Browse the real estate section of your local newspaper, and you’re likely to see the phrase “lack of inventory.” In simple terms, this means there are more people seeking to buy (or rent) homes than houses currently available for sale. If inventory remains tight for several months, prices will rise and housing affordability—already a concern in many areas—will get worse. But why? In advanced market economies like the U.S., we don’t often worry about supply shortages for basic goods and services. On the contrary, a “lack of inventory” for jeans or haircuts or washing machines would be front-page news. Policymakers across the U.S. would like to encourage more housing development, but do not have a clear metric of how much housing their communities need. Federal statistical agencies collect and publish several datasets that track housing, but there is no readily available, timely, easily interpreted metric that assesses the balance between supply and demand. Below we address several challenges to developing such a metric with existing data. Our future work will further explore how to construct and use better metrics.
Housing is both a stock and a flow. Most datasets capture only one dynamic.
The housing inventory available for sale or rent at any point in time depends on three components: the existing stock of homes, additions to the stock from new development, and removal of homes through demolition, destruction, and other channels. Additions and removals together make up the flow. The stock is much larger than the flow: there are approximately 134 million housing units in the U.S., while annual additions and removals are roughly 1 to 2 percent of the stock. To assess the capacity of housing supply to meet demand, we need to understand all three components, but our current data sources are not well integrated for this purpose.
The most comprehensive and frequently used measure of housing stock is the Decennial Census, which provides a snapshot of the total number of housing units every 10 years. The annual American Community Survey provides estimates of housing counts between census years. Both surveys are household based, so the focus is mostly on occupied housing in traditional structures (houses, apartment buildings, and mobile homes). Housing unit counts are most accurate for single-family houses, where one structure equals one unit. Many older structures have been reconfigured to contain different numbers of units than originally built. For instance, the townhouses built in many East Coast central cities in the early 20th century were originally designed as single-family homes. Many of these are now configured as multiple apartments, which is not always reflected in the address lists used by census takers. Because these structures are more concentrated in some locations—especially older neighborhoods in central cities—the accuracy of housing unit counts in the census will vary across different parts of the U.S.
The best available estimate of additions to the stock through new construction is the U.S. Census Bureau’s New Residential Construction Survey. Released monthly, quarterly, and annually, the survey estimates new housing units permitted, started, and completed. These estimates are constructed from surveying local governments (the entities responsible for issuing permits). This survey only tracks increases in housing, not the removal of units, and focuses on ground-up development of new structures. As Figure 1 shows, new units permitted are highly cyclical, but make up around 1 to 2 percent of the housing stock during non-recession years.
Alternately, the Bureau of Economic Analysis measures the value of residential construction activity for the purpose of estimating GDP. But monetary values and housing units are not easily comparable: $10 million worth of construction could represent 40 $250,000 houses or ten $1 million homes—which have very different implications for changes in housing supply.
New construction is only one part of the change in housing supply.
In estimating changes in supply, we mostly rely on counts of new construction, but there are many other ways that supply adjusts. The earlier example of single-family homes converted to multiple apartments is one common means of creating additional housing units within an existing structure. Similarly, “accessory” apartments or granny flats are created by converting basements or building an extension onto a garage. Conversely, a subdivided apartment building may be turned back into a single-family home, which decreases total housing units. Other housing units are lost through demolition, natural disasters, or simply gradual deterioration of the structure until it can no longer be used. A little-known but highly useful companion report to the U.S. Department of Housing and Urban Development’s American Housing Survey documents how many units are gained or lost to the stock through each of these mechanisms.
At a national scale, more than 70 percent of additional units arrive through new construction, and more than one quarter are from reconfiguration of existing structures. The two largest components of losses are movement of mobile homes and loss through demolition or disaster.
How important each housing flow component is to overall supply changes varies across regional and local housing markets. For instance, building permit data from Washington, D.C. show that new building accounts for a small share of residential construction activity, relative to renovation and reconfiguration of the existing housing stock. Units can be created or lost through addition, alteration, or new building permits.
Most supply metrics focus on quantity, but location and quality also matter.
The monthly data releases of total U.S. housing starts, completions and sales are widely covered by the economics-related media. We pay less attention to data below the aggregate counts, although the location, characteristics, and quality of new and existing units have important implications for whether supply will satisfy demand. Building more homes in Texas won’t satisfy growing demand in San Francisco (at the margin, some people may move to find cheaper homes, but most are tied to jobs and social networks in their current location). Similarly, if most new households are families with young children, they are unlikely to move into studio or one-bedroom apartments. Because housing varies widely in quality and underlying characteristics, local inventory shortfalls reflect more than a mismatch in quantity of production.
Better housing supply metrics will help us answer key questions.
Policymakers rely on clear performance metrics to help them achieve their desired outcomes. Education experts use value-added metrics to assess how much students learn from each school. Transportation policymakers use traffic counts and ridership numbers to guide their investments in roads and transit. Mayors and governors who want to improve affordability and encourage economic growth of their cities, regions, and states would benefit from having an accurate, timely, clearly interpretable metric of housing supply. Our current public data sources provide a wealth of detail, but not in a ready-to-use form. Better measures of housing supply would enable us to understand the scope of inventory problems and devise better policy approaches.
Key questions that need to be addressed are:
- Where are the gaps in housing supply? What geographic areas, product types, or price segments have the greatest imbalance between supply and demand?
- Are there local markets where there is underused capacity among existing houses, but where homes could be improved or reconfigured to better match current consumer preferences?
- Which markets have the greatest shortfall in new housing development, relative to demand? Is that due to market factors, regulatory constraints, or both?
- Which of the inputs to housing construction—land, materials, labor, or capital—contribute most to tight inventory?
- Which local areas have demonstrated success in increasing housing supply? How did they succeed?
In the coming months, we’ll examine these questions in more depth. Developing better housing supply metrics will equip policymakers with the information they need to improve housing affordability and choice for American families.