JURNAL BISNIS DAN EKONOMI, September 2004
ANALYSIS OF ACCOUNTING BETAS MEASUREMENT
Oleh : Junaidi
Dosen STIE Widya Manggala Semarang
Abstract
The purpose of this study is to determine the significant of accounting beta (i.e., earning beta, fund flow beta and cash flow beta) in relationship with market risk. The results indicate that operating cash flow risk measure (beta) is significantly relate to market risk, while the others, earnings beta and fund flow beta, do not possess relate with market risk. Adjusting with the issue of a thin trading bias in beta measurement, this study adjusted all the betas using Vasicek Bayesian Method. The results are still consistent. The implication of the result is to prove that accounting beta, especially cash flow beta, can be a surrogate to market beta or market risk.
Keywords: accrual earnings, fund flow, operating cash flow, market risk,
Vasicek Bayesian’s
INTRODUCTION
The Indonesian institute of accountants (IAI) conceptual framework considers providing information on an enterprise’s accrual earnings generating ability as the underlying objective of financial reporting (IAI, 1999, par.22). While the IAI stresses the importance of accrual earnings, it nevertheless, adopts the view that cash flows is basic to assessment performance of the firm in generating cash and cash equivalent (IAI, 1999, statement of financial accounting standard No. 2).
The Financial Accounting Standards Board’s (FASB) conceptual framework considers providing information on an enterprise’s cash flow generating ability as the underlying objective of financial reporting. The FASB has recently issued SFAS No. 95 establishing standards for reporting cash flows in the financial statements, superseding APB Opinion No. 19 (FASB1987). Similar with the IAI, while the FASB stresses the importance of cash flows, it, nevertheless, adopts the views that accrual earnings may provide a better indicator of enterprise performance than information about current cash flows (FASB, 1978, PAR. 44).
Jurnal Bisnis dan Ekonomi – Vol.11 – No.2 – September 2004
The hypothesis that accounting earnings is an empirical surrogate for (or indicator of) future cash flows underlies the alleged the usefulness of earnings in assessing security value/risk. The idea that earnings is a surrogate for future cash flows is consistent with the evidence that conclusively supports the relation between earnings and security prices (Lev and Ohlson, 1982).
Ball and Brown (1969) investigated the usefulness of earnings in assessing security risk. Research then tested the association between the market based beta and an accounting beta. Accounting beta (â^{A}) is expressed as the covariability of a firm’s accounting earnings with the accounting earnings of the market portfolio:
â^{A}_{i }= cov (X_{i} , X_{m}) / ó^{2} (X_{m})
where is X_{i} an accounting earnings and X_{m} is an accounting the market portfolio._{ }Spearman rank order from Ball and Brown’s study show that for correlations between operating earnings and market beta is 0,46 and between earnings after tax and market beta is 0,39 and between earnings for common shares and market beta is 0,41. The result is accounting beta significantly associated with market beta as proxy security risk.
Beaver et al., (1970) investigated factors influence market risk. One’s of the factor is accounting beta. Beaver et al., found, in a model using accounting variables to forecast market risk, earnings variability was the most significant variable and accounting beta did not make a statistically significant contribution.
THEORETICAL FRAMEWORKS AND HYPOTHESIS FORMULATION
The idea that current earnings may be a better surrogate for future cash flows than current cash flows is not without critics. Questions are also raised about the motivation behind the selection of accounting procedures, from within those generally accepted, by managers of firms with earnings based compensation plans. According to the financial press, if a firm has a loss, managers increase the loss by including all possible future losses that they ca write off – take a "big bath" – so that future period’s earnings are higher (Healy, 1985). The big bath phenomenon is consistent with the bonus formula. If earnings are below the target, any attempt to increase earnings above the target by shifting earnings forward in time involves the permanent loss of some of the bonus.
A number of researchers, therefore, have suggested a fundamental change in financial reporting toward more emphasis on cash flow based measures. Cash flow reporting system properly captures the inflation impact given that an appropriate time value rate is applied to discount periodic cash flow. Lawson (1980) uses the tax neutrality principle to argue that a cash flow based tax system could have lessened the impact of the 1974 financial crisis in Britain. This argument have supported with Lawson (1980).
Early empirical work that tried to assess the relative information content of earnings and cash flows employed cash flow proxies by adjusting earnings for some accruals. In particular, the funds flow variables often tested were either earnings plus depreciation or earnings plus depreciation and deferred taxes. Ball and Brown (1968) evaluation of accounting income numbers requires agreement as to what real word outcome constitutes an appropriate test of usefulness. Both the content and the timing of existing annual net income numbers will be evaluated since usefulness could be impaired by deficiencies in either. The results demonstrate that the information contained in the annual income number is useful in that it is related to stock prices.
