National Sun Yat-sen University Institutional Repository:Item 987654321/33485

行政院國家科學委員會專題研究計畫 成果報告

我國企業盈餘管理行為之時間序列與橫斷面分析(2/2)

中 華 民 國 95 年 10 月 24 日

Understanding Earnings Management: A Time-series and

Cross-sectional Analysis

Abstract

Corporations operate under state charters and subject to rules and regulations of governing bodies. This study investigates effects of rules and regulations issued by the governing bodies on earnings management. The evidence shows decreases in earnings management as the volume of rules or regulations increases. Increases in rules or regulations appear to have constrained managers from taking discretionary actions in choosing accounting methods. Not all firms, however, display similar earnings management behavior. Managers of either mature or stagnant firms are likely to manage earnings upward while managers of growth firms are indifferent between upward and downward earnings management.

Introduction

Earning management compromises financial reporting. To improve financial reporting quality regulatory bodies on accounting principles and procedures have been attempting to contain and eliminate earnings management. Successes in these efforts require a clear understanding on rationales or motivations underlying earnings management. The literature has identified several of these rationales or motivations including desires for better performance evaluation and/or higher compensation (Healy, 1985; Holthausen et al., 1995), enhancing job security (Fudenberg & Tirole, 1995), complying with terms of loan or bond covenants (DeFond & Jiambalvo, 1994; Sweeney, 1994), and securing favorable transaction terms with stakeholders (Bowen et al., 1995).

Until all earnings management cease, users of financial statement need to be able to recognize effects of earnings management on reported financial data and make proper adjustments in using these figures for the intended purposes and regulators need to issue proper and effective rules and regulations to hold or prohibit earnings management. These objectives can be achieved only if users and regulators have a good and clear understanding of earnings management behavior, including rationales for and characteristics conducive to earnings management.

Firms are required to follow rules and regulations issued by regulatory bodies in the financial reports they issued. Regulatory bodies issue rules and regulations to, among other intended effects, limit choices and discretionary powers of managers on accounting methods and procedures, reduce earnings management, and increase comparability of financial statements (Schipper, 1989). Although studies have reported that financial reporting has become more conservative over the years (Givoly & Hayn, 2000; Holthausen & Watts, 2001), no research has been reported on effects of increases in rules and regulations on earnings management and quality of financial reports. A good understanding on effects of rules and

regulations and the way in which rules and regulations improve financial reporting quality is an important step toward better regulations and improved financial reports.

In examining effects of accounting rules and regulations on earnings management we need to understand the characteristics, operating environment, strategic objectives, among others, of the firms. Studies have shown that firms evolve through life-cycle stages.1 Firms at different life-cycle stages often exhibit different financial characteristics, require disparate management skills, and have dissimilar priorities and strategies (Porter, 1980). Furthermore, earnings levels and volatilities are likely to be different among firms at various life-cycle stages (Black, 1998; Martinez, 2003). Earnings levels and volatilities are two of the important factors that affect assessment of firms by investors, analysts, lenders, or other users of financial reports and, as a result, may affect earnings management behavior. Furthermore, the literature has documented reported earnings to have effects on managers’ performance evaluations and compensations (e.g., Fudenberg & Tirole, 1995; Skinner, 1993). Managers manage reported earnings to attain favorable performance evaluations and/or compensations (Fudenberg & Tirole, 1995; DeFond & Park, 1997). Since firms at different life-cycle stages are likely to have different earnings characteristics (level and/or volatility), the desires for and types of earnings management may differ among firms at different life-cycle stages or over time by the same firm. A better understanding on the differences, extents, and directions of earnings management of firms at different life-cycle stages is a first step toward better rules and regulations to eliminate or contain earnings management.

This study assesses effects of increases in rules and regulations on earnings management of publicly listed firms in the Taiwan Stock Exchange over the period of 1987-2002. We use discretionary accruals (DA hereafter), determined based on the modified Jones model, as the proxy for earnings management and life-cycle descriptors of Anthony and Ramesh (1992) to identify life-cycle stages of sample firms in each of the years to gain a better understanding of earnings management behavior.

Empirical results support the hypothesis that earnings management decreases as the number of rules and regulations issued by the regulatory body increases. The results also show firms at different life-cycle stages displaying significantly distinct earnings management behavior. Managers of mature or stagnant firms are more likely to manipulate earnings upward while managers of growth firms tend to be indifferent between upward and downward earnings management.

The remainder of the paper contains four sections. Section II reviews relevant literature and develops hypotheses to be tested. Section III discusses the methodology employed to examine the research question. Section IV presents the results of the study. Section V concludes this paper and discusses implications and limitations of the study.

Literature Review and Hypothesis Development

Earnings Management

Reported earnings often play important roles in determining firm values and performance evaluations of managers. Managers benefit from having high reported earnings, either directly or indirectly (Sunder, 1997). Negative earnings surprises, on the other hand, often have adverse effects on stock prices (Brown et al., 1987) and images (Matsumoto, 2002) of the firms. Schipper (1989) points out that direct financial stakes in reported numbers and other incentives often prompt managers to manipulating reported earnings upwards and avoiding unfavorable earning surprises.

Following Schipper, we consider earnings management as “a purposeful intervention in the external financial reporting process, with the intent of obtaining some private gains” (Schipper, 1989, p. 92). Earnings management often implies some misdeeds, misdemeanors, or other transgressions that alter earnings to be reported (Kirschenheiter & Melumad, 2002). Altered earnings mislead and deceive users of financial reports. Dye (1988) suggests external demand to increase stock prices and internal demand to attain optimal contracting among

reasons (incentives) for firms to engage in earnings management. He shows that as long as accounting data are used in compensation contracts, incentives exist for managers to manage accounting data. Other studies, though slightly different in their terminologies, suggest similar reasons for earnings management - capital market motivation and contract motivation (Healy, 1985; DeFond & Jiambalvo, 1994; Teoh et al., 1998a; Burgstahler & Eames, 1998; Kasznik, 1999; Abarbanell & Lehavy, 2003).

Capital Market Motivation.

Financial statements play important roles in capital markets and investment decisions. Anthony and Ramesh (1992) find stock prices responding to sales growth and investments. Many studies have shown that investors’ actions, as reflected in stock prices, respond to reported earnings (DeAngelo et al., 1996; Barth et al., 1999). Teoh et al. (1998a; 1998b) find managers overstating earnings prior to public offerings, especially at the time of initial public offerings (IPO) – a result of apparent belief on the important roles earnings play in determining share prices.

Several studies also find similar relationships between earnings and analyst expectations (Burgstahler & Eames, 1998; Abarbanell & Lehavy, 2003). Analysts often raise their predictions of stock prices when firms report higher earnings and lower their predictions when firms report “surprised” lower earnings. Furthermore, analyst expectations often become benchmarks in earnings manipulations. Burgstahler and Eames (1998) find managers manipulating earnings upward to avoid reporting earnings below the analysts’ consensus expectation. Kasznik (1999) finds managers increase DA to avoid reporting of earnings below the forecasted amount.

Contract Motivation.

lenders, suppliers, and customers (Bowen et al., 1995). Desires to secure favorable contracts with stakeholders and terms specified in debt covenants requiring the debtor firm to meet certain financial thresholds may prompt managers to manage reported earnings. DeFond and Jiambalvo (1994) report that managers are more likely to make income-increasing accounting choices when their firms are close to violating debt covenants. Sweeney (1994) finds significant number of covenant violators making income-increasing accounting choices immediately after the violation.

Not all earnings management increases reported earnings. DeFond and Park (1997) find managers engaged in shifting earnings to the future. Several writers show managers engaging in income-decreasing earnings management to “save” earnings for the future when the expected current earnings exceed that of the upper bound for bonus (Healy, 1985; Holthausen et al., 1995)2.

