Formats and related files
This paper seeks to improve our understanding of household saving behaviour. It is based on an analysis of unit record data from March years 1984 to 1998 taken from the Household Economic Survey (HES). There are limitations of the data set but it provides the only available estimates of income and expenditure, from which saving is estimated as a residual. The HES is a series of cross-sectional surveys rather than a true panel, so we construct synthetic cohorts rather than tracking individual households. We use a range of regression models to separate out the effect of age, birth-year cohort and year on saving rates. The typical age profile for savings is hump-shaped, peaks around age 57 and does not become negative at older ages. Such a profile appears to have shifted down for the cohorts born between 1920 and 1939 relative to the younger and older cohorts studied. This pattern of cohort effects is robust to the inclusion of conditioning variables and to the trimming from the sample of households with either negative or very large ratios of savings to consumption. Preliminary investigation supports the hypothesis that changes in the economic and policy environment help explain the different saving behaviour of different birth cohorts. Tentative results suggest that more favourable environments are associated with lower rates of lifetime saving, although more research is needed to confirm this finding.