An Analysis of Using Time-Series Current and Deferred Income Tax Expense to Forecast Income Taxes Paid

Abstract

Prior research, using cross-sectional data, concluded that interperiod income tax allocation is useful in forecasting income tax payments (Murdoch, Costa, & Krause, 1994 and Cheung, Krishnan, & Min, 1997). Both these articles suggested that future research should focus on investigating whether time-series data are also useful in forecasting income tax payments. This paper uses time-series data from 235 Compustat firms over a 20-year period to evaluate whether income tax expense is useful in forecasting one-, two-, and three-year ahead income tax payments. We conclude that firms’ predictions are more accurate for shorter forecast horizons. Additionally, we determine that deferred income tax expense enhances the ability of current income tax expense to predict future tax payments for approximately 40% of firms across all three forecast horizons. Furthermore, we find that the prediction accuracy of a firm’s one-year ahead forecasts is significantly related to the prediction accuracy of its two- and three-year ahead forecasts.