A model proposal for estimating banks’ future value: Evidence from Turkey

Investors make solid decisions when evaluating their investments based on positive indicators the firm may show in the future, rather than based on its past performance. Accordingly, this study aims to investigate the relationship between performance criteria and the most significant value-based criterion; Economic Value Added (EVA). Further, it evaluates the impact of future EVA values on the bank value. Panel Data Analysis and the OLS Regression model are used to estimate the regression equation. The analysis is performed using data of 10 banks on the BIST Banks Index over the period 2011 to 2020. Furthermore, the EVA criterion was converted into standardized EVA(SEVA) by dividing EVA by total assets. The OLS regression analysis results revealed that the model’s explanatory power for the SEVA variable is 71.92%. The three variables that have positive correlation with SEVA are earnings per share (EPS) and TOBINQ rates at the 1% significance level and the price to sales growth rate with a degree of significance at 10%. Regarding the Panel Data Analysis results, while the explanatory power of the SEVA variable is 72.14%, its association with the EPS and TOBINQ criteria was found to be significant at the 1% significance level. The empirical investigations reveal that the model developed using the future SEVA as a proxy for bank value is found to be promising, and it is accepted that the SEVA variable can be used instead of the bank value.


INTRODUCTION
Nowadays, investors mainly focus on the real value of an asset, which is identified by comparing the firm value with its market value. Defining the real value and how it should be quantified realistically is considered a tricky problem, since a firm's value will vary greatly depending on the firm's state, its competitive position, the experts who will conduct the valuation, the purposes of the valuation and the valuation techniques. Prior studies that have noted the firm value are mostly based on the relationship between the firm value and performance (operating, financial, etc.). According to the assumption underlying these studies, the value of a firm increases with an improvement in its activities. The firm value is closely related to the firm administration, capital structure, mergers, and a country's legal system. As a result, each factor affecting the firm's cash flows and cost of capital will have an impact on firm value although to different degrees. Hence, predetermining the effects of these factors will contribute to the realistic calculation of the firm value.
Rather than relying on the firm's past outcomes, investors evaluate their investments based on positive performance signs the firm might show in the future. According to basic finance theory, a firm's real value is defined as the present value of the cash flows that the firm will generate from its investments, discounted at an appropriate rate. Therefore, the value of the firm mainly depends on its future investment cash flows and the risk level the firm will undertake to generate cash flows.
Banks that act as intermediaries by connecting the savers with a surplus of funds and those in need of funds in the financial markets are the main actors of the financial system. Due to their indispensable role in the financial systems, banks are constantly kept under government supervision. The objective of banks, like that of other commercial institutions in the economy, is to generate profits. However, macroeconomic factors posing risk to the financial system tend to have a considerably more rapid effect on banks, through their significant role in the economy. In addition to macroeconomic factors, such as monetary and fiscal policies, interest rates, exchange rates, and inflation rates, microeconomic factors originating from the bank balance sheet structure also affect the banking sector. The enormous impact of these threats on the banking industry may lead to a financial crisis in countries.
Aside from their key functions, Turkish banks are also aiming to increase their profits by diversifying into a variety of profitable sectors and financial instruments in a mixed structure. Banks need to perform well, since competition among banks in Turkey and throughout the world has become extremely intense. Furthermore, like other commercial enterprises, banks must continually enhance their value in order to increase the shareholders' wealth. Thus, they will be capable of serving both the interests of their shareholders and their customers.
EVA (Economic Value Added), one of the value-based criteria in creating value, is used to represent the value of banks regarding their importance in the financial system In reviewing the literature, the EVA value was used as an absolute value, but it was not taken into account on a firm basis. In contrast to earlier studies, in this paper, the EVA variable refers to standardized (SEVA), which is calculated by dividing EVA by total assets and leading to the development of a new model proposal. To investigate the criteria that best explain the future EVA measure as a proxy for bank value, the following hypothesis has been tested:

LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
H 1 : Performance criteria have a significant impact on EVA.

Data set and model development
During the model development process, performance criteria were developed using the data from ten banks on BIST Banks Index between 2011-2020, and all collected bank-specific data turned into an acceptable format for analysis. The banks used in the analysis are presented in Table  1. This study aims to investigate the relationship between EVA and performance criteria that were used as proxies for bank values. In addition, the ability of the one-year lagged performance criterion to explain the EVA value was analyzed by using an econometric model. Based on the econometric model developed in this study, the dependent variable is the standardized EVA value with the total assets ratio at time t, and the independent variables are the values of the eight performance criteria in Table 1 at time t -1. A total of nine criteria were used to develop an econometric model in which the power of the criteria to explain the future EVA value as a dependent variable was investigated. Panel data was used for the application of analysis to develop econometric models by STATA 16.0 and E-VIEWS 9.0 package programs.

