In the modern world, the effectiveness of an economy is heavily dependent on the functionality of its banking sector. Credit-based loans make up a large majority of the money supply, and many institutions rely on these loans for funding operations and expansion. When such institutions collapse, their outstanding loans default. Non-performing loans deter lenders from loaning money, threatening the overall stability of the banking system. To that end, the welfare of the banking sector and by extension the economy as a whole can be measured by the percentage of loans that default.

This paper aims to identify the institutional and macroeconomic indicators that drive NPLs, and to evaluate the extent to which these indicators impact NPLs both individually and jointly with other indicators. This paper analyzes these indicators in a post-COVID context, following a period of economic crisis where significant government intervention was necessary to sustain international banking systems. In an environment where economic variables may act differently than they did previously, understanding of NPL indicators offers valuable insight for institutions in helping maintain financial stability of the banking system and the overall economy.

The findings of this study indicate that institutional quality and fiscal policy play an important role in determining the level of non-performing loans across countries. Rule of law, government effectiveness, and political stability have strong statistically significant negative relationships with respect to NPLs. Generally, a stable political environment with effective institutions and a strong government is conducive to decreasing NPLs. Fiscal policy variables, particularly subsidies and government debt, exhibit a negative and statistically significant effect as well. Government intervention, through increasing government spending and lowering taxes and cutting costs for businesses, helps control the quantity of NPLs within an economy. Interest payment is the only variable that shows a positive relationship with NPLs. The more money that needs to be allocated towards interest payments, the less money there is available to support the economy. When analyzed under conditions of an effective government making active fiscal policy changes, however, higher interest payments now become linked with a decrease in NPLs.

Other macroeconomic indicators such as GDP growth, inflation, climate risk, and unemployment, have no significant impact on NPLs. Overall, institutional quality and fiscal policy hold more importance in predicting and limiting NPLs than macroeconomic indicators, especially during the post COVID-19 period.

This paper contributes towards existing literature on NPLs in several important ways. First, this paper adopts a comprehensive cross-country dataset from 122 countries, providing broader and more generalizable insight on the indicators of NPLs. This paper also integrates both macroeconomic and institutional variables for analysis, providing a more extensive understanding of the factors that influence performance and stability of banking systems. Most importantly, this paper offers unique insight into the behavior of NPLs within the post COVID-19 period. Much of the existing literature of NPLs is based on data from before the COVID-19 pandemic. This study specifically focuses on the post-COVID-19 period, a new macroeconomic environment for banking institutions where it becomes important to reassess which factors now drive NPLs.

This paper proceeds as follows: Section 2 provides background information on existing literature. Section 3 highlights the empirical findings. Section 4 concludes the paper.

Various methodologies have been used and employed to analyze non-performing loans, such as regression analysis ([10]; [16]), dynamic panel estimation ([3]; [12]), panel data regression ([19]; [6]; [2]), GMM ([17]), ARDL ([11]), and GLS model ([7]).

The literature identifies various determinant variables for non-performing loans, including GDP ([20]), inflation rate ([1]), lending rate ([14]), exchange rate ([19]), liquidity ([11]), housing price index ([6]), stock market capitalization ([3]), broad money to GDP ([1]), government stability ([8]), leverage ratio ([18]), bank z-score ([15]), trade openness ([5]), and retail & corporate lending ([4]).

Existing studies focus on various countries and regions, such as Pakistan ([10]), China ([20]), United States ([6]), Ghana ([2]), Nigeria ([1]), Republic of Macedonia ([11]), and Bangladesh ([14]), as well as groups of countries and regions such as MENA ([8]), CESEE ([19]), Latin America ([17]), Central Europe ([18]), Asia ([7]), and transition economies ([9]).

The literature shows that non-performing loans are shaped by both bank-specific and macroeconomic factors. For example, [10] and [12] highlight the role of international bank efficiency, profitability, and management practices in reducing NPLs. In contrast, [3], [19], and [17], emphasize macroeconomic indicators, particularly GDP growth, inflation, and unemployment as a determinant of NPLs. [6] and [15] show that external shocks, pausing markets, and institutional collapse can also lead to a higher ratio of NPLs. Meanwhile, [8] and [4] suggest that in addition to bank stability, government and political stability can also have a significant impact on NPL quantity.

Within developed countries, indicators of NPLs identified by the literature remain relatively consistent. [3], [6], and [18] show the role of lending rate, specifically the percentage of loans that are high risk or high interest, in determining NPL quantity. Additionally, both [12] and [15] recognize capital adequacy as a strong indicator of NPLs. By contrast, studies analyzing NPLs within developing countries yield mixed conclusions. For instance, [2] and [9] identify inflation as an indicator for NPLs, while [19] and [11] explain that inflation is only impactful in certain contexts, [17] and [1] conclude that inflation has no effect on NPLs.

The extensive summary of existing literature on the determinant indicators of NPLs is presented in Table 1 below.

Table 1. Key studies on indicators of NPLs.

