Kingdom of the Netherlands—Netherlands: Selected Issues
Author: International Monetary Fund
Publication Date:
14 Jul 2005
eISBN:
9781451829518
Language:
English
Keywords:
ISCR; CR; house price; wealth effect; consumer confidence; consumption data; Amsterdam stock exchange price index; utility function; Disposable income; Private consumption; Income; Unemployment rate
Download PDF (342.4 KB)

This Selected Issues paper for the Netherlands highlights that Dutch cost competitiveness deteriorated significantly in the years prior to 2004. The relatively poor economic performance during the new millennium contributed to these concerns generally, as did restrained export performance more specifically. The deceleration of private consumption growth started earlier than the deceleration of per capita real disposable income growth, and occurred at a time when the unemployment rate has been at a historic low and interest rates had declined.

Abstract

This Selected Issues paper for the Netherlands highlights that Dutch cost competitiveness deteriorated significantly in the years prior to 2004. The relatively poor economic performance during the new millennium contributed to these concerns generally, as did restrained export performance more specifically. The deceleration of private consumption growth started earlier than the deceleration of per capita real disposable income growth, and occurred at a time when the unemployment rate has been at a historic low and interest rates had declined.

II. Long-Run Household Consumption Equilibrium in the Netherlands4

A. Overview and Introduction

10. Private consumption growth in the Netherlands has been sluggish in recent years. Since the beginning of 2001, the average quarterly growth rate has been about ¼ percent as opposed to an average of nearly 1 percent in the second half of the 1990s. Given its share in final demand, changes in consumption have a large impact on GDP growth. This, as well as the disappointing growth performance so far this decade, explain why consumer behavior has moved to center stage in policy discussions.

11. Recent consumption behavior cannot be explained by developments in disposable income alone. The deceleration of private consumption growth started earlier than the deceleration of per capita real disposable income growth, and occurred at a time when the unemployment rate was at a historic low and interest rates had declined (Figures 1 to 3). A key related observation is that the average share of disposable income in GDP fell by over 2 percentage points when comparing 2001–04 with 1995–2000. This largely reflected a decline in property income. Corresponding to this, Dutch consumers cut consumption and savings in relation to GDP in roughly equal amounts.5

Figure 1.
Figure 1.

Netherlands: Consumption, Income, and Net Worth

(Average Annual Growth Rates)

Citation: IMF Staff Country Reports 2005, 225; 10.5089/9781451829518.002.A002

Sources: DNB; and IMF staff calculations.
Figure 2.
Figure 2.

Netherlands: Unemployment Rate

Citation: IMF Staff Country Reports 2005, 225; 10.5089/9781451829518.002.A002

Sources: DNB; and IMF staff calculations.
Figure 3.
Figure 3.

Netherlands: Interest Rates and Stock Market Returns

Citation: IMF Staff Country Reports 2005, 225; 10.5089/9781451829518.002.A002

Sources: DNB; and IMF staff calculations.

12. This background suggests looking at the role of wealth as an influence on private consumption behavior. In this connection, the Amsterdam stock exchange price index started to fall in the second quarter of 2001 (Figure 4). In addition, the ratio of per capita household net worth to income clearly started to decline toward end-2001, partly influenced by a slowing in the pace of house price increases (Figures 5 and 6). Consistent with historical evidence, consumer confidence rapidly deteriorated, more or less in tandem with the decline in stock prices (Jansen and Nahuis, 2003). The Dutch central bank has also associated the decline in consumer confidence to doubts about pension funds solvency, dissatisfaction with public services, and overall policy and political uncertainty (De Nederlandsche Bank, 2004).

Figure 4.
Figure 4.

Netherlands: Stock Market Price Index

Citation: IMF Staff Country Reports 2005, 225; 10.5089/9781451829518.002.A002

Sources: DNB; and IMF staff calculations.
Figure 5.
Figure 5.

Netherlands: Real House Prices

(Average Annual Growth Rates)

Citation: IMF Staff Country Reports 2005, 225; 10.5089/9781451829518.002.A002

Source: Kadaster.
Figure 6.
Figure 6.

Netherlands: Real Household Net Worth/ Income 1/

Citation: IMF Staff Country Reports 2005, 225; 10.5089/9781451829518.002.A002

Sources: DNB; and IMF staff calculations.

