Hedging policy in oil sector: empirical analysis

Размещено на http://www.allbest.ru/

Introduction

Why do companies hedge the different proportion of their production? Does hedging imply the same benefits and costs regardless the company characteristics?

These questions can easily arise after analysis of the heterogeneous empirical data on commodity hedging decisions. The empirical data gives us significant space for investigation: there is a sufficient variation not only among the volumes that are hedge by different companies in one period, but also among the volumes that are hedged by one company in different periods.

All companies are supposed to operate in an optimal way and to make the decisions that maximize their profit. Significant variation of companies' behavior means that the factors that determine their hedging activities can potentially differ among companies.

Existing literature proposes that the companies' hedging decisions depend on companies' willingness to decrease the costs of financial distress and optimize the investment program,on existing tax regime and degree of management risk-aversion.

In our research we suppose that proportion of production hedged depends on the spread between the forward commodity price and historical average commodity price and on and on average costs of the company. Difference in price spread can potentially explain why one company hedges different proportion of production in different periods while variance of average costs can potentially explain why different companies hedge different proportion of their production in one period.

We expect that wider spread between the forward commodity price and historical average price and higher average costs of sales lead to increase of proportion of production hedged. Higher price spread means that the hedging level is relatively high and derivatives can fix the attractive price for the commodity. Higher costs mean that in case of price drop the company has a risk of getting low or even negative margin. Thus, companies with higher costs are supposed to hedge higher proportion of their production in order to secure a reasonable margin.

We will not discuss the in this paper what proportion of production it is optimal to hedge. Determining of optimal hedging strategy requires special analysis which goes beyond the scope of this work. We just assume that if a company undertakes hedging, it should provide certain benefits. We try to determine which companies' characteristics affect the hedging policy and what factors stimulate companies to hedge less or more.

The structure of the paper is following. In Section I we present existing literature review, in Section II we reveal data details, methodology and model estimation. Section III presents potential explanation of main results and Section IV presents conclusion.



1. Literature review

Why do companies hedge? The basic incentive for commodity hedging is stabilizing the cash flows. If company does not undertake commodity hedge it will get extra profit it times of commodity price burst and will get low profits in periods of commodity price decrease. Hedging instruments allow to fix the price of commodity and thus decrease profit volatility.

But does hedging imply the same benefits and costs regardless the company characteristics? And what factors incentivize companies to hedge different proportion of their production?

The empirical data shows that companies are heterogeneous in terms ofcommodity price hedging decisions. Moreover, the proportion of production hedged differs not only within various companies in one period, but also within different time periods for one company.Will be discussed in more details in the section “Sample and Data” If we suppose that companies operate in an optimal way there should be factors that justify such a variation.

In terms of benefits we can distinguish two main groups of factors that can influence hedging decisions: operational benefits and firm value benefits. Firm value benefits are connected with effect of hedging decision on the stock price. Operational benefits are the factors that allow company to get better conditions in terms of its operational activities: ability to get higher debt financing, ability to finance its investment programs even in time of commodity price decline.

The classic Modigliani and Miller approach (1958, 1961, 1963) considers hedging as a value-neutral activity. The hedging policy of the company should not matter if shareholders can undertake hedging by themselves at the same cost. Dufey and Srinivasulu (1983)considered hedging as the activity that does not create additional benefits, but moves the company along risk/reward curve to archive the certain tradeoff between volatility and return.

But the research on value-neutrality presumes that Modigliani and Miller theorem assumptions hold. But in the real world Modigliani and Millerassumptions are violated: hedging implies transaction costs, taxes do exist and capital markets are not completely perfect. Thus, the companies' behavior can be significantly different from the one that is suggested by the theory.

The existing literature that analyzes the effect of hedging on firm's value is mixed.

Allauannis and Weston (2001) propose that hedging does bringvalue to a firm. The scholars conducted empirical research on a wide sample of US companies and concluded that undertaking the currency risk management on average leads to 4% increase of firm value. Carter (2003) analyzed the data on US airlines companies and concluded that currency risk management on average leads to 14% increase of firm value.

Nevertheless, such an effect can be the result of incorrect model specification. It can be the case that there is another variable that simultaneously influences hedging activity and firm value. In this case we can reveal the empirical connection which is not a result of causal relationship.Lookman (2003) considered hedging activity as a proxy variable for unobserved company's characteristics that have an effect on the company's value.

There is also empirical research that detects no empirical connection between intensity of hedging activities and firm value. Jin and Jorion (2006) conducted the empirical analysis on a sample of 119 firms for the period from 1998 to 2001 and found no significant effect of hedging activities on firm value.

It is important to mention that intensity of hedging activity can influence firm value via operational advantages such as higher debt capacity and lower costs of financial distress.

Considering operational side, the main incentives for hedging are reduction of financial distress costs and optimizing investment programs, decrease of expected tax payments and mitigation of managerial risk exposure.

Financial distress costs reduction and optimizing of investment financing

Cash flows of oil and gas companies are highly volatile as the majority of transactions are executed at a floating price that is not known in advance. On the contrary, the costs do not generally depend on the commodity price and have to be paid regardless the market conditions.

