Spring 2019: Nadia Kotova (PhD '22) and Anthony Lee Zhang (PhD ‘19) - Search Frictions and Idiosyncratic Price Dispersion in the US Housing Market

SPIRE was proud to support Stanford PhD student in Economic Analysis and Policy Nadia Kotova (PhD ’22) analyzing the inefficiencies in the US housing market. By sponsoring Nadia SPIRE helped her and her fellow student, Anthony Lee Zhang (PhD ‘19) to purchase a dataset that contains historic information on house listings for most of the US. In particular, it has data on houses' original listing price, closing price, and their time spent on the market. This dataset allowed them to test their hypotheses at the level of individual houses rather than at the aggregated market level.

Read Nadia and Anthony's account of the project below:

Residential real estate is a very illiquid asset class: houses take many months to sell, and realtor commissions for each sale are around 6% of total house prices. In this project, we analyze in detail another metric for housing market illiquidity: price dispersion. Similar houses can sell for very different prices at similar points in time. Early work by Case and Shiller (1988) shows that the price volatility of individual houses is much higher than the volatility of city-wide average prices. However, we do not yet have a strong understanding of how idiosyncratic house price dispersion behaves in the cross-section and over time, and whether it is associated with other measures of housing market liquidity.

Measuring price dispersion is difficult because houses are all different. Therefore, measuring price dispersion requires either modelling house prices, or grouping comparable houses together. This paper develops a new strategy to measure price dispersion: we combine two commonly used classes of models for house prices, hedonic and repeat-sales models, to build a flexible model that can estimate the magnitude of price dispersion at the level of individual house sales. We find that price dispersion is quite large: home buyers are exposed to additional idiosyncratic risk of around 15.8% to 32.0% of house prices, relative to the volatility of zipcode house price indices.

Understanding idiosyncratic price dispersion is important from the perspective of household portfolio choice. Real estate is not necessarily a poor investment choice. Jordà, Schularick and Taylor (2019) argue that the risk-adjusted total rate of return on house price indices is higher than that of stock market indices in many countries. We replicate this result and find that a diversified portfolio of US residential real estate over a 10-year holding period outperforms the S&P 500 over our data sample period, 2000-2017. However, the average household in the US only owns a single house, their primary place of residence. Under our estimates, the return variance of an individual house purchase over a 10-year holding period is around 2.1-3.3 times higher than the variance of US-wide average returns. Accounting for transaction costs such as realtor fees, purchasing a single house only slightly outperforms the S&P 500 over 2000-2017 time period.

We then show a number of new stylized facts about price dispersion. Price dispersion is countercyclical: it decreases during the housing boom, increases during the bust, and decreases again during the subsequent recovery. Price dispersion is also seasonal: it systematically decreases during the summer hot season, when volume and prices are high and time-on-market is low. Third, it is negatively correlated with prices and sales volume, and is positively correlated with time-on-market, vacancy rates, and other measures of market tightness. This implies that price dispersion is affected by housing market conditions: ``sellers' markets'', where listed houses sell quickly and there are many bids, tend to have low price dispersion, whereas ``buyers' markets'' tend to have high price dispersion.

We build a simple theoretical model of house price dispersion. In the model, prices of similar houses vary for three reasons. First, houses may differ in quality, in ways that are hard for us to observe in the data. Second, some sellers are less patient than others, so they sell faster, but get lower average prices as a result. Third, some buyers are willing to pay higher prices than other buyers, perhaps because they specifically desire certain features of a house that are hard to find in other houses. Our model allows us to quantitatively determine how much each of these components contributes to house price dispersion. We find that 80.2% of price variance is caused by unobserved house features, 4.5% by differences in seller patience, and 15.3% by differences in buyer values.

Together, our results shed new light on the magnitude of price dispersion and market frictions in housing markets, how they depend on market conditions, and what their fundamental drivers are. These results, thus, help us understand the microstructure of housing markets and speak to the classic question of risk and return for real estate investments.