Abstract

This paper presents the insights from the experimental study on the use of web data from real estate platforms in official statistics. It discusses possible applications of the information derived from real estate sales and rental offers in the production of flash estimates of prices, as well as the observation of the phenomena which are not covered by traditional data sources. The results of the study demonstrate that web data appear particularly useful in monitoring growing rental market, especially in large cities with a large number of immigrants and students. Moreover, taking Poland as a reference point, the study proves that web data may serve as a reliable source for the validation of traditional data, i.e. surveys and administrative registers. The results demonstrate a high degree of convergence in price trends, making web data a promising data source for flash estimates of the price index. Finally, web data may provide additional information on the standard of buildings, surrounding area, and elements of amenities available in the property, which can be applied in more advanced models of real estate prices (i.e. hedonic models).

While web data appear to have a significant potential to augment real estate statistics, it needs to be stressed that their use in the official statistical production poses a number of technical, methodological and quality-related challenges. The paper discusses selected issues related to the use of data from online real estate platforms, such as landscaping of web data sources, technical aspects of web scraping, the stability of web sources, dealing with deduplication of offers and data completeness.