This paper examines the usefulness of Google Trends data for forecasting mortgage lending in the Czech Republic. While the official monthly statistics on mortgage lending come with a publication lag of one month, the data on how often people search for mortgage-related terms on the internet are available without any lag on a weekly basis. Growth in searches for mortgages and growth in mortgages actually provided are strongly correlated. The lag between these two growth rates is two months. Evaluation of out-of-sample forecasts shows that internet search data improve mortgage lending predictions significantly. In addition to forecasting performance evaluation, an experimental indicator of restrictively tight mortgage credit standards and conditions is proposed. Nowadays many countries run bank lending surveys to monitor the tightness of bank lending standards and conditions. The proposed indicator represents a complementary tool to such a survey.
JEL codes: C22, C82, E27, E51
Keywords: Credit demand, credit standards and conditions, credit supply, forecast evaluation, forecasting, Google econometrics, Internet search data, mortgage, smoothing
Issued: December 2014
Download: CNB WP 14/2014 (pdf, 568 kB)