As baby boomers continue to retire from the workforce over the next decade, there will be fewer workers to support children and the elderly, and many fear that economic growth will lag as a result. But the authors of a recent article in Demography, the flagship journal of the Population Association of America, would have us relax: dependency ratios, they argue, have little to do with economic growth.

This is surprising because most of us intuitively know that these ratios matter. My husband is the only child of his mother, so none of the costs of her support can be shared with siblings. Plus our three-year-old daughter would be in Montessori school if she didn’t have a younger sister: we could invest in her education now if we didn’t still need a babysitter in order to work. It doesn’t take much imagination to abstract family issues up to the societal level in a country where Medicaid will pay for nursing care for destitute elderly and where all kids over age five can receive publicly funded education.

In short, it makes sense that a lot of dependents could necessitate a lot of expenditures, and that the number of dependents per paycheck—not simply the number of dependents—would be crucial. Theories about the “demographic dividend” build on this idea: When less is spent on dependents, more can be invested (in physical and human capital) to contribute to future growth, meaning a favorable age structure boosts economic development.

But coauthors Jesus Crespo Cuaresma, Wolfgang Lutz, and Warren Sanderson challenge us to think not just about the dependency ratio, but also about the role of education in determining both the dependency ratio and economic growth. Imagine two families, both with independently employed husbands and working wives, but the one family has one child and the other three. The logic of the demographic dividend would predict that the one-child family would have an easier time saving because they would buy less food and fewer shoes, and therefore they could more easily accumulate a surplus to invest in things like advertising for the husband’s business.

What we’ve called a demographic dividend may really be an education dividend.

Crespo Cuaresma and his co-authors would re-write this scenario to involve education, too. Let’s say the wife in the one-child family is an investment banker—more educated women both command higher wages and choose lower fertility—and the wife in the three-child family is a daycare worker. Now think about what their jobs mean for investment in their husbands’ businesses: the families’ varying expenditures on food and shoes are small compared to the differences in the wives’ earnings. It isn’t that expenditure doesn’t matter; it’s just that productivity matters more.

Thus the authors conclude that what we’ve called a demographic dividend is really an education dividend. That is, when education reduces fertility and creates a relatively small young generation, the resulting economic growth is mostly from education itself rather than the favorable age structure. Shifts in the age structure can help and hurt, but relatively little. An educated population has fewer young dependents, and even though they have more elderly dependents, they are productive enough to afford them.

This theory, if it is correct, has major implications for richer countries. First, growth in the proportion of elderly would not create a large drag on economic growth. Second, a drop in U.S. fertility from 1.9 children per woman (its current rate) to Europe’s level of 1.6 children per woman, which would reduce the number of young dependents, would provide only a moderate boost to the economy. Similarly, the fact that Europe’s fertility is unlikely to fall further does not portend economic difficulties—and neither would a rebound in European fertility.

The role played by favorable age structures in helping foster education and launch growth remains untested.

But Cuaresma, Lutz, and Sanderson based their investigation on a test for the importance of a demographic dividend in a sample of 105 heterogeneous countries, whereas the theory was developed to explain how the departure from high fertility (6–8 children per woman) could provide a one-time boost to economic development. Improvements in infant and child mortality (e.g., childhood deaths dropping from 200 to 100 per 1000 births) created a relatively large young-age population, and the resulting “demographic overhead” presented an obstacle to further economic development. Fertility decline had the opposite effect, creating a relatively small young-age population and with it fewer maintenance requirements and more opportunity to invest in human or physical capital: the demographic dividend. These changes take place in the initial transition from high to low fertility; we would not have expected the relatively small changes in fertility that characterize today’s wealthy countries to have effects of large magnitude.

Extending the concept of a demographic dividend to favorable age structures more generally does not answer the question of whether the demographic dividend helps countries escape the poverty trap. Even the authors’ finding that past education levels and current improvements in education both promote growth avoids the question of what helped produce the past education levels. Nobody would argue against the importance of investing in education for economic growth, but the role played by favorable age structures in helping foster education and launch growth remains untested.

Moreover, even if we accept the authors’ conclusions for the contemporary world, we would still do well to consider how healthy families can create both favorable age structures and favorable education trends. Put simply, below-replacement fertility is associated with the retreat from marriage, and children from intact families have better educational outcomes. Stable marriage can help keep fertility close to the replacement rate, which makes old-age support easier, while simultaneously making educational progress easier. Even if education contributes far more to economic growth than a favorable age structure does, both means can be pursued together.