PLSR and MIMIC-PLS Model Approach to Determinants of Capital Structure Choices and Comparative Analysis


Başar Yılmaz S., Çelik H. E.

Yüzüncü Yıl Üniversitesi Sosyal Bilimler Enstitüsü Dergisi, sa.58, ss.156-169, 2022 (Hakemli Dergi)

Özet

In this study, the determinants of capital structure choices, which are very important for companies. For this reason, the data of 167 companies traded in ISE and BIST between 2008 and 2017 were collected. 11 variables as 8 indicators, and 3 causes, which are thought to have an impact on the capital structure, were determined. In the study, 2 analyses were made and the results were compared. In the first analysis, 3 models were established by changing the dependent variable and analysed with the Partial Least Squares Regression (PLSR) method. Afterwards, the MIMIC Model, which can provide simultaneous solution to these 3 models, is again analysed with Partial Least Squares (PLS). A Comparison of the results of the two analyses is given. According to the results of the analysis, it has been seen that the most effective indicator variables on the capital structure in all models are profitability and liquidity. Among the reason variables, which are thought to represent the capital structure best in the model, total debt is the variable that best represents the capital structure, which is included as a latent variable. Again, according to the MIMIC model result, it has been seen that the effective theory for the companies traded in the BIST in Turkey is the trade-off theory. Finally, when the PLSR and MIMIC models are compared, it can be said that the MIMIC model is more preferable in terms of both examining the relationships in more detail and the compatibility of the results with the theories.