In response to:
Destiny's Tots from the October 8, 1981 issue
To the Editors:
Christopher Jencks’ review of Richard Easterlin’s Birth and Fortune: The Impact of Numbers on Personal Welfare [NYR, October 8] is seriously misleading. He takes Easterlin to task for ignoring data sets that could have been used to test his hypotheses regarding cohort size and relative income influences on fertility. But two of the three data sets that Jencks cites are inappropriate. One compares occupations of fathers and sons rather than incomes; the other compares son’s income to current father’s income, rather than to father’s past income as Easterlin’s hypotheses would require. Only the Kalamazoo data are directly pertinent (though obviously limited in scope), and these are weakly supportive of Easterlin.
A more fundamental issue is whether the type of individual-level data that Jencks promotes can provide a thorough test of Easterlin’s model. To the extent that material aspirations depend on general economic circumstances in childhood, and not simply on one’s own father’s income, direct comparisons of fathers to sons are inadequate. One must use some kind of aggregate-level data, presumably of the type that Easterlin employs. Easterlin makes it clear that he is using “past income of young man’s parents” as a notional proxy for “material aspirations of the couple.” He makes operational this variable by using income of the father’s generation, a very different variable than the one available in the data sets that Jencks advertizes. Easterlin’s language uses father-son pairs for convenience but this language is not essential to the relative income-generation size argument. At most he should be accused of imprecision; the charge of shamanism is grotesquely out of line.
That fertility change can’t be adequately understood through micro-level socioeconomic data is implicitly admitted by Jencks in his own attempt to explain fertility swings in terms of “collective decisions…growing out of the mood of the times.” He goes on to suggest that the demand for children might be comparable to the demand for fads and luxuries such as hula-hoops, designer jeans, and cocaine. The reader can judge for himself whether Jencks’ approach is more satisfying than Easterlin’s. There is at least macro-level empirical support for the latter, while the former seems inherently untestable. Easterlin could be wrong, but he’s stuck his neck out with a proposition that will be severely tested through a massive “natural experiment” to occur within the next decade. In the process, he’s sparked new interest in economic-demographic relations and stimulated a wave of research. Jencks’ “honorable” but myopic technician thinks he is blowing the whistle but he’s succeeded only in fouling the air.
Samuel H. Preston
University of Pennsylvania
Christopher Jencks replies:
The “Easterlin hypothesis” appears to come in two variants: Easterlin’s own version, as described in Birth and Fortune, and the variant set forth by his Pennsylvania colleague Samuel Preston.
Easterlin’s own hypothesis asserts that if a couple is having trouble matching their parents’ standard of living, they will have fewer children than if they find it easy to live better than their parents. Preston’s variant asserts that it does not matter how a specific couple is faring relative to their parents. What matters is how the couple’s generation as a whole is faring relative to the previous generation. If the generation as a whole is doing well, all couples will have lots of children, even if they themselves are worse off than their parents. If a generation does badly relative to its predecessor, all couples will restrict their fertility, even if they themselves have more money than they know what to do with. Easterlin’s version of his hypothesis is theoretically elegant, but fails to fit the available evidence. Preston’s version fits the evidence, at least for the United States from 1940 to 1980, but its theoretical rationale is weaker than Easterlin’s.
The theoretical problem posed by Preston’s variant of the Easterlin hypothesis is why changes in the relative income of a generation should affect all members of the generation equally, regardless of their own relative income. The only plausible answer I can see is that the relative income of a generation affects “macro” phenomena like “the national mood” or “cultural values.” While feelings of affluence or hard times must surely exert such effects, they are hardly unique in this respect. The affluence of the late 1940s and 1950s relative to the 1930s may have boosted fertility by making people “optimistic,” for example. But much else was also involved. Thus, there is no compelling reason to suppose that the modest improvement in young people’s relative incomes that is likely to be associated with shrinking generation size in the 1980s and 1990s will produce another fertility surge unless there is a more general economic upturn, and even that might not prove sufficient.
For precisely this reason Easterlin’s book tries to minimize the importance of “national mood,” “cultural values,” and other squishy, unpredictable explanations of fertility. Instead, Easterlin offers a theory in which couples’ behavior depends on their own life histories. By tying fertility to individual experience, Easterlin seeks to make it predictable, in a way that the national mood can never be. This seems to me an admirable goal. I only wish it had succeeded.
The difficulty, as I noted in my review, is that Easterlin’s theory requires precisely the sort of “micro” data that Preston claims are unnecessary. Easterlin himself recognizes this in Birth and Fortune, saying that while such data would be ideal for testing his theory, they were not available. Preston supports Easterlin’s claim in his letter, claiming that two of the three studies I cited are inappropriate for testing Easterlin’s theory.
Preston first suggests that we cannot test Easterlin’s theory about the effects of relative income with data on relative occupational status. While occupation is not the only determinant of a man’s income, it is a major determinant. Other things being equal, therefore, relative occupational status should serve as a moderately good proxy for relative income. If relative income has a large effect on fertility, as Easterlin claims, relative occupational status should have some effect too. Since no such effect is visible, the relative income hypothesis must be treated skeptically until better evidence becomes available.
Fortunately, better evidence is available. Preston objects to the second study I cited because he says it measured the incomes of fathers and sons at the same point in time. In fact, the study in question measured parental income when sons were finishing high school. It measured the sons’ income and fertility in each of fourteen years after high school. It found no relationship between relative income and fertility.
Preston is also wrong, in my judgment, when he claims that the Kalamazoo brothers study “weakly” supports the Easterlin hypothesis. It is true that the Kalamazoo study shows that the brother with the higher income tends to have slightly more children than the brother with the lower income, as Easterlin’s theory suggests he should. But the issue is the size of the difference, not its direction. The difference between Kalamazoo brothers is far too small to account for a significant fraction of the observed fertility swing between 1940 and 1980.
In this connection I should note an unfortunate error in my review. The review claimed that between 1957 and 1977 fertility fell “at least 40 and perhaps more than 100 percent.” This is plainly absurd. Fertility cannot fall more than 100 percent—though only one reader seems to have noted this obvious fact. The correct figures are 30 to 50 percent.
Finally, I would like to apologize for the suggestion that I thought Easterlin guilty of “shamanism.” I used the term to describe “supply-side” economics, as practiced by the Reagan administration. If the context implied that I meant to tar Easterlin with the same brush, I am sorry.