13. This paper focuses on the long-run determinants of and short-term outlook for Dutch total private consumption. A cointegration model is estimated to test the relevance of a number of factors suggested by the literature on consumption behavior, such as disposable income, net wealth, interest rates, inflation, unemployment (used as a proxy for uncertainty), and demographic variables.6 The empirical results suggest that the long-run determinants of Dutch total private consumption are, indeed, disposable income, net wealth, interest rates, and unemployment. Therefore, the reduction in consumption growth witnessed in the recent period would seem to have resulted from the deceleration of income growth, the fall in households’ net wealth, and economic uncertainty.7 The short-run results indicate a very small positive role for inflation (as measured by the consumption deflator).

B. The Standard Consumption Model

14. This paper uses the standard permanent income framework to explain consumption. It assumes that current consumption is explained by life-time resources or wealth, i.e., labor income and wealth resulting from financial and nonfinancial sources. The representative agent maximizes:

MaxEt{Vt=Στ≥0βτUt+τ}(1)

subject to

Σi=0∞(11+r)iCt+τ+Σi=1∞Bi,t+τ≤Σi=0∞(1+r)iBi,t+τ+Σi=0∞(11+r)iYt+τ=0(2)

valid for all τ≥0, and with a utility function as follows:

Ut=Ct1−γ−11−γ.(3)

Ct is the level of per capita real consumption, Bt are bonds that proxy here the stock of financial wealth, is the rate of return on each unit of that wealth between periods t-1 and t, and Yt is per capita real disposable income.8 Et is the mathematical expectation operator, Ut is an instantaneous utility function that is continuous, differentiable, and concave. The parameter β is the discount factor, and γ measures relative risk aversion (the curvature of the utility function).

15. The first order conditions combined with the budget constraint provide the backbone of the estimated consumption function. The first order conditions can be represented as follows:

Et[β(Ct+1Ct)−γ]=(11+r).(5)

Combining the first order conditions with the budget constraint, and linearizing around the steady state, leads to the solved-out consumption function:

Ct=A0+A1Yt+A2Bt+A3rt.(6)

To the extent that the economy is not at full employment, the estimated equation will test whether unemployment plays a role in explaining consumption. Changes in productivity, fiscal policy, and labor market measures act through real disposable income. In the estimation, the impact of all other variables, as discussed below, is captured by the vector Xt and the associated vector of coefficients ω.

C. Data and Data Issues

16. This study uses per capita real total private consumption. Nondurables consumption data are only available starting in 1995Q1. However, starting the estimation in 1995 would severely reduce its quality. Moreover, total private consumption is the relevant series in analyzing links to the stock market and wealth.9

17. Some proxies and data transformations were necessary. Per capita real household wealth was proxied using per capita real household net worth.10 Because household per capita disposable income and wealth are only available at an annual frequency, the series were transformed by the Dutch central bank into a quarterly frequency using the Lisman filter.11 Household net worth and 3-month interest rates were deflated by the private consumption deflator. All series were kindly provided by the Dutch central bank, except the unemployment rate, which was obtained from the OECD database.12 The sample period comprises 1983Q2–2004Q4. All series were seasonally adjusted using X11.

18. The series were tested for unit roots. In particular, the Elliot, Rothenberg, and Stock (1996) test was used. With a 95 percent confidence, all series except the unemployment rate contained a unit root (Table 1). In contrast, the first differences of the series were stationary. The unit root tests on the levels of the series were done including a constant and a deterministic trend; only a constant was included when testing the first differences of the series for unit roots. The number of lags for all tests were optimally chosen using the Schwarz information criterion.

Table 1.

Netherlands: Elliot, Rothenberg, and Stock Test for Unit Roots 1/

Statistics for ρ=0 (1983:Q2-2004:Q4)

article image
Source: IMF staff estimates.

Private consumption, disposable income, and household net worth are in per capita terms. These variables are also expressed in natural logarithms. Lags are determined according to Schwarz information criterion and by checking that the residuals are white noise.

The DFGLSτ has a null of unit root with a constant and a linear trend. The 5 percent critical value is -2.89.

The DFGLSτ has a null of unit root with a constant. The 5 percent critical value is -1.95.

19. Next, consumption, disposable income, wealth, interest rates, and the unemployment rate were tested for cointegration. The Johansen-Juselius test, corrected for small sample bias, strongly rejected the no-cointegration hypothesis using the λmax or the trace statistic at the 95 percent level (Table 2).13 Residuals were white noise. While the λmax statistic could not reject the null hypothesis of one cointegrating vector against the alternative of two, its value (25.42) was very close to the corrected 95 percent confidence value (25.83). This is consistent with the difference in results between the univariate unit root test, which suggests that the unemployment rate is stationary, and the multivariate test of stationarity, which rejects the hypothesis of stationarity for all variables, including the unemployment rate. The estimation of the long-run consumption equation included the unemployment rate; it improved the overall fit of the equation.

