Analysts’ Recommendation Revisions and Subsequent Earnings Surprises: Pre- and Post-Regulation FD Journal of Accounting, Auditing & Finance 26(3) 475–501 Ó The Author(s) 2011 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0148558X11401556 http://jaaf.sagepub.com Dan Palmon1 and Ari Yezegel2 Abstract This study examines the extent to which analyst recommendations were useful in identifying earnings surprises during the pre- and post-Regulation Fair Disclosure (FD) periods A comparative analysis of the association between recommendation revisions and subsequent earnings surprises suggests a significant decline in the predictive value of analysts’ recommendations after Regulation FD took effect Recommendation revisions are roughly 55% less useful in predicting earnings surprises in the post-Regulation FD period Furthermore, the average abnormal return earned by investors following analysts’ advice to exploit earnings surprises is approximately 70% lower in the post-Regulation FD period Overall, this article’s findings are consistent with Regulation FD having considerably reduced analysts’ comparative advantage in identifying earnings surprises Keyword Regulation FD, analyst recommendations, earnings surprises, portfolio analysis Introduction Earnings-related selective disclosure was one of the most publicized cases of unfair disclosure that contributed to Regulation Fair Disclosure’s (FD) acceptance.1 The Securities and Exchange Commission (SEC) received several thousand comment letters expressing frustration on the basis of the belief that corporations were giving earnings-related information only to a select group of financial analysts and institutional investors Relying on this information, analysts then made recommendations to their clients prior to earnings announcements, thus giving them an unfair advantage over other investors Inevitably, such disclosure policies helped certain selected investors earn profits or avoid losses at the expense of other investors In response, the SEC passed Regulation FD and listed earningsrelated disclosure on top of the list of potential material information that needs to be disclosed simultaneously to all market participants Rutgers Business School, NJ, USA Bentley University, Waltham, MA, USA Corresponding Author: Ari Yezegel, Bentley University, 175 Forest Street, Waltham, MA 02452 Email: ayezegel@bentley.edu Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 476 Journal of Accounting, Auditing & Finance 26(3) This study examines the extent to which analysts’ recommendations helped their clients identify firms with earnings surprises before and after Regulation FD took effect.2 If analysts, as alleged, were privately communicating earnings-related information with managers and providing consistent investment advice to their clients, then analysts’ recommendations should have exhibited predictive power of upcoming earnings surprises in the preRegulation FD period Furthermore, to the extent that Regulation FD was effective in curtailing selective disclosure, the predictive value of recommendations should have declined after Regulation FD took effect We estimate recommendations’ predictive value of upcoming earnings surprises using the association between recommendation revisions and unexpected earnings calculated based on (a) time-series earnings expectations, (b) analysts’ earnings expectations, and (c) earnings announcement returns In addition, we examine the association between earnings surprises and recommendation revisions using regression analysis controlling for postearnings announcement drift, return momentum, accruals anomaly, and institutional trading Finally, we construct a trading strategy designed to capture recommendation revisions’ predictive power of earnings surprises and compare the abnormal returns accrued by this portfolio during the pre- and post-Regulation FD periods We find that prior to Regulation FD’s acceptance, upgraded firms exhibited 3-day market-adjusted earnings announcement returns that were on average 0.93% higher than downgraded firms After Regulation FD took effect, the return differential between upgraded and downgraded firms declined 55% to 0.43% The regression analysis provides similar results and supports the inference that the association between recommendation revisions and subsequent earnings surprises declined after Regulation FD took effect Finally, the trading strategy analysis reveals an approximately 70% decline in the portfolio performance after Regulation FD took effect Overall, our results are consistent with Regulation FD having been effective in limiting selective disclosure and reducing analyst recommendations’ predictive power of upcoming earnings surprises Regulation FD was preceded with intense objection that the rule would harm the level of corporate disclosure Consistently, prior studies on Regulation FD focused primarily on whether the rule damaged corporate disclosure level, increased earnings volatility, and reduced forecast accuracy The literature suggests that Regulation FD did not have significant adverse effects on corporate disclosure This article contributes to the extant literature by comparing the extent to which recommendations were valuable to analysts’ clients in identifying earnings surprises before Regulation FD and how much of this predictive value was eliminated after Regulation FD took effect We also examine the impact of Regulation FD on the abnormal performance of investors who followed analysts’ recommendation revisions with the intent of benefiting from analysts’ earnings-related private information Hence, our analysis provides important insights relating to the fundamental concern of certain investors earning abnormal profits at the expense of other investors based on selective disclosure Literature Review and Hypotheses Development In response to growing concerns of select individuals obtaining access to inside information, the SEC passed Regulation FD, which was concerned with the fair disclosure of nonpublic material information Regulation FD required managers to disseminate any material information simultaneously to all market participants and prohibited selective disclosure Many securities markets professionals and institutional investors argued that bringing Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Palmon and Yezegel 477 additional restrictions on corporate disclosure would reduce the quantity and quality of information available to capital markets After Regulation FD took effect, there was increased concern by practitioners that Regulation FD increased return volatility, adversely affected analysts’ earnings forecasts, reduced corporate disclosure, and increased information asymmetry However, academic research investigating Regulation FD’s impact provided mixed inferences Table presents studies that investigated Regulation FD’s impact and summarizes their findings On the return volatility aspect, while Heflin, Subramanyam, and Zhang (2003), Eleswarapu, Thompson, and Venkataraman (2004), and Sinha and Gadarowski (2010) found a significant decline after Regulation FD, Bailey, Li, Mao, and Zhong (2003) and Francis, Nanda, and Wang (2006) found no change in volatility after Regulation FD took effect Analogous to studies on return volatility, no consensus was reached on whether analysts’ forecasts suffered or improved after Regulation FD Heflin et al., Bailey et al., and Francis et al found no significant impact of Regulation FD, whereas Agrawal, Chadha, and Chen (2006) and Mohanram and Sunder (2006) documented deterioration Similarly, Heflin et al and Francis et al reported no change in forecast dispersion after Regulation FD, whereas Bailey et al., Irani and Karamanou (2003), and Agrawal et al found forecast dispersion to have increased Results on corporate disclosure were also mixed Heflin et al and Bailey et al documented an increase in conference call frequency, and Irani (2004) found that conference calls became more useful in helping analysts increase forecast accuracy In contrast, Bushee, Matsumoto, and Miller (2004) found that corporate conference call policy did not change after Regulation FD Studies examining Regulation FD’s impact on