Banking Sector Profitability: Does Household Income Matter?

Abstract

Household incomes, their level and dynamics are one of the factors that ensure the achievement of the Sustainable Development Goals. At the same time, stable development of the banking sector, which is impossible without steady earnings, determines economic growth, which also positively affects reaching the Sustainable Development Goals. The paper examines the impact of household income on the return on assets of the banking sector in Russia using annual time series from 2003 to 2019. The study was conducted using formalized economic and mathematical methods of analysis by linear regression with least squares tests on the significance of the model, with tests for redundancy of insignificant variables (Wald test), Ramsey test on the functionality of the model, White and Breusch Pagan test for heteroscedasticity (heterogeneity of observations) and multicollinearity by method of inflation factors, graphic method. The Multiple Linear Regression (MLR) model was used. The results show that (1) an increase in the growth rate of household income and deposits in the non-financial sector has a positive effect on the return on assets of the banking sector; (2) an increase in the growth rate of the price of Brent crude oil and non-performing loans negatively affect the dependent variable; and (3) the regions that have the greatest (positive and negative) impact on profitability of whole bank sector in Russia were identified. Increasing household incomes and eliminating inequality in the incomes of the population of different regions will have a positive impact not only on social well-being but will also provide the banking sector with the opportunity for profitable operation and create conditions for sustainable growth. Our conclusions are useful for the regulator and individual banks and can be taken into account when developing and implementing policies aimed at sustainable development.

1. Introduction

Households are important counterparties to the banking sector. The dynamics of households’ welfare determines their activity in relations with banks. Household wealth is based on income, which changes over time. Sufficient income, its fair distribution among members of society as a whole, the population of a particular country or group of countries, contributes to the elimination of poverty, reduction in inequality, and forms the conditions for well-being and health care [1], and receiving quality education. Income of the population forms the demand for all the variety of goods and services of other sectors of the economy, including the banking sector. Stable development of the banking sector contributes to economic growth [2,3] and is accompanied by the creation of jobs with decent conditions, wages [4], the formation of sources for the economy’s transition to renewable energy [5], as well as the reorientation of the system of education, morality with new civilization values, greening the consciousness and worldview of population [6]. Banking crises as the opposite state of stability have a negative impact on society, and contribute to increasing inequality [7]. Thus, the stable development of the banking sector is a necessary, though not the only, condition in achieving a balance between economic growth, social and environmental development. The stable development of the banking sector forms the financial conditions for the transformation of banking policy in accordance with the concept of sustainable development [8], the development of banking products and services of environmental and social orientation [9], financing of green projects [10], charitable programs for environmental conservation, staff development, and improvement of environmental culture in society. This balance is achieved through mechanisms that should ensure the simultaneous growth of household incomes, the reduction in household inequality, and the stable development of the banking sector, which generally contributes positively to economic growth. An important source of the latter are steady earnings.

The situation in Russia is characterized by the functioning of a single banking sector over a large territory with significant regional differentiation of household income, which should be taken into account by banks when developing and implementing banking policy [11]. By accruing deposits from the population and providing loans to households in different regions, the banking sector shapes the macroeconomic balance and generates risks, which are further projected into financial results. The most important indicators for assessing the financial result are indicators of profitability of bank assets and capital.

At the macrolevel, bank profitability indicators are included in the macroprudential indicators of the banking sector, based on which the regulator monitors financial stability and decides on the implementation of regulatory measures. Since the banking sector is interconnected with other sectors of the economy, it is also affected by macroeconomic factors such as the key rate, inflation, currency exchange rate, monetization, real estate prices, and the state of the financial market. The price of Brent crude oil is additionally important for the Russian economy.

The purpose of this paper is to assess the impact of household income on banking profitability as an indicator of macroeconomic financial stability, and to identify the regions that have the greatest positive and negative impact on the profitability of the whole banking sector in terms of differentiated regional income.

The novelty of this study is determined, firstly, by the inclusion in the forecast model as an independent variable of household income, along with macroeconomic and macroprudential variables. Secondly, the dynamics of income is taken into account, for which the rate of income growth is used, by year. Thirdly, the study takes into account the uniqueness of the Russian situation, which is determined by the geographical differentiation of household incomes in different regions of Russia during the functioning of a single banking sector. Fourthly, on the basis of the constructed model, the regions are identified that have the greatest impact on the profitability of the assets of the unified banking sector in terms of differentiated regional incomes.

The structure of the article includes Section 2, “Materials and Methods,” Section 3, “Results”, Section 4, “Discussion”, and Section 5, “Conclusions”.

2. Materials and Methods

We use data on macroeconomic indicators, aggregate cash incomes of the Russian household as a whole and by region, indicators of the banking sector for 2003–2019 [12]. The choice of the time range of measurements is fundamental in connection with the extraordinary influence of atypical factors of the 2020 COVID-19 pandemic, which should be excluded in order to clear the trend.

