Analyst cash flow forecasts and pricing of accruals

ADIAC-00238; No of Pages 11 Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Advances in Accounting, incorporating Advances in International Accounting journal homepage: www.elsevier.com/locate/adiac

Analyst cash flow forecasts and pricing of accruals☆,☆☆ Linna Shi a,⁎, Huai Zhang b,1, Jun Guo c,2 a b c

School of Management, State University of New York at Binghamton, 4400 Vestal Pkwy E, Binghamton, NY 13902, United States Nanyang Business School, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore School of Business, Rutgers University-Camden, 227 Penn Street, Camden, NJ 08102, United States

a r t i c l e

i n f o

Available online xxxx Keywords: Accruals Accrual anomaly Cash flow forecasts Pricing Earnings manipulation

a b s t r a c t This paper investigates how analyst cash flow forecasts affect investors' valuation of accounting accruals. We find that the strength of the accrual anomaly documented in Sloan (1996) is weaker for firms with analyst cash flow forecasts, after controlling for idiosyncratic risk, transaction costs and firm characteristics associated with the issuance of cash flow forecasts. We further show that this reduction in mispricing of accounting accruals is at least partially attributed to the improved ability of investors to price earnings manipulations imbedded in accruals. We investigate several non-mutually exclusive alternative explanations for this improvement in investors' ability and demonstrate that the increased investor attention and the improved accuracy of analyst earnings forecasts both contribute to the mitigation of the accrual anomaly. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction DeFond and Hung (2003, 2007) show that analysts provide cash flow forecasts in response to investor demand in both the U.S. and international settings. After the shocking accounting scandals of the early 2000s such as Enron and WorldCom, investor demand for cash flow information increased dramatically, leading to the growing popularity of analyst cash flow forecasts in recent years. The literature on analyst cash flow forecasts can be divided into two streams. The first stream of literature demonstrates the role that analyst cash flow forecasts play in monitoring managerial behavior. Both Call (2008) and McInnis and Collins (2011) show that analyst cash flow forecasts help to reduce earnings manipulations and improve earnings quality. The second stream of literature investigates the usefulness of the information in analyst cash flow forecasts. Research by Call, Chen, and Tong (2009, 2013) suggests that analyst cash flow forecasts contain information that is helpful for analysts and investors. In response to claims that

☆ Data availability: Data used in this study are available from the sources identified in the study. ☆☆ We thank Philip Reckers (editor), the anonymous referee, Bradley Lail, Nan Zhou, the seminar participants at Nanyang Business School, and the conference participants of the 2013 American Accounting Association Annual Meeting for their helpful comments. Any errors are our own responsibility. ⁎ Corresponding author. Tel.: +1 607 777 4640; fax: +1 607 777 4422. E-mail addresses: [email protected] (L. Shi), [email protected] (H. Zhang), [email protected] (J. Guo). 1 Tel.: +65 6790 4097; fax: +65 6793 7956. 2 Tel.: +1 856 225 6800; fax: +1 856 225 6231.

cash flow forecasts are of low quality (Givoly, Hayn, & Lehavy, 2009), Call et al. (2013) refute the validity of such claims. However, no direct evidence exists on how analyst cash flow forecasts affect the ability of investors to accurately price accounting accruals. We seek to fill this void. Previous work suggests that investors do not price accruals appropriately (Sloan, 1996). As a result, accruals systematically predict future abnormal returns. Sloan (1996) further documents that an accrual strategy that buys firms with low accruals and shorts firms with high accruals yields positive and significant returns. This is perhaps one of the most robust empirical findings in accounting research and has been confirmed by numerous papers (for example, Collins & Hribar, 2000; Hirshleifer, Hou, Teoh, & Zhang, 2004; Hribar & Collins, 2002; Lev & Nissim, 2006; Mashruwala, Rajgopal, & Shevlin, 2006; Pincus, Rajgopal, & Venkatachalam, 2007; Richardson, Tuna, & Wysocki, 2010; Shi & Zhang, 2012; Xie, 2001; Zhang, 2007). In a later study, Xie (2001) shows that the accrual strategy based on discretionary accruals yields significantly positive abnormal returns while the strategy based on non-discretionary accruals yields non-significant abnormal returns. To the extent that discretionary accruals are representative of earnings management, the findings suggest that investors do not “see through” accounting manipulations. In this paper, we predict that analyst cash flow forecasts reduce investors' mispricing of accruals. Our empirical results are consistent with our prediction: we find that the accrual strategy yields a hedge return of 12%, significant at the 0.01 level, when applied to firms without analyst cash flow forecasts. In contrast, the strategy does not yield any significant hedge return when applied to firms with such forecasts.

http://dx.doi.org/10.1016/j.adiac.2014.04.006 0882-6110/© 2014 Elsevier Ltd. All rights reserved.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

2

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx

This univariate analysis, however, does not consider the fact that firm characteristics may differ across firms with and without analyst cash flow forecasts. It is possible that differences in firm characteristics, rather than the existence of analyst cash flow forecasts, affect the accrual anomaly. Following DeFond and Hung (2003), we identify a set of firm characteristics associated with the issuance of analyst cash flow forecasts and control for these firm characteristics in our regression. We also control for idiosyncratic volatility and transaction costs, given prior observations (Mashruwala et al., 2006) that accrual mispricing tends to be concentrated in firms with high idiosyncratic volatility and high transaction costs. After we control for these factors, our regression results suggest that the relation between accounting accruals and subsequent size-adjusted returns is significantly less negative for firms with analyst cash flow forecasts. The finding that analyst cash flow forecasts reduce the accrual anomaly can be attributed to the reduced tendency of management to manipulate accruals, to the improved ability of investors to price earnings manipulations, or to both. On one hand, the issuance of cash flow forecasts makes accrual-based earnings manipulations more transparent and thus reduces managerial incentives to manipulate earnings. Consistent with this conjecture, Call (2008) and McInnis and Collins (2011) find that managers are more likely to report high-quality earnings and less likely to manipulate earnings through accruals when analysts issue cash flow forecasts. Given Xie (2001), the reduction in earnings manipulation and higher-quality earnings will reduce accrual mispricing even if the ability of investors to price earnings manipulations remains the same. On the other hand, analyst cash flow forecasts encourage both investors and analysts to pay more attention to both the cash and accrual components, rather than fixating on aggregate earnings. In this case, investors are more likely to improve their ability to price earnings manipulation imbedded in accruals. To disentangle these two possibilities, we examine the ability of investors to price discretionary accruals — the earnings manipulations component in accruals. Our evidence is consistent with the notion that analyst cash flow forecasts improve investors' ability to accurately price managerial manipulations. Therefore, we conclude that the reduction in accrual mispricing associated with analyst cash flow forecasts is not totally due to reduced earnings manipulations in accruals, but is at least partly due to the improved ability of investors to price earnings manipulations. Lastly, we attempt to understand why analyst cash flow forecasts bring about the improved ability of investors to price earnings manipulations. We provide several non-mutually-exclusive explanations for this finding. The first explanation is directly related to the limited attention capacity of investors. Hirshleifer and Teoh (2003) point out that human beings have limited attention and limited information processing capacities and, thus, tend to focus on and react to salient information (such as accounting earnings), while ignoring information that is equally relevant but less salient (such as accounting accruals). Hirshleifer and Teoh (2003) argue that it is this behavioral bias that gives rise to the accrual anomaly. In the same spirit, Hirshleifer, Lim, and Teoh (2011) use a theoretical model to further point out that the strength of the accrual anomaly decreases when investors pay more attention to earnings components. It is possible that analyst cash flow forecasts alleviate the limited attention problem and directly improve the ability of investors to “see through” earnings manipulations in accruals. The second explanation is related to the behavior of financial analysts. Bradshaw, Richardson, and Sloan (2001) find that analysts tend to be overly optimistic about firms with high accruals and overly pessimistic about firms with low accruals. These results suggest that financial analysts' erroneous forecasts are at least partially responsible for the mispricing of accruals. Call et al. (2009, 2013) show that analysts incorporate accrual information in their forecasts and that earnings forecasts accompanied by cash flow forecasts tend

