determinants of corporate capital structure: Evidence from companies in
the United States of America
Master Thesis by
Bilyana Georgieva Kirilova
S3190404
Supervisor: Assoc. Prof. Silviu Ursu
MSc International Financial Management Faculty of Economics and Business
University of Groningen
May 29th, 2017
___________________________________________________________________________
ABSTRACT
This paper examines the role of political uncertainty in corporate capital structure decisions. In addition, the study investigates the moderating effects of firms’ level of internationalization and accessibility to public debt markets on the main relationship of interest. Fixed effects OLS estimation method is used for the regression model organized in two steps and conducted on quarterly based panel dataset of 2012 U.S. listed firms over the period of Quarter 1, 2005, to Quarter 4, 2014. Based on traditional capital structure theories and empirical findings of existing literature I test the relationship between total, short-term and long-short-term leverage, on the one hand, and political uncertainty, on the other hand, with the inclusion of both individual and combined moderating effects of two interaction variables at the first step of the panel regression model. The second step tests the same relationships in the presence of a time dummy variable accounting for the effects of the latest recession in the USA. I find a negative relationship between political uncertainty and firms’ leverage ratio which is weakened in the presence of access to public bond markets as an alternative source of financing and with the increase in the level of internationalization of a firm. The inclusion of the effects of the recession provide further support to the established relationships and do not outweigh the significance of political risk at times of economic uncertainty. My findings are robust with respect to data frequency and seasonality issues when conducting a robustness check of the fixed effects OLS regression model using annual based sample.
___________________________________________________________________________ Field Key Words: Corporate Capital Structure, Political Uncertainty, Source of Capital, Credit rating, Level of Internationalization
1 INDEX
INTRODUCTION………..2
LITERATURE REVIEW AND THEORETICAL BACKGROUND………4
Capital structure………4
Political uncertainty and capital structure……….6
Source of capital and capital structure………...8
Level of internationalization and capital structure……….10
Research question and hypotheses development……….11
METHODOLOGY………...13 Sample selection………13 Variable definition………14 Sample statistics………18 Empirical model………19 EMPIRICAL RESULTS………..22 Correlation analysis……….22
Step 1 FE OLS regression analysis results………..24
FE OLS and pooled OLS regression analysis comparison………30
Step 2 FE OLS regression analysis results – controlling for the effects of the latest recession...32
Robustness check………..34
CONCLUSION………38
Main conclusions………..38
Contributions and implications………..39
Limitations and recommendations for further research………....40
REFERENCES……….43
2
1. INTRODUCTION
Financial markets and firm behavior tend to be influenced by governmental economic policies which regulate the business environment. They can generate uncertainty on country level due to the characteristics of the political decision-making process and the duration of the enforcement of new regulations which as a result may not always benefit firms’ business interests. Consequently, political instability creates frictions in the financing choices of firms. Existing literature has focused on examining the relationship between political uncertainty and corporate investment decisions (Julio and Yook, 2012; Gulen and Ion, 2013), assets prices (Pastor and Veronesi, 2013) and cost of debt (Weisman et al., 2015; Qi, Roth and Wald, 2010), however, there is limited understanding of a significant research question in the field of corporate finance regarding the impact of political risk on corporate capital structure decisions. Since prior studies have profoundly investigated the firm-level determinants of leverage, my study aims at providing an insight into the external determinants of companies’ debt-financing by investigating how political risk as a country-level characteristic along with firm’s source of capital and intensity of its international activities could affect the choice of capital structure.
3
the combined roles of both firm and country-level variables in corporate capital structure adjustments.
The essence of the research is to explain whether political risk influences the financing decisions of companies and to what extend the accessibility to public debt markets and the level of internationalization of firms could alter the effects political uncertainty imposes on firm leverage. Literature offers opposing views regarding the role of political uncertainty in capital structure. On the one hand, high political risk could drive a company to raise external capital as hedging mechanism in case of expropriation of funds; on the other hand, political uncertainty is associated with bankruptcy costs, risk of default, reduced credit supply and increased borrowing rates, hence, shareholders are incentivized to minimize the leverage in the capital structure. In light of the mentioned different views and in order to identify unbiasedly the type of relationship between leverage and political risk, I create a hypothesis to examine the effects political instability imposes on corporate capital structure decisions. Furthermore, building upon the findings of Faulkender & Peterson (2006) and their understanding of credit rationing effect which suggests that private debt markets tend to be more risk-intolerant as opposed to public ones along with the fact that firms without access to public bond markets are limited by the discretion of private financial intermediaries in regards to the amount of leverage that could be raised, I hypothesize as well that companies with better access to financing alternatives, in particular to the public debt market, are less influenced by the effects of political uncertainty when it comes to capital structure decisions. Further, I hypothesize that highly internationalized companies despite being challenged by greater exposure to various political regimes would be less affected by political risk in their capital structure choices since international firms tend to maintain a well-developed internal capital network that provides cheaper financing opportunities leading to lower levels of debt (Doukas and Pantzalis, 2003). Lastly, the combined effect of being internationally active and having access to alternative sources of capital is hypothesized to have a weakening influence on the main investigated relationship.
4
than 75 000 firm-quarter observations covering a period of fluctuating levels of political risk. The data is organized in an unbalanced panel dataset. I use fixed effects ordinary least squares (FE OLS) regression model organized in two steps over the assumed linear relationship between political uncertainty and leverage used by Desai et. al. (2008) and Zhang et. al. (2015). My study builds on their models by including two interaction variables for the effects of level of internationalization and having access to financing alternatives in addition to controlling for the effects of the latest recession. The results at the first step of the regression model indicate that an increase in the lagged political uncertainty index associates with a decrease in firms leverage. Furthermore, this relationship holds when total leverage is replaced by short-term and long-term leverage. In the three extensions of the baseline model both the individual and simultaneous presence of the interaction variables causes the disappearance of the negative relationship between the main two variables. The incorporation of the additional control variable for the effects of the latest recession at the second step of the panel regression supports and further strengthens the conclusions derived from the main models.
The remainder of the paper is structured as follows: first, theoretical background and review of existing literature presenting capital structure choices related to political uncertainty, source of capital concepts and level of internationalization are outlined followed by the development of the research question and hypotheses. Section Methodology addresses data collection specificity and justification of the used methodology. The paper concludes with empirical results, conclusions, contributions and limitations of the current research.
