Capital Structure Theory in the Transport Sector: Evidence from Visegrad Group

Abstract

Capital structure plays an important role in corporate finance, especially in the period of restrictive monetary policy in many developed countries. This paper aims to estimate the debt ratio based on five selected financial indicators: tangibility, return on assets, size of total assets, current ratio, and size of total sales using multiple linear regression for four countries, such as the Czech Republic, Hungary, Poland, and Slovakia, as well as the V4 region. The total sample consists of 3828 small- and medium-sized enterprises from the transport sector in the Central European area. These data are drawn from Amadeus by Bureau van Dijk from 2019. The results show that three of the five variables are statistically significant in all models. These findings indicate that transport companies prefer the pecking order theory. We find that the increase in tangibility, return on assets, as well as current ratio, reduce the debt ratio. The outputs provide new theoretical and empirical knowledge regarding transport companies in V4.

1. Introduction

Capital structure describes the company’s combination of long-term debt, specific short-term debt, common equity, and preferred equity. A company’s performance can be affected by its capital structure, as debt levels, cost of capital, and financial risk affect a company’s profitability and growth prospects. In general, a company with a higher debt-to-equity ratio is considered riskier than a company with a lower ratio, but the optimal capital structure is a trade-off between the cost of capital and financial risk. Companies need to find a balance between financing growth through debt or equity and ensuring stability through a sustainable capital structure.

Capital structure is important for several aspects such as cost of capital, risk management, ownership and control, or growth opportunities. First, capital structure affects the cost of capital because businesses relying on debt financing have a higher cost of capital for debt interest payments. Second, risk management is essential for securing financial obligations. Third, the capital structure determines the ownership or control of the business. If a company issues a significant amount of equity, it may distribute the ownership interest to current shareholders. On the other hand, if a company issues debt, it can increase the power of creditors in making financial decisions. Finally, the capital structure helps in gaining access to growth opportunities. Overall, capital structure is an essential aspect of a company’s financial management and can impact its long-term success. Therefore, companies must carefully consider the mix of capital used to finance their operations and investments.

This paper aims to identify statistically significant predictors describing various areas of corporate governance in capital structure affecting the debt ratio in small- and medium-sized transport companies from the Visegrad Group using multiple regression analysis. These potential predictors with possible impacts on capital structure are summarized from previous research. We find that tangibility, the return on assets, and others have been the most frequently used financial indicators in scientific research from Jaworski and Czerwonka (2021) [1], Pacheco and Tavares (2017) [2], Daskalakis et al. (2017) [3], Li (2015) [4], Wellalage and Locke (2015) [5], Nazir et al. (2012) [6], Kędzior (2012) [7], Shahzad et al. (2021) [8], Acedo-Ramírez and Ruiz-Cabestre 2014) [9], Matemilola et al. (2013) [10], and Jõeveer (2013) [11].

The added values are the output models for all countries and the entire V4 region. This model can predict the debt ratio in the Central European region for the transport sector. Moreover, the model indicates the relationship between the predictors and capital structure theory. This research brings new knowledge about an atypical industry like the transport industry in the Central European region, unlike other industries. Research shows that tangibility, profitability, and current ratio are key indicators in all countries of the Visegrad Group. We find that all these variables have a positive impact on reducing the debt level. However, these variables have a different impact in each country. Moreover, we found that profitability reduces the debt ratio more in Czech and Slovak enterprises in the transport sector.

The paper consists of the literature review, materials, methods, results, discussion, and conclusion. The literature review summarizes key findings on capital structure theory and highlights potentially significant predictors affecting the debt ratio expressed as the ratio of total liabilities to total assets. Materials and data describe the initial sample of small- and medium-sized enterprises from the Visegrad Group. In addition, this section presents potentially significant independent variables based on a wide range of theoretical and empirical findings from previous research from around the world. Finally, we explain the procedure and methods used in identifying the relationship between the debt ratio and other indicators. The results demonstrate models for all countries, such as the Czech Republic, Hungary, Poland, and Slovakia, but also the V4 region using multiple linear regression. Finally, we describe and compare these models. The discussion summarizes the key findings from our research, which are compared with other theoretical and empirical findings from other previous research from around the world. The conclusion contains a summary of the main findings.

