observed for Royal Dutch Shell, the relative percentage of total assets in derivatives compared to total assets is smaller than for Royal Dutch Shell.

2005 2006 2007 2008 2009 2010 Total Assets 206914 217601 236076 228238 235968 272262

Total Assets in Derivatives 13378 13398 10062 13564 8932 8566

Percentage 6.47 6.16 4.26 5.94 3.79 3.15

Table 18: Total Assets in Derivatives / Total Assets BP There is no apparent increase in hedging activity based on the total assets in derivates over total assets equation. BP seems to spread their hedging over different types of derivatives more than Royal Dutch Shell. Nevertheless, BP appears slightly more inept at equating assets and liabilities in derivatives than Royal Dutch Shell. i.e. the difference between assets and liabilities in derivatives deviates from zero much more on the BP balance sheet compared to the Royal Dutch Shell balance sheet. We do observe that in 2007 and 2008 BP held more liabilities in derivates than in the years before, but in 2009 the sign of net contracts held in derivatives changes. In terms of income we see a steady increase leading up to 2007, after which income and comprehensive become more volatile (see Appendix: Annual Report Outtakes: BP).

(46)

The ExxonMobil Corporation

Differences between BP and Royal Dutch Shell on the one hand and ExxonMobil on the other hand have accumulate, as here, ExxonMobil’s approach to hedging exposure risk differs substantially from the previous two firms. According to ExxonMobil’s 2010 annual report, the corporation only makes limited use of derivative instruments. ExxonMobil does not engage in speculative derivative or derivative trading activities. ExxonMobil only started reporting comprehensive income in 2008. For further details see Appendix: Annual Report Outtakes: ExxonMobil. The comprehensive income statement only lists a few items related to hedging, namely foreign exchange translation adjustments, changes in fair value of cash flow hedged and realized gains/losses from settled cash flow hedges, from which it also becomes evident that ExxonMobil engages in hedging less elaborately than BP and Royal Dutch Shell. ExxonMobil lost a significant amount of money in 2008 on foreign exchange translation adjustments, but reports no gains/losses from cash flow hedges in until 2010.

Intermediate Conclusions

Based on the firms’ annual reporting and news coverage, we can conclude that Royal Dutch Shell is most involved in hedging exposure risk, closely followed by BP. The

ExxonMobil Corporation engages in very little hedging, and thus chooses to remain relatively unhedged against currency fluctuations.

The Effectiveness of Hedging

Results

(47)

theory, taking a higher risk. Additionally, Royal Dutch Shell engages in slightly more elaborately in hedging than BP based on figures taken from the 2005 – 2010 annual reports.

Regression analyses ran over the 2005 – 2010 period provide an insight into what might possibly be the effect of these policy differences.

2005 - 2010 γ Coefficient Significance Royal Dutch Shell A 0.026 0.763

Royal Dutch Shell B 0.037 0.683

BP 0.061 0.572

ExxonMobil 0.349 0.000

Table 19: Comparison of gamma coefficients and significance (2005 – 2010) It is surprising that where Royal Dutch Shell and BP have extremely low and statistically insignificant coefficients for their sensitivity to dollar fluctuations, ExxonMobil exhibits a much stronger relationship between dollar fluctuations and share price fluctuations. Contrary to expectations however, during both the 2005 – 2010 period and the across the whole 2001 – 2010 period on average, the sign of the gamma coefficient is positive. This means an appreciation of the value of the dollar positively impacts the stock returns of ExxonMobil. The coefficient is much smaller than 1, so the magnitude of the effect on share prices will be much smaller than the magnitude of the change in the value of the dollar. Due to the highly volatile macro-economic conditions during the 2007 – 2010 economic crisis, and other factors that could not be controlled for the purpose of these regression analyses, the exact cause of this positive relationship cannot be found. We also noted that during economically benign periods, the sign of the coefficient indicating sensitivity to dollar course fluctuations is negative.

(48)

risk, but hedging against this risk also means that if exposure has unanticipated positive effects, the firm cannot fully benefit from these positive effects. It seems that the risk taken by ExxonMobil during the period investigated has not impacted the firm negatively, but positively.

Intermediate Conclusions

H3a: The firms remaining unhedged against foreign exposure will be more sensitive to exchange rate volatility than the firms that choose to hedge their foreign exposure.

Based on the results presented above, the conclusion must be drawn that indeed firms that remain unhedged against exchange rate volatility exhibit much higher and more significant sensitivity to exchange rate fluctuations. H3a is accepted based on relatively strong evidence. This means that hedging reduces the presumed negative effect that exchange rate exposure has on firm value and is therefore a moderating variable.