1. LITERATURE REVIEW AND THEORETICAL BACKGROUND 2.1.Capital structure
5
structure by either issuing debt or reducing its equity. The authors continue with the statement that “leverage is completely a function of a firm’s demand for debt” (Faulkender & Peterson, 2006, p.45). Dynamic models of capital structure (Leland, 1994, 1998) demonstrate a relation to the static tradeoff models by outlining the continuous adjustments of companies’ capital structure towards the target optimal ratio by repurchasing shares after a share price increase. According to the underlying assumption of trade-off theory, the availability of capital remains continuously elastic and the cost of debt depends only on the risk of the firms’ projects. Myers (1984) draws a distinction between two types of firm behavior. Less risky firms are prone to borrow more before costs of financial distress outweigh the tax benefits of borrowing whereas more risky firms tend to borrow less. The author concludes that cost of debt is a function of the firm risk exposure and the value of a potential loss. The decision between debt and equity balance could be explained by agency theory as well since agency costs of debt comprise an essential part of the trade-off. Issuing debt acts disciplinary to insiders as it reduces the free cash at their discretion and it leads to increased monitoring by outsiders, thereby minimizing the appearance of opportunism and conflicts with shareholders (Jensen, 1986). On the other hand, agency problems could appear between shareholders and debtholders. If a company is highly levered, the returns of an investment project are utilized in servicing the company’s debt, thereby increasing the value of the bondholders. This motivates insiders either to invest in more risky projects which have lower value to the company or to underinvest where both alternatives result in detrimental effect on shareholders’ value. All in all, the optimal capital structure which allows maximization of shareholders’ wealth promotes the balance between agency costs of debt and the benefits of borrowing.
6
not exist. However, studies conducted on the US market demonstrate that less than 20 % of the companies follow pecking order theory assumptions in terms of capital structure.
Traditional approaches to analyzing the determinants of capital structure have mainly emphasized on the effects of firm-level characteristics. Fairly recently, a new approach has been adopted which takes into account external factors, namely characteristics of the company’s environment on a country-level. The study of Gungoraydinoglu and Őztekin (2011) implies that country factors may affect the trade-off between agency costs and benefits of borrowing when determining the leverage ratio of a company.
In light of the mentioned approaches by which proxies of traditional models have been proved to be legitimate determinants of capital structure, Faulkender and Petersen (2006) argue that firms’ leverage should expand from being a “function of firm characteristics” to accounting also for the factors outside a company which may impose limits on its ability to issue debt. The authors develop the idea of supply-side factors of financing by addressing the providers of debt, namely the private debt market (financial intermediaries such as banks) and public debt market (arm’s length lenders such as bond markets). They found that firms with access to public bond markets are highly levered and more inclined to raise debt as opposed to companies without access to alternative source of capital. As a reaction to the commonly observable findings in capital structure literature which are mostly related to firm-level characteristics of leverage in addition to the fact that it remains debatable which are the external determinants of debt financing choices, in this paper I aim to investigate the effect of combined external and internal factors in the face of political uncertainty, level of internationalization and source of capital on corporate capital structure. Later on, I continue with reviewing of findings of previous research in regards to the relationship between the country-specific factor political uncertainty and company leverage choices.
2.2.Political uncertainty and capital structure
7
8
In line with the idea of supply and demand side developed by Faulkender and Petersen (2006), Zhang et. al. (2015) discuss similarly the supply and demand factors as two channels by which economic policy uncertainty could affect firms’ capital structure. On the one hand, the supply side is associated with exacerbation of the external financing environment which leads to an increase in the volatility of company’s future cash flows and give rise to information asymmetry between shareholders and lenders. In turn, this results in an increase in the borrowing costs leaving companies with the option to minimize the leverage in their capital structure so as to achieve financial flexibility. This view is supported by recent studies on US publicly traded companies stating that high political risk reduces the debt in the capital structure of firms on average (Cao et. al, 2013), and overall, leads to more constraining bank loan conditions (Francis et. al; 2014). On the other hand, the demand effects relate to the willingness of a company to minimize its debt at times of political uncertainty, to suppress its “financing demands” and become conservative in its investment decisions. All in all, regardless of the dominance of either of the channels, both of them suggest a negative impact of political risk over firms’ leverage ratios.
2.3.Source of capital and capital structure
Literature has depicted private lenders as being skillful at investigating firms with high level of information asymmetry through monitoring procedures when deciding on extending credit. Consequently, it can be assumed that the source of capital is connected to the firm’s ability to access the debt market (Faulkender and Petersen, 2006). Opaque firms with high investment opportunities at their discretion are predicted by theory to be limited in their access to external financing. Stiglitz and Weiss (1981) discuss that market frictions such as asymmetric information and investment distortions which navigate the capital structure decision-making process may also suggest that companies are constrained by their lenders’ discretion. Therefore, it could be assumed that the link between where the capital is obtained and a firm’s leverage ratio is defined by the trade-off between choosing a financial intermediary (the private debt market) as financing source that provides monitoring benefits but remains expensive, or the arm’s length public bond market as an alternative.
9
10
2.4.Level of internationalization and capital structure
Literature provides evidence for a positive relationship between level of internationalization and capital structure. The geographical diversification leads to minimization of companies’ risks due to subsidiaries synergies, tax rates benefits and lower distress costs, therefore, this enables internationalized companies to carry more debt. Mittoo and Zhang (2008) find that MNCs with better access to international bond markets have in turns higher leverage ratios. MNCs by definition operate cross-borders, thereby these companies have better access to financing opportunities by being active in countries with more developed financial markets. Operating on many developed financial markets is debated to minimize the financial constraints of companies. Khurana et. al. (2006) and McLean et. al. (2012) highlight that the financial markets development provides better access to alternative and low-cost external funds in addition to reducing the information asymmetry problem between lenders and shareholders. The study of De Jong et. al. (2008) argues that a highly developed bond market stimulates the use of debt in a country, hence, it is expected that companies will be more levered.
11
debt vanishes in the presence of political uncertainty. To this end, it remains debatable what the moderating effect of internationalization would be on the capital structure choice.
2.5. Research question and hypotheses development
In light of the mentioned controversial views regarding the relationship between political uncertainty, source of capital, level of internationalization and capital structure of firms, it remains ambiguous what effects could the mentioned determinants impose on companies’ financing decisions, therefore, I formulate a research question as such:
Does political uncertainty affect the leverage ratio of a company and to what extend the source of capital and level of internationalization moderate the relationship between
political uncertainty and capital structure choices?