information asymmetry reached inferences ranging from an increase to a decrease in information asymmetry after Regulation FD took effect On one hand, Eleswarapu et al (2004), Chiyachantana, Jiang, Taechapiroontong, and Wood (2004), and Ahmed and Schneible (2007) documented evidence consistent with an improvement after Regulation FD On the other hand, Sidhu, Smith, Whaley, and Willis (2008) and Gomes, Gorton, and Madureira (2007) reported evidence suggesting an increase in information asymmetry Finally, Charoenrook and Lewis (2009) and Collver (2007) found no change in information asymmetry Another strand of literature examined the informativeness of analyst reports before and after Regulation FD took effect Gintschel and Markov (2004) studied the value of analysts’ information outputs using the return volatility surrounding analysts’ announcements They found that the absolute price impact of financial analysts’ forecasts and recommendations declined by 28% after Regulation FD consistent with Regulation FD having curtailed selective disclosure Cornett, Tehranian, and Yalcin (2007) provided further evidence by evaluating the impact of Regulation FD on affiliated versus unaffiliated analysts Their results suggested that the market reaction to affiliated analysts’ recommendation changes decreased significantly after the passage of Regulation FD Francis et al (2006) provided supporting evidence to Gintschel and Markov’s (2004) results using American Depositary Receipt (ADR) firms to control for confounding events The prior literature provides extensive evidence on the effect of Regulation FD on the preearnings announcement informational efficiency, analysts’ forecast accuracy and dispersion, the informativeness of analysts’ reports, and corporate disclosure However, little is known about Regulation FD’s impact on the usefulness of analysts’ advice in identifying earnings surprises The unfair disclosure of information causing some investors to take positions prior to earnings announcements with advance knowledge of the outcome was one of the central issues in the debate surrounding Regulation FD We contribute to the Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 478 Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 1995Q4–2001Q3 Irani and Karamanou (2003) Bushee, Matsumoto, and Miller (2004) 10/1999–7/2001 10/1999–10/2001 1998–2001 3/1995–6/2004 Chiyachantana, Jiang, Taechapiroontong, and Wood (2004) Gintschel and Markov (2004) Irani (2004) Agrawal, Chadha, and Chen (2006) 3/1999–10/2001 1999Q4–2001Q2 1999Q4–2001Q2 Sample period Heflin, Subramanyam, and Zhang (2003) Bailey, Li, Mao, and Zhong (2003) Study Table Review of Prior Studies - Conference call relevance measured as the call’s ability to improve forecast and consensus accuracy - Forecast accuracy - Forecast dispersion - Return volatility surrounding earnings forecasts and recommendations - Abnormal return volatility and trading volume - Absolute consensus and time-series forecast error - Analyst information advantage - Forecast dispersion - Market reaction to earnings announcements - Forecast accuracy - Forecast dispersion - Voluntary disclosure - Analyst following - Forecast dispersion - Timing of conference calls - Number of conference calls - Price volatility and trading volume - Liquidity - Adverse selection component - Retail and institutional trades Variables of interest - After Regulation FD, forecast accuracy decreased and dispersion increased particularly for early forecasts (continued) - Decrease in analyst following - Increase in analyst forecast dispersion - No substantial impact on conference call policies - The information content of conference calls does not appear to have declined - Increase in liquidity - Decrease in information asymmetry - Decrease in preannouncement period institutional trading and increase in retail trading activity after earnings announcements - Return volatility associated with analyst announcements is on average 28% lower in the in the post-Regulation FD period - The difference in price impact between ‘‘The Leaders’’ and other brokerage houses is 65% smaller in the post-Regulation FD period - Relevance of conference calls to analyst forecasts increased in the post-Regulation FD period - No change in return volatility around earnings announcements - No change in forecast accuracy and increase in forecast dispersion - No evidence that Regulation FD prevents or enhances information leakage during the preannouncement period - Increase in voluntary disclosure - No significant change in forecast accuracy or dispersion - Some improvement in stock price efficiency - Substantial increase in voluntary disclosure frequency Main findings 479 Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 10/1998–11/2002 1997–2002 1999Q4–2001Q2 Cornett, Tehranian, and Yalcin (2007) Gomes, Gorton, and Madureira (2007) Ahmed and Schneible (2007) Charoenrook and Lewis (2009) - Analyst coverage - Forecast error - Return volatility on earnings announcements - Cost of capital - Trading volume - Earnings announcement return Analysts’ responsiveness to: - Earnings guidances - Press releases - Earnings announcements - Hasbrouck summary informativeness statistic - Market reaction to analysts’ recommendation changes Variables of interest - No change in the aggregate amount of firm-specific information conveyed by public disclosure events - Reduction in information quality differences among investors - No substantial change in informed trading that can be solely attributed to Regulation FD - Return volatility surrounding analysts’ recommendation changes decreased significantly after the passage of Regulation FD - Investors’ differential reaction to downgrades by affiliated and unaffiliated analysts declined; no significant change was evident for upgrades - Increase in forecast error and earnings announcement return volatility particularly for small firms Main findings Note: This table provides a summary of the prior studies that investigate the impact of Regulation FD The sample period, variables of interest, and main findings are presented for each study 2000–2002 8/1999–1/2002 Sample period Collver (2007) Study Table continued 480 Journal of Accounting, Auditing & Finance 26(3) literature by examining the extent to which analysts’ recommendation revisions were useful in predicting earnings surprises in the pre- and post-Regulation FD periods and estimating the level of abnormal returns that investors could have earned by following analysts’ recommendations preceding earnings announcements Analysts disclose forecasts and recommendations in their reports to their clients Forecasts represent analysts’ predictions of various financial statement line items (e.g earnings, sales) and not necessarily convey information about analysts’ assessments of companies’ intrinsic values relative to their stock prices (e.g., overvalued or undervalued) Conversely, recommendation ratings represent direct indication of analysts’ assessment of the valuation of the company Consistently, investors react more strongly to recommendation revisions than they to earnings forecast revisions.3 Hence, analysts who intend to give early warnings to their clients about earnings announcements are more likely to communicate this through their recommendation ratings rather than their earnings forecasts Furthermore, analysts can more effectively communicate information they received through selective disclosure via recommendation ratings rather than earnings forecasts because they may not have received a precise forecast from management Many analysts argued against Regulation FD on the basis that they only receive ‘‘soft’’ information from managers, which could not be communicated to the public in a form other than selective disclosure Managers are likely to be reluctant to give precise figures about upcoming earnings to analysts via selective disclosure because they themselves may not have precise knowledge of upcoming earnings at the time Anecdotal evidence suggests that managers often limited their communications to general directional guidance such as ‘‘earnings are likely to be better than expected’’ or ‘‘the current earnings expectations are unrealistic.’’4 