This paper investigates whether sellers in the housing market are loss averse, which is argued to be the source for the positive correlation between prices and volume in housing markets. I use transaction data with information about individual sellers and buyers provided by Eiendomsverdi AS, a private firm owned by the largest commercial banks in Norway, and Statistics Norway. The data cover the Oslo housing market in the period 2005-2020, with 32,044 repeated observations (purchases-to-listings), including unique identifiers for housing units and individuals acting as sellers or buyers, prices, housing attributes, and debt. Using a sample of repeated purchases-to-listings that include appraisal values from surveyors to capture the price expectations of sellers, I find that sellers are not loss averse in their list price choice. However, using hedonic predictions to capture price expectations yields a significant loss aversion effect, highlighting the issue of substituting price expectations with predictions. I interpret the findings as indicating that sellers who receive appraisals and know that buyers can observe these, are not only better informed than others, but also act as being constrained in their list price choice.
I estimate multiple linear regressions of list price models, with either hedonic predictions of selling prices or appraisal values, and prospective losses and gains, to identify whether sellers facing prospective losses are putting higher list prices than they otherwise would have done. I also use different combinations of predictions and unobserved heterogeneity in appraisal values, and unit fixed-effects, to tease out what causes the possible bias in the baseline hedonic-based model estimates. Using hedonic predictions as substitute for price expectations, give results encompassing the existing literature. However, using observed appraisal values provided to the sellers shortly before listing gives no loss aversion effect. A more in-depth investigation of the two models suggests that the hedonic-based model can give spurious results.
There is a positive price-volume correlation in the Oslo housing market, but this cannot be explained by loss aversion. The model framework used in the existing literature may give spurious results, highlighting the importance of dealing with unobserved heterogeneity.