UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Essays on firms' accounting quality Yun, Ke 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2015_september_ke_yun.pdf [ 1.44MB ]
Metadata
JSON: 24-1.0132741.json
JSON-LD: 24-1.0132741-ld.json
RDF/XML (Pretty): 24-1.0132741-rdf.xml
RDF/JSON: 24-1.0132741-rdf.json
Turtle: 24-1.0132741-turtle.txt
N-Triples: 24-1.0132741-rdf-ntriples.txt
Original Record: 24-1.0132741-source.json
Full Text
24-1.0132741-fulltext.txt
Citation
24-1.0132741.ris

Full Text

      ESSAYS ON FIRMS'  ACCOUNTING QUALITY  by   Yun Ke  Ph.D., Washington University in St. Louis, 2009 M.S., New Mexico State University, 2004 B.E., Tsinghua University, 1998  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY   in  The Faculty of Graduate and Postdoctoral Studies (Business Administration)        THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2015 © Yun Ke, 2015  ii  Abstract This dissertation consists of two essays that present new evidence on the determinants and con-sequences of accounting quality. The first essay examines the consequences of earnings quality on a firm’s use of trade credit. I use earnings smoothness, asymmetric timeliness of earnings (conservatism), and earnings management to proxy for earnings quality. Consistent with high accounting quality reducing information asymmetry between firms and stakeholders, I hypothe-size and find that firms with higher accounting quality are able to obtain more trade credit from their suppliers.  Using a customer-supplier paired subsample, I show that the results are robust after controlling for suppliers’ characteristics. Moreover, using the 2007–2008 financial crisis as an exogenous shock to credit supply, I show that the positive relation between trade credit and accounting quality is more pronounced during a period of credit tightening. Furthermore, I find that the characteristics of transacted products also impact the relation—the association is stronger when companies purchase services or differentiated goods. Finally, I show that the positive asso-ciation is concentrated in small firms and firms without credit ratings on senior debt. Overall, the evidence suggests that high earnings quality facilitates firms’ access to trade credit from suppli-ers. The second essay documents the effect of stock underpricing on firms’ financial reporting quality. I use mutual fund fire sales to identify relatively underpriced stocks and use perfor-mance-matched discretionary accruals to proxy for earnings management. Using difference-in-differences tests, I find that firms subjected to mutual fund fire sales increase their level of earn-ings management relative to unaffected firms. I also show that the effect is greater for firms experiencing more severe underpricing, firms with higher information asymmetry, and lower stock liquidity. In addition, earnings management is more pronounced in financially constrained iii  firms. Finally, I examine whether earnings management helps stock price recovery, but find no evidence to support this hypothesis. In sum, the second essay finds that stock underpricing ad-versely affects firms’ financial reporting quality, an indirect effect of the stock market that has been previously overlooked. iv  Preface This dissertation is original, unpublished, independent work by the author, Yun Ke.    v  Table of contents  Abstract .............................................................................................................................. ii Preface ............................................................................................................................... iv Table of contents ................................................................................................................ v List of tables ..................................................................................................................... vii List of figures .................................................................................................................. viii Acknowledgements........................................................................................................... ix Dedication ........................................................................................................................... x Chapter 1: Introduction .................................................................................................... 1 1.1 Accounting quality overview .............................................................................................. 2 1.2 Earnings quality and its consequences .............................................................................. 5 1.3 Earnings quality and its determinants .............................................................................. 7 Chapter 2: Does accounting quality matter for short-term financing? Evidence from trade credit .................................................................................................................... 10 2.1 Introduction ....................................................................................................................... 10 2.2 Background, literature, and hypothesis development ................................................... 16 2.2.1 Background and related literature ................................................................................ 16 2.2.2 Hypothesis development .............................................................................................. 18 2.3 Empirical proxies and research designs ......................................................................... 23 2.3.1 Measure of accounting quality ..................................................................................... 23 2.3.2 Regression models ....................................................................................................... 25 2.4 Sample selection and descriptive statistics ..................................................................... 29 2.4.1 Sample selection .......................................................................................................... 29 2.4.2 Descriptive statistics .................................................................................................... 30 2.5 Empirical results ............................................................................................................... 31 2.5.1 The association between accounting quality and trade credit ...................................... 31 2.5.2 The effect of financial crisis ........................................................................................ 33 2.5.3 The effect of product characteristics ............................................................................ 34 2.6 Additional tests .................................................................................................................. 35 2.6.1 The effect of firm size .................................................................................................. 35 2.6.2 The effect of credit rating ............................................................................................ 36 2.7 Conclusions ........................................................................................................................ 36 vi  Chapter 3: Stock underpricing and earnings management—evidence from mutual fund fire sales ................................................................................................................ 62 3.1 Introduction ....................................................................................................................... 62 3.2 Background, literature, and hypothesis development ................................................... 68 3.2.1 Background and related literature ................................................................................ 68 3.2.2 Hypothesis development .............................................................................................. 70 3.3 Data, variables, and research design ............................................................................... 73 3.3.1 Measuring mutual fund capital flows .......................................................................... 73 3.3.2 Identifying stocks experiencing fire sales .................................................................... 75 3.3.3 Earnings management metrics ..................................................................................... 77 3.3.4 Research designs .......................................................................................................... 78 3.3.5 Descriptive statistics and preliminary evidence ........................................................... 79 3.4 Empirical results ............................................................................................................... 80 3.4.1 Regression results ........................................................................................................ 80 3.4.2 Additional test .............................................................................................................. 83 3.5 Does earnings management correct stock underpricing? ............................................. 84 3.5.1 Long-term effect of earnings management .................................................................. 85 3.5.2 Short-term effect of earnings management .................................................................. 86 3.5.3 Discussion .................................................................................................................... 87 3.6 Conclusions ........................................................................................................................ 88 Chapter 4: Conclusions ................................................................................................. 116 References ....................................................................................................................... 120 Appendices ...................................................................................................................... 128 Appendix A: An example of suppliers refusing to increase trade credit ............................ 128 Appendix B: Variable definitions for Chapter 2 ................................................................. 129 Appendix C: Product characteristics classification for Chapter 2 ...................................... 130 Appendix D: Variable definitions for Chapter 3................................................................. 132   vii  List of tables Table 2.1: Sample selection procedures and sample distribution ................................................. 38 Table 2.2: Descriptive statistics .................................................................................................... 40 Table 2.3: Correlation matrix ........................................................................................................ 43 Table 2.4: The relation between trade credit and accounting quality ........................................... 44 Table 2.5: The impact of financial crisis on the relation between trade credit and accounting quality ....................................................................................................................................... 49 Table 2.6: The impact of product characteristics on the relation between trade credit and accounting quality ..................................................................................................................... 53 Table 2.7: The effect of firm size on the relation between trade credit and accounting quality ... 57 Table 2.8: The effect of credit rating on the relation between trade credit and accounting quality for low and high credit risk firms ............................................................................................. 59 Table 3.1: Descriptive statistics .................................................................................................... 90 Table 3.2: The effect of stock underpricing on firm’s financial reporting quality ....................... 93 Table 3.3: The effect of underpricing severity.............................................................................. 95 Table 3.4: The effect of analyst coverage ..................................................................................... 97 Table 3.5: The effect of stock liquidity ......................................................................................... 99 Table 3.6: The effect of institutional ownership ......................................................................... 101 Table 3.7: The effect of financial constraint ............................................................................... 103 Table 3.8: Additional test: investment and intangible ................................................................ 105 Table 3.9: The effect of earnings management on stock price recovery .................................... 107  viii  List of figures Figure 2.1: The use of trade credit in the U. S. ............................................................................. 61 Figure 3.1: Sample selection ....................................................................................................... 113 Figure 3.2: Cumulative average abnormal returns around mutual fund fire sales ...................... 115  ix  Acknowledgements Switching from science to business is not mission impossible; it is, however, not easy. This is especially true for a person who is introverted. Nevertheless, with the continuous support from my co-advisors, Kin Lo and Sandra Chamberlain, I am coming to the end of this journey and seeing the light at the end of the tunnel. The work that I present here would not be possible with-out their help. They were always patient with my rough, and sometimes naïve ideas. They con-stantly provided me valuable feedback and suggestions. More importantly, they gave me endless encouragement: Kin urged me to “aim high,” while Sandra wanted me to be “grimly determined.” I thank my committee member, Professor Hernan Ortiz-Molina, for his constructive advice on my work. I would also like to thank Professors Joy Begley, Russell Lundholm, Rafael Rogo, Dan Simunic, and Jenny Zhang for their helpful comments on my work and for their teaching. My thanks are extended to other doctoral students in the Accounting Division, and other divisions as well. Office life with other doctoral students is truly enjoyable. Finally, I thank Jackie, my wife, for being there with me all the time. She always encourages me to become a better man, be a better researcher, and to do interesting research.    x  Dedication To my wife Jackie, for being the one I love the most and  for being the one I “hate” the most.   I love you!  1  Chapter 1: Introduction Stakeholders in almost any organization rely on the accounting system to translate and summa-rize the economic transactions that have taken place over a period of time and use the resulting information as the basis for decisions. These decisions can generally be expected to be more judicious if the quality of the accounting is high. While Generally Accepted Accounting Princi-ples (GAAP) in the United States and International Financial Reporting Standards (IFRS) help to establish high quality accounting numbers, it is well understood that these rules give firm man-agers considerable level of discretion to alter the underlying quality of information released through the system. This dissertation examines both the consequences and determinants of ac-counting quality. The first essay (Chapter 2) finds that a firm with higher accounting quality can obtain more trade credit financing from its suppliers, and the second essay (Chapter 3) shows that stock underpricing can adversely affect a firm’s financial reporting quality. In the first chap-ter, I discuss the proxies of accounting quality, and then review the related literature on the con-sequences and determinants of accounting quality. I also provide more specifics about my findings. This dissertation, like many other papers in this genre, confronts a major challenge in measuring accounting (or earnings) quality.1 To do so, I use several proxies that have appeared in prior research. Dechow, Ge, and Schrand (2010) provide an excellent review on accounting quality and summarize the strengths and weaknesses of different proxies of accounting quality. More importantly, they point out that depending on the context of decision-making, earnings quality has different meanings for different users. In Chapter 2, I use earnings smoothness,                                                   1 In this dissertation, “accounting quality”, “earnings quality”, and “financial reporting quality” are interchangeable. 2  asymmetric earnings timeliness (conservatism), and earnings management to proxy for account-ing quality because it has been shown that in the relationship between suppliers and customers, suppliers care about these aspects of their customer firms’ earnings. In Chapter 3, I use perfor-mance-matched abnormal accruals to proxy for earnings management.  1.1 Accounting quality overview The concept of accounting quality (or earnings quality) is fundamental in accounting research, yet, there is no consensus on how best to define and measure accounting quality.  This state of affairs is, in part, a natural outcome of the fact that desirable characteristics of accounting num-bers differ depending on the specific decision-context (Dechow et al. 2010).  While a desirable earnings pattern for an equity holder might engender smooth growth, debtholders are likely to prefer earnings that are conservative, responding quickly to bad news, and therefore, not smooth. A standard list of earnings quality measures includes constructs such as predictability, smooth-ness, persistence, conservatism, lack of manipulation, and timeliness.2   An additional frustration for readers and researchers alike is the fact that accounting quality measures are determined by the confluence of economic transactions and the application of accounting standards and estimates to these transactions. We are often most interested in “quality” as it is reflected in the application of accounting standards, managerial accounting choices, and managerial estimates. However, these drivers of accounting quality are very diffi-cult to separate from economic fundamentals.  Consider one of the accounting quality measures used in Chapter 2, earnings smoothness (typically measured as the standard deviation of account-ing earnings relative to the standard deviation of cash flows).  Earnings are simply cash flows                                                   2 Francis, LaFond, Olsson, Schipper (2004) examine the relation between the cost of equity capital and seven attrib-utes of earnings quality, including accrual quality, persistence, predictability, smoothness, relevance, timeliness, and conservatism. 3  plus accruals, and accruals in turn are determined by the transactions that generate the cash flows, the accounting standards that dictate the accruals, and managerial choices and assumptions. While I can borrow from past research (e.g., Jones 1991; Dechow, Sloan, and Sweeney 1995; Kothari, Leone, and Wasley 2005) to disentangle the managed portion of the earnings smooth-ness measure, from the unmanaged portion, it is unclear whether managerial discretion improves or distorts accounting quality. It is also unclear the degree to which these models affect Type I and Type II errors. For example, using simulation data, Dechow, Sloan, and Sweeney (1995) show that when earnings performance is correlated with the error in measuring discretionary accruals, testing of earnings management can have excessive type I errors (if the hypothesized event correlates with earnings performance but does not lead to earnings management). On the other hand, Kothari, Leone, and Wasley (2005) recognize that the design of using performance-matching to estimate discretionary accruals may increase the type II error rate, leading to a lower probability of detecting earnings management. As it turns out, I will take two different approaches to measure earnings quality in this dissertation.  In Chapter 2, where I study the links between accounting quality and the level of trade credit customer firms are able to access, I use three different measures of earnings quali-ty—earnings smoothness, conservatism, and the magnitude of earnings management. The choice of these measures is supported by prior research, which shows that suppliers care about these earnings attributes. Graham et al.’s (2005) survey shows that executives believe that smooth earnings is perceived by suppliers to be associated with low business risk. Hui et al. (2012) find that suppliers demand conservatism from their customer firms. Finally, Raman and Shahrur (2008) present evidence that suppliers are aware of their customer firms’ opportunistic earnings 4  management. Thus, I examine how these three different measures of earnings quality influence suppliers’ trade credit decision. In Chapter 2, I take the measures of earnings quality (particularly smoothing) as given. Readers may wonder if firms can achieve smooth earnings through earnings management; if so, my measure of earnings smoothness can be artificially induced and calls into question its true quality. With regard to this issue, as mentioned, I do not claim to find the equilibrium or opti-mum level of earnings quality (e.g., smoothness) without any manipulation. That being said, I appeal to the fact that earnings management cannot be done forever. Analytical models of earn-ings smoothing (e.g., Trueman and Titman 1988) typically assume that 1) a firm’s ability to shift income from one period to another depends on its operations, and 2) firm outsiders are not cer-tain of the flexibility a firm has to smooth earnings. In essence, they assume that managing earn-ings is costly (thus implying that earnings smoothing is a viable signaling mechanism). For example, firms that recognize income early potentially incur a cost of accelerating their tax pay-ments. In addition, accounting is such that earnings and cash flows must add up over the life of a firm. Consequently upwards earnings management in any given period implies to a decrease in earnings in subsequent periods. Barton and Simko (2002) use this observation to show that bal-ance sheets which contain accruals to date can be used to detect when earnings management flexibility is most likely to be constrained.  In contrast to Chapter 2 where some of my earnings quality measures are a mix of discre-tionary and non-discretionary components, in Chapter 3, I only focus on discretionary accruals. In Chapter 3 I attempt to see if earnings are managed in order to correct mispricing. To control for firm performance in estimating the level of earnings management, I follow the approach described in Kothari et al. (2005) which provides the highest level of controls for non-5  discretionary accruals. First, to measure discretionary accruals, I regress total accruals on a set of firm characteristics, including the inverse of total assets, the change in sales, and property, plant, and equipment (as in Jones 1991). The model is estimated cross-sectionally each year in the same two-digit SIC industry. Then, each firm-year observation is matched with another from the same two-digit SIC industry and year with the closest return on assets (ROA). The resulting performance-matched discretionary accrual is the difference between two discretionary accruals. Kothari et al. (2015) show that this measure is reliable in terms of Type I error rates in detecting earnings management.  1.2 Earnings quality and its consequences A number of studies have attempted to establish the consequences of variation in earnings quali-ty. Dechow et al. (2010) divide the consequences into several categories, including litigation likelihood, market valuation, executive compensation, and cost of capital. My dissertation is most concerned with the cost of capital. Therefore, I only focus on the links between earnings quality and cost of capital. The idea that high-quality accounting information can reduce information asymmetry be-tween investors and firms, and therefore lower cost of capital is intuitive and appealing. It has motivated both theoretical and empirical studies. Theoretical models on the relation between earnings quality and cost of equity, however, are controversial. Easley and O’Hara (2004) devel-op a model to show that more private information enables informed investors to better adjust their portfolios, and uninformed investors, facing this disadvantage, demand higher returns to compensate for information risk. Thus, higher information asymmetry leads to higher cost of capital. Hughes, Liu, and Liu (2007), however, argue that for large economies, information asymmetry can be diversified away and should not affect cost of capital. 6  Empirical studies also do not yield conclusive evidence on the relation between earnings quality and cost of equity. Bhattacharya, Daouk, and Welker (2003) conduct a cross-country study and show that an increase in earnings opacity (measured by aggressive accruals and loss avoidance) is associated with an increase in the cost of equity. Focusing on the U. S. firms only, Francis, LaFond, Olsson, and Schipper (2005) use accrual quality to proxy for information risk and find that higher accrual quality is associated with lower cost of equity. However, Core, Guay, and Verdi (2008) argue that Francis et al.’s (2005) tests are not appropriate; their tests find no evidence that accrual quality is a priced risk factor and so the link between accrual quality and cost of capital has not yet been established.  There are fewer studies on the relation between earnings quality and debt financing. Bha-rath, Sunder, and Sunder (2008) show that firms with poor accounting quality prefer private debt (i.e., bank loans) and the loan terms are more stringent for these firms. Focusing on reporting conservatism, Zhang (2008) shows that firms reporting more conservatively can get a lower interest rate from their lenders.  Studies on the relation between accounting quality and trade credit financing (i.e., suppli-ers’ lending to their customer firms), however, has been limited.3 This is surprising given that (1) trade credit is a very important source of short-term financing and plays a critical role in provid-ing liquidity to firms and the economy (Bernanke and Gertler 1995), and (2) financial accounting textbooks frequently mention the importance of accounting information to support suppliers’ trade credit decisions. Chapter 2 of this dissertation examines this less explored consequence of accounting quality.                                                   3 Shortly before my final oral defense, I found a recently published paper (Garcia-Teruel, Martinez-Solano, and Sanchez-Ballesta 2014) that studies the same question using Spanish data. Their findings suggests high accounting quality facilitates supplier financing, which is similar to mine. I will further describe their study and compare the difference between their study and mine in the future working paper version.  7  More specifically, in Chapter 2, I examine the relation between a firm’s trade credit and its accounting quality. I use earnings smoothness, asymmetric timeliness of earnings, and earn-ings management to proxy for accounting quality. Theory suggests that high accounting quality reduces information asymmetry between firms and stakeholders, therefore, I hypothesize that firms with higher accounting quality are able to obtain more trade credit from their suppliers. My results are consistent with this prediction. Moreover, the results are robust after controlling for suppliers’ characteristics.  To corroborate my findings, I further examine the temporal and cross-sectional variation of this relation. Using the 2007–2008 financial crisis as an exogenous shock to credit supply, I show that the positive relation between trade credit and accounting quality is more pronounced during the period of credit tightening. Cross-sectionally, I find evidence that the characteristics of firms’ products impact the relation—the association is stronger when companies purchase services or differentiated goods.  