4 Exploring the Cohort Patterns (continued)

Figure 8 – Real Payments under New Zealand Superannuation: 1970-2001

Source: NZ Treasury.

It has been argued that the universal New Zealand superannuation scheme has had, and continues to have, a significant influence on the saving behaviour of households. In the 1970’s, when the scheme was perceived as both more generous and certain, saving rates of the cohorts directly affected were markedly lower. However it is recognised that there is still a debate about this response.

Arguably a pivotal study that examined the effect of public pensions on private saving was Feldstein (1974), and subsequently Feldstein (1996). Based on a life-cycle consumption function including a term for social security wealth, Feldstein argued that the presence of the US social security scheme significantly lowered the level of personal saving in the USA. However, Meguire (2001) challenges these findings and demonstrates that relatively modest changes to the sample period and the definition of the wealth variables can reverse the conclusion.

Accepting for the moment that the USA evidence does support the hypothesis that a public pension scheme reduces the level of personal saving, would we expect this to carry over to New Zealand? The US social security system is funded by a payroll tax and the benefits are earnings related (although not strictly proportional). It is arguable that while the system has in effect operated as a pay-as-you-go scheme (as distinct from a funded scheme), contributors may have perceived the scheme more as an individual retirement account. In that case, the public saving may have been seen as a closer substitute for personal saving than the defined benefit system funded from general revenue that operates in New Zealand.

In short, it appears the evidence is mixed. We would argue however that unit record data of the type used in this study are potentially more revealing of the underlying household responses than aggregate time series data of the type used by Feldstein. It is quite possible that households in the upper two or three deciles of lifetime income, show no response at all to changes in social security wealth. They discount the public pension as either a small share of their retirement income or subject to political uncertainty about whether they would eventually be beneficiaries in any event. When it is recalled that these households provide the majority of the total household saving (over 80 percent in the case of New Zealand) then it should not be surprising if aggregate data failed to reveal any effect of the superannuation scheme on personal saving. In contrast, for the 50 percent of the households in the lower income deciles, there may be some response toward reduced private saving[38], but the aggregate effect of this is likely to be quite small.

In addition to the direct effect of the public pension scheme on the savings of households, there is evidence that the US social security scheme has led to lower retirement ages and labour force participation among older workers (Venti and Wise (1996) and Lumsdaine and Wise (1990)). We would expect that the New Zealand superannuation system especially during the 1970s and 1980s when it became more generous, has played a similar role in contributing to some of the observed decline in the labour force participation rates of older male workers in New Zealand.

For many households, saving for housing is an important element of voluntary saving. If there was a significant probability of being allocated a state house at a subsidised loan rate with generous conditions for purchase it is likely that the incentive for voluntary saving would be commensurately diminished. In fact, lower savers faced a housing market in which the number of state houses being built was much greater than in later years,[39] thus increasing the probability that a family would get access to a state house on concessional terms.

Added to this, the Family Benefits (Home Ownership) Act 1958 provided the option to capitalise a universal child benefit and apply that to equity in a first home. This would have reduced the incentive to save for an initial deposit (although lifetime income would not have altered). However the fact that a certain cash grant was replacing a future stream of benefits subject to political risk would suggest that the certainty equivalent of the capitalisation scheme would have reduced the incentive to save. Those born from 1925 to 1934 would have been in their household formation years when this policy was in place. They correspond to Cohorts 4 and 5 (Table 4) with saving rates significantly below those for the reference group. Finally, low savers who were paying off mortgages in the 1970s and early 1980s faced negative real interest rates, further reducing the cost of housing and permitting consumption levels to be higher than they would have been had housing costs taken a greater share of disposable incomes.

In the past family sizes were larger. Having more children increases the probability that the parents will receive support in retirement for their children. Reduced family size in later years would be consistent with later cohorts being higher savers than the older cohorts with larger families.

Up to this point our consideration of the different saving behaviour of older and younger cohorts has focussed largely on the state provided benefits that each could have expected. Of course to meet these costs taxes had to be paid, so a full intergenerational accounting requires us to incorporate not only market earnings and state benefits, but taxation payments as well.[40] This has not been attempted here. Thomson (1991) compares a prototypical “early” family (born 1930) with a “late” family (born 1955) and traces their lifetime earnings, taxes and benefits. He finds that the benefit:tax ratio for the so-called early family was 2.3 while the corresponding ratio for the late family was 0.6 to 0.8 (p.172). This result is strikingly consistent with the pattern of lifetime saving behaviour displayed by different birth cohorts in the present study.