The consequences of shifting the IPO offer pricing power from securities regulators to market participants in weak institutional environments: Evidence from China ^(2){ }^{2} 将首次公开募股（IPO）定价权从证券监管机构转移到市场参与者在弱制度环境中的后果：来自中国的证据 ^(2){ }^{2}
Jun Chen ^(a){ }^{\mathrm{a}}, Bin Ke^(b)\mathrm{Ke}^{\mathrm{b}}, Donghui Wu^(c,**)\mathrm{Wu}^{\mathrm{c}, *}, Zhifeng Yang ^(d){ }^{\mathrm{d}} 陈俊 ^(a){ }^{\mathrm{a}} , 宾 Ke^(b)\mathrm{Ke}^{\mathrm{b}} , 东辉 Wu^(c,**)\mathrm{Wu}^{\mathrm{c}, *} , 杨志峰 ^(d){ }^{\mathrm{d}}^(a){ }^{a} Department of Finance and Accounting, School of Management, Zhejiang University, Hangzhou, China 浙江大学管理学院财务与会计系，中国杭州^("b "){ }^{\text {b }} Department of Accounting, NUS Business School, National University of Singapore, Mochtar Riady Building, BIZ 1, #07-53, 15 Kent Ridge Drive, 119245, Singapore 新加坡国立大学商学院会计系，莫克塔尔·里亚迪大楼，BIZ 1，#07-53，肯特岭道 15 号，邮政编码 119245，新加坡^("c "){ }^{\text {c }} School of Accountancy, The Chinese University of Hong Kong, Sha Tin, Hong Kong, China 香港中文大学会计学院，沙田，香港，中国^("d "){ }^{\text {d }} Department of Accountancy, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon Tong, Kowloon, Hong Kong, China 香港城市大学会计系，九龙塘达知道 83 号，九龙，香港，中国
A R T I C L E I N F O 文章信息
Article history: 文章历史：
Received 30 June 2016 收到日期：2016 年 6 月 30 日
Received in revised form 1 October 2016 修订版收到于 2016 年 10 月 1 日
Accepted 8 October 2016 接受日期：2016 年 10 月 8 日
Available online 11 October 2016 可在线获取 2016 年 10 月 11 日
We examine the consequences of shifting the IPO offer pricing power from securities regulators to market participants in a representative weak investor protection country, China. We show IPO offer prices relative to reported earnings are less depressed when determined by market participants than by securities regulators. IPO firms are also less likely to select a low quality auditor or inflate the pre-IPO earnings when IPO offer prices are determined by market participants. However, we find no evidence that IPO offerings are more likely to be overpriced when offer prices are determined by market participants. Furthermore, IPO firms’ financial reporting choices made at the time of the IPO have a long lasting impact on the firms’ subsequent financial reporting quality. Overall, our results contribute to the ongoing debate on the appropriate roles of securities regulators versus market forces in protecting public investors in markets with weak institutional environments. 我们研究了在一个代表性弱投资者保护国家——中国，将首次公开募股（IPO）定价权从证券监管机构转移到市场参与者的后果。我们发现，当 IPO 定价由市场参与者决定时，IPO 发行价格相对于报告的收益不那么低迷，而不是由证券监管机构决定。IPO 公司在 IPO 发行价格由市场参与者决定时，也不太可能选择低质量审计师或夸大 IPO 前的收益。然而，我们没有发现证据表明，当发行价格由市场参与者决定时，IPO 发行更可能被高估。此外，IPO 公司在 IPO 时做出的财务报告选择对公司后续的财务报告质量有长期影响。总体而言，我们的结果为关于在弱制度环境下保护公众投资者的证券监管机构与市场力量的适当角色的持续辩论做出了贡献。
There is a large and growing finance literature on initial public offerings (IPOs) around the world (e.g., Chaney and Lewis, 1998; Chan et al., 2004; Banerjee et al., 2011; Clarke et al., 2016; Nielsson and Wojcik, 2016). Most studies in this literature examine various issues resulting from the strategic interaction between corporate insiders (sellers) and outside investors (buyers). While the conduct of IPOs (including eligibility, timing, offer size, and pricing) is subject to various degrees of government regulation in almost all countries, there is little empirical research on the costs and benefits of such government regulation. This is surprising given the controversies surrounding the government’s intervention in the IPO process. The objective of this study is 全球范围内关于首次公开募股（IPO）的金融文献庞大且不断增长（例如，Chaney 和 Lewis，1998；Chan 等，2004；Banerjee 等，2011；Clarke 等，2016；Nielsson 和 Wojcik，2016）。该文献中的大多数研究考察了公司内部人士（卖方）与外部投资者（买方）之间战略互动所产生的各种问题。尽管几乎所有国家的 IPO 行为（包括资格、时机、发行规模和定价）都受到不同程度的政府监管，但关于这种政府监管的成本和收益的实证研究却很少。考虑到围绕政府干预 IPO 过程的争议，这一点令人惊讶。本研究的目标是
to contribute to this important debate by examining the impact of different IPO offer pricing regimes on the behavior of IPO firm insiders. 通过研究不同 IPO 发行定价制度对 IPO 公司内部人士行为的影响，为这一重要辩论做出贡献。
Broadly speaking, IPO offer pricing regimes around the world can fall within the following two extremes. At one extreme, referred to as the market-based approach, underwriters and investors, especially institutional investors, play the lead role in determining IPO offer prices and the securities regulator’s primary responsibility is to establish the necessary supporting institutions such as the rule of law that facilitate the market’s pricing of IPO shares. At the other extreme, referred to as the governmentbased approach, a country’s securities regulator directly sets IPO offer prices based on IPO firms’ financial performance indicators. 广义而言，全球的首次公开募股（IPO）定价机制可以分为以下两个极端。在一个极端，称为市场导向的方法，承销商和投资者，特别是机构投资者，在确定 IPO 发行价格方面发挥主导作用，而证券监管机构的主要责任是建立必要的支持机构，例如促进市场对 IPO 股票定价的法治。在另一个极端，称为政府导向的方法，一个国家的证券监管机构根据 IPO 公司的财务绩效指标直接设定 IPO 发行价格。
While it has been widely adopted in the U.S. and many other developed countries, the market-based approach is controversial and often resisted by regulators in many less developed financial markets. ^(1){ }^{1} One frequently cited justification for rejecting the mar-ket-based approach is that many important market supporting institutions such as reputable and independent underwriters and institutional investors either do not exist or are very weak in weak investor protection countries and therefore there is a high perceived risk that IPO offerings could be over priced under a market-based approach (Meng et al., 2004; CSRC 2009; Sheokand, 2012). Furthermore, even if a market-based approach is preferred, the transition from a government-based approach to a mar-ket-based approach may not be a sure success due to the resistance of vested interests and potential unintended negative consequences resulting from the transition. On the other hand, government regulators may not necessarily be as motivated and informed about general market and specific firm conditions as private investors (Jackson and Roe, 2009). Hence, a priori it is not obvious which approach is more effective in protecting the interests of IPO investors in countries with less developed financial markets. 尽管在美国和许多其他发达国家得到了广泛采用，但基于市场的方法在许多欠发达金融市场中是有争议的，往往受到监管机构的抵制。一个常被引用的拒绝基于市场的方法的理由是，在投资者保护较弱的国家，许多重要的市场支持机构，如声誉良好的独立承销商和机构投资者，要么不存在，要么非常薄弱，因此人们普遍认为在基于市场的方法下，首次公开募股的定价可能会过高（Meng et al., 2004; CSRC 2009; Sheokand, 2012）。此外，即使基于市场的方法更受欢迎，从政府主导的方法转向基于市场的方法也不一定会成功，因为既得利益的抵制和转型可能带来的潜在负面后果。另一方面，政府监管机构可能并不一定像私人投资者那样对一般市场和特定公司状况有足够的动机和了解（Jackson and Roe, 2009）。 因此，事先并不明显哪种方法在保护金融市场不发达国家 IPO 投资者的利益方面更有效。
In this study we examine the consequences of a shift in the IPO offer pricing regime from a government-based approach to a market-based approach in China, the largest emerging market economy with weak investor protection (Allen et al., 2005). There has been an ongoing debate since the start of China’s modern stock market in the 1990s on whether China should follow a market-based approach or government-based approach to regulating IPO offer prices (Luo, 2013; Deng, 2015). 在本研究中，我们考察了中国这一最大的新兴市场经济体在首次公开募股（IPO）定价制度从政府主导转向市场主导的转变所带来的后果，尤其是在投资者保护较弱的背景下（Allen et al., 2005）。自 1990 年代中国现代股市启动以来，关于中国应采取市场主导还是政府主导的方式来监管 IPO 定价的争论一直持续（Luo, 2013; Deng, 2015）。
We examine four specific research questions. First, we examine how IPO offer prices change when the IPO offer pricing approach shifts from a government-based approach to a market-based approach. Consistent with the hypothesis that the Chinese securities regulators suffer from severe agency problems that lead to a strong aversion to the risk of IPO offer overpricing, we find that the IPO offer prices relative to reported earnings are more significantly depressed under the government-based approach than under the market-based approach. 我们研究了四个具体的研究问题。首先，我们考察当 IPO 定价方法从政府主导转变为市场主导时，IPO 发行价格如何变化。与中国证券监管机构面临严重代理问题、导致对 IPO 发行价格过高风险强烈厌恶的假设一致，我们发现，在政府主导的方法下，IPO 发行价格相对于报告的收益被显著压低，低于市场主导的方法。
Second, we examine how a shift from the government-based approach to the market-based approach affects IPO firms’ financial reporting quality at the time of the IPO. We use two complementary approaches to assess financial reporting quality at the time of the IPO: (a) IPO firms’ likelihood of hiring a low quality auditor at the time of the IPO; and (b) the magnitude of upward earnings management in the pre-IPO period. We find that when the IPO offer pricing approach changes from the governmentbased approach to the market-based approach, IPO firms are less likely to hire low quality auditors and conduct upward earnings management in the years immediately prior to the IPO. 其次，我们考察从政府主导的方法转向市场主导的方法如何影响 IPO 公司在 IPO 时的财务报告质量。我们使用两种互补的方法来评估 IPO 时的财务报告质量：（a）IPO 公司在 IPO 时聘用低质量审计师的可能性；（b）IPO 前期的盈余管理幅度。我们发现，当 IPO 定价方法从政府主导转变为市场主导时，IPO 公司在 IPO 前的几年内聘用低质量审计师和进行盈余管理的可能性较低。
Our third question examines whether IPO offer prices are more likely to be overstated under the market-based approach than under the government-based approach. As noted above, one frequently cited argument against the market-based approach in less developed financial markets is that the market-based approach makes it easier for IPO firms’ insiders to sell over-priced shares to less informed and less sophisticated public investors. Since the stock prices of overvalued shares eventually have to reverse with the arrival of new information such as earnings announcements, the above argument implies a greater long-term stock price reversal relative to the IPO offer price for the IPOs under the market-based approach than under the government-based approach. We find no evidence that the long-term abnormal stock price performance relative to the offer price reverses to a greater extent for the IPOs under the market-based approach than the IPOs under the government-based approach. 我们的第三个问题考察了在市场导向方法下，IPO 发行价格是否更可能被高估，而不是在政府导向方法下。如上所述，针对市场导向方法在欠发达金融市场中的一个常被引用的论点是，市场导向方法使得 IPO 公司的内部人士更容易将高价股票出售给信息较少和不够成熟的公众投资者。由于被高估股票的股价最终必须随着新信息（如盈利公告）的到来而反转，上述论点暗示，在市场导向方法下，IPO 的长期股价相对于发行价格的反转程度大于在政府导向方法下的 IPO。我们没有发现证据表明，在市场导向方法下，IPO 的长期异常股价表现相对于发行价格的反转程度大于在政府导向方法下的 IPO。
Our final question examines IPO firms’ financial reporting quality over the long run, defined as three-five years after the IPO. Consistent with Stein (1989) and Shivakumar (2000), we conjecture that once a firm’s insiders adopt a low (high) quality financial reporting strategy at the IPO time, the insiders will find it in their best interest to continue to do so in subsequent financial reporting periods because of the stock market’s rational expectation. Consistent with this conjecture, we find IPO firms’ auditor choices made at the time of the IPO are sticky. We find no evidence that IPO firms that hire a low (high) quality auditor at the time of the IPO are more likely than a control sample of non-IPO firms to switch to a high (low) quality auditor three years after the IPO. Furthermore, we find that earnings quality continues to be higher in the three-five year period after the IPO, for the firms that went public under the market-based approach than for the firms that went public under the govern-ment-based approach. 我们的最后一个问题考察了 IPO 公司在长期内的财务报告质量，定义为 IPO 后三到五年。与 Stein（1989）和 Shivakumar（2000）的观点一致，我们推测，一旦公司的内部人士在 IPO 时采用低（高）质量的财务报告策略，内部人士会发现继续在后续财务报告期间这样做符合他们的最佳利益，因为股市的理性预期。与这一推测一致，我们发现 IPO 公司在 IPO 时的审计师选择是粘性的。我们没有发现证据表明，在 IPO 时聘请低（高）质量审计师的 IPO 公司在 IPO 三年后更可能转向高（低）质量审计师，相较于非 IPO 公司的对照样本。此外，我们发现，在 IPO 后的三到五年期间，采用市场导向方法上市的公司，其盈利质量持续高于采用政府导向方法上市的公司。
Any observed changes in corporate behavior resulting from a shift in IPO offer pricing from the government-based approach to the market-based approach could be due to confounding events (e.g., a general improvement in China’s investor protection environment). We perform a variety of robustness checks to rule out such alternative explanations. More importantly, an important strength of our study’s research design is that China’s IPO offer pricing regimes changed from a government-based approach to a market-based approach and then reversed back to a government-based approach during our sample period. The flip-flop creates a unique opportunity for us to more clearly identify the causal impact of the two contrasting IPO offer price regulatory approaches on our dependent variables of interest. 任何观察到的企业行为变化，可能是由于政府主导的首次公开募股（IPO）定价方式转变为市场主导的定价方式所引起的，这可能受到混杂事件的影响（例如，中国投资者保护环境的普遍改善）。我们进行了多种稳健性检验，以排除这种替代解释。更重要的是，我们研究设计的一个重要优势在于，在我们的样本期间，中国的 IPO 定价制度经历了从政府主导到市场主导再回到政府主导的变化。这种反复变化为我们提供了一个独特的机会，更清晰地识别这两种对比鲜明的 IPO 定价监管方式对我们关注的因变量的因果影响。
Overall, we can draw the following conclusions from our empirical analyses. First, our results suggest that even in a less developed financial market like China with the dominance of less sophisticated retail investors, public investors can still price protect by avoiding IPO offer overpricing under the market-based approach, consistent with Ekkayokkaya and Pengniti (2012). Second, our results suggest that due to the CSRC’s own agency problems, the government-based IPO offer pricing approach provides IPO firm insiders with a stronger incentive to manage earnings not only at the IPO time but also in the years subsequent to the IPO. Given that financial reporting quality is important to the efficient functioning of financial markets and the fact that earnings management is costly, our results suggest that the government-based IPO offer pricing approach causes greater damages than the market-based approach to the long term health of China’s financial markets. For these reasons, we conclude that the marketbased approach is more effective than the government-based approach in protecting the interests of public investors in China. 总体而言，我们可以从我们的实证分析中得出以下结论。首先，我们的结果表明，即使在像中国这样发展较少的金融市场中，零售投资者占主导地位，公众投资者仍然可以通过避免在市场导向的方法下的首次公开募股（IPO）定价过高来进行价格保护，这与 Ekkayokkaya 和 Pengniti（2012）的研究一致。其次，我们的结果表明，由于中国证监会自身的代理问题，政府导向的 IPO 定价方法使得 IPO 公司内部人士在 IPO 时以及 IPO 后的几年中更有动力进行盈余管理。考虑到财务报告质量对金融市场的有效运作至关重要，以及盈余管理的成本，我们的结果表明，政府导向的 IPO 定价方法对中国金融市场的长期健康造成的损害大于市场导向的方法。因此，我们得出结论，市场导向的方法在保护中国公众投资者的利益方面比政府导向的方法更有效。
Our study is directly related to the IPO literature (see Ritter 2003 and 2011 for reviews of the general IPO literature and Yong, 2007 for a review of the literature on Asian IPOs). Most studies in this literature focus on IPO underpricing and the long term performance of IPOs, but very few studies examine the IPO offer pricing per se. More importantly, while there are significant variations across both countries and time in IPO regulations, there has been little research that compares the costs and benefits of different IPO regulatory regimes. Notable exceptions include Ekkayokkaya and Pengniti (2012) that examine impact of governance reform on IPO underpricing and Cattaneo et al. (2015) that examine how regulatory changes affect IPO firm survival. Our study’s contribution is to use the flip-flop of the two IPO offer pricing regimes to demonstrate the consequences of shifting from a gov-ernment-based to a market-based IPO offer pricing approach in a country with a less developed financial market, China. Our results are important for many emerging market economies because many policy makers and the general public have yet to learn to trust markets (The Economist, 2015). 我们的研究与首次公开募股（IPO）文献直接相关（参见 Ritter 2003 年和 2011 年的一般 IPO 文献综述，以及 Yong 2007 年对亚洲 IPO 文献的综述）。该文献中的大多数研究集中在 IPO 定价不足和 IPO 的长期表现上，但很少有研究专门考察 IPO 的发行定价。更重要的是，尽管在不同国家和时间的 IPO 监管方面存在显著差异，但对不同 IPO 监管制度的成本和收益进行比较的研究却很少。值得注意的例外包括 Ekkayokkaya 和 Pengniti（2012 年）研究治理改革对 IPO 定价不足的影响，以及 Cattaneo 等（2015 年）研究监管变化如何影响 IPO 公司的生存。我们研究的贡献在于利用两种 IPO 发行定价制度的反转，展示在金融市场不发达的国家（中国）从政府主导的 IPO 发行定价方法转向市场主导的 IPO 发行定价方法的后果。我们的结果对许多新兴市场经济体具有重要意义，因为许多政策制定者和公众尚未学会信任市场（《经济学人》，2015 年）。
Our study is also related to the broad earnings management literature in emerging markets. While prior studies have provided evidence of aggressive earnings management by emerging market firms (e.g., Aharony et al., 2000; Chen and Yuan, 2004; Haw et al., 2005), less research is done to identify effective mechanisms to curtail such aggressive earnings management. The evidence from DeFond et al. (2000) shows that many proposed solutions by government regulators are often counterproductive. We contribute to this literature by demonstrating that a market-based approach is more effective than a government-based approach in reducing aggressive earnings management in a weak institutional environment like China. 我们的研究也与新兴市场中广泛的盈余管理文献相关。虽然之前的研究提供了新兴市场公司进行激进盈余管理的证据（例如，Aharony 等，2000；Chen 和 Yuan，2004；Haw 等，2005），但关于识别有效机制以遏制这种激进盈余管理的研究较少。DeFond 等（2000）的证据表明，政府监管机构提出的许多解决方案往往适得其反。我们通过证明在像中国这样制度环境薄弱的情况下，市场导向的方法比政府导向的方法更有效地减少激进盈余管理，为这一文献做出了贡献。
The rest of the paper is organized as follows. Section 2 describes the sample and data sources. Section 3 explains the institutional details of the three IPO regimes and analyzes the CSRC’s incentives. Section 4 discusses the hypothesis development, research design, and regression results on the differences in IPO firms’ auditor choice and pre-IPO earnings management across the three IPO regimes. Section 5 examines IPO firms’ long term abnormal stock price performance relative to the IPO offer price across the three regimes. Section 6 analyzes the IPO firms’ financial reporting quality three-five years after the IPO. Section 7 concludes. 本文的其余部分组织如下。第二节描述样本和数据来源。第三节解释三种 IPO 制度的制度细节，并分析证监会的激励。第四节讨论假设发展、研究设计以及关于三种 IPO 制度下 IPO 公司审计师选择和 IPO 前盈余管理差异的回归结果。第五节考察 IPO 公司相对于 IPO 发行价的长期异常股价表现。第六节分析 IPO 公司在 IPO 后三到五年的财务报告质量。第七节总结。
2. The sample 样本
Many aspects of China’s IPOs, including eligibility, timing, offer size, and pricing, are heavily regulated. However, as part of China’s economic reforms, China’s IPO regulatory environment has experienced several interesting changes over time. In this paper we examine three IPO regulatory regimes over the period January 1, 1997-December 31, 2004. Regime 1 covers January 1, 1997-February 11, 1999; Regime 2 covers February 12, 1999-November 6, 2001; and Regime 3 covers November 7, 2001-December 31, 2004. During the three regimes the way IPO offer prices were determined changed significantly while the other key features of IPO regulation (e.g., eligibility, timing, offer size) were kept relatively stable and still heavily regulated. 中国首次公开募股（IPO）的许多方面，包括资格、时机、发行规模和定价，受到严格监管。然而，作为中国经济改革的一部分，中国的 IPO 监管环境随着时间的推移经历了几次有趣的变化。本文考察了 1997 年 1 月 1 日至 2004 年 12 月 31 日期间的三个 IPO 监管制度。制度 1 涵盖 1997 年 1 月 1 日至 1999 年 2 月 11 日；制度 2 涵盖 1999 年 2 月 12 日至 2001 年 11 月 6 日；制度 3 涵盖 2001 年 11 月 7 日至 2004 年 12 月 31 日。在这三个制度期间，IPO 发行价格的确定方式发生了显著变化，而 IPO 监管的其他关键特征（例如，资格、时机、发行规模）则保持相对稳定，仍然受到严格监管。
Regime 1 started earlier than January 1, 1997, but our sample period for Regime 1 starts on January 1, 1997 because our subsequent empirical analyses require a matched control sample of seasoned firms (defined as firms that have been publicly listed for at least two complete fiscal years) for the IPO firms. Because mainland China’s modern stock markets started only in December 1990, the number of seasoned companies available for matching purposes would be too small if our sample includes the IPOs from the years prior to 1997. While the CSRC’s approach to regulating IPO offer prices continues to alternate between the govern-ment-based approach and market-based approach subsequent to the end of Regime 3 on December 31, 2004, we decided to exclude the post-2004 IPOs from our sample for the following reasons. regime 1 的开始时间早于 1997 年 1 月 1 日，但我们对 regime 1 的样本期从 1997 年 1 月 1 日开始，因为我们后续的实证分析需要一个与 IPO 公司匹配的成熟公司对照样本（定义为至少公开上市两完整财年的公司）。由于中国大陆的现代股票市场仅在 1990 年 12 月开始，如果我们的样本包括 1997 年之前的 IPO，能够用于匹配的成熟公司数量将会太少。尽管中国证监会在 2004 年 12 月 31 日结束 regime 3 后，继续在政府主导和市场主导的方式之间交替监管 IPO 发行价格，但我们决定将 2004 年后 IPO 排除在我们的样本之外，原因如下。
