Financial Statement Comparability and Valuation of Seasoned Equity Offerings

Transcription

1 Financial Statement Comparability and Valuation of Seasoned Equity Offerings Philip Shane McIntire School of Commerce University of Virginia Charlottesville, VA David B. Smith College of Business Administration University of Nebraska-Lincoln CBA 387 Lincoln, NE (402) Suning Zhang School of Management George Mason University 4400 University Drive, MS 5F4 Fairfax, VA (703) Draft date: May 1, 2013 Abstract This study examines the relationship between financial statement comparability and the SEO valuation. We argue that financial statement comparability allows investors to better assess the quality of the firms that tap into the season equity market through better comparison with the peer firms, thus reducing investors adverse selection risk related to the information asymmetry of the SEO firms and management s ability to sell overvalued equity. As a result, we find that that SEO firms with better comparability experience less underpricing at the time of seasoned equity offering. We also find that SEO firms with better comparability are less likely to have positive earnings surprises and to issue overvalued equity. Furthermore, we find that better financial statement comparability mitigates management s ability to sell overvalued equity especially for SEO firms with positive earnings surprises and higher real earnings management. Findings in this paper provide the empirical evidence to support the decision usefulness of financial statement comparability. 1

2 Financial Statement Comparability and Valuation of Seasoned Equity Offerings 1. Introduction In this paper, we investigate how the cost to raise seasoned equity capital and the seasoned equity post-issue stock return performance are associated with financial statement comparability. Prior literature has documented that there is a significant underpricing (Smith 1977, Bowen et al. 2008, Corwin 2003) and negative post-issue performance (Spiess and Graves, 1995, Loughran and Ritter, 1995, Teoh et al., 1998; Cohen and Zarowin, 2011, and Kothari, and Mizik and Roychowdhury, 2012) associated with season equity offerings. In this paper, we extend the prior literature by investigating how financial statement comparability is related to the valuation of season equity offerings. We argue that financial statement comparability allows investors to better assess the quality of the firms that tap into the season equity market through better comparison with the peer firms, thus reducing investors uncertainty about the SEO firms and management s ability to sell overvalued equity. In other words, financial statement comparability nurtures better capital allocation. Examining the relationship between SEO firm s financial statement comparability and the pricing at the time of offerings and their post-issue performance is especially critical for us to better understand the emphasis of comparability by the Securities Exchange Commission (SEC) and the Financial Accounting Standards Board (FASB). The SEC concept release [2000] 1 emphasizes that comparability should be one of the necessary elements of financial reporting framework. The FASB indicates in its conceptual framework that comparability enriches the usefulness of information for making decisions. The FASB believes that users decisions involve choosing among alternatives such as investing in one reporting entity or another. Concept 1 SEC Concept Release {release NOS , , File No. s } 2

3 Statement # 8 indicates that information about a reporting entity is more useful to these investors if it can be compared with similar information about other entities. Comparability is the qualitative characteristic that enables users to identify and understand similarities and differences among items. Unlike the other qualitative characteristics found in the FASB s conceptual framework, comparability does not relate to a single item. A comparison requires at least two items which means, for cross-sectional comparability, the financial information of at least two companies. 2 Findings in our paper support that enhanced comparability nurtures better capital allocation by reducing the underpricing around SEOs and post-issue negative performance as more underpricing represents higher costs in raising capital and negative post-issue performance represents investors over valuation of the SEOs. We follow De Franco, Kothari, and Verdi (2011) in conceptualizing financial statement comparability as the degree to which two firms have similar mappings from proxies for economic performance into measures of accounting performance. It is based on the concept that, as two firms experience more of the same economic events, they should have more comparable accounting and similar financial statement information. One implication of this view of financial statement comparability that is emphasized in De Franco et al. (2011) is that analysts consider a firm s comparability with other firms when choosing the cohort of firms to follow. For each SEO firm in our sample, we first identify its peers as the group of 5 firms that are most closely followed by the same group of analysts. We then estimate our De Franco et al (2011) comparability score for each SEO firm relative to its peers and use the mean comparability score as our measure of financial statement comparability. Past research suggests that firms with mappings that correlate more strongly with their peers mappings experience more scrutiny by sophisticated investors, which is associated with a 2 This paragraph paraphrases QC 19 QC 25 of the FASB s Concepts Statement # 8. 3

