Audit qualification and timing of earnings announcements: evidence from China.
INTRODUCTIONThis study investigates the effects of audit opinions and earnings surprises on the timeliness of annual earnings announcements of listed Chinese firms, after controlling for firm size, the presence of losses, financial distress, auditor switches, and changes in the Chinese regulatory environment. (1) Our research contributes to the existing literature in two dimensions. It extends the research on the determinants of the timeliness of annual earnings announcements made in mature economies to the emerging capital market in China. Moreover, in addition to the main effects scrutinized in previous studies, it includes the interaction effect between audit opinions and earnings surprises in regression analysis. Such an enhancement offers additional insights on whether the previously documented good news early and bad news late phenomenon is conditional on audit opinions.
This study builds on previous research findings in the U.S. and Australian markets that suggest that both audit opinions and earnings surprises are significant determinants of the timeliness of annual earnings announcements. Whittred (1980), Keller (1986), and Bamber et al. (1993) report that qualified audit opinions delay the release of preliminary earnings and auditor reports. Pastena and Ronen (1979), Givoly and Palmon (1982), Kross and Schroeder (1984), Chambers and Penman (1984), and Penman (1984) find that managers release good news earlier and bad news later than expected, and that timely filing is one of the most consistent determinants of the informativeness of disclosures. The issue of timeliness has recently become popular again. (2) For example, Begley and Fischer (1998) reassess whether prior findings still hold for the period of 1983-1992, when the litigation environment changed significantly. Carter and Soo (1999) also investigate the timeliness of Form 8-K reports.
This study extends this line of timing research into the Chinese market. Our motivation is threefold. First, the Chinese stock market has developed rapidly in the last decade, from being virtually nonexistent to the second largest in Asia (after Japan) in terms of market capitalization. More involvement by international investors is expected in light of China's continuing economic growth and recent entry into the World Trade Organization. These international investors are likely to be interested in the information dissemination process in the Chinese market, including what factors determine the timeliness of annual earnings announcements.
Second, annual earnings announcements are more important as a source of firm-specific information in China than in mature markets, due to the absence of alternative timely sources of information. Unlike mature markets such as the U.S., little firm-specific information is available to investors prior to annual earnings releases. Listed Chinese firms seldom provide voluntary information or earnings forecasts. (3) Moreover, financial analysts and the financial press are still in their infancy and cannot act effectively as the information intermediaries between listed firms and investors. Consequently, investors have little information available to predict earnings and especially have no warnings of negative earnings surprises before their official releases. (4) The timing pattern of good news early, bad news late, if it exists, could be a valuable source of firm-specific information in China.
Third, a recent study indicates that the quality and independence of Chinese audit firms have significantly improved in recent years (DeFond et al. 1999). Since the first batch of independent auditing standards became effective in January 1996, a significant number of listed firms have received modified opinions. The relative inexperience of listed firms and auditors in dealing with modified opinions may make the delay of annual earnings announcements more pronounced than in mature markets. Moreover, listed Chinese firms announce annual earnings concurrently with the filing of their annual reports. This institutional feature enables us to capture simultaneously the effects of audit opinions and earnings surprises on the timeliness of annual earnings announcements and annual reports.
We develop three hypotheses based upon prior research to examine the determinants of the timeliness of annual earnings announcements in China. Specifically, we hypothesize that: (1) firms with negative earnings surprises announce their annual earnings later than firms with positive earnings surprises; (2) firms with modified audit opinions are expected to announce their annual earnings later than firms with clean opinions, holding the nature of earnings surprises constant; and (3) the timing pattern of annual earnings announcements varies by the magnitude of the earnings surprises. To test the hypotheses, our sample consists of 2,921 annual report releases during 1995-1999, including 430 modified audit opinions. Our primary variables of interest are audit opinions and earnings surprises. Control variables include those in studies of U.S. firms, such as firm size, the presence of losses, financial distress, and an auditor switch. We also include a year dummy variable to control for the possible impact of the 1997 regulatory change in China.
Consistent with previous literature, we observe that both audit opinions and earnings surprises are important determinants of the timeliness of annual earnings releases. Having controlled for the impact of size, the presence of losses, financial distress, and an auditor switch, we find that positive earnings surprises are announced earlier than negative earnings surprises by 9.9 days if the audit opinion is unqualified, and 4.9 days if it is modified. On the other hand, the audit qualification process delays the announcements of both positive and negative earnings surprises, by 13.3 and 8.3 days, respectively. Consistent with Begley and Fischer (1998), we find a significant effect attributable to the magnitude of negative earnings surprises. We also document a significant interaction effect between audit opinions and earnings surprises. Modified positive earnings surprises are released 3.4 days later than unqualified negative earnings surprises.
We also find that the Chinese Securities Regulatory Commission (CSRC), the Chinese equivalent of the Securities and Exchange Commission in the U.S., has succeeded in its effort to accelerate the annual earnings announcements by listed Chinese firms through its 1997 decree. On average, holding other variables constant, the annual earnings announcement lag has been 69.4 days shorter since 1997 than before. In addition, we find that smaller firms tend to have shorter reporting lags than larger firms after 1997, counter to the results in the U.S. This is most likely due to the CSRC regulation requiring smaller firms to release their annual reports early.
The next section provides institutional background on the development of the Chinese stock and audit markets. The third section develops hypotheses. The fourth section describes the sample and variables. The fifth section provides research design and empirical results, and the last section provides the summary and conclusion.
INSTITUTIONAL BACKGROUND
The Disclosure Framework in China
Since the reactivation of stock exchanges in the early 1990s, the stock market in China has grown rapidly. There were only eight and five stocks listed, respectively, on the Shanghai and Shenzhen Stock Exchanges in 1991. By the end of 1999, the total number of firms listed on these two exchanges had increased to 947 (as shown in Panel A of Table 1). The total market capitalization of listed Chinese firms reached RMB 2,647 billion yuan, equivalent to U.S.$319 billion (as shown in Panel B of Table 1).
To facilitate the development of the stock market, the Chinese Ministry of Finance promulgated the Accounting System for Selected Shareholding Companies in May 1992. This marked the first time that listed Chinese firms were required to prepare balance sheets, income statements, and statements of change in financial positions in conformity with international accounting practices. Since then, China has promulgated a conceptual framework for accounting standards and 17 detailed standards. In January 1998, the Ministry revised the accounting system and formalized it as the Accounting System for Shareholding Companies. Currently, all listed firms prepare financial statements in accordance with this new system along with the promulgated detailed accounting standards. (5)
At present, listed Chinese companies are required to release audited financial statements within four months following the fiscal year-end Since all firms are required by the Chinese Accounting Law to use the same calendar fiscal year-end of December 31, they must release their audited annual reports no later than April 30 of the next calendar year. (6) The CSRC has required all listed firms to publish the abstracts of annual reports in the three CSRC-designated newspapers: China Securities Daily, Shenzhen Securities Times, and Shanghai Securities Daily. Because these firms are not required to make preliminary earnings announcements, they rarely disclose earnings before filing reports to the stock exchanges and the CSRC. Therefore, the publication of the abstracts of annual reports normally represents the first public announcement of listed firms' annual results. (7) Currently, the annual report is the most significant and reliable public source of information on listed firms for investors. The abstracts published in newspapers disclose for the first time the firms' annual performance results, audit opinions, and any other significant events. (8)
As the stock market develops, more listed firms tend to release their annual reports in the last week of April, especially when their operating performance is poor (Haw et al. 2000). To mitigate the problem, the CSRC issued a circular in December 1997 instructing the two stock exchanges to arrange for the even releases of annual reports. The regulation also stipulated that the stock exchanges should assign the order of annual report releases based on the asset size of listed firms.
Development of the Auditing Market in China
The concept of professional auditing and certified public accountancy did not emerge until 1980 in China, when China allowed foreign companies to set up joint venture firms. The Regulations of the People's Republic of China on Certified Public Accountants was promulgated in 1986 to guide and regulate the new industry. A total of 500 CPA licenses were granted and 80 CPA firms were set up accordingly. The Chinese Institute of Certified Public Accountants (CICPA) was established at the end of 1988 to administer the CPA examination, which took place for the first time in 1991. The rapid expansion of the Chinese stock market has led to the increasing demand for competent and independent auditing. By the end of 2001, there were about 58,000 CPAs and 6,700 CPA firms in China.
Although the CPA designation is granted by the CICPA, the CSRC imposed its own licensing requirement for CPAs and CPA firms to audit listed Chinese firms to ensure the quality of the audit in 1996. A CPA would be granted a permit to audit listed firms' annual reports if he or she had passed the licensing examination administrated by the Ministry of Finance. A CPA firm would be granted the permit if it had at least eight CPAs on staff. (9) This requirement has effectively excluded most small CPA firms from auditing listed firms. At the end of 1998, only 103 CPA firms were eligible to do so. Panel A of Table 2 describes the development of the auditing market in China. The number of audit clients increased from 53 in 1992 to 952 in 1999. (10) The number of CPA firms auditing listed firms increased accordingly, from 16 to 90. The average number of listed clients also rose from 3 to about 11 per auditor.
As the stock market has developed, a group of large audit firms has emerged. Panel B of Table 2 provides the frequency of listed firms audited by the 10 largest audit firms during 1992-1999. The top five firms audited 34 listed firms in 1992 and 169 in 1999, which represents 64 percent and 18 percent of total listed firms in the market, respectively. We notice that the market share of the top 5 audit firms has declined over time. The average number of listed clients per top 5 auditor reached more than 36 by 1999, substantially higher than that for the non-top 5 auditors. As the number of listed firms has significantly increased, the identity of the 5 largest audit firms has changed. Only three auditors have consistently appeared among the top 5 during 1992-1999: Dahua CPAs, Shanghai CPAs, and Shenzhen Zhonghua CPAs. The fourth and fifth largest audit firms have changed over time. For example, Sichuan CPAs (top 5 during 1996-1997) was dissolved in 1998 by the CSRC due to violations of securities regulations and most of its clients were picked up by another local firm, Sichuan Deyang.
The Auditing Framework in China
Independent auditing standards have been promulgated and implemented to facilitate the development of the stock market in China. In accordance with CPA law, the CICPA issued the first batch of exposure drafts for independent auditing standards in January 1995. The first batch of new auditing standards became effective on January 1, 1996. (11) Since all firms are required by the Accounting Law to use the same calendar fiscal year-end of December 31, financial statements of fiscal year 1995 became the first batch subject to the new auditing standards. The second batch of new standards became effective a year later, followed by the third batch on July 1, 1999. (12)
The first batch includes a standard on audit opinions: Independent Auditing Standard No. 7: Audit Opinion. It specifies four types of audit opinions: unqualified, qualified, disclaimer, and adverse opinion. The standard specifies the same conditions as those of the U.S. in requiring each of the four types of opinions, except that Chinese auditors are required to issue qualified opinions if a consistency principle is violated. In the U.S., the lack of consistent application of GAAP constitutes one of the circumstances for an unqualified opinion with an explanatory paragraph. We provide various examples of qualified, disclaimer, and adverse opinions in the Appendix.
