The Treasury

Global Navigation

Personal tools


New Zealand's Fiscal Policy Framework: Experience and Evolution - WP 01/25

6.  The Evolution of Fiscal Forecasting and Budget Processes

In response to the experience with the framework in the mid-1990s, there have been refinements to forecasting and Budget processes. The following description on the top-down management of government spending is drawn from Barnes and Leith (2001). The two key tools for ensuring overall fiscal control and effectiveness are fixed nominal baselines for departmental spending and the fiscal provisions framework.

6.1  Fixed nominal baselines

The early 1990s saw a change whereby Government spending could be characterised as being split into two tracks: “formula-driven” (i.e. indexed) and “fixed” (i.e. no change to nominal baseline amounts). Previously, departmental funding was split into three main input-based streams: personnel, operating costs and capital. Personnel costs were regularly adjusted for movements in wages, and the other two streams were generally adjusted annually to reflect expected cost movements. The Public Finance Act enabled a baseline approach.

Formula driven indexation applies to non-departmental spending on benefits (e.g., inflation indexation of unemployment payments) and to New Zealand Superannuation. Health and education spending are adjusted through formulas that take into account demographic change. A specific policy decision is required to change the amount spent on non-indexed spending.

6.2  Forecasting assumptions

A key issue to emerge from these changes was the relationship between fixed nominal baselines and the short-term fiscal forecasts. Three-year budget forecasts prepared under GAAP between 1994 and 1996 would include increases in government spending only for those areas affected by indexation. All other spending was assumed to remain constant over time.

This approach provided what might be described as a policy neutral forecast. Cost pressures and new initiatives were “assumed” in the forecasts to be funded from savings and efficiency gains. However, because the fiscal forecasts did not allow for new spending in future Budgets, they understated the likely spending profiles. An example of this “forecast bias” is illustrated in Table 2 below.[23] The left-hand column sets out the forecasts for the 1997/98 year operating balance at different points in time, starting from the first time it was forecast through to the actual result. The right-hand column decomposes the change into its forecasting and policy components. The “forecasting” component includes changes attributable to different macroeconomic conditions than forecast, and revised tax and welfare bases.[24]

The analysis indicates that there was significant policy change ($3.7 billion) with respect to the forecast assumption. This “slippage” against forecast reflects the tension mentioned in Section 5.2.3 – between setting realistic assumptions and the political economy of incorporating a specific amount for new spending.

Table 2 - Operating balance for 1997/98: forecast and policy changes
Forecast 1997/98 operating balance ($billion) Policy and Forecast changes from 1994 DEFU to actual result ($billion)*
1994 DEFU 7.6 Revenue: Policy – 1.0
1995 Budget 7.8 Revenue: Forecasting – 1.1
1996 Budget 3.3 Expenses: Policy – 2.7
1997 Budget 1.5 Expenses: Forecasting – 0.6
1998 Budget 2.8 SOE/CE surplus: Policy
1998 Actual 2.5 SOE/CE surplus: Forecasting 0.2
    Total: Policy – 3.7
    Total:Forecasting – 1.5
Actual less DEFU forecast –5.1   – 5.2

* Change is expressed in terms of the impact on the operating balance. DEFU refers to December Economic and Fiscal Update. CE refers to Crown entity. Totals do not sum due to rounding. Source: Adapted from Table 1.4, OECD (1999).

The approach resulted in optimistic projections of progress towards the long-term fiscal objectives.[25] This created a number of issues, including those mentioned previously around macroeconomic management as well as discipline on the annual Budget process. Further, the approach raised credibility problems about likely progress towards long-term fiscal goals (see for example, OECD, 1999).

On the political side, there were also pressures to find a better way to represent spending intentions. For example, New Zealand’s first coalition government sought a mechanism to demonstrate fiscal prudence and reduce the possibility that portfolio Ministers from different coalition parties would bid up spending in their sector. The response was a statement incorporated into the Coalition Agreement committing to a (cumulative) $5 billion cap on new spending over a three-year term of government to 1999/2000.

Importantly, this cap was on top of expenses already included in the fiscal forecasts (i.e. on top of the fixed nominal baselines and formula-driven indexed items). The cap evolved into a mechanism now known as the fiscal provisions.


  • [23]If fixed nominal baselines and indexed spending are strictly interpreted as representing current policy, then increases in spending reflect a change in fiscal policy rather than “forecast bias”.
  • [24]Some of the forecasting change may reflect changes in fiscal policy and so could arguably be allocated to the policy change component. The decomposition used does not allow for these effects.
  • [25]The Progress Outlooks in Fiscal Strategy Reports did include higher spending scenarios and so provided some indication of alternative paths towards stated long-term fiscal objectives.
Page top