4 higher overall quantity, comparability and quality of information available about the firm (De Franco, et al. 2011: p. 896). We hypothesize that firms with greater financial statement comparability have less underpricing, which translates into a lower seasoned equity offering (SEO) cost of capital and a lower post-issue underperformance measured in economic terms. We also hypothesize that economic effects due to financial statement comparability may be separated from those related to other accounting quality, as indicated by less post-issue underperformance for firms with low quality earnings(e.g., earnings with high real earnings management) but with higher comparability prior to the SEO. Our investigation is divided into four parts. First, consistent with De Franco et al. (2011), we posit an inverse relationship between financial statement comparability and information asymmetry as reflected in SEO underpricing. One outcome we expect and find is that there is less underpricing for high comparability firms than for low comparability firms since investors have less uncertainty about the SEO firms. Second, Lougran and Ritter (1995) suggest that management tend to offer seasoned equity when their equity is overvalued. One way to achieve this goal is to show positive earnings surprise (Kothari, and Mizik and Roychowdhury, 2012). We expect and find that firms with better comparability are indeed less likely to show positive earnings surprises and we also find that firms with better comparability are less likely to be overvalued. In order to document that our financial statement comparability measure captures a distinctive accounting characteristics, we build our hypothesis basing on past research (Cohen and Zarowin, 2011, and Kothari et al., 2012) that suggests that real management is the preferred method of earnings management around SEOs because it is more difficult to detect than accrual earnings management. An outcome we expect and find is a negative relation between financial statement comparability and overvaluation at the time of the SEO, leading to better long run 4

5 equity performance for firms with better comparability. After sorting our firms based on levels of earnings surprise and real earnings management, we find that this relationship remains even in firms with low earnings quality as indicated by high real earnings management and positive earnings surprise. This final outcome provides evidence that financial statement comparability increases the decision usefulness of accounting information in ways that are separate and distinct from increases in other accounting quality measures. Seasoned equity offerings provide a unique setting to investigate financial statement comparability and allow us to contribute to at least fiver streams of accounting research. First, our paper contributes to the literature about how accounting information contributes to capital allocation. Research (Bushman and Smith, 2001; Healy and Palepu, 2001, Lamber, Leuz and Verracchia, 2007; Beyer, Cohen, Lys and Walther, 2010, Bushman, Piotroski, and Smith, 2011) suggest that accounting information facilitates capital allocation by reducing information asymmetry and agency problem. We add this literature by showing that financial statement comparability reduces cost of new equity issuance and long run underperformance. We also show that financial statement comparability is a distinctive accounting characteristic of a firm and investigating financial statement comparability is a separate stream of research from research investigating other accounting quality such as accrual quality and earnings quality. 3 Secondly, our investigation extends the research stream that investigates SEO underpricing. Significant SEO underpricing is documented by a number of studies (Parsons and Raviv, 1985, Loderer, Sheehan and Kadlec, 1991, Loughran and Ritter, 1997, Mola and Loughran, 2004, Corwin, 2003; and Bowen, Chen and Cheng, 2008). The underpricing represents a substantial cost of raising capital. This past research suggests that the underpricing 3 This contribution is particularly important because the FASB in its conceptual framework differentiates between accounting quality and financial statement comparability. 5

6 may be the result of an increase in the adverse selection risk related to asymmetric information around the period of the SEO. We hypothesize that increased comparability is associated with decreased underpricing. Third, our research extends past research investigating the relation between information precision and equity cost of capital. Past studies (Lambert, Leuz, and Verrecchia 2007, 2012, Botosan, Plumlee, and Xie 2004) investigating the relationship between information and cost of capital suggest that cost of capital decreases as the precision of information about company future cash flows increases. We suggest that greater financial statement comparability may be associated with increased precision in estimating firms payoffs (i.e., future cash flows and returns) that are used for valuation. Fourth, a stream of research suggests that management s behavior changes around the time the firm makes a seasoned equity offering. Past studies investigating earnings management prior to SEOs suggest that managers tend to sell overvalued equity at the time of SEOs (Loughran and Ritter, 1995). This line of research documents that SEOs are associated with earnings management and with operating underperformance subsequent to the SEO (e.g., Rangan, 1998; Teoh et al., 1998, Shivakumar, 2000, DuCharme et al., 2004). More recent research (Cohen and Zarowin 2011 and Kothari, and Mizik and Roychowdhury, 2012) provides evidence that positive earnings surprises and subsequent equity under-performance are primarily related to firms engaging in real earnings management. We contribute to this research by investigating whether more comparable firms have fewer positive earnings surprises in the period preceding an SEO. This investigation provides evidence whether managers of more comparable firms behave differently with regard to managing earnings than managers from less comparable firms in the period prior to the SEO. In addition we focus specifically on firms with 6

7 high real earnings management in the period prior to the SEO and we investigate whether firms with more comparable financial reporting systems achieve less long-run underperformance than less comparable firms. This investigation provides evidence whether investors are better able to see though earnings distorted by real earnings management if the financial reporting systems producing those earnings are more comparable. Finally, our paper answers the call from Schipper (2003) for more research to investigate the usefulness of financial comparability. De Franco et al. (2011) provides a comparability metric that measures how the mapping of economic events into the equity returns of one firm are captured by the accounting systems of other firms. It uses this measure to show how increased comparability benefits analyst s ability to improve forecast accuracy. Research (Barth, Landsman, Lang, Williams 2012, Brochet, Jagolinzer, and Riedl 2012) uses this measure to investigate the change in comparability of firms accounting systems that transition to International Financial Reporting Standards and suggests that comparability increases the quality of financial information. We add to the comparability research by providing evidence that financial statement comparability can improve capital allocation at the time of seasoned equity offering. We show that companies can reduce the cost of issuing equity from decreased adverse selection. We also show that investors can benefit from financial statement comparability by better seeing through management s manipulation to sell overvalued equity. The remainder of the paper is divided into five parts. We provide background and develop our hypotheses in section 2, followed by our sample selection and comparative statistics in section 3. Section 4 contains a description of our methodology and our variables. In section 5 we present our results and section 6 is our conclusion. 2. Background and Hypothesis Development 7

