Abstract
This Selected Issues paper examines the external linkages of the New Zealand economy. Empirical results from vector autoregressive models suggest that economic activity in Australia tends to have more of a significant direct impact on New Zealand than does activity in the United States. Fluctuations in the U.S. GDP, however, appear to be transmitted to New Zealand indirectly through their effects on the Australian economy. Financial linkages also have been important components in transmitting shocks from Australia and the United States to the New Zealand economy.
II. Explaining New Zealand’s Savings Behavior1
1. The persistence of current account deficits (averaging about 4½ percent of GDP over the past decade) and the unfavorable comparison of New Zealand’s average national savings rate to the OECD average (15½ percent versus 21 percent) have prompted concerns that New Zealand has a “savings problem”. With rising government saving over most of the last decade, the main culprit is alleged to be private savings, particularly household savings. However, it is not clear whether savings in New Zealand can be characterized as “low”. Cross-country comparisons of savings rates can be misleading because such factors influencing savings as institutional arrangements, financial market development, and societal preferences (especially time preference) may vary sharply between countries. A better test of the existence of a “savings problem” in New Zealand’s case would be whether private and household savings behavior can be adequately explained by their fundamental determinants.
2. The analysis presented here based on econometric estimates of savings equations suggests there do not appear to be significant impediments or distortions in New Zealand that would give rise to a saving problem. A trend decline in the savings rate of New Zealand households is reasonably well explained by such fundamentals as higher public saving (i.e., partial Ricardian equivalence holds), higher government pension and income support transfers to individuals, increased household wealth, and improved access to credit. The long-term behavior of the private savings rate in New Zealand also is relatively well explained by its fundamental determinants, including higher net private foreign liabilities. However, the analysis illustrates the important effect that government pensions and income support transfers have on the level of savings. The government also has a significant effect on saving through the public provision of health care services.2 Hence, it could be the case that some kind of “savings problem” might exist if households were mistaken in their expectations that future pension and health care costs would be met by the government without major increases in taxes or reductions in benefits.
A. Recent Trends in Savings
3. Throughout the period since 1980, New Zealand’s gross national saving rate has been substantially below the average for OECD countries, with the differential widening during the 1990s (Figure 1). New Zealand, however, has not been the only important outlier from the OECD average. During the 1990s, in particular, both Australia and the United States had gross national saving rates significantly below the OECD average, although these countries managed to raise national savings over the period, in contrast to New Zealand. The key difference among the three countries was the behavior of public savings. New Zealand managed to raise the level of the public saving rate faster and to a higher level than Australia or the United States in the first half of the 1990s. Through the end of the 1990s, while the public saving rate leveled off in New Zealand, it continued to grow in Australia and the United States. Meanwhile, throughout the decade, all three countries experienced a somewhat similar trend decline in the household saving rate, which was offset in large part by some improvement in corporate savings. Although cross country comparisons of savings are fraught with difficulties owing especially to differences in institutional arrangements, financial market development, and time preferences, the similarities in the trend in the household saving rate in New Zealand, Australia, and the United States is interesting and may suggest that these trends reflect the influence of similar factors.
New Zealand: International Comparision of Savings
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
New Zealand: International Comparision of Savings
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
New Zealand: International Comparision of Savings
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
4. The private saving rate in New Zealand exhibited no significant trend over the period since 1980; however, as noted above, this masks sharp differences in the behavior of households and corporations. Household savings as a share of GDP has fallen throughout the period, declining from some 7 percent of GDP in 1980 to -1 percent of GDP in 2000 (Figure 2). In contrast, the corporate saving rate has been gradually increasing during this period, with a sharp spike upward between late 1980s and early 1990s during the period when major economic reforms were being implemented. Overall, the gap between the household and corporate saving rates has been increasing over time.
