Incentives for and consequences of initial voluntary asset write-downs in the emerging Chinese...
Publication: Journal of International Accounting Research
Date: Thursday, January 1 2004
ABSTRACT: Understanding the incentives for and consequences of accounting method choices is important to various constituents of accounting. In 1998, a Chinese accounting regulation allowed listed companies to voluntarily write-down assets through their income statements. The regulation was
I. INTRODUCTION
Accounting choice has been an important area of research over the past three decades because accounting information is widely used in firms' formal and informal contracts as well as investors' investment decisions. The majority of accounting choice research examines income-increasing accounting methods. The scant amount of research on income-decreasing choices may partially reflect the existing literature's focus on managerial opportunism as managers' primary motive for selecting particular accounting policies. However, Holthausen (1990) and, more recently, Fields et al. (2001) call for investigation of managerial motives other than opportunism. This study contributes to this stream of literature by investigating the incentives for and consequences of voluntary asset write-downs using a unique research opportunity in the Chinese equity market.
A limited number of asset write-down studies exist, but the results of these studies are mixed due to sample selection differences and inadequate research methods (Alciatore et al. 1998). This study makes use of the insights and findings of the existing literature and capitalizes on a unique institutional setting in China to provide additional evidence of why companies voluntarily write-down assets and whether and how this income-decreasing choice matters to investors in the Chinese stock market.
Two recent standard-setting events in China allow us to unambiguously identify a test sample of firms that voluntarily wrote down assets and a control sample of firms that suffered an asset impairment but chose not to write-down the impaired assets. Together, our test and control samples constitute a research setting analogous to an experimental environment. This research setting enables us to mitigate problems arising from not having an adequate control sample that may contaminate the empirical results documented in prior literature (Alciatore et al. 1998).
In 1998 the Chinese authorities issued the "Accounting Regulation for Listed Companies." The regulation gave all listed A-share companies (1) the option to voluntarily write-down to market value short-term investments, inventory, and long-term investments and to record the unrealized loss in income. (2) Of the 745 A-share companies listed in 1998, we identify 134 companies that voluntarily wrote down at least one of these three assets. Our sample is carefully constructed to exclude any compulsory write-downs by companies with both A- and B- or H-shares. Because the write-downs were both initial and voluntary, and triggered by a single event in the same year, we contend that the firms in our test sample likely present fewer factors that might otherwise confound our examination of the economic incentives for and consequences of the income-decreasing accounting choice.
In 1999, Chinese authorities amended the 1998 regulation and mandated that all listed companies write-down to market value any overvalued short-term investments, inventory, and long-term investments. The amendment required managers to charge that portion of the unrealized loss suffered during 1999 to income, and to charge the remainder, presumably attributable to losses suffered prior to 1999, against equity. The 1999 amendment was not anticipated in 1998, (3) and so it provides us with a unique opportunity to identify a control sample of firms that had asset impairments in 1998 but chose not to recognize the loss.
Overall, we find that the firms that voluntarily wrote down their assets in 1998 experienced a positive valuation effect. We also find that firms with CEO changes and firms reporting losses in 1998 even before the write-down were more likely to write-down their assets and to do so by a larger amount. Finally, we show that the positive valuation effect of the write-down exists only for firms with CEO changes and/or losses prior to taking the write-down. Accordingly, our evidence is consistent with the market responding positively to voluntary write-downs taken by firms that appear to be "cleaning the slate."
This evidence is consistent with managers behaving opportunistically if the write-off is motivated by a desire to manufacture higher accounting earnings in the future by taking a "big bath" coincident with a new CEO's arrival or with the unavoidable reporting of a loss. Alternatively, our evidence is consistent with a manager using the write-off to signal to the market that the firm recognizes its problems and is taking necessary actions to increase its chances of generating economic earnings in the future.
To gain some insight into these alternative interpretations, we examine the relationship between the voluntary asset write-down and one-year-ahead performance improvements as measured by return on assets and operating cash flows. While our net income results demonstrate a performance improvement, our operating cash flow results do not. We provide possible alternative explanations for this inconsistency but recognize that the performance improvement in net income may be an artifact of accrual accounting reversal. Although the performance results seem consistent with the opportunistic interpretation of managers' motives, it is not clear why the market responds positively to the managers' actions if the managers are indeed acting opportunistically. It is possible that the voluntary write-down firms are firms for which the market overestimated the magnitude of the loss. Such firms have strong incentives to correct the market's estimate by voluntarily writing down their assets. Upon observing the managers' disclosure, the market revises its assessment of the loss, yielding a positive valuation effect. Even so, while this may explain the market reaction to the write-off, it does not explain why the positive valuation effect is limited to firms with CEO changes and/or losses prior to the write-down. Accordingly, we believe our results taken together are most consistent with the signaling explanation.
To summarize, our study contributes to the accounting choice literature and to our understanding of accounting and equity markets in China. Although managers in China, as in other countries, have been found to engage in opportunistic earnings management (Haw et al. 1998; Aharony et al. 2000; Chen et al. 2001b), our evidence suggests that opportunism may not be the only explanation for accounting method choices in this emerging market.
H. PRIOR LITERATURE AND RESEARCH QUESTIONS
Many studies on accounting method choice have been undertaken over the past three decades, but progress slowed in the 1990s for three reasons (Fields et al. 2001). First, the literature failed to provide compelling or consistent evidence of whether and how accounting choices affect contracting parties or firm value. Second, the literature failed to explore alternative explanations for managers' motives for selecting certain accounting policies, focusing instead on a single opportunistic motive. Third, the statistical techniques and research designs employed were not powerful enough to distinguish between alternative motives for accounting choice such as managerial opportunism, shareholder wealth maximization, and informational motives. These more recent observations are generally consistent with those made a decade ago by Holthausen (1990) and Watts and Zimmerman (1990). Clearly, more evidence is needed.
In part, the emphasis placed on managerial opportunism explains why the majority of accounting choice studies focus on income-increasing methods. Notably, voluntary asset write-downs are one of the few income-decreasing accounting choices studied in the literature. However, existing studies of voluntary asset write-downs fail to document consistent evidence of the incentives for and consequences of this income-decreasing accounting choice.
Strong and Meyer (1987) identify a change in senior management as the primary reason for write-down decisions and document a positive stock price effect, which they interpret as write-downs being a signal of future events, similar to stock splits. Elliot and Shaw (1988) find that firms disclosing asset write-offs have greater leverage and earnings performance problems throughout the three years before the write-off announcement. They also document significant one- and two-day industry-adjusted negative returns around the write-off announcement. Furthermore, the magnitude of the negative stock return is associated with the relative size of the write-off. Zucca and Campbell (1992) reveal no consistent results or pattern either in the financial characteristics of voluntary write-down firms or in the stock market reaction to the write-downs they study. Francis et al. (1996) examine both the determinants of and stock price reaction to write-off announcements. They provide evidence intended to distinguish between the determinants that measure earnings manipulation and those that reflect asset impairments. Overall, their tests of shareholder wealth effects document a negative market reaction to write-off announcements, which suggests that the valuation effect is driven by write-offs revealing asset impairments rather than conveying positive private information about future performance. However, after partitioning the sample, Francis et al. (1996) find that investors' responses to restructuring charges are positive, which they interpret as write-downs being a signal of future performance improvement.