When political risk is present, a company faces the likelihood of expropriation of assets in addition to possible governmental restraints which may negatively affect a firm’s overall profitability. In a risky environment agency theory implies that shareholders may prefer to minimize the potential losses of their ownership by choosing to increase debt as a hedging mechanism and reduce their stake at the company. In this way, they transfer the risk to the debtholders who in turn will suffer the costs in case of undesirable political events. This is consistent with the agency problem between shareholders and debtholders. Pecking order theory predicts that information asymmetry and poor law system at times of political uncertainty may incentivize managers to raise debt in order to avoid adverse selection costs. However, political uncertainty is related to high bankruptcy costs which require low levels of debt in the capital structure. At times of political risk creditors are uncertain about the creditworthiness of companies in an unstable political environment therefore the rationing effect will lead to reduced credit supply and increased borrowing rates as a hedging mechanism of the lenders. Shareholders in turn choose a regime of low leverage. Furthermore, both channels (supply- and demand-side effects) of political risk suggest a negative effect on capital structure choices. Having considered the results of recent studies on the US market in respect to the conservative investment behavior of firms at times of political distress, the prolonged regimes of low leverage in the corporate capital structure as a mechanism for a firm to stay financially flexible and independent when exposed to high political risk, I hypothesize that:
12
At times of political uncertainty it might be more difficult for a company without access to public debt market to borrow as financial intermediaries (private debt market) are often influenced by the political environment and in turns companies keep low leverage ratio. For instance, tight financial regulations on commercial banks imposed by the government may limit the functions of an intermediary and as a result of this dependence, private debt market appears more likely to be influenced by political uncertainty. Therefore, I hypothesize that companies with access to financing alternative, namely public debt market, are less influenced by the effects of political uncertainty when it comes to capital structure decisions, hence they keep high leverage ratio.
Hypothesis 2: Access to public debt market weakens the relationship between political uncertainty and firms’ leverage
The impact of level of internationalization and leverage ratio remains debatable in terms of its positive or negative influence, nevertheless, the main characteristics of an international companies is the ability to operate on many developed financial markets which is found to minimize the financial constraints of firms. At times of political uncertainty, being active in countries with developed financial markets provides companies with access to financing alternatives. Therefore, I hypothesize that the level of internationalization of a company has a weakening effect on the relationship between political uncertainty and company debt.
Hypothesis 3: Level of internationalization weakens the relationship between political uncertainty and firms’ leverage
Lastly, in light of the proposed mitigating effects both the access to public debt market and level of internationalization impose individually on the relationship between political risk and corporate capital structure, I hypothesize that their combined impact weakens the mentioned link. Hence, the last hypothesis is formulated as follows:
13
3. METHODOLOGY
3.1.Sample selection
A sample of financial statement data of listed companies in the USA is used for the purpose of the current research. The collected data is on quarterly basis in order to examine more time-series variations as indicated in the studies of Zhang et al. (2015) and Leary & Roberts (2005). The sample period covers the timeline of Quarter 1, 2005, to Quarter 4, 2014, generating a sample size of more than 75 000 firm-quarter observations organized in an unbalanced panel dataset. Having considered the influence of the latest recession, I select the mentioned timeline so as to include the political uncertainty effects of the recent financial crisis on the data thereby enhancing the validity of the results and the inferences that follow. The used secondary financial data is gathered from the databases of Orbis and Datastream, while firms’ credit ratings are obtained from Compustat and, lastly, data on the EPU index as a proxy of political uncertainty developed by Baker et al. (2013) is collectable from the authors’ official website. For the purpose of using regression model approach, I use EViews econometric software.
14
3.2.Variable definition
The variables incorporated in the model are represented by proxies used in prior research in the selected field. All proxies are obtained from the mentioned data sources and all financial ratio variables are winsorized at the bottom 5% and at the top 95% levels to eliminate the effect of outliers. In compliance with Zhang et al. (2015) and Gonenc & de Haan (2014), all control variables along with the main explanatory variable are one-quarter lagged in order to avoid endogeneity and reverse causality problem in the regression models. The lagging is also justifiable from economic perspective since capital structure decision-making process takes time to anticipate and adjust to both external and internal changes in determinants.
Consistent with variables used in prior research such as the studies of Zhang et al. (2015) and Gonenc and de Haan (2014), I use total Leverage, Long-term leverage and Short-term leverage as dependent variables on a firm level in my models in order to construct a greater understanding of the effects of political uncertainty on capital structure. I refer the reader to Table 1 for detailed description of all included variables.
Table 1.Variable definitions
Variable
EPU
LEVERAGE LONG-TERM LEVERAGE SHORT-TERM LEVERAGE
CREDIT RATING DUMMY
INTERNATIONALIZATION MODERATOR1 MODERATOR2 RECESSION DUMMY SIZE LIQUIDITY TOBINSQ PROFITABILITY TANGIBILITY FIRMTAX NONDEBTTAXSHIELD CAPEX RD TC TC DEBT RATIO
Trade credit, defined as accounts payable divided by total assets Accounts payable divided by the sum of accounts payable and total debt Firm-level effective tax rate measured as the ratio of corporate income taxes to EBIT
Ratio of depreciation and amortization to total assets
Ratio of research and development expenses to total assets
Capital expenditure divided by one period lagged total assets. Capital expenditure is defined as the amount paid for the acquisition of fixed assets, intangible assets and other long-term assets
Defined as fixed assets (property, plant and equipment) divided by total assets Long-term debt (>1year) divided by total assets
Short-term debt (≤ 1year) divided by total assets
Time dummy variable indicating the effects of the latest recession which spread from December 2007 (Q42007) to June 2009 (Q2 2009) as reported by NBER; 1 indicates the presence of the recession in a quarter, 0 indicates otherwise.
ROA, measured as net income divided by total assets
Interaction variable defined by multiplying EPU(t-1) and Credit rating dummy Interaction variable defined by multiplying EPU(t-1) and Internationalization
Logarithm of total assets
Current ratio defined as total current assets divided by total current liabilities Proxy for qrowth opportunities, calculated as a ratio of market capitalization of a company
to total assets.
Ratio of foreign sales to total sales
Definition
Based on the economic political uncertainty index of Baker et.al. (2013). The monthly data is transformed into quarterly observations by following the formula of Gulen and Ion (2013): EPUt=(3EPUm+2EPUm-1+EPUm-2)/6 and then divided by 100. Annual EPU equals the average of quarterly EPU
Total leverage ratio computed by dividing total debt to total assets
15
As a proxy of political uncertainty, I adopt the Economic Policy Uncertainty index (EPU) developed by Baker et al. (2013). This measure is a news-based index which reflects the frequency of referring to economic policy and uncertainty in ten leading U.S. newspapers. In addition, it accounts for the expirations of federal tax provisions and the level of mismatch between forecasts of consumer price index and government purchases. The EPU index is sensible in taking into account political shocks such as 9/11, Lehman Brothers default, the Eurozone crisis and the US debt-ceiling dispute (Baker et al., 2013).
The index is constructed on monthly basis. Following the method of Gulen and Ion (2013) assumed later in the study of Zhang et. al. (2015), I adjust the variable to quarterly observations by using the following formula:
EPUt = (3EPUm+2EPUm-1+EPUm-2)/6 and to further scale it, I divide it by 100.
