Conditioned tests of the predictions of the trade-off and pecking order theories of capital structure: a UK study.

Empirical studies testing the predictions of the trade-off and pecking order theories have invariably produced inconsistent and contrary results when employing a multiple linear regression model. These erratic and non-inferential results are likely due to the failure to account for the conditionality of the factors that affect firms’ financing decisions, and also the implementation of methodologies that are either inappropriate or biased by researchers’ idiosyncratic methodological approaches. As such, this study investigates whether the adoption of a conditioned multiple linear regression model when testing the predictions of the trade-off and pecking order theories can provide greater insight into the true determinants of firms’ capital structure ratios once model specification has been appropriately determined and variation in variable construction accounted for. The study finds that the adoption of a conditioned multiple linear regression model can provide significant insight into the true determinants of firms capital structure ratios, by enabling more reliable inferences to be made with regards the factors driving observed variable coefficients, and by identifying variable effects that are not observable when a single large dataset is employed. In particular, the results indicate that a number of factors associated with the trade-off theory significantly affect the financing decisions of certain types of firms, whilst factors associated with the pecking order theory appear to be of little concern to decision makers. In addition, the results show that whilst the coefficients of a number of variables cannot be inferred to arise from factors associated with either theory, they do vary significantly across sub-samples of observations, indicating that the factors driving these coefficients vary depending on the type of firm included in the sample. Furthermore, the study demonstrates that the decision to specify a dynamic model is inappropriate when testing the predictions of the trade-off and pecking order theories, and that the specification of such a model can lead to significantly different inferences being made than when the more appropriate static model is employed. Finally, the study shows that the determinants of long- and short-term debt-to-value ratios often differ, and thus should be investigated separately when attempting to identify the factors affecting firms’ capital structure ratios, and that varying the manner in which firm value and firm size are measured can significantly affect the parameter estimates generated, and hence must be accounted for if reliable inferences are to be made.