Working Paper: CEPR ID: DP1251
Authors: Tullio Jappelli
Abstract: Existing estimates of the age-wealth profile use panel or cross-sectional data. Panels with wealth data are rare, and plagued by measurement errors and sample attrition. Cross-sectional data require strong identifying assumptions. This paper represents the first attempt to use repeated cross-sectional data to test one of the central implications of the life-cycle theory, i.e. the extent to which the elderly run down accumulated assets. Using the 1984-93 Italian Survey of Household Income and Wealth, the paper shows that failing to control for the influence of cohort effects leads to substantial bias in the estimate of the age-wealth profile. Once cohort effects are taken into account, the results indicate that households accumulate assets until they are 70 years old; afterwards the estimated average annual rate of wealth decumulation is about 6%. A basic prediction of the life-cycle model, that the cohort effect increases from older to younger cohorts, is strongly supported by the data. The results also uncover considerable population heterogeneity: the rates of wealth decumulation are much lower for rich households and households headed by individuals with higher education.
Keywords: wealth accumulation; lifecycle model; repeated cross-sections
JEL Codes: E21
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| year of birth (J19) | wealth accumulation (E21) |
| older cohorts (J14) | lower wealth accumulation (E21) |
| younger cohorts (J19) | higher wealth accumulation (G51) |
| age 70 (J26) | wealth decumulation (G51) |
| wealthier households (G59) | lower rates of decumulation (D15) |
| higher educational attainment (I23) | lower rates of decumulation (D15) |
| failing to control for cohort effects (C92) | bias in estimates of age-wealth profile (D15) |