The methods and approaches used to form the scientific results of the study are based on the methodology of economic, statistical analysis, the information base of Rosstat and the Bank of Russia for 2003–2019, According to the dynamics of income of the banking sector and households (total cash income of the population by regions of Russia) [13]. The choice of the method of forming a mathematical model is dictated by the conditions for its construction, the objective function and limitations, and the need to assess the relationships between variables corresponding to empirical data [14]. The specification of the regression equation determines the choice of the formula itself, the definition of the parameters of the selected equation, and the analysis of the quality of the equation with a check of its adequacy to empirical data. Further, the authors presented the improvement of the equation using tests [14].

In an empirical study of the dependence of the objective function on variables, formalized economic and mathematical methods of analysis using linear regression were applied with least squares testing for the significance of the model, then tests for redundancy of insignificant variables (Wald’s test), Ramsey’s test for the functionality of the model, White’s test and Breusch–Pagan for heteroscedasticity (heterogeneity of observations), and multicollinearity by the method of inflation factors [15]. The formalization of the equation was carried out using the specialized statistical package, Gretl.

We study the impact of household income on the return on assets of the banking sector as the most important indicator of financial stability [16,17,18]. When modeling the performance of the banking sector, the nominal variables used as regressors are significant [19,20]. We proceed from the assumption that household income determines the indicators of the bank balance in terms of deposits and loans, as well as risks affecting bank profitability.

Various indicators are used in the literature devoted to identifying factors that affect bank profitability. To assess the impact of the resource base, it is common to use the share of customer funds—non-financial organizations in bank liabilities [21], as well as a similar indicator—the share of interbank loans in bank liabilities [22]. Diversification of bank liabilities in the context of individual sectors from which banks raise funds is necessary to reduce risks, in particular concentration risks. The funds of the population as part of bank obligations are stable, in the conditions of the functioning of deposit insurance systems they are characterized by a low risk of withdrawal [23], and therefore a minimal threat to bank liquidity. Diversification of bank liabilities, allocation of household funds as an object of banking policy contributes to the improvement of resource management, which positively affects bank profitability [24,25]. The share of household funds attracted by banks in the structure of bank liabilities shows the importance of household funds for the banking sector. In turn, the ability of the population to place funds in banks determines household income.

The indicators of non-performing loan (NPL), market risk, and liquidity risk are used as variables reflecting the risk of the banking sector. In addition to households, the counterparties of the banking sector are non-financial organizations [26,27,28,29]. Deposits to assets and loans to assets ratios are used as two broad indicators of bank balance sheet characteristics [30,31,32]. We also use the ratio of household loans to household income due to the fact that its growth negatively affects the macroeconomic situation in which the banking sector operates [33].

Since we are examining the role of household cash income in the profitability of the banking sector, we introduce the price of residential real estate into the model as an indicator that determines the financial behavior of households and affects financial stability [16]. In addition, unemployment affects the risks of bank transactions with households [19,34,35,36].

To cover risks, adequate capital is required, the amount and sufficiency of which are the most important indicators of financial stability [16,37,38], in connection with which the variables of dynamics and capital adequacy of the banking sector were used.

The dependent variable is determined on the basis of the aggregate profit of the banking sector as the difference between profitable and loss-making credit institutions [39]. When the economic situation is stable, the volume of profitable banks’ profits is close to the volume of the banking sector’s profit; under ideal conditions, these values coincide. We use the ratio of profitable banks’ profits to total banking profit as an indicator of a favorable situation in the banking sector: the closer the value is to 1, the fewer uncovered losses in the banking sector, the more accurate the reflection of the impact of bank profitability on banking stability.

The level of monetization for M2 was introduced to take into account the ability of the banking sector to generate money supply adequate to the level of GDP [40]. This, in turn, affects the balance sheet indicators of the banking sector, as it determines the provision of households and non-financial organizations with payment instruments.

Between 2003 and 2019, Russia experienced two devaluations of the ruble: in the fourth quarter of 2008–2009 and in the fourth quarter of 2014–2015, due to the crisis manifestations in the economy, and in late 2014—additional changes in the exchange rate regime. The depreciation of the ruble was accompanied by a rise in inflation and the regulator’s key rate. Households and many non-financial organizations receive their income and revenues in rubles. The increase in the volume of liabilities to banks as a result of ruble revaluation of debt in foreign exchange determined the growth of overdue debt under the principal debt and interest payment, required additional formation of reserves, and had a negative impact on the banking sector’s income. To take into account the effect of ruble devaluation and related regulatory measures on bank profitability, indicators of the dynamics of the ruble–dollar exchange rate, inflation and the key rate are introduced.

The state of the Russian stock market is shown by the RTS Index, the index dynamics being a market risk factor and determining the banking sector revenues and expenditures. Our choice of this indicator was based on the fact that the RTS Index, while having the same base of calculation as the Mosbirzh Index, demonstrates different dynamics from the latter, because it is calculated in US dollars; therefore, its use is more acceptable in terms of expressing the dimension of a number of other variables, as well as the chosen methodology.

