On the time-varying behavior of household credit in Brazil

Abstract

We add to the debate on access to finance by proposing an empirical exercise to address how credit, monetary, and macroeconomic variables can explain the time-specific behavior of the real variation in earmarked and non-earmarked credit issued to households in Brazil. We test an economic model by reconciling Brazil's limited availability of monthly data with the insights from the theoretical model suggested by Rubaszek and Serwa (2014) and the empirical literature applied to emerging markets. First, we estimate a standard Vector Autoregression (VAR) model. To capture the dynamic nature of household credit, we employ a Bayesian Time-Varying Coefficient VAR model (BTVC-VAR). About the messages of this paper, first we provide a methodological contribution because the joint behavior of both types of credit is crucial to understand their dynamics in the sense of a general equilibrium approach. Second, when we compare the standard VAR and BTCV-VAR, we can see how important it is to allow flexibility of parameters since the dynamics seem to have changed over time, from April 2011 to December 2023. We highlight the significant role played by non-earmarked spread, the average term of both types of credit, economic activity (IBC-BR), and the R$/US$ exchange rate. This exercise contributes to the discussion on financial stability and monetary policy and can be replicated for other emerging markets.