2. Literature Review

Trade-off theory decides on optimal debt by comparing the tax benefits and bankruptcy costs. Bankruptcy costs increase for higher leverage. On the other hand, businesses have tax benefits because interest is a tax-deductible expense (tax shield). In other words, the capital structure is optimal if the tax benefits are higher than the costs. Moreover, Kopecky et al. (2018) [12] demonstrated a model explaining the lack of strong empirical confirmation in the trade-off model. Second, pecking order theory advocates a hierarchical funding strategy according to the degree of information asymmetry between managers and investors. This theory prefers capital with the lowest degree of information asymmetry to achieve minimum borrowing costs. In other words, internal capital is preferred over external capital, but debt financing plays a more significant role than equity financing in external sources. Pecking order theory does not define the optimal capital structure, unlike the trade-off theory. Third, market timing theory prefers issuing shares during periods with high market-to-book values and debt when market-to-book values are low. In other words, the development of the capital structure is influenced by the development of the stock markets. Leverage decreases during bull markets as opposed to bear markets. Market timing and pecking order theories do not define the optimal capital structure. Jahanzeb et al. (2013) [13] compared three theories on capital structure trade-off theory, pecking order theory, and market timing theory. They found that trade-off theory and pecking order theory played a dominant role in capital structure decision-making, but market timing theory could change managerial decisions based on theoretical and empirical studies. According to Kuc and Kalicanin (2021) [14], trade-off theory explains that higher profitability supports higher levels of debt because profitable businesses are less at risk of financial distress. In addition, these businesses benefit from the tax shield on interest. On the other hand, pecking order theory assumes that an increase in profitability as one of the sources of financing reduces the level of debt. In other words, there is an inverse relationship between profitability and debt. Second, the size of the business reflects strength and stability. According to the trade-off theory, larger firms with stable cash flows are less threatened by financial distress. There is a positive relationship (positive sign) between the size and the level of indebtedness in contrast to the pecking order theory. Third, tangibility has a negative impact on the debt ratio in the largest companies. Fourth, this research uses annual sales growth as a growth indicator. This indicator does not have a statistically significant impact on any of the three capital structures. Fifth, volatility is measured as the standard deviation of return on assets (EBIT/total assets). The results show that the average volatility reaches about 2%. However, this indicator is not a statistically significant variable like the indicator of revenue growth in Serbia. Sixth, Kuc and Kalicanin (2021) [14] found that liquidity has a statistically significant effect in all capital structure models; however, two models using the dependent variable as short-term indebtedness and total indebtedness show a negative relationship according to the pecking order theory. On the other hand, the model with the dependent variable as long-term indebtedness demonstrates a positive impact according to the trade-off theory. Seventh, research has shown that the cash gap has a statistically significant negative impact on short-term indebtedness and total indebtedness. In other words, businesses with a negative cash gap do not prefer interest-bearing financing due to the transfer of this financial burden to business partners as suppliers. Finally, the effective interest rate has no impact on the capital structure of larger Serbian enterprises. In addition to these financial indicators, the research also deals with gross domestic product, inflation, and the banking sector. El-Chaarani [15] examined determinants affecting bank liquidity in the Middle East region based on four specific factors, such as assets quality, performance level, capitalization ratio, and bank size, and three macroeconomic factors, such as economic growth, unemployment, and inflation. Moreover, the author described the impact of capital structure on the performance of listed enterprises from eight European countries such as France, Italy, Spain, Germany, Austria, Switzerland, the United Kingdom, and Ireland based on various legal protection systems [16].

Li (2015) [4] analysed the significant variables affecting the capital structure in 256 American companies from 21 manufacturing industries from 1974 to 1992 using ordinary least square. The capital structure depends on key variables, such as asset growth, tangibility, research and development, and advertising. The results show that asset growth has a negative coefficient because companies expecting higher asset growth prefer equity finance. In other words, the long-term debt ratio decreases. On the other hand, own, tangibility increases this indebtedness, because businesses can provide higher security for creditors. Finally, research and development expenditures reduce the long-term debt ratio, and the increase in these expenditures benefits the reduction of long-term indebtedness. All these variables are statistically significant in contrast to other indicators, while this model achieves adj. R-square less than 10%. These results differ from the model based on data obtained from 1982. This model shows statistics of significant variables such as tangibility, research and development expenditure, advertising expenditure, but also profit margin. However, asset growth is not a statistically significant variable. All these variables reduce long-term indebtedness, except for tangibility. Finally, research has identified a quadratic relationship between risk and leverage. Castro et al. (2015) [17] explained the influence of selected factors on the capital structure expressing the long-term debt ratio. Research has demonstrated European enterprises’ different capital structures and life cycles for technology and non-technology industries from several countries such as Austria, Belgium, France, Germany, Italy, the Netherlands, Spain, and the United Kingdom from 2000 to 2012. Descriptive statistics have shown significant differences between technology and non-technology industries because non-tech industries have higher average values for leverage, profitability, size, tangibility, and age. Castro et al. (2015) [17] found that, unlike tangibility, profitability has a negative effect on the debt ratio at all levels of the company’s life cycle, such as introduction, growth, maturity, shake-out, and decline, regardless of the industry, technological or non-technological enterprises.

