麻豆视传媒免费观看

Over the past two decades, housing costs have outpaced income growth in the United States, increasing the rent burden and heightening barriers to homeownership (). Policy experts and academics widely agree that these trends reflect a long-run housing supply shortage, which is a key driver of housing unaffordability (, ; ). However, there is less consensus on the scale of the shortage. Recent estimates range from 1.5 to 5.5 million units, with variation driven by a combination of methodological differences in calculating the shortage and different characterizations about what constitutes equilibrium in the housing market. In this brief, we examine the assumptions underlying existing shortage estimates and offer an updated estimate based on our preferred methodology. Our calculations show that the U.S. housing market was short 4.9 million housing units in 2023 relative to mid-2000s.聽

Fixing ideas: Households, housing supply, and housing shortages

A household, as defined by the U.S. Census Bureau, refers to all people who jointly occupy a single housing unit. In 2023, there were roughly 131 million households. Households can be comprised of a single person who lives alone, a family living together, or a group of unrelated individuals occupying a shared space. This definition includes many types of living arrangements, including rented apartments, owned homes, and shared residences. The U.S. Census counts households using a combination of the Decennial Census, the Current Population Survey (CPS), and the American Community Survey (ACS) (). Understanding what constitutes a household and how households are accounted for is essential in the context of measuring housing shortages. Figure 1 depicts the total number of households in the U.S. since 1960 together with the total population. Trends in the number of observed households can be driven by population growth, changing social dynamics, and the price of housing.

Typically, a shortage refers to a market situation where the quantity supplied is less than the quantity demanded at the current market price. Neoclassical economic models predict this outcome; for example, in the case of a minimum wage in the labor market, the demand for labor at the minimum wage is predicted to be less than the quantity of labor supplied. However, when discussing housing, a shortage is more complex than a simple mismatch between supply and demand at prevailing prices. In the housing market, a 鈥渟hortage鈥� often refers to the inability of supply to meet demand at affordable prices. This, in turn, can distort the number of observed households, as individuals or families may combine into shared housing arrangements, not because of a lack of units but because they cannot afford to live separately. In other words, the prevailing market price of housing influences both the availability of housing units and the observed number of households, creating a disparity between the number of actual households and the number of desired households.

Estimates of the housing shortage generally refer to any gap between the existing stock of housing and the number of housing units needed to accommodate the desired number of households. The U.S. Census Bureau measures the housing stock, either currently habitable or new construction, through the Housing Vacancy Survey and the Survey of Construction. However, measuring the number of housing units needed to satisfy demand, or the 鈥渢arget housing stock,鈥� is more complex and often involves different estimation methods. In some cases, researchers project future housing needs based on historical trends in construction or household formation, and in other cases, researchers directly estimate housing demand based on factors such as population growth or housing affordability. Figure 2, below, illustrates the range of recent housing shortage estimates that emerge from different approaches to determining the target stock.

Data sources for Figure 2: , , , , ,

Understanding the housing vacancy rate is an important precursor to analyzing different approaches to calculating the supply shortage. Just as the labor market has a natural rate of unemployment to allow for efficient job matching, a healthy housing market requires a certain vacancy rate to align buyer and seller preferences. Therefore, supply shortage estimates must account for some vacancies, making the assumed vacancy rate a critical component in determining the gap between housing stock and demand. Vacancies can be categorized into three main types: units available for sale/rent, those held off the market, and seasonal vacancies. Seasonal vacancies typically reflect the unique needs and patterns of specific regions or market segments, like a vacation or snowbird property, short-term rentals, or property occupied by seasonal workers. Units held off the market may be undergoing renovations, tied up in legal issues, or kept vacant for speculative or personal reasons. Figure 3 illustrates trends in the vacancy rate over the last sixty years, broken down by type. Notably, the overall vacancy rate has been declining since the financial crisis and has remained just below 10% for the past three and a half years.

The Great Recession marked a turning point in the U.S. housing market. Just before this economic crash, housing production was at a 30-year high and home ownership had risen by nearly 8% during the previous decade (, ). However, the subsequent economic turmoil caused a sharp decline in new housing construction, with production levels dropping significantly and never fully recovering to pre-recession levels (). As a result, the early- to mid-2000s are often referenced as the last period when supply was better aligned with demographic needs and trends (). For these and other technical reasons related to data availability from the ACS, we will utilize 2006 as our base year for estimating the housing shortage.

