A study of the factors influencing the capital structure choice for russian financial organizations
The study of the capital structure determinants for the banking organizations operating in the Russian Federation. The goal of this study is to determine factors that influence the capital structure decisions made by the management of the Russian banks.
| Рубрика | Финансы, деньги и налоги |
| Вид | дипломная работа |
| Язык | английский |
| Дата добавления | 28.11.2019 |
| Размер файла | 321,0 K |
Отправить свою хорошую работу в базу знаний просто. Используйте форму, расположенную ниже
Студенты, аспиранты, молодые ученые, использующие базу знаний в своей учебе и работе, будут вам очень благодарны.
Размещено на http://www.allbest.ru/
The Government of the Russian Federation
Federal State Autonomous Institution for Higher Professional Education National Research University Higher School of Economics
St. Petersburg Branch
St. Petersburg School of Economics and Management
A STUDY OF THE FACTORS INFLUENCING THE CAPITAL STRUCTURE CHOICE FOR RUSSIAN FINANCIAL ORGANIZATIONS
Master's dissertation
Area of studies 38.04.08 «Finance and Credit»
Master Programme “Finance”
VERONIKA BURAN
Reviewer If any
academic degree, position, department ____________________
Name
Research Supervisor
PhD, Professor, Finance Department
Department of Finance
Rogova Elena Moiseevna
Saint Petersburg - 2019
Table of Content
Table of Content
Abstract
Key words
Introduction
Chapter 1. Literature review
1.1 Trade-off theory
1.2 Pecking order theory
1.3 Market timing theory
1.4 Signaling theory
Chapter 2. Methodology
2.1 Data and descriptive statistics
Chapter 3. Results and Discussion
3.1 Data preparation
3.2 Fitting the model
3.3 Fitted models' consistency check
3.4 Results
Conclusion
Conclusion
Contribution and practical implications
References
Appendix
Heteroskedasticity test
Hausman test
Breusch-Pagan Lagrange multiplier test
Results
Significance level
Arellano-Bond test
Sargan test
Abstract
This paper is an empirical study of the capital structure determinants for the banking organizations operating in the Russian Federation. The goal of this study is to determine the factors that influence the capital structure decisions made by the management of the Russian banks and to check if there are special features that differs from the results of the previously developed theories tested on the developed markets behavior. The study will examine the relationships of nine factors with the leverage level of banking organizations (listed and not listed). The analysis is built on a sample of 413 companies observed for the six years (2013 - 2018). The study employs a multivariable dynamic panel regression model (GMM) constructed on the classical variables obtained from the theoretical research on the capital structure as well as market imperfections and a set of the country-specific factors. The results obtained after testing a sample of the six-year period of historical data for the 413 Russian banks reveal that there are significant differences in the key factors influencing the capital structure and their magnitudes from the results of the studies conducted on samples of non-financial organizations or on samples of companies, operating in the developed countries.
Key words
Capital structure, Russian financial organizations, leverage, debt, equity, banking sector.
Introduction
capital structure russian bank
“Capital structure decision is the mix of debt and equity that a company uses to finance its business” (Damodaran, 2001). The target capital structure is the one that maximizes the company's value, or minimizes the cost of capital. The access to sources of financing and its cost are the crucial dimensions of the competition on the market and it has long-term strategic impact on the company's performance on the market and survival. Despite of the fact that the linkages between the corporate strategy and the capital structure decisions made within it and the corporate performance has been widely investigated from the different viewpoints in the past decades, the question of the optimal capital structure and its influence on the corporate governance is still one of the key problems in the corporate finance. In this research the main focus is capital structure, therefore, two terms need to be distinguished - capital structure and financial structure. Capital structure refers to the ratio of the long-term part of debt to the equity, whereas financial structure is a balance between the total amount of interest-bearing debt and the equity (Schmidt, 2019). Since the influence of factors is different to long-term debt and short-term debt levels, influence of the factors affecting the debt-equity mix, capital and financial structures should be tested separately (Rauh, J., Sufi, A., 2010).
The capital structure choice of the company is affected by two different types of factors - internal and external factors. External factors reflect country-specific macroeconomic situation (interest rate and inflation rate), whereas internal factors are specific for each of the observed companies (profitability, size, growth opportunities, etc.). Even though the external factors will be the same for all the companies in the sample, it will be included in the model, since one of the goals of the study is to find out if there are any country-specific features that may lead to differences in the capital structure determinants for Russian companies comparing to the other countries' entities.
The goal of this research is to study the influence of the different factors on the leverage level and to find out the capital structure determinants for the Russian financial organizations taking into account the operational differences between financial and non-financial organizations as well as government regulations and capital requirements. The lack of research in this field as well as the changing regulations for the banking sector in Russia after the 2008 crisis provides a solid informational basis for this study. The research starts with analysis of the theoretical basis and previous studies' results, continues with description of the data sample and methodology and the last part is discussion based on the obtained results.
