Determinants of non-performing loans in North Macedonia

Diagnostics
F-statistic
Probability
BG
1.2226
0.3030
BPG
1.3102
0.2407
Ramsey RESET test
0.1537
0.6966
Jarque-Bera
1.827103
0.401019
*** denotes 0.01 significance level, ** 0.05 significance level, * 0.10 significance level. The R-squared and the 
diagnostics are based on the long-run form, also known as the Error Correction form. (Detailed results are available 
at the appendix; see tables 6-9). 
Table 5. Estimated ARDL Coefficients (in the short-run only)
Variable
Coefficient
Standard Error
t-statistic
Probability
D(UN(−2)
0.4627
0.1895
2.4411**
0.0180
D(UN(−3)
0.3972
0.1804
2.2011**
0.0321
D(INT(−1))
1.5288
0.7892
1.9370*
0.0581
D(INT(−2))
2.1233
0.8049
2.6381**
0.0109
D(LGDP)
−2.6863
1.0354
−2.5944**
0.0122
DLGDP(−1)
−2.8648
1.1537
−2.4832**
0.0162
DLGDP(−2)
−1.8279
1.0223
−1.7880*
0.0795
Variable(s) 
F-value 
of Wald test
Chi
2
-value 
of Wald test
Probability value of 
F
Probability value of 
Chi
2
C(2),C(3),C(4),C(5)
3.1101**
12.4404**
0.0226
0.0144
C(6),C(7),C(8)
4.2936***
12.8808***
0.0087
0.0049
C(9),C(10),C(11),
2.7848**
8.3543**
0.0497
0.0392
*** denotes 0.01 significance level, ** 0.05 significance level, * 0.10 significance level. The short-run ARDL is a 1,3,2,2,0. 
The ordering of the variables, in relation to the reported lags, is as stated on the table above. Moving to joint 
significance under the Wald test, C(2)−C(5) on Unemployment, C(6)−C(8) on interest rates, and C(9)−C(11) on GDP. C 
(1) is the lagged NPL and C(13) are the Gross loans C(12) and are insignificant (available at the appendix). Also, 
Residual and Stability diagnostics are consistent with the econometric expectations and this model is robust as well 
(for more details see the Appendix; Tables 10-11, respectively). 
Golitsis et al., Cogent Business & Management (2022), 9: 2140488