Analysis of financial performance determinants: evidence from Slovak agricultural companies

Source: compelled by authors.

Year-wise means of selected dependent variables (ROA, NPM, ROE, ROS) for the period 2015-2019 for Slovak agricultural companies are presented in Figure 3.

Figure 3. The trend of ROA, NPM, ROE, and ROS during 2015-2019

Source: own elaboration.

Figure 3 shows the trends in changes of the dependent variables ROA, NPM, ROE and ROS over time during the period 2015-2019. The results of analysis enabled to establish that ROE and ROS had the same behavior during 2015-2019, which is characterized by a periodic change in the growth of the average values of indicators observed in 2016 and 2018, and their decline (respectively, 2017 and 2019). The exactly opposite trend, if compared to ROE and ROS, is found for NPM, where average values decrease in 2016 and 2018 and increase in 2017 and 2019. As seen from the Figure 3, the average values of ROA had a constant general downward trend from 2015 to 2019.

Research models. To analyze the financial performance determinants of the Slovak agricultural companies this study examined four following models (Models 1-4):

Model 1: ROAit = a +Lev it + в2 LTDtoA it+ @3 STDtoA it + в4 DtoE it + в5 CapInt it + в 6

AsTangit + в 7 CurRatit + в 8 lSize it + в 9 DVTO it + в 10 DVLFit + at;

Model 2: NPMit = a + в1 Lev it + в2 LTDtoA it+ в3 STDtoA it + в4 DtoE it + в5 CapInt it + в 6 AsTang it + в 7 CurRat it + в 8 lSize it + в 9 DVTO it + в 10 DVLF it + at;

Model 3: ROEit = a + в1 Lev it + в2 LTDtoA it+ в3 STDtoA it + в4DtoE it +в5CapInt it+в 6

AsTang it + в 7 CurRat it + в 8 lSize it + в 9 DVTO it + в 10 DVLF it + at;

Model 4: ROS it = a + в1 Lev it + в2 LTDtoA it+ в3 STDtoA it + в4DtoE it +в5CapInt it+в 6

AsTang it + в 7 CurRat it + в 8 lSize it + в 9 DVTO it + в 10 DVLF it + at,

where ROA, NPM, ROE, ROS - dependent variables, i = entity and t = time;

a - identifier;

p - variance introduced by the unit-specific effect for unit i;

в - regression coefficient;

Lev, LTDtoA, STDtoA, DtoE, CapInt, AsTang, CurRat, l Size, DVTO, DVLF - independent variables, i = entity and t = time;

s_it - error term.

Results and discussion

Descriptive statistics and correlations. Descriptive statistics analysis is shown in Table 3.

Table 3 Descriptive statistics (based on observations 1:1 - 527:5)

Assumption test results. Aiming at the adequacy test of the suggested in this study Models 1 -4 to the collected panel data of agricultural companies three classical assumption tests were used, i.e., Normality, Autocorrelation and Heteroscedasticity. Normality test for Models 1 -4 made it possible to determine non-normality of the distribution of errors. For example, for Model 1 for chi-square (2) = 5534.534 p-value = 0, which is less than 0.05, and therefore, the residuals are normally distributed. Applying the Wooldridge test for autocorrelation detected its presence for Models 1 and 4, since the p-values obtained are greater than 0.05 (0.51478; 0.101421), at the same there is its absence for Models 2 -3, since the p-values are less than 0.05 (0.038806 and 0.0313246, respectively). Since FEM was selected as estimate parameter for all Models 1-4, the Wald test was applied to check the level of heteroscedasticity. As a result, the existence of heteroscedasticity for Models 1 -4 was confirmed, as in all models for the corresponding chi-square (527) p-value = 0 was obtained, in its turn, indicating the presence of heterogeneity of observations and different dispersion random error for all Models 1-4.

For the purpose of eliminating the problems of non-normality of error distribution, partial autocorrelation (in relation to Model 1 and Model 4) and heteroscedasticity, we propose to apply robust standard errors. To be precise, since we do not know the structure of heteroscedasticity, for large sample size it is advisable to use robust standard errors, which will allow to eliminate the negative impact of outliers. A similar approach is also used in the studies of Ocak & Findik (2019), Ozkan (2022), Serpeninova et al. (2022).

