REAL-TIME FORECASTS OF STATE AND LOCAL GOVERNMENT BUDGETS WITH AN APPLICATION TO THE COVID-19 PANDEMIC

Abstract

Using a sample of the 48 contiguous US states, we consider the problem of forecasting state governments’ revenues and expenditures in real time using models that feature mixed-frequency data. We find that mixed-data sampling (MIDAS) regressions that predict low-frequency fiscal outcomes using high-frequency macroeconomic and financial market data outperform traditional fiscal forecasting models in both a relative and an absolute sense. We also consider an application of forecasting fiscal outcomes in the face of the economic un-certainty induced by the coronavirus pandemic. Overall, we show that MIDAS regressions provide a simple tool for predicting fiscal outcomes in real time.