Earnings Persistence and Market Reaction : Evidence from Korea

The main purpose of this paper is investigating the persistence of earnings and market reaction in Korean stock market. Most countries have adopted accrual accounting system to reflect correct financial conditions of business in reporting accounting information. But, because accrual accounting system can provide CEO of business with many opportunities to exercise discretions over the financial statements, the value relevance of accruals and cash flows have been main research concern in accounting field.


Introduction
The main purpose of this paper is investigating the persistence of earnings and market reaction in Korean stock market.Most countries have adopted accrual accounting system to reflect correct financial conditions of business in reporting accounting information.But, because accrual accounting system can provide CEO of business with many opportunities to exercise discretions over the financial statements, the value relevance of accruals and cash flows have been main research concern in accounting field.
Many prior research shows that earnings persistence is more attributed to cash flow than accruals components of earnings, but they also report that market participants usually overact more on accruals than earnings in real investing (Sloan, 1996;Subramanyam and Wild, 1996;DeFond and Park, 2001;Xie, 2001;Beneish and Vargus, 2002;Fairfield, Whisenant and Yohn, 2003;Hanlon, 2005;Pincus et al., 2005).
For example, Sloan (1996) reports that earnings persistence is more attributed to cash flows than accruals by using empirical model of Mishkin (1983).He also provides the evidence of accruals anomaly that market participants usually fixate on accruals in stock pricing.
Even though the persistence of earnings, cash flows and accruals have been investigated by many researchers in developed countries, Korean financial market has few evidence on the value relevance of earnings persistence and market participant's reaction on the cash flows and accruals components of earnings.This paper has two main purposes.First, the paper examines the persistence of earnings and its main components (cash flows and accruals) from current year and next year.Furthermore, the paper also tests whether Korean investors react more on cash flows or accrual components of earnings.
Empirical results of this paper show that Korean stock market has a high level of earnings persistence and this is more attributed to cash flows than accruals.The study also suggests that Korean stock market participants usually react more on cash flows than accruals.The results are consistent with prior studies.
The rest of the paper proceeds as follows.The next section discusses previous recent studies that investigate the persistence of earnings and its' components and market participants' reaction to it.Section 3 develops the assumptions and empirical models and data used in the study.Section 4 discusses the empirical results of the study regarding the persistence of earnings and its components and market participant's expectations and reactions on them.Section 5 concludes the study.

Literature Review
Following Dechow (1994) and Sloan (1996), the paper investigates the information content of earnings, cash flows, and accruals and market reaction on them.Dechow (1994) report that cash flows present the information of future earnings and cash flows.Dechow (1994) also show that current earnings have more information about future potential cash flows that cash flows large due to accrual components of earnings.
In line with Dechow (1994), Sloan (1996) also reports that the cash flow component of earnings is more persistent than accrual component of earnings.He shows that market investors usually fail to correctly distinguish the different level of persistence from both cash flows and accrual component of earnings.This is so called 'accrual anomaly' in securities returns.
For example, Subramanyam (1996) demonstrates that the stock market overprices accruals due to abnormal accruals.Xie (2001) demonstrates that stock market overestimates the persistence of earnings and it also overprices accrual components of earnings.Xie (2001) suggest that the overpricing of earnings and accrual components of earnings is largely attributed to managerial discretions.Hanlon (2005) also demonstrates that market participants constantly overprice the accrual component of earnings and underprice cash flows component of earnings.
In contrast to these previous researches, Dechow, Rechardson, and Sloan (2004) demonstrate that market participants correctly anticipate the persistence of earnings and cash flows components of earnings.
Other literatures in emerging markets report compelling results.For example Navissi, Mirza, and Yao (2006) also report that market participants in china fail to estimate correct information of earnings' persistence and they generally underestimate both accrual and cash flows component of earnings performance.But Na (2007) demonstrates that Korean investors usually overestimate the persistence of accrual component of earnings and they underprice accruals in the Korean stock markets.

Study Hypothesis
This paper investigates the information contents of earnings, cash flows, and accruals component of earnings performance and market reaction on them.To investigate whether earnings, cash flows and accruals component of earnings have a positive relationship to future earnings performance, this study tests the following hypotheses; Hypothesis 1 (H-1): Current earnings have a positive function of future earnings performance.
Hypothesis 2 (H-2): Current cash flows component of earnings has a positive function of future earnings performance.
Hypothesis 3 (H-3): Current accruals component of earnings has a positive function of future earnings performance.
And then the study examines market investors' reaction on the persistence information of current earnings, cash flows and accrual component of earnings performance.To test whether market participants overestimate the persistence of earnings, cash flows and accruals component of earnings, the paper test following hypotheses; Hypothesis 4 (H-4): Market participants overestimate the information of earnings' persistence Hypothesis 5 (H-5): Market participants overestimate the information of cash flows' persistence Hypothesis 6 (H-6): Market participants overestimate the information of accruals' persistence