Alciatore et al. (1998) discuss several reasons for the mixed results outlined above. First, studies often include involuntary write-downs, such as lower-of-cost-or-market adjustments to inventory. Second, researchers normally identify write-down samples via a keyword search of databases such as the National Automated Accounting Research System (NAARS) and the Wall Street Journal Index. As a result, companies that voluntarily write-down assets but do not publicly disclose the information are automatically excluded. This exclusion may explain why, with an average of fewer than 200 write-down cases a year, sample sizes are small relative to the size of the U.S. market. Third, existing research does not distinguish between initial versus repeated write-downs, which presumably have different implications for both the company's incentives and the information content of the write-down. Fourth, all of the studies discussed above use pooled cross-sectional and intertemporal data over four to six years. Intertemporal differences are not controlled for and no specific event triggers the voluntary asset write-down decision. Finally, while a control sample may be included, it is unclear whether the control firms choose not to write-down assets because managers choose to withhold the information or because their firms had not suffered an asset impairment.
This study is free from these problems. The 1998 regulation is a unifying event that triggers the initial voluntary asset write-down. Moreover, we clearly identify a control sample of firms that should have written down impaired assets in 1998 but chose not to do so. Figure 1 summarizes the changes in accounting standards on asset write-downs in China and shows why we can identify a control sample without any ambiguity. As noted earlier, both the test and control samples are comprised of companies that issue only A-shares.
Li (2001) also examines recent asset write-down practice in China, but with a different objective. Focusing on incentive variables that lead to the over- or underestimation of asset write-downs, Li (2001) finds that a desire to report higher (lower) earnings led companies to under- (over-) estimate asset impairments when the recording of write-downs became compulsory in 1999, while companies that voluntarily wrote down their assets in 1998 tended to underestimate the write-down.
This study concentrates on companies' motives to voluntarily write-down assets and compares the consequences of this choice for stock values and future performance to the consequences of a mandatory write-down. Specifically, we analyze four research questions. First, does the voluntary write-down matter to firm valuation? Second, what incentive variables are associated with the asset write-down choice? Third, is the valuation implication of an asset write-down dependent on firm characteristics? Fourth, is the asset write-down systematically associated with subsequent performance improvement? Accordingly, this study provides evidence that Li (2001) does not address.
III. MARKET VALUATION EFFECT OF VOLUNTARY ASSET WRITE-DOWNS
The relationship between stock price and accounting choice is generally not well articulated (Fields ct al. 2001), and research designs are often ambiguous as to why particular stock price effects are anticipated. With respect to asset write-downs, positive or negative valuation effects may be anticipated. For example, if the write-down conveys information about asset impairment only, then share prices should fall in response to the write-down. However, if the write-down is perceived by investors to reveal or signal something valuable about the company, then share prices should rise.
There is evidence in the literature supporting both views, but we believe our study is in a better position to articulate and test for a positive valuation effect for several reasons. The public in China knew that both the test and the control firms had suffered asset impairments by the end of 1998. (4) More importantly, because the control firms were required to write-down assets in 1999 with a retroactive adjustment of the pre-1999 asset impairment to opening equity, we can demonstrate ex post that these firms had impaired assets in 1998. Consequently, we do not expect the stock price response to the 1998 voluntary write-down to reflect only asset impairment, which to some extent was already impounded in stock price. Rather, we expect that the managers' decisions to take action voluntarily conveys positive information to the market by signaling the managers are aware of the problem and are taking the necessary actions to resolve it. Our test and control samples allow a direct test of this expectation. (5)
Sample Selection and Descriptive Statistics
We construct both our test and control samples from the TEJ CD-ROM database and 1999 published annual reports. A firm is included in the write-down sample if it has a positive balance at the end of 1998 in at least one of the three provision accounts, but these three provision accounts have zero balances at the beginning of the year. This procedure ensures only initial voluntary write-downs are included in the test sample. To select the control sample that should have written down assets in 1998 but did not, we examine the three provision accounts in both 1999 and 1998. A firm is identified as a control firm if it does not have any balances in the three provision accounts at the end of 1998, but made at least one provision in 1999 and adjusted part of the provision retroactively to opening equity. Table 1 describes both the sample selection process and the magnitude of write-downs in both years.
As shown in Panel A of Table 1, 134 firms voluntarily wrote down at least one of the three assets in 1998. In our analyses, each firm represents one observation. Panel A also shows that 456 firms wrote down at least one of the three assets in 1999 as a result of the mandatory write-down requirement. The data indicate that inventory write-downs are observed most frequently,
Panel B of Table 1 presents descriptive statistics pertaining to the voluntary write-downs taken in 1998. The data indicate that inventory write-downs are largest in magnitude. The average total write-down amount is 10,882,000 RMB, which accounts for 0.89 percent of total assets and 10.15 percent of net income. Though these percentages may appear small in comparison to write-offs reported in the literature, such as restructuring charges and goodwill write-offs, the write-offs represent a significant market event in China because such a large proportion of firms are affected (18 percent in 1998 and 55 percent in 1999).
Panel C of Table 1 reports similar statistics for the control sample. These statistics are based on the portion of the mandatory write-downs adjusted to opening equity. Accordingly, these data document the asset impairment the control firms should have recognized in 1998 but did not. Panel C shows that the mandatory write-downs in 1999 are greater in magnitude than the voluntary write-downs in 1998. The amount charged to opening equity by the mandatory write-down firms may be biased upward, however, if managers have incentives to avoid recognizing the charge in net income for 1999 by claiming more asset impairment occurred before 1999 than is strictly true.
Models and Analyses
Based on evidence that supports the value relevance of accounting information in the Chinese equity market (Chen et al. 2000; Chela et al. 2001a; Etcher and Healy 2000), we employ a return model and a price model, to examine the market value effect. Although earlier studies of asset write-downs tend to use event study methodology to assess the valuation effect (e.g., Strong and Meyer 1987; Elliott and Shaw 1988; Zucca and Campbell 1992), we employ the association study methodology because it is more commonly used in recant studies of the value-relevance of accounting method choice (e.g., Subramanyam 1996; Barth and Clinch 1998; Aboody et al. 1999). (6)
The return and price models we employ address related but different value-relevance questions. The return model provides information about whether the write-down is reflected in changes in value over the one-year return period, while the price model provides information about whether the write-down is value relevant with respect to its association with firm value (Barth et al. 2001). Kothari and Zimmerman (1995) specifically suggest the use of both models to permit more definitive inferences, and it is common practice in the literature to do so (e.g. Easton et al. 1993; Barth and Clinch 1998; Aboody et al. 1999).
The return and price models are standard models with the addition of several write-down variables:
(1) RET = [[beta].sub.0] + [[beta].sub.1] WD98 + [[beta].sub.2] EPS + [[beta].sub.3] [DELTA]EPS + [[beta].sub.4] SIZE + [[beta].sub.5] WD + [[beta].sub.6] WD_98 + [epsilon]
(2) P = [[beta].sub.0] + [[beta].sub.1] WD98 + [[beta].sub.2] BVPS + [[beta].sub.3] NI + [[beta].sub.4] SIZE + [[beta].sub.5] WD + [[beta].sub.6] WD_98 + [epsilon]
PET is a 12-month buy-and-hold return ending in the financial statement announcement month. WD98 is a dummy variable with a value of 1 for companies that voluntary wrote down assets in 1998. EPS is net income per share before the 1998 write-down over beginning stock price. [DELTA]EPS is the change in net income per share from 1997 to 1998 over beginning stock price with 1998 net income computed before the write-down. SIZE is the natural logarithm of market value of equity. P is the stock price as of the end of the announcement month. WD is the amount of asset impairment suffered prior to the end of 1998 scaled by beginning market value. For the test sample, the amount of asset impairment suffered prior to the end of 1998 is the voluntary write-down amount charged against 1998 income; and for the control sample, it is the mandatory write-down amount charged directly to opening equity in 1999. WD_98 is an interaction term between WD98 and WD. BVPS is the book value of assets per share, and NI is 1998 net income per share before the write-down. Table 2 presents results from the two models with and without industry dummies. (7)
We focus our discussion on the results with the industry dummies because our results do not differ significantly with or without industry dummies. Both models are statistically significant (p-values of less than 0.05). Adjusted [R.sup.2] is 0.296 for the return model, and 0.526 for the price model. The earnings variables are significantly positive in both models. These results are consistent with prior studies that employ similar models.