Hedging can be used to fix the selling price in order to ensure stable cash inflow that is sufficient to cover all company's costs. Smith and Stulz (1985) outlined that hedging allows to decrease the probability of default and thus significantly improves the company's credit quality. Credit quality improvement allows company to increase its debt capacity. Mayer and Smith (1990)also propose that as hedging allows to negotiate the favorable condition of borrowing the money as itreduces costs of financial distress.

Haushalter (2000) and Graham and Rogers (2002)claim that companies with high leverage and low margins are more tend to hedge. They argue that such companies have higher expected costs of financial distress which incentivize them to reduce the cash flow volatility. The scholars conducted the empirical analysis which showed that leverage on average is associated with hedging, but firm profitability is not a significant factor.

There are numer of research that shows that hedging is associated with higher debt capacity (Ross, 1998; Hentschel and Kothai, 1995, Graham and Rogers, 1999). The explanation is very intuitive: hedging allows to decrease the probability of default of the companies and lenders are ready to grant higher debt amount. Here the causal relationship can start from lenders: they can oblige companies with volatile cash flows to hedge. Such covenants are constructed in order to improve the counterparty's credit quality and ensure that company will be able to serve the debt even in case of significant commodity price decline.

Another issue that is addressed by hedging is optimizing of investment financing. Stalz (1990) and Froot (1993)claim that stable cash flows in each period is crucial for investment program realization. If company is highly dependent on volatile commodity price it is either miss the investment opportunities or realize them in suboptimal way. Hedging allows to solve the problem because it allows company to achieve higher debt capacity withlower costs of debt and to ensuresufficient cash flow in periods of the most favorable investment opportunities.

The first factor allowing optimizing investment program realization is more favorable conditions on the external financing in case of hedging. This point is strongly connected with decrease of costs of financial distress in case of hedging which was discussed earlier. But here we discuss not only higher debt capacity, but also lower costs of debt.If company has lower probability of default, it can negotiate a better interest rate on financing. Adam (2002)conducted an empirical analysis on a sample of 111 gold companies in North America and found the negative correlation between hedging intensity and cost of debt. Moreover, according to the research firms that hedge not only have lower costs of funds, but also have higher capital expenditure.

The second factor allowing optimizing investment program realization in case of hedging is sufficient cash flows for investment in periods of commodity price declines. When commodity prices are low, the majority of commodity companies have no opportunities to invest in new assets or even sell their own assets. Thus, the price of buying such assets decreases and various favorable investment opportunities are presented at the market. But if company does not hedge it has no funds to finance such investment in the optimal moment. The issue was firstly introduced by Scharfstein (1993). Allayannis and Mozumdar (2001) analyzed the similar problem andconcluded that hedging positively influences investment program realization. During the periods of unfavorable market conditions and cheap investment opportunities companies that do not hedge can neither finance the projects with its own cash flow no get external financing. Companies that hedge, on the contrary, have their own funds to finance the projects and can easily get money from lenders.

Reduction of expected tax costs

Fixing the commodity selling price via hedging can also influence company's tax payments.

The issue was firstly introduced by Smith and Stulz (1985) and followed by Santomero (1995) and Bartram (2002). If tax function is convex (represents progressive tax system, where higher profits are taxed at higher rate) the firms have an incentive to stabilize the cash flows. The logic under the phenomenon is simple: a company with average profit during two period, pays less taxes than a firm that gets extra profit during one year (that is subject to higher tax rate) and low profit during the second year (that is subject to lower tax rate). Figure I shows such an example.

company hedging financial optimizing

Размещено на http://www.allbest.ru/

Figure I/ Reduction of expected costs with convex tax function using hedging

The figure shows how hedging can help to decrease tax payments. Suppose in year 1 the firm gets income before tax 1 in the year 2 the firm gets and income before tax 2. If the firm hedges it has income before tax 3 during 2 years. In presence of convex tax function a firm that undertakes hedging pays less taxes

Graham and Smith (1999) conducted an empirical research on the sample of companies in different tax systems (75% of the companies in the sample operate in system with convex tax function). The researchers also accounted for tax loss carryforwards, which result into local convexity. The data shows that the opportunity to benefit from optimizing the tax payments does exist, but the companies do not widely exploit such opportunities. Geczy et al. (1997) and Fok et al. (1997) conducted the similar research and also didn't find significant dependency between tax saving benefits and hedging intensity.

Shanker (2000) extended the model with opportunity of companies to carry losses forward and back. The research accounts not only for derivatives, but also for the contracts that are executed at a fixed price (which is not a common practice in the industry). Under such a methodology companies fix prices more intensively in case of higher potential tax benefits.

Another tax benefit that can be gained if company undertakes hedging is higherpotential tax shield. This advantage is a result of higher debt capacity in case of hedging that was discussed before. As hedging decreases the probability of default, the company can get a higher debt amount. If company has more debt on its balance sheet, it pays higher interest payments that are deductible from its taxable income. Graham and Rogers (1999) claim that tax shield incentive for hedging prevails the incentive of exploiting tax function convexity because its economic effect is more significant.