New Zealand: Household Savings Rates
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
New Zealand: Household Savings Rates
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
New Zealand: Household Savings Rates
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
5. Part of the explanation for the decline in household saving may relate to measurement issues. Household saving based on the National Income and Outlay Accounts (NIOA) is derived as a residual of current income and expenditure, and consequently, may be subject to substantial bias.3 Income and expenditure are very large components; therefore, measurement errors in each of these will be transmitted to and can be potentially magnified in the savings measure. There is also a potential bias owing to the classification of certain household expenditures as consumption and not investment. For example, expenditures on education and durable household goods are treated entirely as consumption, when they might be better classified as investment. Certain investments in land and natural resources, particularly by unincorporated businesses which are considered to be part of the household sector, are treated as expenditures, biasing household saving downward.4
6. Household saving can be defined more broadly to capture the change in real net wealth. Accordingly, an alternative savings measure can be derived from the household balance sheet accounts. On this basis, the saving rate expressed as the change in household net wealth as a share of disposable income is significantly larger than the NIOA measure (Figure 2). However, the balance sheet measure of household saving is also subject to significant measurement bias. As noted in Thorp and Ung (2001), New Zealand household liabilities are likely to be exaggerated since some portion of home mortgages effectively are used to fund small businesses.5 This measure of household savings may also be biased downward because: (i) real assets do not include household durables, (ii) investment in non-financial assets—such as forestry—are not captured, (iii) net farm wealth is not included, and (iv) the value of household investment not priced through the stock exchange are excluded.6 Nevertheless, both measures of the household saving rate suggest a similar trend decline over the period 1980-2001.
B. Determinants of Savings
7. An empirical test for the existence of impediments or distortions that may affect savings is whether the savings rate can be adequately explained by its long-term fundamental determinants. Empirical models of household savings are typically based on some form of life-cycle hypothesis.7 Generally, the household saving rate is expected to be positively correlated with such determinants as inflation and possibly with real interest rates. It is negatively correlated with the budget balance (or government savings, reflecting some degree of “Ricardian equivalence”), pension and other income support transfers from government, household net financial wealth, innovations in the financial markets that improve household access to credit, and an aging population. A model for private savings can be seen as an extension of the household model. Accordingly, the private saving rate could be specified as being related to most of the same variables as the household savings rate, plus some additional ones such as net foreign liabilities of the private sector (to capture access to credit of the corporate sector).
8. Separate equations were estimated to explain New Zealand’s household and private saving rates applying cointegration theory.8 These equations were estimated using annual data for the period 1982-2001.9 The estimation results suggest that the fundamental determinants adequately explain the long-term behavior of the household saving rate (Table 1 and Figure 3).10 The results indicate that tighter fiscal policy, increased government transfers to households through pensions and other income support programs, improved access by households to credit, higher household net financial wealth, and higher housing value explain most of the trend decline in the household saving rate.11
New Zealand Savings Equations
Estimated Equation for the Household Saving Rate Estimation period: 1982-2001
New Zealand Savings Equations
Estimated Equation for the Household Saving Rate Estimation period: 1982-2001
| Coefficient | Std. error | Prob. Value | |
|---|---|---|---|
| Intercept | 30.820 | 2.252 | 0.000 |
| Government balance | -0.283 | 0.089 | 0.007 |
| Pension and income support | -0.062 | 0.011 | 0.000 |
| Household net financial wealth | -0.082 | 0.009 | 0.000 |
| Housing value | -0.020 | 0.009 | 0.041 |
| Credit card debt outstanding | -0.033 | 0.002 | 0.000 |
New Zealand Savings Equations
Estimated Equation for the Household Saving Rate Estimation period: 1982-2001
| Coefficient | Std. error | Prob. Value | |
|---|---|---|---|
| Intercept | 30.820 | 2.252 | 0.000 |
| Government balance | -0.283 | 0.089 | 0.007 |
| Pension and income support | -0.062 | 0.011 | 0.000 |
| Household net financial wealth | -0.082 | 0.009 | 0.000 |
| Housing value | -0.020 | 0.009 | 0.041 |
| Credit card debt outstanding | -0.033 | 0.002 | 0.000 |
New Zealand: Estimates of Saving Rates
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
New Zealand: Estimates of Saving Rates
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
New Zealand: Estimates of Saving Rates
Citation: IMF Staff Country Reports 2003, 122; 10.5089/9781451830248.002.A002
9. The equation for the private savings rate also adequately explains its long-term movements. The equation suggests that private sector net foreign liabilities act as a fairly significant substitute for domestic savings. Coefficients for the variables for pension and other income support transfers to the private sector, dependency ratio, and easier access to credit from financial innovation have the expected signs and are statistically significant. Inflation has a positive effect as the private sector tries to offset a decline in the real value of non-indexed assets and to compensate for the increased uncertainty regarding future income. The estimated values from the equation suggest that there remains some short-term noise in the movements in the private saving rate, particularly during the period from the late-1980s to the early-1990s—when economic reforms were underway. The fitted values are lower than the actual saving rates during this period which could suggest that corporations were spending more cautiously given the uncertainties associated with the reforms.12
10. The empirical evidence points to the significant influence on savings stemming from government pension and other income support transfers. Government financed health care also can be thought of as having a significant impact on household savings, but it cannot be easily modeled empirically. Thus, current household savings decisions can be seen as being heavily conditioned on the expectation that the government will meet its pension and health care obligations without significant changes in the future burden of taxation or a reduction in benefits. 13 Hence, although the empirical evidence indicates that household and private saving behavior are well explained by their long-term determinants, a significant distortion could exist and savings could be “too low” if the government were not to meet its future pension and health care obligations on the terms that households expect.