Analysts, when not provided a point-forecast from managers about upcoming earnings, are likely to limit their earnings forecast revisions In contrast, through recommendations, analysts can signal upcoming negative or positive earnings news without giving precise information about the degree of the earnings surprise If analysts, as alleged, were privately communicating earnings-related information with managers and providing consistent investment advice to their clients, then analysts’ recommendations should have possessed predictive power of upcoming earnings surprises during the pre-Regulation FD period Furthermore, to the extent that Regulation FD was successful in reducing selective disclosure, the association between recommendation revisions and earnings surprises should have weakened in the post-Regulation FD period Hypothesis (H1): The association between changes in analysts’ recommendations and subsequent earnings surprises declined after Regulation FD took effect Supporters of Regulation FD argued that selective disclosure helped select analysts and their clients reap economic benefits at the expense of other investors If, as alleged, analysts guided their clients to earn profits based on the private earnings guidance they received from management, we should observe significant abnormal returns associated with implementing a trading strategy that follows analysts’ recommendations and liquidates after earnings announcements Furthermore, to the extent that Regulation FD limited earnings-related selective disclosure, we should observe a reduction in the abnormal performance of the trading strategy after Regulation FD took effect Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Palmon and Yezegel 481 Hypothesis (H2): The profitability of a trading strategy intended to capture the private earnings-related information conveyed by recommendation revisions declined after Regulation FD took effect Research Design The period before an earnings announcement corresponds to a time in which firms are likely to have prepared financial statements and managers have the greatest knowledge of the current quarter’s earnings If firm executives selectively disclose earnings-related information to analysts, then this information is most likely to be privately communicated to analysts during the period before earnings announcements To isolate recommendations that may be associated with analysts’ communications with managers about upcoming earnings, we limit our sample to analysts’ recommendation revisions made during the 3-week period ending days before the earnings announcement.5 We then examine Regulation FD’s impact on these recommendations’ predictive value of upcoming earnings surprises by estimating and comparing the association between recommendation revisions and subsequent earnings surprises during the pre- and post-Regulation FD periods using univariate and regression analysis.6 Figure provides an illustration of the timeline and the main tests of this study Univariate Analysis The univariate analysis examines the mean earnings surprise that follows upgrades and downgrades and tests whether a significant change is evident after Regulation FD took effect For robustness, earnings surprise is computed using four alternative methods: (a) standardized unexpected earnings based on time-series expectations, (b) standardized unexpected earnings based on analyst expectations, (c) 3-day earnings announcement abnormal returns, and (d) 2-day earnings announcement abnormal returns Standardized unexpected earning based on time-series expectation is computed as follows: SUEit ¼ et À etÀ4 st;tÀ8 where et is firm i’s earning (excluding extraordinary items) for quarter t and st,t–8 is the standard deviation of unexpected earnings (et et–4) in the past eight fiscal quarters SUE deciles are constructed and transformed to range between 20.5 and 10.5 The construction of SUE deciles controls for common marketwide effects and reduces the influence of extreme values on the results Standardized unexpected earning based on analysts’ expectations is computed as follows: ASUEit ¼ et À e^t ; pt where et is firm i’s actual earnings for quarter t, e^t is the earnings forecast consensus defined as the median of all analysts’ latest earnings forecasts made after the previous quarter’s earnings announcement, and pt is the firm’s stock price at the end of quarter t ASUE deciles are constructed and transformed to range between 20.5 and 10.5 Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 482 Journal of Accounting, Auditing & Finance 26(3) Panel A Fiscal Fiscal Quarter Quarter Begins Begins Fiscal Fiscal Quarter Quarter Ends Ends Firm generates revenues Firm Firmgenerates generatesrevenues revenues and incurs expenses and andincurs incursexpenses expenses Management obtains rough Management Managementobtains obtainsaaarough rough idea of of upcoming upcoming earnings earnings idea idea of upcoming earnings Potential period period for earningsearningsPotential Potential periodfor for earningsrelated selective selective disclosure disclosure related related selective disclosure Analysts Analystsreceiving receivingearningsearningsrelated selective disclosure related selective disclosure revise recommendations revise recommendations (–22,–2) (-22,-2) Upgrades and Downgrades Upgrades & Downgrades Earnings Earnings are are reported reported Earnings Earnings Surprise Surprise 1.1.CAR CAR 2.2.SUE SUE 3.3.UE UE Panel B Earnings Surprise Pre-Regulation FD Upgrade Upgrade Association Return Difference CAR (0,1) CAR (–1,1) Unexpected Earnings Difference (Time-Series) Unexpected Earnings Difference (Analyst) Downgrade Downgrade Post-Regulation FD Earnings Surprise Upgrade Upgrade Association Downgrade Downgrade Return Difference CAR (0,1) CAR (– 1,1) Unexpected Earnings Difference (Time-Series) Unexpected Earnings Difference (Analyst) Figure Research design Panel A: Timeline of the empirical analysis Panel B: Association test for the pre- and post-Regulation periods Note: This figure provides a visual overview of the research design Panel A illustrates which analyst recommendation revisions are included in the analysis in relation to the firm’s quarterly cycle Panel B shows the main test of the article and lists which earnings surprise measures are used Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Palmon and Yezegel 483 Earnings announcement abnormal returns are computed as follows: CARit ¼ X ðRmit À Mktm Þ; m¼À1 where m is equal to on firm i’s quarter t earnings announcement date Rm,i,t is firm i’s daily return on the mth day of quarter t’s earnings announcement, and Mktm is the Center for Research in Security Prices (CRSP) New York Stock Exchange (NYSE)/American Stock Exchange (AMEX)/NASDAQ value-weighted daily return for day m.7 For each quarter, we compute the mean earnings surprise for upgraded and downgraded firms As recommendation revisions can be driven by marketwide information that affects numerous firms, earnings surprises that follow revisions may be correlated across firms Therefore, we follow the Fama and Macbeth (1973) procedure and first compute average quarterly earnings surprises and then calculate time-series averages and t-statistics of quarterly average surprises Regression Analysis The regression analysis allows us to examine the change in recommendations’ predictive value while controlling for confounding factors To measure the change in the association between recommendation revisions and earnings surprises, we estimate the following regression models: SSUEit ¼ a þ b1 REVit þ b2 FDit þ b3 REVit FDit þ b4 SOXit þ b5 REVit SOXit þ b6 LANCRETit þ b7 LRETit þ b8 ACCRit þ b9 CHNG IOit þ eit : SASUEit ¼ a þ b1 REVit þ b2 FDit þ b3 REVit FDit þ b4 SOXit þ b5 REVit SOXit þ b6 LANCRETit þ b7 LRETit ð1Þ ð2Þ þ b8 ACCRit þ b9 CHNG IOit þ eit : CARðÀ1; þ1Þit ¼ a þ b1 REVit þ b2 FDit þ b3 REVit FDit þ b4 SOXit þ b5 REVit SOXit þ b6 LANCRETit þ b7 LRETit þ b8 ACCRit þ b9 CHNG IOit þ eit : ð3Þ CARð0; þ1Þit ¼ a þ b1 REVit þ b2 FDit þ b3 REVit FDit þ b4 SOXit þ b5 REVit SOXit þ b6 LANCRETit þ b7 LRETit þ b8 ACCRit ð4Þ þ b9 CHNG IOit þ eit : In the regression analyses above, we regress alternative earnings surprise measures on the recommendation revisions made during the 3-week period before earnings announcements (REV), a Regulation FD indicator variable that takes a value of for calendar quarters during the post-Regulation FD period (FD) and the interaction of REV and FD variables (REV FD) The coefficient of the REV variable tests whether recommendation Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 606 Journal of Accounting, Auditing & Finance 26(3) ownership structure, characteristics of the issuer, characteristics of the issued bonds, and market conditions Thus, MO has an incremental explanatory power for the cost of debt when the other ownership structure and control variables are given The result suggests that prospective bondholders interpret an increase in MO as an increase in the conflict of interest between bondholders and shareholders This supports our first hypothesis Furthermore, we find that the coefficient on FSTABLE is significantly negative at less than the 01 level The result is consistent with Hypothesis 3—stable shareholdings by financial institutions have a favorable impact on the cost of debt through efficient monitoring.14 This finding suggests that firms facing stronger external and effective monitoring by financial institutions are rewarded with lower yield spreads The stable shareholdings by financial institutions could prevent firm managers from engaging in opportunistic behaviors due to efficient monitoring, thereby decreasing the cost of debt However, the coefficient on CROSS is not significant, which suggests that crossshareholdings have no impact on the cost of debt in the presence of other ownerships and control variables Therefore, Hypothesis is not supported It should be noted that stable shareholdings by financial institutions and cross-shareholdingss have contradictory predictions The results suggest that financial institutions have a stronger incentive to monitor opportunistic managerial behaviors that increase the ACD than cross-shareholders With respect to control variables, they have their expected signs, except for INCR and MATUR, and most variables are statistically significant at conventional levels ACD, MO, and Interest Rate Spread Table shows the regression result of Model to test Hypothesis To support Hypothesis 2, we expect the coefficient of MO ACD to be positive in the model In Model 2, the coefficient of MO ACD is 4.432 and significantly positive at the less than 01 level, as hypothesized We also find that the coefficient of MO is no longer positive, which suggests that in contrast to the results in Table 3, MO has no positive effect on the interest rate spread after controlling for the effect of ACD on the relationship between MO and SPREAD The result indicates that MO has a stronger effect on interest rate spread when the ACD at the time of corporate bond issue is larger, considering that the ACD at the time of bond issue is critically important for explaining the relationship between MO and the cost of debt This finding is consistent with Hypothesis We observe that our control variables have their expected signs, except for MATUR, and that most of the variables are statistically significant at conventional levels Overall, the evidence from the ‘‘Main Results’’ suggests that prospective bondholders use MO information to anticipate a firm’s future ACD and default risk, and then they incorporate this prediction in the pricing of new corporate bond issues Furthermore, bond investors are likely to estimate a firm’s future ACD and default risk higher when managers have already engaged in an action that transfers wealth from the bondholders to the shareholders or when managers have a wide range of options for this purpose at the time of bond issue Additional Analyses The Nonlinearity of MO and the Cost of Debt As discussed in the section titled ‘‘Hypotheses Development,’’ some prior studies suggest the possibility that MO is nonmonotonically related to the cost of debt (Bagnani et al., Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Shuto and Kitagawa 607 Table Result of the Differential Effect of the Agency Cost of Debt on the Relationship Between Managerial Ownership and Bond Spreads Dependent variable = SPREAD Model Independent variable Constant MO MO ACD CROSS FSTABLE MARGIN DER INCR LOAN LNASIZE BSIZE MATUR BCFIRM RISKP Adjusted R2 n Expected sign + + +/2 2 + + 2 + + + Coefficient t-value 2.892*** 21.028 4.432*** 0.096 20.155* 21.660*** 0.025*** 20.001 0.823*** 20.093*** 20.053** 20.024*** 0.015 0.468*** 0.617 589 6.783 21.200 3.529 0.738 21.855 24.061 4.023 20.580 3.488 25.887 22.398 27.006 0.404 7.985 Note: SPREAD = the interest rate spread on the first straight bond issued of the fiscal year t—the spread is the difference between the interest rate on the bond issued by the firm and that on government bonds; MO = fraction of the shares owned by directors at the end of fiscal year t – 1; ACD = the agency cost of debt, computed using factor analysis based on six financial variables at the end of fiscal year t – 1: (1) R&D expenditures / sales, (2) (fixed assets / total assets), (3) cash and marketable securities / total assets, (4) common dividends / total assets, (5) the standard deviation of ROA (net income / total assets) for the past years, (6) the standard deviation of leverage (total debt / total assets) for the past years; CROSS = fraction of the shares owned by cross-shareholders at the end of fiscal year t – 1; FSTABLE = fraction of the stable shareholdings by financial institutions at the end of fiscal year t – 1; MARGIN = the operating income divided by net sales at the end of fiscal year t – 1; DER = the debt equity ratio at the end of fiscal year t – 1; INCR = the interest coverage ratio at the end of fiscal year t – 1; LOAN = the bank loan divided by total assets at the end of fiscal year t – 1; LNASIZE = the natural log of the total assets at the end of fiscal year t – 1; BSIZE = the log of the issue size; MATUR = the years to maturity; BCFIRM = an indicator variable that takes the value of if a bond management company is established, and otherwise; RISKP = the risk premium: the average values of SPREAD on Rating and Investment Information Inc.’s A bonds for the month of issue For details of ownership variables (CROSS and FSTABLE), see the section ‘‘Sample Selection.’’ Indicator variables for the year (Year) are included but not reported t-statistics are based on White’s (1980) heteroskedasticity-consistent standard errors and covariance *Statistically significant at the level of significance using a two-tailed t test **Statistically significant at the 05 level of significance using a two-tailed t test ***Statistically significant at the 01 level of significance using a two-tailed t test 1994; Ortiz-Molina, 2006) To address the possibility of the nonlinearity of MO and the cost of debt, we estimate the following model: SPREAD ¼ C þ b1 MO þ b2 MO2 þ b3 CROSS þ b4 FSTABLE þ b5 MARGIN þ b6 DER þ b7 INCR þ b8 LOAN þ b9 LNASIZE þ b10 BSIZE þ b11 MATUR þ b12 BCFIRM þ b13 RISKP þ YEAR þ e; Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 ð3Þ 608 Journal of Accounting, Auditing & Finance 26(3) where MO2 = the square of the fraction of the shares owned by directors at the end of fiscal year t – Although a piecewise regression model is used in the prior studies (Bagnani et al., 1994; Ortiz-Molina, 2006), we use the quadratic form mentioned above Short and Keasey (1999) argue that the empirical application of the piecewise regression model has a drawback: It allows the coefficients of the MO variables to change only at predetermined levels of ownership As there is no theoretical guidance for the choice of the turning points on the piecewise regression model, we test the relationship between MO and the cost of debt using the quadratic form, which allows the turning points to be determined endogenously Table indicates the regression results It shows that while the coefficient of MO is significantly positive, the coefficient of MO2 is not significant It also reveals that the explanatory power (adjusted R2) of Model is 580, which is slightly lower than that of Model in Table (.581) These results suggest that MO2 has no incremental explanatory