Thus, despite all the theoretical benefits of economizing on tax payments exists, the empirical evidence is mixed and tax incentives are not supposed to be the major factor of hedging policy choice.

Mitigation of managerial risk exposure

If the ownership is separated from control, the agency problem can arise. The classic principal-agent problem arises due to different attitude towards risk of shareholders and managers.

Shareholders in theory are supposed to be risk-neutral because they can diversify their portfolio on the market. Managers, on the contrary, tend to be risk-averse, because all their welfare depends on the results of the company. Moreover, managers worry about their reputation and tend to have short-term planning horizon (Brochet, Loumioti, Serafeim, 2015; Chowdhury, Sonaer, 2015). Thus, risk-averse managers try to stabilize the company's financial results via hedging (Brenner, 2015).

There is a number of empirical research that supports the theory. Turfano (1996) investigated the hedging behavior of the companies in a gold mining industry and in North America and concluded that managerial ownership leads to decreasing of hedging policy intensity. Another interesting finding ofTurfano (1996) research is empirical evidence that young managers are more tend to hedge. The scientist explains this phenomenon by the fact that young managers are less experiences and it is more costly for them to manage uncertainty. But after they get the industry experience the cost of hedging (including hedge-provider commission, limitation of potential profit, etc.) become bigger than the costs of managing the uncertainty and managers decrease hedging volumes. This theory is in line with ideas, proposed by Breeden and Viswanathan (1996), who claimed that hedging is undertaken only by the agents who have high costs of uncertainty managing.

Considering the issue, it is important to take into account Board composition as it can control the management behavior. Dionne and Triki (2005) conducted the research based on the data on gold mining companies in North America and showed that independent directors in the Board significantly influence firms' hedging behavior. The scientists did not determine the optimal hedging level and thus it is hard to determine whether independent directors influenced the hedging behavior in a positive or in a negative way. The case is that companies with higher proportion of independent directors in the Board are more tend to use hedging.

Price level and production costs

Two fundamental factors that are supposed to matter in determining the corporate hedging policy are the price level that determines the revenue of the company and production costs (De Angelis, Abraham, 2011).

The fact is that different companies own reserves and natural resources of different quality. That creates the heterogeneity in exploration and drilling costs. Moreover, companies can bear other costs of financing. That means that for different firms the average costs per one barrel of the oil with the same physical characteristics can be different.

The price for an oil of specified type is homogeneous and it is determined by an Exchange (for example, ICE Intercontinental Exchange that was established in 2000 and is widely used for commodities trading ) or information Agency (Platts, Argus Biggest international data providers of commodity benchmark price). The prices depend on different factors and they are highly volatile.

The case is that companies get different profit per barrel, which can create heterogeneity in hedging policy. While assessing the hedging opportunities on the derivative market (swap levels, option premiums for various strikes) the companies with different costs can implement different hedging strategies and hedge different production volumes.

Corporate Hedging versus Individual Hedging

In classic Modigliani and Miller world investorsare able to undertake any hedging decisions by themselves. Due to the fact that investors can have different attitudes towards risk, it can be more optimal for them to conduct individual hedging comparing to corporate hedging that implies homogeneity of attitude towards risk (Mayers and Smith, 1990). If it were true, we would possibly observe lowhedging intensity on corporate level. But empirical data shows that corporate hedging is widely spread. That means that individual hedging is suboptimal as compared to corporate hedging. There are several reasons why it is so. Firstly, individual investors have less information about firm's risk. Secondly, managers have private information that is not disclosed to shareholders and outsiders. Third, firms can use internal hedging techniques that are not available for shareholders.

Managers know in the very detail all the risks that the company faces. They are aware of current risk exposure, company's development plan and business strategy. Thus, they can undertake hedging policy better than outsiders (DeMarzo and Duffie, 1991).

The managers can be aware of specific private information that can influence market. If the company have hedge on a corporate level, it can make it in an optimal way. Revealing such information to market and shareholders can negatively affect the firm.

On the firm level the company can also use the internal hedging. Under internal hedging we can define the managerial decisions that mitigate or offset risk. For instance, the managers can transfer production to another country to reduce currency risk. Considering oil and gas sector, the company can be organized in a vertically integrated way (production, refinery and consumer). The oil price decline has a negative effect on oil producer because it leads to lower revenue. In such situation refinery can buy crude oil at a lower price, but the price of oil products is highly correlated with oil price and is also expected to go down. Thus, consumer is able to by oil products at a lower price. In general, for such group the risk is mitigated.

Moreover, hedging implies transaction costs. For a large company such costs are more easy-to-bear than for an individual investor. Brealy and Myers (2003) outlined two major types of transaction costs. The first one is fixed component which includes setting the infrastructure for hedging transactions. The second one is the cost component which comprises transaction costs associated with each particular instrument used by the firm.

As the first component of transaction costs of hedging is rather big, only the companies can bear. Hedging on the corporate level becomes more beneficial for investors as in this case each of them does not bear separate fixed transaction costs.