Regulatory Actions and Earnings Management

Regulations and official guidelines such as GAAP, legal rules, and opinions of regulatory agencies have restraining effects on discretionary actions that affect financial reporting (Schipper, 1989). Researchers have documented effectiveness of pronouncements, rules, or opinions in restricting managers’ choices of accounting methods and discretions on reported earnings (Schipper, 1989). Several studies document that financial reporting has become more conservative and choices of accounting methods and procedures have become more restricted over the years (Givoly & Hayn, 2000; Holthausen & Watts, 2001). This is particularly true after the spectacular failures and scandals of several large U.S. firms including Enron, Tyco, WorldCom, and HealthSouth. These well-publicized fiascos raised public’s attention and scrutiny on corporate governance. Researchers have documented increasing conservatism in financial reporting of U.S. firms after the Sabona-Cox act, which was passed after the aforementioned fiascos. Increasing conservatism over the years also

limits managers’ discretion on the interpretation and choice of accruals and other items. These evidences lead to the following hypothesis, stated in alternative form:

Hypothesis 1: Earnings management decreases as rules and regulations

increase.

Firm Life Cycle and Earnings Management

Operating environments such as market growth, investment opportunity, and competitions often differ for firms at different life-cycle stages. Firms at different life-cycle stages are also likely to have different operating characteristics such as strategy, cash flow pattern, financial need, requisite management skill, priority, and profitability. A growth firm, for example, may emphasize continuous revenue growths, increases in market share, or both (Porter, 1980; Spence, 1977; Karnani, 1984; Wernerfelt, 1985; Hong et al., 2002). A firm at a mature stage may focus on garnering positive net cash inflow, sustaining cost cuttings, maintaining stable market share, and sustaining dividend, among others - aspects of operations that differ from those of major concerns of growth firms. Capital markets also react differently to the same event for firms at different life-cycle stages. Anthony and Ramesh (1992) find that the reaction of the stock market to unexpected sales growth or unexpected capital expenditure is the strongest for growth firms and the weakest for stagnant firms3. Several studies find firms at different life-cycle stages to have engaged in dissimilar activities and reported disparate financial performances (White et al., 2003). Selling and Stickney (1989) find life-cycle stages useful in understanding variations of ROA over time and across companies.

White et al. (2003) argue that a company’s financial performance changes as it passes through life-cycle stages. The amounts and importance of cash flows from operating, investing, and financing activities differ among firms at different life-cycle stages (Stickney & Weil, 1996). Savich and Thomposon (1978) find life-cycle stages affect components of

balance sheet, income statement, and statement of cash flows. Consequently, firms at different life cycle stages may manage earnings differently, if they are so incline.

A growth firm is likely to stresses sales increases and experience fluctuating income levels and tightness in cash from operations at times owing to constant increases in operating expenditures to meet continuous growth and/or the needs or commitments in new investments to sustain sales growth. In contrast, a mature firm is likely to enjoy stable and positive net income and cash flows as its market stabilized and needs for and commitments to new investment decreases. Stagnant firms, facing stiff competition and few new investments, are likely to have reduced sales, profits, or both and have low or even negative earnings. Nevertheless, stagnant firms frequently enjoy positive cash flows because of low investment needs and cash inflows from disposal of properties, plants, and equipments no longer needed in operations.

Growth Firm and Earnings Management

Having increases in earning, growth firm managers may take discretionary actions to reduce reported earnings if they perceive current earnings to be excessive as observed by AlNajjar and Riahi-Belkaoui (2001). They attribute the tendency for growth firms with large investments to elect accounting choices that reduce reported earnings to the desire to reduce the perceived excessive returns, especially if the growth firm is also deemed to possess monopolistic power. A growth firm deemed as a monopolist may manipulate earnings downward to avoid paying political costs (AlNajjar & Riahi-Belkoui, 2001).

Growth firms are constantly in need of funds for new investments, expansion, and other needs in operations, which makes preserving funds a continuous concern for growth firms and a rationale for many growth firms to reduce reported earnings. Low reported earnings afford growth firms justifications not to increase, or even reduce, cash dividends (DeAngelo et al. 1994). Reduced dividends increase funds available. Thus, growth firms with increasing

earnings may elect to adopt accounting choices to lower reported earnings.

Growth potentials often are the most important determining factor for market values of growth firms. Near-term earnings, especially those of the current period, are likely to play a secondary role in market valuation of growth firms (Black, 1998). Low current earnings, though undesirable, are likely not to have significant negative impacts on market capitalizations of growth firms. Thus, a growth firm is more likely to manage its earnings downward, rather than upward, if it manages its reported earnings, especially when the growth firm has a large amount of new and intangible assets that increase their market values (Collins et al., 1992).

Abundant investment opportunities often increase information asymmetries between managers and shareholders (Smith & Watts, 1992; Bizjak et al., 1993). To mitigate potential agency problems as a result, growth firms often have compensation contracts that include long-term rewards such as stock options or restricted stock grants (Smith & Watts, 1992; Gaver & Gaver, 1995)4. Compensation contracts based primarily on stock options or restricted stock grants as often seen in growth firms encourage longer-term perspectives and reduce incentives for managers to engage in earnings management to raise short-term earnings.

We test the downward earnings management in two sub-hypotheses. The first sub-hypothesis tests directions while the second sub-hypothesis examines the magnitudes of earnings management. These sub-hypotheses stated in alternative form are:

Hypothesis 2a: Growth firms are more likely to manage earnings downward than upward.

Hypothesis 2b: Growth firms have larger amount of downward than upward earnings management.

If the evidence rejects Hypotheses 1 and 2a or 2b, we further test the effect of increases in rules and regulations on earnings management of growth firms, as stated below in

alternative forms.

Hypothesis 2c: The frequency of growth firms engaging in downward earnings management decreases as rules and regulations increase.

Hypothesis 2d: The magnitude of downward earnings management of growth firms decreases as rules and regulations increase.

Mature Firm and Earnings Management

In most instances, investment activities of mature firms are less intensive and their operating environments are more stabilized than those of growth firms (Balkin and Gomez-Mejia, 1987). Firms operating in these environments often use earnings-based bonus plans constituting primarily of fixed-pay components (Skinner, 1993) and implement their plans with formal performance evaluation processes (Balkin & Gomez-Mejia, 1987). Consequently, accounting numbers, especially reported earnings, play important roles in performance evaluation processes of mature firms (Smith & Watts, 1992).

Limited growth opportunities available to mature firms often shift the weight for determining market value of mature firms toward earnings (Black, 1998). Mature firms can ill-afford to reduce reported earnings as mature firms that reported low earnings often see their firm values curtailed. To maintain or increase firm values, mature firms need to have high earnings. Furthermore, important roles that accounting earnings often play in performance evaluations may inadvertently encourage earnings management. The desires to maintain or increase market value and to gain favorable performance evaluations may prompt managers of mature firms to manipulate earnings upward. The third hypothesis to be tested, stated in alternative form, is:

Hypothesis 3a: Mature firms engage in more upward than downward earnings management.

earnings managements.

If the evidence rejects Hypotheses 1 and 3a or 3b, we further test the effect of increases in rules and regulations on earnings management of mature firms, as stated below in alternative forms.

Hypothesis 3c: The frequency of mature firms engaging in upward earnings management decreases as rules and regulations increase.

Hypothesis 3d: The magnitude of upward earnings management by mature firms decreases as rules and regulations increase.

Hypothesis 4 – Stagnant Firm and Earnings Management

Most likely, firms operating in a stagnant environment have limited investment opportunities (Stickney & Weil, 1996; Black, 1998) and earnings play an important role in determining firm values. Furthermore, a typical reward structure of firms operating in a stagnant business environment is the predominance of fixed salaries and bonuses in its reward or performance evaluation system, with accounting numbers play important roles in determining rewards or performance evaluation (Smith & Watts, 1992; Gaver & Gaver, 1995). Significant roles accounting numbers play in either the determination of firm values or rewards of managers increase the likelihood for managers of stagnant firms to adopt income-increasing accounting procedures to raise firm values as well as their compensations. Furthermore, having low or negative earnings that stagnant firms are likely to have experienced, accounting choices are likely to have large impacts on reported earnings (Black, 1998).