The empirical work to evaluation relation between accounting information and stock prices are Patell and Kaplan (1977), Baran et al., (1980), and Beaver and Landsman (1983). With the exception of Baran et al., these studies generally failed to find evidence that the funds flow variables possess information beyond that of earnings.
Bowen et al., (1986) have reported that the similarity in the properties of earnings and funds flows has been offered as an explanation for the failure of early research to find incremental information in funds flows over that provided by earnings. Further, because of the dissimilarity in their characteristics, it has been hypothesized that cash flows have potential information not captured by either earnings or funds flows.
Recent stock return association studies by Rayburn (1986), Wilson (1986 and 1987), and Bowen et al. (1987) have incorporated properly specified (i.e., fully adjusted) cash flow variables into the analysis. These studies have generally concluded that earnings provide information that is not contained in cash flows alone. Consistent with the early work, however, the recent research fails to find evidence that funds flows have incremental information relative to earnings.
The issue of the relative ability of cash flows and earnings to explain market risk has not been addressed in prior research. Moreover, notwithstanding the central role that market risk plays in determining a security’s expected return, the results of the recent security return based research about the information in cash flows versus earning may not be applicable to risk assessment. Accruals can offset or cancel the transitory component of current cash flows so that unexpected earnings is a better proxy of unexpected permanent earnings than is the current amount of unexpected cash flows.
Rayburn (1986) reports that accounting earnings exhibit lower variability than cash flows. The dampening effect of the accounting accrual process may mean that the earnings series will produce a risk measure that is a poorer proxy of market risk than would the corresponding cash flow series.
Ismail and Kim (1989), investigates whether cash flow based measures of risk have crosssectional variation in market beta beyond that provided by earnings based risk measures. The results indicate that funds and cash flow risk measures (betas) provide significant incremental explanatory power over that provided by the earnings risk measures (beta) in explaining the variability in market betas. Regardless of the specifics of variable definition, the issues in previous research can be summarized by three questions: do the accrual earnings, fund flow and cash flow betas relate with market beta? These questions can be formulated as the following null hypotheses:
H_{1}: The measurement of Earning Beta has significant influence toward market risk
H_{2}: The measurement of fund flow beta has significant influence toward market risk
H_{3}: The measurement of cash flow beta has significant influence toward market risk
RESEARCH METHOD
This study is conducted in the spirit of the early works by Ball and Brown (1969), establishing the empirical/theoretical relation between accounting variables and market risk. The research design is structured primarily on the basis of a simple regression model with market beta as the dependent variable and accounting betas as the independent variables. The analysis is directed at testing the significance of the incremental explanatory power, with respect to the variability in market beta, of the earnings, fund flows, and operating cash flow risk measures relative to each other.
Fundamentally, the method of analysis is a market risk association test. The empirical work in this paper was conducted using both the unadjusted beta estimates and the Bayesian adjusted betas.
1. Market Beta
Monthly stock returns over January 1992 – January 2000 period were used in deriving empirical estimates of market beta from the familiar market model:
R_{i}_{ }= a_{i} + b_{i}R_{M} + e_{i} (3.1)
Where:
R_{i} = rate of return on security i in period t,
a_{I} = intercept,
b_{i} = market beta for security i,
R_{M} = rate of return on market portfolio (IHSG value weighted index), and
e_{i} = disturbance term with ì (U_{it}) = 0 and constant variance.
2. Accounting Betas
Annual observations, over the same periods (19922000), were relied on to estimate accounting betas using the time series regression:
r_{i}_{ }= a_{i} + b_{i }r_{m} + e_{i} (3.2)
Where:
r_{i} = an accounting return variable for firm i in period t,
a_{I} = intercept for firm i,
b_{i}_{ }= accounting beta for firm i,
r_{m} = market index for accounting return computed as the simple average of sample accounting returns, r_{m} in period t, and
e_{i} = disturbance term with ì (U_{it}) = 0 and constant variance.
Beta estimation errors have been noted in earlier studies (Blume 1971) due to the effect of nonstationarity of the beta coefficients.
Approaches were followed in an attempt to circumvent the problems associated with these errors. This study uses Vasicek’s Bayesian technique (1973) to adjust the initial estimates of both market and accounting betas.
Although not designed to adjust for a thin trading, a Bayesian correction does account for differences in the statistical reliability of beta values. It allows for the fact that estimates for thinly traded shares are less reliable. Using Vasicek (1973) weightings, the estimator is:
(3.3)
where:
b_{1} = Averages from share betas values in the sample historical period,
s^{2 }b_{1}_{ } = Variance from share beta in the sample historical period,
s ^{2}b_{i1 }= Variance from share beta i,
b^{ }_{i1} = Historical share beta i.
