Page 261 - Special Topic Session (STS) - Volume 3
P. 261
STS543 Veronica B. B. et al.
fixed effects, data are transformed into first difference. Moreover, residuals are
clustered by banks.
Empirical analysis. The empirical analysis includes two parts. The first part
estimates the impact of each prudential tool or measure on bank lending to
household borrowers as well as the impact on monetary policy conditions and
economic cycles. The second part looks at the impact on the loan quality using
nonperforming loans by banks.
Impact on bank lending to household borrowers. Using a panel
methodology, the impact at the loan level can be seen in equation 1,
∆ log , = + ∑ ∆ log ,− + ∑ =1 ∆ − + ,−1 +
=1
, +
,
(eq. 1)
where ∆ log is the quarterly change in the logarithm of loan
,
committed by bank to a household borrower based on the acquisition cost
of the property in real prices over a given period after the introduction or
change in a macroprudential tool, are bank fixed effects, ,−1 are bank
characteristics, , are macro-financial indicators. The main
coefficient of interest is ∑ which represents the impact of changes in a
=1
domestic macroprudential policy on bank loan commitment to household
borrowers.
1
In arriving at the main dependent variable, this study considered the
quarterly growth of appraised value of the residential unit, the appraised value
of the lot and the average acquisition cost of the property. These indicators
are converted in real terms using the Implicit Price Deflator for Real Estate
Activities from the National Income Accounts. Among these three variables,
the real average acquisition cost of the property proved to be statistically
stable and reliable as the main dependent variable. Moreover, this study
considers two separate dummies for tightening actions and loosening actions.
Such an approach could help verify asymmetric effects of each prudential tool
(Kuttner and Shim, 2016; Bruno et al. 2017). The exercise also considers the
intensity or the total number of times each prudential tool has been used,
classified by tightening and loosening measures and by resilience- and
cyclical- based measures. Moreover, the estimation uses the net intensity on
the use of each prudential tool, that is, net tightening and net loosening.
The exercise also estimates how much it takes for a given prudential tool
to propagate its effects on lending or the optimal k in equation 1. Equation 1
considers only the effect after one quarter. However, the propagation effects
could be longer, especially with respect to the implementation of a bank loan
1 In the empirical estimation, the lag effects included contemporaneous impact. However, these
contemporaneous estimates did not yield significant results.
250 | I S I W S C 2 0 1 9