Article 17 of the standard also allows Chinese auditors to attach an explanatory note to their unqualified opinion. While the standard does not specify the circumstances that require an explanatory note to an unqualified opinion, our examination of audit reports in our sample suggests that the following conditions, which are similar to those in the U.S., justify an explanatory note to an unqualified opinion: (1) auditor agrees with a departure from a promulgated principle, (2) substantial doubt about going concern, and (3) emphasis of a matter, including the existence of significant related-party transactions, important events occurring subsequent to the post balance sheet date, and material uncertainties such as contingency and litigation disclosed in the footnotes. (13) Thus, the explanatory notes usually carry a negative view from the auditor. Furthermore, the financial statement impact of the circumstances in the notes is generally significant. Therefore, even though this type of audit opinion is unqualified, it is perceived in a similar manner as modified opinions. We provide various examples of explanatory notes attached to unqualified opinions in the Appendix.
In a recent study, DeFond et al. (1999) document that the most frequent modified opinions in China are related to a GAAP violation, asset realization, and scope limitation associated with earnings management. They report that the independence of auditing practice in China has improved significantly, as evidenced by the increasing frequency of modified opinions. Panel C of Table 2 shows that prior to 1995, only a handful of companies received modified opinions. However, 10 percent of listed firms received modified audit reports for their 1995 financial statements when the first batch of new auditing standards became effective on January 1, 1996. Since then, the number of listed firms receiving modified opinions rose steadily to a record number of 193 in 1999, representing 20 percent of total clients. Panel D of Table 2 provides evidence that the number of firms switching auditors increased over time, and more than 5 percent of listed firms did so in both 1998 and 1999.
HYPOTHESES
Previous studies in the U.S. and Australia indicate that managers release good news earlier than bad news. (14) Pastena and Ronen (1979) analyze the relative timing of Form 8-K disclosures and suggest that management acts as if it intends to delay the release of negative information. Givoly and Palmon (1982) and Kross and Schroeder (1984) find a significant relationship between the accuracy of earnings forecast and the timeliness of earnings announcements. Chambers and Penman (1984) report that the earliest announcements exhibit significant positive abnormal returns and, in general, observe that early (late) announcements are associated with positive (negative) abnormal returns at the announcement date. Kalay and Loewenstein (1986) obtain similar findings in the context of dividend announcements.
Recently, Begley and Fischer (1998) reassess the relation between earnings news and timing, due to the possibility that intensified litigation risks faced by management and auditors since the 1980s may have eliminated the good news early, bad news late phenomenon. Skinner (1994) argues that litigation concerns induce firms to preempt formal bad news earnings disclosures with voluntary disclosures. However, Begley and Fischer (1998) find consistent and robust evidence to support the good news early, bad news late phenomenon for a sample drawn from the 1980s and early 1990s.
Whittred (1980) investigates whether qualified audit opinions have an impact on the timeliness of Australian annual reports and finds that a qualified audit report delays the release of preliminary and final profit reports. Several other studies examine the determinants of audit delay (i.e., the elapsed time between the close of a fiscal year and the audit report date), such as auditor structure, auditor size, direction of earnings surprises, the presence of a loss, financial distress, and industry membership (e.g., Ashton et al. 1987, 1989; Newton and Ashton 1989; Bamber et al. 1993; Kinney and McDaniel 1993). In general, they report that the qualification process delays the release of audit reports. Keller (1986) finds that preliminary earnings releases for a sample of U.S. firms are delayed when companies are about to receive a subject-to qualification, but fails to find evidence that qualification affects the timing of annual report publication.
While previous studies in the timing literature are primarily based on the mature U.S. and Australian markets, our study extends this line of research to an emerging capital market. The following hypotheses investigate the effects of audit opinions and earnings surprises on the timeliness of annual earnings announcements (in the alternative form):
H1: Firms with negative earnings surprises announce their annual earnings later than firms with positive earnings surprises, holding audit opinions constant.
H2: Firms with modified or worse audit opinions announce their annual earnings later than firms with clean opinions, holding the nature of earnings surprises constant.
Givoly and Palmon (1982) and Kross and Schroeder (1984) indicate that the good news early, bad news late phenomenon may be driven predominantly by bad news being announced late relative to expectations, while good news does not appear to be announced earlier than anticipated. Begley and Fischer (1998) test the effects of both the direction and magnitude of earnings surprises on the timeliness of announcements. They report that while worse news is significantly delayed in comparison to bad news, better news is not announced any earlier than good news). (15) The following hypothesis (in the alternative form) examines the impact of the magnitude of earnings surprises on timeliness in the emerging Chinese market:
H3: The timing pattern of annual earnings announcements varies by the magnitude of the earnings surprises.
SAMPLE AND VARIABLES
Our sample consists of annual earnings announcements by listed Chinese firms during fiscal years 1995-1999. We start our sample from fiscal year 1995 to rule out the possible impact of different auditing practices prior to the application of the first batch of new auditing standards, effective on January 1, 1996. Starting with 3,498 announcements available for 1995-1999, we exclude 499 announcements by B- and H-share firms since they are different from A-share firms in terms of their information environment, reporting requirements, investor sophistication, and information dissemination process. (16) We further delete 78 firm-years with missing financial data from the China Database of the Taiwan Economic Journal (TEJ). This reduces our final sample to 2,921 annual earnings announcements made by 858 A-share firms during 1995-1999. We then hand-collect their announcement dates of annual earnings, names of auditors, and auditor opinions from the annual report abstracts published in the China Securities Daily, Shenzhen Securities Times, and Shanghai Securities Daily.
Following existing research, we measure the timeliness of earnings announcements by the reporting lag and the unexpected reporting lag. The reporting lag (LAG) is defined as the number of days from the fiscal year-end to the annual earnings announcement date (Whittred 1980; Whittred and Zimmer 1984; Chambers and Penman 1984; Schwartz and Soo 1996). Assuming that the reporting lag follows a random walk process, the unexpected reporting lag is then the difference between the actual and the previous year's reporting lags and is defined as the announcement delay (DEL) (Givoly and Palmon 1982; Chambers and Penman 1984; Begley and Fischer 1998). Due to the IPO firms and missing reporting dates in the previous years, the final sample for DEL is reduced to 2,256 firm-years. (17)
We measure the magnitude of earnings surprises by unexpected earnings (UE), which is defined as:
U[E.sub.it] = (N[I.sub.it] - N[I.sub.it-1])/T[A.sub.it-1],
where N[I.sub.it](N[I.sub.it-1]) is the net income for firm i in year t (t-1) and T[A.sub.it-1] is the total assets for firm i at the end of year t-1. For IPO firms, we use net income in the pre-IPO year provided in the prospectus. We use the random walk model to proxy for the market expectation because analyst earnings forecasts are generally not publicly available in China. We scale the earnings change with the lagged total asset to control for the difference in the magnitude of earnings surprises attributable to the difference in firm size. (18) We define D+ as a dummy variable that equals 1 for positive earnings surprises (UE > 0) and 0 otherwise. We also define [D.sub.LOSS] as a dummy variable that equals 1 if the firm experiences a loss in the year and 0 otherwise. In addition, we use the logarithm of total assets at the beginning of the year to measure firm size, and the debt-to-total-asset ratio (D/TA) at the beginning of the year to proxy for financial distress. All the financial data are retrieved from the TEJ database.
We define a dummy variable, [D.sub.SWITCH], to control for the possible impact of an auditor switch on reporting lag, which equals 1 if the auditor for the firm is different between year t and t-1, and 0 otherwise. For IPO firms, we compare the auditor in the IPO year with the auditor in the IPO prospectus who audited the financial statements in the pre-IPO year. In case of missing observation in year t-1, we compare the auditor in year t with the auditor in year t-2 and assume there is no switch if they are the same auditor, and vice versa. The total number of auditor switches is 120 firm-years, which is 4.2 percent of the sample.
Panel A of Table 3 presents the distribution of earnings surprises and timing measures. The mean (median) reporting lag (LAG) is 89.86 (91) days. There is a significant mean (median) announcement acceleration (DEL) of 1.53 (1) days relative to the previous year. Listed Chinese firms experience significantly positive mean (median) changes in UE during the sample period. They also vary significantly in terms of size and financial leverage.
Panel B of Table 3 presents the distribution of LAG, DEL, UE and control variables by auditor opinions. Since the disclaimer and adverse opinions are both serious in nature and less frequent, we merge them into one category. As a result, there are four types of audit opinions including, in order of severity: unqualified opinion, unqualified opinion with an explanatory note, qualified opinion, and disclaimer/adverse opinion. Of the 2,921 auditor reports, 2,491 are unqualified opinions and the remaining 430 are modified audit opinions received by 249 firms. They include 232 (7.9 percent of the sample) unqualified opinions with an explanatory note, 175 (6 percent) qualified opinions, and 23 (0.8 percent) disclaimer or adverse opinions. The distribution of mean LAG suggests that the reporting lag monotonically increases with the severity of audit opinions. A similar pattern exists for the reporting lag delay, even though the relationship is not monotonic for the disclaimer/adverse audit opinion group. Overall, modified opinions delay annual earnings announcements. Moreover, the more serious the modification, the greater is the delay. The distribution of mean UE also suggests a monotonic relationship between audit opinions and earnings changes. The F-statistics indicate that LAG, DEL, and UE are significantly different among the audit opinion groups. As the audit opinions become worse, firms experience greater financial distress and more incidences of losses and auditor switches. On the other hand, firm size does not differ significantly among different audit opinion groups.
Panel C of Table 3 compares the distributions of LAG, DEL, and other variables among the sample years to provide evidence on the impact of regulatory changes. Because the CSRC has required the two stock exchanges to arrange for the even releases of annual reports since 1997, listed companies in general were required to announce their annual earnings on a more timely basis. Indeed, the distribution of mean LAG indicates that annual earnings were announced significantly earlier in 1997-1999 than in 1995-1996, less than 90 days versus longer than 100 days. A significant mean DEL of -13.92 days in 1997 is consistent with the significant impact of the CSRC regulation on shortening the reporting lag. Consequently, we include a dummy variable in our regression to control for the effect of this regulatory change.
Panel D of Table 3 reports the Pearson correlation coefficients among the variables. As expected, the reporting lag is significantly negatively correlated with both the direction and magnitude of earnings surprises, and significantly positively correlated with the incidence of modified opinions. It is also significantly positively correlated with other variables such as size, the presence of losses, financial distress, and auditor switches. However, the results for the announcement delay, (DEL) are different. Of the three types of modified audit opinions, only qualified opinions significantly correlate with DEL. Furthermore, DEL does not correlate with financial distress (D/TA) and auditor switches. This casts some doubt on the appropriateness of DEL as the measure of the timeliness of earnings announcements in China.
