An econometric analysis of banking financial results in Ukraine

An econometric analysis of banking financial results in Ukraine

Oleg Vasyurenko


This paper explores the building of an econometric model with a binomial dependent variable for analysis of banking activities financial results in Ukraine. Identified are salient factors that exert substantial impact on banking financial results in the country. Statistical data for the period 2000 to 2002 were employed in this study. The extracted factors are useful in explicating the division of Ukrainian banks into respective clusters based upon their financial characteristics. The bank clusters statistically derived by the present study might be considered functional peer-groupings by financial professionals, policy makers and regulators when conducting comparative analyses among banks in Ukraine.


A study of factors which potentially impact the dynamics (including capital flows) and sustainable development of a banking system is integral given the decisive role that banks play in the virtually any economy. What can be learned from the incorporated analyses contributes to the germane literature in a timely fashion, given Ukraine’s present status as a transitional economy–a nation punctuated by unpredictable economic events and pervasive governmental amendments to economic policies. Even in stable economies, banks may fall victim to unintended negative consequences owing to frequent systemic financial and credit irregularities: economic crises; bankruptcies; insolvencies; oscillations in the money and capital markets; and other manifestations of an unbalanced market economy.

This study prospectively bears further implications; specific testing outcomes may arguably be regarded as signals to an early detection of unreliable banks. In such cases, timely sanitation procedures could then be strategically exacted to remedy, for example, a bank’s unpredictability–a common symptomatic precursor to bank liquidation. An undertaking of this sort is generally carried out by a country’s national bank which is charged with the monitoring of development dynamics of individual banks as well as the banking system as a whole based upon a certain system of indices and regulations. The development of like banking control is in the offing in Ukraine.

Notwithstanding extant controls, this paper advocates for broader applications of various analytical model constructions and qualitative analyses of banking system parameters; the development of versatile econometric models and studies can contribute to more fairly-weighted decision-making in the banking sector, and, hence, to a potentially more stable and efficient banking system. Accordingly, this paper constructs a statistical model aimed to detect indices that may differentiate functional stability from functional instability within the Ukraine banking system.


Bank condition diagnostics are widely employed by banking supervision authorities in developed countries under the name of Early Warning Systems (EWS). For a review of such systems see Sahawala and Berg (2000).

The most common international banking valuation technique, and one which is adhered to in the Ukrainian Banking system, is referred to as CAMEL, a banking mnemonic for the five principal areas of bank assessment: Capital, Assets, Management, Earnings and Liquidity. Though understood to be germane to banking, the CAMEL criteria have been applied in assessing even broader economic strata (Lavelle, 2004).

Three risks that have consistently produced financial distress in commercial banks are leverage risk, credit risk, and liquidity risk. Bank size and CAMELS-related variables capture the impact of other factors that may affect downgrade risk (Putnam, 1983; and Cole and Gunther, 1998). Leverage risk is the risk that losses will exceed capital, rendering a bank insolvent; Credit risk is the risk that borrowers will fail to make promised interest and principal payments; and liquidity risk is the risk that a bank will be unable to fund loan commitments or meet withdrawal demands at a reasonable cost (Gilbert, Meyer and Vaughan, 2002).

With respect to countries comprising the Former Soviet Union (FSU), some aspects of banking systems, including functional efficiency, reliability and development efficacy, are treated in works by, for instance, Matovnikov (2000), Garshin (2003), Vasyurenko and Azarenkova (2003), and Scannell, Safdari and Newton (2003). Additionally, a key aspect of banking activities analysis, bank stability, is studied in works by Garshin (2003), Kolary, Glennon, Shin and Caputo (2002), and Cole and Gunther (1998). These papers focus on factors that essentially contribute to bankruptcies.


The matter of bankruptcy is a pertinent concern for a developing Ukrainian banking system. Figure 1 presents a chronology from 1992 to 2002 of Ukraine’s registered (shaded) versus liquidated banks.