Moreover, empirical evidence shows that bigger companies tend to hedge more than the smaller ones. The scientists partly explain that fact by difference in transaction costs.

According to Stulz (1996) and Graham and Rogers (1999) firms with bigger assets have more stable cash flows than the companies with smaller assets, but they are more tend to hedge.

Thus, there is no clear understanding of the factors that influence hedging and the measure of their effect. The companies behavior is highly heterogeneous and should be further investigated.

2. Hypotheses, Data Analysis and Model

At the beginning of the empirical research on corporate hedging activities we will discuss the market standards of selling commodities at floating price, how the hedge works and how the decision-making process is organized on corporate level.

Market standards of selling commodities at floating price

Selling commodities at a floating price is a common practice in oil / gas / gas liquids industry. It is standard for producers to enter the binding sale contracts under which they commit to sell the certain volume of oil in the future at market prevailing reference price.

Reference price is a benchmark index that reflects a quote for commodity of specific characteristics (delivery bases and conditions etc.). It works like a market indicatorwith values released on a daily basis by data providers (the main data providers are Platts and Argus). Data providers calculate reference prices based on fundamental data that is received from market participants (both sellers and buyers of the commodity) during the regular market survey.

Such market design implies significant market risk as the price can decrease substantially between the signing of the contract andfixing the price. Hedging instrument allow to construct the strategy that removes or mitigates the market risk.

Corporate hedging activities

To remove the price risk, oil companies can fix the selling price via derivatives. The most common instrument used for commodity hedge is commodity swap.

Commodity swap is a derivative contract that allows the company to fix the future commodity price at some fixed level. It is the OTC instrument that is executed between the company and a bank and it is commonly not embedded in the sale contract. For example if the company sells the commodity swap with fixed level X, it ensures gettingX regardless the market price prevailing on the market.

The mechanics is following:

At T0:

1) The company enters the sale contract under whichat T1 it will sell the oil at price prevailing in the market and equal to average value of the price indexduring the month before T1;

2) The company sells the commodity swap underlying price index with averaging during the month before T1(and settlement at T1).

At T1:

1) The company gets the market prevailing price for the commodity (average value of the price index during the month before T1, can be higher or lower than the swap level);

2) The company gets the payment on the swap:

· If market price (p) is below the swap level (x), the company receives from the bank the payment equal to(x-p) * volume

· If market price (p) is above the swap level (x), the company makes the payment to the bank equal to(x-p) * volume

Figure II Payoff on the swap (sell)

The graph shows the payoff on the commodity swap. If market price is below the Swap level, the company that sold the swap gets Profit equal to (Swap level - market price) * volume. If market price is above the Swap level, the company that sold the swap gets Loss equal to (market price - Swap level) * volume

The payoff of the producer on both the sale of commodity and commodity swap is fixed and equal to swap level (x)multiplied by volume:

· If market price (p) is below the swap level (x), the company sells the commodity on the market at p and receives the payment on the swap; final result equals p * volume + (x - p) * volume = x * volumeIf market price (p) is above the swap level (x), the company sells the commodity on the market at p and makes the payment on the swap; final result equals p * volume - (p - x) * volume = x * volume

The companies can also use options and option strategies for hedging. Buying option put allows the company to fix the minimum selling price.

· If market price (p) is below the option strike (k), the company sells the commodity on the market at p and gets the payment on the option equal to (k - p) * volume; final result equals p * volume + (k - p) * volume = k * volume

· If market price (p) is above the option strike (x), the company sells the commodity on the market at p and doesn't execute the option; final result equals p * volume.

The options are less common for hedging as the companies has to pay the option premium for getting the right (not the obligation) to sell the commodity at strike level.

How the decision-making process is organized

The main parties involved in decision-making process on the hedging activities are:

· Managers that directly make the decision to enter the derivative agreements;

· Shareholders that participate in establishment of company's hedging policy and estimate the managers' performance;

· Debtholders that influence the hedging decisions via putting the covenants into loan documentation.

Managers

Managers are supposed to act for the benefit of shareholders and maximize the profit of the company, but they have their personal incentives to avoid the risk (Brenner, 2015) and prefer short-term benefits to long-term benefits (Brochet, Loumioti, Serafeim, 2015). A number of research shows that manages have short-terms planning horizon as they are hired for a limited term and they get bonuses / are punished for the results that can be observed at the moment, not in the future (Chowdhury, Sonaer, 2015).

Profit maximization implies that managers try to achieve the maximum selling price of commodity and minimum possible costs.

In terms of price the managers have a choice between execution the transaction at floating prices or fixing the price via swaps or other derivative strategies. The decision is complicated as managers cannot forecast the future market price and cannot know in advance whether the swap will ensure higher commodity price or limit the potential profit. The results of the decision in different states of the world are shown in FigureIII. The general idea is that the companies can potentially hedge higher volume if they expect that future market price will be possibly lower than swap levels.

Figure III Payoff for the company in case of hedging / selling at the market price

The scheme shows that the company benefits from hedging if it fixes the price that is higher than the floating market price.