Chapter 1. Literature review
Optimal capital structure for the company is about finding and maintaining the optimal mix of sources of finance for the company. According to J. Dean the problem of balancing these financial tools is important for three classes of economists: financial specialists of the company concerned with financing firms so as to ensure their survival and growth, managerial economists concerned with capital budgeting and economic theorists concerned to explain investment behavior at both micro and macro levels (Dean J., 1951).
The modern capital structure theory started with the Modigliani, F. and Miller, M. influential proposition of conditions for the firm value independence from the capital structure (1958, Harris and Raviv, 1991). According to the Modigliani, F. and Miller, M. H. “capital structure irrelevance” proposition, if the certain restrictive assumptions hold, the value of the company is not affected by its capital structure, hereby the worth of the company is determined solely by the worth of its real assets (1958). The following conditions are necessary for the proposition to hold: lack of corporate income taxes, no bankruptcy and transaction costs, contract terms equality and strong market efficiency. In real markets none of those assumptions hold because there is always a place for information asymmetry, companies are subjects to corporate taxes (and therefore can enjoy benefits of the tax shield), for all of the companies there are transaction and bankruptcy costs, borrowings have different costs for different firms and it affects firm's earnings. Nevertheless, this proposition is still the basement of the modern corporate finance theory, it provides the market imperfections that can be exploited in order to explain the relationship between the market value of the firm and its leverage level. Based on the Modigliani and Miller's proposition further studies were conducted in order to explain the influence of the market imperfections on the capital structure decisions.
As an attempt to explain the correlation between the factors influencing leverage level and the debt-to-equity ratio 4 major theories were developed. Each of them proposes companies an optimal choice of the capital structure mix, but none of them is completely satisfactory, since the suggested results are controversial (summary of each theory is provided in the table 1) (Myers, C, 1993). Four theories that will be tested in this study are the trade-off theory, the pecking order theory, the market timing theory and the signaling theory.
1.1 Trade-off theory
In the later works, Modigliani, F. and Miller, M. H. had included the influence of corporate taxation in the neutrality axiom, proposing that the company's value does not solely depend on the value of the real assets, but the tax savings from debt (the tax shield) should also be considered (1963). The tax-deductibility of interest may prepossess organizations to finance its operations and investment opportunities with debt level increase rather than new equity issuance; however, the existence of the bankruptcy costs should be considered. Bankruptcy costs sets the optimal balance between the benefits provided by the tax shield and the maximum attainable firm value - it is the point of the optimal leverage level, where the maximum tax savings are equal to the marginal bankruptcy costs (Stiglitz, J., 1969).
Therefore, the trade-off theory, or the theory of leverage, suggests that the companies tend to increase the level of the debt financing in their capital structure in order to exploit the tax benefits of the tax shield up to the point where the cost of financial distress for excessive borrowings will outweigh the benefits (Kraus and Litzenberger, 1973).
1.2 Pecking order theory
The pecking order theory is an alternative to the trade-off theory, which focuses mainly on the asymmetric information rather than financial distress and the corporate taxes. It states that there is a specific order for the source of financing choice - first choice is retained earnings, or the internal cash flow of the company, as it is the cheapest financing source for the company, the second choice is debt, and the last and the most expensive source of finance for companies is equity (Myers and Majluf, 1984).
The main feature of the pecking order theory is that it does not focus on setting the target debt ratio (Myers, S., 1993). According to this theory, debt ratio is not defined by the trade-off between the tax shield and financial distress, it is assumed to change as the result of imbalance between the internal cash flows and investment opportunities for a given period of time. Firms with high profitability and lack of investment opportunities will decrease their leverage, whereas companies with lack of internal resources to finance its investments will be driven to increase borrowings.
1.3 Market timing theory
The third theory is the market timing theory or the windows of opportunity theory and it proposes that firms make their financing choice based on the market performance at the point of time when they need to raise money, and choose the source, that is cheaper or considered undervalued at that point of time (Lucas and McDonald, 1990). This theory appeared in the beginning of 1990's separate from the existing theories on the capital structure matter, therefore, comparing to the first two theories discussed above, the theoretical part of it is underdeveloped, what leads to different approaches of market timing interpretation (Miglo, A., 2010).
1.4 Signaling theory
Signaling theory is the fourth most commonly used theory and it is similar to the pecking order theory as it considers the influence of the asymmetric information on the capital structure choice, but, unlike the pecking order theory, it suggests the positive relationships between the leverage level and the cash flow under asymmetric information (Ross, 1977). According to the signaling theory, firms can use leverage as a sign of an optimistic future for the shareholders and they are credible because bad type firms will not afford to use this costly signal to mimic high quality firms.