We examine WD and WD_98 in each model to test for the positive valuation effect of the voluntary asset write-down we expect to observe. With WD98 distinguishing the voluntary write-down firms (i.e., test firms) from the mandatory write-down firms (i.e. control firms), WD captures the valuation effect during 1998 of the write-down impairment loss suffered by the test and control firms prior to the end of 1998 and WD_98 measures the incremental valuation effect of choosing to voluntarily record the loss in 1998. A joint F-test of the coefficients on these two variables (i.e., [[beta].sub.5] + [[beta].sub.6]) reveals whether the valuation effect of the 1998 voluntary write-down is positive.
Table 2 provides the results of estimating Regression Equations (1) and (2), and it documents three interesting findings related to our first research question. First, WD is not statistically significant except in the price model estimated with industries dummies where the coefficient is significantly negative (a one-tailed p-value of 0.054). These results suggest that, for the most part, the information contained in the write-down amount was impounded in stock price prior to 1998, yielding little association between the write-down amount and returns earned during 1998. However, the write-down amount is value relevant in as much as it is associated with stock price. The negative sign on the coefficient indicates that the market interprets the write-down amount as "bad news" consistent with the write-down amount capturing asset impairment. Second, the coefficient on WD_98 is significantly positive in both models, regardless of whether the model is estimated with or without industry dummies. The return result indicates that the voluntary write-down communicated "good news" to the market during 1998 and the price result suggests that the valuation implication of the write-down is significantly less negative than that of the firms that chose not to take the write-down voluntarily in 1998. Finally, our joint F-test of [[beta].sub.5] + [[beta].sub.6] indicates that the sum of the coefficients is highly significantly positive in both models. This result is consistent with our expectation that the write-down amount recorded as a charge to earnings in 1998 by the voluntary write-down firms communicated "good news" to the capital market during 1998, which was impounded in the 1998 year-end stock price. In summary, our results suggest that the mandatory write-down firms' stock prices impounded the "bad news" associated with the write-down some point prior to 1998. In contrast, the voluntary write-clown firms' stock price responded positively to the voluntary write off during 1998 such that, by year end, the overall effect of the write-down amount on stock price is positive for these firms.
There may be alternative explanations as to why the 1998 voluntary asset write-down carries positive valuation information to the market. First, the voluntary write-down decision may signal to the market the managers' determination to deal with the problems underlying the asset losses. By voluntarily writing down assets, these companies may be stating that their worst problems are behind them and a better future lies ahead. This is the main argument in the literature for predicting a positive valuation effect of voluntary asset write-downs (e.g., Strong and Meyer 1987; Francis et al. 1996). Alternatively, disclosing the write-down amount may reduce uncertainties surrounding the severity of the asset impairment suffered and may allow the market to adjust its expectation of the magnitude of the impairment losses. Given fluctuations in China's economy and rules that disallowed A-share companies to write-down their assets to market value before 1998, all A-share companies are believed to have impaired assets. In the absence of disclosure, the market may over-or underestimate impairment losses. Firms whose asset impairment is overestimated by the market have strong incentives to voluntarily write-down assets, and upon observing disclosed asset losses, the market responds positively as it revises its estimate of the magnitude of the loss downward)
The positive valuation effect we document is consistent with either of these explanations. To assess which explanation is more plausible, we conduct additional analyses to (1) ascertain what factors motivate voluntary asset write-downs, (2) examine how these factors affect the results of our valuation analyses, and (3) determine whether voluntary asset write-downs are associated with subsequent performance improvement.
IV. FIRM CHARACTERISTICS AND VOLUNTARY ASSET WRITE-DOWNS
Prior studies report that firms with financial difficulties are more likely to choose income-decreasing accounting methods. For example, DeAngelo ct al. (1994) does find that accounting choices made by financially troubled firms are income increasing to avoid debt covenants. Instead, the accounting choices are primarily income decreasing to recognize financial difficulties. Elliott and Shaw 0988) show that firms that report discretionary asset write-offs have falling earnings-to-assets and earnings-to-market values in the three years before the write off.
One interpretation of these findings is that firms with financial difficulties signal their preparedness to deal with their problems to the market by choosing income-decreasing accounting methods. However, there are reasons to believe other incentives also motivate companies to voluntarily write-down assets. For example, companies reporting profits that exceed investors' expectations may be motivated to take advantage of the slack this situation affords and may write-down assets to allow for above-target profits in future periods as well. Li (2001) reports evidence that asset write-downs in China are associated with various profitability motives. To explore alternative incentives for the voluntary write-down, we estimate the following model that includes variables selected from the extant literature:
(3) WD = [[beta].sub.0] + [[beta].sub.1] SIZE + [[beta].sub.2] DR97 + [[beta].sub.3] MGTSHR + [[beta].sub.4] DELIST + [[beta].sub.5] GOOD + [[beta].sub.6] LOSS + [[beta].sub.7] CEO + [epsilon].
Equation (3) is estimated using both logistic and least-squares regression. In the logistic regression the dependent variable, WD, is an indicator variable for the voluntary asset write-down choice. In the least-squares regression, WD is a continuous variable measured as the dollar amount of the voluntary write-down over total assets. SIZE, measured as the natural logarithm of total assets, is a control variable. Given prior studies find that larger firms are more likely to voluntarily write-down assets (Elliott and Shaw 1988; Francis et al. 1996), we expect to observe a positive coefficient on this variable.
DR97 is the debt-to-asset ratio at the end of 1997, and is included in our model as an additional control variable. While the debt covenant hypothesis predicts an inverse relation between the debt ratio and income-decreasing accounting choices, the literature also provides evidence of a positive relationship (Elliott and Shaw 1988). This positive relationship may be interpreted as a consequence of increasing debt monitoring. High-leverage firms are more likely subject to a careful scrutiny by lenders, which may motive managers of these firms to take discretionary asset write-offs as a signal to lenders of their willingness to deal with the firm's problems. DeAngelo et al. (1994) provide evidence consistent with this explanation.
DELIST, GOOD, and LOSS are included in the model to examine whether profitability in general, and the incentives induced by governmental profitability regulations in particular, affect write-down decisions (Chenet al. 2001b; Li 2001). As discussed earlier, poor-performing companies may have incentives to take a "bath," while well-performing companies may be motivated to use asset write-downs to smooth earnings (Zucca and Campbell 1992; Francis et al. 1996).
Chinese profitability regulations in force during our test period may exacerbate these motives as they explicitly require listed companies to meet target profit levels. First, to be eligible for raising additional capital through rights offering, a company must report a minimum return on equity (ROE) of 6 percent each year and a minimum average ROE of 10 percent for three consecutive years (Chen and Yuan 2001; Li 2001). Second, if a company reports negative ROE for three consecutive years (Company Law of the People's Republic of China 1994, Articles 157 and 158; Li 2001), then it is delisted. Consequently, companies with different levels of pro-write-down earnings may have different incentives for writing down assets. For example, companies reporting losses regardless of the asset write-down have a stronger motive to record the write-down than those reporting marginal profits. Similarly, companies reporting an ROE above 6 percent even after the asset write-down charge is recorded may be more likely to write-down to improve the probability of meeting future profit targets.