Wellalage and Locke (2013) [5] investigated the relationship between corporate governance and capital structure in New Zealand using conditional quantile regression. The most important factors influencing the capital structure are specific corporate indicators such as tangibility, liquidity, risk, growth, tax shield, and size. The results show that these variables have a different impact on debt ratio, for example, a positive significant relationship between size and debt ratio, on other hand, a negative significant relationship between foreign share ownership or risk and debt ratio. Finally, Wellalage and Locke (2013) [5] found that industry as a dummy variable significantly impacts companies with different debt ratios. Serrasqueiro and Caetano (2015) [18] investigated how capital structure decisions are influenced by the basic assumptions of the trade-off theory or the pecking order theory. Research has shown that profitable and older Portuguese SMEs do not use debt; in other words, businesses prefer the pecking order theory. On the other hand, these firms are interested in increasing leverage for larger firm size; in other words, firms draw insights from both theories. The results show that profitability, size, and age are three statistically significant indicators affecting total indebtedness. In addition, the authors found that more profitable and older firms tend to be less indebted than larger firms by the logarithm of total sales. In other words, Portuguese SMEs prefer internal rather than external finance. On the other hand, firm size indicates that larger firms diversify their capital structure to reduce the likelihood of financial distress. The profitable and older businesses have less tendency to take on debt; these results support the Pecking order theory. On the other hand, small- and medium-sized enterprises approach the optimal level of debt according to the trade-off theory. Pacheco and Tavares (2017) [2] explained that profitability, tangibility, firm dimension, liquidity, and risk are significant attributes, unlike growth, tax benefits, and age, influencing the capital structure of Portuguese SMEs in the hospitality sector. One of the main outputs is a look at trade-off theory and pecking order theory. These theories of capital structure should not be seen in isolation. First, Pacheco and Tavares (2017) [2] found that return on equity has a different impact compared to expectations on capital structure; these results do not indicate pecking order theory. However, the return on assets shows the expected negative impact. In other words, businesses prefer internal over external finance. Second, on the other hand, tangibility and growth do not have a statistically significant impact on the capital structure expressed as total debt/total assets. Third, size is a significant indicator for short-term debt ratio (negative relationship) and long-term debt ratio (positive relationship) based on the fixed effects model (FEM) and random effects model (REM). These results indicate that short-term financing is more affordable than long-term financing. Fourth, the results reveal that businesses with low liquidity prefer short-term liabilities. The results show that there is a negative relationship between liquidity and short-term total debt, or total debt. These findings indicate that firms with liquidity problems have more short-term liabilities. Fifth, other tax benefits besides debt are not relevant indicators for capital structure. Sixth, age has a negative impact on the capital structure representing debt ratio. Finally, the risk is expressed as solvency ratio and structure ratio; risk is positively related to debt ratio. In other words, companies with higher risk can reduce their agency costs. On other hand, Mota and Moreira (2017) [19] found that age, asset structure, return on assets, and tangibility have a positive impact, unlike non-debt tax shield and liquidity, on the capital structure of 26 Portuguese companies investing in Angola. These findings indicate that corporate investment as an internationalization process does not change financial policy. The profitability coefficient supports the trade-off theory because profitability is positively associated with debt in contrast to the pecking order theory. In Ireland, Bhaird and Lucey (2010) [20] found that age, size, intangible activity level, ownership structure, and collateral provision are significant variables determining the capital structure of SMEs. Moreover, age, size, ownership structure, and collateral provision have a universal effect on information asymmetry across all industries. Businesses overcome asset shortages by providing personal assets as collateral for business indebtedness or external capital to support research and development projects. The results show that collateral is important for eliminating information asymmetry and supporting long-term financing.

Hang et al. (2018) [21] found that tangible assets, market-to-book ratio, and profitability are significant variables of the capital structure representing the debt level. Tangible assets have a positive sign, unlike others. Hang et al. (2018) [21] highlighted three key findings; first, tangibility reduces the cost of financial distress; second, enterprises with many investment opportunities should focus on equity financing; third, more profitable enterprises should prefer internal project financing to reduce leverage. Similarly, Gómez et al. (2014) [22] described that the marginal effects model is better for describing the relationship between capital structure and financial indicators from 2004 to 2008. The results show that profitability is one of the significant factors influencing the capital structure of Peruvian enterprises. Moreover, Gómez et al. (2014) [22] explained that businesses with higher debt levels have tax advantages under the trade-off theory as opposed to the pecking order theory. This theory prefers internal financing to debt financing or issuing shares. For-profit businesses can reduce debt levels using retained earnings. Second, larger businesses are less likely to fail. In addition, these companies have a higher level of debt than smaller companies; long-term debt is less popular than short-term debt because of the possible conflict between shareholders and creditors. Third, the pecking order theory explains that a high level of risk increases the probability of financial failure. Fourth, businesses with growing sales need to increase their capital. In general, growth generates more future revenue, but also more profits. Fifth, non-debt tax shields are used as tax protection from debt, and an inverse relationship is assumed between NDTS and debt ratio. Finally, available internal resources are expressed by liquidity and profitability. According to the pecking order theory, these indicators can be negatively correlated with capital structure. However, the trade-off theory explains that businesses can increase their debt levels for tax benefits. These benefits can build a positive relationship between profitability and leverage.

Table 1. Summary.

Authors	TANG	ROA	SZTA	LIQU	SZTS
Jaworski and Czerwonka (2021) [1] 1	+ TOT	− POT	+ TOT	− POT
Pacheco and Tavares (2017) [2]		− POT	+ TOT	− POT
		− POT	+ TOT	+ TOT
		− POT	+ TOT	+ TOT
Daskalakis et al. (2017) [3]	− POT	+ TOT			+ TOT
		+ TOT			− POT
Li (2015) [4]			+ TOT	− POT
			+ TOT	− POT
Wellalage and Locke (2015) [5] 2	− POT		+ TOT	− POT
Nazir et al. (2012) [6] 3	− POT	− POT	+ TOT
Kędzior (2012) [7] 4		− POT
Shahzad et al. (2021) [8] 5	− POT	− POT	+ TOT	− POT
	− POT	+ TOT	− POT	− POT
	− POT	− POT	+ TOT	− POT
	− POT	− POT	− POT	− POT
Acedo-Ramírez and Ruiz-Cabestre (2014) [9] 6					+ TOT
					+ TOT
					− POT
					+ TOT
					+ TOT
Matemilola et al. (2013) [10] 7	+ TOT	− POT	+ TOT
	+ TOT	− POT	+ TOT
Jõeveer (2013) [11] 8	− POT	− POT
	− POT	− POT

Note: tangibility (TANG, fixed assets/total assets), return on assets (ROA, EBIT/total assets), size (SZTA, natural logarithm of total assets), current ratio or liquidity (LIQU, current assets/current liabilities), size (SZTS, natural of total sales), trade-off theory (TOT), pecking order theory (POT). ¹ According to 8–9. Model; ² According to quantile regression (Q 0.05); ³ ROA = EBT/TA; ⁴ ROA = EBT/TA. According to the model with the highest R-square; ⁵ According to GMMs for Pakistan, India, Sri Lanka, and Bangladesh; ⁶ According to models for France, Germany, Italy, Spain, and UK; ⁷ According to pooled OLS and FEM; ⁸ ROA = EBT/TA. According to the model with the highest R-square for listed and unlisted companies.

