Table 1.
Description of the determinants and proxies of NPLs and a representative sample of their use in the literature
| Variable | Expected impact on NPL | Basis of the NPL influence sign | Relevant literature |
|---|---|---|---|
| Inflation rate | + | The state of high inflation hinders running a business and reduces the ability of households to settle their liabilities. | Curak et al., 2013; Klein, 2013; Kjosevski & Petkovski, 2021 |
| Unemployment rate | − | Unemployment reduces the ability of households to pay their liabilities. | Klein, 2013; Messai & Jouini, 2013; Wdowiński, 2014; Louzis et al., 2015; Kjosevski & Petkovski, 2021 |
| Banking sector development | − | A large number of loans means greater availability of financing. In addition, the high value of loans means that banks are mature and can properly assess creditworthiness. | Keeton & Morris, 1987;Abid & Zouari, 2014; Petkovski et al., 2021 |
| State debt | + | High state debt limits the availability of financing, which contributes to higher loan-servicing costs. | Louzis et al., 2015; Kjosevski & Petkovski, 2021 |
| GDP growth rate | − | Lower GDP growth means limited ability to settle liabilities. | Messai & Jouini, 2013; Wdowiński, 2014; Beck et al., 2015; Louzis et al., 2015; Kjosevski & Petkovski, 2021 |
| Interest rate | +/− | A high interest rate may limit lending, i.e., only those entities with high creditworthiness receive loans. On the other hand, the high cost of money may make it difficult to settle liabilities. | Curak et al., 2013; Messai & Jouini, 2013; Wdowiński, 2014; Beck et al., 2015 |
| Currency depreciation | + | The depreciation of the domestic currency means an increase in loans in foreign currency. This is important in the case of loans with a significant value in a foreign currency. | Klein, 2013; Wdowiński, 2014; Beck et al., 2015 |
[i] Source: Author's own research and design.
Table 2.
Data specification and symbols
| Variable | Symbol |
|---|---|
| Non-performing loan ratio in 2019 | NPL |
| Natural logarithm of non-performing loan ratio in 2019 | ln_NPL |
| Inflation rate is expressed as an index in 2019, where the base year is 2008 | INF |
| Unemployment rate is expressed as the average unemployment rate in 2009–2019 | UN |
| Banking sector development indicator as a relation between total loans granted to the private sector and GDP in 2019 | DCTOPS |
| State debt as a ratio of public debt to GDP in 2019 | DEBT |
| GDP growth rate as a relation of the value of GDP expressed in constant prices to the national currency in 2019 to 2008 | GDP |
| Interest rate as the average interest rate on loans granted in 2009–2019 | IR |
| OECD membership (dummy variable) | OECD |
| Currency depreciation as a ratio of the value of the currency expressed in SDRs in 2019 in relation to 2008 | CurDep |
[i] Source: Designed by the author
Table 3.
Descriptive statistics of variables
| NPL | INF | UN | DCTOPS | DEBT | GDP | IR | CurDep | |
|---|---|---|---|---|---|---|---|---|
| Mean | 6.44 | 1.60 | 8.37 | 61.40 | 53.60 | 1.24 | 10.16 | 1.31 |
| Maximum | 46.82 | 6.19 | 31.65 | 236.75 | 194.11 | 2.16 | 54.48 | 8.58 |
| Minimum | 0.20 | 1.00 | 0.71 | 3.23 | 2.58 | 0.79 | 0.52 | 0.76 |
| Standard deviation | 6.92 | 0.71 | 6.21 | 42.32 | 31.36 | 0.23 | 7.43 | 0.90 |
| Skewness | 3.38 | 3.92 | 1.55 | 1.37 | 1.40 | 0.89 | 2.78 | 6.20 |
| Kurtosis | 15.23 | 20.04 | 2.09 | 2.38 | 3.28 | 1.54 | 12.94 | 44.56 |
| Jarque-Bera test | 1.119.70 | 1.868 | 57.70 | 53.63 | 75.08 | 22.27 | 799.40 | 8,610.70 |
| JB Probability | 0.0000 | 0.0000 | 0.0000 | 0.0000 | 0.0000 | 0.0000 | 0.0000 | 0.0000 |
| Number of observations | 106 | 106 | 106 | 106 | 106 | 106 | 106 | 106 |
[i] Source: Designed by the author in R software
Table 4.
Top 10 of the highest and lowest NPL ratios in 2019
| Country | NPL as assets % | Country | NPL as assets % |
|---|---|---|---|
| Ukraine | 46.8 | Macao | 0.2 |
| Greece | 39.4 | South Korea | 0.4 |
| Cyprus | 30.5 | Micronesia | 0.4 |
| Angola | 17.7 | Hong Kong | 0.7 |
| Burundi | 17.7 | Luxembourg | 0.8 |
| Iraq | 14.1 | Australia | 0.9 |
| Moldova | 13.1 | Norway | 1.0 |
| Italy | 12.9 | United Kingdom | 1.0 |
| Albania | 12.8 | United States | 1.1 |
| Vanuatu | 12.7 | Singapore | 1.2 |
[i] Source: Designed by the author based on World Bank and IMF data
Table 5.