1.3 Earnings quality and its determinants While prior research and my study in Chapter 2 document the consequences of high accounting quality, not all firms can achieve high-quality reporting. Firm managers often have incentives to distort their firms’ reporting quality. For example, there is a large body of literature associating managerial incentives with different components of CEO compensation packages, including bonuses and equity incentives. Seminal work by Healy (1985) shows that managers use discre-tionary accruals to maximize their current and future bonuses. More specifically, he finds that when an earnings target is still reachable, managers use income-increasing discretionary accruals to hit earnings target to get their bonuses. However, when earnings target is unreachable, they use income-decreasing accruals to save earnings for the next period to maximize next period’s 8  bonuses. Recently, both Cheng and Warfield (2005) and Bergstresser and Philippon (2006) doc-ument that the use of discretionary accruals to manage earnings is associated with managers’ equity incentives. Several studies also suggest that executives manage earnings upwards to avoid violating debt covenants (DeFond and Jiambalvo 1994) and to alleviate their career concerns. For instance, Sweeney (1994) finds that managers use income-increasing accounting changes when their firms approach debt-covenant violations. DeFond and Park (1997) show that managers use discretion-ary accruals to borrow earnings from the future to the current period to alleviate career concerns associated with declined earnings performance. Finally, it has been suggested that executives manage earnings upwards to maintain or to increase their firms’ stock prices, although there is no direct evidence suggesting this motivation. Indirectly, however, two lines of inquiry suggest this might be the case. First, empirical evidence suggests that stock markets punish firms whose earnings miss benchmarks. For example, Sloan and Skinner (2002) find that stock prices decline upon disappointing earnings announcements, and this is more pronounced for growth firms. Second, previous studies find that managers avoid reporting earnings that are below analysts’ forecasts. Managers also avoid reporting earnings declines and negative earnings (Burgstahler and Dichev 1997; Degeorge, Patel, and Zeckhauser 1999).  However, it is difficult to test whether the stock price itself can motivate earnings man-agement because price is endogenously related to firms’ future performance. In Chapter 3, I sidestep this challenge by using an exogenous shock to stock price, the negative price pressure caused by mutual fund fire sales. Coval and Stafford (2007) provide an initial study suggesting 9  that stock mispricing caused by mutual fund fire sales is relatively exogenous to firm fundamen-tals.  More specifically, using mutual fund transaction data from 1996 to 2006, I identify fire sale stocks that experienced mutual fund forced sales due to large capital outflows from the funds. To construct my control firms, I select firms that are traded moderately by financially unconstrained mutual funds. I use performance-matched discretionary accruals as the proxy for earnings management (Kothari, Leone, and Wasley 2005). By using difference-in-differences tests, I find evidence that is consistent with stock underpricing having an effect on firms’ level of earnings management. In addition, I also examine whether the severity of fire sales affects the degree of earnings management by partitioning the sample using firms’ abnormal returns during the fire sale quarter. My regression results show that the more abnormal prices decline due to fire sales, the more firms manage their earnings upwards. Finally, I find that the effect is more pro-nounced in firms with higher information asymmetry, lower stock liquidity, and when firms are more financially constrained. However, I find no evidence that earnings management helps stock prices to recover. The remainder of the dissertation is structured as follows: In Chapter 2, I examine wheth-er firms with higher quality financial reporting can obtain more trade credit financing from their customer firms. In Chapter 3, I investigate whether stock underpricing can adversely affect firms’ financial reporting quality. I summarize the findings and conclude in Chapter 4, in which I also discuss potential future research.     10  Chapter 2: Does accounting quality matter for short-term financing? Evidence from trade credit 2.1 Introduction The intuition that the quality of accounting information plays an integral part of the efficient operation of capital markets has motivated both theoretical and empirical studies in accounting.4 While the precise theoretical underpinning of the relation between accounting quality and capital market resource allocation is somewhat controversial, empirical studies have provided evidence to support this relation. For example, Bharath, Sunder, and Sunder (2008) find that high account-ing quality appears to facilitate firms’ access to the public debt market, and Francis, LaFond, Olsson, and Schipper (2005) show that higher accounting quality is associated with lower cost of equity. However, aside from recent work by Garcia-Teruel, Martinez-Solano, and Sanchez-Ballesta (2014), the link between accounting quality and trade credit financing (i.e., suppliers’ lending to their customer firms) remains relatively unexplored.5 In fact, trade credit is the largest and most important source of short-term financing and plays a critical role in providing liquidity to the economy (Bernanke and Gertler 1995).6 Perhaps this omission in the literature is due to the perception that trade credit decisions are less likely to rely on accounting information than other forms of financing decisions. Yet financial accounting textbooks frequently mention the im-portance of accounting information to support suppliers’ trade credit decisions.                                                   4 For examples of theoretical models, see Easley and O’Hara (2004), Hughes, Liu, and Liu (2007), and Lambert, Leuz, and Verrecchia (2007). For empirical literature, see surveys by Armstrong, Guay, and Weber (2010) and Dechow, Ge, and Schrand (2010). 5 As I mentioned in Chapter 1, Garcia-Teruel, Martinez-Solano, and Sanchez-Ballesta (2014) is a concurrent work, which focuses on Spanish companies. Hui, Klasa, and Yeung (2012) is another related study, in which they find that a firm’s reporting conservatism increases with its suppliers’ bargaining power. 6 According to the U.S. Flow of Funds Account historical data, the total amount of trade credit extended by non-financial businesses was $77 billion in 1993. More recently, Barrot (2013) documents that accounts payable reached about $2 trillion for non-financial U.S. corporate businesses, about three times as large as bank loans, as of Septem-ber 2012 (based on the U.S. Flow of Funds Account) (See Figure 2.1).    11 This paper explores whether a firm’s accounting quality affects the cost of its trade credit. There is no doubt that trade credit differs from standard debt because it comingles an operating decision with a financing decision. The fact that a customer firm and a supplier rely on each other in an operating context implies that the supplier’s financing decision is made in a setting with inherently lower information asymmetry compared with other creditors. Indeed, Petersen and Rajan (1997) argue that a supplier, in comparison to financial institutions, has superior in-formation about its customer firms because it can collect information faster and at a lower cost through daily business activities.7 This argument suggests that accounting information plays a smaller and perhaps insignificant role in the suppliers’ trade credit decision-making, so that ac-counting quality would not matter. Although suppliers can, to some extent, predict customer firms’ future performance using their operational information, they are not immune to information asymmetry. As customer firms are closer to end consumers, they usually have better demand information and market forecasts than suppliers. The resulting bullwhip effect has been extensively discussed in the supply chain environment. Anecdotes, such as the one wherein customers of Hewlett-Packard Co. (HP) use “phantom orders” in order to get more products from HP (Zarley and Damore 1996), also sug-gest that optimistic operational information does not necessarily lead to superior future perfor-mance. On the other hand, it is difficult to verify that customer firms manipulates their operational information. A rosy forecast can be due to managers’ intentionally betting on future prospects, or, as a result of an intrinsically volatile market. Accounting information, however, can be used as a supplement to infer customer firms’ performance and thereby reduce the overall                                                   7 Suppliers can develop intimate knowledge of their customers by accessing customer firms’ information systems. For example, being the battery supplier of Tesla electric automobile, Panasonic can access Tesla’s information system.    12 information asymmetry. Furthermore, when the number of customers increases, suppliers can understand customers’ operations more effectively with accounting information instead of com-pletely relying on summarizing different sources of operational information. In summary, even though suppliers can have some direct information about their customer firms’ operational per-formance, suppliers are still prone to information asymmetry, and accounting information can complement other information. Accounting quality, therefore, can help to facilitate suppliers’ trade credit decisions. The importance of accounting quality is likely to vary across firms and over time. In par-ticular, the recent financial crisis highlighted the crucial role that trade credit can play when bank credit is scarce. Researchers show that during the crisis, firms that were able to obtain more trade credit from their suppliers experienced less negative consequences (Bastos and Pindado 2013; Coulibaly, Sapriza, and Zlate 2013). On the other hand, cash-rich firms and firms with better access to bank credit extended more trade credit to their customer firms, and in return, experi-ence better performance afterwards (Kestens, Van Cauwenberge, and Vander Bauwhede 2012; Garcia-Appendini and Montoriol-Garriga 2013). I conjecture that variability in accounting quali-ty can assist supplier firms to allocate their scarce capital during periods of the crisis-induced credit demand and economic uncertainty. This conjecture draws upon prior research on standard debt. For example, Watts and Zuo (2012) find that during the financial crisis, firms with greater accounting conservatism were able to issue more debt to undertake more investment projects. Extending this literature to trade credit financing, I predict that high accounting quality benefits firms more in terms of obtaining more trade credit from their suppliers during the 2007-2008 financial crisis.    13 The cost of trade credit is also likely to vary with product attributes that increase or de-crease the likelihood of repayment. Prior studies suggest that trade credit decisions are affected by the collateral value and the liquidation value of the transacted products (Mian and Smith 1992; Cuñat 2007; Giannetti, Burkart, and Ellingsen 2011). For example, when customers cannot make their payments, suppliers can repossess goods and resell them, which is not applicable to services. In addition, it will be easier for suppliers to resell standardized goods than differentiated goods. Therefore, I examine whether and how product characteristics affect the relation between ac-counting quality and trade credit.    My research methodology regresses the amount of trade credit on three standard proxy variables for accounting quality. The quality measures include: earnings smoothness, asymmetric timeliness of earnings (conditional conservatism), and abnormal accruals.8 After controlling for other determinants of trade credit usage by a customer firm, I find that a firm can obtain more trade credit from its suppliers when its earnings are smoother, when it has greater reporting con-servatism, and when it has less accumulated abnormal accruals. As prior work shows that the amount of trade credit received by a customer firm depends on the suppliers’ liquidity and will-ingness to lend (Petersen and Rajan 1997; Garcia-Appendini and Montoriol-Garriga 2013), I use a customer-supplier paired subsample from 2006 to 2008 to simultaneously control for the de-mand and supply of trade credit. After controlling for suppliers’ characteristics, I find that the positive association between a firm’s accounting quality and its amount of trade credit is robust. Using the 2007-2008 financial crisis as an exogenous shock to the supply of credit, I also show                                                   8 This research design choice is influenced by prior studies in which a variety of measures have been used. For example, Graham, Harvey, and Rajgopal (2005) reveal that managers use smoother earnings to lower their firms’ perceived risk. Hui et al. (2012) use the analogy between trade credit and debt financing to argue that suppliers prefer their customer firms to report conservatively. Raman and Shahrur (2008) show that suppliers respond to their buyers’ opportunistic earnings management by shortening the duration of customer-supplier relationships. Finally, Dechow et al. (2010) survey several widely used measures of accounting quality and suggest that the choice of measure depends on the decision context.    14 that the positive relation between accounting quality and trade credit is more pronounced during the crisis. This suggests the heightened importance of accounting information in trade credit decisions during times of tight credit supply and high macro-economic uncertainty. Finally, my results suggest that the positive relation between accounting quality and trade credit is stronger when the transacted products are services instead of goods, and when the products are differenti-ated goods instead of standardized goods. I perform additional tests to examine whether the rela-tion between accounting quality and the cost of trade credit varies with firm size and credit rating, as prior research finds that small firms rely more on trade credit and firms without credit ratings have poorer access to credit markets (Petersen and Rajan 1997). My results reveal that the posi-tive association between trade credit and accounting quality is concentrated in small firms and firms without credit ratings.  The paper makes several contributions to the literature. First, it provides new and direct evidence that accounting quality matters for short-term financing. Several studies have docu-mented that high accounting quality eases debt and equity financing (Bharath et al. 2008; Bhattacharya, Daouk, and Welker 2003), but few studies consider the role of accounting quality in trade credit financing.9 This paper provides compelling evidence that accounting quality af-fects suppliers’ trade credit decision-making. In addition, this paper shows how the relation be-tween accounting quality and trade credit financing varies with economic condition (e.g., credit tightening) and product characteristics. Furthermore, this paper documents a new and real benefit of high accounting quality by showing that firms reporting more conservatively can obtain more trade credit from their suppli-ers, especially during the financial crisis. Hui, Klasa, and Yeung (2012) only argue that suppliers                                                   9 One exception is Garcia-Teruel, Martinez-Solano, and Sanchez-Ballesta (2014), which is a concurrent work of my study.    15 demand conservative reporting from their buyers and find that a firm’s conditional conservatism increases with its suppliers’ bargaining power. However, they do not explore whether and how a firm can benefit from its suppliers by reporting conservatively. My paper complements their study by showing that firms with greater reporting conservatism actually obtain more trade credit from their suppliers. In another related study, Zhang (2008) shows that firms with greater condi-tional conservatism obtain lower interest loans from banks. While she focuses on bank credit channel only, my paper extends our understanding to trade credit channel and documents that high accounting quality also eases firms’ access to trade credit financing. Finally, this paper extends two recent studies by Watts and Zuo (2012) and Garcia-Appendini and Montoriol-Garriga (2013) by showing that the importance of accounting quality varies with macro-economic conditions. Watts and Zuo (2012) find that firms with greater ac-counting conservatism experienced less negative stock returns during the financial crisis because they were able to issue more public debt to undertake investment projects. This paper comple-ments their study by showing that conservative firms were also able to obtain more trade credit from their suppliers to alleviate the negative impact of credit tightening. Garcia-Appendini and Montoriol-Garriga (2013) show that firms with sufficient liquidity were able to help their cus-tomer firms by extending more trade credit. However, they do not explore what firm characteris-tics, in terms of financial reporting, can help customer firms get more trade credit from their suppliers. This paper complements their study by revealing a positive relation between customer firms’ accounting quality and the amount of trade credit. This chapter proceeds as follows. Section 2.2 is the background, related literature, and hypothesis development. Section 2.3 describes empirical proxies for accounting quality and    16 research designs. Section 2.4 is sample selection and descriptive statistics. Section 2.5 reports empirical results. Section 2.6 is additional tests. Section 2.7 concludes. 2.2 Background, literature, and hypothesis development 2.2.1 Background and related literature Trade credit arises when a firm buys goods and (or) services from its suppliers without paying cash immediately. Survey evidence by Ng, Smith, and Smith (1999) shows that suppliers usually do not directly charge interest on their trade credit sales; instead, they offer a discount to promote early payment.10 If customer firms are unable to take advantage of the discount, the implicit interest on trade credit can be very high. Using 2-10 net 30 as common terms of sales, Petersen and Rajan (1994) estimate the interest rate to be 44.6 percent annually.11 Once the discount peri-od ends, the later customers pay, the lower the effective interest rate. While delayed payment can reduce the rate significantly, the interest rate is still higher than that of alternative short-term financing sources.12 Ng et al. (1999) also reveal that when a supplier decides to extend trade credit to its customer firms, the terms of trade credit are relatively uniform across all customer firms. This suggests that the main credit decision that a supplier makes is the amount of trade credit to extend, rather than the interest rate (see Appendix A for an example in which suppliers refuse to adjust the amount of trade credit).13 As a result, this paper uses the amount to measure a firm’s use of trade credit.                                                    10 For example, common terms of sale are 2-10 net 30, which means the full purchase price is due in 30 days and the customer can receive a two-percent discount if payment occurs within 10 days of the sale.  11 By taking the early payment discount at the 10th day after the sale, the firm is effectively borrowing at 2/98 per-cent per 20-day period, which is equivalent to an annual rate of 44.6 percent ([1 + 2/98](365/20) - 1) (Petersen and Rajan 1994). 12 Survey evidence shows that 40-60 percent firms delay their payments (Altunok 2012). Wilner (1997) estimates that delayed payment reduces the implicit interest rate by two-fifths. 13 Theoretically, suppliers can change the interest rate by adjusting either the cash discount or the discount period and full payment due date.    17 While there are almost no studies formally incorporating accounting information into trade credit decision, existing theoretical models do leave room for the role of accounting quality. For example, Biais and Gollier (1997) develop a model of trade credit decisions in which there are three players: a bank, a buyer (a customer firm), and a seller (a supplier).14 They classify the buyer into two types, good and bad, based on whether the buyer has access to a positive NPV project. Due to information asymmetry, the seller does not perfectly know the buyer’s type. The seller, however, has private information to infer the buyer’s type and makes a trade credit financ-ing decision.15 The seller is more likely to extend trade credit to the buyer if the probability of the buyer being a good type is high. Accounting information can be used as a supplemental sig-nal in this model. A buyer’s good accounting performance can increase the likelihood of it being inferred as good type by supplier firms. Higher accounting quality increases the precision of the accounting signal. Instead of specifically linking accounting quality to the amount of trade credit, there have been two studies linking accounting quality to the customer-supplier relationship. Bowen, Du-Charme, and Shores (1995) examine firms’ inventory and depreciation method choices and find that firms are more likely to adopt income-increasing choices when suppliers’ implicit claims are higher. Hui et al. (2012), on the other hand, argue that suppliers prefer their customer firms to report more conservatively and find evidence that customer firms’ reporting conservatism in-creases with suppliers’ bargaining power. However, both studies only examine the impact of the                                                   14 Biais and Gollier (1997) cover a complex relation among the three players, while my paper only focuses on the relation between the seller and the buyer. 15 The seller can develop this type of information from its relationship with the buyer or from the buyer’s order. In Biais and Gollier’s (1997) model, if the buyer is good, the signal received by the seller () always takes the value 1. If the buyer is bad, the signal can take the value 1 with probability  and 0 with probability 1-. If among all the buyers, the proportion of good buyers is , then conditionally on the signal received by the buyer is 1, the probabil-ity of the buyer being good is Pr (good | =1) = /(+(1-)).     18 customer-supplier relationship on customer firms’ financial reporting quality, and do not provide any evidence on the benefits that customer firms can receive by having high accounting quality. 2.2.2 Hypothesis development As providers of short-term financing, suppliers consider several important aspects when they extend trade credit to their customers: in the short term, the ability of the customer firm to pay its obligations on a timely basis, and in the long-run, the customer firm’s long-term financial viabil-ity and future prospects as a source of customer revenue.16  Timely payment is a big concern in practice because customers often delay their pay-ments to their suppliers. Based on the data from the National Survey of Small Business Finance (NSSBF), about 40 to 60 percent of firms pay their suppliers late (after the due date); among these firms, the amount being paid late is about 30 to 36 percent of their purchases on account (Altunok 2012). This payment delay problem is not limited to small businesses, as big companies are also slow in paying their suppliers. According to recent news in the Wall Street Journal, some big firms (e.g., P&G and DuPont) exploit their bargaining power and delay their payments to suppliers by several weeks, which helps these big firms to free up cash.17 Studying a sample of small suppliers linked with large customers, Murfin and Njoroge (2015) show that the delay of payment by large customers can cause small suppliers’ contraction in investment in plant and equipment and reduction in operating expenditures if these suppliers do not have enough liquidi-ty. Therefore, a supplier has to consider the possibility of late payment and, in some extreme cases, the default of customers when it extends trade credit.                                                    16 See Appendix A for an example in which a supplier refused to increase trade credit based on the customer firm’s financial results. 17 See the WSJ report “P&G, big companies pinch suppliers on payments” at http://online.wsj.com/article/SB10001424127887324010704578418361635041842.html. Another WSJ report is at http://online.wsj.com/article/SB10001424052702303296604577450561434496668.html.      19 A firm with high accounting quality can help to alleviate its suppliers’ concerns so that they are willing to extend more trade credit to the firm. Suppliers’ asymmetric payoff function suggests that they prefer their customer firms to have stable operations. Customer firms with high accounting quality are perceived to be less risky, suppliers would prefer to deal with these customer firms. Consistent with this perception, Graham, Harvey, and Rajgopal’s (2005) survey results show that customer firms’ managers prefer to report smoother earnings, one aspect of accounting quality, to assure their suppliers a stable business. Not only are smoother earnings perceived to be less risky, but they can also reduce information asymmetry between suppliers and customer firms.18 Empirical evidence suggests that changes in the current stock price are more informative and contain more information about the future earnings of firms when earnings are smoother (Tucker and Zarowin 2006). Thus, suppliers can better forecast customer firms’ earnings and the ability to meet their short-term obligation, which can facilitate trade credit deci-sions. Long-term financial viability is also important for a supplier as it can be costly to build up a long-term relationship with one customer and then switch to another customer. For example, suppose a supplier signs a long-term contract with a customer and then acquires specialized equipment (customer-specific investment) to manufacture products.19 In such cases, the supplier not only is concerned about the buyer’s ability to meet short-term trading obligations, but also with the buyer’s long-term viability (because of the significant customer-specific or relation-                                                  18 Several studies argue that managers intentionally use income smoothing to reveal their private information about future earnings and thereby reduce information asymmetry (Kirschenheiter and Melumad 2002; Sankar and Subramanyam 2001). Other models suggest that shareholders demand managers to smooth earnings to reduce the informational advantage between informed and uninformed investors caused by greater earnings volatility (Goel and Thakor 2003). 19 See Williamson (1975, 1979).    20 specific investment). If the customer goes out of business, the supplier’s investment loses value. Although the supplier might be able to find other customers, the switching cost can be high.. To protect themselves from potential losses, suppliers can demand high accounting quali-ty from their customer firms. For example, a theoretical model suggests that accounting conserv-atism can improve debt contract efficiency by reducing managers’ opportunities to manipulate earnings (Gao 2013). Suppliers’ asymmetric payoff function and their long-term implicit claims in customer firms make trade credit provision similar to debt financing. This similarity leads suppliers to demand accounting conservatism from customers (Hui et al. 2012). In addition, accounting conservatism helps to reduce bankruptcy risk, which protects suppliers from down-side risk (Biddle, Ma, and Song 2012).20 Finally, suppliers can infer their customers’ opportunis-tic behavior from accounting quality and take corresponding action. Raman and Shahrur (2008) find that when customer firms use earnings management opportunistically to influence the per-ception of suppliers, reducing accounting quality, suppliers react to this opportunistic behavior by shortening the duration of customer-supplier relationships.  To summarize, my first hypothesis is that a firm’s use of trade credit increases with its accounting quality.  H1: Other things equal, a firm’s amount of trade credit is positively associated with its accounting quality.  I may not find an association between a firm’s amount of trade credit and its accounting quality if suppliers rely less on accounting quality as a screening tool. While suppliers are one type of creditor, they are not banks and trade credit financing also differs from debt financing. Petersen and Rajan (1997) argue that suppliers, relative to banks, have information and monitor-                                                  20 Zhang (2008) also shows that conservatism benefits lenders by accelerating covenant violation, which signals default risk earlier.    21 ing advantages as they can collect customer firms’ operating information during normal business transactions. If this is the case, accounting information, and consequently, accounting quality may play no role in trade credit decision. In addition, while debtholders have asymmetric payoff functions and would like their clients to avoid risky projects, suppliers may enjoy their customer firms’ upside growth opportunities when customer firms make more purchases. This makes the supplier’s claim in customer firms more like equity. These factors may work against me in terms of finding the relation between the amount of trade credit and some accounting quality measure (such as conservatism, which has been documented to be related to debt contracts.) The 2007-2008 global financial crisis provides an ideal opportunity to study how the as-sociation between accounting quality and trade credit varies with macro-economic conditions. The financial crisis created a shock to the credit markets. Banks reduced their lending signifi-cantly (Ivashina and Scharfstein 2010). Prior research argues that while firms appear to prefer bank credit over trade credit, the shortfall in bank credit forces firms reliant on bank loans to turn to their suppliers for more trade credit (Bernanke and Gertler 1995). Empirical studies find evi-dence generally consistent with the substitution relation between bank credit and trade credit. For example, using the 1997 Asian crisis as a shock to bank credit supply, Love, Preve, and Sarria-Allende (2007) show that the amount of trade credit provided and received increases immediate-ly after the crisis. More recently, Garcia-Appendini and Montoriol-Garriga (2013) show that, indeed, cash-rich firms and firms with better access to credit extended more trade credit to their credit-constrained customers during the crisis.  I hypothesize that, during the financial crisis, the positive relation between accounting quality and trade credit is more pronounced for two reasons. First, as the economic uncertainty heightens and credit supply tightens, the opportunity cost of suppliers’ fund also increases    22 (Bernanke and Gertler 1995). As a result, suppliers become more stringent on extending trade credit to their customers. If suppliers use earnings quality to screen their buyer firms, then they will reduce trade credit to firms with lower accounting quality. Second, the scarcity of credit gives suppliers more bargaining power. When suppliers have greater bargaining power, they are able to demand higher accounting quality from their customers (e.g., Hui et al. 2012). Those firms that cannot meet this higher demand of accounting quality are unlikely to obtain more trade credit from suppliers. Therefore, my second hypothesis is: H2: The positive association between accounting quality and the amount of trade credit is more pronounced during the financial crisis.  Product characteristics can also affect the association between accounting quality and the use of trade credit. If a firm files for bankruptcy, its suppliers are entitled to seize the goods they sold to the firm. Suppliers can repossess and sell the goods to other buyers on the market. There-fore, suppliers are (at least partially) secured by the liquidation value of the goods. On the other hand, service providers are not able to reverse the services they have already performed for their customers. Thus, the consequence of information asymmetry becomes more severe and it is more risky for service providers to extend trade credit. Therefore, I hypothesize that the positive asso-ciation between accounting quality and trade credit is stronger when the transacted products are services. Furthermore, within the goods category, standardized goods and differentiated goods have different liquidation values. Suppliers can repossess standardized goods and sell them to other buyers quickly. This is less true for differentiated goods, which are normally tailored to the needs of a specific buyer or a fewer buyers. As differentiated goods have a higher liquidity risk, a supplier will try to reduce the overall risk of extending trade credit by demanding higher quali-ty accounting information. Partially in line with this intuition, Mian and Smith (1992) argue that    23 suppliers of durable goods are more likely to extend trade credit to their customers compared with suppliers of non-durable goods. To summarize the above discussion, my last set of hypothe-ses is: H3a: The positive association between accounting quality and the amount of trade credit is stronger when firms purchase services than when they buy goods. H3b: The positive association between accounting quality and the amount of trade credit is stronger when purchased goods are differentiated than when the goods are standard-ized. 2.3 Empirical proxies and research designs 2.3.1 Measure of accounting quality The first measure of accounting quality is earnings smoothness. Dechow et al. (2010) define it as the variability of earnings relative to that of cash flows from operations. The implicit assumption is that managers have limited ability to change the variability of cash flows from operations. Following their definition, I use a firm’s past five-year earnings and cash flows data to calculate earnings smoothness: ( ) / ( )it it itEARNINGS SMOOTHNESS EARNINGS OPERATING CASH FLOWS      (1)  where  denotes standard deviation, EARNINGS is net income before extraordinary items, and OPERATING CASH FLOWS is cash from operations (all scaled by total assets). Under this defi-nition, the higher the number, the smoother the earnings. The use of a five-year period is in line with prior research and represents a trade-off between sample size and a sufficiently long time-series (e.g. Tucker and Zarowin 2006). The second measure of accounting quality is the C-SCORE developed in Khan and Watts (2009). It captures firm-specific and time-varying asymmetric earnings timeliness (conditional conservatism). Khan and Watts argue that a firm’s conditional conservatism is a linear function    24 of three characteristics (size, book-to-market, and leverage) and estimate the G- SCORE and C- SCORE as follows:21  1 2 3 4it it it it itEARNINGS D RET D RET          (2)  3 1 2 3 4/it it it itG SCORE MKV M B LEV          (3)  4 1 2 3 4/it it it itC SCORE MKV M B LEV          (4)  where EARNINGS is net income before extraordinary items scaled by the market capitalization at the beginning of the year, RET is the one-year stock return, D is an indicator variable that equals to 1 when RET is non-positive, MKV is the natural logarithm of the market capitalization, M/B is the market-to-book ratio, and LEV is the leverage. Equation (2) is the original Basu (1997) re-gression model. Note that equations (3) and (4) are definitions, not regression models. Substitut-ing equation (3) and (4) into the regression model (2) yields: 1 2 1 2 3 41 2 3 41 2 3 4 5 6( / )( / )( / / )it i i it it iti i it it itit it it i it i it i it itEARNINGS D RET MKV M B LEVD RET MKV M B LEVMKV M B LEV D MKV D M B D LEV                             (5)  I estimate this equation using annual cross-sectional regressions to obtain empirical esti-mators i  and i  (i = 1 to 4), which are assumed constant across firms, but can vary over time. Then I calculate the firm- and time-specific C-SCORE using equation (4) and average over the past 5 years to obtain the measure of asymmetric earnings timeliness. Higher values of C-SCORE indicate more conservative financial reporting. Finally, I use abnormal accruals to proxy for the level of earnings management. Abnor-mal accruals are estimated by the model in Francis et al. (2005). Dechow and Dichev (2002)                                                   21 G-SCORE measures the timeliness of earnings reflecting good news, while C-Score measures the timeliness of earnings reflecting bad news (Khan and Watts 2009).     25 originally developed the model to estimate the degree to which earnings are backed by past, current, and future cash flows. Francis et al. modify the model by including the change in sales and PPE (the gross amount of property, plant, and equipment), as proposed by McNichols (2002): 0 1 1 2 3 1 4 5 ,it i i it i it i it i it i it itTCA CFO CFO CFO REV PPE               (6)  where TCA is total current accruals measured by the balance sheet approach (TCA = CA - CL - CASH + STDEBT).22 CFO is cash flow from operations in past, current, and future years (CFO=EARNINGS – TA). EARNINGS is net income before extraordinary items, and TA = CA - CL - CASH + STDEBT - DEPN. CA is change in current assets, CL is change in current liabilities, CASH is change in cash, STDEBT is change in debt in current liabilities, DEPN is depreciation and amortization expense, and REV is the change in sales. Unexplained accruals are abnormal accruals. I accumulate abnormal accruals over the past five years as the measure of earnings management. A higher number means that the firm engages in more upward earnings management. 2.3.2 Regression models I use the following regression model to test H1: 0 1 2 ,it it it itAP ACCOUNTING QUALITY X Industry FE            (7)  where APit is the ratio of accounts payable to total assets (Petersen and Rajan 1997; Giannetti et al. 2011). Scaling by total assets controls for the systematic effect of size (Lev and Sunder 1979). ACCOUNTING QUALITYit is one of the three measures of accounting quality, and Xit is a vector                                                   22 I use the balance sheet approach because my sample dates back to 1977. Collins and Hribar (2002) point out that the cash flow statement approach, first reported in 1986, gives less measurement error in calculating accruals. I also use this measure to test my hypotheses and the results remain qualitatively unchanged.    26 of control variables for the firm. Industry fixed effects are included. H1 predicts that the use of trade credit increases with accounting quality.  Following prior literature (Petersen and Rajan 1997; Giannetti et al. 2011; Garcia-Appendini and Montoriol-Garriga 2013), I include several variables to control for firm character-istics. The first control variable is SIZE. Large firms are able to access external financing more easily, so they are less likely to rely on trade credit. Thus, I expect the coefficient on SIZE to be negative. Next, I include two variables related to firm age, AGE and AGE2, where AGE is the natural logarithm of one plus a firm’s age (Petersen and Rajan 1997). When a firm is young and small, it has limited access to the credit markets and, therefore, uses more trade credit to grow its business. However, when the firm becomes mature, it substitutes the use of trade credit with other sources of finance. Petersen and Rajan (1997) show that this is indeed the case for their small business sample. Therefore, I expect the coefficients on AGE and AGE2 to be positive and negative, respectively. To control for profitability, I include two variables: PROFIT MARGIN and ROA. Both coefficients are expected to be negative because firms with higher profitability are more likely to be able to support their operations with internally generated cash (Giannetti et al. 2011) and rely less on trade credit. I also include SALES GROWTH and expect the coefficient to be positive because higher growth firms normally require more trade credit to support their business growth (Petersen and Rajan 1997). To control for firms’ cash collection ability, I in-clude AR (accounts receivable scaled by total assets) and expect the coefficient to be positive because firms that extend more trade credit are more likely to rely on trade credit from their suppliers (Bougheas, Mateut, and Mizen 2009). As Garcia-Appendini and Montoriol-Garriga (2013) find that firms with higher NET WORTH and higher TOBIN’S Q use less trade credit, I also include them as control variables and expect the coefficients on them to be negative. Firms    27 with more CASH and DEBT are able to use their own cash or tap the bond market to reduce the use of trade credit. Therefore, I expect the coefficients on them to be negative. Finally, I include MARKET SHARE, a firm-level measure to control for bargaining power; it is a firm’s sales over total sales of the industry (defined by two-digit SIC code). I expect the coefficient to be positive as firms with greater bargaining power are more likely to take this advantage and get more trade credit from their suppliers.   Finally, it is important to control for suppliers’ characteristics because the use of trade credit represents an equilibrium outcome between the demand and supply of trade credit. How-ever, it is impossible to find all customer-supplier relationships because a firm is only required by SFAS No. 14 and No. 131 to disclose a customer’s identity information when the sales to the customer exceeds 10 percent of the firm’s total annual sales. Despite the limitation of the data-base, for a subsample that I can identify a firm’s suppliers, I also include suppliers’ characteris-tics as control variables in the following regression model in the same fashion as for customer firms’ characteristics: 0 1 2 3ijt it it jt ijtAP ACCOUNTING QUALITY X X Industry FE               (8)  where  Xjt is a vector of same control variables for a supplier j. To test H2, I adopt the following regression model from Garcia-Appendini and Montori-ol-Garriga (2013): 0 12 3 4,it itt it t ititAP ACCOUNTING QUALITYCRISIS ACCOUNTING QUALITY CRISIS XIndustry FE                (9)  where CRISIS is an indicator variable for the financial crisis period. It equals one if the reporting date is in the crisis period, and zero otherwise. Following Garcia-Appendini and Montoriol-   28 Garriga (2013), I define the crisis period from July 1, 2007 to June 30, 2008.23 Other studies use a longer crisis period. For example, Watts and Zuo (2012) define the crisis period from August 1, 2007 to August 31, 2009. A longer crisis period can increase the size of the sample, and hence, the testing power. However, Garcia-Appendini and Montoriol-Garriga point out that if a longer crisis period is defined, during the later stage of the financial crisis, the product market demand effects can affect customer firms’ demand and use of trade credit. Therefore, in this study, I use a short crisis period to focus on the supply of trade credit. To test hypothesis H3, I also need information about the products that suppliers sell to their customers. Therefore, the test is again restricted to the customer-supplier paired subsample. I obtain product characteristics information from Giannetti et al. (2011), which is based on Rauch (1999). Whether a supplier is classified as a provider of services or goods depends on its two-digit SIC code. Appendix C presents detailed classification information. To investigate the effect of product characteristics, I first use the following regression model: 0 12 4,it itit ititAP ACCOUNTING QUALITYSERVICE ACCOUNTING QUALITY XIndustry FE             (10)  where SERVICE equals to 1 if the transacted product is service, and 0 if the product is goods. Goods providers are further classified as providers of differentiated or standardized goods. The following model is used:                                                   23 Garcia-Appendini and Montoriol-Garriga (2013) use COMPUSTAT quarterly data to study the change in trade credit during the financial crisis. I use annual data in this study, which potentially reduces my chance of finding the relation between trade credit and accounting quality if a firm’s accounts payable turnover is faster so that my annual measure does not capture its variation. In addition, Garcia-Appendini and Montoriol-Garriga argue that a shorter crisis period captures the drop of bank credit supply more accurately.    29 0 12 4,it itit ititAP ACCOUNTING QUALITYDGOODS ACCOUNTING QUALITY XIndustry FE             (11)  where DGOODS equals to 1 if the transacted goods is differentiated, and 0 if the goods is stand-ardized. The classification is based on Appendix provided in Giannetti et al. (2011), which is obtained from Rauch (1999).  2.4 Sample selection and descriptive statistics  2.4.1 Sample selection Table 2.1 Panel A summarizes my sampling procedure for the full sample. I start with the CRSP-COMPUSTAT Merged Database from 1977 to 2010.24 Following prior studies, I exclude finan-cial, insurance, and real estate firms (SIC 6000-6999) and regulated utilities (SIC 4900-4999). I also exclude firms with negative values of total assets, cash, or sales, as well as firms reporting cash greater than total assets, and those with missing values of accounts payable (Garcia-Appendini and Montoriol-Garriga 2013). Finally, I exclude firms with missing variables required for the regression analysis. My full sample has 8,392 firms and 79,783 firm-years. For my re-gression analyses, the exact number of observations varies depending on which measure of ac-counting quality is used. My second and smaller sample contains suppliers' information (Table 2.1 Panel B). I ob-tained a customer-supplier paired dataset from Professor Garcia-Appendini.25 Using the Custom-er Segment File in COMPUSTAT and hand-collected data, Garcia-Appendini and Montoriol-Garriga (2013) collected information about firms’ key customers and compiled this dataset from                                                   24 One of the accounting quality measures, asymmetric earnings timeliness (conservatism), requires stock return data. Therefore, I use the CRSP-COMPUSTAT Merged Database. 25 The data is downloaded from https://sites.google.com/site/mariemigarcia/research/data. It can also be obtained from Journal of Financial Economics at http://jfe.rochester.edu/data.htm.     30 2006 to 2008.26 The original dataset has 5,303 unique customer-supplier pairs. After data re-strictions discussed above, I end up with 1,592 unique customer-supplier pair-years. Finally, Table 2.1 Panel C reports the fiscal year distribution of my full sample. The sample distributes evenly in my sample period.  2.4.2 Descriptive statistics Table 2.2 reports descriptive statistics for all variables in the two samples. To mitigate the influ-ence of outliers, I winsorize all continuous variables at the 1% and 99% levels in each fiscal year. Panel A is the full sample, and Panel B is the customer-supplier paired sample. In Panel A, the average ratio of accounts payable to total assets is 9.4 percent, which is lower than Rajan and Zingales’ (1995) number (15.0 percent in their Table II). The median is 7.5 percent; so the sam-ple is skewed to the right, which means that some firms use trade credit heavily. In Panel B, for the customer-supplier paired sample, the mean and median ratios of accounts payable to total assets are 11.8 percent and 8.5 percent, respectively. These numbers are higher than those of the full sample, which suggests that the use of trade credit is greater when supplier firms have con-centrated customers.  In terms of the accounting quality measures, firms in the customer-supplier paired sample have smoother earnings than firms in the full sample. This is consistent with conventional wis-dom that larger firms have greater flexibility to smooth earnings than smaller firms. Firms in the paired sample seem to use more negative discretionary accruals to manage their earnings down-wards. However, the asymmetric timeliness of earnings in the full sample is on average greater than that of the paired sample.                                                   26 For detailed information on the construction of the dataset, please see Garcia-Appendini and Montoriol-Garriga (2013). They look at the trade credit usage in the financial crisis, which limits their data period.    31 Next I compare firm characteristics between two samples. The comparison shows that firms in the customer-supplier paired sample are generally larger and older, which suggests a survivorship bias. This means one should be cautious about generalizing the findings using this paired sample. The full sample has higher sales growth, accounts receivable, net worth, cash, and debt levels. The customer-supplier paired sample, on the other hand, has higher profit margins, ROA, Tobin’s Q, and larger market share.  Another point worth mentioning is that within the customer-supplier paired sample (Pan-el B), suppliers are, on average, smaller and younger than customers, which suggests that cus-tomers’ greater bargaining power may bias the results. However, ex ante, it is not clear in which direction it will bias the results. Table 2.3 reports the correlation coefficient matrix. Spearman correlations are above the diagonal, and Pearson correlations are below the diagonal. All correlations significant at the 10% level (two-tailed) are highlighted in bold. Consistent with H1, there is a significant positive cor-relation between AP and EARNINGS SMOOTHNESS. The correlation between AP and EARNINGS CONSERVATISM is also significantly positive. Finally, there is a significant nega-tive correlation between AP and EARNINGS MANAGEMENT. The correlations between AP and control variables generally agree with prior studies. 2.5 Empirical results 2.5.1 The association between accounting quality and trade credit Table 2.4 presents the results of testing H1. Panel A shows the results of using the full sample without controlling for suppliers’ characteristics. For all columns, the dependent variable is AP (accounts payable divided by total assets). In column (1), the measure of accounting quality is earnings smoothness. The coefficient is 0.002 and statistically significant at the 1 percent level (t    32 = 4.51). This positive coefficient is consistent with my prediction that firms with smoother earn-ings are able to obtain more trade credit from their suppliers. The magnitude suggests that a one standard deviation increase (std. dev. = 0.980) in earnings smoothness is associated with a 0.002 increase in trade credit. This is 2.1 percent of the mean level of trade credit.  Column (2) uses asymmetric earnings timeliness (conditional conservatism) as the meas-ure of accounting quality.27 The coefficient is also significantly positive at the 1 percent level (0.055, t = 6.45). The result is consistent with the idea that firms reporting more conservatively are able to obtain more trade credit from their suppliers. A one standard deviation increase (std. dev. = 0.109) in earnings conservatism corresponds to a 0.006 increase in trade credit, which is about 6.4 percent of the mean level of trade credit.  In column (3), I use the extent of earnings management to proxy for accounting quality. The coefficient is significantly negative (-0.019, t = -5.87). This suggests that suppliers are aware of their customers’ earnings management so that customer firms with more income-increasing abnormal accruals have a reduced ability to obtain trade credit. A one standard deviation increase (std. dev. = 0.142) in earnings management translates to a 0.003 decrease in trade credit, which is 2.9 percent of the mean level of trade credit. The coefficients on control variables are generally consistent with prior studies. Con-sistent with Petersen and Rajan (1997) and Garcia-Appendini and Montoriol-Garriga (2013), the use of trade credit is negatively associated with firm size. The level of trade credit has a non-linear inverted “U” shape relation with firm age, as shown by the positive coefficient on AGE and the negative coefficient on AGE2, which is also consistent with Petersen and Rajan (1997).                                                   27 Notice that I do not include SIZE, TOBIN’S Q, and DEBT in this regression. This is to avoid multicollinearity as CONSERVATISM (C-SCORE) is a function of size, book-to-market ratio, and leverage. My results keep qualitative-ly similar if I including them in the regression.    33 The negative coefficients on profit margin and return on assets suggest that profitable firms use less trade credit (Petersen and Rajan 1997; Garcia-Appendini and Montoriol-Garriga 2013). The results also confirm that firms with higher sales growth use more trade credit to support their business growth (Petersen and Rajan 1997). In addition, firms with more cash and debt use less trade credit. Finally, firms with more market share use more trade credit. The use of trade credit not only depends on a firm’s demand for credit but also depends on its suppliers’ ability to extend credit. Previous study suggests that it is important to control for the supply of trade credit (Garcia-Appendini and Montoriol-Garriga 2013). Table 2.4 Panel B presents the regression results after controlling for supplier’s characteristics. The coefficients on EARNINGS SMOOTHNESS and EARNINGS MANAGEMENT remain statistically significant at the 1 percent level, while the coefficient on EARNINGS CONSERVATISM becomes less signifi-cant at the 5 percent level. The magnitude of the coefficients, however, is greater than that of full sample (Panel A). Interestingly, the signs on ROA and sales growth are opposite to what I expect. In this subsample, customer firms with greater ROA and less sales growth tend to use more trade credit. As customer firms in this subsample are usually larger than their suppliers, it is possible that they use their bargaining power to exact rent in terms of trade credit from suppliers while they do not have sales growth. 2.5.2 The effect of financial crisis Table 2.5 provides the results of testing H2, which shows that the relation between the use of trade credit and accounting quality is more pronounced during the financial crisis (2006-2008). Panel A shows the results without controlling for suppliers’ characteristics. I am interested in the coefficients on the interaction terms (CRISIS*ACCOUNTING QUALITY). In column (1), ac-counting quality is measured by EARNINGS SMOOTHNESS and the coefficient on the interac-   34 tion term is positive, though only significant at the 10 percent level (0.001, t = 1.72). In column (2), the proxy of accounting quality is EARNINGS CONSERVATISM. The coefficient on the interaction term is positive and significant at the 5 percent level (0.023, t = 2.23). Finally, ac-counting quality is measured by EARNINGS MANAGEMENT in column (3), and the coefficient on the interaction term is negative and significant at the 10 percent level (-0.014, t = -1.77). Con-sistent with H1, the base coefficients on accounting quality are still significant in all three col-umns. The coefficient on CRISIS is negative and statistically significant at different levels in all columns. This result is consistent with Garcia-Appendini and Montoriol-Garriga’s (2013) finding that during the financial crisis there was a drop in the amount of trade credit. Table 2.5 Panel B presents similar testing results of H2 with the inclusion of suppliers’ characteristics. In general, the results are similar to those for the full sample in Panel A. Alt-hough the coefficients are less statistically significant, their signs are consistent with the hypoth-esis. Overall, the results in this subsection suggest that, during the financial crisis, the positive association between accounting quality and trade credit is stronger.  2.5.3 The effect of product characteristics I show the results of testing H3 in Table 2.6. My H3 predicts that the positive relation between accounting quality and trade credit is more pronounced when the transacted products are services. Therefore, I am interested in the coefficient on the interaction between SERVICE and ACCOUNTING QUALITY, which should have the same sign as the coefficient on ACCOUNTING QUALITY. In column 1, accounting quality is measured by earnings smoothness. The coefficient on it is positive, however, not significant. The coefficient on the interaction term (SERVICE*EARNINGS SMOOTHNESS) is statistically significant (0.001, t = 2.04), and the sign agrees with that of coefficients on EARNINGS SMOOTHNESS (both positive), which is con-   35 sistent with my prediction. The results are similar for two other accounting quality measures (columns 2 and 3). Thus, the results seem to suggest that service providers demanding higher quality accounting information from their customers to compensate for the higher risk of service provision.  