Core Crown - Revenue

Sources: Inland Revenue, The Treasury

Inland Revenue’s total tax forecasts are higher than the Treasury’s in all years, although the gap between the two sets of forecasts is not as large as it was in the Budget Update. Most of the differences occur in the corporate and PAYE tax types.

At the time of the Budget Update, growth in PAYE exceeded that which could be easily explained by growth in aggregate wages and salaries, even after allowing for the impact of fiscal drag. Inland Revenue assumed that this unexplained growth would continue into the forecast period. This assumption opened up a wedge between Treasury and Inland Revenue forecasts of PAYE; this wedge grew to more than $1 billion by the end of the forecast period.

Recent labour market data have shown that employment growth through the first half of 2006 was higher than expected in the Budget Update. PAYE in the September quarter was close to Inland Revenue's forecast, but $200 million above Treasury's.

Treasury's labour market forecast is now much higher than in the Budget Update in the near term and is now more readily reconciled against recent PAYE receipts. In response to recent outturns and to the new labour market forecast, Treasury's PAYE forecasts have increased in the order of $800 million per year. Inland Revenue's PAYE forecasts have increased by much smaller amounts in the near term, and have decreased from 2009 onwards. Thus the difference between the two departments' PAYE forecasts is now much smaller than in the Budget Update.

Detailed comparisons of Treasury and Inland Revenue tax forecasts can be found at www.treasury.govt.nz/forecasts/hyefu/2006.

Treasury or Inland Revenue: Which tax forecasts have been the more accurate?

Ever since the early 1990s, both Treasury and Inland Revenue have been producing tax forecasts for each Budget Update and Half-Year/December Update. This is to ensure that no single agency prepares tax forecasts in isolation and that the official tax forecasts are subjected to some sort of quality assurance. We have examined the accuracy of both agencies’ forecasts over the last decade or so.

Root-mean square error is the square-root of the mean of the squares of the forecast errors. It is a measure of the accuracy and variability of the forecast errors.

Table 2.9 shows that, with the possible exception of the four-year-ahead forecasts, the root-mean square errors of both agencies’ forecasts are very similar to each other. This result is not all that surprising. Both agencies use the same macroeconomic forecast as the basis for their tax forecast. At each forecasting round, there is much discussion between the two tax forecasting teams to test assumptions and judgements, and to ensure that both agencies use all available information in their forecasts. This tends to bring the two sets of forecasts closer together, rather than further apart.Large differences (of $1 billion or more) between the two sets of forecasts have been evident only in the last two forecasting rounds and have mainly affected the longer-horizon forecasts. Therefore, it will be several years before these differences start to affect the statistics presented in table 2.6. In short, over the historical period these forecasts have practically been the same. We will not be able to judge the accuracy of the differences now existing between the two agencies for another 2-3 years.