DELIST is coded 1 for companies reporting a marginal non-negative ROE (0 < ROE < = 1 percent) before asset write-down. We expect these companies are less likely to write-down assets than the companies outside this band. GOOD is coded I for companies reporting ROE above 6 percent after considering the write-down. These companies may have the incentive to recognize the write-down because it allows them to transfer earnings to future periods while maintaining the current ROE target. LOSS is coded I if a firm's ROE is negative regardless of its write-down decision. When reporting positive ROE becomes impossible, companies should be more willing to recognize the write-down in an attempt to increase the probability of meeting future profitability targets. Furthermore, for fear of social turmoil, the government often helps listed companies facing imminent bankruptcy by transferring profits from SOEs, offering local tax cuts, or restructuring solutions that favor the listed company (Wei et al. 2000). This de facto government guarantee of no bankruptcy encourages managers of companies with large losses to select income-decreasing accounting methods to increase the probability of government assistance.
CEO is an indicator variable with a value of 1 for companies that changed their CEO in 1998. Prior studies find that a change in senior management is an important determinant of asset write-down decisions because new managers have the incentive to "clear the deck" of impaired assets to improve the firm's accounting performance in the future (Strong and Meyer 1987; DeAngelo et al. 1994; Francis et al. 1996). Alternatively, the new manager may change the strategic focus of the firm and write-down assets to gain a fresh start. In addition, we include the percentage of management shareholding, MGTSGR, in the model and expect a positive coefficient on the consideration that a larger ownership may motivate management to seize the opportunity provided by the new asset-write regulation to deal with problems behind the asset loss. Table 3 reports our univariate and multivariate results.
The parametric t-test and the nonparametric Wilcoxon test in Panel A of Table 3 reveal no differences in the three variables (DR97, MGTSHR, and SIZE) between the test and control samples. To a certain degree, these results may reflect some unique institutional arrangements in China. For example, bank loan decisions result mostly from government policies. In the absence of a debt capital market, the level of leverage may not reflect the tightness of a firm's debt covenants. Moreover, the low percentage of management shareholdings in China is widely documented, and at less than 0.3 percent, the average percentage of management shareholdings in our sample is consistent with prior studies (Chen et al. 2001b). This low managerial ownership, along with the lack of executive stock options, may explain why the percentage of managerial shareholdings is not different between the two groups.
Panel B of Table 3 presents the observed frequency of LOSS, GOOD, CEO, and DELIST along with the results of a Chi-squared test of differences between the two groups. For the test sample, the observed frequency of LOSS and CEO is significantly higher than the expected frequency (22 and 31 as compared to 8.79 and 21.32, respectively). In addition, the observed frequency of GOOD and DELIST is significantly lower than the expected frequency (80 and 2 as compared to 92.29 and 6.37, respectively). These results suggest that companies reporting losses, regardless of the write-down decision, and companies with CEO changes are more likely to recognize the write-down voluntarily, while companies just meeting the minimum profitability requirement to avoid being delisted are more reluctant to do so. Although we expect companies that perform well, as measured by GOOD, to be more likely to write-down assets, we find the opposite. Based on this finding, we conclude that income smoothing does not play a significant role in motivating the managers of our sample firms to recognize asset write-downs voluntarily, possibly because other concerns, such as the managers' desire for career advancement, dominate.
Panel C of Table 3 contains the results of estimating our logistic and OLS regression equations. Both models are highly significant with p-values less than 0.000, which suggests that the independent variables explain variation in the write-down decision (in the logistic regression) and the write-down amount (in the least-squares regression). Second, the logistic and OLS regression results are consistent with those of the frequency analysis. Companies with CEO changes (CEO) and big losses (LOSS) are more likely to write-down assets voluntarily and in larger amounts, whereas companies with above-target profit performance (GOOD) are less likely to voluntarily write-down their assets and tend to record smaller write-down charges. Finally, companies reporting only marginal profits (DELIST) tend to avoid write-downs, although the association between DELIST and write-down amount is not statistically significant
V. FIRM CHARACTERISTICS AND DIFFERENTIAL VALUATION IMPLICATIONS
In this section of the paper, we examine whether the positive valuation effect of the voluntary write-down interacts with CEO changes or big losses. Such evidence is potentially important for distinguishing the alternative explanations we discussed earlier. If the positive valuation effect is a result of the market's adjustment to the overestimation of a firm's asset impairment before the voluntary write-down, then this positive stock price effect should exist for all voluntary write-down firms regardless of CEO changes or big losses. On the other hand, if the positive valuation effect is better explained by signaling, then we should expect to find differential valuation effects for firms with and without CEO changes or big losses. Recall that Strong and Meyer (1987) conclude that the primary reason for a write-down decision is a senior management change, which, when combined with a positive stock price effect, implies that the write-down is perceived by the capital market as a signal of future events. In addition, firms suffering large losses may also signal their preparedness to deal with their financial difficulties to the market by choosing the voluntary write-down.
We expand the two valuation models (Models 1 and 2) to include the interaction between WD and CEO or LOSS. We then construct F-tests to compare the valuation effect of a voluntary write-down for firms with and without these two characteristics. CEO and LOSS are coded I only for the voluntary write-down firms with these characteristics. Table 4 presents the results of this refined analysis.
Overall, both the return and price models provide supporting evidence for the signaling explanation. The F-tests of WD + WD_98 are generally insignificant, which suggests that the positive valuation effect of the voluntary write-down decision disappears after we control for CEO and/or LOSS. Instead, the voluntary write-down decision has a positive valuation effect only for firms with CEO changes and/or big losses, as evidenced in both Panels A and B of Table 4 by the highly significant F-statistics related to our tests of various combinations of these variables. For example, the significant F-statistics of WD + WD_98 + WD_CEO from the regressions without and with industry dummies (p = 0.017 and p = 0.015 in Panel A and p = 0.000 and p = 0.002 in Panel B, respectively) suggest that the voluntary write-downs have significantly positive valuation implications only for firms with CEO changes.
The interaction term WD_LOSS is positive and significant in the return model. This result suggests that the market reaction to the write-down is significantly more positive for firms that recognized the voluntary write-down and would report a loss regardless of whether the write-down is recorded. If markets are rational, then this result supports the signaling explanation that the disclosure of asset impairment by firms that report losses with or without the write-down is perceived to indicate a turnaround. However, the interaction term WD_CEO is insignificant, which is not consistent with our expectation.
VI. ASSET WRITE-DOWNS AND FUTURE PERFORMANCE
Finally, we compare firm performance in the year following the asset write-down to examine whether the voluntary write-down firms outperform the control firms. Such evidence would certainly strengthen the signaling interpretation for the positive market valuation effect. Our basic model that links the write-down decision to future performance improvement is modified based on the Aboody et al. (1999) model as follows:
(4) [DELTA]PER[F.sub.t+1]=[[beta].sub.0] + [[beta].sub.1] [DELTA]PER[F.sub.t] + [[beta].sub.2]SIZE + [[beta].sub.3] CEO + [[beta].sub.4]LOSS + [[beta].sub.5]WD98 + [member of].
We use two different accounting metrics for the performance variable PERF: return on assets (NIROA) and operating cash flows over total assets (CASH). Both measures are calculated as an industry median-adjusted change over two years. We define the dependent variable [DELTA]PER[F.sub.t+1] as the change in the performance measure during the year following the asset write-down. Since the test sample voluntarily wrote down their assets in 1998, [DELTA]PER[F.sub.t+1] is computed between 1998 and 1999 for these firms. The control sample recorded the mandatory write-down in 1999 and hence, [DELTA]PER[F.sub.t+1] is computed between 1999 and 2000 for this sample, which places each sample on an equal footing in event time.
