This Selected Issues paper on France underlies public intervention in financial markets. Econometric analysis indicates that in the long term, consumption tracks disposable income closely but is also affected by wealth effects. A counterfactual exercise suggests that a lower return to experience could be responsible for lower early wage growth in France. Increased training could enhance the employment experience of the low-skilled young worker in France provided that its cost is shared between the employer and the employee.


This Selected Issues paper on France underlies public intervention in financial markets. Econometric analysis indicates that in the long term, consumption tracks disposable income closely but is also affected by wealth effects. A counterfactual exercise suggests that a lower return to experience could be responsible for lower early wage growth in France. Increased training could enhance the employment experience of the low-skilled young worker in France provided that its cost is shared between the employer and the employee.

I. Household Consumption in France1

A. Overview and Introduction

1. Household consumption has made a steady contribution to growth in France not only because of its size but also thanks to its resilience with respect to fluctuations in income. Nevertheless, some economists and policy makers in France have argued that consumer spending has been suboptimally low over the past ten years. A number of countries experienced a significant decline in the aggregate private household saving rate, often explained by gains in wealth—the strong increases in stock market values in the second half of the 1990s, and, subsequently, in real estate prices. In France, consumers seem to have reacted differently. The saving rate continued to increase throughout the 1990s and the beginning of the 2000s. Moreover, despite its recent drop, it remains among the highest within the group of industrialized countries.

2. The focus of this chapter is on the determinants and short-term outlook of private consumption, approximated by spending on nondurables and services. An error-correction model was estimated to test the relevance of a number of factors suggested in the literature,2 such as disposable income, wealth, interest rates, inflation, demographic variables, and the unemployment rate (as a measure of uncertainty). The results suggest that private consumption tracks disposable income closely, but that wealth effects are also significant. However, overall wealth effects in France are smaller than in the United States, and the evidence of an impact of nonfinancial assets (predominantly housing) is weak. Nevertheless, the built-up and burst of the equity-price bubble had a sizeable impact on the propensity to consume out of current income. The evidence with respect to the other factors is mixed: real interest rates and fiscal variables lack significance, or robustness, while the unemployment rate is important for the short-term dynamics of consumption, but not for its trend.

3. The acceleration of private consumption since the second half of 2003 is compatible with the predictions of the estimated equation. Short-term factors, such as the perception of stronger-than-reported price increases after the introduction of the euro notes and coins, and uncertainties about the future of social security had delayed spending, opening up a gap between long-term equilibrium consumption and expenditures. Consumers seem to be catching up on delayed spending, bringing household spending back toward its long-term equilibrium path (“error-correction mechanism”). While the gap between measured and perceived inflation has not yet narrowed significantly, uncertainty has diminished with the economic recovery and the implementation of pension and health care reforms, and income support and wage increases at the bottom of the income distribution have buttressed the short-run propensity to consume. Overall, near-term prospects for consumption look bright.3 Finally, the stock market recovery and the strong housing market are boosting wealth, raising the long-run equilibrium consumption path. Nevertheless, to attain this higher path, it will be necessary to improve the working of consumer credit markets to raise the degree of spendability of housing wealth and strengthen confidence through the pursuit of growth-oriented structural reforms.

B. Modeling Consumption Behavior

4. Financial market imperfections, which prevent consumers from borrowing against future income, are commonly seen as a key reason for incomplete consumption smoothing. An alternative, but not incompatible explanation, focuses on the effects of uncertainties on spending (Carroll, 2001). The precautionary savings model produces behavior similar to the one generated by liquidity constraints: self-imposed reluctance to borrow replaces binding borrowing limits.

5. There have been a number of studies on the importance of excess sensitivity of consumption to current income in France. Campbell and Mankiw (1991) estimated the sensitivity of consumption to current income to be close to one and interpreted this result to imply that virtually all households face liquidity constraints. Sefton and In’t Veld (1999) got similar results and concluded that the permanent income model fails to explain French consumer behavior. Evidence is mixed in more recent studies covering the period of the run-up to and after the introduction of the euro, which was also characterized by accelerated financial liberalization. The joint Minefi-INSEE econometric model MESANGE does not include wealth in its consumption function. Likewise, Barell and Davis (2004) concluded that the evidence of wealth effects in France was very weak. Other studies, however, got opposite results. Boone and others (1998, 2001, and 2002) found significant effects on household consumption of changes in the value of equities, real estate, and other assets, with elasticities varying between 0.01 and 0.09. Beffy and Monfort (2003) and Flandrin-Le Maire (2004) confirmed the existence of a long-term cointegration relation between consumption, income, and wealth.4

