What Would Have Been: Exploring Counterfactuals in Demography and Health

volume 1, issue 9 October 1, 2006

By Robert Engelman

Whatever one’s view about population as an issue, few people fervently wish the world were home to a lot more human beings than it is. Some may wonder if another Mahatma Gandhi or an Albert Einstein or a Mother Theresa missed out on being born due to the declining global birthrates of the past few decades. But most know that such a question is fundamentally unanswerable and don’t stay awake at night thinking about it.

By contrast, it’s entirely feasible to construct and ponder alternative realities in which human fertility had not fallen as sharply as it has over the past few decades, and, as a result, the world has 9 billion people rather than today’s 6.5 billion. Or, perhaps, a specific country would have to push much harder to achieve key international goals for reducing poverty or maternal and child mortality. Such exercises, which demographer Patrick Heuveline called “seemingly gratuitous,”1 actually can help demonstrate the value of the investments that governments and societies have made in improving access to family planning services over the years. As such, the exercises are a valuable tool, although so far only rarely used, for those advocating the expansion and improvement of sound reproductive health policies and programs based on voluntarism and rights.

These artificially constructed alternative realities are dubbed counterfactuals. Their building blocks are carefully selected trends and developments that run counter to those that actually unfolded. In one sense, of course, true counterfactuals are as unknowable as the non-birth of a future Gandhi. The factors that influence events are too numerous and too interwoven for us ever really to know what would have been if one key trend had been different. The mathematics of demography do allow us, however, to construct simple “if-then” thought experiments on specific quantitative outcomes derived from the same sorts of inputs. Among the more interesting of such exercises is to simply hold average families in a population at some size they had in the past, or decrease family size more slowly than actually occurred. Then we can see what would have happened, all else being equal, to population size and to key health and survival outcomes.

One approach to alternative demographic outcomes was popular in the 1970s and 1980s, when concern about world population was a bigger driver of interest in international family planning than it is today. A term that appeared occasionally in research on family planning was “births averted.” With some specified increase in the use of contraception over time, analysts calculated, a certain number of births would not happen that otherwise would have, leading to lower rates of population growth. Higher numbers or proportions of births averted implied greater success for a particular level of program effort or perhaps a particular method of contraception. As the field of population and family planning evolved toward the more holistic one of reproductive health in the 1990s, however, the image of averted births became less appealing for advocacy. It hinted to some not only at the idea of babies that “would have been,” but at the idea, albeit not the reality, of abortion. Today the phrase appears only occasionally in demographic research.

Counterfactuals remain useful for advocacy if we consider broader aspects of demography and health and new areas of interest in the field. This may be as simple as an overall number indicating what the population of the world or a nation would have been had fertility trends remained high instead of falling as they did. Or it may be as sophisticated as an illustration of what a country’s age structure (a depiction of a population indicating how it is divided among different age groups) would have been—at a time when age structure is much more interesting to demographers and other analysts than in years past. In one new set of calculations, counterfactuals are even being used to document how investments in family planning have eased the challenge governments face in reaching the United Nations’ Millennium Development Goals related to poverty reduction, maternal and child health, access to safe water, the reduction of HIV/AIDS prevalence, and environmental sustainability.

A Different World

Demographers have been experimenting with alternative demographic scenarios at least since 1956, when Princeton’s Ansley J. Coale showed how age composition might have been different at that time had mortality and fertility trends evolved differently in earlier decades.2 In the late 1990s, Heuveline, of the University of Chicago, developed a model that allowed him to project age structures and populations forward from the year 1955 to the year 2000, based not only on the reality of what actually unfolded, but on the alternative fictions that either global mortality rates, or global fertility rates, or both, had remained unchanged at their 1955 levels. It was the demographic equivalent of imagining that everything is the same this afternoon on Fifth Avenue in Manhattan, except that all the cars have fins.

In the demographic case, however, the differences were much more than cosmetic. Had fertility not fallen below its 1955 levels of five children per woman worldwide (instead of falling almost to half that value by today), Heuveline calculated, the world would have had 8.3 billion people in the year 2000 instead of 6 billion. This suggests almost 9 billion people today, instead of slightly more than 6.5 billion, assuming no change in the mortality-rate trends that actually unfolded.