An important factor and macroeconomic indicator of Russia’s economic development is the price of Brent crude oil. Since Russia is one of the main suppliers of oil to the world market, oil export revenues make up a significant part of Russian budget revenues [41]. The level of income as one of the most important sources of economic growth will be determined by the global price environment, which is based on the price of Brent oil [42].

The implementation of the concept of sustainable development in the state policies of the countries of the world community provides for the decarbonization of the economy, including the transition to renewable energy sources [43]. Such processes form the risks of economic changes for countries exporting energy resources, including oil [44], which affects the interests of all economic agents [45], including the banking sector. At the level of an individual, economic changes can act as factors of negative emotions, interfere with a person’s well-being [46], thus jeopardizing the achievement of the sustainable development goals. The presence of an established relationship between the price of oil and economic growth in Russia [47] should be taken into account when modeling the profitability of the most important sector of the economy—the banking sector. In this connection, we introduce the price of oil into the composition of the regressors of our model.

Descriptive statistics are presented in Appendix A 

Table A1. Variables, calculation, and descriptive statistics.

Group	Factor	Calculation	Variable	Standard Deviation	Min	Max
Macroeconomic factors	Household income	The growth rate of total nominal money income of the population, %	Income_nom	1.02	1.30	0.09
	Brent oil price	Brent oil price growth rate for the year, %	Brent_price	0.57	1.48	0.25
	Key intrest rate	Key rate, %	Key_rate	0.06	0.16	0.03
	Inflation	Inflation rate, %	Inflation	0.03	0.13	0.04
	Unemployement	Unemployment rate, %	Unemployment	0.05	0.08	0.01
	Residential real estate prices	Rate of growth of the cost of 1 sq.m. of housing space, %	Real_estate	0.91	1.43	0.14
	Money aggregator M2	M2/ GDP, %	Monetization	0.20	0.46	0.08
	Currency exchange rates	Ruble exchange rate to the US dollar, %	USD_RUR	0.88	1.58	0.17
	Index RTS	RTS Index growth rate	IRTS	0.28	2.75	0.58
Banking sector specific factors	Credit risk	Share of problem and bad loans in the total amount of loans (NPL), %	Credit_risk	0.40	248.57	73.19
	Market risk	Percentage of total capital, %	Market_risk	0.40	51.79	13.64
	Liquidity	The ratio of liquid assets to total assets, %	Liquidity	0.40	41.04	8.84
	Equity	Growth rate of banking sector capital, %	Capital_banks	1.00	1.58	0.16
	Capital adequacy	Equity (capital) adequacy H1.0, %	Capital_adequacy	0.40	19.62	4.30
	Household loans	Growth rate of debt on loans to individuals, %	Debt_individuals	0.89	2.11	0.36
	Household loans to income ratio	The ratio of outstanding loans to individuals to the total nominal monetary income, %	Debt_to_income	0.02	0.24	0.07
	Loans to bank capital ratio	The ratio of loans to individuals to banks’ assets, %	Credit_ individuals	0.03	0.17	0.04
	Loans to non-financial institutions in banks’ assets	Loans to non-financial institutions as a percentage of assets, %	Credit_nonfin	0.01	0.46	0.10
	Loans to financial institutions in banks’ assets	Loans to credit institutions as a percentage of assets, %	Credit_fin	0.05	0.12	0.02
	Deposits of individuals in liabilities	The ratio of deposits and other funds attracted from natural persons to liabilities, %	Deposit_individuals	0.21	0.31	0.03
	Deposits of non-financial organizations in liabilities	The share of funds borrowed from enterprises and organizations in liabilities, as %	Deposit _nonfin	0.25	0.35	0.03
	Ability of credit institutions to generate profit	The ratio of the volume of profit of profitable credit institutions to the total volume of profit of the banking sector, %	BankPr_Fin_res	1.01	15.57	3.72

.

Regressors are checked for adequacy and significance by correlation coefficients [48], regressors correlated by more than 0.75 in modulus are excluded [49].

Furthermore, the correctness is determined by the method of least squares and the significance of the compiled model is checked. The model is then tested for redundancy in non-significant variables (with a p-value > 0.05) [50], and their joint significance is determined. To confirm the obtained data, non-zero values of the regression parameters are checked using the Wald test [51]. Then, using the Ramsey test (RESET) [52], the specification (functionality) of the model is checked and the correctness of the inclusion of parameters in it is assessed by the auxiliary regression of the dependent variable on the factors of the original model, taking into account various degrees of the values estimated from the original model dependent variable. Finally, the final model is assessed for multicollinearity using the variance inflation factor (VIF) method.

To identify the regions that have the greatest impact (both positive and negative) on the regressor of the Income_nom model, the model includes differentiated income indicators by regions using the OLS method according to the dependent variable Income_nom. Regional data cover 17 points in 85 regions, standard deviation = 0.11, min = 0.94, max = 1.86. After excluding redundant variables based on the results of testing the null hypothesis, the reduced model is assessed for adequacy based on regional data, and the corresponding regions are identified.

