Time varying tax-spend nexus: evidence from Brazil

Abstract

The main purpose of this paper is to examine the causal relationship between government revenue and expenditure in Brazil over the period January 1995 to May 2019. Unlike previous studies, which assume constant relationship through the years, this work attempts to use Granger causality test to endogenously capture structural breaks in the model and to determine the direction of causal relationship for each subperiod. Such assumption is of great importance for testing revenue-expenditure nexus in countries like Brazil, which experienced several events that have significantly changed fiscal policy over a long-time horizon and the causality is likely to change over time. This empirical assessment has both theoretical and practical relevance and it can help properly address fiscal imbalances. In this study we employ two Wald test statistics recently proposed by Shin et al. (2018), the recursive rolling window Wald test and the rolling window Wald test of Swanson (1998). The results reflect the prevailing paradigms in different periods and provide support for Fiscal Synchronization hypothesis for the period up to 2008. For the period 2008-2011, marked by the subprime crisis, the Fiscal Synchronization hypothesis has prevailed. The results also offer support for Tax-Spend hypothesis for the period 2011-2013 and Institutional Separation for the periods 2008-2011 and 2013-2015. Finally, from 2015 onwards, when the government made utmost efforts to cut expenditures and to find new sources of revenue, the Spend-Tax hypothesis has prevailed. The new fiscal framework suggests that Brazil is in the right path to control its fiscal imbalance.