Abstract
This Selected Issues paper analyzes the effect of international migration on unemployment in New Zealand. The empirical results in this paper suggest that net migration inflows give rise to a fall in the unemployment rate. The paper estimates a system of equations including the unemployment rate, real wage, net migration rate, and labor force participation rate, taking into account the interdependence of the variables. It also examines the impact of exchange rate volatility on export firms’ decisions to hedge foreign exchange exposure.
IV. Household Indebtedness and Monetary Policy23
“…the fact that the household sector is carrying so much more debt means that monetary policy is, in some sense, a more powerful weapon and has to be used sparingly and delicately.” 24
A. Introduction
1. Household debt has increased substantially in New Zealand, which may affect macroeconomic behavior, including the transmission of monetary policy. The indebtedness of households has doubled in the last 12 years, to reach an estimated 137 percent of disposable income at the end of 2004. Research by the Reserve Bank of New Zealand (RBNZ) concludes that household cash flows will be more interest sensitive, and hence, consumption will likely be more responsive to interest rates (Hull, 2003). A recent BIS survey similarly concluded that rising household debt makes the household sector more sensitive to shocks to interest rates, income, and asset prices (Debelle, 2004). This suggests that one implication of the rise in household leverage is that central banks may be able to achieve their macroeconomic policy goals with smaller or more gradual adjustments in interest rates.
2. This chapter finds some empirical evidence that consumer spending has become more interest sensitive as household indebtedness has increased. In contrast with most previous models of consumption in New Zealand, this paper finds a significant effect from real interest rates on consumption. Consistent with the hypotheses of Hull and Debelle, the strength of this effect appears to have increased as household debt has risen—a model allowing for the interaction of household indebtedness with real interest rates provides greater explanatory power for consumption than a model without this interaction.
3. Nonetheless, the recent rise in households’ equity in housing could moderate the interest rate sensitivity of consumption looking forward. As discussed below, recent U.S. research confirms that households can better smooth the impact of income shocks on their consumption when collateral for borrowing is more abundant. Housing equity has risen by some 100 percentage points of income in recent years in New Zealand, so the scope for households to smooth the effects of shocks to their disposable income due to changes in interest rates may have risen significantly. While housing equity was not found to affect interest rate responses in the econometric analysis, this may be due to the fact that the large increase in housing equity occurred only at the end of the sample period.
B. What Has Happened to Household’s Balance Sheets?
4. Boosted by rising housing prices, household net worth has improved even as debt burdens have risen.25 Household debt was about 50 percent of income in the first half of the 1980s. Following the financial deregulation in the mid-1980s, including the removal of quantitative credit controls, the household debt ratio rose steadily to reach 100 percent by the end of the 1990s. The pace of increase picked up in 2002–04, lifting debt to an estimated 137 percent of disposable income at the end of 2004. In contrast, household financial assets, including listed equities, have been quite stable at about 180 percent of disposable income since 1993. Housing loans account for almost 90 percent of household debt, and the recent strong growth in debt has been linked to the rise in housing prices. But as only one-fifth of the rise in housing values was offset by an increase in household debt, household net worth also rose substantially.
Household Balance Sheet
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New ZealandHousehold Balance Sheet
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New ZealandHousehold Balance Sheet
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand5. The trends in New Zealand’s household indebtedness are similar to those in a number of other countries. Differences in data compilation suggest that cross-country comparisons should focus on large differences in levels, and differences in trends over time.26 Nonetheless, it can be said that compared with Canada, the U.K., and the U.S., household debt levels in Australia and New Zealand were relatively low in the early 1990s. Debt burdens have risen in all these countries, but rose most quickly in Australia and New Zealand, so household debt levels in 2004 are more similar.27
Household Debt to Disposable Income 1/
(In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Reserve Bank of New Zealand and Reserve Bank of Australia.1/ Only data for Australia and USA exclude unincorporated enterprises.Household Debt to Disposable Income 1/
(In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Reserve Bank of New Zealand and Reserve Bank of Australia.1/ Only data for Australia and USA exclude unincorporated enterprises.Household Debt to Disposable Income 1/
(In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Reserve Bank of New Zealand and Reserve Bank of Australia.1/ Only data for Australia and USA exclude unincorporated enterprises.6. In contrast, debt service burdens have been relatively stable since the early 1990s, as in other countries. Debt service burdens peaked in the late 1980s and early 1990s in countries other then the United States. But since the early 1990s there has been little change in the trend level of interest burdens. In New Zealand, even though the household debt ratio almost doubled from 1990 to 2001, household’s interest payments remained just under 9 percent of disposable income. Nonetheless, interest payment ratios of both Australian and New Zealand households have risen recently, to levels somewhat above earlier highs.