to be more accurate. Given the importance of analysts as a financial intermediary in general and their documented contribution to the accrual anomaly in particular, it is reasonable to conjecture that the improved ability of investors to price earnings manipulations is related to more accurate earnings forecasts made by financial analysts. In addition, when analyst earnings forecasts are accompanied by cash flow forecasts, accrual information is indicated implicitly. Therefore, we conjecture that more accurate cash flow forecasts will help investors to detect earnings management imbedded in accruals, and thus reduce the mispricing of earnings manipulations. The third explanation is related to voluntary disclosures of cash and accrual information. DeFond and Hung (2003) find that firms with analyst cash flow forecasts tend to have larger accruals, more heterogeneous accounting choices, high earnings volatility, high capital intensity and poor financial health. All of these traits reduce the usefulness of accounting earnings of those firms for forecasting and valuation. It is thus conceivable that, to satisfy the demand of investors and financial intermediaries, those firms voluntarily disclose more information, which enables investors to better price earnings manipulations. Consistent with this view, Levi (2008) and Louis, Robinson, and Sbaraglia (2008) find that the accrual strategy is less effective for firms that voluntarily provide accrual information at earnings announcements. To investigate these explanations, we use the proportion of analysts following the firm also issuing cash flow forecasts as a proxy for investor attention, and analyst forecast error in earnings and in cash flows as measures of the accuracy of analyst forecasts. We investigate how these variables affect the mispricing of discretionary accruals in a subsample of firms with analyst cash flow forecasts. Our empirical test shows that the increased investor attention and the improved accuracy in analyst earnings forecasts help investors to price discretionary accruals more accurately. We could not empirically test the third explanation due to the limited resources to manually collect data for large sample and the incompleteness of the current machine readable data on voluntary cash flow disclosure (Chuk, Matsumoto, & Miller, 2013). Our paper integrates the literature on the accrual anomaly and analyst cash flow forecasts by investigating the impact of analyst cash flow forecasts on the accrual anomaly. We document that the accrual strategy tends to be less effective for firms with analyst cash flow forecasts, after controlling for idiosyncratic volatility, transaction costs and firm characteristics associated with the issuance of analyst cash flow forecasts. Our investigations also show that the mitigated accrual anomaly for firms with analyst cash flow forecasts is at least partially due to the improved ability of investors to price earnings manipulations. We provide further evidence that the impact of cash flow forecasts on the ability of investors to price accruals is attributed to investor attention and the accuracy of analyst earnings forecasts. Thus, our paper not only contributes to our understanding of the accrual anomaly, but also provides evidence on how investors use the information in analyst cash flow forecasts. Our paper also has implications for investors. We find that the accrual strategy is more profitable for firms without analyst cash flow forecasts, which implies that investors should avoid firms with analyst cash flow forecasts to maximize trading profits from the accrual strategy. Another current study by Mohanram (in press) shares the same implication as our study. However, it is important to note that our paper differs from Mohanram (in press)3 in three ways. First, unlike Mohanram (in press), who uses the Heckman second stage test and the propensity score test, we comprehensively control for all the variables that might be related to the availability of analyst cash flow forecasts and the accrual anomaly documented by the previous 3 While our paper overlaps with Mohanram (2013) in showing that analyst cash flow forecasts reduce the accrual anomaly, our paper is developed independently of and concurrently with Mohanram (2013).

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx

literature such as DeFond and Hung (2003) and Mashruwala et al. (2006).4 This research design alleviates the concern that idiosyncratic volatility, transaction costs and firm characteristics are the factors driving the differential effectiveness of accrual strategy between firms with and without cash flow forecasts. Second, following Hribar and Collins (2002), our study computes accruals using cash flow statement items instead of balance sheet items; this method is immune to errors due to merger and acquisition, divestiture and foreign exchange transactions. Last and most importantly, we not only ask whether analyst cash flow forecasts reduce the accrual anomaly, as does Mohanram (in press), but we also attempt to explain the mitigation effect on the accrual anomaly with analyst cash flow forecasts. The remainder of the paper is organized as follows. Section 2 covers our research design and variable definitions. Section 3 discusses the sample formation and presents empirical results. Section 4 further investigates reasons for the reduction in accrual mispricing with analyst cash flow forecasts. Section 5 presents our conclusions based on the study findings.

2. Research design and variable definitions To examine whether the accrual anomaly is less pronounced for firms with analyst cash flow forecasts, we first examine the hedge returns generated by the accrual strategy among firms with and without analyst cash flow forecasts. Following Hribar and Collins (2002), we compute accruals as [Income before extraordinary items (IBC) − Net cash flow from operating activities (OANCF) + Extraordinary items and discontinued operations (XIDOC)] / Average total assets (AT).5 Similar to Sloan (1996), we first form ten accrual deciles at the end of the fourth month after current fiscal year end, because calculations of accruals require information from financial statements that are publicly available by that time (Alford, Jones, & Zmijewski, 1994). Then, for each decile, we track the subsequent size-adjusted returns, SARt + 1, computed as the firm's 12-month buy-and-hold returns starting four months after the current fiscal year end minus the buy-and-hold returns in the same window on a value-weighted portfolio of firms in the same CRSP size decile.6 The accrual strategy return is the return to the arbitrage portfolio, buying the lowest accrual decile and short-selling the highest accrual decile. We determine whether a firm has analyst cash flow forecasts, DUMMYC, by examining the availability of cash flow forecasts for the firm in the month of the current fiscal year end. DUMMYC takes

4 We do not use the Heckman (1979) two-stage self-selection model to control for the endogeneity of analysts' decision to issue cash flow forecasts, for two reasons. First, the Heckman (1979) two-stage approach requires the inclusion of some variables in the choice model (i.e., the model focusing on analysts' decision to issue cash flow forecasts) that do not appear in the treatment model (i.e., the model focusing on the mispricing of accruals). However, firm characteristics related to the issuance of cash flow forecasts, such as earnings volatility, may affect investors' valuation of accruals and cannot be excluded from the treatment model. As Wooldridge (2006) points out, violation of the exclusion restriction leads to inefficient estimates with overstated standard errors. Furthermore, Lennox, Francis, and Wang (2012) suggest the importance of identifying an instrument variable that is related to the likelihood being modeled in the first-stage choice model but unrelated to the dependent variable in the second-stage treatment model. An inappropriate instrument variable leads to fragile inferences and severe multicollinearity problems. Given the difficulty of identifying an appropriate instrument variable, we choose not to rely on the Heckman procedure but to control for firm characteristics associated with the issuance of cash flow forecasts in a multivariate regression. 5 Compustat mnemonics are presented in parentheses on the right hand side of the variable definition equation. 6 Following Sloan (1996), when a firm delists, we use the delisting return in the delisting month and assume a return equal to the firm's size-matched portfolio for the remainder of the year. If a delisting is due to liquidation or a forced delisting and the delisting return is missing, the delisting return is set to −30%, following Shumway (1997). Our inferences do not change without this treatment of delisting returns.

3

the value of one if analyst cash flow forecast is available at the month of current fiscal year end, and zero otherwise. This definition ensures that the availability of analyst cash flow forecasts is known prior to portfolio formation and precludes foresight bias.7 We predict that the accrual strategy would generate lower returns for firms with analyst cash flow forecasts. To test our prediction, we compare returns from the accrual strategy across firms with and without analyst cash flow forecasts. To consider the effect of idiosyncratic volatility, transaction costs and firm characteristics associated with the issuance of analyst cash flow forecasts, we control for two sets of variables. First, DeFond and Hung (2003) find that analysts are more likely to issue cash flow forecasts for firms with a larger magnitude of accruals, more heterogeneous accounting choices, higher earnings volatility, higher capital intensity, poorer financial health and larger firm size. Therefore, our model controls for these six firm characteristics. Second, Mashruwala et al. (2006) find that the accrual anomaly is concentrated in firms with high idiosyncratic volatility, low stock price, and low trading volume. This finding, in addition to the possibility that firms with and without analyst cash flow forecasts may differ in idiosyncratic risk and transaction costs (proxied by stock price and trading volume), necessitates controlling for these variables. We control for these variables and define them following DeFond and Hung (2003) and Mashruwala et al. (2006). Detailed definitions are provided in Appendix A. Following Mashruwala et al. (2006), we estimate the following model: SARtþ1 ¼ β0 þ β1 DUMMYC t þ β2 RACCRt þ β3 RACCRt  DUMMYC t

ð1Þ

þβ4 RACCRt  RCONTROL þ β5 RSIZEt þ β6 RBETAt þβ7 RBTOM t þ β8 RETOP t þ εt where:SARt + 1 is the size-adjusted return for the 12-month return window starting four months after the current fiscal year end; DUMMYC is a dummy variable equal to one if a firm has available analyst cash flow forecast in the month of current fiscal year end and zero otherwise; RACCR is the decile rank based on accruals; RCONTROL is the decile rank of control variables, including RMAG (the decile rank of “the magnitude of accruals”), RCHOICE (the decile rank of “accounting choice heterogeneity”), RCAP (the decile rank of “capital intensity”), REVOL (the decile rank of “earnings volatility”), RALTZ (the decile rank of “Altman's Z-score”), RSIZE (the decile rank of the firm's market value of equity), RIDVOL (the decile rank of “idiosyncratic volatility”), RPRICE (the decile rank of “price”), and RVOL (the decile rank of “volume”); RSIZE, RBETA, RBTOM and RETOP are, respectively, the decile rank of firm size (size), CAPM beta (beta), the book-to-market ratio (B/M), and the earnings-to-price ratio (E/P). These variables are known predictors of future returns (Fama & French, 1992, 1995, 1996; Lakonishok, Shleifer, & Vishny, 1994). All deciles are formed annually and scaled between zero and one. If analyst cash flow forecasts reduce mispricing of accruals, the coefficient on the interaction term, RACCR ∗ DUMMYC, should be positive and significant, indicating that returns from the accrual strategy are lower for firms with analyst cash flow forecasts. The reduced accrual mispricing for firms with analyst cash flow forecasts can be attributed to reduced earnings manipulations (McInnis & Collins, 2011), to the improved ability of investors to “see through” earnings manipulations, or to both. To evaluate these possibilities, we further test whether investors exhibit improved abilities to price