As is evident from Figure 1, and in concert with findings in a study by Stone and Rasp (1991), the data samples are not sufficiently large enough to warrant tests of significance. The small number of observations may lead to unsubstantiated statistical interpretation of econometric results. It is worth specifying that parallel works by Russian scholars are based on sample sizes upwards of 1000 registered banks and liquidated banks.


Given aforementioned data deficiencies, this paper proposes to investigate instead the effect of certain parameters characterizing the activities of Ukrainian banks based on their financial results. The suitability of such analysis is based on the rationale that profit is a chief index of operationalization for virtually any business entity and on the fact that Ukrainian banking system is characterized, in spite of a substantial increase of issued loans, attracted assets, and its own capital, by very ambiguous values of financial results as indicated in Figure 2.


As evidenced in Figure 2, in spite of a trend to increase the financial results since year 2000, the more recent values remain substantially below those seen in previous years, 1995 to 1997. Notwithstanding the disparities, this item comprises only 0.3% of GNP. Moreover, a typical feature of the Ukrainian banking system is a stable group of up to 8% of banks with negative values of current year financial result.


This study employs a statistical model with a binomial dependent variable, not uncommon to the present type of analysis. (For a study of U.S. banks, which employs a probit regression where the dependent variable takes a value of one (1) for any bank whose CAMELS rating falls from satisfactory to unsatisfactory … and a value of zero (0) if the bank is examined but not downgraded, see Gilbert, Meyer and Vaughan, 2002).

In general form, such models may be written as follows:

P(y) = F(x, b)


P(y) is the probability of the dependent variable taking a binomial value. In this study the dependent variable is the variation of the current year financial results value for particular Ukrainian banks within the period 2000 to 2002.

The dependent variable takes on the value of:

one (1) if a bank has maintained an exclusively positive trend of current year financial results during the entire reported period (in other words, demonstrating a sustainable growing trend in financial results year to year);


zero (0), if such a trend was not detected.

F (…) is the function of a random value distribution assessment:

x represents independent variable regressions. More than 30 indices were used as assessment data which reflected the activities of 157 continuously operating Ukrainian banks (data from sources published in the press);

b the regression parameters to be evaluated.

After a preliminary econometric study, the following preliminary parameters (numbers correspond to those depicted in Figure 3) were selected as independent regressors for a model with a binomial independent variable:

1. KR – total amount of bank credit portfolio;

2. VP – percentage of securities portfolio in bank assets structure;

3. D1 – amount of deposits by businesses;

4. D2 – amount of deposits by individuals;

5. RA – amount of reserves for bank active operations;

6. CB- bank authorized stock;

7. DZ – amount of bank debtors’ liabilities;

8. NA – basic assets and intangible assets of bank;

9. VB – amount of nominal accounts in balance sheet of a particular bank.

Statistical significance of the data under study is confirmed by the diagram as presented in Figure 3, which reflects statistically significant twofold correlations between certain factors. The corresponding matrix of coupled correlation coefficients is presented in Table 1.

In the model, rather than using absolute values of said parameters, employed are parameter ratios with respect to the amount of nominal accounts in the balance sheet of a particular bank. Such an approach characterizes both the percentage of various types of banking operations and the absolute value of respective values as independent of bank size. (Relatively larger banks can potentially reduce risk by diversifying across product lines and geographic regions (Gilbert, Meyer and Vaughan, 2002)).

Statistical characterization of variables as selected for subsequent analysis (as per 01.01.2003) is presented in Table 2.

The normalization for average values of indices for the totality of banks is captured in Figure 4 which depicts relative average values for each of the sub-samples of banks. Noteworthy is the symmetry demonstrated in the distribution of bank groups selected for index LN(VB). This is indicative of the fact that all Ukrainian banks, irrespective of their distinctiveness in terms of scope of assets (Ukraine’s 10 largest banks possess 54.1% of all the nation’s banking assets as per 01.01.2003), suffer from certain common difficulties in attaining consistently stable financial results.