Panel B further separates goods into standardized goods and differentiated goods. The re-sults are very similar to those presented in Panel A. Consistent with the idea that the liquidation cost of differentiated goods is higher and suppliers demand greater accounting quality, I find that the positive association between accounting quality and trade credit is stronger when the product is differentiated goods.  2.6 Additional tests 2.6.1 The effect of firm size Prior studies also show that the use of trade credit is concentrated in small firms because these firms have limited access to the credit markets and are likely to be credit-constrained (Petersen and Rajan 1997). In addition, Nilsen (2002) shows that small firms borrow more from their sup-pliers during monetary contractions. For these reason, I examine the effect of firm size on the relation between accounting quality and trade credit. I partition the full sample by the median size in each fiscal year. If a firm’s size is smaller than the median in a year, it is a small firm and I assign SMALL to be one; otherwise, it is zero. I am interested in the coefficient on the interac-tion between SMALL and ACCOUNTING QUALITY. The results are shown in Table 2.7. Again, three columns are for three different accounting quality measures. The coefficients on the inter-action term (SMALL*ACCOUNTING QUALITY) are significant on all three measures, suggest-ing that the effect is indeed stronger for smaller firms.    36 2.6.2 The effect of credit rating Firms’ credit risk can potentially affect the relation between accounting quality and trade credit. Prior studies document that having a credit rating increases a firm’s ability to obtain external financing (Faulkender and Petersen 2008) and firms without credit ratings have limited access to credit markets and use more trade credit (Petersen and Rajan 1997). If a firm has a credit rating, its demand for suppliers’ trade credit is expected to decrease. On the other hand, when a firm has a credit rating, its suppliers are also likely to use the rating to evaluate the firm’s creditworthi-ness and reduce the use of accounting information to screen customers. Therefore, I predict the positive association between accounting quality and trade credit to be stronger for firms without ratings.  Following Jiang (2008), to test the effect of credit rating, I collect firms’ senior debt rat-ings from the annual COMPUSTAT file available between 1985 and 2010. I then classify firms based on whether the firm has its senior debt rated. If there is no senior debt rating, then NORATING equals one; otherwise, it is zero. The results are reported in Table 2.8. Consistent with my prediction, the effect is more pronounced among firms with no rated senior debt, as the coefficients on the interaction term (NORATING*ACCOUNTING QUALITY) are statistically significant.  2.7 Conclusions  Trade credit is an important source of short-term financing and this paper seeks to understand whether accounting quality affects firms’ ability to obtain trade credit. Consistent with the intui-tion that higher quality accounting information mitigates information asymmetry between a firm and its outsiders, I find that firms with higher earnings quality (measured by earnings smooth-ness, earnings conservatism, and earnings management) are able to obtain more trade credit from    37 their suppliers. The positive relation between trade credit and accounting quality is robust after controlling for suppliers’ characteristics. Additional analyses show that the positive association is concentrated in small firms and firms without credit ratings on senior debt. Using the 2007-2008 financial crisis as an exogenous shock to the supply of bank credit, I investigate the impact of the liquidity shock on the association between accounting quality and trade credit. Consistent with suppliers becoming more stringent on their lending during the crisis and using accounting information to more intensely scrutinize their buyers, I find that the posi-tive association is more pronounced during the financial crisis.  Finally I predict that product characteristics also affect the relation between accounting quality and trade credit. Specifically, if the transacted products are services instead of goods, the positive relation between accounting quality and trade credit is stronger because the collateral value of services is lower. Furthermore, within the goods category, the positive association is stronger for differentiated goods since the liquidation of differentiated goods takes more time.      38 Table 2.1: Sample selection procedures and sample distribution  This table describes my sample selection procedures (Panel A and Panel B) and the frequency distribution of sample firms by fiscal year (Panel C). Panel A is for the full sample. Starting with the CRSP-COMPUSTAT Merged database from 1977 to 2010, I exclude financial, insurance, and real estate firms (SIC 6000-6999) and regulated utilities (SIC 4900-4999). I also exclude firms with negative values of total assets, cash, or sales, as well as firms reporting cash greater than total assets, and those with missing values of accounts payable. Finally, I exclude firms with missing variables required for regression analysis. My full sample has 8,392 firms and 79,783 firm-years. Panel B is for the customer-supplier paired sample from year 2006 to 2008. I start with 5,303 customer-supplier pair-years from Professor Garcia-Appendini’s dataset, in which financial firms have already been removed. After further removing utility firms and firms with missing variables, I have 1,592 customer-supplier pair-years left. Panel C is the frequency distri-bution of full sample firms by fiscal year from 1977 to 2010.   Panel A:   Sampling procedure Sample A. Full sample  (1977-2010)  Firms Firm-years  CRSP-COMPUSTAT universe 11,065 121,745  Excluding financial, insurance, and real estate firms  (SIC 6000-6999) -2,132 -21,488  Excluding regulated utilities (SIC 4900-4999) -238 -5,213  Excluding firms with missing variables  -303 -15,261      Sample A: Whole COMPUSTAT firms 8,392 79,783     Panel B:  Sampling procedure Sample B. Customer-supplier paired sample  (2006-2008) Pair-years   Initial dataset 5,303  Excluding regulated utilities (SIC 4900-4999) -93  Excluding firms with missing variables  -3,618     Sample B: Customer-supplier pair-years 1,592         39 Table 2.1 (continued)               Panel C:              Fiscal year Firms Percent   Fiscal year Firms Percent 1977 2,113 2.65  1994 2,502 3.14 1978 2,384 2.99  1995 2,527 3.17 1979 2,486 3.12  1996 2,578 3.23 1980 2,397 3.00  1997 2,622 3.29 1981 2,343 2.94  1998 2,635 3.30 1982 2,238 2.81  1999 2,620 3.28 1983 2,228 2.79  2000 2,536 3.18 1984 2,172 2.72  2001 2,541 3.18 1985 2,093 2.62  2002 2,510 3.15 1986 2,097 2.63  2003 2,515 3.15 1987 2,051 2.57  2004 2,552 3.20 1988 2,172 2.72  2005 2,564 3.21 1989 2,148 2.69  2006 2,402 3.01 1990 2,157 2.70  2007 2,239 2.81 1991 2,331 2.92  2008 2,068 2.59 1992 2,440 3.06  2009 2,045 2.56 1993 2,476 3.10   2010 2,001 2.51         Total 79,783 100     40 Table 2.2: Descriptive statistics  This table presents descriptive statistics for all variables. AP is accounts payable scaled by total assets. EARNINGS SMOOTHNESS is the variance of earnings divided by the variance of cash flows over the past five years and then multiply by (-1). EARNINGS CONSERVATISM is the average C_Score over the past five years, where C_Score is developed in Khan and Watts (2009). EARNINGS MANAGEMENT is the accumulated abnormal accruals over the past five years, where abnormal accruals are estimated using Francis et al.’s (2005) model. SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is ac-counts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. Sample A is my full sample, while sample B is the customer-supplier paired sample. All continuous variables are winsorized at 1% and 99% level in each fiscal year. See Appendix B for detailed variable definitions.   41 Table 2.2 (continued)  Panel A: Full sample Variables N Mean Std. Dev. Min Q1 Median Q3 Max Dependent variable:         AP 79,783 0.094 0.073 0.002 0.044 0.075 0.121 0.449 Accounting quality:         EARNINGS SMOOTHNESS 79,783 -1.055 0.980 -11.126 -1.175 -0.823 -0.546 -0.097 EARNINGS CONSERVATISM 52,002 0.120 0.109 -0.290 0.047 0.122 0.196 0.507 EARNINGS MANAGEMENT 63,350 -0.002 0.142 -0.502 -0.079 -0.004 0.072 0.530 Firm characteristics:         SIZE 79,783 5.088 1.990 0.130 3.618 4.969 6.472 10.671 AGE 79,783 2.625 0.572 0 2.197 2.639 3.045 3.892 AGE2 79,783 7.219 2.972 0 4.828 6.965 9.269 15.146 PROFIT MARGIN 79,783 0.297 0.633 -3.424 0.214 0.318 0.456 0.928 ROA 79,783 0.013 0.138 -0.911 -0.001 0.042 0.079 0.327 SALES GROWTH 79,783 0.128 0.394 -0.743 -0.019 0.082 0.196 6.768 AR 79,783 0.189 0.124 0 0.096 0.176 0.260 0.662 NET WORTH 79,783 0.518 0.203 0.032 0.373 0.510 0.668 0.966 TOBIN'S Q 79,783 1.625 1.220 0.413 0.977 1.252 1.798 25.289 CASH 79,783 0.139 0.171 0 0.023 0.069 0.187 0.943 DEBT 79,783 0.221 0.179 0 0.060 0.203 0.338 0.784 MARKET SHARE 79,783 0.020 0.064 0 0.001 0.003 0.012 1       42 Table 2.2 (continued)  Panel B: Customer-supplier paired sample Variables N Mean Std. Dev. Min Q1 Median Q3 Max Dependent variable:         AP 496 0.118 0.103 0.003 0.043 0.085 0.160 0.404 Accounting quality:         EARNINGS SMOOTHNESS 496 -0.973 0.909 -6.930 -1.113 -0.766 -0.476 -0.124 EARNINGS CONSERVATISM 368 0.036 0.086 -0.085 -0.036 0.029 0.089 0.312 EARNINGS MANAGEMENT 445 -0.043 0.124 -0.405 -0.094 -0.033 0.022 0.373 Firm characteristics:         SIZE 496 8.511 1.659 2.719 7.558 8.757 9.962 10.504 AGE 496 3.118 0.622 0.693 2.639 3.135 3.761 3.850 AGE2 496 10.107 3.730 0.480 6.965 9.831 14.147 14.824 PROFIT MARGIN 496 0.325 0.478 -5.496 0.196 0.308 0.510 0.905 ROA 496 0.046 0.108 -0.853 0.030 0.062 0.094 0.280 SALES GROWTH 496 0.119 0.298 -0.508 0.021 0.079 0.147 2.705 AR 496 0.145 0.118 0 0.055 0.127 0.198 0.572 NET WORTH 496 0.451 0.175 0.032 0.339 0.441 0.557 0.911 TOBIN'S Q 496 1.777 0.777 0.557 1.215 1.566 2.152 5.924 CASH 496 0.125 0.156 0 0.024 0.063 0.168 0.943 DEBT 496 0.210 0.146 0 0.093 0.197 0.310 0.713 MARKET SHARE 496 0.077 0.116 0 0.009 0.034 0.094 0.659 Supplier's characteristics:       SIZE 1,124 5.951 1.890 2.041 4.494 5.846 7.307 10.504 AGE 1,124 2.843 0.565 0.693 2.398 2.773 3.258 3.850 AGE2 1,124 8.402 3.261 0.480 5.750 7.687 10.615 14.824 PROFIT MARGIN 1,124 0.311 0.655 -7.703 0.230 0.364 0.536 0.922 ROA 1,124 -0.012 0.172 -0.853 -0.038 0.037 0.078 0.288 SALES GROWTH 1,124 0.130 0.337 -0.614 -0.028 0.074 0.214 2.705 NET WORTH 1,124 0.560 0.218 0.032 0.397 0.566 0.750 0.943 TOBIN'S Q 1,124 1.797 1.118 0.432 1.096 1.491 2.100 8.148 CASH 1,124 0.220 0.219 0 0.038 0.143 0.346 0.943 DEBT 1,124 0.180 0.174 0 0.007 0.140 0.302 0.750 MARKET SHARE 1,124 0.017 0.065 0 0.000 0.002 0.007 1     43 Table 2.3: Correlation matrix  This table presents the correlation matrix. Spearman correlations are above the diagonal, and Pearson correlations are below the diagonal. All correlations significant at the 10% level (two-tailed) are highlighted in bold.    (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (1) AP 1 0.10 0.03 -0.04 0.05 0.20 0.20 -0.41 0.03 0.01 0.43 -0.26 -0.12 -0.23 0.03 0.22 (2) EARNINGS SMOOTHNESS 0.08 1 -0.05 0.00 0.11 0.09 0.09 -0.12 0.14 0.02 0.04 -0.07 -0.06 -0.20 0.05 0.18 (3) EARNINGS CONSERVATISM 0.01 -0.05 1 0.13 -0.87 -0.16 -0.16 -0.09 -0.26 -0.08 0.16 0.25 -0.23 0.10 -0.20 -0.64 (4) EARNINGS MANAGEMENT -0.08 -0.02 0.12 1 -0.10 -0.04 -0.04 -0.02 -0.04 0.03 0.11 0.11 -0.08 -0.09 0.04 -0.08 (5) SIZE 0.08 0.07 -0.87 -0.09 1 0.23 0.23 -0.06 0.16 0.06 -0.16 -0.47 -0.04 -0.31 0.43 0.78 (6) AGE 0.12 0.08 -0.16 -0.04 0.24 1 1.00 -0.11 0.11 -0.11 0.11 -0.13 -0.09 -0.17 0.09 0.25 (7) AGE2 0.12 0.08 -0.17 -0.04 0.25 1.00 1 -0.11 0.11 -0.11 0.11 -0.13 -0.09 -0.17 0.09 0.25 (8) PROFIT MARGIN -0.04 0.02 -0.05 -0.01 0.07 0.05 0.04 1 0.19 0.09 -0.23 0.27 0.31 0.26 -0.15 -0.25 (9) ROA 0.01 0.15 -0.20 0.00 0.23 0.13 0.13 0.36 1 0.29 0.11 0.19 0.51 0.03 -0.19 0.20 (10) SALE GROWTH -0.01 0.01 -0.02 0.02 0.02 -0.10 -0.10 -0.03 0.09 1 0.05 0.04 0.31 -0.01 -0.01 0.00 (11) AR 0.48 0.05 0.17 0.12 -0.15 0.09 0.09 0.05 0.08 0.00 1 -0.01 0.00 -0.09 -0.14 0.01 (12) NET WORTH -0.28 0.00 0.23 0.10 -0.45 -0.13 -0.15 0.03 0.10 0.02 -0.04 1 0.22 0.50 -0.78 -0.46 (13) TOBIN’S Q -0.11 0.01 -0.15 -0.07 -0.13 -0.12 -0.12 -0.02 0.17 0.20 -0.02 0.24 1 0.32 -0.27 -0.10 (14) CASH -0.22 -0.12 0.14 -0.09 -0.36 -0.20 -0.21 -0.17 -0.20 0.05 -0.18 0.47 0.33 1 -0.54 -0.39 (15) DEBT -0.07 -0.04 -0.15 0.05 0.37 0.04 0.04 -0.03 -0.11 0.01 -0.18 -0.77 -0.26 -0.42 1 0.36 (16) MARKET SHARE 0.13 0.06 -0.26 -0.02 0.32 0.11 0.11 -0.01 0.07 -0.02 0.04 -0.18 -0.05 -0.13 0.10 1       44 Table 2.4: The relation between trade credit and accounting quality  Panel A:  This table tests the association between firms’ use of trade credit and accounting quality using the full sample from fiscal year 1977 to 2010. The OLS regression model is:  0 1 2it it it itAP ACCOUNTING QUALITY X Industry FE Year FE              The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (col-umn 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. Stand-ard errors are clustered at the firm and year level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, respectively. See Appendix B for detailed variable definitions.       45 Table 2.4 (continued)    Predicted Dependent variable: AP Independent variable sign (1) (2) (3) INTERCEPT  0.290*** 0.038*** 0.281***   (17.56) (2.67) (14.09) EARNINGS SMOOTHNESS + 0.002***     (4.51)   EARNINGS CONSERVATISM +  0.055***     (6.45)  EARNINGS MANAGEMENT -   -0.019***     (-5.87) SIZE - -0.007***  -0.007***   (-14.12)  (-11.86) AGE + 0.012*** 0.016* 0.011   (2.91) (1.84) (1.58) AGE2 - -0.003*** -0.003** -0.003**   (-3.77) (-1.99) (-2.46) PROFIT MARGIN - -0.004*** -0.006*** -0.004***   (-4.04) (-3.03) (-3.32) ROA - -0.012*** -0.003 -0.013***   (-3.87) (-0.58) (-3.99) SALES GROWTH + 0.008*** 0.004*** 0.009***   (10.52) (3.40) (8.84) AR + 0.112*** 0.205*** 0.125***   (10.10) (14.79) (9.60) NET WORTH - -0.288*** -0.090*** -0.278***   (-20.65) (-16.68) (-18.90) TOBIN'S Q - -0.001**  -0.001**   (-2.37)  (-2.46) CASH - -0.036*** 0.003 -0.037***   (-10.66) (0.73) (-9.46) DEBT - -0.267***  -0.256***   (-22.66)  (-20.31) MARKET SHARE + 0.056*** 0.076*** 0.062***   (4.36) (4.25) (4.45)      INDUSTRY FE Yes Yes Yes YEAR FE  Yes Yes Yes      Number of observations 79,783 52,002 63,350 R-squared 0.509 0.521 0.521        46 Table 2.4 (continued)  Panel B:  This table tests the association between trade credit and earnings quality controlling for suppli-er’s characteristics. Customer-supplier pairs between 2006 and 2008 are obtained from Professor Garcia-Appendini. The OLS regression model is: 0 1 2 3 .ijt it it jt ijtAP ACCOUNTING QUALITY X X Industry FE Year FE                 The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (col-umn 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. The variable definitions for suppliers are the same. Standard errors are clustered at the firm and year level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, respectively. See Ap-pendix B for detailed variable definitions.         47 Table 2.4 (continued)    Predicted Dependent variable is AP Independent variable sign (1) (2) (3) INTERCEPT  0.057 -0.403*** -0.306*   (0.39) (-3.23) (-1.83) EARNINGS SMOOTHNESS + 0.021***     (2.71)   EARNINGS CONSERVATISM +  0.077**     (1.94)  EARNINGS MANAGEMENT -   -0.139***     (-3.20) Firm variables:    SIZE - -0.010**  -0.006   (-2.34)  (-1.29) AGE + 0.238*** 0.364*** 0.409***   (3.31) (5.09) (4.08) AGE2 - -0.046*** -0.069*** -0.071***   (-3.80) (-6.14) (-4.37) PROFIT MARGIN - -0.076*** -0.176*** -0.095***   (-7.64) (-8.85) (-5.07) ROA - 0.094*** 0.176*** 0.085*   (2.59) (5.13) (1.75) SALES GROWTH + -0.056* -0.027 -0.057   (-1.87) (-1.00) (-1.60) AR + 0.310*** 0.355*** 0.408***   (10.47) (7.98) (10.06) NET WORTH - -0.324*** -0.082** -0.261***   (-12.27) (-2.09) (-7.78) TOBIN'S Q - 0.006  0.013   (0.77)  (1.26) CASH - -0.093 0.028 -0.114**   (-1.29) (0.51) (-2.46) DEBT - -0.342***  -0.248***   (-9.51)  (-7.78) MARKET SHARE + 0.177*** 0.340*** 0.149***   (13.35) (4.92) (8.08)            48 Table 2.4 (continued)    Predicted Dependent variable is AP Independent variable sign (1) (2) (3) Supplier variables:     SIZE  0.008*** 0.010*** 0.004***   (6.35) (4.30) (3.09) AGE  0.016 0.016 0.010   (0.65) (0.33) (0.30) AGE2  -0.005 -0.006 -0.003   (-1.01) (-0.62) (-0.46) PROFIT MARGIN  0.012*** 0.031*** 0.010***   (5.78) (3.97) (4.58) ROA  -0.028** -0.061*** -0.008   (-1.99) (-3.96) (-0.74) SALES GROWTH  -0.003 -0.002 -0.007   (-0.94) (-0.20) (-1.63) NET WORTH  0.033** 0.058 0.019   (2.44) (1.52) (1.47) TOBIN'S Q  0.002 -0.002 0.001   (1.25) (-0.68) (0.52) CASH  0.027*** 0.027 0.011   (3.18) (1.64) (1.32) DEBT  0.028* 0.028 0.023   (1.79) (0.55) (1.32) MARKET SHARE  -0.007 -0.051 0.081**   (-0.11) (-0.39) (2.40)      INDUSTRY FE Yes Yes Yes YEAR FE  Yes Yes Yes      Number of observations 1,592 752 1,465 R-squared   0.718 0.747 0.755     49 Table 2.5: The impact of financial crisis on the relation between trade credit and accounting quality  Panel A:  This table tests the association between trade credit and earnings quality in the financial crisis period using the full sample between 2006 and 2008. The OLS regression model is: 0 1 23 4,it it tit t ititAP ACCOUNTING QUALITY CRISISACCOUNTING QUALITY CRISIS XIndustry FE                 The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (col-umn 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. For simplicity, the coefficients on control variables are not reported. Robust standard errors are clus-tered at the firm level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, re-spectively. See Appendix B for detailed variable definitions.   50 Table 2.5 (continued)    Predicted Dependent variable: AP Independent variable sign (1) (2) (3) INTERCEPT  0.028 0.009 0.169   (1.47) (0.37) (5.62) EARNINGS SMOOTHNESS + 0.001***     (4.19)   EARNINGS CONSERVATISM +  0.049***     (2.96)  EARNINGS MANAGEMENT -   -0.025**     (-2.33) CRISIS - -0.003** -0.003*** -0.005*   (-2.28) (-2.74) (-1.65) CRISIS*EARNINGS SMOOTHNESS + 0.001*     (1.72)   CRISIS*EARNINGS CONSERVATISM +  0.023**     (2.23)  CRISIS*EARNINGS MANAGEMENT -   -0.014*     (-1.77)      Firm control variables  Yes Yes Yes Industry FE  Yes Yes Yes      Number of observations  6,709 4,809 5,790 R-squared   0.539 0.575 0.556       51 Table 2.5 (continued)  Panel B:  This table tests the impact of financial crisis on the relation between trade credit and earnings quality controlling for supplier’s characteristics. Customer-supplier pairs between 2006 and 2009 are obtained from Professor Garcia-Appendini. The OLS regression model is: 0 1 23 4 5.ijt it tt it it jtijtAP ACCOUNTING QUALITY CRISISCRISIS ACCOUNTING QUALITY X XIndustry FE                   The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (col-umn 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. For simplicity, the coefficients on control variables are not reported. Robust standard errors are clus-tered at the firm level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, re-spectively. See Appendix B for detailed variable definitions.   52 Table 2.5 (continued)    Predicted Dependent variable: AP Independent variable sign (1) (2) (3) INTERCEPT  0.051 0.247 0.025   (0.48) (1.23) (0.15) EARNINGS SMOOTHNESS + 0.004***     (3.25)   EARNINGS CONSERVATISM +  0.213*     (1.72)  EARNINGS MANAGEMENT -   -0.023*     (-1.74) CRISIS - -0.003 -0.009*** -0.002   (-1.43) (-2.19) (-0.36) CRISIS*EARNINGS SMOOTHNESS + 0.002*     (1.79)   CRISIS*EARNINGS CONSERVATISM +  0.055*     (1.77)  CRISIS*EARNINGS MANAGEMENT -   -0.027*     (-1.73)      Firm control variables  Yes Yes Yes Supplier control variables  Yes Yes Yes Industry FE  Yes Yes Yes      Number of observations  1,592 752 1,465 R-squared   0.894 0.805 0.813    53 Table 2.6: The impact of product characteristics on the relation between trade credit and account-ing quality  Panel A:  This table tests the impact of product characteristics on the relation between trade credit and accounting quality controlling for supplier’s characteristics. Customer-supplier pairs between 2006 and 2008 are obtained from Professor Garcia-Appendini. The OLS regression model is: 0 1 24 ,it it i itit itAP ACCOUNTING QUALITY SERVICE ACCOUNTING QUALITYX Industry FE              The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (col-umn 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). SERVICE equals to 1 if the transacted product is service, and 0 if the product is goods. SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. For simplicity, the coefficients on control variables are not reported. Robust standard errors are clus-tered at the pair level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, respec-tively. See Appendix B for detailed variable definitions.                54 Table 2.6 (continued)    Predicted Dependent variable: AP Independent variable sign (1) (2) (3) INTERCEPT  0.287*** 0.040*** 0.059***   (17.32) (2.78) (6.93) EARNINGS SMOOTHNESS + 0.001     (1.41)   SERVICE*EARNINGS SMOOTHNESS + 0.001**     (2.04)   EARNINGS CONSERVATISM +  0.044     (1.54)  SERVICE *EARNINGS CONSERVATISM +  0.011***     (3.28)  EARNINGS MANAGEMENT -   -0.027     (-1.24) SERVICE *EARNINGS MANAGEMENT -   -0.014***     (-3.85)      Firm control variables  Yes Yes Yes Industry FE  Yes Yes Yes      Number of observations  1,592 752 1,465 R-squared   0.508 0.402 0.218      55  Table 2.6 (continued)  Panel B:  This table tests the impact of goods differentiation on the relation between trade credit and ac-counting quality controlling for supplier’s characteristics. Customer-supplier pairs between 2006 and 2008 are obtained from Professor Garcia-Appendini. The OLS regression model is:  0 1 24 ,it it i itit itAP ACCOUNTING QUALITY DGOODS ACCOUNTING QUALITYX Industry FE             The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (column 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). DGOODS equals to 1 if the transacted goods is differentiated, and 0 if the goods is standardized. SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. For simplicity, the coefficients on control variables are not reported. Robust standard errors are clus-tered at the pair level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, respec-tively. See Appendix B for detailed variable definitions.                56 Table 2.6 (continued)    Predicted Dependent variable: AP Independent variable sign (1) (2) (3) INTERCEPT  0.103 0.120 0.128   (17.32) (2.78) (6.93) EARNINGS SMOOTHNESS + 0.002*     (1.83)   DGOODS*EARNINGS SMOOTHNESS + 0.003**     (2.57)   EARNINGS CONSERVATISM +  0.069     (1.23)  DGOODS *EARNINGS CONSERVATISM +  0.057**     (2.25)  EARNINGS MANAGEMENT -   -0.087     (-1.57) DGOODS *EARNINGS MANAGEMENT -   -0.091**     (-2.83)      Firm control variables  Yes Yes Yes Industry FE  Yes Yes Yes      Number of observations  1,196 580 1,105 R-squared   0.473 0.378 0.203      57 Table 2.7: The effect of firm size on the relation between trade credit and accounting quality  This table tests the effect of firm size on the relation between trade credit and accounting quality. The OLS regression model is: 0 1 23it it it itit itAP ACCOUNTING QUALITY SMALL ACCOUNTING QUALITYX Industry FE Year FE                . The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (col-umn 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). For each year, if the firm size is below the median, SMALL equals to one. Otherwise, it is zero. SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. For simplicity, the coefficients on control variables are not reported. Standard errors are clustered at the firm and year level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, re-spectively. See Appendix B for detailed variable definitions.                  58 Table 2.7 (continued)    Predicted Dependent variable: AP Independent variable sign (1) (2) (3) INTERCEPT  0.287*** 0.040*** 0.059***   (17.32) (2.78) (6.93) EARNINGS SMOOTHNESS + 0.001**     (2.53)   SMALL*EARNINGS SMOOTHNESS + 0.001**     (2.04)   EARNINGS CONSERVATISM +  0.044***     (7.57)  SMALL*EARNINGS CONSERVATISM +  0.011***     (4.28)  EARNINGS MANAGEMENT -   -0.027***     (-5.53) SMALL*EARNINGS MANAGEMENT -   -0.015***     (-4.85)      Firm control variables  Yes Yes Yes Industry FE  Yes Yes Yes Year FE  Yes Yes Yes      Number of observations  79,783 52,002 63,350 R-squared   0.508 0.402 0.218             59 Table 2.8: The effect of credit rating on the relation between trade credit and accounting quality for low and high credit risk firms  This table tests the effect of credit rating on the relation between trade credit and accounting quality. The OLS regression model is: 0 1 23it it it itit itAP ACCOUNTING QUALITY NORATING ACCOUNTING QUALITYX Industry FE Year FE                . The dependent variable is AP (the ratio of accounts payable to total assets). The independent variable is ACCOUNTING QUALITY, which is measured by EARNINGS SMOOTHNESS (col-umn 1), EARNINGS CONSERVATISM (column 2), and EARNINGS MANAGEMENT (column 3). I collect firms’ senior debt ratings from the annual COMPUSTAT file available between 1985 and 2010. If a firm has no rating on its senior debt, NORATING equals to one. Otherwise, it is zero. SIZE is the natural logarithm of total assets. AGE is the natural logarithm of (1+ firm age), where firm age is the number of years since a firm’s first appearance in the CRSP database. AGE2 is the square of AGE. PROFIT MARGIN is gross profit scaled by total sales. ROA is return on assets. SALE GROWTH is this year total sales divided by last year’s total sales and then minus 1. AR is accounts receivable scaled by total assets. NET WORTH is total book value of equity scaled by total assets. TOBIN’S Q is the sum of total market capitalization and liabilities scaled by total assets. CASH is total cash and cash equivalent scaled by total assets. DEBT is the sum of long-term debt and debt in current liabilities scaled by total assets. MARKET SHARE is a firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code. For simplicity, the coefficients on control variables are not reported. Standard errors are clus-tered at the firm and year level. ***, **, * denote significance level at the 1%, 5%, and 10% levels, respectively. See Appendix B for detailed variable definitions.             60 Table 2.8 (continued)    Predicted Dependent variable: AP Independent variable sign (1) (2) (3) INTERCEPT  0.288*** 0.039*** 0.059***   (17.47) (2.77) (6.97) EARNINGS SMOOTHNESS + 0.005***     (3.09)   NORATING*EARNINGS SMOOTHNESS + 0.001**     (2.27)   EARNINGS CONSERVATISM +  0.123***     (3.10)  NORATING*EARNINGS CONSERVATISM +  0.011***     (3.63)  EARNINGS MANAGEMENT -   -0.215***     (-5.23) NORATING*EARNINGS MANAGEMENT -   -0.011***     (-4.29)      Firm control variables  Yes Yes Yes Industry FE  Yes Yes Yes Year FE  Yes Yes Yes      Number of observations  61,422 32,004 44,344 R-squared   0.509 0.402 0.219      61  Figure 2.1: The use of trade credit in the U. S.   This figure shows the total amount of trade credit (in $ trillion) extended by U. S. non-financial businesses from 2003 to 2012. Data are from the U.S. Flow of Funds Account at http://www.federalreserve.gov/releases/z1/.      0.000.200.400.600.801.001.201.401.601.802.002003 2004 2005 2006 2007 2008 2009 2010 2011 2012   62 Chapter 3: Stock underpricing and earnings management—evidence from mutual fund fire sales 3.1 Introduction Prior studies suggest that executives engage in earnings management in an attempt to increase stock prices when their firms’ performance deteriorates.28 However, less is known about whether executives also manage earnings when their firms’ underlying performance displays no problem but stock prices nevertheless deviate from fundamentals and their firms become undervalued. Although in both cases stock prices experience the same downturn, the former is a reflection of a firm’s underlying performance, while the latter is a consequence of stock market mispricing. This study focuses on the latter context and examines whether companies manage earnings up-wards in an attempt to correct stock underpricing. Rather than being opportunistic, the potential earnings management in this context is undertaken with good intentions, which has rarely been documented. Some scholars argue that stock mispricing quickly disappears as sophisticated traders are able to identify underpricing and trade against it (e.g., Friedman 1953); this would make earnings management for the purpose of correcting the underpricing unnecessary. However, theoretical model suggests that mispricing can persist when the trading costs are higher than potential profits (Shleifer and Vishny 1997). For example, when a particular stock becomes undervalued, there may not be enough buyers who are interested in the stock. In this case, underpricing can last longer since there is not enough demand for the stock.                                                    