6. This study is based on the life-cycle and permanent income theories, which assume that current consumption is related to life-time resources: the sum of human and nonhuman wealth. The stock of human wealth is defined as the present value of expected life-time labor income. Nonhuman wealth, the stock of tangible assets (A) resulting from decisions made in the previous period, includes financial (F) and nonfinancial assets (H), such as housing.5 The marginal propensity to consume (MPC) out of life-time resources may vary over time, depending on a number of factors, such as the age structure of the population, income distribution, taxes, interest rates, and uncertainty (for a more detailed presentation see Appendix I). The following relation is derived by maximizing utility under an intertemporal resource constraint:


7. Life-time labor income is unobservable, as is the discount rate used to convert it into present value human capital (human wealth). A common assumption is that, at the aggregate level, current real disposable labor income6 is proportional to expected life-time real labor income7 (Human Wealtht = kt * Yt). With this assumption, the consumption function becomes:


Changes in potential output (trend productivity), as well as taxes, social security contributions, and benefits impact on expected permanent disposable income. All these factors will be reflected in the estimated coefficients of the reduced form equation, captured either in a convoluted way by income and wealth elasticities, or more straightforwardly by other “control” variables. The following log-linear approximation of (2) will be used for the estimation of long-run consumer demand:


where lower case Latin letters represent logs (x=ln(X)), Zt is a vector of all other variables,8 and ΨP the vector of associated coefficients. Consumption, income, and wealth variables are in per capita terms.

Data issues

Consumption versus consumer expenditures

8. While household spending and consumption of nondurables and services largely coincide, durables consumption (defined as services out of the stock of durables) and purchases of durable consumption goods are typically separate events in time. Aggregation over individual durables purchases removes a substantial part of this difference. Nevertheless, the time-series properties of aggregate durables are very distinct from both consumption and income.9 Because expenditures on durables are typically made in longer, discrete intervals, similar to investment decisions, habit persistence, and convex adjustment cost are likely to play an important role, and the impact of some determinants (such as financing conditions) on the timing of purchases is likely to differ substantially from its relevance for nondurables consumption. Durables consumption can be expected to be much smoother than respective expenditures on durable goods. Consequently, total consumption can be assumed to follow closely the pattern of spending on nondurables and services. The approximation of total consumption by nondurables spending is unproblematic, as long as the share of durables in total spending is not changing systematically.10

Labor and property income

9. The income variable in the consumption equation will be “nonproperty disposable income (YNP)”—defined as private households’ disposable income minus income derived from asset holdings: distributed profits, other investment income, net interest, and rent income. This is consistent with the theoretical approach (¶7 and Appendix I).11 In addition, there are reasons to assume that the short-run MPC out of property and nonproperty income differ significantly, with a short-run MPC out of nonproperty income being by far more important. Assets and derived property incomes are distributed unequally across the population. Households with significant asset holdings are less likely to be credit-constraint or strongly risk-averse. They will be able and willing to smooth consumption over time (the optimal behavior under intertemporal utility maximization). Consequently, current consumption will be little sensitive to fluctuations in property income. In the long run, however, unspent income will accumulate and raise the stock of wealth, increasing consumption, given current income. These wealth effects will be captured by the inclusion of assets in the estimated long-run equation (see also ¶¶11 and 12).

10. The share of property income in total disposable income increased from about 3 percent at the beginning of the 1980s to some 9 percent in the early 2000s, when widening margins in the private business sector led to an increase in distributed profits, and income from other investments (including through asset funds) increased strongly. Net interest and rent income played a minor role (Figure 1, upper panel). It is interesting to compare the savings rate as conventionally measured in the national accounts, with the savings rate defined in terms of nonproperty income. The latter did not increase in the 1980s and 1990s, but actually fell (Figure 1, lower panel). This suggests that the increasing share of property income might be among the reasons for the observed increase in the conventional savings rate.12

Figure 1.
Figure 1.

