Prudential Policies in the Eurozone: A Propensity Score Matching Approach

This paper studies the effectiveness of micro- and macroprudential policy tools in the euro area. The established empirical literature on macroprudential policy generally considers panel estimations that suffer from two estimation biases, i.e., a selection bias and a time bias. We control for the former by a propensity score matching approach. Based on a logit model, we estimate the probability of a policy tightening for every country at each point in time. Matching procedures then find one or more matching partners for every tightening event with a similar likelihood of a tightening but no shift in the prudential policy stance. An iterative approach ensures that we offset the time bias, which exists if the estimation does not control for effects of preceding and subsequent prudential policy changes. We find that the announcement of a prudential policy tightening reduces credit growth significantly by about 1% on average. We further differentiate between effects along three dimensions. First, we observe that lending is more affected when policymakers have not communicated the implementation of measures before. Second, the effects are more substantial when EU/EA institutions are behind changes in the prudential policy stance. Third, microprudential policy measures have a bigger impact than macroprudential policies.