H3b: The firms that hedge their foreign exposure will enjoy more positive stock returns than the firms that remain unhedged.

Even though exchange rate sensitivity is much higher in firms that remain unhedged, the effects are not necessarily negative. The example of ExxonMobil has proven that the outcome of remaining unhedged depends on the specific micro- and macro-economic conditions. H3b must be rejected, based on the evidence found and presented in this thesis. It may however be rephrased as follows: Firms that hedge their foreign exposure will find that their stock returns do not fluctuate as much as the currencies it is exposed to.

(49)

Model Summary

As a control test, and also to create an overview of the different models built during this research, the analyses were repeated using dummy variables to identify these different firms and effects as opposed to investigating the different firms by specifying one dependent variable for each firm. This means that there will always be k-1 dummy variables to allow for an investigation of the different effects we observe. Given the focus of this research, the model summary only comprises the exchange rate (i.e. dollar rate) coefficients. The stock price indices of the different case firms have been recoded into a single dependent variable, where the individual firms are identified using dummy variables for the purpose of testing our different models. To investigate period effects the dummy variables are coded so that the global economic crisis period was given a value of 1, whereas the preceding period was given a value of 0. For investigations of a single firm, this firm was given a dummy variable value of 1, where all other firms were represented by a value of 0. Similarly, to investigate country of origin effects, U.S. nationals were given a dummy variable value of 1, and non-U.S. nationals a value of 2. The effects of hedging were explored by giving firms that hedged a significant amount of exposure a value of 1, and the firms that did not a value of 0.

(50)

Model Covariates 1 2 3 4 5 6 7 8 Period 2003 – 2006 0.091 (0.079) 0.013 (0.110) -0.159* (0.106) 0.065 (0.062) -0.159* (0.106) 0.009 (0.054) 2007 - 2010 0.034 (0.079) 0.063 (0.136) 0.499*** (0.092) 0.044 (0.068) 0.499*** (0.092) 0.158** (0.056) 2003 - 2010 0.055 (0.054) 0.049 (0.090) 0.227*** (0.069) 0.053 (0.047) 0.227*** (0.069) 0.109** (0.039) Hedging 2005 - 2010 0.041 (0.055) 0.349*** (0.082) 0.118*/** (0.046) N 4176 2088 2088 2088 6264 6264 2088 8352

* P<0.2, ** P<0.1, *** P<0.01; Standard errors are in parentheses.

Table 20: Model Summary The results presented in table 20 strongly correspond with the results presented earlier, which strengthens the intermediate conclusions drawn. All models are statistically significant, with relatively high percentages of explained variance and high F-statistics. Though many coefficients are statistically insignificant, the changes observed in their values indicate that there may be evidence to support several hypotheses, as was outlined above. It is noteworthy that model 8 carries many more significant coefficients than the other models, though not surprising as model 8 constitutes a combination of the preceding models. The coefficients for ExxonMobil over the 2003 – 2010 period are significantly different from the coefficients over the 2001 – 2006 period, indicating that the economic crisis or other events have had a significant impact on firm value.

(51)

Discussion

Based on the statistical significance and high F-statistics of the devised models, conclusions may be derived from the results with a fair amount of certainty. We only find weak evidence to support H1a. This is consistent with existing literature, which explains that other researchers have encountered similar problems regarding statistical significance. We do find some evidence to suggest that the firms exhibited increased levels of sensitivity to exchange rate changes as a result of increased exchange rate volatility. The evidence presented to support H1a is mostly conceptual and therefore not very strong. Conceptual evidence supports the assumption that the firms experienced increased share price volatility and also supports the negative effect it had on the stock returns of Royal Dutch Shell, BP and ExxonMobil, so there is evidence to support H1b. It is important to acknowledge that other factors beyond the scope and control of this research may have impacted stock returns. We found evidence to support H2a, and can thus say that indeed an appreciation of the value of the dollar appears to have a negative impact on ExxonMobil’s stock returns. It must be noted that this condition appears to hold true only during relatively stable economic periods, and so this test should be repeated for other periods than the periods investigated here before these findings can fully be confirmed. H2b is more difficult to prove since all gamma coefficients produced by regression analyses for BP and Royal Dutch Shell over varying periods of time are always statistically insignificant. Their sign is nevertheless positive, which means that movements in the Broad Dollar Index may also be seen in the course of BP’s and Royal Dutch Shell’s stock prices. There is not enough evidence to conclusively accept H2b.

