Comment on W. Whitney Hicks’ “economic development and fertility change in Mexico, 1950–1970”

Abstract

W. Whitney Hicks has examined a puzzling problem of modern economicdemographic history: the continuation of high fertility in Mexico in the presence of rapid and persistent economic growth (Hicks, 1974). The theory of the demographic transition (DT) implies that birth rates should be falling in Mexico, and DT optimists suggest that a decline is imminent (Oechsli and Kirk, 1974). While I concur with Hicks' conclusion that a rapid fertility decline is not imminent (p. 419), his results and analysis, particularly his findings on the influence on fertility of various independent variables, are subject to question. An important initial issue is whether Mexican age-specific fertility fell between 1960 and 1970 as Hicks claims (p. 412). Hicks analyzes a number of his results in terms of this decline, and yet there is evidence that age-specific fertility may have risen between 1960 and 1970. Table 1 presents evidence on mean children ever born per woman in specified age groups. These increases among women in the most fertile age groups are not easily reconciled with Hicks' assertion of falling fertility in the period 1960-1970. Hicks' measure of age-specific fertility is derived from vital statistics data which have been adjusted for underreporting and delayed registration. One cannot know what errors remain after this adjustment process. Similarly, census data on children ever born, employed by Hicks in much of his analysis, may be flawed by resnonse error. While a total fertilitv rate cannot be derived directly from census data on children ever born, it is unlikely that period fertility would decline at the same time that the number of children ever born to young women was rising. A more complex question is whether or not various factors hypothesized to influence fertility are statistically significant. Hicks notes that his results should be interpreted with caution (p. 414), but I do not believe this customary caveat is sufficient. Utilizing similar techniques to analyze Mexican fertility data in the 1960 and 1970 censuses, I obtained radically different results from those of Hicks. Child-woman ratios for the states of Mexico were chosen as the dependent variable for cross-sectional fertility analysis. I believe these ratios, when corrected by estimates of underenumeration and infant mortality, are as reliable as Hicks' measures of adjusted period fertility. Although official statistics of infant mortality are not reliable, Roberts (1973) has recently reported on corrected levels of infant mortality by state in Mexico in 1960. These corrected rates can be used to directly inflate the reported childwoman ratios to account for infant mortality. Regional variations in underenumeration, which are substantial, are a more difficult problem. Roberts (1973) has also measured the degree of underreporting of infant mortality in Mexico in 1960, and these estimates are used as proxies for the degree of underenumeration of children 0-4. This adjusted de-