Table 2.

Netherlands: Johansen-Juselius Maximum Likelihood Test for Cointegration

article image
Source: IMF staff estimates. The models include a drift term in the variables but not in the cointegration space.

Column r refers to the number of cointegrated vectors.

The λ max and the trace statistics critical values are corrected for small samples using Cheung and Lai (1993).

This is a test of long-run exclusion of the relevant variable from the cointegration space. It is distributed as a chi square variable with r degrees of freedom.

The LM are Lagrange multiplier tests. The p values are in parentheses.

D. The Long-Run Elasticities of the Model

20. The econometric results support the view that sluggish private consumption so far this decade has been consistent with the behavior of some key consumption fundamentals (Figure 7). Equation (6) was estimated by the nonlinear dynamic least squares estimator of Phillips and Loretan (1991).14 The estimated form was:

Ct=A+θYt+ωXt+Σt=−kk(φΔYt+γΔXt)+ρ(A+θYt−1+ωXt−1)+ϵt,(7)

where Yt includes per capita real disposable income, per capita real net wealth and the interest rate, and Xt includes the unemployment rate and the inflation rate (measured by the consumption deflator). The last term is a serially uncorrelated error term with constant variance.

Figure 7.
Figure 7.

Netherlands: Consumption and Estimated Long-Run Consumption

(Levels; in natural logs)

Citation: IMF Staff Country Reports 2005, 225; 10.5089/9781451829518.002.A002

Sources: DNB; and IMF staff calculations.

21. The long-run estimates of consumption are reported in the equation below.

Ct=−3.82+0.10incomet+0.21wealtht−0.01intt−0.01unmeployementt,withρ=0.35.(8)

The t-statistics are -18.13, 2.19, 4.76, -2.62, -4.22, and 3.71, respectively. The equilibrium correction term (ρ) is strongly significant.15 All the variables are significant at the 95 percent level. While there is some evidence of multicollinearity between real per capita disposable income and the real per capita net worth measure, this does not seem to be a serious problem.16 Inflation is never significant in the long run.17 The real interest rate is always significant and negative. Uncertainty, as proxied by the unemployment rate, plays a statistically significant role in explaining long-run consumption, though its economic magnitude is much lower than income or wealth.

22. While the long-run income elasticity is low, wealth effects are economically and statistically significant. The low long-run income elasticity likely reflects the inclusion of property income in real disposable income and the use of total consumption rather than nondurables consumption in the estimation.18 This explanation is all the more likely because the share of durables in total consumption has been consistently increasing since 1995Q1.19 The wealth elasticity is somewhat higher than usual. The relatively large role of wealth suggests a liquid and efficient consumer credit market. Based on the sample average wealth-to-consumption ratio, the Dutch marginal propensity to consume out of real net wealth is about 5¼ cents to the euro. This is somewhat higher than the 4 cents to the dollar for the United States (for gross stock market wealth) and close to the United Kingdom’s 5 cents to the pound (for gross stock market wealth). The estimated wealth elasticity is consistent with results in Ludwig and Sløk (2002) who show that the Netherlands belongs to “market-based economies” to the extent that its wealth elasticity is comparable to Anglo-Saxon economies’ rather than to continental Europe’s bank-based economies (e.g., the wealth elasticity is less than 3 cents to the euro in France).20

23. Long-run consumption behavior has closely followed the behavior of its fundamentals, including the proxy for uncertainty. A comparison of long-run estimated consumption with actual consumption suggests that the latter caught up with the favorable fundamentals of the second half of the 1990s sometime in 2000. Since mid-2001, negative shocks to the factors that explain consumption reduced equilibrium long-run consumption, which reached a trough in 2003Q3. By the end of the sample period (Q4 2004), actual consumption appeared to have caught up with its long-run fundamentals.21

E. Concluding Remarks

24. This paper found that income, wealth, interest rates, and uncertainty (proxied by the unemployment rate) are the main determinants of total private consumption in the Netherlands. The relatively large role of real net wealth suggests a liquid and efficient credit market, facilitating consumption smoothing. It also suggests that Dutch consumers adjust their consumption quickly to persistent changes in their wealth.

25. With consumption having caught up with its fundamentals by the end of 2004, there are reasonable prospects for a pick up in consumption. In particular, private consumption can now be expected to accelerate with growth prospects improving and to the extent that the stock market and house prices boost wealth, and unemployment (and other sources of uncertainty generally) subside.