Operating in a stagnant business environment with deteriorating earnings also increases the likelihood of violating debt covenants. To avoid or delay violations of debt covenants, stagnant firms may resort to choices of accounting policies or procedures. DeFond and Jiambalvo (1994) and Sweeney (1994) find many firms adopted income-increasing

accounting choices around the time of debt-covenant violations. Peltier-Rivest (1999) notes that firms not in financial distresses but with high debt ratios manipulate earnings upward in the year of covenant violation in attempting to improve upcoming negotiation positions with lenders.

These evidences suggest that stagnant firms are likely to engage in upward earnings management. We test the following hypothesis, stated in alternative form.

Hypothesis 4a: Stagnant firms engage in more upward than downward earnings management.

Hypothesis 4b: Stagnant firms undertake larger amount of upward than

downward earnings management.

We further test the effect of increases in rules and regulations on earnings management of mature firms, as stated below in alternative forms.

Hypothesis 4c: The frequency of stagnant firms engaging in upward earnings management decreases as rules and regulations increase.

Hypothesis 4d: The magnitude of upward earnings management by stagnant firms decreases as rules and regulations increase.

Life Cycle Stage and Discretionary Accrual

The underlying reasons for the hypotheses above also suggest that the amount of DAs is likely to be the lowest for growth firms, higher for mature firms, and the highest for stagnant firms.

Hypothesis 5a: The discretionary accruals are the lowest for growth firms and the highest for stagnant firms.

Increases in rules and regulations would decrease differences in the magnitudes of DAs among firms at different life-cycle stages. We tested the following hypothesis:

different life-cycle stages decrease as rules and regulations increase.

Research Method

This study uses discretionary accruals as a proxy for earnings management and examines firms at various life-cycle stages. This section explains first estimation of discretionary accrual and determination of life-cycle stages and discusses statistical methods, and sample selection procedures for the empirical analyses conducted in this study.

Discretionary Accruals

Researchers have used several different measures to gauge effects of accounting choices on reported earnings. Some focus on the choice of a single account. Others examine net effects of multiple accounts. McNichols and Wilson (1988) examine the effect of accounting choices for allowance for bad debts on reported earnings. Sweeney (1994) investigates managers’ responses to violations of debt-covenants. Becker et al. (1998) conclude that DA captures the net effect of accounting choices on reported earnings and is the best measures to seize the net effect of accounting choices. Healy (1985), DeAngelo, (1986), and Jones (1991) use DA to capture effects of multiple accounting choices and document significant positive associations between DA and earnings manipulation. Following these studies, this study also uses DA as the proxy for effects of earnings management.

Different studies have used different models to measure DAs. Many studies, however, have concluded the modified cross-sectional Jones model to be the best method for measuring earnings management (Dechow et al. 1995; Bartov et al., 2001). Thereby this study uses the modified cross-sectional Jones model, as stated in Equation (1), to estimate the year-specific parameters, a1、a₂ and a₃:

TAit/Ait-1= a1 (1/ Ait-1) + a2 (△REVit/ Ait-1) + a3 (PPEit/ Ait-1) + εit (1) Where,

TAit = total accruals of sample firm i in year t

Ait-1 = total assets of sample firm i at the end of year t-1 △REVit = change in net revenues of sample firm i in year t

PPEit = gross property plant and equipment of sample firm i at the end of year t εit =

A firm’s total accrual for a given period is the difference between the firm’s net income and the total operating cash flows of the period. Equation (1) states that a firm’s total accrual for a given period is a function of the change in total revenue from the previous period, the total gross amounts of property, plant, and equipment at the end of the current period, a constant term, and an error term. All measures in the equation, except the error term, are deflated by the total assets at the end of the previous period.

The discretionary accrual (DA) is the difference between the total accrual and the non-discretionary accrual:

DAit = TAit －NDAit (2)

A firm’s non-discretionary accrual (NDA) of a given period is a function of the change of revenue from the previous period and the amount of total gross property, plant, and equipment of the current period, as stated in the following model:

NDAit = α1 (1/ Ait-1) +α2 [(△REVit －△RECit) / Ait-1] + α3 (PPEit/ Ait-1) (3) Where,

NDAit = total non-discretionary accruals for sample firm i in year t △RECit = change in net receivables for sample firm i in year t

Life-Cycle Stages

Anthony and Ramesh (1992) show that firm life-cycle stages can be identified using three descriptors: dividend payout ratio (DP), sales growth percentage (SG), and firm age (AGE). Using these descriptors in a two-step procedure this study estimates the life-cycle stage of each firm in each of the sample years. First, we use each of the life-cycle descriptors in a univariate procedure to identify the life-cycle stage of a firm in each of the sample periods. In the event that not all three descriptors yield consistent classifications for a firm-year, we use the composite score from a multivariate procedure that incorporates all three descriptors to determine the life-cycle stage of the firm-year. Black (1998) also uses a similar multivariate classification procedure to identify characteristics of firms at different life-cycle stages. Table 1 presents the scheme for life-cycle classification.

[Insert Table 1 here]

Empirical Test

Earnings may be managed upward or downward. To prevent earnings managed in opposite directions from canceling out each other and shows, erroneously, no or a minimal perceptible effect on DA, this study uses absolute values of DA to measure effects of earnings management.

We use two tests to examine differences in DA of the sample firms at different life-cycle stages: Chi-square and Student-t tests. The Chi-square tests examine frequencies of upward and downward earnings management while t-test investigate differences in magnitudes between upward and downward earnings management of firms in each of the life-cycle stages. In addition, we also examine differences in DAs between firms at different life-cycle stages.

Previous studies have found several factors to have effects on DA (Becker et al., 1998; Krishnan, 2003). To eliminate or minimize confounding effects these factors may exert on

DAs and the test results in our examinations of effects of rules and regulations and life cycle stages, we incorporate these factors in statistical tests. Reynolds and Francis (2000) show DA to associate significantly with firm size. We include the log of total assets of each of the sample firms as a proxy for firm size to control size effect. DeFond and Jiambalvo (1994) and Sweeney (1994) document DA to correlate significantly with leverage. We use leverage ratios, defined as total debts over total assets, as the control variable for differences in leverage among sample firms. Balsam et al. (2003) report increased likelihoods for firms with large total accruals to also have high DAs. We include the absolute value of total accruals to control the size effect of total accruals. DeFond and Subramanyam (1998) observe that auditor changes often have a significant, positive correlation with magnitudes of DAs. The two years around the time of change have the most effects: the last year of the old auditor and the first year of the new auditor. The effect before the change, however, is not always the same as that of after the change. This study includes a dummy variable to denote the last year before auditor change and another dummy variable to represent the first year of a new auditor. The literature reports that DAs often differ with differences in auditor size (Becker et al., 1998; Krishnan, 2003). We use a dummy variable to represent whether the auditor is a Big 6 or a Non-Big-6 firm. In addition, firms may engage earnings management through either issuing new equities or repurchasing outstanding shares (Teoh et al., 1998a; Becker et al., 1998). Following Becker et al. (1998), the regression model includes two dummy variables to control changes in outstanding shares of more than 10 percent during the year, one for increases and the other for decreases. The following is the final regression model:

DAit = β0 + β1X1it+ β2X2it + β3SIZEit+ β4LEVit + β5ABSTAit + β6ShareIncrit + β7 ShareDecr it + β8 OldAud it + β9 NewAud it + β10 B6it + εit (1) Where:

DAit = total discretionary accruals of sample firm i in year t

otherwise.4

X2it = dummy variable, 1 if the sample firm-year is at a stagnant stage and 0 otherwise.5

SIZEit = log of total assets of sample firm i at the end of year t

LEVit = total debts of sample firm i at the end of year t divided by the total assets of the same firm-year

ABSTAit = absolute value of total accruals for sample firm i in year t

ShareIncr it = dummy variable, 1 if the total outstanding shares of sample firm i increases by more than 10 percent during year t, 0 otherwise.