3. Results

The results of calculating the correlation coefficients of the regressors are presented in

Table 1. Correlation matrix in observations of 2003–2019, at 5% critical values (two-sided).

Indicators	Income_Nom	Brent_Price	Credit_Risk	Market_Risk	Liquidity
Income_nom	1.00	0.37	0.23	−0.08	0.60
Brent_price		1.00	0.34	0.29	0.61
Credit_risk			1.00	0.79	0.66
Market_risk				1.00	0.46
Liquidity					1.00
Indicators	Capital_adequacy	Key_rate	Unemployment	Inflation	Real_estate
Income_nom	0.43	0.80	0.72	0.72	0.74
Brent_price	0.63	0.31	0.27	−0.06	0.39
Credit_risk	0.61	0.41	−0.07	0.30	0.46
Market_risk	0.64	0.04	−0.25	−0.037	0.15
Liquidity	0.903	0.67	0.22	0.33	0.69
Capital_adequacy	1.00	0.41	0.10	0.26	0.50
Key_rate		1.00	0.63	0.64	0.63
Unemployment			1.00	0.49	0.28
Inflation				1.00	0.375
Real_estate					1.00
Indicators		Debt_individuals	Debt_to_income	BankPr_Fin_res	Credit_individuals
Income_nom		0.80	−0.91	−0.10	−0.67
Brent_price		0.36	−0.58	−0.68	−0.48
Credit_risk		0.24	−0.30	−0.13	−0.31
Market_risk		−0.23	−0.08	−0.29	−0.06
Liquidity		0.51	−0.64	−0.40	−0.72
Capital_adequacy		0.26	−0.51	−0.42	−0.52
Key_rate		0.65	−0.79	−0.19	−0.77
Unemployment		0.44	−0.82	0.01	−0.71
Inflation		0.57	−0.55	0.27	−0.30
Real_estate		0.72	−0.66	−0.35	−0.49
Debt_individuals		1.00	−0.72	−0.22	−0.48
Debt_to_income			1.00	0.31	0.80
BankPr_Fin_res				1.00	0.19
Credit_individuals					1.00

.

After checking the regressors for multicollinearity, Market_risk, Capital_adequa, Liquidity, Debt_individua, Debt_to_income, Key_rate, Monetization, USD_RUR, and Real_estate were excluded from the model; the assessment showed the significance of the resulting model, since the R-squared = 0.998397, and the p-value (p-value) is less than 0.05 (

Table 2. OLS model, observations 2003–2019 (period T = 17 years, dependent variable Fin_result).

Regressors	Coefficient	Standard Error	t-Statistic	p-Value	Significance
const	−16.0072	1.81136	−8.837	0.0031	***
Income_nom	5.96731	0.411961	14.49	0.0007	***
Brent_price	−0.296927	0.111478	−2.664	0.0761	*
Credit_risk	−0.00182742	0.000415794	−4.395	0.0218	**
Unemployment	6.32273	4.07850	1.550	0.2189
Inflation	2.96172	1.37032	2.161	0.1194
BankPr_Fin_res	0.0162323	0.00971058	1.672	0.1932
Credit_individuals	−7.82751	1.41552	−5.530	0.0117	**
Credit_nonfin	0.143502	0.267428	0.5366	0.6288
Credit_fin	7.84625	3.18738	2.462	0.0907	*
Capital_banks	1.32610	0.192622	6.884	0.0063	***
Deposit_individuals	23.9462	3.38565	7.073	0.0058	***
Deposit_nonfin	6.99547	1.05906	6.605	0.0071	***
IRTS	0.0188063	0.0282861	0.6649	0.5537

*, **, *** These symbols indicate the degree of significance of the coefficients for the model (the more symbols, the more significance).

and

Table 3. OLS estimations, observations from 2003–2019 (period T = 17 years, dependent variable Fin_result).

Name	Value	Name	Value
Average dependence of variables	0.894575	Standard deviation of the dependent variable	0.507795
Sum of square residuals	0.006613	Model standard error	0.046949
R-square	0.998397	Corrected R-square	0.991452
Fisher distribution F (13, 3)	143.7496	p-value (F)	0.000841
Log likelihood	42.62000	Akaike criterion	−57.23999
Schwarz criterion	−45.57500	Hennan-Quinn test	−56.08047
Rho parameter	−0.363433	Darbin-Watson statistics	2.725513

).

As a result of testing the reduced model, testing the null hypothesis, the insignificant regressors Unemployment, Inflation, Credit_nonfin, BankPr_Fin_res and IRTS were excluded from the model.

The insignificance of the unemployment rate in the profitability of the banking sector indicates the absence of a critical situation with the employment of bank borrowers, the implementation by banks of timely preventive measures to reduce credit risks caused by the loss of income by borrowers—households (household income shocks). This confirms the correctness of the choice of the observation period, which made it possible to exclude the impact of sharp but temporary changes in the level of unemployment.