To further explore this issue, we replicate Table 1 in Chambers and Penman (1984) in an attempt to investigate the stability of reporting patterns of individual Chinese firms over time. Table 4 presents summary statistics of the within-firm time-series variability of reporting lag, in comparison to the measures for U.S. firms reported by Chambers and Penman (1984). (19) As shown in Panel A of Table 4, Chambers and Penman (1984) report a mean absolute announcement delay, |DEL|, of 7.07 days in the U.S. This suggests that annual reporting dates in the U.S. can be predicted within a week with a reasonable level of accuracy. This allows previous studies in mature markets to use the announcement delay (DEL) as a proxy for the timeliness of earnings announcements. (See, for example, Begley and Fischer 1998.)
However, the mean |DEL| for our sample is 20.47 days. Panel B of Table 4 partitions the sample into pre-regulation (1995-1996), regulation (1997), and post-regulation (1998-1999) subsamples, indicating that the |DEL| for the three periods is 17.51, 19.48 and 21.96 days, respectively. Therefore, the greater variation in reporting dates in China is not due to the change in the reporting regulation in 1997. Overall, the results suggest that the reporting pattern of Chinese firms is much less predictable compared with that of mature markets. Consequently, we use LAG, instead of DEL, as the primary variable to measure the timeliness of earnings announcements in China. In the remainder of the paper, we focus our empirical analysis on LAG and provide information about DEL for purposes of comparison where appropriate.
DETERMINANTS OF TIMELINESS OF EARNINGS ANNOUNCEMENTS
Descriptive Statistics on the Determinants of Timeliness
Table 5 presents the timing pattern of annual earnings announcements by audit opinions and earnings surprises. Since there are only 23 disclaimers or adverse opinions, we merge them into qualified opinions to increase the power of statistical tests. Panel A presents the mean reporting lag (LAG) by sign of UE and audit opinion. For positive earnings surprises, the LAG increases monotonically with the severity of audit opinions when the earnings surprise is held constant. The mean LAG is 82.6, 92.3 and 103.1 days, respectively, for the unqualified opinion, unqualified opinion with explanatory notes, and qualified opinions. Both the F- and [chi square]-statistics to test the differences in LAG among the three audit opinion groups are significant at the 1 percent level. The pattern of mean LAG for negative earnings surprises is similar. The results indicate that audit modification delays earnings announcements, holding earnings surprises constant. Our results also suggest that negative earnings surprises are announced later than positive surprises within each audit opinion group. The F- and Z-statistics for the difference in LAG between positive and negative earnings surprise groups are significant at the 1 percent level for all three opinion groups. Overall, results are consistent with H1 and H2. (20)
Previous studies examining the effects of audit opinions and earnings surprises simultaneously only include the main effects of these variables in their analysis (Bamber et al. 1993). This implicitly assumes that the effect of qualified opinions is the same for firms with both positive and negative earnings surprises, and that the good news early, bad news late phenomenon is unconditional on audit opinions. A glance at Panel A of Table 5 suggests otherwise. The mean reporting lag for unqualified negative earnings surprises is 95.7 days, but 103.1 days for positive earnings surprises with qualified, disclaimer, or adverse opinions. Indeed, as reported in Panel B of Table 5, the 911 unqualified negative earnings surprises are announced significantly earlier than the 52 positive earnings surprises with qualified, disclaimer, or adverse opinions at the 1 percent level, with a mean of 95.7 and 103.1 days, respectively. When we merge unqualified opinions with explanatory notes into the modified opinion group, both the parametric and nonparametric tests indicate that the 911 unqualified negative earnings surprises are, on average, not announced later than the 164 positive earnings surprises with modified opinions. (21) These results suggest a possible interaction effect between audit opinions and earnings surprises on the timing of annual earnings announcements. We will address this issue in our regression analysis.
Regression Analysis
Even though results reported in Table 5 are consistent with H1 and H2, they fail to control for other variables that affect the earnings announcement lag in previous studies (e.g., Bamber et al. 1993; Begley and Fischer 1998). Correlation analysis in Panel D of Table 3 indicates that the reporting lag is significantly positively correlated with control variables such as size, the presence of losses, financial distress, and auditor switches. We thus control these variables and analyze the effects of audit opinions and earnings surprises on the earnings announcement lag with the following regression (firm-year subscripts omitted):
(2) LAG = [[alpha].sub.1] + [[alpha].sub.2][D.sub.97] + [[beta].sub.1][D.sub+] + [[beta].sub.2][D.sub.Q] + [[beta].sub.3][D.sub.+.sup.*][D.sub.Q] + [[beta].sub.4]UE + [[beta].sub.5]U[E.sup.*][D.sub.+] + [[gamma].sub.1]SIZE + [[gamma].sub.2][D.sub.97.sup.*] SIZE + [[gamma].sub.3][D.sub.Loss] + [[gamma].sub.4[D/TA] + [[gamma].sub.5][D.sub.SWITCH] + [epsilon]
where:
LAG = earnings announcement lag, measured as the number of days between fiscal year-end and the earnings announcement date;
[D.sub.97] = a dummy variable that equals 1 for observations in 1997 and after, and 0 otherwise;
[D.sub.+] = a dummy variable that equals 1 for positive earnings surprises (UE > 0), and 0 otherwise;
[D.sub.Q] = a dummy variable that equals 1 if the audit opinion is modified (unqualified with an explanatory note, qualified, disclaimer, or adverse), and 0 otherwise;
UE = scaled unexpected earnings as defined in Equation (1);
SIZE = the logarithm of total assets at the beginning of the year;
[D.sub.LOSS] = a dummy variable that equals 1 if the firm experiences a loss in the year, and 0 otherwise;
D/TA = debt-to-total-asset ratio at the beginning of the year, to proxy for financial distress;
[D.sub.SWITCH] = a dummy variable that equals 1 if the auditor for the firm is different between year t and year t-1, and 0 otherwise; and
[epsilon] = an error term.
To simplify the exposition, we combine unqualified opinion with explanatory notes with qualified, disclaimer, and adverse opinions, denoted by [D.sub.Q] = 1. We include [D.sub.+.sup.*][D.sub.Q] to examine the possible interaction effect between audit opinions and earnings surprises on the timing of annual earnings announcements. Hypothesis 1 predicts [[beta].sub.1] ([[beta].sub.1] + [[beta].sub.3]) to be negative for unqualified (modified) opinions. Hypothesis 2 predicts [[beta].sub.2] ([[beta].sub.2] + [[beta].sub.3]) to be positive for negative (positive) earnings surprises. Hypothesis 3 predicts [[beta].sub.4] ([[beta].sub.4] + [[beta].sub.5] to be negative for earnings decreases (increases). We include [D.sub.97] * SIZE in the regression because the two stock exchanges have been instructed by the CSRC to require smaller firms to announce their annual earnings earlier since 1997, and [[gamma].sub.2] would be positive if the regulation is effective. Consistent with prior studies, [[gamma].sub.3], [[gamma].sub.4], and [[gamma].sub.5] are all expected to be positive.
Panel A of Table 6 presents the regression results with White's t-statistics. The adjusted [R.sup.2] is 25.95 percent. (22) As expected, [[beta].sub.1] is -9.9 days and significantly negative. [[beta].sub.2] is 8.25 days and significantly positive. So are [[beta].sub.1 + [[beta].sub.3] and [[beta].sub.3] + [[beta].sub.3] (discussed more in Panel B of Table 6). Thus, the results are consistent with H1 and H2. For firms experiencing earnings decreases, [[beta].sub.4] is significantly negative at the conventional level, consistent with H3. The bigger the earnings decrease, the later the earnings announcement. However, the coefficient for earnings increase, [[beta].sub.4] + [[beta].sub.5], is 2.84 and statistically insignificant (not reported). That is, bigger earnings increases are not reported earlier than smaller earnings increases. The results are consistent with Begley and Fischer's (1998) finding for American firms. (23)
Panel A also presents coefficients and the corresponding significance levels of control variables. The coefficient for [D.sub.97] ([[alpha].sub.2]) is -69.42 days and significantly negative at the 1 percent level, indicating that Chinese regulators have succeeded in prompting listed firms to accelerate their annual earnings announcements through its 1997 decree. While the coefficient for SIZE ([[gamma].sub.1]) is negative but not statistically significant, the coefficient for [D.sub.97] * SIZE ([[gamma].sub.2]) is significantly positive. The impact of size on reporting lag in and after 1997, [[gamma].sub.1] + [[gamma].sub.2], is 3.75 and significant at the 1 percent level (not reported). The results indicate that larger firms announced earnings later than smaller firms in 1997 and after. This is counter to the findings in the U.S. that larger firms tend to have shorter reporting lags than smaller firms, and is most likely due to the CSRC regulation requiring smaller firms to release their annual reports early. Finally, the coefficients for [D.sub.Loss] and DITA are all significantly positive at conventional levels. The coefficient for [D.sub.SWITCH] is also positive but insignificant.
Panel B of Table 6 presents a 2 x 2 table containing the reporting lags for each of the audit opinions by earnings surprise types, after controlling for all of the other variables included in the regression. The average reporting lag for the group of unqualified positive earnings surprises ([[alpha].sub.1] + [[beta].sub.1]) is the shortest (92.85 days), as expected, and that for the group of negative earnings surprises with modified audit opinion ([alpha].sub.1] + [[beta].sub.2]) is longest (111 days). The modified positive earnings surprises are, on average, announced 106.14 days ([[alpha].sub.1] + [[beta].sub.1] + [[beta].sub.2] + [[beta].sub.3]) after the fiscal year-end, in comparison to the 102.75 days ([alpha].sub.1]) for unqualified negative earnings surprises. Pair-wise comparisons of these reporting lags indicate that, when the audit opinion is unqualified (modified), positive earnings surprises are announced significantly earlier than negative earnings surprises by 9.9 (4.86) days, consistent with H1. Similarly, the audit qualification process significantly delays the announcements of both positive and negative earnings surprises by 13.29 and 8.25 days, respectively, consistent with H2.
Panel B of Table 6 also demonstrates that the difference in reporting lags between positive and negative earnings surprises is reduced significantly by 5.04 (9.90-4.86) days when the audit opinion changes from unqualified to modified. That is, the impact of earnings surprises on reporting lag becomes less profound when the situation of audit qualification arises. Furthermore, when we compare the differences in reporting lags between unqualified and modified audit opinions for positive (13.29 days) and negative (8.25 days) earnings surprises, we notice that the delay caused by audit qualification is greater for positive earnings surprises than for negative surprises. In addition, positive earnings surprises with modified audit opinions are announced, on average, 3.39 (106.14 -102.75) days later than unqualified negative earnings surprises. The difference between these two groups ([[beta].sub.1] + [[beta].sub.2] + [[beta].sub.3]) is significantly different from zero at the 10 percent confidence level (not reported), corroborating the significant positive coefficient for the interaction term between audit opinions and earnings surprises ([beta].sub.3]) in Panel A of Table 6. We thus document for the first time a significant interaction effect between audit opinions and earnings surprises on reporting lags. (24) The result is consistent with Choi and Jeter's (1992) study that documents a significant decline of the market's responsiveness to earnings announcements after the issuance of qualified audit reports for a sample of "subject to" qualifications and consistency qualifications.
