1
The impact of subsidies on the capital
structure of energy-related companies in
the United States
Haijo Schipper - S2467801
MSc. Finance - Energy Focus Area Supervisor: Dr. Egle Karmaziene
__________________________________________________________________________________ 1. Abstract
The evidence of the effects of subsidies on firms is mixed. The key to finding an academic basis for this debate is leverage. Since in the United States subsidies are tax credits, the effectiveness of tax credits should be measurable in leverage ratios, since effective tax credits reduce the attractiveness of debt tax shields. For this analysis, I use a sample of firms from the United States in both the renewable and fossil fuel energy sector. This study explores relationships between three specific energy subsidy types (Investment Tax Credit, Production Tax Credit and drilling cost deductibility) and leverage. Lastly, for this study I hand-collect a previously unavailable database on the use of the three main energy related federal subsidies by U.S. energy related firms. This thesis reveals a positive and highly significant relationship between renewable energy subsidies in the United States and leverage with pooled OLS. When taking firm level fixed effects into consideration, there is a significant negative effect of renewable energy subsidies on firm leverage. Therefore, apparently U.S. firms in the renewable sector are more leveraged than those only active in the fossil fuel sector. Nevertheless, when renewable energy-related companies start using subsidies, they significantly lower their debt levels, indicating functioning renewable energy tax credits.
Keywords: Capital structure, Production Tax Credit, Investment Tax Credit, drilling cost deductibility,
subsidies, energy.
2 2. Introduction
On the 1st of June, 2017 President Donald Trump announced that the United States would withdraw from the Paris climate change agreement1. This decision marked a political shift by the federal government of the United States in the energy market and especially the market for renewable energy. It was the start of the repeal and reduction of renewable energy subsidies and the repeal of federal laws and requirements for drilling of oil and gas. The political decision of President Trump is supported by empirical research on subsidy efficiency and effectiveness (e.g. Kalkuhl et al. (2013)) and criticized by environmental organizations and climate change research (Karl et al. (2009)).
Kalkuhl et al. (2013) find in their empirical study on carbon emissions that subsidies on green energy without carbon pricing do not cause a significant change in carbon emission. This result provides a key argument for the theory that subsidies are an inefficient or even ineffective government measure to achieve their policy targets. When researchers analyze the effects of subsidies, it is usually from a macroeconomic or environmental perspective (Fölster and Nyström (2010); Czymanski (1987); Kalkuhl et al. (2013); Herguera and Lutz (2003) among others). However, very few researchers have focused on the effects on the firms involved. The available data on subsidies is limited; companies report whether they use subsidies and generally which subsidies they use, but no generalized database on subsidies in the United States exists. Therefore, I decided to collect data on energy related subsidies used from annual reports and test the effects of energy related subsidies companies.
Reinartz and Schmid (2016) find a positive causal relation of production flexibility of energy plants on the leverage of energy related companies. This finding has motivated me to explore the effect of subsidies on leverage ratios of energy related companies further. The pecking order theory of Myers and Majluf (1984) suggests a preference of internal financing of companies over external debt issues over seasoned equity offerings (SEOs). Government subsidies are essentially zero return requiring capital (though there arguably are some application costs for subsidies in terms of salaries). Free capital will always be preferred over return-requiring capital. In the pecking order theory, subsidies would thus be first preference. It logically follows that subsidies would be used instead of reducing the raised funds of equity offering, thereby increasing the leverage ratio.
3 The United States’ government does not pay out all subsidies in cash, but as tax-deductible items. Tax-deductible items increase the amount of non-debt tax shields (NDTS) a company possesses. According to the empirical summary of existing literature on company leverage by Kumar, Colombage, and Rao (2017) there is a negative correlation between NDTS and leverage in the United States. This result is intuitive, since debt creates a tax shield. When non-debt tax shields increase, the advantages of debt financing decrease. If firms would use U.S. subsidies mainly to reduce tax expenses, subsidies are a substitute for a debt tax shield. Therefore, the NDTS theory suggests that subsidies would decrease the amount of leverage of a company and the pecking order theory suggests the exact opposite relationship. In this thesis, I test empirically which of the two effects are more prevalent in practice collecting a newly constructed dataset on energy related U.S. firms.