7 Foresight bias is a serious concern in our analysis given evidence from McNichols, and O'Brien, (1997) and Das, Guo, and Zhang (2006). Because cash flow forecasts issued subsequent to portfolio formation can be influenced by the firms' return performance, considering such forecasts would create an endogeneity issue in our analysis.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

4

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx

discretionary accruals, which measure managerial earnings manipulations, for firms with analyst cash flow forecasts. We replace the decile rank of accruals (RACCR) with the decile rank and the continuous value of discretionary accruals in Eq. (1). We have the following equation: SARtþ1 ¼ β0 þ β1 DUMMYC t þ β2 DACC t þ β3 DACC t  DUMMYC t

ð2Þ

þβ4 DACC t  RCONTROL þ β5 RSIZEt þ β6 RBETAt þβ7 RBTOMt þ β8 RETOP t þ εt where:DACC is the decile rank or the continuous value of discretionary accruals estimated from the Dechow and Dichev Model (Dechow & Dichev, 2002) or from the Allen, Larson and Sloan Model (the ALS Model hereafter; Allen, Larson, & Sloan, 2013). Please refer to Appendix A for the definition of discretionary accruals for the two measures.8 The definitions of other variables are the same as in Eq. (1). If analyst cash flow forecasts reduce the mispricing of accruals at least partially through the improved ability of investors to price earnings manipulations, the coefficient on the interaction term, DACC ∗ DUMMYC, will be positive and significant. To further explain the improved ability of investors to “see through” earnings manipulations with analyst cash flow forecasts, we investigate several non-mutually-exclusive explanations in a subsample of firms with available analyst cash flow forecasts (where DUMMYC = 1). Specifically, we conjecture that investor attention, the accuracy of analyst forecasts, and voluntary disclosures of cash or accrual may help investors to improve their pricing abilities. For the subsample tests, the regression model is modified from regression (2): SARtþ1 ¼ β0 þ β1 RDACC t þ β2 EXP þ β3 RDACC t  EXP

ð3Þ

þβ4 RDACC t  RCONTROL þ β5 RSIZEt þ β6 RBETAt þβ7 RBTOMt þ β8 RETOPt þ εt where:RDACC in Eq. (3) is the decile rank of the discretionary accrual from the ALS Model.9 EXP represents the potential explanatory variables for the improved ability of investors to price earnings manipulations. Specifically, we use the proportion of analysts following the firm also issuing cash flow forecasts (PERCFF) to proxy for investor attention, and the absolute forecast error in earnings (AFEE) and that in cash flows (AFECF) to measure the accuracy of analyst forecasts. If these variables explain the improved ability of investors to price earnings manipulation with analyst cash flow forecasts, we expect the coefficients on RDACC ∗ PERCFF to be positive and significant, and the coefficients on RDACC ∗ RAFEE and RDACC ∗ RAFECF to be negative and significant. 3. Empirical results 3.1. Initial sample

sample spans from 1993 to 2009. The initial sample consists of 55,961 firm-year observations. Panel A in Table 1 presents the frequency distribution of firms with analyst earnings forecasts and cash flow forecasts from 1993 to 2009 by year. Consistent with the findings of DeFond and Hung (2003), we observe that both the number and the proportion of firms with analyst cash flow forecasts increase steadily over time. Specifically, the number of analyst cash flow forecasts increased from 31 in 1993 to 1338 in 2009, and the proportion increased from 1% to nearly 50%. Both the absolute number and the proportion of firms with cash flow forecasts increased dramatically in 2002. It is possible that the accounting scandals of the early 2000s resulted in elevated investor suspicion of accounting earnings and, subsequently, an increased demand for analyst cash flow forecasts. Overall, our results suggest that analyst cash flow forecasts have become increasingly popular, consistent with DeFond and Hung (2003). Panel B presents the frequency distribution of firms with analyst cash flow forecasts across industries during the sample period. We follow the industry definitions in Fama and French (1997). Of the whole sample, 14.35% represents the petroleum and natural gas industry. The business service industry, retail industry, telecommunications industry, electronics industry and transportation industry each contribute to over 5% of our sample. 3.2. Main results 3.2.1. Descriptive statistics To test our hypothesis, we follow DeFond and Hung (2003) and trim the top and bottom 0.5% of the sample population with respect to sizeadjusted returns, accruals, cash flows, earnings, natural log of the market value, Altman's Z-score, capital intensity, and earnings volatility. We do not trim the accounting choice heterogeneity variable because it takes on only a limited number of values. Following Gutierrez and Kelly (2008), we require the stock price at the beginning of current fiscal year to be greater than $5 per share, given that the bid–ask bounce contaminates the return information of low-price firms. The final sample consists of 32,443 observations. Table 2 presents summary descriptive statistics for the key variables for firms with and without analyst cash flow forecasts. The two far-right columns show ##p-values for t-tests and Wilcoxon tests comparing the mean and median of firms with and without cash flow forecasts. We find that firms with analyst cash flow forecasts have significantly more assets, higher market value, more heterogeneous accounting choices, higher capital intensity, and poorer financial health than those without analyst cash flow forecasts.10 Moreover, firms with analyst cash flow forecasts have significantly lower idiosyncratic volatility, higher price, and higher trading volume, which predict lower accrual strategy return and highlight the importance of controlling for these variables in our multivariate tests (Mashruwala et al., 2006).

We obtain return and accounting data from CRSP and Compustat, respectively, for NYSE/AMEX/NASDAQ firms. We require total accruals and size-adjusted returns for the next year to be available. To avoid the potential foresight bias documented in Kraft, Leone, and Wasley (2006), earnings for the following year are not required for inclusion in our sample. We obtain forecast data from the I/B/E/S summary file and require that at least one earnings forecast is available in the month when the current fiscal year ends. Given that the data on analyst cash flow forecasts prior to 1993 are not available from I/B/E/S, our

3.2.2. Univariate analysis: Accrual anomaly and cash flow forecasts To test our hypothesis that analyst cash flow forecasts reduce the accrual anomaly, we first form ten accrual deciles based on all the sample firms each year and then examine the accrual strategy return separately for firms with and without analyst cash flow forecasts. Panel A in Table 3 reports the size-adjusted returns across the ten accrual deciles for the two groups of firms. The mean values of sizeadjusted returns for the ten accrual deciles are reported in the first

8 In untabulated robustness tests, we find similar results using alternative discretionary accrual measures from the Jones Model (Jones, 1991), the Modified Jones Model (Dechow, Sloan, & Sweeney, 1995), and the Kothari Model (Kothari, Leone, & Wasley, 2005). 9 In untabulated robustness tests, we find similar results when using alternative measures of discretionary accruals.

10 Our results are slightly different from those of DeFond and Hung (2003) in that (1) the mean value of the magnitude of accruals is significantly lower for firms with analyst cash flow forecasts, and (2) the differences in the mean value of earnings volatility between the two groups is not significant. This difference is likely due to the different sample periods. Using their sample period, we are able to replicate their findings.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx Table 1 (continued)

Table 1 Frequencies of firms with analyst cash flow forecasts by year and by industry. Panel A: Inter-temporal frequencies of firms with analyst earnings forecasts and analyst cash flow forecasts. Year

Number of firms with analyst earnings forecasts

Number of firms with analyst cash flow forecasts

Percent of firms with analyst cash flow forecasts

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Total

3071 3457 3439 4151 4281 4078 3898 3663 3107 2793 2732 2896 2909 2982 2970 2794 2740 55,961

31 179 51 220 307 220 654 682 565 881 1126 1252 1368 1404 1449 1367 1338 13,094

1.01 5.18 1.48 5.30 7.17 5.39 16.78 18.62 18.18 31.54 41.22 43.23 47.03 47.08 48.79 48.93 48.83 23.40

Panel B: Frequency distribution of analyst cash flow forecasts by Fama–French 48-industry classification. Industry description Petroleum and natural gas Business service Retail Telecommunications Electronic equipment Transportation Pharmaceutical products Computers Machinery Chemicals Wholesale Medical equipment Steel works Measuring and control equipment Food products Business supplies Restaurants, hotels, motels Healthcare Entertainment Construction Apparel Consumer goods Automobiles and trucks Construction materials Personal services Precious Metals Trading Utilities Printing and publishing Non-metallic and industrial metal mining Aircraft Other Coal Insurance Beer & liquor Shipping containers Recreation Candy & soda Rubber and plastic products Electrical equipment Agriculture Tobacco products Shipbuilding, railroad equipment Defense Banking Textiles Real estate

N

Percent 1879 1299 888 828 807 771 579 570 475 387 349 286 270 255 226 225 219 212 201 189 183 178 167 152 151 127 126 125 109 101 91 78 65 57 56 56 47 42 42 40 36 36 32 23 20 19 13