Sensitivity Analysis
First, we estimate Regression (2) without the interaction term between earnings surprises and audit opinions to facilitate the comparison of our study with the existing timing literature in the following model:
(3) LAG = [[alpha].sub.1] + [[alpha].sub.2][D.sub.97] + [[beta].sub.1][D.sub.+] + [[beta].sub.2][D.sub.Q] + [[beta].sub.4]UE + [[beta].sub.5]UE * [D.sub.+] + [[gamma].sub.1]SIZE + [[gamma].sub.2][d.sub.97] * SIZE + [[gamma]sub.3][D.sub.LOSS] + [[gamma].sub.4]D/TA + [[gamma].sub.5][D.SWITH] + [epsilon].
The results are reported in Panel C of Table 6, which are consistent with those reported in previous studies (for example, Bamber et al. 1993). While [[beta].sub.1] is significantly negative, [[beta].sub.2, [[gamma].sub.3], [[gamma].sub.4], and [[gamma].sub.5] are all significantly positive at the conventional levels. While the coefficients of [D.sub.97] and [[gamma].sub.1] + [[gamma.sub.2] remain significant with predicted signs, [[beta].sub.4] is no longer significantly different from zero at the conventional level. The results suggest the usefulness of controlling for the interaction effect of audit opinions and earnings surprises in examining the relationship between the magnitude of the earnings surprises (UE) and the timing pattern of earnings announcements.
Second, we estimate the following regression separately for the early (1995-1996) and latter (1997-1999) periods to examine whether our results are sensitive to regulatory changes:
(4) LAG = [[alpha].sub.1] + [[beta].sub.1][D.sub.+] + [[beta].sub.2][D.sub.Q] + [[beta].sub.3][D.sub.Q] + [[beta].sub.4]UE + [[beta].sub.5]UE*[D.sub.+] + [[gamma].sub.1]SIZE + [[gamma].sub.2][D.sub.LOSS] + [[beta].sub.3]D/TA+ [[gamma].sub.4][D.sub.SWITCH + [epsilon]
Model (2) forces the coefficients to be the same between the two different sample periods. As reported in Table 7, the explanatory power of the model is much higher in the latter period (adjusted [R.sup.2] = 20.51 percent) than in the early period (adjusted [R.sup.2] = 13.74 percent). While results for both periods are in general consistent with our hypotheses and those reported in Table 6 for the full sample, results for 1997-1999 are generally stronger. In particular, coefficients for the interaction term of [D.sub.+] * [D.sub.Q] [[beta].sub.3]), UE ([[beta].sub.4]), and SIZE ([gamma].sub.1] are significant only in the latter period.
Third, we separate [D.sub.Q], the modified opinions variable, in Regression (2) into two groups by the severity of audit opinions, as in Panel A of Table 5: (l) unqualified opinion with explanatory notes, and (2) qualified, disclaimer, or adverse opinion. The test results using White's ACOV estimates (not reported in tables) are qualitatively the same as those in Panel A of Table 6. There is also evidence that announcement lags increase as the audit opinions become more severe, especially during 1997-1999. This is consistent with the improved quality and independence of Chinese auditing firms as reported in DeFond et al. (1999).
Fourth, we conduct two subsample tests that provide reassurance that the observed patterns of reporting lags are driven by the differences in audit opinions and earnings surprises and not by uncontrolled cross-sectional differences in the sample. Results are presented in Table 8. First reported is the subsample comparison of reporting lags between firms that experienced a change in earnings surprises from positive to negative (837 firm-years), and vice versa (988 firm-years) between year t-1 to year t. The mean reporting lag for the former group is 97.1 days, and the median is 102 days. For the latter group, the mean and median are 88.5 and 89 days, respectively. Both the parametric and nonparametric tests indicate that the difference is significant at the 1 percent confidence level. The results between firms that experienced a change in audit opinions from unqualified to qualified and vice versa are similar.
Fifth, we replace LAG in model (4) with DEL and estimate the following model separately for the pre-regulation (1995-1996) and post-regulation (1998-1999) periods:
(5) DEL = [[alpha].sub.1] [[beta].sub.1] [D.sub.+] + [[beta].sub.2] + [D.sub.+] + [[beta].sub.3][D.sub.Q] [D.sub.Q] + [[beta].sub.4] UE*[D.sub.+] + [[gamma].sub.1] SIZE + [gamma].sub.2][D.sub.LOSS] + [[gamma].sub.3.sup.D/TA]+ [[gamma].sub.4 [D.sub.SWITCH] + [epsilon]
The latter period does not include the 1997 sample because of the structural change in DEL due to the 1997 CSRC regulation. The regression results (not reported) indicate that the explanatory power (adjusted [R.sup.2] of the model is 4.43 percent and 5.98 percent, respectively, for the pre- and post-regulation periods. For the latter period, the coefficients for earnings surprise dummy and magnitudes ([[beta].sub.1], ([[beta].sub.4] + ([[beta].sub.5], are statistically significant and similar to those in Table 6. The of SIZE ([[gamma].sub.1]) is significantly positive. However, for the pre-regulation period, [[beta].sub.1] is the only significant coefficient with the predicted sign. For both pre- and post-regulation periods, the coefficients for other control variables ([D.Loss], D/TA, [D.SWITCH]) have opposite signs to their prediction and are frequently significant at the 1 percent level. Thus, the results suggest that DEL does not appear as good as LAG to proxy for the timeliness of earnings announcements in the emerging market of China.
SUMMARY AND CONCLUSIONS
This study extends the line of timing research to China's emerging capital market by investigating the effects of audit opinions and earnings surprises on the timeliness of annual earnings announcements. Based on a sample of listed Chinese firms for 1995-1999, we observe that both audit opinions and earnings surprises have significant effects as reported in the literature. In addition, we find evidence consistent with Begley and Fischer (1998) that the bigger the earnings decrease, the later the earnings announcement. In summary, our findings are consistent with results reported in previous studies based on data from the U.S. and Australia.
Furthermore, we document a significant interaction effect between audit opinions and earnings surprises on reporting lags. Firms with positive earnings surprises but modified opinions have significantly longer reporting lags than firms with negative earnings surprises and unqualified opinions. The result is consistent with the notion that unexpected earnings increases are not necessarily good news unless certified as clean. It would be interesting to further investigate whether this empirical regularity is unique to China or also exists in mature markets.
Our results are important in that, even though announcement dates are generally less predictable in the emerging market of China than in the mature markets, we have identified a set of determinants of the timing of earnings announcements. Our analysis also indicates that CSRC regulations are effective in changing the behavior of listed firms in releasing their annual reports. In particular, we find that smaller firms tend to have shorter reporting lags than larger firms after 1997, which is counter to the results in the U.S. This is most likely due to the CSRC regulation requiring smaller firms to release their annual reports early.
APPENDIX
EXAMPLES OF MODIFIED AUDIT REPORTS IN CHINA
Qualified Opinions
Departures from GAAP
Company: Jiangsu Shanshan
Auditor: Shenzhen Zhonghua, top 5 auditor in 1993-1999
Stock exchange: Shenzhen
Fiscal Year: 1997
Jiangsu Shanshan incurred losses in 1995 and 1996. Had losses appeared again in 1997, the firm would be the first listed Chinese firm to experience three consecutive losses and its stock subsequently could have been suspended from trading. In 1997, the firm recorded sales of RMB 160.03 million to a subsidiary of its controlling shareholder and booked a net profit of RMB 20 million. This entry turned the company's bottom line from red to black. However, its auditor believes the procedures to close the sales to be incomplete. (The 1998 interim report revealed that the sales were at RMB 9.65 million with a cost of RMB 9.46 million. Jiangsu Shanshan became the first stock to be suspended from trading by CSRC.)
Scope limitation
Company: Huanghe Machinery
Auditor: Xian
Stock exchange: Shanghai
Fiscal Year: 1995 and 1996
In 1995 Huanghe sold its products across 24 regions in China. Due to the vast distance and its audit firm's limited resources, its audit firm did not properly audit the company's RMB 75 million of inventory and RMB 31.44 million of trade receivables. The 1996 scope limitation was related to the above two items plus the fact that the company failed to provide relevant information to allow the auditors to audit an RMB 21 million long-term investment.
Inconsistent Application of GAAP
Company: Guangdong Shaoneng
Auditor: Guangzhou, top 10 auditor in 1996
Stock exchange: Shenzhen
Fiscal year: 1996
Guangdong Shaoneng was listed in Shenzhen in August 1996. On June 1, 1996, the Shaoneng changed the useful lives of several of its categories of depreciable assets. Office buildings were changed from 20 to 40 years, factory buildings were changed from 20 to 30 years, machinery was changed from 10 to 20 years, and transportation equipment was changed from 5 to 10 years. These changes in estimates increased the 1996 net income of Shaoneng by RMB 10.7 million.
Disclaimer Opinions
Company: SAIC Multiple Trading
Auditor: Shanghai Lixin, top 10 auditor from 1993-1998
Stock exchange: Shanghai
Fiscal Year: 1998, the first disclaimer opinion issued in China
In 1998, SAIC Multiple Trading incurred another loss of RMB 109.67 million after incurring losses in the past two years, with a negative book value of equity of RMB 181.88 million. The total liability amounts to RMB 1,218.84 million, of which 56 loans of RMB 454.14 million are past due, among which 32 loans of RMB 299.53 million are one year past due. The firm has also lost 53 lawsuits involving RMB 202.67 million. Other receivables amount to RMB 551.92 million, of which RMB 378.19 million are from subsidiaries of the firm's holding company.
Adverse Opinions
Company: Chongqing Yu-Gang Tioxide
Auditor: Chongqing
Stock exchange: Shenzhen
Fiscal Year: 1997, the first adverse opinion issued in China
In 1997, Chongqing Yu-Gang Tioxide capitalized RMB 80.64 million of interest expenses and did not accrue U.S.$0.898 million of interest expenses. The debt burden was heavy and production lines were halted.
Unqualified Opinions with Explanatory Notes
The Auditor Agrees with a Departure from a Promulgated Principle
Company: Mingdong Electric Group
Auditor: Fujian Huaxing
Stock exchange: Shenzhen
Fiscal Year: 1997
Mingdong Electric set up a joint venture with Honda in 1995. The registered capital of the joint venture is U.S.$ 6 million, equivalent to RMB 50 million. Mingdong holds 40 percent of the joint venture. By the end of 1997, the net assets of the joint venture were RMB 31.84 million. In 1997 Mingdong used the cost method to account for the investment, which is unchanged since the inception of the joint venture.
Substantial Doubt about Going Concern
Company: Beijing Light Bus Manufacturing
Auditor: Beijing Jingdu
Stock exchange: Shanghai
Fiscal Year: 1999
Beijing Light Bus Manufacturing had incurred losses for the past three consecutive years. The company's financial leverage is extremely high. The auditor reminded shareholders to pay attention to the ability of the firm to continue as a going concern.
