A Composite Approach to Forecasting State Government Revenues

Fiscal uncertainty has led many states to utilize a variety of economic and revenue forecasting methods. To improve accuracy, many budget analysts use alternative projections generated by different techniques. Composite modeling goes a step further and allows users to systematically combine two or more forecasts. This paper examines the effectiveness of composite forecasting of sales tax revenues in Idaho. Baseline projections are provided by an econometric model and by a univariate time series equation. Composite forecasts are found to outperform both baseline models. Combined forecasts are also found to be more precise than executive branch forecasts actually adopted from 1982 through 1985.