后安然审计报告的隐藏性标准：证据筛选【外文翻译】.doc

1、 外文翻译 原文 Post-Enron implicit audit reporting standards: sifting through the evidence Material Source: Open Access publications from Katholieke Universiteit Leuven Author: Marleen Willekenst, Heidi Vander Bauwhede and Piet Sercu When, after Enrons collapse (October-December 2001), the less than glamo

2、rous role played by Arthur Andersen gradually became evident, analysts and the press suddenly eyed audit reports much more critically. This meant a higher probability that audit failures, if any, might be detected. And when, a few months later, the unthinkable happened and a BIG5 auditor sank in a m

3、atter of months, in the minds of the profession the expected damage to the auditor if and when a failure is discovered was revised upwards by several notches. Some response to the resulting increase in auditors business risk was natural, under the circumstances. SarbanesOxley raised the explicit aud

4、it standards, by imposing new rules. We argue that auditors simultaneously raised their implicit audit standards-that is, they increased compliance with the existing explicit audit reporting standards, or lowered the materiality threshold. l The tell-tale sign of a rise of the implicit standards, ev

5、erything else being the same, would be a higher incidence of non-clean audit opinions; and such a higher incidence is exactly what we observe, even after sifting out effects from increased risk and changing clienteles in the sample.In fact, the sifted evidence is in many ways stronger than the prima

6、 facie one. In the remainder of this paper we first discuss the auditors alternatives and the practical difficulties we face when measuring their response. We next describe the sample in greater detail, we estimate to what extent shifts in company risks have affected auditor behavior, and test wheth

7、er there remains a residual schift that may be attributable to an Enron-Andersen effect. I.The prima facie evidence Toughening the implicit standards would surely have been a sensible response to the auditors increased business risk. First, a prior non-clean opinion does reduce the chance that the a

8、uditor gets sued if the auditee keels over (Carcello and Palmrose, 1994). Second, the auditors alternative ways to handle the increased business risk must have looked far less appealing. For instance, they could have bought outside insurance, or increased fees so as to auto-insure, or stepped up the

9、 auditing effort itself. But each of these would have meant higher fees;2 and getting more money would have been a hard sell in the critical climate of the time. Next to increasing the standards, the auditors might also have weeded out the riskier audit clients,3 but we do not explicitly study that

10、possible reaction. In short, raising implicit auditing reporting standards should have made sense. And sure enough, among the Worlds cope-covered non-financials in our sample, the number of non-clean opinions rose from 5.3 percent in 1999 to 6.9 percent in 2002, the first unambiguously postEnron/And

11、ersen year4 (Table 1). Upon closer inspection of the figures, however, the story no longer appears as cut and dry, and the main task we set ourself in this paper is to show that the simple univariate analysis, above, does get us the right picture. II.Tougher standards or tougher circumstances? Table

12、 1 shows that, first, the higher incidence of non-clean opinions seems to be confined to big-5 auditees. Among the Worlds cope-covered non-financials with full data and a non-BIG5 auditor, the relative number of non-clean opinions did not change all that much over the 1999-2002 period, while for BIG

13、5 auditees the number rose by 2 percent (i.e. from 4.1 percent to 6.1 percent). Did the smaller players not feel any pressure? One could indeed argue that the amount of reputational capital at risk for a BIG5 firm was higher (DeAngelo, 1981), but thats a far cry from saying that smaller firms stood

14、nothing to lose at all. Or were the smaller ones already so perfectionist-“we try harder“?-that no extra action was called for? Adherents of this view might point out the four times higher incidence of non-clean opinions;5 but critics would reply that the non-BIG5 clientele may be different, too. Th

15、e BIG5/non-BIG5 divide is not the only puzzling figure in Table 1. Equally mystifying,from the Enron/ Andersen perspective, for non-BIG5 firms the peak year within the 1999-2002 window actually is 2000, which is unambiguously prior to Enron. So there must have been other factors at work. A change in

16、 economic circumstances may have mattered, for one. 2000 was the year the leT bubble burst, and one would expect relatively more of these companies to have a non-BIG5 auditor. But the economic context could also have been responsible for part of the 2001 rise in BIG5 qualifications and modifications

17、: notorious events in 2001 beside Enron include the widening of the stock-market slump, IX-H, and a drastic slowdown in general economic activity. In short, the deterioration of many auditees financial situations should have made the auditors more careful even without the Enron-Andersen event. Anoth

