Curva de Phillips salarial novo Keynesiana para a economia brasileira: identificação com dados estaduais a partir de uma análise de vetores autorregressivos para dados em painel

Abstract

The present work aims to empirically verify the New Keynesian wage Phillips curve for the Brazilian economy using state data. The theoretical support is based on the New Keynesian Phillips curve developed by Galí (2011) and its breakdown for many regions in a monetary union, as proposed by Levy (2019). The empirical strategy uses data from the Continuous National Household Sample Survey (PNADC) and IPEADATA between the second quarter of 2012 and the last quarter of 2019, in a structure of Autoregressive Vectors for Panel Data (PVAR). In our favorite estimate, the results suggest that in the second period after the shock there is a significant −0.27 percentage point decrease in nominal wage growth, which remains below its equilibrium value for about 6 quarters. At the end of 8 quarters, wage growth has a cumulative impact of −0.70 percentage points after an innovation in the unemployment rate. Furthermore, the inclusion of productivity measures in the labor market, such as education levels, did not represent significant changes between nominal wage growth and the unemployment rate for the period considered, whereas alternative measures of wage inflation, unemployment and the output gap, expose significant changes between nominal wage growth and marginal cost for the Brazilian economy with disaggregated data. It was also possible to observe, for the period considered, that estimates at the aggregate level were used, for that purpose, aggregated data referring to Brazil as the weighted average of its Federative Units and estimates of the 5 large regions, North, Northeast, Southeast, South and Center – West separately, in a vector structure – autoregressive – VAR, did not show statistical and economic significance.