Baby boomers and the housing market on the cusp of COVID-19
Marijn Bolhuis, Judd N. L. Cramer 02 April 2020
The disproportionately large baby boomer cohort (born in the US from 1946 to 1964) has consistently been central to culture, politics, and the economy. The significance of their combined demand and purchasing power in terms of the housing market has been a frequent topic of conversation since at least the early 1990s.
Assuming that it is first-time homebuyers that mainly drive the housing market, and most baby boomers had purchased their first homes by the late 1980s, Mankiw and Weil (1989) famously predicted that housing demand and prices would peak in the 1990s. ‘Generation X’ (the generation following the baby boomer cohort) was much smaller, to the extent that their first-home purchases in the 1990s would entail a relative decline in housing demand growth. Mankiw and Weil’s (1989) predictions became somewhat infamous as the housing market boomed from the late 1990s until 2006.1
In a recent paper (Bolhuis and Cramer 2020), however, we revisit Mankiw and Weil’s hypothesis about the effects of ‘the baby boomer lifecycle’ on the US housing market. We suggest that Mankiw and Weil actually had the correct intuition but spoke too soon. Rather than their first-home purchases, it is the baby boomers’ retirements that may slow the housing market.
As baby boomers enter retirement, they are increasingly downsizing, a trend that affects different segments of the US housing market. We construct a novel dataset by combining neighbourhood-level housing market data with census information on demographics and housing structures. Our empirical analysis reveals that, since the housing trough in 2012, larger homes (and those in neighbourhoods with relatively more baby boomers) underperform substantially in terms of price growth, home sales, and liquidity. In the next two decades, as more and more baby boomers look to downsize, more than a quarter of the US homes occupied by their owner(s) will likely hit the market (Romem 2019). If the trends we document continue, the underperformance of these homes on the housing market may put increased pressure on baby boomers’ savings, at the exact time that they move into retirement.
Demographics and the lifecycle of housing demand
Housing demand tends to follow a predictable pattern over the lifecycle, as we summarise in Figure 1. We use data from the 2017 American Community Survey to plot (for each age group) the share of recently purchased homes that have less than three or more than three bedrooms. Generally, home buyers start out in smaller homes. During their child-rearing years, the demand for larger homes increases. Lifecycle demand peaks around age 40, when close to 50% of buyers move into homes with four or more bedrooms. The mean value of recently purchased homes is $400,000, as shown on the righthand vertical axis in Figure 1. As buyers move closer to retirement, they tend to downsize, moving from the homes where they raised their families into considerably smaller homes, often with only one or two bedrooms.
Figure 1 Lifecycle demand for different housing types
Notes: Data from the 2017 American Community Survey. Recent buyers are those that bought a home in last two years. Age refers to the age of the self-reported household head.
When applied to the baby boomer cohort, these lifecycle dynamics explain the relative increase in the demand for smaller homes in recent years. Between 2000 and 2020, the share of Americans over the age of 60 increased from 23% of the population in 2000, to 31% in 2020. This is expected to reach 34% in 2030. Retiring baby boomers tend to ‘sell large’ and ‘buy small’, which means that their preferences increasingly overlap with those of the ‘millennial’ generation. Currently the largest cohort in American society, millennials often seek out urban amenities and smaller living spaces (Couture and Handbury 2017), features for which they will increasingly have to compete with baby boomers on the housing market.
We compute the resulting demand dynamics using a shift-share analysis, presented in Figure 2. Holding the buying shares of each age group constant, the relative demand for homes with more than four bedrooms fell by 7% between 2000 and 2020, while the reverse is true for smaller homes.
Figure 2 Implied demand for different housing types
Notes: Population data from the NBER Census U.S. Intercensal County Population Data. Initial choice shares of housing types from the 2000 American Community Survey. Projections from CDC State Population Projections.
Older Americans’ homes lagging the market
Lifecycle-related changes in relative demand for different home types has already resulted in large adjustments in the US housing market, both within and across neighbourhoods. We combine neighbourhood-level housing market data from Zillow with detailed census information on demographics and housing structures. Figure 3 shows that since 2012 the price growth of houses with more than four bedrooms has lagged the price increases of one- and two-bedroom homes by about one percentage point per year (within the same zip code).2
Figure 3 Relative annual price growth of home types, within zip codes, 2012-19
Notes: Data from Zillow. Price index refers to the zip code-specific Zillow Home Price Index (HPI). We compute relative price growth by subtracting the HPI from the 2012-19 annual price growth of different home types in a neighbourhood. The figure plots the mean of this statistic across zip codes.
To measure local exposure to demographic change, we construct a demographic demand shock by using a neighbourhood’s composition of home types as shares, and county-level changes in the age composition as shifts. This allows us to construct within-county changes in demand for housing that are driven by the initial housing distribution in specific zip codes.3 We find that the home prices in baby boomer-rich zip codes have declined substantially, relative to those in millennial-rich areas. This means that, within local labour markets, neighbourhoods that experienced a positive relative demand shock (more millennials) have experienced substantially faster home price growth over the latest cycle. This is true both across the country and within counties, and these trends are driven by areas with lower housing supply elasticities.4 For example, within Los Angeles County (the most populous county in the US), zip codes where 15% of the inhabitants are over the age of 60 have experienced about three percentage points higher annual home price growth than neighbourhoods where 30% of the population is older than 60 (Figure 4).
