The Census Bureau's 1989 Projections of Future U.S. Population

Which Scenario is Reasonable?

By CIS March 1989

Which Scenario is Reasonable?

"In 1933 Herbert Hoover's Committee on Social Trends predicted that 'we shall probably attain a population of between 145 and 150 million during the present century.' This forecast was off by 100 million. In 1938 Franklin Roosevelt's Committee on Population Problems opined: 'Even with the highest rates that can be reasonably assumed, there would be a natural increase of less than 50 million from 1935 to 1980.' Off by 50 million.

A few years later, demographers were talking about a 'population explosion.'"1

Arthur Schlesinger, Jr.

The Census Bureau's just-released population projections to the year 20822 have again triggered warnings in the media and among the demographic doomsayers. Their somber view is that the country faces demographic decline unless Americans have more children and allow in more immigrants.

As often happens when the Census Bureau publishes a series of projections, the media single out Census's middle range scenario and rectify its numbers. The verb "would," carefully chosen by Census, is replaced by the definitive "will." In this instance, the Census Bureau prepared 30 scenarios. But Level 14 became for many commentators the "most likely" outcome for what will transpire, demographically, over the next ninety years.

Changing Assumptions, Changing Projections

The projections are, by definition, correct. The Census Bureau demographers are among the world's most talented and their projection techniques are particularly sophisticated. The Census report itself explains that: "The projections represent the mathematical outcome of assumptions about future trends in fertility, morality, and net immigration." What most of the media neglect to mention are the assumptions underlying the projections. Eventually these often prove to be incorrect. Indeed, that is one of the reasons why the Census Bureau periodically updates its projections — the assumptions must be altered based on new evidence about fertility, mortality, and migration.

Given the wide range of possibilities in fertility, mortality, or migration, the Census Bureau develops a variety of different projection scenarios that cover all reasonably possible outcomes, from very low to very high. The middle scenario (Level 14) is just that — it falls in the middle. That is the scenario which grabs the attention of the media, with no mention about how those numbers were derived. (To be fair to the media, the Census Bureau encourages this middle range emphasis by concentrating on Level 14.)

Projections Developed From Assumptions?

First, a base population is selected. In this case, the estimated July 1, 1986 population serves as the base. This population, in turn, was projected from the 1980 Census and the estimated undercount in that census was included in the 1986 estimate. However, the Census Bureau had no way at that time of knowing how many persons, here illegally, would opt for the general and farm workers amnesties under the 1986 Immigration Reform and Control Act (IRCA). In fact, about 3 million have come forward. How many had been counted in 1980 is impossible to determine.

Next, three alternative rates for fertility, mortality, and migration were chosen. These form the basis for all projections. The three TFRs are 1.5, 1.8, and 2.2. The three life expectancy rates are 77.9, 81.2, and 88.0. The three levels of net immigration are 300,000, 500,000, and 800,000.

What are the assumptions used to develop the Level 14 scenario — the so-called "most likely" projection of the nation's population for the next ninety years? The answer is simple: the middle level for each demographic variable. If the TFR is 1.8; if life expectancy is 81.2 years; if net immigration is 500,000 annually, then the numbers widely reported in the media will be the Untied States population in future years — 268 million at the turn of the century; peaking at just under 302 million in 2040 and dropping to 292 million by 2080.

The Volatility of Fertility Trends

Are these the most reasonable and realistic assumptions to use about our demographic future? No one really knows. Mortality trends are fairly easy to project. However, fertility rates and migration levels can be quite volatile. The medium fertility rates and migration levels need to be examined in more detail.

The U.S. population is divided into three subgroups: white, black, and "other." Hispanics are not now counted as a distinct racial category — a surprise in view of the Census Bureau's recently published report projecting the Hispanic population.3 Presumably, a substantial majority of Hispanics are included in the white subgroup.

The TFR for white is assumed to be 1.749 in 1987, and then gradually increases to 1.8 by 2020. census's report points out that: "The TFRs for white women exhibit no particular trend [in recent years] even though the white Hispanic share of all white births ahs been growing quickly. This indicates that white non-Hispanic fertility rates probably have fallen slightly. Given the assumption of eventual convergence between Hispanic and white fertility levels, it is a reasonable assumption to reduce the ultimate fertility level."4 (That is, from 1.9 in an earlier report to 1.8 in this report.) However, the current fertility rate of Hispanic women is about 2.8 and, as its share of the white population increases, will its assumed fertility decline be rapid enough to compensate for its growing share of the total white subgroup? The TFR for blacks is currently 2.156 and is projected to fall to 1.8 by 2050. Similarly, the rate for "other races" is now 2.147 and is also projected to fall to 1.8 by 2050.