The insignificance of the inflation indicator confirms the connection between banking indicators in Russia precisely with the nominal values of other variables, which is consistent with the conclusions of other authors [19,20]. In Russia, the change in the inflation rate is quite expected, is controlled by the bank management, and is reflected in a balanced way in the pricing of bank liabilities and assets with interest payments.

The insignificant share of loans to non-financial organizations in the banking assets is explained by the orientation of the banking sector to other areas of placement of funds, which is due to the difficult situation in the economy. Banks do not consider these assets as acceptable in terms of profitability to risk ratio, which, in fact, should cause concerns of the national regulator.

The irrelevance of the RTS Index in ensuring profitability of the banking sector as an indicator of financial stability indicates insufficient development of the Russian stock market, which does not have a significant impact on the macroeconomic environment in which the banking sector functions.

The insignificance of the ratio of profit of profitable credit institutions to the aggregate result in the profitability of the banking sector testifies to the insignificance of aggregate losses incurred by banks, which correlates with the statements of the Bank of Russia [53].

After excluding insignificant regressors, carrying out the Wald test, Ramsey (RESET), White and Breusch–Pagan tests, the final model was obtained (

Table 4. Reduced OLS model (observations 2003–2019, period T = 17, dependent variable Fin_result).

Regressors	Coefficient	Standard Error	t-Statistic	p-Value	Significance
const	−12.5585	0.725722	−17.30	<0.0001	***
Income_nom	5.91298	0.370524	15.96	<0.0001	***
Brent_price	−0.210205	0.0809251	−2.598	0.0317	**
Credit_risk	−0.000988597	0.000243788	−4.055	0.0037	***
Credit_individuals	−5.73146	0.792736	−7.230	<0.0001	***
Credit_fin	3.23712	1.73959	1.861	0.0998	*
Capital_banks	1.12619	0.150923	7.462	<0.0001	***
Deposit_individuals	16.6969	1.02713	16.26	<0.0001	***
Deposit_nonfin	5.29934	0.838867	6.317	0.0002	***

and

Table 5. Resulting OLS estimates in the reduced model, (period T = 17 years, dependent variable Fin_result).

Name	Value	Name	Value
Average dependence of variables	0.894575	Standard deviation of the dependent variable	0.507795
Sum of square residuals	0.022177	Model standard error	0.052651
R-square	0.994625	Corrected R-square	0.989249
Fisher distribution F (8, 8)	185.0312	p-value (F)	2.88 × 10−8
Log likelihood	32.33415	Akaike criterion	−46.66830
Schwarz criterion	−39.16938	Hennan-Quinn test	−45.92289
Rho parameter	−0.051175	Durbin–Watson statistics	2.078856

). These (*, **, ***) symbols indicate the degree of significance of the coefficients for the model (the more symbols, the more significance).

The results of the Wald test confirm the obtained data that the p-value > 0.05, which means the regressors Unemployment, Inflation, Credit_nonfin, BankPr_Fin_res, and IRTS can really be excluded (Test statistics: F (5,3) = 1.41229, p-value 0.412693).

The results of checking the specification (functionality) of the model and evaluating the correctness of the inclusion of parameters in the model using the Ramsey test show that the specification of the original model is defined correctly, and there is no need to add the squares of some variables.

The results of checking the model for heteroscedasticity or heterogeneity of observations, the presence of unequal variances of the random error of the regression model using the White and Breusch–Pagan tests indicate the absence of heteroscedasticity, and the normal distribution of errors in the regression.

As can be seen from the model (

Table 6. The final model for assessing the impact of household cash income on the return on assets of the banking sector as an indicator of financial stability, taking into account macroenvironmental factors.

Regressors	Coefficient	Standard Error	t-Statistic	p-Value	Significance
const	−12.5585	0.725722	−17.30	<0.0001	***
Income_nom	5.91298	0.370524	15.96	<0.0001	***
Brent_price	−0.210205	0.0809251	−2.598	0.0317	**
Credit_risk	−0.000988597	0.000243788	−4.055	0.0037	***
Credit_individuals	−5.73146	0.792736	−7.230	<0.0001	***
Credit_fin	3.23712	1.73959	1.861	0.0998	*
Capital_banks	1.12619	0.150923	7.462	<0.0001	***
Deposit_individuals	16.6969	1.02713	16.26	<0.0001	***
Deposit_nonfin	5.29934	0.838867	6.317	0.0002	***
Average dependence of variables	0.894575		Standard deviation of the dependent variable		0.507795
Sum of square residuals	0.022177		Model standard error		0.052651
R-square	0.994625		Corrected R-square		0.989249
Fisher distribution F (8, 8)	185.0312		p-value (F)		2.88 × 10−8
Log likelihood	32.33415		Akaike criterion		−46.66830
Schwarz criterion	−39.16938		Hennan-Quinn test		−45.92289
Rho parameter	−0.051175		Durbin–Watson statistics		2.078856

) and the regression equation, the greatest positive influence on the financial result of the banking sector as an indicator of financial stability is exerted by the growth rates of the total nominal household cash income, the share of loans to financial organizations in bank assets, and the share of deposits. These (*, **, ***) symbols indicate the degree of significance of the coefficients for the model (the more symbols, the more significance).

