Market Microstructure of Luxury Properties: A Data-Driven Approach

Abstract

I undertake a data-driven approach to identify the sense of luxury vis-a-vis residential properties and also investigate the market microstructure from the dynamics of Bid-Ask spread and Years on Market for these properties. I identify Luxury properties using 1,174proprietary features of owners, demography, finances, property-level hedonic features, etc., for 35,000 properties (between 1 Mil and 50 Mil listing/sold price) from First American Public MLS database and find fascinating insights into what defines luxury across these geographies consistently. I am able to predict Bid-Ask Spread with high (84.01%)accuracy which is very valuable, given the tremendous opportunities and lack of comparabilities (i.e., Market Microstructure Invariance is invalid) in this niche market. I find evidence that the hedonic features of Luxury properties are the most significant in shaping the bid-ask spread and not the buyer-seller attributes. The Years on Market driven by seller attributes, is impossible to predict but I can provide upper bounds on the time from listing to closing.