France: Property and Nonproperty Income and Savings

Citation: IMF Staff Country Reports 2004, 346; 10.5089/9781451813623.002.A001

Sources: INSEE; and IMF staff calculations.

11. Notwithstanding the adjustment, the French savings rate is among the highest ones across industrialized countries, though such comparisons are subject to caveats,13 so that the question remains which other factors might have prevented a fall in the conventional saving ratio, despite the rise in wealth, and the fall in real interest rates. Moreover, the increase in both saving ratios at the beginning of this decade requires some additional explanations. Candidates include the burst of the stock market bubble, perceived inflation surprises, demographics and increased awareness that welfare promises might not be fulfilled in the future, and hence private precautionary savings might have increased with the risks faced by individual households.

Wealth and consumption

12. Disaggregated annual household wealth data, starting in 1977, have been constructed by the Banque de France and INSEE. They were transformed into quarterly figures, following closely the approach presented by Beffy and Monfort (INSEE, 2003).14 Quarterly financial saving flows and price data for stocks, bonds, and the existing housing stock were used to account for intra-annual wealth accumulation and valuation changes. The upper panel on Figure 2 plots the ratio of total household assets, financial assets, and nonfinancial assets to income. The value of total household asset holdings in percent of annual disposable income has increased from a factor of 3 at the beginning of the1980s to a factor of 6 in 2000 (Figure 2, upper panel).15 It has roughly stayed at that level as the decline in stock values from its peak in 2000 was fully offset by increasing real estate prices.

Figure 2.
Figure 2.

France: Wealth and Consumption

Citation: IMF Staff Country Reports 2004, 346; 10.5089/9781451813623.002.A001

Sources: Beffy and Monfort (INSEE); INSEE National Accounts; and IMF staff calculations.

13. From the end-1970s to the mid-1990s, the ratio of consumer expenditures to income and the asset to income ratio moved closely together. However, the stock market boom of the 1990s was not followed by a significant increase in spending, and by the end of the 1990s, a sizeable gap between the two ratios had opened up (Figure 2, bottom panel). There are several possible explanations for this divergence: (i) the long-term relation between wealth and consumption might have broken down. A possible reason for this could be that the very sharp increase in stock prices, and later in real estate prices, has been perceived by households as a temporary aberration, not as an increase in permanent resources; (ii) financial assets are more likely to lead to consumption because they are easier to liquidate, therefore the two wealth components should be ‘spendability’ weighted;16 (iii) there is no structural break in the long-term consumption-wealth relation, but due to factors driving the short-term adjustment process, a significant equilibrium gap might have opened up and could be followed by a fall in the savings ratio, if consumption follows an error-correction process.

Estimating a consumption function

14. The following long-run consumption equation in log levels was estimated using the Stock-Watson procedure with a standard k=2 lag and lead structure.17 The estimation period was 1982:1 to 2003:4. The standard errors (in brackets) are adjusted for long-term variance.18. The coefficients for income and the two wealth components were restricted to sum to unity, a restriction that was not rejected at the 5 percent level.19



  • cnt = real consumer expenditures on nondurables and services per head

  • yt = real nonproperty disposable income per head

  • ht = nonfinancial assets per head

  • ft = financial assets per head

  • πt = annual inflation rate, implicit price deflator of cn

  • a65t = age group of 65 years and above in percent of population

15. The estimated short-run equation shows the growth rate of consumption as a function of lagged income and consumption growth, the change in the rate of unemployment, and the lagged error correction term (ECT), which is significant and correctly signed, but the adjustment speed is unusually high, implying that about half of the disequilibrium gap will be closed after one quarter.20


with UR = unemployment rate

ECT = error term from long-run equation

16. The main results are:

  • Current income has been the most important determinant of consumption, which is suggestive of either the presence of liquidity constraints faced by consumers in imperfect credit markets (due to regulation or asymmetric information), or strong risk aversion on the side of French consumers leading to precautionary behavior in the presence of uncertainties with respect to future income and needs.

  • Wealth was found to be another important determinant of consumption. Its importance can be gleaned from Figure 3, which shows the error terms of long-run consumption equations including and excluding wealth. Adding wealth significantly improves the fit of the equation.

