1
Empirical Implication of Hedging Effects on Debt,
Investment, and Sensitivity of Stock Return:
Evidence from US Oil and Gas Industry
Master Thesis Finance
University of Amsterdam | Business Economics: Finance
Abstract
This paper studies the hedging implication of oil and gas producing firms from 2010 to 2014, and evaluates the differences in the implication between financially distressed firms and non-distressed firms. The evidence suggests that the extent of a fraction-hedged production is positively related to debt capacity, negatively related to financing costs in terms of bond rating, and loan spread. Moreover, the result verifies
hedging activities are associated with tendencies of increasing the investment expenditures, lowering production costs, and indirectly reducing the sensitivity of
stock return to oil prices. Except for the hedging – debt ratio relation, all results demonstrate larger extent of hedging to financially distressed firms. Findings are subject to the conditions, while this study provides additional insights to oil and gas
producers, creditors, hedgers, and stock market investors.
Author: Joo Mi Park
Student number: 10418318 Supervisor: dr. J. E. Ligterink Submission Date: 07. July. 2016
2
Table of Content
I. Introduction II. Literature Review
2.1 Hedging Theory
2.1.1 Financial Economics Approach
2.1.2 Agency and Shareholder Maximization Theory 2.2 Oil and Gas Price Hedging Rationale
2.2.1 Commodity Market
2.2.2 Investment, Financing, and Hedging in Oil and Gas Industry 2.2.3 Stock Market Return and Hedging in Oil and Gas Industry 2.3 Summary and Hypotheses
III. Methodology
3.1 Financial Constraint Measure 3.2 Hedging Variable
3.3 Dependent Variables
3.3.1 Hedging and Financing Costs
3.3.2 Hedging and Investment Opportunities 3.3.3 Hedging and Sensitivity of Stock Returns 3.4 Control Variables
IV. Data and Descriptive Analysis
4.1 Sample Selection and Data Collection 4.2 Descriptive Analysis
V. Empirical Results
5.1 Hedging and Financing Costs – Debt Ratio 5.2 Hedging and Financing Costs – Bond Rating 5.3 Hedging and Financing Costs – Loan Spread
5.4 Hedging and Investment Opportunities – Investment Expenditure 5.5 Hedging and Investment Opportunities – Production Cost
5.6 Hedging and Sensitivity of Stock Returns VI. Robustness Checks
VII. Conclusion Bibliography
3
I.
Introduction
The oil and gas industry went through booms and busts throughout its history and currently facing the most deteriorated situation since 1990s. Earnings and profits of the oil and gas producing firms have greatly dropped and a number of companies have gone bankrupt1. Meanwhile, the open interest on energy products derivatives in the United States, such as over-the-counter crude oil futures, grew gradually over the last years2. Interesting questions arise among the varying motivations of participants in oil and gas commodity market, in which extent the oil and gas producers react to those market fluctuations, and what the impacts of risk management on its future investments and firm value are. To address these questions, this paper extensively examines the risk management activities of US oil and gas producing companies between 2010 and 2014. Particularly, this research investigates if the ratio of total crude oil and gas production hedged against the price variations relates to its external financing decisions and costs, investment opportunities, and stock returns.
Figure 1. Monthly S&P GSCI crude oil index and price fluctuation in period of 2010-2014
While both crude oil spot price and index variants in a large range over the period, especially from mid of 2014 there was a radical decrease in both measures and still has not fully recovered.
Corporate risk management has been widely discussed in current literature both in theories and empirical researches. Risk management through financial
hedging has received the most attention in the academic, where researches are mainly done across the industries. There are two main streams in the empirical research; one is testing whether a specific type of financial securities hedges effectively against the specific risk. The second stream studies the hedging premium on firm value, and the effect of financial hedging on firm financing, investment and tax liability.
1. Source from http://www.nytimes.com/interactive/2016/business/energy-environment/oil-prices.html?_r=0
2. Source from U.S. Energy Information Administration: https://www.eia.gov/finance/markets/financial_markets.cfm 0 20 40 60 80 100 120 0 100 200 300 400 500 600 700
Jan-10 Jan-11 Jan-12 Jan-13 Jan-14
O il Pri ce ($/Bbl ) Mo nt h ly S& P GS CI Cr u d e O il Ind ex
Monthly S&P GSCI Crude Oil
4
The results though do not consistently support the theories. Research papers related to the hedging activities in the oil and gas industry mostly focused on the first stream, whether in particular oil and gas price risk, exchange rate risk, or interest rate risk was hedged appropriately with derivatives and whether it had effects on firm value. The research in this paper incorporates the risk types and hedging instruments, and demonstrates the hedging effect on several aspects. Moreover, it revisits the hedging theories and shows the importance of corporate risk management especially to such a capital-intensive industry for securing sustainable financing to their projects.
This paper performs an empirical analysis of hedging implication on the determinants of the financing costs, on the investment opportunities, and on the sensitivity of stock returns in US oil and gas producing industry. The test of hedging and external financing costs’ implication highly resembles to the method of
Haushalter (2002), which specifically examines on oil and gas producers. This research designs distinctively from Haushalter (2002), as external financing costs such as bond rating and debt constraints are not measured on binary variables but on nominal variables to provide a specific degree of hedging implication. Moreover, the analysis contributes to Campello et al. (2011) who test the cross-industry firms’ positive effects of hedging policies on accessibility to capital in terms of loan prices, and on their capability of investment. While following the similar method, this paper incorporates all the hedging activities against the total risk that an oil and gas
producing firm could face and focuses on one industry. Finally, the relation between hedging and stock return sensitivity to oil price is examined. Previous similar research is from Jin and Jorion (2006) and this research adopts the pooled cross-sectional time-series regressions to measure how indirectly a firm’s hedging activities affect its stock return by hedging against oil price fluctuations.
Importantly, this research focuses on comparison of the degree of hedging between less financially distressed and relatively more distressed firms. This is based on the classic argument from Smith and Stulz (1985) that hedging is possibly more valuable for financially constrained firms. Thus, the hedging implication of all hypotheses represents the whole oil and gas producing firms, as well as a comparison between the firms. The results of this paper provide an additional line of evaluation over the oil and gas producers’ hedging activities to the creditors, investors, and corporate risk managers in firms.
The remainder of the paper presents as follows. Section II provides general corporate risk management theories and the empirical evidence, which leads to the hypotheses. Section III elaborates the method of this research including the
explanation of all the variables. Section IV describes the data collection process and descriptive statistics of the variables. Section V examines the empirical results of three hypotheses, and Section VI includes the robustness check of the regression models. Finally, Section VII concludes the findings.
5
II.
Literature Review
2.1 Hedging Theory
Corporate risk management is enabled through different channels for instance, liquidity management, operating flexibility, and hedging policies. This research mainly focuses on the implication of financial hedging, since hedging allows managing both financial liquidity and operating flexibility. Financial hedging is defined as taking a financial position that offsets the future risks against the expected cash flow fluctuation. Moreover, hedging is also taking an action to reduce the dependence of firm value on future risk factors.
Corporate risk was initially considered merely as a factor of market risk-return estimation, however, it developed into a financial theory in the 1950s (Klimczak, 2008). Theories have been developed from the opposition of Modigliani-Miller paradigm in the real world. In a classic Modigliani and Miller (1958) world with perfect capital markets, which means no taxes, information asymmetries, or
transaction costs exist, corporate risk management is irrelevant. In practice, however, markets are imperfect and firms may benefit from hedging policies due to the market frictions such as taxes, expected financial distress costs, information asymmetries, and costs of external financing. These hedging strategies ultimately affect the investment decisions and consequently the firm value. The following section extensively explains the risk management theories and the empirical evidences.