Data

Table A1 presents various indicators sourced from the World Bank World Development Indicators. These variables include non-performing loans as a percentage of total gross loans (Bank Loans), GDP growth as an annual percentage (GDP Growth), change in CPI as an annual percentage (Inflation), rule of law (Rule of Law), climate risk as a percentage change in CO₂ emissions excluding LULUCF since 1990 (Climate Risk), interest payments as a percentage of expense (Interest Payments), political stability and absence of violence/terrorism (Political Stability), unemployment as a percentage of total labor force (Unemployment), subsidies and other transfers as a percentage of expense (Subsidies), central government debt as a percent of GDP (Gov Debt), and government effectiveness (Gov Effectiveness). All variable data is taken from the most recent available year in 2022, which reflects the post-COVID-19 time period.

Descriptive Statistics

Table 2 provides descriptive statistics for variables used in the regression analysis. Bank Loans has 122 observations, with a mean of 4.389 and a standard deviation of 4.178, ranging from a minimum of 0.242 to a maximum of 27.704. GDP

Table 2. Descriptive statistics.

Growth has 122 observations, with a mean of 4.095 and a standard deviation of 4.408, ranging from a minimum of −19.609 to a maximum of 19.793. Inflation has 117 observations, with a mean of 10.054 and a standard deviation of 8.883, ranging from a minimum of 1.045 to a maximum of 72.309. Rule of Law has 122 observations, with a mean of 0.134 and a standard deviation of 0.908, ranging from a minimum of −1.7 to a maximum of 1.958. Climate Risk has 119 observations, with a mean of 216.531 and a standard deviation of 503.013, ranging from a minimum of −70.277 to a maximum of 3914.458. Interest Payments have 87 observations, with a mean of 7.597 and a standard deviation of 7.381, ranging from a minimum of 0.31 to a maximum of 41.317. Political Stability has 122 observations, with a mean of 0.012 and a standard deviation of 0.797, ranging from a minimum of -2.205 to a maximum of 1.25. Unemployment has 121 observations, with a mean of 7.076 and a standard deviation of 6.138, ranging from a minimum of 0.231 to a maximum of 35.359. Subsidies have 88 observations, with a mean of 46.193 and a standard deviation of 19.078, ranging from a minimum of 6.506 to a maximum of 84.994. Gov Debt has 44 observations, with a mean of 65.18 and a standard deviation of 42.367, ranging from a minimum of 15.393 to a maximum of 216.213. Gov Effectiveness has 122 observations, with a mean of 0.179 and a standard deviation of 0.898, ranging from a minimum of −1.747 to a maximum of 2.049.

Models

To determine the main individual indicators of NPLs, the following cross-section regression model will be applied. Many variables in the dataset are only available for one recent year for most countries. Additionally, this cross-sectional regression allows for broad and global comparisons during the specific economic recovery period.

Y_i = β₀ + β₁X_i + ε_i(Equation 1)

where Y_i is the dependent variable for observation i, which refers to Bank Loans. β₀ is the constant term representing the expected value of the dependent variable when the independent variable is zero. β₁ is the coefficient for the independent variable X_i which includes Growth, Inflation, Rule of Law, Climate Risk, Interest Payments, Political Stability, Unemployment, Subsidies, Gov Debt, and Gov Effectiveness individually. The coefficient shows the change in the dependent variable when the independent variable changes by one unit. ε_i represents the error term, which is the difference between actual value and the predicted value from the model.

In addition to the simple regression model, the following multiple regression model will be applied to determine the combined impact of indicators. After estimating several multiple regression models, only the most significant one is reported here.

Y_i = β₀ + β₁Gov Effectiveness_i + β₂Subsidies_i + β₃Interest Payments_i + ε_i (Equation 2)

where β₁, β₂, and β₃ are coefficients for the independent variables Gov Effectiveness_i, Subsidies_i, and Interest Payments_i, respectively. The coefficients show the change in the dependent variable when their associated independent variable changes by one unit. Several combinations of variables were tested, and the model with the most variables where all variables maintained statistically significant coefficients is reported.

Figure 1. Scatterplot graphics between Bank Loans, Rule of Law, Gov Effectiveness, Political Stability, GDP Growth, Subsidies, and Gov Debt.

Figure 2. Scatterplot graphics between Bank Loans, Inflation, Climate Risk, Interest Payments, and Unemployment.

A variety of interesting conclusions can be drawn from Figure 1 and Figure 2. Figure 1 shows the indicators (Rule of Law, Gov Effectiveness, Political Stability, GDP Growth, Subsidies, and Gov Debt) that have negative relationships with NPLs, suggesting that a higher value of these variables leads to a lower level of NPLs. In contrast, Figure 2 shows the indicators (Inflation, Climate Risk, Interest Payments, and Unemployment) that have positive relationships with NPLs, suggesting that a higher value for these variables correlates with a higher quantity of NPLs.

Table 3. Regression analysis.

Robust standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1.

Table 3(a) presents the result of three different regression models, where the dependent variable is Bank Loans and the independent variables are Rule of Law, Government Effectiveness, and Political Stability. A statistically significant result at one percentage level is marked with (***), the five percentage level is marked with (**), and the ten percentage level is marked with (*). The sign in front of the β coefficient shows whether the independent variable has a positive or negative impact on the dependent variable. A positive impact is indicated with a “+”, while a negative impact is indicated with a “−”. R-Squared shows the extent to which the independent variable explains the change in the dependent variable.

