House Price Modeling under Covid-19: Analysis of parameters on online listing platforms

This study tried to examine how Covid-19 affected house prices in online listing platforms for the Istanbul Metropolitan area. In all online listing platforms in Turkey, net living area, building age, being in a gated community, the number of floors, and floor of the apartment are the primary filtering and evaluation criteria. We analyzed how and in what direction these parameters affect house prices, depending on people's preferences, from the beginning of 2020, which is considered the beginning of Covid-19, to June of 2021, the period when life began to continue relatively independent from covid-19. While doing this, we had 635,234 observations of house sales from online listings. We divided the data into three groups for houses with lower, middle and upper-income level prices, running them in a split model would be a better option when considering Istanbul's metropolitan structure. For each dataset, we have created regression models on a monthly basis and tracked the change of parameter coefficients. While all parameters in the model gave meaningful results for the lowest price segment, the significance level decreased as the prices increased. During pandemic, the tendency of the low-income group has evolved towards a modern form of housing in gated communities. As a result, the tendency to live in old buildings has decreased and the "large space requirement" related to size has left its place to "more room" houses in these preferences. When we run two co-models constructed at the beginning and end of the period, the coefficients for living in the gated communities increased by 58%, the coefficients for the number of rooms increased by 41% and the coefficients for the net living area increased by 21%. The building age coefficient changed its sign to negative as expected. Furthermore, none of the parameters except the net living area in the highest price group yielded to a significant result.