2.1.1 Financial Economics Approach
Most of the theories of corporate hedging are an extension of the classic Modigliani-Miller paradigm and the most extensively proven in theoretical models and empirical research (Klimczak, 2008). This approach provides the implication of hedging from the incurrence of irrelevance conditions: tax shield, lower bankruptcy cost (Smith & Stulz, 1985), higher leverage capacity (Modigliani & Miller, 1963), reduces external financing (Froot et al., 1993), and ensures internal financing (Froot et al., 1993). In this perspective, hedging should be indeed beneficial to firms, which in addition should lead to a hedging premium.
Smith and Stulz (1985) show that the probability of financial distress decreases and debt capacity increases as hedging reduces the volatility of income. Expected distress costs account for both the financial distress probability and corresponding costs if a firm defaults (Graham & Rogers, 2002). Debt ratio used in several papers to measure the expected distress costs (Fazzari et al., 1998; Haushalter, 2000; Graham & Rogers, 2002; Campello et al., 2011). Most of the studies shown a positive debt ratio coefficient on derivative hedging that could be interpreted also as; a hedging firm with higher debt ratio decreases the default possibility. Positive correlation between leverage and hedging supports the theory that corporate risk management is a tool for alleviating financial contracting costs (Haushalter, 2000).
Froot et al. (1993) suggest, hedging affects firms through two mechanisms, external financing and spend in investment. Several empirical research conduct the
6
implications of hedging in financing costs and investment. Haushalter (2000) argues based on their evidence of the US oil and gas producers, in which more financially leveraged firms with less financial flexibility manage the commodity price risks more vigorously. Moreover, Haushalter (2000) observes that hedging costs are related to economies of scale, so that larger firms tend to actively hedge to manage risks with relatively less hedging costs. For some occasions, hedging may be concerned as additional and substantial to the firms’ cost of operations. Contrary to those concerns, financial economics approach suggests that hedging should ease the access to capital by increasing its credit. Indeed, hedging firms pay less interest spreads and have few restrictions on capital expenditure in their loan agreements (Campello et al., 2011). Thus, favorable terms in financing covenants for hedging firms allow them to invest more and provide possibility of increasing firm’s revenue. Moreover, Giambona et al. (2011) show the liquidity level of a firm is important to the credit lines especially when the credits are limited, reversely speaking, if one has great liquidity to internally finance, the importance of credit line decreases. Therefore, a firm’s ability to invest and the credibility in the capital market are closely related and it is important how the hedging activities mediate in the relations.
Corporate risk management also focuses on the necessary cash flow level for investments. The hedging implication between cash flow variability and financing costs relates to the underinvestment problem. Hedging activities help to prevent costly external finance and enable firms to make fund investments that otherwise is not possible (Froot et al., 1993). Almeida et al. (2004) suggest that financially constrained firms in terms of a firm’s cash, manage liquidity to maximize their cash flow
sensitivity and firm value. Giambona et al. (2011) also indicate that firms with more cash holdings have greater opportunities of choosing between internal and external liquidity. Meanwhile, firms with less accessibility to credit lines tend to substitute cash holdings to investment. Originally this stems from the influential paper from Smith and Stulz (1985) that hedging lowers the financial distress costs by reducing the volatility of cash flows. Cash flow fluctuation is one of the largest risk factors to a firm, since it leads to variations in external financing and consequently affects the operating leverage. In other words, in case of costly financial distress, hedging may ease external funding by lowering volatility of income or cash flow. Thus, sufficient internal financing arises its importance, as external funding is possibly more
expensive with market imperfections. Kaplan and Zingales (1997) have shown that financially less constrained firms appear to mostly rely on internal cash flow to invest regardless of the availability of low cost external funds. Alternatively explaining, internal cash flow variations matters to both investment and variability in external financing (Froot et al., 1993).
Along with the fact that hedging eases financial distress costs and risks, hedging creates an additional insight to creditors for evaluating firms. Firms in
financial difficulty tend to engage in asset substitution, which is exchanging assets for riskier investments (Jensen & Meckling, 1976). Thus, one’s investment and hedging relation might be adjusted by the possibility of firms engaging in risk shifting. There is no standard measure for a firm’s risk-shifting behavior; finance theory explains that
7
firms with greater growth opportunities are likely to be riskier than less growing firms (Jensen & Meckling, 1976). From this expectation, Campello et al. (2011) expect and verify from their results that hedging – investment relation shows greater effect on the firms with more growth chances. Especially, they use the loan spread term for the investment opportunity proxy. The result embraces the risk shifting, as when the negative hedging effect on loan spread is larger, in which the hedging is worthwhile, firms with growth opportunities possibly do risk shifting.
This section introduced the theoretical framework of financial economics and explained that hedging is related to investment, financing costs, cash flow, debt capacity, and risk shifting. This paper also examines whether hedging addresses similar results in oil and gas industry and following hypotheses are driven in section 2.3.
2.1.2 Agency and Shareholder Maximization Theory
Another approach in the hedging theories is the agency theory. Agency theory extends the field of corporate risk management as taking into account the information asymmetries between the debt-holders and management including shareholders (Klimczak, 2008). Agency theory also explains how managers’ attitude affects the risk-taking behavior and hedging policies (Smith & Stulz, 1985). It implies that hedging signals to the debt-holders as their debt is protected and hedging also yields greater effects with the structure of individual block ownership.
Managers also react to the shareholders’ interests not only to the debt-holders. This is due to the informational asymmetries existing also in the relationship of managers and shareholders (DeMarzo & Duffie, 1991; Viswanathan, 1998). Shareholder theory focuses explicitly on shareholders’ interests as a managerial instrument and as the main consideration of hedging policy (Freeman, 1984). They argue that highly qualified managers have incentive to hedge to exhibit its ability to the market. The informational asymmetry degree is measured primarily the proportion of institution ownership of the share, and results that higher institution ownership hedged less. However, Geczy et al (1997) and Klimczak (2008) find incompatible result, as firms with high institutional ownership are possibly hedge more against currency exposures.
In this paper, financial economic approach is focused and verified through this empirical research, while shareholder and agency theory will also interactively
considered to explain the part of the results. 2.2 Oil and Gas Price Hedging Rationale 2.2.1 Commodity Market
Commodities are raw or primary products that typically categorized as hard commodities, which are mined or extracted natural sources, and soft commodities, which are agricultural products. Commodities are important, as they are primary goods for production in various industries. Thus, buying and selling commodities have been crucial not only to the investors, but also to the commodity related
8
industries. A commodity market is a virtual or physical marketplace for trading commodities, and traded on exchange via futures market. The primary factors of quoting the commodity market price are the supply and demand for the commodity. However, commodity markets are highly and directly influential to macroeconomic factors such as weather in the crop market, political controls in the oil market. Thus, as the other financial securities, commodity price hedging comes as an important matter to the businesses in respect of financing.
2.2.2 Investment, Financing, and Hedging in Oil and Gas Industry Oil and gas market has been spotlighted in academic world and in the
investment market due to its complex business for several reasons. Oil and gas sector is extremely capital-intensive industry where large investments are necessary to mine minerals and extract oil and gas for the producers, and to initiate new projects. The products in these markets are fairly homogeneous and hard to differentiate their products. Thus, the best performing producers in this industry can be explained as producing at lower costs with efficiency and protecting the product prices (Sadorsky, 2001). Producers do not have much power to control the prices and herein the
necessity of active price risk management emerges. Thus, oil and gas firms should heavily take the hedging instruments into account to cover any losses against price volatility. Moreover, return on investments requires long time to realize it, thus, the investment is illiquid unlikely to the other industries. Sustainable financing to this capital-intensive industry is crucial and corporate risk management may matters to the firms regarding to the financing in the oil and gas industry.