Projecting the fertility of immigrants, the Census report states: "Immigrants are assumed to immediately begin bearing children at the same rate as the equivalent age-race group in the U.S. population that year."5 This suggests that Mexican or Filipino women immigrating at age 20, for example, would exhibit the same fertility rates respectively as white and other races women would in any given year. As the report itself points out: "Recent survey information indicates that this fertility assumption may be too low."6 Thus, it is possible that the medium fertility level, widely publicized as being 1.8, may be somewhat misleading.

Projecting Future Immigration in a Crowded World

Medium net immigration is assumed to be 685,000 in 1985, falling to 575,000 in 1990, and standing at half a million by 2000. According to the report, the 1985 figure includes 100,000 undocumented immigrants. This is the lowest level of Census's assumed yearly illegal immigration of 100,000 to 300,000. Emigration is assumed to total 160,000 yearly. Also included in the net immigration total are 33,000 Puerto Ricans who, of course, are not international migrants.

Because of the failure to separate Hispanics as a subgroup, the net immigration levels by race are somewhat confusing. The report assumes that of the 500,000 annual new settlers, 273,000 will be white, 54,000 black, and 173,000 other races. Presumably, a large portion of the white immigrants will be Hispanic. It is difficult to second guess these numbers. Projections are simply the mathematical results of certain stated demographic assumptions. However, one may question their reasonableness as the medium scenario.

Legal immigration has climbed consistently since World War II and shows no signs of leveling off. The unrestricted immigration of immediate family members has grown by more than 6 percent yearly since 1970. Two major legal immigration reform measures now under consideration in the 101st Congress would further boost legal immigration.


  • The Kennedy-Simpson bill (S 358) would increase worldwide immigration to 590,000 yearly, exclusive of refugees and asylees — about 100,000 more than current non-refugee immigration. The bill would cap non-refugee immigration at 590,000 but would allow for periodic reviews and adjustments.

  • The Berman bill (HR 672) in the House, mirrored in many aspects by the Simon bill in the Senate (S 448), would boost non-refugee immigration immediately to over 700,000. By entitling immediate family of permanent residents and special immigrants to unlimited entry (a privilege now allowed only to immediate family of U.S. citizens), the Berman bill would encourage robust future growth.

Refugee admissions, after declining in the early and mid-1980s, are rising again, in large part because of liberalized Soviet emigration policies and continued heavy outflow from southeast Asia. Nearly 95,000 refugee admissions are authorized for 1989, up 23 percent from 1988.

The legalization of 3 million predominantly young and Hispanic former illegal immigrants may well stimulate sizable follow-on immigration for family reunification. And there is still little solid evidence that the 1986 IRCA is producing a rapid decline in the number of illegal entries. Given the circumstances, a higher projected level of net immigration than 500,000 might be more realistic for the most likely scenario. Census's high projection of 800,000 yearly now seems more reasonable for a medium projection.

How Likely is the "Likely Scenario"?

The so-called "most likely" scenario may not be most likely after all. Fertility could rise; it could even fall. Immigration levels could rise. The numbers could also stabilize if a backlash resulted from increasing numbers. The Census report's 30 different projections allow for individual judgment in selecting which may be the "most likely" scenario.

Let's examine two of these, beginning with the high side. Some "neo-doomsayers" have expressed concern about a possible reduction in U.S. population around the middle of the next century. Two years ago, the pleas was for American women to have more children. Now another approach is stressed. As a report in U.S. News and World Report states: "Immigration should be raised."7

Fortunately, this new Census Bureau report allows us to assess the potential impacts of shifts in demographic behavior. Under the high fertility assumptions, the TFR for white women would gradually climb and reach 2.2 by 2025. For blacks and other races, which are already close to that level, the rate would climb slightly before settling at 2.2 in 2050. This is not a particularly high fertility rate — it is barely above the level needed to replace the population. Under the high immigration assumptions, the net amount would rise to 800,000 per year, representing immigration of 1 million and emigration of 200,000. That amount may not be too far from current levels depending on our assumptions about the actual number of illegal entries in any given year.

What would be the result of a combination of high fertility (2.2 live births per woman), high immigration (net of 800,000 per year) and mortality at the same medium level as in the "most likely" scenario? According to the Census report, the population would reach 276 million in 2000; 394 million in 2050, and by 2080, the U.S. population would be rapidly approaching the half billion mark (471 million people). Some economists, social scientists, and business leaders may argue that such growth would be good for the country; but it raises clear questions about the quality of life, political institutions, and environmental purity in such a society. Consider, for example, 50 million people living in California, with perhaps another 30 million each in Texas, New York, and Florida. Without any movement to such under-populated places as the Dakotas and Montana, such regional population concentrations would be a distinct possibility.