5

14.35 9.92 6.78 6.32 6.16 5.89 4.42 4.35 3.63 2.96 2.67 2.18 2.06 1.95 1.73 1.72 1.67 1.62 1.54 1.44 1.40 1.36 1.28 1.16 1.15 0.97 0.96 0.95 0.83 0.77 0.69 0.60 0.50 0.44 0.43 0.43 0.36 0.32 0.32 0.31 0.27 0.27 0.24 0.18 0.15 0.15 0.10

(continued on next page)

Panel B: Frequency distribution of analyst cash flow forecasts by Fama–French 48-industry classification. Industry description

N

Percent

Fabricated products Total

7 13,094

0.05 100.00

The sample consists of 55,961 firm-year observations. All those firms are listed on NYSE/ AMEX/NASDAQ with non-missing total accruals, analyst one-year-ahead earnings forecasts and one-year-ahead size-adjusted returns. Analyst earnings forecasts and analyst cash flow forecasts are obtained from I/B/E/S.

ten columns. The next two columns, “Hedge” and “pa”, present the pooled accrual strategy return and the associated p-value from a two-tailed t-test, respectively. For firms without analyst cash flow forecasts, the accrual strategy return is 12.12% and significant at the 0.01 level. For firms with analyst cash flow forecasts, the return is 0.19% and not significant. The next two columns, “Mean” and “p b”, report the mean value of the annual accrual strategy return and the associated p-value from a two-tailed t-test, respectively. The mean annual accrual strategy return is 10.91% and significant at the 0.01 level for firms without analyst cash flow forecasts but is only − 3.81% and not significant for firms with available cash flow forecasts. In the last column, we report the number of years when the accrual strategy works. The accrual strategy generates positive returns in 12 out of 17 years for firms without cash flow forecasts but only 7 out of 17 years for firms with cash flow forecasts.11 As a robustness check, we form accrual deciles separately for the two groups of firms, and the results in Panel B of Table 3 continue to hold.12

3.2.3. Multivariate analysis: Accrual anomaly and cash flow forecasts Table 4 presents the results of multivariate regressions after we control for the possible confounds such as idiosyncratic volatility, transaction costs and firm characteristics associated with the issuance of cash flow forecasts as discussed earlier. The regression results of Model 1 are consistent with our findings presented in Table 3. The coefficient on RACCR is − 0.115, indicating that the accrual strategy yields an annual size-adjusted buy-and-hold return of 11.5% for firms without analyst cash flow forecasts. The coefficient on RACCR ∗ DUMMYC is 0.070, significant at the 0.01 level, suggesting that analyst cash flow forecasts significantly reduce the accrual strategy return by 7%.13 To address the concern that returns might be correlated cross-sectionally in each industry and year, we control for industry fixed effect, year fixed effect, and both in Model 2, Model 3 and Model 4 and find results consistent with those of Model 1. Next, we further investigate whether accrual mispricing is completely mitigated for firms with analyst cash flow forecasts by testing whether β2 + β3 = 0. Consistent with our findings in the univariate test in Table 3, β2 + β3 is not significantly different from zero, indicating that the accrual strategy does not work for firms with analyst cash flow forecasts. To summarize, our empirical results indicate that the accrual strategy is significantly less effective for firms with analyst cash flow forecasts after we control for idiosyncratic volatility, transaction costs and firm characteristics associated with the issuance of cash flow forecasts. 11 Our untabulated results for annual analysis are consistent with Green, Hand, and Soliman (2011). We find that the accrual strategy return is either insignificant or takes on an unexpected sign during 2003–2006, even for firms without cash flow forecasts. 12 Untabulated results show that all our findings in later regressions continue to hold when portfolios in each group of firms are formed separately. 13 As Mashruwala et al. (2006) point out, it is reasonable that the coefficient estimate on RACCR does not exactly match the hedge portfolio return of buying firms in the top accrual decile and of shorting firms in the bottom accrual decile.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

6

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx

Table 2 Summary descriptive statistics. Variable

N

Mean

Std

Median

p-Value of test of difference t-Testa

Wilcoxon testb

Average total assets Without cash flow forecasts With cash flow forecasts

23,030 9413

1133 5363

3916 11,215

303 1827

b0.01

b0.01

Market value of equity Without cash flow forecasts With cash flow forecasts

23,030 9413

1457 6072

4661 11,157

356 1944

b0.01

b0.01

Magnitude of accruals Without cash flow forecasts With cash flow forecasts

23,030 9413

7.93% 7.59%

8.90% 7.15%

5.84% 5.96%

0.00

0.03

Accounting choice heterogeneity Without cash flow forecasts With cash flow forecasts

23,030 9413

0.09 0.11

0.11 0.12

0.00 0.00

b0.01

b0.01

Capital intensity Without cash flow forecasts With cash flow forecasts

23,030 9413

0.76 1.27

1.76 2.04

0.41 0.55

b0.01

b0.01

Earnings volatility Without cash flow forecasts With cash flow forecasts

23,030 9413

1.91 1.89

5.04 5.41

0.62 0.55

0.75

b0.01

Altman's Z-score Without cash flow forecasts With cash flow forecasts

23,030 9413

6.27 4.69

7.66 5.28

4.16 3.30

b0.01

b0.01

Idiosyncratic volatility Without cash flow forecasts With cash flow forecasts

23,030 9413

2.25% 1.49%

6.15% 2.19%

1.30% 0.90%

b0.01

b0.01

Price Without cash flow forecasts With cash flow forecasts

23,030 9413

b0.01

b0.01

Volume Without cash flow forecasts With cash flow forecast

23,030 9413

b0.01

b0.01

21.66 31.43

20.44 29.39

71,568 343,010

17.00 26.54

240,170 712,446

22,199 129,445

The sample consists of 32,443 firm-year observations listed on NYSE/AMEX/NASDAQ from 1993 to 2009, after (1) trimming the top and bottom 0.5% of the sample population with respect to accruals, cash flows, earnings, natural log of the market value, Altman's Z-score, capital intensity, and earnings volatility; (2) requiring all the variables listed in this table to be nonmissing; and (3) requiring the price at the beginning of current fiscal year ≥$5. Definitions of all variables are presented in Appendix A. a The t-test tests the null hypothesis that the mean difference between observations with and without cash flow forecasts is zero. b The Wilcoxon test, a non-parametric statistical method, tests the null hypothesis that the median difference between observations with and without cash flow forecasts is zero.

4. Explanations for the mitigated accrual anomaly 4.1. The ability of investors to price earnings manipulations The reduction in the mispricing of accounting accruals among firms with analyst cash flow forecasts can be attributed to the

reduced tendency of managements to manipulate earnings, as shown in Call (2008) and McInnis and Collins (2011), to the improved ability of investors to “see through” earnings manipulations, or to both. In this sub-section, we test whether investors improve their ability to price discretionary accruals, the earnings manipulation component in accruals, for firms with analyst cash flow forecasts.

Table 3 Group comparison on portfolio size-adjusted returns on different accrual deciles. Hedge

pa

Mean

pb

# of years

−8.55% 4.43%

12.12%*** 0.19%

0.000 0.945

10.91%*** −3.81%

0.009 0.553

12 7

Panel B: Accrual deciles are formed based on each group of firms (with vs. without analyst cash flow forecasts) each year. Without cash flow forecasts 3.39% 5.85% 1.85% 3.37% 1.01% 0.62% −0.18% −2.60% −2.50% −10.25% With cash flow forecasts 5.78% 6.09% 7.15% 5.43% 3.56% 1.53% 4.18% 2.22% 1.37% 2.44%

13.63%*** 3.35%

0.000 0.175

12.12%*** −5.81%

0.004 0.321

13 7

Accrual deciles

Lowest accrual

2

3

4

Panel A: Accrual deciles are formed based on all firms each year. Without cash flow forecasts 3.57% 5.87% 1.86% 3.34% With cash flow forecasts 4.62% 6.59% 6.95% 4.68%

5

6

0.35% 4.29%

1.62% 2.79%

7

0.15% 2.12%

8

9

−2.83% 3.62%

−3.06% −0.97%

Highest accrual

Decile ranks are based on accruals in the entire sample of firms each year. Accruals are defined in the Appendix A. Analyst cash flow forecasts are obtained from I/B/E/S. The symbols *, **, and *** denote significance at the 10%, 5% and 1% levels, respectively. Some of the cells report missing values because there are no observations in that portfolio. Hedge: the return to the hedge portfolio with a long position in the bottom accrual decile and a short position in the top accrual decile. pa: the p-value associated with “hedge”, using a two-tailed t-test. Mean: the mean value of annual hedge portfolio returns. pb: the p-value associated with “mean”, using a two-tailed t-test. # of years: the number of years when accrual strategy generates positive hedge returns.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx Table 4 The effect of analyst cash flow forecasts on the mispricing of total accruals (Dependent variable = SARt + 1).