From the diagram in Figure 4, it is discernible that the most prominent differentiations between pairs of average values of indices for selected groups of banks are evident for the following rations: percentage of securities in amount of bank assets (VP/VB), reserves for bank active operations (RA/VB), and debtors’ liabilities (DZ/VB). Furthermore, as can be observed in Figure 4, the respective indices are much larger for banks belonging to the group where a positive trend in current year financial result is not detected.

Economic interpretation of this fact follows quite readily. Too high a percentage of securities in the structure of banking assets (VP/VB) for banks with independent variable y=0 is, for one, a reflection of lack of efficiency and under-development ascribed to the Ukrainian stock market, whereas the underestimation of such a scenario by banks within this group is relegated to that of a secondary factor.

The dissimilarity of percentages for the reserves for bank activities operations index readily signals a high risk level of bank credit policy as waged by respective groups of banks, which, nevertheless, in the final analysis does not affect the stability of obtained financial results. In other words, scrutiny of this index confirms the thesis that risky financial policy is not favorable for financial stabilization under a transitional economy regime. Furthermore, as is evident from the ratio DZ/DB, the limiting factor of growing financial stability in Ukraine remains the non-payments between particular business entities. A substantial difference between values of debtors’ liabilities percentage in structure of assets for groups of banks is also indicative of the mounting impact of non-income assets on the results of banking activities.


Following from the above analysis is the basic construction of a statistically significant model for disaggregation of the banking population into smaller, cohesive groupings. Selected as basic parameters capable of explaining this disaggregation of banks into prescribed, in effect, clusters (y=1 and y=o) are the following:

VP/VB – percentage of securities portfolio in bank assets structure;

RANB – amount of reserves for bank active operations;

DZ/VB – amount of bank debtors’ liabilities.

With respect to the assessment function of the distribution for random value F(…), preference was assigned to a linear combination of selected parameters which had generated the most statistically significant results among the multiplicity of functions analyzed. The model to be analyzed was bolstered and supplemented in the process of study in terms of the identification of independent variables that could serve to explain the partitioning of banks into prescribed clusters.

Results from an assessment employing the Levenberg-Marquardt method (a popular alternative to the Gauss-Newton method) yielded an econometric model with a binomial dependent variable as characterized in Table 3. The statistical significance of the obtained model is confirmed by respective t-statistics and plevel values.

The obtained signs in the regression equation are in full conformity with their economic interpretation. This is wholly true for debtors’ liabilities as well, which during certain periods of time was utilized as a bona fide financial resource in Ukraine.

The adequacy of the model is further confirmed by the spatial location of observed and predicted values as depicted in Figure 5.

Though some of the data points in Figure 5 are obscured by overlap due to scale limitations of the diagram, readily discernible is the convergence of respective banks into distinctive groupings. Thus, the obtained econometric model using a binomial dependent variable may be deemed as credibly adequate to explain the delineation of banks into respective clusters.

Regulatory decisions are routinely executed on the basis of performance. Hence, it is imperative that an entity’s performance is judged within the context of its apposite peer group. This, accordingly, necessitates careful specification of both the salient variable(s) and the variable’s numerical ranges, in order to appropriately demarcate discrete peer groups within a given population. Managerial decisionmaking with regard to a given financial institution essentially necessitates the establishment of a construct conducive to fair comparative analyses, which peer-based modeling affords to a large degree.

Corporate and regulatory decisions are routinely executed on the basis of institutional (or divisions thereof) performance. Hence, it is imperative that an entity’s performance is judged within the context if its apposite peer group. Managerial decision-making with regard to a given financial institution essentially necessitates the establishment of a construct conducive to fair comparative analyses, which peer-based modeling affords to a large degree.


