Abstract
As the COVID-19 pandemic came unexpectedly, many real estate experts claimed that the property values would fall like the 2007 crash. However, this study raises the question of what attributes of an apartment are most likely to influence a price revision during the pandemic. The findings in prior studies have lacked consensus, especially regarding the time-on-the-market variable, which exhibits an omnidirectional effect. However, with the rise of Big Data, this study used a web-scraping algorithm and collected a total of 18,992 property listings in the city of Vilnius during the first wave of the COVID-19 pandemic. Afterwards, 15 different machine learning models were applied to forecast apartment revisions, and the SHAP values for interpretability were used. The findings in this study coincide with the previous literature results, affirming that real estate is quite resilient to pandemics, as the price drops were not as dramatic as first believed. Out of the 15 different models tested, extreme gradient boosting was the most accurate, although the difference was negligible. The retrieved SHAP values conclude that the time-on-the-market variable was by far the most dominant and consistent variable for price revision forecasting. Additionally, the time-on-the-market variable exhibited an inverse U-shaped behaviour.
Highlights
The emergence of the COVID-19 pandemic and its detrimental consequences to the global financial system were unexpected and affected millions of people by descending economic activity into a partial shutdown
Research results In accordance with previous studies on the topic of pandemics and real estate, this paper found a significant but adequate apartment price response during the COVID-19 pandemic
The COVID-19 pandemic has dramatically affected many economic operations, and within these circumstances, real estate experts have claimed that real estate prices might fall
Summary
The emergence of the COVID-19 pandemic and its detrimental consequences to the global financial system were unexpected and affected millions of people by descending economic activity into a partial shutdown. Since many authors have found the TOM variable significantly predicting price drop, a heatmap of TOM values according to the Vilnius city boroughs was created for all four months and both sell and rent operations. One thing to consider is that machine learning processes have a stochastic feature, meaning that in different iterations, the models changed accuracy positions [8, 38] This is especially true when SMOTE oversampling or stratified cross-validation that splits data into different sets is used. Number of rooms Sq.m Apartment floor Number of floors in the building Year Distance to shop Distance to school Distance to kinder Built_type Heating Time on the market (TOM) Initial listing price If located at city center If price change occurred. Number of rooms Sq.m Apartment floor Number of floors in the building Time on the market (TOM) Initial listing price If located at city center If price change occurred
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
