Türk Bankacılık Sektöründe Ürün Çeşitlendirmesinin Karlığa Olan Etkisi

Abstract

Türk Bankacılık sektöründe son 10 yılda gerçekleşen önemli yasal düzenlemeler ve yoğunlaşan sektörün bir sonucu olarak, bankalar karlılıklarını koruyabilmek için yeni stratejiler uygulamakta. Bu noktada, dünya bankacılık literatüründe de son yıllarda sık işlenen bir konu olan ürün çeşitlendirme stratejisi gündeme gelmektedir. Türkiye bankacılık sektörü incelendiğinde, son 5 yılda vergi sonrası özkaynak karlılığı gibi temel performans göstergelerinde önemli düşüşler görülmektedir. Bankaların karlılıklarının bu düşüşten etkilenmemesi için odaklanma ve çeşitlenme gibi stratejileri kullandıkları görülmektedir. Bu bilgiler ışığında, bu çalışmada çeşitlenme stratejisinin Türkiye’deki bankaların performanslarına olan etkisi incelenerek, performans artırıcı bir strateji olarak Türk Bankacılık Sektöründe kullanılıp kullanılamayacağının belirlenmesi amaçlanmaktadır. Yapılan bu çalışmada Türk bankacılık sektöründe özellikle son yıllarda azalan aktif ve özkaynak karlılıkları için ürün çeşitlendirmesinin bir çözüm olup olamayacağı değerlendirilmiştir. Çalışma 2007-2012 yılları arasında Türkiye’de faaliyet gösteren 41 bankanın 3 aylık dönemler itibariyle bilanço ve gelir tablosu verileri kullanılarak gerçekleştirilmektedir. Ürün detayındaki veriler için bankalar birliği verileri esas alınmıştır. Daha önceki çalışmalar göz önünde bulundurularak, tez çalışmasının hipotezi; bankanın çeşitlenme miktarı ile karlılığı arasındaki pozitif ilişki bulunmaktadır, şeklinde belirlenmiştir. Bu çalışmada, Türkiye bankacılık pazarında sıklıkla kullanılan ve banka performansını arttıran üst düzey yönetim stratejisi olduğu düşünülen çeşitlenmenin nicel ölçümleri yapılmıştır. Literaturde çokca adı geçen ve kapsamlı bir ölçüm yöntemi olan entropi ve Herfindahl endeksleri olmak uzere iki farklı ürün çeşitlenme ölçümü yapılmışır. Daha sonra çeşitlenme ölçümleri ve bazı banka özelliklerinin karlılığa olan etkisi panel veri regresyon analizi yöntemi kullanılarak incelenmiştir. Bu araştırma kapsamında karlılığı en fazla Sermaye Yeterliliği ve Şube Başına Toplam Aktif etkilemiştir. Sermaye yeterliliği için pozitif etki, Şube başına toplam aktif için ise negati etki söz konusudur. Aktif kalitesi ile performans arasında anlamlı ilişki gözlemlenmemiştir. Herfindahl endeksiyle bulunan çeşitlenme ROA ve RAPROA’yı negatif yönde etkilerken, ROE ve RAPROE için anlamlı etki gözlemlenmemiştir. Tam ters bir etki ile Entropi ile ölçümlenen çeşitlenme ROA ve RAPROA’yı pozitif yönde etkilerken, ROE ve için anlamlı etki gözlemlenmemiştir. Her iki endeks ile de hesaplanan çeşitlenmeler için, banka özelliklerinin etkileri benzer olarak bulunmuştur.