To control for the time-series properties of performance, Model (4) includes [DELTA]PER[F.sub.t], the change in performance one year before the write-down. [DELTA]PER[F.sub.t] is computed between 1997 and 1998 for the voluntary write-down sample and between 1998 and 1999 for the control sample. The logarithm of total assets at the end of 1997 (SIZE) serves as a control for the effect of firm size on profitability. CEO and LOSS are the two incentive variables that are believed to indicate future improvement, and they are significantly associated with the voluntary write-down decision. WD98 is the indicator variable to distinguish the voluntary write-down sample from the control sample. A significantly positive coefficient on WD98 suggests that the voluntary write-down firms have a greater improvement in performance one year after the write-down than the mandatory write-down firms. Such a finding would support the conclusion that the voluntary write-down signals the future improvement in performance after controlling the effect of CEO and LOSS. We then expand the basic performance model to include interactions between WD98 and CEO or LOSS to examine whether these incentive variables affect the association between the voluntary write-down and performance improvement. The results of this analysis are presented in Table 5. Panel A presents the results of estimating the basic model and Panel B presents the results of estimating the expanded model.
As revealed in Panel A (the column for [DELTA]NIROA) of Table 5, we find that the voluntary write-down is significantly associated with future profitability improvement as measured by [DELTA]NIROA (p = 0.017) after controlling prior year performance change and the potential impact of size. In addition, the coefficient on LOSS is significantly positive in the [DELTA]NIROA model, while CEO is not a significant explanatory factor in this model.
The first two columns of Panel B of Table 5 provide evidence of whether the positive relationship between the voluntary write-down and [DELTA]NIROA is stronger for firms with CEO changes or big losses. After including the two interaction terms (WD98_CEO and WD98_LOSS), the coefficient on WD98 becomes insignificant, which suggests that the association between the voluntary write-down and NIROA improvement does not exist for firms without CEO changes or big losses. The association between the voluntary write-down and future performance is significantly more positive for firms with CEO changes and losses, although the level of significance is higher for WD98_LOSS (p = .000 for one- as well as two-tailed p-values), but marginal for WD98_CEO (one-tailed p = .091). Finally, the joint F-tests achieves statistical significance for the test of WD98 + WD98_LOSS (p = 0.000) only, which suggests that the positive relationship between the voluntary write-down and performance improvement exists only for firms with big losses, not for firms with CEO changes.
While measuring performance improvement by ROA is vulnerable to the possibility that the performance improvement we observe is an artifact of accrual basis accounting, our research design is nonetheless biased downward against finding such an improvement. In the years following a write-off, future expenses are lower as is the firm's asset base, resulting in higher ROA. Since we find that the voluntary write-down amounts are smaller than the mandatory write-down amounts (see the descriptive statistics in Table 1), the mandatory write-down firms stand to gain the larger mechanical improvement in ROA. However, even with this downward bias, we cannot completely rule out the effect of accrual accounting reversal. To provide additional evidence to bolster our position that the market responded positively to voluntary write-down (because it signals future improvements in "real" economic performance), we further examine changes in operating cash flow. Unfortunately, the cash flow-based results do not provide any support for the hypothesized association of the voluntary write-down and performance improvement. As presented in the column for [DELTA]CASH in both panels, neither the WD98 variables nor the joint F-tests are significant.
Given the insignificant results based on operating cash flows our findings regarding the ex post association between the voluntary write-down and performance improvement are inconclusive at best. Two explanations are possible for the inconsistency between net income and operating cash flows. First, even with a downward bias, we cannot eliminate the possibility that the observed relationship between the voluntary write-down and the net income improvement is an artifact of accrual accounting. However, under such circumstances we should not observe a positive market response to the write-down if market participants are rational. Second, our one-year ahead time horizon may not be sufficiently long to capture any real performance improvement. Whatever the actual reason may be, more conclusive evidence awaits future research.
VII. SUMMARY AND CONCLUSION
The incentives for and consequences of accounting method choice are central issues to many constituents of accounting. This study examines an income-decreasing accounting choice in the form of initial asset write-downs by firms in China. Two recent standard-setting events provide us with a unique research opportunity to clearly identify a test sample of firms that voluntarily wrote down assets for the first time in 1998, and a control sample of firms that suffered from asset impairments before 1998 but chose not to write-down until the following year when write-downs became mandatory. Hence, our research setting is free from many problems identified in the asset write-down literature.
Overall, we find the following. First, the voluntary asset write-downs have a positive valuation effect. Second, firms with CEO changes or big losses are more likely to write-down assets and tend to write-down assets by a larger amount. Third, our refined valuation analysis reveals that the positive valuation effect exists only for firms with CEO changes and/or big losses. Finally, we document an expost association between the voluntary asset write-down and subsequent performance improvement in terms of return on assets but not in terms of cash flows. Taken together we believe these results are most (albeit not entirely) consistent with a signaling explanation for managers' decision to voluntarily write-down assets. Specifically, firms with large losses and CEO changes recognize the write-off to signal to the market that they are prepared to deal with their firms' problems. The market, observing the signal, reacts positively. A key assumption implicit in this interpretation of our results is that participants in the Chinese stock market are rational.
In addition to taking advantage of a unique research opportunity, this study adopts a broad approach, based on insights and findings from the extant literature, to examining the incentives for and consequences of voluntary asset write-downs. Our findings add new evidence to the accounting choice literature in an international setting. However, this study is not without limitations. While we attribute the positive valuation effect to the signaling explanation and provide some evidence consistent with tins explanation, our research design and empirical results do not allow us to completely rule out alternative explanations. (9) In addition, conclusions drawn from a one-time-only event in China may not be generalizable to other markets and/or to other write-down decisions made under different conditions. Also, the exclusion of accounts receivable write-downs due to data unavailability may limit our ability to make full use of this unique research opportunity. Although not every limitation can be fully addressed, some certainly point out directions for future research.
TABLE 1
Sample Selection Description
Panel A: Sample Selection
Total number of A-share company in 1998 727
(Companies with both A- and B-shares are
excluded)
Companies that voluntarily wrote down assets in
1998:
Write-down case: Short-term investment 8
Inventory 116
Long-term investment 25
Total number of voluntary write-down companies 134 (18%) (a)
Companies that had asset impairment before 1998
but did not write-down until 1999
Write-down case: Short-term investment 29
Inventory 421
Long-term investment 173
Total number of mandatory write-down companies 456 (55%) (a)
Total number of voluntary and mandatory
write-down companies 590
Number of companies with H-shares (b) 14
Number of companies with missing values 39
Total number of sample companies 537
Panel B: Magnitude of Voluntary Write-Down in 1998
[absolute value
RMB % Total % Specific of % of Net
Type (000) Assets Assets income] (c)
Short-term investment 4,433 0.90 32.92 4.63
Inventory 11,741 0.90 4.83 10.36
Long-term investment 2,431 0.33 6.14 5.88
Total write-down 10,882 0.89 3.32 10.15
Panel C: Magnitude of Mandatory Write-Down in 1999 (adjusted
retroactively)
[absolute value
RMB % Total % Specific of % of Net
Type (000) Assets Assets income] (c)
Short-term investment 7,334 0.57 34.59 6.01
Inventory 10,835 1.03 7.38 32.66
Long-term investment 7,370 0.69 21.79 9.39
Total write-down 13,266 1.25 5.94 48.05
(a) Percentage of A-share companies at year end.
(b) These are A-share companies that also issue H-shares. They are
excluded due to the mandatory write-down requirement for these
companies in 1998 (see note a).
(c) Before write-down effect.
TABLE 2
Assets Write-Down and Stock Valuation
Panel A: Annual Return as Dependent Variable
Without Industry Dummies
t-stat.
