Local government revenue forecasting methods: competition and comparison

Abstract

This study examines forecast accuracy associated with the forecast of 55 revenue data series of 18 local governments. The last 18 months (6 quarters; or 2 years) of the data are held-out for accuracy evaluation. Results show that forecast software, damped trend methods, and simple exponential smoothing methods perform best with monthly and quarterly data; and use of monthly or quarterly data is marginally better than annualized data. For monthly data, there is no advantage to converting dollar values to real dollars before forecasting and reconverting using a forecasted index. With annual data, naïve methods can outperform exponential smoothing methods for some types of data; and real dollar conversion generally outperforms nominal dollars. The study suggests benchmark forecast errors and recommends a process for selecting a forecast method.

The Status of Budget Forecasting

“…Autocorrelation is the theoretical justification for the use of time-series methods, which can be either simple or complex. Simple time-series methods include moving average, simple exponential smoothing, Holt exponential smoothing, and damped trend exponential smoothing (see D. W. Williams and Kavanagh [2016] for a complete description of these methods and the formulas by which they are produced). Frank and Zhao (2009) suggest that most quantitative forecasting at the local government level is likely simple moving averages or trend analysis.…”

“…R, however, requires a considerable learning curve. D. W. Williams and Kavanagh (2016) find that Forecast Pro and Autobox generally produce results that modestly outperform typical spreadsheet approaches. Both of these products implement ARIMA and other sophisticated techniques that the modestly skilled forecaster is unlikely to successfully use without assistance.…”

“…This section uses a number of technical terms, which are defined in the Appendix. For those interested in the equations used for many of these techniques see D. W Williams and Kavanagh (2016). for simpler methods orMakridakis, Wheelwright, and Hyndman (1998) for an extensive treatment.9 We use "simulation" to refer to any approach that uses math to imitate real world processes.…”

“…Terms related to bias are defined diffusely through the forecasting literature. Formulae for many of these methods can be found inMakridakis et al (1998) or in D. W Williams and Kavanagh (2016)…”

An Open Systems Model of Local Government Forecasting

“…This stream of literature typically reviews best practices or conducts surveys of forecasters. It has established that forecasters in state and local governments are less inclined to use advanced quantitative econometric and time-series methods relative to the qualitative and contextual knowledge of budgeting (Cirincione, Gurrieri, & Van de Sande, 1999; Frank, 1992, 1993; Frank & Zhao, 2009; Gianakis & Frank, 1993; Reddick, 2004; Williams & Calabrese, 2016; Williams & Kavanagh, 2016). In contrast, state and local governments with more educated budgeters and advanced technology are more likely to use technical (e.g., quantitative econometric or time-series) forecasting methods (Frank & McCollough, 1992; Reddick, 2004).…”