First, the Chinese IPO market experienced two prolonged IPO suspensions (i.e., mid 2005-mid 2006 and October 2008-May 2009) and more importantly the distinction between the government-based approach and market-based approach in the post2004 years is not as clear cut as in Regimes 1-3. For example, during the period from the mid 2006 to September 2008, IPO offer prices were determined by the market but the CSRC capped the PE multiple used in the offering price determination at 30, significantly higher than the cap of 20 during Regime 3. After using a market-based IPO offer pricing approach during the period from June 2009 to April 2012, the CSRC required the PE multiple used in IPO offer pricing to be no more than 25% higher than the average PE multiple for the industry peers after April 2012. Because the PE multiple caps mandated by the CSRC in the above two periods were much higher than before, it is difficult to say whether they should belong to the market-based approach or the government-based approach. 首先，中国的 IPO 市场经历了两次较长时间的 IPO 暂停（即 2005 年中-2006 年中和 2008 年 10 月-2009 年 5 月），更重要的是，2004 年后的政府主导模式与市场主导模式之间的区别并不像第 1-3 阶段那样明显。例如，在 2006 年中到 2008 年 9 月期间，IPO 发行价格由市场决定，但中国证监会对发行价格确定中使用的市盈率倍数设定了 30 的上限，显著高于第 3 阶段的 20 的上限。在 2009 年 6 月至 2012 年 4 月期间采用市场主导的 IPO 发行定价方法后，中国证监会要求 2012 年 4 月后 IPO 发行定价中使用的市盈率倍数不得超过行业同行平均市盈率倍数的 25%。由于中国证监会在上述两个时期规定的市盈率倍数上限远高于之前，因此很难判断它们应归属于市场主导模式还是政府主导模式。
Second, mainland China’s stock market experienced a significant regime shift in 2005 referred to as the split-share structure reform. Prior to this reform, mainland listed Chinese firms had two classes of common shares referred to as tradable shares and non-tradable shares, respectively. The common shares held by the original shareholders were not tradable on the stock exchanges after the IPO while the new common shares issued to public investors (referred to as A shares) were freely tradable on the domestic stock exchanges. The split-share structure reform made both classes of common shares publicly tradable. Chen 其次，中国大陆的股市在 2005 年经历了一次重大的制度变革，称为股权分置改革。在此改革之前，大陆上市的中国公司有两类普通股，分别称为可交易股和不可交易股。原股东持有的普通股在首次公开募股（IPO）后无法在证券交易所交易，而新发行给公众投资者的普通股（称为 A 股）则可以在国内证券交易所自由交易。股权分置改革使两类普通股均可公开交易。
Table 1 表 1
Sample selection procedures. 样本选择程序。
Year 年
1997
1998
1999
2000
2001
2002
2003
2004
Total 总计
Initial sample 初始样本
188
102
91
135
64
70
66
98
Minus: 负：
quad\quad Firms with B shares outstanding quad\quad 发行 B 股的公司
et al. (2012a) and Liao et al. (2014) find that the split-share structure reform helps better align the incentives between controlling shareholders and minority shareholders. et al. (2012a) 和 Liao et al. (2014) 发现，股权分置改革有助于更好地协调控股股东与小股东之间的激励。
Because of the significant regime change associated with the split-share structure reform and the reduced distinction between the government-based approach and market-based approach in the post-2004 period, it could be more difficult to design a clean experiment to detect the predicted effects resulting from IPO offer pricing regime changes post 2004. However, we believe that the fundamental institutional forces considered in this study are equally relevant today as evidenced by the CSRC’s intervention in the IPO offer pricing in 2014 (Hu, 2014) and the Chinese Government’s massive intervention in the stock market in response to the stock market rout in the summer of 2015 (Wei, 2015). 由于与股权分置改革相关的重大体制变化，以及 2004 年后政府主导与市场主导方法之间的区别减少，设计一个干净的实验以检测 2004 年后首次公开募股（IPO）定价制度变化所带来的预期效果可能会更加困难。然而，我们相信，本研究中考虑的基本制度力量在今天同样具有相关性，正如 2014 年中国证监会对 IPO 定价的干预（胡，2014）以及中国政府在 2015 年夏季股市暴跌后对股市的大规模干预（魏，2015）所证明的。
Our sample includes all IPOs on the two mainland stock exchanges during the period from January 1, 1997 to December 31, 2004. Table 1 shows the distribution of our sample IPOs by calendar year. We obtained the list of IPOs from the CSMAR database. There are a total of 814 IPOs listed on the mainland stock exchanges in the form of A shares during our sample period. Twenty four of the 814 IPOs issued either B shares (which are listed on mainland exchanges but could be owned only by foreigners prior to February 19, 2001) or H shares (which are listed in Hong Kong) prior to the public listing of the A shares on the mainland stock exchanges. Because these 24 IPOs are not genuine IPOs, they are excluded from our final sample. We also exclude five IPO firms in the financial industry which is heavily regulated. Finally, we exclude one IPO firm that was listed on the mainland through backdoor listing. ^(2){ }^{2} The final sample contains 784 IPOs. 我们的样本包括 1997 年 1 月 1 日至 2004 年 12 月 31 日期间在两个大陆证券交易所上市的所有首次公开募股（IPO）。表 1 显示了我们样本 IPO 按日历年的分布。我们从 CSMAR 数据库获得了 IPO 的列表。在我们的样本期间，大陆证券交易所共计有 814 个以 A 股形式上市的 IPO。在 814 个 IPO 中，有 24 个在 A 股公开上市之前发行了 B 股（在大陆交易所上市，但在 2001 年 2 月 19 日之前只能由外国人持有）或 H 股（在香港上市）。由于这 24 个 IPO 不是真正的 IPO，因此它们被排除在我们的最终样本之外。我们还排除了五家受到严格监管的金融行业的 IPO 公司。最后，我们排除了一家通过借壳上市在大陆上市的 IPO 公司。最终样本包含 784 个 IPO。
We obtain most of the required data from the CSMAR database. For the data that are either missing or unavailable in CSMAR (e.g., audit firms’ names, the headquarters of the audit firms and their audit clients), we hand collected them from the firms’ IPO prospectuses and annual reports. 我们从 CSMAR 数据库获取大部分所需数据。对于在 CSMAR 中缺失或不可用的数据（例如，审计公司的名称、审计公司的总部及其审计客户），我们从公司的 IPO 招股说明书和年度报告中手动收集。
3. IPO offer pricing under the government-based approach versus market-based approach 3. 政府主导方法与市场主导方法下的 IPO 定价
3.1. Institutional background 3.1. 机构背景
We first discuss how IPO offer prices are determined under the government-based approach versus the market-based approach during our sample period. During Regime 1 IPO offer prices were largely determined by the CSRC. Specifically, IPO offer prices were determined as the product of an EPS and a relatively fixed PE multiple over the range of 12-15.^(3)12-15 .^{3} Although the PE multiple was relatively fixed throughout this sub-period, the definition of EPS used in the offer price formula varied over time. Over the period January 1, 1997-March 17, 1998, the EPS was the average EPS over the three years prior to the IPO year. On March 17, 1998 the CSRC issued a new regulation CSRC (1998) that required the definition of EPS to be the forecasted EPS in the IPO year. 我们首先讨论在我们的样本期间，政府主导的方法与市场主导的方法如何确定首次公开募股（IPO）发行价格。在第一阶段，IPO 发行价格主要由中国证券监督管理委员会（CSRC）决定。具体而言，IPO 发行价格是每股收益（EPS）与相对固定的市盈率（PE）倍数的乘积，范围在 12-15.^(3)12-15 .^{3} 之内。尽管在这一子期间市盈率倍数相对固定，但用于发行价格公式的每股收益定义随着时间而变化。在 1997 年 1 月 1 日至 1998 年 3 月 17 日期间，每股收益是 IPO 年份前三年的平均每股收益。1998 年 3 月 17 日，CSRC 发布了新的规定（CSRC 1998），要求每股收益的定义为 IPO 年份的预测每股收益。
Regime 2 started on February 12, 1999 because in December 1998 China passed the nation’s first comprehensive Securities Law that signaled a significant shift of the IPO offer price determination process from a government-based approach to a mar-ket-based approach. The new Securities Law stipulates that an IPO’s offer price be negotiated between the issuer and the underwriter and then authorized by the CSRC (Article 28 of the 1999 Securities Law). Based on the spirit of the Securities Law, on February 12, 1999, the CSRC issued a new regulation (CSRC, 1999), which requires the IPO offer price to be determined based on the IPO firm’s growth potential and industry and market valuation rather than a regulator-prescribed formula with a fixed PE multiple. The new regulation requires that an IPO company submit a report that should include a detailed analysis of the prospect of the industry it is affiliated with, its current status and growth potential, and stock valuation in the secondary and primary markets, and the offer price determined based on the above analysis. ^(4){ }^{4} We confirmed with an anonymous source from the CSRC that the CSRC didn’t intervene in the IPO offer price determination during Regime 2. regime 2 于 1999 年 2 月 12 日开始，因为在 1998 年 12 月，中国通过了国家首部综合证券法，标志着首次公开募股（IPO）发行价格确定过程从政府主导转向市场主导。新的证券法规定，IPO 的发行价格由发行人和承销商协商确定，并由中国证券监督管理委员会（CSRC）授权（1999 年证券法第 28 条）。根据证券法的精神，1999 年 2 月 12 日，CSRC 发布了一项新规（CSRC，1999），要求 IPO 发行价格应基于 IPO 公司的成长潜力以及行业和市场估值来确定，而不是由监管机构规定的固定市盈率公式。新规要求 IPO 公司提交一份报告，报告应包括其所属行业的前景、当前状况和成长潜力的详细分析，以及在二级市场和一级市场的股票估值，并根据上述分析确定发行价格。 我们从中国证监会的一位匿名消息源确认，在第二阶段期间，中国证监会没有干预首次公开募股的发行价格确定。
Regime 3 represents a reversal from the market-based approach to the government-based approach. The starting date of Regime 3 is not very clear because the CSRC never issued any explicit regulation stipulating the starting date of Regime 3 . One regime 3 代表了从市场导向转向政府导向的逆转。 regime 3 的起始日期并不明确，因为中国证监会从未发布任何明确规定 regime 3 起始日期的法规。
Table 2 表 2
The PE multiples used in IPO offer prices. IPO 发行价格中使用的市盈率倍数。
Panel A. The mean (median) [standard deviation] 面板 A。均值（中位数）[标准差]
Regimes 政权
The PE multiples of IPO firms IPO 公司的市盈率倍数
The PE multiples of IPO firms as a fraction of the median PE multiple of non-IPO firms in the same industry IPO 公司的市盈率倍数与同一行业非 IPO 公司的中位数市盈率倍数的比率
Panel A. The mean (median) [standard deviation]
Regimes The PE multiples of IPO firms The PE multiples of IPO firms as a fraction of the median PE multiple of non-IPO firms in the same industry
1 "14.791
(14.600)
[1.313]" "0.358
(0.355)
[0.066]"
2 "28.926
(28.050)
[11.127]" "0.531
(0.506)"
3 "18.293
(19.735)
[2.595]" "0.423
(0.403)
[0.107]"
The two-tailed p-value from a two-sample t-test (Wilcoxon rank sum test)
Regime 1 vs. Regime 2: " < 0.0001
( < 0.0001)" " < 0.0001
( < 0.0001)"
Regime 3 vs. Regime 2: " < 0.0001
( < 0.0001)" " < 0.0001
( < 0.0001)"| Panel A. The mean (median) [standard deviation] | | |
| :---: | :---: | :---: |
| Regimes | The PE multiples of IPO firms | The PE multiples of IPO firms as a fraction of the median PE multiple of non-IPO firms in the same industry |
| 1 | $\begin{aligned} & 14.791 \\ & (14.600) \\ & {[1.313]} \end{aligned}$ | $\begin{aligned} & 0.358 \\ & (0.355) \\ & {[0.066]} \end{aligned}$ |
| 2 | $\begin{aligned} & 28.926 \\ & (28.050) \\ & {[11.127]} \end{aligned}$ | 0.531 <br> (0.506) |
| 3 | $\begin{aligned} & 18.293 \\ & (19.735) \\ & {[2.595]} \end{aligned}$ | $\begin{aligned} & 0.423 \\ & (0.403) \\ & {[0.107]} \end{aligned}$ |
| The two-tailed $p$-value from a two-sample t-test (Wilcoxon rank sum test) | | |
| Regime 1 vs. Regime 2: | $\begin{aligned} & <0.0001 \\ & (<0.0001) \end{aligned}$ | $\begin{aligned} & <0.0001 \\ & (<0.0001) \end{aligned}$ |
| Regime 3 vs. Regime 2: | $\begin{aligned} & <0.0001 \\ & (<0.0001) \end{aligned}$ | $\begin{aligned} & <0.0001 \\ & (<0.0001) \end{aligned}$ |
Panel B. The Pearson correlation coefficient between an IPO firm’s PE multiple and the median PE multiple of non-IPO firms in the same industry 面板 B。IPO 公司的市盈率倍数与同一行业非 IPO 公司的中位市盈率倍数之间的皮尔逊相关系数
The two-tailed pp-value from a test of difference in the correlation coefficients ^(a){ }^{\mathrm{a}} 双尾 pp -值来自相关系数 ^(a){ }^{\mathrm{a}} 的差异检验
0.296
Regime 1 vs. Regime 2: 制度 1 与制度 2：
( < 0.001)(<0.001)
Regime 3 vs. Regime 2: 政权 3 与政权 2：
< 0.0001<0.0001
Regimes "Correlation coefficient
(two-tailed p-value)"
1 -0.155
2 (0.009)
3 0.540
( < 0.001)
The two-tailed p-value from a test of difference in the correlation coefficients ^(a) 0.296
Regime 1 vs. Regime 2: ( < 0.001)
Regime 3 vs. Regime 2: < 0.0001| Regimes | Correlation coefficient <br> (two-tailed $p$-value) |
| :--- | :--- |
| 1 | -0.155 |
| 2 | $(0.009)$ |
| 3 | 0.540 |
| | $(<0.001)$ |
| The two-tailed $p$-value from a test of difference in the correlation coefficients ${ }^{\mathrm{a}}$ | 0.296 |
| Regime 1 vs. Regime 2: | $(<0.001)$ |
| Regime 3 vs. Regime 2: | $<0.0001$ |
The median PE multiple of non-IPO firms in the same two-digit industry code is computed at the beginning of the issuance month of the corresponding IPO firm. When computing the median PE multiple, we exclude non-IPO firms with negative earnings or PE multiples larger than 300 . All the continuous variables are winsorized at the 1 st and 99th percentiles. 在对应 IPO 公司的发行月份开始时，计算同一两位数行业代码下非 IPO 公司的中位数市盈率倍数。在计算中位数市盈率倍数时，我们排除了盈利为负或市盈率倍数大于 300 的非 IPO 公司。所有连续变量在第 1 和第 99 百分位数处进行了温莎化处理。 ^(a){ }^{a} The difference in the correlation coefficients between two independent samples can be tested using the following ZZ-statistic: Z=(z_(1)-z_(2))//sigma(z_(1)-z_(2)).z_(1)Z=\left(z_{1}-z_{2}\right) / \sigma\left(z_{1}-z_{2}\right) . z_{1} and z_(2)z_{2} are the Fisher’s zz transformation of the sample correlations r_(1)r_{1} and r_(2)r_{2}, respectively and sigma(z_(1)-z_(2))\sigma\left(z_{1}-z_{2}\right) is the standard deviation of ( z_(1)-z_(2)z_{1}-z_{2} ) and is equal to sqrt(1//(N_(1)-3)+1//(N_(2)-3))\sqrt{1 /\left(N_{1}-3\right)+1 /\left(N_{2}-3\right)}, where N_(1)N_{1} and N_(2)N_{2} are the sample sizes of the two samples, respectively. ^(a){ }^{a} 两个独立样本之间相关系数的差异可以使用以下 ZZ -统计量进行检验： Z=(z_(1)-z_(2))//sigma(z_(1)-z_(2)).z_(1)Z=\left(z_{1}-z_{2}\right) / \sigma\left(z_{1}-z_{2}\right) . z_{1} 和 z_(2)z_{2} 是样本相关性 r_(1)r_{1} 和 r_(2)r_{2} 的费舍尔 zz 变换， sigma(z_(1)-z_(2))\sigma\left(z_{1}-z_{2}\right) 是 ( z_(1)-z_(2)z_{1}-z_{2} ) 的标准差，等于 sqrt(1//(N_(1)-3)+1//(N_(2)-3))\sqrt{1 /\left(N_{1}-3\right)+1 /\left(N_{2}-3\right)} ，其中 N_(1)N_{1} 和 N_(2)N_{2} 分别是两个样本的样本大小。
obvious reason for not issuing a formal regulation is that such a regulation would be in direct conflict with the 1999 Securities Law, which grants IPO companies the right to set IPO offer prices based on their own, industry, and market conditions. We select November 7, 2001 as the starting date of Regime 3 because the CSRC suspended all IPO activities over the period from August 7, 2001 to November 6, 2001 in response to the bearish stock market condition. More importantly, consistent with the governmentbased approach, starting from November 7, 2001, IPO offer prices were determined based on the EPS in the year prior to the IPO year using a PE multiple capped at 20 (a very low threshold as seen in Table 2 reported later) and fluctuated in a narrow band. ^(5){ }^{5} 不发布正式规定的明显原因是，这样的规定将与 1999 年证券法直接冲突，该法赋予首次公开募股（IPO）公司根据自身、行业和市场条件设定 IPO 发行价格的权利。我们选择 2001 年 11 月 7 日作为第三阶段的起始日期，因为中国证监会在 2001 年 8 月 7 日至 2001 年 11 月 6 日期间暂停了所有 IPO 活动，以应对熊市的市场状况。更重要的是，符合政府主导的做法，从 2001 年 11 月 7 日起，IPO 发行价格是根据 IPO 年之前一年的每股收益（EPS）确定的，使用的市盈率倍数上限为 20（如后面表 2 所示，这是一个非常低的门槛），并在一个狭窄的区间内波动。
3.2. Why does the CSRC prefer the government-based IPO offer pricing approach? 3.2. 为什么中国证监会更倾向于政府主导的 IPO 定价方法？
For readers unfamiliar with the history of China’s economic development over the past three decades, it may come as a surprise that China’s IPO regulatory regime changes so frequently. We believe that such flip-flops are usually due to ideological differences among China’s different political factions and the self-interest of government agencies (Li, 2013). More specifically, the CSRC’s adoption of the government-based approach in Regimes 1 and 3 can be explained by the CSRC’s lack of experience in regulating financial markets and asymmetric loss function (Peltzman, 1976). Specifically, China’s modern stock markets started only in early 1990s, after more than four decades of socialist planning economy. Hence, almost no government officials had adequate experience in regulating financial markets. More importantly, the CSRC is averse to the risk of IPO share overpricing because China’s institutional investor clientele is underdeveloped and a significant portion of IPO subscribers are small retail investors. Hence, there is a perceived risk that significant losses from overpriced IPOs could lead to retail investors’ loud complaints and even protests, which could cost government officials’ jobs and promotion prospect. On the other hand, retail investors would be less likely to complain about underpriced IPOs. Therefore, the CSRC has a natural incentive to depress IPO offer prices. 对于不熟悉中国过去三十年经济发展历史的读者来说，中国的 IPO 监管制度变化如此频繁可能会令人感到惊讶。我们认为，这种反复无常通常是由于中国不同政治派别之间的意识形态差异以及政府机构的自身利益所致（李，2013）。更具体地说，证监会在第一和第三阶段采用政府主导的方法可以解释为证监会在监管金融市场方面缺乏经验以及不对称损失函数的影响（佩尔茨曼，1976）。具体而言，中国的现代股票市场仅在 1990 年代初期开始，经过四十多年的社会主义计划经济。因此，几乎没有政府官员在监管金融市场方面有足够的经验。更重要的是，证监会对 IPO 股票定价过高的风险持谨慎态度，因为中国的机构投资者客户群尚不成熟，且相当一部分 IPO 认购者是小型散户投资者。 因此，人们认为，来自高价首次公开募股（IPO）的重大损失可能导致散户投资者的强烈投诉甚至抗议，这可能会影响政府官员的职位和晋升前景。另一方面，散户投资者对低价 IPO 的投诉可能较少。因此，中国证监会自然有动力压低 IPO 发行价格。
Fig. 1. PE multiples used in IPO offer prices. This figure plots the price-to-reported earnings (PE) multiples used in the IPO offer prices during the sample period. The PE multiples are directly obtained from IPO firms’ prospectuses. 图 1. IPO 发行价格中使用的市盈率倍数。该图绘制了样本期间 IPO 发行价格中使用的市盈率（PE）倍数。市盈率倍数直接来自 IPO 公司的招股说明书。
Therefore, we expect the CSRC to have a strong incentive to depress the PE multiples used in the IPO offerings under the govern-ment-based approach. 因此，我们预计中国证监会将有强烈的动机来压低政府主导方式下首次公开募股所使用的市盈率倍数。
This prediction is born out in our data. As shown in Fig. 1 and Panel A of Table 2, both the mean and median of the IPO firms’ PE multiples used in the offer price formulas are significantly lower in Regimes 1 and 3 than in Regime 2 . The variance of the IPO firms’ PE multiples is also significantly lower in Regimes 1 and 3 than in Regime 2 (two-tailed pp-values < 0.0001<0.0001 ). In addition, IPO firms’ PE multiples derived from the IPO offer prices deviate significantly more from the PE multiples of non-IPO firms in the same two-digit industry code during Regimes 1 and 3 than during Regime 2. As shown in Panel A of Table 2, the median IPO firm’s PE multiple as a fraction of the median PE multiple of all non-IPO firms in the same industry is 0.355 in Regime 1, 0.506 in Regime 2, and 0.403 in Regime 3. While the median ratios are all below one, the median ratio is significantly higher in Regime 2 than in Regimes 1 and 3. Furthermore, as shown in Panel B of Table 2, the Pearson correlation between an IPO firm’s PE multiple and the median PE multiple of non-IPO firms in the same industry is significantly more positive during Regime 2 than during Regimes 1 and 3. Overall, the evidence in Table 2 and Fig. 1 strongly suggests that the CSRC severely depressed IPO offer prices during Regimes 1 and 3 . 这一预测在我们的数据中得到了验证。如图 1 和表 2 的 A 面所示，首次公开募股（IPO）公司的市盈率倍数在定价公式中，第一和第三阶段的均值和中位数显著低于第二阶段。IPO 公司的市盈率倍数的方差在第一和第三阶段也显著低于第二阶段（双尾 pp -值 < 0.0001<0.0001 ）。此外，IPO 公司从 IPO 定价中得出的市盈率倍数在第一和第三阶段与同一两位数行业代码的非 IPO 公司市盈率倍数的偏差显著大于第二阶段。如表 2 的 A 面所示，IPO 公司的中位市盈率倍数占同一行业所有非 IPO 公司中位市盈率倍数的比例在第一阶段为 0.355，在第二阶段为 0.506，在第三阶段为 0.403。虽然中位数比例均低于 1，但第二阶段的中位数比例显著高于第一和第三阶段。此外，如表 2 的 B 面所示，IPO 公司市盈率倍数与同一行业非 IPO 公司中位市盈率倍数之间的皮尔逊相关性在第二阶段显著高于第一和第三阶段。 总体而言，表 2 和图 1 中的证据强烈表明，在第 1 和第 3 阶段，CSRC 严重压低了 IPO 发行价格。
Given the CSRC’s strong preference for the government-based approach (i.e., using low and fixed PE multiple in IPO offer prices), why did we observe a switch to the market-based approach in Regime 2? As noted above, this switch was forced by China’s passage of the first comprehensive Securities Law in 1998 that mandated the adoption of a market-based approach in IPO offer pricing. Then, one may wonder why we still observe a reversal from the market-based approach in Regime 2 to the gov-ernment-based approach in Regime 3. We believe this is again due to the CSRC’s asymmetric loss function mentioned above. Specifically, the Shanghai stock market index lost 25%25 \% of its value from June 12, 2001, the date when the Chinese Government announced its intent to sell down state-owned non-tradable shares of mainland listed firms, to November 6, 2001, the ending date of Regime 2. Because of such strong negative market-wide sentiment, the Chinese Government cancelled its plan to sell down state-owned non-tradable shares on June 24, 2002. The CSRC must also have heightened its perceived risk of IPO blowups during the second half of 2001 and therefore resorted to administrative means to not only stop the flow of new IPOs for three months from August 7, 2001 to November 6, 2001 but also cap the maximum PE ratio at 20 for IPOs launched in Regime 3, a blatant violation of the Securities Law. ^(6){ }^{6} 鉴于中国证监会对政府主导方法（即在首次公开募股（IPO）报价中使用低且固定的市盈率倍数）的强烈偏好，为什么我们在第二阶段观察到了向市场主导方法的转变？如上所述，这一转变是由于中国在 1998 年通过了第一部综合证券法，该法要求在 IPO 报价中采用市场主导方法。那么，人们可能会想，为什么我们在第二阶段的市场主导方法又回到了第三阶段的政府主导方法。我们认为这再次是由于中国证监会上述提到的非对称损失函数。具体来说，从 2001 年 6 月 12 日中国政府宣布计划出售大陆上市公司国有非流通股到 2001 年 11 月 6 日第二阶段结束，上海股市指数损失了 25%25 \% 的价值。由于这种强烈的市场负面情绪，中国政府在 2002 年 6 月 24 日取消了出售国有非流通股的计划。 中国证监会在 2001 年下半年必须也提高了对首次公开募股（IPO）失败的感知风险，因此采取了行政手段，不仅在 2001 年 8 月 7 日至 2001 年 11 月 6 日之间停止了新 IPO 的流入，还将第三阶段 IPO 的最高市盈率限制在 20，这明显违反了证券法。 ^(6){ }^{6}
4. Financial reporting quality at the time of IPO 4. IPO 时的财务报告质量
4.1. Hypothesis development 4.1. 假设发展