Independent Variable Coefficient (two-tailed p)
Intercept 2.975 9.55 (0.000)
WD98 0.039 1.02 (0.308)
EPS 4.635 6.27 (0.000)
[DELTA]EPS 1.940 2.45 (0.015)
SIZE -0.211 -9.71 (0.000)
WD -0.326 -0.20 (0.840)
WD_98 12.059 3.49 (0.000)
F-test (p-value)
WD + WD_98 14.74
n 527
Adjusted [R.sup.2] 0.286 (0.000)
With Industry Dummies (a)
t-stat.
Independent Variable Coefficient (two-tailed p)
Intercept 3.286 9.11 (0.000)
WD98 0.035 0.90 (0.370)
EPS 4.973 6.55 (0.000)
[DELTA]EPS 1.637 2.01 (0.045)
SIZE -0.223 -9.58 (0.000)
WD -0.473 -0.29 (0.775)
WD_98 12.508 3.55 (0.000)
F-test (p-value)
WD + WD_98 15.17
n 527
Adjusted [R.sup.2] 0.296 (0.000)
Panel B: Year-End Price Per Share of Equity as Dependent Variable
Without Industry Dummies
t-stat.
Independent Variable Coefficient (two-tailed p)
Intercept 38.490 11.13 (0.000)
WD98 0.207 0.54 (0.589)
BVPS 0.364 1.79 (0.074)
NI 10.611 14.76 (0.000)
SIZE -1.921 -8.77 (0.000)
WD -23.163 -1.16 (0.249)
WD_98 41.652 2.04 (0.042)
F-test (p-value)
WD + WD_98 18.93 (0.000)
n 522
Adjusted [R.sup.2] 0.408 (0.000)
With Industry Dummies (a)
t-stat.
Independent Variable Coefficient (two-tailed p)
Intercept 42.279 13.15 (0.000)
WD98 0.241 0.67 (0.053)
BVPS 0.498 2.66 (0.503)
NI 10.132 15.41 (0.008)
SIZE -2.350 -11.18 (0.000)
WD -30.190 -1.61 (0.018)
WD_98 45.968 2.41 (0.016)
F-test (p-value)
WD + WD_98 16.56 (0.000)
n 522
Adjusted [R.sup.2] 0.526 (0.000)
(a) Industry dummies are not presented.
WD98 = dummy variable with a value of 1 for voluntary asset
write-down companies in 1998;
[DELTA]EPS = change in earnings per share from 1997 to 1998
before write-down in 1998 over beginning price;
EPS = earnings per share before write-down in 1998 over
beginning price;
SIZE = natural logarithm of market value;
BVPS = book value of assets per share;
NI = net income per share;
WD = the amount of asset impairment up to 1998 over beginning
market value; and
WD_98 = interaction term between WD98 and WD.
TABLE 3
Determinants of Write-Down Decisions
Panel A: Univariate Analysis
t-test of Mean Difference
Independent Test Control t-stat
Variable Sample Sample (p-value)
DR97 0.464 0.451 -0.72 (0.474)
MGTSHR 0.002 0.002 -0.35 (0.726)
SIZE 13.295 13.392 1.2 (0.232)
Wilcoxon Test of Median Difference
Independent Test Control z-stat
Variable Sample Sample (p-value)
DR97 0.476 0.459 0.6213 (0.534)
MGTSHR 0.005 0.004 -0.4548 (0.649)
SIZE 13.235 13.282 -1.0031 (0.316)
Panel B: Frequency Test
Observed Frequency (Expected Frequency)
Variable Test Sample Control Sample Chi-Square (p-value)
LOSS 22 (8.790) 18 (31.210) 27.494 (0.000)
GOOD 80 (92.29) 340 (327.71) 9.628 (0.002)
CEO 31 (21.32) 66 (75.685) 6.884 (0.001)
DLIST 2 (6.37) 27 (22.628) 4.065 (0.044)
Panel C: Multivariate Analysis
Logistic Regression (a)
Independent
Variable Prediction Coefficient Wald [chi square] (p)
Intercept ? 1.355 0.471 (0.493)
SIZE ? -0.192 1.655 (0.198)
DR97 + 0.561 0.690 (0.406)
MGTSHR + 4.824 0.020 (0.887)
DELIST - -1.805 5.036 (0.025)
GOOD + -0.59 2.923 (0.087)
LOSS + 0.942 4.216 (0.040)
CEO + 0.468 3.140 (0.076)
n (test
sample) 537 (118)
[R.sup.2] 0.096 (0.000)
Least-Squares Regression (b)
Independent t-value
Variable Coefficient (Pr > |t|)
Intercept 0.011 1.51 (0.132)
SIZE -0.001 -1.22 (0.223)
DR97 0.003 1.28 (0.202)
MGTSHR -0.066 -0.52 (0.603)
DELIST -0.001 -0.63 (0.532)
GOOD -0.003 1.91 (0.056)
LOSS 0.009 4.27 (0.000)
CEO 0.002 1.86 (0.064)
n (test
sample) 537 (118)
Adjusted [R.sup.2] 0.102 (0.000)
(a) Dependent variable is coded 1 for voluntary write-down companies.
(b) Dependent variable is total voluntary write-downs in 1998 over
beginning assets.
SIZE = natural logarithm of beginning market value;
DR97 = beginning debt ratio;
MGTSHR = percentage of shares held by management;
DELIST = dummy variable, coded 1 if 0 < ROE <= 1% before write-downs,
and 0 otherwise;
GOOD = dummy variable, coded 1 for reporting ROE > 6% irrespective of
write-downs;
LOSS = dummy variable coded 1 for reporting loss with or without
write-downs in 1998; and
CEO = dummy variable, coded 1 if the company changed its CEO in
1998, and 0 otherwise.
TABLE 4
Differential Valuation Implication of Asset Write-Down
Panel A: Annual Return as Dependent Variable
Without Industry Dummies
Independent Variable Coefficient t-value Pr > |t|
Intercept 2.860 9.20 0.000
WD98 0.018 0.42 0.678
EPS 5.646 7.20 0.000
[DELTA]EPS 2.138 2.70 0.007
SIZE -0.204 -9.46 0.000
LOSS 0.023 0.28 0.776
CEO 0.136 1.28 0.203
WD -0.277 -0.17 0.862
WD 98 -1.845 -0.21 0.833
WD-CEO 10.484 1.26 0.210
WD-LOSS 19.349 2.49 0.013
F-test (p-value)
WD+WD_98 0.06 (0.805)
WD+WD_98+WD_CEO 5.73 (0.017)
WD+WD_98+WD_LOSS 13.49 (0.000)
WD+WD-98+WD-CEO+WD_LOSS 0.300 (0.000)
Adjusted [R.sup.2]
Panel B: Year-End Price Per Share of Equity as Dependent Variable
Without Industry Dummies
Independent Variable Coefficient t-value Pr > |t|
Intercept 33.803 11.44 0.000
WD98 -0.0781 -0.21 0.836
BVPS 0.209 1.07 0.287
NI 15.153 16.78 0.000
CEO 0.292 0.78 0.435
LOSS 5.603 7.14 0.000
SIZE -1.953 -9.37 0.000
WD -18.115 -0.96 0.336
WD_98 33.804 1.58 0.114
WD_CEO 4.033 0.46 0.644
WD-LOSS 5.110 0.59 0.557
F-test (p-value)
WD+WD_98 2.63 (0.105)
WD+WD_98+WD_CEO 13.78 (0.000)
WD+WD 98+WD_LOSS 9.40 (0.002)
WD+WD_98+WD_CEO+WD LOSS 9.50 (0.002)
Adjusted [R.sup.2] 0.466 (0.000)
With Industry Dummies
Independent Variable
Intercept 3.145 8.69 0.000
WD98 0.021 0.45 0.650
EPS 5.873 7.33 0.000
[DELTA]EPS 1.892 2.31 0.021
SIZE -0.215 -9.20 0.000
LOSS 0.001 0.02 0.987
CEO 0.129 1.19 0.235
WD -0.341 -0.21 0.836
WD 98 -0.120 -0.01 0.989
WD-CEO 9.155 1.09 0.276
WD-LOSS 18.667 2.40 0.017
WD+WD_98 0.00 (0.957)
WD+WD_98+WD_CEO 5.99 (0.015)
WD+WD_98+WD_LOSS 8.05 (0.005)
WD+WD-98+WD-CEO+WD_LOSS 13.12 (0.000)
Adjusted [R.sup.2] 0.308 (0.000)
Independent Variable
Intercept 40.789 13.18 0.000
WD98 0.007 0.02 0.985
BVPS 0.339 1.86 0.063
NI 13.989 16.64 0.000
CEO 0.268 0.78 0.436
LOSS 4.670 6.48 0.000
SIZE -2.341 -11.6 0.000
WD -23.763 -1.32 0.188
WD_98 40.861 2.04 0.042
WD_CEO -1.706 -0.21 0.831
WD-LOSS 3.404 0.43 0.668
WD+WD_98 3.91 (0.048)
WD+WD_98+WD_CEO 10.22 (0.002)
WD+WD 98+WD_LOSS 11.26 (0.000)
WD+WD_98+WD_CEO+WD LOSS 6.68 (0.010)
Adjusted [R.sup.2] 0.566 (0.000)
n = 527 in Panels A and 522 in Panel B. Industry dummies are not
presented.