  • The wealth effect is smaller than for instance in the United States. The MPC out of total wealth is 2.7 cents per euro, compared to 4 cents per dollar frequently estimated for the United States.21 The estimated elasticity of consumption with respect to nonfinancial wealth (predominantly housing) was only about a third of the elasticity with respect to financial wealth and statistically not significant at the 5 percent level. The implied MPC out of financial wealth alone was 2 cents per euro (over the 1990-2003 period, the average share of financial in total assets was 47 percent).

Figure 3.
Figure 3.

Wealth Effect on the Error-Correction Term 1/

Citation: IMF Staff Country Reports 2004, 346; 10.5089/9781451813623.002.A001

1/ Residuals (estimation period 1982 Q1 to 2003 Q4).

Nevertheless, given the large size of changes in wealth, the effect on consumption has been important.

  • The results for real interest rates were inconclusive; coefficients were often insignificant, unstable, and sensitive to the calculation of real interest rates.22

  • The negative coefficient of inflation was robust with respect to its definition as well as over time. The inclusion of inflation resulted in a weaker wealth effect and a more rapid speed of adjustment to the equilibrium error in the short-run equation.23

  • Aging, measured by an increasing share of people above the age of 65, had a significant but not very robust, small negative effect on consumption.

  • The level of unemployment was not significant in the long-run equation, but its changes had a negative effect on the growth rate of consumption.

  • Neither real public debt nor the public deficit-to-income ratio were found to be significant.24

Reported and perceived inflation

17. There are a number of possible explanations for the negative impact of inflation on consumption: first, the level and variability of inflation are positively correlated, therefore the inflation rate might capture uncertainties regarding real income and wealth; second, higher inflation rates cause nonindexed assets to depreciate; third, with low variability in nominal interest rates of a large part of the portfolio of private households, changes in inflation dominate the variations in the real interest rate. In this case, the negative sign for inflation would indicate that the income effect of real interest rates outweighs the intertemporal substitution effect for French consumers. However, the ambiguous results on real interest rates cast doubt on this interpretation. And finally, the negative inflation effect might be partially spurious, due to distortions produced by the interplay of inflation and the measurement of income in the national accounts. However, these distortions should be largely absent in nonproperty income.

18. Whatever the reason for the significance of inflation might be, the estimation reflects the average reaction of consumption on measured inflation in the past. Obviously, the rate of inflation perceived by consumers sometimes differs significantly from officially measured rates. The perception of an inflation rate significantly above the reported one after the introduction of the euro notes and coins on January 1, 2002 (Figure 4) has been attributed to sharp price increases in some highly visible consumption goods and services. It is likely to have discouraged consumption temporarily. Given the size of the inflation coefficient, the effect might have been large: a 1 percentage point transitory increase in perceived inflation could have resulted in a 0.5 percent drop in target consumption. Once the high inflation perception disappears, consumption is likely to catch up with the level determined mainly by income and wealth.

Figure 4.
Figure 4.

Inflation Perception

Citation: IMF Staff Country Reports 2004, 346; 10.5089/9781451813623.002.A001

Sources: European Commission, Consumer Survey; and INSEE.

19. Given the way the consumer survey covering price-trend perceptions over the past 12 months is designed, it is difficult to link the survey data to a specific inflation rate. Consumers are asked the question: “How do you think that consumer prices have developed over the last 12 months?” They are then asked to choose between five possible answers, ranging from “risen a lot” to “fallen”. No specific price index is mentioned, nor do the answers give a precise number of the rate of inflation as seen by the households. However, the close correlation between the perceived price trends over the past 12 months and actual inflation rates until 2001 suggests that perceived inflation in 2002-03 has been higher than reported inflation. On the basis of an ad hoc equation, the forecast for 2002-03 inflation rates as perceived by households were 3.5 percent and 4.3 percent.25 This combined with the estimated semi-elasticity in Equation (4) implies that consumption was depressed by 0.9 percentage points in 2002, and 1.3 percentage points in 2003 due to the perception of a hike in prices after the introduction of euro notes and coins.

Demographic developments

20. In line with the life-cycle theory, Beffy and Monfort (2003) found evidence for an increase in the saving ratio, caused by a demographic shift in the French population, when the cohort of baby boomers was reaching prime earning age. Earlier studies looking at the relation between age cohorts and consumption in France, however, could not confirm the predictions of the life-cycle theory. Using survey data, Fall, Loisy, and Talon (2000) found that total assets increased steadily in relation to age, even after retirement. Moreover, they noticed an increase in the savings ratio after retirement in the 1980s, and for the 1995 survey a decline in the savings ratio around the age of 60, but a rebound for the age group of 70 and higher. De Seres and Pelgrin (2003) also found a negative relation between the old age dependency ratio 26 and consumption in a system of dynamic panel equations. Bodier (1999) estimated a small effect of aging on overall consumption, but a larger effect on the structure of demand for goods and services.