This paper employs probabilistic statistical assessments and models in the analysis of factors affecting financial results of activities of Ukrainian commercial banks. The methodology incorporates the application of an econometric model with a binomial dependent variable. Results of the implementation of said approach include an outwardly appropriate, preliminary division of banks into distinctive clusters. In the process of study, revealed are the most significant factors affecting both the division of banks into clusters and the qualitative characteristics of financial results within clusters under study. Such factors are: percentage of credit portfolio and securities portfolio in the structure of securities; percentage of reserves for active operations; percentage of authorized stock in the structure of nominal accounts of balance sheets at particular banks; and percentage of debtors’ liabilities as a component of non-income bank assets.

The implemented approach affords not only a better understanding of operational mechanisms inherent in the Ukrainian banking system, but it has the potential to function as a principal instrument to develop banking control procedures both within particular banks as well as at the system level.

This study demonstrates and reinforces the efficacy of statistical procedures in the classification of entities. Clusters of banks in Ukraine may be regarded are peer groups, in one sense polarized based on their financial characteristics. Notwithstanding institutional volatilities, the financial analyst derives merit from segregating entities into groups, accomplished in this study by employing cluster analysis. Hence, the peer group becomes an important benchmarking tool for the financial professional or analysts in Ukraine.




Para- 1 2 3 4


values KR/VB VP/VB D1/VB D2/VB

1 1.00

2 -0.41 1.00

3 0.15 -0.14 1.00

4 0.10 -0.04 0.15 1.00

5 0.22 0.19 -0.20 -0.14

6 -0.24 0.14 -0.57 -0.52

7 -0.36 0.21 -0.16 -0.10

8 -0.45 0.05 -0.06 0.12

9 0.16 -0.06 0.40 0.41

Para- 5 6 7 8 9


values RA/VB CB/VB DZ/VB NA/VB LN(VB) **





5 1.00

6 0.17 1.00

7 0.41 0.12 1.00

8 0.08 0.06 0.32 1.00

9 -0.20 -0.75 -0.21 -0.12 1.00

** logarithmic scale chosen to minimize variation range of VB



Statistics KR/ VP/ D1/ D2/ RA/



average 0.743 0.023 0.235 0.194 0.040

variance 0.031 0.007 0.025 0.020 0.004

maximum 1.121 0.503 0.934 0.641 0.366

asymmetry -0.901 2.789 0.709 0.292 2.366

excess 1.651 9.708 1.196 -0.767 7.064


average 0.734 0.018 0.239 0.191 0.035

variance 0.024 0.004 0.028 0.021 0.002


average 0.748 0.037 0.234 0.197 0.053

variance 0.041 0.009 0.019 0.018 0.006


Statistics CB/ DZ/VB NA/VB LN(VB)



average 0.180 0.009 0.057 11.858

variance 0.041 0.004 0.007 1.661

maximum 0.998 0.358 0.632 15.639

asymmetry 1.482 3.303 2.616 0.605

excess 2.008 11.506 11.672 0.314


average 0.184 0.007 0.057 11.908

variance 0.037 0.004 0.008 1.699


average 0.177 0.014 0.056 11.835

variance 0.046 0.003 0.005 1.569


Dependent Variable: y

Level of confidence: 95.0% (alpha=0.050)

Variable Coefficient Std. Error t-value

KR 0.88733 0.089048 9.96459

RA -2.74578 0.739635 -3.71234

CB 0.23735 0.182829 1.29823

DZ 1.39808 0.696008 2.00872

Variable p-level Lo. Conf Up. Conf

KR 0.000000 0.71141 1.06325

RA 0.000287 -4.20699 -1.28456

CB 0.196161 -0.12384 0.59855

DZ 0.046326 0.02306 2.77311


Cole R.A., and J.W. Gunther. “Predicting Bank Failures: A Comparison Of On- And Off-Site Monitoring Systems”, Journal of Financial Services Research, 13:2, 1998.

Deppler, Michael. “Beyond Integration; Squaring Europe’s Social Preferences with Robust Growth”, Finance and Development, June 2004, pp. 8-13.