In this section we analyze how variations in the IPO pricing mechanisms across Regimes 1-3 affect IPO firms’ financial reporting quality in the pre-IPO years. During Regime 1, we expect IPO firms’ insiders (i.e., founding shareholders and their appointed managers) to have a strong incentive to inflate reported earnings in the pre-IPO period for two reasons. First, as noted in Section 3, regardless of an IPO firm’s growth prospect, the PE multiple imbedded in the offer price prescribed by the CSRC was always fixed in the narrow range of 12-1512-15, which is much lower than the median PE multiple commanded by nonIPO firms in the same industry. In fact, as noted in Table 2, the IPO firms’ PE multiples in Regime 1 are negatively correlated with the median PE multiple of non-IPO firms in the same industry. 在本节中，我们分析了第 1-3 阶段中 IPO 定价机制的变化如何影响 IPO 公司在 IPO 前几年的财务报告质量。在第 1 阶段，我们预计 IPO 公司的内部人士（即创始股东及其任命的管理者）有强烈的动机在 IPO 前期夸大报告的收益，原因有二。首先，如第 3 节所述，无论 IPO 公司的增长前景如何，证监会规定的发行价格中嵌入的市盈率倍数始终固定在 12-1512-15 的狭窄范围内，这远低于同一行业非 IPO 公司所拥有的中位市盈率倍数。实际上，如表 2 所示，第 1 阶段 IPO 公司的市盈率倍数与同一行业非 IPO 公司的中位市盈率倍数呈负相关。
Second, the determination of IPO offer prices was directly based on earnings numbers. During the first half of regime 1 over 1/1/1997-3/17/1998 (denoted Regime 1a), the EPS used in the PE multiple was the average reported earnings per share in the three years prior to the IPO. During the second half of regime 1 over 3/18/1998-2/11/1999 (denoted Regime 1b), the EPS used in the PE multiple was the forecasted earnings per share in the IPO year. 其次，IPO 发行价格的确定直接基于盈利数据。在 1997 年 1 月 1 日至 1998 年 3 月 17 日的第一阶段（称为阶段 1a）中，市盈率中使用的每股收益（EPS）是 IPO 前三年报告的每股平均收益。在 1998 年 3 月 18 日至 1999 年 2 月 11 日的第一阶段后半段（称为阶段 1b）中，市盈率中使用的每股收益是 IPO 年度的预测每股收益。
Because the reported earnings in the pre-IPO period were directly used in the IPO pricing formula during Regime 1a, IPO firms’ insiders should have a clear incentive to inflate earnings in the pre-IPO period in order to increase the IPO proceeds and avoid unwarranted undervaluation of their firms’ IPO shares. ^(7){ }^{7} During Regime 1b, we also expect IPO firms’ insiders to have a strong incentive to inflate their earnings forecast for the IPO year because earnings forecasts were directly used to determine IPO offer prices. However, the insiders’ incentive to inflate earnings in the pre-IPO period is less clear because the reported earnings in the pre-IPO period were not directly used in the IPO pricing formula. We argue that during Regime 1b IPO firms’ insiders should also have an incentive to inflate reported earnings in the pre-IPO period. The reason is that if the reported earnings in the pre-IPO period were substantially below the forecasted earnings in the IPO year, IPO firms’ insiders would find it more difficult to justify to the CSRC a high earnings forecast in the IPO year. 由于在首次公开募股（IPO）前期报告的收益直接用于 Regime 1a 中的 IPO 定价公式，IPO 公司的内部人士应该有明确的激励在 IPO 前期夸大收益，以增加 IPO 收益并避免对其公司 IPO 股票的不当低估。 ^(7){ }^{7} 在 Regime 1b 期间，我们也预计 IPO 公司的内部人士会有强烈的激励来夸大 IPO 年度的收益预测，因为收益预测直接用于确定 IPO 发行价格。然而，内部人士在 IPO 前期夸大收益的激励不太明确，因为在 IPO 前期报告的收益并未直接用于 IPO 定价公式。我们认为，在 Regime 1b 期间，IPO 公司的内部人士也应该有激励在 IPO 前期夸大报告的收益。原因是，如果在 IPO 前期报告的收益远低于 IPO 年度的预测收益，IPO 公司的内部人士将更难向中国证券监督管理委员会（CSRC）证明 IPO 年度的高收益预测。
The CSRC could be aware of IPO firms’ earnings management incentives and therefore could have taken actions to limit IPO firms’ upward earnings management. However, this may not happen for two reasons. First, earnings management can take many different forms and due to the information asymmetry between an IPO firm and the regulator, the CSRC staff may not be able to easily detect earnings inflation. Second, even if the CSRC staff had the ability to detect earnings inflation, they may not have the incentive to correct such earnings management because China is a weak investor protection country and therefore the CSRC could be captured by IPO firms’ insiders. Ultimately, it is an empirical question whether the CSRC can fully constrain IPO firms’ upward earnings management in Regime 1. 中国证监会可能意识到 IPO 公司的盈余管理动机，因此可能采取措施限制 IPO 公司的盈余上调管理。然而，这可能不会发生，原因有二。首先，盈余管理可以采取多种不同形式，由于 IPO 公司与监管机构之间的信息不对称，中国证监会的工作人员可能无法轻易发现盈余膨胀。其次，即使中国证监会的工作人员有能力发现盈余膨胀，他们也可能没有纠正这种盈余管理的动机，因为中国是一个投资者保护较弱的国家，因此中国证监会可能会受到 IPO 公司内部人士的影响。最终，是否中国证监会能够完全约束 IPO 公司的盈余上调管理是一个实证问题。
Relative to Regime 1, we expect IPO firms’ insiders to have a weaker incentive to inflate reported earnings in the pre-IPO period during Regime 2. During Regime 2, offer prices were determined by market forces rather than by a mechanical formula prescribed by the CSRC and therefore were less likely to be undervalued. We expect the market to consider multiple factors before deriving an IPO firm’s final offer price. While important, reported earnings were not the only factor considered by the market. In addition, any suspicion of earnings manipulation could be viewed as a negative signal about management’s integrity and the firm’s corporate governance and therefore would be punished by the market by discounting the firm’s offer price. Due to investors’ ability to price protect, efficient contracting theory (Jensen and Meckling, 1976) would predict that IPO firms’ insiders have a strong incentive to bond to high-quality earnings by investing in monitoring technologies such as hiring a high quality auditor and avoiding aggressive earnings management. 相对于制度 1，我们预计在制度 2 期间，IPO 公司的内部人士在 IPO 前期对虚增报告收益的动机会较弱。在制度 2 期间，发行价格是由市场力量决定的，而不是由证监会规定的机械公式，因此不太可能被低估。我们预计市场在确定 IPO 公司的最终发行价格时会考虑多个因素。虽然报告收益很重要，但并不是市场考虑的唯一因素。此外，任何对收益操纵的怀疑都可能被视为对管理层诚信和公司治理的负面信号，因此市场会通过降低公司的发行价格来惩罚这种行为。由于投资者具备价格保护的能力，效率契约理论（詹森和梅克林，1976）预测，IPO 公司的内部人士有强烈的动机通过投资监控技术（如聘请高质量审计师和避免激进的收益管理）来绑定高质量收益。
Of course, one may argue that the market mechanism described above may not work well in a weak investor protection country like China. For example, bonding may be just too costly due to China’s weak law and enforcement (Doidge et al., 2007). Alternatively, Chinese stock market investors may not be very sophisticated and are fixated on reported earnings in IPO offer price determination. As a result, due to weak regulatory enforcement against corporate wrongdoings, IPO firms’ insiders may find it more beneficial to inflate reported earnings in order to send to the market a false positive signal about the firm’s future prospect. Overall, it is an empirical question whether the degree of upward earnings management is smaller in Regime 2 than in Regime 1. 当然，有人可能会争辩说，上述市场机制在像中国这样投资者保护较弱的国家可能效果不佳。例如，由于中国法律和执法的薄弱，担保可能成本过高（Doidge 等，2007）。另外，中国股市投资者可能不够成熟，过于关注首次公开募股（IPO）定价中的报告收益。因此，由于对企业不当行为的监管执法薄弱，IPO 公司的内部人士可能会发现夸大报告收益更有利，以向市场传递关于公司未来前景的虚假积极信号。总体而言，收益管理的上升程度在制度 2 中是否小于制度 1 是一个实证问题。
As explained in Section 2, the IPO pricing mechanism in Regime 3 is more similar to that in Regime 1 than that in Regime 2. Specifically, during Regime 3, the PE multiple was capped at 20 and the earnings per share used in the PE multiple were the reported earnings per share in the year prior to the IPO. While the PE multiple allowed by the CSRC is slightly higher in Regime 3 than in Regime 1, as noted in Section 3, it is still lower than the median PE multiple of non-IPO firms in the same industry. In addition, the correlation between an IPO firm’s PE multiple and the median PE multiple of non-IPO firms in the same industry is significantly less positive in Regime 3 than in Regime 2. Hence, we expect IPO firms’ insiders to have a stronger incentive in Regime 3 than in Regime 2 to inflate reported earnings at least in the year prior to the IPO year. 如第 2 节所述，第三阶段的 IPO 定价机制与第一阶段更为相似，而不是与第二阶段。具体而言，在第三阶段，市盈率倍数被限制在 20，市盈率中使用的每股收益是 IPO 前一年报告的每股收益。虽然第三阶段中国证监会允许的市盈率倍数略高于第一阶段，但如第 3 节所述，仍低于同一行业非 IPO 公司的中位市盈率倍数。此外，IPO 公司的市盈率倍数与同一行业非 IPO 公司的中位市盈率倍数之间的相关性在第三阶段显著低于第二阶段。因此，我们预计在第三阶段，IPO 公司的内部人士比在第二阶段有更强的激励在 IPO 前一年至少夸大报告的收益。
One countervailing force that may limit upward earnings management in Regime 3 is the improving overall investor protection in China over time (see Section 4.3 for a detailed discussion). Therefore, it is possible the degree of earnings inflation is smaller in Regime 3 than in Regime 1 due to the strengthening of other monitoring mechanisms over time. However, if we find an increase in upward earnings management from Regime 2 to Regime 3, the increase is more likely due to the IPO regime change rather than confounding factors. 在第三阶段，可能限制向上盈余管理的一个对抗力量是中国整体投资者保护的逐步改善（详见第 4.3 节）。因此，由于其他监控机制的加强，第三阶段的盈余膨胀程度可能小于第一阶段。然而，如果我们发现从第二阶段到第三阶段向上盈余管理的增加，这种增加更可能是由于 IPO 制度的变化，而不是混杂因素。
To test how IPO firms’ financial reporting quality changes across the three regimes, we consider two complementary dimensions of an IPO firm’s financial reporting environment. First, we consider IPO firms’ auditor selection. Second, we consider the extent of upward earnings management in the year prior to the IPO (denoted as year t-1t-1 ). ^(8){ }^{8} To the extent that IPO firms’ insiders have a stronger (weaker) incentive for high quality financial reporting in Regime 2 than in Regimes 1 and 3, we should expect the insiders to be more likely (less likely) to select a high-quality auditor, and less likely (more likely) to inflate earnings in year t-1t-1. We state the above predictions in the form of null hypotheses as follows ^(9){ }^{9} : 为了测试 IPO 公司在三个制度下财务报告质量的变化，我们考虑 IPO 公司财务报告环境的两个互补维度。首先，我们考虑 IPO 公司的审计师选择。其次，我们考虑 IPO 前一年（记作年份 t-1t-1 ）的向上盈余管理程度。 ^(8){ }^{8} 如果 IPO 公司的内部人士在制度 2 中对高质量财务报告的激励比在制度 1 和 3 中更强（更弱），我们应该预期内部人士更有可能（不太可能）选择高质量审计师，并且在年份 t-1t-1 中不太可能（更可能）夸大盈余。我们将上述预测以零假设的形式表述如下 ^(9){ }^{9} ：
H1. Managers of IPO firms in Regimes 1, 2 and 3 are equally likely to select a high-quality auditor. H1. 在第 1、2 和 3 个制度下，IPO 公司的管理者选择高质量审计师的可能性相同。
H2. Managers of IPO firms in Regimes 1, 2 and 3 are equally likely to inflate earnings in the year prior to the IPO. H2. 在第 1、2 和 3 类制度下，首次公开募股（IPO）公司的管理者在 IPO 前一年同样可能会虚增收益。
In addition to inflating reported earnings, IPO firms could have a variety of alternative responses to the CSRC’s rigid control over IPO offer prices during Regimes 1 and 3. For example, IPO firms could opt to stay private and seek alternative funding sources. IPO firms could also cut the size of the shares to be sold in the IPO followed by larger issuances of seasoned equity offerings (SEOs) at higher prices. Unfortunately, during our sample period these options were not readily available. Staying private was not an attractive option because alternative financing sources were severely limited. We find no evidence in Table 5 reported later that growth firms were more likely to stay private in Regimes 1 and 3. Both the IPO offer size and SEOs were heavily regulated, too. During our sample period the mean IPO offer size as a percentage of total shares outstanding after the IPO was 31.8%31.8 \% with a small standard deviation of 6.6%^(10)6.6 \%{ }^{10} Chinese listed firms also required the explicit approval of the CSRC for SEOs in our sample period. We find no evidence that the IPO offer size (the SEO offer size) is significantly smaller (larger) in Regimes 1 and 3 than in Regime 2 (untabulated). In summary, IPO firms have very few alternative responses other than earnings management to mitigate the rigid IPO offer pricing regulations during Regimes 1 and 3. 除了夸大报告的收益外，IPO 公司在第一和第三阶段对证监会对 IPO 发行价格的严格控制可能有多种替代应对措施。例如，IPO 公司可以选择保持私有并寻求替代融资来源。IPO 公司还可以减少 IPO 中出售的股份规模，然后以更高的价格进行更大规模的增发（SEO）。不幸的是，在我们的样本期间，这些选项并不容易获得。保持私有并不是一个有吸引力的选择，因为替代融资来源严重有限。我们在后面的表 5 中没有发现证据表明，成长型公司在第一和第三阶段更可能保持私有。IPO 发行规模和增发也受到严格监管。在我们的样本期间，IPO 发行规模占 IPO 后总流通股的平均比例为 31.8%31.8 \% ，标准差为 6.6%^(10)6.6 \%{ }^{10} 。在我们的样本期间，中国上市公司还需要证监会对增发的明确批准。我们没有发现证据表明，第一和第三阶段的 IPO 发行规模（增发规模）显著小于（大于）第二阶段（未列出）。 总之，IPO 公司在第 1 和第 3 阶段几乎没有其他替代方案来应对严格的 IPO 发行定价规定，除了收益管理。
4.2. Regression results 4.2. 回归结果
4.2.1. Test of H1 4.2.1. H1 测试
We use the following logistic regression to test H 1 : 我们使用以下逻辑回归来检验 H 1：
SL=beta_(0)+beta_(1)REG1+beta_(2)REG3+" Controls "+epsiS L=\beta_{0}+\beta_{1} R E G 1+\beta_{2} R E G 3+\text { Controls }+\varepsilon
See Appendix A for all variable definitions. The regression is estimated using the IPO firms during our sample period. SL is a dummy variable that measures the quality of an IPO firm’s hired auditor, with one for a low-quality auditor. The definition of SL follows Wang et al. (2008) who argue and find that small local auditors are less independent and hence provide lower quality audits than other Chinese auditors (see also Chan et al., 2006). 请参见附录 A 以获取所有变量定义。回归分析使用了我们样本期间的 IPO 公司。SL 是一个虚拟变量，用于衡量 IPO 公司聘请的审计师的质量，低质量审计师为 1。SL 的定义遵循 Wang 等人（2008）的观点，他们认为并发现小型地方审计师的独立性较差，因此提供的审计质量低于其他中国审计师（另见 Chan 等人，2006）。
The variables REG1 and REG3 are used to test H1. To the extent that the market-based approach (i.e., Regime 2) provides managers of IPO firms with a stronger incentive to supply high-quality financial reporting, the coefficients on REG1 and REG3 should be significantly positive. ^(11){ }^{11} 变量 REG1 和 REG3 用于测试 H1。在市场导向的方法（即，制度 2）在多大程度上为 IPO 公司的管理者提供了更强的激励以提供高质量的财务报告时，REG1 和 REG3 的系数应该显著为正。
Following prior research (Copley and Douthett, 2002; Willenborg, 1999; Weber and Willenborg, 2003; DeFond et al., 2000), we control for the following common determinants of auditor choices. We use CURRENT, LEV, ROA, and VOLATILITY to control for audit risk, and ASSETS and ARINV to control for audit complexity. To the extent that more risky clients hire high quality auditors as a bonding mechanism to reduce agency costs, coefficients on LEV and VOLATILITY should be negative while the coefficients on CURRENT and ROA should be positive. We expect the coefficients on ASSETS and ARINV to be negative since small local audit firms are less able to undertake large and/or complex audits. 根据之前的研究（Copley 和 Douthett, 2002；Willenborg, 1999；Weber 和 Willenborg, 2003；DeFond 等, 2000），我们控制以下审计选择的共同决定因素。我们使用 CURRENT、LEV、ROA 和 VOLATILITY 来控制审计风险，使用 ASSETS 和 ARINV 来控制审计复杂性。在风险较高的客户聘请高质量审计师作为减少代理成本的担保机制的情况下，LEV 和 VOLATILITY 的系数应为负，而 CURRENT 和 ROA 的系数应为正。我们预计 ASSETS 和 ARINV 的系数为负，因为小型地方审计公司较难进行大型和/或复杂的审计。
Growth firms have a greater need for external financing, and thus should have a stronger incentive to hire a high quality auditor (Wang et al., 2008). Hence, we also include GROWTH as a control and its coefficient is expected to be negative. While the IPO offer size is heavily regulated in China, we follow U.S. research by including two IPO characteristics, PROCEEDS and RETAINED_OWN, as additional control variables. Prior U.S. research (Dye, 1993; Willenborg, 1999) suggests that large auditors provide greater insurance coverage in the event of securities litigation than small auditors and therefore IPO firms with greater IPO proceeds should be more likely to appoint large auditors. Because securities litigation rarely occurs in China, we doubt that auditors play a similar insurance role in Chinese IPOs. Nevertheless, we still include PROCEEDS to be consistent with prior research. Prior U.S. research also argues that IPO firms’ insiders can use RETAINED_OWN to signal firm value. However, the coefficient on RETAINED_OWN is hard to predict because RETAINED_OWN could be either a complement to or a substitute for high-quality auditors (Titman and Trueman, 1986; Datar et al., 1991). 成长型公司对外部融资的需求更大，因此应该更有动力聘请高质量的审计师（王等，2008）。因此，我们也将 GROWTH 作为控制变量，其系数预计为负值。虽然在中国，首次公开募股（IPO）的发行规模受到严格监管，但我们仍然借鉴美国的研究，将两个 IPO 特征 PROCEEDS 和 RETAINED_OWN 作为额外的控制变量。之前的美国研究（Dye，1993；Willenborg，1999）表明，与小型审计师相比，大型审计师在证券诉讼发生时提供更大的保险保障，因此，IPO 公司如果获得更多的 IPO 收益，应该更有可能聘请大型审计师。由于在中国证券诉讼很少发生，我们怀疑审计师在中国 IPO 中是否发挥类似的保险作用。尽管如此，我们仍然包括 PROCEEDS 以与之前的研究保持一致。之前的美国研究还认为，IPO 公司的内部人士可以利用 RETAINED_OWN 来传递公司价值。然而，RETAINED_OWN 的系数难以预测，因为 RETAINED_OWN 可能是高质量审计师的补充或替代品（Titman 和 Trueman，1986；Datar 等，1991）。
Following Wang et al. (2008), we further control for SOE and MARKETIZATION. Wang et al. (2008) argue that Chinese SOEs tend to select small local auditors because the latter often have special knowledge about local SOEs but this tendency is moderated in regions with stronger institutional environments. Consistent with Wang et al. (2008), we measure the strength of institutions by MARKETIZATION, which is the marketization index for the province in which the IPO firm is located (higher index suggests stronger market institutions). We expect the coefficients on SOE and MARKETIZATION to be positive and negative, respectively. We include HHI because Chan and Wu (2011) show that the audit market structure changed significantly over time due to audit firm mergers. When the audit market is more concentrated in the hands of a few large audit firms (i.e., the value of HHI is higher), there is a lower supply of audit services provided by small audit firms. We therefore expect the coefficient on HHI to be negative. Finally, we include industry fixed effects. ^(12){ }^{12} 根据王等人（2008）的研究，我们进一步控制国有企业（SOE）和市场化（MARKETIZATION）。王等人（2008）认为，中国的国有企业倾向于选择小型地方审计师，因为后者通常对地方国有企业有特殊的了解，但这种倾向在制度环境较强的地区会有所减弱。与王等人（2008）的观点一致，我们通过市场化来衡量制度的强度，市场化是指 IPO 公司所在省份的市场化指数（指数越高，市场制度越强）。我们预计国有企业和市场化的系数分别为正和负。我们还包括了 HHI，因为 Chan 和 Wu（2011）表明，由于审计公司合并，审计市场结构随着时间的推移发生了显著变化。当审计市场集中在少数大型审计公司手中时（即 HHI 值较高），小型审计公司提供的审计服务供应量较少。因此，我们预计 HHI 的系数为负。最后，我们还包括行业固定效应。
This table reports the regression results of IPO firms’ auditor choice (SL). Panel A shows the descriptive statistics of the regression variables. The t-statistics (Z-statistics) are from the two-sample t-test (Wilcoxon rank sum test) for the differences in mean (median) values. ^(**),^(****){ }^{*},{ }^{* *}, and ^(******){ }^{* * *} denote two-tailed statistical significance at the 10%,5%10 \%, 5 \%, and 1%1 \% level. Panel B reports the logistic regression results of SL. Regression A uses the IPO firms only. Regression B uses the IPO firms and all the seasoned firms in the IPO years (defined as firms that have been publicly listed for at least two complete fiscal years). See Appendix A for all variable definitions. GROWTH, CURRENT, ARINV, LEV, and ROA are measured in the year immediately prior to the IPO year. ASSETS is measured at the end of the IPO year. All the continuous variables are winsorized at the 1st and 99th percentiles. The reported pp-values are two-tailed and clustered by year. 该表报告了首次公开募股（IPO）公司的审计师选择（SL）的回归结果。面板 A 显示了回归变量的描述性统计数据。t 统计量（Z 统计量）来自于均值（中位数）差异的两样本 t 检验（Wilcoxon 秩和检验）。 ^(**),^(****){ }^{*},{ }^{* *} 和 ^(******){ }^{* * *} 表示在 10%,5%10 \%, 5 \% 和 1%1 \% 水平上的双尾统计显著性。面板 B 报告了 SL 的逻辑回归结果。回归 A 仅使用 IPO 公司。回归 B 使用 IPO 公司和 IPO 年份内所有的成熟公司（定义为至少公开上市两完整财年的公司）。有关所有变量定义，请参见附录 A。GROWTH、CURRENT、ARINV、LEV 和 ROA 的测量时间为 IPO 年份的前一年。ASSETS 在 IPO 年份末进行测量。所有连续变量在第 1 和第 99 百分位数处进行了温莎化。报告的 pp -值为双尾并按年份聚类。
Panel A of Table 3 shows the descriptive statistics of model (1)'s regression variables by the three regimes. The values of several explanatory variables differ significantly across the three regimes. Therefore, it is important to include these as control in the regression. 表 3 的面板 A 显示了模型（1）回归变量在三个制度下的描述性统计。几个解释变量在三个制度之间的值差异显著。因此，在回归中将这些变量作为控制变量是很重要的。
Panel B of Table 3 shows the regression results of auditor choice using different approaches. Regression A in the first two columns of Panel B, Table 3 shows the result of model (1) using the 784 IPO firms during our sample period. To control for calendar time cross sectional dependence, the reported pp-values for the regressions in Tables 3-7 are clustered by year with the appropriate degree of freedom adjustment (Cameron et al., 2008; Cameron and Miller, 2011). ^(13){ }^{13} The coefficients on REG1 and REG3 are both significantly positive. This evidence suggests that managers of IPO firms are more willing to hire a high-quality auditor in Regime 2 than in Regimes 1 and 3. 表 3 的面板 B 显示了使用不同方法的审计选择回归结果。表 3 面板 B 前两列的回归 A 显示了在我们的样本期间使用 784 家 IPO 公司的模型（1）的结果。为了控制日历时间的横截面依赖性，表 3-7 中回归的报告 pp -值按年份进行聚类，并进行了适当的自由度调整（Cameron 等，2008；Cameron 和 Miller，2011）。 ^(13){ }^{13} REG1 和 REG3 的系数均显著为正。这一证据表明，IPO 公司的管理者在第二种制度下比在第一和第三种制度下更愿意聘请高质量的审计师。
