The impact of political, institutional, methodological, and economic factors on forecast error

Error in state revenue forecasting has negative effects on government resulting in re-budgeting and budgetary cutbacks. Minimizing forecast error and the disruptions the error causes is of paramount importance to state governments if they are to properly plan and execute policy. This research considers the sources of state revenue forecast error. Four sources of forecast error, economic stability, political party dominance, institutional structures, and forecasting methodology, are considered and tested for their effects. Economic stability refers to the cyclic nature of the economy, when the economy is erratically expanding or contracting, forecasting should be more difficult. Economic instability was found to be positively related to forecast error, although significance among the various models tested were mixed. Political party dominance, or a unified government, is found to reduce forecast error. It is theorized poor forecasts result in less political capital loss to split governments than to a unified government. Forecast methodology utilized by the forecasters is considered for both sales and income taxes. Three methodologies, qualitative, extrapolative and econometric, are tested for effects with only sales tax econometric methods showing a significant reduction in error. Several institutional factors are also found to influence forecasts errors. Forecast frequency is positively related to forecast errors suggesting that too many forecasts blur the distinctions between random changes and trend changes. University consultation on the forecast is also positively forecast error suggesting that many states have highly qualified staff. Two other institutional factors, balanced budget requirements and presence of economic advisory councils were not significant. In addition to the effects found in these four areas, it was also discovered that there exists a consistent underestimation bias attributable to an asymmetric loss function.