Haushalter (2000) shows that hedging policy is associated with financial leverage to US oil and gas producers. It shows oil and gas producers hedging against price risk relates positively to the debt ratio and even greatly affects for the companies having less financial flexibility. Carter et al. (2004) research the case of jet fuel
hedging for American airlines, and conclude hedging is sufficiently important in this industry in terms of resolving underinvestment problem. Not enough papers show how total hedging activities affect the leverage degree specifically in the oil and gas industry. Only recently, Kumar and Rabinovitch (2013) find robust evidence that reduction in financial distress costs and borrowing costs motivates managers to hedge in the oil and gas industry where hedging was measured as the hedging intensity. In addition, intrinsic cash flow risk and spikes in commodity price also causes to hedge. Overall, the expectation of commodity hedging effects in oil and gas industry is positive to the leverage.
2.2.3 Stock Market Return and Hedging in Oil and Gas Industry
Considerable empirical papers focus on the hedging premium, which is subjected to the firm value and shareholder maximization theory. Empirical studies focus on both indirect effects of hedging through using derivatives and direct effects of hedging on firm value. An influential paper of Allayannis and Weston (2001) based on cross-industry firms; examine the relationship between foreign currency derivatives hedging and Tobin’s Q ratio meaning the firm value, and conclude hedging positively relates
9
to firm value. Evidence from a large sample across countries and industries, Bartram et al (2011) find significantly high value of using derivatives in reducing total and systemic risks, increasing firm value and returns, especially in economic downturn of 2001-2002. However, not all the results consistently support the value maximization theory. Guay and Kothari (2003) analyze the potential gains of using derivatives of non-financial firms and conclude derivatives have minor effects on equity values and cash flows. The interpretation provides specific risks such as currency and interest exposures are secondary risks and do not significantly have effects on the firm. Moreover, studies also concentrate more on the endogenous variables of testing the derivatives usage effects. Debates are still continuing to find appropriate variables for specific risks. In this case, the relationship of derivative usage and firm value is meretricious.
Firm value is not the only index to examine the hedging premium but also the sensitivity of stock returns could be a key proxy. Research of hedging premium in oil and gas industry highly used the stock return as a measure of hedging premium. Jin and Jorion (2006) verify the positive relationship between stock return sensitivity to oil and gas prices and hedging, while the hedging effect on firm market value was insignificant. Moreover, Aleisa et al. (2003) examine the relationship between the US oil industry equity indices and oil prices, similar results also hold significantly for UK (El-Sharif et al., 2005). There is ample evidence of showing the link between oil prices and equity returns in cross-industry firms and country level, but only few studies investigate at the firm level on oil and gas industry (Mohanty et al., 2010). Not enough prior studies examine how the hedging affects the relationship between oil and gas prices and firm level of equity returns as Jin and Jorion (2006) conduct.
Abundant papers about risk management in oil and gas industry, mostly explain how the commodity derivatives effectively hedge against on specific risk, such as foreign exchange risk, price risk, and market risk. Moreover, the hedging premium on oil and gas industry is widely examined; they mostly use the firm market value indexes such as Q ratios and return on asset. However, studies on hedging effect on specific risk type and on market values of firm yield contradictory results. Thus, this research endeavors to incorporate all types of hedging and examine the hedging impact on sensitivity of stock market return to oil price.
2.4 Summary and Hypotheses
To summarize the literature review, corporate risk management theories suggest that oil and gas price hedging activities are predicted to be beneficial by lowering external financing costs and securing investment opportunities. Hedging activities are also expected to stabilize cash flow from the oil and gas price fluctuations, and result better market evaluation to those well-hedged firms. Essentially, hedging may be particularly valuable for firms that are more likely to distress (Smith and Stulz, 1985). Therefore, the research of hedging theories will be extended to test the difference of hedging effects dependent on the degree of financial distress.
10
H1. Financially distressed US oil & gas producers are more beneficial from hedging by decreasing financing cost and increasing debt capacity
H2. Financially distressed US oil & gas producers are more beneficial from hedging by attracting more investment opportunities
H3. Financially distressed US oil & gas producers are more beneficial from hedging by realizing less stock return sensitivity
The main comparison of this research is between the financially distressed firms and non-distressed firms among the hedging oil and gas producers. Overall implications would provide additional explanation to evaluate the corporate hedging both to creditors and shareholders, and reasoning of hedging against risks to the firms. The following section provides what the determinants of testing hypotheses are, and how this incorporated research will be conducted.
11
III.
Methodology
3.1 Financial Constraint Measure
To examine all the hypotheses, firms should be classified by the degree of financial distress. Kaplan-Zingales index (1997) is one of the efficient measures for financial constraints where several papers adopted (Kaplan & Zingales, 1997; Fazzari et al., 1998; Lamont et al., 2001; Whited and Wu, 2006). In addition, the index is initially developed based on the US firms, which also suits to use for the sample of this paper. KZ index is defined as follows,
𝐾𝑍 𝑖𝑛𝑑𝑒𝑥 = −1.002 ∗ 𝐶𝐹 + 0.283 ∗ 𝑄 + 3.139 ∗ 𝐿𝑒𝑣 − 39.368 ∗ 𝐷𝑖𝑣 − 1.315 ∗ 𝐶𝐻
KZ index is measured with five components where CF represents cash flow to total capital, Q as market to book ratio, Lev as debt to total capital, Div as dividends to total capital, CH as cash holdings to capital. The higher KZ index means that the firms are more constrained. In this research, a firm’s KZ index is defined as an average KZ index over the period of 2010 to 2014. This study uses a binary variable of KZ index to set two different groups of less and more financially distressed US oil and gas producers based on the median of sample firms’ KZ index.
3.2 Hedging Variable
Hedging variable is the main variable of all models to answer the hypotheses.
Evidence of the empirical studies shows that the construction of the hedging variable is not fixed but varies by the choice of the author. Nonetheless, mostly the binary proxy method was adopted in risk management research due to the belief of unreliable disclosed notional value (Greczy et al., 1997; Graham & Rogers, 2002; Carter et al., 2004; Klimczak, 2008). However, the binary proxy does not measure the magnitude of hedging activity. This may matter especially when the firm size is included in the analysis as smaller firms are likely to have stronger interaction between financing decisions and hedging (Lin et al., 2012). For this reason, several studies have employed the quantitative hedging variable such as using face value of contracts (Haushalter, 2000; Allayannis & Weston, 2001; Campello et al., 2011), using derivative portfolio delta (Tufano, 1996; Jin & Jorion, 2006).
This paper adopts the quantitative hedging variable, which is calculated as, 𝑇𝑜𝑡𝑎𝑙 ℎ𝑒𝑑𝑔𝑒𝑑 𝑎𝑚𝑜𝑢𝑛𝑡 𝑜𝑓 𝑦𝑒𝑎𝑟 𝑡 𝑇𝑜𝑡𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑦𝑒𝑎𝑟 𝑡⁄
and indicates the proportion of hedged value to the total oil and gas production. Most of the firms disclose either explicitly the percentage of hedging amount against their yearly production or reveals the hedged volume in BOE. Meanwhile, some firms only disclose the hedged amount in dollars representing the face value of hedging
instruments. In this case, the hedged amount is compared to the total production value in dollars instead. Taking this approach confronts few important assumptions. Firstly, firms are assumed to hedge only for nontrading and non-speculative purpose.
Numerous firms state about their risk management as hedging activities are used only for the purposes of protecting the price, interest rate, cash flow risks, while none of
12
them state as they hedge for speculation. Following from the reasoning of Graham and Rogers (2002), if the sample firms hedge on both liability and assets that
counterpart the risks, those firms are hedging appropriately. Furthermore, firms often enter a long-term security such as five-year swaps, which is excluded from the yearly hedging amount since it does not directly hedge to the production of next year.
Inclusion of the interaction variables within the test model expands the interpretation of hedging effects within the hypothesis. Related approaches used Altman’s Z-score and market-to-book value ratio (M/B) interaction with hedging (Graham & Rogers, 2002; Bartram et al., 2009; Campello et al., 2011). Altman’s Z-score indicates the level of default risks and higher Z-Z-score means the firm has better financial health. The interaction term of hedging and Z-score is expected to have a negative coefficient within the models, meaning that hedging alleviates the link between the dependent variable and Z-score. Moreover, hedging-M/B interaction term reflects that hedging provides firms with substantial growth options to invest more, and expected to have positive effect on the models.
3.3 Dependent Variables
3.3.1 Hedging and Financing Costs Debt Ratio
The base test of hedging effect on external financing costs is the debt capacity. Literature explains the hedging ability to increase debt capacity and the degree of leverage level can be measured in debt ratio. Financial leverage is defined as the ratio of total debt to the market value of asset (Haushalter, 2000; Campello et al., 2011; Whited, 2006). According to the literature, debt ratio is expected to have a positive coefficient in the first model as well as the degree of hedging impact on debt capacity is greater for more financially constrained group. The theory also suggests that hedging and liability causal relationship goes both ways (Graham & Rogers, 2002), which implies higher leverage capacity is one of the incentives to hedge. Thus, the model constructs the hedging and liability as a simultaneous structure using the OLS regression method as following,
𝐷𝑒𝑏𝑡 𝑟𝑎𝑡𝑖𝑜 = 𝑓(𝐻𝑒𝑑𝑔𝑖𝑛𝑔 & 𝐻𝑒𝑑𝑔𝑖𝑛𝑔 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠, 𝐾𝑍 𝑖𝑛𝑑𝑒𝑥 𝑑𝑢𝑚𝑚𝑦, 𝐹𝑖𝑟𝑚 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠, ) Bond Rating
The role of hedging related to the leverage is also alleviating the financial contracting costs. Particularly, bond rating is related to the contracting costs, as it is one of the determinants for the loan spread, loan quantity and default risks. The S&P bond rating is allocated a cardinal scale of range from 1 for AAA rated bonds to 20 for D rated bonds (Chen et al., 2007). Unlikely to the other hedging related papers where mostly binary proxy used for the bond rating, this research scales the hedging effect on bond ratings quantitatively. It is expected to have a negative coefficient, which actually
13
conveys positive relation with hedging, since the better financially healthy firms with lower numeric values have more incentives to hedge. Thus, bond rating is an
additional dependent variable for the first hypothesis and the regression’s framework is the same as following,
𝐵𝑜𝑛𝑑 𝑅𝑎𝑡𝑖𝑛𝑔 = 𝑓(𝐻𝑒𝑑𝑔𝑖𝑛𝑔 & 𝐻𝑒𝑑𝑔𝑖𝑛𝑔 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠, 𝐾𝑍 𝑖𝑛𝑑𝑒𝑥 𝑑𝑢𝑚𝑚𝑦, 𝐹𝑖𝑟𝑚 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 ) Loan Spread
Differences in the costs of external loans influence the value of hedging, and the next model examines the effects of hedging on the cost of debt financing. The external financing costs of a firm could be explained by loan spreads (Haushalter, 2002). Loan spread is the borrowing rate paid to the bank group annually. It is a loan interest rate in basis points excess of LIBOR. In the model, natural logarithm of loan spread is used to mitigate the skewness effect in data following from Graham et al. (2008) and Campello et al. (2011). As stressed in both papers, controlling the firm and loan characteristics that possibly affect the loan spreads is important. Thus, an additional variable related to the loan size will be included in the base model. Moreover, firms with higher credit ratings are likely to receive more favorable loan terms, so the bond ratings variable is also included as a control variable. Regarding to the theories, hedging and loan spread relationship is predicted to be negative. Thus, the OLS regression model for loan spread is structured as,
𝐿𝑜𝑎𝑛 𝑆𝑝𝑟𝑒𝑎𝑑 = 𝑓(𝐻𝑒𝑑𝑔𝑖𝑛𝑔 & 𝐻𝑒𝑑𝑔𝑖𝑛𝑔 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠, 𝐾𝑍 𝑖𝑛𝑑𝑒𝑥 𝑑𝑢𝑚𝑚𝑦, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠
− 𝐹𝑖𝑟𝑚 & 𝐿𝑜𝑎𝑛 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠, 𝐵𝑜𝑛𝑑 𝑅𝑎𝑡𝑖𝑛𝑔 ) 3.3.2 Hedging and Investment Opportunities
Based on the assumption that hedging positively affects the external financing costs, it relates to the advantages of hedging that directly impacts on investment expenditures and opportunities. This approach is modeled from Haushalter (2002), Graham et al. (2008), and Campello et al. (2011). The model structures identically to the first hypothesis and uses OLS regression method.
Investment Expenditure
Investment expenditure variable is measured as the ratio of capital expenditure to the market value of asset. Capital expenditure is an important component of loan
agreements, which affects the investment restriction. Accordingly, a firm’s
investment decision of containing capital expenditure in their loan contracts provides creditors’ assessment of the firm. Moreover, a firm’s investment opportunities may matter to the extent of expected bankruptcy costs and underinvestment costs (Myers, 1984). Thus, the logic follows that the hedging extent is expected to be positively correlate with the expenditure ratio.
14 Production Cost
Production cost per barrel of oil equivalent (BOE) is a measure of operating leverage and refers to the cost of extracting oil and natural gas including taxes. This measure is unique to obtain only from oil and gas producers where the firms disclose their production costs in 10-Ks either in the unit of BOE or MCFE (thousand cubic feet equivalent). The implication of the relation between production cost and hedging is, the greater cost of production per BOE, the more sensitive the cash flow is to the fluctuations in oil and gas prices. Haushalter (2002) assumes that companies are more prone to encounter financial distress with higher production costs. Thus, hedging should alleviate the risk of financing in production, which predicts negative
relationship between the production cost variable and hedging. The OLS regression model for both investment expenditure and production cost is constructed as follows,
𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 𝑂𝑝𝑝𝑜𝑟𝑡𝑢𝑛𝑖𝑡𝑖𝑒𝑠 = 𝑓(𝐻𝑒𝑑𝑔𝑖𝑛𝑔 & 𝐻𝑒𝑑𝑔𝑖𝑛𝑔 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠, 𝐾𝑍 𝑖𝑛𝑑𝑒𝑥 𝑑𝑢𝑚𝑚𝑦, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 − 𝐹𝑖𝑟𝑚 & 𝐿𝑜𝑎𝑛 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠) 3.3.3 Hedging and Sensitivity of Stock Return
Hedging activities may directly affect and protect the profit and loss from oil and gas prices. Consequently, the return of the oil and gas prices affects a firm’s net income and thus, the oil and gas producing firms’ market valuation. To estimate how hedging herein affects indirectly to a firm’s annual stock return, the model uses the two-stage least squares method.
Using a two-factor model from Jin and Jorion (2006), in the second stage of this examination between stock return and oil price return structures as following,
𝑅𝑖,𝑡 = 𝛼𝑖 + 𝛽𝑚𝑘𝑡,𝑡∗ 𝑅𝑚𝑘𝑡,𝑡+ 𝛽𝑜𝑖𝑙,𝑖,𝑡∗ 𝑅𝑜𝑖𝑙,𝑡+ 𝜀𝑖,𝑡
where 𝑅𝑖,𝑡 is the average stock return for firm 𝑖 in year 𝑡, 𝑅𝑚𝑘𝑡,𝑡 is the yearly rate of change in the stock market index especially the S&P 500 index, and 𝛽𝑜𝑖𝑙,𝑖,𝑡 is the return on the oil sales prices of firm 𝑖 in year 𝑡.
In the first stage, the potential impact of hedging activities on the oil sale price exposure is examined.
