![Sample our Built Environment journals, sign in here to start your access, Latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5926/TOC-Subject-Sample-2021-Built-Environment.jpg"})
ABSTRACT
The aim of this paper is to explain private consumption as a function of income and wealth with data from European Union countries. To examine how the developments in housing and stock markets may have affected consumption behaviour, we adopt two econometric procedures. First, we use the Stock–Watson procedure to account for wealth effects on consumption over the long run. Second, through an error-correction model we measure wealth effects on consumption over the short run. We found significant albeit mixed values for the long-run elasticities of consumption with respect to real residential and equity prices. We also found strong evidence that consumption exhibits error-correction behaviour in the short run, with the value of the error-correction term signifying that household consumption takes several quarters to completely respond to changes in the markets.
Acknowledgements
Financial support granted by the Fundação para a Ciência e Tecnologia (FCT), the Programa POCTI and Universidade Portucalense is gratefully acknowledged. The authors would also like to thank to Michael Gail, Peter van Els and two anonymous referees for extremely valuable comments. The usual disclaimer applies.
Notes
1. As stated by Green (2002, p. 775): ‘The issue is one on which academics and the financial community seem to disagree. Rigorous academic works (CitationFama, 1981; CitationFisher & Merton, 1984; CitationBarro, 1990; CitationPoterba & Samwick, 1995), have generally found that the wealth effect on consumption is quite small, while research reports from financial service companies generally assert that the effect is real and important.’
2. Bayoumi and Edison (2002, pp. 7–9). See also the discussion in CitationLudwig and Sloek (2004).
3. The workhorse of modern consumption theory is the rational-expectations version of the permanent income hypothesis, as derived by CitationHall (1978). CitationGali (1990) has shown that that type of aggregate consumption function can also be derived from the dynamic optimizing behaviour of consumers with finite horizons and life-cycle saving.
4. A difficulty with estimating the first equation) directly is that planned consumption does not always equal actual consumption owing inter alia to lags in adjustment and liquidity constraints. For instance, individuals may not be able to adjust within each period their spending on house services, given large searching, moving and finance costs. Also, if there is considerable habit persistence in consumption behaviour, then individuals may adjust their spending slowly to bring it in line with the level suggested by the economic determinants in the first equation. Another reason may be the presence of liquidity-constrained individuals, who cannot smooth their consumption by borrowing against their future income due to capital market restrictions. All this suggests using an error–correction approach. Additionally, for a simple illustration of the ‘life-cycle theory’ see CitationDavis and Palumbo (2001).
5. Boone et al. (1998, p. 6): ‘(…) the life cycle model takes no account of uncertainty in the future stream of revenues (CitationDeaton [1991] and Carroll [1992]), or bequest motives (CitationWilhelm [1996], and Laitner and Juster [1996]). Furthermore, CitationZeldes (1989) argued that the strength of any wealth effect should also be linked to the distribution of wealth and the existence of liquidity constraints.
6. The traditional life-cycle model in CitationAndo and Modigliani (1963) simply implies that aggregate consumption is linearly related to labour income and wealth. It says nothing about the cointegration properties of the variables. In contrast, the theoretical life-cycle model in CitationGali (1990) implies that consumption, income and wealth variables may share a common trend. But whether this theoretical implication is consistent with actual data still needs to be tested, so one must test for the presence of cointegration.
7. By using a logarithmic functional form, we are implicitly assuming that elasticities are fixed over the sample period.
8. For the limitations in using asset prices as proxies, see Bertaut (2002, p. 10). The (in)direct impact of stock market prices on aggregate consumption has been studied, for instance, by CitationRomer (1990) and CitationPoterba and Samwick (1995). The role of housing prices on consumption is the focus in, among others, CitationMiles (1992), CitationBrady et al. (2000) and CitationGirouard and Blondal (2001).
9. Since the theoretical model states that it is the quantity of wealth that affects consumption, not the price of wealth, the stock market capitalization was also tested but the results were inconclusive. The real equity and residential price indices were obtained from the Bank of International Settlements (BIS) (using national data) and were first presented in a paper by CitationBorio et al. (1994). The remaining data are from Eurostat's NewCronos database. The data for Germany begins after 1990, owing to breaks in the series arising from the German unification. The BIS database does not include Portugal, so instead we use a ‘share price index’ taken from Eurostat's NewCronos database and a real estate prices index published by ‘Confidencial Imobiliário’.
10. The only exceptions are the unemployment rate (ur) in France and Germany, the short-term interest rate (str) in Portugal and the inflation rate (inf) in Belgium, Germany and the UK. The results are almost in all cases unchanged if the ADF model includes an intercept and/or a trend or a different lag structure. ADF results could be made available upon request.
11. It would be interesting to relate these results with the different financial structures prevailing in each country, albeit due to space limitations we do not perform that analysis here. Notice also that Denmark presents the puzzling result of a significant negative effect. Nevertheless, that result is strongly reversed when we estimate the sample only for the 1990s.
12. Notice that the conclusions for Portugal and Germany should be taken with some caution due to the shorter time period covered.
13. We can refer here to CitationPoterba (2000), who states that the growing importance of retirement accounts may have reduced the marginal propensity to consume out of stock market wealth in the US. Indeed, CitationLudvigson and Steindel (1999) and CitationMehra (2001), estimate a lower marginal propensity to consume out of total wealth for more recent periods. The other possible factor contributing to a lower marginal propensity to consume out of wealth is the falling cost of leaving bequests.
14. An idea supported by CitationPoterba and Samwick (1995) and CitationBoone et al. (1998), among others. Nevertheless, since our sample does not include the 70s, it is better to interpret the results with some caution.
15. For a complete exposition of this problem see the stylized example presented in CitationDavis and Palumbo (2001).
16. Davis and Palumbo (2001, p. 27): For example, the unemployment rate or consumer sentiment indexes are intended to capture the precautionary behaviour of households, while including variables that predict the growth rate of income can proxy for the potential effects of borrowing constraints on consumer spending. Meanwhile, including lagged growth rates of consumption, income, and wealth help to capture additional short-run dynamics in the reactions of these variables to transitory shocks that do not affect the target level of consumption (and, thus, do not involve error correction,per se).
17. This result is consistent with the sluggish responses noted by CitationCase et al. (2005). The speed of adjustment is usually measured by ‘half-lives’, which are computed as ln (0,5)/ln(1 + τ), where τ is the coefficient on the error – correction term.
18. A noticeable difference is the different behaviour of Denmark and the UK, two non-euro economies, from the rest of the EU. In this line, CitationCarruth et al. (1999) found no evidence of a common consumption function across the EU countries.
19. Bertaut (2002, p. 18): ‘For changes in both financial and non-financial wealth, these short-run effects die out over time, and the only lasting effect of wealth on the level of consumption comes from a permanent change to the level of wealth through the error-correction mechanism.’