In order to examine the effect of internationalization and source of capital on companies’ leverage ratio, two interaction variables are included in the models. The first moderator reflects the product between political uncertainty and firms’ access to public debt market. The latter component is defined as credit rating dummy variable following Faulkender & Petersen (2006) and Cao et al.(2013). The data of S&P issuer’s long-term credit ratings is generated on monthly basis in Compustat and converted to quarterly observations. 1 indicates the presence of bond rating of a company in either of the months comprising a quarter period and 0 marks otherwise. The second moderator represents the multiplication of EPU and level of internationalization variable. A ratio of foreign sales to total sales is selected as a measure of level of internationalization following the paper of Gonenc and de Haan (2014). I refer the reader to the appendix of this study for a graphic representation of the relationships between the moderators, the main independent and dependent variables.
16
therefore, Tobin’s Q is included as a control variable in my regression models. The study of Zhang et al (2015) discusses that the sensitivity of company’s leverage to political instability could be a result of the reduced financing needs, explained by Faulkender and Petersen (2006) as demand side of the firm. As examined in prior studies, political risk is negatively related to corporate investment levels because of the fluctuating discount rates. This prompts the incorporation of investment ratio change Capex as a control variable in my models in order to control for variations in investment. The paper of Zhang et al. (2015) motivate the usage of TC (trade credit) and TCDebtRatio (trade credit to total debt ratio) as control variables with the conclusion that informal financing through trade credits increases with the upsurge of EPU index and it indirectly affects capital structure decisions. All mentioned, control variables are one-quarter laggedin the regression models. Table 2 provides an overview of both theoretically expected and empirically supported relationships between commonly used firm-level determinants of capital structure included in my study and leverage.
At the second step of the analysis, I control for the effects of the latest recession via adding a Recession dummy variable to the models following Cao et. al.(2013). Overall economic conditions, such as the business cycles, can influence the financing decisions of a company and they can associate with political uncertainty. As indicated by a report of the National Bureau of Economic Research (NBER), the recession spread from December 2007 to June 2009. To ensure the objectivity of the estimated results and the inferences that follow, I include the Recession dummy variable in my models in order to outline the effects of both the political instability and economic conditions on the capital structure of firms. Figure 1 represents the time series plot of the EPU index on quarterly basis along with the recession period as dated by NBER. In line with Baker et al. (2013), it is evident that political instability is not necessarily
Variables Theoretically predicted relationship Empirically reported relationship Theories
Firms size Positive Positive
Negative
Asset tangibility Positive Positive
Negative
Profitability Positive
Negative Negative
Growth opportunities Negative Negative
Positive
Liquidity Negative Negative Pecking order theory
Trade-off theory, Agency theory, Financing hierarchy theory
(pecking order model) Trade-off theory, Agency theory,
Financing hierarchy theory (pecking order model) Trade-off theory, Agency theory,
Financing hierarchy theory (pecking order model) Trade-off theory, Agency theory,
17
related to an economic one since the index is seen to be peaking at both extreme periods of booms and busts in the business cycle. Thereby, economic downturn should not be assumed as natural association to political uncertainty. For instance, the increase of the index in 2009, considered the end of the recession period, is related with events such as the Lehman Brothers bankruptcy and the U.S. presidential elections. Its consequential plummeting happened during the Euro crisis in 2010. The highest point of the index was reached during the debt ceiling disputes at the end of 2011. Therefore, an assumption could be made that most of the events associated with an increase in the EPU index are not necessarily related to business cycles.
The empirical models at both steps of the regression analysis use cross section fixed effects to control for industry specifics of the companies included in my sample. Prior research on the relationship between political uncertainty and capital structure has included period fixed effects to control for overall macroeconomic factors and seasonality in the financing decisions of firms (Zhang et al, 2015). Time fixed effects by definition capture the unobservable characteristics of the data that change over time, but not cross-sectionally, in other word, the intercept of the regression models is allowed to differ over the selected time period, but not across the different entities. In this respect, a multicollinearity concern rises with respect to the main independent variable, namely the EPU index. The statistical feature of the main regressor characterizes it as such that varies over time but not cross-sectionally. Consequently, time fixed effects appear perfectly multicollinear in the baseline models in terms of the main regressor causing the
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Q1 2005 Q3 2005 Q1 2006 Q3 2006 Q1 2007 Q3 2007 Q1 2008 Q3 2008 Q1 2009 Q3 2009 Q1 2010 Q3 2010 Q1 2011 Q3 2011 Q1 2012 Q3 2012 Q1 2013 Q3 2013 Q1 2014 Q3 2014 0 0,5 1 1,5 2 2,5
Figure 1. Time series plot of quarterly EPU index and
recession
18
occurrence of near singular matrix. To avoid multicollinearity issues, time fixed effects are not included where not permitted. The inclusion of the fixed effects in the regressions is in accordance with the outcome of performed Hausman and Redundancy tests which support the
usage of panel regression approach. Later on I conduct a comparison between the outputs of
pooled OLS and FE OLS regressions for which I refer the reader to Table 6 and section Empirical results for further details.
3.3.Sample statistics
19
average of 44,4 % suggesting a significant presence of trade credit in the debt-financing plans of the companies.
3.4.Empirical model
Scheme 1. Schematic representation of the main investigated relationship and the interaction effects of the two moderators
Observations Mean Median
Standard
deviation Maximum Minimum
EPU 79800 1.181 1.156 0.432 2.363 0.513
LEVERAGE 78306 0.209 0.177 0.195 0.642 0
LONG-TERM LEVERAGE 78192 0.174 0.131 0.179 0.584 0 SHORT -TERM LEVERAGE 76689 0.029 0.006 0.046 0.172 0
CREDIT RATING DUMMY 47371 0.374 0 0.484 1 0
INTERNATIONALIZATION 70096 0.220 0.096 0.260 0.785 0 MODERATOR1 45824 0.431 0 0.607 2.038 0 MODERATOR2 68519 0.265 0.099 0.349 1.855 0 RECESSION DUMMY 79800 0.175 0 0.380 1 0 CAPEX 76180 0.049 0.033 0.047 0.183 0.003 FIRMTAX 76250 0.188 0.241 0.186 0.468 -0.243 NONDEBTTAXSHIELD 78105 0.041 0.036 0.024 0.101 0.008 SIZE 78410 13.297 13.348 2.068 16.999 9.595 TOBINSQ 75358 1.457 1.052 1.205 4.781 0.227 RD 50793 0.066 0.023 0.094 0.347 0 LIQUIDITY 77995 2.658 2.034 1.897 7.889 0.690 PROFITABILITY 78366 0 0.040 0.152 0.179 -0.457 TANGIBILITY 78257 0.267 0.184 0.237 0.789 0.015 TC 77825 0.074 0.056 0.061 0.234 0.008 TC DEBT RATIO 77638 0.444 0.305 0.364 1 0.028
Table 3. Sample statistics table . This table reports descriptive statistics of all variables used in the
models. Sample period is from Quarter 1, 2005, to Quarter 4, 2014. All corporate financial variables are winsorized at the bottom 5% and at the top 95% levels
20
I conduct a firm-level analysis as demonstrated in the studies of Zhang et. al. (2015) and Desai et. al. (2008) and build on their models by incorporating the effects of two moderator variables. The same authors have assumed a linear relationship between political uncertainty and company leverage, therefore, I follow the same assumption. The used estimator in the current study is fixed effects ordinary least squares (FE OLS) and it comprises of two steps. In order to examine the relationship between political uncertainty and corporate capital structure and for the purpose of testing the outlined hypotheses, I have constructed the following baseline empirical model. It reflects the linear relationship assumed by Zhang et al (2015) and Desai et. al. (2008) between political risk and firm’s leverage. In the first step of my model the first hypothesis is tested via the following baseline model:
1) LEVERAGEit = α0 + β1*EPU(t-1)+ βn*control variables i(t-1) + α𝑖+ eit
EPU index adjusted by the indicated formula represents one-quarter lagged economic policy uncertainty. Control variables are one-quarter lagged as well.