8 The relationship between comparability and SEO valuation in terms of underpricing and post-issue underperformance has not previously been investigated in the extensive SEO literature. Prior literature has documented a significant SEO underpricing (Parsons and Raviv, 1985; Loderer, Sheehan and Kadlec, 1991; Loughran and Ritter, 1997; Mola and Loughran, 2004; Corwin, 2003; and Bowen et al., 2008). Underpricing is viewed in this research to be a consequence of an increase in the adverse selection risk related to asymmetric information around the period of the SEO. Bowen et al. (2008) suggest that two ways to decrease this adverse selection risk are through public disclosure and financial intermediaries. Lang (2008) states that lower information asymmetry connected to higher analyst coverage identified in Bowen et al. (2008) may be a result of these financial intermediaries being sensitive to a firm s characteristics that make it less costly to follow. It uses past research (Lang and Lundholm 1996, Francis, Schipper, and Vincent 2002, and De Franco 2007) as the foundation for its argument that analyst coverage is increasing in firm disclosure transparency. One implication of this argument identified by De Franco et al. (2011) is that analysts seek out comparable companies because the strategy allows them to more efficiently use insights they discover across a number of related companies. This argument is consistent with past research (Ramnath 2002, Gleason, Jenkins and Johnson 2008, and Durnev and Mangen 2009) that suggests that comparability among firms enhances information transfers. A company s greater financial statement comparability may be associated with lower analyst information search cost and enhanced intuitions about the underlying economics that determine company value due to a better awareness of the economic event mapping into the accounting information. So arguments made in past research are consistent with the case where comparability enhances the transparency of a company s financial information to analysts because extracting consequential financial relationships from a 8

9 company s accounting data may not be very effective without a comparable yardstick. We suggest that the first step to identify two or more firms that have comparable accounting systems is to focus on analyst following. In our second step we use a comparability metric found in De Franco et al. (2011) to calculate the comparability of firms that are followed the same analysts. We suggest that this measure is associated with a firm s cost of capital and so with the underpricing at the time of the SEO as well as the underperformance in the post SEO period. The connection between the De Franco et al. (2011) comparability measure and cost of capital is supplied by past research (Lambert, Leuz, and Verrecchia 2007, 2012, Botosan, Plumlee, and Xie 2004). This research defines cost of capital as the expected return on a firm s common stock expressed in terms of the CAPM. It shows cost of capital to be increasing in the covariances between the firm s cash flows with other firms cash flows and suggests that as the precision of the information about these covariances increases, the covariances decrease. We suggest that greater financial statement comparability may be associated with increased precision in estimating firms payoffs (i.e., future cash flows and returns) that are used for valuation. This implies that a firm s cost of capital decreases as the comparability of its accounting information increases. Investors are able to depend on more precise signals about the firm s covariances and less on noisy information from the wider market. 4 So, we combine De Franco et al. s (2011) findings with those of Bowen et al. (2008) as interpreted by Lang (2008) and suggest that there is an association between comparability and the seasoned equity offering discount. We state our first hypothesis in the alternative form as follows: 4 Past research (Feltham Xie 1994, and Holmstrom 1979) infers that information precision can be improved by using a second measure that is jointly observable and conveys information beyond the information conveyed by the first measure that is useful for predicting the payoff. 4 This implies having a second, comparable measure as well as measures supplied by the firm However, caution is needed here since Holmstrom [1979] and Scott [2012] examine usefulness in the context of improving contract efficiency. 9

10 H1A: Relative to SEO-firms without close comparables, SEO-firms with close comparables exhibit less underpricing at the time of the offering. A voluminous literature indicates managers have the motivation and means to manage earnings around seasoned equity offerings in order to maintain a higher stock price. One stream of research concentrates on the influence of earnings management as a basis of mispricing (Teoh et al. 1998, Ragan 1998 and Ducharme et al. 2004). Another stream of this research (Kothari, Mizik, and Roychowdhury 2012, Gunny 2010, and Cohen and Zarowin 2010) provides evidence that accrual earnings management is too obvious to investors and suggests that managers may focus on manipulating real activities because this manipulation is more difficult to detect. We add to this research by suggesting that when a company is making a seasoned equity offering, it becomes more restricted in its ability to manage its earnings as its financial statements become more comparable to financial statements of other firms not making seasoned equity offerings. The reason is that investors can more readily detect their manipulations by comparing its financial statements to its peers. Since past research finds managers concentrate on real earnings management, we also concentrate on this form of potential manipulation. This leads to our second hypothesis stated in the alternative form: H2A: Relative to SEO-firms with close comparables, SEO-firms without close comparables are more likely to achieve a positive earnings surprise. Past research (Loughran and Ritter 1995, 1997, and Spiess and Affeck-Graves 1995) documented that SEO firms have below-average returns in the years following an SEO than the non SEO issuers. Loughran and Ritter (1995) suggest that the poor post-issue performance is attributable to management s ability to sell over-valued stock at the time of SEO. We argue that the lack of comparable yards sticks may contribute to management s abilities to promote this 10

