Online Real Estate Agencies and Their Impact on the Housing Market

Talk by Oleksandr Talavera (University of Birmingham) as part of the Research Seminar Series of the IOS Economics Department.

Online platforms have transformed many markets, as evidenced by the rise of firms such as Amazon, Uber, and Airbnb. However, the recent emergence of online real estate agencies has not yet received much attention. We use regression, matching, and machine learning (meta learner) methods to investigate the impact of online agencies on the housing market, with a particular focus on time on market and sale margin (the difference between asking and paid prices). Our dataset consists of 1,274,214 properties in the UK, for which we have matched Zoopla listings with actual transactions from the Land Registry. We find that time on market and the sale margin are shorter for properties listed with online agents. However, the size of this reduction is much smaller when meta learners are used. We attribute this finding to meta learners being better able to uncover nonlinear relationships between variables. Our overall finding is that time on market is, on average, four days shorter and the sale margin is 0.35% less for properties listed with online agents. The shorter time on market of online agents, combined with an average fee less than one third of that charged by traditional agents, explains why online agents are rapidly gaining market share. The market share of online agents has risen, particularly for properties in the mid-price range and in regions with younger demographics. Also, we find that the rise of online agents has caused traditional agents to change their behavior – both time on market and sales margin are lower for traditional agents in regions with a higher share of online agents.