The first extension of the baseline model builds on the previous model by including an interaction variable, namely the access to public debt market. This model illustrates the impact of the first moderator variable on the main relationship. MODERATOR1 is computed by multiplying the proxy variable of source of capital with the one-quarter lagged proxy of political risk. The second hypothesis is tested by:
2) LEVERAGEit = α0 + β1*EPU(t-1) + β2*CREDIT RATING DUMMY i(t-1) +
β3*MODERATOR1it + βn*control variables i(t-1) + α𝑖 + λ𝑡 +eit
The second expanded model describes the impact of the second moderator on the main relationship. MODERATOR2 is computed by multiplying the proxy variable of level of internationalization with one-quarter lagged proxy of political risk. The third hypothesis is examined by the following extended model:
3) LEVERAGEit = α0 + β1*EPU(t-1) + β2*INTERNATIONALIZATIONi(t-1) +
21
Lastly, the combined effect of both moderators on the main relationship and the fourth hypothesis is tested via the third extended regression model.
4) LEVERAGEit = α0 + β1*EPUt-1 + β2*CREDIT RATING DUMMYi(t-1) +
β3*MODERATOR1it + β4*INTERNATIONALIZATIONi(t-1) + β5*MODERATOR2it +
βn*control variables i(t-1) +α𝑖 + λ𝑡 + eit
In each of the presented models, I use total leverage, short-term leverage and long-term leverage as dependent variables in the regressions.
The second step of the empirical model includes the control of a recession dummy variable to each of the presented regression equations, thereby generating the following additional four models. The control effect is examined only on total leverage as dependent variable in the regressions.
i. LEVERAGEit = α0 + β1*EPU(t-1) + β2*RECESSION DUMMYt + βn*control variables i(t-1)
+α𝑖 + eit
ii. LEVERAGEit = α0 + β1*EPU(t-1) + β2*CREDIT RATING DUMMYi(t-1)+
β3*MODERATOR1it + β4*RECESSION DUMMYt + βn*control variables i(t-1) + α𝑖 + λ𝑡 +
eit
iii. LEVERAGEit = α0 + β1*EPU(t-1) + β2*INTERNATIONALIZATIONi(t-1) +
β3*MODERATOR2it + β4*RECESSION DUMMYt + βn*control variables i(t-1) + α𝑖 + λ𝑡 + eit
iv. LEVERAGEit = α0 + β1*EPU(t-1) + β2*CREDIT RATING DUMMYi(t-1) +
β3*MODERATOR1it + β4*INTERNATIONALIZATIONi(t-1)+ β5*MODERATOR2it +
β6*RECESSION DUMMY t + βn*control variables i(t-1) +α𝑖 + λ𝑡 + eit
22
4. EMPIRICAL RESULTS
4.1.Correlation analysis
24
In order to check the severity of multicollinearity in my data, I have computed Variance Inflation Factor (VIF) estimates in EViews. Centered VIF values higher than 4 indicate potential multicollinearity problems. All regressors used in the study with the exception of the interaction variables, have centered VIF coefficients no higher than 2.77 thereby eliminating the raised concern about multicollinearity.
4.2.Step 1 FE OLS regression analysis results
Within the first step of the research model, twelve regression equations are run in order to obtain results of the testable hypotheses. The estimated coefficients of all regressions are divided in 4 sections as presented in Table 5 where each of the sections indicate the four regression models used to investigate the main relationship between political risk and corporate capital structure reflected in the baseline model and the three extended models. In each of the four sections, leverage, short-term leverage and long-term leverage are used as dependent variables. The table reports in parentheses robust standard errors corrected for heteroskedasticity. A priori to outlining the results of the conducted regressions, it should be mentioned that Wald test is performed on the joint significance of all variables incorporated in the models. The output corresponds with the expectations by showing probability value of 0.000 which demonstrates the high relevance of the used variables in the panel regressions.
25
26
the negative coefficient of NonDebtTaxShield related to short-term leverage is reasonably expected. The fewer the non-debt tax shields, the greater the leverage ratio since it would enable the creation of tax shield to replace the non-debt ones in the short-run. However, the control variable loses its effect with total leverage and relates positively with long-term leverage. This indicates that the presence of sufficient amount of non-debt tax shields enables a company to carry more debt in the long-run. RD expenses appear to cast a significant negative effect only on short-term debt ratio and this tendency is observable in the following three extensions. The negative relationship translates into lower levels of debt due to asymmetric information problems. Capex estimates turn out to be insignificant which is in line with the findings of Zhang et al. (2015). According to the authors, this result translates into a dominance of the supply side, the external factors determining the financing choices of a company as Faulkender and Petersen (2006) describe them in their study, over the demand side, the firm-specific characteristics driving the financing needs. TC and TCDebtRatio coefficients appear to be highly significant; trade credit is positively associated with long-term and total leverage, and negatively with short-term debt ratio. According to De Jong et. al. (2008), trade credit occupies a significant part of short-term debt structure thereby its influence over capital structure in the short-run is justifiable. Moreover, this is in line with the findings of Zhang et al. (2015) where the presence of political instability is found to be positively related with trade credit meaning that companies may choose informal channels of financing at politically uncertain times. The presence of a positive association between the control variable and long-term and total leverage is persistent in the extended three models. In comparison, TCDebtRatio relates negatively to leverage ratios meaning that the weight of importance of trade credit in the financing decisions of companies is non-negligible. Consistent with Zhang et al. (2015), the association of trade credit with leverage measures provides further support of the supply-side factors of capital structure, nevertheless it should be mentioned that due to the high financing costs and the lower flexibility related with an actual purchase of goods, trade credit may not be the primary choice as a financing channel when compared to the formal ones.
27
The effects of all control variables remain relatively the same in the three additional extended models with few minor exceptions which do not change the basic inferences. Here forth, I emphasize on the impact of the main independent variable on the three leverage variables in the presence of the moderators in the following analysis and lastly, I conclude with the results of the cumulative effect of both interaction variables over the main investigated relationship.