I focus on three specific types of subsidies: the renewable energy-related production tax credit (PTC), the renewable energy-related investment tax credit (ITC), and the oil-related drilling cost deductibility. These three subsidies are the independent variables in the model. Leverage is measured as debt as a percentage of total market value. The relevant sample range is the years 2008 until 2017. In fact, the tax authorities link these three types of subsidies directly to investment in production of energy. I exclude non-energy related fiscal tax-deductions and legal entities from the analysis, since the aim of this research is to analyze energy-specific subsidies. In order to measure the individual effect of energy-specific subsidies, I include a control variable for other NDTS. Subsidies are company-related and only listed firms are required to report their level of debt and equity. Therefore, I use company data of listed firms. I choose for U.S firms, because the United States has relatively many energy-related firms for a single country and the United States has centralized subsidies, unlike European countries. Centralized subsidies provide the opportunity to analyze subsidies for firms in a comparable situation.
Therefore, the NDTS theory suggests that subsidies would decrease the amount of leverage of a company and the pecking order theory suggests the exact opposite relationship. This
apparent contraction leads to the research question of my paper:
What is the effect of federal subsidies on the leverage of energy related firms in the United States?
4 smart firm’s projects or the poor projects that need subsidies to survive? Empirical research suggests poorly performing firms have higher debt levels, due to historical profits (Kumar, Colombage and Rao, 2015). If this is the case, does that mean green energy firms have poor projects and thus both higher debt levels and higher subsidy levels? I would expect so and test this with regression analysis. My research is based on a regression model conducted by Ordinary Least Squares (OLS) and panel data on companies related to the oil- and gas-industry and the renewable energy sector. The data on total debt, market value, and deferred tax assets are obtained from Datastream. I analyze data from annual reports to obtain the dummy variable for ITC (used or not) and the values of the PTC, as well as the drilling cost deductibility.
5 3. Theoretical framework
In the past, many researchers analyzed the nature of corporate capital structures. Modigliani and Miller (1958) started with the theory that capital structure is irrelevant in a ‘perfect market’. The perfect market assumptions are individually relaxed. Jensen and Meckling, (1976) relax the assumption of no information asymmetry by including agency cost. The static-tradeoff theory of Myers (1984) states the optimal leverage ratio increases with the size of tax shields and decreases with the cost of bankruptcy, since higher leverage increases bankruptcy risk (Harris and Raviv, 1991). Taxes relax the frictionless trading assumption and cost of bankruptcy the bankruptcy cost assumption. The assumption of perfect competition on the financial markets is the only assumption I keep in this thesis.
3.1 Pecking order theory
According to the pecking order theory (Myers and Majluf, 1984), firms prefer using retained earnings to issuing debt and prefer issuing debt to issuing equity. Since subsidies are basically free capital, subsidies are the most desirable type of funding. After subsidies, retained earnings are a desirable source of funds, because it does not require additional issuance of debt or equity. After retained earnings, one would argue loans are desirable. Shareholders prefer to earn as much of the profits as possible and debt has a lower required return than equity, thus new debt is desirable. More profitable firms would also like to take on more debt, because there is a tax shield to benefit from (La Rocca et al., 2009). The pecking order theory suggests from a company perspective both high quality and low-quality firms prefer as much subsidies as possible.
3.2 Research on capital structure
6 are more stable and are able to obtain more debt. A negative relationship between the age of the company and leverage is observed. Young companies are funded by mostly owner’s equity and have less and less attractive debt options at banks. A positive relationship between growth and leverage is found. A growing company requires more funds than just its retained earnings. Since bigger companies have more excess to bank loans, they are more likely to apply for them in order to gain debt tax shields and reduce cost of capital. A negative relation between liquidity and leverage exists. Loss making firms will see both their equity values decrease and their liquidity decrease.
A negative relationship between NDTS and leverage is found by Bradly, Jarrell and Kim (1984). Debt tax shields are seen as one of the main advantages of debt. When this advantage is left out, debt becomes less attractive and thus there is a lower target debt-to-equity ratio. Lastly, a strongly significant negative relationship between risk and leverage is found. There are two theoretical arguments for this, firstly in line with trade-off theory (Kraus and Litzenberger, 1973) an increase in bankruptcy cost times bankruptcy chance makes debt more expensive. The beta of debt increases, increase cost of debt and decreasing attractiveness of debt. Secondly, higher risk causes higher cost of debt compared to firms with lower risk, since we live in a risk-averse world. There are several other researchers conducting research factors that correlate with capital structure that were not included in Kumar, Colombage and Rao (2015). Wald and Long (2007) found that in the US states with the strictest antitakeover laws, firms had higher debt-to-market-value ratios than in states with less strict antitakeover laws. Johnson (1998) found that the level of bank debt has a significant positive correlation with the total leverage of firms, even when the bank debt is excluded from total leverage. This implies that firms which are able to get more bank debt are also able to get more other leverage. This is consistent with the pecking order theory. However, these are not the more profitable firms. Firms with lower debt levels seem to yield higher returns.
3.3 Research in the energy field
7 to become riskier, because the probability of bankruptcy increases. Moreover, the likelihood that not all obligations to debt holders can be fulfilled increases as well. Chmelíková and Somerlíková (2014) show that start-ups in the Czech Republic are more leveraged than other companies. In their sample of a few thousand startups the percentage of debt was approximately 67% for start-ups and 50% for the average Czech company. It is therefore wise to use a control variable for size of the company behind the power plant in the research. These articles suggest size and leverage are negatively correlated. When companies already have high leverage ratios it seems unlikely that there is a strong negative effect of subsidy yield on these companies. Therefore, it is interesting to see whether there is a mediation effect of size of the correlation between subsidies and leverage.
The energy sector uses real option theory (theory Pindyck (1988); application i.e. Murto and Nese (2003)) to analyze net present value (NPV) of different investment opportunities. In real options, investment decisions between two types of energy plants are valued as having two different call options on real assets (the energy plant). The decision to optimally invest in either one is based on prices of underlying resources. Generally, there are two barriers on resource prices. If the price is high enough, the company should invest in the plant that does not require this resource. If the price is low enough, the company should invest in the plant that does require this resource. Between the barriers, the option value of waiting is higher than the earnings from either one of the plants. In this case, it is optimal to wait. In the theoretical framework of Murto and Nese (2003), subsidies do influence choice of power plant as a fixed income to the renewable plant.
3.4 Research on subsidies
8 Czamanski (1987) uses a microeconomic approach to find the effect of location subsidies on corporate decisions in an industrial environment. In this article he uses an algebraic way to find that subsidies change the relation between the value of labor and capital and the wages and cost of capital in the following way:
Before subsidies: Formula 1 𝐹𝐿 𝐹𝐾 =𝑤 𝑟
With subsidy on capital:
Formula 2
𝐹𝐿 𝐹𝐾
= 𝑤
𝑟(1 − 𝑡𝑟)
In these formulas 1 and 2 F is value, L is labor, K is capital, w is wage level, r is cost of capital, and t is percentage of debt. Czamanski (1987) finds that firms with subsidies face a different cost of capital than the market. However, these subsidies only cause inefficiency when they are location-bound. This is not the case for ITC, PTC and drilling cost deductibility.
Herguera and Lutz (2003) use a macroeconomic approach and find that subsidies can help to increase domestic consumer surplus by increasing product quality and ‘leapfrog’ foreign competitors. The effect fades out when the foreign government subsidizes in the same way. On the national level, the surplus can increase, whereas on the international level, this is likely not the case. On a national level, subsidized firms tend to improve product quality (in this case a reduction of CO2 emission can be considered), while internationally this effect might fade out. However, in the energy sector global CO2 emission reduction is the target of the subsidies. This objective differs greatly from the target of other subsidies, which is to boost the local economy. When applying this theory to the case of subsidized firms, emission quality of the industry as a whole might increase.
3.4.1 Trends in energy transition policies in the Western World
9 from coal to renewable (and partly natural gas-related) energy, better isolation in housing, homeowners installing solar panels on their roofs, and innovative ideas from start-ups. The combination of ideas has let EU governments to provide decentralized subsidies.