ShareDecrit = dummy variable, 1 if the total outstanding shares of sample firm i decreases by more than 10 percent during year t, 0 otherwise.

OldAud it = dummy variable, 1 for the last year before an auditor change, 0 otherwise.

NewAud it = dummy variable, 1 for the first year with a new auditor, 0 otherwise. B6it = dummy variable, 1 if the auditor is a Big 6 auditor, 0 otherwise.

Sample Selection

The sample firms are publicly listed firms (including OTC) in the Taiwan Stock Exchange that have data available in the Taiwan Economic Journal (TEJ) data bank,6 excluding firms in banking, insurance, or security industries7. For the time period included in the time-series test, 19878 to 2002, 154 firms have complete data required for this study. The cross-sectional analyses use data from 1992 to 2002 to allow for the most firms with complete data. Since the determination of life-cycle stages requires three years data prior to the first testing year, the testing period is further narrowed to 1995 to 2002. After deleting firms with incomplete data, 264 firms were selected for the cross-sectional analysis. Table 2 reports industry classifications of the sample firms. The sample firms come from seventeen

industries and all but two industries, electronics and textile, have less than 7 percent of the sample firms.

[Insert Table 2 here]

Results

Time-series Result

Figure 1 shows that the absolute values of means and medians of DA of the 154 sample firms have decreased over the years 1987 to 2002. The year 1989 has the highest means and medians of DA. Both means and medians decrease gradually from 1989 to 1992, pick up gradually from 1992 to 1994, and go down again in each of the years after 1994, except in 1997. The general decreasing trends of the absolute mean and median values of the DA suggest decreases in the extent of earnings management as rules and regulations increase over the years, as stated in the alternative form of Hypothesis 1.

[Insert Figure 1 here]

A one-time event might have led to the spike of DAs in 1997. The government introduced a new imputation tax in 1997 to be levied starting January 1998. The newly enacted tax law led many firms to defer 1997 earnings in attempts to reduce tax burden (Lin et al., 2004). Thus, the observed spike in 1997 might represent an abnormal phenomenon.

Descriptive Statistics

Table 3 presents summary statistics of life-cycle descriptors of firms identified by life-cycle stages. Both MDP and AGE increase while MSG decreases as firms shift from growth, to mature, and to stagnant stages. Anthony and Ramesh (1992) also report similar observations.

[Insert Table 3 here]

Table 4 reports means, medians, quartiles, and standard deviations of each of the variables included in model (1) for the 2,112 firm-year observations (264 sample firms from 1995 to 2002). Of the 2,112 observations, 27% are from growth firms (X1), 24.4% from stagnant firms (X2), and 48.6% from mature firms. The average DA is 2.1% of the total assets. The mean firm size as measured by log of total assets (6.885), average leverage ratio (40.6%), and average absolute value of total accruals (6.6%), are all approximately the same as those of all publicly listed firms. On changes in outstanding shares, approximately 36.3% of the sample firms increase by more than 10 percent (ShareIncr) while only 2.7% decrease by more than 10 percent (ShareDecr) of their outstanding shares. Of the entire sample firm-years, 2.5% changed the incumbent auditor (OldAud) in the year following the year of observation and 2.7% have new auditors in the year of observation. The Big 6 auditors (B6) audit 75% of the sample firms.

[Insert Table 4 here]

Table 5 reports correlation coefficients between variables. The lower left corner presents Pearson correlation coefficients and the upper right corner reports non-parametric Spearman rank correlation coefficients. The number in parentheses represents the significant level of the correlation coefficient.

All correlation coefficients between the dependent variable, DA, and each of the independent variables are statistically significant at .10 or lower, except those of the independent variables SIZE and OldAud in both the Pearson and the Spearman correlations and SharDecr in the Spearman correlation. The correlation coefficients between firms at growth stages (X1) and DAs are negative and statistically significant (p<.001), suggesting that growth firms and DAs move in opposite directions – an indication that growth firms manage DAs downward. These results reject the null hypothesis and support the direction prescribed in Hypothesis 2a. In contrast, the correlation coefficients between firms at stagnant

stage (X2) and DAs are significant (p<.001) and positive, indicating that stagnant firms and DAs move in the same direction - an indication that stagnant firms manage earnings upward. The results are in accordance with the direction stated in Hypothesis 4a.

The correlation coefficients between LEV and DA are negative and statistically significant (p<.001), suggesting that firms with higher leverages tend to have lower DAs, a result differs from those of DeFond and Jiambalvo (1994) and Peltier-Rivest (1999)9. The significant positive correlation coefficients between the absolute values of total accruals and DAs indicate that firms with higher absolute values of total accruals also have higher DAs, as Balsam et al. (2003) suggested. The correlation coefficients between DA and ShareIncr are positive and statistically significant (p<.001) in both Pearson and Spearman correlations. The Pearson correlation coefficient between DA and SharDecr is positive and statistically significant at, however, a lower lever (p<0.087) while the Spearman correlation is not statistically significant10. These results suggest that changes in the number of outstanding shares (in either direction) are positively correlated with DA, as concluded in DeFond and Subramanyam (1998). The significant negative correlation coefficient for NewAud (p<0.002) says that firms in their first year with a new auditor tends to have a lower DA than those in other years. The significant negative B6 correlation coefficient suggests that clients of the big 6 auditors have lower DAs than clients of non-big 6 auditors, a result consistent with the findings of Becker et al. (1998) and Krishnan (2003).

[Insert Table 5 here]

Table 6 reports the means of DAs for firms at different life-cycle stages in each of the sample years and for the entire sample period, grouped by the directions of the DAs (upward and downward). In general, growth companies have higher downward DAs than either mature or stagnant firms. Over the eight-year sample period (1995-2002) the average

downward DAs are 0.079, 0.062, and 0.039 for firms at growth, mature, and stagnant stages, respectively. The differences in means for firms at different stages are statistically significant11. Similar phenomena also are observed for upward discretionary accruals for firms at different stages, from 0.086, to 0.079, and to 0.068 for firms at growth, mature, and stagnant stages, respectively12. The differences in means of upwardDAs are also statistically significant. Overall, growth firms have the highest discretionary accruals (earnings management), followed by mature firms, and stagnant firms have the lowest discretionary accruals in each of the sample years and in total. These results are in the opposite to the direction stated in the alternative form of Hypothesis 5a.

[Insert Table 6 here]

Table 7 describes the frequencies (in percentages) of firms at each of the life-cycle stages engaged in upward and downward earnings management for each of the years and for the entire period. Of the firm-years observed to have earnings management, more growth firms (46.8%) made downward earnings management than firms at the other two stages, followed by mature firms (39.9%), and then stagnant firms (32.9%). The frequencies of firms at different life-cycle stages having upward discretionary accruals, however, are in the opposite direction. Stagnant firms have the highest percentage of upward earnings management (67.1%), followed by mature firms (60.1%), with growth firms (53.2%) the last. Similar patterns also are observed in each of the years. The differences between upwards and downwards earnings management for growth firms are not statistically significant and fail to reject the null Hypothesis 2a. The results show significant differences for firms at both mature and stagnant stages, however, as stated in the directions prescribed in Hypotheses 3a and 4a.

more growth firms with upward earnings management than those with downward management in each of the early years from 1995 to 1998. For example, in 1995, 78.7% of the growth firms have upwards DAs while 21.3% have downwards DAs. These patterns are opposite of the prediction prescribed by Hypothesis 2a. Starting 1999, the pattern reversed, however, and follows the direction stated in Hypothesis 2a. Similar patterns reversing are also observed for firms at mature and stagnant stages. The patterns in each of the years from 1995 to 1999 for both mature and stagnant firms show the patterns called for in both Hypotheses 3a and 4a. Starting 2000, however, the patterns reversed.