Household Interest Payments 1/
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Reserve Bank of New Zealand and Reserve Bank of Australia.1/ Only data for Australia and USA exclude unincorporated enterprises.Household Interest Payments 1/
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Reserve Bank of New Zealand and Reserve Bank of Australia.1/ Only data for Australia and USA exclude unincorporated enterprises.Household Interest Payments 1/
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Reserve Bank of New Zealand and Reserve Bank of Australia.1/ Only data for Australia and USA exclude unincorporated enterprises.7. The decline in nominal interest rates made possible by lower inflation accounts for these divergent trends in debt and debt service ratios. From an average of almost 18 percent in the second half of the 1980s, nominal mortgage lending rates fell to 11 percent in the first half of the 1990s, and have averaged 7¾ percent in the first half of this decade. This 10 percentage point reduction in nominal interest rates is the primary factor accounting for the broad stability in the interest burden despite the rise in debt levels.28 Indeed, without this decline in interest rates, the large scale rise in debt levels would probably not have been feasible, as the higher resulting debt service burden might have increased default rates significantly.29
Household Interest Burden and Interest Rates
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New ZealandHousehold Interest Burden and Interest Rates
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New ZealandHousehold Interest Burden and Interest Rates
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand8. Financial liberalization also contributed to the rise in debt levels. Innovations in mortgage financing followed the financial deregulation and elimination of capital controls in the mid-1980s. These innovations include lower down payments—up to 95 percent of a property’s value can be borrowed—and lower transactions costs, e.g., line-of-credit home equity mortgages.30 However, increases in household gearing ratios have accounted for a relatively modest part of the increase in debt, e.g., the average loan-to-value ratio of housing mortgages of about 28 percent in mid-2004 is only a little above its level in the early 1990s. Rather, by relaxing financing constraints on those purchasing housing, and by making equity in housing more accessible for financing other needs, financial innovations have likely raised the demand for housing, raising housing prices and thereby providing additional collateral for household borrowing.
Household Gearing Indicators
(In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand and staff estimates.1/ Housing mortages in percent of the value of housingHousehold Gearing Indicators
(In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand and staff estimates.1/ Housing mortages in percent of the value of housingHousehold Gearing Indicators
(In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand and staff estimates.1/ Housing mortages in percent of the value of housing
Household Debt and Housing Equity
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand and staff estimates.1Value of housing net of debt secured by housing.Household Debt and Housing Equity
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand and staff estimates.1Value of housing net of debt secured by housing.Household Debt and Housing Equity
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Reserve Bank of New Zealand and staff estimates.1Value of housing net of debt secured by housing.C. How Might Higher Household Debt Affect Monetary Policy Transmission?
9. Household debt levels could affect the strength of a number of the channels through which monetary policy influences consumer spending:
Inter-temporal substitution effects may weaken with higher debt levels. Interest rates affect the tradeoff between consuming now or in the future, where spending on consumer durables may be especially interest sensitive. In practice, the strength of the substitution effects is likely to depend on household’s ease of access to credit. A higher debt level could reduce access to new credit, reducing the strength of the response of consumption to interest rate cuts. But such a moderation in interest sensitivity will be smaller if lenders focus more on debt service than debt levels, as an interest rate cut would increase the exposure lenders are willing to accept.