Emphasis of a Matter
The existence of significant related-party transactions
Company: Shanghai Automation Instrumentation
Auditor: Dahua, top 5 auditor in 1993-1999
Stock exchange: Shanghai
Fiscal Year: 1998
Shanghai Mcline Electronics Ltd, which is a subsidiary of Shanghai Automation Instrumentation, had an outstanding accounts receivable of RMB 160.45 million from Mcline (America) Inc. Mcline (America) issued a written consent to use its assets as the collateral. Shanghai Mcline Electronics Ltd. is a joint venture between Shanghai Automation Instrumentation and Mcline (America) with a registered capital of U.S.$77.1 million, equivalent to RMB 640 million.
Important events occurring subsequent to the post balance sheet date
Company: Beijing Light Bus Manufacturing
Auditor: Beijing Jingdu
Stock exchange: Shanghai
Fiscal Year: 1999
Beijing Light Bus Manufacturing had incurred losses for the past three consecutive years. It incurred a loss again in the first quarter of 2000. The auditor report was issued on April 17, 2000.
Material uncertainties disclosed in the footnotes
Company: Shenzhen Properties Development
Auditor: Shenzhen
Stock exchange: Shenzhen
Fiscal Year: 1997
In 1997, contingent losses disclosed in the footnotes 8.3, 8.4, 8.5, 8.8, and 8.9 amount to RMB 146.31 million.
TABLE 1
Development of the Chinese Stock Market
Year 1991 1992 1993
Panel A: Number of Listed Firms
Shanghai Stock Exchange 8 29 105
Shenzhen Stock Exchange 5 23 76
Total 13 52 181
Panel B: Market Capitalization (a)
Shanghai Stock Exchange 2.9 71.5 218.8
Shenzhen Stock Exchange 8.0 49.0 133.5
Total 10.9 120.5 352.3
Panel C: The Composite Index of Chinese Stock Market
Shanghai Stock Exchange 293.7 780.4 833.8
Shenzhen Stock Exchange 110.4 241.2 238.3
Year 1994 1995 1996
Panel A: Number of Listed Firms
Shanghai Stock Exchange 171 188 293
Shenzhen Stock Exchange 120 135 237
Total 291 323 530
Panel B: Market Capitalization (a)
Shanghai Stock Exchange 259.7 252.6 547.8
Shenzhen Stock Exchange 109.1 94.9 436.5
Total 368.8 347.5 984.3
Panel C: The Composite Index of Chinese Stock Market
Shanghai Stock Exchange 647.9 555.3 912.0
Shenzhen Stock Exchange 140.7 113.3 312.9
Year 1997 1998 1999
Panel A: Number of Listed Firms
Shanghai Stock Exchange 383 438 484
Shenzhen Stock Exchange 362 413 463
Total 745 851 947
Panel B: Market Capitalization (a)
Shanghai Stock Exchange 921.8 1,078.8 1,458.0
Shenzhen Stock Exchange 831.1 898.0 1,189.1
Total 1,752.9 1,976.8 2,647.1
Panel C: The Composite Index of Chinese Stock Market
Shanghai Stock Exchange 1,194.1 1,146.7 1,366.6
Shenzhen Stock Exchange 381.3 343.9 392.9
Source: Securities Markets Herald, 1995-1999.
(a) Market capitalization is in billion Chinese yuan. One U.S.
dollar is about RMB 8.3 Chinese yuan.
TABLE 2
Development of the Auditing Market in China
Fiscal Year 1992 1993 1994 1995
Panel A: Profile of the CPA Firms That Audit Listed Chinese Firms
Number of Audit Reports Issued 53 248 296 336
Number of CPA Firms 16 58 61 69
Number of Audit Reports per CPA Firm 3.3 4.3 4.9 4.9
Panel B: Profile of the 10 Largest CPA Firms in China
Dahua 4 19 19 21
Shanghai 11 31 31 33
Shenzhen Zhonghua 7 16 20 19
Shenzhen Tongren 0 0 0 2
Zhejiang Tianjian 1 6 9 10
Sichuan Deyang 0 1 1 1
Shanghai Lixin 3 19 20 20
Shanghai Zhonghua 7 13 13 13
Shenzhen Shoukou Zhonghua 5 12 14 15
Sichuan 0 6 9 12
Number of Firms Audited by the Top 5 CPA
Firms (a) 34 98 104 108
Percentage of Firms Audited by the Top 5
CPA Firms (a) 64% 40% 35% 32%
Average Audit Reports by the Top 5 CPA
Firms (a) 6.8 19.6 20.8 21.6
Panel C: Issuance of Modified Opinions
(Including Unqualified Opinion with an Explanatory Note)
Number of Firms Receiving Modified
Opinions 0 4 6 34
Percentage of Firms Receiving Modified
Opinions 0 2% 2% 10%
Number of Modified Opinions by the Top 5
CPA [Firms.sup.+] 0 1 3 17
Panel D: Auditor Switches
Number of Auditor Switches 0 2 10 9
Fiscal Year 1996 1997 1998 1999
Panel A: Profile of the CPA Firms That Audit Listed Chinese Firms
Number of Audit Reports Issued 571 696 865 952
Number of CPA Firms 83 87 90 90
Number of Audit Reports per CPA Firm 6.9 8.0 9.6 10.6
Panel B: Profile of the 10 Largest CPA Firms in China
Dahua 31 34 40 42
Shanghai 37 37 36 34
Shenzhen Zhonghua 21 24 27 33
Shenzhen Tongren 12 19 21 30
Zhejiang Tianjian 16 20 24 30
Sichuan Deyang 2 5 25 28
Shanghai Lixin 22 22 25 23
Shanghai Zhonghua 16 17 17 22
Shenzhen Shoukou Zhonghua 12 15 15 17
Sichuan 18 22 -- --
Number of Firms Audited by the Top 5 CPA
Firms (a) 129 139 153 169
Percentage of Firms Audited by the Top 5
CPA Firms (a) 23% 20% 18% 18%
Average Audit Reports by the Top 5 CPA
Firms (a) 25.8 27.8 30.6 36.3
Panel C: Issuance of Modified Opinions
(Including Unqualified Opinion with an Explanatory Note)
Number of Firms Receiving Modified
Opinions 51 98 150 193
Percentage of Firms Receiving Modified
Opinions 9% 14% 17% 20%
Number of Modified Opinions by the Top 5
CPA Firms + 21 25 20 37
Panel D: Auditor Switches
Number of Auditor Switches 18 25 51 49
(a) The identity of the five largest audit firms has changed over
the years as the number of listed firms has significantly increased.
Only three auditors have consistently appeareamong the top 5 during
1992-1999: Dahua, Shanghai, and Shenzhen Zhonghua. The fourth and
fifth largest audit firms have changed over time.
TABLE 3
Sample Description (a)
Panel A: Descriptive Statistics of 2,921 Annual Earnings Announcements
Variable Mean Std. Dev. 1%
Reporting Lag (LAG) 89.86 24.81 26
Announcement Delay (DEL) -1.53 * 28.03 -75
Unexpected Earnings (UE) 0.01 * 0.07 -0.19
Total Assets (in million yuan) 1,066.53 150.12 147.62
Debt-to-Asset Ratio (DITA) 0.44 0.19 0.07
Variable Q1 Media
Reporting Lag (LAG) 75 91
Announcement Delay (DEL) -17 -1 *
Unexpected Earnings (UE) -0.012 0.01 *
Total Assets (in million yuan) 418.31 690.27
Debt-to-Asset Ratio (DITA) 0.32 0.44
Variable Q3 99%
Reporting Lag (LAG) 110 120
Announcement Delay (DEL) 13 76
Unexpected Earnings (UE) 0.03 0.24
Total Assets (in million yuan) 1,173.57 7,190.90
Debt-to-Asset Ratio (DITA) 0.56 0.91
Panel B: Distribution of the Means of Variables by Audit Opinions
Audit Opinion Sample % LAG
Unqualified 2,491 85.3 87.4
Unqualified with an
Explanatory Note 232 7.9 98.3
Qualified 175 6.0 108.9
Disclaimer or Adverse 23 0.8 124.0
Total Sample 2,921 100%
F-statistics 70.6 (+)
Audit Opinion DEL UE D/TA
Unqualified -2.30 * 0.014 * 0.42
Unqualified with an
Explanatory Note 0.59 0.005 0.50
Qualified 4.04 * -0.03 * 0.58
Disclaimer or Adverse 3.91 -0.08 0.94
Total Sample
F-statistics 3.2 (++) 36.4 (+) 106.8 (+)
Audit Opinion SIZE [D.sub.LOSS] [D.sub.SWITCH]
Unqualified 13.51 3.9% 3.5%
Unqualified with an
Explanatory Note 13.39 17.7% 6.0%
Qualified 13.52 30.9% 9.1%
Disclaimer or Adverse 13.53 95.7% 26.1%
Total Sample
F-statistics 1.57 191.5 (+) 14.7 (+)
Panel C: Distribution of the Means of Variables by Sample Year
Year Sample LAG DEL
1995 246 108.5 13.54 *
1996 479 100.1 0.47
1997 654 83.9 -13.92 *
1998 747 87.7 3.88 *
1999 795 84.9 -3.63 *
Year UE [D.sub.NOTE] [D.sub.QLY]
1995 -0.012 * 0.0% 14.2%
1996 0.016 * 4.4% 4.6%
1997 0.020 * 7.2% 5.5%
1998 0.007 * 9.5% 4.9%
1999 0.006 * 11.7% 5.7%
Year [D.sub.ADV] [D.sub.LOSS] [D.sub.SWITCH]
1995 0.0% 6.9% 3.7%
1996 0.4% 6.3% 3.1%
1997 0.2% 7.0% 2.9%
1998 1.5% 8.8% 5.9%
1999 1.1% 7.2% 4.4%
Panel D: Pearson Correlation Coefficients and Related p-Values
LAG DEL UE
DEL 0.5897
<.0001
UE -0.2200 -0.1800
<.0001 <.0001
[D.sub.+] -0.2948 -0.2367 0.5484
<.0001 <.0001 <.0001
[D.sub.NOTE] 0.1003 0.0242 -0.0222
<.0001 0.2504 0.2315
[D.sub.QLY] 0.1844 0.0542 -0.1426
<.0001 0.0101 <.0001
[D.sub.ADV] 0.1225 0.0193 -0.1169
<.0001 0.3605 <.0001
SIZE 0.0310 0.0536 -0.0243
0.0948 0.0103 0.7781
[D.sub.LOSS] 0.2914 0.0686 -0.4125
<.0001 0.0011 <.0001
D/TA 0.2161 0.0032 -0.1566
<.0001 0.8809 <.0001
[D.sub.SWITCH] 0.0535 -0.0156 -0.0071
0.004 0.4594 0.7019
[D.sub.+] [D.sub.NOTE] [D.sub.QLY]
DEL
UE
[D.sub.+]
[D.sub.NOTE] -0.0685
<.0001
[D.sub.QLY] -0.1632 -0.0742
<.0001 <.0001
[D.sub.ADV] -0.0848 -0.0262 -0.0225
<.0001 0.1574 0.2243
SIZE -0.0052 -0.0401 0.0043
0.9508 0.0302 0.8182
[D.sub.LOSS] -0.3013 0.1154 0.2263
<.0001 <.0001 <.0001
D/TA -0.1168 0.0901 0.1772
<.0001 <.0001 <.0001
[D.sub.SWITCH] -0.0204 0.0273 0.0627
0.2709 0.1405 0.0011
[D.sub.ADV] SIZE [D.sub.LOSS] D/TA
DEL
UE
[D.sub.+]
[D.sub.NOTE]
[D.sub.QLY]
[D.sub.ADV]
SIZE 0.0029
0.7874
[D.sub.LOSS] 0.3005 -0.0572
<.0001 0.0020
D/TA 0.2315 0.2110 0.3135
<.0001 <.0001 <.0001
[D.sub.SWITCH] 0.0975 -0.0075 0.0783 0.0787
<.0001 0.6844 <.0001 <.0001
* indicates statistical significance at the 1 percent level based
on the two-tailed t-test for the means and the Wilcoxon test for
the medians.