DUMMYC RACCR RACCR ∗ DUMMYC RACCR ∗ RSIZE RACCR ∗ RMAG RACCR ∗ RCHOICE RACCR ∗ RCAP RACCR ∗ REVOL RACCR ∗ RALTZ RACCR ∗ RIDVOL RACCR ∗ RPRICE RACCR ∗ RVOL RSIZE RBETA RBTOM RETOP Intercept p-Value for F-test β2 + β3 = 0 Industry fixed effect Year fixed effect Number of observations Adjusted R2

Model 1

Model 2

Model 3

Model 4

0.003 (0.799) −0.115*** (0.000) 0.070*** (0.002) 0.067* (0.100) −0.012 (0.474) −0.022 (0.217) −0.011 (0.528) −0.001 (0.506) 0.056*** (0.002) 0.001 (0.705) −0.007*** (0.006) −0.004 (0.123) 0.001 (0.951) 0.032*** (0.002) 0.046*** (0.000) 0.093*** (0.000) −0.021 (0.110) 0.2013 No No 32,443 0.0095

−0.006 (0.645) −0.113*** (0.000) 0.079*** (0.001) 0.075* (0.068) −0.004 (0.788) −0.004 (0.829) −0.015 (0.449) −0.002 (0.378) 0.033* (0.078) −0.001 (0.723) −0.007*** (0.004) −0.006* (0.064) 0.005 (0.811) 0.024** (0.017) 0.065*** (0.000) 0.137*** (0.000) −0.043*** (0.002) 0.3531 Yes No 32,443 0.0162

−0.011 (0.422) −0.112*** (0.000) 0.068*** (0.003) 0.059 (0.149) −0.011 (0.489) −0.022 (0.216) −0.010 (0.560) −0.001 (0.556) 0.055*** (0.002) 0.001 (0.733) −0.007*** (0.004) −0.004 (0.131) 0.014 (0.475) 0.032*** (0.001) 0.047*** (0.000) 0.094*** (0.000) −0.023* (0.078) 0.2193 No Yes 32,443 0.0167

−0.019 (0.179) −0.107*** (0.000) 0.080*** (0.001) 0.068* (0.096) −0.004 (0.789) −0.006 (0.762) −0.016 (0.409) −0.002 (0.418) 0.032** (0.085) −0.001 (0.699) −0.008*** (0.003) −0.006* (0.059) 0.015 (0.423) 0.025*** (0.016) 0.064*** (0.000) 0.137*** (0.000) −0.033* (0.079) 0.4534 Yes Yes 32,443 0.0234

The symbols *, **, and *** denote significance at the 10%, 5% and 1% levels, respectively. p-Values are reported in the parentheses under the coefficients based on a two-tailed t-test. SARt + 1: Size-adjusted return for firm in year t + 1. Please refer to Appendix A for detailed definition. DUMMYC: Dummy variable equal to one if analyst cash flow forecast is available and equal to zero otherwise. RACCR: Decile rank of “Accruals” for each year t, scaled between zero and one. Please refer to Appendix A for the definition of “Accruals”. RSIZE: Decile rank of “market value of equity”, scaled between zero and one. Please refer to Appendix A for the definition of “market value of equity”. RMAG: Decile rank of “magnitude of accruals”, scaled between zero and one. Please refer to Appendix A for the definition of “magnitude of accruals”. RCHOICE: Decile rank of “accounting choice heterogeneity”, scaled between zero and one. Please refer to Appendix A for the definition of “accounting choice heterogeneity”. RCAP: Decile rank of “capital intensity”, scaled between zero and one. Please refer to Appendix A for the definition of “capital intensity”. REVOL: Decile rank of “earnings volatility”, scaled between zero and one. Please refer to Appendix A for the definition of “earnings volatility”. RALTZ: Decile rank of “Altman's Z-score”, scaled between zero and one. Please refer to Appendix A for the definition of “Altman's Z-score”. RIDVOL: Decile rank of “idiosyncratic volatility”, scaled between zero and one. Please refer to Appendix A for the definition of “idiosyncratic volatility”. RPRICE: Decile rank of “price”, scaled between zero and one. Please refer to Appendix A for the definition of “price”. RVOL: Decile rank of “volume”, scaled between zero and one. Please refer to Appendix A for the definition of “volume”. RBETA: Decile rank of “Beta”, scaled between zero and one. Please refer to Appendix A for the definition of “Beta”. RBTOM: Decile rank of “B/M”, scaled between zero and one. Please refer to Appendix A for the definition of “B/M”. RETOP: Decile rank of “E/P”, scaled between zero and one. Please refer to Appendix A for the definition of “E/P”.

7

We first replace the decile rank of accruals with the decile rank of discretionary accruals in Eq. (1) and run the regression with all the control variables and the industry and year fixed effects. The first column of Table 5 presents the results when discretionary accruals are estimated by the Dechow and Dichev Model (Dechow & Dichev, 2002). Consistent with Xie (2001), the coefficient of the rank of discretionary accruals is negative and significant at the 0.01 level. More importantly, the interaction term of DACC ∗ DUMMYC is positive and significant at the 0.01 level, suggesting that the ability of investors to correctly price earnings manipulations is stronger for firms with analyst cash flow forecasts. It is possible that the magnitude of discretionary accruals can be different for firms with and without analyst cash flow forecasts in the same decile rank and that this difference can affect the accrual strategy return. In order to address this concern, we use the continuous value of discretionary accruals estimated from the Dechow and Dichev Model in the second column. The coefficient on the interaction term of DACC ∗ DUMMYC is still positive and significant at the 0.01 level, consistent with our findings using the decile rank of discretionary accruals. The next two columns show the results when we use the alternative measure of discretionary accruals from the ALS Model; the results do not change. 14 In addition, the p-values for F-test β2 + β3 = 0 in Table 5 are not statistically significant, indicating that the mispricing of discretionary accruals is not significantly different from zero for firms with analyst cash flow forecasts. In summary, from the results in Table 5, we conclude that the reduction in accrual mispricing associated with analyst cash flow forecasts is at least partially attributed to the improved ability of investors to price accrual manipulations.

4.2. Alternative explanations for the improved investor ability To understand the improved ability of investors to price earnings manipulations, we further investigate several non-mutually-exclusive alternative explanations with a subsample of firms with available analyst cash flow forecasts.

4.2.1. Investor attention Psychological evidence (e.g., Fiske, 1995; Hirshleifer & Teoh, 2003; Peng & Xiong, 2006) shows that attention is a scarce cognitive resource. Proposing a theoretical model that human beings tend to focus on and react to information that is salient (such as accounting earnings) while ignoring information that is equally relevant but less salient (such as accounting accruals), Hirshleifer and Teoh (2003) try to use this behavioral bias to explain the accrual anomaly. In the same spirit, Hirshleifer et al. (2011) point out that the strength of the accrual anomaly decreases when investors pay more attention to earnings components. According to Hirshleifer and Teoh (2003), “the salience of a stimulus is its ‘prominence’ tendency to ‘stand out’, or its degree of contrast with other stimuli in the environment.” When more analysts issue cash flow forecasts for a firm, the information about cash flows and accruals is more likely to ‘stand out’ and become more ‘prominent’ to investors, making investors pay more attention to cash flows and accruals and fixate less on earnings. Therefore, the number (or the proportion) of analyst cash flow forecasts is an appropriate proxy for investor attention. To control for firm size, we use the proportion of analysts following the firm that also issue cash flow forecasts (PERCFF) as a proxy for investor 14 In untabulated robustness tests, we also check other measures of discretionary accruals estimated from the Jones Model (Jones, 1991), the Modified Jones Model (Dechow et al., 1995), and the Kothari Model (Kothari et al., 2005). The results are consistent.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

8

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx

Table 5 The effect of analyst cash flow forecasts on the mispricing of discretionary accruals (dependent variable = SARt + 1). DACC_DD

DACC_ALS

Decile rank Continuous Decile rank Continuous −0.015 (0.299) DACC −0.103*** (0.001) DACC ∗ DUMMYC 0.070*** (0.002) DACC ∗ RSIZE 0.173*** (0.000) DACC ∗ RMAG −0.010 (0.5553) DACC ∗ RCHOICE 0.008 (0.655) DACC ∗ RCAP −0.012 (0.537) DACC ∗ REVOL −0.024 (0.207) DACC ∗ RALTZ −0.007 (0.685) DACC ∗ RIDVOL −0.008 (0.687) DACC ∗ RPRICE −0.105*** (0.000) DACC ∗ RVOL −0.087*** (0.001) RSIZE −0.013 (0.510) RBETA 0.040*** (0.000) RBTOM 0.045*** (0.000) RETOP 0.107*** (0.000) Intercept −0.016 (0.430) p-Value for F-test β2 + β3 = 0 0.3736 Industry fixed effect Yes Year fixed effect Yes Number of observations 30,542 Adjusted R2 0.0254 DUMMYC

0.020** (0.017) −0.128** (0.013) 0.068*** (0.005) −0.007 (0.909) 0.006 (0.856) 0.045 (0.207) 0.022 (0.523) 0.023 (0.521) −0.021 (0.513) −0.039 (0.332) 0.035 (0.390) 0.004 (0.942) 0.028** (0.011) 0.035*** (0.000) 0.065*** (0.000) 0.110*** (0.000) −0.120*** (0.000) 0.3003 Yes Yes 30,542 0.0205