From Table 6 it follows that the equation of the regression model can be represented as:

Fin_result = −12.6 + 5.91 ∗ Income_nom − 0.210 ∗ Brent_price − 0.000989 ∗ Credit_risk − 5.73

∗ Credit_individuals + 3.24 ∗ Credit_fin + 1.13 ∗ Capital_banks + 167

∗ Deposit_individuals + 5.30 ∗ Deposit_nonfin

The greatest negative impact on the financial result of the banking sector as an indicator of financial stability is exerted by the growth rate of the price of Brent crude oil, the share of loans to households in bank assets, and the indicator of credit risk as the share of problem and bad loans in the total volume of loans.

The positive influence of factors is due to the following:

With the growth of the aggregate nominal household cash income, the banking profitability increases, that is, the theoretical proposition on the connection between the state of the banking sector and the state of bank counterparties is confirmed.

The growth of profitability with the growing share of loans to financial institutions in bank assets is due to the fact that Russian banks consider such loans as profitable investments with low risk and high liquidity, which, if necessary, can be easily converted into cash. The conclusion is consistent with the findings [22] that there is a positive relationship between the optimal level of cash as the share of cash in assets, which balances costs and benefits and maximizes bank profitability. In the case of Russian banks, banking profitability is maximized by the share of highly liquid loans in assets.

The established positive impact of the share of deposits of households and non-financial organizations in bank liabilities is consistent with the findings [54] and corresponds to the situation when the banking sector is limited in external borrowing and focuses on domestic sources that are cheap and stable [55], which is typical for Russia [56], especially after 2008. The positive impact on the profitability of the Russian banking sector of an increase in equity (capital) is also consistent with the findings for other countries [57,58,59].

The negative impact on the bank profitability of factors is due to the following:

The higher the volatility of the price of Brent oil, the less accurate the forecast values of economic indicators, indicators of bank balance sheets, the higher the cost of preventing and covering losses from shocks [60]. The negative impact on the profitability of the banking sector’s assets of the share of loans to households in bank assets is explained by the expansion of lending to households, increased competition between banks in search of solvent borrowers, and cheaper credit products, which correlates with the findings for Latin American countries [55], as well as [61]. Fully consistent with the findings presented in the literature, the negative impact of the growth in the share of problem and bad loans in total loans (NPL) on bank profitability.

Actual and calculated data observed and predicted values according to the resulting final model (Figure 1 and Figure 2) confirm the adequacy of the identified dependencies.

Figure 2 shows the proximity of the calculated forecast points for the final model and real data in the period under review; the regression errors are distributed normally, the standard error of the regression = 0.0526515 (Figure 2).

Data on the analysis of the confidence intervals are presented in

Table 7. Results of the analysis of confidence intervals for the regression coefficients of the final model.

Variable	Coefficient	95 Confidence Interval
const	−12.5585	(−14.2320, −10.8849)
Income_nom	5.91298	(5.05855, 6.76741)
Brent_price	−0.210205	(−0.396819, −0.0235918)
Credit_risk	−0.000988597	(−0.00155077, −0.000426422)
Credit_individuals	−5.73146	(−7.55952, −3.90341)
Credit_fin	3.23712	(−0.774389, 7.24863)
Capital_banks	1.12619	(0.778160, 1.47422)
Deposit_individuals	16.6969	(14.3284, 19.0655)
Deposit_nonfin	5.29934	(3.36491, 7.23377)

.

Since all the regression coefficients fell into the 95% confidence interval (t (8, 0.025) = 2.306,

Table 8. Evaluation of regressors for the presence of multicollinearity.

Regression Parameters	Assessment of Multicollinearity
Income_nom	6.038
Brent_price	2.372
Credit_risk	1.837
Credit_individuals	4.856
Credit_fin	5.887
Capital_banks	3.361
Deposit_individuals	3.918
Deposit_nonfin	3.073

, the model was constructed correctly. More detailed results of data analysis by points of predicted and real values, as well as their residuals, are shown in Figure 3.

As a result of evaluating the model using the method of inflation factors (variance inflation factor, VIF), the absence of multicollinearity was found (Table 8).

To identify the regions that have the greatest impact (both positive and negative) on the profitability of the whole banking sector, the model includes regional data (

Table 9. Model based on regional data (observations 2003–2019, period T = 17 years, dependent variable: Income_nom).