summarizes the impact of five indicators on debt ratio from previous research. We find that, according to tangibility, most companies prefer the pecking order theory in contrast to the trade-off theory by Jaworski and Czerwonka (2021) [1] and Matemilola et al. (2013) [10]). Similarly, return on assets demonstrates that many businesses prefer the pecking order theory by Jaworski and Czerwonka (2021) [1], Pacheco and Tavares (2017), Nazir et al. (2012), Kędzior (2012) [7], Matemilola et al. (2013) [10], and Jõeveer (2013) [11]. Acedo-Ramírez and Ruiz-Cabestre (2014) [9] found that Italian businesses prefer the pecking order theory, unlike others such as French, German, Spanish, or British businesses. These businesses prefer the trade-off theory, according to which higher sales indicate potential expansion requiring a higher debt ratio. Similarly, Daskalakis et al. (2017) [3] presented contradictory results. On the other hand, liquidity is mainly related to POT. These results show that higher liquidity, and more available funds, reduce the debt ratio. However, Pacheco and Tavares (2017) [2], as the only one of the compared outputs, claimed that higher liquidity contributes to a higher debt ratio, unlike Jaworski and Czerwonka (2021) [1], Li (2015) [4], Wellalage and Locke (2015) [5], and Shahzad et al. (2021) [8]. In addition, we find that company size by the natural logarithm of total assets indicates the trade-off theory, except for Shahzad et al. (2021) [8]. Only the models for Indian and Bangladeshi firms yield different results. Finally, we summarize that most indicators such as tangibility, return on assets, and liquidity are associated with the pecking order theory in contrast to company size as the natural logarithm of total assets or total sales. This summarizes the empirical findings from the previous research.

We formulate five hypotheses about the relationship between capital structure as debt ratio (DERA, calculated by total liabilities to total assets) and selected independent predictors.

Hypothesis 1 (H1).

TANG has a negative impact on DERA.

Hypothesis 2 (H2).

ROA has a positive impact on DERA.

Hypothesis 3 (H3).

SZTA has a negative impact on DERA.

Hypothesis 4 (H4).

LIQU has a positive impact on DERA.

Hypothesis 5 (H5).

SZTS has a negative impact on DERA.

3. Materials and Methods

This paper aims to model the debt ratio of transport companies in all countries of the Visegrad Group such as the Czech Republic, Hungary, Poland, and Slovakia, as well as the region based on financial indicators selected from a wide range of theoretical and empirical knowledge from ongoing research using multiple linear regression.

Sample. The sample consists of small- and medium-sized enterprises, these businesses play an important role in every sector, creating most of all businesses. In addition, the transport industry creates a dynamic environment for the relocation of goods, services, and passengers; movement is a driving force in every economy. The total sample includes 3828 small- and medium-sized enterprises from the transport sector in the Visegrad Group (Czech Republic, Hungary, Poland, and Slovakia). As can be seen, almost 85% of all samples are medium-sized (see

Table 2. Sample.

		Category of Company		Total
		Small	Medium-Sized
Country	CZ	138	432	570
	HU	228	1022	1250
	PL	85	1222	1307
	SK	143	558	701
	V4	594	3234	3828

Note: Czech Republic (CZ), Hungary (HU), Poland (PL), and Slovakia (SK). Source: [23].

).

Variable. We draw financial data on transport companies from the Amadeus database from Bureau van Dijk/Moody’s Analytics. The input data are used to calculate selected financial indicators from previous research worldwide. We apply debt ratio (DERA) as the dependent variable calculated total liabilities to total assets by Nazir et al. (2012) [6], Matemilola et al. (2013) [10], Wellalage and Locke (2013) [5], Acedo-Ramírez and Ruiz-Cabestre (2014) [9], Daskalakis et al. (2017) [3], Pacheco and Tavares (2017) [2], Jaworski and Czerwonka (2021) [1], and Shahzad et al. (2021) [8].

Table 3. Independent variables and their impact according to the capital structure theory.

Acronym	Variable	Formula	Reference	TOT	POT	Our Expectations
TANG	tangibility	FA/TA	Kędzior (n.d.) [7], Matemilola et al. (2013) [10], Wellalage and Locke (2013) [5], Li (2015) [4], Sikveland and Zhang (2020) [24], Jaworski and Czerwonka (2021) [1]	+	−	+
ROA	return on assets	EBIT/TA	Matemilola et al. (2013) [10], Daskalakis et al. (2017) [3], Pacheco and Tavares (2017) [2], Jaworski and Czerwonka (2021) [1], Shahzad et al. (2021) [8]	+	−	−
SZTA	company size	ln of TA	Matemilola et al. (2013) [10], Wellalage and Locke (2013) [5], Jaworski and Czerwonka (2021) [1], Shahzad et al. (2021) [8]	+	−	+
LIQU	liquidity	CA/CL	Wellalage and Locke (2013) [5], Pacheco and Tavares (2017) [2], Jaworski and Czerwonka (2021) [1], Shahzad et al. (2021) [8]	+	−	−
SZTS	company size	ln of TS	Kędzior (2012) [7], Acedo-Ramírez and Ruiz-Cabestre (2014) [9], Daskalakis et al. (2017) [3]	+	−	+

Note: fixed assets (FA), total assets (TA), EBIT (earnings before interest and tax), current assets (CA), current liabilities (CL), total sales (TS). TOT and POT according to author and Rahman (2019) [25].

shows five independent variables such as tangibility (TANG), return on assets (ROA), company size as the natural logarithm of total assets (SZTA), liquidity (LIQU), and company size as the natural logarithm of total sales (SZTS) with calculation and reference to several authors. Moreover, we present the expected results compared to the usual impacts of the indicator depending on the capital structure theory.