11 over valuation and subsequent below-average returns. This leads to our third hypothesis stated in alternative returns: H3A: Relative to SEO-firms with close comparables, SEO-firms without close comparables exhibit lower long-term stock performance after the offering. Past research (Kothari et al and Cohen and Zarowin 2010) suggests that the reason why management has the ability to sell overvalued stock is from using real earnings management to achieve positive earnings surprise at the time of the SEO. By doing so, management sacrifices future economic benefits that are manifested in lower average returns following the SEO. This setting provides us with a way to investigate whether financial statement comparability increases the decision usefulness of accounting information in ways that are separate and distinct from increases in accounting quality. Our thesis is that companies are more likely to successfully over-price their SEOs when they are low comparable firms. We suggest that to the extent that we control for real earnings management then comparability provides benessfits to investors that are not related to the quality of earnings. This leads to our fourth set of hypotheses stated in the alternative form: H4A: After controlling for positive earnings surprise, relative to SEO-firms with close comparables, SEO-firms without close comparables exhibit lower long-term stock performance after the offering. H4B: After controlling for real earnings management and positive earnings surprise, SEO-firms without close comparables exhibit lower long-term stock performance after the offering than SEO-firms with close comparables. 3. Sample 3.1 Sample Selection We gather our sample from the Securities Data Company (SDC Platinum) database. Our initial sample includes all primary seasoned equity offerings firms from 1984 through 2010, 11

12 yielding 9,566 observations. Following the existing literature, we require issuing firms to be traded on NYSE, AMEX, and NASDAQ stock exchanges by eliminating 509 observations. Consistent with past research on earnings management, we exclude 2,705 offerings by financial firms. Finally we only include SEO firms for which we can calculate comparability measures (see details in section 2.1) and with closing price greater than $3. Our initial SEO sample contains 4,327 SEO firm years. To test our hypothesis on the relationship between financial statement comparability and underpricing, we need offer prices, closing prices and values for our control variables. This resulted in eliminating 241 firm years from our sample. The underpricing sample contains 4,086 observations. To test our hypothesis on the relationship between financial statement comparability and post-issue long run performance, we eliminate 751 observations due to missing financial variables needed to calculate earnings surprise and real earnings. We eliminate returns at top and bottom 1 percent levels. Because some firms may issue more than one seasoned equity offerings in a five year period, we eliminate 1122 issues that are offered with in five year window after the first issue. The final sample used in these tests contains 2,203 observations. Insert Table 1 about here 3.2 Key Variable construction Comparability measure: We use a comparability measure developed in De Franco et al. (2011). It is based on the concept that, as two firms experience more of the same economic events, they should have more comparable accounting and similar financial statement information. One implication of this view of comparability that is emphasized in De Franco et al. (2011) is that analysts consider a 12

13 firm s comparability with other firms when choosing the cohort of firms to follow. Analysts do this because greater comparability among firms decreases their workloads. The greater comparability among the firms in their cohorts increases the ease with which they can assess each firm s intrinsic value with reference to that firm's financial statements and the financial statements of other firms. Another related implication is that analysts are likely to follow firms that are closely comparable with each other as a way to increase the quality of their forecasts, which we suggest means that more comparable firms will have less long-run underperformance after the SEO. We use these implications to identify the groupings of firms to estimate our comparability measures. Figure 1 provides a description of the procedure we use to estimate the grouping of firms (e.g., five related firms and the SEO firm) that we then use to estimate our comparability measure. First, we utilize I/B/E/S to find all financial analysts following the SEO firm i at time t, where t represents the 90-day period prior to the date of the SEO. Second, we identify all other firms (i.e., F1, F2,.F11) followed by all of the analysts (i.e., A1, A2, A3, A4) that follow firm i during the same time period. Third, we rank the other firms by the number of firm i analysts following each firm. In Figure 1, for example, we have one firm (F4) followed by three analysts and four firms (F2, F3, F5, and F9) that are followed by two analysts. To estimate our comparability measure for each SEO firm i we use the five most closely related firms, determined by the number of analysts following each related firm that are also following firm i. So, in our Figure 1 example the cohort for firm i that we use to estimate the comparability measure consists of firm i, firm 2, firm 3, firm 4, firm 5, and firm 9. After we have identified the grouping of other firms that are most closely comparable to each of our SEO firms, we estimate our De Franco et al comparability measure for each SEO 13