The coefficients on LEV (a proxy for audit risk), SOE (a proxy of state ownership), and HHI (a proxy for audit market concentration) are consistent with our predictions and significant at the 10%10 \% two-tailed level or better. The coefficient on RETAINED_OWN is significantly negative, suggesting a complementary relation between the choice of audit quality and RETAINED_OWN. Contrary LEV（审计风险的代理）、SOE（国有产权的代理）和 HHI（审计市场集中度的代理）上的系数与我们的预测一致，并在 10%10 \% 双尾水平或更好处显著。RETAINED_OWN 上的系数显著为负，表明审计质量选择与 RETAINED_OWN 之间存在互补关系。相反
This table reports the OLS regression result of IPO firms’ earnings management (PMDACC and NCROA). Panel A reports the descriptive statistics of the regression variables. The t-statistics (Z-statistics) are from the two-sample t-test (Wilcoxon rank sum test) for the differences in mean (median) values. *, ^(****){ }^{* *}, and ^(******){ }^{* * *} denote two-tailed statistical significance at the 10%,5%10 \%, 5 \%, and 1%1 \% level. Panel B reports the OLS regression result of PMDACC and NCROA. See Appendix A for variable definitions. ASSETS, GROWTH, CURRENT, LEV, and OCF are measured in the year immediately prior to the IPO year. All the continuous variables are winsorized at the 1 st and 99th percentiles. The reported pp-values are two-tailed and clustered by year. 该表报告了首次公开募股（IPO）公司的收益管理（PMDACC 和 NCROA）的 OLS 回归结果。面板 A 报告了回归变量的描述性统计。t 统计量（Z 统计量）来自于均值（中位数）差异的两样本 t 检验（Wilcoxon 秩和检验）。*, ^(****){ }^{* *} 和 ^(******){ }^{* * *} 表示在 10%,5%10 \%, 5 \% 和 1%1 \% 水平上的双尾统计显著性。面板 B 报告了 PMDACC 和 NCROA 的 OLS 回归结果。有关变量定义，请参见附录 A。ASSETS、GROWTH、CURRENT、LEV 和 OCF 的测量时间为 IPO 年之前的年份。所有连续变量在第 1 和第 99 百分位数处进行了温莎化。报告的 pp -值为双尾并按年份聚类。
to the argument that higher ROA reduces the demand for high quality audits, the coefficient on ROA is significantly negative. The coefficients on the other control variables are all insignificant. 对于“更高的资产回报率降低对高质量审计的需求”这一论点，资产回报率的系数显著为负。其他控制变量的系数均不显著。
A potential weakness of Regression A in Panel B of Table 3 is that the observed effects for REG1 and REG3 could be due to unobservable confounding forces that affect all publicly listed Chinese firms. To deal with this alternative explanation, we use seasoned firms (defined as firms that have been publicly listed for at least two complete fiscal years) as a control group by estimating the following modified model (1a): 回归 A 在表 3 的面板 B 中的一个潜在弱点是，REG1 和 REG3 的观察效应可能是由于影响所有上市中国公司的不可观察的混杂因素。为了应对这种替代解释，我们使用成熟公司（定义为至少公开上市两完整财年的公司）作为对照组，通过估计以下修改后的模型（1a）：
SL=beta_(0)+beta_(1)REG1+beta_(2)REG3+beta_(3)IPO+beta_(4)IPO xx REG1+beta_(5)IPO xx REG3+" Controls "+epsiS L=\beta_{0}+\beta_{1} R E G 1+\beta_{2} R E G 3+\beta_{3} I P O+\beta_{4} I P O \times R E G 1+\beta_{5} I P O \times R E G 3+\text { Controls }+\varepsilon
IPO is a dummy variable that equals one for IPO firms and zero for seasoned firms. The regression result of model (1a) is shown as Regression B in the last two columns of Panel B, Table 3. The sample includes 784 IPO firms during our sample period and 4386 contemporaneous seasoned firms. For each calendar year, the matched seasoned firms are randomly assigned to the three IPO regimes based on the proportion of IPO firms in each regime. For example, there are a total of 89 IPO firms and 297 matched seasoned firms in 1999. Since six of the 89 IPOs (6.74%) fall in Regime 1 and the remaining 83 ( 93.26%93.26 \% ) IPOs fall in Regime 2, we randomly allocate 20 (i.e., 6.74%xx2976.74 \% \times 297 ) of the seasoned firms to Regime 1 and the remaining 277 seasoned firms to Regime 2 in 1999. IPO 是一个虚拟变量，对于 IPO 公司等于 1，对于成熟公司等于 0。模型(1a)的回归结果显示在表 3 的 B 面板最后两列的回归 B 中。样本包括我们样本期间的 784 家 IPO 公司和 4386 家同期成熟公司。对于每个日历年，匹配的成熟公司根据每个制度中 IPO 公司的比例随机分配到三个 IPO 制度中。例如，1999 年共有 89 家 IPO 公司和 297 家匹配的成熟公司。由于 89 家 IPO 中有 6 家（6.74%）属于制度 1，其余 83 家（ 93.26%93.26 \% ）IPO 属于制度 2，因此我们在 1999 年随机将 20 家（即 6.74%xx2976.74 \% \times 297 ）成熟公司分配到制度 1，其余 277 家成熟公司分配到制度 2。
Our variables of interest in model (1a) are IPO xx\times REG1 and IPO xx\times REG3. The coefficients on IPO xx\times REG1 and IPO xx\times REG3 continue to be significantly positive. This evidence suggests that the results of Regression A in Panel B of Table 3 are unlikely due to confounding factors. ^(14){ }^{14} 我们在模型（1a）中关注的变量是 IPO xx\times REG1 和 IPO xx\times REG3。IPO xx\times REG1 和 IPO xx\times REG3 的系数仍然显著为正。这一证据表明，表 3 的 B 面板中的回归 A 结果不太可能是由于混杂因素造成的。 ^(14){ }^{14}
Table 5 表 5
IPO firms’ growth rates. IPO 公司的增长率。
Panel A shows the descriptive statistics of IPO firms’ sales growth measured using SALESGROWTH and ADJ_SALESGRWOTH and Tobin’s Q measured by GROWTH and ADJ_GROWTH. See Appendix A for variable definitions. The pp-values for the differences in the means (medians) are based on the two-sample t-test (Wilcoxon rank sum test). Panel B shows the OLS regression results. See Appendix A for definitions of independent variables. All the continuous variables are winsorized at the 1st and 99th percentiles. The reported pp-values are two-tailed and clustered by year. 面板 A 显示了首次公开募股（IPO）公司的销售增长的描述性统计数据，使用 SALESGROWTH 和 ADJ_SALESGROWTH 进行测量，以及使用 GROWTH 和 ADJ_GROWTH 测量的托宾 Q。有关变量定义，请参见附录 A。均值（中位数）差异的 pp -值基于两样本 t 检验（Wilcoxon 秩和检验）。面板 B 显示了 OLS 回归结果。有关自变量的定义，请参见附录 A。所有连续变量在第 1 和第 99 百分位数处进行了温莎化。报告的 pp -值为双尾并按年份聚类。
4.2.2. Test of H2 4.2.2. H2 测试
We use the following OLS regression model to test H 2 : 我们使用以下 OLS 回归模型来检验 H 2：
EM=beta_(0)+beta_(1)REG1+beta_(2)REG3+" Controls "+epsiE M=\beta_{0}+\beta_{1} R E G 1+\beta_{2} R E G 3+\text { Controls }+\varepsilon
See Appendix A for all variable definitions. EM represents the degree of upward earnings management in the year prior to the IPO year and is measured using both PMDACC (signed performance matched modified Jones model discretionary accruals) per 请参见附录 A 以获取所有变量定义。EM 代表首次公开募股（IPO）前一年向上盈余管理的程度，使用 PMDACC（签名绩效匹配的修正琼斯模型的自由裁量应计）进行测量。
Table 6 表 6
IPO firms’ long-term stock returns using the calendar time portfolio approach. IPO 公司的长期股票回报使用日历时间投资组合方法。
The table shows IPO firms’ mean monthly abnormal stock return (alpha)(\alpha) relative to the IPO offer price over the three years subsequent to the IPO offering. To control for cross-sectional dependence of long-term stock returns, we follow Mitchell and Stafford (2000) by using the calendar-time portfolio approach to compute the abnormal return and the significance of the abnormal stock return for the IPOs in each IPO regime. Specifically, for all the IPO firms in each IPO regime, a stock portfolio is formed each calendar month to include all companies that have completed the IPOs within the past 36 months. The portfolios are rebalanced monthly to drop firms that reach the end of their 36-month period and add all firms that have just conducted an IPO. We require a minimum of 10 observations for each portfolio month. Then, we use the Carhart (1997) four-factor model to compute the average monthly abnormal return (i.e., the estimated intercept alpha\alpha from the Carhart model). Annualized alpha\alpha is simply alpha\alpha multiplied by 12 . 表格显示了 IPO 公司在 IPO 发行后三年内相对于 IPO 发行价格的平均月异常股票收益 (alpha)(\alpha) 。为了控制长期股票收益的横截面依赖性，我们遵循 Mitchell 和 Stafford（2000）的做法，使用日历时间投资组合方法来计算异常收益及其在每个 IPO 阶段的显著性。具体而言，对于每个 IPO 阶段的所有 IPO 公司，每个月形成一个股票投资组合，包含过去 36 个月内完成 IPO 的所有公司。投资组合每月重新平衡，剔除达到 36 个月期限的公司，并添加刚刚进行 IPO 的所有公司。我们要求每个投资组合月份至少有 10 个观察值。然后，我们使用 Carhart（1997）四因子模型计算平均月异常收益（即 Carhart 模型中的估计截距 alpha\alpha ）。年化 alpha\alpha 仅为 alpha\alpha 乘以 12。
Table 7 表 7
Auditor choice three years after the IPO. 首次公开募股后三年审计师选择。
Panel A. Do IPO firms switch from SL auditors to non-SL auditors three years after the IPO? 面板 A。IPO 公司在 IPO 后三年内是否从 SL 审计师转向非 SL 审计师？
The dependent variable in both panels is DIFFERENT_AUDITOR. The sample in Panel A includes only the IPO firms that employed SL auditors at the time of the IPO and their matched seasoned firms that also employed SL auditors in the year prior to the IPO. The sample in Panel B includes only the IPO firms that employed non-SL auditors at the time of the IPO and their matched seasoned firms that also employed non-SL auditors in the year prior to the IPO. The control variables other than Audit firm experiences a merger and Audit firm is closed by regulators are the same as those in Table 3 except that they are changes from the IPO year to three years after the IPO. See Appendix A for the other variable definitions. All the continuous variables are winsorized at the 1st and 99th percentiles. The reported pp-values are two-tailed and clustered by year. 两个面板中的因变量是 DIFFERENT_AUDITOR。面板 A 的样本仅包括在首次公开募股（IPO）时聘用 SL 审计师的公司及其在 IPO 前一年也聘用 SL 审计师的匹配成熟公司。面板 B 的样本仅包括在 IPO 时聘用非 SL 审计师的公司及其在 IPO 前一年也聘用非 SL 审计师的匹配成熟公司。除了审计公司经历合并和审计公司被监管机构关闭的控制变量外，其他控制变量与表 3 中的相同，只是它们是从 IPO 年份到 IPO 后三年的变化。有关其他变量定义，请参见附录 A。所有连续变量在第 1 和第 99 百分位数处进行了温莎化。报告的 pp -值为双尾并按年份聚类。
Kothari et al. (2005) and NCROA (signed non-operating earnings) per Chen and Yuan (2004). ^(15){ }^{15} PMDACC represents the extent of upward earnings management using operating activities while NCROA represents the extent of upward earnings management using non-operating activities. ^(16){ }^{16} As explained in Appendix B, PMDACC is the difference in abnormal accruals (DACC) between IPO firm year it and a matched seasoned firm in the same two-digit industry as firm ii with the closest ROA in year tt. Hence, regression model (2) for PMDACC is essentially similar in spirit to regression model (1a). Kothari 等（2005）和 NCROA（签署的非经营收益）根据 Chen 和 Yuan（2004）。 ^(15){ }^{15} PMDACC 代表使用经营活动进行向上收益管理的程度，而 NCROA 代表使用非经营活动进行向上收益管理的程度。 ^(16){ }^{16} 如附录 B 所解释，PMDACC 是 IPO 公司年份与同一两位数行业中与公司 ii 在年份 tt 中 ROA 最接近的匹配成熟公司之间的异常应计差异（DACC）。因此，PMDACC 的回归模型（2）在精神上与回归模型（1a）基本相似。
REG1 and REG3 are used to test H2. To the extent that managers of IPO firms have a stronger incentive to inflate reported earnings in the period prior to the IPO year during Regimes 1 and 3 than during Regime 2, the coefficients on REG1 and REG3 are predicted to be positive. REG1 和 REG3 用于测试 H2。在第一和第三阶段，IPO 公司的管理者在 IPO 年份之前的报告收益膨胀动机比在第二阶段更强，因此 REG1 和 REG3 的系数预计为正。
Controls is a list of standard control variables from prior research, including operating cash flow (OCF), firm size (ASSETS), leverage (LEV), growth (GROWTH), current ratio (CURRENT), and industry fixed effects (e.g., DeFond and Jiambalvo, 1994; Becker et al., 1998; Menon and Williams, 2004; Francis and Yu, 2009; Chen and Yuan, 2004; Chahine et al., 2012; Chahine et al., 2015; Lee and Masulis, 2011). We also include SOE because ownership structure may affect the incentives of earnings management (Chen et al., 2011). 控制变量是来自先前研究的标准控制变量列表，包括经营现金流（OCF）、公司规模（资产）、杠杆（LEV）、增长（GROWTH）、流动比率（CURRENT）和行业固定效应（例如，DeFond 和 Jiambalvo，1994；Becker 等，1998；Menon 和 Williams，2004；Francis 和 Yu，2009；Chen 和 Yuan，2004；Chahine 等，2012；Chahine 等，2015；Lee 和 Masulis，2011）。我们还包括国有企业（SOE），因为所有权结构可能会影响盈余管理的激励（Chen 等，2011）。
Table 4 reports the test results of H2. Panel A shows the descriptive statistics of model (2)'s regression variables by the three regimes. The values of several explanatory variables differ significantly across the three regimes. Therefore, it is important to include these as control in the regression. 表 4 报告了 H2 的测试结果。面板 A 显示了模型（2）回归变量在三个状态下的描述性统计。几个解释变量在三个状态下的值差异显著。因此，在回归中将这些作为控制变量是很重要的。
Panel B shows the OLS regression results of model (2). When the dependent variable is PMDACC, the coefficients on REG1 and REG3 are both significantly positive. When the dependent variable is NCROA, neither of the coefficients on REG1 and REG3 is significantly positive. These results suggest that managers of IPO firms have a greater incentive to use discretionary operating activity accruals but not non-operating activity earnings to inflate reported earnings at the IPO time during Regimes 1 and 3 than during Regime 2. The fact that managers of IPO firms do not use NCROA to inflate earnings may not be surprising because as noted in footnote 15 , such earnings management could be too obvious to the CSRC and thus would raise immediate red flags in the CSRC’s IPO application review process. ^(17){ }^{17} 面板 B 显示了模型（2）的 OLS 回归结果。当因变量为 PMDACC 时，REG1 和 REG3 的系数均显著为正。当因变量为 NCROA 时，REG1 和 REG3 的系数均不显著为正。这些结果表明，IPO 公司的管理者在第 1 和第 3 阶段比在第 2 阶段更有动力使用可自由支配的经营活动应计项来夸大 IPO 时的报告收益，而不是使用非经营活动收益。IPO 公司的管理者不使用 NCROA 来夸大收益可能并不令人惊讶，因为正如脚注 15 所述，这种收益管理可能对证监会来说过于明显，因此在证监会的 IPO 申请审查过程中会引起立即的警觉。
The coefficients on the control variables GROWTH, and OCF are significant at the 10%10 \% level in the regressions of PMDACC. The positive coefficient on GROWTH is consistent with the hypothesis that growth firms have greater incentives to manage earnings (Dechow et al., 2011). The negative coefficient on OCF is consistent with Dechow (1994) who finds a negative correlation between accruals and operating cash flows. The coefficients on SOE are negative at the 10%10 \% level, consistent with Chen et al. (2011) who suggest that state-controlled firms have weaker incentives to manage earnings. 控制变量 GROWTH 和 OCF 的系数在 PMDACC 的回归中在 10%10 \% 水平上显著。GROWTH 的正系数与增长型企业有更大动机管理收益的假设一致（Dechow 等，2011）。OCF 的负系数与 Dechow（1994）的发现一致，即应计项目与经营现金流之间存在负相关。SOE 的系数在 10%10 \% 水平上为负，与 Chen 等（2011）的观点一致，他们认为国有企业管理收益的动机较弱。
4.3. Competing explanations 4.3. 竞争性解释
During our sample period China observed many significant institutional reforms that could also affect IPO firms’ financial reporting behavior. In this section we examine how these competing institutional forces may affect the interpretation of our empirical results reported in Tables 3 and 4. A key strength of our research designs for Tables 3 and 4 is the availability of three IPO offer pricing regimes and therefore any alternative explanation has to explain not only the initial change of results from Regime 1 to Regime 2 but also the reversal of results from Regime 2 to Regime 3. Overall, we find no evidence that the confounding factors discussed below can explain our regression results across the entire three regimes. 在我们的样本期间，中国观察到了许多重要的制度改革，这些改革也可能影响首次公开募股（IPO）公司的财务报告行为。在本节中，我们将探讨这些相互竞争的制度力量如何影响我们在表 3 和表 4 中报告的实证结果的解释。表 3 和表 4 研究设计的一个关键优势是有三个 IPO 定价机制，因此任何替代解释不仅需要解释从机制 1 到机制 2 的初始结果变化，还需要解释从机制 2 到机制 3 的结果反转。总体而言，我们没有发现证据表明下面讨论的混杂因素可以解释我们在整个三个机制中的回归结果。
4.3.1. Accounting and auditing standards 4.3.1. 会计和审计标准
The first potential confounding factor is the modernization of China’s financial accounting and auditing standards. To serve the immediate need of China’s financial market development, Chinese regulators overhauled the accounting standards twice during our sample period (i.e., 1998 and 2001). China’s auditing standards and audit profession have also experienced significant changes since early 1990s. Regulators’ oversight of audit firms have also increased significantly since 2001 due to a few high profile accounting and auditing scandals around 2001. Overall, during our sample period China’s accounting and auditing standards have become more complete, rigorous, and compatible with the rising demand for information by capital market investors. Hence, to the extent that improving accounting and auditing standards result in a positive impact on listed firms’ financial reporting quality, we should expect Chinese IPO firms’ financial reporting quality to increase monotonically during our sample period. In addition, changes in accounting and auditing standards should affect all listed firms. Since we use seasoned firms as a control in all of our regression analyses, the confounding influences of accounting and auditing standards, if any, should be further minimized. Therefore, it is unlikely that the gradual strengthening of China’s accounting and auditing standards can explain our results across the three regimes, especially the reversal of financial reporting quality in Regime 3. 第一个潜在的混淆因素是中国财务会计和审计标准的现代化。为了满足中国金融市场发展的迫切需求，中国监管机构在我们的样本期间内对会计标准进行了两次全面改革（即 1998 年和 2001 年）。自 1990 年代初以来，中国的审计标准和审计行业也经历了重大变化。由于 2001 年发生的一些高调的会计和审计丑闻，监管机构对审计公司的监督自 2001 年以来也显著增加。总体而言，在我们的样本期间，中国的会计和审计标准变得更加完善、严格，并与资本市场投资者对信息日益增长的需求相兼容。因此，在改善会计和审计标准对上市公司财务报告质量产生积极影响的情况下，我们应该预期中国 IPO 公司的财务报告质量在我们的样本期间内单调增加。此外，会计和审计标准的变化应影响所有上市公司。 由于我们在所有回归分析中使用成熟公司作为对照，因此会计和审计标准的混杂影响（如果有的话）应该进一步最小化。因此，中国会计和审计标准的逐步加强不太可能解释我们在三个制度下的结果，特别是在制度 3 中财务报告质量的逆转。
4.3.2. Alternative monitors of listed firms' financial reporting 4.3.2. 上市公司财务报告的替代监测者
There were significant changes in the monitoring environment of publicly listed Chinese firms during our sample period. One important change is the perceived increase in litigation risk facing listed firms and audit firms since 2001 due to stepped up regulatory enforcement. Another important change is the gradual development of China’s nascent institutional investor community, mutual fund management companies in particular since 2001. A third important change is the restructuring of the CSRC’s Stock Issuance Examination and Verification Committee in December 2003 that significantly increased the professionalism and transparency of the Committee. ^(18){ }^{18} We expect these institutional changes to increase the overall monitoring environment of publicly listed 在我们的样本期间，上市中国公司的监测环境发生了显著变化。一个重要变化是自 2001 年以来，由于监管执法的加强，上市公司和审计公司面临的诉讼风险感知增加。另一个重要变化是自 2001 年以来，中国新兴的机构投资者群体，特别是共同基金管理公司的逐步发展。第三个重要变化是 2003 年 12 月中国证监会的股票发行审核委员会的重组，这显著提高了委员会的专业性和透明度。我们预计这些制度变化将增强上市公司的整体监测环境。
Chinese firms throughout our entire sample period and therefore they are unlikely the drivers of our results across the three regimes, especially the reversal of financial reporting quality in Regime 3. 在我们整个样本期间，中国公司因此不太可能是我们在三个制度中结果的驱动因素，特别是在第三个制度中财务报告质量的逆转。
4.3.3. The stock market sentiment 4.3.3. 股票市场情绪
As noted in Section 3, the switch from Regime 2 to Regime 3 was prompted by the steep drop of the Shanghai stock market index. In fact, the Shanghai stock market composite index hovered around 1200 in Regime 1 and around 1500 in Regime 3. In contrast, the Shanghai stock market composite index rose from around 1100 all the way up to around 2200 in Regime 2, a boom market. Hence, one may wonder whether the documented results in Tables 3 and 4 are due to changes in the overall stock market sentiment rather than changes in the IPO pricing regimes. Our analyses indicate that changing market sentiment is unlikely to be an alternative explanation for the reported results in Tables 3 and 4 for the following reasons. First, Rajgopal et al. (2007) find that earnings management is higher during high stock market sentiment periods. In contrast, we find earnings management to be lower in Regime 2, a high stock market sentiment period, inconsistent with this alternative explanation. Second, there is no reason to believe that the effect of market sentiment on earnings management differs for IPO firms and seasoned firms. Since we use seasoned firms as a control in both Table 3 and Table 4 (note we use performance-matched discretionary accruals in Table 4), the effect of market sentiment should have already been neutralized in our reported results. Third, we directly control for the market returns in the IPO quarter and the two quarters prior in Tables 3 and 4 and find similar inferences (untabulated). 如第 3 节所述，从第 2 阶段转向第 3 阶段是由于上海股市指数的急剧下跌。实际上，上海股市综合指数在第 1 阶段徘徊在 1200 左右，在第 3 阶段徘徊在 1500 左右。相比之下，上海股市综合指数在第 2 阶段（一个繁荣市场）从 1100 左右一路上涨到 2200 左右。因此，人们可能会想知道表 3 和表 4 中记录的结果是否是由于整体股市情绪的变化，而不是 IPO 定价机制的变化。我们的分析表明，市场情绪的变化不太可能是表 3 和表 4 中报告结果的替代解释，原因如下。首先，Rajgopal 等（2007）发现，在高股市情绪时期，盈余管理水平较高。相反，我们发现第 2 阶段（一个高股市情绪时期）的盈余管理水平较低，这与这一替代解释不一致。其次，没有理由相信市场情绪对盈余管理的影响在 IPO 公司和成熟公司之间存在差异。 由于我们在表 3 和表 4 中使用成熟公司作为对照（请注意，我们在表 4 中使用了业绩匹配的自由裁量应计），市场情绪的影响应该已经在我们报告的结果中被中和。第三，我们在表 3 和表 4 中直接控制了首次公开募股季度及前两个季度的市场回报，并得出了类似的推论（未列出表格）。
The empirical analyses performed so far have taken as given the three IPO regulatory regimes and the composition of the IPO firms within each regime. In this section we examine whether our inferences in Tables 3 and 4 are sensitive to this assumption by analyzing two types of potential selection biases: (a) the selection bias due to the government intervention in the IPO selection process; and (b) the selection bias due to individual firms’ timing of IPOs. 到目前为止进行的实证分析已将三种首次公开募股（IPO）监管制度及每种制度下 IPO 公司的组成视为既定。在本节中，我们将通过分析两种潜在的选择偏差来检验表 3 和表 4 中的推论是否对这一假设敏感：（a）由于政府干预 IPO 选择过程而导致的选择偏差；（b）由于个别公司 IPO 时机的选择而导致的选择偏差。