The tree shows the payoff for the company in case of hedging / selling commodity at market price in different states of the world. The company can either sell commodity at market price(Floating price, the left branch [L]) or execute the swap and sell the commodity at Fixed price (the right branch [R]). If company do not hedge and Market price is higher than the Fixed price (LL), company get the higher profit [if company expects this case it is not expected to hedge]. If company do not hedge and Market price is lower than the Fixed price (LR), company get the lower profit [if company expects this case it is expected to hedge].If company hedges and Market price is higher than the Fixed price (RL), company get the lower profit [if company expects this case it is not expected to hedge]. If company hedges and Market price is lower than the Fixed price (RR), company get the higher profit [if company expects this case it is expected to hedge].

The main idea is that if future market price is lower than the swap level the hedging is beneficial. Thus, if the manager thinks in terms of payoff he would hedge a higher amount if he considers the hedge level attractive and thinks that future price can be potentially lower than this level.

One of the measures of attractiveness of swap price levels that can be considered by managers is spread between forward commodity price and historical average price. Significant spread means that at the moment the commodity price is on the high level (as compared to historical dynamics) and hedging on such a level potentially can be beneficial for the company.

Hypothesis 1: Spread between the forward commodity price and historical average price positively influences proportion of production hedged other factors being equal.

Figure IV shows the historical price dynamics and the swaps levels, which could be fixed at the beginning of 2015, 2016 and 2017, 2018. If the hypothesis 1 is true, firms should hedge bigger share of production in 2017 and 2018 than in 2015 and 2016.



Figure IV ICE Brent price dynamics and swap levels for commodity price fixing

The figure shows the historical price dynamics and the Asian swaps levels (with monthly averaging), which could be fixed at the beginning of 2015, 2016 and 2017 and 2018

As mentioned above, another factor that influences the company profit is costs. If the costs of the company are high, the company has a risk of losing the margin in case of negative price dynamics. Thus, we expect that a producer with higher costs will hedge on average a higher proportion of production in order to decrease the risk of financial distress.

Hypothesis 2:

Average costsof sales per unit produced positively influence proportion of production hedgedother factors being equal.

Control variables

Market value of total assets

This factor is used to indicate for the firm size and used as a proxy for transaction costs. Here we follow Stulz (1996) and Graham and Rogers (1999) who claim that companies with higher total assets value are in a better position in respect to transaction costs (fixed element of costs is reallocated to higher production level). Thus, for such companies average transaction costs per unit of production is lower. We expect to detect positive correlation with share of production hedged.

Financial leverage

The factor reflects the company's indebtedness. In this research we use net debt/EBITDA ration as proxy for company's indebtedness. As outlined before (Whited,1992), companies with high debt level have to meet more strict constraints with regards to stability of their cash flow and more tend to hedge. We expect to detect positive correlation with share of production hedged.

Sample and data

To reveal the determinants of corporate hedging policy we are to conduct the empirical research of the oil and gas companies' behavior. The behavior of production companies, refineries and traders is different due to different risk profile. In this research we focus on the behavior of production companies and thus will analyze the sample of oil and gas/gas liquids producers.

It is also important to control whether the production company is vertically integrated (whether the production company is part of the group that containsproducers, refineries and consumers of oil / gas / gas liquids / oil products). Vertically integrated companies have a different risk profile as the price change can be a negative event for one company of the group and positive event for other company of the group. For example, the decline of oil prices leads to lower revenue of oil producers, but also implies lower costs of buying the crude oil for refineries. Going further, as the prices of oil and oil products are highly correlated, the decline of oil prices is accompanied by oil products prices' decline. But this means that oil products consumers included in the group will buy oil products cheaper. Thus, the commodity price risks of the vertically integrated companies are subject to natural hedging on the group level.

We are to analyze the companies that operate in USA and Canada for two major reasons. First, we want to reveal the general principles that are not distorted by market inefficiency and thus we will construct the sample that contains companies that operate at developed markets. Second, a number of companies from USA and Canada discloses the information about their hedging activities in 10-K SEC filings what allows getting the empirical data.

The total number of public oil and gas companies in Canada and USD is equal to 2050. After we excluded refineries, traders and vertically integrated companies the sample shortened to 780 companies.

After we collected the available data on the hedging activities and excluded the companies that sell their products at fixed prices the number of companies in the sample decreased to 139 companies.

The analysis is based on information on derivatives contracts with settlement in the period from 4th quarter of 2017 to 4st quarter of 2018. As the hedging decisions are taking on average 1 year before the settlement period we analyze the data starting from 4th quarter of 2016.Such time window is taken because the market practice if hedging information disclosure started only in the 4th quarter of 2017 and there is no data for more early periods.

As data on the hedging activities is very limited and not available in databases, the research implies a lot of hand-collected data. Tickers of the companies and general fundamental data were downloaded from Bloomberg and Thomson Reuters databases, data on hedging activities was mainly hand-collected from 10-K reports and Exploration and Production Data for Oil, Natural Gas and Natural Gas Liquids made by Bloomberg New Energy Finance.

The sample includes 50 oil companies, 48 gas companies and 37 companies that produce gas liquids.