14 firm. The first step is to estimate earnings using the proxy for the accounting function for the SEO firm and the accounting proxy for each of the comparable firms. For each SEO firm, we estimate the following equation using the 16 quarters as of the most recent quarter before the SEO offering date (for example, AMR Corp had an SEO on 1/24/1991 and the most recent fiscal quarter ends in 12/31/1990 so we collect 16 quarters as of 12/31/1990): Earnings it = α i + βi Return it + ε it. (1) Earnings is the ratio of quarterly net income before extraordinary items to the beginning-ofperiod market value of equity, and Return is the stock return during the quarter. Under the framework in equation (1), ˆ α i and ˆ βi proxy for the accounting function for firm i. For each SEO firm i, we then estimate the following equation using the 16 quarters as of the fiscal quarter before the firm issue seasoned equity for each of a cohort of the 5 closest firms: Earnings jt = α j + β j Return jt + ε jt. (2) Consistent with De Franco et al. (2011), the accounting function for firm j is surrogated by ˆ α j and ˆ β j. To estimate the closeness, we use firm i s and firm j s estimated accounting functions to predict their earnings, assuming they both had the same return (i.e., if they had experienced the same economic events, Return it ). Specifically, we use the two estimated accounting functions for each firm with the economic events of a single firm, firm i. We calculate predicted earnings for firm j and firm i as follows: E(Earnings) iit = ˆα i + ˆβi Return it, (3) E(Earnings) ijt = ˆα j +ˆβj Return it. (4) E(Earnings) iit is the predicted earnings of firm i given firm i s function and firm i s return in period t; and, E(Earnings) ijt is the predicted earnings of firm j given firm j s function and firm i s 14

15 return in period t. We calculate expected earnings based on SEO firm s (i.e., firm i s) returns for each of its 5 closest cohort of firm j1 through j5. In our final step we use our estimated earnings numbers to calculate the comparability measure for each SEO firm i. We define financial statement comparability between SEO firm i and its cohort of 5 closest firms j1 though j5 as the average of the mean absolute difference between the predicted earnings using SEO firm i s and its cohort of 5 firms j1 through j5 s functions. ( ) ( ) lower values indicate poorer financial statement comparability. We also compute a second measure of comparability which we call COMP. First we rank all SEO firms based on Comparability. COMP takes the value of 0 for companies with Comparability values that are equal to and below the median value and a value of 1 for companies with Comparability values that are greater than the median value Earnings Surprise measures: We next show how we calculate our earnings surprise variable. Following Kothari et al. (2012), we assume ROA follows a first order autoregressive with firm and time fixed effects. ( ) (5) is defined as operating income before depreciation divided by total assets. represents the firm fixed effect and ( ) represents time fixed effect. is the first order autoregressive coefficient that represents the persistence of ROA series. Also following Kothari et al. (2012), we construct the time-series panel of by using the intersection of all SEO firms with Compustat during For example, AMR Corp had an SEO in 1991, we keep 15

16 ROAs of AMR Corp from 1984 through We use the time-series panel of to calculate the parameters that we use to obtain the expected for SEO firm i during the fiscal year when SEO was conducted. For example, AMR Corp had an SEO in Jan We estimate the expected ROA for fiscal year 1991). The difference between actual ROA and expected ROA is ROA surprise (ROAS). We also compute a second measure of ROA surprise which we call ES. ES takes the value of 0 for companies with ROAS values that are equal to and less than 0 and a value of 1 for companies with ROAS values that are greater than Cohen and Zarowin Real Earnings Management measure: Our measures used to test our hypotheses related to real earnings management follow Cohen and Zarowin (2010). Our real earnings management measure is depending on three abnormal components: (1) abnormal CFO, (2) abnormal PROD and (3) abnormal DISC, where abnormal CFO is the actual CFO minus the expected CFO. Expected CFO is calculated using equation (8). Abnormal PROD is the actual PROD minus the expected PROD. We calculate expected PROD using the equation (9). Finally, abnormal DISC is the actual DISC minus the expected DISC. We calculate expected DISC using equation (10). All of the expectation models are regressed for each industry and 2-digit industry SIC code. (8) (9) (10) is cash flow from operation (COMPUSTAT data item 308-data item 124). SALES are the sales revenue (data item 12) and ASSETS is total assets (data item 6). PROD is the sum of costs 16

17 of goods sold (data item 41) and changes in inventory during the year (data item3). DISC is the sum of advertising expenses (data item 45), R&D expenses (data item 46) and SG&A (data item 189). Finally, to measure the firm s real earnings management activities through the above three strategies, we define REM=abnormalCFO-1*abnormal PROD+abnormalDISC (11) We also compute a second measure of real earnings management from REM which we call REMS. REMS takes the value of 1 for companies with REM values that are equal to and less than 0 and a value of 0 for companies with REMS values that are greater than Abnormal post-issue stock returns In order to measure SEO firms post issue long run stock returns, we follow Barber and Lyon (1997) approach by matching each SEO firm with a control non SEO firm of similar sizes and book-to-market ratios. As one of our hypotheses is to examine the relationship comparability and long run performance when controlling for earnings surprise and real earnings management level, we follow Kothari et al.(2012) to start cumulating the returns from fiscal year end 5. The control firm is chosen first from a pool of non SEO firms from the same 2-didgit SIC group. This pool of non SEO firms is defined as firms that have not issued seasoned equity in the 5 year period prior to the sample SEO firm fiscal year end. We first choose the control firms with market value of equity between 70% and 130% of that of the sample SEO firm. We then rank the control firms by the closeness of the difference between control firm s book-to-market ratio and that of the sample SEO firm. The best control firm is selected as the one with book-to-market ratio being closest to that of the sample firm. When we are not able to find the matching firm, we 5 Alternatively, we also start cumulate the returns from the SEO date until 6o months after SEO date. Our results are not sensitive to how we cumulate the returns. 17

