Xiaofen Lin

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Economics / Economic Policy


Professor John Burbridge


Professor Martin Browning

Committee Member

Professor Lonnie Magee


The conventional life-cycle model predicts that households save period to retirement and use their savings to finance consumption during retirement. Thus whether households dissave after retirement or at older ages is critical to the validity of life-cycle theory. This question is also of concern to policy makers as aggregate saving and investment, in Canada and elsewhere, may fall as the population ages. Many studies based on aggregate data for a cross-section of OECD countries provide support for the life-cycle hypothesis, since the data suggest that countries with a greater proportion of elderly people have lower household saving rates. On the other hand, most empirical research directed at household consumption and saving behaviour based on household data has found little evidence that supports the life-cycle hypothesis. The three essays comprising this thesis attempt to establish and explain the micro evidence on saving behaviour of older households, and also try to overcome some of the usual barriers to using cross-section survey data in empirical research in the field. All three essays employ Canadian FAMEX data from 1969 to 1992. Like much other research in the field, the empirical work reported in Essay 1 in this thesis provides evidence against the prediction of life-cycle theory that households dissave at older ages. It is found that the median saving rate for older households exhibits a distinct age pattern: it drops sharply at retirement age, but then rises, thereby forming a saving dip. The most important contribution of Essay 1 is to address two well-known problems, cohort bias and differential mortality bias, arising from using cross-section survey data. Cohort bias is dealt with by forming a synthetic longitudinal sample from repeated cross-section data, and a new method is developed to correct the differential mortality for the age profile of the saving rate. However, the puzzle in the saving pattern of the elderly still remains after the corrections; the median saving rates are positive and rising with age after retirement. Essay 2 focuses on the estimation of the saving rate as a function of various income sources as well as age. During the transition from work to retirement, households experience a dramatic change in composition as well as in the level of their income. Following the permanent income hypothesis, if consumers have different perceptions of the permanence of different income sources, they would react differently to income changes, depending on which income component changed. The main finding of the essay is that, in the after-retirement period, the "pure" aging effect is solely responsible for the rising trend of the saving rate. However, for a given level of total income, higher pension income is associated with a higher saving rate, while higher transfer income is associated with higher consumption. Essay 3 examines the change in the composition of consumption demands as households age, and how the effects of three factors - age, total expenditure and retirement - contribute to this change. It is found that the "savings puzzle" comes largely because reductions in food consumption at home, private transportation expenditure and perhaps tobacco/alcohol spending due to age alone are larger than the offsetting age effect associated with an increase in gifts to other households by the elderly. On the other hand, for some reason, older households largely "obey" the cross-section income elasticities in reducing their consumption of most goods as their incomes fall with age.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Included in

Economics Commons