These variables, such as tangibility, profitability, liquidity, company size as the natural logarithm of total assets, and the natural logarithm of total sales, are among the most frequently used indicators. We assume that the increase in profitability (negative coefficient) compared to the natural logarithm of total assets (positive coefficient) has a positive impact on reducing the debt ratio in transport companies because the company can rely on its resources rather than bank loans. Moreover, an increase in the share of non-current assets in total assets (positive coefficient) or the company size according to the natural logarithm of total sales (positive coefficient) can lead to an increase in debt ratio because the company can guarantee its assets to the bank when obtaining new bank loans. Finally, we assume that liquidity (negative coefficient) has a positive impact on reducing the debt ratio because if the company has enough cash to cover short-term liabilities, bank loans are not needed to cover short-term trade liabilities. In other words, profitability and liquidity reduce the debt ratio according to the pecking order theory, in contrast to the natural logarithm of total assets, the natural logarithm of total sales, and tangibility, which support an increase in the debt ratio according to the trade-off theory.

Table 4. Descriptive statistics for the total sample.

		DERA	TANG	ROA	SZTA	LIQU	SZTS
N	Valid	3390	3560	3556	3828	3544	3427
	Missing	438	268	272	0	284	401
Mean		0.77	0.45	0.06	7.37	7.44	7.51
Median		0.72	0.46	0.05	7.26	1.47	7.72
Mode		0.02 a	0.00	0.00	6.64 a	0.12 a	7.98
Std. Deviation		0.58	0.28	0.18	0.73	129.02	1.21
Skewness		5.84	−0.01	−4.84	0.74	45.83	−2.04
Kurtosis		0.04	0.04	0.04	0.04	0.04	0.04
Range		83.99	−1.07	162.72	0.19	2289.44	7.10
Minimum		0.08	0.08	0.08	0.08	0.08	0.08
Maximum		−0.28	−0.27	−4.50	6.21	−5.71	−0.76
Percentiles	25	0.41	0.20	0.01	6.80	0.91	7.04
	50	0.72	0.46	0.05	7.26	1.47	7.72
	75	1.01	0.67	0.11	7.82	2.82	8.27

^a Multiple modes exist. The smallest value is shown. Note: debt ratio (DERA), tangibility (TANG), return on assets (ROA), size of total assets (SZTA), liquidity (LIQU), and size of total sales (SZTS).

shows that the average DERA of transport companies in V4 is 77%; this average does not differ much from the median of 72%. However, 25% of companies have a DERA lower than 41%. On the other hand, 25% of enterprises have a DERA of at least 101%. The average TANG in transport companies is 45%, and the median is 46%. Descriptive statistics show that 25% of transport companies have at least 67% of assets in tangible assets. The average ROA is 6%, and the median is only 5%. However, one-quarter of businesses have an ROA of less than 1%, and the minimum is less than (-) 4%. The average and median LIQU are significantly different from other indicators, as the average LIQU is more than 7.4, but the median LIQU is less than 1.47. In other words, LIQU differs between transport companies. Firm size by the natural logarithm of total assets or sales shows that the averages reach relatively similar values of around 7 ± 1. Table 4 reveals extra data on selected financial variables.

Table 5. Correlation matrix for the total sample.

	DERA		TANG		ROA		SZTA		LIQU		SZTS
DERA	1		−0.019		−0.295	**	−0.090	**	−0.131	**	−0.043	*
TANG			1		−0.142	**	0.266	**	−0.032		−0.240	**
ROA					1		0.011		0.079	**	0.140	**
SZTA							1		0.007		0.350	**
LIQU									1		−0.035	*
SZTS											1

* Correlation is significant at the 0.01 level (2-tailed). ** Correlation is significant at the 0.05 level (2-tailed).

demonstrates the correlation is weak for all pairs. The strongest correlation is between SZTA and SZTS (0.350). The results show that 11 out of 15 pairs are statistically significant, e.g., there is no correlation between DERA and TANG, ROA and SZTA, TANG and LIQU, or SZTA and LIQU. Moreover, there are nine weak indirect correlations between variables. The results show that DERA will decrease if ROA, SZTA, LIQU, or SZTS increase. Finally, the correlation table indicates that, in general, these indicators are appropriately selected for modelling the dependence between the DERA and other indicators using multiple linear regression.

Methods. First, we did not use one-way ANOVA because normality and homogeneity of variance as one of the key assumptions for application are not met. If normality was observed, we would use Welch statistics. However, these assumptions are not met; the Kruskal-Wallis test is used to identify the difference in debt ratio depending on the country. These results will point out the need to create one universal model or model for each country. Then, we used multiple linear regression to identify the relationship between the debt ratio and independent variables. This method is a supervised machine learning algorithm containing a broader data spectrum. The purpose of multiple regression is to predict dependent variables based on potential input variables. The main output is a formula explaining the factor impact on the dependent variable. Multiple linear regression requires assumptions such as homogeneity of variance (homoscedasticity), no multicollinearity, independence of observations, normality, and the linear relationship between dependent and independent variables. The model selects statistically significant variables using backward elimination, forward selection, or stepwise selection. We apply stepwise selection. This method combines forward and backward processes. The stepwise method is a procedure that attempts to find the best multiple linear regression model including statistically significant variables from a more extensive set of potential variable datasets.

$$y={\beta }_0+{\beta }_1{X}_1+\cdots +{\beta }_n{X}_n+\epsilon$$

(1)

where:

• $y$—dependent variable,
• ${\beta }_0$—intercept,
• ${\beta }_1$—the regression coefficient of the first independent variable,
• $X_1$—the first independent variable,
• ${\beta }_n$— the regression coefficient of the last independent variable,
• $X_n$—last independent variable,
• $n$—the number of predictor variables,
• $\epsilon$—model error.