The summary of the discussed theories is represented in the table 1 below.
Table 1
Summary of main theories
|
Trade-off theory |
Pecking order theory |
Market timing theory |
Signaling theory |
||
|
Profitability |
Positive |
Negative |
Positive |
||
|
Firm size |
Positive |
Negative |
Positive |
||
|
Growth opportunities |
Negative |
Positive |
|||
|
Tangibility of assets |
Positive |
Positive |
|||
|
Non-debt tax shield |
Negative |
||||
|
Stock market return |
Negative |
||||
|
Average lending rate |
Positive |
Negative |
Negative |
The extensive empirical studies have revealed several key determinants of the company's leverage preferences such as the firm size, growth potential, collateral value of assets and the firm's historical performance (Rajan and Zingales, 1995; Titman and Wessels, 1988). According to the empirical corporate finance literature, size of the firm and collateral value of assets are positively correlated with the leverage (this supports the trade-off theory), while profits and amount of dividends have negative correlation with leverage. However, these studies are mostly focusing on the non-financial organizations, therefore the application of the results for evaluating the capital structure of financial organizations is somewhat questionable.
In the 2015, John R. Graham conducted a research of changes in capital structure of the corporations over the 20th century; according to the results, in 1946 the median firm had no debt in its capital structure, but by 1970, the leverage ratio rose up to 31% (John R. Graham, 2015). In that paper one of the relationships studied was the relationships between the corporate tax increase in the United States (it grew from 10% in 1920 to 52% in 1950) and the gradual substitution of the preferred stock by the debt in the capital structure. This happened because of the tax benefit conferred by the debt - the interest payments on the debt are deducted from the taxable income and provide a tax shield for corporations while dividends do not. Florian Heider in his study examined whether these relationships between tax level and leverage exists or this is just the coincidence and revealed that taxes are an important determinant of capital structure choices - for every percentage point increase of the tax, corporations increase long-term leverage by about 40 basis points (Heider F., 2015). However, those findings are true only for non-bank organizations. The reaction to tax increase of financial institutions is slightly different. A study conducted by T. Berg and J. Gider revealed that banks tend to have more leverage than non-banking organizations and up to 90% of this difference can be explained by the asset risk because banks tend to have a large fraction of assets composed of a well-diversified portfolio of the non-bank debt, therefore the asset risk for banking organizations is lower than for non-banks (2017). Consequently, only well-capitalized banks can afford to increase their leverage after the tax increase without violating the minimum capital and maximum leverage requirements (Schandlbauer A., 2013).
In the literature, there is the evidence that banks tend to have leverage ratio that differs from the non-bank organizations (Berg, T. and Gider, J. 2017). In most of the researches, conducted before the study of Gropp and Heider (2010), financial organizations were excluded from the sample, because the high leverage, that is usual for banking organizations, is an indicator of financial distress for nonfinancial firms (Fama and French, 2010). One of the specific conditions that need to be considered analyzing financial organizations are the capital requirements, which influence the cost of capital for banks but are not applicable for non-financial organizations (Baker M. P., and Wurgler J., 2015). Currently in Russian Federation those requirements are set on the level from 1% up to 8% for different asset classes. Before the research, performed by Gropp and Heider in 2010 most academics agreed that capital requirements are a restriction for such organizations from choosing its capital structure and, therefore, usual capital structure determinants implemented for nonfinancial organizations are not applicable (Mishkin, 2000). However, later this assumption was empirically tested by many researches and the bottom line was that banks tend to hold its equity at levels that are significantly higher than the regulatory minimum reserve requirements (Berger et al, 2008). This results indicate that, regardless of the capital requirements, the capital structure of financial and nonfinancial organizations could be influenced by the same set of factors.
Another important factor that should be considered in evaluating the behavior of the Russian banks regarding their capital structure decisions according to previous research are different reaction of banks and non-banks on the tax raise as there is evidence that banks react less on tax raise as the determinant of their capital structure than non-banks (Schandlbauer A., 2017).
There are several studies examining the leverage preferences of Russian firms: E. Ilyukhin was examining a set of the factors determining the capital structure on a sample of 48 Russian firms for the period 2009-2015, I. V. Ivashkovskaya and M. S. Solntseva in 2007 were testing two capital structure theories, the trade-off theory and the pecking order theory, on a set of 62 Russian firms operating on the different industries in 2002-2005 (Ilyukhin, 2017) (Ivashkovskaya and Solntceva, 2007). The third and the most extensive study in terms of the sample (99 firms) and the timeframe (2002-2011) was a study performed by E. D. S. Silva et al (Silva et al., 2016). It is reasonable to compare the results revealed in these three papers with the results of this study as they are done for the firms operating in the same country and economic realities, however the difference in the sample choice is expected to lead to some differences in the results as here the sample consists of only banking organizations. The results of the discussed studies and the effects of the tested determinants are represented in the table 2 below.