Regions	Coefficient	Standard Error	t-Statistic	p-Value	Significance
const	0.465119	0.0663713	7.008	0.0902	*
Moscow_obl	−0.152706	0.112927	−1.352	0.4054
Chukotka_ao	0.397748	0.0346844	11.47	0.0554	*
Evreysk_ao	−0.508221	0.112010	−4.537	0.1381
Tyva	0.412239	0.128164	3.216	0.1919
Altay	0.191633	0.0757855	2.529	0.2397
Moscow	−0.966784	0.0744743	−12.98	0.0489	**
Saint Petersburg	0.393694	0.0459430	8.569	0.0740	*
Adygeya	−0.915095	0.0940421	−9.731	0.0652	*
Kalmykia	0.588974	0.0759471	7.755	0.0816	*
Ingushetia	−0.240073	0.0459673	−5.223	0.1204
Sverdlov_obl	1.89714	0.234836	8.079	0.0784	*
Tumensk_s	−0.776683	0.0983948	−7.894	0.0802	*
Kabardino_balk	−0.479785	0.0892513	−5.376	0.1171
Karachaev_cherk	0.839427	0.273332	3.071	0.2004
Sev_Osetia	−0.0850522	0.0566995	−1.500	0.3743

) using the OLS method using the dependent variable Income_nom or an increase in nominal household income. These (*, **, ***) symbols indicate the degree of significance of the coefficients for the model (the more symbols, the more significance).

The indicators p-value, R-square, and conclusions on assessing the significance of the regional model were detailed in

Table 10. Resulting OLS estimates in the model based on regional data (period T = 17 years, dependent variable Income_nom).

Name	Value	Name	Value
Average dependence of variables	1.141721	Standard deviation of the dependent variable	0.087297
Sum of square residuals	0.000021	Model standard error	0.004625
R-square	0.999825	Corrected R-square	0.997193
Fisher distribution F (15, 1)	379.9770	p-value (F)	0.040237
Log likelihood	91.35775	Akaike criterion	−150.7155
Schwarz criterion	−137.3841	Hennan-Quinn test	−149.3903
Rho parameter	−0.039397	Durbin–Watson statistics	2.041759

.

After the test for the redundancy of variables in order to improve the information criteria about the significance of factors, a reduced model was obtained based on regional data (

Table 11. Reduced model based on regional data (observations 2003–2019, period T = 17 years, dependent variable: Income_nom).

Regions	Coefficient	Standard Error	t-Statistic	p-Value	Significance
const	0.366719	0.112084	3.272	0.0097	***
Chukotka_ao	0.212423	0.112655	1.886	0.0920	*
Moscow	−0.797693	0.227230	−3.511	0.0066	***
Saint Petersburg	0.477883	0.163597	2.921	0.0170	**
Adygeya	−0.536294	0.154714	−3.466	0.0071	***
Kalmykia	0.465731	0.133621	3.485	0.0069	***
Sverdlov_obl	1.20471	0.244511	4.927	0.0008	***
Tumensk_s	−0.361463	0.192036	−1.882	0.0925	*

) and its resulting estimates (

Table 12. Resulting OLS estimates in the reduced model based on regional data (period T = 17 years, dependent variable Income_nom).

Name	Value	Name	Value
Average dependence of variables	1.141721	Standard deviation of the dependent variable	0.087297
Sum of square residuals	0.005983	Model standard error	0.025784
R-square	0.950928	Corrected R-square	0.912761
Fisher distribution F (7, 9)	24.91490	p-value (F)	0.000034
Log likelihood	43.46987	Akaike criterion	−70.93974
Schwarz criterion	−64.27404	Hennan-Quinn test	−70.27716
Rho parameter	0.036598	Durbin–Watson statistics	1.853009

). These (*, **, ***) symbols indicate the degree of significance of the coefficients for the model (the more symbols, the more significance).

As can be seen from the model according to Table 11 and Table 12, the Chukotka Autonomous Okrug, the Republic of Adygea, and the Republic of Kalmykia have the greatest negative impact on the dependent variable, and the greatest positive impact is made by Moscow and Saint Petersburg (the values of each of the factors).

The observed data and predicted parameters according to the model with regional data are presented at Figure 4.

Thus, our results are consistent with the conclusions [62] about the need for geographic diversification of the bank’s strategy.

4. Discussion

As a result of the study, it was shown that the profitability of assets of the banking sector in Russia is most positively influenced by the positive dynamics of nominal household cash income, the rate of capital growth, the share of deposits of households and non-financial organizations in bank liabilities, and the share of loans to financial institutions in bank assets. The strongest negative impact on the profitability of the Russian banking sector is exerted by such macroeconomic factors as the increase in the price of Brent crude oil, the share of problem and bad loans in the banking portfolio, and the share of loans to households in bank assets.

Parallel positive effects for the banking sector and social welfare in society can be achieved through appropriate policies aimed at eliminating inequality on any grounds.

When developing policy, banks and the regulator should take into account additional factors that affect the amount of income. In the literature, a direction of research is developing, called ecofeminism. According to the concept of ecofeminism, the eradication of the imbalance between nature and humanity is possible with the eradication of gender inequality. Gender inequality can manifest itself, among other things, in the difference in income by gender. Per capita income of women participating in microfinance schemes [63], household size has an impact, while age negatively affects the income of people with low and medium incomes.