Income effects on consumption will become more negative at higher debt levels. A rise in household indebtedness means that an increase in interest rates will result in a larger increase in household interest payments than before. Hence, the negative income effect on the consumption of debtor households will be larger, while the size of the positive impact on the consumption of creditor households is unchanged. The extent to which this income effect becomes more negative will be greater if debtors have a higher marginal propensity to consume than creditors, which seems likely as the spending of some debtors may be constrained by limits on their access to credit
Wealth or financing effects may be affected asymmetrically by higher debt. In both New Zealand and Australia, the link between consumption and household net worth has been evident in recent years.31 Hence the effects of interest rates on house prices may be an important monetary transmission mechanism. Financial liberalization may make asset prices more sensitive to interest rates (Iacoviello and Minetti, 2004.) But higher debt levels could have an asymmetric effect on asset price sensitivity to interest rates. High debt may tend to moderate the increase in housing prices in response to interest rate cuts, by limiting the resulting expansion in credit and hence in demand for housing. But high debt could possibly exacerbate the negative impact on housing prices of interest rate hikes, especially if it is associated with a larger share of households who treat housing as an investment.32
Household Interest Flows
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Statistics New Zealand.1/ Includes income from pension and insurance accounts.Household Interest Flows
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Statistics New Zealand.1/ Includes income from pension and insurance accounts.Household Interest Flows
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Source: Statistics New Zealand.1/ Includes income from pension and insurance accounts.
Australia: Household Saving and Net Worth
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Australian Bureau of Statistics, RBA, and staff estimates.Australia: Household Saving and Net Worth
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Australian Bureau of Statistics, RBA, and staff estimates.Australia: Household Saving and Net Worth
(In percent of disposable income)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Australian Bureau of Statistics, RBA, and staff estimates.10. The short-term structure of household debt in New Zealand increases the potential importance of income effects. While only 26 percent of mortgage lending was on a floating rate basis in November 2004, the bulk of mortgage loans have rates which are fixed for relatively short periods, with 60 percent of all mortgages on rates fixed for 2 years or less. Thus interest payments by New Zealand households are relatively sensitive to changes in official interest rates. Nonetheless, it is observed that some households make larger principal payments when interest rates are low, giving them a buffer when interest rates rise (Debelle, 2004). Hence, the impact of rate hikes may depend on the extent to which they are within the range of cyclical variation allowed for by such buffers.
11. The capacity of households to smooth their consumption in the face of higher interest payments also depends on the level of housing equity. Lustig and Nieuwerburgh (2004) find that households in the U.S. are better able to smooth their consumption in the face of shocks to regional income if housing wealth is higher, increasing households’ collateral for borrowing. In addition, Hurst and Stafford (2004), find that U.S. households are more likely to refinance their mortgage to access home equity if they receive a negative income shock and they have few liquid assets. It therefore seems likely that households with greater equity in housing will be better able to borrow to smooth the impact of higher interest payments on their consumption, so higher housing equity could partly offset the impact of rising indebtedness on the interest sensitivity of consumption.
D. Has Consumption Become More Interest Sensitive in New Zealand?
12. There is currently little international evidence on the impact of changes in household debt on the interest sensitivity of household spending. There is a literature analyzing the impact of financial liberalization on household consumption. For example, Barrell and Davis (2004) find a shift in consumption behavior in G-7 countries due to financial liberalization, with a decline in short-run income elasticities, and a rise in short-run wealth and interest rate elasticities. Before discussing the econometric analysis of consumption in New Zealand, the following presents some graphical analysis for Australia and New Zealand, as results for Australia are also likely to be relevant for New Zealand.
13. Household interest payments appear to have significant effects on savings in both Australia and New Zealand. In both countries, changes in the interest payment burden are found to be positively correlated with changes in the household savings rate with a lag of about 5 quarters. This indicates the lags in the effect of interest rates on consumption are relatively long, as interest payments themselves respond with a lag to changes in official interest rates. Changes in savings rates in New Zealand have a correlation of 0.61 with changes in the interest payment burden, which is modestly higher than the 0.56 correlation with changes in the interest rate. Since changes in the interest burden depend on the level of indebtedness, this is suggestive that the impact of interest rates on consumption spending has risen with the level of household indebtedness.
Australia: Household Savings and Interest Payments
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Australian Bureau of Statistics, RBA, and staff estimates.Australia: Household Savings and Interest Payments
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Australian Bureau of Statistics, RBA, and staff estimates.Australia: Household Savings and Interest Payments
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Australian Bureau of Statistics, RBA, and staff estimates.
New Zealand: Household Savings and Interest Payments
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Statistics New Zealand, RBNZ, and staff estimates.New Zealand: Household Savings and Interest Payments
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Statistics New Zealand, RBNZ, and staff estimates.New Zealand: Household Savings and Interest Payments
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Sources: Statistics New Zealand, RBNZ, and staff estimates.14. That said, previous econometric analysis has typically not found consumption to be very interest sensitive in New Zealand. Goh and Downing (2002) report that earlier analysis of consumption in New Zealand did not find a statistically significant effect from interest rates. Their analysis, which took advantage of newly available RBNZ data on household balance sheets, found no evidence of interest rate effects in the short-run, but did find some evidence of effects after 12–18 months. This finding is consistent with above graphical evidence, and indicates that the econometric analysis of consumption behavior will need to allow for relatively long lags.