Table 2
Summary of previously conducted studies' results
|
E. D. S. Silva et al. |
I. V. Ivashkovskaya and M. S. Solntseva |
E. Ilyukhin |
||
|
Profitability |
Negative |
|||
|
Firm size |
Negative |
Positive |
||
|
Tangibility |
Positive |
|||
|
Non-debt tax shield |
Positive |
|||
|
Inflation rate |
Negative |
|||
|
GDP growth rate |
Positive |
This chapter sums up the modern capital structure theories development history, description and the discussion of results obtained by researchers in the previously conducted studies. The discussed theories are the trade-off, pecking order, market timing and signaling theories. The results of empirical studies testing those theories are contradictive because of differences in the sample choice and the time spans. This provides the need for the further research in the field in order to reveal the relationships between the capital structure and the explanatory variables for the certain types of organizations/samples; the unit of interest in this research is banking sector due to the lack of research of capital structure determinants for the banking organizations. The next chapter contains description of the methodology - the choice of explanatory variables, measures and hypotheses for the chosen variables and the descriptive statistics.
Chapter 2. Methodology
In all of the studies introduced before, authors were testing the factors affecting the capital structure on the samples constructed from the non-financial organizations operating in the developed markets. The goal of this study is to check whether the reaction of Russian banks will differ from the obtained results and define the primary factors influencing the difference between the financing decisions of the banks and non-banks operating in Russia.
The population of application for the research is all the financial companies, which had a banking license from the Central Bank of the Russian Federation by April 16, 2019. The sample for this research is a convenience sample as there are certain limitations - some companies were excluded on the basis of fulfillment of the preferred criteria for the study, therefore, there were some eliminations from the sample - non-banks and companies that did not have available data for the whole period of observation. The final sample consists of 413 companies, the sampling unit is one particular organization; data for the analysis is retrieved from the Spark database.
Selection of explanatory variables is the crucial task while conducting the analysis of the cross-sectional variation in capital structure (Harris and Raviv, 1991, Titman and Wessel, 1988). For the purpose of developing the working hypothesis and the appropriate regression model the following determinants were chosen:
Profitability (return on assets, return on equity);
Tangibility of assets;
Size;
Earnings volatility;
Macroeconomic factors (interest rate, inflation rate);
Pre-tax income and total taxes paid;
Non-debt tax shield;
Growth opportunities;
Asset risk (asset quality);
For the purpose of the empirical research performance, it is crucial to consider that links between the actually chosen explanatory variables and the theoretical dependent variables are complex and there is a need to consider additional empirical studies and theories in order to justify the choice of the variables and the calculation approach. Theoretical background and previous studies' results for each determinant and their measurements are discussed below.
Leverage determinants
Leverage
Leverage is the explained variable, which plays a key role in the model. The most commonly used formulas to calculate the indebtedness of the company are the ratio of total liabilities to total assets (which usually overestimates the level of leverage) and ratio of the total debt to total assets. The difference is that total debt includes only interest-bearing part of the liabilities - long-term and short-term debt. For the purpose of this study the leverage will be split into long-term and short-term parts, each one will be regressed separately on the outlined factors, and after, the total debt will be included in the model to check the interrelations of the factors with the total amount of debt.
Profitability
In this study, two measures of profitability will be exploited: return on equity (ROE), and return on assets (ROA). According to the trade-off theory positive relationships are expected between the profitability and the leverage level of the company, implying that higher profitability is achieved as a result of the increased tax shield (Kraus and Litzenberger, 1973). However, according to the pecking order theory, the relationships between the leverage level and profitability of the company are inverse and many empirical studies have proven this relationship to be inverse (Ivashkovskaya and Solntseva, 2007, Mazur, K., 2007, Myers and Majluf, 1984). In this study the effect of profitability measures and leverage (both long- and short-term) is expected to be negative.
Hypotheses:
H0: Profitability has a negative effect on LtD of the banks.
H0: Profitability has a negative effect on StD of the banks.
Tangibility of assets
The relationship between the amount of the tangible assets in the asset structure of the company and leverage level is based on the agency theory concept - conflict of interest between the shareholders and lenders leads to the certain requirements for the borrowers. Collateral value of assets may serve as a major determinant of the debt available for the company (Harris and Raviv, 1999). Even though this measure most probably is not appropriate for banks, it is included because there is a need to test it to make sure this is true for the defined sample; if it is irrelevant - it will be observed from the results and excluded from the final model. In several studies conducted previously, tangibility of assets was proven to have direct relationship with leverage - study, conducted by Frank and Goyal revealed positive relationships, the research made by Ivashkovskaya and Solntceva in 2007 exposed the same result (2009, 2007). In this study, if the variable will be proven to be relevant, the relationships between leverage and assets tangibility are expected to be inverse.