(+) and (++) indicate statistical significance at the 1 percent
and 5 percent confidence levels based on the F-test.
(a) Results are based on 2,921 firm-years, except for measures
related to DEL Due to missing LAG at t-1, the total number of
DEL is 2,256 firm-years. DEL is missing either because the year
is the IPO year or because the announcement date in the previous
year is missing. For non-IPO firms, we calculate the 1995 DEL
based on the announcement dates available on the 1994 annual
reports. We use financial information in the TEJ China database
to calculate UE, SIZE, and D/TA. When financial data for t-1 are
not available for IPO firms, we use information in the pre-IPO
year provided in the prospectus. Accordingly, as to UE for IPO
firms, we use net income in the pre-IPO year provided in the
prospectus. When auditor information is not available in year
t-1 for [D.sub.SWITCH], we compare the auditor in year t with
the auditor in year t-2 and assume there is no switch if they
are the same auditor, and vice versa. For IPO firms, we compare
the auditor in the IPO year with the auditor in the IPO prospectus
who audited the financial statements in the pre-IPO year.
LAG is the earnings announcement lag, measured as the number of days
between the fiscal year-end and the earnings announcement date; DEL is
the unexpected earnings announcement delay, measured as the difference
between LAGs in year t and year t-1; UE is defined in Equation (1) as
the unexpected earnings scaled by lagged total assets; [D.sub.+] is a
dummy variable that equals 1 for positive earnings surprises (UE > 0),
and 0 otherwise; [D.sub.NOTE] is a dummy variable that equals 1 if the
audit opinion is unqualified with an explanatory note, and 0 otherwise;
[D.sub.QLY] is a dummy variable that equals 1 if the audit opinion is
qualified, and 0 otherwise; [D.sub.ADV] is a dummy variable that equals
1 if the audit opinion is a disclaimer or adverse opinion, and 0
otherwise; SIZE is the logarithm of total assets (in million yuan) at
the beginning of the year; [D.sub.LOSS] is a dummy variable that
equals 1 if the firm experiences a loss in the year, and 0 otherwise;
D/TA is the debt-to total-asset ratio at the beginning of the year,
to proxy for financial distress; [D.sub.SWITCH] is a dummy variable
that equals 1 if the firm has a different auditor in year t from year
t-1, and 0 otherwise.
TABLE 4
Predicative Ability of the Previous Year's Reporting Lag for Timing of
Annual Report Releases
Panel A: Mean Statistics of DEL in Days and within-Firm Time-Series
Variability of the Reporting Lag (LAG) in Days
Chambers and Penman (1984) Our Sample for
Summary Measure Table 1 (Annual) 1995-1999
DEL -1.01 -1.53 *
|DEL| 7.07 20.47
Mean Std. Dev. (LAG) 7.2 17.73
Std. Dev. (Std. Dev.) 5.6 11.03
Mean Range (LAG) 22.4 36.00
Std. Dev. (Range) 16.6 22.46
Panel B: Mean Statistics of DEL in Days by Subsample Periods
Pre-Regulation Year of Regulation Post-Regulation
Subsample Period Subsample Period Subsample Period
in 1995-1996 in 1997 in 1998-1999
DEL 6.24 * -13.92 * -0.13
|DEL| 17.51 19.48 21.96
* indicates statistical significance at the 1 percent level based on
the two-tailed t-test.
For the sample of 2,921 firm-years, the reporting lag (LAG) is defined
as the number of days from the fiscal year-end to the annual earnings
announcement date. Due to missing LAG at t-1, the total number of DEL
is 2,256 firm-years. DEL is missing either because the year is the IPO
year or because the announcement date in the previous year is missing.
For non-IPO firms, we calculate the 1995 DEL based on the announcement
dates available on the 1994 annual reports. DEL, reporting lag's (LAG)
deviation from the previous year, is the announcement delay defined as
DEL = LA[G.sub.t] - LA[G.sub.t-1]. In year 1997, the Chinese Securities
Regulatory Commission (CSRC) promulgated a regulation requiring listed
firms to release their annual reports on a more timely basis.
TABLE 5
Timeliness of Annual Earnings Announcements by Earnings Surprises and
Audit Opinions
Panel A: Mean Reporting Lag (LAG) by Earnings Surprises and Audit
Opinions
No. of Observations Mean LAG in Days
Auditor Opinion Types UE>0 UE<0 UE>0 UE<0
Unqualified 1,580 911 82.6 95.7
Unqualified with an
Explanatory Note 112 120 92.3 104.0
Qualified, Disclaimer,
or Adverse 52 146 103.1 113.3
F-statistics to test H2 24.1 * 50.0 *
[chi square]-statistics
to test H2 59.0 * 121.2 *
Mean LAG in Days
F-stat. Z-stat.
Auditor Opinion Types To test H1 To test H1
Unqualified 178.7 * 13.6 (+)
Unqualified with an
Explanatory Note 17.0 * 4.0 (+)
Qualified, Disclaimer,
or Adverse 10.3 * 2.5 (+)
F-statistics to test H2
[chi square]-statistics
to test H2
Panel B: Subsample Comparison of Reporting Lags (LAG)
Mean Median
Subsamples No. of Observations LAG LAG
Negative UE with
Unqualified Opinion 911 95.7 100
Positive UE with Qualified,
Disclaimer, or Adverse Opinion 52 103.1 114.5
Negative UE with Unqualified
Opinion 911 95.7 100
Positive UE with Modified
Opinion 164 95.7 102
Subsamples Prof>F Prob>|Z|
Negative UE with
Unqualified Opinion 0.015 0.001
Positive UE with Qualified,
Disclaimer, or Adverse Opinion
Negative UE with Unqualified
Opinion 0.991 0.393
Positive UE with Modified
Opinion
* indicates statistical significance at the 1 percent confidence level
based on the F-test or Kruskal-Wallis test.
(+) indicates statistical significance at the 1 percent level based on
the Wilcoxon two-sample test. The modified opinion group in Panel B
consists of unqualified opinions with explanatory notes as well as
qualified, disclaimer, or adverse opinions.
LAG is the earnings announcement lag, measured as the number of days
between the fiscal year-end and the earnings announcement date. UE is
the unexpected earnings scaled by lagged total assets. For IPO firms,
we use net income in the pre-IPO year provided in the prospectus.
TABLE 6
Regression Analysis of the Timeliness of Annual Earnings Announcements
(2) LAG = [[alpha].sub.1] + [[alpha].sub.2][D.sub.97] + [[beta].sub.1]
[D.sub.+] + [[beta].sub.2][D.sub.Q] + [[beta].sub.3][D.sub.+.sup.*]
[D.sub.Q] + [[beta].sub.4]UE + [[beta].sub.5]U[E.sup.*][D.sub.+] +
[[gamma].sub.1]SIZE + [[gamma].sub.2][D.sub.97.sup.*]SIZE +
[[gamma].sub.3][D.sub.LOSS] + [[gamma].sub.4]D/TA + [[gamma].sub.5]
[D.sub.SWITCH] + [epsilon]
Variable Intercept [D.sub.97] [D.sub.+]
Coefficient [[alpha].sub.1] [[alpha].sub.2] [[beta].sub.1]
Predicted Sign + - -
Panel A: Model (2), Adjusted [R.sup.2] = 25.95%
Coefficient 102.75 -69.42 -9.90
White's t-statistics (7.39) * (-4.53) * (-10.06) *
Variable [D.sub.Q] [D.sub.+.sup.*] UE
[D.sub.Q]
Coefficient [[beta].sub.2] [[beta].sub.3] [[beta].sub.4]
Predicted Sign + ? -
Panel A: Model (2), Adjusted [R.sup.2] = 25.95%
Coefficient 8.25 5.04 -21.91
White's t-statistics (6.68) * (2.21) ** (-1.82) ***
Variable U[E.sup.*] SIZE [D.sub.97.
[D.sub.+] sup.*]SIZE
Coefficient [[beta].sub.5] [[gamma].sub.1] [[gamma].sub.2]
Predicted Sign ? - +
Panel A: Model (2), Adjusted [R.sup.2] = 25.95%
Coefficient 24.75 -0.08 3.83
White's t-statistics (1.57) (-0.10) (3.33) *
Variable [D.sub.LOSS] D/TA [D.sub.SWITCH]
Coefficient [[gamma].sub.3] [[gamma].sub.4] [[gamma].sub.5]
Predicted Sign + + +
Panel A: Model (2), Adjusted [R.sup.2] = 25.95%
Coefficient 15.77 9.21 3.12
White's t-statistics (10.31) * (3.81) * -1.59
Panel B: A Summary of Coefficients for Each of the Categories after
Controlling for All of the Other Variables in Model (2)
Earnings Surprise
[D.sub.+] = 1 [D.sub.+] = 0
Unqualified Opinion, [D.sub.Q] = 0 92.85 102.75
Modified Opinions, [D.sub.Q] = 1 106.14 111.00
Difference 13.29 8.25
[chi square]-statistics to test H2 46.4 (+) 44.6 (+)
[chi square]-statistics
Difference to test H1
Unqualified Opinion, [D.sub.Q] = 0 9.90 101.3 (+)
Modified Opinions, [D.sub.Q] = 1 4.86 4.6 (++)
Difference 5.04 4.9 (++)
[chi square]-statistics to test H2 4.9 (++)
Panel C: Sensitivity Analysis: Model (3) without the Interaction Term
of [D.sub.+.sup.*][D.sub.Q], Adjusted [R.sup.2] = 25.86%
Variable Intercept [D.sub.97] [D.sub.+]
Coefficient 102.07 -68.37 -9.38
White's t-statistics (7.32) * (-4.47) * (-9.83) *
Variable [D.sub.Q] [D.sub.+.sup.*] UE
[D.sub.Q]
Coefficient 10.49 -- -18.40
White's t-statistics (9.35) * -- (-1.52)
Variable U[E.sup.*] SIZE [D.sub.97.
