5 Summary and Conclusions

“No one saving rate measure is the answer to all the questions one might pose about saving and wealth accumulation”

Boskin (1988), p.31

It has long been recognised that saving rates in New Zealand are low, and below those of most other OECD countries. From a macroeconomic perspective, a dearth of saving raises concerns of over reliance on the possibly fickle supply of foreign saving. If national savings are not sufficient to meet the desired level of investment, foreign savings in the form of capital inflows will be required to make up the shortfall. In 2012 some 85% of New Zealand's net investment was met by foreign rather than national savings.

Low levels of domestic saving may have implications for the level of investment. Too little capital investment can be associated with lower labour productivity and a slower growth of real incomes than might otherwise be the case. From a micro perspective, retirement income adequacy and unsustainable household debt are often aired as concerns.

Successive governments have introduced a range of pro-saving policies. Similar approaches have been adopted by other countries. A necessary condition for ensuring such interventions are truly welfare enhancing is that evidence base on which they rest is solid. This paper contributes to that goal by examining the measurement of saving. It builds on a number of existing pieces published by the Treasury.

The paper makes three significant contributions. First it documents the existing measures of savings from the national accounts from 1972 to 2012. Second it summarises the evidence on the level and trends of both flow and stock measures of saving. Third, it makes a series of adjustments which illuminate important conceptual issues in the measurement of saving. Throughout the focus is on the measurement of savings; no attempt is made to address the question of adequacy.

The long run average national saving rates are 17.7% of GDP (gross) and 3.4% (net). For the household sector the average long run rate is -0.1%. As the boundary between households and business in not one that is clearly defined in practice, it is sometimes preferable to look at the combined household and business sectors, denoted the private sector. Here the long run average rate is 2.4% of GDP.

Changes in net wealth are a useful alternative for estimating savings rates. Such changes however embody both and active and passive component of saving, the latter arising from asset revaluations. In applying the stock measure to aggregate household sector data on assets and liabilities we find the long run average annual saving rate was 16.1% of household disposable income (from 1979 to 2011). After stripping out the effect of house price revaluations (the passive component) which amounted to an annual average of 8.9% we are left with an estimate of the active annual average household saving rate of 7.1%. In contrast, over the same period the flow measure of the average annual household saving rate as measured by the national accounts was -0.7% of household disposable income.

A completely independent approach to measuring household saving rates from net wealth data can be applied to household survey data. Over the years 2004 to 2006, the annual average saving rate derived from the Survey of Family Income and Employment was 19%, which was virtually identical to the saving ratio from the aggregate Reserve Bank of 18% (adjusted to be comparable with the SoFIE calculations), suggesting these estimates are reasonably robust. In contrast, over the same period the flow measure of the average annual household saving rate as measured by the national accounts was -7.3% of household disposable income.

Regarding changes in wealth, it is useful to note that changes in wealth may occur for a variety of reasons, including permanent changes in the future earning potential of the asset, as well as changes in the way these earnings are valued. A fruitful avenue for future work may be an examination of the extent to which asset revaluations reflect increases in earning potential, and therefore may reduce the need for future saving, as opposed to transitory changes in the way in which these earnings are discounted (eg, Lettau and Ludvigson, 2004).

This study finds that when the flow measures of both household and national saving rates are adjusted for spending which may more appropriately be defined as investment and the impact of inflation, estimated saving rates are significantly increased. The unadjusted rate for household saving was -4.1% of household disposable income between 1996 and 2011; for the same period the net national saving rate was 2.8% of GDP. After incorporating the adjustments the estimated saving rates were 0.3% and 11.8% respectively.

Similar findings have emerged from other countries. For example, in the case of the USA, Boskin (1998, pp.30-31) observes:

While the United States has a saving rate which is low by historical and international standards, that saving rate is substantially higher when more comprehensive measures of saving are developed. While there are substantial difficulties in developing such augmented measures of national saving, various data sources and estimation methodologies all conclude that adjustments for net saving in durables, government capital, capital gains and losses, revaluations, etc. are all substantial.

A potentially significant adjustment arises from the stock of wealth that has accumulated in the New Zealand Superannuation Fund. To the extent that this is viewed as funds contributed by households through taxation for the provision of retirement income in future, it represents savings by households. On average the stock measure of savings is some 2.1 percentage points of household disposable income higher between 2002 and 2011, when allowance is made for the net wealth accumulated in the NZSF.

Statistics New Zealand is constantly seeking improvement in the collection and coverage of the data it publishes. Important revisions have been made to the household saving rates. The consequence of these is that some of the very extreme levels of household dissaving seen between 2004 and 2009 have been revised such that the current estimates of negative saving by households from the household income and outlay account are very much more modest. Over this period the annual average change was an improvement in the saving rate of households of over 7 percentage points of disposable income. These revisions underscore the importance for the policy debate to be grounded in solid evidence, and for full cognizance of the limitations of the underlying data.

The finding that existing flow measures may not be capturing the full extent of savings in all likelihood does little to alter the relative position of New Zealand in international comparisons. Arguably, were similar adjustments made to other countries a largely comparable ranking would remain. The exception is when New Zealand (a significant debtor country) is compared with significant creditor countries (eg Norway, China, Hong Kong). In this case adjusting for the impact of inflation is likely to lead to a closing of the apparent gap in national saving rates by as much as 4 percentage points of GDP.

In short, this paper concludes that the actual level of the saving rate may not be quite as low as the standard estimates imply. It is not that they are in any sense “wrong:” It is simply that they may not necessarily capture the full richness of saving behaviour.

Bollard and Barrow (2012) review a number on measurement issues relating to national income. They conclude that potential revisions could add as much as 10 percent to New Zealand's official GDP (p.15). Other things equal this would imply a significant change in the estimated rate of national saving.[35]

This is not to imply that the augmented measures presented here are a complete picture of saving rates. Conceptual and practical issues remain, whether they be in aggregation across heterogeneous households, the impact of changes in the age distribution and the composition of households, the role of bequests, the measurement of all forms of physical, natural, human and social capital, the choice of appropriate deflators or the comprehensive revaluation of all forms of assets and liabilities. Judgement calls are still needed to deal with many of these issues.

Overall, however the results presented here suggest that broader measures of saving might well be considered as complements to the official series when analysing the saving behaviour of firms and households, and the net saving of the nation as a whole.