Hypotheses:
H0: Tangibility has a negative effect on LtD of the banks.
H0: Tangibility has a negative effect on StD of the banks.
Size
The influence of the firm size on its leverage is proven by number of studies. It is suggested that large firms have more bargaining power and, therefore, they can benefit from economies of scale issuing long-term debt (Marsh, 1982). Furthermore, large firms have less exposure to the risk of bankruptcy because of better diversification. However, the sign of this relationship is unclear. A study, conducted by Fama and Jensen (1983) proven the positive relation between the firm size and its leverage, while Titman and Wessels (1988) found negative correlation between these two indicators. In this study, size is expected to reveal positive relationships with the amount of long- and short-term debt in the capital structure.
There are several indicators that are commonly used to measure the size of the organization - market capitalization, value of the total assets, revenue, amount of the equity capital, etc. According to the Deutsche Bank research performed by Jan Schildbach in 2015, the best indicator to measure the size of bank organizations is revenue, as revenue is a common denominator for all of the diversified activities bank can be engaged in, whereas equity value represents book value of the bank, market capitalization does not represent purely the size of the firm, but the success and market expectations (and is not applicable for the chosen sample - only 67 out of 413 banks are publicly traded), and value of total assets measures nominal gross value of the firm's activities, but it may suffer from the valuation problems and it does not take into account possible different business models individual banks use.
Hypotheses:
H0: Size has a positive effect on LtD of the banks.
H0: Size has a positive effect on StD of the banks.
Earnings volatility
As a measure for the economic trend the earnings volatility (or income variability) parameter was chosen. It is a measure for the business risk and in the Russian economy it may represent the effect of the sanctionary restrictions on the level of the leverage. Variability in earnings implies the higher probability of the bankruptcy, therefore the expected relations between indicators is negative - firms with high variability of the income will have less leverage in their capital structure. The income variability is calculated as the ratio of the pre-tax interest income standard deviation over the total assets. It is expected to prove that companies with higher earnings volatility tend to have lower leverage level.
Hypotheses:
H0: Earnings volatility has a negative effect on LtD of the banks.
H0: Earnings volatility has a negative effect on StD of the banks.
Interest rate and inflation rate
Interest rate and inflation rate represent the external macroeconomic factors, which are included in the model as banks have larger exposure to the fluctuations of the business cycle than non-financial firms. Pecking order theory suggest negative relationships with leverage for both factors, whereas trade-off theory suggests positive relationships. Based on the results of the previously conducted researches and taking into account specific features of the country and the sample type it is expected to reveal positive relationships of macroeconomic factors with the leverage level (Barry et al., 2008; Frank and Goyal, 2009).
Hypotheses:
H0: Interest rate has a positive effect on LtD of the banks.
H0: Interest rate has a positive effect on StD of the banks.
H0: Inflation rate has a positive effect on LtD of the banks.
H0: Inflation rate has a positive effect on StD of the banks.
Pre-tax income and total amount of taxes paid
These two factors will be tested and may not be included in the final model because of the chance to cause the multicollinearity problem in the model, but their influence will be checked individually as correlation with other factors.
Hypotheses:
H0: Total taxes paid has a positive effect on LtD of the banks.
H0: Total taxes paid has a positive effect on StD of the banks.
Non-debt tax shield
The non-debt tax shield, commonly proposed as a capital structure determinant in the most of the previously conducted studies is not considered in the final model, because of the sample specific - banks asset structure differs from the asset structure of the non-bank organizations, therefore the non-debt tax shield, primarily consisting of the asset depreciation benefits is not expected to add a value as a leverage factor for banks. In the previously conducted studies, the influence of the non-debt-related tax shields (e.g. tax deductions for depreciation and amortization or investment tax credit) was proven to be negative (Wanzenried, 2002).
The variable is included in the preliminary analysis (descriptive statistics and the correlation table), however, as well as pre-tax income and the amount of taxes paid, it causes the multicollinearity bias in the sample (represented in the appendix - variance inflation factor) and, therefore, is excluded from the further analysis.
Growth opportunities
The growth opportunities is a very questionable factor in terms of its relationships with the leverage level, as well as in terms of the measurement. It adds value to a firm as an asset, but it is intangible, it cannot be collaterized and it is not charged under the taxable income (Titman and Wessels, 1988). Usually, the positive relations are expected between growth and leverage, because if firm is expected to grow, it will need more resources, which should be raised through the additional debt issuance, according to the pecking order theory.
However, in order to have better justification of the expected influence of the growth opportunities on the capital structure, researchers need to look beyond the classical theories addressing the capital structure issue and consider the influence of the agency theory. According to the agency theory there are two possible ways the leverage level can be influenced by the growth opportunities - either positive or negative. The type of the relationships is expected to follow one of the approaches - underinvestment or overinvestment theory. Under the first approach, there is evidence that management of the companies that have high leverage may forego investment opportunities with positive NPV because of bondholders' priority for the cash flows of the firm. In this case, firms with growth opportunities are not expected to increase the share of debt in their capital structure. Under the overinvestment view leverage is expected to have direct relationships with growth opportunities as raising debt instead of equity in case if there are no productive projects is the way to safeguard the firm from the wrong decisions of the management (Sualehkhattak and Hussain, 2017).
Different researches, conducted previously, revealed different outcomes, for example, Myers (1977), determined that firms with lack of growth opportunities favor debt financing (it goes along with the underinvestment view), while Fama and French (2002) concluded that firms with growth opportunities tend to rely on debt less (it supports the underinvestment view). Other researches suggested different relations with growth opportunities for long-term debt and for short-term debt (Rajan and Zingales, 1995, Titman and Wessels, 1988).
Most commonly implemented measure of growth opportunities are PVGO (present value of growth opportunities), which is calculated as the difference between the stock price and ratio of earnings over the cost of equity and the market-to-book ratio. Because of the limitations caused by the sample choice (most of the banks are not publicly traded) the PVGO and market-to-book measures are not applicable. However, there are other measures available, for instance, Titman and Wessels used percentage change in total assets as the measure for growth opportunities, furthermore, in case of the banking sector, specific growth opportunities measures are available, such as growth rate of deposits and growth rate of advances (1988). In this study the chosen measure of the growth potential is percentage growth in advances as it can also represent the future (expected) gains of the bank.
Hypotheses:
H0: Growth opportunities has a negative effect on LtD of the banks.
H0: Growth opportunities has a positive effect on StD of the banks.
Asset risk
Banks tend to have more leverage than non-banking organizations of the same size (Berg, T. and Gider, J. 2017). According to the study, performed by Berg and Gider on a sample of all companies, listed in the stock exchanges of the U.S. from 1965 to 2013, up to 90% of the differences in the leverage level between banks and non-banks can be explained by the asset risk (Berg, T. and Gider, J. 2017). Asset structure of financial organizations differs from the asset structure of non-banks - assets of banks mainly consist of the well-diversified portfolio of loans provided to the non-financial organizations, therefore, the asset risk for such organizations is much lower that for non-financial organizations of the same size. In this study asset risk will be included as explanatory variable in order to check its relationships with the capital structure construction.
Commonly preferred measure for the asset risk is unlevered equity volatility. Another commonly used measure is the accounting-based ROA volatility over the observed period. This calculation method allows to avoid influence of the market expectations and it is constructed independently from the leverage measures, however, it is somewhat similar to the earnings volatility measure, introduced earlier, furthermore, adding such measure might lead to the multicollinearity bias in the model (with the ROA variable).
According to the research, performed by Beltrame et al. on a sample of 97 banks, there is a significant evidence of the influence of the asset quality (non-performing asset coverage ratio) on the systematic risk of the bank (2018). Considering the results of study of Beltrame et al., specific features of the sample and the inability to exploit common asset risk measures, the non-performing asset coverage ratio is chosen as the measure of asset risk (asset quality).
Hypotheses:
H0: Asset risk has a negative effect on LtD of the banks.
H0: Asset risk has a negative effect on StD of the banks.
The summary table containing the list of variables, calculation methods for each variable and its expected relationships with the capital structure is represented in the table 3 below.
Table 3
Variables and hypotheses summary
|
Variable |
Variable name |
Measure/Proxy |
Expected effect on long-term debt |
Expected effect on short-term debt |
|
|
Leverage (long-term) |
LtD |
= Long-term debt/equity |
|||
|
Leverage (short-term) |
StD |
= Short-term debt/equity |
|||
|
Leverage (total) |
TD |
= LtD + StD |
|||
|
Profitability |
ROA |
= After-tax income/Total Assets |
- |
- |
|
|
ROE |
= After-tax income/Total Equity |
- |
- |
||
|
Tangibility of assets |
AT |
= Fixed assets/Total Assets |
- |
- |
|
|
Size |
S |
= Interest Revenue |
+ |
+ |
|
|
Earnings volatility |
Vol |
= Standard deviation of pre-tax income/Total Assets |
- |
- |
|
|
Interest rate |
Int |
= Average annual interest rate |
+ |
+ |
|
|
Inflation rate |
Inf |
= Average annual inflation rate |
+ |
+ |
|
|
Pre-tax income |
- |
||||
|
Total taxes paid |
Tot_tax |
= Amount of total taxes paid |
+ |
+ |
|
|
Non-debt tax shield |
- |
||||
|
Growth opportunities |
GO |
= Growth rate of advances |
- |
+ |
|
|
Asset risk (asset quality) |
AR |
= Non-performing loans/(Equity + loan loss reserves) |
- |
- |
2.1 Data and descriptive statistics
After the data, required for the analysis was collected, it was observed, that some values are missing in the reports available in the spark database. For some organizations some tax or revenue information was not available, and also, there were banks that were established after 2013 and, therefore the data for such organizations is available for the period less that the observed in this study. The missing values create complications for the researcher and challenges the validity of the observed results. Some of the missing data was obtained from the other available resource databases (Central Bank website, etc.) and organizations that still had missing data, were excluded from the final sample. After all the adjustments made, the final sample contains 413 banks observed over 6 years without any missing values.
