Determinants of Affecting Level from Systematic Risk : Evidence from BIST 100 Companies in Turkey

The main purpose of this paper is to examine the impact of accounting variables on systematic risk of firms. By using data of 58 companies from BIST-100 Index for the period between 2006 and 2015, panel data analysis is employed. The results of the study indicate a statistically significant and positive effect of asset size, asset turnover, previous term equity to total debt and previous term cash ratio on systematic risk. On the other hand, negative influence of profitability, equity to total debt, cash ratio and previous term debt to total assets on systematic risk is detected. Also the study determines that consumer price index, previous term beta and previous term GDP per capita affects the systematic risk negatively and increase the explanatory power of the model significantly.


Introduction
Modern finance describes the aim of firms as the maximization of shareholders' wealth.To achieve this aim, firms desire to maximize the sum of their securities' market value and current cash flow (Auerbach, 1979).However, only raising return is not enough to increase value of securities and so value of firm.Investors also deal with the risk of the securities (Choudhary & Choudhary, 2010).There is a positive relationship between risk and required rate of return (Gallagher & Andrew, 2002).Net present value of securities is calculated by discounting cash flows with a discount rate.Discount rate is shaped by risk which refers to the likelihood to receive a return on an investment that is different from excepted (Damodaran, 2012).Increasing risk results to higher discount rates (expected returns), so for securities with equal cash flows, higher risk results to lower value.For this reason, investors want to recognize risk of the securities to determine the expected return.
The risk of the securities can be classified as systematic and unsystematic risk.Unsystematic risk which arises from the firm itself and can be controlled by the firm consists of components as operational risk, management risk and financial risk.Unsystematic risk can be eliminated by portfolio diversification (Ercan & Ban, 2005).On the other hand, systematic risk is due to non-firm reasons such as interest, inflation, economic stagnation and political events.For this reason it is not possible to remove systematic risk (Karan, 2004).Although systematic risk is derived from non-firm reasons, internal factors of the firms can influence the affecting level from systematic risk.In other words, firm-specific factors determine the extent that firms are affected by systematic risk.
The capital asset pricing model (CAPM) of Sharpe (1964), Linther (1965) and Mossin (1966) explains the expected return of assets according to exposed systematic risk for a well-diversified portfolio.According to model, excepted return of an asset can be formulized as: : the expected return of the asset : the risk-free rate of interest (the beta): the sensitivity of the expected excess asset returns to the expected excess market returns : the expected return of the market : the market premium (the difference between the expected market rate of return and the risk-free rate of return) Expected return of the asset consists of two components: time value of money and risk premium.The time value of money is represented by the risk-free rate (R f ) in the formula and the risk-free rate is customarily the yield on government bonds.The other half of the CAPM formula represents risk and calculates the amount of compensation the investor needs for taking on additional risk.This is calculated by taking a risk measure (beta) that compares the returns of the asset to the market over a period of time and to the market premium (E(R m ) -R f ).Because a rational investor eliminates the unsystematic risk by portfolio diversification, the beta (β i ) reflects only the sensitiveness of systematic risk faced by asset.
For the investors, to determine the factors affecting risk is crucial for evaluating risk and return relation.So determinants of the beta (βi) as an indicator of systematic risk are of prime importance for investors.The aim of the study is to determine the effects of the accounting variables on affecting level from systematic risk of the assets.

Literature Review
Studies which investigate the effect of accounting variables on systematic risk carried out for different markets are in the literature.When the studies are examined, it is seen that mostly studies carried out for developing or least developed markets have become more intense in recent years.Authors, data set, methodology and conclusion of the studies which focus on this subject are shown in Table 1.

Multiple Regression Analysis
Empirical findings suggest that the degrees of operating and financial leverage explain a large portion of the variation in beta.Both operating and financial leverage have a positive effect on systematic risk.Empirical results of the study show that profitability ratio and activity ratio have a statistically significant and negative effect on systematic risk.However, effect of liquidity and leverage ratio on systematic risk is statistically insignificant.Martikainen (1991) 28 firms listed in the Helsinki Stock Exchange for the whole 1975-1986 period.

Regression, Factor and Transformation Analysis
In the study, the effect of profitability, financial leverage, operating leverage, and corporate growth, measured as growth in earnings and dividends on the systematic risk is investigated.The most important factor explaining stock returns is found to be highly related to the leverage of the firm Hamid Prakash and Anderson (1994) 651 large companies which their data included in Compustat data tape

Coefficients of Correlation
Empirical evidence reveals that the growth rate, measured in either net income or operating income, is positively related to the relative systematic risk of the firm.As a result of the study, sales growth, leverage, ratio of short-term debt in total debt, asset size and ratio of long-term debt in total debt affect the systematic risk (beta) positively.However there is negative effect of price earning ratio, sales size, the ratio of tangible fixed assets to permanent capital, the ratio of total debt to the shareholders' equity on systematic risk.

Panel Regression Analysis
According to the study, quick ratio and asset turnover referred as the measure of liquidity and operating efficiency respectively affect the systematic risk negatively.On the other hand, debt ratio and return on assets used to define leverage and profitability respectively have a positive effect on beta.

