Uma metodologia de construção de séries de alta frequência das finanças municipais no Brasil com aplicação para o IPTU e o ISS: 2004-2010

Taxes levied by Brazil's 5.563 municipalities are significant components of the country's aggregate gross tax burden. However, official high-frequency aggregate data on these revenues are unavailable as of this moment. This paper develops a methodology for estimating them. Two procedures are adopted: i) cluster and interpolation techniques, in order to construct a panel with the annual revenues of all 5.563 Brazilian municipalities (and Brasilia, the country's capital) from 2002 to 2010; and ii) stratified random sampling and temporal disaggregation procedures, in order to estimate aggregate monthly revenues from a sample of 297 municipalities. Two primary sources of data are combined: annual data from Finanças do Brasil (Finbra) - Dados Contábeis dos Municípios from the Ministry of Finance and monthly data from the still relatively unknown Relatórios Resumidos de Execução Orçamentária (RREOs). The methodology was employed in the construction of monthly estimates of the revenues of the two most important Brazilian local taxes from 2004 to 2010. The paper also discusses some stylized facts found in the disaggregated annual data and aggregated high frequency data on the tax revenues of Brazilian municipalities.