- Finally, with the exit of the last baby boomers from a particular age bracket, there was a crash in demand for products or services that were targeted to that age group, and a new problem of excess supply emerged.
To trace these effects, we follow the 1950 cohort, who were born in the fourth year of the baby boom generation. For housing markets, the 1950 cohort reached age 20 in 1970, and began the phase that found large numbers of the cohort first leaving home and entering rental apartment markets. Their rising numbers not only stimulated apartment construction in the suburbs, but also led to gentrification and rejuvenation in portions of central cities that were popular with young adults. In the late 1970s, the 1950 cohort entered the prime homebuying ages of 24-34, triggering surges in demand and prices. More recently, the 1950 cohort passed age 45, after which mobility usually slows down and home purchases lessen. Their progress forward from this age is the matter in dispute. Under the cohort interpretation, housing consumption flattens out after this age, but under the traditional interpretation, demand declines markedly.
Population Projections
Examination of Census Bureau population projections (Day 1993) suggests the importance of future changes in each age group. In 1995, the baby boom generation is clearly visible as a hump between ages 30 and 45 (see Exhibit 1). While the 25-29 age group contains only 18.5 million persons, the 30-34 bracket contains 22.5 million. Conversely, there are fewer people in the older age groups than in the advancing baby boom. The projections show how the large hump will move up the age scale, with slight augmentation by immigration and depletion by mortality.
Differences expected in the coming decades (1995 to 2005 and 2005 to 2015) are plotted in the bottom portion of Exhibit 1.
In the first decade, the 30-34 age group will lose more than three million people, decreasing the ranks of entry level home buyers. At the same time, huge increases are projected for the 40-59 age range. If the 40-59 group will be characterized by declining housing demand, the prospects for the housing industry are not good. In the following decade, these age losses and increases ripple forward ten years in age. There should be an increase in potential home buyers in the 25-34 age bracket, but this increase is far outweighed by the increase in aging baby boomers in the 50-69 age range.
Another significant change, although of less importance than changes in age groups, is in the racial composition of the population. Although the African-American or black population will grow slightly from 12 to 13 percent between 1995 and 2015, other minority populations will grow dramatically (see Exhibit 2).
The percentage of Hispanics will rise from 10 to 15 percent, although this population is heavily concentrated in California, Texas, and the southwestern states. The Asian population will also rise, from 3.5 percent to 6 percent, more in California than in other states. The bottom portion of Exhibit 2 shows how the growing minority population is concentrated more in the younger age groups than among the elderly.
The Question of Future Impacts
How will the dramatic increases in the older age groups affect the housing market? A much-publicized study by Mankiw and Weil (1989) projected that house prices would decline by 47 percent in 20 years, based on declines at young adult ages and the expectation that the large baby boom generation would reduce its consumption as it grew older. This study was roundly criticized on several grounds, but most observers accepted the demographic interpretation that housing demand of aging boomers would decline. More recently, a study by Pitkin and Myers (1994) marshaled evidence against the traditional interpretations accepted by both Mankiw-Weil and their critics.
The declining number of potential young homebuyers is not in dispute, as demonstrated in Exhibits 1 and 2. The question is how to predict the future housing demand of the aging baby boom generation. The essence of the problem is that the data reflect the present, but analysts want to be able to predict what future data would show. The traditional view is that the behavior of the baby boomers when they reach older ages can be predicted by reference to the current older age group. The newer cohort view tests this assumption by assembling data from the past and computing the trajectories of cohorts as they age, projecting those trajectories into the future. The results of this method can be surprisingly different from the traditional view.
The heart of the difference is whether we compare age groups at one point in time or follow the same people over time. The age group comparison assumes that all the differences are due purely to advancing age. Thus, young men should become just like their fathers when they reach a certain age. While partly true, there are additional generational differences that will persist when young men reach their fathers' present age. Over time successive cohorts travel through the age groups, replacing their predecessors, and raising or lowering the behavior at each age. The cohort approach tracks not only age, but also the distinct differences and momentum established by each cohort as it advances upward. The importance of a cohort perspective is especially valid for housing market analysis.
One measure of housing consumption is the proportion of households that own large, single-family homes (with seven or more rooms). This measure detects whether older households may be moving to condominiums, mobile homes, smaller single-family units, or switching to renting all together. As shown in Exhibit 3, there is a pronounced age profile visible for large home ownership, with the peak ages in the 40s and a sharp drop-off thereafter. According to the traditional interpretation, the aging baby boomers will move out of their large houses and move to smaller units, potentially glutting the market with resales.
Superimposed on the graph in Exhibit 3 are actual cohort trajectories observed between 1960 and 1990. Cohort A was aged 40-44 in 1960 and was poised for dropping out of large home ownership, according to the traditional view. This cohort reached age 50-54 in 1970, age 60-64 in 1980, and age 70-74 in 1990. Strikingly, Cohort A's proportion of large owned homes held steady for this entire period. In a similar fashion, Cohort B (aged 30-34 in 1960) is traced forward to 1990. Because it was younger at the start of the study period, this cohort experienced a substantial rise in ownership of large houses. In the most recent period, ownership declined slightly, but far less than would have been suggested by the age cross-sectional profile.