28 While there is no study explicitly arguing that firms do so to increase stock prices, empirical evidence from two lines of literature suggests that is likely the case. One line of literature shows that executives manage earnings up-wards to meet/beat earnings benchmarks, such as analysts’ forecasts, and to avoid reporting losses (e.g., Burgstahler and Dichev 1997; Degeorge, Patel, and Zeckhauser 1999). The other line of literature finds that stock market re-wards firms meeting or beating earnings targets (Bartov, Givoly, and Hayn 2002) and punishes firms missing earn-ings targets (Skinner and Sloan 2002). For general review on managers’ opportunistic earnings management, see Fields, Lys, and Vincent (2001).    63 Persistent underpricing is associated with detrimental outcomes. Hau and Lai (2013) show that undervalued firms significantly lower their investment and employment levels relative to their unaffected peers. Moreover, underpricing diminishes executives’ compensation and job security. If there are long-lasting negative effects of underpricing, managers have sufficient mo-tivation to help stock prices increase to their fundamental values. Prior literature suggests that managers can influence stock prices, at least partially, by managing earnings through discretion-ary accruals (e.g., Xie 2001). Following this line of thinking, this paper conjectures that stock undervaluation can induce an increase in earnings management that is measured by discretionary accruals.  The empirical challenge for testing earnings management in cases of stock undervalua-tion is to identify such cases where the undervaluation is unrelated to firms’ performance. This paper responds to this challenge by using the setting of mutual fund “fire sales.” Mutual fund fire sales occur when several funds react to their investors’ capital withdrawals by selling a limited number of stocks simultaneously.29 Coval and Stafford (2007) provide evidence suggesting that this selling pressure pushes stock prices below their fundamental values, causing stock under-pricing. In particular, they observe that there is a major difference between the prices of stocks subject to fire sales and the prices of stocks being sold voluntarily by unconstrained funds: over the 12-month period after fire sales, the former’s prices rebound 6.13%, while the latter’s prices do not. They interpret this finding as evidence that mutual fund fire sales are driven by the need for liquidity, as opposed to information, and that the resulting mispricing is unrelated to firms’                                                   29 The selling can be concentrated in a limited number of stocks because of similar investment strategies (e.g., mer-ger arbitrage) among several funds.    64 performance. The setting of mutual fund fire sales has been used recently in finance and account-ing research to examine the impact of stock mispricing on firms and managers.30 Using mutual fund transaction data from 1996 to 2006, I identify fire sale stocks that ex-perienced mutual fund forced sales due to large capital outflows. To construct my control firms, I select firms that are traded moderately by financially unconstrained mutual funds. I use perfor-mance-matched discretionary accruals as the proxy for earnings management (Kothari, Leone, and Wasley 2005).  By using difference-in-differences tests, I find that firms experiencing fire sales have a higher level of income-increasing earnings management. On average, fire sale firms increase their discretionary accruals level by 0.3 percent of total assets after they experience fire sales.  Next, I examine whether the relation between underpricing and earnings management is affected by the extent and the duration of underpricing. First, I show that the severity of fire sales affects the degree of earnings management: firms manage their earnings upwards more after steeper stock price declines. Second, prior literature suggests that when information asymmetry is high, trading cost is also high (Diamond and Verrecchia 1991), which delays the correction of mispricing. Thus, I predict that income-increasing earnings management is also concentrated in firms with high information asymmetry. Using analyst coverage to proxy for the level of infor-mation asymmetry, I find that earnings management is stronger in firms with lower analyst cov-erage, which is consistent with my prediction. Third, Oehmke (2009) shows that stock liquidity can affect arbitrage cost. As the negative consequences from fire sales can persist for illiquid stocks, I predict that earnings management is concentrated in firms with low stock liquidity. Consistent with my prediction, I show that after fire sales, the increased level of earnings man-                                                  30 See related literature in the next section.    65 agement is only significant in firms with low stock liquidity. Fourth, from the perspective of investor composition, institutional investors are more sophisticated than individual investors and more likely to understand that fire sale stocks’ prices do not completely reflect firms’ fundamen-tals but are linked to mutual funds’ capital withdrawals. Therefore, for firms with higher institu-tional ownership, managers face less pressure to push stock prices back and are less likely to conduct earnings management.31 Consistent with my prediction, I find that earnings management following fire sales is pronounced in firms with low institutional ownership.  Finally, I investigate whether financial constraints affect the relation between underpric-ing and earnings management. Recent research on stock underpricing provides evidence that undervalued firms reduce their investment, and that these effects are concentrated in financially constrained firms (Hau and Lai 2013; Lou and Wang 2014). In light of these findings, I test if financially constrained firms are more sensitive to the consequences of underpricing so that managers are more likely to manage earnings. The evidence is consistent with my prediction.  I also perform additional tests to examine if the response of earnings management to un-derpricing is concentrated in firms with more investment or intangible assets, as previous re-search shows that the level of investment and intangible assets can also proxy for information asymmetry (e.g., Barth, Kasznik, and McNichols 2001; Gu and Wang 2005). I partition my sam-ple based on the level of investment and intangible assets, respectively. The regression results confirm that there is more earnings management subsequent to mutual fund fire sales in firms with more investment and intangible assets. While prior studies show that managers can, through their insider trading, behave as arbi-trageurs to correct mispricing, and the market understands the signals sent out by managers (Ali,                                                   31 In addition to their informational advantage, institutional investors can also monitor management more effectively and discipline managers’ earnings management (Rajgopal, Venkatachalam, and Jiambalvo 1999).    66 Wei, and Zhou 2011), I investigate if the market understands that firms’ income-increasing earn-ings management is a way to guide stock prices back to fundamental values. If so, we would expect earnings management to be associated with positive abnormal returns. However, I do not find evidence that earnings management helps stock price recovery. The market, however, does not punish firms using higher discretionary accruals either. I interpret the result as a natural re-sponse to earnings management, as it is usually regarded as an opportunistic rather than credible signal, and investors are unable to distinguish one from another.  I also examine whether managers use other signals to reinforce earnings management. Previous study suggests that because earnings management is a low cost signal, firm managers can only credibly signal their private information when they combine earnings management with other signals (e.g., stock splits in Louis and Robinson 2005). In this study, I consider stock re-purchase as another possible signal that managers can use. However, my results suggest that its effect on stock price recovery is limited. This paper makes several contributions to the literature. First, this study examines capital market incentives for earnings management from a new perspective–underpricing (Dechow and Skinner 2000). Prior studies demonstrate that due to agency contracting problems, managers inflate earnings when their firms underperform (see reviews by Healy and Wahlen 1999; Dechow and Skinner 2000). Some researchers disagree with this point of view and show that managers in troubled companies make their accounting choices in order to reflect their firms’ financial difficulties (e.g., DeAngelo, DeAngelo, and Skinner 1994). In these cases, it is impos-sible to disentangle stock price changes from firms’ performance, so it is unclear whether stock price decline itself, without deteriorating firm performance, can motivate manager to manage earnings. In this study, mutual fund fire sales provide a unique setting in which exogenous    67 shocks to stock prices is unrelated to firms’ performance. Under this setting, this paper focuses purely on whether stock price declines give managers incentives to manage earnings, and it pro-vides evidence that managers do change their financial reporting in response to underpricing. This study complements recent papers that document real effects that are associated with mispricing connected to fire sales (see review by Bond, Edmans, and Goldstein 2012). Edmans, Goldstein, and Jiang (2012) show that stock price declines lead to a heightened likelihood of firms being acquired, and Hau and Lai (2013) find stock price declines result in a reduced level of investment and employment. This paper investigates whether the decline in stock prices af-fects managers’ financial reporting incentives and consequently, firms’ reporting behavior. I present evidence that firms use discretionary accruals to manage their reported earnings upwards after experiencing mutual fund fire sales.  Finally, this paper improves our understanding of how the market responds to managers’ behavior. The earnings management literature shows that on some occasion managers can credi-bly signal favorable information using discretionary accruals. For example, Louis and Robinson (2005) find that managers can combine stock splits and discretionary accruals to signal good performance. Linck, Netter, and Shu (2013) show that a financially constrained firm with valua-ble projects can also use discretionary accruals to signal good prospects. This paper complements these studies by demonstrating that the signaling function of earnings management is situation-dependent. Although managers seem to try to correct stock underpricing by taking a financial reporting approach, the market does not effectively respond to it. Even when earnings manage-ment is combined with stock repurchase, it still does not help stock price recovery. This chapter proceeds as follows. Section 3.2 is background, related literature, and hy-pothesis development. Section 3.3 describes data, variables, and research design. I report empiri-   68 cal results in section 3.4. Section 3.5 examines the effect of earnings management. Section 3.6 concludes. 3.2 Background, literature, and hypothesis development 3.2.1 Background and related literature In financial markets, a “fire sale” of a publicly traded stock occurs when some investors concen-trate their selling in that stock and there are a limited number of investors who are interested in buying that same stock. For example, hedge funds specializing in merger arbitrage can concen-trate their holdings in a limited number of stocks, creating significant overlap of holdings among hedge funds. When one fund has to sell those stocks to meet investors’ withdrawal of capital, if the reason for the capital withdrawal is common across funds, then other funds are likely to face the same situation and sell their holdings as well. With a limited number of buyers, the prices of the liquidated stocks fall.   Coval and Stafford (2007) empirically investigate the costs of fire sales (and purchases) for the liquidated stocks. They focus on open-end mutual funds that experienced severe capital flows.32 Then they identify the stock transactions associated with these funds to find stocks being heavily sold. These stocks are labeled “fire sale stocks”. Coval and Stafford draw two conclu-sions. First, mutual fund fire sales exert significantly negative price pressure and push stock prices below their fundamental values (-7.9% abnormal return over the two quarters of fire sale). Second, there is a major difference between the price of stocks subject to fire sales and the price of stocks being sold voluntarily by unconstrained funds: the former’s prices rebound over the 12-month period after fire sales, while the latter’s prices do not. Based on these findings, they argue                                                   32 Open-end mutual fund is a type of mutual fund that imposes no restrictions on the amount of shares the fund issues. If investors’ demand is high, the fund can issue more shares. If investors wish to sell, the fund will buy back shares.    69 that the price effect is liquidity-driven, instead of information-driven. In other words, mutual fund fire sales create exogenous and negative shocks to stock prices. Building on Coval and Stafford’s conclusion that fire sale stocks are underpriced, several recent studies use fire sales to examine the effect of stock underpricing on firms and explore managers’ response to this underpricing. First, some research has examined whether firms incor-porate or ignore stock price information in their investment and employment decisions. Using fire sales linked to funds’ overexposure to bank stocks in the 2007-2009 financial crisis, Hau and Lai (2013) show that firms whose stocks are subject to fire sales significantly reduced their in-vestment and employment levels (-20% and -4.7%, respectively) relative to their unaffected peers in the same industry. These effects are concentrated in financially constrained firms, sug-gesting that stock underpricing affects investment and employment through the channel of exter-nal financial constraints. Lou and Wang (2014) conduct a similar study, but expand the time period beyond the financial crisis to the period from 1990 to 2010 and use mutual fund fire sales. They confirm that fire sale pricing results in a reduction in corporate investment. Furthermore, they provide some evidence that the effect is stronger when the stock price is more informative. Edmans et al. (2012) extend the line of inquiry to corporate takeovers. They find that an interquartile range change in the price discount due to fire sales leads to a seven percent increase in takeover probability in the following year.  Gao and Lou (2013) document how fire sales affect firms’ financing decisions. They show that mutual fund fire sales affect both equity and debt prices, but to different degrees. They also find that after stock price declines, firms that depend heavily on external finance experience reductions in both equity and debt issuances. Meanwhile, firms with greater internal cash re-sources reduce equity issuances but increase debt issuances. They conclude that both equity and    70 debt can be jointly mispriced and the impact on firms’ financing decisions depends on whether firms are financially constrained.  Fire-sale-related price declines also seem to affect suppliers’ investment decisions. Williams and Xiao (2014) find that if a customer firm experiences a fire sale and becomes un-dervalued, its suppliers decrease subsequent relationship-specific investments. In particular, supplier firms spend less on R&D and produce fewer patents related to the customer firm’s tech-nology. Finally, Ali et al. (2011) ask how managers respond to stock underpricing. They find that managers take the advantage of stock underpricing to obtain personal benefits. Firm managers engage in more insider trading and grant themselves more options. Ali et al. conclude that corpo-rate insiders enhance their personal benefits by timing their trading and option grants in response to stock underpricing.  3.2.2 Hypothesis development As argued in the previous subsection, prior studies have found that firms experiencing mutual fund fire sales incur significant price declines. Moreover, this stock underpricing has real conse-quences. Following fire sales, firms reduce their investment and employment levels. They expe-rience a higher likelihood of being taken over, and they are less likely to issue equity.  Managers have two additional reasons to correct stock underpricing. First, equity is typi-cally the largest portion of executive compensation packages. While this compensation design motivates managers to achieve better firm performance (e.g., Lambert and Larcker 1987; Morck, Shleifer, and Vishny 1988), high equity incentives can create an excessive focus on stock price. When stock prices fall below their fundamental value due to mutual fund fire sales, managers suffer a financial loss, which motivates them to correct underpricing. Second, stock price de-   71 clines also lead to heightened career concerns. Prior research shows that poor stock performance is associated with a high probability of CEO turnover (Weisbach 1988).  To correct this underpricing, I hypothesize that managers will believe that upward earn-ings management can help. This belief would be supported by the association between stock prices and earnings, which has been long documented in the literature since Ball and Brown (1968). This literature includes studies that find executives manage firms’ reported earnings in an attempt to increase stock prices prior to stock for stock merger (Erickson and Wang 1999) and seasoned equity offering (Teoh, Welch, and Wong 1998). A recent survey by Dichev, Graham, Harvey, and Rajgopal (2013) further confirms that CFOs consider earnings management as a tool to influence stock prices. Therefore, my first hypothesis is: H1: Firms experiencing mutual fund fire sales will increase upward earnings manage-ment relative to control firms that are not subject to fire sales. I also explore the cross-sectional determinants that affect the degree of earnings man-agement. First, I expect the severity of stock underpricing to affect the extent to which execu-tives manage earnings. The intuition is simple. The more stock prices deviate from fundamental values, the more firms suffer from underpricing in terms of investment, employment, and equity financing, which should give managers more incentives to engage in earnings management.  Next, I expect firms with higher information asymmetry to be more likely to manage earnings upwards. The market microstructure literature predicts that when information asym-metry is high, market makers set up high trading costs to protect themselves against better in-formed traders (Glosten and Milgrom 1985; Kyle 1985). Hence, high information asymmetry leads to longer underpricing. In addition, firms with high levels of information asymmetry are held by investors who are less likely to understand the reasons behind the decline in stock prices.    72 These investors will have difficulty understanding if poor stock returns are due to poor perfor-mance or noise (e.g., mispricing due to fire sales). This effect of information asymmetry would delay the convergence of stock price to their fundamentals.  Stock illiquidity also affects the degree to which executives manage earnings. The ex-pected duration of stock underpricing increases with stock illiquidity. Prior research, both theo-retical and empirical, finds that low stock liquidity increases trading costs, impedes arbitrage, and prolongs stock underpricing (Oehmke 2009; Sadka and Scherbina 2008). Therefore, ex ante, if managers think that stock underpricing will last long, they are more likely to manage earnings upward.  The extent to which a firm is held by sophisticated investors should speed up the price correcting mechanism. Institutional investors are generally considered as sophisticated investors (Walther 1997). They have an informational advantage over individual investors and therefore, are more likely to understand that the decline in stock prices is unrelated to firms’ fundamentals. Ex ante, stocks with more institutional ownership should be less mispriced. For example, Collins, Gong, and Hribar (2003) show that more institutional investors help to mitigate the mispricing of accruals. Therefore, when a firm has more institutional investors, mispricing caused by fund fire sales is expected to be short-lived, and the firm’s manager will have less incentive to manage earnings. Finally, prior studies suggest that the negative impact of stock mispricing concentrates in firms that are financially constrained. Hau and Lai (2013) argue that a stock price decline usually signals negative information regarding a firm’s investment opportunity set. Without knowing that the decline is caused by fund fire sales, investors will find it rational to temporarily suspend their supply of capital. This will force firms to reduce their investment if they rely heavily on    73 external financing. On the other hand, if firms are not financially constrained, they can maintain their investment plans without worrying about stock underpricing. Empirical evidence suggests that the reduction in corporate investment after fund fire sales is mainly driven by firms that are more financing-dependent (Hau and Lai 2013; Lou and Wang 2014). As the cost of stock under-pricing increases with financial constraint, the benefit of earnings management to mitigate stock underpricing is likely to be greater for financially constrained firms.  The discussion above is the rationale for my second hypothesis: H2: Upward earnings management will be more pronounced for (a) firms experiencing greater undervaluation due to fire sales, (b) firms that have higher level of information asymmetry, (c) firms that have lower level of stock liquidity, (d) firms with less institu-tional ownership, and (e) firms that are more financially constrained. 3.3 Data, variables, and research design Similar to prior studies (Coval and Stafford 2007; Khan, Kogan, and Serafeim 2012), I consider fire sale stocks as those stocks sold heavily by mutual funds due to liquidity needs rather than due to information regarding firm performance. In other words, I need to identify stocks sold heavily by constrained mutual funds but not by unconstrained funds. Figure 3.1 depicts the idea of my sample selection process. The subsections below describe my sample identification strate-gy in detail.  3.3.1 Measuring mutual fund capital flows I collect relevant fund information from 1996 to 2006. Mutual fund information is from the CRSP Survivorship Bias Free Mutual Fund Database, which reports monthly data of total net assets (market value) under management and net return for each fund. I choose the beginning year to be 1996 because option grant data are limited before the mid-1990s (Ali et al. 2011). The    74 sample period ends in year 2006 because I want to avoid the impact of the 2007-2008 financial crisis (e.g., Ali et al. 2011; Khan et al. 2012). Following prior literature on mutual funds (Wermers 2000; Coval and Stafford 2007), I only focus on actively managed, diversified U.S. domestic equity funds, excluding index funds, international funds, municipal bond funds, “bond and preferred” funds, and sector funds.33 I then calculate mutual fund quarterly capital flows. Following Coval and Stafford (2007), I first find monthly capital flow, which is the change in total net assets (TNA) during the month adjusted for investment return: , , , 1 ,(1 ),j t j t j t j t flowmonthly TNA TNA R    (1)  where TNAj,t is the total net assets of the fund j at the end of month t, and Rj,t is the monthly re-turn of the same fund during the month t. Note that a fund may have several different classes of units (or shares). If that is the case, I combine assets across all share classes to get total net assets, and use the weighted average returns as monthly return, where the weight is the beginning-of-month assets.34 Next, I aggregate monthly flows within a quarter to get quarterly capital flows. As larger funds tend to have greater capital flows, I scale quarterly capital flows by total net assets at the beginning of the quarter to get a percentage flow measure.  To identify mutual funds experiencing extreme capital flows, I follow prior studies (e.g., Coval and Stafford 2007; Ali et al. 2011) to sort percentage flows in each quarter and use the 10th/90th percentile of flow as the threshold levels. A fund is considered to be experiencing se-vere capital outflows (inflows) if the percentage flow is below the 10th percentile (above the 90th                                                   33 Detailed fund selection procedure is similar to that in Kacperczyk, Sialm, and Zheng (2008). 34 This is the same procedure used in Kacperczyk et al. (2008) and Ali et al. (2011).      75 percentile), and is otherwise experiencing normal flows (see Figure 3.1 Step 1 in the middle column). 3.3.2 Identifying stocks experiencing fire sales After finding funds’ flow levels, I merge the dataset with the Thomson Reuters Mutual Fund Holdings Database, which contains stock holdings information of each fund at report dates, so that one can find fund purchases and sales between two consecutive report dates and calculate trading pressures exerted on stocks. Following prior work (Wermers 2000; Coval and Stafford 2007; Kacperczyk, Sialm, and Zheng 2008), I find stock holdings by matching funds primarily through their names. Occasionally, when it is difficult to match with names, I use the investment objective, management company name, and total net assets to conduct the matching.   To identify fire sale stocks, I calculate two types of trading pressure, pressure and upres-sure. The pressure measure is the third measure developed in Coval and Stafford (2007) as . Their first measure is defined as follows. “We sum the difference between flow-induced pur-chases and flow-induced sales in a given quarter and then divide this difference by by the aver-age trading volume of the stock from prior quarters. Flow-induced sales (purchases) are identified as reductions (increases) in shares owned by funds experiencing severe outflows (in-flows). Severe flows are those below/above the 10th/90th percentile of flow” (p. 489). Their third measure replaces the denominator by the number of shares outstanding (see Eq. 2 below):    76    inflow-driven purchases -  outflow-driven sales,max(0, ) | (90 ), , ,SharesOutstanding, 1max(0, ) | (10 ), , ,SharesOutstanding,pressurei tHoldings flow percentile thj i t j tji tHoldings flow percentile thj i t j tji t  ,1 (2)  where Holdingsj i t is the change in fund j’s holding of stock i in quarter t, and flowj,t is capital flow for fund j in quarter t. This measure captures the extent to which a stock’s trading is driven by those mutual funds that are experiencing extreme capital flows. I then sort stocks into deciles based on pressure. Stocks in the top decile are subject to large buying pressure by mutual funds with severe inflows (pressure 9 in Figure 3.1 left column), while stocks in the bottom decile are subject to large selling pressure by mutual funds with severe outflows (pressure 0 in Figure 3.1 left column). To distinguish flow-induced trading from potential information-driven trading, I follow Khan et al. (2012) to calculate unforced pressure, upressure, which is:  , , ,,, 1) | (10 ) (90 ).SharesOutstandingj i t j tji ti tHoldings percentile th flow percentile thupressure   (3)  This variable captures a stock’s net trading pressure caused by mutual funds without severe capi-tal flows (Flow 1 to Flow 8 in Figure 3.1). After being sorted into deciles based on upressure, stocks in the top decile are information-driven purchases (upressure 9 in Figure 3.1 right col-umn), and stocks in the bottom decile are information-driven sells (upressure 0 in Figure 3.1 right column).     