18、er interfering factor might have been shifts in the sample, generating noise and, possibly, bias. From Table 1 we indeed note that between 1999 and 2002 the number of non- BIG5 audit clients in the sample dropped by over one third. True, most of the deleted firms dropped from the sample because of a

19、n incomplete.Worlds cope record, not because of a take-over or bankruptcy. Still, such a drastic shift in the composition of the sample may have masked a possible toughening of non- BIG5 audit reporting standards, especially if incomplete data records would go together with generally more shoddy rep

20、orting practices and, therefore, more objections from the auditor. Such a correlation between data incompleteness and poor reporting practices is not inconceivable: in our 大型 -customer sample, which supposedly tends to contain the safer and more careful firms,the shrinkage because of missing data is

21、 just 11 percent. Our study is, to the best of our knowledge, the only one thus far studying the auditors reaction to Enron/ Andersen, but several other articles indirectly support our logic.Studies of the reaction of the stock market (Asthana et al. 2003; Chaney and Philipich 2002; Krishnamurthy et

22、 al. 2002; Callen and Morel 2002) show that stock prices of clients of both Andersen and other BIG5/4 auditors dropped significantly, especially after the release of negative information concerning the role of Andersen in the Enron case. This clearly suggests that auditor brand names and reputations

23、 were severely damaged and that serious doubt existed concerning audit quality and, consequently, the credibility of audited financial statements. There are also studies showing auditors reaction to a relaxation of business risk rather than a rise in it. Notably, Geiger and Raghunandan (2001, 2002)

24、and Francis and Krishnan (2002) report that auditors became less likely to issue a going-concern modified opinion after the Private Securities Litigation Reform Act of 1995, which reduced the likelihood of litigation against auditors. Our prima facie findings-a rise of implicit audit standards follo

25、wing an increase in risk-is consistent with this literature. III.The sample and the variables Our sample includes all U.S. listed firms for which data are available on Worldscope for the fiscal years 1999, 2001, 2002 or 2002. This provides about 30,000 records. As in prior audit reporting research (

26、Reynolds and Francis, 2001; DeFond, Raghunandan and Subramanyam, 2002) we exclude financial institutions, i.e. 5,000 records relating to SIC codes 6000-6999, because the measures of financial distress that will be used in our empirical model are not applicable to the financial sector. We also exclud

27、e observations with missing values (about 12,500 records).This process results in a sample of 12686 firm-years relating to 3367 different companies. The main test variable is the post-Enron indicator, I Enr. There is some a priori ambiguity whether the full reaction to Enron/ Andersen should already

28、 be expected in the audits of the 2001 statements. To steer clear of that, we treat 1999-2000 as pre-Enron (that is, lEnT = 0), skip 2001 in the tests, and treat 2002 as the post-Enron sample (IEnT = 1). In light of the evidence from Table 1, we let the intercept and the coefficient for I Enr differ

29、 across the BIG5 and non- BIG5 samples. This approach assumes that the coefficients for the control variables are the same for the two sets of auditors. We test this assumption by running separate regressions for the BIG5 and non- BIG5 observations, respectively. IV.Results and discussion The statis

30、tical insignificance of the post-Enron shift for non- BIG5 firms warrants some comments. This insignificance is the result of a largeish standard error (caused, in turn, by the smaller sample and more coming-and-goings in the audit client list), not of a small estimated effect. In fact, also the hyp

31、othesis that there is no difference between the post-Enron shifts in non- BIG5 and BIG5behavior is statistically acceptable. In unclear cases like this, ones priors should playa major role, and there are no good reasons why smaller auditors would not have reacted at all. Note that, algebraically spe

32、aking, the estimated effect is positive. That is, the impression of no effect we get from the univariate data in Table 1 appears to be caused by shifts in the sample; taking into account these aspects, one actually observes a rise in the incidence of non-clean opinions. True, Table 3 gives the impre

33、ssion that the effect is still smaller than for BIG5 firms, but that is an illusion too. The coefficients in the table measure the sensitivity of the log odds ratio, which is non-linear (S-shaped) in the arguments. In fact, in view of the higher probabilities, the typical non- BIG5 auditee is higher

34、 up the Scurve, where the sensitivities to the regressors are higher too. To estimate the change in probabilities holding constant the sample, we compute the 2000 and 2001 fitted odds for each firm, freezing the sample composition and company characteristics at their 2000 level. Table 4 shows the re