26. Further structural reforms could give a positive lift to consumption through other than mainstream channels. Consumer confidence is sensitive to developments in the stock market in the Netherlands. Thus, to the extent that structural reforms boost potential growth, consumption could not only be boosted directly, but also indirectly through confidence effects. In addition, enhancing the transparency of pension fund rules should reduce uncertainty about the present value of consumers’ expected income. Reduced uncertainty would increase the observed marginal propensity to consume out of income and wealth.

APPENDIX I: Detailed Econometric Results

Nonlinear Least Squares - Estimation by Gauss-Newton

Convergence in 3 Iterations. Final criterion was 0.0000009 < 0.0000100

Dependent Variable CONS

Quarterly Data From 1983:02 To 2004:04

Usable Observations 83 Degrees of Freedom 61

Total Observations 87 Skipped/Missing 4

Centered R**2 0.995528 R Bar **2 0.993988

Uncentered R**2 0.999999 T x R**2 83.000

Mean of Dependent Variable -6.015491650

Std Error of Dependent Variable 0.110469073

Standard Error of Estimate 0.008565472

Sum of Squared Residuals 0.0044754056

article image
article image
Cumulated Periodogram Test for Series RESGER2Maximum Gap = 0.1399 At Frequency 1.8162Approximate Rejection Limits1% = 0.2037 5% = 0.1700 10% = 0.1525Variables are:A0 = ConstantA1 = Real per capita disposable incomeA2 = Real household’s net worth per capitaA4 = 3-Month real interest rateA6 = Unemployment rateB1 = Change in incomeB2 = Change in net worthB3 = Consumption deflator inflationB4 = Change in 3-month interest ratesB6 = Change in unemployment rateC1 = Change in income, lagged onceC3 = Inflation, lagged onceC4 = Change in 3-month interest rates, lagged onceC6 = Change in unemployment rate, lagged onceD1 = Change in income, lagged twiceD3 = Inflation, lagged twiceD4 = Change in 3-month interest rates, lagged twiceD6 = Change in unemployment rate, lagged twiceE1 = Change in income, lead onceE4 = Change in 3-month interest rates, lead onceE6 = Change in unemployment rate, lead onceDELTA = Long-run equilibrium adjustment parameter

References

  • Allais, O., L. Cadiou, and S. Dées, 2002, Defining Consumption Behavior in a Multi-Country Model, Centre d’Etudes Prospectives et d’Informations Internationales, WP 01–02.

    • Search Google Scholar
    • Export Citation

  • Bayoumi, T., and H. Edison, 2003, Is Wealth Increasingly Driving Consumption? De Nederlandsche Bank NV, DNB Staff Reports No.11.

  • Boone, L., and N. Girouard, The Stock Market, the Housing Market and Consumer Behaviour, OECD Economic Studies No. 35.

  • Cheung, Y. W., and K. S. Lai, 1993, Finite Sample Sizes of Johansen’s Likelihood Ratio Test for Cointegration, Oxford Bulletin of Economics and Statistics 55, pp. 313–28.

    • Search Google Scholar
    • Export Citation

  • Deaton, A., 1992, Understanding Consumption, Oxford: Clarendon Press.

  • De Nederlandsche Bank NV, 2004: The Dutch Consumer: from Shopaholic to Enthusiastic Saver, De Nederlandsche Bank Quarterly Bulletin, pp. 57–68, September.

    • Search Google Scholar
    • Export Citation

  • Elliott, G., T. J. Rothenberg, and J. Stock, 1996, Efficient Tests for an Autoregressive Unit Root, Econometrica 64, pp. 813–36.

  • Jansen, W. J., and N. J. Nahuis, 2003, The Stock Market and Consumer Confidence: European Evidence, Economics Letters 79, pp. 89–98.

  • Ludwig, A., and T. Sløk, 2002, The Impact of Changes in Stock Prices and House Prices on Consumption in OECD Countries, International Monetary Fund, WP/02/01.

    • Search Google Scholar
    • Export Citation

  • Muellbauer, J,. and R. Lattimore, 1995, The Consumption Function: A Theoretical and Empirical Overview, in H. Pesaran and M. Wickens (eds.), Handbook of Applied Econometrics, pp. 221–301, Oxford: Blackwell.

    • Search Google Scholar
    • Export Citation

  • Phillips, P., and M. Loretan, 1991, Estimating Long-Run Economic Equilibria, Review of Economic Studies 58, pp. 407–436.

  • Schule, W., 2004, Household Consumption in France, France: Selected Issues, International Monetary Fund, pp. 5–26.

