Page 421 - Special Topic Session (STS) - Volume 3

STS552 Natalia Nehrebecka
borrowers. In contrast to debt securities that represent concentrated markets,
with several banks holding large shares of debt securities. Over half of the off-
balance sheet exposure was, however, open credit lines.
For the comparison, I consider two models: (1) Quantile regression (QR)
and (2) Bayesian Model Averaging (BMA). Table 2 shows the estimation results
of QR for the 5th, 25th, 50th, 75th and 95th per centile and the corresponding
BMA estimates.

Table 2: Parameter estimates Quantile Regression and Bayesian Model
Averaging

PIT
BMA

Variables Post
Q(0.05) Q(0.25) Q(0.50) Q(0.75) Q(0.95)
Mean
(Post SD)
PIP

0.0028*** 0.000192 -0.0048*** -0.0061*** -0.0004*** -0,0031

ln_EAD (0,0004)
(0.0002) (0.0003) (0.0003) (0.0002) (0.0000)
1

0.0673*** 0.163*** 0.117*** 0.0491*** 0.0011*** 0,1213
Collateral
Indicator (0,0013)
(No, Yes) (0.0002) (0.0003) (0.0003) (0.0002) (0.0000)
1

-0.335*** -0.120*** -0.0720*** -0.0361*** -0.0054*** -0,0699
Days Past
Due (less (0,0053)
than 90 (0.0032) (0.0048) (0.0051) (0.0031) (0.0003)
days) 1

-0.506*** -0.343*** -0.178*** -0.0595*** -0.0072*** -0,224
Days Past
Due (more
(0,0032)
than 90 (0.0019) (0.0029) (0.0030) (0.0018) (0.0002)
days)
1
-0.0808*** -0.0114*** -0.00277*** 0.00175*** 0.0000 -0,01
Time spent
in default (0,0015)
(months)) (0.0005) (0.0007) (0.0008) (0.0004) (0.0000)
1
0.00913*** 0.0381*** 0.0366*** 0.0122*** -0.0000 0,0353

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