Empirical studies of auctions of non-distressed residential real estate

Abstract: Real estate auctions have become a popular sales mechanism in Sweden in the past couple of decades when the internet facilitated the marketing of properties for sale. With the possibility to display several exterior and interior pictures of the property rather than the single façade picture style of traditional newspapers advertisements, and a more detailed description of the object, brokers significantly increased the number of potential buyers at the showings of properties for sale. The increase in the number of visitors has, especially in the larger cities in Sweden, increased the frequency of bidding wars, in which the winning bid ends far above the list price. Many auction participants, especially those who are inexperienced, perceive the auction-based sales mechanism as very stressful considering that the purchase of a home is for most buyers the largest investment they will ever make. Therefore, the auction mechanism for selling homes is regularly debated in media, among politicians and by regulatory bodies. The scientific literature analyzing auctions of non-distressed homes is, however, very limited. Hence, the debate and analysis of the Swedish auction-based sales mechanism has little empirical evidence to lean on. A main reason for the lack of scientific work in this area is the difficulty to obtain sufficiently detailed auction data to study how sellers’ list price strategy and auction participants bidding strategies affect the outcome of real estate auction auctions and the occurrence of a winner’s curse.The overall aim of this thesis is to study how sellers’ list price strategy and bidding strategies of auction participants affect the outcome of auctions of non-distressed residential homes. The thesis is empirical and comprises four papers that analyze the housing markets in Stockholm and Gothenburg. Papers 1 and 2 analyze sales of condominium apartments in Central Stockholm, while Paper 3 analyzes sales of single-family homes in Stockholm County. Paper 4 analyzes sales of condominium apartments in Gothenburg.Papers 1, 2 and 3 contribute to the empirical auction literature by studying how the number of bidders, the seller’s list price strategy and auction participants bidding strategies affect the unfolding of auctions and the resulting transaction price. Using a hedonic model Paper 1 analyzes how the number of bidders affects the transaction price. Paper 2 extends the database in Paper 1 to cover a longer time period and develops the hedonic model in Paper 1 by adding explanatory variables describing the seller’s list price strategy and auction participants bidding strategies. Paper 3 develops the hedonic model in Paper 2 by incorporating a more thorough analysis of possible omitted variable bias and further employs methods used in the previous empirical literature to mitigate omitted variable bias by including as explanatory variable the residuals from a price regression of previous sales, as well as the residuals from a list price regression. Furthermore, the paper estimates how list price and bidding strategies affect the probability of a winner’s curse.Paper 4 studies auctions of condominiums in Gothenburg, but with a different focus than Papers 1, 2 and 3 since detailed auction data was not available. The paper instead focuses on how sellers’ list price strategy affect sales price and time-on-market, and how information asymmetry between brokers and sellers, combined with brokers’ commission fee structures, affect the choice of list price.