Estimation of effects of recent macroprudential policies in a sample of advanced open economies
Ragnar Nymoen, Kari Pedersen, Jon Ivar Sjåberg
We analyse a quarterly panel data set consisting of ten advanced open economies that have introduced macroprudential policy measures: caps on loan to value and income (LTV and LTI), and debt service to income (DSTI) requirements in particular, but also risk weights (RW), amortization (Amort) and, less used, countercyclical buffer (CCyB). Estimation of dynamic panel data models, that also include the central bank rate, and controls for common nominal and real trends, gives support to the view that several of the measures may have reduced credit growth when they were introduced.The estimated impact effects are most significant for LTV, LTI and RW. For Amort, the long-run effect on credit growth is significant, and the same is found for RW. The estimation results when house price growth is the dependent variable are in the main consistent with the results for credit growth. The results do not support that CCyB has reduced lending (as a consequence of higher financing costs), and we suggest that the variable is mainly a control in our data set. In that interpretation, it is interesting that the estimated coefficients of the other five instruments are robust with respect to exclusion of CCyB from the empirical models. The results are also robust to controls in the form of impulse indicator saturation (IIS).