Ramsey's RESET test results
| RESET | 5.3029 |
|---|---|
| df1 | 2 |
| df2 | 95 |
| p value | 0.0065 |
[i] Source: Designed by the author based on R software results
Table 6.
Correlation matrix
| ln NPL | INF | UN | DCTOPS | DEBT | GDP | IR | OECD | CurDep | |
|---|---|---|---|---|---|---|---|---|---|
| ln NPL | 1 | ||||||||
| INF | 0.30 | 1 | |||||||
| UN | 0.25 | −0.20 | 1 | ||||||
| DCTOPS | −0.54 | −0.35 | −0.08 | 1 | |||||
| DEBT | 0.33 | 0.02 | 0.14 | 0.08 | 1 | ||||
| GDP | −0.11 | −0.05 | −0.13 | 0.00 | −0.26 | 1 | |||
| IR | 0.27 | 0.41 | −0.11 | −0.47 | −0.04 | −0.02 | 1 | ||
| OECD | −0.28 | −0.32 | −0.01 | 0.34 | 0.13 | −0.27 | −0.48 | 1 | |
| CurDep | 0.28 | 0.91 | −0.07 | −0.28 | 0.08 | −0.10 | 0.27 | −0.16 | 1 |
[i] Source: Designed by the author in R software
Table 7.
Regression statistics for the assessment of non-performing loans
| Variable | Model (a) | Model (b) | Model (c) | Model (d) | Model (e) |
|---|---|---|---|---|---|
| INF | 0.1624 | 0.1499 | 0.1690 | n/a | n/a |
| (0.1526) | (0.1783) | (0.1328) | |||
| UN | 0.0268* | 0.0279* | 0.0277* | 0.0227 | 0.0243* |
| (0.0258) | (0.0187) | (0.0190) | (0.0520) | (0.0318) | |
| DCTOPS | −0.0110*** | −0.0106*** | −0.0110*** | −0.0116*** | −0.0114*** |
| (0.0000) | (0.0000) | (0.0000) | (0.0000) | (0.0000) | |
| DEBT | 0.0105*** | 0.0104*** | 0.01071*** | 0.0107*** | 0.0110*** |
| (0.0000) | (0.0000) | (0.0000) | (0.0000) | (0.0000) | |
| GDP | −0.1737 | −0.1430 | n/a | −0.2259 | n/a |
| (0.5958) | (0.6577) | (0.4903) | |||
| IR | −0.0075 | n/a | −0.0066 | −0.0045 | n/a |
| (0.5274) | (0.5744) | (0.7007) | |||
| OECD | −0.3360 | −0.2906 | −0.3035 | −0.3856* | −0.3268* |
| (0.0789) | (0.0992) | (0.0921) | (0.0419) | (0.0468) | |
| p-value: | 0.0000*** | 0.0000*** | 0.0000*** | 0.0000*** | 0.0000*** |
| Residual standard error | 0.7050 | 0.7028 | 0.7024 | 0.7088 | 0.7037 |
| White test p-value: | 0.3654 | 0.3014 | 0.406 | 0.3185 | 0.2484 |
| Goldfeld–Quandt test p-value: | 0.9114 | 0.8307 | 0.9144 | 0.8789 | 0.8730 |
| Multiple R-squared | 0.4922 | 0.4901 | 0.4907 | 0.4814 | 0.4785 |
| Adjusted R-squared | 0.4559 | 0.4592 | 0.4598 | 0.45 | 0.4578 |
| AIC | 236.3774 | 234.8116 | 234.6833 | 236.6023 | 233.1925 |
| BIC | 260.3484 | 256.1191 | 255.9909 | 257.9098 | 249.1731 |
| RESET | 1.2344 (0.2956) | ||||
| Shapiro | 0.9841(0.2376) | ||||
| Jarque Bera | 3.0718 (0.2153) | ||||
| Observations | 106 | 106 | 106 | 106 | 106 |
Table 8.
VIF testing results
| INF | UN | DCTOPS | DEBT | GDP | IR | OECD | |
|---|---|---|---|---|---|---|---|
| vif(a) | 1.3699 | 1.1468 | 1.4102 | 1.1042 | 1.2153 | 1.6610 | 1.5495 |
| vif(b) | 1.3278 | 1.1215 | 1.2615 | 1.1038 | 1.1885 | X | 1.3277 |
| vif(c) | 1.3531 | 1.1194 | 1.4102 | 1.0576 | X | 1.6245 | 1.3884 |
| vif(d) | 1.0817 | 1.3429 | 1.0971 | X | 1.2004 | 1.6100 | 1.4983 |
| vif(e) | 1.0280 | 1.1399 | 1.0400 | X | X | X | 1.1428 |
[i] Source: Designed by the author in R software