In the United States, subsidies are more centralized at the federal government (Energy Investment Agency, 2013). In order to keep subsidies measurable, I will focus on the United States in this thesis. Moreover, empirical evidence suggest that non-location bound subsidies are more efficient than location-bound subsidies. The debate of what subsidies are, is addressed in the theoretical background.
3.4.2 United States’ subsidies
10 the indirect subsidies and tax advantages reported in Oil Change International only drilling cost deductibility is directly enough related to core activities of energy related companies. Therefore, I include it as a subsidy.
The US government motivates subsidizing both green and grey energy in several ways. Firstly, the government claims the US should not be dependent on foreign countries for primary needs (Edwards, 2016). Heating of houses and to a lower extend electricity is seen as a primary need. Secondly, energy is a primary need that requires low prices, in order to make sure even the poorest Americans have access to it (Oppenheim, 2016). This way, subsidies are supported (to some extent) by both political parties. The federal Department of Energy (DOE) had a budget of $27 billion, of which $17 billion went to nuclear weapons, and $10 billion to research and subsidies in the energy sector (Edwards, 2016). These are the more direct subsidies to the energy sector. This is $4 billion more than in 2013.
3.5 Research question and hypothesis
In this research, I explore the relationship between subsidies to energy plants and the capital structures of those plants. To summarize, the empirical research in the United States on energy subsidies supports a number of contradicting theories. There are those who believe subsidies are bad and inefficient investments (Edwards, 2016), those who believe subsidies can help ‘leapfrog’ companies to be more competitive than foreign competitors (Herguera and Lutz (2003)), those who believe subsidies are necessary to keep the poor population warm (Oppenheim, 2016) and those who believe green subsidies should be expanded and all grey energy subsides should be terminated (Oil Change International, 2016). On the basis of the theoretical background discussed above, I propose the following research question:
What effect do subsidies have on leverage of energy related companies in the United States?
11 thus increase the amount of equity, reducing both risk and increasing expected return on the projects. This would cause energy related firms to be more profitable and have more debt possibilities. However, subsidies are equity and subsidized firms are expected to have poorer projects. The availability of debt will likely not increase by the same percentage as the equity increases due to paid out subsidies. Moreover, subsidies are often only reputed for a year and are thus uncertain, especially during elections. Therefore, I hypothesize that there is a negative correlation between subsidies and leverage.
In order to test this, I will use the following hypotheses:
Hypothesis 1: There is a negative effect of all three types of United States’ subsidies level on leverage.
Hypothesis 2: When firms without subsidies apply for subsidies successfully, they reduce their leverage ratio.
Hypothesis 1 implies that the effect of NDTS on leverage would be stronger than the pecking order effect. This effect would be stronger, since subsidies are practically 100% NDTS when they are in the form of a tax deductibility. The other effect of issuing less equity only happens after using retained earnings. Retained earnings are also higher when the effective tax rate is lower thanks to the NDTS. Since retained earnings are higher in the pecking order than debt, the company uses retained earnings first. This causes a relatively small increase in issued debt, making the NDTS effect theoretically stronger than the pecking order effect.
12 4. Data and Methodology
4.1 Sample construction
The sample covers energy related companies for the United States. In the United States, there are several sub-sectors in the energy sector. I include sub-sector renewables and oil- and gas-related companies. These are the two sectors where companies apply for the ITC, PTC, and drilling cost deductibility. For the methodology, I use parts of Reinartz and Schmid (2016) to lay the empirical framework. In contrast to Reinartz and Schmid (2016), I focus on the company level of the subsidies, rather than plant level. Therefore, I use the DataStream database to import my financial data from several sources, rather than the Platts World Electric Power Plants (WEPP) database. The data that is required in this research is data on debt/total assets ratios of companies and subsidies on their power plants. Since data on subsidies is not provided by DataStream, I analyze annual reports for ITC, PTC and drilling cost deductibility. Since they are three different forms of subsidies, I report three variables. I measure ITC as a dummy variable, since tax authorities allow companies to note the use of ITC without reporting the actual value. PTC is reported as values, so I include company yearly values of PTC in the sample. Some companies do not mention either tax credit in their annual reports. When they mention deferred taxes, but do not specify these, I removed them from the sample. Some companies specify tax credits, others consolidate all kind of deferred tax measures. When ITC, PTC and deferred taxes are aggregated, I give 1 for the dummy value and not available for the specific tax credit used. In order to measure energy-related subsidies more precisely, I will focus on data from the more centralized federal United States subsidies, rather than other countries with less centralized subsidies. My final sample on market leverage includes 1278firm-year observations from147firms.