3. Accounting Return Variables
Three accounting return variables were investigated: an earnings measure, a funds flow, and a operating cash flow measure. For each year, t, these returns are defined from PACAP and Indoexchange data items as follows:
 Earnings: Operating income divided by beginning of the period market value of common equity (operating income divide multiple result closing price with outstanding share) . [(OI)_{t} / (CP x OS)_{t1}]
 Funds flow: Operating income plus depreciation, amortization, and depletion divided by beginning of the period market value of common equity. [(OI + DEP)_{t} / (CP X OS)_{t1}]
 Operating cash flow: Cash flows generated from continuing operations divided by beginning of the period market value of common equity, where cash flows are defined as operating income plus depreciation, amortization, and depletion and the change in non cash working capital. Non cash working capital is change between current asset less cash and short term investment, less current liabilities this period and current asset less cash and short term investment, less current liabilities beginning period. [(OI + DEP)_{t} + (CA – C&CE – CL)_{t} – (CA – C&CE – CK)_{t1}] / [(P x OS) _{t1}]
The definition of the earnings return variable is in agreement with Ball and Brown (1969) and Beaver et al., (1970). An alternative method to estimate accounting betas as a direct proxy for the market beta is shown in Beaver et al., (1970). Accounting betas, according to this alternative, are calculated on a before tax and interest basis and then adjusted for taxes and financial leverage.
4. Data Source and Sample Selection
The data used are drawn from the PACAP and Indoexchange file for the 19922000 periods. The screening process required manufacturing firms to have a complete set of the required monthly and annual data for that period and to have fiscal year ends in December. The screening process resulted in a sample of 58 firms as show in table 1. Specification of the accounting return variables required on a oneyear lagged measure, thus losing one year. The total period becomes 8 year of data.
Table 1
Manufacturing Firms Sample
No.  Summary  Firms  No.  Summary  Firms 
1  ALKA  Alakasa Industrindo  30  KLBF  Kalbe Farma 
2  MYTX  APAC Centertex Corp.  31  KKGI  Kurnia Kapuas Utama Glue Ind. 
3  AQUA  Aqua Golden Missisippi  32  LMSH  Lionmesh Prima 
4  ARGO  Argo Pantes  33  MLPL  Multipolar Corporation 
5  ASGR  Astra Graphia  34  MLBI  Multi Bintang Indonesia 
6  ASII  Astra Internasional  35  MTDL  Metrodata 
7  BYSB  Bayer  36  MDRN  Modern Photo 
8  BRNA  Berlina  37  NIPS  Nipress 
9  BRAM  Branta Mulia  38  TKIM  Pabrik Kertas Tjiwi Kimia 
 BAT  BAT Indonesia  39  HDTX  Panasia 
11  CTBN  Citra Tubindo  40  POLY  Polysindo 
12  DNKS  Dankos Laboratories  41  RDTX  Roda Vivatex 
13  DLTA  Delta Djakarta  42  SHDA  Sari Husada 
14  DPNS  Duta Pertiwi Nusantara  43  SCPI  Schering Plough Indonesia 
15  DYNA  Dynaplast  44  BATA  Sepatu Bata 
16  EKAD  Ekadarma Tape Industries  45  SMCB  Semen Cibinong 
17  ERTX  Eratex Djaja Limited  46  SMGR  Semen Gresik 
18  GJTL  Gajah Tunggal  47  IKBI  Sumi Indokabel 
19  GRIV  Great River International  48  SCCO  Supreme Cable Manufacturing 
20  GDYR  Goodyear Indonesia  49  TOTO  Surya Toto Indonesia 
21  GGRM  Gudang Garam  50  SQBI  Squibb Indonesia 
22  HMSP  Hanjaya Mandala Sampoerna  51  SUBA  Suba Indah 
23  MYRX  Hanson Industri Utama  52  TFCO  Teijin/Tifico 
24  IGAR  Igarjaya  53  TBMS  Tembaga Mulia semanan 
25  INKP  Indah Kiat Pulp&Paper  54  TRPK  Trafindo Perkasa 
26  INTP  Indocement Tunggal Perkasa  55  TRST  Trias Sentosa 
27  INDR  IndoRama synthetics  56  ULTJ  Ultrajaya Milk Ind. & Trading 
28  INCI  Intan Wijaya Chemical Industry  57  UNIC  Unggul Indah Corporation 
29  ITMA  Itamaraya Gold Industri  58  UNVR  Unilever Indonesia 
RESULTS AND DISCUSSIONS
1. Descriptive Statistics
The empirical work in this paper was conducted using both the unadjusted beta estimates and the Bayesian adjusted betas. The two sets of results were found to be quite similar and, thus, we report and discuss the results based on the before and after Bayesian adjusted betas.