The next steps after proper preparation and handling of the data are to evaluate the selected variables using the econometrical techniques including correlation, normality test, multicollinearity and heteroscedasticity tests and select the final model.
The data is strongly balanced. The descriptive statistics for the data is represented in the Table 4. It contains means and standard deviations for all of the variables that were considered for the model with breakdown for within and between variations in the variables. From the table it is easy to observe the individual-invariant regressors, which have within variation equal to zero (inflation rate, interest rate, year) and the time-invariant regressors with zero between variation (bank identification number).
Table 4
Descriptive statistics
|
Variable |
Variation |
Mean |
Std. Dev. |
Min |
Max |
Observations |
|
|
Bank |
overall |
207 |
119.2465 |
1 |
413 |
N = 2478 |
|
|
between |
119.3671 |
1 |
413 |
n = 413 |
|||
|
within |
0 |
207 |
207 |
T = 6 |
|||
|
Year |
overall |
2015.5 |
1.70817 |
2013 |
2018 |
N = 2478 |
|
|
between |
0 |
2015.5 |
2015.5 |
n = 413 |
|||
|
within |
1.70817 |
2013 |
2018 |
T = 6 |
|||
|
TD |
overall |
.0940494 |
.3107881 |
0 |
5.234429 |
N = 2478 |
|
|
between |
.2407232 |
0 |
3.252732 |
n = 413 |
|||
|
within |
.196872 |
-1.912792 |
4.440822 |
T = 6 |
|||
|
LtD |
overall |
.0378495 |
.203296 |
0 |
4.03202 |
N = 2478 |
|
|
between |
.1864427 |
0 |
3.252732 |
n = 413 |
|||
|
within |
.0814773 |
-.8275628 |
1.101239 |
T = 6 |
|||
|
StD |
overall |
.0561999 |
.229479 |
0 |
5.234429 |
N = 2478 |
|
|
between |
.1525759 |
0 |
2.221672 |
n = 413 |
|||
|
within |
.1715465 |
-1.950641 |
4.402973 |
T = 6 |
|||
|
ROE |
overall |
-.0416328 |
3.646127 |
-180.1419 |
19.4656 |
N = 2478 |
|
|
between |
1.476873 |
-29.72678 |
3.317908 |
n = 413 |
|||
|
within |
3.33429 |
-150.4568 |
30.39712 |
T = 6 |
|||
|
ROA |
overall |
.0066568 |
.032708 |
-.4336793 |
.3332298 |
N = 2478 |
|
|
between |
.017875 |
-.0666647 |
.0968451 |
n = 413 |
|||
|
within |
.0274034 |
-.3955669 |
.2874428 |
T = 6 |
|||
|
S |
overall |
13168.91 |
103588.3 |
-208.8202 |
2363635 |
N = 2478 |
|
|
between |
101716.8 |
25.42855 |
1912582 |
n = 413 |
|||
|
within |
20127.58 |
-560407.8 |
464222 |
T = 6 |
|||
|
AT |
overall |
.0437486 |
.05987 |
.0000147 |
.6340694 |
N = 2478 |
|
|
between |
.0521058 |
.0002755 |
.3443267 |
n = 413 |
|||
|
within |
.0295784 |
-.2445147 |
.4774696 |
T = 6 |
|||
|
Vol |
overall |
.020145 |
.0317106 |
.0005976 |
.7653951 |
N = 2478 |
|
|
between |
.0245072 |
.0007307 |
.2180024 |
n = 413 |
|||
|
within |
.0201538 |
-.1169368 |
.5675377 |
T = 6 |
|||
|
Int |
overall |
.092 |
.0247908 |
.055 |
.135 |
N = 2478 |
|
|
between |
0 |
.092 |
.092 |
n = 413 |
|||
|
within |
.0247908 |
.055 |
.135 |
T = 6 |
|||
|
Inf |
overall |
.0716667 |
.0375028 |
.025 |
.129 |
N = 2478 |
|
|
between |
0 |
.0716667 |
.0716667 |
n = 413 |
|||
|
within |
.0375028 |
.025 |
.129 |
T = 6 |
|||
|
Pre_tax_inc |
overall |
2183.704 |
34390.55 |
-218907.8 |
1002588 |
N = 2478 |
|
|
between |
30938.75 |
-54325.53 |