𝛽𝑜𝑖𝑙,𝑖,𝑡= 𝛼𝑖+ 𝛽𝑖,𝑡∗ 𝐻𝑒𝑑𝑔𝑖𝑛𝑔 + 𝜂𝑖,𝑡 ∗ 𝐹𝑖𝑟𝑚 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐𝑠 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 + 𝜀𝑖,𝑡 Incorporating these two equations to one form yields,
𝑅𝑖,𝑡= 𝛼 + 𝛽𝑚𝑘𝑡 × 𝑅𝑚𝑘𝑡,𝑡+ (𝛾1+ 𝛾2 × 𝐻𝑒𝑑𝑔𝑖𝑛𝑔 + 𝛾𝑛 × 𝐶𝑜𝑛𝑡𝑟𝑜𝑙) × 𝑅𝑜𝑖𝑙,𝑡+ 𝜀 Main interest is the coefficient 𝛾2, which delivers the answer of indirect hedging effect on stock return to oil price. Yearly term is applied for all variables since the hedging data is only available on yearly basis. Theories suggest that oil and gas
15
hedging reduces the sensitivity of a firm’s stock return to oil and gas prices, thus the coefficient of hedging is expected to be negative.
3.4 Control Variables
The control variables are divided to firm characteristics and loan characteristics as follows (Allayannis & Weston, 2001; Jin & Jorion, 2006; Campello et al., 2011). Firm Characteristics
a. Firm Size: Large firms are more possibly to engage in hedging than small firms due to large fixed costs for hedging. The proxy is widely used as log of total assets.
b. Profitability: Profitable firms are more likely to have higher firm value Q ratio and trade at a premium. Thus, return on assets defined, as the ratio of net income to total assets, will be used to control profitability.
c. Liability: Capital structure of a firm could possibly relate to the firm value. To control for capital structure, a variable explained in the long-term debt over the shareholders’ equity. This is also a main interest dependent variable for testing the first hypothesis.
d. Market-to-Book (M/B): Similar to the reasoning of firm size and it is measured by the sum of market value of equity and the book value of debt divided by the total assets.
e. Production Mix: Differences between firms’ operation in the oil industry versus gas industry exist depending on which area a firm mainly focuses. Thus, this is controlled for using the proportion of revenues from oil specific revenue over the oil and gas total revenue.
Loan Characteristics
a. Log loan spread: The natural loan spread is controlled for the loan contracts where the economics expansion and recessions affect the term spreads as well. This is also a main interest dependent variable for testing the first hypothesis. b. Log loan size: Similar to the reasoning of log maturity where the loan size is
controlled for the loan spread and measured in natural log of the loan amount in millions of dollars.
Control variables vary marginally by the hypothesis as described above. Bond rating variable, which is one of the main dependent variables for the first hypothesis, includes as a control variable in the hedging and loan spread relation test.
16
IV.
Data and Descriptive Analysis
4.1 Sample Selection and Data Collection
The sample of US oil and gas producing firms is selected from COMPUSTAT using the Standard Industry Classification (SIC) code of 1311 Crude Petroleum and Natural Gas, and 2911 Petroleum Refining. Selected firms’ annual report components of the period of 2010 to 2014, were downloaded from COMPUSTAT as well. This brings sample firms of 209. Some firms did not have full 5 years of data due to several reasons such as bankruptcy, mergers, and etc., thus only firms with data of 5 consecutive years are chosen. Moreover, following from Jin & Jorion (2006), firms with less than $20 million of total assets are excluded. Those firms are classified as “small business issuers”, which normally require less disclosure in Form-10K filings. Even if small firms might have hedged without disclosure, those firms are excluded from the sample, since this research needs an exact annual hedged amount in dollars. It senses to be disambiguate to assume those firms hedge zero amounts, thus, it falls to category of exclusion. Throughout this process, 174 firms are retained that match the above-mentioned criteria. Furthermore, necessary information for the study is downloaded from the COMPUSTAT database, which derives the firm characteristics variables, closing stock prices, oil sales prices, and bond rating variables per firm.
After the basic selection process, hedging and production information for each firm and each year is obtained manually from the annual reports or 10-K filings. Keywords to search the hedging and production information was, “Item 7”, “Item 8”, “hedg”, “risk management”, “derivative”, “risk”, “commodity risk” and to search the production information was, “production volume”, “volume”, “production cost”, “production expense”, “per BOE”, “average produc”, “results of operations”. Some firms had different units in their reporting such as using the natural gas unit of cubic feet (CF), or British thermal unit (BTU) and those data had to be calculated to the barrel of oil equivalent units (BOE) to have a consistent data unit. Through the data collection process, firms that did not unveil the hedging information were omitted from the data list and finally results a sample of 80 firms. From the hand-collected data, the variables for hedging and operating characteristics could be derived, such as production price per BOE and revenue fraction of oil sales to total sales.
The final sample is non-random data, which means that they are selected under few classifications. The firms without data for 5 consecutive years were excluded from initial sample. This might be subject to survivorship bias that causes optimistic statistical significance of tests, leading to distorted results. There is no normalized performance measure for quantifying the possible bias, yet few
suggestions (Brown et al., 1992). It is crucial to recognize the characteristics and roles of survived firms in the market within the chosen period, and how it differs from the firms that failed. The firms that were omitted from the sample are relatively small firms in service of crude oil and natural gas exploration rather than in refining sector. Moreover, those firms are mostly merged to the larger firms that are listed in the final sample firms, thus it can be assumed that the effect of excluded firms are absorbed to
17 Table 1. Sample Description
This table contains descriptive statistics for 80 US oil and gas producing firms from 2010 to 2014. Firms included in the sample provided the hedged amount data and abled to calculate the fraction of production hedged. The proportion of annual production hedged is defined as total hedged against commodity price risk in BOE over the total volume of production in BOE. The production mix is defined as revenue from oil sales over the total revenue from oil and gas sales. Investment expenditure is measured as the ratio of capital expenditure to the market value of asset. Bond rating is in a cardinal scale of range from 1 for AAA rated bonds to 20 for D rated bonds. Log loan spread is defined as natural logarithm of interest rate in basis points excess of LIBOR. Log loan size is measured as natural logarithm of loan in million dollars. Annual stock return, oil sale price return, and S&P500 oil future index are all measured as natural logarithm of year n-1 data subtracting year n data.
Variables No. of Obs Mean Std. Dev. First Quartile Median Third Quartile Hedging Fraction Hedged 400 0.497 0.243 0.314 0.51 0.7 Hedged x Z-score 399 0.596 0.612 0.317 0.575 0.854 Hedged x M/B 387 0.384 0.32 0.155 0.325 0.524 Firm characteristics Profitability 399 0.108 0.149 0.071 0.12 0.174 Firm size (log) 399 2.053 0.251 1.898 2.057 2.218 Cash holding 398 0.036 0.046 0.003 0.018 0.056 Z-score 399 1.368 1.317 0.857 1.22 1.751 M/B 387 0.786 0.541 0.417 0.656 1.004 KZ index 383 1.52 2.996 1.019 1.936 2.596 Debt ratio 399 0.56 0.209 0.447 0.551 0.634 Operating characteristics
Production cost (/BOE) 400 12.387 6.408 8.232 11 15.41 Production mix 399 0.561 0.287 0.329 0.586 0.793
Investment & Loan
Investment expenditure 399 0.222 0.123 0.125 0.208 0.311
Bond rating 400 9.837 5.582 6.5 11 15
Log loan spread 186 5.246 0.474 5.01 5.298 5.521 Log loan size ($M) 186 6.893 1.162 5.991 6.907 7.649
Returns
Annual Stock Return 307 0.885 1.069 0.88 0.98 1.041 Oil Sale Price Return 308 1.009 0.028 0.99 1.002 1.026 S&P500 Oil Future
18
Table 2. Fraction of Production Hedged and Oil Sales Price by Year
This table presents the proportion of production hedged for each year in period of 2010 to 2014. The proportion of annual production hedged is defined as total hedged against commodity price risk in BOE over the total volume of production in BOE.
the final sample. Another possible way to reduce this matter is to check if the error terms are robust to the firms’ fixed effect. Yet, due to the potential selection bias issue, there might be upwards bias in the results regarding to the fact that succeeded firms over the period are only remained.