Table 2
Descriptive Statistics
 Unadjusted Beta  Adjusted Beta  
 N  Mean  SD  Mean  SD  
Earn_Beta  58  0,99  1,95  0,89  1,36  
Fund_Beta  58  0,99  1,89  0,84  1,20  
Cash_Beta  58  1,00  5,20  0,96  4,77  
Market _Beta  58  0,58  1,66  0,57  0,31  
Valid N (listwise)  58 




Table 2 presents descriptive statistics, mean and standard deviation, for market and accounting betas. As expected, the Bayesian adjustment procedure reduced the variability of the risk measures, particularly for accounting betas. As observed in other studies (e.g., Beaver and Manegold (1975) and Baran et al. (1980), accounting betas have higher crosssectional standard deviations relative to that of the market beta. This observation is due, in part, to the number of observations (annual vs. monthly) used in obtaining the estimates.
Table 3
Pearson Correlation Result
After Vasicek
Before Vasicek 
 Earn(prob.)  Fund(prob.)  Cash(prob.)  Market(prob.) 
Earn_Beta  1,000  0,929(0,000**)  0,056(0,675)  0,132(0,322)  
Fund_Beta  0,929(0,000)  1,000  0,081(0,546)  0,056(0,679)  
Cash_Beta  0,056(0,675)  0,081(0,546)  1,000  0,257(0,052*)  
Market_Beta  0,132(0,322)  0,056(0,679)  0,257(0,052)  1,000  
Earn_Beta  1,000  0,918(0,000**)  0,069(0,607)  0,164(0,219)  
Fund_Beta  0,918(0,000**)  1,000  0,053(0,693)  0,062(0,643)  
Cash_Beta  0,069(0,607)  0,053(0,693)  1,000  0,332(0,011*)  
Market_Beta  0,164(0,219)  0,062(0,643)  0,332(0,011*)  1,000 
** Correlation is significant at the 0,01 level (2tailed).
* Correlation is significant at the 0,05 level (2tailed).
From Table 3, it could be seen that all variables have positive relation with market risk. This result gives a prior conclusion that there was a positive relation between variables of accounting beta and a market risk.
Operating cash flow beta had a value of 0.257 after adjustment and 0.332 before adjustment, correlated with market risk. Other accounting beta likes earnings beta and fund flow beta not significant correlated with market beta. This result gave a prior indication that the information existing in the operational cash flow is more capable in explaining the market risk variable.
There was a very high correlation between profit beta and cash flow beta of 93% after adjustment was done, and 92% before beta adjustment was done. Viewed from significance value in two direction testing, it was drawn a conclusion that this relation was significant at the level of 1%. IT gave a reflection that there was multicolinearity between profit beta and cash flow beta.
The term of multicolinearity was used to show the existence of linear relation among independent variables in regression model. Multicolinearity in this study could be avoided because the assessment of independent profit variables correlated linearly with cash flow independent variable measurement. The characteristic existing in both variables changed simultaneously all the time, in which the value was affected by the same factors. Gujarati (1995, pp 344345) said that if the goal of prediction was only to predict dependent variables’ value, then, the multicolinear problem can be avoided, in condition that the collinearity style continuously the same in prediction period as observed in sample period.
Ismail and Kim (1989) stated that there was a collinearity problem between the accounting risk measurements, especially between profit and cash flow, and cash flow with operational cash flow. The researcher concluded that with the existence of this co linearity.
2. The Result of Hypothesis Testing
The result of hypothesis testing H1, H2, and H3 was gained from the simple regression result by using each explanatory variable of profit beta measurement, cash flow, and operational cash flow on variable related market risk. Table 4, explained that the regression result between variable related to market risk with explanatory variable of profit beta. The result was as follow:
Table 4
Simple Regression Y = a + b_{i}X_{i + }e_{i}
Earn_Beta  R^{2}  S,E,of Regretion  t_test (probabilitas)  F_test (probabilitas)  Durbin_Watson 
Bayesian Adjusted Beta  0,0175  0,3116  0,99(0,32)  0,99(0,32)  2,1592 
Unadjusted Beta  0,0268  0,6569  1,24(0,22)  1,54(0,22)  1,9116 
From the coefficient mark in table 4, it was gained a result that profit beta had positive impact direction or suitable with market risk. The number of R^{2} (determination coefficient) after the adjustment was of 0.0175, which meant that only 1.75% variables related to market risk that can be explained by profit beta variable. While, the rest of it, 98.25%, explained by other variables. This determination coefficient value was seen indifferent with the result before Vasicek adjustment was carried out.
The value of standard errors of estimate of 0.311639 was bigger than market risk deviation standard value, of 0.311631 that gave a conclusion that the market average risk was better as a predictor of market risk, than regression model in the testing after adjustment was carried out. It gave an illustration that profit beta was worse in explaining market risk than the market risk average. But, the contrary happened when Vasicek adjustment was not carried out, either in market beta or accounting beta.