Outliers. Mahalanobis distance is an effective metric to find the distance between a point and a distribution. We remove all outliers for each input variable by calculating outlier bounds or identifying outliers using boxplots for each variable. After removing these outliers, we apply the Mahalanobis distance to determine the multivariate distance. These final samples for each country and region are used to model the dependent variable using multiple linear regression.

$$D^2={\left(x-m\right)}^T{C}^{-1}\left(x-m\right)$$

(2)

where:

• $D^2$—Mahalanobis distance,
• $x$—vector or data,
• $m$—vector of the mean of independent variables,
• $C^{-1}$—the inverse covariance matrix of independent variables,
• $T$—indicates the vector should be transposed.

Multicollinearity. First, we detect multicollinearity using VIF. Multicollinearity is a typical phenomenon when an independent variable is highly correlated with one or more other independent variables in a regression model. This phenomenon is undesirable because the possible results are distorted. If the VIF exceeds 10, the independent variables are highly correlated. This metric is used in creating an initial regression model to determine the appropriate combination of independent variables without mutual correlation with each other.

$$VIF=\frac{1}{1-R_i^2}=\frac{1}{tolerance}$$

(3)

where:

• $VIF$—variation inflation factor,
• $R_i^2$—unadjusted coefficient of determination for regressing the ith independent.

However, we identify potentially hidden multicollinearity using the Condition Index. If multicollinearity was detected, we gradually remove the variables contributing to a high Condition Index. If this metric exceeds 15, the result indicates high multicollinearity.

$$C{I}_i=\sqrt{\frac{{\lambda }_{max}}{{\lambda }_i}}$$

(4)

where:

• CI—Condition index,
• $\lambda$—Eigenvalue.

Durbin-Watson statistics assumes values from 0 to 4. Autocorrelation refers to the correlation between the values of variables across different data sets. If the values approach 0, we detect positive autocorrelation. On the other hand, if the values are close to 4, the result indicates a negative autocorrelation. Finally, if the values are around 2, we do not detect autocorrelation.

$$DW=\frac{{{\displaystyle \sum }}_{t=2}^T{\left(e_t-e_{t-1}\right)}^2}{{{\displaystyle \sum }}_{t=1}^t{e}_t^2}$$

(5)

where:

• $e_t$—residual figure,
• $T$—number of observations of the experiment.

Coefficient of determination explains the proportion of variation in the dependent variable explained by multiple linear regression. Adjusted R-squared is a modified version of R-squared with an adjustment for the number of predictors in the model. We use adj. R-square for comparison with other models.

$$R^2=1-\frac{SSR}{SST}$$

(6)

$$adj. R^2=1-\left(\frac{n-1}{n-\left(p+1\right)}\right)\left(1-R^2\right)$$

(7)

where:

• SSR—sum of squares regression,
• SST—total sum of squares,
• $R^2$—sample R-squared,
• $n$—the sample size,
• $p$—number of the independent variable.

4. Results

Table 6. Hypothesis Test Summary—Kruskal-Wallis Test.

Null Hypothesis	N	Test	Df.	Sig.
The distribution of TL/TA is the same across categories of country.	3.299	107.987	3	0.000

Asymptotic significances are displayed. The significance level is 0.05.

shows that the debt ratio statistically significantly differs depending on the selected country of the Visegrad Group using the non-parametric Kruskal-Wallis test. Pairwise comparisons demonstrate that the debt ratio differs in four out of six pairs (see Figure 1). As can be seen, the debt ratio does not statistically significantly differ between Polish and Czech companies, as well as between Czech and Hungarian companies using the Bonferroni correction. These results indicate that a multiple regression model for modelling the debt ratio must be created for all member countries of the Central European group.

Table 7. Multiple linear regression models.

	TOT	POT	CZ	HU	PL	SK	V4
intercept			1.267	1.680	1.252	1.468	1.380
TANG	+	−	−0.444	−0.783	−0.393	−0.311	−0.420
ROA	+	−	−0.817	−0.466	−0.255	−0.764	−0.465
SZTA	+	−
LIQU	+	−	−0.129	−0.246	−0.237	−0.287	−0.223
SZTS	+	−
Sample size			436	918	778	491	2686
Outliers based on Mahalanobis distance			2	0	0	8	23
F			115.532	453.898	272.916	133.291	755.184
ANOVA			0.000	0.000	0.000	0.000	0.000
Condition Index			5.244	8.109	5.996	7.627	6.489
Excluded variables			SZTA SZTS	SZTA	−	−	SZTA SZTS
Durbin-Watson statistics			1.896	1.920	1.850	1.874	1.881
R			0.667	0.773	0.717	0.671	0.677
R-square			0.445	0.598	0.514	0.451	0.458
adj. R-square			0.441	0.597	0.512	0.447	0.457
Stepwise method			3	3	3	3	3

Note: Trade-off theory (TOT), pecking order theory (POT), Czech Republic (CZ), Hungary (HU), Poland (PL), Slovakia (SK), tangibility (TANG), return on assets (ROA), size of total assets (SZTA), liquidity (LIQU), size of total sales (SZTS). The dependent variable is the debt ratio (DERA).