The composition of the IPO firms could vary across the three regimes because the Chinese government has always played an important role in selecting IPO firms since the establishment of China’s modern financial markets in 1990. First, the Company Law always requires IPO firms to satisfy certain minimum requirements such as firm size or profitability. However, the stipulated IPO eligibility requirements were pretty much the same over the three IPO regimes and thus shouldn’t affect our inferences. IPO 公司的组成可能在三个制度之间有所不同，因为自 1990 年中国现代金融市场建立以来，中国政府在选择 IPO 公司方面一直发挥着重要作用。首先，公司法始终要求 IPO 公司满足某些最低要求，例如公司规模或盈利能力。然而，规定的 IPO 资格要求在三个 IPO 制度中基本相同，因此不应影响我们的推论。
Second, to support the central government’s five-year national economic plans, ^(19){ }^{19} the CSRC issued explicit IPO regulations (e.g., CSRC, 1996 and CSRC, 1997) that favored key strategic state-owned enterprises or business enterprises in selective industries such as agriculture, energy, transportation, telecommunication, and raw materials during our entire sample period. To address this concern, we have controlled for the differences in IPO firms’ financial, ownership, and industry characteristics in our regressions. ^(20){ }^{20} In addition, given that many of the regulatory IPO selection policies favored state-owned enterprises, we also repeat all of our reported regressions by limiting the IPO sample to SOE firms (untabulated). All of our inferences for H1-H2\mathrm{H} 1-\mathrm{H} 2 are similar. 其次，为了支持中央政府的五年国家经济计划， ^(19){ }^{19} 中国证监会在整个样本期间发布了明确的 IPO 规定（例如，1996 年和 1997 年），这些规定有利于关键的战略国有企业或在农业、能源、交通、通信和原材料等特定行业的商业企业。为了解决这一问题，我们在回归分析中控制了 IPO 公司在财务、所有权和行业特征方面的差异。 ^(20){ }^{20} 此外，考虑到许多监管 IPO 选择政策偏向国有企业，我们还通过将 IPO 样本限制为国有企业（未列出）来重复所有报告的回归分析。我们对 H1-H2\mathrm{H} 1-\mathrm{H} 2 的所有推论都是相似的。
Third, the way IPO firm candidates were selected changed during our sample period. Prior to 2003, mainland China’s IPOs followed a rigid quota system under which the amount of new shares or the number of IPOs to be issued in a given year were allocated by the CSRC to the provinces and central government ministries while the provincial governments and central government ministries were directly responsible for recommending IPO firm candidates under their control to the CSRC for approval (Pistor and Xu, 2005). During the post-quota system period the role of provincial governments and central government ministries in the IPO firm selection was significantly reduced, but the CSRC’s control over the IPO firm selection still remained intact. We do not believe that the abolishment of the quota system can explain the differential regression results across the three regimes, especially the reversal in Regime 3. The reason is that the IPO firm selection procedures were the same in Regimes 1 and 2 and hence the difference in results between Regimes 1 and 2 cannot be explained by the change of the quota system. In addition, the CSRC was directly responsible for the IPO firm selection in Regime 3. If the government-based approach is more effective in protecting investors, we should expect IPO firms’ financial reporting quality to be higher in Regime 3 than in Regime 1 or Regime 2, a prediction not supported by our results. ^(21){ }^{21} 第三，在我们的样本期间，IPO 公司候选人的选择方式发生了变化。2003 年之前，中国大陆的 IPO 遵循严格的配额制度，新的股票数量或每年发行的 IPO 数量由中国证监会分配给各省和中央政府部门，而各省政府和中央政府部门直接负责向中国证监会推荐其管辖下的 IPO 公司候选人以供批准（Pistor 和 Xu，2005）。在后配额制度时期，省政府和中央政府部门在 IPO 公司选择中的作用显著减少，但中国证监会对 IPO 公司选择的控制仍然保持不变。我们认为，配额制度的废除无法解释三个制度之间的差异回归结果，特别是制度 3 中的反转。原因在于，制度 1 和制度 2 中的 IPO 公司选择程序是相同的，因此制度 1 和制度 2 之间结果的差异无法通过配额制度的变化来解释。 此外，证监会在第三阶段直接负责首次公开募股（IPO）公司的选择。如果政府主导的方法在保护投资者方面更有效，我们应该预期第三阶段的 IPO 公司财务报告质量会高于第一阶段或第二阶段，但我们的结果并不支持这一预测。 ^(21){ }^{21}
The composition of the IPO firms could also vary across the three regimes because the founding shareholders of IPO firms may have an incentive to time their firms’ IPOs. For example, during Regime 1 or 3, the founding shareholders of an IPO firm, growth firms in particular, may decide not to go public if they find the regulator-prescribed offer price to be unattractive even after earnings inflation. However, financing opportunities for unlisted Chinese firms were quite limited in our sample period. Typically, going public was the most viable financing option. In addition, it was difficult for any firm to predict the government’s future IPO policies. Hence, if given an opportunity to go public in Regime 1 or 3, a growth firm may not be willing to give up the opportunity because the firm does not know when the next opportunity will show up. IPO 公司的组成在三个制度下也可能有所不同，因为 IPO 公司的创始股东可能有动力来选择其公司的 IPO 时机。例如，在制度 1 或 3 期间，IPO 公司的创始股东，特别是成长型公司，可能会决定不上市，如果他们发现监管机构规定的发行价格即使在盈利膨胀后也不具吸引力。然而，在我们的样本期内，未上市中国公司的融资机会相当有限。通常，上市是最可行的融资选择。此外，任何公司都很难预测政府未来的 IPO 政策。因此，如果在制度 1 或 3 中有机会上市，成长型公司可能不愿意放弃这个机会，因为该公司不知道下一个机会何时会出现。
To directly assess the severity of the IPO timing bias by growth firms across the three regimes, we next examine whether IPO firms are more likely to be growth firms during Regime 2 than during Regimes 1 and 3. Firm growth is measured using the average sales growth in the two years prior to the IPO year (SALESGROWTH) and Tobin’s Q (GROWTH). 为了直接评估在三个阶段中成长型公司首次公开募股（IPO）时机偏差的严重性，我们接下来检查在第二阶段中，IPO 公司是否比在第一和第三阶段更可能是成长型公司。公司增长通过首次公开募股年份前两年的平均销售增长（SALESGROWTH）和托宾 Q（GROWTH）来衡量。
An IPO firm’s SALESGROWTH is affected not only by a firm’s IPO timing strategy but also by the industry and regional economic conditions at the time of IPO. To control for industry effects, we compute an industry-adjusted average sales growth (ADJ_SALESGROWTH) by subtracting the median SALESGROWTH of all seasoned firms in the same two-digit industry year from the IPO firm’s SALESGRWOTH. Similarly, an IPO firm’s Tobin’s Q (GROWTH) could be also affected by the prevailing stock market sentiment unrelated to the IPO firm’s true growth. Hence, we also use ADJ_GROWTH. To control for an IPO firm’s regional economic conditions at the time of IPO, we also run regressions of our dependent variables on REG1, REG3, and a list of control variables measuring the provincial economic conditions of the IPO firm’s headquarters. 一家 IPO 公司的销售增长不仅受到公司 IPO 时机策略的影响，还受到 IPO 时行业和地区经济状况的影响。为了控制行业效应，我们通过从 IPO 公司的销售增长中减去同一两位数行业年份所有成熟公司的销售增长中位数，计算出行业调整后的平均销售增长（ADJ_SALESGROWTH）。同样，IPO 公司的托宾 Q（增长）也可能受到与 IPO 公司真实增长无关的市场情绪的影响。因此，我们还使用 ADJ_GROWTH。为了控制 IPO 公司在 IPO 时的地区经济状况，我们还对我们的因变量进行回归分析，使用 REG1、REG3 以及一系列衡量 IPO 公司总部省级经济状况的控制变量。
As shown in Panel A of Table 5, we find no evidence that the mean and median SALESGROWTH or the mean and median ADJ_SALESGROWTH are significantly higher for the IPOs in Regime 2. In fact, based on the medians, there is evidence that IPO firms’ growth is lower in Regime 2 than in Regimes 1 and 3. In addition, as shown in Panel B of Table 5, after controlling for the provincial economic conditions of IPO firms’ headquarters, the coefficients on REG1 and REG3 are never significantly negative, suggesting that there is no evidence that growth IPO firms are more likely to go public in Regime 2 than in Regime 1 or 3 . We reach similar inferences when using GROWTH and ADJ_GROWTH as the dependent variables (see the right columns of Panels A and B). The only exception is that we find some evidence consistent with the self-selection bias if the dependent variable is GROWTH but this effect is purely driven by the overall high stock market sentiment during Regime 2 rather than unique to the IPO firms in Regime 2. 如表 5 的面板 A 所示，我们没有发现证据表明，第二阶段的 IPO 公司在平均和中位数销售增长（SALESGROWTH）或调整后的销售增长（ADJ_SALESGROWTH）方面显著高于其他阶段。事实上，根据中位数的数据，有证据表明，IPO 公司的增长在第二阶段低于第一和第三阶段。此外，如表 5 的面板 B 所示，在控制了 IPO 公司总部的省级经济条件后，REG1 和 REG3 的系数从未显著为负，这表明没有证据表明增长型 IPO 公司在第二阶段比在第一或第三阶段更可能上市。当使用 GROWTH 和 ADJ_GROWTH 作为因变量时，我们得出的推论也类似（见面板 A 和 B 的右侧列）。唯一的例外是，如果因变量是 GROWTH，我们发现一些与自我选择偏差一致的证据，但这一效应纯粹是由第二阶段整体高涨的股市情绪驱动，而不是第二阶段 IPO 公司所特有的。
As noted in the Introduction, the debate on the government-based approach versus the market-based approach partially reflects regulators’ concern that market forces in weak investor protection countries may not be as effective in pricing IPOs and as a result public investors may end up overpaying for IPOs. In this section we directly test the validity of this concern by examining our sample IPOs’ long-term abnormal stock returns relative to the IPO offer prices across the three regimes. To the extent that IPO offerings are overpriced, there should be a reversal of IPO firms’ long-term abnormal stock returns as investors gradually learn more about the firms’ true values over time. 正如引言中所提到的，政府主导的方式与市场主导的方式之间的辩论部分反映了监管者的担忧，即在投资者保护较弱的国家，市场力量在定价首次公开募股（IPO）方面可能不够有效，因此公众投资者可能会为 IPO 支付过高的价格。在本节中，我们直接测试这一担忧的有效性，通过考察我们样本中的 IPO 在三种体制下相对于 IPO 发行价格的长期异常股票收益。若 IPO 发行价格被高估，随着投资者逐渐了解公司的真实价值，IPO 公司的长期异常股票收益应该会出现反转。
Table 6 shows the results of abnormal stock returns relative to the IPO offer price over a three year period subsequent to the IPO offering. Inferences are similar if we use a one-year or two-year horizon. It is important to note that the long term abnormal returns based on the Carhart (1997) four-factor model in Table 6 are defined relative to the IPO offer price rather than the IPO’s first-trading day closing price because our interest here is to test whether IPO offer prices are appropriately determined. ^(22){ }^{22} To control for cross sectional dependence of long-term abnormal returns, we compute the abnormal returns and associated significance levels using the calendar time portfolio approach per Mitchell and Stafford (2000). 表 6 显示了 IPO 发行后三年内相对于 IPO 发行价格的异常股票收益结果。如果我们使用一年或两年的时间范围，推论是相似的。需要注意的是，表 6 中基于 Carhart（1997）四因子模型的长期异常收益是相对于 IPO 发行价格而不是 IPO 首日收盘价来定义的，因为我们在这里的兴趣是测试 IPO 发行价格是否被适当地确定。为了控制长期异常收益的横截面依赖性，我们根据 Mitchell 和 Stafford（2000）的方法，使用日历时间组合方法计算异常收益及相关显著性水平。
As shown in Table 6, the mean monthly abnormal return for the IPOs is 0.093 for Regime 1, 0.068 for Regime 2, and 0.054 for Regime 3. Hence, the average long-term abnormal return of Chinese IPOs defined relative to the offer price is always positive for the IPOs in all three regimes. ^(23){ }^{23} This evidence suggests that there are some common institutional factors that may cause significant IPO underpricing in all three regimes (see Guo et al., 2014). More importantly, there is no evidence that the abnormal returns across the three regimes differ significantly from each other. Hence, we find no evidence that the IPO offer prices are more likely to be over priced in Regime 2 than in Regimes 1 and 3. 如表 6 所示，IPO 的平均月异常收益率在第一阶段为 0.093，在第二阶段为 0.068，在第三阶段为 0.054。因此，相对于发行价格，中国 IPO 的平均长期异常收益在所有三个阶段始终为正。 ^(23){ }^{23} 这一证据表明，可能存在一些共同的制度因素导致在所有三个阶段中 IPO 的显著低估（见 Guo 等，2014）。更重要的是，没有证据表明三个阶段的异常收益率之间存在显著差异。因此，我们没有发现证据表明在第二阶段 IPO 的发行价格比第一和第三阶段更可能被高估。
Given that IPO shares were priced using much lower PE ratios in Regimes 1 and 3 than in Regime 2, the similar long term abnormal returns of the IPOs in the three regimes may seem surprising. The explanation for this surprising result is that the earnings numbers used in the IPO pricing were significantly overstated in Regimes 1 and 3 than in Regime 2 as shown in Section 4.2. Therefore, the PE ratios based on the firms’ true earnings for the IPOs in Regimes 1 and 3 are actually much higher than the reported ones. Hence, it is not totally surprising to observe similar long term abnormal returns for the IPOs in the three regimes. 考虑到在第 1 和第 3 阶段，首次公开募股（IPO）股票的定价使用的市盈率（PE）远低于第 2 阶段，因此这三个阶段的 IPO 长期异常收益相似可能令人惊讶。对此惊人结果的解释是，在第 1 和第 3 阶段，用于 IPO 定价的盈利数字显著高于第 2 阶段，如第 4.2 节所示。因此，基于公司真实盈利的第 1 和第 3 阶段 IPO 的市盈率实际上远高于报告的市盈率。因此，观察到这三个阶段的 IPO 具有相似的长期异常收益并不完全令人惊讶。
Overall, the evidence in Table 6 suggests that Chinese regulators’ concern over IPO overpricing seems misplaced because even in the absence of regulators’ intervention, Chinese IPO investors know how to price protect, even during a high market sentiment period, as manifested by the significant IPO underpricing in Regime 2. On the other hand, regulators’ intervention in IPO pricing in Regimes 1 and 3 results in at least one important side effect: costly upward earnings management at the time of the IPO. 总体而言，表 6 中的证据表明，中国监管机构对 IPO 定价过高的担忧似乎是多余的，因为即使在没有监管干预的情况下，中国的 IPO 投资者也知道如何进行价格保护，即使在市场情绪高涨的时期，这在第二阶段的显著 IPO 低估中得到了体现。另一方面，监管机构在第一和第三阶段对 IPO 定价的干预导致了至少一个重要的副作用：在 IPO 时成本高昂的向上盈余管理。
6. Financial reporting quality in the post-IPO period 6. IPO 后时期的财务报告质量
In this section we examine whether IPO firms’ financial reporting quality at the time of the IPO also affects their financial reporting quality long after the IPO, defined as three-five years after the IPO. The conjecture we wish to test is whether a firm has an incentive to maintain the same reporting strategy in the years after the IPO, once the firm has selected a low (high) quality financial reporting strategy at the time of the IPO. We test our conjecture using three complementary approaches discussed in the following subsections. 在本节中，我们考察首次公开募股（IPO）公司在 IPO 时的财务报告质量是否会影响其在 IPO 后很长一段时间内的财务报告质量，定义为 IPO 后三到五年。我们希望检验的假设是，一旦公司在 IPO 时选择了低（高）质量的财务报告策略，是否有动力在 IPO 后的几年中维持相同的报告策略。我们将使用以下小节中讨论的三种互补方法来检验我们的假设。
Our conjecture is consistent with Stein’s (1989) rational expectation model (see also Shivakumar, 2000). Specifically, according to the model, once a firm has selected a low quality financial reporting strategy at the IPO, rational investors would expect the firm’s management to continue to have an incentive to manage earnings subsequent to the IPO and therefore would rationally discount the firm’s post-IPO earnings accordingly. Taking investors’ conjecture as given, a rational corporate insider would find it optimal to manipulate reported earnings in the post-IPO periods because the firm cannot credibly signal the absence of earnings management. In other words, earnings management by the firm and the resulting discounting by investors is a unique Nash equilibrium in a prisoner’s dilemma game between the firm and investors. 我们的推测与斯坦因（1989）的理性预期模型一致（另见 Shivakumar，2000）。具体而言，根据该模型，一旦公司在首次公开募股（IPO）时选择了低质量的财务报告策略，理性投资者会预期公司的管理层在 IPO 之后仍然有动机进行盈余管理，因此会相应地理性地折扣公司的 IPO 后盈余。在假设投资者的推测为前提下，理性的公司内部人士会发现，在 IPO 后的时期操纵报告盈余是最优的，因为公司无法可信地发出没有盈余管理的信号。换句话说，公司的盈余管理和投资者因此产生的折扣是在公司与投资者之间的囚徒困境游戏中的一个独特纳什均衡。
Similarly, once a firm has committed to a high quality reporting strategy at the IPO, rational investors would expect the firm’s management to continue to do so subsequent to the IPO. Taking investors’ conjecture as given, a rational corporate insider would also find it optimal to maintain the high quality reporting strategy in the post-IPO period because trust by the market is a valuable intangible asset and any deviation from the bonding to a high quality reporting strategy would destroy the market’s trust and therefore would be costly to corporate insiders. 同样，一旦公司在首次公开募股（IPO）时承诺采用高质量的报告策略，理性的投资者会期望公司管理层在 IPO 之后继续这样做。考虑到投资者的推测，理性的公司内部人士也会发现，在 IPO 后维持高质量的报告策略是最优的，因为市场的信任是一种宝贵的无形资产，任何偏离高质量报告策略的行为都会破坏市场的信任，因此对公司内部人士来说是有成本的。
6.1. Auditor quality 6.1. 审计师质量
We first use auditor quality as a proxy for a firm’s reporting strategy. We are interested in the following questions: 我们首先使用审计师质量作为公司报告策略的代理。我们对以下问题感兴趣：
(a) Conditional on an IPO firm selecting a low quality auditor at the time of the IPO in Regimes 1 and 3, would the IPO firm have an incentive to select a higher quality auditor three years after the IPO? （a）在第一和第三种情况下，如果 IPO 公司在 IPO 时选择了低质量审计师，IPO 公司在 IPO 后三年是否会有动力选择更高质量的审计师？
(b) Conditional on an IPO firm selecting a high quality auditor at the time of the IPO in Regime 2, would the IPO firm have an incentive to select a lower quality auditor three years after the IPO? (b) 在第二种情况下，如果 IPO 公司在 IPO 时选择了一位高质量的审计师，那么在 IPO 后三年，IPO 公司是否会有动机选择一位低质量的审计师？
To test each of the above two questions, we adopt the following regression model: 为了测试上述两个问题，我们采用以下回归模型：
" DIFFERENT AUDITOR "=beta_(0)+beta_(1)" IPO "+" Controls "+epsi\text { DIFFERENT AUDITOR }=\beta_{0}+\beta_{1} \text { IPO }+ \text { Controls }+\varepsilon
We run the model for the three IPO regimes separately. Because auditor choice could change for a variety of other reasons, we match each IPO firm in the IPO year tt with all the seasoned firms in the same year tt that employ the same type of auditor (i.e., SL=1S L=1 or SL=0S L=0 ). DIFFERENT_AUDITOR is one if an IPO firm or a matched non-IPO firm selects a low quality auditor (i.e., SL=1S L=1 ) at the IPO time but a high quality auditor (i.e., SL=0S L=0 ) three years after the IPO or vice versa, and zero otherwise. The control variables in Controls are the same as those in Table 3 except that they are changes from the IPO year to three years after the IPO. Our variable of interest is IPO, which equals one for IPO firms and zero for non-IPO firms. We have no ex ante predictions for the coefficients on IPO. 我们分别对三个 IPO 制度运行模型。由于审计师选择可能因其他多种原因而变化，我们将每个 IPO 公司在 IPO 年份 tt 与同一年 tt 中所有采用相同类型审计师（即 SL=1S L=1 或 SL=0S L=0 ）的成熟公司进行匹配。如果 IPO 公司或匹配的非 IPO 公司在 IPO 时选择了低质量审计师（即 SL=1S L=1 ），但在 IPO 后三年选择了高质量审计师（即 SL=0S L=0 ），或者反之，则 DIFFERENT_AUDITOR 为 1，否则为 0。控制变量 Controls 与表 3 中的相同，只是它们是从 IPO 年份到 IPO 后三年的变化。我们关注的变量是 IPO，对于 IPO 公司等于 1，对于非 IPO 公司等于 0。我们对 IPO 的系数没有事先的预测。
Table 7 shows the regression results. We first focus on the IPO firms that select a low quality auditor at the IPO year (see Panel A of Table 7). The coefficients on IPO are insignificant for both Regimes 1 and 3. Thus, there is no evidence that IPO firms in Regimes 1 and 3 that selected a low quality auditor in the IPO year are more likely than the control firms to switch to a high quality auditor three years after the IPO. 表 7 显示了回归结果。我们首先关注在 IPO 年份选择低质量审计师的 IPO 公司（见表 7 的面板 A）。在第 1 和第 3 阶段，IPO 的系数均不显著。因此，没有证据表明在 IPO 年份选择低质量审计师的第 1 和第 3 阶段的 IPO 公司比对照公司更可能在 IPO 后三年内转向高质量审计师。
The coefficient on IPO is significantly negative for the IPOs in Regime 2. This evidence suggests that the IPO firms in Regime 2 that selected a low quality auditor in the IPO year are less likely than the control firms to switch to a high quality auditor three years after the IPO. 在第二阶段，首次公开募股（IPO）的系数显著为负。这一证据表明，在第二阶段选择低质量审计师的 IPO 公司，在 IPO 后第三年更不可能转向高质量审计师，相较于对照公司。
We next examine the IPO firms that select a high quality auditor at the IPO year (see Panel B of Table 7). The coefficients on IPO are always insignificant for each of the three regimes. Therefore, we find no evidence that IPO firms that selected a high quality auditor in the IPO year are more likely than the control firms to switch to a lower quality auditor three years after the IPO. 我们接下来检查在首次公开募股（IPO）年份选择高质量审计师的 IPO 公司（见表 7 的 B 面）。在三种情况下，IPO 的系数始终不显著。因此，我们没有发现证据表明在 IPO 年份选择高质量审计师的 IPO 公司比对照公司更可能在 IPO 后三年内更换为低质量审计师。
Overall, the results in Table 7 strongly suggest that the auditor choice at the IPO time is sticky and therefore will likely have a long lasting impact on IPO firms’ financial reporting quality in the post-IPO period. 总体而言，表 7 中的结果强烈表明，首次公开募股时的审计师选择是固定的，因此可能会对首次公开募股公司在首次公开募股后时期的财务报告质量产生持久影响。
Table 8 表 8
Discretionary accrual management over the three-five year period after the IPO. 首次公开募股后三到五年期间的自由裁量应计管理。
This table presents the OLS regression results of earnings management to avoid reporting losses in the post-IPO period from year t+3t+3 to t+5t+5. The dependent variable is PMDACC. See Appendix A for variable definitions. All the continuous variables are winsorized at the 1st and 99th percentiles. The reported pp-values are two-tailed and clustered by firm. 该表展示了从 t+3t+3 年到 t+5t+5 年在首次公开募股后期间，为避免报告亏损而进行的盈余管理的 OLS 回归结果。因变量为 PMDACC。变量定义见附录 A。所有连续变量在第 1 和第 99 百分位数处进行了温莎化。报告的 pp -值为双尾且按公司聚类。
6.2. Earnings management 6.2. 盈利管理
We next examine whether the firms that went public during Regimes 1 and 3 manage earnings to a greater extent in the three-five years post the IPO than the firms that went public during Regime 2 using the following model: 我们接下来检查在第一和第三阶段上市的公司在首次公开募股后 3 到 5 年内是否比在第二阶段上市的公司更大程度地操纵收益，使用以下模型：
" PMDACC "=beta_(0)+beta_(1)REG1+beta_(2)REG3+beta_(3)REG1xx" LOSS "+beta_(4)REG3xx" LOSS "+beta_(5)" LOSS "+" Controls "+epsi.\text { PMDACC }=\beta_{0}+\beta_{1} R E G 1+\beta_{2} R E G 3+\beta_{3} R E G 1 \times \text { LOSS }+\beta_{4} R E G 3 \times \text { LOSS }+\beta_{5} \text { LOSS }+ \text { Controls }+\varepsilon .
