Using online job postings to predict key labour market indicators

Authors

ŠTEFÁNIK Miroslav LYÓCSA Štefan BILKA Matúš

Year of publication 2023
Type Article in Periodical
Magazine / Source Social Science Computer Review
MU Faculty or unit

Faculty of Economics and Administration

Citation
Web https://journals.sagepub.com/doi/full/10.1177/08944393221085705
Doi http://dx.doi.org/10.1177/08944393221085705
Keywords vacancy statistics; online data; time series; predictive modelling; unemployment; employment
Attached files
Description We explore data collected as an administrative by-product of an online job advertisement portal with dominant market coverage in Slovakia. Specifically, we process information on the aggregate quarterly registered number of online job vacancies. We assess the potential of this information in predicting official vacancy, employment and unemployment statistics. We compare the characteristics of the online job posting data with those reported in comparable studies conducted for the Netherlands and Italy. Several differences are identified; most notably, our data are more persistent and stationary around a linear time trend. Additionally, we assess the predictive potential of the online job posting data by comparing in- and out-of-sample estimates of three regression models that predict job vacancy statistics and employment and unemployment levels one to four quarters ahead. Irrespective of the predictive horizon and labour market indicator, the online job posting data always provide a statistically significant predictor. These results are further solidified in an out-of-sample study that shows that forecast errors are lowest for predictions generated by models incorporating online job posting data. In general, the usefulness of the data seems best for longer forecast horizons.

You are running an old browser version. We recommend updating your browser to its latest version.