The value of F counted was of 0.99 with significance level of 0.32, which was much bigger than 0.05, then, regression model could not be used again to predict the market risk on the testing after the adjustment. The similar thing happened in the testing before beta adjustment was carried out.
The testing of regression coefficient significance of profit beta variable was seen from the comparison of t counted value with t table. If t counted was smaller than t table, then, the regression coefficient was not significant as the determinant of market risk. If on the adverse condition happened, then, regression coefficient was significant as the market risk determinant.
The value of t counted gained from the regression result of 0.99, was compared with t table in the two sides testing with the significance level of 5%, and the explanatory degree n2 was 56. The gained t table value from the t distribution table, the result of value interpolation of degree of freedom (df) 40 was of 2.021 and df 60 was of 2, then, from 2 + ((^{4}/_{20}) x (2.021 – 2)), it was gained a value of 2.0042. Because t counted was smaller than t table, thus, it could be concluded that the earning beta determinant coefficient was not significant as the explanatory of market risk. The result of this conclusion, that hypothesis 1 did not success to be supported by data, either in the beta testing before or after the beta adjustment was carried out by using Vasicek method.
To prove the existence of autocorrelation, it was compared the value of DurbinWatson of 2.1592 after beta adjustment was carried out and 1.9116 before beta adjustment was carried out, with table value. The value of d_{1} in the sample n = 55 was of 1.53 and sample n = 60 was of 1.55. From the result of interpolation n = 57, it was gained a value d_{1} of 1.562 from the measurement of 1.53 + ((^{3}/_{5} x (1.55 – 1.53)). To test the autocorrelation, it should be find the result of 4 – d_{1} was of 2.438. The value d_{u} sample n = 55 was of 1.6 and sample n = 60 was of 1.62. From the result of interpolation n = 57, it was gained d_{u} value of 1.612 and 4 – d_{u} of 2.388. Because D_{u} <>_{u}), then, the model in this equation did not face autocorrelation problem in the testing, either before or after the beta adjustment was carried out.
The testing on the heteroscedasticity assumption was carried out by using White method (appendix 1). This assumption was happened because the probability distribution of disturbance was assumed to be the same (constant) for all observation of X, that was the variant of each U_{1} was the same for all independent variable value, which in fact had inconstant variant. From the result of White testing, it was gained R^{2} observed, which was not significant in the significance level of 5%. This result gave a conclusion that there was no heteroscedasticity fault in model or variant in model having heteroscedasicity character.
Table 5, explained regression that the regression result between variable related to market risk with explanatory variable of fund flow beta. The result was as follow:
Table 5
Simple Regression Y = a + b_{i}X_{i + }e_{i}
Fund_Beta  R^{2}  S,E,of Regretion  t_test (probabilitas)  F_test (probabilitas)  Durbin_Watson 
Bayesian Adjusted Beta  0,0031  0,3139  0,42(0,67)  0,17 (0,67)  2,1781 
Adjusted Beta  0,0038  0,6646  0,46(0,64)  0,21(0,64)  1,9468 
From the coefficient mark in table 5, it was gained a result that fund flow beta had positive impact direction or suitable with market risk. The number of R^{2} (determination coefficient) after the adjustment was of 0,0031, which meant that only 0,3% variables related to market risk that can be explained by profit beta variable. While, the rest of it, 99.97%, explained by other variables. This determination coefficient value was seen indifferent with the result before Vasicek adjustment was carried out.
The value of standard errors of estimate of 0.3139 was bigger than market risk deviation standard value, of 0.3116 that gave a conclusion that the market average risk was better as a predictor of market risk, than regression model in the testing after adjustment was carried out. It gave an illustration that profit beta was worse in explaining market risk than the market risk average. But, the contrary happened when Vasicek adjustment was not carried out, either in market beta or accounting beta.
The value of F counted was of 0.17 with significance level of 0.67, which was much bigger than 0.05, then, regression model could not be used again to predict the market risk on the testing after the adjustment. The similar thing happened in the testing before beta adjustment was carried out.
The testing of regression coefficient significance of fund flow beta variable was seen from the comparison of t counted value with t table. If t counted was smaller than t table, then, the regression coefficient was not significant as the determinant of market risk. If on the adverse condition happened, then, regression coefficient was significant as the market risk determinant.
The value of t counted gained from the regression result of 0.42, was compared with t table in the two sides testing with the significance level of 5%, and the explanatory degree n2 was 56. The gained t table value from the t distribution table, the result of value interpolation of degree of freedom (df) 40 was of 2.021 and df 60 was of 2, then, from 2 + ((^{4}/_{20}) x (2.021 – 2)), it was gained a value of 2.0042. Because t counted was smaller than t table, thus, it could be concluded that the fund flow beta determinant coefficient was not significant as the explanatory of market risk. The result of this conclusion, that hypothesis 2 did not success to be supported by data, either in the beta testing before or after the beta adjustment was carried out by using Vasicek method.