Section 2 presents the first extension of the baseline model by the incorporation of the first moderator variable. In its presence, the EPU index continues to be negatively related to leverage. The individual effect of Moderator1 which is used in the model as a proxy indicating whether a company has an alternative for debt financing besides the private debt market, shows a significant positive coefficients when associated with total leverage and long-term leverage, however negative estimates appear when related to Short-term leverage. This relationship to both total and long-term debt measures could be explained by the characteristics of the dummy variable incorporated in the formation of Moderator 1 which reflect the long-term credit rating of the issuer. Evidently, it is sensible to expect that it will relate more prominently to the long-term measure of debt as opposed to short-long-term leverage. This finding is consistent with the results of Faulkender and Petersen (2006). Overall, in the presence of the moderator and the stand-alone lagged variable Credit rating dummy used in the formation of the first interaction variable, the coefficients of the EPU index have become insignificant compared to those in the baseline model, hence, the moderator has the effect of not only weakening the negative relationship between political uncertainty and firms leverage, but also causing its complete vanishing. Consequently, it could be concluded that the second hypothesis finds support by the data. R-squared and Adjusted R-squared are relatively high in this section as well with coefficients between 0.88 and 0.60 which demonstrate the explanatory power of the model on the variation of the movements in the dependent variables.
28
moderator on long-term and total leverage is negative and highly significant which provides evidence to the perception that the more internationalized a firm is with its well-developed operating network of subsidiaries and internal capital markets for less expensive financing alternative, the less levered it is. In addition, this finding is in line with agency cost theory which predicts a negative effect of level of internationalization on corporate leverage ratio. Consistent with the models in the previous three sections, the model used in section 3 provides a good fitness to the data with Adjusted R-squared coefficient per each of the three regression equations of 0.87, 0.61 and 0.83 respectively.
30
4.3. FE OLS and pooled OLS regression analysis comparison
32
4.4.Step 2 FE OLS regression analysis results – controlling for the effects of the latest recession
Table 7 reports the estimates of panel regressions of the baseline model and the three additional extended models indicated in sections (1) to (4) respectively. Results in section (1) are obtained from the baseline model which tests the relationship between political risk and company's leverage ratio. Section (2) reports the output from testing the relationship between political uncertainty and leverage ratio in the presence of interaction variable Moderator1 indicating the access to public debt market. Section (3) outlines the impact of Moderator2 on the main relationship with the interaction variable signifying the presence of level of internationalization. The last section (4) reports the estimates of the effect of both moderators on the main relationship. All sections represent the examined effect of political uncertainty on total leverage with additional control included - a time dummy variable for the effects of the latest recession, namely Recession dummy. The sample period is from Quarter 1, 2005, to Quarter 4, 2014. Robust standard errors corrected for heteroskedasticity are reported in parentheses under the coefficient estimates.
Section 1 reports the output from the baseline model with the inclusion of a dummy variable to control for the effects of the latest recession. The coefficient of the recession dummy variable is highly significant at 1% level. Consistent with the results of Cao et al (2013), the positive effect on total leverage is expected since the impact of the recession led to economic downturn during which increase in leverage is considered the only way for companies to finance their operations. EPU index remains negatively related to total leverage in the presence of the recession dummy variable. In comparison to the estimates of the first step, the coefficient in this section is smaller but with stronger significance than the one in the first step regression analysis. This could be interpreted that political risk does not lose its strength of significance in companies’ capital structure decisions in the presence of economic crisis where the default risk is supposed to outweigh the political one. Consistent with the model in the first step, section 1 provides high coefficients of the Adjusted R-squared (0.86) indicating for the high explanatory power of the model.
33
the additional three sections. The recession dummy loses its influence in the extended models, nevertheless, its presence provides a second support to the established inferences from the first step of the regression analysis.
(1) (2) (3) (4) EPU -0.0025*** -0.0049 0.0020 -0.0018 (0.0008) (0.5913) (0.1778) (0.8062) RECESSION DUMMY 0.0040*** 0.0029 0.0001 0.0302 (0.0008) (0.5055) (0.2784) (0.8394) FIRMTAX -0.0164*** -0.0150*** -0.0123*** -0.0124 (0.0024) (0.0036) (0.0025) (0.0037) CAPEX -0.0045 0.0140 -0.0399** -0.0135 (0.0175) (0.0271) (0.0190) (0.0299) LIQUIDITY -0.0003 -0.0003 -0.0006 -0.0008 (0.0005) (0.0008) (0.0005) (0.0008) NONDEBTTAXSHIELD 0.0058 0.0190 -0.0215 0.0153 (0.0486) (0.0775) (0.0532) (0.0835) PROFITABILITY -0.0730*** -0.0776*** -0.0811*** -0.0851*** (0.0050) (0.0076) (0.0055) (0.0085) RD -0.0009 -0.0139 0.0090 -0.0136 (0.0174) (0.0244) (0.0191) (0.0277) SIZE 0.0089*** 0.0003 0.0034** 0.0026 (0.0015) (0.0024) (0.0017) (0.0026) TANGIBILITY 0.0321*** -0.0234 0.0234** -0.0204 (0.0118) (0.0173) (0.0132) (0.0198) TOBINSQ -0.0052*** -0.0070*** -0.0075*** -0.0079*** (0.0007) (0.0011) (0.0007) (0.0012) TC 0.3645*** 0.3162*** 0.3033*** 0.2828*** (0.0213) (0.0301) (0.0215) (0.0303) TC TO DEBT RATIO -0.2861*** -0.2931*** -0.2810*** -0.2900*** (0.0030) (0.0045) (0.0031) (0.0046)
CREDIT RATING DUMMY - 0.0190*** - 0.0171***
(0.0060) (0.0061) MODERATOR1 - 0.0138*** - 0.0145*** (0.0029) (0.003) INTERNATIONALIZATION - - 0.0147** 0.0228** (0.0058) (0.0103) MODERATOR2 - - -0.0142*** -0.0270*** (0.0029) (0.0061)
Cross section fixed effect Yes Yes Yes Yes
Time fixed effects No Yes Yes Yes
Observations 44929 18355 40749 16975
R-squared 0.8682 0.8844 0.878533 0.890275
Adjusted R-squared 0.8635 0.8761 0.873749 0.881777
F-statistic 186.7686 106.2383 183.6458 104.7727
De pe nde nt variable : Total le ve rage
Table 7. Step 2 FE OLS regression results. The table reports the estimates of panel regressions of the baseline
34
The demonstrated results provide evidence for the negative impact of political uncertainty on corporate capital structure decisions which disappears with the increase of a firms’ level of internationalization and when a company has access to alternative source of capital. My findings of a negative relationship are consistent not only with previous studies documenting the effect of political risk on firms’ financing choices in the USA (Desai et al., 2008; Cao et. al., 2013.) but also with conducted research on Chinese firms as well (Zhang et al., 2015.) In the following section, I conduct a robustness check of the results and conclusions previously reported in order to identify supporting channels of the already established relationship between EPU index and leverage.
