Housing prices are a key input for economic research and policy evaluation—but they are not always measured in the same way. A new study shows that commonly used alternatives to sales prices can lead to systematically different conclusions, depending on how and why they are used.
In “Mind the lag: Using assessed and list prices as proxies for housing market values,” Gabriel M. Ahlfeldt, Hans R. A. Koster, and Tu Giang Vu investigate whether list prices and assessed values can reliably substitute for actual transaction prices in housing market research. Their central question is simple but consequential: when do these alternative price measures accurately reflect market values, and when do they fail to do so?
Sales prices—the prices at which homes are actually sold—are widely considered the best measure of housing values. Yet such data are often unavailable, incomplete, or difficult to access, especially for small areas or long time spans. As a result, researchers frequently rely on list prices (asking prices set by sellers) or assessed values (prices determined for tax purposes). The study examines how this choice of data affects empirical results.
To answer this question, the authors use large, matched property-level datasets from the Netherlands and New York City, covering millions of homes over several decades. By observing sales prices, list prices, and assessed values for the same properties, they can directly compare how closely these measures align—across locations and over time. This rare data setup allows them to move beyond assumptions and test, in a systematic way, how different price measures behave in practice.
The analysis begins by separating housing prices into two components: a spatial component, which captures price differences across neighborhoods, and a time component, which captures how prices change over months and years. This distinction is important because many studies rely either on cross-sectional comparisons across space or on changes over time, such as before-and-after evaluations of new infrastructure or policy reforms.
The first main finding is reassuring. Across space, both list prices and assessed values track sales prices very closely. In both countries, areas with higher sales prices also tend to have higher list prices and higher assessed values. As a result, when researchers study static questions—such as how much people are willing to pay for certain neighborhood characteristics or amenities—using list prices or assessed values leads to estimates that are very similar to those obtained with sales prices. In these cross-sectional settings, the choice of price measure mainly affects precision, not the substance of the results.
The picture changes once time enters the analysis. The study shows that both list prices and assessed values adjust more slowly to market shocks than sales prices do. This sluggish adjustment is particularly pronounced for assessed values, which are often updated only once a year and may rely on past transactions or rental income rather than current market sales. In New York City, assessed values for larger residential buildings—based on income capitalization methods—lag especially far behind actual transaction prices.
List prices adjust more quickly than assessed values, but they also show delays, especially at high frequencies. At the monthly or quarterly level, list prices often fail to fully reflect recent changes in sales prices. Over longer horizons, such as several years, they eventually catch up, but short-run movements in the housing market are only imperfectly captured.
These timing differences have clear consequences for applied research. When the authors replicate well-known studies that rely on difference-in-differences designs—methods that compare price changes before and after an event—they find that estimates based on list prices are typically smaller than those based on sales prices. Estimates based on assessed values are often even more attenuated and can, in some cases, become statistically insignificant or economically implausible. This pattern appears in applications ranging from the impact of wind turbines on nearby house prices in the Netherlands to the effects of a major subway extension in New York City.
The broader message of the paper is therefore nuanced. List prices and assessed values are not “bad” data. They are, in fact, highly informative in cross-sectional analyses and can be especially useful when transaction data are sparse or unavailable. Assessed values, with their broad coverage, are well suited for mapping spatial price differences or for calibrating static quantitative spatial models.
At the same time, the study cautions against using assessed values—and, to a lesser extent, list prices—in research designs that rely on short- to medium-run price changes. When the timing of adjustment matters, sales prices remain the most reliable measure of housing market values.
To the Study
About the Authors
Gabriel M. Ahlfeldt
Professor at Humboldt University in Berlin where he holds the Chair of Econometrics; Also Visiting Professor at the London School of Economics, faculty of the Berlin School of Economics, and an affiliate of LSE-CEP, CESifo, and CEPR. As a quantitative spatial economist his primary field is urban economics, but his research cuts across many fields such as environment, finance, labour, political economy, and real estate.
Hans R. A. Koster
Professor of Urban Economics at Vrije Universiteit Amsterdam and affiliated with the Tinbergen Institute. His research concerns the economic analysis of cities, regions and the environment.
Tu Giang Vu
PhD researcher at Department of Spatial Economics at Vrije Universiteit. Her main interest is housing market and spatial data.