−0.014 (0.333) −0.092*** (0.002) 0.067*** (0.004) 0.103** (0.015) −0.022 (0.182) 0.007 (0.693) −0.025 (0.190) −0.034* (0.072) 0.007 (0.683) −0.011 (0.603) −0.101*** (0.000) −0.049* (0.071) 0.007 (0.725) 0.042*** (0.000) 0.052*** (0.000) 0.109*** (0.000) −0.026 (0.185) 0.5161 Yes Yes 30,107 0.0269

0.021** (0.016) −0.256*** (0.004) 0.105** (0.016) −0.274** (0.012) 0.001 (0.989) 0.114* (0.077) −0.009 (0.880) −0.031 (0.628) 0.063 (0.255) −0.149** (0.032) 0.089 (0.213) 0.238*** (0.009) 0.025** (0.023) 0.036*** (0.000) 0.067*** (0.000) 0.111*** (0.000) −0.141*** (0.000) 0.1203 Yes Yes 30,107 0.0240

The symbols *, **, and *** denote significance at the 10%, 5% and 1% levels, respectively. p-Values are reported in the parentheses under the coefficients based on a two-tailed t-test. DACC: Discretionary accruals, DACC_DD estimated from the Dechow and Dichev (2002), and DACC_ALS estimated from the Allen et al. (2013), for each industry and year. Decile rank are scaled between 0 and 1. Please refer to Appendix A for a detailed definition of “DACC_DD”, and “DACC_ALS”. The other variables are defined as in Table 4.

attention.15 Firms with a large proportion of analysts issuing cash flow forecasts are expected to increase investor attention on both cash flows and accruals and thus show less mispricing of accruals. The empirical results of this analysis are shown in Model 1 of Table 6. RDACC is the decile rank of discretionary accruals estimated from the ALS Model for each year, scaled between 0 and 1. The coefficient on RDACC ∗ PERCFF is positive and significant at the 0.05 level, indicating that investors are less likely to misprice earnings manipulations in accruals when investors pay more attention to cash flows and accruals.

15 The number of analyst cash flow forecasts is significantly related to firm size while the proportion of analysts following the firms also issuing cash flow forecasts, controlling for firm size, is a better proxy for investor attention. For example, one can expect investors to be more attentive to cash and accrual when 4 out of 5 analysts (80%) issue cash flow forecasts for Firm A than when 6 out of 20 analysts (30%) issue cash flow forecasts for Firm B. We thank the anonymous referee for comments on this issue.

4.2.2. Analyst forecast accuracy Bradshaw et al. (2001) find that analysts tend to be overoptimistic for firms with high accruals and over-pessimistic for firms with low accruals. Their results suggest that erroneous forecasts made by financial analysts are at least partially responsible for the mispricing of accruals. Call et al. (2009) find analyst earnings forecasts to be more accurate when the same analysts also forecast cash flows, suggesting that directly forecasting earnings components alleviates the limited attention problem of analysts. In addition, Call et al. (2013) show that analyst cash flow forecasts include adjustment for accruals and the market reacts to the information in cash flow forecasts issuance and revisions. Given the importance of analysts as a financial intermediary in general and their documented contribution to the accrual anomaly in particular, it is possible to attribute our main findings to the increased accuracy in earnings forecasts made by financial analysts.16 We investigate whether the increased accuracy of analyst earnings forecasts might explain the improved ability of investors to price earnings manipulations for firms with cash flow forecasts. We calculate the analyst forecast error in earnings (AFEE) as the absolute value of actual earnings minus median consensus analyst forecasts, deflated by the stock price as of the fourth month after current fiscal year, with both actual earnings and the stock price from the I/B/E/S summary file. We then form deciles based on the absolute forecast error in earnings annually, and the decile rank is scaled between zero and one. The empirical results of analyst forecast accuracy are reported in Model 2 of Table 6. The coefficient on RDACC ∗ RAFEE is − 0.250, significant at the 0.01 level, suggesting that the mispricing of discretionary accruals is significantly smaller (larger) when analyst earnings forecast error is low (high). To interpret this result economically, for firms with analyst cash flow forecasts, the hedge strategy based on discretionary accruals generates no returns when analyst earnings forecast error is in the lowest decile, but the strategy generates 25% more hedge returns when analyst earnings forecast error is in the highest decile. Since accrual information is indicated implicitly when analyst earnings forecasts are accompanied by cash flow forecasts, we also conjecture that more accurate cash flow forecasts will help investors to detect earnings manipulations imbedded in accruals, and therefore to reduce mispricing of earnings manipulations. We calculate the analyst forecast error in cash flows (AFECF) as the absolute value of actual cash flows minus median consensus analyst forecasts, deflated by the stock price as of the fourth month after current fiscal year, and then form deciles based upon AFECF. The decile rank of AFECF is formed annually and scaled between zero and one. In Model 3 of Table 6, the coefficient on RDACC ∗ RAFECF is − 0.108, significant at the 0.10 level, indicating that less analyst forecast error in cash flows also helps to reduce mispricing of discretionary accruals. In sum, we find that the accuracy of analyst forecasts improves the ability of investors to price earnings manipulations.

4.2.3. Voluntary disclosures DeFond and Hung (2003) report that firms with analyst cash flow forecasts have larger accruals and higher earnings volatility, implying that the quality of earnings is low for those firms. Conceivably, to satisfy investor demand, these firms voluntarily disclose more information, which contributes to the improved ability of investors to price accruals. Consistent with this view, Levi (2008) 16 We note that this explanation is consistent with investors' limited attention being responsible for the accrual anomaly. The difference between this explanation and the previous explanation in Section 4.2.1 is that this explanation focuses on the impact of financial analysts, whereas the previous one focuses on how analyst cash flow forecasts influence investors directly.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx Table 6 Explanations for the improved investors' ability to price earnings manipulations (dependent variable = SARt + 1).

RDACC PERCFF RDACC ∗ PERCFF

Model 1

Model 2

Model 3

Model 4

−0.044 (0.491) −0.029** (0.453) 0.144** (0.014)

0.150** (0.031)

0.073 (0.301)

0.106 (0.158) −0.043 (0.273) 0.168*** (0.004) 0.081** (0.034) −0.254*** (0.000) 0.058 (0.109) −0.013 (0.832) 0.175** (0.031) −0.001 (0.985) −0.008 (0.829) −0.058 (0.135) −0.032 (0.397) −0.013 (0.724) 0.047 (0.224) −0.201*** (0.000) −0.095* (0.082) 0.047 (0.195) 0.092*** (0.000) 0.054** (0.011) 0.052** (0.014) −0.114*** (0.006) Yes Yes 6052 0.0616

RAFEE

0.094*** (0.008) −0.250*** (0.000)

RDACC ∗ RAFEE RAFECF RDACC ∗ RAFECF RDACC ∗ RSIZE RDACC ∗ RMAG RDACC ∗ RCHOICE RDACC ∗ RCAP RDACC ∗ REVOL RDACC ∗ RALTZ RDACC ∗ RIDVOL RDACC ∗ RPRICE RDACC ∗ RVOL RSIZE RBETA RBTOM RETOP Intercept Industry fixed effect Year fixed effect Number of observations Adjusted R2

0.233*** (0.004) −0.003 (0.928) −0.014 (0.702) −0.066* (0.087) −0.041 (0.265) −0.011 (0.758) 0.041 (0.291) −0.179*** (0.000) −0.104* (0.055) 0.015 (0.661) 0.096** (0.000) 0.060*** (0.006) 0.052** (0.014) −0.034 (0.334) Yes Yes 6052 0.0577

0.144* (0.071) 0.001 (0.976) −0.000 (0.998) −0.038 (0.317) −0.025 (0.500) −0.022 (0.539) 0.054 (0.167) −0.196*** (0.000) −0.101* (0.062) 0.053 (0.129) 0.094*** (0.000) 0.062*** (0.004) 0.056*** (0.008) −0.116*** (0.002) Yes Yes 6052 0.0594

0.090*** (0.008) −0.108* (0.057) 0.179** (0.024) −0.002 (0.944) −0.008 (0.820) −0.042 (0.270) −0.039 (0.295) −0.016 (0.659) 0.042 (0.279) −0.174*** (0.000) −0.116** (0.033) 0.048 (0.173) 0.094*** (0.000) 0.053** (0.012) 0.049** (0.021) −0.101*** (0.006) Yes Yes 6052 0.0574

The symbols *, **, and *** denote significance at the 10%, 5% and 1% levels, respectively. pValues are reported in the parentheses under the coefficients based on a two-tailed t-test. RDACCt: Decile rank of DACC_ALS, discretionary accruals estimated from the Allen et al. (2013) Model for each year and industry, scaled between 0 and 1. Please refer to Appendix A for the detailed definition of “DACC_ALS”. PERCFF: The proportion of analysts following the firm also issuing cash flow forecasts, equal to the number of analyst cash flow forecasts relative to the number of analyst earnings forecasts for each company. RAFEE: Decile rank of “Analyst Forecast Error in Earnings” for year t + 1, scaled between zero and one. Please refer to Appendix A for the definition of “Analyst Forecast Error in Earnings”. RAFECF: Decile rank of “Analyst Forecast Error in Cash Flows” for year t + 1, scaled between zero and one. Please refer to the Appendix A for the definition of “Analyst Forecasts Error in Cash Flows”. The other variables are defined as in Table 4.

and Louis et al. (2008) find that the accrual strategy is less effective when firms voluntarily provide accrual information at earnings announcements. As data on voluntarily disclosed accrual information at the time of earnings announcement are not immediately available for our large sample under investigation,17 we must rely on machine readable databases, such as First Call Company Issued Guidance (CIG 17 Both Levi (2008) and Louis et al. (2008) manually collect information about voluntary disclosure of accruals at the time of earnings announcements.

