3.3  Approaches to identifying discretionary fiscal policy

There are a number of approaches that can be used to estimate the effect of the economic cycle and so calculate the structural fiscal balance (see Giorno, Richardson, Roseveare and van den Noord, 1995; Bank of Italy, 1999; van den Noord, 2000). A common approach is a two-step methodology that utilises an estimate of the output gap together with a set of elasticities of tax and spending to output. The New Zealand Treasury regularly publishes a structural OBERAC estimated using this two-step method (see the Appendix). The OBERAC and estimated structural OBERAC are shown in Figure 1. Tam and Kirkham (2001) find that the calculation of the structural balance for New Zealand using the two-step method is sensitive to the output gap calculation.

Figure 1 - OBERAC and estimated structural OBERAC

Note: As at Budget 2002

Source: The Treasury

There are two disadvantages in using Treasury’s method as an input into the estimation of fiscal impulse. First, the current methodology does not make any adjustment for cyclical variations in interest rates and inflation. However, this problem has become less important as inflation rates have declined and become more stable. Moreover, since we are interested in discretionary policy, then net interest payments would be excluded in order to focus on the primary structural balance. Second, the current methodology is sensitive to estimates of potential output, which are uncertain, especially toward the end of the sample period and into the forecast horizon. These are the periods where discretionary policy changes and fiscal impulse may be of most interest. Blanchard (1993) points out that the CAB was not designed as an indicator of changes in discretionary fiscal policy, and it relies needlessly on the uncertain calculation of potential output.

Instead, Blanchard suggests the use of an indexed approach. His suggested indicator of discretionary fiscal policy has become known as the Blanchard Fiscal Impulse (BFI) (see for example, Alesina and Perotti, 1995, 1997; and various papers in Bank of Italy, 1999). The BFI is defined as the value of the primary surplus which would have prevailed, were unemployment at the same value as in the previous year, minus the value of the primary surplus in the previous year, both as a ratio to GDP in each year (see Blanchard, 1993). Blanchard removes net interest payments as a simple way of adjusting the balance for changes in inflation and real interest rates. Fluctuations in net interest payments are also considered non-discretionary. The BFI is essentially a cyclical adjustment that eliminates from the fiscal balance changes in taxes and transfers due to changes in the unemployment rate (Alesina and Perotti, 1997).

Although the BFI approach avoids the need to estimate potential output, Kearney, McCoy, Duffy, McMahon and Smyth (2000) note that it assumes a stable relationship between changes in unemployment and economic activity, which may not be appropriate (especially during periods of structural change).

Another approach is to use a structural VAR, which takes into account any feedback between fiscal policy and the economic cycle (see Bouthevillain and Quinet, 1999; Kearney et al., 2000). The structural balance estimated via the two-step method attempts to remove the effect of the economic cycle on the fiscal balance, but ignores the fact that the fiscal balance also may affect the economic cycle. Bouthevillain and Quinet (1999) and Kearney et al. (2000) estimate a two-variable structural VAR model that decomposes fluctuations in the deficit-to-GDP ratio into those arising from shocks to output (assumed to have permanent effects) and changes in the deficit itself (assumed to have transitory effects).

However, Kearney et al. suggest that the structural VAR method can be unreliable in the presence of structural change. Other disadvantages of this method are that the identifying procedure is subjective so that the estimates are sensitive to small changes in the restrictions. In addition, the structural VAR method is not simple to update when new information becomes available, making it difficult to monitor an indicator based on a structural VAR approach on a regular basis.