The differences in age structure are equally intriguing. Had fertility remained at 1955 levels throughout the 20th century, the same number of people would have been 60 years old or older—of course, since they were all born before 1955. Similar numbers of people under both scenarios would have been from ages 30 to 59 for a slightly more complicated reason: the fertility decline that actually happened after 1955 didn’t really get rolling significantly until much later, in the 1970s. That means, not surprisingly, that many hundreds of millions more people would have occupied a significant global youth bulge, under age 30, and especially under age 15. [See Figure 1.]

FIGURE 1: WORLD POPULATION AND AGE STRUCTURE IN 2000, ACTUAL AND COUNTERFACTUAL

Source: Heuveline, P. 1999. The Global and Regional Impact of Mortality and Fertility Transitions (1950-2000). Population and Development Review 25(4): 681-702, December.

 

 

For certain regions, and at least one country, Heuveline’s alternative scenario is especially impressive. In constructing his model, the demographer needed to divide the world into regions to account for the geographically diverse fertility and mortality trends of the last five decades. One of his regions was made up of a single country—China—simply because that country includes more than a fifth of the world’s people. Had China’s fertility not fallen from more than 6 children per woman in 1955 to slightly more than 1.7 in 2000 (and this is no endorsement of the country’s one-child policy, which was only partly responsible for that decline), the massive country’s population would have been 2.34 billion, compared to its actual population of 1.3 billion today.

The underlying message of this exercise? We can be confident that if family planning had not risen in popularity and availability as it did in the last third of the 20th century, world population would have been significantly greater than it is today, and significantly younger. The implications of that may be arguable, but not the fact itself.

A Different Egypt

In the past few months, research investigator Scott Moreland at Constella Futures, an international health policy group, has developed a counterfactual scenario for a single country, Egypt. Moreland’s scenario demonstrates not just population outcomes, but how much harder the government of Egypt would have had to work to achieve the Millennium Development Goals by 2015 if fertility had fallen much more slowly than it did after the 1970s.3

“The study in Egypt was commissioned by the Egyptian Ministry of Health and Population because they knew they would have to take over from the declining USAID-funded program, and they had to defend the request for a budget request in front of skeptics,” Moreland reports, in describing the importance of this effort for advocacy.

“With global funding for reproductive health under threat, especially to fight HIV/AIDS, it is important that decision makers know what has already been achieved and therefore what can continue to be achieved. Even with the fuzziness of establishing a counterfactual scenario, it may be more convincing to look at the past rather than gaze into the future as we usually do with such modeling exercises.”

Moreland used different assumptions than did Heuveline. Rather than hold fertility steady at some past rate, Moreland imagined that use of contraception in Egypt had climbed as slowly as in sub-Saharan Africa—at roughly 0.5 percent a year—rather than at the much more rapid rate recorded in Egypt, 1.8 percent a year. Under this simple assumption, Egypt today would have a fertility rate of 4.6 children per woman, rather than its actual rate of 3.1. The country’s population would be approaching 84 million people, 12 million more than the actual number. And as with Heuveline’s alternate human population, the additional human numbers would be weighted overwhelmingly among the youngest age groups, especially those under five years old.

In large part because of this counterfactual weighting towards the youngest, the real Egypt has about one third the rate of infant mortality calculated by Moreland for his “alternate Egypt,” avoiding 3.1 million infant deaths that would have been expected to occur under the lower-contraception, higher-fertility scenario. This is in large part because there would have been a much higher proportion of closely-spaced and other high-risk pregnancies. The real rate of under-five mortality is also impressively low—well under a third what it would have been counterfactually—while a calculated 17,000 maternal deaths did not occur between 1980 and 2005 due to the higher use of contraception and consequent lower fertility.

Similarly, Moreland was able to calculate significantly higher expenditures for schooling, child immunization, and food subsidies per capita that would have had to occur, all else being equal, under his alternative high-fertility scenario. In effect, he found, for each Egyptian pound of expenditure for the country’s family planning program during the 25-year period, the government saved nearly 20 pounds it would otherwise have had to spend on education, child health and food subsidy costs. [See Figure 2.]