In the last decade, because of regulations and increasing concentration in Turkish banking sector, banks began to look for new strategies in order to maintain their profits. At this point, product diversification that is popular topic in the banking literature comes into question. In the last 5-year period of Turkish Banks, a sharp fall in main performance indicators like return of equity of is continued. Banks are using focus and diversification strategies in order not to be effected from the decrease in profits. Because of this information, effects of diversification strategy on Turkish Banks performance are needed to be analyzed. This study aims to answer if product diversification strategy is a solution for the decreasing trend of the return of assets and return of equities in Turkish banking sector. The study seeks to examine the effect of product diversification on bank profitability ratios by using a quarterly panel data of 41 banks that are active in the Turkish banking sector over the period 2007-2012. Data is derived by using financial statements of banks. By using active banks during the period 2007 and 2012, the research data became balanced. Total asset size of the 41 banks used in statistical analysis is adequate to 95% of total assets in the Turkish banking sector. Research data includes banks with different diversification degrees and asset sizes. As a result of different asset sizes and diversification degrees variables are not normally distributed. Despite this situation data is used without making any difference because, it has 861 observations and can be called a large data, which is expected to include different populations as a result of its size. In this study, data didn’t divided to smaller samples to see the big picture of the population. In research model, relation between diversification and performance is analyzed by using diversification degree and control variables to explain change in performance. Thus, performance is not analyzed directly by using economies of scale and scope. Performance is defined by using four different independent variables: return on equity (ROE), return on assets (ROA), risk adjusted return on equity (RAPROE) and risk adjusted return on assets (RAPROA).RAPROA and RAPROE is used for eliminating seasonal effects of banks financial data. As a diversification strategy, product diversification strategy is taken into account. Product diversification is a commonly used top management strategy in Turkish banking sector. It is believed to increase bank performance and quantitatively analyzed in the study. The study measures the degree of product diversification in banks through the Herfindahl-Hirschman Index (HHI) and Entropy Index calculated on the product lines offered by banks on the market. In Herfindahl-Hirschman Index, if a firm operates in a single classified group, HHI index of diversification is zero and it becomes close to one if the firm’s total sales are divided equally among any number of classified groups. Entropy Index measure is designed to decompose the total diversification measure into managerially meaningful elements of total diversification: unrelated and related diversification, international (related and unrelated) market diversification. The HHI diversification index cannot be decomposed as directly as entropy measure in additive elements that define the contribution of diversification at each level of classified group aggregation to the total. Like HHI index of diversification, the entropy index of total diversification also yields a score of zero for single classified group firms and becomes greater with increasing levels of diversification. Product sales volumes are used for calculation of in indexes. After calculating both of the indexes that are widely used in the literature, panel data regressions analyze applied for determining the effect of diversification and control variables. By taking into consideration studies in the literature, hypothesis of the study is; “Banks’ level of product diversification is positively correlated with the banks profitability”. At the beginning of the analysis, five control variables are included in the regression: asset quality (AK), capital adequacy ratio (SY), total assets per branch (SA), bank size (BB) and gross domestic product (GDP). Asset quality, capital adequacy ratio, total assets per branch and bank size is bank specific endogenous variables. For considering the effects of macroeconomic changes, GDP is used as a control variable. By using correlation matrix it is determined that bank size is highly correlated with other control variables. Because of this finding, we make two regressions with including banks size and excluding bank size for examining the effect of bank size. Two panel data regression analysis show that bank size can be excluded from analysis for better results. During these, two test panel data regressions it is clearly seen that GPD has no significant effect on performance and this variable has a distribution that may affect analysis. It is found that bank specific endogenous variables are more effective on banks performance than macroeconomic variables. Thus, banks size and gross domestic product variables are excluded from the rest of the analysis. Asset quality is rate of total loans on total assets of the bank. Loans are main profit source of banks so; this variable is expected to have positive relation with performance. Capital adequacy ratio (SY) is equities over assets ratio. Increase in capital adequacy ratio makes bank becomes less volatile for shocks as a result if capital adequacy ratio increases performance is expected to increase too. Total assets per branch are indicator of operational excellence. When total assets per branch increases, it means customers either are getting bank services in channels than branches like mobile banking or call center. As total asset per branch increases operational cost decreases, so performance is expected to be effected positively. For different combinations of four independed variables (ROA, ROE RAPROA, RAPROE) and two diversification measures (HHI, DT) eight models used in panel data regression. In the first step of the analysis, whether the series are stationary or not was determined by applying unit root test to each variable. A great number of panel unit root tests were developed in order for searching the stationary between panel series. In the study, of the root tests, Levin, Lin, and Chu, Im, Pesaran, and Shin , Generalized Dickey Fuller (ADF) tests were used. While null hypothesis about the existence of common unit root is tested in Levin, Lin, and Chu test, null hypothesis about the existence of individual unit root is tested in Im, Peseran, and Chin test. In addition to this, the existence of panel unit root is tested by ADF test in the series. EViews 7.0 econometric package is used in the analysis. When the panel unit root test results are examined, it is generally seen that unit root does not exist in series. Since p values are lower than critical value 0.05, the H0 hypothesis that states the series includes unit root has been declined. Secondly, to select to most appropriate effect on regression, Breusch-Pagan LM test applied to all models for deciding between pooled OLS and Random effect. Results of the Breusch-Pagan Lm test show probabilities fewer than 5% significance levels thus Random effects preferred to pooled OLS. There are two methods assuming that the fixed may change in accordance with sections. These methods are Fixed Effects Method and Random Effects Methods. Which of these two methods “fixed effect” (the prediction of a different fixed coefficient for each unit taking place in the panel) or “random effect” (obtaining the prediction of a different fixed coefficient for each unit in the panel randomly) is going to be valid is determined by Hausman Test. Results of the Hausman test show probabilities fewer than 5% significance levels thus fixed effects preferred to random effects. Further diagnostics on data is made by Breush-Godfrey LM test to examine autocorrelation. In regression analysis using time series data, autocorrelation of the errors is a problem. Autocorrelation of the errors, which themselves are unobserved, can generally be detected because it produces autocorrelation in the observable residuals. Results of the Breush-Godfrey LM test show that probabilities fewer than 5% significance levels thus autocorrelaions exist in the all models. Moreover, White test is applied for heteroskedasticity. In statistics, the White test is a statistical test that establishes whether the residual variance of a variable in a regression model is constant: that is for homoscedasticity. The results of the all models for white test indicate heteroskedasticity due to probabilities fewer than 5% significance levels. The existence of heteroskedastic in this model means that misestimating of the variances by parameter predictors and this leads the interval estimation, t and F tests (that are going to be done) to be incorrect. Autocorrelation and hereroskesasticity problem in the model is solved by correcting the standard errors by White’s cross-section coefficient covariance method. By means of this method, not only different error variances in each cross-section but also the correlation problem between cross-sections are to be solved. Within this study, the results show that, capital adequacy ratio and assets per branch office are mainly correlated with the performance. Capital adequacy is positively and an asset per branch office is negatively correlated with performance. However, there is no significant correlation between asset qualities, Product diversification calculated by Herfindahl Hirshman index is found negatively correlated with ROA and RAPROA, but not significantly correlated with ROE and RAPROE. In direct contradiction, product diversification calculated by entropy index is positively correlated with ROA and RAPROA, but not significantly correlated with ROE and RAPROE. For both calculation of product diversification, bank specific effects on profits are same.