Determinants of non-performing loans in North Macedonia
Funding The authors received no direct funding for this research. Author details
Download 1.78 Mb. Pdf ko'rish
|
Determinants of non performing loans in North Macedonia
- Bu sahifa navigatsiya:
- Citation information
Funding
The authors received no direct funding for this research. Author details Petros Golitsis 1 Khurshid Khudoykulov 2 E-mail: kh.khudoyqulov@tsue.uz Savica Palanov 1 , 3 1 Business Administration & Economics Department, CITY College, University of York Europe Campus, Thessaloniki, Greece. 2 Finance Department, Tashkent State University of Economics, Tashkent, Uzbekistan. 3 Komercijalna Banka AD, Skopje, Republic of North Macedonia. Citation information Cite this article as: Determinants of non-performing loans in North Macedonia, Petros Golitsis, Khurshid Khudoykulov & Savica Palanov, Cogent Business & Management (2022), 9: 2140488. Notes 1. The unemployment rate measures the number of peo- ple actively looking for a job as a percentage of the labour force. 2. The variable used is the weighted interest rates on total extended denar credits (average for the period, in %, p.a.). Other macroeconomic and bank-specific explanatory variables were tested, including nominal effective exchange rate, the real effective exchange rate, household final consumption expenditure, inflation, banks’ capital to assets, and foreign exchange reserves, etc., but were statistically insig- nificant and thus were not used in the final specification. 3. The reported minimum value of unemployment is 14.8 and not 14.7. The end date of this variable is the first Golitsis et al., Cogent Business & Management (2022), 9: 2140488 https://doi.org/10.1080/23311975.2022.2140488 Page 16 of 40 quarter of 2022; we have used extrapolation to acquire the Q2 of 2022 value. 4. On the final specification, the dummy variable gener- ated statistically significant results and the sign is as expected (i.e., it has an adverse impact on NPLs). 5. It has to be noted that even though the use of the cointegrating ARDL approach is reasonable, as stated, when there is a mix of I(0) and I(1) series, this does not apply when it is the dependent variable that is I(0). In our case, the dependent variable, NPL, is I 1 ð Þ accord- ing to the ADF unit root test, but according to the Zivot-Andrews is not. Thus, according to the asymp- totic theory developed in the paper of Pesaran, Pesaran et al. ( 2001 ), which provides a simple univari- ate framework for testing the existence of a single level relationship between y t and x t when it is not known with certainty whether the regressors are purely I(0), purely I(1) or mutually cointegrated, our results may not be fully reliable because a “degenerate case” might arise. This apart from leading to a difficulty in interpretation (e.g., as mani- fest in the 100% reversion each period), could question the reliability of the statistical results. 6. Available at the Appendix. 7. Prior to the implementation of the model, a decision on the maximal number of lags to be included is required. We have chosen 4 (note that our data is quarterly). The decision on the optimal number of lags is made based on Akaike Information Criterion, AIC. Still, using the general-to-specific modeling frame- work, we developed models with varying numbers of lags for the determinants differentiated (equal number of lags for all determinants) ending up to the reported specification. Also, in terms of the post-estimation tests, it has to be noted that the BG test for serial correlation failed to reject the respective null hypoth- esis of autocorrelation, and furthermore, the residuals follow a normal distribution (see the JB statistic on Table 4 ), do not display heteroscedasticity (according to BPG test), and there is no specification bias, according to Ramsey Reset test, as well. The CUSUM test supports further the stability of the model; the CUSUM and the CUSUM of squares remains between the 5% critical bounds, which provides statistical evi- dence on the stability of the parameters. The results are provided on Table 4 . We have also applied a short- run ARDL(1,3,2,2,0). For more details see the appendix. Download 1.78 Mb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling