Political and organizational influences on the accuracy of forecasting state government revenues

This paper tests a general theory of the factors influencing the accuracy of state government revenue forecasts. Besides the more familiar hypotheses on forecasting techniques and randomness of dependent variable time series, our theory includes hypotheses on the political environment and organizational procedures used in forecasting. The primary data are from three surveys of state governments and include percentage forecasts errors for total and sales tax revenues. The analysis uses two measures of forecast accuracy, the mean and median absolute percentage errors. These are estimated in a linear model that uses ordinary least squares and least absolute value regressions. The results confirm most parts of the theoretical model, subject to the caveats of field data. Forecast accuracy increases when there are independent forecasts from competing agencies. It increases even more when formal procedures exist to combine competing forecasts. It decreases when outside expert advisors are used and when there is a dominant political party or ideology. Finally, it increases when simple regression models and judgmental methods are used as opposed to univariate time series methods or econometric models.