15. An error correction equation for consumption is estimated based on a standard theoretical model. Consumption (c) is determined by net wealth (nw), labor income (le), and real interest rates (r) in the long run. The effect of interest rates may depend on the level of indebtedness (d):
16. To test whether the level of household indebtedness affects the sensitivity of consumption to interest rates, two versions of the model are estimated (see Annex IV.1 for details). Model A includes the real interest rate as in a standard consumption equation, while model B allows for the interaction between the real interest rate and household debt relative to disposable income. Real interest rates are found to affect consumption with lags of 3 to 5 quarters in both models. There is little sign of instability in the estimated interest rate coefficient based on recursive estimates of model A. However, model B (R2=0.734) fits modestly better than model A (R2=0.707). When both the real interest rate (r) and the interaction between interest rates and debt (r*d) are included in the same equation, the coefficient on r has an incorrectly positive coefficient, and is statistically insignificant, so model B is statistically preferred.33
Actual and Fitted Values, Model B
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Actual and Fitted Values, Model B
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Actual and Fitted Values, Model B
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
17. The sensitivity of consumption to interest rates has increased substantially according to these estimates. The figure presents the simulated effect on consumption in Model B of a 1 percentage point increase in real interest rates made in 25 basis point steps per quarter. With debt approaching 140 percent of disposable income, a 1 percentage point rise in real interest rates is estimated to reduce consumption by 2.3 percent after 8 quarters from the initial rate increase. The final impact is up from an estimated reduction of 0.8 percent when the household debt ratio was 50 percent in the late 1980s. If these estimates are broadly correct, the earlier failure to find significant interest rate effects on consumption may be because these effects were relatively small in earlier samples.
Simulated Impact on Consumption of Raising Interest Rates by 1% in Total in Quarters 1 to 4 (In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Simulated Impact on Consumption of Raising Interest Rates by 1% in Total in Quarters 1 to 4 (In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
Simulated Impact on Consumption of Raising Interest Rates by 1% in Total in Quarters 1 to 4 (In percent)
Citation: IMF Staff Country Reports 2005, 153; 10.5089/9781451830309.002.A004
References
Basdevant, Olivier, Nils Björksten, and Özer Karagedikli, 2004, “Estimating a time varying neutral real interest rate for New Zealand,” Reserve Bank of New Zealand Discussion Paper 2004/01.
Barrell, R. and E.P. Davis, 2004, “Financial Liberalization, Consumption, and Wealth Effects in 7 OECD Countries”, NIESR Discussion Paper No. 247.
Debelle, Guy, 2004, “Macroeconomic implications of rising household debt,” BIS Working Paper No. 153.
Goh, Khoon and Richard Downing, 2002, “Modeling NZ Consumption Expenditure Over the 1990s,” New Zealand Treasury Working Paper 02/19.
Hull, Leslie, 2003, “Financial deregulation and household indebtedness,” Reserve Bank of New Zealand Discussion Paper 2003/01.
Hurst, Eric and Frank Stafford, 2004, “Home is Where the Equity Is: Mortgage Refinancing and Household Consumption,” Journal of Money Credit and Banking Vol. 6, No. 6 (December).
Iacoviello, M. and R. Minetti, 2004, “Financial Liberalization and the Sensitivity of House Prices to Monetary Policy: Theory and Evidence,” Mimeo, Boston College.
Lustig, Hanno and Stijn Van Nieuwerburgh, 2004, “Housing Collateral and Consumption Insurance Across US Regions,” National Bureau of Economic Research Working Paper 10505.
Thorp, Clive and Bun Ung, 2000, “Trends in household assets and liabilities since 1978,” RBNZ Bulletin, June.
Whitley, John, Richard Windram, and Prudence Cox, 2004, “An empirical model of household arrears,” Bank of England Working Paper No. 214.