To obtain the loan information namely, loan spread and loan size, Thomson ONE database was the optimal option to use. Thomson ONE database features
financial fundamentals, transaction data, market quotes and importantly the loan deals that entail over 92,000 syndicated loans including the global transactions. Thus, the loan related information per firm by year was available in this database. Moreover, to construct the last hypothesis to test the stock return sensitivity to hedging, the market index variable is used as a control variable, thus, the S&P500 Oil Future index was used and it is publicly available in S&P500 by yearly. These loan data and market index data was gathered to the hedging, operating, and firm characteristics data. After collecting the whole data, all variables for this study could be calculated and prepared to test the hypotheses.
4.2 Descriptive Analysis
Table 1 contains descriptive statistics of the whole variables of 80 US oil and gas producing firms between 2010 and 2014. The extent of hedging among the 80 firms is an average percentage of 49.7 over the whole period, and a mean percentage of 51. Though the average fraction of hedged production seems to be relatively constant over the period, the hedged amounts of each firm widespread depending on their risk management policies. Table 2 displays the yearly trend of hedging ratio over
production, which has been increasing over the period from 2010 at an average of 41.8 percent to 55.4 percent in 2013. The average percentage of hedging to
production slightly decreased in 2014 to 53.5 where this tendency resembles to the average oil sales price of entire firms. Table 2 also presents the flow of average and mean price in dollars of oil sales by year and clearly shows in 2010 selling at 73.54 $ per barrel of oil equivalents (BOE) on average, and gradually increasing till 2013 to
Hedged Oil Sales Price Forward – Spot Price
No. of Firms Mean Percentage of Production Hedged (%) Median Percentage of Production Hedged (%) No. of Firms Mean Oil Sales Price ($) Median Oil Sales Price ($) Mean Oil Forward Price ($) Mean Oil Forward Price - Spot Price ($) 2010 80 41.8 44.5 78 73.54 74.62 79.61 4.99 2011 80 48.1 50 78 91.05 90.42 95.11 4.69 2012 80 49.7 50 78 93.01 92.95 94.15 1.2 2013 80 55.4 57 78 93.15 93.75 98.05 4.3 2014 80 53.5 55 78 86.22 86.93 92.91 5.98
19
93.15 $ on average. The oil sales price decreased in 2014 to 86.22 $, at the same time where the firms also modified their hedging amount compared to the other years. Observing from the dollars difference between annual mean crude oil forward price and spot price, the difference disperses greatly each year, except in 2012. As Caporale et al. (2014) find that there are a weak causal relationship between the future and spot oil prices, but varies by time significantly, The financial market over-estimates the expected spot price than what actually the value delivers. This is consistent with Mello & Parsons (2000), which investigate time-varying hedging intensity pattern and conclude firms do not hedge systematically, but financial constraints more vary with time. In conclusion, merely observing this yearly tendency of hedging against the market risk is anecdotal.
To test the hypotheses, the finalized sample will be grouped into two based on the mean of KZ index, which is 1.52. The firms scored below 1.52 means relatively less financially constrained than the others, vice versa. One of the interested variables for this research is the debt ratio, which is not much diffused among the sample as one fourth of the firms’ debt ratio is 44.7 and the third-fourth is 63.4 with a standard deviation of 0.209. The production cost per BOE is 12.38 $ on average, with one-fourth spend less than 8.23 $ and the third-one-fourth spend more than 15.41 $. The production cost depends mostly on the efficiency and economies of scales of a firm, which also represents the operational ability. The revenue fraction of oil sales to the total sales denotes how much firms make profit from oil sales. More than half (56.1%) of the sales are from oil sales on average, while it widely disperses by each firm, from 33% to 79%.
Regarding to the bond rating of firms, the average nominal scale is 9.8
representing about BBB credit rating. The bond rating of gas and oil producers widely spreads from the rating of AA to CCC. Both annual stock price return and oil price return variables show how much current year’s price relatively differs to the previous year. An average of annual stock price return is 0.885 implies that in general, US oil and gas producers’ stock return decreased over the five year of sample period, meanwhile the oil sales price increased about 0.009 on average. Through the summary statistics, this empirical evidence clearly demonstrates the oil and gas market and the firms’ situations.
Table 3 consists the Pearson correlation between the variables. The
correlations show how the relevant independent variables have interconnection with hedging. For instance, the debt ratio is positively related to the hedging variable as 0.0404, and the investment expenditure also has a positive coefficient of 0.0397, which were already predicted from the theories. These variables are not consistent with the hedging interaction variables, which are not explainable merely with this result. Moreover, the bond rating, loan spread, and production cost, variables also have mixed results of correlation coefficients with hedging variable. Thus, the
following section necessarily provides the in-depth relations between the variables on interest.
20 Table 3. Pearson Correlation Coefficients
This table contains correlation coefficients for the entire variables and sample
Fractio n Hedged Hedged x Z-score Hedged x M/B Profita bility Firm size (log) Cash holding Z-score M/B KZ index Debt ratio Product ion cost Fractio n Oil Revenu e Invest ment expendi ture Bond rating Loan spread (log) Loan size (log) Annual stock price Oil sale price S&P 500 Oil Future Index Fraction Hedged 1 Hedged x Z-score 0.2629 1 Hedged x M/B 0.5981 0.3378 1 Profitability -0.1650 0.7668 0.031 1
Firm size (log) 0.0163 0.2616 -0.0056 0.0538 1
Cash holding -0.1402 0.1215 0.0909 0.0254 0.2068 1 Z-score -0.2180 0.8169 0.0359 0.7406 0.2925 0.2666 1 M/B -0.0126 0.2041 0.7360 0.1345 -0.0017 0.2213 0.2115 1 KZ index -0.2048 -0.5042 -0.2576 -0.2234 -0.2232 -0.0805 -0.4198 -0.1695 1 Debt ratio 0.0404 -0.6472 -0.1377 -0.4727 -0.2982 -0.0940 -0.6563 -0.2476 0.6530 1 Production cost -0.0099 -0.1487 -0.1490 -0.1256 -0.0999 -0.1410 -0.2536 -0.2691 0.0542 0.2663 1
Fraction Oil Revenue -0.1028 0.0709 -0.0709 0.2319 -0.1541 0.0239 0.1035 0.0333 -0.1136 -0.1543 -0.1016 1 Investment
expenditure 0.0397 -0.2165 0.0537 -0.0090 -0.4334 -0.2258 -0.3307 0.0580 0.2917 0.1749 -0.0150 0.1175 1
Bond rating 0.2191 0.0499 0.0659 0.1260 -0.3599 -0.3283 -0.1069 -0.0692 -0.0897 -0.0758 0.0123 0.1564 0.4158 1
Loan spread (log) 0.0865 -0.1377 -0.0236 -0.0522 -0.5145 -0.1063 -0.2004 -0.1062 0.1822 0.2441 0.0385 0.0183 0.1750 0.1831 1
Loan size (log) 0.0190 0.2316 0.0018 0.1412 0.6646 0.1477 0.2426 0.0151 -0.1578 -0.1773 -0.0419 -0.0894 -0.2643 -0.1966 -0.4211 1
Annual stock price -0.0810 0.1984 0.3026 0.2210 0.0419 0.1262 0.2406 0.3607 -0.0619 -0.0815 -0.0563 0.0721 -0.1040 0.0125 -0.1340 0.0381 1
Oil sale price -0.1514 0.1102 0.0910 0.1610 -0.1200 0.0018 0.1819 0.2321 0.1342 -0.0468 0.1087 -0.0455 0.0193 0.0352 0.1406 -0.0454 0.0853 1 S&P 500 Oil Future
21
V.
