Econometric models for fractional response variables with applications to corporate financing choices

In this project we make several contributions for the theoretical and empirical literature on fractional regression models. With regard to theoretical developments, we: (i) perform a Monte Carlo comparison of several alternative fractional regression models and propose a suitable testing and model selection strategy; (ii) develop a generalized method of moments estimator that accounts appropriately for the likely presence of unobserved heterogeneity in the data generating process; (iii) show how to deal with multinomial data structures; (iv) deal with some sampling issues which may affect fractional data, including excess of zeros and/or ones, endogenous stratification and nonresponse; and (v) extend the analysis of regression models for dependent integers ratio variables, explicitly allowing and testing for specific breakdowns of binomial models assumptions. To illustrate the usefulness of the techniques proposed, we apply some of them to the analysis of capital structure decisions.