77 I follow Khan et al. (2012) to select fire sale stocks and control firms.35 I consider stocks that are in the middle three deciles (40th, 50th, and 60th percentile) of upressure as stocks that are normally traded by unconstrained mutual funds (see Figure 3.1 Step 3 in the right column). These are the candidates of my control stocks. They are not subject to information-driven pur-chases or sales. Within this group of stocks, some stocks also reside in the bottom decile of pres-sure (Step 2). These stocks are deemed as fire sale stocks (FIRESALE = 1). That is to say, fire sale stocks are stocks that are normally traded by unconstrained mutual funds, but they are also unfortunately subject to large selling pressure by some mutual funds experiencing severe capital outflows. The decline in these stocks’ prices is not driven by information, but by the liquidity consideration of some funds (Coval and Stafford 2007). All the other stocks in the middle three deciles of upressure belong to my control group (FIRESALE = 0). Following Ali et al. (2011), I require a stock to be owned by at least five funds. Notice that stocks in the group of upressure 0 and 9 (Step 3) are not used in my research design as they are most likely traded for information reason by unconstrained mutual funds. 3.3.3 Earnings management metrics I use performance-matched abnormal accruals to proxy for earnings management (Kothari et al. 2005). Following Kothari et al. (2005), I calculate total accruals (TA) as the change in non-cash current assets minus the change in current liabilities excluding the current portion of long-term debt, and then minus depreciation and amortization. Total accruals is scaled by lagged total as-sets. I estimate the Jones model discretionary accrual cross-sectionally within the same two-digit                                                   35 Khan et al. (2012) use the following approach: “We sort stock-quarters into deciles of Pressure and UPressure, and identify IBP stocks as those in the top decile of Pressure but in the middle three deciles (deciles four, five, and six) of UPressure. In other words, IBP stocks are those that are subject to large buying pressure by mutual funds with extreme inflows, but that are not subject to widespread net trading pressure by other mutual funds.” (p. 1378)    78 SIC industry (with more than ten observations) each quarter and a firm’s abnormal accruals is the residual from the regression model:  0 1 1 2 3(1/ ) .it it it it itTA ASSETS SALES PPE          (4)  For each firm, a matched firm with the closest ROA is selected from the same two-digit SIC industry and quarter. Performance-matched abnormal accruals is the difference between a firm’s abnormal accruals and its matched firm’s abnormal accruals. 3.3.4 Research designs To empirically test how stock underpricing affects firms’ earnings management level, I use mu-tual fund fire sales as exogenous shocks to stock prices and run a multivariate difference-in-differences (DiD) regression as follows: , 0 1 , 2 , 3 , ,, , ,i t i t i t i t i ti t i tABNACC FIRESALES POST FIRESALES POSTX              (5)  where i indexes firm and t indexes time. ABNACC i,t is the quarterly performance-matched ab-normal accruals that measures a firm’s earnings management level. FIRESALEi,t is an indicator variable that equals one if a firm is subject to mutual fund fire sales and zero otherwise. POSTi,t is another indicator variable that equals one in the post-fire sale four-quarter period and zero in the pre-fire sale four-quarter period.  I am interested in the coefficient β3 as it captures the DiD effect, which is the effect of stock underpricing on a firm’s earnings management level. I also include a vector of control variables, Xi,t. These variables are adopted from prior literature to control for the factors that have been shown to motivate earnings management (e.g., closeness to debt covenant violation, bankruptcy risk). SIZE and book-to-market ratio (BM) con-trol for firm size and growth; return on assets (ROA) controls for operational performance. I use leverage (LEV) and Altman’s (1968) Z-score (ALTZ) to control for financial distress. Also, as    79 prior study suggests that CEO’s bonus and equity incentives are strongly associated with earn-ings management (Cheng and Warfield 2005), I include two measures to control for it: BONUS is the ratio of CEO’s bonus to total compensation, and OPTIONS is the ratio of Black-Scholes value of CEO’s option compensation to total compensation. Both are measured in the year prior to fire sales. Finally, I include a set of corporate governance variables because these features can affect managers’ opportunities to manage earnings. CHAIR is an indicator variable that equals to one if the CEO is also the chair of the board. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership. BIGN equals to one if the firm is audited by a global auditing firm and zero otherwise.  The sample size is limited by data availability. I exclude firm-quarters with missing data from CRSP and COMPUSTAT. I also require firms to have the same number of quarters in both the pre- and post-fire sales periods. The final sample includes 30,402 firm-quarters from 298 fire sale firms and 3,711 control firms. Within the final sample, 92% of the firms have eight quarters in pre- and post-fire sales periods. 3.3.5 Descriptive statistics and preliminary evidence Panel A of Table 3.1 reports the summary statistics of main variables in the sample, including all fire sale firms and control firms in both pre- and post-fire sales periods. The mean (median) abnormal accrual (ABNACC) is -0.004 (-0.002) for the whole sample. While the numbers are smaller than those numbers (-0.02 and 0, respectively) in Kothari et al (2005), it is likely due to the fact that this study estimates abnormal accruals quarterly while they use annual data. The average book-to-market ratio (BM) is about 0.5, and the average return-on-assets (ROA) is 1.3 percent. In terms of CEOs’ compensation, bonus is close to 20 percent of total compensation, and the percentage of options in total compensation is higher, about 31 percent of total compen-   80 sation. In about 80 percent of the observations, CEOs are also chairman of the board. About 67 percent of board members are independent. The mean institutional ownership is close to 64 per-cent. Finally, more than 63 percent of observations are audited by big N auditors. All variables (except CHAIR and BIGN) are winsorized at 1% and 99% to mitigate the influence of outliers. To verify that fire sale firms and control firms are statistically similar to each other before mutual fund fire sales, I compare the characteristics between them in the quarter before fire sales. Panel B of Table 3.1 shows the results of this comparison. The two groups show similar mean values of abnormal accruals, book to market ratio, return-on-assets, Altman’s Z-score, leverage and size in the quarter before fire sales. These measures for two groups are not statistically dif-ferent from each other. For the abnormal returns in the last quarter before fire sales, fire sale firms have a slightly higher value than that of control firms, but it is not statistically significant. The only exceptions are CEO compensations and institutional ownership. For fire sale firms, CEOs’ bonus is low, option value is high, and institutional ownership is high. Overall, these measures suggest fire sale firms are similar to control firms in many dimensions.  3.4 Empirical results 3.4.1 Regression results Table 3.2 presents the results of testing my main hypothesis (H1) with regression model (Eqn. 5). This study is mainly interested in the interaction term (FIRESALE*POST), and predicts the coef-ficient to be positive. The dependent variable is performance-matched abnormal accruals (ABNACC) across all columns. Column (1) presents a parsimonious specification without includ-ing any control variables. While the coefficient on the interaction term (FIRESALE*POST) is positive, it is not statistically significant. In column (2), I include control variables and firm fixed effect. The coefficient on the interaction term stays positive and becomes significant at the 10%    81 level (0.003, t-statistics = 1.30, one-tailed test). Finally, column (3) reports the results by further including time fixed effect. The coefficient on FIRESALE*POST remains positive and signifi-cant (0.003, t-statistics = 1.35, one-tailed test).  The magnitude of the coefficient estimate suggests that fire sale firms, on average, in-crease their abnormal accruals by 0.3 percent, which corresponds to 5% of the standard deviation of ABNACC (0.058). The signs of the coefficients on control variables are generally consistent with those documented in prior studies. However, the coefficient on BIGN has the sign opposite to what I expected based on prior literature. It is possible because this study uses quarterly finan-cial reporting data. As the four-quarter period does not necessarily end at fiscal year-end and quarterly reporting is not subjected to auditing, how big N auditors affect the level of accumulat-ed abnormal accruals is, ex ante, unclear. Overall, the results in Table 3.2 weakly support my main prediction that fire sale firms use abnormal accruals to increase their reported earnings relative to firms not subject to fire sales. Hypothesis H2 predicts that earnings management should be stronger in firms who suffer more from fire sales or, more specifically, whose stock prices are more adversely impacted by fire sales. I focus on the abnormal return in the quarter of fire sales and calculate the abnormal return adjusted by value-weighted market return (ABNRET). Based on the abnormal return, I separate the sample into two subgroups: low and high abnormal returns. Then I run regression model (Eq. 5) on two subgroups. Table 3.3 reports the results. Column (1) is for the subgroup that has been impacted more severely by fire sales. The coefficient on the interaction term is positive and significant at the 1% level (0.008, t-statistics = 2.55, one-tailed test). The magnitude is also much greater than the average effect (0.003) reported in Table 3.2. In column (2), where    82 firms suffer less from fire sales, the coefficient is negative and not significant, suggesting that there is no impact of underpricing on earnings management for this subgroup.  Tables 3.4 - 3.7 report the results of testing H2. My H2 first predicts that earnings man-agement is stronger in fire sale firms with higher information asymmetry. I use analyst coverage to proxy for the level of information asymmetry. Analyst coverage is the average number of unique analysts who issued quarterly earnings forecast for the stock in the four quarters prior to fire sales. I use the median of analyst coverage to separate my sample into two subsamples. Ta-ble 3.4 presents the results of running regression model (Eq. 5) on two subsamples. Consistent with my prediction, firms with low analyst coverage have greater discretionary accruals (column 1). As column (1) shows, the coefficient on FIRESALE*POST is significantly positive at the 10% level (0.005, t-statistics = 1.48, one-tailed test) among firms with low levels of analyst coverage, and the magnitude of the effect is bigger than that reported in Table 3.2 for the full sample. On the other hand, column (2) shows that there is no significant effect among firms with high levels of analyst coverage. H2 also predicts that earnings management is more pronounced in fire sale firms with low stock liquidity. To perform this test, I partition the sample into two subsamples based on stock liquidity measured in the four quarters prior to fire sales, where liquidity is average daily trading volume divided by shares outstanding. Then I run regression model (Eq. 5) on two sub-samples separately and report my results in Table 3.5. For the low liquidity subsample, column (1) shows that the coefficient on the interaction term (FIRESALE*POST) is statistically signifi-cant at the 5% level (0.007, t-statistics = 1.82, one-tailed test), and the magnitude is more than double the coefficient reported for full sample (0.003) in Table 3.2. However, for the high liquid-ity group, the coefficient is not significant, as shown in column (2). In sum, the result in Table    83 3.5 is consistent with my prediction that stock liquidity negatively affects fire sale firms’ earn-ings management.  Table 3.6 shows the results of testing the prediction that the impact of fire sales on earn-ings management is stronger among firms with lower institutional ownership. The sample is partitioned by average institutional ownership in the four quarters before fire sales. Column (1) shows that the effect is statistically strong (0.005, t-statistics = 1.47) for firms with low institu-tional ownership, and the magnitude of the effect is also bigger than that reported in Table 3.2 for the full sample. Column (2), however, indicates that the effect is not significant among firms with high institutional ownership.  In Table 3.7, I reclassify firms by using the degree of financial constraint, which is meas-ured by KZ index (Kaplan and Zingales 1997).36 Column (1) is for firms that are less financially constrained (low KZ score). The coefficient shows the increase in earnings management is not significant for this group. Column (2) is for financially constrained firms. As predicted, the coef-ficient on FIRESALE*POST is positive and statistically significant at the 10% level (0.005, t-statistics = 1.51, one-tailed test). Overall, the results in Table 3.7 suggest that consistent with my prediction, earnings management is stronger in firms whose operations are more financially constrained. 3.4.2 Additional test Besides analyst coverage, I also use two other metrics to measure information asymmetry. Prior literature finds that information asymmetry increases with firms’ investment and intangible as-                                                  36 I calculate the KZ index by following Lamont, Polk, and Saa-Requejo (2001), which uses the following: −1.002(CashFlow/K) + 0.283(Q) + 3.139(Debt/Capital) − 39.368(Div/K) − 1.315(Cash/K). K is property, plant, and equipment. Cash flow is operating income plus depreciation. Cash is cash plus marketable securities. Dividends are annual dividend payments. Tobin’s Q is calculated as (book value of assets minus book value of common equity minus deferred taxes plus market value of equity) / book value of assets. Debt is the sum of short-term and long-term debt. Capital is debt plus total stockholders’ equity.    84 sets (e.g.,Barth et al. 2001; Gu and Wang 2005). Even when the market is efficient, firms with substantial investment or intangible assets are more likely to have their stock prices deviate from their fundamental values. Therefore, firms may suffer more from fire sales when they have high levels of investment or intangible assets. I use the median level of investment and intangible assets measured in the four quarters before fire sales to partition the sample into two subsamples.  Table 3.8 reports the results. Columns (1) and (2) are for the test using investment level to partition sample, while columns (3) and (4) are for the test using intangible assets for partition. As one can see, the coefficients are significantly positive in firms with high investment (0.006, t = 1.91, in column 2) and intangible assets (0.010, t = 2.61, in column 4). For firms with low levels of investment and intangible assets, the impact is not significant (columns 1 and 3). Again, these results are consistent with my prediction that earnings management is stronger for firms with higher information asymmetry.  3.5 Does earnings management correct stock underpricing? Whether earnings management is effective in influencing stock prices remains unclear as prior research finds mixed evidence. Several studies on equity issuance suggest managers can tempo-rarily influence stock prices because investors do not see through earnings management in a timely or complete manner. For example, research shows that firms use income-increasing earn-ings management to inflate earnings prior to seasoned equity offerings (SEOs) and to increase offering prices (Teoh et al. 1998; Cohen and Zarowin 2010).   The prior literature on the accrual anomaly also suggests that investors fail to understand the valuation implication caused by earnings management through accruals management. For example, by decomposing accruals into normal accruals and abnormal accruals, Xie (2001) finds that the abnormal returns associated with Sloan’s (1996) trading strategy based on accruals are    85 mainly attributable to abnormal accruals, which is subject to management manipulation. This also suggests that accrual management can, to a certain degree, influence stock prices. There are studies, however, arguing that investors anticipate and discount earnings man-agement so that management’s attempt to influence stock prices is ineffective. For example, Baber, Chen, and Kang (2006) show that when balance sheet and/or cash flow disclosures are provided at earnings announcements, for firms that appear to manage earnings upwards, their stock price reactions to earnings decrease. Therefore, ex ante, it is unclear whether earnings management is able to influence stock prices effectively. 3.5.1 Long-term effect of earnings management To examine whether earnings management helps to increase stock prices after fund fire sales in the long term, I employ the following pooled OLS regression (similar to that of Ali et al. 2011): 0 1 2 3 45(0,365)_ ,i i i ii iCAAR FIRESALE EM EM SIZE BMLAG CAAR                (6)  where CAAR(0, 365) is cumulative annual abnormal return in the one year (four quarters) after the fire sale quarter, EM is accumulated performance-matched abnormal accruals (ABNACC) in the four quarters after fire sales, SIZE is the natural logarithm of total sales at the end of fire sale quarter, BM is the book-to-market ratio in the fire sale quarter, LAG_CAAR is lagged cumulative annual abnormal return in the four quarters before fire sales. If earnings management helps stock price recovery, the coefficient on FIRESALE*EM should be positive. The results are shown in Table 3.9. Panel A presents the descriptive statistics of variables for both fire sale firms and control firms. Panel B reports the result of testing the effectiveness of earnings management in the long term. In model (1), earnings management is measured by ac-cumulated abnormal accruals to measure, which is continuous. In model (2), I use an indicator    86 variable to measure earnings management. More specifically, if accumulated abnormal accruals are positive, then EM equals one, otherwise zero. In both models, the coefficients on EM are positive, and it is statistically significant in model (1), which suggests that for the whole sample, earnings management does lead to positive abnormal returns. However, I am more interested in the coefficient on the interaction term FIRESALE*EM, as the coefficient suggests whether earn-ings management by fire sale firms can help price recovery. While the coefficient is positive in both models, it is only statistically significant at 10 percent level in model (2). Therefore, I do not find strong evidence that managers’ earnings management behavior after experiencing mutu-al fund fire sales is able to influence stock price correction in the long term.  3.5.2 Short-term effect of earnings management It is possible that long-term abnormal returns are too noisy for me to detect the effect of earnings management on prices. Therefore, I also test the short-term effect. In this test, I use abnormal returns around earnings announcements to capture investors’ short-term reaction to earnings managements. In addition, previous studies suggests that managers can use other signals to rein-force the private information they want to communicate through earnings management (e.g., Louis and Robinson 2005). Earnings management is regarded as a low-cost signal because it can either reflect managers’ opportunistic behavior or signal managers’ insider information (i.e., managers without insider information can mimic those who have such information), and inves-tors may not be able to distinguish one from the other. If this is the case, then managers will need to use other means to reinforce the earnings management signal. One potential method is stock repurchase (Bartov 1991). Here, I partition my sample into two groups: one with stock repur-chase and one without repurchase, and run the following pooled regression model separately on two subsamples:    87 0 1 2( 1,1) ,i i i i i iCAR FIRESALES ABNACC ABCACC X              (7)  where CAR(-1, 1) is the cumulative abnormal return around the earnings announcement day (t = [-1,1]), ABNACC is quarterly performance-matched discretionary accrual, and Xi is a vector of control variables. Detailed variable definitions are in Appendix D. Descriptive statistics are re-ported in Table 3.9 Panel C. The regression results are shown in Table 3.9 Panel D. My prediction is that the coeffi-cient on the interaction term should be positive. For the repurchase subsample, the coefficient on the interaction term is only significantly positive in the first quarter. It is significant at the 10% level (t = 1.59). For non-repurchase subsample, they are not significant at all. Therefore, I only find weak evidence that stock repurchases reinforce the signal that managers try to convey through earnings management.  3.5.3 Discussion Given that I find no evidence that earnings management helps stock price recovery, in this sub-section I discuss the possible reason behind it. The testing results in previous subsections suggest that it is difficult for the market to understand the use earnings management to correct stock underpricing. This can actually be an equilibrium outcome of earnings management. In the con-text of seasonal equity offerings, Shivakumar (2000), using a rational expectations model, argues that because it is difficult for firms to credibly signal the absence of earnings management, inves-tors simply assume all firms manage their earnings and discount the information accordingly. Anticipating this reaction from investors, firms rationally manage earnings upwards prior to equity offerings. He provides empirical evidence that is consistent with his argument. From the owner’s point of view, Liang (2004) shows in his model that by allowing some earnings man-   88 agement, the owner can reduce agency costs because the compensation risk can be allocated more efficiently. Capital market research also provides some evidence that investors are suspicious of earnings management. For example, Keung et al. (2010) focus on firms that report earnings sur-prises in the range [0, 1 cent] as these firms appear to have more incentives to manage earnings upwards to avoid reporting losses. Meanwhile, investors are more likely to be suspicious of earn-ings management behind small positive earnings surprises and react negatively. Their evidence suggests that these firms have low earnings response coefficients (ERCs) upon earnings an-nouncements.  Based on the discussion above, it is possible that investors anticipate executives to man-age earnings upwards after experiencing fire sales. Consequently, they may not react to earnings management, and stock prices do not increase. This can explain why my testing finds no evi-dence that earnings management helps stock price recovery. 3.6 Conclusions Firms incur substantial costs from stock underpricing, including forgoing investment projects, hiring fewer employees, and reducing external financing. As prior literature finds that managers can use earnings management to influence stock prices, I hypothesize that undervalued firms manage earnings upwards in an attempt to correct stock underpricing. I examine the relation by using the setting of mutual fund fire sales, which creates exogenous shocks to stock prices. I find evidence that firms subject to mutual fund fire sales use discretionary accruals to increase their reported earnings. I also show that the effect is more pronounced when firms experience more negative abnormal returns due to fire sales. In addition, the degree of earnings management is greater for firms with low liquidity and high information asymmetry, which are the factors that    89 can ex ante predict the duration of mispricing. Finally, I show that earnings management is con-centrated in financially constrained firms. These results complement the research on the real effects of financial markets by showing that market inefficiency can also distort firms’ reporting quality. However, earnings management does not appear to influence stock prices in this study.      90 Table 3.1: Descriptive statistics  Panel A:  This panel provides summary statistics of the fire sale and control firms, including both pre and post fire sales periods. The sample period is from 1996 to 2006. Stocks are held by actively man-aged, diversified U.S. domestic equity funds, and are held by at least five funds. I require all firms to have non-missing records from CRSP and COMPUSTAT. Finally, I require all firms have the same number of quarters in both pre and post periods. ABNACC is performance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Altman’s (1968) Z-score. LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compensation. CHAIR is an indicator variable that equals to 1 if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independ-ent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to 1 if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D. All variables (ex-cept CHAIR and BIGN) are winsorized at 1% and 99% level.    Variables N Mean Std. Dev. Min P25 Median P75 Max ABNACC 30,402 -0.004 0.058 -0.206 -0.030 -0.002 0.025 0.173 BM 30,402 0.508 0.359 -0.116 0.263 0.449 0.659 2.039 ROA 30,402 0.013 0.024 -0.104 0.005 0.013 0.023 0.077 ALTZ 30,402 3.634 3.717 -1.985 0.995 2.829 5.125 20.428 LEV 30,402 0.244 0.166 0 0.109 0.252 0.365 0.679 SIZE 30,402 6.031 1.465 2.425 5.046 5.928 7.019 9.593 BONUS 30,402 0.194 0.165 0 0.061 0.176 0.286 0.754 OPTION 30,402 0.311 0.271 0 0.000 0.285 0.519 0.915 CHAIR 30,402 0.808 0.394 0 1 1 1 1 IND 30,402 0.668 0.169 0.182 0.556 0.700 0.800 0.923 INSTOWN 30,402 0.635 0.196 0.133 0.497 0.646 0.785 1 BIGN 30,402 0.637 0.481 0 0 1 1 1  91  Table 3.1 (continued)  Panel B: This panel compares summary statistics of fire sale firms and control firms in the last quarter before mutual fund fire sales. Firm characteristics are measured at the fiscal quarter end before each fire sale quarter. ABNACC is performance-matched discretionary (abnormal) accruals esti-mated from Kothari et al.’s (2005) model. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Altman’s (1968) Z-score. LEV is leverage. SIZE is the natu-ral logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compensation. CHAIR is an indicator variable that equals to 1 if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the per-centage of institutional ownership in a firm. BIGN is an indicator variable that equals to 1 if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D. All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. ***, ** and * indicate significance at the 1%, 5%, and 10% levels using two-tailed tests.      92  Table 3.1 (continued)    Fire Sales Firm   Control Firm   Between-Group Variables N Mean Median Std. Dev.   N Mean Median Std. Dev.   t-Test ABNACC 283 -0.008 -0.007 0.056  3,533 -0.005 -0.003 0.057  -0.66 BM 283 0.498 0.437 0.331  3,533 0.518 0.464 0.362  -0.89 ROA 283 0.013 0.013 0.022  3,533 0.012 0.013 0.024  0.55 ALTZ 283 3.860 3.331 3.490  3,533 3.588 2.751 3.706  1.20 LEV 283 0.234 0.237 0.163  3,533 0.246 0.256 0.167  -1.16 SIZE 283 5.929 5.777 1.253  3,533 6.014 5.910 1.477  -0.95 BONUS 283 0.166 0.137 0.144  3,533 0.195 0.176 0.166  -2.83*** OPTION 283 0.358 0.363 0.278  3,533 0.314 0.289 0.271  2.64*** CHAIR 283 0.788 1 0.409  3,533 0.810 1 0.393  -0.88 IND 283 0.674 0.692 0.159  3,533 0.662 0.667 0.171  1.12 INSTOWN 283 0.698 0.722 0.180  3,533 0.624 0.633 0.196  6.13*** BIGN 283 0.551 1 0.498  3,533 0.592 1 0.491  -1.35 ABNRET 283 0.002 0.002 0.181   3,533 -0.006 -0.014 0.175   0.70     93  Table 3.2: The effect of stock underpricing on firm’s financial reporting quality  This table reports the results of the multivariate difference-in-differences (DiD) tests on how stock underpricing affects firms’ financial reporting quality, using a balanced panel sample. I use the following regression: 0 1 2 3 ,it i t i t itABNACC FIRESALES POST FIRESALES POST             in column (1), and then include BM, ROA, ALTZ, LEV, SIZE, BONUS, OPTIONS, CHAIR, IND, INSTOWN, and BIGN in columns (2) and (3). Firm fixed effect is included in columns (2) and (3), and time fixed effect is included in column (3). The dependent variable is ABNACC, which is performance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. FIRESALE equals to one if a firm is subject to mutual fund fire sales, and zero otherwise. POST equals one in the post-fire sales four quarters, and zero in the pre-fire sales four quarters. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Alt-man’s (1968) Z-score as implemented in Begley, Ming, and Watts (1996). LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compen-sation. CHAIR is an indicator variable that equals to one if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to one if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D.  All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.    94  Table 3.2 (continued)  Variables Pred. sign Dependent Variable: ABNACC (1) (2) (3) FIRESALE*POST + 0.003 0.003* 0.003* (0.96) (1.30) (1.35) FIRESALE ? -0.003 -0.002 -0.002 (-1.64) (-1.49) (-1.61) POST ? 0.001 0.001 0.002** (0.74) (1.42) (2.00) BM ?  -0.000 -0.001  (-0.12) (-0.28) ROA -  -0.064 -0.059  (-1.10) (-1.01) ALTZ +  0.002*** 0.002***  (2.61) (2.71) LEV +  0.042*** 0.043***  (3.58) (3.57) SIZE -  -0.015*** -0.017***  (-5.30) (-5.43) BONUS +  -0.001 -0.001  (-0.23) (-0.29) OPTIONS +  -0.003 -0.002  (-0.80) (-0.65) CHAIR +  0.002 0.001  (0.81) (0.67) IND -  0.008 0.006  (1.16) (0.78) INSTOWN -  0.007 0.004  (1.06) (0.62) BIGN -  0.006*** 0.003*  (3.39) (1.55) INTERCEPT  -0.004*** 0.059*** 0.107***  (-8.34) (3.56) (5.45)      Firm FE  NO YES YES Time FE  NO NO YES # of Observation  30,402 30,402 30,402 Adjusted R-square   0.0% 3.6% 3.6%  95  Table 3.3: The effect of underpricing severity  This table reports the results of the multivariate difference-in-differences (DiD) tests on how stock underpricing affects firms’ financial reporting quality. The sample is partitioned into two subsamples (low and high) based on the abnormal return in the quarter of fire sales. I use the following regression: 0 1 2 3 ,it i t i t it itABNACC FIRESALES POST FIRESALES POST X                and include firm fixed and time fixed effects. The dependent variable is ABNACC, which is per-formance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. FIRESALE equals to one if a firm is subject to mutual fund fire sales, and zero otherwise. POST equals one in the post-fire sales four quarters, and zero in the pre-fire sales four quarters. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Alt-man’s (1968) Z-score as implemented in Begley, Ming, and Watts (1996). LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compen-sation. CHAIR is an indicator variable that equals to one if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to one if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D.  All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.    96  Table 3.3 (continued)  Variables Pred. sign Dependent Variable: ABNACC (1) (2) LOW ABNRET HIGH ABNRET FIRESALE*POST + 0.008*** -0.002 (2.52) (-0.40) FIRESALE ? -0.005* 0.000 (-1.76) (0.13) POST ? -0.000 0.004*** (-0.20) (3.62) BM ? -0.001 -0.000 (-0.22) (-0.04) ROA - -0.068 -0.056 (-0.94) (-0.83) ALTZ + 0.003*** 0.002** (3.35) (2.10) LEV + 0.060*** 0.040*** (4.31) (2.55) SIZE - -0.018*** -0.021*** (-4.87) (-5.24) BONUS + -0.001 -0.001 (-0.21) (-0.23) OPTIONS + 0.002 -0.007* (0.47) (-1.58) CHAIR + 0.000 0.002 (0.10) (0.76) IND - 0.006 0.009 (0.64) (0.95) INSTOWN - 0.008 0.003 (0.99) (0.38) BIGN - 0.005** 0.003 (1.65) (1.15) INTERCEPT  0.054** 0.130***  (2.47) (5.38)     Firm FE  YES YES Time FE  YES YES # of Observation  15,198 15,204 Adjusted R-square   2.9% 3.1%  97  Table 3.4: The effect of analyst coverage  This table reports the results of the multivariate difference-in-differences (DiD) tests on how information asymmetry affects the relation between stock underpricing and firms’ financial re-porting quality. I use analyst coverage to proxy for information asymmetry and partition my sample using the median value of the mean analyst coverage in four quarters prior to fire sales. I use the following regression:  0 1 2 3 ,it i t i t it itABNACC FIRESALES POST FIRESALES POST X                and include firm fixed and time fixed effects. The dependent variable is ABNACC, which is per-formance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. FIRESALE equals to one if a firm is subject to mutual fund fire sales, and zero otherwise. POST equals one in the post-fire sales four quarters, and zero in the pre-fire sales four quarters. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Alt-man’s (1968) Z-score as implemented in Begley, Ming, and Watts (1996). LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compen-sation. CHAIR is an indicator variable that equals to one if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to one if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D.  All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.   98  Table 3.4 (continued)  Variables Pred. sign Dependent Variable: ABNACC (1) (2) LOW COVERAGE HIGH COVERAGE FIRESALE*POST + 0.005* 0.002 (1.48) (0.64) FIRESALE ? -0.001 -0.004 (-0.54) (-1.54) POST ? 0.002 0.002 (1.41) (1.41) BM ? -0.001 -0.002 (-0.26) (-0.41) ROA - -0.075 -0.038 (-0.87) (-0.50) ALTZ + 0.004*** 0.001 (3.39) (1.13) LEV + 0.046*** 0.049*** (2.76) (3.33) SIZE - -0.026*** -0.008** (-5.67) (-1.86) BONUS + -0.006 0.003 (-0.87) (0.43) OPTIONS + -0.009** 0.003 (-1.79) (0.64) CHAIR + 0.005** -0.002 (1.65) (-0.62) IND - 0.001 0.009 (0.12) (0.89) INSTOWN - 0.006 0.001 (0.74) (0.07) BIGN - 0.004 0.003 (1.07) (0.95) INTERCEPT  0.079*** 0.056**  (3.20) (2.10)     Firm FE  YES YES Time FE  YES YES # of Observation  15,152 15,250 Adjusted R-square   4.7% 2.7%     99  Table 3.5: The effect of stock liquidity  This table reports the results of the multivariate difference-in-differences (DiD) tests on how stock liquidity affects the relation between stock underpricing and firms’ financial reporting quality. I use the median of the mean daily liquidity in four quarters prior to fire sales to partition my sample. The daily liquidity is the daily trading volume scaled by shares outstanding. I use the following regression: 0 1 2 3 ,it i t i t it itABNACC FIRESALES POST FIRESALES POST X                and include firm fixed and time fixed effects. The dependent variable is ABNACC, which is per-formance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. FIRESALE equals to one if a firm is subject to mutual fund fire sales, and zero otherwise. POST equals one in the post-fire sales four quarters, and zero in the pre-fire sales four quarters. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Alt-man’s (1968) Z-score as implemented in Begley, Ming, and Watts (1996). LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compen-sation. CHAIR is an indicator variable that equals to one if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to one if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D.  All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.    100  Table 3.5 (continued)  Variables Pred. sign Dependent Variable: ABNACC (1) (2) LOW LIQUIDITY HIGH LIQUIDITY FIRESALE*POST + 0.007** 0.001 (1.82) (0.27) FIRESALE ? -0.003 -0.002 (-1.20) (-0.81) POST ? 0.001 0.002* (0.99) (1.92) BM ? 0.001 -0.001 (0.12) (-0.36) ROA - -0.177** -0.003 (-1.72) (-0.04) ALTZ + 0.003** 0.002** (2.23) (1.91) LEV + 0.062*** 0.036** (3.08) (2.18) SIZE - -0.016*** -0.019*** (-3.64) (-4.37) BONUS + 0.002 -0.004 (0.33) (-0.53) OPTIONS + -0.002 -0.002 (-0.52) (-0.47) CHAIR + 0.002 -0.001 (0.68) (-0.23) IND - 0.006 0.007 (0.49) (0.72) INSTOWN - 0.008 0.000 (0.75) (0.05) BIGN - 0.001 0.007** (0.19) (2.06) INTERCEPT  0.129*** 0.116***  (4.41) (4.77)     Firm FE  YES YES Time FE  YES YES # of Observation  15,174 15,228 Adjusted R-square   3.7% 3.3%  101  Table 3.6: The effect of institutional ownership  This table reports the results of the multivariate difference-in-differences (DiD) tests on how institutional ownership affects the relation between stock underpricing and firms’ financial re-porting quality. I partition my sample using the median of average institutional ownership level in the four quarters prior to fire sales. I use the following regression:  0 1 2 3 ,it i t i t it itABNACC FIRESALES POST FIRESALES POST X                and include firm fixed and time fixed effects. The dependent variable is ABNACC, which is per-formance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. FIRESALE equals to one if a firm is subject to mutual fund fire sales, and zero otherwise. POST equals one in the post-fire sales four quarters, and zero in the pre-fire sales four quarters. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Alt-man’s (1968) Z-score as implemented in Begley, Ming, and Watts (1996). LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compen-sation. CHAIR is an indicator variable that equals to one if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to one if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D.  All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.   102  Table 3.6 (continued)   Variables Pred. sign Dependent Variable: ABNACC (1) (2) LOW INSTOWN HIGH INSTOWN FIRESALE*POST + 0.005* 0.002 (1.47) (0.79) FIRESALE ? -0.003 -0.005** (-1.13) (-2.31) POST ? 0.002 0.001 (1.52) (1.01) BM ? -0.007 0.001 (-1.51) (0.34) ROA - 0.047 -0.156** (0.55) (-2.31) ALTZ + 0.001 0.003*** (1.14) (3.11) LEV + 0.052*** 0.045*** (2.89) (2.78) SIZE - -0.021*** -0.014*** (-5.35) (-2.67) BONUS + -0.002 -0.002 (-0.32) (-0.31) OPTIONS + -0.001 -0.005 (-0.18) (-1.08) CHAIR + 0.001 0.001 (0.39) (0.41) IND - 0.016* -0.006 (1.59) (-0.54) INSTOWN - 0.009 -0.008 (1.07) (-0.79) BIGN - 0.005** 0.004 (1.68) (1.09) INTERCEPT  0.070*** 0.093***  (3.44) (3.06)     Firm FE  YES YES Time FE  YES YES # of Observation  15,222 15,176 Adjusted R-square   3.9% 3.3%  103  Table 3.7: The effect of financial constraint   This table reports the results of the multivariate difference-in-differences (DiD) tests on how financial constraint affects the relation between stock underpricing affects firms’ financial re-porting quality. I partition my sample using the median of the KZ score during the fiscal year before fire sales. I calculate the KZ index by following Lamont, Polk, and Saa-Requejo (2001), which is the following: KZ score = −1.002(CashFlow/PP&E) + 0.283(Q) + 3.139(Debt/Capital) − 39.368(Div/PP&E) − 1.315(Cash/PP&E). Cash flow is operating income plus depreciation. Cash is cash plus marketable securities. Dividends are annual dividend payments. Tobin’s Q is calculated as (book value of assets minus book value of common equity minus deferred taxes plus market value of equity) / book value of assets. Debt is the sum of short-term and long-term debt. Capital is debt plus total stockholders’ equity. I use the following regression:  0 1 2 3 ,it i t i t it itABNACC FIRESALES POST FIRESALES POST X                and include firm fixed and time fixed effects. The dependent variable is ABNACC, which is per-formance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. FIRESALE equals to one if a firm is subject to mutual fund fire sales, and zero otherwise. POST equals one in the post-fire sales four quarters, and zero in the pre-fire sales four quarters. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Alt-man’s (1968) Z-score as implemented in Begley, Ming, and Watts (1996). LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compen-sation. CHAIR is an indicator variable that equals to one if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to one if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D.  All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.    104  Table 3.7 (continued)  Variables Pred. sign Dependent Variable: ABNACC (1) (2) LOW KZ SCORE HIGH KZ SCORE FIRESALE*POST + 0.002 0.005* (0.63) (1.51) FIRESALE ? -0.005* -0.002 (-1.95) (-1.11) POST ? 0.002* 0.001 (1.65) (0.83) BM ? -0.003 0.003 (-0.51) (0.79) ROA - -0.034 -0.051 (-0.40) (-0.68) ALTZ + 0.002** 0.005*** (2.28) (4.11) LEV + 0.058*** 0.049*** (3.75) (2.92) SIZE - -0.023*** -0.015*** (-3.74) (-3.95) BONUS + -0.012* 0.008 (-1.40) (1.07) OPTIONS + -0.007* -0.001 (-1.29) (-0.30) CHAIR + 0.002 -0.002 (0.69) (-0.70) IND - 0.006 0.012 (0.52) (1.37) INSTOWN - -0.011 0.013** (-1.03) (1.79) BIGN - 0.002 0.007** (0.44) (2.21) INTERCEPT  0.142*** 0.121***  (4.30) (4.43)     Firm FE  YES YES Time FE  YES YES # of Observation  13,944 13,974 Adjusted R-square   3.2% 5.2%  105  Table 3.8: Additional test: investment and intangible  This table reports the results of the multivariate difference-in-differences (DiD) tests on how the level of investment and intangible assets affect the relation between stock underpricing and firms’ financial reporting quality. I use the median of mean firms’ investment and intangible assets in four quarters prior to fire sales to partition my sample, respectively. I use the following regres-sion:  0 1 2 3 ,it i t i t it itABNACC FIRESALES POST FIRESALES POST X                and include firm fixed and time fixed effects. The dependent variable is ABNACC, which is per-formance-matched discretionary (abnormal) accruals estimated from Kothari et al.’s (2005) model. FIRESALE equals to one if a firm is subject to mutual fund fire sales, and zero otherwise. POST equals one in the post-fire sales four quarters, and zero in the pre-fire sales four quarters. BM is book-to-market ratio at the end of fiscal quarter. ROA is return on assets. ALTZ is the Alt-man’s (1968) Z-score as implemented in Begley, Ming, and Watts (1996). LEV is leverage. SIZE is the natural logarithm of a firm’s sales at the end of fiscal quarter. BONUS is the ratio of CEO bonus to total compensation. OPTION is the ratio of CEO option compensation to total compen-sation. CHAIR is an indicator variable that equals to one if CEO is also the chairman of the board, and zero otherwise. IND is the percentage of independent directors on the board. INSTOWN is the percentage of institutional ownership in a firm. BIGN is an indicator variable that equals to one if the firm’s auditor is one of the big N auditors, and zero otherwise. Detailed definitions of variables are listed in Appendix D.  All variables (except CHAIR and BIGN) are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.   106  Table 3.8 (continued)  Variables Pred. sign Dependent Variable: ABNACC (1) (2) (3) (4) LOW INVESTMENT HIGH INVESTMENT LOW INTANGIBLE HIGH INTANGIBLE FIRESALE*POST + 0.000 0.006** 0.001 0.010*** (0.01) (1.91) (0.18) (2.61) FIRESALE ? -0.001 -0.005** -0.001 -0.005* (-0.52) (-2.33) (-0.42) (-1.67) POST ? 0.004*** -0.000 0.002** 0.001 (3.43) (-0.30) (2.41) (0.92) BM ? 0.002 -0.008* -0.002 -0.001 (0.57) (-1.70) (-0.59) (-0.10) ROA - -0.027 -0.078 -0.051 -0.051 (-0.37) (-0.89) (-0.72) (-0.67) ALTZ + 0.002** 0.002*** 0.002*** 0.003** (2.05) (2.34) (2.48) (2.03) LEV + 0.088*** 0.018 0.062*** 0.030* (4.61) (1.23) (3.92) (1.56) SIZE - -0.023*** -0.013*** -0.018*** -0.024*** (-5.52) (-2.91) (-4.61) (-3.56) BONUS + 0.000 -0.005 0.001 -0.004 (0.02) (-0.65) (0.20) (-0.51) OPTIONS + 0.001 -0.006 -0.004 0.001 (0.32) (-1.12) (-1.02) (0.33) CHAIR + 0.003 0.001 -0.000 0.004 (0.78) (0.25) (-0.06) (0.90) IND - 0.003 0.003 0.004 0.000 (0.27) (0.31) (0.47) (0.02) INSTOWN - -0.002 0.005 0.003 -0.002 (-0.17) (0.57) (0.36) (-0.15) BIGN - 0.001 0.006** 0.003 0.008 (0.17) (2.14) (1.27) (1.23) INTERCEPT  0.149*** 0.089*** 0.099*** 0.128***  (5.00) (3.74) (4.43) (3.32)        Firm FE  YES YES YES YES Time FE  YES YES YES YES # of Observation  15,218 15,184 21,700 8,702 Adjusted R-square   0.041 0.029 0.04 0.04  107  Table 3.9: The effect of earnings management on stock price recovery  Panel A: This table reports the descriptive statistics of variables in testing the long-term effect of earnings management on stock price recovery. CAAR(0, 365) is cumulative abnormal return in the one year (four quarters) after fire sales, where normal return is the value-weighted market return. EM is the measure of earnings management, which is the accu-mulated abnormal accruals (ABNACC) in four quarters after fire sales. LAG_SIZE is the last reported SIZE prior to fire sales. LAG_BM is the last BM ratio prior to fire sales. LAG_CAAR is the one-year cumulative abnormal return prior to fire sale quarter. De-tailed definitions of variables are listed in Appendix D. All variables are winsorized at 1% and 99% level.  Variables N Mean Std. Dev. Min P25 Median P75 Max CAAR(0, 365) 3,667 0.015 0.133 -0.372 -0.057 0.011 0.089 0.415 EM 3,667 -0.014 0.106 -0.624 -0.077 -0.009 0.052 0.438 LAG_SIZE 3,667 6.031 1.462 2.465 5.057 5.914 7.030 9.605 LAG_BM 3,667 0.513 0.352 -0.104 0.267 0.457 0.667 1.884 LAG_CAAR 3,667 -0.009 0.394 -0.825 -0.275 -0.026 0.197 1.453     108  Table 3.9 (continued)  Panel B: This table reports the results of testing the long-term effect of earnings management on stock price recovery. Specifically, it tests how earnings management affects firms’ cumu-lative abnormal returns in four quarters after the fire sales. I use the following regression:  0 1 2(0,365) ,i i i i i iCAAR FIRESALES EM EM X             and include firm fixed and time fixed effects. CAAR(0, 365) is cumulative abnormal return in the one year (four quarters) after fire sales, where normal return is the value-weighted market return. EM is the measure of earnings management, which is the accu-mulated abnormal accruals (ABNACC) in four quarters after fire sales in model (1), which is continuous. In model (2), EM is a binary variable that equals to 1 if accumulated abnormal accruals is positive, and 0 otherwise. LAG_SIZE is the last reported SIZE prior to fire sales. LAG_BM is the last BM ratio prior to fire sales. LAG_CAAR is the one-year cumulative abnormal return prior to fire sale quarter. Definitions of variables are listed in Appendix D. All variables are winsorized at 1% and 99% level. Standard errors are clus-tered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a pre-dicted sign) significance at the 1%, 5%, and 10% levels.    109  Table 3.9 (continued)  Variables Pred.  sign Dependent Variable: CAAR(0, 365) (1) (2) CONTINUOUS BINARY FIRESALES*EM + 0.023 0.021* (0.25) (1.31) EM + 0.067** 0.008 (1.82) (1.18) LAG_SIZE ? -0.020* -0.021* (-1.68) (-1.78) LAG_BM ? 0.016 0.016 (0.70) (0.69) LAG_CAAR  -0.013 -0.013 (-1.21) (-1.18) INTERCEPT  0.130* 0.130* (1.88) (1.88)     Firm FE YES YES Time FE  YES YES # of Observation  3,798 3,782 Adjusted R-square   5.2% 9.1%  110  Table 3.9 (continued)  Panel C: This table reports the descriptive statistics of variables in testing the short-term effect of earnings management on stock price recovery. CAR(-1, 1) is cumulative abnormal return in 3-day ([-1, 1]) around earnings announcement day for four quarters after fire sales, where normal return is the value-weighted market return. ABNACC is quarterly abnormal accruals, the measure of earnings management. LAG_SIZEq is lagged SIZE in last quarter. LAG_BMq is lagged BM ratio in last quarter. LAG_CARq is last quarter’s cumulative ab-normal return. Detailed definitions of variables are listed in Appendix D. All variables are winsorized at 1% and 99% level.  Variables N Mean Std. Dev. Min P25 Median P75 Max CAR(-1,1) 15,151 0.004 0.066 -0.196 -0.028 0.003 0.036 0.209 ABNACC 15,151 -0.004 0.058 -0.206 -0.030 -0.002 0.025 0.173 LAG_SIZEq 15,151 6.060 1.464 2.454 5.076 5.962 7.050 9.605 LAG_BMq 15,151 0.513 0.366 -0.114 0.264 0.451 0.667 2.075 LAG_CARq 15,151 0.012 0.188 -0.434 -0.098 0.002 0.104 0.671       111  Table 3.9 (continued)  Panel D: This panel reports the results of testing the short-term effect of earnings management on stock price recovery by partitioning sample into two subsamples: with stock repurchase and without stock repurchase. It tests how earnings management affects firms’ cumulative abnormal returns around earnings announcement day in each of the four quarters after the fire sales. I use the fol-lowing regression:  0 1 2( 1,1) ,i i i i i iCAR FIRESALES ABNACC ABCACC X              and include firm fixed and time fixed effects. CAR is cumulative abnormal return in 3-day ([-1, 1]) around earnings announcement day for four quarters after fire sales, where normal return is the value-weighted market return. ABNACC is quarterly abnormal accruals, the measure of earn-ings management. LAG_SIZEq is lagged SIZE in last quarter. LAG_BMq is lagged BM ratio in last quarter. LAG_CARq is last quarter’s cumulative abnormal return. Detailed definitions of variables are listed in Appendix D. All variables are winsorized at 1% and 99% level. Standard errors are clustered at the firm level. ***, ** and * indicate two-tailed (one-tailed when there is a predicted sign) significance at the 1%, 5%, and 10% levels.    112  Table 3.9 (continued)  Variables Pred.  sign Dependent Variable: CAR(-1,1) WITH REPURCHASE NO REPURCHASE (1) (2) (3) (4) (5) (6) (7) (8) 1st QTR 2nd QTR 3rd QTR 4th QTR 1st QTR 2nd QTR 3rd QTR 4th QTR FIRESALES*ABNACC + 0.217* 0.170 0.118 -0.059 -0.163 0.001 0.132 0.073 (1.59) (1.24) (0.85) (-0.38) (-1.34) (0.00) (0.67) (0.46) ABNACC + 0.014 -0.042 -0.028 -0.013 0.097** 0.009 0.006 -0.040 (0.35) (-1.07) (-0.76) (-0.42) (2.11) (0.17) (0.11) (-0.67) LAG_SIZEq ? -0.011 0.007 -0.013 0.004 -0.008 -0.009 -0.002 0.004 (-1.01) (0.59) (-1.23) (0.38) (-0.97) (-1.15) (-0.19) (0.48) LAG_BMq ? 0.005 -0.008 0.002 0.010 0.008 -0.005 -0.002 0.007 (0.26) (-0.47) (0.10) (0.55) (0.60) (-0.35) (-0.12) (0.42) LAG_CARq ? -0.021 -0.022 -0.007 -0.014 -0.004 -0.015 -0.004 -0.003 (-1.25) (-1.31) (-0.46) (-0.96) (-0.22) (-0.87) (-0.21) (-0.19) INTERCEPT  0.075 -0.018 0.106 -0.018 0.046 0.017 0.009 0.036 (1.08) (-0.24) (1.62) (-0.19) (0.90) (0.37) (0.19) (0.75)            Firm FE YES YES YES YES YES YES YES YES Time FE  YES YES YES YES YES YES YES YES # of Observation  1,954 1,936 1,970 1,972 1,841 1,845 1,824 1,809 Adjusted R-square   5.0% 6.2% 8.5% 6.0% 4.2% 2.1% 8.9% 7.1%   113  Figure 3.1: Sample selection  This figure depicts my sample selection process. I first calculate mutual fund capital flows and sort them into deciles for each calendar quarter (Step 1 in the middle). Flow 0 and 9 (pink) corre-sponds to funds with the most severe capital outflow and inflow, respectively. Flow 1 to 8 corre-sponds to funds with normal capital flow. For Step 2 (in the left), I focus on stocks traded by funds with severe flows (Flow 0 and 9) and calculate trading pressures for these stocks. After that, I rank them into deciles based on trading pressures (in the left). Stocks in the lowest decile (trading pressure 0) means they experience highest selling pressures (excessive selling). Next, for Step 3 (in the right) I turn back to funds with normal capital flows and look at stocks traded by these funds (Flows 1 to 8). Once again, I calculate trading pressures for these stocks and rank them into deciles (in the right). Following prior literature (Khan et al. 2012), I regard stocks within the middle trading pressures (40th, 50th, and 60th percentile, or trading pressure 4, 5, and 6) as normally traded stocks. Within this group of stocks, if a stock also resides in the bottom trad-ing pressure (10th percentile) caused by mutual funds experiencing severe flows, I treat it as a fire sale stock. Other stocks in this group are control stocks. 114  Figure 3.1 (continued)      115  Figure 3.2: Cumulative average abnormal returns around mutual fund fire sales  This figure shows the cumulative average abnormal returns around mutual fund fire sales of fire sale firms and control firms. Raw return is adjusted by the equal-weighted market return. Similar to Coval and Stafford (2007), I find the reversal of stock price decline caused by fire sales takes longer than 12 months in my sample.      0%1%2%3%4%5%6%7%8%9%10%11%12%-12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18CARMonthControl Firms Fire Sale Firms  116 Chapter 4: Conclusions  High-quality accounting information facilitates efficient capital allocation in financial markets. The rationale is that high accounting quality reduces the level of information asymmetry between firms and investors. Evidence shows that higher accounting quality is associated with lower cost of debt and equity. Less known, however, is whether and how high accounting quality facilitates short-term financing, specifically trade credit financing by suppliers. While prior studies suggest that suppliers demand their customer firms to report conservatively, it is unclear how customer firms can benefit from conservative reporting (high-quality reporting).  In the first essay, I build upon prior research that finds higher accounting quality is asso-ciated with lower costs of debt and equity and show that firms with higher-quality reporting appear to benefit from trade credit financing. Using earnings smoothness, asymmetric timeliness of earnings, and earnings management to proxy for accounting quality, I find that firms with higher accounting quality are able to obtain more trade credit from their suppliers. The results are robust after controlling for suppliers’ characteristics. Moreover, I show that the positive relation between trade credit and accounting quality is more pronounced during the period of credit tight-ening. Finally, I find that the characteristics of firms’ products also impact the relation in such a way that the association is stronger when companies purchase services or differentiated goods. In sum, Chapter 2 presents evidence that higher-quality financial reporting facilitates suppliers’ trade credit financing decisions. An important caveat should be noted in interpreting my findings. This study only docu-ments a general association between accounting quality and trade credit financing instead of a causal relation. To strengthen my tests, I employ the 2007–2008 financial crisis as a shock to credit supply to examine how the degree of association varies. Still, without an experimental   117 setting, it is impossible to completely rule out the possibility that a firm with more trade credit from its suppliers can perform better and, therefore, have higher reporting quality.  Overall, the first essay makes three contributions. First, it provides evidence that account-ing quality matters for short-term financing by showing that firms with higher accounting quality can receive more trade credit from their suppliers. The result complements prior studies’ findings that high accounting quality eases debt and equity financing. In addition, I also show the relation between accounting quality and trade credit financing varies with credit tightening and product characteristics. Second, it complements several recent studies by showing that firms with greater reporting conservatism can obtain more trade credit from their suppliers. Hui et al. (2012) argue that suppliers demand conservative reporting from their buyers but do not explore whether and how a firm can benefit from its suppliers by reporting conservatively. Zhang (2008) focuses on bank loans and shows that firms with greater conditional conservatism can get lower interest loans from banks. Finally, this essay extends recent studies by Watts and Zuo (2012) and Garcia-Appendini and Montoriol-Garriga (2013). Watts and Zuo find that firms with greater accounting conservatism experienced less negative stock returns during the financial crisis because they were able to issue more public debt to undertake investment projects. Garcia-Appendini and Montoriol-Garriga find that firms with sufficient liquidity were able to help their customer firms by extending more trade credit. This essay complements both studies by revealing that conserva-tive firms were also able to obtain more trade credit from their suppliers to alleviate the negative impact of credit tightening.  While many studies, including my first essay, document that firms can benefit from high-quality financial reporting, there are factors that motivate managers to manipulate earnings and hence, reduce the quality of financial reporting. One important factor is capital market motiva-  118 tion, namely stock price consideration. Previous studies show that executives manage earnings opportunistically in an attempt to increase stock prices. Less examined is whether executives manage earnings with good intentions, taking actions to stabilize stock prices or to correct stock underpricing.  