Empirical Results
5.1 Hedging and Financing Costs – Debt Ratio
Based on the assumption of firms with higher debt level confront greater possibilities of financial distress, most studies understand that higher expectation of a firm’s financial distress encourages more hedging activities (Graham & Rogers, 2002). The results of models linking the extent of hedging and debt ratio estimated with OLS regression prove the empirical findings are similar to theoretical expectations. Results under column (1) of Table 4, which is the main result of whole sample firms, present a positive 0.0715 coefficient of hedging for debt ratio. This implies that average hedging oil and gas producing firms could increase debt capacity by about 6.34% contrast to the sample mean of annual debt level (=0.0715×0.14/0.56). Prior studies find similar result as Haushalter (2000) shows 0.252 coefficient of debt to asset ratio with a tobit regression model constructed as the hedging variable to be the dependent variable. Using the similar constructed model, Graham and Rogers (2002)
demonstrate 0.1031 coefficient of debt to asset ratio with OLS regression model. This research also results that hedging causes to increase debt capacity and firms could use this additional external funding to invest in other projects. Also, Whited (1992) states that greatly levered firms have strong premiums for external funds.
The regression model with interaction variables of whole sample firms in column (2) and (3), hedging still shows a significant and positive coefficient while the interaction variables with hedging are insignificant. However, the basic line models with interaction term of Z-score and M/B ratio are insignificant and even weakens the effect of hedging on debt ratio with Z-score interaction variable, while strengthens the extent of hedging with M/B ratio term. Yet, it is anecdotal to conclude if those
interaction terms have great effect on the constructed model with whole sample. The main interest of the first hypothesis is whether more distressed firms are more beneficial from increasing hedging to add leverage for few motivations such as tax deduction, compare to those less distressed firms. The basic line model result with the KZ index binary variable indicating 1 represents 128 relatively greater distressed firms contains in column (4) and indicating 0 for 57 less distressed firms contains in column (7). Unlikely to the author’s expectation, relatively well operating firms have more advantage from hedging to increase debt capacity, showing the coefficient of 0.1168 and the opposite group reports the coefficient of 0.0248 each with the significance level of 5% and 10%. The interpretation could be that the degree of hedging effect on liability depends on the economies of scales and this context is in line with oil and gas producing firms’ characteristics. Higher earning firms pay more corporate tax, which explains that the greater extent of adding leverage in those firms intensifies the tax deduction and lowers the expected default rate from hedging.
The regression models with hedging and M/B ratio interaction term for both groups are insignificant, which represent under column (6) and (9), while the hedging variables itself are still significant. Meanwhile, the prior study of Graham and Rogers (2002) have contrasted result as showing significant 0.0024 coefficient for hedging
22
and M/B ratio interaction term, which seems a marginal effect. Instead in this
research, hedging and Z-score interaction variable is highly significant at confidence level of 1% under column (5) and (8). This implies that regardless of the distress level, lower Z-score and hedging firms among the sample groups have more incentives to increase its debt capacity than with higher Z-score and hedging firms. Simply comparing the hedging impact on debt capacity shows greater level in less distressed firms. On the contrary, firms that are financially healthier among more distressed firms have greater effect on increasing hedging activities compare to those higher Z-score firms among less distressed firms. Overall, hedging activities do increase the debt capacity regardless of the motivation to hedge; while, the magnitude of the effect depends on the financial state of a firm.
5.2 Hedging and Financing Costs – Bond Rating
The results of bond rating with the hedging hypothesis predictions are in line with the corporate hedging theory. Assuming that bond or debt rating reflects the capital market’s thorough inspection, and enables firms to have better accessibility to the debt market; greater hedging extent should increase a firm’s creditability for better financing costs where the bond rating represented in a descending cardinal scale should be lower.
The regression results of entire sample show significant and negative hedging coefficient to bond rating. Result under column (1) of Table 5 proves that increase in a unit of hedging increases approximately 3.3454 levels in bond rating. Haushalter (2000) also prove based on the oil and gas producers that bond rating shows significantly positive effect on hedging with a coefficient of - 0.054. The level increases by introducing the interaction term of hedging with Z-score in column (2), which explains more financially stable and hedging firms are tend to score higher bond rating than lower Z-score firms and hedging. Meanwhile, there is no direct effect of firm value in Q ratio on hedging to bond rating, shown in column (3). Importantly, the hedging percentage variable exhibits greater effect when including the interaction term with Z-score as negative 6.5303 from negative 3.3454 when it did not include the interaction term. Thus, the result supports Smith and Stulz (1985), hedging is indeed particularly valuable for distressed firms.
Dividing the sample groups bring slightly different results from the whole sample. The hedging effect on bond rating turns to be insignificant in less distressed firms group in column (7) and (9), while only the regression with Z-score interaction term was significant. Financially distressed firms group validates positive and greatly significant hedging effect on bond rating, described under column (4) to (6). The coefficient of hedging variable of distressed group is more influential as negative 3.5214 with lower significance level compare to the result of whole sample as negative 3.3454. Regarding to the control variables, firms that are generally greater size, have more cash, less leveraged and hold smaller loan size graded better bond rating.
The result suggests that hedging activities could be an appropriate measure of signaling to the creditors especially when scrutinizing less performing oil and gas
23
producing firms due to its illiquid investment. The observation of relationship
between hedging and bond rating explains prior to the investment expenditure, which creditability matters the yield spread and liquidity for future investment.
5.3 Hedging and Financing Costs – Loan Spread
Continuing from the credibility and hedging relationship, Table 6 gauges the effect of hedging on loan spread conducted on the measure of natural logarithm. According to the theory, efficient hedging positions commit firms to meet their loan agreements and improve the contract terms, which transfer to lowering loan spread.
The estimation of hedging and loan spread on whole sample group was insignificant yet showing negative tendency of hedging to loan spread, under column (1) to (3). Meanwhile, only the group of relatively more distressed firms result a significant and negative coefficient of hedging variable. This means that the
hypothesis exactly verifies that financially distressed firms are more advantageous to hedge to decrease the financing costs in the context of loan spread. The estimate of column (4) with the base regression model implies that hedging oil and gas producing firms’ imposed loan spreads are 11.39% lower than non-hedging firms
(=−0.2292×0.497, where 0.497 is the mean of hedged fraction in Table 1). In other words, the average loan spread of whole firms and period is 209 basis points and hedging enables to reduce about 24 basis points. Campello et al. (2011) implies the same through the relationship between interest rate, foreign exchange hedging and loan spread, and the model using Z-score interaction term shows 0.496 positive and significant coefficient. The conclusion of this research also verifies the corporate hedging theory once again that hedging is particularly more valuable for financially unhealthy borrowers.
None of the extended model with hedging – M/B ratio term is significant; in which the firm value measure does not particularly match to the hypotheses.
Meanwhile, only the interaction term with the Z-score brings more solid explanation, as hedging is even more incremental value for those with higher Z-score firms within more distressed firms group. Under column (5), hedging decreases the loan spread by reducing the default risk costs with the interaction variable. With a similar inference, less distressed firms also show a significant and negative coefficient 1.3020 of
hedging percentage variable, shows in column (8). However, the rest of the regression results are inconsistent that could be interpreted as low default risk firms have weak relation of hedging and loan spread. Thus, financially healthy firms may have marginal benefit of hedging when it comes to borrowing and lowering the loan interests.
Only recently the first paper in the literature validate that hedging is related to loan spreads (Campello et al., 2011), this study also adds complementary results that hedging provides favorable credit valuation and terms. Incorporating the findings on hedging effect to debt ratio, bond rating, and loan spreads; the results moderately and broadly suggest that hedging generally eases the external financing costs and
24 Table 4. Hedging and Debt Ratio
The dependent variable is total debt to the market value of assets (Haushalter, 2000). The results are obtained from OLS regression with random fixed effects. First three columns of (1) to (3) present the results of whole sample, and column (4) to (6) present the results of relatively more constrained firms, while column (7) to (9) represents the results of less constrained firms. Standard errors are reported in parentheses. ***, **, and * describe the significance level at 1%, 5%, and 10%, respectively.