Panel data regression results. Table 4 presents the results of the regression analysis (p-value and level of significance) carried out using FEM. The results show, which of the model regressors significantly influence on the dependent variables, and what is the impact. In table 4, the independent variables DVTO and DVLF proposed for Models 1-4 are missing, as they were removed from their composition due to exact collinearity while evaluating the models using FEM.

Table 4

Model 1-4 (ROA, NPM, ROE, ROS). FEM (Robust standard errors), using the observations 1-2635

Note: *Significant at the 10 % level; **Significant at the 5 % level; ***Significant at the 1 %level.

Source: calculated via Gretl software package.

Model 1 (ROA) can be interpreted through the following equation:

у = 5.81318 - 0.0585255x1 - 0.112433x2 + 0.0781087x3 - 0.000451223x4 -

- 2.80577e-06x5 - 0.0414539x₆ - 0.00278093x₇ - 0.360796x8,(1)

where ? - ROA;

We can conclude that the regressors such as const, CurRat and l_Size є statistically significant (marked with * in the Table 4). Herewith, the regressors - const and l_Size - have the highest level of significance at the 1 % level, hence significantly influencing on ROA. Accordingly, we can state that ROA of agricultural companies is influenced by their size. And because this impact is an inverse one, businesses with lower size have higher ROA values. For example, if l_Size is increased by 1, then ROA will decrease by 0.360796. Based on the obtained value of LSDV R-squared for Model 1, it was established that almost 76 % of the variation of ROA can be explained by the variation of the independent variables. The largest share in the total variation among all variables is l_Size, which explains 64.64 % of the variation of ROA and LTDtoA, which explains 5.23 %. Other independent variables explain less than 5 % of the variation of ROA. The value of R-squared shows Model 1 accounts for 8.51 % of the variance which are within the panel units.

Model 2 (NPM) can be interpreted through the following equation:

? = - 2.68633 - 0.335790x1 + 0.0396243x2 + 0.0215960x3 + 0.00145156x₄ -- 0.00467558x5 + 0.350433x₆ - 0.000867961x₇ + 0.189368x8,(2)

where: ? - NPM;

x1 - x8 - the same as in Model 1.

Based on the Table 4, only CapInt has a significant influence on NPM, with a significance level at the 1 % level. As CapInt increases by 1, NPM values decrease by

0. 00467558, indicating the presence of an inverse effect. The LSDV R-squared value shows that 62.93 % of the variation of the NPM can be explained by the variation of regressors. The largest share in the total variation among all variables is l_Size, which explains 33.16% of the variation of NPM and LTDtoA, which explains 11.11 %. Other independent variables explain less than 10 % of the variation of NPM.

Model 3 (ROE) can be demonstrated through the following equation: у = 25.9517 - 0.874200x1 - 0.0157958x2 + 1.65217x3 + 0.378300x₄ - 0.00183331x5 -

- 1.12827x6 - 0.0213116x7 - 1.67891x8,(3)

where ? - ROE;

x1 - x8 - the same as in Model 1.

The dependent variable ROE is significantly influenced by such regressors as const, DtoE, CapInt, CurRat and l_Size, all of which, except for CurRat, affect the dependent variable with the highest level of significance - at the 1 % level. CapInt, CurRat and l_Size inversely affect ROE. As shown, like in Model 1, an increase in the volume of agricultural companies negatively affects the value of NPM. The LSDV R-squared value shows that 68 % of the variation of the ROE can be explained by the variation of regressors. The largest share in the total variation among all variables is DtoE, which explains 23.02 % of the variation of ROE, l_Size - 17.16 %, STDtoA - 17.03 %, AsTang - 7.34 %. Other independent variables explain less than 5 % of the variation of ROE. The obtained value of R-squared, similar in Model 2, is quite high compared to Model 1, since as much as 49.6 % of the variance within Model 3 panel units account for.

The analysis of Model 4 by applying FEM revealed the absence of regressors that could significantly affect ROS.

The panel data regression analysis results both confirm and deny the conclusions of the previous studies on financial performance determinants of agricultural companies. This proves to be the case for all the studied dependent variables used for the Models in this paper (ROA, NPM, ROE, ROS). This confirms the hypothesis about the existence of a unique influence of financial performance determinants due to the presence of conditions and features specific for the Slovak agricultural sector.