4.5.Robustness check
35
their presence lead to insignificant coefficients of the EPU index, thereby providing evidence to the weakened negative relationship between political uncertainty and capital structure. The inclusion of both interaction variables in section 4 demonstrates a clear mitigating power over the negative impact political uncertainty imposes over leverage. The coefficients of the EPU related to the three measures of leverage remain insignificant marking the disappearance of any effect of EPU. The re-estimation overall provides support to the robustness of the main results in terms of data frequency and strengthens the inferences of the performed estimations.
When the focus rests on the comparison between the control variables used in the four sections of the robustness check and their estimates presented in Table 5, no significant deviations from the main estimated model appear. The coefficients of Capex in the original model are insignificant at all levels in all four sections and they remain insignificant in the robustness check. This gives rise to the support of the established reasoning of the supply-side effect shaping the role of political uncertainty in firms’ financing decisions. As for Size the results are consistent with those obtained from the main regression model confirming the positive relationship found with leverage. The same holds for Profitability. In regards to Tangibility coefficients, they are significant and negative in the models only when related to short-term leverage which is consistent with the negative coefficients presented in Table 5. TobinsQ is negatively related to all three measures of leverage in all sections which is in line with the results of the regressions on quarterly basis. TC and TCDebtRatio do not deviate from the already established relationships with capital structure demonstrated in Table 5.
37 (1) (2) (3) (4) EPU -0.0057** 0.0063 -0.0126 -0.0032 (0.0022) (0.4811) (0.6763) (0.3703) RECESSION DUMMY 0.0015 0.0021 0.0046 -0.0029 (0.0015) (0.5691) (0.3404) (0.3917) FIRMTAX -0.0143*** -0.0151*** -0.0100** -0.0114** (0.0048) (0.0051) (0.0048) (0.0053) CAPEX 0.0215 0.0468 -0.0038 0.0282 (0.0309) (0.0338) (0.0329) (0.0364) LIQUIDITY -0.0016 -0.0011 -0.0018* -0.0015 (0.0010) (0.0011) (0.0011) (0.0011) NONDEBTTAXSHIELD 0.0724 0.0207 0.0220 0.0507 (0.0983) (0.1141) (0.1063) (0.1216) PROFITABILITY -0.0804*** -0.0839*** -0.0901*** -0.0945*** (0.0096) (0.0104) (0.0106) (0.0115) RD -0.0239 -0.0223 -0.0078 -0.0169 (0.0325) (0.0350) (0.0364) (0.0396) SIZE 0.0067** -0.0018 0.0031 0.0030 (0.0031) (0.0038) (0.0036) (0.0041) TANGIBILITY 0.0225 -0.0399 0.0028 -0.0474* (0.0238) (0.0248) (0.0259) (0.0277) TOBINSQ -0.0069*** -0.0084*** -0.0091*** -0.0093*** (0.0014) (0.0016) (0.0014) (0.0016) TC 0.3778*** 0.3145*** 0.3285*** 0.2897*** (0.0415) (0.0430) (0.0410) (0.0427) TC TO DEBT RATIO -0.3260*** -0.3141*** -0.3172*** -0.3094*** (0.0058) (0.0064) (0.0058) (0.0063)
CREDIT RATING DUMMY - 0.0100 - 0.0030
(0.0092) (0.0094) MODERATOR1 - 0.0118*** - 0.0170*** (0.0040) (0.0042) INTERNATIONALIZATION - - 0.0675*** 0.0615*** (0.0144) (0.0154) MODERATOR2 - - -0.0492*** -0.0506*** (0.0079) (0.0083)
Cross section fixed effect Yes Yes Yes Yes
Time fixed effects No Yes Yes Yes
Observations 10733 8746 9813 7504
R-squared 0.8982 0.9011 0.9060 0.9063
Adjusted R-squared 0.8813 0.8850 0.8888 0.8900
F-statistic 53.0894 56.0363 52.8744 55.4118
Table 9. Robustness check of Step 2 FE OLS panel regression by using annual observations
The table reports the estimates of panel regressions of the baseline model and the three additional extended models indicated in sections (1) to (4) respectively. Results in section (1) are obtained from the baseline model which tests the relationship between political risk and company's leverage ratio. Section (2) reports the output from testing the relationship between political uncertainty and leverage ratio in the presence of interaction variable MODERATOR1 indicating the access to public debt market. Section (3) outlines the impact of MODERATOR2 which signifies the effect of level of internationalization on the main relationship. The last section (4) reports the estimates of the effect of both moderators on the main relationship. All sections represent the examined effect of political uncertainty on total leverage with additional control included - a time dummy variable for the effects of the latest recession. The sample period is from 2005 to 2014 inclusive. Robust standard errors corrected for heteroskedasticity are reported in parentheses under the coefficient estimates. R-squared and Adjusted R-squared are presented accordingly for each regression. Significance at 1%, 5% and 10% are indicated by ***, ** and *, respectively.
38
5. CONCLUSION
5.1.Main conclusions
This study highlights the importance of a significant field in the corporate finance literature, namely the determinants of corporate capital structure. Using a prominent measure of political uncertainty, in this paper I investigate whether political risk influences firm’s financing decisions by conducting a fixed effects OLS panel regression model on quarterly data of 2012 U.S. listed firms over the period of Quarter 1, 2005, to Quarter 4, 2014. Based on traditional capital structure theory and empirical findings of existing literature I test the relationship between total, short-term and long-term leverage on the one hand and political uncertainty, on the other hand with the inclusion of both individual and combined moderating effects of two interaction variables at the first step of the panel regression model. The second step tests the same relationships in the presence of a time dummy variable accounting for the effects of the latest recession in the USA.
39
negative impact political uncertainty imposes on firms’ leverage. The more internationally active a company is, the more it has the chance to rely both on its internal capital market for cheaper financing through its developed network of subsidiaries and units on the one hand, and on the many financial markets in which it operates that further provide an ample of financing opportunities, on the other hand. Therefore, an internationalized firm is less susceptible to the negative effects political uncertainty unveils on capital structure decisions regardless of the fact that an internationally active firms function in many political regimes.
Lastly, the inclusion of the effects of the recession provide further support to the established relationships and provides no evidence for outweighing the significance of political risk at times of economic uncertainty.