Multiple Regression Analysis
In the study, the effect of liquidity (current and quick ratio), leverage (debt to equity ratio and long-term debt to total asset), total asset turnover and asset growth rate on systematic risk is examined.The results of the study reveal that no one of the variables influences systematic risk statistically.Chen (2013)

Method of Moments
Analysis results suggest that size affects systematic risk positively, while asset turnover has a negative impact on systematic risk.In addition, effect of acidtest ratio, leverage ratio and return on assets on systematic risk is statistically insignificant.
In addition to accounting variables, effect of macroeconomic factors on systematic risk has been the subject of different studies.Robichek and Cohn (1974) detected that economic growth and inflation have the ability to explain systematic risk of firms.Also industrial production growth (Andersen et al., 2005) and interest rate (Kazi, 2008) are the other macroeconomic factors which their effect on systematic risk is studied.The evaluated factors are not limited with accounting or macroeconomic variables, for example by progressing further democratic politics is considered as an affecting factor as well (Bechtel, 2009).

Data Set
To determine the influence of accounting variables on affecting level from systematic risk, data of 58 companies listed on BIST-100 index is used for 10 term period between 2006 and 2015.Companies which are banking sector companies and sport sector companies are excluded from data set because of different financial tables' structure and different financial terms respectively.Also only companies with available data for all of the period are included to data set.
The dependent variable of the study is BETA as a measure of systematic risk.BETA is calculated by regression analysis.For each companies and each year, daily stock returns and daily market returns regressed according to following model: Y = βo + β1 X Where Y is daily average returns of company; X is daily average returns of market while coefficient β1 is estimated BETA on yearly bases.Returns of the companies and market are derived by following formula; Return = Ln (P t / P t-1 ) Where P t is the price of company or market at t day, P t-1 is the price at the day before t and Ln is the natural logarithm.BIST-100 index is selected to represent the market as generally accepted.
In the study, 27 accounting variables are determined as independent variables.Also one period lagged variables are created for each variable.The independent variables and acronym are shown in Table 2.
For panel data analysis like all-time series variables, the stationary of the series is crucial.When the variables in a regression are nonstationary, correct R-square values and t-statistics cannot be generated by the analysis (Greene, 2012).For preventing spurious relations between the variables, unit root tests are applied and nonstationary series are included the model with first or second differences which provides stationary series.In this study, Im, Pesaran and Shin and ADF panel unit root tests are used to test the stationarity of series.If result of the each one test reveals that the serial is not stationary at level, first or second difference is used according to stationarity.Tests results are shown in Table 3. Variables which are not stationary at level are added to model as their stationary first difference.
To confirm the determinants of beta, independent variables are selected by stepwise backward elimination and best suitable model is determined.Ten independent variables from the listed variables and their one period lagged form are chosen according to their statistical significance and the following model is established: In addition to Model 1, one period lagged BETA is added as an independent variable and Model 2 is formed.Also by adding GDP per capita and consumer price index (CPI) as control variables Model 3 is performed.Although the natural logarithm of GDP per capita is stationary at level, the natural logarithm of CPI is not stationary at level and its stationary first difference is used.

Panel Data Analysis
The data set of the study include time series and cross section together, so to analyze the influence of accounting variables on affecting level from systematic risk panel data analysis is used.Using panel data has several advantages over crosssectional and time series data sets.Panel data usually give the researcher a large number of data points which allow a researcher to analyze a number of economic questions that cannot be addressed using cross-sectional or time-series data sets.
To use panel data sets increases the degree of freedom and reducing the collinearity among explanatory variables and improves the efficiency of econometric estimates (Hsiao, 2003).
When panel data analysis is used, to select suitable panel estimation model is crucial.To make a selection between mainly three models (pooled ordinary least squares, random effects and fixed effects) is necessary.For all three models, the null hypothesis of F test is not rejected although the null hypothesis of Breusch-Pagan LM Test is rejected.So the best suitable estimation model for all three models is determined as random effects.Also the results of Hausman test support this choice.Panel data analysis for all three models is performed in the form of random effects.