Table 1 Descriptive statistics
Mean Median St. dev. Observations Leverage ratio 0.47 0.31 0.55 1278 Size (total assets) 11262 1244 40291 1278 Deferred taxes 1,849,837 201,705 4,255,405 1099 Drilling dummy 0.62 1.00 0.49 1279 Investment tax credit dummy 0.25 0.00 0.43 1264 Renewable energy dummy 0.26 0.00 0.44 1279 Production tax credit 2363 0.00 12016 1137
13 specific year and is 0 if the firm uses neither.
4.1.1 Time period and frequency
Companies report value of their debt either quarterly or annually. In order to include more companies and thereby reduce sample selection bias, I include annual data on companies. In order to correct for possible financial crisis effects in my sample, I take a 9-year sample of the years between 2009 and 2017. This sample excludes the burst market of 2007 and 2008 and thus represents the post-crisis recovering and boom economy. In order to correct for annual differences in leverage ratios I run multiple OLS regressions, namely pooled OLS regression without year- and firm fixed effects, with just year fixed effects and with both year and firm fixed effects.
4.2 Subsidies
For the variable subsidies, I observe annual reports of all the companies in the sample. The ITC, PTC, and drilling cost deductibility are reported in several different ways. Some companies report them as a constituted item such as tax assets and liabilities. Luckily, these companies specify which tax credits they use. Other companies report the subsidies they use separately, which enables me to construct not only a dummy variable, but also a variable which is interval data. When companies report specific credits, I assign a value of one to that dummy variable and zero otherwise. When they report absolute amounts, I include it in the interval data for in this case Production Tax Credit. This way I collect a database of dummy variables for ITC, PTC, renewable energy (ITC + PTC) and drilling cost deductibility. ITC is generally reported as a balance sheet item, which companies write off to use it. PTC is generally reported as subsidies used in a specific year. Since companies can choose either PTC or ITC for a renewable project, most companies use either one. Other companies use ITC for one project and PTC for another project. Therefore, it is possible for companies to use both.
4.2.1 Energy Information Agency database
14 projects can claim $0.012/kWh. The Investment Tax Credit is the alternative for renewable energy by offering a 30% rebate on investment in solar, biofuel or wind energy and a 10% rebate on other renewable investments. The deduction for the intangible cost of oil and gas drilling costs allows oil- and gas-related companies to deduct all their intangible drilling costs over 5 years (Committee for a Responsible Federal Budget, 2013).
4.3 Statistical relationship
In order to test effects of subsidies on leverage of energy related companies, I analyze the subsidies themselves. There are two general ways of conducting this research. I decided to use a multivariable regression analysis for this data. The first is a regression analysis in which subsidy level is the independent variable and the leverage ratio measured as debt/total assets as the dependent variable. Here, some firm fixed effects are desirable, since it shows the effect of increased or decrease use of subsidies within the same firm. The second way to test for the effects of subsidies on leverage is an event study in which we take certain government subsidy announcements and check for changes in leverage after the announcement. In order to decide upon which of the two to measures to use, in this part previous literature and research conducted on these topics is analyzed.
Table 2 Correlations among the variables
Size Total debt Deferred taxes
ITC
dummy PTC DDD RED Leverage Size (Market value) 1,000 Total debt 0,528 1,000 Deferred taxes 0,835 0,776 1,000 ITC dummy -0,027 0,343 0,158 1,000 PTC -0,004 0,210 0,061 0,366 1,000 DDD 0,102 -0,129 -0,030 -0,663 -0,269 1,000 RED -0,035 0,324 0,142 0,965 0,424 -0,697 1,000 Leverage -0,040 -0,015 -0,050 0,016 0,023 -0,103 0,048 1,000 where DDD is drilling deductibility dummy and RED is renewable energy dummy.
15 4.4 Leverage
In this thesis, I calculate leverage as debt as a percentage of market value of the total assets. The reason to take debt/value rather than debt/equity is that debt/value has a shorter range, allowing more accurate estimations.