[D.sub.+] sup.*]SIZE
Coefficient 22.80 -0.05 3.75
White's t-statistics (1.45) (-0.01) (3.27) *
Variable [D.sub.LOSS] D/TA [D.sub.SWITCH]
Coefficient 15.41 9.10 3.30
White's t-statistics (9.87) * (3.75) * (1.69) ***
* and *** indicate statistical significance at the 1 percent and 10
percent levels based on the two-tailed White's t-test.
(+) and (++) indicate statistical significance at the 1 percent and 5
percent confidence levels based on the F-test.
When financial data for t-1 are not available for IPO firms, we use
information in the pre-IPO year provided in the prospectus. Thus, as to
UE for IPO firms, we use net income in the pre-IPO year provided in the
prospectus. When auditor information is not available in year t-1 for
[D.sub.SWITCH], we compare the auditor in year t with the auditor in
year t-2 and assume there is no switch if they are the same auditor,
and vice versa. For IPO firms, we compare the auditor in the IPO year
with the auditor in the IPO prospectus who audited the financial
statements in the pre-IPO year.
Results are based on 2,921 earnings announcements of A-share firms
during 1995-1999. LAG is the earnings announcement lag, measured as the
number of days between the fiscal year-end and the earnings
announcement date; [D.sub.97] is a dummy variable that equals 1 for
observations in 1997 and after, and 0 otherwise; [D.sub.+] is a dummy
variable that equals 1 for positive earnings surprises (UE > 0), and 0
otherwise; [D.sub.Q] is a dummy variable that equals 1 if the audit
opinion is modified, and 0 otherwise; UE is the unexpected earnings
scaled by lagged total assets; SIZE is the logarithm of total assets
(in million yuan) at the beginning of the year; [D.sub.LOSS] is a dummy
variable that equals 1 if the firm experiences a loss in the year, and
0 otherwise; D/TA is the debt-to-total-asset ratio at the beginning of
the year, to proxy for financial distress; [D.sub.SWITCH] is a dummy
variable that equals 1 if the firm has a different auditor in year t
from t-1, and 0 otherwise. Modified audit opinions ([D.sub.Q] = 1)
include unqualified opinion with an explanatory note, qualified
opinion, disclaimer, or adverse opinion.
TABLE 7
Regression Analysis of the Timeliness of Annual Earnings Announcements:
Sensitivity Analysis
(4) LAG = [a.sub.1] + [[beta].sub.1][D.sub.+] + [[beta].sub.2][D.sub.Q]
+ [[beta].sub.3][D.sub.+.sup.*][D.sub.Q] + [[beta].sub.4]UE +
[[beta].sub.5]U[E.sup.*][D.sub.+] + [[gamma].sub.1]SIZE +
[[gamma].sub.2][D.sub.LOSS] + [[gamma].sub.3]D/TA + [[gamma].sub.4]
[D.sub.SWITCH] + [epsilon]
Variable Intercept [D.sub.+] [D.sub.Q]
Coefficient [[alpha].sub.1] [[beta].sub.1] [[beta].sub.2]
Predicted Sign + - +
Panel A: Results Based on 725 Announcements in 1995-1996, Adjusted
[R.sup.2] = 13.74%
Coefficient 97.84 -7.52 7.35
White's t-statistics (12.62) * (-4.20) * (4.70) *
Panel B: Results Based on 2,196 Announcements in 1997-1999, Adjusted
[R.sup.2] = 20.51%
Coefficient 31.18 -10.69 8.29
White's t-statistics (8.37) * (-9.16) * (5.44) *
Variable [D.sub.+.sup.*] UE U[E.sup.*]
[D.sub.Q] [D.sub.+]
Coefficient [[beta].sub.3] [[beta].sub.4] [[beta].sub.5]
Predicted Sign ? - ?
Panel A: Results Based on 725 Announcements in 1995-1996, Adjusted
[R.sup.2] = 13.74%
Coefficient -5.09 43.05 -119.78
White's t-statistics (-0.99) (1.62) (-2.61) *
Panel B: Results Based on 2,196 Announcements in 1997-1999, Adjusted
[R.sup.2] = 20.51%
Coefficient 6.74 -27.63 43.17
White's t-statistics (2.66) * (-2.06) ** (2.67) *
Variable SIZE [D.sub.LOSS] D/TA
Coefficient [[gamma].sub.1] [[gamma].sub.2] [[gamma].sub.3]
Predicted Sign + + +
Panel A: Results Based on 725 Announcements in 1995-1996, Adjusted
[R.sup.2] = 13.74%
Coefficient 0.72 12.54 0.37
White's t-statistics (0.69) (5.40) * (0.10)
Panel B: Results Based on 2,196 Announcements in 1997-1999, Adjusted
[R.sup.2] = 20.51%
Coefficient 3.85 17.35 9.53
White's t-statistics (6.05) * (9.35) * (3.47) *
Variable [D.sub.SWITCH]
Coefficient [[gamma].sub.4]
Predicted Sign +
Panel A: Results Based on 725 Announcements in 1995-1996, Adjusted
[R.sup.2] = 13.74%
Coefficient 6.88
White's t-statistics 2.26) *
Panel B: Results Based on 2,196 Announcements in 1997-1999, Adjusted
[R.sup.2] = 20.51%
Coefficient 1.68
White's t-statistics (0.73)
* and ** indicate statistical significance at the 1 percent and 5
percent levels based on the two-tailed White's t-test.
When financial data for t-1 are not available for IPO firms, we use
information in the pre-IPO year provided in the prospectus. Thus, as to
UE for IPO firms, we use net income in the pre-IPO year provided in the
prospectus. When auditor information is not available in year t-1 for
[D.sub.SWITCH], we compare the auditor in year t with the auditor in
year t-2 and assume there is no switch if they are the same auditor,
and vice versa. For IPO firms, we compare the auditor in the IPO year
with the auditor in the IPO prospectus who audited the financial
statements in the pre-IPO year.
LAG is the earnings announcement lag, measured as the number of days
between the fiscal year-end and the earnings announcement date;
[D.sub.+] is a dummy variable that equals 1 for positive earnings
surprises (UE > 0), and 0 otherwise; [D.sub.Q] is a dummy variable that
equals 1 if the audit opinion is modified, and 0 otherwise; UE is the
unexpected earnings scaled by lagged total assets; SIZE is the
logarithm of total assets (in million yuan) at the beginning of the
year; [D.sub.LOSS] is a dummy variable that equals 1 if the firm
experiences a loss in the year, and 0 otherwise; D/TA is the debt to
total asset ratio at the beginning of the year, to proxy for financial
distress; [D.sub.SWITCH] is a dummy variable that equals 1 if the firm
has a different auditor in year t from t-1, and 0 otherwise.
TABLE 7
Industry Estimates
Panel A: Coefficient Estimates (t-statistics) for the Specialization
Measures for the DAC Models, by Major Industry Groups
Industry LEADER DOMINANCE SHARE
Construction -0.0001 0.005 0.012
and Mining (-0.058) (0.766) (1.225)
Manufacturing -0.004 0.0002 0.001
(-0.402) (0.055) (0.177)
Transportation -0.003 ** -0.014 *** -0.021 ***
(-1.748) (-8.103) (-3.283)
Trade -0.004 ** -0.005 * -0.006
(-2.208) (-1.643) (-1.212)
Finance and -0.002 0.003 0.002
Insurance (-0.933) (0.464) (0.132)
Services -0.027 *** -0.039 *** -0.058 ***
(-7.170) (-12.086) (-6.691)
Industry MOSTCL SHARECL NCLIENTS
Construction -0.001 0.011 0.0000
and Mining (-0.396) (0.493) (0.166)
Manufacturing -0.0031 *** -0.021 ** 0.0002 ***
(-3.246) (-2.285) (12.028)
Transportation 0.002 -0.024 *** 0.0003 ***
(1.063) (-2.714) (8.200)
Trade -0.002 -0.012 0.001 ***
(-0.783) (-0.787) (6.738)
Finance and -0.007 *** -0.052 *** -0.0001 ***
Insurance (-3.085) (-2.991) (-2.807)
Services -0.017 *** -0.010 0.0005 ***
(-5.210) (-0.441) (20.144)
Panel B: Coefficient Estimates (t-statistics) for the Specialization
Measures for the ERC Models, by Major Industry Groups
Industry LEADER DOMINANCE SHARE
Construction 0.047 *** -0.027 0.126 ***
and Mining (3.628) (-0.812) (4.501)
Manufacturing -0.004 -0.028 ** -0.005
(0.358) (-1.700) (-0.163)
Transportation 0.028 0.042 0.055
(1.126) (1.118) (0.478)
Trade 0.019 * 0.061 * 0.065
(1.337) (1.386) (1.193)
Finance and 0.052 * 0.135 -0.040
Insurance (1.320) (0.977) (-0.181)
Services 0.036 ** 0.023 * 0.125 **
(2.147) (1.341) (2.288)
Industry MOSTCL SHARECL NCLIENTS
Construction 0.022 ** 0.395 *** 0.002 ***
and Mining (1.817) (3.383) (5.683)
Manufacturing -0.015 * -0.084 -0.0002 **
(-1.468) (-0.956) (-1.685)
Transportation 0.035 * 0.197 0.0004
(1.438) (1.071) (1.104)
Trade 0.022 ** 0.258 ** -0.0002
(1.651) (2.081) (-0.267)
Finance and 0.022 -0.230 -0.001 **
Insurance (0.615) (-0.782) (-1.860)
Services 0.038 *** 0.053 -0.000
(2.352) (0.432) (-0.699)
***, **, and * indicate significance at the 1 percent, 5 percent,
and 10 percent levels, respectively, one-tailed.
We thank the editor and two anonymous reviewers for their valuable comments and suggestions. We also acknowledge the financial assistance of The Chinese University of Hong Kong and Charles Tandy American Enterprise Center at Texas Christian University.
(1) In this paper, an earnings surprise relates to whether unexpected earnings are positive (good news) or negative (bad news). When we discuss the magnitude of the earnings surprise, we will make it explicit.
(2) Many recent studies have referred to the timing pattern of earnings announcements as the motivation for their analysis (Dye 1991), to provide a contrast with other aspects of their research (Easton and Zmijewski 1993), or to provide a possible explanation for their results (Sloan 1996).
(3) Listed Chinese companies do not make preliminary earnings announcements. Their annual operating results are first released to the public through the publication of the abstracts of their annual reports in newspapers designated by the Chinese Securities Regulatory Commission (CSRC).
(4) The following case sheds light on this issue. Chengdu Hongguang Co. raised RMB 410.2 million yuan in its May 1997 initial public offering. It released its 1997 annual report on April 30, 1998, the statutory deadline for the release of its annual results, showing a loss of RMB 198.4 million yuan in fiscal year 1997. The market was surprised by the extent of the loss and the company's share price dropped by more than 16 percent immediately after the announcement.