613221.3 |
n = 413 |
|||
|
within |
15080.98 |
-329156.2 |
391549.9 |
T = 6 |
|||
|
Tot_tax |
overall |
557.6501 |
7418.138 |
-28319.59 |
192320.2 |
N = 2478 |
|
|
between |
7097.772 |
-5538.315 |
143012.9 |
n = 413 |
|||
|
within |
2179.93 |
-53942.66 |
49864.94 |
T = 6 |
|||
|
ND_TaxSH |
overall |
55203.33 |
511628.1 |
1.258 |
9536142 |
N = 2478 |
|
|
between |
491485.8 |
9.095833 |
9062899 |
n = 413 |
|||
|
within |
143849.1 |
-3115946 |
4356514 |
T = 6 |
|||
|
GO |
overall |
.1466678 |
1.381694 |
-.9554095 |
41.29249 |
N = 2478 |
|
|
between |
.703359 |
-.2876777 |
11.01718 |
n = 413 |
|||
|
within |
1.18969 |
-10.38337 |
30.42198 |
T = 6 |
|||
|
AR |
overall |
.2290147 |
.7884139 |
0 |
18.03852 |
N = 2478 |
|
|
between |
.5512822 |
0 |
4.786591 |
n = 413 |
|||
|
within |
.564179 |
-4.424585 |
13.59887 |
T = 6 |
Table 5 represents the correlations between the main variables. According to the table, large banks tend to have better profitability and less short-term leverage than the smaller ones. Income variability correlates negatively with the leverage level, while external factors, such as interest rate and the inflation level have weak interrelations with the income variability, therefore it can be concluded that it may not be a good measure for the economic trend if it has no relations with the macroeconomic factors. Surprisingly, the tangibility of assets have a negative correlation with leverage. High correlation between the interest rate and inflation rate can potentially cause the multicollinearity problem, as well as correlation between size, pre-tax income, total taxes paid and the non-debt tax shield.
Table 5
Correlation matrix
|
LtD |
StD |
TD |
ROE |
ROA |
S |
AT |
Vol |
Int |
Inf |
Pre_tax_ inc |
Tot_tax |
ND_Tax SH |
GO |
AR |
||
|
LtD |
1.000 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
StD |
0.028 |
1.000 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TD |
0.675 |
0.757 |
1.000 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ROE |
0.006 |
-0.167 |
-0.120 |
1.000 |
|
|
|
|
|
|
|
|
|
|
|
|
|
ROA |
0.033 |
-0.038 |
-0.006 |
0.302 |
1.000 |
|
|
|
|
|
|
|
|
|
|
|
|
S |
0.034 |
-0.005 |
0.019 |
0.000 |
0.013 |
1.000 |
|
|
|
|
|
|
|
|
|
|
|
AT |
-0.104 |
-0.072 |
-0.121 |
-0.002 |
-0.066 |
-0.054 |
1.000 |
|
|
|
|
|
|
|
|
|
|
Vol |
-0.007 |
-0.023 |
-0.022 |
-0.094 |
0.033 |
-0.022 |
0.159 |
1.000 |
|
|
|
|
|
|
|
|
|
Int |
-0.008 |
-0.035 |
-0.031 |
-0.008 |
-0.078 |
0.009 |
0.030 |
-0.057 |
1.000 |
|
|
|
|
|
|
|
|
Inf |
0.028 |
0.050 |
0.055 |
-0.021 |
0.049 |
-0.007 |
-0.001 |
0.024 |
0.565 |
1.000 |
|
|
|
|
|
|
|
Pre_tax_inc |
0.011 |
-0.019 |
-0.006 |
0.079 |
0.101 |
0.885 |
-0.025 |
-0.029 |
-0.017 |
-0.025 |
1.000 |
|
|
|
|
|
|
Tot_tax |
0.009 |
-0.008 |
0.000 |
0.001 |
0.029 |
0.886 |
-0.032 |
-0.018 |
-0.013 |
-0.023 |
0.926 |
1.000 |
|
|
|
|
|
ND_TaxSH |
0.010 |
-0.005 |
0.003 |
-0.005 |
0.014 |
0.950 |
-0.027 |
-0.018 |
-0.001 |
-0.014 |
0.851 |
0.811 |
1.000 |
|
|
|
|
GO |
-0.003 |
0.013 |
0.008 |
-0.053 |
0.017 |
-0.001 |
-0.050 |
0.147 |
0.008 |
0.041 |
-0.003 |
0.001 |
0.001 |
1.000 |
|
|