WD98 = dummy variable with a value of 1 for voluntary asset
write-down companies in 1998;
[DELTA] EPS = change in earnings per share from 1997 to 1998 before
write-down in 1998 over beginning price;
EPS = earnings per share before write-down in 1998 over
beginning price;
SIZE = natural logarithm of market value;
BVPS = book value of assets per share;
NI = net income per share;
WD = the amount of asset impairment up to 1998 over beginning
market value;
WD_98 = interaction term between WD98 and WD;
LOSS = coded 1 for reporting loss with or without write-downs
and wrote down assets in 1998;
CEO = coded 1 if the company changed its CEO and wrote down
assets in 1998, and 0 otherwise;
WD_CEO = interaction term between WD and CEO; and
WD_LOSS = interaction term between WD and LOSS.
TABLE 5
Assets Write-Down and Future Performance
Panel A: Basic Results
[DELTA]NIRO[A.sub.t+1]
Dependent Variable Estimate t-value (Pr > |t|)
Intercept -0.048 -1.12 (0.263)
SIZE 0.003 0.86 (0.388)
[DELTA]NIRO[A.sub.t] -0.333 -8.72 (0.000)
[DELTA]CAS[H.sub.t]
CEO 0.006 0.95 (0.344)
LOSS 0.068 6.94 (0.000)
WD98 0.014 2.40 (0.017)
Adjusted [R.sup.2] 0.345 (0.000)
Panel B: Expanded Results with Interaction Terms
Intercept -0.043 -1.07(0.284)
SIZE 0.001 0.44 (0.662)
[DELTA]NIRO[A.sub.t] -0.314 -8.64 (0.000)
[DELTA]CAS[H.sub.t]
CEO 0.026 2.45 (0.015)
LOSS -0.004 -0.55 (0.580)
WD98 -0.006 -0.90 (0.367)
WD98-CEO 0.018 1.34 (0.181)
WD98-LOSS 0.117 7.12 (0.000)
Adjusted [R.sup.2] 0.423 (0.000)
F-test for WD98 + WD98_CEO 0.94 (0.332)
F-test for WD98 + WD98_LOSS 47.51 (0.000)
[DELTA]CAS[H.sub.t+1]
Dependent Variable Estimate t-value (Pr > |t|)
Intercept 0.165 0.97 (0.332)
SIZE -0.013 -1.00 (0.316)
[DELTA]NIRO[A.sub.t]
[DELTA]CAS[H.sub.t] -0.719 -15.89 (0.000)
CEO -0.016 -0.61 (0.540)
LOSS 0.003 0.08 (0.940)
WD98 0.015 0.54 (0.586)
Adjusted [R.sup.2] 0.322 (0.000)
Panel B: Expanded Results with Interaction Terms
Intercept 0.190 1.11 (0.266)
SIZE -0.014 -1.07 (0.284)
[DELTA]NIRO[A.sub.t]
[DELTA]CAS[H.sub.t] -0.715 -15.82 (0.000)
CEO -0.216 -2.08 (0.038)
LOSS 0.008 0.07 (0.948)
WD98 0.000 0.00 (1.000)
WD98-CEO 0.214 1.99 (0.047)
WD98-LOSS -0.01 -0.08 (0.932)
Adjusted [R.sup.2] 0.324 (0.000)
F-test for WD98 + WD98_CEO 0.01 (0.931)
F-test for WD98 + WD98_LOSS 1.65 (0.199)
n = 535 in all cells.
[DELTA]NIRO[A.sub.t+1] = industry median-adjusted change in net income
ROA between 1999 (2000) and 1998 (1999) for
test control) companies;
[DELTA]CAS[H.sub.t+1] = industry median-adjusted change in operating
cash flow over total assets between 1999
(2000) and 1998 (1999) test (control)
companies;
[DELTA]NIRO[A.sub.t] = industry median-adjusted change in net income
ROA before asset write-down between 1998
(1999) and 1997 (1998) for test (control)
companies;
[DELTA]CAS[H.sub.t] = industry median-adjusted change in operating
cash flow over total assets between 1998
(1999) and 1997 (1998) for test (control)
companies;
CEO = dummy variable, coded 1 if the company changed
its CEO in 1998, and 0 otherwise;
LOSS = dummy variable coded 1 for reporting loss with
or without write-downs in 1998;
WD98 = dummy variable with a value of 1 for voluntary
asset write-down companies in 1998;
WD98_CEO = interaction term between WD98 and CEO;
WD98_LOSS = interaction term between WD98 and LOSS.
FIGURE 1
Changes of Accounting Standards in 1998 and 1999: Asset Write-Downs
Pre-1998 1998 1999
Historical Cost Voluntary Write-Downs Mandatory Write-Downs
134 A-share companies 456 A-share companies
(test sample) (control sample)
The entire write-down The write-down amount
amount recorded as expenses divided into two parts:
in the Income Statement 1: Prior period
adjustments to the
beginning Retained
Earnings (for asset
impairment before 1999)
2: Expenses in the
Income Statement (for
asset impairment during
1999)
We thank the editor, the anonymous associate editor, and two reviewers for their helpful comments and suggestions that have improved this manuscript. The late editor, Professor R. S. Olusegun Wallace, was responsible for the majority of the review process, and we are grateful for many of his insights. Professor Shimin Chan thanks the University of Louisiana at Lafayette for its financial support. Professor Charles Chen and Professor Su gratefully acknowledge the financial support from Strategic Research Grant No. 7001307 provided by City University of Hong Kong.
(1) A-shares are issued to domestic investors, while B-shares are issued to investors outside the Chinese mainland. There were 921 A-share companies and 108 B-share companies at the end of 1999. Most of these B-share companies issued A-shares as well. In addition, some Chinese companies issued shares in overseas markets such as Hone Kung (known as H-shares). Although the 1998 regulation allowed A-share companies to voluntarily write-down assets, the asset write-down was mandated for companies with B- or H-shares. Hence, this study includes only these companies that issue A-shares exclusively.