Figure 4 Annual price growth of zip codes in Los Angeles County, 2012-19
Notes: Binned scatterplot of zip code-level 2012-19 annual price growth against its share of population that is over 60. Each bin represents about 5 zip codes. Home price data refer to Zillow’s Home Price Index (HPI). Population data from U.S. Census Data Tables, averaged over 2013-17 to preserve confidentiality.
We find that the homes of baby boomers have not only experienced slower price growth but have also become significantly harder to sell. Within the same county, a 1% lower demographic demand shock in a neighbourhood leads to an 11% relative decline in home sales. Even though more and more houses in neighbourhoods with relatively greater numbers of older Americans are put up for sale, with the number of homes on the market increasing by eight percentage points more in these neighbourhoods, buyers are not following. The expected time on the market, the percentage of homes sold with a cut, and the median price cut on sold homes all increase substantially in these neighbourhoods.
Not only are large housing market corrections already underway, the combination of lifecycle effects and demographic developments means that these adjustments are likely to continue over the next decade. If the trends we document persist, they will have significant macroeconomic effects. The asset position of the middle class tends to move closely with home prices (Kuhn et al. forthcoming), and older Americans are central to these trends. Currently, Americans over the age of 55 hold almost 60% of total housing wealth, with 60-75% of their assets held in real estate (Davis and Van Nieuwerburgh 2015). Asset-wise, the middle class is almost completely dependent on housing, with retirement portfolios a distant second. Since many Americans in the middle class will retire within the next few years, they stand to exert demographic pressure on their neighbourhoods. This pressure might significantly affect the $16.5 trillion in real estate wealth held in these households, precisely when they will need those assets most.
Covid-19 presents something of a wildcard in this context. The pandemic can reverse or accelerate these adjustments, and it is too early to tell what the wider effects of the virus may be. Confined to their urban apartments, millennials may come to prefer the distancing of the suburbs again, reversing current trends. In the tragic case that the lives of older Americans are lost at higher rates than their younger compatriots, the trends we document might accelerate. As we begin to think through the economic fallout of the pandemic, the effects of demographic distribution on the US housing market will have to be taken into account. Clearly, the stakes are high.
Autor, D H, D Dorn and G H Hanson (2013), “The China Syndrome: Local Labor Market Effects of Import Competition in the United States”, The American Economic Review 103(6): 2121-2168.
Bartik, T J (1991), Who benefits from state and local economic development policies? Kalamazoo, MI: W E Upjohn Institute for Employment Research.
Bolhuis, M A and J N L Cramer (2020), “The Millennial Boom, the Baby Bust, and the Housing Market”, arXiv preprint arXiv: 3102262.
Couture, V and J Handbury (2017), “Urban revival in America, 2000 to 2010”, NBER Working Paper w24084.
Davis, M A, and S Van Nieuwerburgh (2015), “Housing, finance, and the macroeconomy”, Handbook of Regional and Urban Economics 5: 753-811.
Hamilton, B W (1991), “The baby boom, the baby bust, and the housing market. A second look”, Regional Science and Urban Economics 21(4): 547-552.
Kuhn, M, M Schularick and U I Steins (forthcoming), “Income and wealth inequality in America”, Journal of Political Economy, forthcoming.
Mankiw, N G and D N Weil (1989), “The baby boom, the baby bust, and the housing market”, Regional Science and Urban Economics 19(2): 235-258.
Mankiw, N G and D N Weil(1992), “The Baby Boom, the Baby Bust, and the Housing Market”, Regional Science and Urban Economics 21(4), 541-52.
Romem, I (2019), The Silver Tsunami: Which Areas will be Flooded with Homes once Boomers Start Leaving Them? Available at: www.zillow.com
Saiz, A (2010), “The geographic determinants of housing supply”, The Quarterly Journal of Economics 125(3): 1253-1296.
1 See Mankiw & Weil (1989) for the original prediction that changing demographics would lead to a fall in housing prices in the 1990s. See Hamilton (1991) for a critical evaluation of their empirical methods. In a later reply to their original paper, Mankiw & Weil (1992) concluded that demographics were still a central driving force behind the housing market, and that housing demand would remain depressed during the 1990s. In hindsight, the growth of home sales and prices did falter in 1989 and remained low during the early 1990s. Growth picked up from 1996 onwards, continuously increasing into the 2006 housing bust that led to the Great Recession.
2 The U.S. is divided into almost 42 thousand zip codes, with about eight thousand citizens per zip code. Even metropolitan areas contain many zip codes. Los Angeles County, for example, is divided into 505 zip codes.
3 This Bartik (1991) shift-share shock is analogous to a labour demand shift-share instrument (e.g. Autor et al. 2013) that uses within-region (e.g. local labour markets) industry composition as shares and regional (e.g. state) industry growth rates as shifts.
4 We exploit variation across U.S. metropolitan areas in the housing supply elasticity (Saiz 2008). Almost all predictive power of neighbourhoods’ share of baby boomers is accounted for by the 50% of MSAs with the lowest supply elasticities. These results support our conjecture that the stark differences in price growth across neighbourhoods and home types are driven by relative demand shocks to homes with fewer bedrooms.