To prove the existence of autocorrelation, it was compared the value of DurbinWatson of 2.1781 after beta adjustment was carried out and 1.9468 before beta adjustment was carried out, with table value. The value of d_{1} in the sample n = 55 was of 1.53 and sample n = 60 was of 1.55. From the result of interpolation n = 57, it was gained a value d_{1} of 1.562 from the measurement of 1.53 + ((^{3}/_{5} x (1.55 – 1.53)). To test the autocorrelation, it should be find the result of 4 – d_{1} was of 2.438. The value d_{u} sample n = 55 was of 1.6 and sample n = 60 was of 1.62. From the result of interpolation n = 57, it was gained d_{u} value of 1.612 and 4 – d_{u} of 2.388. Because D_{u} <>_{u}), then, the model in this equation did not face autocorrelation problem in the testing, either before or after the beta adjustment was carried out.
The testing on the heteroscedasticity assumption was carried out by using White method (appendix 2). This assumption was happened because the probability distribution of disturbance was assumed to be the same (constant) for all observation of X, that was the variant of each U_{1} was the same for all independent variable value, which in fact had inconstant variant. From the result of White testing, it was gained R^{2} observed, which was not significant in the significance level of 5%. This result gave a conclusion that there was no heteroscedasticity fault in model or variant in model having heteroscedasicity character.
Table 6, explained regression that the regression result between variable related to market risk with explanatory variable of operating cash flow beta. The result was as follow:
Table 6
Simple Regression Y = a + b_{i}X_{i + }e_{i}
Cash_Beta  R^{2}  S,E,of Regretion  t_test (probabilitas)  F_test (probabilitas)  Durbin_ Watson 
Bayesian Adjusted Beta  0,0658  0,3038  1,98 (0,05)  3,94 (0,05)  2,3203 
Unadjusted Beta  0,1104  0,6281  2,63(0,01)  6,95(0,01)  2,0822 
From the coefficient mark in table 6, it was gained a result that operating cash flow beta had positive impact direction or suitable with market risk. The number of R^{2} (determination coefficient) after the adjustment was of 0,0658, which meant that only 6,58% variables related to market risk that can be explained by profit beta variable. While, the rest of it, 93,42%, explained by other variables. This determination coefficient value was seen indifferent with the result before Vasicek adjustment was carried out.
The value of standard errors of estimate of 0.3036 was smaller than market risk deviation standard value, of 0.3116 that gave a conclusion that the market model was better as a predictor of market risk, than regression model in the testing after adjustment was carried out. It gave an illustration that operating cash flow beta was good in explaining market risk than the market risk average.
The value of F counted was of 3,94 with significance level of 0,05, which was much smaller than 0.10, then, regression model could be used to predict the market risk on the testing after the adjustment. The similar thing happened in the testing before beta adjustment was carried out.
The testing of regression coefficient significance of profit beta variable was seen from the comparison of t counted value with t table. If t counted was smaller than t table, then, the regression coefficient was not significant as the determinant of market risk. If on the adverse condition happened, then, regression coefficient was significant as the market risk determinant.
The value of t counted gained from the regression result of 1,98 with significance 0,05. Because probability counted was smaller than level significance 0,1 thus, it could be concluded that the operating cash flow beta determinant coefficient was significant as the explanatory of market risk. The result of this conclusion, that hypothesis 3 did success to be supported by data, either in the beta testing before or after the beta adjustment was carried out by using Vasicek method.
To prove the existence of autocorrelation, it was compared the value of DurbinWatson of 2.3203 after beta adjustment was carried out and 2,0822 before beta adjustment was carried out, with table value. The value of d_{1} in the sample n = 55 was of 1.53 and sample n = 60 was of 1.55. From the result of interpolation n = 57, it was gained a value d_{1} of 1.562 from the measurement of 1.53 + ((^{3}/_{5} x (1.55 – 1.53)). To test the autocorrelation, it should be find the result of 4 – d_{1} was of 2.438. The value d_{u} sample n = 55 was of 1.6 and sample n = 60 was of 1.62. From the result of interpolation n = 57, it was gained d_{u} value of 1.612 and 4 – d_{u} of 2.388. Because D_{u} <>_{u}), then, the model in this equation did not face autocorrelation problem in the testing, either before or after the beta adjustment was carried out.
