Page 273 - Special Topic Session (STS) - Volume 3

STS543 Muizz A. et al.
distribution of income of approved borrowers, using the Kolmogorov-Smirnov
test (Tzur-Ilan, 2018).
Second, prospective borrowers who were able to obtain financing after the
implementation of the policies, may have opted to take on a smaller loan that
is more in line with their affordability. Consequently, we expect borrowers’ DSR
for personal financing and residential property loan to decline after the
implementation of the policies. To measure this impact, we estimate the
following regression using ordinary least squares for personal financing and
residential property loan borrowers, respectively (Bekkum et al., 2019):

,, = + + + ,, (1)

,

∗
,, = + + + + + ,, (2)

,
2
3

1

where indexes individual borrowers, indexes state of residence, and
indexes time (quarters). The dependent variable, ,, , is the borrower’s DSR.
is a policy index, which would increase by 1 if a new tightening

measure is introduced and decline by 1 if a measure is relaxed (Cerutti et al.,
2017). However, it is important to note that no relaxation of measures took
place during the period of estimation. The state fixed effects ( ) control for

fixed differences across regions, such as housing affordability. We also control
for changes in the overnight policy rate (OPR) as this may have a direct impact
on the borrowers’ DSR. is a vector of control variables measured at the
,
borrower-level, and ,, the error term. The main parameter of interest, β,
measures the average borrowers’ DSR in the period after the implementation
of the policies relative to (unconstrained) borrowers acquiring loans in the
period before. We also estimate equation (2) to investigate the impact of the
policies on the low-income group.
For home buyers that were impacted by the policies, they may have
adjusted their housing choices by buying cheaper houses that may be smaller,
farther from the city centre or in a lower quality neighbourhood compared to
the average borrower. To measure this, we estimate equation (1) and (2) with
house price as the dependent variable.

4. Results
A. Changes in the Distribution of Borrower Characteristics
From the Kolmogorov-Sminov test, we found that the distribution of
income of newly-approved borrowers is statistically different after the
implementation of the policies, for both personal financing and residential
property loan (Chart 5 and 6). Notably, the average income of the population
is higher after the introduction of the measures. This is within expectations -
with the improvement in loan affordability assessment which takes into

262 | I S I W S C 2 0 1 9