[Insert Table 7 here]

Table 8 cross-tabulates frequency percentages between earnings management and life-cycle stages. The Pearson Chi-square, 21.775, is statistically significant at p < 0.000. Post hoc tests (Appendix) show that both percentage differences in upward discretionary accruals between growth (53.2%) and mature (60.1%) firms and between mature (60.1%) and stagnant (67.1%) firms are statistically significant at p<0.05. Likewise, both percentage differences in downward discretionary accruals between growth (46.8%) and mature (39.9%) firms and between mature (39.9%) and stagnant (32.9%) firms are statistically significant at p<0.05.

Of firms that engaged in downward earnings management, growth firms have the highest adjusted residual value (3.8, p < .01)13. This suggests that the downward earnings management phenomenon is more perspicuous for growth firms than firms at either the mature or stagnant stage. Similarly, among firms that engaged in upward earnings management, stagnant firms have the highest adjusted residual value (3.8, p < .01). This suggests that the upward earnings management phenomenon is more lucent for stagnant firms than those at the other two stages.

[Insert Table 8 here]

Table 8 also shows 53.2% of the growth firms engaged in upward earnings management, while 46.8% practiced downward earnings management. The Chi-square value (2.398) reported in Table 9, however, shows the difference is not statistically significant (p = 0.122) and fails to support the alternative form stated in Hypothesis 2b. In contrast, the upward earnings management frequency percentages for firms at both the mature and the stagnant stages are significantly higher than those of downward earnings management (p<0.000 for both). These results follow the directions described in Hypotheses 3b and 4b.

[Insert Table 9 here]

Table 10 shows that the difference in DA between upward and downward earnings management for growth firms is not statistically significant (p > 0.1). Again, the result fails to reject the null Hypothesis 2b. Both mature and stagnant firms, however, engage in significantly higher upward earnings management than downward earnings management. These results support the predicted directions of both Hypotheses 3b and 4b.

[Insert Table 10 here]

ANOVA results reported in Table 11 show significant differences in DAs among firms at different life-cycle stages at p < 0.001 (F-value = 7.503). The Scheffe’s post hoc test shows that the DAs of both mature firms and stagnant firms are significantly higher than those of growth firms (p = 0.040 and 0.001, respectively). However, the DAs of stagnant firms are not significantly higher than those of mature firms (p = 0.178).

[Insert Table 11 here]

The univariate analyses reported above examine only one independent variable at a time. Other, unexamined, factors may have contributed to the observed results. To control for potential confounding effects other factors may exert on DAs, we conducted a multivariate analysis that includes all potential confounding variables identified earlier in the literature section. Table 12 presents the OLS regression results. The F-value, 27.912, is statistically significant (p<0.000). The significant and negative X1 coefficient, which signifies growth firms, suggests that growth firms have significantly lower DAs than those of mature firms. The significant and positive X2 coefficient, the dummy variable representing stagnant firms, suggests that stagnant firms have significantly higher DAs than those of mature firms. Overall, growth firms have the lowest DAs, followed by mature firms, and stagnant firms have the highest DAs. The result is consistent with the alterative form stated in Hypothesis 5a.

[Insert Table 12 here]

Tables 13 through 16 present results of logistic regressions on frequency and magnitude of earnings management over the years for mature and stagnant firms14. The variable Year in all of the results is significant and negative, indicating that the frequency and magnitude of earnings management decrease as time passes (rules and regulation increases), as stated in Hypotheses 3c, 3d, 4c, and 4d.

[Insert Table 13 through 16 here]

management over the years between firms at different life cycle stages. The results fail to show statistically significant difference as stated in Hypothesis 5b, except for the mean difference in mean discretionary accruals between stagnant and growth firms.

[Insert Table 17 and 18 here]

The collinearity diagnostics indicate that all variance inflation factor (VIF) values of the variables are less than 1.3, suggesting that the multicollinearity is likely not an issue for the regression model.

Conclusion

This study investigates effects of increases in rules and regulations on uses of accounting choices in earnings management. Using DA as the proxy for earnings management the result shows that, over a fifteen-year period (1987-2002), the amount of absolute value of DA decreases as the number of accounting rules and regulations issued by regulatory agencies or institutions increases. This suggests that increases in rules and regulations reduce uses of discretionary accounting choices in earnings management and accounting rules and regulations issued by accounting governing bodies and regulatory agencies are effective in reducing earnings management.

As long as firms manage their reported earnings, users of financial reports including investors, financial analysts, and lenders need to adjust or allow for earnings management in using financial data in decisions to minimize erroneous decisions or undesirable results. Not all firms, however, engage in similar earnings management (in either direction or magnitude) and different adjustments may be needed for different firms. This study also examines earnings management behavior for firms at different life-cycle stages and finds significant differences in earnings management behavior among firms at different life-cycle stages.

Mature or stagnant firms are likely to manage earnings upward, rather than downward. Nevertheless, both the frequency and magnitudes of earnings management decrease for firms at mature or stagnant as accounting rules and regulations increase.

This study examines only publicly traded companies. Results may differ for non-publicly held companies. All sample firms are located in Taiwan. Firms operate in different cultures or under different regulations may not display similar behavior. The sample screening process requires companies to have complete data for the entire sample period (1987-2002). Firms not having complete data may be a result of increases or changes in rules or regulations and, were they included in the study, might yield different results than those reported above. Unfortunately, data unavailability prevents further analyses to ensure congeniality between sampled and excluded firms. Furthermore, data from a different time period may exhibit different behavior. This study uses data from 1995 through 2002. Firms may have changed their behavior in a post-Enron era. The data in the study are taken from a relatively small market; the results may not be applicable to a large market or market with a different culture. This study uses DA as the proxy for earnings management. Other proxies may yield different results. Furthermore, this study examined effects of all accounting rules and regulations on discretionary accruals. Not all rules or regulations, however, may have the same effect on earnings management. Some rules or regulations may even have opposite effects and an aggregation all of them as though all rules or regulations having similar effects, at least in directions, may have arrested significant effects that the evidence would have showed otherwise.

Nevertheless, the evidence shows that rules and regulations are effective in reducing earnings management. However, not all earnings managements are similar. To the extent possible, regulatory bodies of accounting rules and procedures need to take into considerations differences in earnings management behavior to improve effectiveness of new or changed rules and regulations.

Appendix

Because the SPSS software does not provide the value of the test, we calculated the post hoc test of the differences in upward earnings management percentage between the growth and mature stages and between the mature and stagnant stages as follows (Chen, 2004):

Growth Stage VS. Mature Stage

(0.601-0.532)

±

× （

_）

=

±

The confidence interval estimates for the percentage difference are between 0.006 and 0.132, without including zero. Thus these two percentages were significantly different.

Mature Stage VS. Stagnant Stage

(0.671-0.601)

±

× （

_）

=

±

The confidence interval estimates for the percentage difference are between 0.008 and 0.132, without including zero. Thus these two percentages were significantly different.

Notes

1. The literature includes several classification schemes on corporate life cycles. Anthony & Ramesh, (1992) classify firms into growth, mature, and stagnation stages. Adizes (1997) identifies stages of corporations from grow to die as courtship, infancy, go-go, adolescent, prime, stable, aristocracy, recrimination, bureaucracy, and death. Others have different schemes.

2. However, there is no consistent result with regard to the direction of manipulation when earnings are below the lowest bound for receiving bonus.

3. Anthony and Ramesh (1992) use a three-stage life-cycle scheme for firms: growth, mature, and stagnation. Stickney & Weil (1996), Black (1998), White et al., (2003) and many other researchers also use this or a similar scheme on firm life-cycle stages in their researches. The literature has many other schemes for identifying firm life cycle stages. Among them the more popular systems are ten-stage system by Adizes (1997): Courtship, Infancy, Go-go, Adolescent, Prime, Stable, Aristocracy, Recrimination, Bureaucracy, and Death and Boston Consulting Group’s four-stage strategic positioning (Henderson, 1979): Build, Hold, Harvest, and Divest.

4. A near-term compensation mechanism discourages long-term investments – a situation not beneficial to a growth firm (Bizjak et al., 1993).

