5  Impact of Varying Assumptions

This section explores the impact on the projected levels of social expenditure of variations to the assumptions used in the benchmark case. In each case, the benchmark model is taken and one assumption or associated set of assumptions is varied. In this manner the following sections show the impact, relative to the benchmark case, of changes in each of these critical assumptions.

5.1  Higher Rate of Migration

The effect of higher net migration was examined by raising the annual net migration level from 5,000 to 20,000. This latter figure corresponds to the upper level used by Statistics New Zealand (2000). Higher rates of migration are variously seen as adding to the burden of social expenditure on the one hand, or providing more workers and enhancing the tax base on the other. The result depends on the age distribution of the migrants relative to the resident population. In this study we assume that the higher migration levels would simply reflect the mix by age and gender of migrants between 1997 and 2001. The effect (see Table 6) is to lower the projected rise in social expenditure by 2051 rather modestly from 31.0% to 29.5%. Migrants are only slightly younger on average than the New Zealand population. Given the estimates of the standard errors this change cannot be regarded as statistically significant.[30]

5.2  Changes in Labour Market Parameters

Two adjustments were made to the labour market parameters. The first raised the average growth in labour productivity to 1.75% from 1.50% per year. This, as expected, reduces the projected mean ratio of social expenditure, from 31.0% to 28.7% by 2051. Details are given the final column of Table 6. In the second case higher rates of female labour force participation were used. All female rates were increased to match those of the males (where male rates were higher), with the exception that for women between 20 and 29 their participation rates were held at a maximum of 90% of the male rates for those age groups. The results are also shown in Table 6, where it can be seen that the higher participation resulted in only a modest and statistically insignificant decline in the mean projected share of social expenditure in GDP.

5.3  Changes in Standard Deviations of Social Expenditures

The standard deviations of the growth rates of major categories of social expenditure were estimated from historical data, as discussed earlier (see Table 1). Their size reflects considerable uncertainty. They contribute significantly to the overall confidence band around social expenditure projections. By 2051, the 5 and 95 percentile bands around the mean estimate of 31.0% are from 21.1% to 44.7%. To explore what would happen if the variability in social expenditure programmes were lower in the future, two sets of simulations were made. For the first simulation, all the standard deviations were set to zero; this provides a useful point of comparison since the variation arises entirely from demographic and labour market distributions. In the second simulation, the standard deviations were reduced by 50 percent.

The effects on the final distributions of social expenditures are shown in Table 7. When the standard deviations of social expenditures are set to zero, the 5 and 95 percentile bands for the ratio of social expenditure to GDP in 2051 are reduced to 25.3% to 37.4%. The confidence intervals therefore remain substantial, reflecting the high standard deviations on participation and unemployment rates.

Note: ^a Variability only contributed by population, labour force participation rate, unemployment rates and productivity