21. Aging in France is already visible (Figure 5). The share of citizens aged over 65 in the total population has been increasing since the mid-1980s, while the share of people of working age passed its peak at the end of the 1980s. Nevertheless, the labor force has continued to grow due to a significant rise in the participation rate. The empirical investigation looked into the impact of three variables reflecting the shift in the population structure: the old age dependency ratio (the age group of 65 and above in percent of the working-age population), the working-age population in percent of the total population, and the labor force in percent of the total population. The coefficient on the age 65+ group was small, significantly negative, but somewhat fragile. The results for the group of working-age people and the labor force variable were positively signed, but the coefficient was neither very significant nor robust across different specifications or over time.

Figure 5.
Figure 5.

France: Demographic Developments

Citation: IMF Staff Country Reports 2004, 346; 10.5089/9781451813623.002.A001

Source:AMECO Database.

22. The negative impact of an increase in the share of older persons in total population contradicts the implications of the life-cycle theory, according to which aging or a shift in the distribution of assets towards the older generation increases the aggregate MPC out of total assets.27 However, bequest motives and the precautionary asset holdings due to uncertainties of the remaining life time are likely to significantly reduce this effect. In addition, consumption and leisure may be substitutes, increasing the likelihood of a negative effect on the MPC out of income and wealth. To the extend that early retirement is motivated by a favorable utility value of leisure compared to consumption, raising the effective retirement age, through incentives to work longer, would not only increase income (and hence consumption), but could also result in a higher MPC out of this higher income level.

C. Conclusions

23. Income and wealth are the main determinants of consumption in France. The close relation between current income and consumption is indicative of shallow consumer credit markets. This conclusion is confirmed by the relatively small, and only weakly significant, impact of nonfinancial—mainly housing—wealth. Facilitating access to U.S. style home equity loans, relaxing ceilings on lending rates, and lowering transactions costs (e.g., on mortgage financing) would improve households’ ability to smooth consumption.

24. With more efficient financial markets, expectation of future income and consumption needs will likely play an increasing role in current consumer expenditure decisions. The likelihood of a negative impact of real interest rates would rise with the increasing importance of intertemporal decision-making, improving the effectiveness of monetary policy.

25. A determined structural reform program, aimed at raising potential output, could succeed in strengthening confidence. Raising the effective retirement age would presumably increase the MPC out of both current income and wealth.

26. The weakness of consumption in 2002 has probably been accentuated by the perception of sharp price increases after the introduction of the euro notes and coins, attributed to the experience with daily, small expenditure items, such as certain food products and services. The decline in stock market values also depressed consumption. The fall in the savings ratio in 2003-04 is consistent with households catching up with equilibrium consumption, equity prices well off their lows, and support for low incomes. Once inflation perceptions return to normal, the savings ratio is likely to decline further. However, special and confidence factors-which are difficult to capture-may also have played a role.

APPENDIX I The Solved-out Consumption Function

27. The workhorse model of consumer behavior is derived from the basic idea that maximizing utility under resource constraints will lead to consumption smoothing, relative to current income.28 The first-order condition for maximization, the Euler equation, states that the marginal rate of substitution between period one and period two consumption (c1 and c2) should be equal to the relative price of today’s compared to future consumption, the ratio of the subjective discount factor (1+δ), and the market interest rate factor (1+r). A useful illustration of this condition is derived under the assumption of inter-temporal additivity of utilities and a CES utility function (U):


28. The two-period budget constraint, with no bequests, states that life-time consumption equals life-time resources, the initial wealth endowment plus present and discounted future income:


The first-order condition Euler equation takes the form:


where the inter-temporal elasticity of substitution σ=1/(1+δ) the reciprocal of the coefficient of relative risk aversion p, measures the response of consumption to changes in intertemporal prices. If the incentive to wait (1 + r) overcomes impatience (1 + δ), it pays to postpone consumption, and consumption will grow over time, given permanent income. 29