Researchers [25] argue that there is a difference in the use of loans received by women and those received by men: women mainly use loans for consumption. On the one hand, the expansion of lending for consumption stimulates demand, respectively, production, as a result, contributes to economic growth. However, on the other hand, consumption in excess of real needs is contrary to responsible consumption, and can lead to excessive spending of resources, conflict with the achievement of the Sustainable Development Goals. In conditions of low financial literacy, which, as a rule, is inherent in low-income households, the situation may be complicated by additional risks for creditor banks [64] as a result of the population’s debt load [65], which requires efforts to improve financial literacy and form a financially responsible culture.

The results obtained on the positive impact on the profitability of the share of household deposits in banking liabilities should be reflected in banking policy, provide for the development of various banking service channels that will ensure financial inclusion for different groups of the population [59]. In the context of the development of digital technologies, banks are able to strengthen the deposit base by attracting funds from a wider range of people, including people with disabilities, as well as households located in geographically remote settlements. Our findings are consistent with those on the positive impact of digital financial inclusion on economic growth by [66], the development of which strengthens the possibility of a positive impact of household deposits on bank profitability. At the same time, increasing the financial accessibility of banking services for different groups of the population contributes to strengthening the social well-being of members of society, which together contributes to the achievement of the Sustainable Development Goals.

5. Conclusions

The study revealed a significant impact of household income on the profitability of the Russian banking sector. Increasing household incomes not only brings social well-being closer, but also creates the conditions for economic growth [66], including the strengthening of the stability of the banking sector, which is confirmed by the identified positive relationship between the income of the population and the profitability of banking assets as the most important characteristic of banking stability. The value of the population’s income goes beyond the formation of conditions for social well-being. Achieving the goals of sustainable development in terms of improving the well-being of the population is impossible without decent wages for all members of society, regardless of geographical differences. The development and implementation of the economic policy of the state [67] and the policies of individual banks should take into account the whole variety of possible consequences due to the established positive impact on the profitability of the banking sector of increasing incomes of the population.

Factors that have a negative impact on the profitability of the banking sector deserve special attention from the state and the regulator, since they pose a threat to banking stability and, as a result, can act as a deterrent to economic growth. When implementing the banking policy, the identified relationship between the price of oil and the profitability of the Russian banking sector should be taken into account. The resulting conclusion about the direct and negative impact of oil prices on the profitability of the banking sector is consistent with the conclusions [43] for Turkey. When conducting stress testing of the banking sector, the state regulator should take into account the impact of global economic and geopolitical events on world oil prices [44], and provide options for an adequate macroprudential response to possible risks in order to prevent threats to banking stability [68], which, in turn, is a necessary condition for the profitability of responsible banking [10], which ensures the achievement of the Sustainable Development Goals through a balance between economic development and the interests of society.

The results obtained have some limitations. Thus, in the study, households are differentiated on a regional basis, which is of particular importance for Russia as a country with a significant extent of territory and geographical differentiation in incomes of the population in different regions [65]. However, the study does not take into account differences due to gender inequality [63,66].

The use of the indicator of average per capita income also has some disadvantages, since it does not fully reveal the impact of the incomes of different groups of the population, for example, those grouped by income level in rich and poor households [24].

A possible limitation on the use of the results may be an additional factor not only of a macroeconomic but also a global scale, such as a pandemic [69]. During a pandemic, national governments of different countries implement unprecedented measures to support the economy and society, the population receives additional payments to compensate for the loss of part of their income, bank regulators use the credit holiday regime [70]. The results obtained can be assessed for sustainability after the end of the pandemic.

Further research can focus not only on identifying the positive and negative impact of certain factors on the financial performance of the banking sector, but also on assessing the balance of interests of banks and its counterparties under the influence of significant factors, their perception by society. An increase in the profitability of banking assets can be ensured by banks raising interest rates on loans at a higher rate compared to rates on deposits, for example, in the context of the development of inflationary processes [71]; however, this limits the ability of contributors to receive fair payment for their funds. As a result, imbalances may form between the development of the banking sector, which is provided at the expense of the interests of creditors and depositors, and a reduction in the opportunities for improving the welfare of households that place funds in banks. Imbalances do not contribute to maintaining a positive perception of their interaction with banks among banking counterparties, can cause negative psycho-emotional consequences and become obstacles in ensuring the sustainability of society [46]. Further research can be aimed at identifying possible imbalances of interests of various participants in relations with the banking sector and developing proposals for eliminating imbalances, which should be reflected in the banking policy of [72]. In this regard, banks should develop corporate governance, since corporate governance has a positive effect on bank profitability [73], good corporate governance practices improve financial results [74]. The implementation of the principles of responsible banking in the policy of each individual bank will contribute to the sustainable development of the economy and society.

Our conclusions are useful for banks in implementing adequate banking policies, as well as for the regulator in deciding the issue of differentiating the parameters of regulatory instruments aimed at ensuring financial stability, in the context of the pursuit of the Sustainable Development Goals.