5. Using the mature stage as the reference stage, a positive X1 coefficient indicates that the amount of DA at the growth stage is higher than that of the mature stage. Similarly, a negative X2 coefficient denotes that the amount of DA in the stagnant stage is lower than that of the mature stage.

6. The TEJ data bank in Taiwan is equivalent to the Compustat and CRSP in the U.S.

7. These firms are excluded because of their special industry practices that may render data incompatible with firms in other industries.

9. Other than differences such as sample firms and time periods studied, further analyses failed to reveal other contributors factors for the differences in findings.

10. The differences in results could be due to the much smaller number of firms decreased their outstanding shares (2.7%) than firms that increased shares outstanding (36.3%).. 11. ANOVA (F = 21.077, p < 0.000)

12. ANOVA (F = 3.593, p = 0.028)

13. According to Haberman (1978), an adjusted residual value higher than 1.96 is significant at .05 and at .01 if the value is greater than 2.56.

14. No further analyses were conducted for firms at growth stage because of the failure to reject the null hypotheses pertaining to growth firms observed and reported earlier.

References

Abarbanell, J., and R. Lehavy, R., “Can Stock Recommendations Predict Earnings Management and Analyst’S Earnings Forecast Errors?”, Journal of Accounting Research 41 (1), 1-31, (2003).

AlNajjar, F., and Riahi-Belkaoui, A., “Growth Opportunities and Earnings Management”, Managerial Finance 27(12), 72-81,(2001).

Anthony, J. H., and Ramesh, K., “Association between Accounting Performance Measures And Stock Prices: A Test Of The Life-Cycle Hypothesis”, Journal of Accounting and Economics 15(2/3), 203-227, (1992).

Adizes, I., Managing Corporate Lifecycles, Prentice-Hall, 1997.

Balkin, D. B., and Gomez-Mejia, L. R., “Toward A Contingency Theory of Compensation Strategy”, Strategic Management Journal 8(2), 169-182,(1987).

Balsam S., Krishnan, J., and Yang, J. S., “Auditor Industry Specialization and Earnings Quality”, Auditing: A Journal of Practice and Theory 22(2), 71-97, (2003).

Barth, M. E., Elliott, J.A., and Finn, M.W., “Market Rewards Associated with Patterns of Increasing Earnings”, Journal of Accounting Research 37(2), 387-413,(1999).

Bartov, E., Gul, F. A., and Tsui, J.S., “Discretionary-Accruals Models and Audit Qualifications”, Journal of Accounting and Economics 30(3), 421-452,(2001).

Becker, C. L., DeFond, M.L., Jiambalvo, J., and Subramanyam, K.R., “The Effect of Audit Quality on Earnings Management”, Contemporary Accounting Research 15(1), 1-21, (1998).

Bizjak, J. M., Brickley, J.A., and Coles, J.L., “Stock-Based Incentive Compensation and Investment Behavior”, Journal of Accounting and Economics 16(1-3), 349-372,(1993). Black, E., “Life Cycle Impacts on the Incremental Value-Relevance of Earnings and Cash

Flow Measures”, Journal of Financial Statement Analysis 4(1), 40-56,(1998).

Accounting Method Choice”, Journal of Accounting and Economics 20 (3), 255-295,(1995).

Brown, L.D., Griffin, P., Hagerman, R., and Zmijewski, M., “An Evaluation of Alternative Proxies for The Market’S Assessment of Unexpected Earnings”, Journal of Accounting and Economics 9 (2), 159-194,(1987).

Burgstahler, D., and Eames, M.J., “Management of Earnings and Analysts Forecasts”, Working paper, University of Washington, (1998).

Chen, C.C., Statistics for Behavioral and Social Science. (3rd ed.) Taipei: Great Stream Publishing. (in Chinese), 2004.

Collins, D. W., Kothari, S.P., Shanken, J., and Sloan, R.G.., “Lack of Timeliness Versus Noise as Explanations for Low Contemporaneous Return-Earnings Relation”, Working paper, (1992).

DeAngelo, L. E., “Accounting Numbers As Market Valuation Substitutes: A Study of Management Buyouts of Public Stockholders”, The Accounting Review 61(3), 400-420, (1986).

DeAngelo, H., DeAngelo, L., and Skinner, D. J., “Accounting Choice in Troubled Companies”, Journal of Accounting and Economics 17(1-2), 113-143,(1994).

DeAngelo, H., DeAngelo, L., and Skinner, D., “Reversal of Fortune: Dividend Signaling and The Disappearance of Sustained Earnings Growth”, Journal of Financial Economics 40 (3), 341-371,(1996).

Dechow, P. M., Sloan, R.G., and Sweeney, A.P., “Detecting Earnings Management”, The Accounting Review 70 (2), 193-225, (1995).

DeFond, M. L., and Jiambalvo, J., “Debt Covenant Violation and Manipulation of Accruals,” Journal of Accounting and Economics, 17(1-2), 145-176. (1994).

DeFond, M. L., and Park, C. W., “Smoothing Income in Anticipation of Future Earnings”, Journal of Accounting and Economics 23(2), 115-139, (1997).

DeFond, M. L., and Subramanyam, K.R., “Auditor Changes and Discretionary Accruals”, Journal of Accounting and Economics 25(1), 35-67, (1998).

Dye, R., “Earnings Management in an Overlapping Generations Model”, Journal of Accounting Research 26(2), 195-235,(1988).

Fudenberg, K., and Tirole, J., “A Theory of Income and Dividend Smoothing Based on Incumbency Rents”, Journal of Political Economy 103(1), 75-93, (1995).

Gaver, J. J. and Gaver, K. M., “Compensation Policy and the Investment Opportunity Set”, Financial management 24(1), 19-32,(1995).

Givoly, D., and Hayn, C., “The Changing Time-Series Properties of Earnings, Cash Flows and Accruals: Has Financial Reporting Becoming More Conservative?”, Journal of Accounting and Economics, 29(3), 287-320, (2000).

Healy, P. M., “The Effect of Bonus Schemes on Accounting Decisions”, Journal of Accounting and Economics 7(1-3), 85-107,(1985).

Haberman, S. J., Analysis of Qualitative. New York: Academic Press,1978.

Henderson, B. D., Henderson on Corporate Strategy. Cambridge. MA: Abt Books, 1979. Holthausen, R., and Watts, R., “The Relevance of Value-Relevance Literature for Financial

Accounting Standard Setting”, Journal of Accounting and Economics 31(1-3), 3-75,(2001).

Holthausen, R., Larcker, D.F., and Sloan, R.G.., “Annual Bonus Schemes and the Manipulation of Earnings”, Journal of Accounting and Economics, 19(1), 29-74. (1995).

Hong, K.P., Wu, M., and Xie, H., “Accounting Conservatism: A Life Cycle Perspective”, Working Paper, University of Illinois at Urbana-Champaign. (2002).

Jones, J., “Earnings Management during Import Relief Investigations”, Journal of Accounting Research 29 (2), 193-228,(1991).

Model”, Management Science 30, 696-712,(1984).

Kasznik, R., “On the Association between Voluntary Disclosure and Earnings Management”, Journal of Accounting Research, 37 (1), 57-81,(1999).

Kirshenheiter, M. and Melumad, N.D., “Can “Big Bath” and Earnings Smoothing Co-Exist as Equilibrium Financial Reporting Strategies?”, Journal of Accounting Research 40(3), 761-796, (2002).

Krishnan, G. V., “Does Big 6 Auditor Industry Expertise Constrain Earnings Management?”, Accounting Horizons 17, 1-16, (2003).

Lin, S.M., Lin, T.H., and Tsai, Y.C., “Earnings Management in Taiwan’s Imputation Tax System and Corporate Dividend Policy”, Taiwan Accounting Review 4(2), 127-152,(2004).

Martinez, I., “The Impact of Firm-Specific Attributes on the Relevance in Earnings and Cash-Flows: A Nonlinear Relationship between Stock Returns and Accounting Numbers”, Review of Accounting & Finance 2(1), 16-39, (2003).