9

hereafter), for these data. However, one significant problem is that “[CIG] has multiple biases regarding firm and forecast characteristics as well as changes in coverage over time” (Chuk et al., 2013). Therefore, we cannot safely rely on CIG for voluntary cash flow forecasts.18 In untabulated tests, we use the availability of management forecasts on either earnings or cash flows from the CIG database as a rough proxy for voluntary disclosure but do not find significant results. However, the results could be driven largely by the inappropriate proxy, and we cannot conclusively rule out the possibility that voluntary disclosures about cash flows/accruals explain the reduced mispricing of discretionary accruals in the presence of analyst cash flow forecasts. We leave it to later studies to investigate this possibility. 4.2.4. Overall investigation The last column of Table 6 presents the results including investor attention (PERCFF) and the accuracy of analyst forecasts (RAFEE and RAFECF). The coefficient on RDACC ∗ PERCFF is positive and significant at the 0.01 level, while the coefficient on RDACC ∗ RAFEE is negative and significant at the 0.01 level. However, the coefficient on RDACC ∗ RAFECF is no longer significant after we control for the accuracy of earnings forecast in the same model, indicating that the accuracy of cash flow forecasts is not a major explanation for the improved investor ability to price earnings manipulations. Overall, our result implies that both investor attention and the accuracy of analyst earnings forecasts help to explain the improvement in the ability of investors to price earnings manipulations. Nevertheless, readers should exercise caution when interpreting the findings in Section 4.2. First, the explanations under investigation are not mutually exclusive. For example, the accuracy of analyst forecasts and investor/analyst attention could be highly correlated. Second, the measurements that we use to proxy for the possible explanations may have limitations. Given the limited data availability, we will leave it to later studies to develop more precise proxies for a closer investigation. Finally, there may be other explanations that have not been covered in the above discussions. We encourage further investigation. 5. Conclusions This paper examines how analyst cash flow forecasts affect the accrual anomaly. We predict and find that the accrual strategy return is lower for firms with analyst cash flow forecasts. This result holds after we control for idiosyncratic volatility, transaction costs and firm characteristics associated with the issuance of analyst cash flow forecasts. We further show that the mitigation of the accrual anomaly with analyst cash flow forecasts is not completely attributed to reduced earnings manipulations in firms with analyst cash flow forecasts (as suggested in previous literature), but is at least partially attributed to the improved ability of investors to price earnings manipulation. Finally, we consider three possible explanations for the improved ability of investors to see through earnings manipulations for firms with cash flow forecasts: investor attention, the accuracy of analyst forecasts, and voluntary disclosures. We find that both investor attention and the accuracy of analyst earnings forecasts help to improve investors' ability to correctly price earnings manipulations. Our work takes the field one step closer to a better understanding of the relation between analyst cash flow forecasts and the accrual anomaly. 18 Our empirical evidence confirms the bias mentioned in Chuk et al. (2013): over our sample period of 1993–2009, management cash flow forecasts (CASHEPS and FFO) account for less than 4% of observations in CIG, and we have only 19 observations with management cash flow forecast (CASHEPS and FFO) out of our sample of 6052 in Table 6. Given the data limitation, we choose not to use voluntary cash flow forecasts from CIG as a valid proxy in our test.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

10

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx

Appendix A. Definition of variables DUMMYC SARt + 1

Average total assets Earnings Cash flows Accruals Size Beta B/M E/P DACC_DD

DACC_ALS

CONTROL Magnitude of accruals Accounting choice heterogeneity

Capital intensity Earnings volatility Altman's Z-score

Size Idiosyncratic volatility Price Volume EXP PERCFF AFEE AFECF

= Dummy variable, which takes the value of one if analyst cash flow forecast is available in the month of current fiscal year ends, and zero otherwise.
= ∏ ð1 þ r is Þ−∏ 1 þ r ps , where ris and rps are returns in month s for firm i and size portfolio p, respectively. Size deciles are determined by the s s distribution of market values of all the AMEX/NYSE/NASDAQ firms at the beginning of the calendar year. SARt + 1 is computed over the 12-month holding period, beginning four months after the current fiscal year ends. When a firm delists, we use the delisting return in the delisting month and assume a return equal to the firm's size-matched portfolio for the remainder of the year. If a firm's delisting is due to liquidation or a forced delisting and the delisting return is missing, the delisting return is set to −30%, following Shumway (1997). = Average value of Total Assets (AT) at the beginning and end of current fiscal year. = Income Before Extraordinary Items (OANCF) / Average total assets. = [Cash Flows from Operating Activities (OANCF) − Extraordinary Items & Discontinued Operations (XIDOC)] / Average total assets. = Earnings − Cash flows = [Income Before Extraordinary Items (IBC) − Cash Flows from Operating Activities (OANCF) + Extraordinary Items & Discontinued Operations (XIDOC)] / Average total assets. = The natural log of market value of equity in millions of dollars, where market value of equity = Price-Close-Annual-Fiscal (PRCC_F) ∗ Common Shares Outstanding (CSHO) at the end of current fiscal year. = CAPM beta estimated using 24 monthly return observations ending one month prior to the accrual portfolio formation date (i.e., the third month after current fiscal year end). = Book value of common equity (CEQ) / Market value of common equity (PRCC_F ∗ CSHO) at the end of current fiscal year. = Operating income after depreciation (OIADP) / Market value of common equity (PRCC_F ∗ CSHO) at the end of current fiscal year. = Discretionary accruals estimated from the Dechow and Dichev (2002) Model: Accrualst = β1 ∗ [1 / TAt] + β2 ∗ CFOt − 1 / TAt + β3 ∗ CFOt / TAt + β3 ∗ CFOt + 1 / TAt + εt; where Accrualst is defined as above; CFOt − 1, CFOt and CFOt + 1 is net cash flow from operating activities (OANCF) minus extraordinary items and discontinued operations (XIDOC) for the past, current, and future year; and TAt is average total assets in year t. = Discretionary accruals estimated from the Allen et al. (2013) Model: Accrualst = β1 ∗ [1 / TAt] + β2 ∗ CFOt − 1 / TAt + β3 ∗ CFOt/ TAt + β4 ∗ CFOt + 1 / TAt + β5 ∗ SGRt + β6 ∗ EMPGRt + εt; where Accrualst is defined as above; CFOt − 1, CFOt and CFOt + 1 is net cash flow from operating activities (OANCF) minus extraordinary items and discontinued operations (XIDOC) for the past, current, and future year; and TAt is average total assets in year t; SGRt is the percentage change of sales (SALE); and EMPGRt is percentage change of the number of employees (EMP). = |Accruals| defined as above, at the beginning of current fiscal year. = An index ranging from zero to one that captures the comparability of a firm's accounting choice with its industry peers, at the beginning of current fiscal year. The index is computed by assigning a value of one to each firm which accounting choice differs from the most frequently chosen method in that firm's industry group, for each of the following five accounting choices: (1) inventory valuation; (2) investment tax credit; (3) depreciation; (4) successful efforts vs. full cost for companies with extraction activities; and (5) purchase vs. pooling. If a firm has no information or a missing value for a given accounting choice, the choice is coded as zero (consistent with the firm selecting the most common accounting choice in the industry). The score for each firm is summed and then scaled by the number of accounting choices in the industry: 5 for firms in the petroleum and natural gas industry (because they are eligible for all 5 choices); 3 for firms in banking, insurance, real estate, and trading industries (because they have no inventory choice and are not extractive industries); and 4 for firms in all other industries (because they are not extractive industries). = Gross property, plant and equipment (PPEGT)/sales (SALE), at the beginning of current fiscal year. = The coefficient of variation of EPS measured over the past four years, calculated as standard deviation of EPS / |mean of EPS|, where EPS is Basic Earnings Per Share excluding Extraordinary Items (EPSPX) scaled by Price-Close-Annual-Fiscal (PRCC_F) for the prior fiscal year. = Altman's Z-score measured in previous fiscal year. Following Altman (1968), the Z score equals 1.2 ∗ [Net working capital (WCAP) / Total assets (AT)] + 1.4 ∗ [Retained earnings (RE) / Total assets (AT)] + 3.3 ∗ [Earnings before interest and taxes (EBIT) / Total assets (AT)] + 0.6 ∗ [Market value of equity (PRCC_F ∗ CSHO) / Book value of liabilities (LT)] + 1.0 ∗ [Sales (SALE) / Total assets (AT)]. Lower Altman's Z-scores indicate poorer financial health. = The natural log of market value of equity in millions of dollars, where market value of equity = Price-Close-Annual-Fiscal (PRCC_F) ∗ Common Shares Outstanding (CSHO) at the end of current fiscal year. = The residual variance from a regression of firm-specific returns on the returns of the CRSP equally weighted market index over the 12 months ending one month prior to the accrual portfolio formation date (i.e., the third month after current fiscal year end). = The CRSP closing stock price one month before the accrual portfolio formation date (i.e., the third month after current fiscal year end). = The CRSP daily closing price times CRSP daily shares traded, averaged over the year ending one month prior to the accrual portfolio formation date (i.e., the third month after current fiscal year end). = The proportion of analysts following the firm also issuing cash flow forecasts, equal to the number of analyst cash flow forecasts relative to the number of analyst earnings forecasts for each company. = Analyst Forecast Error in Earnings = |Median of earnings forecast for year t + 1 four months after the current fiscal year end − Actual earnings for year t + 1| / Closing price as of the end of the third month after current fiscal year end. = Analyst Forecast Error in Cash Flows = |Median of cash flow forecasts for year t + 1 four months after the current fiscal year end − Actual cash flows for year t + 1| / Closing price as of the end of the third month after current fiscal year end.