According to the results, the general (Lev) and partial indicators (LTDtoA, STDtoA) characterizing the ratio of debts to assets of agricultural companies do not have a significant impact on all four used indicators characterizing financial performance. These findings contradict the results of studies stated by Liu et al. (2020), Xu et al., (2021), Vukovic et al. (2022), where on the contrary a significant positive effect on such indicators was found. It can be concluded that search for the optimal capital structure should not be considered as an important task for management from the standpoint of improving the financial performance of the Slovak agricultural companies.

In the previous papers, the authors emphasize the decisive role of intellectual capital and intangible assets in ensuring the financial performance of companies (Xu & Zhang, 2021; Vukovic et al., 2022; Lehenchuk et al., 2022; Zavalii et al., 2022), claiming a positive significant relationship between the indicators. However, the results obtained in our study indicate that AsTang does not have a significant impact on the independent variables of Models 1-4, which proves the lack of significant impact of intangible objects on the financial performance of the Slovak agriculture companies.

There is also a discordancy between the results of our study and those views of some authors (Burja & Burja, 2017; Vukovic et al., 2022), who established the presence of positive significant impact of CurRat on financial performance. The obtained results confirmed the significant impact of CurRat only on ROA and ROE, however, this impact was negative. For this reason, improving the liquidity of Slovak agriculture companies will not lead to an improvement, but on the contrary, to a worsening of its financial performance. In relation to the size of the enterprise, a similar effect on the financial performance of Slovak agriculture companies was found as well. Odalo et al. (2016), Pokharel et al. (2020), Liu et al., (2020), Vukovic et al. (2022) determined the positive significant impact of the agricultural companies' size on financial performance, however, we can observe that l_Size influence negatively on ROA and ROE. The above conclusion is meaningful for critical assessment of state policy in the agrarian sphere in the Slovak Republic. To be more specific, it questions the recommendations of the compilers of the Report on Agriculture and Food Sector in the Slovak Republic for 2020 (Report ..., 2021). The document sets up a need to build vertical structures and support land consolidation in the Slovak Republic, which will certainly lead to an increase in size of agricultural enterprises and will have a negative impact on their financial performance as can be concluded from our findings.

Conclusions

The purpose of this study was to examine financial performance determinants of 527 Slovak agricultural companies over the period 2015-2019. In order to objectify the results and test the research hypotheses, panel data regression analysis was applied in the study. Such research is especially relevant in the context of the need to enhance the quality of financial management as well as improve the financial performance of the Slovak agriculture companies.

The study was based on the use of regression analysis to test four models presenting the relationships between financial performance measures and its determinants (regressors). Return on Assets, Net Profit Margin, Return on Equity and Return on Sales were selected as dependent variables that characterize financial performance. Ten independent variables, i.e., Leverage, Long-Term Debt to Assets, Short-Term Debt to Assets, Debt to Equity, Capital intensity, Tangibility of Assets, Current Ratio, Size, Dummy variable for type of ownership, Dummy variable for legal form, were chosen to explain each of four financial performance measures as determined.

Applying F-statistics test, Breusch-Pagan test and Hausman test the feasibility of FEM as estimate panel data parameter for Models 1 -4 has been determined. Having found the incomplete adequacy for Models 1-4 to the generated data assumed from the Normality test, Autocorrelation test and Heteroscedasticity test via GRETL software package, the necessity of applying robust standard errors has been substantiated to eliminate negative impact of outliers.

The obtained results of panel regression analysis showed that the various determinants had significant impact on the different financial performance measures. None of the determinants had a significant effect simultaneously for all formulated Models (1-4). Analysis of the Model 4 did not reveal any significant factors influencing ROS. Such results partially contradict the conclusions of the previous studies and confirm the proposed hypothesis about the existence of a unique influence of financial performance determinants due to the presence of specific conditions and features inherent in the Slovak agricultural sector.

As the study showed, l_Size has negative significant impact on ROA and ROE at the 1 % level ROA and ROE of Slovak agricultural companies. Therefore, an increase in the size of such enterprises will lead to a decrease in the financial performance measures. CapInt also has a negative significant impact on NPM and ROE. Accordingly, the greater the capital intensity of an agricultural enterprise gets,

i. e., the ability to provide the appropriate level of income due to availability of a certain amount of assets, the smaller the related financial performance measures are. In consequence, managers of agricultural enterprises should contribute to the development of “non-capital-intensive” for agricultural companies to enhance the values of NPM and ROE.