Matsumoto, D.M., “Management’s Incentives to Avoid Negative Earnings Surprises”, The Accounting Review 77(3), 483-514,(2002).

McNichols, M., and Wilson, G. P., “Evidence of Earnings Management from the Provision for Bad Debts”, Journal of Accounting Research 26, 1-31,(1988).

Peltier-Rivest, D., “The Determinants of Accounting Choices in Troubled Companies”, Quarterly Journal of Business and Economics 38(4), 28-44,(1999).

Porter, M.E., Competitive Strategy: Techniques for Analyzing Industries and Competitors. Free Press, New York, 1980.

Reynolds, J. K., and Francis, J.R., “Does Size Matter? The Influence of Large Clients on Office-Level Auditor Reporting Decisions”, Journal of Accounting and Economics 30(3), 375-400,(2000).

MSU Business Topics (Autumn), 35-44,(1978).

Schipper, K., “Commentary on Earnings Management”, Accounting Horizons 3(4), 91-106,(1989).

Selling, T. I., and Stickney, C. P, “The Effects of Business Environment and Strategy on A Firm’s Rate of Return on Assets”, Financial Analyst Journal 45(1), 43-52,(1989).

Skinner, D. J., “The Investment Opportunity Set and Accounting Procedure Choice: Preliminary Evidence”, Journal of Accounting and Economics, 16(4), 407-445, (1993). Smith, C. W. and Watts, R.L., “The Investment Opportunity Set and Corporate Financing,

Dividend, and Compensation Policies”, Journal of Financial Economics 32(3), 263-292,(1992).

Spence, A.M., “Entry, Capacity, Investment and Oligopolistic Pricing”, Bell Journal of Economics 8, 534-544,(1977).

Stickney, C. P., and Weil, R.L., Financial Accounting: An Introduction to Concepts, Methods, and Uses. (8th ed.). Orlando: The Dryden Press, 1996.

Sunder, S., Theory of Accounting and Control. Cincinnati. OH: South-Western, 1997.

Sweeney, A. P., “Debt-Covenant Violations and Managers’ Accounting Responses”, Journal of Accounting and Economics 17(3), 281-308,(1994).

Teoh, S. H., Welch, I., and Wong, T.J., “Earnings Management and the Underperformance of Seasoned Equity Offerings”, Journal of Financial Economics 50(1), 63-99,(1998a). Teoh, S.H., Wong, T. J., and Rao, G., “Are Accruals During Initial Public Offerings

Opportunistic?”, Review of Accounting Studies 3, 175-208, (1998b).

Wernerfelt, B., “The Dynamics of Prices and Market Shares over the Product Life Cycle”, Management Science 31, 928-939,(1985).

White, G. I., Sondhi, A.C., and Fried, D., The Analysis and Use of Financial Statements. (2nd ed.). NY: John Wiley & Sons, 2003.

Table 1

Life-Cycle Classification Scheme Using Life-cycle Descriptors of Anthony and Ramesh (1992)

Rankings of Life-cycle Descriptors

Life-cycle Stages MDP* MSG* AGE

Growth Low High Young

Mature Medium Medium Adult

Stagnant High Low Old

* MDP and MSG refer to medians of dividend payout ratio and sales growth percentage, respectively, in the three years preceding year t. AGE is the number of years between the current year and the first year the firm’s stock is traded, as reported by TEJ.

Table 2

Industry Distribution of the Selected Sample Firms

Two digit SIC Industry Numbers Percentage

11 Cement 8 3% 12 Food 15 6% 13 Plastics 15 6% 14 Textiles 37 14% 15 Electric, Machinery 10 4% 16 Appliance, Cable 12 5% 17 Chemical 18 7% 18 Glass, Ceramics 6 2% 19 Paper, Pulp 7 3% 20 Steel, Iron 13 5% 21 Rubber 8 3% 22 Automobile 2 1% 23,24,53 Electronics 44 17% 25 Construction 15 6% 26,56 Transportation 11 4% 27 Tourism 5 2% 29 Department Stores 8 3% Others 30 11% Total 264 100%

Figure 1: Time-series of Absolute Discretionary Accruals (1987– 2002). 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 1987 1989 1991 1993 1995 1997 1999 2001 Mean Median

Table 3: Mean and Median of the Median DP, SG, and AGE for Firms at Different Life-Cycle Stages

Median Values of Life Cycle Descriptors

MDP MSG AGE

Life Cycle Stage Mean Median Mean Median Mean Median Growth (n=571) 39.20 48.14 22.84 20.72 20.27 19.00 Mature (n=1024) 45.71 57.96 4.11 3.98 30.44 30.00 Stagnant (n=517) 81.03 83.71 -1.52 -2.03 34.51 34.00

MDP: dividend payout ratio MSG: sales growth rate AGE: firm age

Table 4

Descriptive Statistics of Discretionary Accruals and Control Variables For the 264 Sample Firms (1995-2002)

Variable First Quartile Median Third Quartile Mean Std-Deviation

DA -0.031 0.018 0.068 0.021 0.104

X1 0.000 0.000 1.000 0.270 0.444

X2 0.000 0.000 0.000 0.244 0.430

LEV 0.291 0.399 0.502 0.406 0.164 ABSTA 0.021 0.046 0.085 0.066 0.076 ShareIncr 0.000 0.000 1.000 0.363 0.481 ShareDecr 0.000 0.000 0.000 0.027 0.163 OldAud 0.000 0.000 0.000 0.025 0.156 NewAud 0.000 0.000 0.000 0.027 0.163 B6 1.000 1.000 1.000 0.756 0.429 DA = discretionary accruals

X1 = dummy variable, 1 if the sample firm-year is at a growth stage and 0 otherwise

X2 = dummy variable, 1 if the sample firm-year is at a stagnant stage and 0 otherwise

SIZE = log of total assets

LEV = total debts divided by total assets ABSTA = absolute value of total accruals

ShareIncr = outstanding shares increase more than 10 percent during the year ShareDecr = outstanding shares decrease more than 10 percent during the year OldAud = the last year is followed by an auditor change

NewAud = the first year with a new auditor

Table 5 Pearson and Spearman Correlation Coefficients between Variables (n=2112)

DA X1 X2 SIZE LEV ABSTA ShareIncr ShareDecr OldAud NewAud B6

DA -0.076 (0.001) 0.082 (0.000) 0.025 (0.246) -0.145 (0.000) -0.210 (0.000) 0.241 (0.000) 0.023 (0.281) 0.019 (0.395) -0.067 (0.002) -0.044 (0.044) X1 -0.074 (0.001) -0.347 (0.000) 0.040 (0.068) 0.046 (0.036) 0.157 (0.000) 0.187 (0.000) 0.041 (0.058) -0.016 (0.466) 0.015 (0.487) 0.080 (0.000) X2 0.063 (0.004) -0.347 (0.000) 0.088 (0.000) -0.112 (0.000) -0.141 (0.000) -0.135 (0.000) -0.001 (0.951) -0.014 (0.523) -0.028 (0.194) -0.061 (0.005) SIZE 0.008 (0.721) 0.041 (0.062) 0.086 (0.000) 0.264 (0.000) -0.043 (0.049) 0.083 (0.000) -0.085 (0.000) -0.023 (0.282) -0.030 (0.166) 0.094 (0.000) LEV -0.121 (0.000) 0.051 (0.020) -0.116 (0.000) 0.217 (0.000) 0.116 (0.000) -0.158 (0.000) 0.087 (0.000) 0.069 (0.001) 0.094 (0.000) 0.048 (0.029) ABSTA 0.160 (0.000) 0.116 (0.000) -0.118 (0.000) -0.018 (0.412) 0.175 (0.000) 0.030 (0.164) 0.062 (0.004) 0.033 (0.134) 0.069 (0.001) 0.045 (0.038) ShareIncr 0.226 (0.000) 0.187 (0.000) -0.135 (0.000) 0.098 (0.000) -0.167 (0.000) 0.074 (0.001) -0.127 (0.000) 0.005 (0.834) -0.037 (0.092) 0.033 (0.133) SharDecr 0.037 (0.087) 0.041 (0.058) -0.001 (0.951) -0.090 (0.000) 0.110 (0.000) 0.092 (0.000) -0.127 (0.000) -0.008 (0.698) 0.007 (0.740) -0.006 (0.791) OldAud 0.028 (0.197) -0.016 (0.466) -0.014 (0.523) -0.010 (0.638) 0.065 (0.003) 0.035 (0.112) 0.005 (0.834) -0.008 (0.698) 0.010 (0.644) -0.029 (0.188) NewAud -0.090 (0.000) 0.015 (0.487) -0.028 (0.194) -0.012 (0.570) 0.114 (0.000) 0.068 (0.002) -0.037 (0.092) 0.007 (0.740) 0.010 (0.643) 0.021 (0.330) B6 -0.040 (0.064) 0.080 (0.000) -0.061 (0.005) 0.116 (0.000) 0.038 (0.085) 0.022 (0.319) 0.033 (0.133) -0.006 (0.791) -0.029 (0.187) 0.021 (0.330) Lower left reports Pearson correlation coefficients and upper right reports Spearman rank correlation coefficients. Amounts in the parentheses denote probability level (2-tail)