18 first allow the non SEO firms from the same 2-digit SIC code and with market value of equity between 50% and 150% of that of the sample SEO firm to enter into the pool. If this procedure does not yield any control firms, we search the matching firms from the same 1-digit SIC code and with market value of equity between 50% and 150% of that of the sample SEO firm. Another complica1tion of the above procedure is that some of the matching control firms may issue SEO in the next five years. If we require the control matching firms to have no SEO issues in the next five years, we may introduce the forward-looking bias. To deal with this problem, we follow Spiess and Afflect-Grave (1995) by replacing the best matching control firm with the next closest matching control firm. Through the above matching procedure, we define the abnormal returns as the difference between the cumulative return of the sample firm and that of the matching control firm. 3.3 Descriptive Statistics Table 2 contains the descriptive statistics on our various measures. Panel A reports variables used in the underpricing prices. Our SEO firms on average have a comparability score of -2.4%. We measure the underwriting discount (UD) in a manner consistent with Bowen et al. (2008) to be one times the close-to-offer return, which is calculated from the previous day s closing price to the offer price. Consistent with prior literature, we find that, on average, the offer price is about 3% below the closing price before the offering. The SEO firms are followed by 10 analysts. Our average of number of analysts following is higher than what is reported by Bowen et al. (2008) because the SEO sample is selected from firms who are followed by at least one analyst. Our mean closing price before offering is $27; for issues with positive cumulative returns prior to the offering, the mean is about 4.5%; for issues with negative cumulative returns prior to the offering, the mean is about -2.6%; mean market value is $2,172 million; mean return 18

19 volatility is about 0.029; about 89.7% of issues are offered after the implementation of Rule 10b- 21; about 46 percent of issues are listed on NASDQ exchange; the IPO underpricing is about 16.9%. Most of these statistics are consistent with prior literature except that our mean market value is slightly higher as we require all the issues to have comparability measure by restricting our sample to firms that are followed by at least one analyst and have been traded on public exchange for at least four years prior to the offerings. Panel B of table 2 reports statistics for our post-issue long run performance sample. We find that our mean earnings surprise (ROAS) is about 0.013, consistent with a mean of in the Kothari et al. (2012). About 57.5% of our sample SEO firms experience a positive earnings surprise. Our mean real earnings management score (REM) is and 44.3% of our sample SEO firms have positive real earnings management (REMS), indicating 44.3 of sample SEO firms have higher real earnings management level. Our sample SEO firms experience about 27.9% long run underperformance commencing from the fiscal year end of SEO year. This is quite in line with about 30% long run underperformance in a five year period commencing from the SEO date (Spiess and Grave 1995). Insert Table 2 about here 4. Research Design and Results 4.1 Tests of Hypothesis 1 We use the following regression to test our first hypothesis: UD = β 0 + β 1 *COMPARABILITY) + β 2 *Analyst coverage + β 3 * Price + β 4 *CAR_postive + β 5 *CAR_negative + β 6 *Relative offer size + β 7 *NASDQ + β 8 *Tick + β 9 *Market value + β 10 *Volatility + β 11 *IPO underpricing + β 12 *Rule10b-21 +ε (12) 19

20 We calculate our dependent variable underpricing, UD, in a manner that is consistent with previous literature to be the close to offer return, which is computed from the previous day s closing price to the offer price. Comparability is based on the De Franco et al. (2011) as detailed in section and more negative values of Comparability indicate lower or poorer comparability. Our control variables are selected based on past literature on SEO underpricing. We control for analyst coverage as Bowen et al. (2008) suggest that more analyst coverage reduces information asymmetry and thus reduces the cost of raising equity. Corwin (2003) suggests that there may be a possible issue day price pressure for security offering, therefore, variable Price is used to control for the closing stock price on the day before SEO is offered. Prior research also suggests that pre-offer price move and issue size may affect underpricing (Gerard Nanda 1993). We construct variable CAR to represent the cumulative stock returns adjusted by value-weighted market returns over the 10 days prior to the offer. When CAR is positive, we set CAR_positive equal to CAR and zero otherwise. When CAR is negative, we set CAR_negative equal to CAR and zero otherwise. Scholes (1972) argues that one could view a seasoned equity offering as a permanent shift in the supply of existing shares or a temporary liquidity shock. Both permanent and temporary price pressure could potentially induce underpricing. Thus, we control for the relative offer size by dividing the SEO offer size relative to outstanding shares. We add indicator variable NASDQ because past research shows that SEO firms traded at NASDQ may experience more underpricing. Corwin (2003) adds variable Tick to reflect that underpricing may result from a tendency to round prices down to the nearest eighth or integer value (Lee, Lochhead, Ritter, and Zhao (1996). As in Corwin (2003), we control for market value and the volatility Finally we control for macro market condition by controlling for IPO market underpricing during the same 20