Garshin V.V., Macroeconomic Factors as Ingredients of Stable Development of a Bank, Russian Economic School, BSP#2003/060R, NES, Nakhimovsky Prospekt, 47, Suite 1721, 117418, Moscow Russian Federation. Tel: (7-095) 129-3911, Fax: (7-095) 129-3722. Ukrainian translation: [TEXT NOT REPRODUCIBLE IN ASCII.], BSP#2003/060R.

Gilbert, Alton R.; Meyer, Andrew P.; and Vaughan, Mark D., “Could CAMELS Downgrade Model Improve Off-Site Surveillance?”, Review, Federal Reserve of St. Louis, Volume 84, Number 1, January/February 2002, pp. 47-63.

Kolary J., D. Glennon, H. Shin, M. Caputo. “Predicting large US Commercial Bank Failures”, Journal of Economics and Business, No 54, 2002, pp. 361-387.

Lavelle, Peter, Introducing UT’s Russia CAMEL Index, Untimely Thoughts, January 19, 2004, A Homepage devoted to Understanding Russia,

Matovnikov M. Yu. Functioning of the Banking System under the Conditions of Microeconomic Instability–Proceedings of IEPP, No. 23, 2000, Ukrainian translation: [TEXT NOT REPRODUCIBLE IN ASCII.], No23, 2000.,

Putnam, Barron H., “Early Warning Systems and Financial Analysis in Bank Monitoring: Concepts of Financial Monitoring,” Federal Reserve Bank of Atlanta Economic Review, November 1983, pp. 6-13, cited by Gilbert, Alton R.; Meyer, Andrew P.; and Vaughan, Mark D., “Could CAMELS Downgrade Model Improve Off-Site Surveillance?”, Review, Federal Reserve of St. Louis, Volume 84, Number 1, January/February 2002, pp. 47-63.

Sahajwala R., Berg van den P. Supervisory Risk Assessment and Early Warning Systems.–Basel Committee on Banking Supervision. Working Papers.–No4, 2000.

Scannell, Nancy J.; Safdari, Cyrus; and Newton, Judy. “An Extended Application of Factor Analysis in Establishing Peer Groups Among Banks In Armenia”, Journal of the Academy of Business and Economics (JABE).–Volume 1, Number 1, 2003.–PP. 129-136.

Stone M. and Rasp J., “Tradeoffs In The Choice Between Logit And OLS for Accounting Choice Studies”.–The Accounting Review, 1991, pp. 170-187.

Vasyurenko, Oleg Vladimirocich and Azarenkova, Galina Michailovna. “Managing Bank Liquidity from the Point of View of Changing the Speed of its Financial Flows”, Banking Business.–2003–No. 1,-C. Pp. 60-63. Ukrainian translation: [TEXT NOT REPRODUCIBLE IN ASCII.].–2003.–No1.–C. 60-63.


The authors are grateful for the excellent language translation assistance provided by Ms. Ms. Larissa Shelegova, Senior Lecturer of English at the Kharkov Banking Institute of the Ukrainian Academy of Banking of the National Bank of Ukraine.

Author Profiles:

Dr. Vasyurenko Oleg Vladimirovich earned his Doctorate in Economics at the Ukrainian Academy of Banking in 1999 and is Vice Rector of the Ukrainian Academy of Banking of the National Bank of Ukraine and Director of the Kharkov Banking Institute. He is credited with scores of publications on the state and development of Ukraine’s banking system.

Dr. Azarenkova Galyna Michailovna earned her Candidate of Economic Sciences at the Kharkov National Economic University in 1997 and serves as the Deputy Director for Science and International Relations and Associate Professor of Finance at the Kharkov Banking Institute of the Ukrainian Academy of Banking of the National Bank of Ukraine.

Dr. Nancy J. Scannell holds a B.A. from Michigan State University and Ph.D. from the University of Illinois at Chicago in Economics. She teaches Finance as an Associate Professor at the University of Illinois at Springfield and served as a Fulbright Specialist at the Ukrainian Academy of Banking.

COPYRIGHT 2004 International Academy of Business and Economics

COPYRIGHT 2005 Gale Group