See Appendix A for variable definitions. Both PMDACC and Controls are defined as in model (2) except that they are measured over the three-five year post-IPO period. 请参见附录 A 以获取变量定义。PMDACC 和控制变量的定义与模型（2）相同，只是它们是在首次公开募股后三到五年的期间内测量的。
While corporate insiders have an incentive to manage earnings upward in the pre-IPO years in order to increase the market valuation of the IPO offering, the direction of earnings management in the post-IPO years could be either upward or downward depending on corporate insiders’ incentives. For example, an insider may manage earnings downward (upward) in a post-IPO year if he expects the unmanaged earnings to exceed (miss) relevant benchmarks. Hence, model (4) includes LOSS, a proxy for upward earnings management incentives. Publicly traded Chinese companies would be downgraded by the mainland stock exchanges to the “special treatment” status with restricted trading if they report two consecutive years of losses, and are forced to de-list if they report three consecutive years of losses (Chen et al., 2001). Given these costs to reporting negative earnings, we expect managers of loss firms to have a strong incentive to manage earnings upward to avoid reporting losses, especially in Regimes 1 and 3 where insiders face lower earnings management constraints. Hence, we expect the coefficients on REG1 xx\times LOSS and REG3 xx\times LOSS to be positive. 虽然公司内部人士在首次公开募股（IPO）前的几年有动机向上管理收益，以提高 IPO 发行的市场估值，但在 IPO 后的几年中，收益管理的方向可能是向上或向下，这取决于公司内部人士的动机。例如，如果内部人士预计未管理的收益将超过（未达到）相关基准，他可能会在 IPO 后的一年中向下（向上）管理收益。因此，模型（4）包括 LOSS，作为向上管理收益动机的代理。中国上市公司如果连续两年报告亏损，将被大陆证券交易所降级为“特别处理”状态，限制交易；如果连续三年报告亏损，则被迫退市（陈等，2001）。考虑到报告负收益的成本，我们预计亏损公司的管理者有强烈的动机向上管理收益，以避免报告亏损，特别是在第 1 和第 3 阶段，内部人士面临较低的收益管理约束。因此，我们预计 REG1 xx\times LOSS 和 REG3 xx\times LOSS 的系数为正。
Column (1) of Table 8 reports the regression results of model (4). Consistent with our prediction, the coefficients on REG1 xx\times LOSS and REG3 xx\times LOSS are both positive but only the coefficient on REG3 xx\times LOSS is significant. 表 8 的第(1)列报告了模型(4)的回归结果。与我们的预测一致，REG1 xx\times LOSS 和 REG3 xx\times LOSS 的系数均为正，但只有 REG3 xx\times LOSS 的系数显著。
Regression model (4) assumes all firms who wish to inflate reported earnings have the slacks (Barton and Simko, 2002). However, this assumption may not hold for the firms that have reported two consecutive years of losses. Hence, we also break LOSS firms into firms that have reported only one year of loss (LOSS1) and firms that have reported two consecutive years of losses (LOSS2). As shown in column (2) of Table 8, the coefficients on REG1 xx\times LOSS2 and REG3 xx\times LOSS2 are insignificant, consistent with the hypothesis that these firms have already exhausted the accrual earnings management slacks. On the other hand, the coefficients on both REG1 xx\times LOSS1 and REG3 xx\times LOSS1 are now significantly positive. 回归模型（4）假设所有希望夸大报告收益的公司都有盈余（Barton 和 Simko，2002）。然而，这一假设可能不适用于已经连续报告两年亏损的公司。因此，我们还将亏损公司分为仅报告一年亏损的公司（LOSS1）和连续报告两年亏损的公司（LOSS2）。如表 8 的第（2）列所示，REG1 xx\times LOSS2 和 REG3 xx\times LOSS2 的系数不显著，这与这些公司已经耗尽应计收益管理盈余的假设一致。另一方面，REG1 xx\times LOSS1 和 REG3 xx\times LOSS1 的系数现在显著为正。
6.3. Earnings response coefficient 6.3. 收益响应系数
Prior research (e.g., Francis and Ke, 2006) shows that the earnings response coefficient is lower for firms with lower earnings quality. Hence, we next examine whether the earnings response coefficient in the three-five years after the IPO is lower for the 先前的研究（例如，Francis 和 Ke，2006）表明，收益响应系数在收益质量较低的公司中较低。因此，我们接下来将研究首次公开募股后三到五年内的收益响应系数是否较低。
Table 9 表 9
The earnings response coefficient over the three-five year period after the IPO. 首次公开募股后三到五年期间的收益响应系数。
Variable 变量
Regression A 回归 A
Regression B 回归 B
Coeff. 系数
pp-value pp -值
Coeff. 系数
pp-value pp -值
Intercept 拦截
-0.006-0.006
0.000
-0.007-0.007
0.000
UE
1.141
0.002
0.652
0.001
UE xx\times REG1
-0.388-0.388
0.006
0.014
0.797
UE xx\times REG3 UE xx\times REG3
-0.324-0.324
0.010
-0.003-0.003
0.947
UE xx IPOU E \times I P O
0.229
0.012
UE xx REG1xx IPOU E \times R E G 1 \times I P O
-0.316-0.316
0.011
UE xx REG3xx IPOU E \times R E G 3 \times I P O
-0.264-0.264
0.043
UE xx LNMVU E \times L N M V
-0.036-0.036
0.562
0.016
0.631
UE xxU E \times GROWTH UE xxU E \times 增长
-0.035-0.035
0.154
0.002
0.935
UE xx LEVU E \times L E V
0.699
0.003
-0.158-0.158
0.125
UE xx\times CLOSS UE xx\times CLOSS
0.107
0.284
-0.025-0.025
0.683
UE xx|UE|U E \times|U E|
-2.874-2.874
0.000
-0.332-0.332
0.464
UE xx\times SECONDHALF UE xx\times 下半场
-0.503-0.503
0.000
-0.307-0.307
0.000
UE xx\times STDRET
-9.289-9.289
0.154
-3.756-3.756
0.424
UE xx\times RESTRUCTURE UE xx\times 重组
0.082
0.742
0.216
0.056
UE xx SOEU E \times S O E
0.056
0.588
0.005
0.940
Industry fixed effects 行业固定效应
Included 包含
Included 包含
Adj. R^(2)\mathrm{R}^{2} 形容词 R^(2)\mathrm{R}^{2}
4.70%
2.84%
N
4526
22,497
Variable Regression A Regression B
Coeff. p-value Coeff. p-value
Intercept -0.006 0.000 -0.007 0.000
UE 1.141 0.002 0.652 0.001
UE xx REG1 -0.388 0.006 0.014 0.797
UE xx REG3 -0.324 0.010 -0.003 0.947
UE xx IPO 0.229 0.012
UE xx REG1xx IPO -0.316 0.011
UE xx REG3xx IPO -0.264 0.043
UE xx LNMV -0.036 0.562 0.016 0.631
UE xx GROWTH -0.035 0.154 0.002 0.935
UE xx LEV 0.699 0.003 -0.158 0.125
UE xx CLOSS 0.107 0.284 -0.025 0.683
UE xx|UE| -2.874 0.000 -0.332 0.464
UE xx SECONDHALF -0.503 0.000 -0.307 0.000
UE xx STDRET -9.289 0.154 -3.756 0.424
UE xx RESTRUCTURE 0.082 0.742 0.216 0.056
UE xx SOE 0.056 0.588 0.005 0.940
Industry fixed effects Included Included
Adj. R^(2) 4.70% 2.84%
N 4526 22,497 | Variable | Regression A | | Regression B | |
| :---: | :---: | :---: | :---: | :---: |
| | Coeff. | $p$-value | Coeff. | $p$-value |
| Intercept | $-0.006$ | 0.000 | $-0.007$ | 0.000 |
| UE | 1.141 | 0.002 | 0.652 | 0.001 |
| UE $\times$ REG1 | $-0.388$ | 0.006 | 0.014 | 0.797 |
| UE $\times$ REG3 | $-0.324$ | 0.010 | $-0.003$ | 0.947 |
| $U E \times I P O$ | | | 0.229 | 0.012 |
| $U E \times R E G 1 \times I P O$ | | | $-0.316$ | 0.011 |
| $U E \times R E G 3 \times I P O$ | | | $-0.264$ | 0.043 |
| $U E \times L N M V$ | $-0.036$ | 0.562 | 0.016 | 0.631 |
| $U E \times$ GROWTH | $-0.035$ | 0.154 | 0.002 | 0.935 |
| $U E \times L E V$ | 0.699 | 0.003 | $-0.158$ | 0.125 |
| UE $\times$ CLOSS | 0.107 | 0.284 | $-0.025$ | 0.683 |
| $U E \times\|U E\|$ | $-2.874$ | 0.000 | $-0.332$ | 0.464 |
| UE $\times$ SECONDHALF | $-0.503$ | 0.000 | $-0.307$ | 0.000 |
| UE $\times$ STDRET | $-9.289$ | 0.154 | $-3.756$ | 0.424 |
| UE $\times$ RESTRUCTURE | 0.082 | 0.742 | 0.216 | 0.056 |
| $U E \times S O E$ | 0.056 | 0.588 | 0.005 | 0.940 |
| Industry fixed effects | Included | | Included | |
| Adj. $\mathrm{R}^{2}$ | 4.70% | | 2.84% | |
| N | 4526 | | 22,497 | |
This table reports the OLS regression result of stock price reactions to earnings announcements in the post-IPO period from year t+3t+3 to t+5t+5, where year tt is the IPO year. The dependent variable is the market-adjusted three-day cumulative abnormal return surrounding the semiannual or annual earnings announcement date (CAR). Regression A examines stock price reactions to earnings announcements for the sample IPO firms only. Regression B examines stock price reactions to earnings announcements for our sample IPO firms versus a matched sample of control firms that went public before the beginning of our sample period, i.e., 1997. See Appendix A for variable definitions. All the continuous variables are winsorized at the 1 st and 99 th percentiles. The reported pp-values are two-tailed and clustered by firm ID. 该表报告了在 t+3t+3 年至 t+5t+5 年期间，首次公开募股（IPO）后股票价格对盈利公告反应的 OLS 回归结果，其中 tt 年为 IPO 年份。因变量是围绕半年度或年度盈利公告日期的市场调整三天累计异常收益（CAR）。回归 A 仅考察样本 IPO 公司的股票价格对盈利公告的反应。回归 B 考察我们的样本 IPO 公司与在我们样本期开始之前，即 1997 年，上市的匹配控制公司之间的股票价格对盈利公告的反应。有关变量定义，请参见附录 A。所有连续变量在第 1 和第 99 百分位数处进行了温莎化。报告的 pp -值为双尾并按公司 ID 聚类。
firms that went public during Regimes 1 and 3 than for the firms that went public during Regime 2. Following Francis and Ke (2006), our basic earnings response coefficient model is as follows: 在第一和第三阶段上市的公司，其收益反应系数模型比在第二阶段上市的公司更高。根据 Francis 和 Ke（2006）的研究，我们的基本收益反应系数模型如下：
CAR_(it)=alpha+beta_(1)UE_(it)+beta_(2)UE_(it)xxX_(it)+epsi_(it)C A R_{i t}=\alpha+\beta_{1} U E_{i t}+\beta_{2} U E_{i t} \times X_{i t}+\varepsilon_{i t}
where ii and tt are firm and time subscripts, respectively. See Appendix AA for variable definitions. CAR is the market-adjusted three-trading day cumulative abnormal return centered on a firm’s earnings announcement date. ^(24){ }^{24} Because analysts’ earnings forecasts are not readily available, the earning surprise UEU E is defined using a random walk model. XX represents a set of common control variables used in prior research, including LNMV, GROWTH, LEV, CLOSS, |UE|, SECONDHALF, STDRET, and RESTRUCTURE. We also include SOE in XX to allow the earnings response coefficient to differ for state-controlled firms and non-state-controlled firms. 其中 ii 和 tt 分别是公司和时间的下标。有关变量定义，请参见附录 AA 。CAR 是以公司盈利公告日期为中心的市场调整三交易日累计异常收益。 ^(24){ }^{24} 由于分析师的盈利预测并不容易获得，盈利意外 UEU E 是使用随机游走模型定义的。 XX 代表在先前研究中使用的一组常见控制变量，包括 LNMV、GROWTH、LEV、CLOSS、|UE|、SECONDHALF、STDRET 和 RESTRUCTURE。我们还在 XX 中包含 SOE，以允许国有企业和非国有企业的盈利响应系数有所不同。
One complication that affects model (5) is that China introduced mandatory quarterly reporting in the first quarter of 2002, which falls within our sample period. Prior to that, Chinese listed firms were only required to report earnings semiannually. In order to use one single unified research design for the entire sample period, model (5) uses only the semiannual and annual earnings announcements and UEU E is defined every six months. In addition, for the semiannual earnings announcements that fall in the mandatory quarterly reporting period, to capture the full market effect of UE,CARU E, C A R for the semiannual earnings announcement is defined as the sum of the three-day abnormal returns for both the semiannual announcement and the previous quarter’s announcement. A similar procedure applies to annual earnings announcements that fall in the mandatory quarterly reporting period. 一个影响模型（5）的复杂因素是，中国在 2002 年第一季度引入了强制季度报告，这恰好在我们的样本期内。在此之前，中国上市公司只需每半年报告一次收益。为了在整个样本期内使用一个统一的研究设计，模型（5）仅使用半年和年度收益公告，并且 UEU E 每六个月定义一次。此外，对于在强制季度报告期间的半年收益公告，为了捕捉 UE,CARU E, C A R 的完整市场效应，半年收益公告被定义为半年公告和前一季度公告的三天异常收益之和。对于在强制季度报告期间的年度收益公告也适用类似的程序。
Table 9 shows the regression results of model (5). Since this is a 3-year panel data set, the reported pp-values are clustered by firm. Regression A of Table 9 shows the regression results using only the 784 IPO firms that went public during our sample period 1997-2004. The sample period for each firm is the three years t+3t+3 to t+5t+5 after the IPO year tt, resulting in a total of 4704 表 9 显示了模型（5）的回归结果。由于这是一个为期 3 年的面板数据集，报告的 pp 值是按公司聚类的。表 9 的回归 A 显示了仅使用在我们样本期间 1997-2004 年上市的 784 家 IPO 公司的回归结果。每家公司的样本期是 IPO 年份 tt 后的三年 t+3t+3 到 t+5t+5 ，总共结果为 4704。
maximum firm half-years ( 784 xx3xx2784 \times 3 \times 2 ) over the period 2000-2009.^(25)2000-2009 .{ }^{25} The final sample used in regression A is 4526 due to missing values. The coefficients on UE xxU E \times REG1 and UE xx REG3U E \times R E G 3 are significantly negative, suggesting that earnings quality is lower in the years t+3t+3 to t+5t+5 for the firms that went public in Regimes 1 and 3 than for the firms that went public in Regime 2. 最大公司半年度（ 784 xx3xx2784 \times 3 \times 2 ）在 2000-2009.^(25)2000-2009 .{ }^{25} 期间。由于缺失值，回归 A 中使用的最终样本为 4526。 UE xxU E \times REG1 和 UE xx REG3U E \times R E G 3 的系数显著为负，这表明在 t+3t+3 到 t+5t+5 年间，处于第 1 和第 3 阶段的公司相比于处于第 2 阶段的公司，其盈利质量较低。
One concern with the regression results of Regression A is that the results could be due to contemporaneous confounding factors that occurred during the three years t+3t+3 to t+5t+5 (i.e., 2000-2009). To address this concern, Regression B of Table 9 adds a matched sample of control firms that went public prior to the beginning of Regime 1 (i.e., 1997) for each of the three years t+3t+3 to t+5t+5. For example, for the 784 IPO firms in year t+3t+3 (corresponding to calendar years 2000-2007), we include the control firms in the same calendar period 2000-20072000-2007 as a control. The control firm years are allocated into the three IPO regimes using the same method as in Table 3 (see Section 4.2.1). We repeat the same control sample matching procedures for year t+4t+4 and year t+5t+5 respectively. Our inferences are not affected because the coefficients of interest on UEU Exx\times REG1 xx IPO\times I P O and UE xx REG3xx IPOU E \times R E G 3 \times I P O continue to be significantly negative. Overall, the evidence from Table 9 suggests that the financial reporting quality selected by IPO firms at the time of the IPO has a long lasting impact on the firms’ long-term financial reporting quality. 对回归 A 的回归结果的一个担忧是，这些结果可能是由于在三年 t+3t+3 到 t+5t+5 （即 2000-2009 年）期间发生的同时混杂因素所导致。为了解决这个问题，表 9 的回归 B 为每个三年 t+3t+3 到 t+5t+5 增加了一组在第 1 阶段开始之前（即 1997 年）上市的匹配控制公司样本。例如，对于 t+3t+3 年（对应日历年 2000-2007）的 784 家 IPO 公司，我们在相同的日历期间 2000-20072000-2007 中包含控制公司作为对照。控制公司年份按照与表 3 中相同的方法分配到三个 IPO 阶段（见第 4.2.1 节）。我们分别对 t+4t+4 年和 t+5t+5 年重复相同的控制样本匹配程序。我们的推论没有受到影响，因为对 UEU Exx\times REG1 xx IPO\times I P O 和 UE xx REG3xx IPOU E \times R E G 3 \times I P O 的关注系数仍然显著为负。总体而言，表 9 的证据表明，IPO 公司在 IPO 时选择的财务报告质量对公司的长期财务报告质量有持久影响。
7. Conclusions 7. 结论
Information asymmetry between corporate insiders and public investors is a constant concern for IPO firms, especially for firms domiciled in weak investor protection countries where corporate insiders have stronger incentives to manipulate the reported information included in IPO prospectuses. To protect public investors from purchasing overpriced IPO shares, securities regulators around the world have adopted different approaches to regulating IPO offer prices, ranging from a government-based approach, under which IPO offer prices are directly determined by government regulators, to a market-based approach, under which IPO offer prices are determined by market forces. 公司内部人士与公众投资者之间的信息不对称是 IPO 公司面临的一个持续关注的问题，尤其是在投资者保护较弱的国家，企业内部人士更有动力操纵 IPO 招股说明书中包含的报告信息。为了保护公众投资者免于购买价格过高的 IPO 股票，全球的证券监管机构采取了不同的方法来监管 IPO 发行价格，从政府主导的方法，即 IPO 发行价格由政府监管机构直接确定，到市场主导的方法，即 IPO 发行价格由市场力量决定。
Consistent with the hypothesis that the Chinese securities regulators have a strong aversion to the risk of IPO offer overpricing, we show that the Chinese securities regulators are more likely to depress IPO offer prices relative to reported earnings during Regimes 1 and 3 than during Regime 2. 与中国证券监管机构对 IPO 发行定价过高风险的强烈厌恶假设一致，我们显示在第一和第三阶段，中国证券监管机构更可能压低 IPO 发行价格相对于报告的收益，而不是在第二阶段。
We next examine the likelihood of hiring a low quality auditor and the extent of upward earnings management in the pre-IPO period for the IPO firms in Regimes 1 and 3 versus the IPO firms in Regime 2. We find that IPO firms’ insiders are more likely to hire a high quality auditor, and less likely to inflate reported earnings in the year immediately prior to the IPO in Regime 2 than in Regimes 1 and 3. Contrary to regulators’ concern, we find no evidence that IPOs’ long-term abnormal stock price performance relative to the offer price is lower during Regime 2 than during Regimes 1 and 3. Furthermore, we find that IPO firms’ financial reporting quality choices made at the IPO time are sticky and have a long lasting impact on the firms’ financial reporting quality. Overall, our results suggest that despite China’s weak investor protection and the underdevelopment of financial market supporting institutions during our sample period, a switch to a (quasi) market-based approach is still more effective than the gov-ernment-based approach in motivating IPO firms’ insiders to adopt high quality financial reporting without increasing the risk of IPO overpricing. Furthermore, IPO firms’ reporting choices at the IPO time have long lasting impact on the firms’ future financial reporting. 我们接下来考察在首次公开募股（IPO）前期，第一和第三制度下的 IPO 公司与第二制度下的 IPO 公司聘用低质量审计师的可能性以及向上盈余管理的程度。我们发现，在第二制度下，IPO 公司的内部人士更有可能聘用高质量审计师，并且在 IPO 前一年内更不可能夸大报告的收益，这与第一和第三制度相比。与监管机构的担忧相反，我们没有发现 IPO 在第二制度下相对于发行价格的长期异常股价表现低于第一和第三制度的证据。此外，我们发现 IPO 公司在 IPO 时做出的财务报告质量选择是粘性的，并对公司的财务报告质量产生持久影响。总体而言，我们的结果表明，尽管在我们的样本期间，中国的投资者保护较弱，金融市场支持机构发展不足，但转向（准）市场导向的方法仍然比政府导向的方法更有效，能够激励 IPO 公司的内部人士采用高质量的财务报告，而不会增加 IPO 定价过高的风险。 此外，IPO 公司在首次公开募股时的报告选择对公司未来的财务报告有着持久的影响。
Our findings carry important implications for the ongoing debate on the relative efficiency of the government versus the market in resource allocation. This issue is particularly relevant in many emerging markets such as China, where the domestic financial markets are expected to play a bigger role in transforming the national economy (Cai, 2012). Opinions differ sharply among regulators and investors on the extent of the government’s involvement in regulating financial markets (World Bank, 2012; Davis, 2012). We contribute to this debate by providing empirical evidence on the effects of the government-based approach versus the (quasi) market-based approach in IPO pricing on IPO firms’ financial reporting quality and the risk of IPO overpricing. Because prior research shows that earnings management is costly and lower financial reporting quality is associated with lower investment efficiency (Biddle et al., 2009; Chen et al., 2013), the evidence from our study implies that the market-based IPO offer pricing approach is more effective than the government-based IPO offer pricing approach in improving the IPO market’s resource allocation. 我们的研究结果对政府与市场在资源配置效率方面的持续辩论具有重要意义。这个问题在许多新兴市场中尤为相关，例如中国，国内金融市场预计将在转型国家经济中发挥更大作用（蔡，2012）。监管者和投资者对政府在金融市场监管中的参与程度意见分歧很大（世界银行，2012；戴维斯，2012）。我们通过提供关于政府主导的方法与（准）市场主导的方法在首次公开募股（IPO）定价对 IPO 公司财务报告质量和 IPO 过度定价风险影响的实证证据，为这一辩论做出贡献。由于先前的研究表明，盈余管理是有成本的，且较低的财务报告质量与较低的投资效率相关（比德尔等，2009；陈等，2013），我们研究的证据表明，基于市场的 IPO 报价定价方法在改善 IPO 市场的资源配置方面比基于政府的 IPO 报价定价方法更有效。