4.5 Control variables
The main estimation model of this exploratory research is pooled OLS. In this OLS estimate I include control variables size, non-debt tax shields (NDTS). Moreover, I control for country effects by taking a sample of U.S. firms. Furthermore, I control for year by including regressions with year fixed effects. In their research on flexibility, Reinartz and Schmid (2016) take county and year effects also in consideration to leave out year specific effects and the effect of tax systems. When some of the factors summarized by Kumar, Colombage and Rao (2015) and leverage are included in the analysis, this yields the following regression equation:
Formula 3
16 5. Results and discussion
5.1 Main results
In this section, I present the main findings of this research in Table 2 and Table 3. In Table 2, we can see the regression outputs when using pooled OLS without firm or year fixed effects. In Table 2, model I depicts the basic model in which only a constant and a clustered dummy for both renewable energy subsidies are included. In model II, both ITC and PTC are included separately, where ITC is a dummy and PTC is measured as a continuous variable measured in million dollars. In models III and IV, I include both the size control variable, which is measured as market value of the assets in million dollars and the deferred tax measured as net tax assets & liabilities in million dollars. Model III includes aggregated renewable subsidies, whereas in model IV the two subsidies are split. In the pooled OLS model, we can see there is no consistent significant effect of drilling cost deductibility on capital structure. However, the renewable subsidies have a consistent (highly) significant positive effect on the leverage ratio. When comparing the ITC and PTC, ITC seems to have a more significant effect. However, there were fewer companies using the PTC than the ITC, which may cause the difference in level of significance.
Table 3: Capital structure and subsidies as individual observations
17
The dependent variable is LEVERAGE. Estimation models are pooled-OLS regression. T-statistics based on Huber/White robust standard errors clustered by firms are presented in parentheses. ***, ** and * indicate significance on the 1%-, 5%- and 10%-level, respectively.
Size, PTC and deferred taxes are in $ millions. Size is measured as market value of assets.
In Table 3, the models are expended in order to control for firm and year fixed effects. Model I and II include the Models with only year fixed effects, where in Model I the renewable subsidies are separately included and in Model II, they are aggregated. In Model I and II we can see that the effect of renewable subsidies remains positive and significant. The adjusted R-squared increases compared to the models without year fixed effects. In Model III we can observe an interesting change in the relationship. When I include firm fixed effects the relationship suddenly changes into a negative and significant (at the 10% level) effect of renewable subsidies on the leverage ratio of companies. This implies that when all company-year observations are measured separately without considering the company to which it belongs, we find companies with renewable subsidies are more leveraged. However, when firm fixed effects are included, the same company tends to lower its leverage ratio when it starts to use ITC or PTC. The explanatory power of the observation also skyrockets from 0.17 to 0.55 of variance explained. A potential explanation for this shift in sign is that leverage ratios of firms in general using renewable subsidies is higher than those not using renewable energy-related subsidies, while when firms start using subsidies their leverage ratio drops. This would imply generally fossil fuel companies have significantly lower leverage ratios than renewable energy-related companies. Nevertheless, when companies start using
renewable energy subsidies as well, they do use it as a NDTS. The increase in NDTS causes equity to be more attractive and thus companies would reduce their leverage ratios.
Table 4: Capital structure and subsidies with fixed effects
18
(0.044) (0.043) ITC-dummy 0.262 ***
(0.054) Production tax credits 0.001
(0.001) Size -0.003 *** -0.003 *** -0.005 *** (0.001) (0.001) (0.001) Deferred taxes 0.022 *** 0.021 *** -0.009 (0.008) (0.007) (0.009) Observations 967 1087 1087 Adjusted R-squared 0.123 0.167 0.549 Year fixed effects Yes Yes Yes Firm fixed effects No No Yes
The dependent variable is Leverage ratio. Estimation models are Year fixed effects and firm fixed effects regression. T-statistics based on Huber/White robust standard errors clustered by firms are presented in parentheses. ***, ** and * indicate significance on the 1%-, 5%- and 10%-level, respectively. Size, PTC and deferred taxes are in $ millions. Size
is measured as market value of assets.