Table 6

Mean of Upward and Downward Discretionary Accruals (1995 – 2002)

Downward (n=846) Upward (n=1266) Growth Mature Stagnant Growth Mature Stagnant

1995 0.047 0.040 0.040 0.115 0.091 0.075 1996 0.073 0.042 0.037 0.069 0.074 0.076 1997 0.065 0.110 0.060 0.112 0.127 0.121 1998 0.066 0.050 0.045 0.083 0.077 0.083 1999 0.068 0.058 0.040 0.079 0.051 0.033 2000 0.082 0.058 0.026 0.078 0.052 0.033 2001 0.101 0.069 0.051 0.050 0.032 0.032 2002 0.078 0.084 0.037 0.040 0.041 0.029 1995-2002 0.079 0.062 0.039 0.086 0.079 0.068 Table 7

Frequency (%) of Upward and Downward Discretionary Accruals (1995 to 2002)

Downward (n=846) Upward (n=1266) Growth Mature Stagnant Growth Mature Stagnant

1995 21.3% 19.3% 13.1% 78.7% 80.7% 86.9% 1996 27.7% 25.7% 17.5% 72.3% 74.3% 82.5% 1997 9.7% 6.8% 5.0% 90.3% 93.2% 95.0% 1998 39.5% 22.9% 19.3% 60.5% 77.1% 80.7% 1999 51.4% 48.6% 48.1% 48.6% 51.4% 51.9% 2000 61.2% 57.2% 59.6% 38.8% 42.8% 40.4% 2001 84.0% 78.6% 70.5% 16.0% 21.4% 29.5% 2002 69.4% 62.2% 44.8% 30.6% 37.8% 55.2% 1995-2002 46.8% 39.9% 32.9% 53.2% 60.1% 67.1%

Table 8

Cross-Tabulation of Upward and Downward Discretionary Accruals and Life-Cycle Stages

Life-cycle Stages

Growth Mature Stagnant Total

Count 304 615 347 1266

Expected Count 342.3 613.8 309.9 1266

(%)Within Life-cycle Stages 53.2% 60.1% 67.1% 59.9% Std. Residual -2.1 0.0 2.1 Upward Discretionary Accruals Adjusted Residual -3.8 0.1 3.8 Count 267 409 170 846 Expected Count 228.7 410.2 207.1 846 (%)Within Life-cycle Stages 46.8% 39.9% 32.9% 40.1% Std. Residual 2.5 -0.1 -2.6 Downward Discretionary Accruals Adjusted Residual 3.8 -0.1 -3.8 Total Count 571 1024 517 2112

Pearson Chi-Square = 21.775 (P-value < 0.000)

Table 9

Percentage Difference Tests of the Upward and Downward Earnings Management in Each Life-cycle Stage

Upward Downward Chi-square Value P-Value

Growth 304(53.2%) 267(46.8%) 2.398 0.122

Mature 615(60.1%) 409(39.9%) 41.441 0.000

Stagnant 347(67.1%) 170(32.9%) 60.598 0.000

Table 10

Mean Difference between Upward and Downward Discretionary Accruals in Different Life-cycle Stages

Upward Downward Difference t-value p-value Growth stage 0.086(n=304) 0.079(n=267) 0.007 1.184 0.237 Mature stage 0.079(n=615) 0.062(n=409) 0.017 3.364 0.001 Stagnant stage 0.068(n=347) 0.039(n=170) 0.029 5.434 0.000

Table 11

ANOVA and Scheffe’s Test of Discretionary Accruals in Different Life-cycle Stages

Life Stage Mean Mean Difference P-value

Growth Stage (n=571) 0.008 0.014(Growth VS. Mature) 0.040 Mature Stage (n=1024) 0.022 0.010(Mature VS. Stagnant) 0.178 Stagnant Stage (n=517) 0.033 0.025(Growth VS. Stagnant) 0.001 F-value = 7.503, P-value = 0.001

Table 12

OLS Regression of Discretionary Accruals

Independent Variables Coefficients t-value P-value VIF

Intercept -0.006 -0.207 0.836

X1 (dummy variable, 1growth, 0 others) -0.026 -4.966 0.000 1.182 X2 (dummy variable, 1 stagnant, 0 others) 0.014 2.658 0.008 1.193 SIZE (log of total assets) 0.004 0.929 0.353 1.127 LEV (total debts/total assets) -0.066 -4.706 0.000 1.193 ABSTA (absolute value of total accruals) 0.246 8.532 0.000 1.070 SharIncr (Outstanding shares increases > 10%) 0.049 10.485 0.000 1.130 SharDecr (Outstanding shares decreases > 10%) 0.043 3.250 0.001 1.047 OldAud (last year is followed by an auditor change) 0.018 1.326 0.185 1.007 NewAud (first year with a new auditor) -0.050 -3.793 0.000 1.018 B6 (Big 6 auditor) -0.008 -1.677 0.094 1.023 Adjust R-square 11.3% F-statistic (p<0.000) 27.912 Number of observations 2112

Table 13

Frequency of Upward Earnings Management Of Mature Firms

Independent Variables Coefficients S.E P-value

Intercept 893.064 78.364 0.000

YEAR -0.449 0.039 0.000

SIZE (log of total assets) 0.676 0.169 0.000 LEV (total debts/total assets) -1.192 0.484 0.014 ABSTA (absolute value of total accruals) -4.589 0.961 0.000 SharIncr (Outstanding shares increases > 10%) 0.438 0.170 0.010 SharDecr (Outstanding shares decreases > 10%) 0.188 0.520 0.717 OldAud (last year is followed by an auditor change) 0.351 0.424 0.408 NewAud (first year with a new auditor) -0.243 0.448 0.587

B6 (Big 6 auditor) -0.108 0.169 0.524

-2Log Likelihood 1125.281

Cox & Snell –R2 0.219

Nagelkerke –R2 0.296

Table 14

Magnitude of Upward Earnings Management Of Mature Firms

Independent Variables Coefficients t-value P-value VIF

Intercept 10.084 5.575 0.000

YEAR -0.005 -5.561 0.000 1.130

SIZE (log of total assets) 0.005 1.340 0.181 1.228 LEV (total debts/total assets) -0.007 -0.636 0.525 1.213 ABSTA (absolute value of total accruals) 0.943 48.22 0.000 1.099 SharIncr (Outstanding shares increases > 10%) -0.002 -0.603 0.547 1.129 SharDecr (Outstanding shares decreases > 10%) -0.007 -0.574 0.566 1.074 OldAud (last year is followed by an auditor change) -0.001 -0.017 0.987 1.012 NewAud (first year with a new auditor) -0.004 -0.376 0.707 1.058 B6 (Big 6 auditor) -0.001 -0.013 0.990 1.032

Adjust R-square 0.810

F-statistic (p<0.000) 291.58

Number of observations 615