Appendix A. Variable definitions 附录 A. 变量定义
Variable 变量
Definition 定义
REG1
=1=1 if the IPO date is between January 1, 1997 and February 11, 1999 (i.e., Regime 1), and 0 otherwise. 如果首次公开募股日期在 1997 年 1 月 1 日和 1999 年 2 月 11 日之间（即，制度 1），否则为 0。
REG3
=1=1 if the IPO date is between November 7, 2001 and December 31, 2004 (i.e., Regime 3), and 0 otherwise. 如果首次公开募股日期在 2001 年 11 月 7 日到 2004 年 12 月 31 日之间（即，制度 3），否则为 0。
IPO
=1=1 if an observation is for an IPO company and 0 otherwise. 如果观察是针对 IPO 公司的，则为 =1=1 ，否则为 0。
SL
=1=1 if an IPO company employs a small local auditor, and 0 otherwise; an auditor is considered a small local one if it is domiciled in the same province (or equivalent in China) as the IPO client and is not ranked among the top quintile of auditors based on audited non-IPO client assets at the year end. 如果一家 IPO 公司聘请了一个小型本地审计师，则为 =1=1 ，否则为 0；如果审计师与 IPO 客户位于同一省（或中国的同等地区），并且在年末审计的非 IPO 客户资产中未排名前五分之一，则被视为小型本地审计师。
PMDACC
Signed performance-matched modified Jones Model discretionary accruals (Kothari et al., 2005). See Appendix B for the details. 签署的与业绩匹配的修订琼斯模型自由裁量应计项目（Kothari 等，2005）。详细信息请参见附录 B。
NCROA
Signed non-operating earnings scaled by lagged total assets. 签署的非经营性收益按滞后总资产进行调整。
SALESGROWTH
A firm's average sales growth rate in the two years prior to the IPO year. 一家企业在首次公开募股（IPO）年份前两年的平均销售增长率。
ADJ_SALESGROWTH
SALESGROWTH for an IPO firm minus the median SALESGROWTH for the seasoned firms in the same two-digit industry code or in the same one-digit industry code if the number of eligible seasoned firms for a two-digit industry is < 3<3. IPO 公司的销售增长减去同一两位数行业代码或在同一位数行业代码中，如果两位数行业的合格成熟公司的数量为 < 3<3 ，则为成熟公司的销售增长的中位数。
ASSETS
The natural logarithm of total assets. 总资产的自然对数。
LEV
The ratio of total liability over total assets. 总负债与总资产的比率。
ROA
Net income over total assets. 净资产收益率。
VOLATILITY
The standard deviation of the market model residual. The market model is estimated over the 52 weeks, starting from one month after the IPO listing; in the market model, weekly returns of a sample firm are regressed on the contemporaneous and lagged weekly market returns. 市场模型残差的标准差。市场模型是在首次公开募股上市后一个月开始的 52 周内估计的；在市场模型中，样本公司的每周收益与同期和滞后的每周市场收益进行回归。
CURRENT
The ratio of current assets over current liabilities. 流动资产与流动负债的比率。
ARINV
The ratio of the sum of account receivables and inventories over total assets. 应收账款和存货总和与总资产的比率。
GROWTH
Tobin's Q. For an IPO firm, GROWTH is computed as the sum of the market value of equity owned by the pre-IPO shareholders at the end of the 60th trading day and the pre-IPO book value of liabilities, divided by the pre-IPO total assets. For a seasoned firm, GROWTH is defined as the sum of the market value of equity and the book value of liabilities divided by total assets measured at the end of fiscal year. 托宾 Q。对于首次公开募股（IPO）公司，GROWTH 的计算方法是将 IPO 前股东在第 60 个交易日结束时所拥有的市场价值和 IPO 前负债的账面价值相加，然后除以 IPO 前的总资产。对于成熟公司，GROWTH 的定义是将市场价值和负债的账面价值相加，然后除以在财政年度结束时的总资产。
ADJ_GROWTH
Market adjusted Tobin's Q for an IPO firm, defined as IPO firm's Q (GROWTH) minus the contemporaneous median Q of the seasoned firms in the market. The seasoned firms' QQ is computed as the sum of the market value of equity and the book value of liabilities divided by total assets measured at the beginning of fiscal year. 市场调整后的托宾 Q 值用于首次公开募股（IPO）公司，定义为 IPO 公司的 Q（增长）减去市场中成熟公司的同期中位数 Q。成熟公司的 QQ 计算为股本市场价值与负债账面价值之和除以在财年开始时测量的总资产。
PROCEEDS
The natural logarithm of the gross proceeds from the IPO. 首次公开募股总收益的自然对数。
RETAINED_OWN
The retained stock ownership by the pre-IPO owners, measured as the number of pre-IPO shares divided by the number of outstanding shares after the IPO. 预 IPO 所有者保留的股票所有权，衡量标准为预 IPO 股份数量与 IPO 后流通股份数量的比值。
OCF
Operating cash flows in the year immediately prior to the IPO year, deflated by lagged total assets; For earlier IPOs when cash flow statements are not available, operating cash flow = operating income - total accruals (TA), where TAT A is defined using the balance sheet approach as follows: ( Delta\Delta Current assets -Delta-\Delta Cash -Delta-\Delta Short-term lending) - ( Delta\Delta Current liability -Delta-\Delta Short term borrowing -Delta-\Delta Current portion of long-term debt) - Depreciation and Amortization expense. 在首次公开募股（IPO）前一年，经营现金流量经过滞后总资产的调整；对于早期的 IPO，当现金流量表不可用时，经营现金流量 = 经营收入 - 总应计（TA），其中 TAT A 使用资产负债表方法定义如下：（ Delta\Delta 流动资产 -Delta-\Delta 现金 -Delta-\Delta 短期借款） - （ Delta\Delta 流动负债 -Delta-\Delta 短期借款 -Delta-\Delta 长期债务的当前部分） - 折旧和摊销费用。
SOE
=1=1 if the largest shareholder is a government institution and controls at least 20%20 \% of the outstanding common shares, and 0 otherwise. 如果最大股东是政府机构并控制至少 20%20 \% 的流通普通股，则为 =1=1 ，否则为 0。
MARKETIZATION
The annual provincial marketization index developed by Fan et al. (2010), with a higher index value representing more developed market institutions. The index is based on the following five sub-indices: the relationship between the governments and markets, the growth of the private sector, the development of product markets, the development of factor markets, and the development of market intermediaries and legal institution environment. 由范等人（2010）开发的年度省级市场化指数，指数值越高表示市场制度越发达。该指数基于以下五个子指数：政府与市场的关系、私营部门的增长、产品市场的发展、要素市场的发展，以及市场中介和法律制度环境的发展。
HHI
The Herfindahl-Hirschman Index of the audit market for the province (or its equivalent in China) in which an IPO firm is domiciled, defined as sums_(i)^(2)\sum s_{i}^{2}, where s_(i)s_{i} is the annual market share of audit firm ii in the province measured at the end of a year using the audited client assets. IPO firms are excluded in computing s_(i)s_{i} to mitigate the effect of new IPO clients on the market share measure. 审计市场的赫芬达尔-赫希曼指数是指 IPO 公司注册所在省（或中国的等效省份）的 sums_(i)^(2)\sum s_{i}^{2} ，其中 s_(i)s_{i} 是审计公司 ii 在该省的年度市场份额，按年末使用审计客户资产进行测量。在计算 s_(i)s_{i} 时排除 IPO 公司，以减轻新 IPO 客户对市场份额测量的影响。
GDP
The natural logarithm of the Gross Domestic Product in the province (or equivalent in China) in which a firm is domiciled. 该公司注册所在省（或中国的同等地区）的国内生产总值的自然对数。
GDP Growth Rate 国内生产总值增长率
The average GDP growth rate for the province (or equivalent in China) in which a firm is domiciled in the two years prior to the IPO year. 公司注册所在省（或中国的同等地区）在首次公开募股（IPO）年份前两年的平均 GDP 增长率。
Credit Market Size 信用市场规模
The natural logarithm of the Gross Domestic Product contributed by the financial industry in the province (or equivalent in China) in which a firm is domiciled. 该公司注册所在省（或中国的同等地区）金融行业对国内生产总值的自然对数。
FDI
The natural logarithm of foreign direct investment received by the province (or equivalent in China) in which a firm is domiciled. 该公司注册所在省（或中国等效地区）收到的外商直接投资的自然对数。
Fiscal Deficit 财政赤字
The ratio of fiscal expenditure over fiscal revenue in the province (or equivalent in China) in which a firm is domiciled. 该公司注册所在省（或中国的同等地区）财政支出与财政收入的比率。
Credit Marketization 信用市场化
The annual provincial credit marketization index developed by Fan et al. (2010). It is measured as the percentage of deposits taken by non-state financial institutions and the percentage of short-term loans to the non-state sector. 由范等人（2010）开发的年度省级信用市场化指数。它的测量方式是非国有金融机构所吸收的存款占比和非国有部门短期贷款占比。
DIFFERENT_AUDITOR
=1=1 if an IPO firm or a matched non-IPO firm selects a low quality auditor (i.e., SL=1S L=1 ) at the IPO time but a high quality auditor (i.e., SL=0S L=0 ) three years after the IPO or vice versa, and 0 otherwise. 如果一家首次公开募股（IPO）公司或一家匹配的非 IPO 公司在 IPO 时选择了低质量审计师（即 SL=1S L=1 ），但在 IPO 后三年选择了高质量审计师（即 SL=0S L=0 ），或者反之，则为 0。
Audit firm experiences a merger 审计公司经历了一次合并
=1=1 if an IPO firm's audit firm experienced a merger event during the period between the IPO year and the earlier of the third year after the IPO or the year of an audit firm switch, and 0 otherwise. 如果一家 IPO 公司的审计公司在 IPO 年份与 IPO 后第三年或审计公司更换年份之间发生合并事件，则为 =1=1 ，否则为 0。
Audit firm is closed by regulators 审计公司被监管机构关闭
=1=1 if an IPO firm's audit firm was forced to be closed by the regulators during the period between the IPO year and the earlier of the third year after the IPO or the year of an audit firm switch, and 0 otherwise. 如果一家 IPO 公司的审计公司在 IPO 年份与 IPO 后第三年之前或审计公司更换年份之间的期间被监管机构迫使关闭，则为 =1=1 ，否则为 0。
LOSS
=1=1 if a firm reported a loss in the previous year, and 0 otherwise. The sum of LOSS1 and LOSS2 is LOSS. 如果一家公司在前一年报告了亏损，则为 =1=1 ，否则为 0。LOSS1 和 LOSS2 的总和为 LOSS。
LOSS1
=1=1 if a firm reported a loss in the previous year but a profit two years before, and 0 otherwise. 如果一家公司在前一年报告了亏损，但在两年前盈利，则为 =1=1 ，否则为 0。
LOSS2
=1=1 if a firm reported a loss in the previous two consecutive years, and 0 otherwise. 如果一家公司在前两年连续报告亏损，则为 =1=1 ，否则为 0。
CAR
The three-trading day market-adjusted cumulative abnormal return centered on the earnings announcement date. 以盈利公告日期为中心的三交易日市场调整累计异常收益。
UE
Unexpected earnings for each half year, determined as the difference between the operating income per share for a half year and that for the corresponding same half year in the previous year, scaled by the stock price at the end of the previous half year. 每半年的意外收益，定义为该半年的每股营业收入与前一年同半年的每股营业收入之间的差额，按前一个半年的末尾股价进行调整。
LNMV
Natural log of market value of equity at the beginning of each half year. 每个半年的开始时，股权市场价值的自然对数。
CLOSS
=1=1 if a firm reports a loss in the current year, and 0 otherwise. 如果一家公司在当前年度报告亏损，则为 =1=1 ，否则为 0。
|UE||U E|
The absolute value of UEU E. UEU E 的绝对值。
SECONDHALF
=1=1 for the second half of a fiscal year, and 0 otherwise. =1=1 在财政年度的下半年为 0，其他情况下为 0。
STDRET
The standard deviation of the market-adjusted daily stock returns over a 180-day window ending seven days prior to the earnings announcement date and with at least 10 non-missing daily returns. 市场调整后的每日股票收益的标准差，计算范围为截至盈利公告日期前七天的 180 天窗口，并且至少有 10 个非缺失的每日收益。
RESTRUCTURE
=1=1 if the non-operating income as a percentage of total assets is -5%-5 \% or below, and 0 otherwise. 如果非经营收入占总资产的比例为 -5%-5 \% 或以下，则为 =1=1 ，否则为 0。
Variable Definition
REG1 =1 if the IPO date is between January 1, 1997 and February 11, 1999 (i.e., Regime 1), and 0 otherwise.
REG3 =1 if the IPO date is between November 7, 2001 and December 31, 2004 (i.e., Regime 3), and 0 otherwise.
IPO =1 if an observation is for an IPO company and 0 otherwise.
SL =1 if an IPO company employs a small local auditor, and 0 otherwise; an auditor is considered a small local one if it is domiciled in the same province (or equivalent in China) as the IPO client and is not ranked among the top quintile of auditors based on audited non-IPO client assets at the year end.
PMDACC Signed performance-matched modified Jones Model discretionary accruals (Kothari et al., 2005). See Appendix B for the details.
NCROA Signed non-operating earnings scaled by lagged total assets.
SALESGROWTH A firm's average sales growth rate in the two years prior to the IPO year.
ADJ_SALESGROWTH SALESGROWTH for an IPO firm minus the median SALESGROWTH for the seasoned firms in the same two-digit industry code or in the same one-digit industry code if the number of eligible seasoned firms for a two-digit industry is < 3.
ASSETS The natural logarithm of total assets.
LEV The ratio of total liability over total assets.
ROA Net income over total assets.
VOLATILITY The standard deviation of the market model residual. The market model is estimated over the 52 weeks, starting from one month after the IPO listing; in the market model, weekly returns of a sample firm are regressed on the contemporaneous and lagged weekly market returns.
CURRENT The ratio of current assets over current liabilities.
ARINV The ratio of the sum of account receivables and inventories over total assets.
GROWTH Tobin's Q. For an IPO firm, GROWTH is computed as the sum of the market value of equity owned by the pre-IPO shareholders at the end of the 60th trading day and the pre-IPO book value of liabilities, divided by the pre-IPO total assets. For a seasoned firm, GROWTH is defined as the sum of the market value of equity and the book value of liabilities divided by total assets measured at the end of fiscal year.
ADJ_GROWTH Market adjusted Tobin's Q for an IPO firm, defined as IPO firm's Q (GROWTH) minus the contemporaneous median Q of the seasoned firms in the market. The seasoned firms' Q is computed as the sum of the market value of equity and the book value of liabilities divided by total assets measured at the beginning of fiscal year.
PROCEEDS The natural logarithm of the gross proceeds from the IPO.
RETAINED_OWN The retained stock ownership by the pre-IPO owners, measured as the number of pre-IPO shares divided by the number of outstanding shares after the IPO.
OCF Operating cash flows in the year immediately prior to the IPO year, deflated by lagged total assets; For earlier IPOs when cash flow statements are not available, operating cash flow = operating income - total accruals (TA), where TA is defined using the balance sheet approach as follows: ( Delta Current assets -Delta Cash -Delta Short-term lending) - ( Delta Current liability -Delta Short term borrowing -Delta Current portion of long-term debt) - Depreciation and Amortization expense.
SOE =1 if the largest shareholder is a government institution and controls at least 20% of the outstanding common shares, and 0 otherwise.
MARKETIZATION The annual provincial marketization index developed by Fan et al. (2010), with a higher index value representing more developed market institutions. The index is based on the following five sub-indices: the relationship between the governments and markets, the growth of the private sector, the development of product markets, the development of factor markets, and the development of market intermediaries and legal institution environment.
HHI The Herfindahl-Hirschman Index of the audit market for the province (or its equivalent in China) in which an IPO firm is domiciled, defined as sums_(i)^(2), where s_(i) is the annual market share of audit firm i in the province measured at the end of a year using the audited client assets. IPO firms are excluded in computing s_(i) to mitigate the effect of new IPO clients on the market share measure.
GDP The natural logarithm of the Gross Domestic Product in the province (or equivalent in China) in which a firm is domiciled.
GDP Growth Rate The average GDP growth rate for the province (or equivalent in China) in which a firm is domiciled in the two years prior to the IPO year.
Credit Market Size The natural logarithm of the Gross Domestic Product contributed by the financial industry in the province (or equivalent in China) in which a firm is domiciled.
FDI The natural logarithm of foreign direct investment received by the province (or equivalent in China) in which a firm is domiciled.
Fiscal Deficit The ratio of fiscal expenditure over fiscal revenue in the province (or equivalent in China) in which a firm is domiciled.
Credit Marketization The annual provincial credit marketization index developed by Fan et al. (2010). It is measured as the percentage of deposits taken by non-state financial institutions and the percentage of short-term loans to the non-state sector.
DIFFERENT_AUDITOR =1 if an IPO firm or a matched non-IPO firm selects a low quality auditor (i.e., SL=1 ) at the IPO time but a high quality auditor (i.e., SL=0 ) three years after the IPO or vice versa, and 0 otherwise.
Audit firm experiences a merger =1 if an IPO firm's audit firm experienced a merger event during the period between the IPO year and the earlier of the third year after the IPO or the year of an audit firm switch, and 0 otherwise.
Audit firm is closed by regulators =1 if an IPO firm's audit firm was forced to be closed by the regulators during the period between the IPO year and the earlier of the third year after the IPO or the year of an audit firm switch, and 0 otherwise.
LOSS =1 if a firm reported a loss in the previous year, and 0 otherwise. The sum of LOSS1 and LOSS2 is LOSS.
LOSS1 =1 if a firm reported a loss in the previous year but a profit two years before, and 0 otherwise.
LOSS2 =1 if a firm reported a loss in the previous two consecutive years, and 0 otherwise.
CAR The three-trading day market-adjusted cumulative abnormal return centered on the earnings announcement date.
UE Unexpected earnings for each half year, determined as the difference between the operating income per share for a half year and that for the corresponding same half year in the previous year, scaled by the stock price at the end of the previous half year.
LNMV Natural log of market value of equity at the beginning of each half year.
CLOSS =1 if a firm reports a loss in the current year, and 0 otherwise.
|UE| The absolute value of UE.
SECONDHALF =1 for the second half of a fiscal year, and 0 otherwise.
STDRET The standard deviation of the market-adjusted daily stock returns over a 180-day window ending seven days prior to the earnings announcement date and with at least 10 non-missing daily returns.
