Page 264 - Special Topic Session (STS) - Volume 3
P. 264

STS543 Veronica B. B. et al.
                  performing loans (NPL). Following Chavan and Gambacorta (2016),   is a
                                                                                       ,
                  logit function of the ratio of gross NPL to total loans for bank b at time t.  In
                  particular, the logit function is given in equation 6 as,

                                                       
                                          ,  = ln [  ]     (eq. 6)
                                                    1− 

                      Given  the  perceived  persistence  of  NPL,  this  study  uses  a  dynamic
                  specification that includes lagged value of the NPL as an explanatory variable.
                  The study considers a specification that takes into account the sum of one lag
                  to  four  quarter  lag  effects  to  capture  the  total  effects  in  t.    Similar  to
                  specifications in equations 1 to 4, the bank-specific characteristics  ,−1  in
                  equation 5 includes the size of a bank (or total resources in real terms), liquidity
                  ratio, capital ratios using capital adequacy ratio and Common Equity Tier 1
                  ratio to total assets, funding composition using outstanding deposits relative
                  to total liabilities, and profitability of banks using real net interest income.
                      Robustness checks. Diagnostic tests are used to check  for normality of
                  residuals across equations at 1%, 5% and 10% levels of significance. The results
                  are  broadly  robust  against  normality  tests  and  different  specifications  of
                  dependent  and  independent  variables.  The  residual  tests  show  that  all
                  estimated coefficients are significant and that the instruments used are not
                  correlated  with  the  residuals  (using  Hansen  test).    The  standard  errors  of
                  regression  are  robust  and  that  the  errors  in  the  first  difference  regression
                  exhibit no second order serial correlation (using serial correlation test).

                  3.  Results
                      Following diagnostic and robustness checks, the results reveal important
                  findings.  First, tightening of domestic prudential, particularly those tightening
                  measures that are meant to preserve resilience of the banking system, are
                  effective in curbing growth of real bank loan commitments to borrowers for
                  acquiring  new  residential  properties.    The  results  show  that  tightening
                  macroprudential policies have direct and negative impact that can last up to
                  four  quarters  on  real  bank  loan  commitments  to  borrowers  based  on  real
                  acquisition  cost  of  new  properties  from  March  2014  to  December  2017.
                  Importantly, results reveal that tightening domestic macroprudential policies
                  vary with both business and financial cycles.  Overall, these findings confirm
                  other  studies’  observation  that  prudential  policies  are  more  likely  to  find
                  effectiveness.
                      Second,  this  study  highlights  the  bigger  negative  impact  of  tightening
                  prudential  measures  on  real  bank  loan  commitments  to  maintain  banks’
                  resilience in by U/KBs compared to TBs, an indication of the presence of bank
                  lending channel. Third, real bank loan commitments to household borrowers
                  are driven by bank deposits (relative to total liabilities), liquidity position and



                                                                     253 | I S I   W S C   2 0 1 9
   259   260   261   262   263   264   265   266   267   268   269