Estimated Equation for the Private Saving Rate Estimation period: 1982-2001
Estimated Equation for the Private Saving Rate Estimation period: 1982-2001
| Coefficient | Std. error | Prob. Value | |
|---|---|---|---|
| Intercept | 123.99 | 14.117 | 0.000 |
| Pension and income support | -0.622 | 0.157 | 0.001 |
| Dependency ratio | -1.567 | 0.230 | 0.000 |
| Inflation | 0.083 | 0.043 | 0.074 |
| Private sector net foreign liabilities | -0.096 | 0.049 | 0.072 |
| Credit card debt outstanding | -0.036 | 0.013 | 0.012 |
Estimated Equation for the Private Saving Rate Estimation period: 1982-2001
| Coefficient | Std. error | Prob. Value | |
|---|---|---|---|
| Intercept | 123.99 | 14.117 | 0.000 |
| Pension and income support | -0.622 | 0.157 | 0.001 |
| Dependency ratio | -1.567 | 0.230 | 0.000 |
| Inflation | 0.083 | 0.043 | 0.074 |
| Private sector net foreign liabilities | -0.096 | 0.049 | 0.072 |
| Credit card debt outstanding | -0.036 | 0.013 | 0.012 |
ANNEX Data Sources and Definitions
The sample period for the estimations is from 1982 to 2001. The definition and sources for each variable are as follows:
Household saving rate is the ratio of household savings to personal disposable income. Sources: National Income and Outlay Accounts, Statistics New Zealand, New Zealand Treasury.
Private saving rate is the ratio of private sector savings (gross national savings minus public savings) to private sector disposable income which is derived as the sum of household disposable income and net of tax operating surplus plus subsidies. Sources: National Accounts, Statistics New Zealand, New Zealand Treasury, and staff estimates.
Government balance is the ratio of central government balance to GDP. Source: IMF World Economic Outlook database.
Pension and other income support is the ratio of social security and welfare payments per recipient to the per capita household disposable income. In the private savings equation, it is computed as a ratio to total government current expenditure. Sources: Social Welfare Benefits, Statistics New Zealand, and IMF Government Financial Statistics database.
Household net financial wealth is the ratio of household financial assets less liabilities to household disposable income. Source: Thorp (2002).
Housing value is the ratio of the value of housing to household disposable income. Source: Thorp (2002).
Credit card outstanding is the ratio of credit card debt outstanding to disposable income. For households, the credit card debt outstanding was estimated by applying the average ratio of household credit card debt outstanding to total credit card debt outstanding during the period for which such data are available (from June 2000 to October 2002). Source: Monetary aggregate data, Reserve Bank of New Zealand.
Dependency ratio is the ratio of population under 15 years and over 65 years to the working age population. Sources: Statistics New Zealand, New Zealand Treasury.
Inflation equals the percentage change in the consumer price index. Source: Statistics New Zealand.
Private sector net foreign liabilities equals the cumulative current account balance net of government overall balance as a ratio to GDP. Sources: Statistics New Zealand, IMF Government Financial Statistics database, and staff estimates.