References Allen, E. J., Larson, A.R., & Sloan, R. G. (2013). Accrual reversals, earnings and stock returns. Journal of Accounting and Economics, 56(1), 113–129. Alford, A. W., Jones, J. J., & Zmijewski, M. E. (1994). Extensions and violations of the statutory SEC form 10-K filing requirements. Journal of Accounting and Economics, 17(2), 229–254. Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. Journal of Finance, 23(4), 589–609. Bradshaw, M. T., Richardson, S. A., & Sloan, R. G. (2001). Do analysts and auditors use information in accruals? Journal of Accounting Research, 39(1), 45–74. Call, A.C. (2008). Analysts' cash flow forecasts and the predictive ability and pricing of operating cash flows. Working paper. : University of Georgia. Call, A.C., Chen, S., & Tong, Y. H. (2009). Are analysts' earnings forecasts more accurate when accompanied by cash flow forecasts? Review of Accounting Studies, 14(2–3), 358–391. Call, A.C., Chen, S., & Tong, Y. H. (2013). Are analysts' cash flow forecasts naïve extensions of their own earnings forecasts? Contemporary Accounting Research, 30(2), 438–465.

Chuk, E., Matsumoto, D., & Miller, G. (2013). Assessing methods of identifying management forecast: CIG vs. researcher collected. Journal of Accounting and Economics, 55, 23–42. Collins, D. W., & Hribar, P. (2000). Earnings-based and accrual-based market anomalies: One effect or two? Journal of Accounting and Economics, 29(1), 101–123. Das, S., Guo, R., & Zhang, H. (2006). Analysts' selective coverage and subsequent performance of newly public firms. Journal of Finance, 61(3), 1159–1185. 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.M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The Accounting Review, 70(2), 193–225. DeFond, M. L., & Hung, M. (2003). An empirical analysis of analysts' cash flow forecasts. Journal of Accounting and Economics, 35(1), 73–100. DeFond, M. L., & Hung, M. (2007). Investor protection and analysts' cash flow forecasts around the world. Review of Accounting Studies, 12(2-3), 377–419. Fama, E. F., & French, K. (1992). The cross-section of expected stock returns. Journal of Finance, 47(2), 427–465.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006

L. Shi et al. / Advances in Accounting, incorporating Advances in International Accounting xxx (2014) xxx–xxx Fama, E. F., & French, K. (1995). Size and book-to-market factors in earnings and returns. Journal of Finance, 50(1), 131–155. Fama, E. F., & French, K. (1996). Multifactor explanations of asset pricing anomalies. Journal of Finance, 51(1), 55–84. Fama, E. F., & French, K. (1997). Industry costs of equity. Journal of Financial Economics, 43(2), 153–193. Chapter in edited book: Fiske, S. T. (1995). Social cognition. In A. Tesser (Ed.), Advanced social psychology (pp. 149–194). New York: McGraw-Hill. Givoly, D., Hayn, C., & Lehavy, R. (2009). The quality of analysts' cash flow forecasts. The Accounting Review, 84(6), 1877–1911. Green, J., Hand, J. R. M., & Soliman, M. T. (2011). Going, going, gone? The apparent demise of the accruals anomaly. Management Science, 57(5), 797–816. Gutierrez, R., Jr., & Kelly, R. K. (2008). The long-lasting momentum in weekly returns. Journal of Finance, 63(1), 415–447. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47(1), 153–161. Hirshleifer, D., & Teoh, S. H. (2003). Limited attention, information disclosure, and financial reporting. Journal of Accounting and Economics, 36(1–3), 337–386. Hirshleifer, D., Hou, K., Teoh, S. H., & Zhang, Y. (2004). Do investors overvalue firms with bloated balance sheet? Journal of Accounting and Economics, 38(1), 297–331. Hirshleifer, D., Lim, S. S., & Teoh, S. H. (2011). Limited investor attention and stock market misreactions to accounting information. Review of Asset Pricing Studies, 1(1), 35–73. Hribar, P., & Collins, D. W. (2002). Errors in estimating accruals: Implications for empirical research. Journal of Accounting Research, 40(1), 105–134. Jones, J. J. (1991). Earnings management during import relief investigations. Journal of Accounting Research, 29(2), 193–228. Kraft, A., Leone, A. J., & Wasley, C. (2006). An analysis of the theories and explanations offered for the mispricing of accruals and accrual components. Journal of Accounting Research, 44(2), 297–339. Kothari, S. P., Leone, A. J., & Wasley, C. E. (2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39(1), 163–197. Lakonishok, J., Shleifer, A., & Vishny, R. W. (1994). Contrarian investment, extrapolation, and risk. Journal of Finance, 49(5), 1541–1578. Lennox, C. S., Francis, J., & Wang, Z. (2012). Selection models in accounting research. The Accounting Review, 87(2), 589–616.

11

Lev, B., & Nissim, D. (2006). The persistence of the accruals anomaly. Contemporary Accounting Research, 23(1), 193–226. Levi, S. (2008). Voluntary disclosure of accruals in earnings press releases and the pricing of accruals. Review of Accounting Studies, 13(1), 1–21. Louis, H., Robinson, D., & Sbaraglia, A. (2008). An integrated analysis of the association between accrual disclosure and the abnormal accrual anomaly. Review of Accounting Studies, 13(1), 23–54. Mashruwala, C., Rajgopal, S., & Shevlin, T. (2006). Why is the accrual anomaly not arbitraged away? The role of idiosyncratic risk and transaction costs. Journal of Accounting and Economics, 42(1–2), 3–33. McInnis, J., & Collins, D. W. (2011). The effect of cash flow forecasts on accrual quality and benchmark beating. Journal of Accounting and Economics, 51(3), 219–239. McNichols, M., & O'Brien, P. C. (1997). Self-selection and analyst coverage. Journal of Accounting Research, 35(supplement), 167–199. Mohanram, P.S. (2014). Analysts' cash flow forecasts and the decline of the accruals anomaly. Contemporary Accounting Research (in press). Peng, L., & Xiong, W. (2006). Investor attention, overconfidence and category learning. Journal of Financial Economics, 80(3), 563–602. Pincus, M., Rajgopal, S., & Venkatachalam, M. (2007). The accrual anomaly: International evidence. The Accounting Review, 82(1), 169–203. Richardson, S., Tuna, I., & Wysocki, P. (2010). Accounting anomalies and fundamental analysis: A review of recent research advances. Journal of Accounting and Economics, 50(2–3), 410–454. Shi, L., & Zhang, H. (2012). Can the earnings fixation hypothesis explain the accrual anomaly? Review of Accounting Studies, 17(1), 1–21. Shumway, T. (1997). The delisting bias in CRSP data. Journal of Finance, 52(1), 327–340. Sloan, R. G. (1996). Do stock prices fully reflect information in accruals and cash flows about future earnings? The Accounting Review, 71(3), 289–315. Book: Wooldridge, J. M. (2006). Econometric analysis of cross section and panel data. Cambridge, Massachusetts: The MIT Press. Xie, H. (2001). The mispricing of abnormal accruals. The Accounting Review, 76(3), 357–373. Zhang, X. F. (2007). Accruals, investment, and the accrual anomaly. The Accounting Review, 82(5), 1333–1363.

Please cite this article as: Shi, L., et al., Analyst cash flow forecasts and pricing of accruals, Advances in Accounting, incorporating Advances in International Accounting (2014), http://dx.doi.org/10.1016/j.adiac.2014.04.006