5.2 Discussion of the results
The empirical results of this paper seem to support both the non-debt tax shield observation (Kumar, Colombage and Rao, 2017) and the pecking order theory (Myers and Majluf, 1984). When firms have more renewable energy subsidies, they are more leveraged. The findings support the opposite of what I expected in hypothesis 1. The evidence suggests subsidized firms have higher levels of leverage, rather than lower. A possible explanation for these empirical results is that debt is higher in the pecking order. An alternative explanation for this observation is that only poor projects are financed with subsidies and thus the quality of subsidized firms is lower. This would be in line with the empirical results on the negative relationship between leverage and profitability of Kumar, Colombage and Rao (2017).
19 A third interesting finding is that in five of the six models whether companies use deductibility of their new exploratory drilling costs or not is statistically insignificant. A potential explanation for this observation is that deductibility of only exploratory drilling costs would not create a strong enough NDTS to change optimal leverage ratios for often highly capital-intensive oil- and gas companies.
5.3 Reliability and validity
For this research, I used several measures to ensure reliability and validity. For the reliability of the sample, I used the entire population of United States energy related firms. However, I did select a survivorship-bias in the sense that only firms that survived the entire measurement period of 9 years are included. Therefore, the population of firms likely has a higher leverage ratio than the sample. This does not interfere with the reliability of the results, since in order to test the effect of subsidies as NDTS, we want to include firms that actually pay taxes. Since firms in default generally make a loss, they do not pay taxes. For the firm fixed effects, I specifically selected fixed effects in order to test the company-specific effects.
The argumentation behind the assumption of causality in my research is that the subsidy system of the United States works in such a way that all subsidies are tax credits. Tax credits are a form of NDTS. Previous researchers among which Bradly et al. (1984) have found NDTS has a negative effect on leverage in the United States, which provides academic ground to assume causality.
5.4 Limitations and further recommendations
20 6. Conclusion
In this thesis, the goal is to explore the effects the three main federal subsidies have on the leverage ratios of companies in the United States. The results show that when firms use renewable subsidies, they consistently have significantly higher debt levels in all models without firm fixed effects. This result suggests that in general the pecking order of financing holds. When firms use subsidies, they have higher leverage ratios than the firms which do not have subsidies.
Furthermore, when the model includes firm fixed effects the sign shifts and gives a small significant negative effect of the renewable subsidies on leverage ratio. This effect is in line with the empirical research on non-debt tax shields. When a firm does not have subsidies in one year and in the next year it does have subsidies, it holds on relatively more equity. The advantage of debt tax shield decreases due to increased non-debt tax shields.
Moreover, the subsidies that do have a significant effect are the renewable subsidies. Especially in the United States, the Investment Tax Credits have a significant effect on leverage. When Production Tax Credits are included as well, the explanatory power increases. Lastly, fossil fuel subsidies in the form of drilling cost deductibility for new locations does not have a significant effect on the leverage ratios of companies.
21 7. References
Bottiglia, R., Pichler, F. (2016) Crowdfunding for SMEs: a European perspective. [Online] Available at:
https://link-springer-com.proxy-ub.rug.nl/book/10.1057%2F978-1-137-56021-6 [Accessed 12 February 2018]
Bradley, M, Jarrel, G.A., Kim, E. (1984) On the existence of optimal capital structure: theory and evidence. Journal of Finance 39(3): 857-878.
Chmelíková, G., Somerlíková, K. (2014) Capital structure in start-up firms in the conditions of the Czech economy. Acta Universitatis Agriculturae et Silviculturae Mendelianae Brunensis 62(2): 363-372.
Committee for a Responsible Federal Budget (2013) The Tax Break-Down: Intangible Drilling Costs. [Online] Available at: http://www.crfb.org/blogs/tax-break-down-intangible-drilling-costs [Accessed 17 May 2018]
Czamanski, D. (1987) The effect of location subsidies on corporate decisions. Regional Science and Urban Economics 17: 411-421.
Dagblad van het Noorden (2017) Ondernemers in Emmen willen omzet en banen halen uit duurzame energie. [accessed 01-02-2018; Available: http://www.dvhn.nl/drenthe/Ondernemers-in-Emmen-willen-omzet-%C3%A9n-banen-halen-uit-duurzame-energie-22625187.html]
Desai, M.A., Hines Jr., J.R. (2008) Market reaction to export subsidies. Journal of International Economics 74(2): 459-474.