The percentiles of proportion of production hedged are shown in Table I. The statistics shows that there is a huge variation among the firms in terms of proportion of production hedged.

The summary statistics is presented in Table II and will be described separately for oil, natural gas and gas liquids.

Oil companies

Oil companies on average hedge 40.00% of their production. It is important to mention that some companies that do not apply the hedge at all (9.00%) and some companies hedge most all the production (6.00% of companies hedge more than 80% of their production). The is a sufficient variation of proportion of production hedged (standard deviation equals 0.28).

The variance among average costs per unit produced is significant. The minimum average costs are equal to USD 2.15 per bbl., while the maximum average costs are equal to USD 63.70 per bbl. The average value of average costs equals USD 19.40 per bbl.

Average spread between forward and historical average price during the analyzed period is negative and equal -USD 1.89. Minimum price spread is equal to -USD 14.52, maximum price spread is equal to USD 9.35.

The average net debt/EBITDA coefficient is equal to 5.71. It is important to mention that some companies have cash on the balance that makes the net debt/EBITDA coefficient negative. The debt burden varies among companies from negative values (-6.04) to particularly high values (the maximum net debt/EBITDA value qualms 133.68).

50.00% of oil companies undertake the strategic hedge that fixes the selling price in two-year advance.

Table I. Descriptive statistics

Percentiles
	Oil	Naturalgas	Gasliquids
1%	0.000	0.000	0.000
5%	0.000	0.000	0.000
10%	0.000	0.000	0.000
25%	0.120	0.000	0.000
50%	0.435	0.161	0.000
75%	0.610	0.526	0.000
90%	0.730	0.728	0.190
95%	0.820	0.817	0.440
99%	0.950	0.974	0.690
Mean	0.396	0.267	0.052
Std. Dev.	0.284	0.286	0.147

Gas companies on average hedge 26.71% of their production. Similar to oil companies, some companies do not apply the hedge (16.00%) and some companies hedge most all the production (5.40% of companies hedge more than 80% of their production). The isalso a sufficient variation of proportion of production hedged (standard deviation equals 0.29).

Average costs per unit produced also have significant variation. The minimum average costs are equal to USD 1.04 per MMBtu., while the maximum average costs are equal to USD 10.04 perMMBtu. The average value of average costs equals USD 3.33 per MMBtu.

Average spread between forward and historical average price is negative and equal -USD 0.13. Minimum price spread is equal to -USD 1.23, maximum price spread is equal to USD 0.75.

The average net debt/EBITDA coefficient is equal to 5.39. 49.16% of gas companies undertake the strategic hedge that fix the selling price in two-year advance.

Gasliquids companies

Gas liquids companies on average hedge much less then oil and gas companies (5.00% of their production). 42.00% of the gas liquids producers do not hedge at all, and 54.00% companies undertake the strategic hedge.

The average costs of sales vary from USD 4.19 per MMBtu to USD 59.96 per MMBtu. The average value of average costs equals USD 18.87 per MMBtu.

Average spread between forward and historical average price is positive and equal to USD 1.53.

The average net debt/EBITDA coefficient is equal to 4.47.

Table II. Descriptive statistics

	Oil	Naturalgas	Gasliquids
Variable	Mean	Std. Dev.	Min	Max	Mean	Std. Dev.	Min	Max	Mean	Std. Dev.	Min	Max
OUTPUT_HEDGED	0.40	0.28	0.00	1.38	0.27	0.29	0.00	1.02	0.05	0.15	0.00	0.79
FWD-HISTORICAL_PRICE	-1.89	3.64	-14.52	9.35	-0.13	0.32	-1.23	0.75	1.53	3.90	-11.92	10.10
AVG COSTS OF SALES	19.40	10.37	2.15	63.70	3.33	1.64	1.04	10.04	18.87	9.19	4.19	59.96
CAPEX	-349.27	292.82	-1366.33	-15.70	-368.27	300.41	-1366.33	-15.70	-387.85	301.52	-1366.33	-15.70
TOTAL_ASSETS	11068.63	12725.34	431.53	73867.00	11657.17	12955.98	431.53	73867.00	10746.40	9898.60	489.56	45564.00
NET_DEBT/EBITDA	5.71	12.83	-6.04	133.68	5.39	12.61	-6.04	133.68	4.47	6.60	-5.07	58.74
REVENUE	768.40	1017.76	24.97	5079.00	802.56	1032.89	24.97	5079.00	732.13	839.84	24.97	3983.00
STR_HEDGE	0.50	0.50	0.00	1.00	0.49	0.50	0.00	1.00	0.54	0.50	0.00	1.00
NO_HEDGE	0.09	0.29	0.00	1.00	0.16	0.37	0.00	1.00	0.41	0.49	0.00	1.00

The model

To investigate the effect of spread between forward level and historical average commodity price and average costs on proportion of production hedged we build the econometric model on the panel data as described above. The general model is following:

,

wherethe dependent variable is - proportion of output hedged and main independent variables are:

- spread between the forward commodity price and historical average price,

- average costs of sales per 1 unit of commodity and controls as described below.