Housing shortage estimates

Back-of-the-envelope

The simplest way to measure the housing supply shortage is to compare the current housing stock to the total number of households, accounting for a natural vacancy rate. We call this a 鈥渂ack of the envelope鈥� housing shortage estimate. At the end of 2023, the total housing stock from the Housing Vacancy Survey was 145.9 million units, and the 2023 American Community Survey recorded 131.3 million households. Using a natural vacancy rate of 12%¹, the back-of-the-envelope estimated housing shortage was about 1 million housing units in 2023, as shown in Figure 4. While 麻豆视传媒免费观看, this estimate likely represents the lower bound of the housing shortage, as the number of households itself can be constrained by the housing market. For instance, high housing costs may lead some individuals to delay forming their own households.

Deviations from historical trends

provides an alternative way to estimate the housing shortage by comparing historical construction rates to current trends. From 1968 to 2000, about 1.5 million new housing units were built each year, but between 2001 and 2020, this rate fell by 18% to about 1.23 million per year. Based on this slower pace, NAR estimates a housing shortage of 5.5 million housing units by the end of 2020. Additionally, the report considers the loss of older homes and the need to meet growth in household formation, which could push the total shortage closer to 6.8 million units. This method focuses on under-building but doesn’t fully account for changes in housing demand, like fewer people forming households due to high prices.

Other methods use trends in household formation to estimate the target housing stock. For example, the number of households increases when a college student moves out of their parents鈥� home to live with friends or someone decides to live alone for the first time. On the other side of the ledger, when aging parents decide to move in with their adult children or when two individuals move in together the number of households decreases. Household formation declined sharply after the Great Recession, a trend that worsened with the onset of the pandemic (). Equilibrium in the housing market might describe the point at which the stock of housing plus new construction equals the demand for housing plus net household formations. uses this idea to arrive at a shortage estimate of 2.5 million by calculating the cumulative gap between household formation growth and construction starts between 2012 and 2023. However, this estimate fails to account for any housing shortage that may have accumulated prior to 2012. In a similar fashion to Jones (2024), find an annual construction shortfall of 100,000 by comparing trends in housing construction and household formation and accounting for broader market dynamics such as vacancy rates and economic conditions.

Accounting for pent-up demand

While previously discussed methods center on household formation rates, a more nuanced approach considers how many individuals might form their own households under different market conditions. This unmet demand, often referred to as 鈥減ent-up鈥� demand, includes those that are living in shared housing due to high costs. For example, estimates a housing supply shortfall of 4.5 million units by focusing on individuals or families living with non-relatives. However, this takes a narrow view of who might otherwise prefer to form their own household. Beyond non-relatives, young adults living with their parents, older parents living with their children, or families in shared households are relevant to pent-up demand but are not the central focus of the Zillow estimate.

Alternatively, estimates pent-up demand by comparing the share of people who self-report being the head of a household (鈥渉eadship rates鈥�) by age in 2000 to those in 2021. The difference between the headship rate of each age cohort across years is aggregated to calculate the housing shortage, which Up For Growth estimates to have been 3.9 million units in 2023. However, this estimate does not account for factors that may have contributed to the change in headship rates during this time-period鈥攂e it shifting preferences for living arrangements or housing market conditions affecting household formation across different age groups.

Finally, uses a statistical method known as a Oaxaca-Blinder decomposition² to analyze how much of the observed decline in household formation is due to high housing costs. The Oaxaca- Blinder decomposition is a statistical technique that decomposes the difference in mean outcomes across two groups into shares attributable to difference in magnitude of each group鈥檚 characteristics and the difference in the effect of those characteristics. In this case, Freddie Mac estimates how the decline in household formation between 2001 and 2020 relates to differences in housing costs during these two periods鈥攁s opposed, for example, to differences in age distribution across the population or changes in income鈥攁nd characterizes this decline as households that are missing due to pent-up demand. This method stands out as the only one that directly accounts for the relationship between the price of household and household formation. They combine this estimate with the back-of-the-envelope shortage calculation that assumes a 13% vacancy rate to conclude that the U.S. housing market was short 3.8 million units in 2020.

An updated estimate

We adapt Freddie Mac鈥檚 approach, detailed in the technical appendix, and use the latest survey data to provide an updated estimate of the housing shortage in 2023. This approach is more robust than simply projecting trends in construction and household formation because it captures specific factors that drive household formation. By using the Oaxaca-Blinder decomposition, we can associate part of the decline in household formation to housing costs to provide a more nuanced projection of the number of households that would have formed under more favorable conditions. This approach provides a clearer connection between the shortage and underlying driving factors.