compares five models for selected countries and regions, and also shows the basic characteristics of the models, such as the size of the final sample, and the number of outliers; these outliers were detected using the Mahalanobis distance. In addition, we verify multicollinearity using Condition Index and autocorrelation using Durbin-Watson statistics. Finally, we compare the models using adj. R-square. All these models were created using a multi-step stepwise method. This method selects statistically significant variables. Table 7 presents five models estimating corporate performance in Central European countries. All these models are composed of statistically significant indicators such as tangibility (TANG), return on assets (ROA), and liquidity (LIQU) in contrast to company size as the natural logarithm of total assets or total sales. All these variables were identified using a stepwise method consisting of three steps. We find that TANG, ROA, or LIQU have a positive impact on reducing the DERA because if the company increases the TANG, ROA, or LIQU, the DERA will decrease. As can be seen, firstly, TANG has the greatest impact on reducing DERA in the model for Hungarian transport companies in contrast to the other models, as the coefficient for TANG is (−) 0.783. Second, ROA has the greatest impact on reducing the DERA in the model for Czech companies in the transport sector, because the coefficient is (−) 0.817, unlike the others. On the other hand, the lowest coefficient is part of the model for Polish companies. Thirdly, LIQU plays the most prominent role in the model for Slovak transport companies in contrast to the model for Czech transport companies. In the three output models for the Czech and Hungarian countries, similar to the regional model, we found a high Condition Index (above 15). In other words, this index has been reduced by removing selected indicators as the natural logarithm of total assets or total sales. All these models have a Condition Index lower than 15, these results demonstrate that multicollinearity is not present, and the input variables are suitable for modelling corporate performance in the transport industry in all countries. In addition, the Durbin-Watson statistic reaches approximately 2, indicating non-autocorrelation. Finally, the results show that all these models explain more than 44% of the variability of the dependent variable using a statistically significant model (see ANOVA). The model for Hungarian transport companies explains almost 60% of the variability of the dependent variable. This model is the best compared to others. On the other hand, the model for the Czech transport sector explains less than 45% of the variability of the dependent variable.

5. Discussion

Table 8. Hypothesis summary.

Acronym	Variable	Formula	TOT	POT	Our Expectations	CZ	HU	PL	SK	V4
TANG	tangibility	FA/TA	+	−	+	−	−	−	−	−
ROA	return on assets	EBIT/TA	+	−	−	−	−	−	−	−
SZTA	company size	ln of TA	+	−	+
LIQU	liquidity	CA/CL	+	−	−	−	−	−	−	−
SZTS	company size	ln of TS	+	−	+

summarizes the results of the hypothesis testing. We find that three of the five hypotheses by country are statistically significant, and the only two indicators showing the size of the company have no effect on the capital structure. On the other hand, all statistically significant variables have a positive impact on the debt ratio expressing the capital structure, because the regression coefficients are negative. In other words, higher tangibility, profitability, and liquidity lowers the debt ratio. These results correspond with the pecking order theory. However, our expectations assumed that profitability and liquidity indicators would have a positive impact on debt reduction, except for tangibility. We expected that higher tangibility motivates obtaining new financial resources from banks for investment and expansion.

Table 8 describes several models for capital structure expressed as debt ratio (DERA, calculated by total liabilities to total assets). We summarize these models using several features such as time frame, sample size, country, industry, regression coefficients, statistically significant indicators with p-value, calculation of the significant indicator, the method used, R-square, or adj. R-square. Research on capital structure is relatively limited in the Central European area, especially for a specific industry. Several models include indicators such as return on assets (ROA), tangibility (TANG), current ratio (LIQU), or company size (SZTS). Other indicators are specific to the selected model.

Moreover,

Table 9. The comparison of our research with previous research.

Authors	Data	Sample	Country	Sector	Coefficients	p-Value	Variable	Method	R-Square	adj. R-Square
Jedrzejczak-Gas (2018) [26]	2009–2016	55	PL	transport spediction logistics	0.452	*	intercept	MLR	0.83	0.80
					−0.581	***	TANG = FA/TA
					0.022	*	SZTS = ln of TS
					−0.075	***	LIQU = CA/CL
					−0.064	*	ETR = tax/gross profit
Enjolras et al. (2021) [27]	2009–2018	>4000	PL	agriculture	−0.045		intercept	GMM	n/a	n/a
					0.720	*	DERA = TLt−1/TAt−1
					−0.198	***	LAND = land/TA
					−0.136	***	ROA = EBIT/TA
					0.043	*	assets = TA
					0.328	***	growth = TI/TA
					−0.215	*	farm orientation (horticulture)
Krištofík and Medzihorský (2022) [28]	2016–2019	n/a	SK	wood	0.522	*	intercept	REM	n/a	n/a
					−0.508	***	ROA = EBIT/TA
					−0.004	***	growth = % annual growth of TA
					−0.281	**	cash ratio = CCE/TA
	2020				1.525	***	intercept	REM	n/a	n/a
					−0.055	**	SZTS = ln of TS
					−0.740	***	cash ratio = CCE/TA
Our research	2019	436	CZ	transport	1.267	***	intercept	MLR	0.45	0.41
					−0.444	***	TANG = FA/TA
					−0.817	***	ROA = EBIT/TA
					−0.129	***	LIQU = CA/CL
		918	HU		1.680	***	intercept		0.60	0.60
					−0.783	***	TANG = FA/TA
					−0.466	***	ROA = EBIT/TA
					−0.246	***	LIQU = CA/CL
		778	PL		1.252	***	intercept		0.72	0.51
					−0.393	***	TANG = FA/TA
					−0.255	***	ROA = EBIT/TA
					−0.237	***	LIQU = CA/CL
		491	SK		1.468	***	intercept		0.67	0.45
					−0.311	***	TANG = FA/TA
					−0.764	***	ROA = EBIT/TA
					−0.287	***	LIQU = CA/CL
		2686	V4		1.380	***	intercept		0.46	0.46
					−0.420	***	TANG = FA/TA
					−0.465	***	ROA = EBIT/TA
					−0.223	***	LIQU = CA/CL

