Unit Root Tests and the Classification of Corporate Financial Behaviour
21 Pages Posted: 16 Mar 2006
Date Written: March 14, 2006
The empirical analysis of corporate financial behaviour show the relevance of an approach able to nest both Trade-Off and Pecking-Order theoretical considerations. While we believe that the good empirical performance of the nested model is due to companies genuinely opting for a mixed behaviour, we also think that the parameters homogeneity implied by dynamic panel estimation techniques applied to longitudinal data (large N, small T) may hide the presence of companies following a specific behaviour, i.e. the pure TO or the pure PO model. Disentangling these pure behaviours and the characteristics of companies following them is important for the implementation of fiscal and/or monetary policies. We thus suggest a parsimonious and reliable approach for assessing the contribution of each of the two theories in explaining the effective debt-ratio patterns. This approach is based on single-company and panel unit root tests. Results show that the behaviour of enterprises is largely heterogeneous in both the theory driving the historical data and, more relevantly, in the model parameters. The single-company unit root test response seems to indicate that 70% of debt-ratios analysed are non-stationary, denoting a Pecking-Order behaviour. Given the surprising result, we try to go further, linking unit root tests response to companies characteristics and operative conditions, and checking how estimates of the nested-mixed model change in the sub-samples of stationary and non-stationary debt-ratios (pure financial behaviours).
Keywords: Unit root tests, Mean-reversion, Panel data, Capital structure, Trade-off, Pecking-order, Modified Pecking Order approach
JEL Classification: C12, G32, C22, C23
Suggested Citation: Suggested Citation