The use of Structural Equation Modelling (SEM) in Capital Structure Empirical Analysis

Abstract

This study synthesizes the capital structure determinants theory and empirically examines both the determinants and the suggested firm behaviour patterns in relation to financing decisions of 651 UK companies between 1985 to 2000. Such analysis is carried out by using a relatively new and innovative factor-analytic structural equation modelling (SEM) methodology. The SEM Methodology allows the use of more than one indicator for a latent variable. It also estimates the latent variables and accommodates reciprocal causation and interdependences among variables. Methodologically, all previous UK capital structure studies used conventional regression estimates. Given the differences in tax regimes, and similarities in economic systems, it is important to find out whether the results will be different from those of the US study, (Titman and Wessels, 1988), and the Australian study, Chiarella et al. (1992), that used SEM methodology. Consistent with the dominant theory, the findings of this study are that non-debt tax shields, business risk and probability of bankruptcy are negatively related to gearing, while tangibility, firm size and current profitability are positively related to gearing. The study does not provide support for any negative impact on debt arising from past profitability and tangibility