(2) In addition to these three asset write-downs, the new regulation also allowed A-share companies to make provisions for doubtful accounts based on their judgment rather than within the allowed percentages (i.e., 0.3-0.5 percent of the outstanding balances of accounts receivable) as previously specified. Collectively, these items are known as the four free-choice provisions. However, two masons prevented us from accurately identifying the voluntary part of the bed debt allowance made in 1998. First, companies were allowed to provide bad debt allowance based on the government approved percentages before 1998; second, we did not have bad debt write-off information to compute the bad debt allowance flowing through the income statement in 1998. Consequently, this study excludes bad debt allowance and focuses on the true initial voluntary write-downs of inventory and short- and long-term investments.
(3) Although the 1998 regulation mandated asset write-downs for companies with B or H shares, the regulation was unambiguous in that A-share companies could "continue to apply the same accounting methods in their financial reports."
(4) Hidden asset losses were widely recognized as one of the most frequently used methods to overstate earnings in China before the 1998 regulation. Empirical Studies on Capital Market Related Accounting Issues (edited by Jiang end Li 1998) documents consistent evidence of unreported asset losses. More evidence of hidden losses in assets before 1998 regulation is reported in a book titled A Study on Earnings Management that was published in 2000 (Wei et al. 2000).
(5) We thank an anonymous reviewer for raising this issue and the editor for suggesting the technique that we adopted to test the valuation effect of the voluntary versus mandatory write-downs (see Models 1 and 2).
(6) Although a significantly positive valuation effect based on the event study method constitutes strong evidence that a voluntary write-down signals good news to the market, the result hinges on an important assumption that no confounding value-relevant events exist during the study window. In this paper, we present results from the return and price model, but an event study of the write-down announcement day return produces similar, though weaker, results.
(7) All regression results, including those in the following tables, are based on data that is winsorized at one percentile before estimating the regression models.
(8) We are grateful to the editor for suggesting this alternative explanation.
(9) We are grateful to both reviewers for directing our attention to this important limitation.
REFERENCES
Aboody, D., M. E. Barth, and R. Kasznik. 1999. Revaluations of fixed assets and future firm performance: Evidence from the U.K. Journal of Accounting and Economies 26: 149-178.
Aharony, J., C. J. Lee, and T. J. Wong. 2000. Financial packaging of IPO firms in China. Journal of Accounting Research 38 (1): 103-126.
Alciatore, M., C. C. Dee, P. Easton, and N. Spear. 1998. Asset write-downs: A decade of research. Journal of Accounting Literature 17: 1-39.
Barth, M. E., and G. Clinch. 1998. Revalued financial, tangible, and intangible assets: Association with share prices and non-market-based value estimates. Journal of Accounting Research 36: 199-233.
--, W. H. Beaver, and W. R. Landsman. 2001. The relevance of the value relevance literature for financial accounting standard setting: Another view. Journal of Accounting and Economics 30: 77-103.
Company Law of the People's Republic of China. 1994. Asian Law Journal 2 (8): 34-40.
Chen, C. J., S. Chen, and X. Su. 2001a. Is accounting information value-relevant in the emerging Chinese stock market? Journal of International Accounting, Auditing, and Taxation 10: 1-22.
--, --, and --. 2001b. Profitability regulation, earnings management, and modified audit opinions: Evidence from China. Auditing: A Journal of Practice & Theory 20 (2): 9-30.
Chen, J. P. C., X. Su, and R. Zhao. 2000. An emerging market's reaction to initial modified audit opinions: Evidence from the Shanghai stock exchange. Contemporary Accounting Research 17 (3): 429-455.
Chen, K. C. W., and H. Q. Yuan. 2001. Earnings management and capital resource allocation: Evidence from China's accounting-based regulation of rights issues. Working paper, Hong Kong University of Science and Technology.
DeAngelo, H., L. DeAngelo, and D. J. Skinner. 1994. Accounting choice in troubled companies. Journal of Accounting and Economics 17:113-143.
Easton, P. D., P. H. Eddey, and T. S. Harris. 1993. An investigation of revaluation of tangible long-lived assets. Journal of Accounting Research 31 (Supplement): 1-38.
Eccher, E., and P. M. Healy. 2000. The role of international accounting standards in transitional economies: A study of the People's Republic of China. Working paper, Massachusetts Institute of Technology.
Elliott, J. A., and W. H. Shaw. 1988. Write-offs as accounting procedures to manager perceptions. Journal of Accounting Research 26 (Supplement): 91-119.
Fields, T., T. Lys, and L. Vincent. 2001. Empirical research on accounting choices. Journal of Accounting and Economics 3 (September): 255-307.
Francis, J., D. Hanna, and L. Vincent. 1996. Causes and effects of discretionary asset write-offs. Journal of Accounting Research 34 (Supplement): 117-134.
Haw, I. M., D. Qi, W. Wu, and W. Zhang. 1998. Earnings management of listed firms in response to security regulations in China's emerging capital market. Working paper, Chinese University of Hung Kong.
Holthansen, R. W. 1990. Accounting method choice: Opportunistic behavior, efficient contracting, and information perspectives. Journal of Accounting and Economics 12:207-218.
Kothari, S. P., and J. L. Zimmerman. 1995. Price and return models. Journal of Accounting and Economics 20: 155-192.
Jiang, Y., and S. Li. 1998. Empirical Studies on Capital Market Related Accounting Issues. Chinese version. Shanghai, China: Shanghai University of Finance & Economies Press.
Li, Z. 2001. Empirical research on the asset write-downs in Chinese listed companies. China Accounting and Finance Review 3 (4): 114-157.
Strong, J. S., and J. R. Meyer. 1987. Asset write-downs: Managerial incentives and security returns. Journal of Finance 42:643-661.
Subramanyam, K. R. 1996. The pricing of discretionary accruals. Journal of Accounting and Economics 22: 249-281.
Watts, R. L., and J. L. Zimmerman. 1990. Positive accounting theory: A ten-year perspective. The Accounting Review 65: 131-156.
Wei, M. H., J. Tan, and S. Lin. 2000. A Study on Earnings Management. Chinese version. Beijing, China: China Economics and Finance Press.
Zucea, L. J., and D. R. Campbell. 1992. A closer look at discretionary write-downs of impaired assets. Accounting Horizons (September): 30-41.
Charles J. P Chen is an Associate Professor at the City University of Hong Kong, Shimin Chen is an Associate Professor at the University of Louisiana at Lafayette and an Associate Professor at Lingnan University, Hong Kong, Xijia Su is an Associate Professor at the City University of Hong Kong, and Yuetang Wang is a Professor at Nanjing University.
In addition, make sure to read these articles:
- Janus Capital Group Announces Fourth Quarter and Year-End 2003 Results.
- Delta Financial Turns Cash Flow Positive for theFourth Quarter.
- DURA Automotive Announces Strategic Actions.
- $2 Bil Charge Fuels Rtl '01 Loss
- The Earle M. Jorgensen Co. announces financial results.
- Eastman reports record 4th quarter and 1995 sales; earnings.
- Opting out of the US.
- Withdrawal from the US.
- Step change programme introduced to North American operations.
- Eastman announces 4th quarter and 1996 sales,earnings.
- Meet SFAS-121: FASB's recent attempt to bring asset values in line with "real" values.
- Companies forecast disappointing first quarter results. (Executive Briefing).
- Total year 2000 results
- HECLA REPORTS ASSET WRITE-DOWNS, ANNOUNCES PREFERRED DIVIDEND AND ELECTS NEW DIRECTOR.
- Worries Weigh Down Vu
Email
Digg It
del.icio.us