(52)

effect. H3b was rejected, given that we find evidence to suggest that firms that hedge their foreign exposure will find that their stock returns do not fluctuate as much as the currencies it is exposed to, but that this effect is not necessarily negative.

(53)

(54)

Conclusion

(55)

be based on the competitive position, health and overall risk tolerance of each specific MNC, as there does not appear to be any evidence for best practice policies.

It should not be neglected that this research has not considered all variables that may influence stock returns and firm value. These may range from firm specific events in the micro- or macro-environment, such as the recent oil spill in the Gulf of Mexico, to larger events in the macro-environment or natural disasters impacting a range of firms. This research briefly touched upon the presumed volatility in the integrated oil and gas industry. To further develop this thesis, an in-depth analysis of this industry can be combined with broader macro-economic data to tentatively deduce other factors that may have influenced the values of the firms investigated. Governments, institutions and organisations like OPEC all exert tremendous pressure on firms operating in this industry, and these pressures should be considered in future research. Other interesting pressures to consider in future research include sustainability, and health and safety pressures.

(56)

References

Adler, M., Dumas, B. (1984) Exposure to Currency Risk: Definition and Measurement.

Financial Management, Volume 13, Issue 2, pp. 41-50.

Allayannis, G., Weston, J.P. (2001) The Use of Foreign Currency Derivatives and Firm Market Value. The Review of Financial Studies, Volume 14, Issue 1, pp. 243-276.

American Federal Reserve Bank (2005) Indexes of the Foreign Exchange Value of the Dollar.

Federal Reserve Bulletin, pp.1-8.

Bartram, S.M., and Bodnar, G.M. (2007). The Exchange Rate Exposure Puzzle. Managerial

Finance, Volume 33, pp. 642-666.

BP (2011) Annual Reports 2005 – 2010. Available at:

http://www.bp.com/extendedsectiongenericarticle.do?categoryId=9036021&contentId=70666 89 [accessed 16-10-2011].

Choi, J.J., and Prasad, A.M. (1995) Exchange risk sensitivity and its determinants: a firm and industry analysis of U.S. multinational. Financial Management, Volume 24, Number 3, pp. 77-88.

Chue, T.K. and Cook, D. (2008). Emerging Market Exchange Rate Exposure. Journal of

Banking and Finance, Volume 32, pp. 1349-1362.

Côté, A. (1995) Exchange Rate Volatility and Trade: A Survey. Bank of Canada, Working Paper, pp. 94-5.

Datamonitor (2005) Global Integrated Oil & Gas – Industry Profile. Datamonitor. Available through: Business Source Premier [Accessed 28-02-2011].

(57)

Datamonitor (2010b) Exxon Mobil Corporation – Company profile. Datamonitor. Available through: Business Source Premier [Accessed 28-02-2011].

Datamonitor (2010c) Royal Dutch Shell Plc – Company profile. Datamonitor. Available through: Business Source Premier [Accessed 28-02-2011].

Dewenter, K.L., Higgins, R.C., Simin, T.T. (2004) Can event study methods solve the currency exposure puzzle? Pacific-Basin Finance Journal, Volume 13, Issue 2, pp. 119-144.

Donnelly, R. and Sheehy, E. (1996) The share price reaction of U.K. exporters to exchange rate movements: an empirical study. Journal of International Business Studies, Volume 27, Number 1, pp.157-165.

Economist, The (2011a) Oil Markets and Arab Unrest – The Price of Fear, The Economist, [online]. Available at http://www.economist.com/node/18285768?story_id=18285768 [Accessed 16-03-2011]

Economist, The (2011b) The World Economy – From Tsunamis to Typhoons, The Economist, [online]. Available at http://www.economist.com/node/18440971?story_id=18440971 [Accessed 24-03-2011]

EIA (2006a) International Natural Gas and Liquefied Natural Gas (LNG) Imports and Exports. U.S. Energy Information Administration - Independent Statistics and Analysis, [online]. Available at http://www.eia.doe.gov/emeu/international/gastrade.html [accessed 02-03-2011].

EIA (2006b) International Petroleum (Oil) Imports and Exports. U.S. Energy Information

Administration - Independent Statistics and Analysis, [online]. Available at

http://www.eia.doe.gov/emeu/international/oiltrade.html [accessed 02-03-2011].

Foreign Exchange Exposure: Risk Management During Crisis