35、sulting marginal probability. For BIG5 auditees, the ceteris-paribus effect is estimated at +1.75 percent, quite close to the observed rise of 1.93 percent. For non- BIG5 auditees, however, the diagnosis is that, holding constant the sample, we would have seen a rise of 2.5 percent rather than a dro

36、p by 1.6 percent. Thus, the apparent lack of reaction on behalf of non- BIG5 auditors seems to be because many of their clients in the sample became less risky. This fits in with the fact that their peak year, 2000, is the year of the bursting of the ICT bubble. We conclude with a brief discussion o

37、f the results for the control variables. The financial distress variables all come in with the expected sign, and quite significantly so: a higher bankruptcy risk and a negative profit or cash flow have a massive impact on the chances of getting a non-clean opinion, while customer size (log assets)

38、lowers the chances. The variables that, in some studies, proxy for complexity, do not pick up anything in our sample: neither AIR nor Inventory are associated with the probability of a non-clean opinion. V.Conclusion The accounting scandals and the demise of Andersen have increased auditors ex ante

39、business risk. As a result, stock markets revised downward the value of the audit firms (Asthana et al. 2003; Chaney and Philipich 2002; Krishnamurthy et al. 2002; Callen and Morel 2002). One commonsensical reaction on behalf of auditors should have been to apply the existing rules more carefully an

40、d, thus, issue more non-clean audit opinions. This is exactly what we see. Closer scrutiny reveals that the higher incidence of non-clean opinions is not due to the (substantial) changes in the audit client list or their balance sheets. Instead, shifts in the client characteristics seem to have mask

41、ed the Enron effect, and especially so in the non- BIG5 sample. This study mirrors earlier results where auditors relaxed their standards following a drop in business risk (Geiger and Raghunandan, 2001, 2002; Francis and Krishnan, 2002). 译文 后安然审计报告的 隐藏 性标准： 证据 筛选 资料来源 : 天主教鲁汶大学开放存取出版物 作者： Marleen Wi

42、llekenst, Heidi Vander Bauwhede and Piet Sercu 当前 ， 在 安然公司的倒闭 后 （ 10 月至 2001 年 12 月）， 因 安达信发挥 的迷惑 作用 而变得 逐渐变得明显， 这 使得分析师和媒体会 更加批判性 的 关注 审计报告 。 如果有风险的话，可能会被检测到， 这 会 意味着审计失败的概率较高 。 几个月后， 不可思议的事情发生了，一名 隶属于 五大会计师事务所 的 审计师在几个月间失败，而在他的失败过程中，假如能发现一些缺口，并及时修正，则是唯一可能避免损失的方法 。 在这种情况下 ，一些在审计业务的风险自然而然会随之增加 。 Sar

43、banesOxley 提出了明确的审计标准 ， 实施新的规则 。 同时我们认为，审计 师 的 审计标准 隐性 的 提高了 ，也就是说，他们增加 现有的审计报告标准明确规定，或降低门槛的重要性。 一个标准上升的 标志 是公告的升级 ， 在一切 相同的 情况下 ， 这 将是一个发生率较低 的审计意见，正如 我们看到 的，就算从风险和不断变化的客户群的 样本 中 进行筛选 ， 而这样的发生率还是较高， 事实 上 ， 筛选 过的证据 在很多方面能表现出更强的一面 。 在本文的其余部分 中，我们首先讨论 我们面临的 审计师在进行选择和测量时的 实际困难， 和他们的反应。接下来 我们更详细地描述了 下一次

44、的样本， 估计到底 什么 样的程度的变化会 影响 公司的风险， 进而 影响审计人员的行为，并测试是否仍有残余 量，这种情况很 可能是由于安然和 安达信的效果。 1 表面证据 增强 隐 性 标准将肯定是一个 能 明智的回应审计员增加审计业务的 风险。 首先， 首先， 如果 是 受审核 方超过了界 线，且 之前不是 无保留意见的话 ，那么会减少了审计人员得到起诉 的机会 （ Carcello 和 Palmrose， 1994 年） 。第二，审计师 用其他方法来提高处理业务的质量而使得风险少得多，这一定很吸引人 。例如， 1.他们可以在买保险 或增加费 用 外， 能 够自动投保，或加强了审计事业本身