Slower Growth, But a More Populous Nation

Now let us look at a slower growth scenario. The fertility remains the same — that is, the medium assumption where all TFRs converge at 1.8; assumed net immigration is reduced to 300,000 per year, and life expectancy remains at the medium level. The results are quite interesting. By 2000 the population would reach 264 million; it would peak at 288 million in 2030, and by 2080, it would total 266 million, or about 20 million more people than the present U.S. population.

There are those who would argue that, with such low levels of immigration, labor shortages would become chronic. (The assumption here is that the high U.S. rate of job creation of the past two decades will continue indefinitely.) But at present rates of participation, the labor force of 2080 would be almost 10 million larger than in 1989. Assuming continuation of the present pace of technological innovation, wide categories of jobs would no longer exist. A tighter labor market in the coming decades would stimulate the United States economy to rationalize its use of labor to better compete with Japan, whose economy has boomed with slow labor force growth and almost zero immigration. A closer balance between the supply and demand for labor would help ease the problems of the underclass by encouraging the training of neglected minorities and their full absorption into the work force.

Beyond the economy, quality of life looms large. Even California's freeways might become less congested places with slower population growth. Perhaps there would be less concern about over-burdened landfills, prospective energy shortages, and further depletion of our natural resources.8

The Population Effect of Longer Life Expectancy

So far we have avoided discussing any changes in mortality. All three scenarios have assumed the moderate level, where life expectancy levels off at 81.2 years. However, some gerontologists have criticized the Census Bureau for using such conservative mortality estimates, and overlooking the prospects of major improvements in life expectancy in the next century. In this Census report, the low mortality pattern assumes that life expectancy will reach 88 years by the year 2080. Let us apply this pattern of mortality to three scenarios: "high," "low," and "most likely."

High

With improvements in longevity, future populations, of course, would be larger. Under the high scenario, the 2080 population would have already surpassed half a billion mark and still be growing.

Low

Under the low scenario, the population would stop growing in 2040 when it would reach 301 million. By 2080, it would total 290 million. This is remarkably similar to the highly-publicized "most likely" scenario.

Most Likely

Those who warn of population decline under the medium scenario might note that with lower mortality (and medium fertility and migration) the population would reach 270 million in 2000, peak at 318 million in 2050, and remain at about that level until 2080. No population decline would occur.

Perhaps, rather than fret about not enough babies and not enough immigration, the neo-doomsayers should join others in advocating measures to increase the life expectancy of all Americans. To be sure, such a society would be "older" than now, but with an elderly population of greater vigor and a longer average working life. But are we as a nation incapable of adjusting to a population where the median age is well over 40? Is 65 a sacrosanct retirement age? If life expectancy were 88 (and it would only reach that point in several decades) would it not follow that a 65 or 70 year-old person would be healthier than his counterpart in 1990?

The Census Bureau study deserves praise for its professionalism. The serious student of America's population future can learn a great deal from the multitude of data it provides. It is risky, however, when the media and other analysts disregard the author's caveats and treat Census population projections as predictions. All of us might well remind ourselves once more that projections are simply the mathematical results of assumptions about future trends in fertility, mortality, and migration. It approaches recklessness to base policy choices with enduring effects, such as immigration or pro-natalist programs, on such projections. They can alert us to certain trends that may or may not be seen as dangerous. But in no way should they be considered predictions of the future.


End Notes

1 Arthur Schlesinger, Jr.: "Future Always Outwits Political Pundits," Wall Street Journal, February 8, 1989.

2 U.S. Bureau of the Census: Current Population Reports, Series P-25, No. 1018. "Projections of the Population of the Untied States, by Age, Race, and Sex: 1988 to 2080," by Gregory Spencer, Washington, D.C., 1989 (hereafter "Census Projections 1989").

3 U.S. Bureau of the Census, Current Population Reports, Series P-25, No. 995. "Projections of the Hispanic Population: 1983 to 2080," by Gregory Spencer, Government Printing Office, Washington, D.C., 1986.

4 Census Projections 1989, p. 21.

5 Census Projections 1989, p. 21.

6 See U.S. Bureau of the Census, Current Population Reports, Series P-20, No. 421, "Fertility of American Women: June 1986," U.S. Government Printing Office, Washington, D.C., December 1987.

7 Ben J. Wattenberg: "The Case for More Immigrants," U.S. News and World Report, February 13, 1989, p. 29.

8 See Leon Bouvier: Will There Be Enough Americans?" Center for Immigration Studies, Washington, D.C., July 1987.