Estimation Results and Discussion
In the study, panel data analysis is performed to determine the effects of accounting variables on systematic risk.Because of heteroscedasticity and autocorrelation in Model 1 and Model 2, the analysis is performed with robust standard errors generated by Arellano (1987), Froot (1989) andRogers (1993).Model 3 is analyzed with Huber (1967), Eicker (1967) and White (1980) heteroscedasticity-consistent standard errors to eliminate only heteroscedasticity problem.Panel data analysis results are represented in Table 6.Three models were generated to analyze the determinants of systematic risk.Only 10 accounting variables which are selected from 27 variables and their one period lagged form are involved in Model 1.One period lagged form of BETA is added to independent variables and Model 2 is created.Also by adding GDP, CPI and their one period lagged form as control variables Model 3 is constituted.
According to panel data analysis, all three models are statistically significant at 1% confidence level.Adjusted R-square which reveals the explanatory power of the model indicates that only 9.79% changes in BETA can be explained by independent accounting variables.When BETA(-1) is added to model, the explanatory power of the model increases significantly and reaches to 31.70%.Highest adjusted R-square (36.28%) is reached by adding macroeconomic control variables to the model.
The results of the models indicate that equity to total debt has a negative effect on systematic risk.This means that increasing leverage results to increase in systematic risk.This result is in compliance with the studies of Chen (2013), Alaghi (2013), Tanrıöven and Aksoy (2011), Eryiğit & Eryiğit (2009), McAlister et al. (2007), Lee and Jang (2007) and Tandelilin (1997).On the other hand one period lagged debt to total assets (DTA(-1)) has a statistically significant and negative effect on BETA.Likewise one period lagged equity to total debt (ETD(-1)) affects BETA positively.This means previous year's leverage affects current year beta negatively.
Although previous term leverage affects systematic risk negatively, current term leverage has a positive effect on systematic risk.
According to analysis results, cash ratio (CRS) as a measure of liquidity share a contradictory feature with leverage.Cash ratio has a negative effect on systematic risk as expected.In other words, increase cash ratio causes to decrease in systematic risk.The result is in accord with the studies of Alaghi (2013) and Iqbal and Shah (2012).On the other hand, previous term cash ratio (CSR(-1)) has a positive effect on BETA.Namely, if previous term cash ratio was high, an additive effect exists in the current term beta.
Another result of the study reveals that asset turnover (TAT) as a measure of operating efficiency affects the systematic risk positively like the study of Eryiğit and Eryiğit (2009).Increase in asset turnover results to rise in BETA.Similarly asset size (LnTA) of the firms affects systematic risk at same direction.Increasing total assets causes systematic risk to rise.This result is consistent with the studies of Hamid et al. (1994), Tandelilin (1997), Lee and Jang (2007), Tanrıöven and Aksoy (2011).
Profitability is another factor that can affect the systematic risk of the firms.The results of the study reveal that net profit margin (NPM) and operating profitability have a negative effect on BETA.If the profitability of the firms increases, the systematic risk of the firms decreases.A similar result is obtained for dividend payout ratio (DPR).Dividend payout ratio also affects systematic risk negatively.
Although results about the dividend payout ratio are consistent with the study of Adhikari (2015), profitability results contradict with this study.
In the study, also the effects of previous term BETA (BETA(-1)), GDP per capita as previous(GDP(-1)) and current term (GDP) and consumer price index as previous (CPI(-1)) and current term (CPI) are investigated.The previous term BETA has a negative effect on current term BETA.Also the previous term BETA raises the explanatory power of the model significantly.Similarly CPI and previous term GDP per capita affects systematic risk negatively.On the other hand, the effects of GDP per capita and previous term CPI cannot be explained statistically.

Conclusion
This study examines the influence of accounting variables on affecting level from systematic risk.Analysis results reveal that current term leverage affects BETA positively.Firms with high leverage face with a higher systematic risk.On the other hand, previous term leverage has an opposite influence on affecting level from systematic risk.Higher previous term leverage causes to decrease in systematic risk.Likewise, current and previous term cash ratios have a contradictory impact on systematic risk.Although cash ratio has a negative effect on systematic risk as expected, previous term cash ratio has a positive effect on systematic risk.In addition, positive influence of asset turnover and asset size on systematic risk is determined.Increasing firm size and operating efficiency causes to rise in systematic risk.However profitability and dividend payout ratio affects systematic risk negatively.Firms with more profit and making dividend payment take the edge off systematic risk.
In the study, in addition to accounting variables, effect of previous term systematic risk, GDP per capita and consumer price index is evaluated.The previous term BETA influences current term BETA negatively and has a high explanatory power for the model.Similarly the negative effect of CPI and previous term GDP on systematic risk is detected.
This study investigates the determinants of affecting level from systematic risk within the context of accounting variables and it expects to contribute related literature with this dimension.However the factors that can influence systematic risk are not limited with accounting or macroeconomic variables.Future studies can make further contrıbution to this field by determining the effect of nonfinancial variables like managerial and proprietary factors.

Table 1 (cont.) Literature Review
R&D/sales, asset size, firms' age and Competitive intensity measured with Herfindahl's four firm concentration ratio have a negative effect on systematic risk.On the other hand, asset growth rate and leverage affects systematic risk positively.

Table 3 . Unit Root Test Results
(Park, 2011)ate model for the data set can be determined by using F Test, Breusch-Pagan LM Test and Hausman Test.F test and Breusch-Pagan LM Test examines the null hypotheses that there is no fixed effect and there is no random effect respectively.If null hypotheses of these tests are not rejected, pooled OLS is the best suitable choice.If H 0 of the F test is rejected and H 0 of Breusch-Pagan LM Test is not rejected, fixed effect model should be selected.Reverse results for both tests suggest random effect model as suitable one.If H 0 of both tests is rejected, Hausman test will be used to choose correct model.If the null hypothesis of Hausman test is rejected, use the fixed effect model; otherwise, go for the random effect model(Park, 2011).
F Test, Breusch-Pagan LM Test and Hausman Test results for the models are represented in Table4.