ANNEX IV.1: An Error-Correction Model of Household Consumption
Data sources:
| c | Household consumption, constant 1995/96 prices, seasonally adjusted, Statistics NZ code: SNCQ.S2RP30ZES | |
| le | Total gross wage and salary earnings, seasonally adjusted, Statistics NZ code: EESQ.SHAZS, | |
| nw | Household net worth, RBNZ. Quarterly data available for March 1991 to June 2004. Covers housing assets in addition to financial assets and liabilities. | |
| h | Housing asset value, RBNZ. Quarterly data available for March 1991 to June 2004. | |
| r | Real mortgage interest rate (i - πe) | |
| i = Effective mortgage interest rate, RBNZ | ||
| πe = 1 year ahead, mean household inflation expectation, Marketscope survey | ||
| d | Ratio of household debt to disposable income, RBNZ | |
| c | Household consumption, constant 1995/96 prices, seasonally adjusted, Statistics NZ code: SNCQ.S2RP30ZES | |
| le | Total gross wage and salary earnings, seasonally adjusted, Statistics NZ code: EESQ.SHAZS, | |
| nw | Household net worth, RBNZ. Quarterly data available for March 1991 to June 2004. Covers housing assets in addition to financial assets and liabilities. | |
| h | Housing asset value, RBNZ. Quarterly data available for March 1991 to June 2004. | |
| r | Real mortgage interest rate (i - πe) | |
| i = Effective mortgage interest rate, RBNZ | ||
| πe = 1 year ahead, mean household inflation expectation, Marketscope survey | ||
| d | Ratio of household debt to disposable income, RBNZ | |
To convert le, nw, and h into real values, they are deflated by the implicit price of household consumption (SNCQ. S2NP30ZES / SNCQ. S2RP30ZES)
Econometric results (OLS estimates):
Dependent: Δlog(c)
Sample: 1993Q2 to 2004Q2
Observations: 45
Dependent: Δlog(c)
Sample: 1993Q2 to 2004Q2
Observations: 45
| Model A: Fixed Interest Effect | Model B: Variable Interest Effect | |||
| Variable | Coefficient | Std. Error | Coefficient | Std. Error |
| log(c)-1 | -0.8214 | 0.1222** | -0.8229 | 0.1135** |
| log(le)-1 | 0.3814 | 0.0801** | 0.3713 | 0.0755** |
| log(nw)-1 | 0.3143 | 0.0512** | 0.3947 | 0.0582** |
| (r-3+r-5)/2 | -0.8854 | 0.1664** | -- | -- |
| ((r*d)-3+(r*d)-5)/2 | -- | -- | -1.3417 | 0.2261** |
| Δlog(le) | 0.4180 | 0.0882** | 0.4166 | 0.0839** |
| Δlog(h) | 0.2191 | 0.0518** | 0.2246 | 0.0493** |
| Δ8log(nw)-1 | -0.1690 | 0.0271** | -0.2145 | 0.0307** |
| Intercept | 2.1481 | 0.3168** | 1.6506 | 0.2184** |
| S.E. | 0.48% | 0.46% | ||
| R2 | 0.707 | 0.734 | ||
| DW | 1.96 | 1.97 | ||
Dependent: Δlog(c)
Sample: 1993Q2 to 2004Q2
Observations: 45
| Model A: Fixed Interest Effect | Model B: Variable Interest Effect | |||
| Variable | Coefficient | Std. Error | Coefficient | Std. Error |
| log(c)-1 | -0.8214 | 0.1222** | -0.8229 | 0.1135** |
| log(le)-1 | 0.3814 | 0.0801** | 0.3713 | 0.0755** |
| log(nw)-1 | 0.3143 | 0.0512** | 0.3947 | 0.0582** |
| (r-3+r-5)/2 | -0.8854 | 0.1664** | -- | -- |
| ((r*d)-3+(r*d)-5)/2 | -- | -- | -1.3417 | 0.2261** |
| Δlog(le) | 0.4180 | 0.0882** | 0.4166 | 0.0839** |
| Δlog(h) | 0.2191 | 0.0518** | 0.2246 | 0.0493** |
| Δ8log(nw)-1 | -0.1690 | 0.0271** | -0.2145 | 0.0307** |
| Intercept | 2.1481 | 0.3168** | 1.6506 | 0.2184** |
| S.E. | 0.48% | 0.46% | ||
| R2 | 0.707 | 0.734 | ||
| DW | 1.96 | 1.97 | ||
Specification tests of Model B:
| P-value | |||
| Autocorrelation: | F(1,36) = | 0.1615 | 0.6902 |
| F(4,33) = | 3.3002 | 0.0222* | |
| ARCH: | F(4,29) = | 0.3245 | 0.8592 |
| Normality test: | Chi2(2) = | 2.5247 | 0.2830 |
| Heteroscedasticity | F(14,22) = | 1.0016 | 0.4843 |
| Functional form: | F(1,36) = | 0.1565 | 0.6947 |
| Out-of-sample Chow-tests for stability: | F(4,33) = | 0.8620 | 0.4969 |
| F(8,29) = | 1.0165 | 0.4458 |
| P-value | |||
| Autocorrelation: | F(1,36) = | 0.1615 | 0.6902 |
| F(4,33) = | 3.3002 | 0.0222* | |
| ARCH: | F(4,29) = | 0.3245 | 0.8592 |
| Normality test: | Chi2(2) = | 2.5247 | 0.2830 |
| Heteroscedasticity | F(14,22) = | 1.0016 | 0.4843 |
| Functional form: | F(1,36) = | 0.1565 | 0.6947 |
