Measurement Matters A Meta-Study of the Determinants of Corporate Capital Structure

Abstract

This study aggregates the mixed empirical evidence of the seven most commonly investigated determinants of corporate capital structure. We apply meta-regression analysis on a data set of 3,890 reported results, manually collected from 100 primary studies covering firm observations from 57 countries over the past 65 years. Our results reveal that – in descending order of importance – tangible assets (positive sign), market-to-book ratio (negative sign), and profitability (negative sign) are significant determinants of corporate debt level. In addition, we identify the presence of publication selection bias in academic literature. Accordingly, specific results are systematically overrepresented, as authors prefer reporting statistically significant estimates in line with theory or corresponding to previous empirical research. Significant determinants as well as publication selection bias are more pronounced for characteristics like market-based measures of capital structure or total debt measures of capital structure or for top articles from highly renowned journals, as compared to book-based measures of capital structure or long-term debt measures of capital structure or randomly selected articles including more unknown and unpublished studies. Overall, these findings highlight the need to relativize existing statistically significant results in this field and instead provide independent analyses in future for scientific progress.