power for interest rate spread and is not consistent with the assumption that the relationship between the cost of debt and MO is nonmonotonic In addition to the discussion regarding the section ‘‘Hypotheses Development,’’ we can suggest two possible reasons for our results being different from those of Bagnani et al (1994) First, as stated above, Bagnani et al use a piecewise regression model to test the hypothesis Second, we also indicate the possibility of sample selection biases because they obtained the sample from the list of Fortune 500 companies, which comprises only those firms whose revenues are extremely high The Effect of Bond Rating on the Relationship Between MO and the Cost of Debt Our results are consistent with Hypothesis 2, indicating that the association between MO and interest rate spread is greater when the ACD at the time of corporate bond issue is larger To verify the robustness of the results, we estimate the regression model using the bond rating instead of ACD because the bond rating is often assumed to reflect an ACD and a firm’s default risk (Bhojraj & Sengupta, 2003; Sengupta, 1998; Shuto et al., 2009) Furthermore, Bhojraj and Sengupta (2003) argue that the influence of corporate governance mechanisms would be more critical when dealing with debts of poor quality than otherwise For high-risk firms, bondholders would rely more on the firm’s governance structure because traditional measures of past profitability and leverage may not be very informative about future cash flows Thus, we expect that the ownership structure should have a greater effect on bond yield spread for poorly rated bonds than on high-quality bonds In particular, we estimate the following regression model: SPREAD ¼ C þ b1 MO þ b2 LOWRATE MO þ b3 CROSS þ b4 FSTABLE þ b5 MARGIN þ b6 DER þ b7 INCR þ b8 LOAN þ b9 LNASIZE ð4Þ þ b10 BSIZE þ b11 MATUR þ b12 BCFIRM þ b13 RISKP þ YEAR þ e; where LOWRATE = an indicator variable that takes the value of if R&I’s bond ratings are A or BBB, and otherwise (i.e., AAA or AA).15 We expect the coefficient on LOWRATE MO to be positive in Model The result of the regression with the interaction term is given in Table In Model 4, the coefficient of Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Shuto and Kitagawa 609 Table Regression Results on the Nonlinear Relationship Between Managerial Ownership and Bond Spreads Dependent variable = SPREAD Model Independent variable Constant MO MO2 CROSS FSTABLE MARGIN DER INCR LOAN LNASIZE BSIZE MATUR BCFIRM RISKP Adjusted R2 n Expected sign 1/2 2 2 1 Coefficient t-value 2.958*** 2.085** 21.432 0.077 20.225*** 21.633*** 0.005 0.001 0.980*** 20.064*** 20.072*** 20.025*** 0.064* 0.505*** 0.580 643 6.885 2.367 20.426 0.597 22.773 24.016 1.271 0.900 4.807 24.216 23.275 27.454 1.782 8.729 Note: SPREAD = the interest rate spread on the first straight bond issued of the fiscal year t—the spread is the difference between the interest rate on the bond issued by the firm and that on government bonds; MO = fraction of the shares owned by directors at the end of fiscal year t – 1; MO2 = square of the fraction of the shares owned by directors at the end of fiscal year t – 1; CROSS = fraction of the shares owned by cross-shareholders at the end of fiscal year t – 1; FSTABLE = fraction of the stable shareholdings by financial institutions at the end of fiscal year t – 1; MARGIN = the operating income divided by net sales at the end of fiscal year t – 1; DER = the debt equity ratio at the end of fiscal year t – 1; INCR = the interest coverage ratio at the end of fiscal year t – 1; LOAN = the bank loan divided by total assets at the end of fiscal year t – 1; LNASIZE = the natural log of the total assets at the end of fiscal year t – 1; BSIZE = the log of the issue size; MATUR = the years to maturity; BCFIRM = an indicator variable that takes the value of if a bond management company is established, and otherwise; RISKP = the risk premium: the average values of SPREAD on Rating and Investment Information Inc.’s A bonds for the month of issue For details of ownership variables (CROSS and FSTABLE), see the section ‘‘Sample Selection.’’ Indicator variables for the year (Year) are included but not reported t-statistics are based on White’s (1980) heteroskedasticity-consistent standard errors and covariance *Statistically significant at the level of significance using a two-tailed t test **Statistically significant at the 05 level of significance using a two-tailed t test ***Statistically significant at the 01 level of significance using a two-tailed t test LOWRATE MO is positive and statistically significant at the 01 level, as expected The result reveals that MO has stronger effects on bond yield spread for lower rated bonds, which is consistent with the results of the previous section and the implication of the theory Endogeneity of MO and the Cost of Debt Our results suggest that shareholdings of managers increase the cost of debt because rational bondholders use MO information to anticipate a firm’s future ACD and default risk While interpreting the results, we should consider the joint determination of MO and the cost of debt Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 610 Journal of Accounting, Auditing & Finance 26(3) Table Result of the Differential Effect of Bond Ratings on the Relationship Between Managerial Ownership and Bond Spreads Dependent variable = SPREAD Model Independent variable Constant MO LOWRATE MO CROSS FSTABLE MARGIN DER INCR LOAN LNASIZE BSIZE MATUR BCFIRM RISKP Adjusted R2 n Expected sign 1 1/2 2 2 1 Coefficient t-value 2.853*** 0.132 2.283*** 0.086 20.224*** 21.900*** 0.007 0.003** 0.946*** 20.067*** 20.068*** 20.020*** 0.030 0.471*** 0.612 520 6.436 0.356 3.490 0.662 22.724 23.621 1.370 2.403 4.692 24.483 23.101 26.074 0.732 7.578 Note: SPREAD = the interest rate spread on the first straight bond issued of the fiscal year t—the spread is the difference between the interest rate on the bond issued by the firm and that on government bonds; MO = fraction of the shares owned by directors at the end of fiscal year t – 1; LOWRATE = an indicator variable that takes the value of if Rating and Investment Information Inc.’s bond ratings are A or BBB, and otherwise (i.e., AAA or AA); CROSS = fraction of the shares owned by cross-shareholders at the end of fiscal year t – 1; FSTABLE = fraction of the stable shareholdings by financial institutions at the end of fiscal year t – 1; MARGIN = the operating income divided by net sales at the end of fiscal year t – 1; DER = the debt equity ratio at the end of fiscal year t – 1; INCR = the interest coverage ratio at the end of fiscal year t – 1; LOAN = the bank loan divided by total assets at the end of fiscal year t – 1; LNASIZE = the natural log of the total assets at the end of fiscal year t – 1; BSIZE = the log of the issue size; MATUR = the years to maturity; BCFIRM = an indicator variable that takes the value of if a bond management company is established, and otherwise; RISKP = the risk premium: the average values of SPREAD on Rating and Investment Information Inc.’s A bonds for the month of issue For details of ownership variables (CROSS and FSTABLE), see the section ‘‘Sample Selection.’’ Indicator variables for the year (Year) are included but not reported t-statistics are provided in parentheses They are based on White’s (1980) heteroskedasticityconsistent standard errors and covariance **Statistically significant at the 05 level of significance using a two-tailed t test ***Statistically significant at the 01 level of significance using a two-tailed t test In our hypothesis development, we assume that MO does affect the cost of debt However, there is a possibility that the magnitude of the cost of debt determines the variation in MO For example, firm managers may consider the cost of the capital of the firms when deciding whether to hold stocks of their firms, which is contrary to our assumption If MO and SPREAD are jointly determined, the estimated results are biased and difficult to interpret To solve this simultaneity problem, we use a simultaneous equation model in which the shareholdings of managers and the interest rate spread are jointly determined Specifically, we consider the following system of equations: Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Shuto and Kitagawa 611 SPREAD ¼ C þ b1 MO þ b2 CROSS þ b3 FSTABLE þ b4 MARGIN þ b5 DER þ b6 INCR þ b7 LOAN þ b8 LNASIZE ð5Þ þ b9 BSIZE þ b10 MATUR þ b11 BCFIRM þ b12 RISKP þ YEAR þ e: MO ¼ C þ b1 SPREAD þ INSTRUMRENTS þ e: ð6Þ We estimate Model by conducting a two-stage regression In the first stage, we regress MO on all exogenous variables from Models to The estimation of this regression requires the construction of a set of variables (INSTRUMENTS) associated with MO We use two variables as the instruments: sales growth and – (fixed assets / total assets) In the second stage, we estimate Model instead of MO using the fitted value from the first stage This value is labeled as MOFIT.16 The results of the estimation of the second-stage regression are summarized in Panel A of Table These results are consistent with those in the previous section: The coefficient of MOFIT is positive and statistically significant We also test Hypothesis by using a two-stage regression.17 Panel B indicates that the coefficient of MO ACD is significantly positive, which supports our second hypothesis These results suggest that our findings not merely reflect the simultaneity between MO and interest rate spread Robustness of the Results We describe the analyses conducted further to verify the robustness of our results First, we conduct a regression analysis on a year-by-year basis for our sample and estimate the tvalue based on the approach used by Fama and MacBeth (1973) As these empirical analyses are based on years of pooled cross-sectional data in which the same firm can appear multiple times in the sample, these observations may not be independent This procedure may involve cross-sectional and autocorrelational problems It is well known that the Fama and MacBeth approach can solve these problems and provide a better inference on the estimates The results are summarized in Table Panel A summarizes the results for Hypothesis We are mostly able to obtain the same results: The coefficient of MO is significantly positive With respect to the test of Hypothesis 2, the regression results are summarized in Panel B of Table 9, which provides evidence supporting Hypothesis that the coefficient of MO ACD is significantly positive Our results are robust under the Fama and MacBeth (1973) approach Second, we also examine the relationship between MO and the cost of equity capital If MO reflects the bondholder–shareholder conflict in our research setting, it is expected that MO would not be positively associated with the cost of equity capital because we cannot correctly predict how the severity of bondholder–shareholder conflict affects the cost of equity capital.18 We use the cost of equity capital measured using the three-factor model based on the study by Fama and French (1993, see the appendix) With regard to control variables, we added the logarithm of the market value of equity (MV) and capital asset pricing model (CAPM) beta estimated using data from the 60 months preceding the most recent month of April (BETA) to Model and deleted the variables of the characteristics of the issued bonds from the model Our untabulated result shows that the coefficient of MO is negative and not significant, which is consistent with our prediction Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 612 Journal of Accounting, Auditing & Finance 26(3) Table The Results From the Estimation of the Second-Stage Regression on MO Panel A: The test of hypothesis Dependent variable = SPREAD Model Independent variable Expected sign Constant MOFIT CROSS FSTABLE MARGIN DER INCR LOAN LNASIZE BSIZE MATUR BCFIRM RISKP Adjusted R2 n 1/2 2 2 1 Coefficient 2.611*** 22.019*** 0.822*** 0.625*** 24.273*** 0.017*** 20.009*** 20.008 20.034** 20.079*** 20.011*** 0.016 0.549*** 0.599 626 t-value 6.002 6.768 4.659 4.344 27.513 3.779 24.045 20.032 22.189 23.469 23.003 0.471 9.754 Panel B: The test of Hypothesis Dependent variable = SPREAD Model Independent variable Constant MOFIT MO ACD CROSS FSTABLE MARGIN DER INCR LOAN LNASIZE BSIZE MATUR BCFIRM RISKP Adjusted R2 n Expected sign 1/2 2 2 1 Coefficient t-value 4.299*** 242.397*** 44.132*** 0.117 20.410*** 23.464*** 0.033*** 20.002 0.473** 20.179*** 20.056** 20.020*** 0.105*** 0.445*** 0.642 586 9.020 25.982 6.436 0.874 24.242 26.809 5.639 21.440 2.112 28.230 22.500 25.734 2.644 7.752 Note: SPREAD = the interest rate spread on the first straight bond issued of the fiscal year t—, the spread is the difference between the interest rate on the bond issued by the firm and that on government bonds; MO = fraction of the shares owned by directors at the end of fiscal year t – 1; ACD = the agency cost of debt, computed using factor analysis based on six financial variables at the end of fiscal year t – 1: (1) R&D expenditures / sales, (2) (fixed assets / total assets), (3) cash and marketable securities/total assets, (4) common dividends / total assets, (5) the standard deviation of ROA (net income / total assets) for the past years, (6) the standard deviation of Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Shuto and Kitagawa 613 Table (continued) _ leverage (total debt / total assets) for the past years; CROSS = fraction of the shares owned by crossshareholders at the end of fiscal year t – 1; FSTABLE = fraction of the stable shareholdings by financial institutions at the end of fiscal year t – 1; MARGIN = the operating income divided by net sales at the end of fiscal year t – 1; DER = the debt equity ratio at the end of fiscal year t – 1; INCR = the interest coverage ratio at the end of fiscal year t – 1; LOAN = the bank loan divided by total assets at the end of fiscal year t – 1; LNASIZE = the natural log of the total assets at the end of fiscal year t – 1; BSIZE = the log of the issue size; MATUR = the years to maturity; BCFIRM = an indicator variable that takes the value of if a bond management company is established, and otherwise; RISKP = the risk premium: the average values of SPREAD on Rating and Investment Information Inc.’s A bonds for the month of issue For details of ownership variables (CROSS and FSTABLE), see the section ‘‘Sample Selection.’’ Indicator variables for the year (Year) are included but not reported t-statistics are based on White’s (1980) heteroskedasticity-consistent standard errors and covariance **Statistically significant at the 05 level of significance using a two-tailed t test ***Statistically significant at the 01 level of significance using a two-tailed t test Finally, to assess the economic significance of MO, we consider the marginal effect, which is measured as the coefficient of MO in Model Table shows that the coefficient of MO is 1.655 This result suggests that for a 1% increase in MO, with everything else remaining constant, SPREAD increases by 1.655 basis points Table also indicates that for a one standard deviation increase in MO (0.031), SPREAD increases by 5.131 basis points (1.655 0.031 100), which appears to be economically significant Furthermore, we consider the economic significance of MO ACD in Model Table shows that the coefficient of MO ACD is 4.432, which indicates that for a one standard deviation increase in MO ACD (0.019), SPREAD increases by 8.421 basis points (4.432 0.019 100) The value is higher than that of MO in Model by 5.131 basis points Therefore, the overall results suggest that the coefficient of MO has economic meaning, and it appears to be more economically significant when the ACD at the time of bond issue is larger, which is consistent with our hypotheses Choice of Debt Financing and Bond Yield Spread Finally, it is noteworthy to focus on the result in our main analyses that the LOAN variable has a significant effect on the interest rate spread In this section, we conduct a more indepth discussion of the results of bank loans as the choice of debt financing (bank loan vs bond issue) is critically important; that is, it is associated with the corporate governance structure of firms and the ACD It is well known that Japanese firms are bank dependent relative to U.S firms, and bank loans are a significant source of long-term capital Previous