45、， 但是，这些都意味着每个更高的费用 。 2.在关键的时间，能够从更容易的工作中赚到更多的钱，在 下一步 中用来提高标准，审计人员也可以淘汰更 高 的风险审计客 户。 3.但是 我们并没有明确的研究 过受害者 可能出现的 反应。 总之，提高审计报告的标准隐性应该是有意义的。果然， 在我们的样本中，世界 应对覆盖的非金融类 非正确 意见人数从 1999 年的 5.3%上升 至 6.9%， 2002年，第一次明确后安然 /安达信（见表 1）。然而，就算经仔细检查的数据，故事已经不再出现了。而我们的 主要任务是 ， 在本论文 中要表明，简单的单因素分析，上面和右边是让我们 的 数据 图片。 2 采取

46、更强硬的标准或者采取更强硬的情况 表 1 显示，第一，不正确 意见发生率较高 的情况 似乎仅限于 大型 审计。 在世界上非 大型 审计员覆盖的数据还不够多， 财务不正确的 意见相对数 在全国各地 并没有 像 1999-2002 年期间改变的这么 多， 而大型的 审计的数量上升 2%（ 即4.1%至 2005 年的 6.1%）。 难道 小公司没有感到任何压力吗？人们可以争辩说，确实声誉资本在一个大型 的 公司 中 风险金额较高（ DeAngelo， 1981），但是这是从规模较小的公司说，站在什么 角度 损失都相距甚远。或者是较小的 公司是完美 者 ，即使我们更努力，没有多余的成效算什么？ 没有

47、额外的行动就是要求吗？这种观点的支持者 可能指出 不正确的 意见的情况 会高四倍； 但批评者会回答说，非 大型 的事务所 可能会有所不同了。 该 小 型 和 大型 事务所 的 鸿沟并不是唯一令人费解的数据在 表 1。 从安然 /安达信 的角度来看 同样都是神秘的 ，非 大型 公司在 1999-2002 年 的高峰窗口实际上是 2000 之前 。 所以一定有其他因素在起作用。改变经济情况可能没关系 ,这是其一。 2000 年是全国 泡沫破灭 的一年 ，人们会期望这些公司相对更多有小 型 审计 师。 但是，经济方面也已在 大型 的资格和修改 以及 2001 年上升的部分负责人 中： 安然事件包括

48、在内的 2001 年臭名昭著的股市暴跌， 第 IX- H 和在一般经济活动急剧放缓 。 总之，审核员即使没有安然、 安达信事件 也应该做的更仔细，但是许多审计单位 状况 却一直 恶化 。另一个 干扰因素可能 是 已经变化的样本、 产生的噪音，可能还有偏见。从表 1中我们确实注意 到，在 1999 年和 2002 年的小型 审计客户样本中有超过三分之一的下降。 的确， 不是因为处理破产或被接管的保险公司 而导致 大部分的删除公司的样稿 , 留下了一个不完整的 记录。 尽管如此 ，这样 成分急剧的改变 的样品有可能掩盖了小 型 审计报告标准的增加量 ， 特别是如果记录不完整的数据通常会配以以次充好

49、的报告行为 ，因此， 更多的反对意见 会投诉给审计人员 。这种数据之间的相关性不完整和以次充好的 做法 并不 是不可思议 的 ：在我们的 大型 事务所的 客户 往 来样 本中 ，据称往往包含更安全和更仔细的公司，因为公司缺少数据的收缩只是 11%。 以我们的 研究 所知 ， 迄今为止唯一的一个研究审计 师的反应， 但有几种安然 /安达信的表现间接地支持我们其他方面的股票市场逻辑研究的反应。（ Asthana 等， 2003； 张秀珍 和 Philipich 2002； 克里希纳穆尔蒂等， 2002。卡朗和莫瑞尔 2002） 这 表明，两个 大型 事务所的 客户 和其他客户的股票价格 相比，1/ 4 的 审计师 有着 显着下降，特别是在有关安达信 /安然案中 发布负面消息 的作用。 这清楚地表明，审计师的品牌和声誉受到严重破坏，是跟审计质量存在着严重疑问有关，因此，要严格对待审计的财务报 表的可信度。也有研究表明审计 师的反应，放 宽企业经营风险，而不是让它在上升。值得注意的是，盖革和Raghunandan（ 2001 年， 2002 年）和弗朗西斯和 Krishnan（ 2002）报告，审计人员可能变得不那么 容易出具持续性经营的意见