Models for forecasting Tax Revenues: The ICMS of the State of Espírito Santo
Name: BERNARDINO JOSAFAT DA SILVA CASTANHO
Type: MSc dissertation
Publication date: 01/04/2011
Advisor:
Name |
Role |
|---|---|
| GUTEMBERG HESPANHA BRASIL | Advisor * |
Examining board:
| Name |
Role |
|---|---|
| GUTEMBERG HESPANHA BRASIL | Advisor * |
| ROGÉRIO ARTHMAR | Internal Examiner * |
Summary: The main objective of this dissertation is the research of a formal model for the monthly forecast of the Value Added Taxes on sales and services (ICMS) collected by the State of Espírito Santo, derived from the term series data analysis of the tax revenue from January 2000 to December 2009 and from the composition basis of the taxation incidence of the tax. The statistical characteristics of the ICMS series were identified and forecasts were drawn up with the use of Holt-Winters exponential smoothing models, of Box-Jenkins methodology, with intervention analysis for structural change detection and of a causal econometric model with dynamic structure. For the specification of the econometric model, the most relevant sectors of the economy that compose the ICMS tax basis and directly influence the tax revenue have been identified. It was chosen the mixed econometric model with multiple regression and only one behavioral equation. The predictive performance of the models was compared through the mean absolute percentage error (MAPE), in order to choose the one that shows the best estimate for the year 2010, used as basis of efficiency evaluation of the generated ex-post forecast. Keywords: ICMS. Forecast. Tax Revenue. Econometric Model.
