Um modelo integrado econométrico+insumo-produto para previsão de longo prazo da demanda de combustíveis no Brasil [An integrated econometric+input-output model for long term forecast of fuel demand in Brazil]

The article presents an integrated econometric + input - output model for generating long-run forecasts of fuel demand in Brazil. The model is based on the integration, by means of a linking strategy, of a vector error correction model with a hybrid input-output model for the Brazilian economy and is capable of forecasting on a yearly basis consumption of four groups of fuels: gas, diesel, fuel oil, and alcohol. In model development, the econometric and the integrated models underwent predictive tests, with the latter being calibrated for optimum performance, using data for the years 2004 to 2007. The integrated model is then used to generate forecasts from 2008 to 2017. The forecasts are based on two alternative scenarios, one predicting a short, and the other a long, duration for the current world economic crisis. The results indicate that under both scenarios a substantial increase of fuel demand is to be expected in the next 10 years.