![Veritas propter investigationem [Truth through research]](images/SubTitle.jpg)
|
PensionReforms’ summary and comments
Until 2007, New Zealand’s private savings operated in a relatively ‘hands-off’ regulatory environment. The government provided a relatively generous, Universal Pension at Tier 1 (‘New Zealand Superannuation’) and left New Zealanders and their employers to decide whether, how and how much to save for retirement. There were no tax breaks for private provision and, unlike Australia, no compulsory private arrangements at Tier 2.
That changed in 2007 with the introduction of KiwiSaver: see here. This was essentially founded on the assumption that the government needed to intervene in individuals’ saving arrangements. KiwiSaver is the world’s first, national, auto-enrolment, tax-subsidised retirement savings scheme. Evidence for that assumption of the need to intervene is actually sparse: see here and here for example. Whether or not this expensive intervention will actually work is also a moot point: see here and here for example. In fact, New Zealanders, before KiwiSaver, may actually have been slightly over-saving for retirement: see here and here.
This latest report did not set out to discover whether KiwiSaver was a justified intervention; its genesis significantly preceded KiwiSaver with the launch in 2002 of a longitudinal study, the Survey of Family Income and Employment (SoFIE) as the
Putting that disappointment to one side, PensionReforms suggests that, before KiwiSaver, there did not seem to be too much reason to worry about what New Zealanders were doing about saving for retirement (or anything else). Although the report did not set out to question whether KiwiSaver was needed in 2007 (indeed, does not even refer to KiwiSaver because that started a year after the second tranche of data was collected), what continues to be discovered about all this makes its introduction even more of a mystery. It was probably the product of policy-based evidence gathering rather than the preferable evidence-based policy formation. (File size 726 KB; 77 pp) 3
first longitudinal study of its kind in New Zealand. SoFIE looks at, amongst other things, New Zealanders’ saving and wealth accumulation habits. The report analyses SoFIE’s first results: the first, that is, that cover more than one year of the eight year study.
“Household saving rates are an important piece of evidence needed for informed policy debate particularly, but not solely, in relation to retirement income policies. Conceptually, estimates of saving rates can be based on a flow measure (income less consumption) or a stock measure (changes in net wealth). The flow measure of saving from the national accounts has shown a strong downward trend, and has been negative since 1993. There appears to be an inverse relation between this measure of savings and net housing wealth. However, regardless of whether the flow or stock method is used, the measurement of saving by households has proved less than straightforward.”
Financial data are gathered each two years over SoFIE’s eight year life. The first of these years (2004) was well before KiwiSaver emerged. In the second year (2006), the first, modest, iteration of KiwiSaver was known about but had not started. The third look will be as of 2008, one year after KiwiSaver began.
“After updating estimates from two the existing sources, this paper presents initial estimates derived from SoFIE. … The estimates were made by comparing net wealth in 2004 with that in 2006 at the individual level and computing the implied real saving rate on an annual basis. This yielded an overall median estimate of 16% of gross income. This is of the same order of magnitude as the long run average annual saving rate measured from the aggregate household balance sheet from [the Reserve Bank of New Zealand], which was 16% of disposable income, equivalent to about 12% of gross income.”
This ‘cross check’ gave the report’s authors some comfort about their overall conclusions. It also confirmed the difficulties of using what the ‘System of National Accounts’ calls ‘household saving’. In New Zealand, this has been substantially negative since 1993 (see here).
But, focusing on SoFIE’s medians disguises some difficulties with the data.
“There is a strikingly wide distribution of saving rates. For example across many categories of individuals around 40% are estimated to have had a decline in net wealth implying a negative rate of saving. Some of this is to be expected as for example, when young individuals invest in education and acquire student loans, and older people draw on past savings in retirement. However, the number of negative savers exceeds that which could be attributed to these two groups.”
The report notes that, at the ‘1st percentile’ (bottom), the ‘real saving rate was -1,723% while at the ‘99th percentile’, the ‘real saving rate was +3,326% while the ‘median’ was +16%.
The report suggests that the dispersion could be explained by dividing ‘observed net wealth’ into two components: a ‘permanent’ component and a ‘random or transitory’ component, which apparently could be quite large.
“Some of the transitory component could have arisen through measurement errors. For example, survey respondents may have reported having a particular asset in wave 2 [2004] and omitted to mention it in wave 4 [2006]. Some changes in wealth could have arisen through marriage dissolutions, or through equity withdrawals for consumption. Much remains to be done to develop a fuller insight into the magnitude of the transitory component of net wealth and its effect on saving rates.”
Other difficulties with the data have prevented the authors from a full analysis of changes in wealth:
“Ideally, we would want to decompose the change in the gross value of all assets into that due to prices and that due to a real change in quantity (together with an interaction effect). The price effect constitutes “passive” saving, while changes in the quantity reflect ‘active’ saving.”
Some of the data can be analysed with relative confidence. Given housing’s importance in the asset holdings of New Zealand households (a net 44.5% of all net assets – Table 5, p24), the report was able to separate housing’s price effect from the ‘active saving’ component.
“Our estimate of the median saving rate for property holders fell from 41% to 6% when we removed an estimate of the effect of house prices. For the longitudinal population as a whole, the effect was to reduce the estimated median saving rate from 16% to 5%. Asset revaluations are therefore a potentially large contributor to changes in household net wealth.”
However, as the report notes, there were difficulties with applying the same treatment to other assets. Also, SoFIE’s treatment of assets owned by family trusts (and the economic relationship between the household and the trust) means there is a less than complete picture of the ‘economic unit’s’ wealth and changes in that where there is a family trust.
The report emphasises the importance of longitudinal studies:
“The ability to hold constant many unobservable characteristics of individuals, by observing them at repeated points in time, offers the opportunity to address a wide range of social policy questions in a manner not previously possible. As additional waves of data from SoFIE become available, the real value of a major longitudinal study will grow markedly. The evidence from long standing surveys of this type in other countries bears testimony to their value.”
PensionReforms agrees with this conclusion. It seems a shame though that there are so many difficulties with the SoFIE data. Despite the report’s best efforts to sort the useful from the distracting, there have to be large caveats about the findings once we move too far away from results that are close to the medians. It is too late to do much about this as the last (2008 and 2010) waves of financial information will have already been, or are being, collected. What should happen next is a replacement, improved version of SoFIE; the risk is that there will be no replacement.
Powered by Website Manager.
©
RightNow Ltd 2002.