5.2.Contribution and implications
40
In addition to providing guidelines to corporate decision makers and investors alike, the contribution of the study from a managerial point of view could be observed in the employing of a novel measure of political uncertainty, namely the EPU index which could be used as a predictor of not only the financing preferences of firms at politically instable times but also it could yield rich insights into the corporate behavior of companies and the duration of particular leverage regime they choose when exposed to political risk. Furthermore, the study sheds light on the industries most susceptible to political uncertainty as outlined above and emphasizes on the alternatives of prevention from the negative effects of political risk. Since political risk insurances are limited and often inefficient in their coverage, all adjustments of firms’ capital structure remain costly and investors usually bear all these costs in politically instable environment. Therefore, the strong effects of both moderating variables in mitigating the negative constraining impact of political risk could be perceived as prevention alternatives for firms exposed to this type of risk. The more internationalized a firm is, meaning that it is less focused on serving only the local markets and more oriented towards export activities, the more its level of internationalization facilitates circumvention of a common manifestation of political risk such as currency convertibility restrictions. Firms actively engaged in international operations are not limited in their bargaining power against a government that has the ability to enforce higher taxes, expropriate assets or control price levels. Orientation towards sales beyond the local market makes a firm less dependent on local political regimes, less constraint and consequently, less susceptible to negative outcomes of political shifts simply because an increase in the level of internationalization of a firm enables a company to shift its operations to more favorable political or economic conditions. The same liberating effect is observable when considering an expansion in the profile of a company beyond public listing, namely an inclusion of credit rating that guarantees access to alternative suppliers of external funds. All in all, being internationally active and rated could provide insurance for a company’s financing needs at times of political distress.
5.3.Limitations and recommendations for further research
41
country coverage, my study suffers from data availability in terms of credit ratings data. The accessibility to historical information on credit ratings is limited which hampers the collection of data in this respect. Furthermore, only 11% of the companies in my sample are rated which even further decreases the testable sample, prominent conclusions to be drawn from which. Moreover, gathering data from several databases presented drawbacks as the financial data varies with respect to year-end reconciliation in Orbis and Datastream creating mismatching effects between selected searching criteria and actual presented data. What is more, the collected data is available in different frequencies the conversion of which into either quarterly or annual basis may be seen as imposing bias to the results and increasing the probability of calculation error. A better understanding could be established if primary data is used from direct contact with companies or by examining their annual reports.
Furthermore, the usage of OLS estimator raises concerns with respect to potential endogeneity issues and omitted unobserved variables problems. On the one hand, endogeneity may result in inconsistent and bias estimates, and consequently, to misleading interpretations and incorrect findings. Omitted unobserved firm factors, on the other hand, are such that vary over time and cross-sectionally which may impact the main investigated relationship if neglected. In order to address both issues, fixed effects OLS panel regression model is used with an incorporation of lagged variables. Cross-section fixed effects contribute to minimizing the bias emerging from omitted unobserved firm variables while lagging of variables is perceived as a commonly used measure for coping with endogeneity problems to a certain extent. As indicated above, period fixed effects could not have been included in all of the models due to near singular matrix which could be considered a major limitation to the estimation process. However, a better estimator model could be used to provide more reliable results such as the generalized methods of moments (GMM) as it is more sensible in accounting for endogeneity and unobserved heterogeneity in the data than the traditional fixed effects OLS model. In addition, GMM is considered more flexible in allowing dynamism of the investigated underlying process in the model. Future research in this field of interest could produce better results by using the suggested alternative in studies.
42
of them clustered around 0 in the matrix, however, the distinguishing peaking shape in both directions towards 0.4 and -0.4 demonstrates an outlined positive slope which indicates the presence of autocorrelation. In addition, the Durbin-Watson statistics reported in the regression outputs lie in both the inconclusive region and the positive autocorrelation area of critical values in the Durbin-Watson statistical distribution. A positive autocorrelation indicates a relationship between a residual and its immediate previous value, meaning that if a residual at time t-1 is positive, it is highly likely that the residual at time t is positive as well. The consequences of using OLS estimator in the presence of autocorrelation of the error terms are similar to those of neglecting heteroskedasticity, namely the estimated coefficients are still unbiased, but inefficient, a probability of Type 1 error could occur and the standard errors are perceived as incorrect. In order to address autocorrelation, heteroskedasticity-consistent standard errors (White (robust) standard errors) are used in the models, however, the autocorrelation problem remain persistent. Future research could transform model variables into logarithms or to use a more sophisticated estimator model to address the mentioned concerns.
Last, sample selection bias could be considered a limitation to the study as well. The exclusion of certain sectors from the sample along with filtering the data according to a settled threshold of 3 million worth of total assets can deliver distorted empirical results and lower statistical significance of the conducted tests. Therefore, further research would be beneficial if the main relationship along with the moderating effects are examined when the sample of companies include public utilities firms. As indicated above, these firms are particularly susceptible to political agendas and therefore, they suffer extensively from political shifts. Adding this sector to the investigated sample could generate new and different insights into the role of political uncertainty in corporate capital structure.
ACKNOWLEDGEMENT
43
REFERENCES
Anderson, D., Sweeney, D., Williams, T., Camm, J., Cochran, J., 2013. Statistics for Business & Economics. Cengage Learning.
Baker, S., Bloom, N., Davis, S., 2013. Measuring Economic Policy Uncertainty. Chicago Booth Research Paper (13-02)
Economic Policy Uncertainty. US Monthly Index (2012) Retrieved on April 14, 2017 from:
http://www.policyuncertainty.com/us_monthly.html
Bekaert. G., Harvey. C., Lundblad. C., Siegel. S., 2014. Political risk spread. Journal of International Business Study 45. 471-493
Cantillo, M., J. Wright, 2000, How Do Firms Choose Their Lenders? An Empirical Investigation, Review of Financial Studies, 13, 155-189
Cao, W., Duan, X., Uysal, V.B., 2013. Does Political Uncertainty Affect Capital Structure Choices? Working Paper. University of Oklahoma.
Chkir, I., Cosset, J., 2001. Diversification strategy and capital structure of multinational corporations. Journal of Multinational Financial Management 11, No.1, 17-37
Degryse, H., Goeij, P., Kappert, P., 2010, The impact of firm and industry characteristics on small firms capital structure”, Small Business Economics 38, No. 4, 431-447
Desai, M.A., Foley, C.F., Hines, J.R., 2008. Capital structure with risky foreign investment. Journal of Financial Economics 88, No.3, 534–553.
Desai, M.A., Foley, C.F., Hines, J.R., 2004. A multinational perspective on capital structure choice and internal capital markets. Journal of Finance 59, 2451–2488.
De Jong, A., Kabir, R., Nguen, T.T., 2008. Capital structure around the world: The roles of firm- and country-specific determinants. Journal of Banking and Finance 32, 1954-1969.
Digium Content Marketing Team, (n.d.) SMB, SME, and Large Enterprise: Why the Size of
Your Business Matters, Retrieved April 13, 2017 from:
http://blogs.digium.com/2016/02/18/smb-sme-large-enterprise-size-business-matters/