Empirical Model for Hypotheses
This paper employs following empirical model used in prior researches (Freeman, Ohlson and Penman, 1982;Sloan, 1996). (1) Where, (next year operating performance) is defined as operating income scaled by total assets in year t.t is operating income in year t-1 (current year operating performance).and are the coefficients of main variables in this model.Equation ( 1) tests earnings persistence by estimating .
Prior literatures show that CEOs usually use discretionary accrual items to make desired earnings in accrual accounting system.These studies report that high degree accruals are perceived as less qualified than low degree of accruals and cash flows are perceived as more qualified and related to firm value.In the Korean stock market context, the assumption of prior studies also can be tested.So this paper examines whether earnings persistence is more attributed to accruals or cash flows by testing follow model. (2) Where, is total accruals in year t-1, and is operating cash flows in year t. and , , are the coefficients of this equation.Equation (2) measures the relative persistence of accruals and cash flows by estimating and .If is bigger than , earnings persistence is more attributed to cash flows than to accruals, it means that Korean investors rely more on cash flows than accruals components of earnings in expecting and anticipating future potential earnings.This paper also employs the empirical model of Mishkin (1983) and Sloan (1996) to test whether Korean investors use the information of earnings, cash flows and accruals persistence in estimating firm value.The empirical model employed in Mishin (1982) and Sloan (1996) is as following; (3) Where, is the holding returns for year t, is the expectation of the holding returns for year t. is earnings in year t and is the forecast of in year t. and is an earnings response coefficient.If represents earnings, equation (3) suggests that in the context of efficient market, the significance of current and next year's earnings persistence exist.
(3.1) Equation (3.1) indicates that if market is efficient, market participants correctly expect the relationship between earnings persistence and firm value, and then they cannot have any abnormal stock returns ( ).
This theoretical model can be restated as the following equation ( 4).The form of equation ( 4) enables the study to test whether Korean investors promptly react on earnings persistence or delay until next year's financial reporting. (4) If Korean investors fully anticipate the information of earnings persistence ( = ), market efficient inevitably exist in the Korean stock markets.This paper examines this assumption for both earnings and earnings components by using following equation ( 5). ( If the firm value is correctly expected by the persistence of cash flows and accruals, = and = in equation ( 5).This implies that stock mispricing does not exist and there are no abnormal securities returns.

Sample Selection and Data Source
The empirical test of this paper employs data from two sources.Insert Table 1

Insert Table 3 about here
This paper carries out yearly cross-sectional regressions to investigate earnings quality in the Korean stock markets.<Table 4> shows the regression analysis on the persistence of earnings over the period of 2000-2008.
The results show that the significant earnings persistence in all individual year regressions rejects the null hypothesis at the 1% level of significance.This indicates that there exists a significant relation between current and one year before earnings performance.The results also suggest that earnings are not transitory in the Korean stock markets.

Insert Table 4 about here
The study also carries out the nonlinear generalized least square regressions with yearly cross-sectional samples to investigate the market participants' reaction on the information of earnings persistence in the Korean security markets.
<Table 5> presents the regression analysis of investors' reaction on the information of earnings persistence over the period of 2000-2008.This paper does the nonlinear generalized least square regression of current abnormal returns on one year before earnings.The coefficients of are between 0.000 and 0.339 and the coefficients of are between 0.000 and 108.592.
The likelihood ratio statistics in the nonlinear generalized least square regression are significance at the 1% or 5% level of significance except 2000 and 2001.The parameter of the nonlinear generalized least square regression in 2005 and 2006 fail to converge after 100 iterations.<Table 5> indicates that the investors in Korean stock market fail to anticipate the information of earnings persistence for security prices.The results also suggest that Korean investors underestimate the persistence of earnings.

Insert Table 5 about here
The paper implements yearly cross-sectional regressions to investigate cash flows and accruals components of earnings performance in the Korean stock markets.<Table 6> reports the regression analysis on the relative persistence of cash flows and accruals components of earnings over the period of 2000-2008.The regression results indicate that the coefficients on accruals are between 0.256 and 0.539 and the coefficients on cash flows are between 0.138 and 0.967 at the 1% level of significance.
Most regression results present that the coefficients on accruals are bigger than that of cash flows except 2001.The results indicate that the accrual component of earnings is more persistent than the cash flows component of earnings.This also suggests that the persistence of earnings is significantly attributed more to accruals than to cash flows.