29. Combining the first-order condition with the budget constraint leads to the solved-out consumption function:




The marginal propensity to consume (MPC) out of total wealth—asset holdings plus human wealth is in general not a constant, as it reflects a geometrically-weighted average of time varying subjective discount rates and market interest rate, where the weights are given by the inter-temporal elasticity of substitution. Assuming expected income is growing at rate g, then the consumption function becomes:


30. The simple two-period model has some interesting implications, which help interpret the result of the estimations. It can easily be extended to the multi-period case, with implications for the effects of the economy’s age composition, and to include a variety of terms, which work predominantly through their impact on the MPC. Uncertainties about noninsured risk to income, for example, could be represented by a factor added to the subjective discount rate, lowering the MPC. Alternatively, the precautionary motive implies that savings increase (given income) when the level of asset holdings is considered no longer adequate to buffer consumption against bad shocks. The tug-of-war between impatience and prudence defines a target level of precautionary wealth. The balance is shifting towards a higher asset-holding target with increased uncertainties.30

31. The impact of the real interest rate on consumption depends on the initial asset position, the rate of income growth, and the degree of intertemporal substitution. The different channels, through which the real interest rate affects consumption work in opposite directions, and the resulting overall effect is likely to be small. Moreover, the ratio of assets to income is shifting over time, making the effect of real interest rates on consumption unstable.

Figure A1.
Figure A1.

France: Wealth and Consumption

Citation: IMF Staff Country Reports 2004, 346; 10.5089/9781451813623.002.A001

Source: IMF staff calculations.
Figure A2.
Figure A2.

France: Durables

Citation: IMF Staff Country Reports 2004, 346; 10.5089/9781451813623.002.A001

Sources: INSEE; and IMF staff calculations.


  • Allard-Prigent, C., C. Audenis, K. Berger, N. Carnot, S. Duchene, and F. Pesin (2002), “Présentation du Modèle MESANGE,” Minefi, document de travail.

    • Search Google Scholar
    • Export Citation
  • Attanasio, O. (1998), “Consumption Demand,” NBER Working Paper 6466.

  • Barrell, R., and E. P. Davis (2004), “Consumption, Financial and Real Wealth in the G-5,” NIESR and Brunel University, mimeo.

  • Beffy, P., and B. Monfort (2003), “Patrimoine des ménages, dynamique d’allocation et comportement de consommation,” INSEE, Direction des Etudes et Synthèses Economiques.

    • Search Google Scholar
    • Export Citation
  • Beffy, P., X. Bonnet, B. Monfort, and M. Darracq-Paries (2003), “MZE, un modèle macroèconometrique pour la zone euro,” Economie et Statististique No. 367.

    • Search Google Scholar
    • Export Citation
  • Berger, K., and A. Daubaire (2004), “Les determinants du taux d’épargne en question,” Problèmes Economiques, No. 2.849.

  • Bodier, M. (1999), “Les effets d’âge et de génération sur le niveau et la structure de la consommation,” Economie et Statistique, pp. 324-5.

    • Search Google Scholar
    • Export Citation
  • Bonnet, X., and E. Dubois (1995), “Peut-on comprendre la hausse imprévue du taux d’épargne des ménages depuis 1990?,” Economie et Prevision, No. 121.

    • Search Google Scholar
    • Export Citation
  • Boone, L., and N. Girouard (2002), “The Stock Market, the Housing Market and Consumer Behaviour,” OECD Economic Studies No. 35.

  • Boone, L., and I. Wanner (1998), “Stock Market Fluctuations and Consumption Behaviour: Some Recent Evidence,” OECD Economics Department Working Papers No, 98.

    • Search Google Scholar
    • Export Citation
  • Boone, L. (2001), Financial Market Liberalisation, Wealth and Consumption, OECD Economics Department, Working Papers No. 308.

  • Browning, M., and A. Lusardi (1996), “Household Saving: Micro Theories and Micro Facts,” Journal of Economic Literature.

  • Campbell, J., and N. G. Mankiw (1991), “The response of consumption to income - A cross-country investigation,” European Economic Review, Vol. 335, pp. 71521.

    • Search Google Scholar
    • Export Citation
  • Carnot, N. (2001), “The Model Manège, Structure and Equations,” Minefi, Working Paper.

  • Carroll, C. (1992), “The Buffer-Stock Theory of Savings: Some Macroeconomic Evidence,” Brookings Papers on Economic Activity, Vol. 1992:2.

    • Search Google Scholar
    • Export Citation
  • Carroll, C. (2001), “A Theory of the Consumption Function, With and Without Liquidity Constraints,” Journal of Economic Perspectives, Vol. 15-3.

    • Search Google Scholar
    • Export Citation
  • Davidson, J.E.H., D. Hendry, F. Srba, and S. Yeo (1978),Econometric modelling of the aggregate time-series relationship between consumers’ expenditure and income in the United Kingdom, Economic Journal, Vol. 88, pp. 66192.

    • Search Google Scholar
    • Export Citation
  • Davis, M., and M. Palumbo (2001), “A Primer on the Economics and Time Series Econometrics of Wealth Effects,” Federal Reserve Board, DP.

    • Search Google Scholar
    • Export Citation
  • Deaton, A. (1992), “Understanding Consumption,” Oxford.

  • De Serres, Alain, and Florian Pelgrin (2003), “The Decline in Private Saving Rates in the 1990s in OECD Countries: How Much Can be Explained by Non-wealth Determinants,” OECD Economic Studies No. 36.

    • Search Google Scholar
    • Export Citation
  • Fall, M., C. Loisy, and G. Talon (2000), “Household Savings in France 1984-1998,” mimeo.

  • Faulkner-MacDonagh, C. (2004), “Canadian Household Saving - Developments and Risks,” IMF, Canada--Selected Issues.

  • Flandrin-Le Maire, G. (2004), “Les effets de richesse et l’arbitrage consommation épargne des ménages en France,” DP, mimeo.

  • Hendry, D. (1994), HUS Revisited, Oxford Review of Economic Policy, Vol. 10, pp. 86106.

  • Hendry, D., von Ungern-Sternberg (1981), Liquidity and Inflation Effects on Consumers’ Expenditure, in: A. Deaton, (ed.) Essays in the Theory and Measurement of Consumers’ Behaviour, CUP.

    • Search Google Scholar
    • Export Citation
  • INSEE (2002), Quel est le taux d’epargne des ménages français ? Les mesures du taux d’epargne et leur interpretation, in: L’Economie Francaise.

    • Search Google Scholar
    • Export Citation
  • Loisy, C. (1999), L’épargne des ménages de 1984 à 1995: disparité et diversité, Economie et Statistique, pp. 3245.

  • Lollivier, S. (1999), La consommation sensible aux variations de revenu, meme sur le court terme, Economie et Statistique, pp. 3245.

  • Ludwig, A., and T. Slok (2002), The Impact of Changes in Stock Prices and House Prices on Consumption in OECD countries, IMF Working Paper 02/1.

    • Search Google Scholar
    • Export Citation
  • Muellbauer, J. (1995), The Assessment: Consumer Expenditure, Oxford Review of Economic Policy.

  • 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.

    • Search Google Scholar
    • Export Citation
  • Norman, B., M. Sebastia-Barriel, and O. Weeken (2002), Equity Wealth and Consumption -the Experience of Germany, France and Italy in an International Context, Bank of England Quarterly Bulletin.

    • Search Google Scholar
    • Export Citation
  • Pichette, L., and D. Tremblay (2003), Are Wealth Effects Important for Canada?, Bank of Canada Working Paper.

  • Sefton, J.A., and J. In’t Veld (1999), Consumption and Wealth: an International Comparison, The Manchester School, Vol. 67.

  • Stock, J.H., and M. Watson (1993), A Simple Estimator of Cointegrating Vectors in Higher Order Integrated Systems, Econometrica, Vol. 61, pp. 783820.

    • Search Google Scholar
    • Export Citation
  • Tzanninis, D. (2002), Why Has UK Household Consumption Been So Strong?, IMF United Kingdom, Article IV Consultation, Selected Issues.


This chapter was prepared by Werner Schule (EUR).


For a comprehensive presentation of theories of consumer behavior, see Deaton (1992). Muellbauer and Lattimore (1995) provide a detailed overview of the theory and estimation of the solved-out consumption function. Browning and Lusardi (1996) survey micro theories and micro evidence. Attanasio (1998) emphasizes the importance of individual level data to estimate preference parameters.