Control variables:

- dummy variable that equals 1 if a company produces natural gas, 0 else;

- dummy variable that equals 1 if a company produces gas liquids, 0 else;

- dummy variable that equals 1 if a company is incorporated in Canada, 0 else;

- netdebt divided by EBITDA, proxy for the debt burden;

- the USD amount for total assets, proxy for company's size;

- the USD amount capital expenses;

- the USD amount of revenue;

- dummy variables that equals 1 for companies that apply strategic long-term hedgeEquals 1 for companies that have open-long term hedging positions (for 2019 year in our sample). Such phenomenon is treated as proxy for more conservative policy that implies more active hedging policy;

- dummy variables that equals 1 for companies that do not execute any commodity hedges.

We also need to account for the fact that the effect of the factors can be different for different products. Moreover, it can differ among the companies that produce one product, but belong to the different groups (for examples, the companies that undertake the strategic hedge and those that do not hedge in advance). To adopt our model for such situations we also include additional group variables:

-product of dummy variable for natural gas and spread between forward commodity price and historical average price( * );

-product of dummy variable for gas liquids and spread between forward commodity price and historical average price ( * );

-product of dummy variable for strategic hedging and spread between forward commodity price and historical average price ( * );

- product of dummy variable for strategic hedging, dummy variable for natural gas and spread between forward commodity price and historical average price ( * * );

- product of dummy variable for strategic hedging, dummy variable for gas liquids and spread between forward commodity price and historical average price ( * * );

-product of dummy variable for natural gas and average costs of sales ( * );

-product of dummy variable for gas liquids and average costs of sales ( * );

-product of dummy variable for strategic hedging and average costs of sales ( * );

- product of dummy variable for strategic hedging, dummy variable for natural gas and average costs of sales ( * );

- product of dummy variable for strategic hedging, dummy variable for gas liquids and average costs of sales ( * );

We also tried to include the polynomial forms of variables (spread between forward and historical average price, average costs of sale per 1 unit produced)in the model in order to check for U-shaped dependency. As the variables for polynomial form are not significant we excluded them from consideration.

We do not include any additional variable for tax regime as all the companies in the sample operate under the same tax conditions.

Variable design

Spread between forward and historical average commodity price (further - price spread)

As the companies in the sample disclose the details on their hedging activities we have the data on the price levels they fixed. These levels can be considered as forward price that were in the market when the companies enter these contracts.

The historical average price is calculated as average price for the commodity for two-year period before entering the contact.

If company does not undertake the hedging,the forward level is calculated based on Bloomberg forward swap prices.

Average costs per unit produced

The variable is calculated as total costs of sales divided by number of commodity units produced in the respective period.

Time lag for independent variables

As the companies hedge the price of their future production, there is time lag between the hedging period and making the hedging decision. To determine the time lag that we should apply in the model we analyzed the average time period between entering the derivative contract and settlements on these contracts. The information presented in quarterly and annual reports shows that the average time lag is approximately 1 year.Thus, all the independent variables are taken with 1 year lag.

Companies can hedge volumes for futureperiods in advance that is more than 1 year.We consider such activities as strategic hedging,but in these situations companies on average hedge the moderate volume. Such strategic hedges do not reflect the final proportion hedged as latercompanies enter additional derivative contracts.On average the final additional hedging contracts with regards to certain period are executed one year before the settlement.

To account for possible difference of companies that execute strategic hedges we added a dummy variable that equals one if company has hedges that fix the selling price for the production for two years in advance.

Model specification

In the analysis we deal with panel data that is two dimensional (across companies and time).The basic econometric models that can be applied to panel data are pooled, fixed effect and random effect models.

Pooled model

The main assumption of the model is absence of the neither significant individual company effects nor significant time effects. In this case the data can be pooled as the coefficients in the model are supposed to be constant.

We believe that the significant variance of volumes hedged depends on a few factors that are not necessarily explained by the variables that are included in the model. Possible omitted variables are attitude towards risk and payoff on the derivatives executed in the past. Thus, we acknowledge that the polling of the data can give us the wrong results and consider its results as a primary one.

Fixed effect model

The main assumption of the fixed effect model is presence of unique company effects that are persistent in time. This models the situation when there are some company-specific effects that are not explained by independent variables in the model. Actually the model is equivalent to the pooled model with dummy variables to account for difference in companies. Such dummy variables can potentially absorb the effect of companies' different attitude towards risk and payoff on the commodity derivatives in the future.

Random effect model

Random effect models also account for individual effect, but in this model individual effect are supposed not to be correlated with the error term. Unlike the fixed effect model the effect of the company individual parameters is supposed to be random.

Empirical results

The results of both fixed and random effect models are presented in in Table III.

The outcome of both models is similar. In order to choose the better specification we conducted the Hausman test that checks whether the correlation between the unobserved time-invariant variable and other independent variables exist. The test shows that the correlation does exist and we should consider fixed effect model as a primary one. This is in line with idea of controlling for unobserved variables such as shareholders' preferences with regards to risk.