studies suggest that firms with better governance systems could issue corporate bonds since such firms’ ACD is expected to be relatively low (Diamond, 1991) In other words, firms with higher ACD are likely to depend on bank loans rather than bond issues as banks are highly able to monitor firms Consistent with that prediction, Hoshi, Kashyap, and Scharfstein (1993) indicates that keiretsu firms with high profitability are more prone to use bond issues and decrease bank loans Furthermore, in our research context, firms with high MO are assumed to bear high agency costs of debt, and they might be highly dependent on bank loans If such firms issue corporate bonds, their interest yield spread might be relatively higher because it is expected that bank loans are not available any more or have become costly Hoshi et al (1993) also find that owner-manager firms with low profitability tend to access the public debt market Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 614 Journal of Accounting, Auditing & Finance 26(3) Table Regression Results on the Relationship Between Managerial Ownership and Bond Spreads: Fama and MacBeth (1973) Approach Panel A: The test of Hypothesis Dependent variable = SPREAD Model Independent variable Expected sign Constant MO CROSS FSTABLE MARGIN DER INCR LOAN LNASIZE BSIZE MATUR BCFIRM RISKP Adjusted R2 n 1/2 2 2 1 Coefficient t-value 2.607*** 3.410** 0.168** 20.238** 21.416*** 0.008 0.001 1.151*** 20.052** 20.064* 20.023*** 0.055* 0.224* 0.462 643 4.412 2.162 2.003 22.527 24.224 1.366 1.182 4.934 22.421 21.959 25.110 1.782 1.773 Panel B: The test of Hypothesis Dependent variable = SPREAD Model Independent variable Constant MO MO ACD CROSS FSTABLE MARGIN DER INCR LOAN LNASIZE BSIZE MATUR BCFIRM RISKP Adjusted R2 n Expected sign 1 1/– 2 2 1 Coefficient t-value 2.610*** 210.067* 12.286*** 0.223** 20.247** 21.089*** 0.032*** 20.002 1.011*** 20.091*** 20.043 20.023*** 0.010 0.156 0.509 589 4.312 21.652 2.749 2.558 22.020 23.668 4.418 20.790 3.022 24.708 21.352 24.551 0.287 1.067 Note: SPREAD = the interest rate spread on the first straight bond issued of the fiscal year t—the spread is the difference between the interest rate on the bond issued by the firm and that on government bonds; MO = fraction of the shares owned by directors at the end of fiscal year t – 1; ACD = the agency cost of debt, computed using factor analysis based on six financial variables at the end of fiscal year t – 1: (1) R&D expenditures / sales, (2) 1(fixed assets / total assets), (3) cash and marketable securities/total assets, (4) common dividends / total assets, (5) Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Shuto and Kitagawa 615 Table (continued) _ the standard deviation of ROA (net income / total assets) for the past years, (6) the standard deviation of leverage (total debt / total assets) for the past years; CROSS = fraction of the shares owned by cross-shareholders at the end of fiscal year t – 1; FSTABLE = fraction of the stable shareholdings by financial institutions at the end of fiscal year t – 1; MARGIN = the operating income divided by net sales at the end of fiscal year t – 1; DER = the debt equity ratio at the end of fiscal year t – 1; INCR = the interest coverage ratio at the end of fiscal year t – 1; LOAN = the bank loan divided by total assets at the end of fiscal year t – 1; LNASIZE = the natural log of the total assets at the end of fiscal year t – 1; BSIZE = the log of the issue size; MATUR = the years to maturity; BCFIRM = an indicator variable that takes the value of if a bond management company is established, and otherwise; RISKP = the risk premium: the average values of SPREAD on Rating and Investment Information Inc.’s A bonds for the month of issue For details of ownership variables (CROSS and FSTABLE), see the section ‘‘Sample Selection.’’ t-statistics are based on White’s (1980) heteroskedasticity-consistent standard errors and covariance *Statistically significant at the level of significance using a two-tailed t test **Statistically significant at the 05 level of significance using a two-tailed t test ***Statistically significant at the 01 level of significance using a two-tailed t test Our results in Tables and indicate that the coefficients on LOAN are all significantly positive, as predicted The results suggest that the bond yield spreads of firms with higher bank loans are also high, which is consistent with the above argument Therefore, we should note that controlling for bank loans is critically important in our analyses, and our main hypotheses are supported after controlling for the choice of debt financing Conclusion Agency theory predicts that shareholdings of managers create a conflict of interest between shareholders and bondholders (Jensen & Meckling, 1976; Myers, 1977) Limited liability shareholders may have an incentive to expropriate bondholders’ wealth by taking investment and financial decisions aimed at reducing the value of the firm’s outstanding debt Rational bondholders would demand a higher interest rate to compensate for the added risk on owner-managers’ behaviors To test the implication of the theory, we investigate the relationship between MO and the cost of debt, as measured by the interest rate spread on corporate bonds for Japanese firms We find that MO is positively associated with interest rate spread on corporate bonds, as expected We also find that stable shareholdings have a significantly positive association with the interest rate spread, whereas cross-shareholdings are not significantly correlated with it Furthermore, we expect that the effect of MO on the cost of debt strengthens when the potential ACD of firms at the time of bond issue is larger By employing factor analysis to measure the current ACD, we find that MO has a higher correlation with interest rate spread when the potential ACD at the time of bond issue is larger The results are robust with respect to additional analyses, including the possibility of a nonlinear relationship, bond ratings, endogeneity problems, and the Fama and MacBeth (1973) approach Consequently, our results suggest that prospective bondholders in the Japanese market anticipate a firm’s future ACD by using MO information and incorporate this prediction in the pricing of new corporate bond issues Further results suggest that the prospective bondholders estimate a higher firm’s future ACD because of managerial behavior that benefits the managers at the expense of bondholders’ wealth when the current ACD at the time of bond issue is already larger Our results also suggest that prospective bondholders perceive stable shareholdings of financial institutions to be mitigating the wealth transfer problem between bondholders and shareholders Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 616 Journal of Accounting, Auditing & Finance 26(3) Overall, our results suggest that MO is an important determinant of bond yield spread in the Japanese bond market The results also suggest that bond investors focus on the current ACD at the time of bond issue to determine the interest rate in the bond contract Furthermore, the results show that accounting information is useful in estimating the ACD Finally, we find that while cross-shareholdings have no impact on the cost of debt, stable shareholdings by financial institutions has an effect that decreases the cost of debt Our study contributes to prior studies by resolving the function of the Japanese ownership structure in a bond contract setting as this is the first study to examine the relationship between the unique features of the Japanese ownership structure and the interest rate spread of issued bonds Our study may be useful for considering the function of the unique ownership structure of Japan in the bond market Appendix Estimation method for the equity cost of capital To calculate a firm’s estimated equity cost of capital (ECC), we estimate the following equation ^ RMRF;i;t ðRM À Rf Þ þ b ^ SMB;i;t SMBt þ b ^ HML;i;t HMLt ECCi;t ¼ Rf ;t þ b t ð1Þ where (Rm – Rf) is the monthly return of the market portfolio in excess of the risk-free rate HML and SMB are the monthly returns to the book-to-market and size factor mimicking portfolios, as