Our Oaxaca-Blinder decomposition compares average household size in 2006³ and 2023, using the 2006 and 2023 American Community Survey. We attribute changes in the average household size to housing costs, age, marital status, number of children, labor force participation, income, education, race, citizenship, geographic residence, and the multigenerational nature of the household. The decomposition technique allows us to estimate how the change in average household size (group mean) between 2006 and 2023 can be attributed to the change in housing costs between the two time-periods, holding other factors fixed. We use this parameter to estimate what the average household size would have been in 2023 if housing costs remained at 2006 levels. We adjust this estimate for a natural equilibrium vacancy rate of 12%, reflecting the vacancy rate that was present in the market in our base year. Finally, we combine this estimate with the back-of-the-envelope estimate of the housing shortage in 2023Q4 to arrive at a 2023 housing shortage estimate of 4.9 million units, as illustrated in Figure 5. In the technical appendix, we show that this estimate varies between 3.4 million and 6.4 million units, depending upon the choice of model and vacancy rate.

There is more work to be done to measure the magnitude of the shortfall at a local level. For example, places like New York City may face more severe shortages compared to less dense cities, suburban areas, or rural areas. Adapting these methodologies to estimate the size of the housing shortage at the local level should be a top priority⁴. Moreover, the housing shortage is likely to be particularly acute in the affordable housing sector. The National Low Income Housing Coalition estimates that in 2022, only 7.1 million rental units were available for the 11 million extremely low-income households nationwide, reflecting a significant supply-demand imbalance (). The shortage of affordable housing is thought to contribute to the fact that over 650,000 people were experiencing homelessness in January 2023 (). Adapting these methodologies to estimate the size of the housing shortage for subpopulations should also be a top priority.⁵

Conclusion

The U.S. housing supply has struggled to meet demand in the last several decades, driving up costs and deepening affordability challenges. Both state and local governments, as well as the federal government, are actively pursuing policies to help boost housing supply as a key strategy to address housing affordability (). However, effective policy solutions require an accurate understanding of the scope of the housing shortage. As highlighted in this brief, the magnitude of the shortage is not straight-forward to measure. Estimates suggest that the shortage has ranged between 1.5 million and 5.5 million during recent years. Using the latest data available by the 2023 American Community Survey and the Housing Vacancy Survey and following the method employed by Freddie Mac, we estimate the U.S. housing market was short 4.9 million units at the end of 2023.

Technical Appendix

Our methodology for calculating the 鈥減ent-up鈥� housing demand closely follows the analysis of the U.S. housing market done by . This technical appendix provides details of how we calculate the missing households.

To begin, our analysis is based on household size as measured by the American Community Survey. We choose this data source because it includes detailed demographic characteristics about households and information about housing costs for both homeowners and renters. Alternatively, one could utilize the Current Population Survey, which also identifies homeowners and renters. However, the CPS provides more limited information about housing costs, and the sample size of the CPS is an order of magnitude smaller than the ACS (roughly 100,000 compared with roughly 2 million).

We estimate a threefold Oaxaca-Blinder to decompose the change in household size between 2006 and 2023. This is a statistical method that allows us to attribute the change in household size between these years to various factors such as housing costs, age, marital status, number of children, labor force participation, income, education, race, citizenship, geographic residence, and multigenerational nature of the household. The inclusion of these factors is guided by our intuition of the variables that impact household formation. The decomposition exercise yields a parameter that can be interpreted as the difference in average household size between 2006 and 2023 that can be attributed to the change in housing costs between the two years. We use this parameter to calculate the counterfactual household size in 2023, and we use the 2023 population data to arrive at the target number of households as follows:

Where HHSize2006 聽is the average household size in 2006 calculated from ACS data. 聽is the parameter yielded by the decomposition exercise as the change in household size between the two time periods attributable to housing costs. 聽is the population of potential heads in the U.S. population and 聽is our assumed equilibrium vacancy rate of 12%. We include all adults (age 18 and above) and those aged 16-18 who are not in school in our calculation of potential heads of households.

We test the robustness of our estimate to the inclusion of different demographic and household characteristics as well as different vacancy parameters. Table A1 summarizes the estimated number of missing households and housing shortage under different specifications of the Oaxaca-Blinder decomposition as well as two different vacancy parameters (12% and 13%). Our preferred specification is model 5 with an equilibrium vacancy parameter of 12%. Different estimates of missing households are within a narrow range of 2.0 to 3.5 million households across all models resulting in the housing shortage estimates to range from 3.4 to 5.1 million households. If we assume a higher vacancy rate of 13%, the housing shortage estimates range from 4.7 to 6.4 million households.