ANALYSIS OF ACCOUNTING BETAS MEASUREMENT

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) / ó² (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 cross-sectional 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:

H₁: The measurement of Earning Beta has significant influence toward market risk

H₂: The measurement of fund flow beta has significant influence toward market risk

H₃: 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_iR_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 (1992-2000), 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:

b₁ = Averages from share betas values in the sample historical period,

s² b₁ = Variance from share beta in the sample historical period,

s ²b_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:

1. 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)_t-1]
   
2. 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)_t-1]
   
3. 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)_t-1] / [(P x OS) _t-1]
   

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 1992-2000 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 one-year 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	Indo-Rama 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 cross-sectional 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 (2-tailed).

* Correlation is significant at the 0,05 level (2-tailed).

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 344-345) 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_iX_{i +} e_i

Earn_Beta	R2	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 R² (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 n-2 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 Durbin-Watson of 2.1592 after beta adjustment was carried out and 1.9116 before beta adjustment was carried out, with table value. The value of d₁ 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 d₁ 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 – d₁ 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 U₁ was the same for all independent variable value, which in fact had inconstant variant. From the result of White testing, it was gained R² 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_iX_{i +} e_i

Fund_Beta	R2	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 R² (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 n-2 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 Durbin-Watson of 2.1781 after beta adjustment was carried out and 1.9468 before beta adjustment was carried out, with table value. The value of d₁ 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 d₁ 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 – d₁ 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 U₁ was the same for all independent variable value, which in fact had inconstant variant. From the result of White testing, it was gained R² 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_iX_{i +} e_i

Cash_Beta	R2	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 R² (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 Durbin-Watson of 2.3203 after beta adjustment was carried out and 2,0822 before beta adjustment was carried out, with table value. The value of d₁ 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 d₁ 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 – d₁ 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 U₁ was the same for all independent variable value, which in fact had inconstant variant. From the result of White testing, it was gained R² 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:

1. 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.

2. 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.

3. 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/dividend-paying 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. 159-178.

___, and ___, "Portfolio Theory and Accounting," Journal of Accounting Research 7 (Autumn 1969), pp. 300-323.

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. 22-35.

Beaver, W., and J. Managold, "The Association Between Market-Determined and Accounting-Determined Measures of Systematic Risk: Some Further Evidence," Journal of Financial and Quantitative Analysis (Juni 1975), pp. 231-284.

Beaver, P. Kettler, and M. Scholes, "The Association Between Market Determined and Accounting Determined Risk Measures," The Accounting Review (October 1970), pp. 654-682.

_____, 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.

1-10.

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.713-725.

______,_______, dan_______, "The Incremental Information Content of Accrual Versus Cash Flows," The Accounting Review (October 1987), pp. 723-747.

Financial Accounting Standards Board, "Objectives of Financial Reporting by Business Enterprises," Statement of Financial Accounting Concepts No. 1 (1978), pp. 5183-5198.

______, "Statement of Financial Accounting Standards No. 95: Statement of Cash Flows (November 1987), pp. 1045-1088.

Gujarati, Damodar N., "Basic Econometries," Third Edition., McGraw Hill International Edition, New York, (1995) pp. 319-354.

Gombola, M. J., and J. E. Ketz, "a Note on Cash Flow and Classification Patterns of Financial Ratios," The Accounting Review (January 1983), pp. 435-452.

Ikatan Akuntan Indonesia, Standar Akuntansi Keuangan, Buku Satu. Salemba Empat, Jakarta, (1999), pp. 2.1-2.20.

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.125-136.

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 Cash-Flow Basis," The International Journal of Accounting: Education and Research (1980)

Lev, B., and J. A., Ohlson, "Market-Based Empirical Research in Accounting: A Review Interpretation, and Extention," Supplement of Journal of Accounting Research (1982), pp. 249-322.

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. 112-133.

Vasicek, O., "A Note on Using Cross-Sectional Information in Bayesian Estimation of Security Betas," The Journal of Finance (Desember 1973), pp. 1233-1239.

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. 165-200.

______, "The Incremental Information Content of the Accrual and Funds Component of Earnings After Controlling for Earnings," The Accounting Review (April 1987), pp. 293-322.

Appendix 1

Output White’s General Heteroscedasticity Test

With Independent Variable Earn_Beta

White Heteroskedasticity Test:
F-statistic	1,368180	Probability		0,263097
Obs*R-squared	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	t-Statistic	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
R-squared	0,047394	Mean dependent var		0,416705
Adjusted R-squared	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	F-statistic		1,368180
Durbin-Watson stat	1,689509	Prob(F-statistic)		0,263097

Appendix 2

Output White’s General Heteroscedasticity Test

With Independent Variable Fund_Beta

White Heteroskedasticity Test:
F-statistic	0,397643	Probability		0,673818
Obs*R-squared	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	t-Statistic	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
R-squared	0,014254	Mean dependent var		0,426550
Adjusted R-squared	-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	F-statistic		0,397643
Durbin-Watson stat	1,747735	Prob(F-statistic)		0,673818

Appendix 3

Output White’s General Heteroscedasticity Test

With Independent Variable Cash_Beta

White Heteroskedasticity Test:
F-statistic	1,736044	Probability		0,185732
Obs*R-squared	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	t-Statistic	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
R-squared	0,059380	Mean dependent var		0,380918
Adjusted R-squared	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	F-statistic		1,736044
Durbin-Watson stat	1,877709	Prob(F-statistic)		0,185732