(5) See xiang (1998) for a detailed discussion on Chinese accounting standards setting process and progress. All accounting standards and regulations in China are available at http://www.snai.edu/asp/news/showfolder.asp?folderid=257, maintained by the Shanghai National Accounting Institute.
(6) In exceptional cases, firms are allowed to apply to the stock exchanges for an extension of up to two months. Compared to U.S. firms (Easton and Zmijewski 1993), listed Chinese firms' level of compliance with the statutory deadline of annual earnings announcements is quite high. We find only 21 firm-year observations with announcement dates later than April 30 during 1995-1999.
(7) Since 1999, the CSRC has required firms expecting a loss to make a preliminary announcement once management is reasonably certain about the outcome. However, the announcement does not contain any specific amount of the possible loss. There have been a few cases in which the expected loss does not occur at all.
(8) Since December 1999, the CSRC has also required firms listed on Shanghai and Shenzhen stock exchanges to post the full report on http://www.sse.com.cn and http://www.cninfo.com.cn, respectively.
(9) In June 2000, the CSRC required a CPA firm to have 20 CPAs on staff in order to audit financial statements of listed firms.
(10) The numbers are larger than those in Table 1 because firms listed during January 1 to April 30 of the following year are also included.
(11) Prior to the adoption of the new auditing standards, Chinese auditors followed "suggested guidelines" issued by the Ministry of Finance in 1992. They were less enforceable than the new standards.
(12) All auditing standards and regulations are available at http://www.cicpa.org.cn/FaguiResult.asp.
(13) Another circumstance requiring an unqualified opinion with an explanatory paragraph in the U.S. is a report involving other auditors' work, but this is rare in China. We have seen only one case so far, for Hainan Nanyang Shipping in 1997. Its auditor issued an unqualified opinion with an explanatory paragraph stating that the firm established two wholly owned subsidiaries in Hong Kong and their 1997 financial statements were audited by a local audit firm in Hong Kong.
(14) Earlier studies suggest two possible theories to explain this empirical regularity. The first one is the "stakeholder theory," which suggests that in the absence of an opportunity to hide bad news because of mandatory disclosure requirements, managers have incentives to delay its release (Watts and Zimmerman 1990). Alternatively, the "internal reporting hypothesis" maintains that managers are concerned about internal performance evaluation. Thus, managers are motivated to delay reporting bad news within the firm until it is verified, justified, and restated (Laurie and Pastena 1975).
(15) Whittred and Zimmer (1984) also examine the relationship between the timeliness of financial reporting and financial distress for a sample of Australian firms.
(16) Listed Chinese firms were originally authorized to issue common shares only to Chinese citizens (which are called "A-shares"), but some of them have since been allowed to issue B-shares that are available only to foreign investors. Both A-and B-shares are listed in Chinese stock exchanges. Some Chinese firms are also allowed to list their shares at the Hong Kong stock exchange (so called "H-shares"). B- and H-share firms are subject to more stringent disclosure and auditing requirements on the basis of International Accounting Standards, while A-share firms follow the Chinese GAAP. See Aharony et al. (2000) for a detailed discussion on the institutional features of B- and H-shares market.
(17) For non-IPO firms, we collect relevant information from 1994 annual reports to calculate DEL (and [D.sub.SWITCH] as defined below) for 1995.
(18) We thank an anonymous referee for the suggestion to use the lagged total asset as the scaler in Equation (1).
(19) Panel A of Table 4 indicates that the mean standard deviation of the reporting lag is 17.73 days and the mean range of reporting lag is 36 days, respectively, both greater than the corresponding 7.2 and 22.4 days in Chambers and Penman (1984). That is, the within-firm variability of reporting lag is greater in China than in the U.S. But the result is partly due to the fact that China's statutory annual earnings announcement period is four months from the fiscal year-end, compared to three months in the U.S.
(20) While the differences in DEL between positive and negative earnings surprise groups are significant at the 1 percent level for unqualified opinion and unqualified opinion with an explanatory note groups, it is only marginally significant for the group of qualified, disclaimer, or adverse opinions. However, the differences in DEL among the three audit opinion groups are statistically insignificant for both positive and negative earnings surprises. The results are consistent with the proposition that DEL is not as good as LAG to proxy for the timeliness of earnings announcements in China.
(21) In contrast, results for DEL indicate that both the mean and median DEL for the group of 754 unqualified negative earnings surprises are significantly positive and larger than the corresponding mean and median for the group of 48 positive earnings surprises with qualified, disclaimer, or adverse opinions. Similarly, both the parametric and nonparametric tests indicate that the 754 unqualified negative earnings surprises are significantly delayed, while the 101 positive earnings surprises with modified opinions (when unqualified opinions with explanatory notes are combined) are significantly accelerated.
(22) This is smaller than the adjusted [R.sup.2] of 43 percent reported by Bamber et al. (1993), partly because we do not control variables such as auditor structure due to the lack of such data in China. On the other hand, it is much greater than the adjusted [R.sup.2] of 4 percent reported by Begley and Fischer (1998).
(23) We exclude the top and bottom 1 percent extreme values of UE and repeat Regression (2). The results are similar to those reported in Panel A of Table 6, except for UE. The coefficient for UE becomes insignificant, which indicates that the significant relation between the magnitude and reporting for earnings decreases may be driven by extreme earnings decreases. It would be interesting to see whether this is unique to China or also exists for Begley and Fischer's (1998) sample in the U.S.
(24) If the regession model is estimated without the interaction variable ([D.sub.+] *[D.sub.Q] .when it is significantly different from zero, then a relevant explanatory variable is excluded. If this variable is correlated with [D.sub.+] and [D.sub.Q], then the estimators of [[beta].sub.1] and [[beta].sub.2] will be biased and inconsistent. Furthermore, the estimators of the variances of these parameters will be biased upward. If the interaction term is uncorrelated with [D.sub.+] and [D.sub.Q], then the estimators of [[beta].sub.1] and [[beta].sub.2] will be unbiased and consistent, but the estimators of their variances will still be biased upward.
REFERENCES
Aharony, J., C. J. Lee, and T. J. Wong. 2000. Financial packaging of IPO firms in a traditional economy: The case of B-shares and H-shares in China. Journal of Accounting Research 38: 103-126.
Ashton, R. H., J. J. Willingham, and R. K. Elliot. 1987. An empirical analysis of audit delay. Journal of Accounting Research 25: 275-292.
--, P. R. Graul, and J. D. Newton. 1989. Audit delay and the timeliness of corporate reporting. Contemporary Accounting Research 5: 657-673.
Bamber, E. M., L. S. Bamber, and M. P. Schoderbek. 1993. Audit structure and other determinants of audit report lag: An empirical analysis. Auditing: A Journal of Practice & Theory 12: 1-23.
Begley, J., and P. E. Fischer. 1998. Is there information in an earnings announcement delay? Review of Accounting Studies 3: 347-363.
Carter, M., and B. Soo. 1999. The relevance of form 8-K reports. Journal of Accounting Research 37: 119-132.
Chambers, A., and S. Penman. 1984. Timeliness of reporting and the stock price reaction to earnings announcements. Journal of Accounting Research 22: 21-47.
Choi, S. K., and D. C. Jeter. 1992. The effects of qualified audit opinions on earnings response coefficients. Journal of Accounting and Economics 15: 229_248.
DeFond, M. L., T. L Wong, and S. H. Li 1999. The impact of improved auditor independence on audit market concentration in China. Journal of Accounting and Economics 28: 269-305.
Dye, R. 1991. Informationally motivated auditor replacement. Journal of Accounting and Economics 14: 347-374.
Easton, P., and M. Zmijewski. 1993. SEC form 10K/10Q reports and annual reports to shareholders: Reporting lags and squared market model prediction errors. Journal of Accounting Research 31:113-129.
Givoly, D., and D. Palmon. 1982. Timeliness of annual earnings announcements: Some empirical evidence. The Accounting Review 57: 486-508.
Haw, I. M., D. Qi, and W. Wu. 2000. Timeliness of annual report releases and market reaction to earnings announcements in an emerging capital market: The case of China. Journal of International Financial Management and Accounting 11: 108-131.
Kalay, A., and U. Loewenstein. 1986. The information content of the timing of dividend announcements. Journal of Financial Economics 16: 373-389.
Keller, S. 1986. Reporting timeliness in the presence of subject to audit qualifications. Journal of Business Finance and Accounting 13:117-130.
Kinney, W. R., Jr., and L. S. McDaniel. 1993. Audit delay for firms correcting quarterly earnings. Auditing: A Journal of Practice & Theory 12: 135-142.
Kross, W., and D. A. Schroeder. 1984. An empirical investigation of the effect of quarterly earnings announcement timing on stock returns. Journal of Accounting Research 22: 153-176.
Laurie, A., and V. Pastena. 1975. How promptly do corporations disclose their problems? Financial Analysts Journal 31: 55-61.
Newton, J. D., and R. H. Ashton. 1989. The association between audit technology and audit delays. Auditing: A Journal of Practice & Theory 8: 22-37.
Pastena, V., and R. Ronen. 1979. Some hypotheses on the pattern of management's information disclosures. Journal of Accounting Research 17: 550-564.
Penmen, S. 1984. Abnormal returns to investment strategies based on the timing of earnings reports. Journal of Accounting and Economics 6: 165-183.
Schwartz, K., and B. Soo. 1996. The association between auditor changes and reporting lags. Contemporary Accounting Research 13: 353-371.
Skinner, D. 1994. Why do firms voluntarily disclose bad news? Journal of Accounting Research 32: 38-60.
Sloan, R. 1996. Do stock prices fully impound information in accruals and cash flows about future earnings? The Accounting Review 71: 289-316.
Xiang, B. 1998. Institutional factors influencing China's accounting reforms and standards. Accounting Horizons 12: 105-119.
Watts, R. L., and J. L. Zimmerman. 1990. Toward a positive theory of the determination of accounting standards. The Accounting Review 65: 131-156.
Whittred, G. 1980. Audit qualification and the timeliness of corporate annual reports. The Accounting Review 55: 563-577.
--, and I. Zimmer. 1984. Timeliness of financial reporting and financial distress. The Accounting Review 59: 287-295.
Submitted: November 2000
Accepted: March 2003
In-Mu G. Haw is a Professor at Texas Christian University and the Chinese University of Hong Kong, Kyungjoo Park is an Associate Professor at The Hong Kong Polytechnic University, Daqing Qi is an Associate Professor at Cheung Kong Graduate School of Business, and Woody Wu is a Professor at The Chinese University of Hong Kong.
 Printer friendly
 Cite/link
 Email
 Feedback
| |
| Author: | Haw, In-Mu G.; Park, Kyungjoo; Qi, Daqing; Wu, Woody |
|---|---|
| Publication: | Auditing: A Journal of Practice & Theory |
| Geographic Code: | 9CHIN |
| Date: | Sep 1, 2003 |
| Words: | 15002 |
| Previous Article: | Management trade-offs of internal control and external auditor expertise. |
| Next Article: | The effectiveness of expectation models in recognizing error patterns and generating and eliminating hypotheses while conducting analytical... |
| Topics: | |