Housing Markets Under the Microscope: City by City, Year by Year

When measuring U.S. economic activity, government agencies gather an array of data that describe our diverse economy. As part of their work, they produce readings of general movements in price levels, or inflation. And within these readings, which reflect the costs consumers pay for everyday necessities, lurks one of the biggest expenses of all: housing. In current inflation indexes, housing is reflected as a broad, national average. But is it possible to create housing indexes that precisely capture city-level prices? In their paper, "The Price of Housing in the United States, 1890-2006," four researchers - Ronan C. Lyons, Allison Shertzer, Rowena Gray, and David Agorastos¹ - present a novel pair of indexes and show how they change our understanding of long-run price behavior in major cities across the country.

To create these indexes - one to measure rents and one to measure sales prices - the paper's authors crafted a unique data set to reflect housing markets across 30 large cities for a span of nearly 120 years. (The data set and the indexes together are referred to as the Historical Housing Prices Project, or HHP.) Whereas traditional indicators of housing markets tend to rely on survey responses from homeowners, tenants, and landlords, these authors turned to real estate listings in local newspapers to more accurately assess each city's unique housing market. Through a manually intensive process, they gathered details about each property's basic attributes, such as size, location, price, and type of dwelling. All told, they collected information from nearly 2.7 million newspaper listings, an effort designed "to obtain a diverse sample in terms of geography and economic trajectory over the 20th century." The resulting indexes, along with interactive tools, are hosted online by the Federal Reserve Bank of Philadelphia; taken together, they provide a dynamic resource for researchers investigating long-run housing prices.

Over time, government statisticians have adjusted their inflation calculations, seeking to better understand the true cost of housing (by accounting for changes in housing quality, for instance). In the spirit of contributing to these improvements, the authors used their HHP data to create more-precise renderings of historical housing prices, noting that government rental index data "[have] never been comprehensively revised."

To understand why the HHP indexes are more accurate, consider what happens in rental markets when departing tenants leave behind vacant apartments. When new tenants eventually move in, rents for these apartments typically "reset" to market levels. But traditional indexes capture this change only if the apartment happens to be included in an inflation survey after the new tenants arrive (which is not guaranteed), and only if the new tenants respond to the survey (likewise far from certain). By basing rents on actual newspaper listings, which accurately reflect rents in the open market, the authors compensate for this shortcoming and arrive at more complete measurements of market real estate prices.

After shoring up real estate data in this way (along with other enhancements), the authors compared their HHP rent index with official government gauges of rent growth. Among other findings, the comparison shows that over the long run, official data have likely understated the magnitude and persistence of steadily changing rents. Between 1914 and 2006, for instance, the authors' rent index for the United States rose by 3.6 percent annually, whereas the official government barometer rose by 2.6 percent - a full percentage point lower per year, implying a twofold difference in price levels in 2006. Meanwhile, the average annual inflation rate for the period excluding housing was 3.3 percent. Given the gap between official rent growth and HHP growth, the authors find that their data "would put housing inflation slightly above inflation in the wider economy" rather than well below it, as official data suggest. This discovery brings to light an important corollary: The long-run increase in the standard of living has probably been overstated.

In a separate part of their analysis, the authors map the trajectories of housing prices in relation to the ebb and flow of economic growth (an oscillation sometimes referred to as the business cycle). They trace the housing market's interaction with economic fluctuations over time, finding a strong link between the two. "More often than not," they report, "real growth rates in [the economy] and housing prices move in the same direction..." There are exceptions, of course - periods during which housing markets do not move in tandem with the economy. After World War I and again after World War II, for instance, housing prices remained relatively stable despite the onset of economic contractions.

Housing market trajectories become even more intriguing when comparing long-term trends in rental rates and sales prices. Market rents are relatively stable; sales prices are substantially less so. The two exhibited a "permanent divergence...in the 1970s as the price of owned housing began thirty years of volatile growth." This striking difference is accentuated by the remarkably steady behavior of rents, which were within 10 percent of their 1890 values almost two-thirds of the time being studied.²

By offering a new set of tools for exploring housing and its role in the U.S. economy, the authors of "The Price of Housing in the United States, 1890-2006" show that it is possible to account for intercity variation over the long run. They acknowledge the foundational work done by researchers who came before them, but they also look ahead, writing that they hope their work "will be of great use to future researchers and spur new inquiry into the evolution of housing markets in the United States."