4

Prepared by Francisco Nadal De Simone. I thank David Hofman for his comments, De Nederlandsche Bank and, in particular, Mostafa Tabbae, for kind assistance with the data, and participants in several meetings in the Netherlands for their insights. Any remaining errors are my own responsibility.

5

Another relevant observation for the Netherlands is that the household saving ratio has been highly correlated with the share of property income in disposable income. This is broadly in line, but in the opposite direction, of developments in France, where the savings ratio increased during most of the recent cyclical downswing as the share of property income in disposable income increased (Werner, 2004).

6

A seminal book on consumption theories is Deaton (1992). Muellbauer and Lattimore (1995) provide a theoretical and empirical overview of the consumption function. The literature on wealth effects on consumption is vast: see, for example, Boone and Girouard (2002), and for multicountry studies including the Netherlands, Ludwig and Sløk (2002), Bayoumi and Edison (2003), and Allais and others (2002).

7

Changes to the pension regime or fiscal policy developments may have been additional sources of consumer uncertainty, but this was difficult to explore empirically in a direct way. Nevertheless, the empirical results are indicative of the role that uncertainty can play in affecting consumption.

8

The interest rate is assumed constant for simplicity. With Arrow-Debreu securities, the intertemporal budget constraint would include risky securities, and the model would be used to price them.

9

Net worth excludes pensions.

10

Net worth excludes pensions. Econometric results including pensions were not significantly different and are available upon request. The wealth elasticity was lower, however, as pensions are less liquid than the rest of net wealth.

11

Real disposable income includes property income. There is no readily available series of real nonproperty disposable income.

12

The share of population 65 years and older was also part of the sample, but was not included in the estimation as it was found to be an I(2) process.

13

The small sample bias correction followed Cheung and Lai (1993).

14

Their single-equation estimator is asymptotically equivalent to the maximum likelihood estimator of a full system of equations under Gaussian assumptions. The technique provides estimates that are statistically efficient and whose t-ratios can be used for inference in the usual way. Most importantly, the method takes into account both the serial correlation of the errors and the endogeneity of the regressors that is present when there is a cointegration relationship.

15

Appendix 1 contains the complete estimates of equation (7). There is no serial correlation in the residuals as indicated by the cumulated periodogram test. As usual with the Phillips and Loretan estimator, the goodness of fit of the equation as measured by the centered R2 is very high.

16

A model that excludes wealth results in a higher income elasticity, but not significantly different from the value reported above. Moreover, the long-run parameters are superconsistent.

17

Inflation enters the short-run consumption dynamics with a positive sign. While empirical work often finds inflation to be significant with a negative sign, this is not always the case; and it is not even necessary from a theoretical viewpoint. A negative sign can be expected to the extent that inflation causes nonindexed assets to depreciate or that changes in inflation dominate changes in nominal interest rates. In any of those two cases, consumption will tend to fall because of a negative income effect. When the intertemporal substitution effect (i.e., ceteris paribus, inflation induces consumers to bring forward their consumption) outweighs the income effect, a positive coefficient may result.

18

Recall from footnote 1 that the household saving ratio is highly correlated with the share of property income in disposable income.

19

While nondurable expenditure and consumption largely coincide, durable expenditure and consumption are separate events in time. Expenditure on durables are typically made in longer discrete intervals, so habit persistence, duration, and convex adjustment costs are likely to play an important role. The impact of financing conditions is likely to also play a role on the timing of those purchases quite different from the role played in nondurables purchases. Durables consumption, defined as services out of the stock of durables, can be expected to be much smoother than expenditures. Therefore, it is to be expected that the marginal propensity to consume out of total consumption (i.e., consumption including durables) will be relatively lower than the marginal propensity to consume out of only nondurables.

20

Wealth elasticity estimates are frequently based on consumption equations that are restricted to nondurables consumption.

21

On average, consumption adjusts to its equilibrium level in about three quarters.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Kingdom of the Netherlands—Netherlands Selected Issues
Author: International Monetary Fund
Volume/Issue:
Volume 2005: Issue 225
Publisher:
International Monetary Fund
ISBN:
9781451829518
ISSN:
1934-7685
Pages:
68
DOI:
https://doi.org/10.5089/9781451829518.002

Netherlands: Consumption, Income, and Net Worth

(Average Annual Growth Rates)

Netherlands: Unemployment Rate

Netherlands: Interest Rates and Stock Market Returns

Netherlands: Stock Market Price Index

Netherlands: Real House Prices

(Average Annual Growth Rates)

Netherlands: Real Household Net Worth/ Income 1/

Netherlands: Consumption and Estimated Long-Run Consumption

(Levels; in natural logs)