An analysis of the accuracy of the Treasury's macroeconomic and tax forecasts.
Tax revenue forecasts
Tax revenue is a measure of the tax that is due for a particular period, regardless of whether or not the tax has actually been paid. Tax revenue is an important component of the Crown's operating balance, accounting for the majority of the Crown's total revenue.
All of the RMSEs have decreased since the last report, indicating that the accuracy of the tax revenue forecasts has increased over the last few years. As the forecast horizon shortens, the RMSEs decline, indicating that the shorter-horizon forecasts were more accurate than the longer-horizon forecasts, a feature that is always desirable in a forecasting process.
There was insufficient evidence to conclude that the tax revenue forecasts were biased at any forecast horizon. However, there was evidence of serial correlation in the tax forecast errors. If errors are serially correlated, then an under/over-forecast of one result increases the probability of an under/over-forecast of the next result. This will be manifested in long runs of solely-positive or solely-negative forecast errors, as was the case from 2000 through to 2008 (see following chart).
In 2009, when the average one-year-ahead tax forecasting error was over 4%, the Treasury set itself a target of achieving tax forecasting errors of less than ±3%. The chart above shows that:
- six of the seven forecast errors from 2003 to 2009 were outside this range, and
- the six forecast errors from 2010 to 2015 were all within the target range.