The Parker-Gallman sample and wealth distributions for the antebellum south: A reply

Abstract

In a recent paper in this Journal, I argued that “the extent to which the wealth distribution in the rural South was more unequal than was that in the rural North was significantly greater than implied by recent quantitative studies” (Yang, 1984, p. 89). I showed that the inequality measure computed from the Parker-Gallman (P-G) sample was biased downward both by sample truncation and the use of farm value as a proxy for real property. Taken together, the Gini coefficient of wealthholdings by Southern farm households in 1860 was estimated to be understated by approximately 10 percentage points. To recapitulate, the bias comes from (i) exclusion of (poor) farmers who were not listed in the agricultural schedules, i.e., categories C and D as defined in that paper’; (ii) exclusion of part of the tenants of category B; and (iii) the usage of farm value as a proxy for real estate, with the consequence of ignoring nonfarm real estate and multiple ownership, and of falsely attributing the ownership of rental farms to tenants. The third point was recognized by, among others, Gallman (1969) and Wright (1978); the second, by Bode and Ginter (1980). According to a recent estimate by Bode and Ginter (1984, their Table 2), out of 64 category-B tenants that should have been included in the Parker-Gailman sample, in 94 five-farm blocks (in 48 selected Georgia counties) that they examined, exactly half were excluded. The extent of the first type truncation (C and D), however, was not yet estimated even for a restricted region.