Edwards, C. (2016) Downsizing the Federal Government. [Online] Available at: https://www.downsizinggovernment.org/energy/energy-subsidies [Accessed 14 March 2018] Energy Investment Agency (2013) Direct Federal Financial Interventions and Subsidies in
Fiscal Year 2013. [Online] Available at:
22 Fölster, S., Nyström, J. (2010) Climate Policy to Defeat the Green Paradox. Ambio 39(3): 223-235.
Fernandes, B., Cunha, J., Ferreira, P. (2011) The use of real option approach in energy sector investments. Renewable and Sustainable Energy Reviews 15(9): 4491-4497.
Harris, M., Raviv, A. (1991) The Theory of Capital Structure. The Journal of Finance 46(1): 297-355.
Herguera, I., Lutz, S.H. (2003) The effect of subsidies to product innovation on international competition. Economics of Innovation and New Technology 12(5): 465-480.
Jensen, M.C., Meckling, W.H., (1976) Theory of the firm: managerial behaviour, agency cost and ownership structure. Journal of Financial Economics 3(4): 305-360.
Myers, S.C. (1984) The capital structure puzzle. The Journal of Finance 39(3): 574-592
Johnson, S.A. (1998) The effect of Bank Debt on Optimal Capital Structure. Financial Management 27(1): 47-56.
Kalkuhl, M., Edenhofer, O., Lessman, K. (2013) Renewable energy subsidies: Second best policy or fatal aberration for mitigation? Resource and Energy Economics 35(3): 217-234. Karl, T.R., Mellilo, J.M., Peterson, T.C., Hassol, S.J. (2009) Global Climate Change Impacts in the United States. 1st ed. Washington D.C., United States of America: Global Change Research Program
Kraus, A., Litzenberger (1973), R.H. A state-preference model of optimal financial leverage. Journal of Finance 28(4): 911-922.
Kumar, S., Colombage, S., Rao, P. (2017) Research on capital structure determinants: a review and future directions. International Journal of Managerial Finance 13(2): 106-132. Lambrecht, B.M. (2017) Real options in Finance. Journal of Banking and Finance 81 (2017): 166-171
23 Modigliani, F. and Miller, M.H. (1958), The cost of capital, corporation finance, and the theory of investment, American Economic Review 48, 261-297.
Murto, P., Nese, G. (2003) Input price risk and optimal timing of energy investment: Choice between fossil- and biofuels. Bergen University Working paper (January, 2003).
Myers, S.C., Majluf, N.S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics. 13 (2): 187– 221.
Oppenheim, J. (2016). The United States regulatory compact and energy poverty. Energy research & Social Science. 18: 96-108.
Pindyck (1988). Irreversible investment, capacity choice, and the value of the firm. American Economic Review. 31: 25-39.
Powers, M. (2013). Sustainable energy subsidies. Environmental Law. 43(2): 211-232.
Redman, J., Trout, K., Douglas, A., Bossong, K. (2017) Dirty oil dominance: dependent on denial. Oil Change International [published October 2017] [available at:
http://priceofoil.org/content/uploads/2017/10/OCI_US-Fossil-Fuel-Subs-2015-16_Final_Oct2017.pdf]
Regionale Energiestrategie, Hoe ondernemers helpen de energietransitie te versnellen. [Published 25-09-2017]. [Accessed 01-02-2018; available at: http://regionale-energiestrategie.nl/hoe-ondernemers-helpen-de-energietransitie-te-versnellen/]
Reinartz, S.J., Schmid, T. (2016), Production Flexibility, Product Markets, and Capital Structure Decisions. The Review of Financial Studies. 29(6): 1501-1548.
La Rocca, M., La Rocca, T., Gerance, D., Smark, C. (2009). Effect of Diversification on Capital Structure. Accounting and Finance 49:799-826.
Sine, W.D., David, R.J., Mitsuhashi, H. (2007) From Plan to Plant: Effects of Certification on Operational Startup in the Emergent Independent Power Sector. Organization Science 18(4): 578-594.