RESTRUCTURE =1 if the non-operating income as a percentage of total assets is -5% or below, and 0 otherwise.| Variable | Definition |
| :---: | :---: |
| REG1 | $=1$ if the IPO date is between January 1, 1997 and February 11, 1999 (i.e., Regime 1), and 0 otherwise. |
| REG3 | $=1$ if the IPO date is between November 7, 2001 and December 31, 2004 (i.e., Regime 3), and 0 otherwise. |
| IPO | $=1$ if an observation is for an IPO company and 0 otherwise. |
| SL | $=1$ if an IPO company employs a small local auditor, and 0 otherwise; an auditor is considered a small local one if it is domiciled in the same province (or equivalent in China) as the IPO client and is not ranked among the top quintile of auditors based on audited non-IPO client assets at the year end. |
| PMDACC | Signed performance-matched modified Jones Model discretionary accruals (Kothari et al., 2005). See Appendix B for the details. |
| NCROA | Signed non-operating earnings scaled by lagged total assets. |
| SALESGROWTH | A firm's average sales growth rate in the two years prior to the IPO year. |
| ADJ_SALESGROWTH | SALESGROWTH for an IPO firm minus the median SALESGROWTH for the seasoned firms in the same two-digit industry code or in the same one-digit industry code if the number of eligible seasoned firms for a two-digit industry is $<3$. |
| ASSETS | The natural logarithm of total assets. |
| LEV | The ratio of total liability over total assets. |
| ROA | Net income over total assets. |
| VOLATILITY | The standard deviation of the market model residual. The market model is estimated over the 52 weeks, starting from one month after the IPO listing; in the market model, weekly returns of a sample firm are regressed on the contemporaneous and lagged weekly market returns. |
| CURRENT | The ratio of current assets over current liabilities. |
| ARINV | The ratio of the sum of account receivables and inventories over total assets. |
| GROWTH | Tobin's Q. For an IPO firm, GROWTH is computed as the sum of the market value of equity owned by the pre-IPO shareholders at the end of the 60th trading day and the pre-IPO book value of liabilities, divided by the pre-IPO total assets. For a seasoned firm, GROWTH is defined as the sum of the market value of equity and the book value of liabilities divided by total assets measured at the end of fiscal year. |
| ADJ_GROWTH | Market adjusted Tobin's Q for an IPO firm, defined as IPO firm's Q (GROWTH) minus the contemporaneous median Q of the seasoned firms in the market. The seasoned firms' $Q$ is computed as the sum of the market value of equity and the book value of liabilities divided by total assets measured at the beginning of fiscal year. |
| PROCEEDS | The natural logarithm of the gross proceeds from the IPO. |
| RETAINED_OWN | The retained stock ownership by the pre-IPO owners, measured as the number of pre-IPO shares divided by the number of outstanding shares after the IPO. |
| OCF | Operating cash flows in the year immediately prior to the IPO year, deflated by lagged total assets; For earlier IPOs when cash flow statements are not available, operating cash flow = operating income - total accruals (TA), where $T A$ is defined using the balance sheet approach as follows: ( $\Delta$ Current assets $-\Delta$ Cash $-\Delta$ Short-term lending) - ( $\Delta$ Current liability $-\Delta$ Short term borrowing $-\Delta$ Current portion of long-term debt) - Depreciation and Amortization expense. |
| SOE | $=1$ if the largest shareholder is a government institution and controls at least $20 \%$ of the outstanding common shares, and 0 otherwise. |
| MARKETIZATION | The annual provincial marketization index developed by Fan et al. (2010), with a higher index value representing more developed market institutions. The index is based on the following five sub-indices: the relationship between the governments and markets, the growth of the private sector, the development of product markets, the development of factor markets, and the development of market intermediaries and legal institution environment. |
| HHI | The Herfindahl-Hirschman Index of the audit market for the province (or its equivalent in China) in which an IPO firm is domiciled, defined as $\sum s_{i}^{2}$, where $s_{i}$ is the annual market share of audit firm $i$ in the province measured at the end of a year using the audited client assets. IPO firms are excluded in computing $s_{i}$ to mitigate the effect of new IPO clients on the market share measure. |
| GDP | The natural logarithm of the Gross Domestic Product in the province (or equivalent in China) in which a firm is domiciled. |
| GDP Growth Rate | The average GDP growth rate for the province (or equivalent in China) in which a firm is domiciled in the two years prior to the IPO year. |
| Credit Market Size | The natural logarithm of the Gross Domestic Product contributed by the financial industry in the province (or equivalent in China) in which a firm is domiciled. |
| FDI | The natural logarithm of foreign direct investment received by the province (or equivalent in China) in which a firm is domiciled. |
| Fiscal Deficit | The ratio of fiscal expenditure over fiscal revenue in the province (or equivalent in China) in which a firm is domiciled. |
| Credit Marketization | The annual provincial credit marketization index developed by Fan et al. (2010). It is measured as the percentage of deposits taken by non-state financial institutions and the percentage of short-term loans to the non-state sector. |
| DIFFERENT_AUDITOR | $=1$ if an IPO firm or a matched non-IPO firm selects a low quality auditor (i.e., $S L=1$ ) at the IPO time but a high quality auditor (i.e., $S L=0$ ) three years after the IPO or vice versa, and 0 otherwise. |
| Audit firm experiences a merger | $=1$ if an IPO firm's audit firm experienced a merger event during the period between the IPO year and the earlier of the third year after the IPO or the year of an audit firm switch, and 0 otherwise. |
| Audit firm is closed by regulators | $=1$ if an IPO firm's audit firm was forced to be closed by the regulators during the period between the IPO year and the earlier of the third year after the IPO or the year of an audit firm switch, and 0 otherwise. |
| LOSS | $=1$ if a firm reported a loss in the previous year, and 0 otherwise. The sum of LOSS1 and LOSS2 is LOSS. |
| LOSS1 | $=1$ if a firm reported a loss in the previous year but a profit two years before, and 0 otherwise. |
| LOSS2 | $=1$ if a firm reported a loss in the previous two consecutive years, and 0 otherwise. |
| CAR | The three-trading day market-adjusted cumulative abnormal return centered on the earnings announcement date. |
| UE | Unexpected earnings for each half year, determined as the difference between the operating income per share for a half year and that for the corresponding same half year in the previous year, scaled by the stock price at the end of the previous half year. |
| LNMV | Natural log of market value of equity at the beginning of each half year. |
| CLOSS | $=1$ if a firm reports a loss in the current year, and 0 otherwise. |
| $\|U E\|$ | The absolute value of $U E$. |
| SECONDHALF | $=1$ for the second half of a fiscal year, and 0 otherwise. |
| STDRET | The standard deviation of the market-adjusted daily stock returns over a 180-day window ending seven days prior to the earnings announcement date and with at least 10 non-missing daily returns. |
| RESTRUCTURE | $=1$ if the non-operating income as a percentage of total assets is $-5 \%$ or below, and 0 otherwise. |
ش. We wish to thank an anonymous reviewer, Jeffry Netter (the editor), Sudipta Basu, Francois Brochet, Karthik Ramanna, Siew Hong Teoh, Gwen Yu, Amy Zang, Huai Zhang, and workshop participants at University of California at Irvine, Harvard University, University of International Business and Economics, Shanghai University of Finance and Economics, University of Toronto, Tsinghua University, Xiamen University, China Journal of Accounting Research 2013 Symposium (Zhuhai, China), and the MIT Asia Conference in Accounting for helpful comments. We thank Jaywon Lee for helping us understand the evolution of Korea’s IPO regulatory regimes and Douglas Miller for help on the implementation of the wild cluster bootstrap method. Jun Chen acknowledges the financial support from the National Nature Science Foundation of China (approval numbers 71572181 and 71102085). 我们感谢一位匿名审稿人、编辑杰弗里·内特、苏迪普塔·巴苏、弗朗索瓦·布罗谢、卡尔蒂克·拉曼纳、谢宏·张、关韵、艾米·臧、怀张，以及加州大学欧文分校、哈佛大学、对外经济贸易大学、上海财经大学、多伦多大学、清华大学、厦门大学、中国会计研究杂志 2013 年研讨会（珠海，中国）和麻省理工学院亚洲会计会议的与会者们提供的宝贵意见。我们感谢李在元帮助我们理解韩国首次公开募股监管制度的演变，以及道格拉斯·米勒在实施野生集群自助法方面的帮助。陈俊感谢中国国家自然科学基金（批准号 71572181 和 71102085）的资助。
^(1){ }^{1} We performed a review of IPO offer price regulations across six representative Asian economies. We find that government regulators played a direct role in IPO offer pricing in India prior to 1992 (Sheokand, 2012), in Korea prior to 2002 (Kim et al., 2004) and in Malaysia prior to 1996 (Mohd, 2004). As of today, regulators in China, Japan and Taiwan continue to impose various degrees of regulatory intervention in IPO offer price determination (Chang et al., 2014). 我们对六个代表性亚洲经济体的首次公开募股（IPO）发行价格法规进行了审查。我们发现，在 1992 年之前，印度的政府监管机构在 IPO 发行定价中发挥了直接作用（Sheokand，2012）；在 2002 年之前，韩国的政府监管机构也发挥了类似作用（Kim et al.，2004）；在 1996 年之前，马来西亚的政府监管机构同样如此（Mohd，2004）。截至目前，中国、日本和台湾的监管机构仍然在 IPO 发行价格的确定中施加不同程度的监管干预（Chang et al.，2014）。
^(2){ }^{2} Specifically, in January 2004, the TCL Group was merged into its publicly listed subsidiary, the TCL Communication, by issuing new shares to the public. This was incorrectly coded as an IPO in CSMAR. ^(2){ }^{2} 具体来说，在 2004 年 1 月，TCL 集团通过向公众发行新股与其上市子公司 TCL 通信合并。这在 CSMAR 中被错误地编码为首次公开募股（IPO）。 ^(3){ }^{3} Interestingly, the requirement of using a fixed PE multiple was never formally stipulated in any regulation, but this was a well-understood regulatory practice as noted in prior research (DeFond et al., 2000; Kao et al., 2009). 有趣的是，使用固定市盈率的要求从未在任何法规中正式规定，但正如之前的研究所指出的，这是一种被广泛理解的监管实践（DeFond 等，2000；Kao 等，2009）。 ^(4){ }^{4} The additional IPO disclosure requirements implemented in Regime 2 continue to hold during Regime 3 and hence cannot explain the variation of our regression results across the three regimes, especially the reversal of results in Regime 3. 在第 3 阶段，实施的第 2 阶段额外 IPO 披露要求仍然有效，因此无法解释我们在三个阶段之间回归结果的变化，特别是在第 3 阶段结果的反转。
^(5){ }^{5} During a conference call on May 22, 2009, a CSRC official confirmed the adoption of the government-based approach during Regime 3 (http://news.xinhuanet.com/ fortune/2009-05/22/content_11420922.htm). In an internal research report by the Shanghai Stock Exchange, Meng et al. (2004) also suggest that starting from late 2001, the CSRC reverted back to the government-based approach by capping IPOs’ maximum PE multiple at 20. ^(5){ }^{5} 在 2009 年 5 月 22 日的电话会议上，证监会官员确认在第三阶段采用了政府主导的方法 (http://news.xinhuanet.com/ fortune/2009-05/22/content_11420922.htm)。在上海证券交易所的一份内部研究报告中，孟等人（2004）也建议，从 2001 年底开始，证监会通过将 IPO 的最高市盈率限制在 20，重新回归政府主导的方法。
^(6){ }^{6} Was the CSRC’s perceived risk of IPO blowups warranted? To answer this question, we examine the stock price performance of IPOs launched over the six-month period ending on November 6, 2001, the ending date of Regime 2, from the perspective of the regulator who faced the task of making a decision of whether to switch to Regime 3 or not on November 6, 2001. For the 29 IPOs listed during this six-month window, we find that the mean cumulative raw stock return from the offer price up to November 5, 2001 is a positive 123%. In fact, there is only one IPO with a negative return during this window. Thus, the CSRC’s switch from Regime 2 to Regime 3 appears to be driven by a perceived future IPO blowup risk resulting from the broad stock market’s weak performance rather than the poor past showing of IPOs per se. 中国证监会对首次公开募股（IPO）崩盘风险的担忧是否合理？为了解答这个问题，我们从监管者的角度考察了截至 2001 年 11 月 6 日（即第二阶段结束日期）六个月内推出的 IPO 的股价表现，以评估其是否应在 2001 年 11 月 6 日决定是否切换到第三阶段。在这六个月内上市的 29 个 IPO 中，我们发现截至 2001 年 11 月 5 日，从发行价到现在的平均累计原始股票回报率为正的 123%。实际上，在此期间只有一个 IPO 出现了负回报。因此，中国证监会从第二阶段切换到第三阶段似乎是由于对未来 IPO 崩盘风险的担忧，这种担忧源于整体股市表现疲软，而不是 IPO 本身的过往表现不佳。
^(7){ }^{7} Could firm insiders use earnings management to cause an overvaluation of IPO shares in Regime 1? We believe this is unlikely assuming public investors have the ability to price protect (see our discussion about Regime 2’s earnings management incentives below). ^(7){ }^{7} 公司内部人士是否可以利用盈余管理导致第一阶段首次公开募股（IPO）股票的高估？我们认为这不太可能，前提是公众投资者有能力进行价格保护（请参见我们下面关于第二阶段盈余管理激励的讨论）。 _(8)^(8){ }_{8}^{8} As a robustness check, we also consider earnings management in year t-2t-2 and obtain similar inferences, though the sample size for year t-2t-2 is smaller due to missing data (untabulated). 作为稳健性检验，我们还考虑了 t-2t-2 年的盈余管理，并得出了类似的结论，尽管由于缺失数据（未列出）， t-2t-2 年的样本量较小。 ^(9){ }^{9} One may argue that the effects of H1-H2 should be stronger for growth firms because these firms suffer more from the CSRC’s depressing of IPO offer prices in Regimes 1 and 3. We find no consistent evidence supporting this prediction. One potential explanation for this no result is that the CSRC’s prescribed PE ratios in Regimes 1 and 3 were so low (see Fig. 1) that all firms had an incentive to inflate their earnings to the maximum possible level in the pre-IPO period. ^(9){ }^{9} 有人可能会争辩说，H1-H2 的影响对于成长型公司应该更强，因为这些公司在第 1 和第 3 阶段受到证监会压低 IPO 发行价格的影响更大。我们没有找到支持这一预测的一致证据。一个可能的解释是，在第 1 和第 3 阶段，证监会规定的市盈率如此之低（见图 1），以至于所有公司都有动力在 IPO 前期将其收益膨胀到最大可能水平。
^(10){ }^{10} As a benchmark, the mean and standard deviation for the U.S. IPOs in the same years were 32.68%32.68 \% and 24.83%24.83 \% respectively (untabulated based on our own calculation). It is obvious that the standard deviation is much larger in the U.S., even though the means are similar for the two markets. 作为基准，同年美国首次公开募股（IPO）的均值和标准差分别为 32.68%32.68 \% 和 24.83%24.83 \% （根据我们自己的计算未列出）。显然，美国的标准差要大得多，尽管两个市场的均值相似。 ^(11){ }^{11} One partner of a large domestic audit firm told us that an IPO audit typically takes between two to six months. To allow for the possibility that IPO firms may not have enough time to change auditors in response to the sudden changes of Regimes 2 and 3, we also perform a robustness check by delaying the starting dates of Regimes 2 and 3 by six months. Untabulated regression results show that none of our inferences are affected using this alternative cutoff. 一家大型国内审计公司的合伙人告诉我们，首次公开募股（IPO）审计通常需要两到六个月。为了考虑到 IPO 公司可能没有足够的时间在应对第 2 和第 3 阶段的突然变化时更换审计师，我们还通过将第 2 和第 3 阶段的开始日期推迟六个月来进行稳健性检验。未列出的回归结果显示，使用这一替代截止日期，我们的推论没有受到影响。 ^(12){ }^{12} We do not control for foreign ownership as in DeFond et al. (2000) because we exclude IPO firms with outstanding H or B shares and thus foreign ownership in our sample IPO firms is trivial. 我们没有像 DeFond 等人（2000）那样控制外资持股，因为我们排除了拥有 H 股或 B 股的 IPO 公司，因此我们样本中的 IPO 公司外资持股微不足道。
^(13){ }^{13} Because the number of years in our sample is relatively small, as a robustness check, we also follow Cameron et al. (2008) and Cameron and Miller (2011) by computing the bootstrapped pp-values for all of our regression results and obtain similar inferences. Following the recommendation of Cameron et al. (2008), we use the wild cluster bootstrap-t method. ^(13){ }^{13} 由于我们样本中的年份数量相对较少，作为稳健性检验，我们还遵循 Cameron 等人（2008）和 Cameron 与 Miller（2011）的做法，计算所有回归结果的自助法 pp -值，并得出类似的推论。根据 Cameron 等人（2008）的建议，我们使用野生聚类自助 t 方法。
^(14){ }^{14} With an aim to increase audit quality, Chinese regulators encouraged small local auditors in mainland China that have the license to audit listed firms to merge with either each other or larger auditors in 2000 and 2001. Chan and Wu (2011) show that such audit firm mergers during this period result in increased auditor independence. To make sure that our results in Table 3 are not driven by such merger-related effects, we also rerun the regressions in Table 3 after excluding all the IPOs whose hired auditors happened to have experienced such a merger in the year prior to the IPO year. None of the inferences in Table 3 are affected (untabulated). 为了提高审计质量，中国监管机构在 2000 年和 2001 年鼓励在中国大陆拥有上市公司审计执照的小型地方审计师事务所相互合并或与更大审计师事务所合并。Chan 和 Wu（2011）显示，在此期间的审计师事务所合并导致审计师独立性提高。为了确保我们在表 3 中的结果不受此类合并相关影响的驱动，我们还在排除所有在首次公开募股（IPO）前一年经历过合并的审计师的情况下重新进行表 3 中的回归分析。表 3 中的推论没有受到影响（未列出）。
^(15){ }^{15} Using a sample of seasoned equity offering applications submitted to the CSRC for approval over the period 1996-1998, Chen and Yuan (2004) and Haw et al. (2005) show that earnings management using NCROA is easy to detect and therefore is discounted by the CSRC. Hence, there is a possibility that IPO firms in our sample period may not use NCROA to inflate earnings in the pre-IPO period. 使用 1996-1998 年期间提交给中国证监会审批的成熟股权发行申请样本，陈和袁（2004）以及霍等人（2005）表明，使用净资本回报率（NCROA）进行盈余管理容易被发现，因此受到中国证监会的折扣。因此，在我们样本期间，首次公开募股（IPO）公司可能不会在 IPO 前期使用 NCROA 来虚增盈余。 ^(16){ }^{16} Inferences are similar if we use the signed performance-matched Jones model discretionary accruals per Kothari et al. (2005). ^(16){ }^{16} 如果我们使用 Kothari 等人（2005）所述的签名绩效匹配 Jones 模型的自由裁量应计，推论是相似的。
17 We also find in untabulated regression analysis that IPO firms in Regimes 1 and 3 are more likely than IPO firms in Regime 2 to experience a reversal of reported earnings in the three-year period subsequent to the IPO year. 我们还发现，在未列出表格的回归分析中，第一和第三阶段的 IPO 公司比第二阶段的 IPO 公司更可能在 IPO 年份后的三年内经历报告收益的逆转。 ^(18){ }^{18} See Yang (2013) for a detailed discussion of the Committee. 请参见杨（2013）对委员会的详细讨论。
^(19){ }^{19} See Chen et al. (2012b) for a detailed historical overview of China’s five-year national economic plans. 请参见陈等人（2012b）关于中国五年国家经济计划的详细历史概述。 ^(20){ }^{20} To further control for the differences in firm characteristics, we also use a propensity score matching (PSM) approach to match the IPOs in Regime 2 with the IPOs in the combined Regimes 1 and 3 based on the observable firm characteristics used in the respective regression models. The observable firm characteristics for the matched IPOs are not significantly different at the 10% significance level between Regime 2 and the combined Regimes 1 and 3. More importantly, the regression results for Tables 3 and 4 continue to differ for Regime 2 versus the combined Regimes 1 and 3 (untabulated). 为了进一步控制公司特征的差异，我们还使用倾向得分匹配（PSM）方法，根据各自回归模型中使用的可观察公司特征，将第二阶段的首次公开募股（IPO）与第一和第三阶段的合并首次公开募股进行匹配。匹配的首次公开募股在 10%的显著性水平下，第二阶段与合并的第一和第三阶段之间的可观察公司特征没有显著差异。更重要的是，表 3 和表 4 的回归结果在第二阶段与合并的第一和第三阶段之间仍然存在差异（未列出表格）。 _(21){ }_{21} There appears to be some confusion in the literature on the exact ending date of the IPO quota system. Even though the IPO quota system officially ended by mid1999 (Cheung et al., 2009), Pistor and Xu (2005) argue that the IPO quota system was in de facto effect till the end of 2002 due to a large backlog of IPO applicants that had been selected under the quota system but Chalmers et al. (2014) argue the quota system ended in March 2001. We agree with Pistor and Xu (2005) based on our own research, but using March 2001 as an alternative cutoff would not make much of a difference because the ending date of Regime 2 is November 6,2001 and there is also an IPO suspension period over August 7, 2001-November 6/2001. In addition, using Cheung et al.'s cutoff doesn’t affect our inferences either because Cheung et al.'s cutoff may explain the difference in results from Regime 1 to Regime 2 but it cannot explain the reversal of results from Regime 2 to Regime 3 because both Regimes 2 and 3 fall under the same post-IPO quota regime. 在文献中，关于 IPO 配额制度的确切结束日期似乎存在一些混淆。尽管 IPO 配额制度在 1999 年中期正式结束（Cheung 等，2009），但 Pistor 和 Xu（2005）认为，由于在配额制度下选定的大量 IPO 申请者，IPO 配额制度实际上一直有效到 2002 年底，而 Chalmers 等（2014）则认为配额制度在 2001 年 3 月结束。根据我们自己的研究，我们同意 Pistor 和 Xu（2005）的观点，但将 2001 年 3 月作为替代截止日期并不会产生太大差异，因为第二阶段的结束日期是 2001 年 11 月 6 日，并且在 2001 年 8 月 7 日至 2001 年 11 月 6 日之间还有一个 IPO 暂停期。此外，使用 Cheung 等的截止日期也不会影响我们的推论，因为 Cheung 等的截止日期可能解释了第一阶段与第二阶段结果之间的差异，但无法解释第二阶段与第三阶段结果的反转，因为第二阶段和第三阶段都属于同一后 IPO 配额制度。
_(22)^(22){ }_{22}^{22} Most of the positive long-term abnormal returns for all three regimes are realized in the first few days of an IPO’s listing. _(22)^(22){ }_{22}^{22} 所有三个阶段的大部分正向长期异常收益都是在首次公开募股（IPO）上市的头几天内实现的。 ^(23){ }^{23} The positive long-term abnormal returns in Table 6 are not contradictory to the evidence on long-term Chinese IPO stock performance (e.g., Chan et al., 2004). This is because the calculation of abnormal returns in Table 6 starts from the offer price while the calculation of abnormal returns in prior long-term performance studies typically starts from the closing price at the end of the first day or first month. Our sample IPOs also exhibit the familiar IPO underpricing, i.e., significantly positive first-day abnormal returns. Consistent with Chan et al. (2004), the mean post-IPO long-term abnormal return starting from the second month of the IPO date is close to zero and statistically insignificant for all three regimes (untabulated). 表 6 中的长期正异常收益与中国 IPO 股票长期表现的证据并不矛盾（例如，Chan 等，2004）。这是因为表 6 中异常收益的计算是从发行价开始，而之前长期表现研究中的异常收益计算通常是从首日或首月末的收盘价开始。我们的样本 IPO 也表现出熟悉的 IPO 低估现象，即显著正的首日异常收益。与 Chan 等（2004）一致，从 IPO 日期第二个月开始的平均 IPO 后长期异常收益对于所有三种情况接近于零且在统计上不显著（未列出）。
24 Individual Chinese stocks could be suspended from trading for a prolonged time period. To avoid potential confounding events resulting from such trading suspensions, we require the number of stock market open days between trading day -1 and trading day +1 to be no > 15>15. We obtain similar inference if we use a cutoff of 10 or 5 . 24 只个别中国股票可能会被暂停交易较长时间。为了避免因这种交易暂停而导致的潜在混淆事件，我们要求交易日-1 和交易日+1 之间的股市开放天数为零 > 15>15 。如果我们使用 10 或 5 的截止值，我们会得到类似的推论。
^(25){ }^{25} Because our IPO firms went public over the period 1997-2004, years t+3t+3 to t+5t+5 span the period 2000-2009. ^(25){ }^{25} 因为我们的 IPO 公司在 1997 年至 2004 年期间上市，年份 t+3t+3 到 t+5t+5 涵盖了 2000 年至 2009 年。