| Out-of-sample Chow-tests for stability: | F(4,33) = | 0.8620 | 0.4969 |
| F(8,29) = | 1.0165 | 0.4458 |
There is negative residual autocorrelation at the 2nd and 3rd lags. This may reflect the measurement error in the levels of consumption, which is evident in the high negative autocorrelation in the differences of consumption. The consequence is that reported standard errors bias are biased up rather than down.
Long-run Solution:
18. The estimated long-run coefficients on income and wealth add to 0.93, but the homogeneity restriction that the elasticities on income and wealth sum to unity is not accepted statistically. The lack of homogeneity may reflect problems with gross labor earnings as a proxy for income, and possibly also the omission of significant assets in the measurement of net household wealth, e.g., unlisted business equity and nonresidential land, especially farms. The long-run elasticity on income appears relatively low, although income which is not consumed will also add to wealth, leading to a larger total impact from higher income in the long-run.
Short-run Dynamics
19. The high error correction coefficient indicates deviations from equilibrium consumption levels are not long lasting. The long lag (8 quarters) on the change in wealth has the effect of delaying the full effect of a rise in wealth on consumption by 2 years, where the initial effect is about half of the long-run effect. Hence the strong rises in net wealth due to housing price increases in 2002–03 may continue to support consumption during 2005 even if housing prices are flat. The significance of the contemporaneous change in the value of housing (Δlog(h)) may reflect a consumer confidence effect, as higher price increases are also associated with greater turnover in housing.
Prepared by Craig Beaumont.
Reserve Bank of Australia Governor Ian Macfarlane, in testimony to House of Representatives, Commonwealth of Australia, February 18, 2005.
Household balance sheet data are prepared by the RBNZ (Thorp and Ung, 2000). Household debt includes financing for small business secured on housing. Nonfinancial assets do not include the value of equity in unlisted companies, or the value of nonresidential land, e.g., farms.
These cross-country data are collected from national sources by the Reserve Bank of Australia. The key differences in the data coverage are that data for Australia and the U.S. exclude small business loans secured on real estate, while data for New Zealand and the other countries do not make this exclusion.
From a broader cross-country perspective, developments in Australia and New Zealand have similarities with those in Spain. In contrast, household debt burdens were relatively stable in the 1990s in France and Japan, increased modestly in Germany and Sweden, and rose to substantially higher levels in Denmark and the Netherlands.
Basdevant, et al, 2004, find that a decline in the neutral or equilibrium real interest rate also contributed to the decline in the trend level of interest rates in New Zealand.
Whitley, et al, 2004, find the debt service burden is found to be a key factor driving arrears on mortgages in the UK.
Leslie, 2003, provides further discussion of financial liberalization in New Zealand.
A chart is presented for Australia, as Statistics New Zealand is reviewing the Household Income and Outlay Account data for New Zealand, but the broad pattern is similar.
It is notable that the 50 basis point tightening by the Reserve Bank of Australia at the end of 2003 was followed by a marked cooling in the housing market in 2004. Anecdotal reports indicate that household consumption responded more strongly due to higher debt levels.
Equations allowing for an interaction between housing equity and real interest rates, to allow for greater ability of households to smooth the income effects of changes in interest rates if housing equity is high, did not produce a statistically significant effect.