Insert Table 6 about here
The study conducts the nonlinear generalized least square regressions with yearly cross-sectional samples to investigate the market participants' reaction on the information of cash flows and accruals components of earnings persistence in the Korean security markets.
<Table 7> shows the regression analysis of investors' reaction on the information of cash flows and accruals persistence over the period of 2000-2008.This paper does the nonlinear generalized least square regression of current abnormal returns on one year before accruals and cash flows components of earnings.
The coefficients of are between -0.041 and 0.339 and the coefficients of are between -0.986 and 74.792.The likelihood ratio statistics( = ) in the nonlinear generalized least square regression are significance at the 1% or 5% level of significance except 2000, 2001 and 2004.The parameter of nonlinear generalized least square regressions in 2006 fails to converge after 100 iterations.
The coefficients of are between -0.929 and 0.499 and the coefficients of are between -3.258 and 35.058.The likelihood ratio statistics( = ) in the nonlinear generalized least square regression are significance at the 1% or 5% level of significance except 2000 and 2007.The parameter of nonlinear generalized least square regressions in 2006 fails to converge after 100 iterations.<Table 7> indicates that the investors in Korean stock market fail to anticipate the information of cash flows and accruals components of earnings persistence for security prices.The results also suggest that Korean investors usually underestimate the persistence of cash flows and accruals.
Insert Table 7 about here

Conclusions
This study investigates the persistence of earnings, cash flows and accruals and market reactions on them in the Korean financial markets over the period of 1999-2008.The paper also examines the efficiency of Korean stock markets by testing investors' reaction on earnings and its' components.
The empirical results of this paper show that while hypothesis 1 (Current earnings have a positive function of future earnings performance), hypothesis 2 (Current cash flows component of earnings has a positive function of future earnings performance), and hypothesis 3 (Current accruals component of earnings has a positive function of future earnings performance) are supported, Hypothesis 4 (Market participants overestimate the information of earnings' persistence), hypothesis 5 (Market participants overestimate the information of cash flows' persistence), and hypothesis 6 (Market participants overestimate the information of accruals' persistence ) are not supported.
From the test result of research hypotheses, the paper reports some evidences on the earnings persistence and market participant reactions in the Korean stock market.Similar to prior literatures, the empirical results of this paper indicates that a high level of earnings persistence exists in the Korean stock markets.But, dissimilar to prior studies the persistence of earnings more contributed to accruals than cash flows components of earnings.
The paper also examines whether Korean investors correctly react on the information of earnings and its components.The results suggest that Korean investors fail to use earnings persistence information and they usually underestimate both accruals than cash flows components of earnings.Subramanyam, K. R., and J. J. Wild.(1996).Going-concern status, earnings persistence, and informativeness.Contemporary Accounting Research, 13, 251-274. Xie, Hong. (2001).The Mispricing of Abnormal Accruals.The Accounting Review, 76, 357-373.= Abnormal returns at the end of fiscal year t, where year t is the event year; E t = Earnings in period t deflated by total assets of year t; CF t = Operating Cash Flows in period t deflated by total assets of year t; ACC t = Total number of shares outstanding in year t deflated by total assets of year t.
Financial statement data are obtained from the KIS-VALUE (Korea Investors Service-Financial Analysis System) and stock return data are obtained from KISRI (Korea Securities Research Institute) stock databases.These data sets span the 10-year period from 1999 to 2008.During the process of sample selection, the study includes firms with accounting earnings, cash flows, accruals and other financial data sufficient for empirical analysis, but the paper excludes financial banking & public business firms and impairment of capital firms on KIS-VALUE and KISRI database.Prior to the tests the paper eliminates outliers with Cook's Distance greater than 0.5 and absolute value of student residuals greater than 1. <Table 1> shows the selection procedure of sample firms over the period of 2000-2008 in the Korean stock markets.
about here <Table 2> shows descriptive statistics for firm-year sample firms used in this study.The paper identifies 13,383 firm-year observations for the period 2000-2008.Total means of (AR t ) is 0.129, and its highest value is 106.578.Total means of E t is 0.007; its highest value is 28.531.Total means of CF t is 0.043; its highest value is 1.215.Total means of ACC t is -0.036; its highest value is 30.901.pearson correlations for sample data used in this study.E t , CF t and ACC t are positively correlated at the 1% level, while AR t , E t , CF t , and ACC t are not significantly correlated.These correlations suggest that CF t and ACC t are related to E t , but AR t is not.

Table 1 .
Selection of sample firms

Table 4 .
Regressions Analyzing the Persistence of EarningsEquation: