The present article studies the determinants of banking spreads, allowing for the possibility that the impact of some of these determinants on spreads may differ according to the particular loan type. This concern is fostered by both theoretical and empirical evidence supporting the general idea that the hetero-geneity of banks’ loan portfolios should be taken into account when studying the drivers of spread. This approach is distinct from previous work in the liter-ature, usually utilizing a single interest margin per bank, in order to measure the impact of its determinants. Using a dataset of observations on various per-sonal loan categories and the Difference GMM approach, the present study es-timates that marginal effects of, respectively, banks’ risk aversion, credit risk, and market share on spreads differ significantly according to whether the loan is a consumer loan, a paycheck-linked credit line or a revolving credit line for individuals. These findings suggest, accordingly, that central banks and regula-tory agencies should observe the composition of banks’ loans portfolios when writing their policies aiming at spread reduction.

The present text is focused on the evaluation of the influence that determinants of loan contracts’ design may have on loan renegotiation. This general purpose is faced with the possible simultaneity of determination of some such factors (namely the use of collaterals and loan interest rate) and the likelihood of contract renegotiation. Naturally, this simultaneity corresponds to an issue of endogeneity of these factors within econometric models of the conditional probability of renegotiation. If neglected, covariates’ endogeneity causes model misspecification, translating into unreliable empirical assessments of the marginal effects of all renegotiation determinants on the probability of loan redesign. Building on the theory of incomplete contracts, suggesting that the possibility of contract renegotiation is already anticipated at the contracting date, the empirical findings of the paper, based on a dataset provided by a Brazilian bank, provide strong indication that this can indeed be the case. Accordingly, the empirical assessment of the marginal contributions of important variables to the likelihood of renegotiation is revisited, thereby enabling a deeper understanding of renegotiation processes.