Debt Ratio (1) (2) (3) (4) (5) (6) (7) (8) (9) Hedging 0.0715* 0.0371* 0.1822* 0.0248* 0.01829*** 0.0804* 0.1168** 0.2919*** 0.0592* (0.061) (0.092) (0.017) (0.082) (0.077) (0.130) (0.061) (0.057) (0.046) Hedging x Z-score 0.0291 -0.2239*** -0.1656*** (0.059) (0.046) (0.030) Hedging x M/B -0.1663 0.1436 0.0896 (0.021) (0.160) (0.051) Profitability -0.3485*** 0.3613* 0.4454** -0.3891*** 0.2606* -0.3968*** 0.1264 0.5705*** 0.0974 (0.061) (0.142) (0.140) (0.068) (0.159) (0.068) (0.213) (0.184) (0.221) Firm Size -0.0208* -0.1240*** -0.1292*** -0.0436** -0.0266** -0.0422** 0.0164 0.0233** 0.0141 (0.012) (0.028) (0.029) (0.017) (0.014) (0.017) (0.013) (0.013) (0.013) Cash Ratio 0.0697 0.0351 -0.0037 0.1723 0.3593 0.1884 -0.4618 -0.2432 -0.5228 (0.277) (0.277) (0.276) (0.333) (0.311) (0.334) (0.408) (0.322) (0.402) M/B -0.1506*** -0.1651*** -0.0913 -0.1514*** -0.0999*** -0.2124*** -0.1341*** -0.0914*** -0.1759** (0.028) (0.035) (0.065) (0.034) (0.032) (0.075) (0.036) (0.029) (0.086) Investment Expenditure 0.0981 0.0197 -0.0021 0.1057 -0.0836 0.1207 -0.0145 -0.0291 -0.0135 (0.105) (0.107) (0.106) (0.140) (0.136) (0.140) (0.108) (0.083) (0.109) Oil Revenue 0.0057 0.0256 0.0436 0.0096 -0.0034 0.0166 0.0012 -0.0088 -0.0135 (0.052) (0.078) (0.075) (0.066) (0.054) (0.067) (0.060) (0.047) (0.061) Production Cost 0.0000 -0.0088*** -0.0084** -0.0025 -0.0003 -0.0026 -0.0005 -0.0009 -0.0003 (0.002) (0.003) (0.003) (0.002) (0.002) (0.002) (0.003) (0.002) (0.003) Loan Size 0.0013 -0.0088 -0.0083 0.0137 0.0188 0.0135 -0.0094 -0.0079 -0.0083 (0.011) (0.011) (0.011) (0.016) (0.015) (0.016) (0.009) (0.007) (0.009)
25
Loan Spread 0.0371* 0.0469* 0.0469 0.0730* 0.0693* 0.0765* -0.0006 -0.0012 -0.0017
(0.028) (0.030) (0.030) (0.042) (0.302) (0.042) (0.025) (0.020) (0.025)
Observation 185 185 185 128 128 128 57 57 57
Firm fixed effect Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year fixed effect Yes Yes Yes Yes Yes Yes Yes Yes Yes
KZ Index Dummy No No No 1 1 1 0 0 0
Adj. R2 0.2814 0.1882 0.1832 0.3704 0.5294 0.3718 0.3483 0.6141 0.4789
Chi Value 82.89 12.39 12.71 72.04 106.14 72.66 37.93 92.81 37.89
Table 5. Hedging and Bond Rating
The dependent variable is bond rating in a cardinal scaleof range from 1 for AAA rated bonds to 20 for D rated bonds (Chen et al., 2007). The results are obtained from OLS regression with year fixed effects. First three columns of (1) to (3) present the results of whole sample, and column (4) to (6) present the results of relatively more constrained firms, while column (7) to (9) represents the results of less constrained firms. Standard errors are reported in parentheses. ***, **, and * describe the significance level at 1%, 5%, and 10%, respectively. Bond Rating (1) (2) (3) (4) (5) (6) (7) (8) (9) Hedging -3.3454*** -6.5303*** -5.5916*** -3.5214** -5.3434*** -4.2837* 0.9556 -4.2224* 0.1395 (1.147) (1.555) (2.322) (1.550) (2.114) (2.964) (1.419) (2.595) (3.532) Hedging x Z-score 3.5104*** 2.1027* 3.5972*** (1.187) (1.664) (1.458) Hedging x M/B -3.4616 -1.2359 -1.3861 (3.112) (4.0903) (4.085) Profitability -0.0226 0.7953 0.4201 -0.5920 -0.0918 -0.4647 3.9308 8.2066* 4.2653 (1.868) (1.847) (1.909) (2.250) (2.278) (2.298) (5.561) (5.514) (5.707) Firm Size -1.1397*** -0.9148*** -1.1414*** -0.9111*** -0.8329** -0.9274*** -1.3415*** -1.0385*** -1.3314*** (0.229) (0.237) (0.229) (0.366) (0.371) (0.372) (0.208) (0.231) (0.212)
26 Cash Ratio -20.0061*** -17.2546*** -19.8904*** -11.5168 -10.6796 -11.519 -52.9108*** -48.2281*** -51.9966*** (6.616) (6.537) (6.613) (8.028) (8.034) (8.061) (9.471) (9.068) (9.946) Liability -6.7614*** -8.4830*** -6.5753*** -7.7475*** -8.4823*** -7.7252*** 3.2039 -2.9001 3.6041 (1.686) (1.748) (1.693) (2.256) (2.324) (2.266) (2.926) (3.653) (3.185) M/B -1.0031 -0.5190 0.2936 -0.5150 -0.2480 -0.0988 -1.8015* -1.5625 -1.0380 (0.772) (0.773) (1.398) (0.960) (0.980) (1.681) (1.063) (1.014) (2.493) Investment Expenditure 10.0255*** 8.0059** 9.6038*** 11.3379*** 10.0961*** 11.1744*** 1.9466 0.6564 1.8985 (2.536) (2.572) (2.562) (2.284) (3.515) (3.440) (2.767) (2.657) (2.801) Oil Revenue 0.9387 0.9436 0.8504 0.6981 0.7993 0.6509 0.6194 0.5840 0.6907 (0.938) (0.917) (0.941) (1.253) (1.252) (1.267) (0.996) (0.936) (1.029) Production Cost -0.0048 -0.0034 -0.0004 0.0588 0.0549 0.0581 -0.1786*** -0.1737*** -0.1737** (0.045) (0.044) (0.045) (0.056) (0.056) (0.056) (0.069) (0.066) (0.071) Loan Size 0.5948** 0.4399* 0.5776** 0.5541 0.4853 0.5572 0.0766 -0.0623 0.0665 (0.284) (0.282) (0.284) (0.414) (0.416) (0.416) (0.258) (0.249) (0.262) Loan Spread -0.0576 -0.0933 -0.1196 -1.2698 -1.3181 -1.3190 0.4682 0.2733 0.4716 (0.654) (0.640) (0.656) (1.099) (1.097) (1.116) (0.524) (0.503) (0.530) Observation 185 185 185 128 128 128 57 57 57
Firm fixed effect No No No No No No No No No
Year fixed effect Yes Yes Yes Yes Yes Yes Yes Yes Yes
KZ Index Dummy No No No 1 1 1 0 0 0
Adj. R2
0.374 0.4014 0.3749 0.2388 0.2429 0.2426 0.8402 0.8578 0.8367
F value 11.30 11.56 10.47 4.95 4.69 4.61 27.60 28.93 24.76
Table 6. Hedging and Loan Spread
The dependent variable is log loan spread in basis points excess of LIBOR and presents in natural logarithm to mitigate the skewness of the data. The results are obtained from OLS regression with random fixed effects. First three columns of (1) to (3) present the results of whole sample, and column (4) to (6) present the results of relatively more constrained firms, while column (7) to (9) represents the results of less constrained firms. Standard errors are reported in parentheses. ***, **, and * describe the significance level at 1%, 5%, and 10%, respectively.
