Unobservable Family and Individual Contributions to the Distributions ofIncome and Wealth

This paper uses a data set composed of combinations of full brothers, half brothers as well as fathers and sons to measure the effect of common family background on households'income and wealth. While the data is drawn from a nineteenth century population, the intra-class correlation (after the effects of age, occupation, nativity, residence and duration in the economy have been removed) for income ranges from .13 to .18 which is similar to that found in modern samples. Intra-class correlations for wealth are significantly higher (.18 to .35) than those for income. The addition of fathers' observed characteristics to the sweeping regressions reduces the unobserved common background effect shared by brothers by about twenty percent.The intra-class correlations of half brothers were lower than those observed for full brothers though the small differences between the two groups suggest that fathers played a dominant role in the transmission of the common family effect. Unobserved background was decomposed into individual and family effects by a variance components procedure. The individual effect was dominant for income while the family effect was dominant for wealth.