Another finding of the paper is that DtoE has a significant positive impact on ROE. This means that the greater the ability of agricultural companies to cover all debts with its shareholder equity, the more it improves ROE. As follows, managers of agricultural companies should be more cautious about attracting additional credit funds for the development of the business activities as it will lead to a negative impact on ROE values. Accordingly, it is crucial to find other ways of financing company's activities either at the expense of investors' funds or subsidies from the state or European financial institutions.

Despite the recent development of the knowledge economy and active use of modern technologies in the agricultural sector, the results of the study related to the role of AsTang in ensuring financial performance showed that intangible assets are not key for the Slovak agricultural companies. Another reason justifying such results may lay in the high level of conservatism within the current accounting system framework regarding the recognition of intangible assets. As a result, a significant part of technological innovations of agricultural enterprises remains “outside” of financial reporting, not being an object for analysis by scientists and, logically, distorting the results of the empirical research.

The use of dummy variables (DVTO, DVLF) did not provide for detecting the influence of qualitative variables, i.e., type of ownership, legal form on financial performance of Slovak agricultural companies. Due to the presence of exact collinearity, the dummy variables (DVTO, DVLF) were excluded from the regressors of Models 1-4.

In general terms, the results of the study can serve as an empiric framework for making recommendations on how to enhance the financial performance of Slovak agricultural companies, to ensure the quality and competitiveness of agricultural production on the European market and domestically.

The study has three limitations to be considered for the objective evaluation of its findings. The first limitation is the sampling period, which covers only 5 years, from 2015 to 2019. However, to obtain more accurate results of panel data regression analysis it is possible to expand the sampling period. The second limitation is a list of independent variables, restricted by financial reporting structure of the Slovak agricultural companies. However, the potentially useful information can be gathered by conducting additional observations, as well as supplemented if requested by the specific independent variables, proven to be the case of ROS, where all Model 4 regressors didn't have any significant impact. The third limitation concerns sample size, as not all the Slovak agricultural companies were included in the study. This situation was caused by companies not disclosing all the necessary data for the period 2015-2019 or some companies had a limited period of functioning (were established after 2015 or liquidated during of the period).

Prospects for further research are conducting comparative studies of the influence of different groups of factors on the financial performance of agrarian companies in different countries. In particular, for countries that are at the same phase of agricultural market development and have similar institutional conditions for the functioning of enterprises, which will allow conducting such comparative studies.

References

1. Ababiya, A. (2018). Financial performance of agricultural enterprises and their determinant factors in Hadiya Zone, Ethiopia. Journal of Poverty, Investment andDevelopment,46,37-46.Availableat:

https://iiste.org/Journals/index.php/JPID/article/view/43881.

2. Bulavinova, N., Burdenko, I., Lehenchuk, S., Tsaruk, I., & Ostapchuk, T. (2021). Trends in research of responsible investment in the context of sustainable development: bibliometric analysis. Agricultural and Resource Economics, 7(3), 179-199. https://doi.org/10.51599/are.2021.07.03.11.

3. Burja, V., & Burja, C. (2017). Financial performance and the business risk in

agricultural sector of Romania. Annals of the “Constantin Brancusi" University of Targu Jiu. EconomySeries,3,75-82. Available at:

https://www.utgjiu.ro/revista/ec/pdf/2017- 03/09_BURJA%20VASILE,%20BURJA%20CAMELIA.pdf.

4. Hampl, F. (2020). A statistical analysis of the financial performance of organic and conventional farms in the Czech Republic with respect to their size. Agricultural Economics - Czech, 66, 1-9. https://doi.org/10.17221/78/2019- AGRICECON.

5. Kravcakova Vozarova, I., Kotulic, R., & Vavrek, R. (2019). Disparities in the

financial performance of agricultural entities according to the legal form: the case of Slovakia.AppliedEconomics,51(56),5999-6008.

https://doi.org/10.1080/00036846.2019.1645285.

6. Lehenchuk, S., Tumpach, M., Vyhivska, I., Makarovych, V., & Laichuk, S.

(2022). The impact of innovation on the profitability of Slovak pharmaceutical companies. Marketing and Management of Innovations,2,184-296.

https://doi.org/10.21272/mmi.2022.2-25.

7. Liu, L., Xu, J., & Shang, Y. (2020). Determining factors of financial

performance of agricultural listed companies in China. Custos e Agronegocio, 16(4), 297-314.Availableat:

http://www.custoseagronegocioonline.com.br/numero4v16/OK%2012%20financial.pdf.