Special factors (higher energy consumption due to cold winter weather, a shift in health spending to households and prolonged sales) have also been supportive. To the extent that these factors subside, the recent acceleration in household spending may prove partly temporary.


None of these authors was able to find evidence of changes in the responsiveness of French consumers to equity prices, despite a rapid increase in stock market capitalization.


Some authors also include consumer durables.


Income derived from the holdings of assets (property income) needs to be substracted from total disposable household income if tangible assets appear as a separate right-hand side variable in the consumption function.


In the two-period example (see Appendix 1), assuming the potential output growth rate equals the real interest rate, the life-time income equals two times current income (with the life time horizon T=2). Alternatively, instead of using the proportionality assumption, Brayton and Tinsley (1996) estimate an econometric model for human wealth. General dynamic equilibrium models calibrate behavioral parameters and solve the model forward. While this allows for a much more rigorous theoretical treatment of consumer behavior, these models are not yet close enough to the data to be used for forecasting.


Such as the unemployment rate, real interest rates, inflation, and demographic variables.


The variance of quarterly changes in durables expenditures is more than four times the variances of either total consumption or income.


This was the case until the mid-1990s, when the decline in relative (and absolute) durables prices provided for an increasing share of real expenditures on durables, at a constant percentage of durables purchases in overall nominal consumer spending. Between 1978 and 2003, the average share of real durables in total consumption was 8.1 percent (see Figure A2 in Appendix I).


Further, there are severe measurement problems linked to the treatment of property income in the national accounts. One of these problems is caused by the distortionary impact of inflation (in particular at high rates) on real after-tax disposable income.


Muellbauer and Lattimore (1995) made a strong case in favor of nonproperty income in modeling consumption, to lower the risk of errors in savings measurement.


Problems of cross-country saving-rate comparisons are discussed in INSEE, “L’économie française 2002/3.” Cross-country comparisons of saving ratios are also discussed in “Comparison of Household Saving Ratios—Euro Area/United States/Japan,” joint ECB-OECD study, mimeo.


For the years 1978-2000, quarterly measures for household wealth—disaggregated into financial and nonfinancial assets—were graciously provided by Beffy and Monfort. For the years up to 2003, Fund staff updated them on the basis of revised national account data and indicators of the evolution of asset prices.


Figure 2 shows the ratio of W to quarterly Y.


The “*” indicates significance at the 5 percent level.


The F-test statistic was F(1,54)= 4.78; (p=0.033).


The equilibrium error was significant and negatively signed in all short-run equations tested. However, the adjustment speed varied between -0.2 and -0.6 depending on which variables were included in the long-run equation. The inclusion of inflation in particular resulted in a shorter reaction time.


The MPC were calculated on the basis of the average wealth-consumption ratios over the period 1990-2003. The impact of nonfinancial wealth is not significant at the 5 percent level.


This may be due to problems in measuring expected inflation, but is also consistent with the ambiguous predictions of theory on the impact of the real interest rate on consumption.


The negative effect of inflation is a well known phenomenon; see Hendry (1994) on the effect in the Davidson-Hendry-Srba-Yeo (1978) consumption function. The reduction in the size of the wealth effect indicates that the inflation variable might be picking up the impact of higher inflation on the real value of nonindexed assets.


De Serres and Pelgrin (2003) have identified public-sector savings as an important determinant of the household savings rate in a dynamic panel study across OECD countries. Cross-country panel equations have generally been more successful in finding evidence of partial Ricardian equivalence than aggregate time series studies. Berger and Daubaire (2004) confirm a Ricardian effect across OECD countries.


The equation estimated from 1985:q1 to 2000:q4 is:πt=α0+0.016*πperceived+0.77πt1 equation (perceived inflation being the only explanatory variable) would forecast even larger inflation discrepancies (2.9 and 3.8 percentage points).


Defined as the age group of 65 and above in percent of the working-age group.


Other demographic variables, such as the age group between 15 and 65, were used alternatively with a significant (but positive) impact on consumption.


Consumption smoothing is the basic underlying idea of both the life-cycle hypothesis and the permanent income theory of consumption.


Increased uncertainties might also lead to higher investment into human capital; Carroll (1992) finds that the increased divorce risk in the United States motivated women to spend more time in education.

France: Selected Issues
Author: International Monetary Fund