The testing on the heteroscedasticity assumption was carried out by using White method (appendix 3). This assumption was happened because the probability distribution of disturbance was assumed to be the same (constant) for all observation of X, that was the variant of each U_{1} was the same for all independent variable value, which in fact had inconstant variant. From the result of White testing, it was gained R^{2} observed, which was not significant in the significance level of 5%. This result gave a conclusion that there was no heteroscedasticity fault in model or variant in model having heteroscedasicity character.
CONCLUSIONS AND LIMITATIONS
1. The Study’s Conclusions
The major purpose of this paper was to determine whether fund flow, cash flows, and earnings beta related with the market risk. The results of this conclusion, that hypothesis 1 and 2 earning and fund flow betas did not success to be supported by data, either in the beta testing before or after the beta adjustment was carried out by using Vasicek method.
The result indicate that operating cash flow risk measure (beta) provide significant association in explaining the variability in market betas. This findings relevan with Ismail and Kim (1989), and confirm the hypothesis presented in the recent literature (e.g., Gombola and Ketz (1983) and Bowen et al. (1986) about earnings and the traditional cash flow proxies (i.e., funds flows) having potentially similar information because of their similar statistic properties. I find that the existence of substantial multicollinearity among the variables does not negate the possibility of each providing explanatory power in the presence of the others
2. The Study’s Limitations
This study had several limitations. Those limitations could be outlined as follows:

It concerns the conformity of accountancy beta measurement with market beta measurement. Accountancy beta measurement uses annual data whereas in counting market risk, it uses monthly data in measuring coefficient value in accountancy variable, to get beta. Data that is used, as independent variable is that mean data of earning rate, fund flow and cash flow of sample company whereas independent variable in counting the value of market variable coefficient, it is used "IHSG". So the result of this research cannot be described as match conformity.

This research also does not use cash flow data that is publicity, but it must count the rate of fund flow and operation cash flow. It happens because supervision period in this research from 1991 to 1999. Whereas accountancy standard makes compulsory to publish new cash flow report on financial report in 1995.

The period that is observed in this researches including economy crisis until beta measurement both accountancy beta and market beta in the crisis period and before and after crisis period.
3. The Study’s Implications
The major implication of the results of this study is that, within the confines of explaining market risk, the information in accrual earnings may be largely a subset of the broader set of information included in operating cash flow. Hence, operating cash flow data has the potential of supplying additional information on a firm’s risk. From this perspective, and give that security prices incorporate investor’s assessments of a firm’s debt/dividendpaying ability, the findings of this paper lend credence to the theoretical arguments about the importance of operating cash flow as indicator of some aspect of a firm’s performance.
References
Ball, R., and P. Brown, "An Empirical Evaluation of Accounting Income Numbers," Journal of Accounting Research (Autumn 1968), pp. 159178.
___, and ___, "Portfolio Theory and Accounting," Journal of Accounting Research 7 (Autumn 1969), pp. 300323.
Baran, A., J, Lakonishok, and A. R., Ofer, "The Information Content of General Price Level Adjusted Earnings: Some Empirical Evidence," The Accounting Review (January 1980), pp. 2235.
Beaver, W., and J. Managold, "The Association Between MarketDetermined and AccountingDetermined Measures of Systematic Risk: Some Further Evidence," Journal of Financial and Quantitative Analysis (Juni 1975), pp. 231284.
Beaver, P. Kettler, and M. Scholes, "The Association Between Market Determined and Accounting Determined Risk Measures," The Accounting Review (October 1970), pp. 654682.
_____, and W. R. Landsman, "Incremental Information Content of Statement 33 Disclosures (Research Report: Financial Accounting Standards Board, 1983).
Blume, M., "On the Assessment of Risk, "Journal of Finance (Maret 1971), pp.
110.
Bowen, R. M., D. Burgstahler, and L. A. Daley, "Evidence on the Relationships Between Earnings and Various Measures of Cash Flow," The Accounting Review (October 1986), pp.713725.
______,_______, dan_______, "The Incremental Information Content of Accrual Versus Cash Flows," The Accounting Review (October 1987), pp. 723747.
Financial Accounting Standards Board, "Objectives of Financial Reporting by Business Enterprises," Statement of Financial Accounting Concepts No. 1 (1978), pp. 51835198.
______, "Statement of Financial Accounting Standards No. 95: Statement of Cash Flows (November 1987), pp. 10451088.
Gujarati, Damodar N., "Basic Econometries," Third Edition., McGraw Hill International Edition, New York, (1995) pp. 319354.
Gombola, M. J., and J. E. Ketz, "a Note on Cash Flow and Classification Patterns of Financial Ratios," The Accounting Review (January 1983), pp. 435452.
Ikatan Akuntan Indonesia, Standar Akuntansi Keuangan, Buku Satu. Salemba Empat, Jakarta, (1999), pp. 2.12.20.
Ismail, E. Badr and Kim, K. Moon, "On the Association of Cash Flow Variables with Market Risk: Futher Evidence," The Accounting Review (Januari 1989), pp.125136.
