Formation of a statistical model of the capital structure

Recent studies on capital structure optimisation show that the significant factors and the degree of their influence on the choice of financing sources vary across companies in different industries. At the same time, the specific type of dependence of the equity multiplier Z on key determinants (ROE, ROS, Y, Ib, Tb, etc.) remains, as a rule, unclear. The article deals with the construction of a regression model that allows predicting the capital structure with a given level of profitability, taking into account the impact of return on sales, resource productivity, cost of debt and taxes on the ratio of financing sources for companies in certain sectors of Russian industry. Methodologically, the research rests on the regression analysis. The evidence base is the data coming from Russian confectionery enterprises and construction companies observed over the last five years (2017–2021). According to the findings, we have proposed a multiplicative five-factor multiple regression model, which makes it possible to conduct a comparative analysis of the degree of influence of the above-mentioned normalised determinants on the Z multiplier. The normalised value of return on sales (ROS) has the greatest impact on the capital structure, then resource productivity and cost of debt follow in terms of the degree of influence, and the normalised value of the tax burden has the least impact. Testing the regression model for forecasting capital structure using the example of a confectionery company makes it possible to find the capital structure of a particular company with a given ROE value. The paper substantiates the quality of constructed capital structure models from different perspectives.