which has happened simply because they have so much more money. When the very rich build bigger mansions, they shift the frame of reference that shapes demands for those with slightly smaller incomes, who travel in overlapping social circles. The near-rich respond by building bigger houses as well, which shifts the frame of reference for others just below them, and so on, all the way down the income ladder.
Figure A. Median wages and median house prices (both in constant 2000 dollars), 1950–2010. Sources: Economic Policy Institute, www.epi.org/publication/ib330-productivity-vs-compensation (pre-1975 median wage growth assumed equal to mean pre-1975 wage growth); U.S. Census Data, www.census.gov/const/uspriceann.pdf.
This cascade is the most parsimonious explanation for the striking fact that the median new single-family house in the United States, which stood at 1,570 square feet in 1970, had grown to more than 2,300 square feet by 2007. That growth cannot be explained by growth in the median wage or median family income, which changed by much smaller amounts during those years.
What changed dramatically was the context in which the median family’s housing choice was made. Any family that failed to rent or purchase a house near the median of its local price distribution would have had to send its children to below-average schools. So a family that was determined not to see its children fall behind had little choice but to keep pace with what others were spending on housing.
I’ll describe a simple measure of the cost of keeping up, one that can be calculated easily with existing data. It rests on the positive link between the average price of a house and the quality of its neighborhood school. This link implies that the median family must outbid 50 percent of all parents to avoid sending its children to a school of below-average quality.
Figure B. Monthly hours of work required for the median earner to rent the median house, 1950–2010. Source: Calculated from data in figure A.
Figure A shows the time profiles of median U.S. house prices and median hourly earnings for American workers in the census years from 1950 to 2010. As discussed, the distribution of income was exceptionally stable in the years up to roughly 1970. Median hourly earnings were rising at a relatively rapid clip, slightly exceeding the rise in median house prices, and incomes elsewhere in the distribution were rising at approximately the same rate. In contrast, most income growth after 1970 accrued to top earners, while median hourly wages increased only slightly. And yet median house prices grew much more rapidly during the latter period.
The upshot is that by 2010, the median earner had to work substantially more hours each month than in 1950 to gain access to a house at the midpoint of the housing price distribution (see figure B). For illustrative purposes, I assume that the implicit monthly cost of a given house is 1 percent of its purchase price. During the immediate postwar decades, when the income distribution was stable, the median burden of homeownership varied little and was actually slightly lower in 1970 (41.5 monthly hours of work) than in 1950 (42.5 hours). But the burden began rising sharply in 1970, and by 2010, the median worker had to work 82.9 hours a month—almost twice as many as in 1970—to put his family into a house of median price.
Housing is of course not the only expenditure that is sensitive to context. Increasing concentration of income at the top has also spawned similar expenditure cascades for items such as clothing, gifts, birthday parties, and other celebrations to mark special occasions.9 In these domains as well, the median earner must spend more than before or else experience significant adverse consequences of one kind or another.
Of course, not all such spending has been purely wasteful. Although the utility conferred by a diamond ring may depend largely on its relative size and quality, for example, even the lone resident of a desert island might take additional pleasure in the way an absolutely larger stone refracts the light. Yet surely much of the extra spending of recent years has been a relatively inefficient source of extra utility. The average American wedding now costs almost thirty thousand dollars, nearly twice as much as in 1990.10 Does anyone believe that the extra spending has made couples and their families any happier?
Although additional outlays for many consumption goods—such as houses beyond a certain size—don’t accomplish much, they crowd out other forms of spending that would produce real improvements in the quality of life. If houses grew less rapidly, for example, we could invest in mass transit systems that would yield shorter, less stressful commutes and thereby free up more time to spend with friends and family. Or we could support medical research and safety investments that would reduce premature death. The list goes on.
Figure C. Per-capita GDP, 1960–2010. Source: http://research.stlouisfed.org.
Wasteful “positional arms races” occur because people take too little account of the costs that certain types of consumption impose on others.11 When one job applicant spends more on an interview suit, for example, others must spend more as well or accept lower odds of getting a callback. Yet, as noted, when all spend more, no one’s odds of landing the job are any higher than before.
Existing policy instruments can easily curtail such waste. In a world of perfect information, the ideal remedy would be to tax goods in proportion to the extent to which their use generates negative side effects.12 In practice, we lack the detailed information necessary to implement this remedy. But in chapter 11,I describe a much simpler instrument that would serve almost as well.
For the past several hundred years, rising per-capita GDP has been interpreted to imply that economic well-being has steadily increased. During the era that figure C depicts, for example, GDP per capita has risen steadily and rapidly, leading many to conclude that there have been significant improvements in economic welfare. This measure of well-being, however, is completely insensitive to the costs imposed by the expenditure cascades just described. From the perspective of the median worker, the contrast between the impressions conveyed by figures B and C could hardly be more striking.
Economic models of human behavior affect how we think about the relationship between income and well-being, which in turn affects the mix of policies we adopt. The current emphasis on maximizing per-capita GDP completely ignores the central role of context in consumption decisions. Alternative welfare measures like the one I have proposed would help focus attention on the economic forces that bear most heavily on well-being. And by so doing, they would strengthen support for policies that would make everyone better off.
NOTES
1. Sarah Anderson, John Cavanagh, Scott Klinger, and Liz Stanton, “Executive Excess 2005,” report for United for a Fair Economy based on annual CEO pay studies conducted by Business Week (1990–2004) and the Wall Street Journal (2005), published August 30, 2005, www.faireconomy.org/files/Executive_Excess_2005.pdf.
2. Thomas Piketty and Emmanuel Saez, “Income Inequality in the United States, 1913–1998,” Quarterly Journal of Economics 118, no. 1 (February 2003): 1–39. Updated to 2010 at http://elsa.berkeley.edu/∼saez/saez-UStopincomes-2010.pdf.
3. Edward Wolff, The Asset Price Meltdown and the Wealth of the Middle Class (New York: New York University, 2012).
4.