Table III. Main results of the regression

		(1)	(2)
VARIABLES	Pridictedsign	Fixedeffectmodel	Randomeffectmodel
FWD-HISTORICAL PRICE	+	0.0222***	0.0201***
		(0.00431)	(0.00407)
AVG COSTS OF SALES	+	0.00910**	0.00677**
		(0.00371)	(0.00270)
CAP EX	+	4.70e-05	1.95e-05
		(6.29e-05)	(5.32e-05)
TOTAL ASSETS	+	7.76e-07	3.74e-07
		(2.57e-06)	(1.93e-06)
NET DEBT/EBITDA	+	0.000448	0.000365
		(0.000396)	(0.000403)
REVENUE	-	-8.40e-05***	-5.72e-05**
		(2.61e-05)	(2.25e-05)
NO HEDGE	-	-0.204***	-0.244***
		(0.0385)	(0.0322)
STR HEDGE	+		0.0323
			(0.0617)
NAT_GAS_D_SPREAD	+	0.0426*	0.120**
		(0.0558)	(0.0520)
GAS_LIQ_D_SPREAD	-	-0.0205***	-0.0186***
		(0.00594)	(0.00571)
STR_HED_D_SPREAD	-	-0.0237***	-0.0298***
		(0.00762)	(0.00709)
NAT_GAS_D_AVG_COSTS	+	0.0341	0.000748
		(0.0258)	(0.0171)
GAS_LIQ_D_AVG_COSTS	-	-0.0101	-0.00123
		(0.00657)	(0.00419)
STR_HED_D_AVG_COSTS	-	-0.0152	-0.00743*
		(0.00500)	(0.00309)
STR HEDGE_NAT_G_D_SPREAD	+	0.166**	0.0817*
		(0.0834)	(0.0787)
STR HEDGE_GAS_L_D_SPREAD	+	0.0216**	0.0293***
		(0.00948)	(0.00894)
STR HEDGE_NAT_G_D_AVG_COSTS	-	-0.105	-0.0264
		(0.0359)	(0.0183)
STR HEDGE_GAS_L_D_ AVG_COSTS	+	0.0185	0.00153
		(0.00886)	(0.00393)
	-		-0.275***
			(0.0768)
	+		0.0491
			(0.102)
CANADA	-		-0.0643
			(0.0867)
CONSTANT	+	0.392***	0.410***
		(0.0446)	(0.0587)
Observations		676	676
R-squared		0.177	0.150
Numberofid		141	141

Standarderrorsinparentheses

*** p<0.01, ** p<0.05, * p<0.1

Regression build on the existing sample is significant (p-value is 0). Explanation power of the model is represented by R2that is close to 17.70%. This means that17.70% of the proportion of production hedged variance can be explained by the factors included in the model.

The model shows that the spread between forward commodity price and historical average price positively influence the proportion of production hedged. And the effect is different for companies that (1) produce different commodities, (2) have established principles to hedge in advance (strategic hedge).

The baseline coefficient that shows the influence of the spread between forward and historical commodity price on the proportion of production hedged is equal to 2.22%. That means that if the spread increases by USD 1, the companies tend to increase the proportion of production hedged on average by 2.22%. The effect is significant at any reasonable significance level.

Companies that produce natural gas are much more sensible to price spread (the variable * is significant, coefficient equals 0.0426).Such phenomenon can be explained by the scale of the price. The price of natural gas fluctuates in nearly 2.5 - 3.5 USD per MMBtu, and price increase by USD 1 is a significant increase, that makes the companies fix such levels. Thus, if gas price spread increases by USD 1, the company tend to increase proportion hedged on average by 6.48%Valid for companies that do not undertake strategic hedge. The coefficients for companies that undertake the strategic hedge are presented in table IV..

Oil liquids companies, on the contrary, tend to hedge less production (as compared to the oil group) with the price spread increase (the variable * is significant, coefficient equals -0.02). If gas liquids price spread increases by USD 1, the company on average tend to increase the proportion hedged by0.17%Valid for companies that do not undertake strategic hedge. The coefficients for companies that undertake the strategic hedge are presented in table IV..

Undertaking the strategic hedge also influences companies' response to change of price spread. If oil companies undertake the strategic hedge they tend to increase the proportion of production hedged by lower amount in response to increase of price spread. This phenomenon can be explained by the fact that companies that undertake strategic hedge are less sensible to price spread. Such companies execute the derivative contracts regardless the market conditions.

Gas companies, on the contrary, tend to increase the proportion of production hedged if they undertake strategic hedge. If gas liquids price spread increases by USD 1, the company that undertake the strategic hedge on average tend to increase the proportion hedged by 18.82%. As mentioned above, the magnitude of reaction can be explained by the scale of price.

Gas liquids companies also tend to increase the proportion of production hedged if they undertake strategic hedge. If gas liquids price spread increases by USD 1, the company that undertake the strategic hedge on average tend to increase the proportion hedged by 4.38%.

Table IV summarizes the coefficients that show by how much the companies tend to increase the proportion of production hedged in response to price spread increase.

Table IV. Increase of proportion of production hedged in response to price spread increase