Callen, T., 1998, Savings Trends and Issues” IMF Staff Country Report No. 98/3.
Cerisola, M. and P. De Masi, 1999, “Determinants of the U.S. Personal Saving Rate”, IMF Staff Country Report SM/99/101.
Choy, W., 2000, “Determinants of New Zealand National and Household Saving Rates: A Cointegration Approach”, Unpublished paper, New Zealand Treasury.
Claus, I. and G. Scobie, 2002, “Saving in New Zealand: Measurement and Trends”, New Zealand Treasury Working Paper, 02/02.
Claus, I. and G. Scobie, 2001, “Household Net Wealth: An International Comparison”, New Zealand Treasury Working Paper, 01/19.
Edwards, S. (1996), “Why Are Latin America’s Saving Rates so Low? An International Comparative Analysis”, Journal of Development Economics, Vol. 51, pp. 5–44.
Gibson, J. and G. Scobie, 2001, “Household Saving Behavior in New Zealand: A Cohort Analysis”, New Zealand Treasury Working Paper, 01/18.
Joint Working Group, 1999, “Saving Rates and Portfolio Allocation in New Zealand”, New Zealand Treasury Working Paper 99/9.
Masson, S., T. Bayoumi and H. Samiei, (1998), “International Evidence on the Determinants of Private Saving”, The World Bank Economic Review, Vol. 12, No. 3, pp. 483–501.
Phillips, P. and B. Hansen, 1990, “Statistical Inference in Instrumental Variables Regression with 1(1) Processes”, Review of Economic Studies, Vol. 57, pp. 99–125.
Thorp, C., 2002, “Developments in Credit Markets Over Two Decades”, Reserve Bank of New Zealand: Bulletin, June, Vol. 65, No. 2, pp. 38–50.
Thorp, C., and B. Ung, 2001, “Recent Trends in Household Financial Assets and Liabilities”, Reserve Bank of New Zealand: Bulletin, June, Vol 64, No. 2, pp. 14–24
Prepared by Uma Ramakrishnan (Ext. 3-5413), who is available to answer questions.
In the absence of publicly provided health care, savings would generally be expected to be higher, as households would have to save more for precautionary purposes.
For detailed discussions of potential bias in measuring household savings, see Joint Working Group (1999), Claus and Scobie (2001 and 2002).
Another technical factor contributing to the low level in household savings is the definitional change in national savings between the 1968 System of National Accounts (SNA) and the revised 1993 version of the SNA. However, comparison of available data by both definitions (for the period from 1987 to 1999) indicates that the savings level shifted but the underlying trend did not change. In this paper, the 1968 SNA definition is used since a longer time series is available.
Some 85 percent of household liabilities are housing loans. Estimates of household loans used for business purposes are in the range of 10-20 percent of housing loans (whereby the business borrowing is secured on residential property).
For 2000, the estimated net real farm wealth, was about 17 percent of current net wealth (around NZ$40 billion).
See Edwards (1996) and Masson, Bayoumi, and Samiei (1998) for a review of the literature. Choy (2000) provides the most recent empirical estimates for New Zealand.
The equations were estimated using the Phillips-Hansen fully modified OLS (FMOLS) procedure. This method estimates the long-run parameters by correcting for serial correlation in the residuals without having to explicitly specify the dynamics of the model. It is a valid procedure when there exists a single cointegration equation and when the explanatory variables are not themselves cointegrated. For details, see Phillips and Hansen (1990).
Explanations on the derivation of the variables used and sources for the data are provided in the Annex.
Inflation and the real interest rate variables were dropped in the final specification of the household savings equation because they were statistical insignificant, to keep a parsimonious specification.
The pension and income support variable is defined in a manner that the dependency ratio is effectively imbedded in it.
Another explanation could be that corporate profits may have been “overstated” in periods of high inflation from larger depreciation allowances and inventory valuation. This might also explain why inflation is a significant determinant of private savings but not household savings.
Gibson and Scobie (2001) find that savings behavior varies across households and a very large share of total household savings comes from a small number of high-income households. There is also some preliminary evidence in the paper that the tax and benefit system has treated different cohorts differently which may partly explain low savings rates among some cohorts.