Healy, P., "The Impact of Bonus Schemes on the Selection of Accounting Principles," Journal of Accounting and Economics 7 (April 1985).
Lawson, G., "The Measurement of Corporate Profitability on a CashFlow Basis," The International Journal of Accounting: Education and Research (1980)
Lev, B., and J. A., Ohlson, "MarketBased Empirical Research in Accounting: A Review Interpretation, and Extention," Supplement of Journal of Accounting Research (1982), pp. 249322.
Patell, J. H., dan R. S. Kaplan, "The Information Content of Cash Flow Data Relative to Annual Earning," Unpublished Working Paper (Stanford University, 1977).
Rayburn, J., "The Association of Operating Cash Flow and Accruals with Security Returns," Supplement to Journal of Accounting Research (1986), pp. 112133.
Vasicek, O., "A Note on Using CrossSectional Information in Bayesian Estimation of Security Betas," The Journal of Finance (Desember 1973), pp. 12331239.
Wilson, G. P., "The Relative Information Content of Accruals and Cash Flows: Combined Evidence at the Earnings Announcement and Annual Report Release Date," Supplement to Journal of Accounting Research (1986), pp. 165200.
______, "The Incremental Information Content of the Accrual and Funds Component of Earnings After Controlling for Earnings," The Accounting Review (April 1987), pp. 293322.
Appendix 1
Output White’s General Heteroscedasticity Test
With Independent Variable Earn_Beta
White Heteroskedasticity Test:  
 
Fstatistic  1,368180  Probability  0,263097  
Obs*Rsquared  2,748855  Probability  0,252984  





Test Equation:  
Dependent Variable: RESID^2  
Method: Least Squares  
Date: 04/07/86 Time: 18:40  
Sample: 1 58  
Included observations: 58  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  0,513960  0,147977  3,473243  0,0010 
Earn_Beta  0,220307  0,136813  1,610276  0,1131 
Earn_Beta^2  0,025931  0,021354  1,214340  0,2298 
Rsquared  0,047394  Mean dependent var  0,416705  
Adjusted Rsquared  0,012754  S.D. dependent var  1,000407  
S.E. of regression  0,994007  Akaike info criterion  2,876194  
Sum squared resid  54,34277  Schwarz criterion  2,982768  
Log likelihood  80,40962  Fstatistic  1,368180  
DurbinWatson stat  1,689509  Prob(Fstatistic)  0,263097 
Appendix 2
Output White’s General Heteroscedasticity Test
With Independent Variable Fund_Beta
White Heteroskedasticity Test:  
Fstatistic  0,397643  Probability  0,673818  
Obs*Rsquared  0,826711  Probability  0,661427  





Test Equation:  
Dependent Variable: RESID^2  
Method: Least Squares  
Date: 04/07/86 Time: 18:42  
Sample: 1 58  
Included observations: 58  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  0,505031  0,165065  3,059582  0,0034 
Fund_Beta  0,120421  0,166033  0,725280  0,4714 
Fund_Beta^2  0,009238  0,022165  0,416786  0,6785 
Rsquared  0,014254  Mean dependent var  0,426550  
Adjusted Rsquared  0,021592  S.D. dependent var  1,050589  
S.E. of regression  1,061870  Akaike info criterion  3,008279  
Sum squared resid  62,01629  Schwarz criterion  3,114854  
Log likelihood  84,24010  Fstatistic  0,397643  
DurbinWatson stat  1,747735  Prob(Fstatistic)  0,673818 
Appendix 3
Output White’s General Heteroscedasticity Test
With Independent Variable Cash_Beta
White Heteroskedasticity Test:  
Fstatistic  1,736044  Probability  0,185732  
Obs*Rsquared  3,444055  Probability  0,178704  





Test Equation:  
Dependent Variable: RESID^2  
Method: Least Squares  
Date: 04/07/86 Time: 18:42  
Sample: 1 58  
Included observations: 58  
Variable  Coefficient  Std. Error  tStatistic  Prob. 
C  0,329086  0,120378  2,733764  0,0084 
Cash_Beta  0,025102  0,022379  1,121691  0,2669 
Cash_Beta^2  0,002785  0,001618  1,721704  0,0907 
Rsquared  0,059380  Mean dependent var  0,380918  
Adjusted Rsquared  0,025176  S.D. dependent var  0,862222  
S.E. of regression  0,851299  Akaike info criterion  2,566232  
Sum squared resid  39,85905  Schwarz criterion  2,672806  
Log likelihood  71,42072  Fstatistic  1,736044  
DurbinWatson stat  1,877709  Prob(Fstatistic)  0,185732 
Tidak ada komentar:
Poskan Komentar