21 month as SEO as in Bowen (2008) and the implementation of Rule 10b-21 as a way to reduce the manipulative short selling prior to SEOs. Variable Rule 10b-21 is set to 1 if SEOs are offered after the Rule 10b-21 becomes effective on August 25, Our results of testing Hypothesis 1 tests are found in table 3. In column 2 of the table we investigate the relationship between underwriting discount (UD) with no control variables. The results show a negative relation between UD and Comparability, indicating that more comparable firms with SEOs have smaller underwriting discounts. This result is consistent with less asymmetric information at the time of the SEO offering and so less change in the cost of capital for comparable firms at the time of the SEO. Our results in column 3 of table 3 are generally consistent with past research. Comparability remains significantly related to UD. Analyst Coverage is also significantly negatively related to UD. This is consistent with past research though we cannot separate whether more analysts are causing the decrease in information asymmetry related to the lower underwriting discount or whether they are attracted to more comparable firms that have less information asymmetry. The other significant variables are in directions consistent with past research. Overall our results suggest that more comparable firms have smaller underwriting discounts at the time of the SEO. This indicates a smaller cost of capital effect related more comparable firms and so is consistent with rejection of our first hypothesis. Insert Table 3 about here 4.2 Tests of Hypothesis 2 Table 4 contains the results of our hypothesis 2 tests. Our investigation is a four step process as indicated by the four regression models presented in the table. In column one, we find 21

22 that higher real earnings management level (REMS) is related to positive relationship. In column two we show the results of our investigation of the relationship between positive earnings surprises and comparability. We use the variable COMP in these tests to be equal to 1 if an observation is more negative than the median comparability amount and 0 if it is equal to or less negative to the median comparability amount. This means that more comparable observations are associated with 0 and less comparable observations are associated with 1. We hypothesize that firms are more likely to successfully manage investor perceptions when comparability is low, and so we expect a1 to be greater than 0 as estimated in the following regression equation: ES = a0 + a1(comp) + u, where ES=1 if positive (13) The results of our tests show a significantly positive relation between COMP and ES. Third we investigate whether the effect related to comparability remains after including real earnings management. Column three contains our results from including both COMP and REMS in the regression. We hypothesize that firms with low comparability and firms using real earnings management are more likely to use real earning management so we expect a positive relation to both COMP and REMS. Our results in table 4 are consistent with our expectations. Fourth, in column 4 of table 4 we investigate whether the interaction term between COMP and REMS is significantly positive. As indicated by our model (3) below, we hypothesize that firms with positive earnings surprises achieve those surprises, consistent with past research, with real earnings management. We also hypothesize that firms are more likely to have positive earnings surprises when they use real earnings management and are low comparability firms. Consistent with hypothesis 2, we expect a3 to be greater than 0, indicating a relatively larger effect of REM on ES when comparability is low. ES = a0 + a1(rems) + a2(comp) + a3(comp)(rems) + u (14) 22

23 Our results in table 4 are consistent with rejection of our hypothesis 2, though the interaction term is only significant at the 10 percent level. Insert Table 4 about here 4.3 Tests of Hypothesis 3 In table 5 we present the results of testing hypothesis 3 where we investigate the relation between comparability and long stock return underperformance. Past research (Loughran and Ritter 1995, 1997, and Spiess and Affeck-Graves 1995) shows an SEO pricing anomaly where SEO firms have below-average returns in the years following an SEO. We hypothesize that investors are less likely to be fooled by more comparable firms and therefore we are less likely to observe post-issue underperformance for these SEO firms. Firms with high and low comparability are sorted into group 1 and 2 respectively. Our results at the bottom of the table indicate that there is a significant difference between the long run underperformance of high comparable firms and low comparability firms at the time of the SEO. These results are consistent with rejection of our hypothesis 3. These results provide us the first glance at the decision usefulness to mitigate management s ability to sell overvalued equity. Insert Table 5 about here 4.4 Tests of Hypothesis 4 Past research by Kothari et al. (2012) indicates that SEO issuing companies with long run under performance may have fooled investors by showing positive earnings surprises during the SEO year. In table 6, we investigate whether this long-run underperformance is mitigated by comparability. We hypothesize that investors are less likely to be fooled by more comparable 23