Note: Czech Republic (CZ), Hungary (HU), Poland (PL), Slovakia (SK), Visegrad Group (V4), tangibility (TANG), fixed assets (FA), total assets (TA), natural logarithm of total sales (SZTS), total sales (TS), liquidity (LIQU), current assets (CA), current liabilities (CL), effective tax rate (ETR), debt ratio (DERA), total liabilities (TL), return on assets (ROA), total investments (TI), earnings before interest and tax (EBIT), cash and cash equivalents (CCE). Note: p < 0.10 (*), p < 0.05 (**), p < 0.01 (***).

displays our five models for the transport sector depending on the country or region with other models describing the debt ratio in Central European countries such as Poland and Slovakia. In addition, these models are created for other industries such as agriculture and the wood processing industry; only one of the three models by Jedrzejczak-Gas (2018) [26] is created for a similar industry, such as transport and logistics. However, this model is based on a smaller sample of transport companies, unlike our models. This sample consists of less than 60 companies. The results demonstrate that indicators such as TANG, ROA, or LIQU are statistically significant indicators for capital structure modelling. All these variables have a positive impact on reducing DERA. Our results correspond to empirical findings from Jedrzejczak-Gas (2018) [26], as TANG has a similarly negative coefficient. In addition, the corporate performance also reduces DERA, because Enjolras et al. (2021) [27] and Krištofík and Medzihorský (2022) [28] found that increasing corporate performance helps the self-financing of the company. However, this indicator has a different impact depending on the sector because profitability plays a more significant role, unlike the agricultural sector. Furthermore, we find that the current ratio also has a greater impact on reducing the debt ratio, as the coefficient of liquidity is (−) 0.075 according to Jedrzejczak-Gas (2018) [26], unlike our output models. Finally, the results indicate that company size calculated as the natural logarithm of total sales has no statistically significant impact on the capital structure in our models, unlike Jedrzejczak-Gas (2018) [26] and Krištofík and Medzihorský (2022) [28]. Even this indicator has a different impact on both presented models for the transport or wood processing industries. This indicator is even removed in three of the five models to reduce multicollinearity between the input variables based on the Condition Index.

6. Conclusions

We identified three statistically significant input variables, such as tangibility, return on assets, and current ratio in contrast to company size as the natural logarithm of total assets or total sales. These variables do not have a statistically significant impact on the capital structure as a proportion of total liabilities to total assets. These findings demonstrate that the primary objective of the paper has been met, as the research offers five regression models estimating the capital structure expressed by the debt ratio using multiple regression analysis. The results show that transport companies prefer the pecking order theory. In other words, three out of five financial indicators predict that increases in tangibility, return on assets, and liquidity reduces debt levels. All these models explain 44–60% of the variability. The model for Polish transport companies achieves the highest adj. R-square, unlike Czech companies.

This paper contributes to the development of current theoretical and empirical knowledge for the transport industry in the Visegrad Group. This research helps to identify three key indicators such as tangibility, profitability, and current ratio reducing the debt ratio in transport companies. The results show that profitability has the greatest impact on debt reduction in the Czech Republic and Slovakia compared to other countries. These results can lead to the comparison and identification of key differences between small- and medium-sized enterprises, which form the key pillar of the Central European economy, compared to large and very large transport enterprises.

The research helps investors, bankers, and other stakeholders in identifying weaknesses as potential failures. One of the benefits is helping to improve awareness of the key determinants in the management of the capital structure in small- and medium-sized enterprises to ensure effective financing. These results can help SMEs to cooperate with banks ahead of time, avoid potential financial losses, as well as build a good reputation with the public.

Limitations. One of the restrictions is limited access to data describing the financial situation of transport companies in the Visegrad Group. This limitation did not allow for expanding the final group of independent variables with selected indicators, such as non-debt tax shield (calculate by depreciation to total assets) by Matemilola et al. (2013) [10], Wellalage and Locke (2013) [5], Daskalakis et al. (2017) [3], and Jaworski and Czerwonka (2021) [1]. These results can contribute to a deeper analysis and more detailed findings about the transport sector in V4. Second, another limitation is the limited data; the study draws data from the database for the year 2019. Third, the sample does not consist of large and very large enterprises. On the other hand, small- and medium-sized companies are present, the majority of all enterprises in central Europe, but large and very large enterprises may differ from them. Finally, the results do not offer general conclusions for all industries. These results are particularly suitable for European transport companies in central Europe. In other words, we cannot generalize these results to all sectors.

Future research. This paper contributes to the development of theoretical and empirical knowledge about capital structure. First, future research can test the hypotheses on a wider sample of SMEs. The initial sample can be expanded with detailed information obtained from additional data obtained through the survey. Second, we can analyse the differences between the selected determinants and short-term or long-term debt [18]. Third, we can compare these findings with new results for large and very large firms to identify differences depending on firm size. This analysis arouses interest in capital structure as one of the important pillars of corporate finance. Finally, we can investigate the impact of qualitative variables such as company size according to the classification of companies into small- and medium-sized companies, legal forms such as private and public transport companies, and the gender of the director on the capital structure. All these variables can be potential indicators of an increase in the overall variability of the dependent variable.