5.3 Has fiscal policy been pro-cyclical?

A key policy question is whether fiscal policy has been counter or pro-cyclical over time. The stylised fact from cross-country studies is that fiscal policy has tended to be a-cyclical or counter-cyclical in industrial countries, but has tended to be pro-cyclical in developing countries (Ilzetzki and Vegh, 2008).

We quantify the impact of fiscal stance on the level of both GDP and long-term interest rates by conducting the following counterfactual exercise. Using the moving average representation of the SVAR model, we simulate an alternative path for each variable by comparing the actual path with a simulated path in which fiscal shocks (both net taxes and government spending) are suppressed. In this way, the model is used to simulate the outcome of alternatives to the observed policy shocks to create a counterfactual with which to compare the observed history and is essentially a different way of presenting the results shown in Figures 13-15.

Figure 16shows actual GDP detrended and the counterfactual output gap that excludes the contribution of the fiscal shocks. The difference comprises the fiscal impulse, defined as the net contribution to GDP from net taxes and governments spending. Discretionary fiscal policy is defined as pro-cyclical (counter-cyclical) if de-trended actual GDP is further away from (closer to) the horizontal axis than the counterfactual simulation of GDP without fiscal shocks. Equivalently, fiscal policy would be pro-cyclical if the fiscal impulse is the same sign as the output gap and the fiscal impulse is not larger in magnitude than the output gap.

Discretionary fiscal policy appears to have avoided pro-cyclical contributions to output over 1984 to 1997, but has acted in a pro-cyclical manner for much of the subsequent period. In other words, this model suggests that over 1998 to 2010, New Zealand has experienced a more pronounced cycle in GDP than would have been the case without unexpected changes in discretionary fiscal policy.

Fiscal shocks are broadly a-cyclical over the late 1980s and early 1990s. From 1994 to 1997, fiscal policy appears to have played a counter-cyclical role by dampening an upturn in GDP. Over 1998 and 1999, fiscal policy dampens GDP which exacerbates the negative output gap. Over 2000 to 2006, fiscal policy is pro-cyclical by exacerbating the upturn in GDP.

Over 2007 to 2010, fiscal policy dampens GDP which is initially counter-cyclical but then pro-cyclical as the output gap turned negative. This outcome is mainly driven by the dampening effects of net taxes as the effect of government spending during this period is slightly positive (see Figure 13).

This rather counter-intuitive result could again be partly explained by the tax puzzle mentioned in previous sections. The comprehensive study by Kearney et al. (2000) also shows that the SVAR measure of fiscal impulse may be biased during periods of major shifts in the economy. By comparing across five alternative measures, they find that the SVAR measure of fiscal impulse can diverge substantially from the others particularly during episodes of major structural changes in the economy. They argue that in such cases, the model cannot capture the major shift that took place in that period and so is applying a common estimate of the whole period.

Figure 16: GDP cycle and fiscal policy

Source: Statistics New Zealand, authors' calculations.

Figure 17 shows the de-trended actual inflation and a counter-factual simulation without contributions from fiscal policy. The pro-cyclicality of fiscal policy with respect to inflation can be assessed in a similar manner to the GDP cycle. Overall, fiscal policy appears to make only a minor contribution to inflation deviations from trend. In the mid 1980s, fiscal policy makes a small pro-cyclical contribution by exacerbating above-trend inflation. Over the early and mid 1990s, fiscal policy appears broadly a-cyclical with respect to inflation. During the late 1990s, fiscal policy dampens inflation in a pro-cyclical manner. Fiscal policy is a-cyclical or modestly counter-cyclical over the 2000s.

Figure 17: Inflation cycle and fiscal policy

Source: Statistics New Zealand, authors' calculations.

Figure 18 shows the actual long-term interest rate and a counter-factual interest rate without fiscal policy shocks. This indicates that fiscal policy induced a higher long-term interest rate over 1984 to 1990 (averaging around 50 basis points) and 2000 to 2007 (averaging around 25 basis points) and dampened the interest rate over 1991 to 1999 (averaging around 60 basis points) and 2007 to 2010 (averaging around 60 basis points).

Figure 18: Interest rate and fiscal policy

Figure 19 shows the counterfactual interest rate which isolates the effect of only government spending shocks, since we may be more confident about the identification of government spending shocks compared with net tax shocks. This shows that government spending dampened the interest rate over 2000 to 2004 by 45 basis points on average and then exerted an upward contribution by an average 30 basis points over 2005 to 2010.

The results suggest that fiscal policy can exert quite a significant impact on the long-term interest rate. As context, the standard deviation of the absolute quarterly change in the 10-year bond yield over 1990:1 to 2010:2 is 35 basis points.

Figure 19: Interest rate and government spending

Source: Statistics New Zealand, authors' calculations.

Brook (2012) argues that pro-cyclical discretionary fiscal policy over the 2000s likely exacerbated the interest rate and exchange rate cycles in New Zealand thereby worsening the current account balance. Our analysis is broadly congruent with Brook (2012) in finding evidence of pro-cyclical fiscal policy and upward pressure on the interest rate during parts of the 2000s.