Column 1 shows the effect of Rule of Law on Bank Loans. The coefficient is negative and statistically significant at one percentage level. The negative sign indicates that as the Rule of Law for a given nation increases, the quantity of non-performing bank loans within that nation decreases. In other words, stronger Rule of Law is linked to better performance of a nation’s banking system. The R-squared value of 0.257 shows that 25% of the variation in the Bank Loans can be explained by Rule of Law. Column 2 and Column 3 present the effect of Government Effectiveness and Political Stability on Bank Loans, respectively. The coefficients are both negative and statistically significant to one percentage point, suggesting that there is negative correlation between these two variables and NPLs. More effective governments with higher political stability result in fewer bank loan defaults.

This relationship can be explained through observation that a nation’s political order and ability to enforce laws is also indicative of the strength of its banking system. Nations that have more stable governments are better able to enforce their policies and control their banking system, thereby decreasing the quantity of NPLs.

Table 3(b) presents the result of three different regression models where the dependent variable is Bank Loans and the independent variables are GDP Growth, Subsidies, and Gov Debt.

Column 4 shows the effect of GDP Growth on Bank Loans. The coefficient is negative and statistically insignificant. Column 5 shows the effect of Subsidies on Bank Loans. The coefficient is negative and statistically significant to one percentage point, suggesting that an increase in subsidies given by a government decreases the amount of bank loan defaults. Column 6 shows the impact of Gov Debt on Bank Loans. The coefficient is negative and statistically significant to ten percentage points. This indicates a weak inverse relation between government debt and bank loan defaults.

Increases in government debt and subsidies are indicators of expansionary fiscal policy. An increase in government debt arises from increased government spending while subsidies largely stem from tax cuts that help lower cost of production. This economic stimulation leads to increased household income, thereby making loans easier to manage and lowering risk of default. Additionally, an increase in government borrowing may increase the total gross loans of an economy without necessarily increasing the quantity of NPLs, leading to a decrease in the value of the variable as a whole.

Table 3(c) shows the result of four different regression models where the dependent variable is Bank Loans and the independent variables are Inflation, Climate Risk, Interest Payments, and Unemployment.

The only statistically significant variable is Interest Payments (Column 9) at one percentage level, suggesting that an increase in interest payments is linked to an increase in NPLs. A higher portion of expense being dedicated to interest payments often means less money contributed towards paying off loan principal, which may be caused by interest rate increases or poor borrowing habits. Both of these are signs of risky lending, which increases the chance of default.

Table 4. Multiple regression analysis.

Robust standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1.

Table 4 shows the result of multiple regression analysis examining the combined effect of the most important indicators on NPLs. The result shows that Gov Effectiveness, Subsidies, and Interest have negative and statistically significant effects on NPLs. When all variables are considered, government effectiveness becomes the strongest driver of NPLs. Interest payments now exhibit a negative relationship with NPLs rather than a positive relationship when considered alongside other factors. In countries with effective governments and active fiscal policy changes, a larger portion of government expense being diverted towards interest is a sign of attempts to limit the quantity of NPLs. While higher interest payments individually are usually linked to increases in NPLs, higher interest payments can also result from government spending and activity which assists in decreasing NPLs. To this end, higher interest payments are indicative of proactive institutions and overall greater government effectiveness. When estimated using equation 1, higher interest payments may not necessarily reflect higher debt payments from government institutions. Only when institutional quality and fiscal policy are accounted for do interest payments show their true predictive effect.

This paper investigates the macroeconomic and institutional determinants of non-performing loans across 122 countries within a post COVID-19 environment. The results emphasize the importance of the government in maintaining banking stability. Specifically, stronger government effectiveness, political stability, and rule of law significantly reduce NPLs. Subsidies and government debt also play an important role in mitigating NPLs, while interest payments are a significant positive driver of NPLs. The results also identify the insignificance of some indicators such as GDP growth, inflation, climate risk, and unemployment. The multiple regression analysis model mostly supports the evidence here.

Designing prudent fiscal policy and strengthening institutional frameworks prove effective in minimizing systematic risk of lending, thereby mitigating the effects of NPLs overall. This may involve adjusting fiscal levers such as government spending and tax rates, thereby increasing government debt and subsidies, two variables that have been identified as having a negative effect on NPLs. Ensuring government effectiveness and institutional stability through political policy also proves effective in controlling NPLs quantity.

Future studies could employ panel data methods that account for change in NPLs policy over time by examining the lagged effect of macroeconomic shocks. Once more data of the overall transition and recovery of the global economy from the COVID-19 crisis becomes available, studies could take advantage of the larger dataset to provide more robust empirical evidence. This allows future studies to validate the findings of this study, and to analyze the change in the significance of determinant factors over time. Future studies may also incorporate region-specific datasets to better aid policymakers in targeting strategies that address region-specific conditions and indicators.

I would like to thank Dr. Abdullah Yalaman for guiding me through the process of conducting this research.

Table A1. Definition of the variable.