In my second essay, I explore this possible motive by examining the effect of stock mis-pricing on firms’ financial reporting quality. Coval and Stafford (2007) show that the selling pressure due to mutual fund fire sales pushes stock prices below their fundamental values, caus-ing stock underpricing. More importantly, they point out that this underpricing is largely exoge-nous to firm performance. Building on their findings, I show that stock underpricing caused by fire sales is associated with firms’ level of earnings management, where earnings management is measured by performance-matched abnormal accruals. Evidence seems to suggest that the asso-ciation is stronger when stock underpricing is more severe. Further evidence suggests that earn-ings management is more pronounced in firms with high information asymmetry, in firms with low stock liquidity, in firms with low institutional ownership, and in firms that are financially constrained. However, I find little evidence that earnings management helps stock price recovery after fire sales. In summary, my second essay also contributes to the literature in several ways. First, it examines capital market incentives for earnings management from a new perspective–stock underpricing. Second, this essay complements several recent papers documenting real effects that are associated with mispricing connected to fire sales. These studies show that stock price declines lead to a heightened likelihood of firms being acquired and a reduced level of invest-ment and employment. My study, on the other hand, reveals the effect of stock underpricing on firms’ financial reporting quality. Finally, this essay improves our understanding of how the   119 market responds to earnings management. My results suggest that managing earnings upwards does not effectively increase stock prices. While prior research suggests that managers can signal their private information through earnings management, this study shows that the signaling func-tion of earnings management is situation-dependent and the market does not necessary respond to it. Given that I find almost no evidence that earnings management helps stock price recov-ery, I plan to further investigate the problem in two ways. First, while prior studies suggest stock underpricing due to mutual fund fire sales is unrelated to firm performance, it is less clear if fund managers pick stocks to liquidate based on their firms’ accounting characteristics, such as accru-als. Capital market studies generally show that institutional investors understand, at least partial-ly, the implication of accruals. Therefore, I will explore if the degree of stock underpricing is related to accounting quality. Second, I will examine whether managers use other methods to influence stock prices. For example, managers can explain stock price declines through voluntary disclosure (e.g., press release or conference call). Alternatively, they can also increase dividend payment to signal their firms’ good performance. Thus, I will further explore managers’ possible reaction to stock underpricing after mutual fund fire sales.          120 References Ali, A., Wei, K. D., & Zhou, Y. (2011). Insider trading and option grant timing in response to fire sales (and purchases) of stocks by mutual funds. Journal of Accounting Re-search, 49(3), 595-632.  Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of corporate bank-ruptcy. Journal of Finance, 23(4), 589-609.  Altunok, F. (2012). Three essays on trade credit. PhD dissertation. North Carolina State Univer-sity.  Armstrong, C. S., Guay, W. R., & Weber, J. P. (2010). The role of information and financial reporting in corporate governance and debt contracting. Journal of Accounting and Eco-nomics, 50(2), 179-234.  Baber, W. R., Chen, S., & Kang, S. H. (2006). Stock price reaction to evidence of earnings man-agement: Implications for supplementary financial disclosure. Review of Accounting Stud-ies, 11(1), 5-19.  Ball, R., & Brown, P. (1968). An empirical evaluation of accounting income numbers. Journal of Accounting Research, 6(2), 159-178.  Barrot, J. N. (2013). Financial strength and trade credit provision: Evidence from trucking firms. Working paper. Massachusetts Institute of Technology.  Barth, M. E., Kasznik, R., & McNichols, M. F. (2001). Analyst coverage and intangible as-sets. Journal of Accounting Research, 39(1), 1-34.  Barton, J., & Simko, P. J. (2002). The balance sheet as an earnings management constraint. The Accounting Review, 77(s-1), 1-27.  Bartov, E. (1991). Open-market stock repurchases as signals for earnings and risk chang-es. Journal of Accounting and Economics, 14(3), 275-294.  Bartov, E., Givoly, D., & Hayn, C. (2002). The rewards to meeting or beating earnings expecta-tions. Journal of Accounting and Economics, 33(2), 173-204.  Bastos, R., & Pindado, J. (2013). Trade credit during a financial crisis: A panel data analy-sis. Journal of Business Research, 66(5), 614-620.  Basu, S. (1997). The conservatism principle and the asymmetric timeliness of earnings. Journal of Accounting and Economics, 24(1), 3-37.    121 Begley, J., Ming, J., & Watts, S. (1996). Bankruptcy classification errors in the 1980s: An empir-ical analysis of Altman's and Ohlson's models. Review of Accounting Studies, 1(4), 267-284.  Bergstresser, D., & Philippon, T. (2006). CEO incentives and earnings management. Journal of Financial Economics, 80(3), 511-529.  Bernanke, B. S., & Gertler, M. (1995). Inside the black box: The credit channel of monetary policy transmission. Journal of Economic Perspectives, 9(4), 27-48.  Bharath, S. T., Sunder, J., & Sunder, S. V. (2008). Accounting quality and debt contracting. The Accounting Review, 83(1), 1-28.  Bhattacharya, U., Daouk, H., & Welker, M. (2003). The world price of earnings opacity. The Accounting Review, 78(3), 641-678.  Biais, B., & Gollier, C. (1997). Trade credit and credit rationing. Review of Financial Stud-ies, 10(4), 903-937.  Biddle, G. C., Ma, M. L., & Song, F. M. (2013). Accounting conservatism and bankruptcy risk. Working paper. University of Hong Kong.  Bond, P., Edmans, A., & Goldstein, I. (2012). The real effects of financial markets. Annual Re-view of Financial Economics, 4(1), 339-360.  Bougheas, S., Mateut, S., & Mizen, P. (2009). Corporate trade credit and inventories: New evi-dence of a trade-off from accounts payable and receivable. Journal of Banking and Fi-nance, 33(2), 300-307.  Bowen, R. M., DuCharme, L., & Shores, D. (1995). Stakeholders' implicit claims and accounting method choice. Journal of Accounting and Economics, 20(3), 255-295.  Burgstahler, D., & Dichev, I. (1997). Earnings management to avoid earnings decreases and losses. Journal of Accounting and Economics, 24(1), 99-126.  Cheng, Q., & Warfield, T. D. (2005). Equity incentives and earnings management. The Account-ing Review, 80(2), 441-476.  Cohen, D. A., & Zarowin, P. (2010). Accrual-based and real earnings management activities around seasoned equity offerings. Journal of Accounting and Economics, 50(1), 2-19.  Collins, D. W., Gong, G., & Hribar, P. (2003). Investor sophistication and the mispricing of accruals. Review of Accounting Studies, 8(2-3), 251-276.  Core, J. E., Guay, W. R., & Verdi, R. (2008). Is accruals quality a priced risk factor? Journal of Accounting and Economics, 46(1), 2-22.   122  Coval, J., & Stafford, E. (2007). Asset fire sales (and purchases) in equity markets. Journal of Financial Economics, 86(2), 479-512.  Coulibaly, B., Sapriza, H., & Zlate, A. (2013). Financial frictions, trade credit, and the 2008–09 global financial crisis. International Review of Economics and Finance, 26, 25-38.  Cuñat, V. (2007). Trade credit: suppliers as debt collectors and insurance providers. Review of Financial Studies, 20(2), 491-527.  DeAngelo, H., DeAngelo, L., & Skinner, D. J. (1994). Accounting choice in troubled compa-nies. Journal of Accounting and Economics, 17(1), 113-143.  Dechow, P. M., & Dichev, I. D. (2002). The quality of accruals and earnings: The role of accrual estimation errors. The accounting Review, 77(s-1), 35-59.  Dechow, P., Ge, W., & Schrand, C. (2010). Understanding earnings quality: A review of the proxies, their determinants and their consequences. Journal of Accounting and Econom-ics, 50(2), 344-401.  Dechow, P. M., & Skinner, D. J. (2000). Earnings management: Reconciling the views of ac-counting academics, practitioners, and regulators. Accounting Horizons, 14(2), 235-250.  Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The Accounting review, 193-225.  DeFond, M. L., & Jiambalvo, J. (1994). Debt covenant violation and manipulation of accru-als. Journal of Accounting and Economics, 17(1), 145-176.  DeFond, M. L., & Park, C. W. (1997). Smoothing income in anticipation of future earn-ings. Journal of Accounting and Economics, 23(2), 115-139.  Degeorge, F., Patel, J., & Zeckhauser, R. (1999). Earnings management to exceed thresh-olds. Journal of Business, 72(1), 1-33.  Diamond, D. W., & Verrecchia, R. E. (1991). Disclosure, liquidity, and the cost of capi-tal. Journal of Finance, 46(4), 1325-1359.  Dichev, I. D., Graham, J. R., Harvey, C. R., & Rajgopal, S. (2013). Earnings quality: Evidence from the field. Journal of Accounting and Economics, 56(2), 1-33.  Easley, D., & O'Hara, M. (2004). Information and the cost of capital. Journal of Finance, 59(4), 1553-1583.  Edmans, A., Goldstein, I., & Jiang, W. (2012). The real effects of financial markets: The impact of prices on takeovers. Journal of Finance, 67(3), 933-971.   123  Erickson, M., & Wang, S. W. (1999). Earnings management by acquiring firms in stock for stock mergers. Journal of Accounting and Economics, 27(2), 149-176.  Faulkender, M., & Petersen, M. A. (2006). Does the source of capital affect capital struc-ture? Review of Financial Studies, 19(1), 45-79.  Fields, T. D., Lys, T. Z., & Vincent, L. (2001). Empirical research on accounting choice. Journal of Accounting and Economics, 31(1), 255-307.  Francis, J., LaFond, R., Olsson, P., & Schipper, K. (2005). The market pricing of accruals quali-ty. Journal of Accounting and Economics, 39(2), 295-327.  Friedman, M. (1953). The case for flexible exchange rates. In “Essays in positive economics”. University of Chicago Press.  Gao, P. (2013). A measurement approach to conservatism and earnings management. Journal of Accounting and Economics, 55(2), 251-268.  Gao, P., & Lou, D. (2013). Cross-market timing in security issuance. Working paper. University of Notre Dame.  Garcia-Appendini, E., & Montoriol-Garriga, J. (2013). Firms as liquidity providers: Evidence from the 2007–2008 financial crisis. Journal of Financial Economics, 109(1), 272-291.  García‐Teruel, P. J., Martínez‐Solano, P., & Sánchez‐Ballesta, J. P. (2014). Supplier Financing and Earnings Quality. Journal of Business Finance & Accounting, 41(9-10), 1193-1211.  Giannetti, M., Burkart, M., & Ellingsen, T. (2011). What you sell is what you lend? Explaining trade credit contracts. Review of Financial Studies, 24(4), 1261-1298.  Glosten, L. R., & Milgrom, P. R. (1985). Bid, ask and transaction prices in a specialist market with heterogeneously informed traders. Journal of Financial Economics, 14(1), 71-100.  Goel, A. M., & Thakor, A. V. (2003). Why do firms smooth earnings? Journal of Business, 76(1), 151-192.  Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of corporate financial reporting. Journal of Accounting and Economics, 40(1), 3-73.  Gu, F., & Wang, W. (2005). Intangible assets, information complexity, and analysts’ earnings forecasts. Journal of Business Finance and Accounting, 32(9‐10), 1673-1702.  Hau, H., & Lai, S. (2013). Real effects of stock underpricing. Journal of Financial Econom-ics, 108(2), 392-408.    124 Healy, P. M. (1985). The effect of bonus schemes on accounting decisions. Journal of Account-ing and Economics, 7(1), 85-107.  Healy, P. M., & Wahlen, J. M. (1999). A review of the earnings management literature and its implications for standard setting. Accounting Horizons, 13(4), 365-383.  Hribar, P., & Collins, D. W. (2002). Errors in estimating accruals: Implications for empirical research. Journal of Accounting Research, 40(1), 105-134.  Hughes, J. S., Liu, J., & Liu, J. (2007). Information asymmetry, diversification, and cost of capi-tal. The Accounting Review, 82(3), 705-729.  Hui, K. W., Klasa, S., & Yeung, P. E. (2012). Corporate suppliers and customers and accounting conservatism. Journal of Accounting and Economics, 53(1), 115-135.  Ivashina, V., & Scharfstein, D. (2010). Bank lending during the financial crisis of 2008. Journal of Financial Economics, 97(3), 319-338.  Jiang, J. (2008). Beating earnings benchmarks and the cost of debt. The Accounting Re-view, 83(2), 377-416.  Jones, J. J. (1991). Earnings management during import relief investigations. Journal of Ac-counting Research, 193-228.  Kacperczyk, M., Sialm, C., & Zheng, L. (2008). Unobserved actions of mutual funds. Review of Financial Studies, 21(6), 2379-2416.  Kaplan, S. N., & Zingales, L. (1997). Do investment-cash flow sensitivities provide useful measures of financing constraints? Quarterly Journal of Economics, 112(1), 169-215.  Kestens, K., Van Cauwenberge, P., & Bauwhede, H. V. (2012). Trade credit and company per-formance during the 2008 financial crisis. Accounting and Finance, 52(4), 1125-1151.  Keung, E., Lin, Z. X., & Shih, M. (2010). Does the stock market see a zero or small positive earnings surprise as a red flag? Journal of Accounting Research, 48(1), 91-121.  Khan, M., Kogan, L., & Serafeim, G. (2012). Mutual fund trading pressure: Firm‐level stock price impact and timing of SEOs. Journal of Finance, 67(4), 1371-1395.  Khan, M., & Watts, R. L. (2009). Estimation and empirical properties of a firm-year measure of accounting conservatism. Journal of Accounting and Economics, 48(2), 132-150.  Kirschenheiter, M., & Melumad, N. D. (2002). Can “Big Bath” and earnings smoothing co‐exist as equilibrium financial reporting strategies? Journal of Accounting Research, 40(3), 761-796.    125 Kothari, S. P., Leone, A. J., & Wasley, C. E. (2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39(1), 163-197.  Kyle, A. S. (1985). Continuous auctions and insider trading. Econometrica, 53(6), 1315-1335.  Lambert, R. A., & Larcker, D. F. (1987). An analysis of the use of accounting and market measures of performance in executive compensation contracts. Journal of Accounting Re-search, 25, 85-125.  Lambert, R., Leuz, C., & Verrecchia, R. E. (2007). Accounting information, disclosure, and the cost of capital. Journal of Accounting Research, 45(2), 385-420.  Lamont, O., Polk, C., & Saa-Requejo, J. (2001). Financial constraints and stock returns. Review of Financial Studies, 14(2), 529-554.  Lev, B., & Sunder, S. (1979). Methodological issues in the use of financial ratios. Journal of Accounting and Economics, 1(3), 187-210.  Liang, P. J. (2004). Equilibrium earnings management, incentive contracts, and accounting standards. Contemporary Accounting Research, 21(3), 685-718.  Linck, J. S., Netter, J., & Shu, T. (2013). Can managers use discretionary accruals to ease finan-cial constraints? Evidence from discretionary accruals prior to investment. The Accounting Review, 88(6), 2117-2143.  Lou, X., & Wang, Y. A. (2014). Flow-induced mispricing and corporate investment. Working paper. University of Delaware.  Louis, H., & Robinson, D. (2005). Do managers credibly use accruals to signal private infor-mation? Evidence from the pricing of discretionary accruals around stock splits. Journal of Accounting and Economics, 39(2), 361-380.  Love, I., Preve, L. A., & Sarria-Allende, V. (2007). Trade credit and bank credit: Evidence from recent financial crises. Journal of Financial Economics, 83(2), 453-469.  McNichols, M. F. (2002). Discussion of the quality of accruals and earnings: The role of accrual estimation errors. The Accounting Review, 77(s-1), 61-69.  Mian, S. L., & Smith, C. W. (1992). Accounts receivable management policy: Theory and evi-dence. Journal of Finance, 47(1), 169-200.  Morck, R., Shleifer, A., & Vishny, R. W. (1988). Management ownership and market valuation: An empirical analysis. Journal of Financial Economics, 20, 293-315.  Murfin, J., & Njoroge, K. (2015). The implicit costs of trade credit borrowing by large firms. Review of Financial Studies, 28(1), 112-145.   126  Ng, C. K., Smith, J. K., & Smith, R. L. (1999). Evidence on the determinants of credit terms used in interfirm trade. Journal of Finance, 54(3), 1109-1129.  Nilsen, J. H. (2002). Trade credit and the bank lending channel. Journal of Money, Credit, and Banking 34(1): 226 - 253.  Oehmke, M. (2009). Gradual arbitrage. Working paper. Columbia University.  Petersen, M. A., & Rajan, R. G. (1994). The benefits of lending relationships: Evidence from small business data. Journal of Finance, 49(1), 3-37.  Petersen, M. A., & Rajan, R. G. (1997). Trade credit: Theories and evidence. Review of Finan-cial Studies, 10(3), 661-691.  Rajan, R. G., & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. Journal of Finance, 50(5), 1421-1460.  Rajgopal, S., Venkatachalam, M., & Jiambalvo, J. J. (1999). Is institutional ownership associated with earnings management and the extent to which stock prices reflect future earn-ings? Working paper. University of Washington.  Raman, K., & Shahrur, H. (2008). Relationship-specific investments and earnings management: Evidence on corporate suppliers and customers. The Accounting Review, 83(4), 1041-1081.  Rauch, J. E. (1999). Networks versus markets in international trade. Journal of International Economics, 48(1), 7-35.  Sadka, R., & Scherbina, A. (2008). Mispricing and costly arbitrage. Working paper. Boston College.  Sankar, M. R., & Subramanyam, K. R. (2001). Reporting discretion and private information communication through earnings. Journal of Accounting Research, 39(2), 365-386.  Shivakumar, L. (2000). Do firms mislead investors by overstating earnings before seasoned equity offerings? Journal of Accounting and Economics, 29(3), 339-371.  Shleifer, A., & Vishny, R. W. (1997). The limits of arbitrage. Journal of Finance, 52(1), 35-55.  Skinner, D. J., & Sloan, R. G. (2002). Earnings surprises, growth expectations, and stock returns or don't let an earnings torpedo sink your portfolio. Review of Accounting Studies, 7(2-3), 289-312.  Sloan, R. (1996). Do stock prices fully reflect information in accruals and cash flows about fu-ture earnings? The Accounting review, 71(3), 289-315.    127 Sweeney, A. P. (1994). Debt-covenant violations and managers' accounting responses. Journal of Accounting and Economics, 17(3), 281-308.  Teoh, S. H., Welch, I., & Wong, T. J. (1998). Earnings management and the underperformance of seasoned equity offerings. Journal of Financial Economics, 50(1), 63-99.  Tucker, J. W., & Zarowin, P. A. (2006). Does income smoothing improve earnings informative-ness? The Accounting Review, 81(1), 251-270.  Trueman, B., & Titman, S. (1988). An explanation for accounting income smoothing. Journal of Accounting Research, 127-139.  Walther, B. R. (1997). Investor sophistication and market earnings expectations. Journal of Ac-counting Research, 157-179.  Watts, R. L., & Zuo, L. (2012). Accounting conservatism and firm value: Evidence from the global financial crisis. Working paper. Massachusetts Institute of Technology.  Weisbach, M. S. (1988). Outside directors and CEO turnover. Journal of Financial Econom-ics, 20, 431-460.  Wermers, R. (2000). Mutual fund performance: An empirical decomposition into stock‐picking talent, style, transactions costs, and expenses. Journal of Finance, 55(4), 1655-1703.  Wilner, B. S. (1997). Paying your bills: An empirical study of trade credit. Working paper. Uni-versity of Iowa.  Williams, R., & Xiao, S. C. (2014). The effect of stock prices on real investment in the vertical supply chain. Working paper. University of Arizona.  Williamson, O. E. (1975). Markets and hierarchies: Antitrust analysis and implications. New York: The Free Pres.  Williamson, O. E. (1979). Transaction-cost economics: The governance of contractual rela-tions. Journal of Law and Economics, 22(2), 233-261.  Xie, H. (2001). The mispricing of abnormal accruals. The Accounting Review, 76(3), 357-373.  Zarley, C., & Damore, K. (1996). Backlogs plague HP: Resellers place phantom orders to get more products. Computer Reseller News, (May 6).  Zhang, J. (2008). The contracting benefits of accounting conservatism to lenders and borrow-ers. Journal of Accounting and Economics, 45(1), 27-54.       128 Appendices Appendix A: An example of suppliers refusing to increase trade credit The following example is from TravelCenters of America LLC (NYSE: TA) 10-K filing for fiscal year 2008 We rely upon trade creditors for a significant amount of our working capital.          Our fuel purchases are our largest operating cost. In 2007, we purchased $4.9 billion of fuel. Historically, we have paid for our fuel purchases after delivery. During the past year as our fuel costs have increased some of our fuel suppliers have been unwilling to adjust the amounts of our available trade credit to accommodate the increased costs of the fuel volumes which we purchase; for example, a $10 million amount of trade credit will allow us to purchase five mil-lion gallons of fuel at $2.00 per gallon, but only 3.33 million gallons at $3.00 per gallon. Also, our financial results and business conditions in the U.S. financial markets generally have caused some fuel suppliers to request letters of credit or other forms of security for our pur-chases. As a result, our investment in our working capital has increased. Any increased invest-ment in working capital decreases our financial flexibility to use our capital for other business purposes and may cause us to continue to experience losses or our losses to increase.       129 Appendix B: Variable definitions for Chapter 2 Variable Definition Dependent variable AP A measure of a firm’s use of trade credit: AP = accounts payable/total assets. Independent variable EARNINGS SMOOTHNESS The variance of earnings divided by the variance of cash flows over the past five years, and then multiply by (-1). EARNINGS CONSERVATISM Average C_Score over the past five years, where C_Score is estimated with Khan and Watts’ (2009) model. EARNINGS MANAGEMENT Accumulated abnormal accruals over the past five years, where abnormal accruals are estimated by Francis et al.’s (2005) model. Control variables SIZE The natural logarithm of the book value of total assets at the end of fiscal year. AGE The natural logarithm of (1 + firm age), where firm age is measured as the number of years since a firm’s first appearance in the CRSP monthly stock return files. AGE2 The square of AGE. PROFIT MARGIN Gross profit divided by total sales. ROA Return on assets, which equals income before extraordinary items divided by total assets. SALE GROWTH This year’s total sales divided by last year’s total sale, and then minus 1. AR Accounts receivable divided by total assets. NET WORTH The total book value of equity divided by total assets. TOBIN’S Q The sum of total market capitalization and total liabilities divided by total assets. CASH Total cash and cash equivalent divided by total assets. DEBT The sum of long-term debt and debt in current liabilities divided by total assets. MARKET SHARE A firm’s total sales divided by its industry total sales, where industry is defined by two-digit SIC code.    130 Appendix C: Product characteristics classification for Chapter 2 The table below is the product characteristics classification from Giannetti, Burkart, and Ellings-en (2011), which is also based on Rauch (1999). According to Rauch’s (1999) classification, standardized goods are those goods with a clear reference price listed in trade publications (e.g., commodity), while differentiated goods are those goods with multidimensional characteristics so that the prices are highly heterogeneous. Services include the remaining industries.            Sector  SIC Services Differentiated Standardized   code   goods goods  Manufacturing           Coal mining   12   0   0   1   Non-metallic minerals   14   0   0   1   Food, kindred products   20   0   0   1   Textile mill products   22   0   0   1   Apparel   23   0   0   1   Lumber, wood products   24   0   0   1   Furniture, fixture   25   0   1   0   Paper, allied products   26   0   0   1   Printing, publishing   27   0   1   0   Chemicals   28   0   0   1   Petroleum, coal products   29   0   0   1   Rubber, plastic products   30   0   1   0   Leather   31   0   0   1   Stone, glass, clay products   32   0   1   0   Primary metal industries   33   0   0   1   Fabricated metal products   34   0   1   0   Machinery   35   0   1   0   Electrical, electronic equipment  36   0   1   0   Transportation, equipment   37   0   1   0   Instruments   38   0   1   0   Miscellaneous products   39   0   1   0        131 Appendix C (continued)  Sector  SIC Services Differentiated Standardized   code   goods goods Transportation, communication, public utilities          Other surface passenger transportation  41   1   0   0   Motor freight transportation, warehousing  42   1   0   0   Water transportation   44   1   0   0   Air transportation   45   1   0   0   Transportation services   47   1   0   0   Communications   48   1   0   0   Electric, gas, sanitary services  49   1   0   0   All wholesale trade           Durable goods   50   1   0   0   Non-durable goods   51   1   0   0   All retail trade           Building materials   52   1   0   0   Department stores   53   1   0   0   Food stores   54   1   0   0   Automotive   55   1   0   0   Apparel, accessory stores   56   1   0   0   Furniture   57   1   0   0   Miscellaneous retail stores   59   1   0   0   Drug and proprietary stores   61   1   0   0   Finance, insurance, real estate           Insurance agents, brokers   64   1   0   0   Real estate   65   1   0   0   Other services           Business services   73   1   0   0   Automobile repair, services, parking  75   1   0   0   Legal services   78   1   0   0   Commercial engineering,  accounting, research  79   1   0   0        132 Appendix D: Variable definitions for Chapter 3 Variable Definition ABNACC Performance-matched abnormal accruals are derived from Kothari et al. (2005). A firm’s abnormal accruals are defined as the residual from the following regression model: 0 1 1 2 3(1/ ) .it it it it itTA ASSETS SALES PPE         A matched firm with the closest ROA is selected from the same industry (two-digit SIC code) and quarter. Performance-matched abnormal accruals is the difference between the firm’s abnormal accruals minus the matched firm’s;  FIRESALE An indicator variable equals to 1 if a firm experiences mutual fund fire sales, and zero otherwise; POST An indicator variable equals to 1 if ABNACC is measured after a firm experiences mutual fund fire sales, and zero otherwise; BM Book-to-market ratio measured at the end of fiscal quarter; ROA Return on assets, equals to income before extraordinary items divided by total assets;  ALTZ The Altman’s (1968) Z-score as implemented in Begley et al. (1996). The model is specified as 1 2 3 4 510.4 1.0 10.6 0.3 0.17 ,ALTZ X X X X X     where X1 is working capital/total assets, X2 is retained earings/total assets, X3 is earnings before interest and taxes/total assets, X4 is market capitalization/total liabilities, and X5 is sales/total assets; LEV Sum of long-term debt and debt in current liabilities, then scaled by total assets. SIZE The natural logarithm of a firm’s sales at the end of fiscal quarter;  BONUS The ratio of CEO bonus to CEO total compensation from EXECUCOMP; OPTIONS The ratio of Black-Scholes value of CEO option compensation to CEO total compensa-tion from EXECUCOMP; CHAIR An indicator variable equals to 1 if the firm’s CEO is also the chairman of the board, and zero otherwise. IND The percentage of independent directors on the board from RiskMetrics; INSTOWN The percentage of institutional ownership in a firm from Thomson Reuters Institutional Holdings database; BIGN An indicator variable equals to 1 if the firm’s auditor is one of the big N auditors, and zero otherwise; ABNRET Abnormal return accumulated in the last quarter before a firm experiences mutual fund fire sales, which is adjusted by value-weighted market return;      133 Variable Definition For the long-term effect test CAAR(0, 365) Annual abnormal return accumulated in four quarters after mutual fund fire sales, which is adjusted by value-weighted market return; LAG_CAAR Annual abnormal return accumulated in four quarters before mutual fund fire sales, which is adjusted by value-weighted market return; EM Performance-matched abnormal accruals accumulated in the four quarters  (ABNACC) after fire sales; LAG_SIZE The last reported SIZE prior to fire sales; LAG_BM The last BM ratio prior to fire sales; For the short-term effect test CAR(-1, 1) Cumulative abnormal return in 3-day ([-1, 1]) around earnings announcement day; LAG_CAR Last quarter’s cumulative abnormal return; LAG_SIZEq Lagged SIZE in last quarter; LAG_BMq Lagged BM ratio in last quarter.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0132741/manifest

Comment

Related Items