The end result of both cohorts' trajectories is that they have pushed the 1990 age profile outward relative to the 1980 profile. For example, in 1980 when Cohort B was aged 50-54, traditional theory would have predicted that its ownership of large homes would drop to the level of 60-64 year-olds when that age was reached in 1990. Instead, Cohort B moved on a relatively flat trajectory, with a 25% level of large home ownership in 1990, rather than dropping to 17% as would be expected from the 1980 cross-section. The effects of this cohort momentum are to push the age-demand curve outward and to elevate the housing consumption in every older age group being entered by new cohorts. Following the traditional interpretation, and relying only on the 1980 age profile, would have resulted in erroneous future projections.
A second measure of consumption is the average value of all single-family homes, regardless of their size. This analysis requires that substantial attention be paid to choosing the appropriate price deflator to construct a constant-dollar time series. The example in Exhibit 4 is drawn from a detailed study by Myers and Pitkin (1995) and shows the alternative that was judged best because it yielded the greatest reduction in price inflation over time. As with ownership of large homes, the data show a decline in home ownership with age, but this decline is an illusion created by the age cross-section at one point in time. Both the cohorts have moved upward in mean house value over time, and their trajectories have pushed out the 1990 age cross-section. Because the house values in these trajectories have been adjusted for inflation, any upward movement reflects a growth in housing quality or in the quantity of consumption. The data show that people upgrade their housing, either by adding amenities or remodeling, as they grow older. Again, a downward projection based on the 1980 cross-section would have been misleading.
From these data it is clear that the large baby boom generation is carrying their housing consumption forward as they grow older and will probably continue to do so as they advance into their elderly years. The lower levels of consumption suggested at older ages in the 1980 cross-section were the product of the fact that earlier cohorts advanced on much lower trajectories. For example, those aged 70-74 in 1980 would have first reached age 30-34 back in 1940, at the end of the Great Depression and just before World War II, hardly an auspicious time to begin one's housing career. Each successive generation of young adults in the 1950s and '60s fared better, because they entered the housing market with the benefit of VA and FHA programs, low interest rates, and relatively low prices. Spurred also by the postwar baby boom, these young parents were pressed to buy single-family homes in ever greater numbers. The result is that by 1980 the age cross-section reflected not simply their age but also the accumulated success that was relatively greater for each successive generation. The difference in housing demandbetween age groups is purely an illusion created by the lower demand of an earlier generation (now elderly) and the higher demand of postwar adults (now middle aged).
Conclusion
The real demographics of U.S. housing revolve around the aging of real people in cohorts, not cross-sections at a moment in time. The cohort trajectories lend themselves to projection into future decades better than do the age cross-sections that have traditionally been used. The clear implication is that the baby boom generation will not experience the loss in housing demand that had been feared. Indeed, it is likely that households will continue to invest in their housing to make it more comfortable as their needs change. Thus the large baby boom generation should continue to wield a positive impact on the housing market, rather than a negative one.
Why have most real estate economists held to the traditional view if it is clearly so inadequate? A contrarian view always deserves explanation, and the authors offer the following thoughts. First, analysts often have data for only one point in time, so the cohorts' changes are not visible to them. Even when data is available for multiple decades, economists have failed to graph it in a fashion that reveals the temporal structure. The worst abuses of the traditional method are embedded in complex econometric models whose age assumptions are obscured. Indeed, so powerful are the demographic effects, that any specification yields results that are highly significant statistically, so economists have not been pressured to find more refined models.
Perhaps the best explanation for the oversight is that before 1990 it had little significance. Both the cohort and traditional approaches suggest growing demand for housing up to about age 45. It is only after that point that the cohort trajectories continue upward and the cross-section turns downward. The leading edge of the baby boom generation first reached this critical turning point only in 1991. Now that the bulk of the baby boomers are arriving at this critical stage, analysts who wish to forecast their future consumption must choose between methods. Given the graphic contrast shown here, hopefully more analysts will appreciate why real demographics require cohort analysis.
References
Day, Jennifer Cheeseman. 1993. Population Projections of the United States, by Age, Sex, Race, and Hispanic Origin: 1993 to 2050. Series P25, No. 1104. Current Population Reports. Washington D.C.: U.S. Bureau of the Census.
Joint Center for Housing Studies. 1995. State of the Nation's Housing. Cambridge: Harvard University.
Mankiw, N. Gregory and David N. Weil. 1989. "The Baby Boom, the Baby Bust, and the Housing Market." Regional Science and Urban Economics 19: 235-258.
Myers, Dowell. 1990. "The Demographic Contribution to Real Estate Market Analysis." Pp. 45-73 in Research in Real Estate Vol. III , ed. Stephen D. Kapplin and Arthur C. Schwartz. Greenwich, Conn.: JAI Press.
Pitkin, John R., and Dowell Myers. 1994. "The Specification of Demographic Effects on Housing Demand: Avoiding the Age-Cohort Fallacy." Journal of Housing Economics 3 (3): 240-50.
Myers, Dowell and John R. Pitkin. 1995. "Evaluation of Price Indices by a Cohort Method." Journal of Housing Research 6(3): 497-518.