24 firms even when they are generating a positive earnings surprise and therefore we are less likely to observe post-issue underperformance. Firms with positive earnings surprises are sorted into group 1 and 2, whereas firms with non-positive earnings surprises are sorted into group 3 and 4. Within firms with positive or nonpositive earnings surprises, we further sort firms into more or less comparable to their closest cohorts, respectively. The most important comparison in table 6 is between group 1 and group 2 as we try to examine the decision usefulness of the financial statement comparability when SEO firms have positive earnings surprises. Our results at the bottom of the table indicate that there is a significant difference between the long run underperformance of high comparable firms and low comparability firms at the time of the SEO. These results are consistent with rejection of our hypothesis 4a. Insert Table 6 about here In testing our hypothesis 4b, we sort SEO-firms on the interaction of ES, REMS and COMP into 8 groups. Firms in groups 1 through 4 have positive earnings (ES=1). Group 1 and 2 include the firms with lower real earnings management activities, while groups 2 and 4 include the firms with higher real earnings management activities. Furthermore, firms in groups 1 and 3 are composed firms that are less comparable to their closest cohorts as compared to firms in groups 2 and 4, respectively. We are particularly interested in group 3 and 4 as firms in these two groups achieve positive earnings surprises and also engage in higher level of real earnings management. We expect that within these two groups, relative to SEO-firms with close comparables, SEO-firms without close comparables have more overvaluation problem, as exhibited by poorer long-run performance. In other words, we expect that firms in group 3 will underperform firms in group 4. 24

25 Our table 7 results indicate a significant difference in the long run underperformance for more comparable firms using real earnings management and with positive earnings surprises (group 4) and less comparable firms using real earnings management and with positive earnings surprises (group 3). These results are consistent with rejection of our hypothesis 4b. Insert Table 7 about here 5. Other considerations Past research has suggested some other explanations for the negative post-issue performance. One of the explanations is offered by Teoh et al. (1998), in which they find that SEO firms with higher discretionary accruals prior to the offerings have lower post-issue stock performance. Therefore, we control for this accounting characteristic in our test of how financial statement comparability can potentially affect the SEO post-issue long run performance. Table 8 presents the effect of financial statement comparability on SEO post-issue long performance by partitioning our SEO sample through interaction of DCAS and COMP into 4 groups. The DCAS equals to 1 for firms in groups 1 and 2 with positive discretional accruals, indicating these SEO higher discretional accruals level. Group 3 and 4 include the firms with non-positive discretional accruals. Furthermore, firms in groups 1 and 3 are composed of firms that are less comparable to their closest cohorts as compared to firms in groups 2 and 4, respectively. We are particularly interested in group 1 and 2 as firms in these two groups have higher discretional accruals level. We expect that within these two groups, relative to SEO-firms with close comparables, SEOfirms without close comparables have more overvaluation problem, as exhibited by poorer longrun performance. In other words, we expect that firms in group 2 will underperform firms in group 1. Our results tabulated in Table 8 support our expectation. Our results tabulated in Table 25

26 8 also indicate that financial statement comparability is a distinctive accounting attribute that is different from accrual quality. Insert Table 8 about here Another explanation for the management s ability to take advantage of windows of opportunity to issue overvalued equity (Choe, Masulis, and Nanda 1993, Lougran and Ritter 1995a, Bayless and Chaplinsky 1996). We define the periods with above (and below) the median monthly average initial IPO returns as hot SEO market (and cold SEO market otherwise). Table 9 presents results by interacting high/low comparability with hot/cold SEO market. We find that higher comparability reduce long run post-issue underperformance in hot and in cold market. More interestingly, we find that comparability mitigate the underperformance problem more profoundly in hot market than in cold market, indicating comparability especially useful when companies take window of opportunity to sell overvalued equity. Insert Table 9 about here 6. Conclusion De Franco et al. (2011) provides a comparability metric that measures how the mapping of economic events into the equity returns of one firm are captured by the accounting systems of other firms. It uses this measure to show how increased comparability is related to improved analyst forecast accuracy as a result of decreased cost of acquiring information about a firm. Research (Barth, Landsman, Lang, Williams 2012, Brochet, Jagolinzer, and Riedl 2012) uses this measure to investigate the change in comparability of firms accounting systems that transition to International Financial Reporting Standards and suggests that comparability increases the quality of financial information. 26

27 Seasoned equity offerings provide an important setting for investigating comparability and allow us to contribute to past research in a number of ways. Our paper contributes to the literature about how accounting information contributes to capital allocation. We add this line of literature by showing that financial statement comparability reduces cost of new equity issuance and long run underperformance. In addition, our research extends past research investigating the relation between information precision and equity cost of capital. We suggest that greater financial statement comparability may be associated with increased precision in estimating firms payoffs (i.e., future cash flows and returns) that are used for valuation. Second, our investigation extends the research stream that investigates SEO underpricing. This past research suggests that the underpricing may be the result of an increase in the adverse selection risk related to asymmetric information around the period of the SEO. We hypothesize and show that increased comparability is associated with decreased underpricing. Third, our investigation extends a stream of research that suggests that management s behavior changes around the time the firm makes a seasoned equity offering. Recent research (Cohen and Zarowin, 2011 and Kothari, and Mizik and Roychowdhury, 2012) provides evidence that positive earnings surprises and subsequent equity underperformance are primarily related to firms engaging in real earnings management. We contribute to this research by investigating whether more comparable firms have fewer positive earnings surprises in the period preceding an SEO. This investigation provides evidence whether managers of more comparable firms behave differently with regard to managing earnings than managers from less comparable firms in the period prior to the SEO. Our results indicate that less comparable firms are more likely to have positive earnings surprises and more importantly, that less comparable firms that use real earnings management are more likely to have positive earnings surprises. 27