Results
From the results of the descriptive statistics given in Table 2 Table 2, the variable with the highest mean and standard deviation is MB, while the lowest variable is SEVA. From the gap of maximum and minimum values of MB, it is possible to understand the existence of great disparity between the banks in the market to book ratio.
The results of correlation coefficients are given in Table 4 in order to examine the structure of the relationship and provide an overview of the correlation between the variables in the research model.
The findings part of the study begins with the OLS regression analysis and in the next stage, continues with the results of panel data analysis. To perform OLS regression and panel data analysis, some assumptions need to be tested. These assumptions are respectively: Inter-Unit Correlation (cross-section dependency), heteroscedasticity, and Autocorrelation. Making predictions about the variables by ignoring these assumptions will cause the t values of the variables to lose their validity (Tatoğlu, 2016, p. 8). It is impossible to dis-  cuss a valid statistical model without testing these assumptions before making a regression analysis. To test the assumptions, firstly the correct regression model must be identified. Furthermore, to make sure whether the model in this study is a random effect or fixed effect, the Hausman test was applied and the Random Effect Model was accepted as the right model at 0.05 significance level as presented in Table 5.  Table 5.
Regarding the existence of autocorrelation in the developed model, the fact that the Durbin-Watson and Baltagi-Wu LBI values are close to the value 2, suggests that there is no autocorrelation in the first order (Wooldridge, 2002). The developed models do not appear to have an autocorrelation problem. At 0.05 level of significance, it was proved that there is heteroscedasticity and cross-section problem based on the results presented in Table 5. These problems were resolved with robust estimators, the results of which are reported in Tables 6 and 7. Furthermore, the variables must be stationary in panel data analysis. To test the stationarity of the variables, the second generation unit root test was applied using the Pesaran (2007) Following the completion of all required diagnostic tests, the variables for the analysis were established using the data from BIST Banks Index for the years 2011-2020 and the analysis model was designed. Table 6 presents the results of the OLS regression analysis of the model developed to investigate future bank values.  Note: *, ** and *** indicate significance levels at 0.10, 0.05 and 0.01.
Regarding the results of panel OLS regression analysis, the power of the model to explain the changes in the SEVA value was R 2 = 71.92%. The variables in the developed model can explain about 72% of changes in the SEVA value, while the remaining 28% is explained by variables outside the model. From the results of Table 6, the SEVA variable was found to be associated with EPS and TOBINQ at 1% and with PSGROWTH at a 10% level of significance. On the other hand, no significant relationship was found between other criteria in the model and SEVA values. EPS, TOBINQ, and PSGROWTH criteria are significantly and positively related to the SEVA value. In other words, with an increase in these criteria, the future SEVA values of banks also increase. This result reveals that the one-year delayed SEVA value can be significantly and positively affected by some criteria.
In general, based on the results of the Panel OLS Regression, it is proved that the model established is successful and effective in explaining the SEVA criterion chosen to represent the bank value. In addition, for interest groups, especially for shareholders, the model developed suggests that they should focus on the EPS, PSGROWTH, and TOBINQ criteria. For the panel model, Hausman Test was applied and, as mentioned before, the analysis was applied with Random Effects Model according to the Hausman Test result. It is assumed that all the problems related to diagnostic tests assumptions shown in Table 5 are resolved with the Arellano, Froot, and Rogers Resistive Estimator. The results of the panel data analysis are presented in Table 7. In general, based on the findings of the analyses, it has been proved that the value of the SEVA criterion at time t, which is one of the leading value-based criteria and was chosen to represent the bank value, is related to the values of the performance measures at time t -1, hence the explanatory power of the model established is quite high.
In this study, it is proven that the claim of Stern & Stewart consultancy company that "EVA is the best criterion related to firm values, among the value-based criteria" is valid for BIST Banks.

DISCUSSION
This study was conducted using data from ten banks on the BIST Banks Index covering the 10-year period from 2011 to 2020 with the objective of developing econometric models based on the SEVA criterion in order to reveal a model that best explains the bank's value and contribute to the current literature by adding a new perspective. SEVA is considered as one of the criteria that best represents the firm's value to shareholders and interest groups.

CONCLUSION
This paper provides a test of the relationship between the performance criteria and the EVA, which is the most often used value-based criterion. It also indicates the contribution of one-year lagged EVA values to firm value, that is, the bank value. As a result, unlike traditional firm valuation criteria or approaches, it was meant to present a new perspective on firm value using the EVA criterion.
In this regard, since investors expect a higher return than the risk they take by their investment, every strategic choice made by a firm's management in response to shareholder demands should be shaped to increase the firm's value. Also, firms have begun to focus on the criteria that will add value to the firm, rather than using the accounting profit created based on the historical cost in order to grow rapidly and maximize firm value. Economic Value Added (EVA) criterion, which emerged first among the value-based performance measures, is the most commonly used criterion to accurately estimate the firm value and to develop the perception of value creation in firms. EVA is a performance measure that can provide more accurate results than other measures in determining the real economic profit of firms. EVA is also the clearest criterion to indicate whether the firm is creating value for shareholders or not and provides significant advantages over traditional accounting criteria since it takes into account both the direct cost of debt and the cost of equity, which is an indirect cost item. This distinction, along with being critical for the firm's partners, demonstrates the amount of the risk related to the partner's investment.
The analyses revealed that the econometric model developed with EVA as an indicator for bank value was quite successful. Accordingly, it has been accepted that the EVA criterion can be used to substitute for bank values. In fact, this study presents a model related to the values created by banks to interest groups such as investors, managers, and shareholders. As a result, it was revealed that the model applied can explain changes in the values that banks will generate in future at a significant level and suc-cessfully estimate the banks' future values. In other words, the present study has achieved its goal by proving that the developed model can be applied by anyone who intends to use value-based criteria and that value-based criteria are highly effective in creating value. Furthermore, it has been indicated that value-based criteria can be used in value determination studies efficiently. This study is expected to add a new perspective to the academic literature in this regard. Therefore, further research is suggested with more focus on developing several models with data, including more performance criteria and longer analysis periods, and subsequent comparative analyzes of these models.