Simona Malovaná, Martin Hodula, Zuzana Gric, Josef Bajzík
The ever-increasing use of borrower-based measures such as loan-to-value, debt-to-income, and debt service-to-income limits has created a demand to better understand the transmission and effectiveness of such policy. In this paper, we collect more than 700 estimates from 34 studies on the effect of borrower-based measures on bank loan provision. A birds-eye view of our dataset points to significant fragmentation of the literature in terms of the estimated coefficients. On average, the introduction or tightening of borrower-based measures reduces annual credit growth by 1.6 pp. Using a battery of empirical tests, we verify the presence of a strong publication bias, especially against positive and statistically non-significant estimates. The bias-corrected coefficient is about half the size of the uncorrected mean of the collected estimates but remains safely negative. Further, we explore the context in which researchers obtain such estimates and we find that differences in the literature are best explained by model specification, estimation method, and underlying data characteristics.
JEL codes: C83, E58, G21, G28, G51
Keywords: Bayesian model averaging, borrower-based measures, macroprudential policy, meta-analysis, publication bias
Issued: August 2022
Download: CNB WP No. 8/2022 (pdf, 1.4 MB)