Wealth accumulation and the life-cycle in economic history: Implications of alternative approaches to data

Using census-linked probate data from nineteenth century Ontario, this paper finds that the age and age-squared specification used in most parametric wealth regressions to capture life-cycle behavior overestimates wealth accumulation compared with estimates constructed using a simple nonparametric technique known as LOWESS. As a result, many of the high annual rates of wealth accumulation estimated by studies of nineteenth century North American wealth may be the result of functional form. Moreover, the nonparametric wealth-age profiles support for the existence of early and middle adulthood life-cycle "squeezes" suggesting the traditional hump-shaped wealth-age profile does not fully capture the dynamics of the economic life-cycle. (C) 1998 Academic Press.