described in Fama and French (1993) First, we calculate each factor’s average monthly return over a period of 60 months before month m Then, we estimate the expected annual factor returns, (Rm – Rf), HML, and SMB, by compounding the resulting average monthly returns over a period of 12 months before the beginning of the fiscal year Second, we estimate the betas associated with the firm’s return to each of the three factors by estimating the following monthly time series regression, as described in Fama and French We estimate the following equation by considering the period of the latest 60 months preceding the beginning of the firm’s fiscal year À Á RETi;m À Rf ;m ¼ þ bRMRF;i RM;m À Rf ;m þ bSMB;i SMBm þ bHML;i HMLm þ ei;m ð2Þ ACKNOWLEDGMENTS The authors appreciate the helpful comments and suggestions received from the editor, two anonymous referees, Kazuyuki Suda, Hidetoshi Yamaji, Hisakatsu Sakurai, Masahiro Enomoto, Nobuyuki Teshima, Takuya Iwasaki, Takashi Yaekura, Takashi Obinata, Mioko Takahashi, Yoshihiro Tokuga, Yoshitaka Fukui, Toshifumi Takada, Fumihiko Kimura, Masaaki Aoki, and the participants of the workshops conducted at Kobe University, Musashi University, and Tohoku University The authors also appreciate the participants of Asian Academic Accounting Association 10th Annual Conference at Kadir Has University All errors are the responsibility of the authors Authors’ Note Data Availability: Data are publicly available from sources identified in the article Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Shuto and Kitagawa 617 Declaration of Conflicting Interests The author(s) declared that they had no conflicts of interests with respect to their authorship or the publication of this article Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Shuto gratefully acknowledges the financial support from the Grant-in-Aid for Scientific Research (#22730354) from the Ministry of Education, Culture, Sports, Science and Technology of Japan Kitagawa gratefully acknowledges the financial support from the Grant-in-Aid for Scientific Research (#21730370) from the Ministry of Education, Culture, Sports, Science and Technology of Japan Notes Another way to reduce conflicts of interest between bondholders and shareholders is to write 10 11 12 bond covenants that restrict the owner-managers’ behaviors to harm bondholders’ wealth (Smith & Warner, 1979) However, McDaniel (1986) indicates that while the use of bond covenants reduces the agency cost of debt (ACD), the protection offered by these covenants cannot totally eliminate the conflicts between bondholders and shareholders With regard to Japanese firms, prior studies have clarified the entrenchment effect from some perspectives: firm performance (Teshima, 2004), earnings management (Teshima & Shuto, 2008), and accounting conservatism (Shuto & Takada, 2008) For details of ownership variables (cross-shareholdings [CROSS] and stable shareholdings [FSTABLE]), see the section titled ‘‘Sample Selection.’’ Most prior studies examining Japanese firms also use the ratio of the shares owned by all directors as a proxy for managerial ownership (MO; Shuto & Takada, 2010; Teshima & Shuto, 2008) Sengupta (1998) examines the effect of disclosure quality on the bond interest rate for U.S firms, and Shuto, Otomasa, and Suda (2009) investigate the bond yield spread for Japanese firms Because the bank loans variable has various implications in the context of debt contracts in Japan, we conduct an in-depth discussion again on the effect size of bank loans on the interest rate spread in the section ‘‘Choice of Debt Financing and Bond Yield Spread.’’ Sengupta (1998, p 464) calculate this variable on the basis of the interest rate of Moody’s AAA bonds However, we could not calculate the average interest rate spread on straight bond issues (SPREAD) for the month of issue because there were insufficient issued bonds with the AAA rating to calculate the risk premium (RISKP) in our sample Following Shuto et al (2009), we then used SPREAD on Rating and Investment Information Inc.’s A bonds to compute RISKP Although we also conduct factor analysis on a year-by-year basis for our sample and calculate the ACD by each year (1996–2003), the results are generally consistent with those of the body We also estimate the alternative model that includes ACD as an independent variable in regression Model The regression results on the main variable are consistent with those of the body These also include parent companies We also conducted robustness tests by using unconsolidated financial statements because consolidated financial statements were not required for primary financial statements under the Securities and Exchange Law of Japan before March 2000 (Shuto, 2009) The results based on unconsolidated financial statements data are consistent with those of the analyses of the body Prior studies that examine Japanese firms indicate that MO is negatively correlated with firm size, which supports our consideration For example, Shuto and Takada (2008), who examine Japanese firms from 1990 and 2005, indicate that MO is negatively and significantly correlated with firm size (the coefficient of Spearman correlation = –0.438, p value = 000) In our sample, Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 618 13 14 15 16 17 Journal of Accounting, Auditing & Finance 26(3) Table also indicates that MO is negatively correlated with the natural log of total assets (LNASIZE) For regression analysis in this study, we calculate the variance inflation factor (VIF) to detect the multicollinearity problem due to high correlation among some of the independent variables We find that the VIF values obtained from regression Model are all less than Furthermore, we find that the VIF values of regression Model are all less than Considering that the standard VIF value is 10 for multicollinearity detection, it is possible to conclude that there is no multicollinearity problem in our models It should be noted that the simple correlation between SPREAD and FSTABLE in Table was positive However, we find that after controlling for other variables, the coefficient on FSTABLE is significantly negative The bond rating used in this analysis is from R&I, the most comprehensive and popular database on bond ratings in Japan Using the Hausman (1978) test, we also assess whether the two variables (MO and SPREAD) are jointly determined To conduct this test, we run the second-stage regression, while including both the actual variables and the predicted value from the first-stage regression The test rejects the null hypothesis that the coefficient of the predicted value is (p value = 000), which implies that the simultaneity problem does exist Specifically, we consider the following system of equations: SPREAD ¼ C þ b1 MO þ b2 MO ACD þ b3 CROSS þ b4 FSTABLE þ b5 MARGIN þ b6 DER þ b7 INCR þ b8 LOAN þ b9 LNASIZE þ b10 BSIZE þ b11 MATUR þ b12 BCFIRM þ b13 RISKP þ YEAR þ e: MO ¼ C þ b1 SPREAD þ INSTRUMENTS þ e: 18 If the incentive alignment effect on MO dominates in this setting, we can expect that MO is negatively associated with the cost of equity capital because the effect has the potential to increase the shareholder value of firms and decrease the cost of equity capital References Ahmed, A S., Billings, B K., Morton, R M., & Stanford-Harris, M (2002) The role of accounting conservatism in mitigating bondholder-shareholder conflicts over dividend policy and in reducing debt costs Accounting Review, 77, 867–890 Anderson, R C., Mansi, S A., & Reeb, D M (2003) Founding family ownership and the agency cost of debt Journal of Financial Economics, 68, 263–285 Aoki, M., & Patrick, H T (1994) The Japanese main bank system: Its relevance for developing and transforming economies New York, NY: Oxford University Press Bagnani, E S., Milonas, N T., Saunders, A., & Travlos, N G (1994) Managers, owners, and the pricing of risky debt: An empirical analysis Journal of Finance, 49, 453–477 Berglo¨f, E., & Perotti, E (1994) The governance structure of the Japanese financial keiretsu Journal of Financial Economics, 36, 259–284 Bhojraj, S., & Sengupta, P (2003) Effect of corporate governance on bond ratings and yields: The role of institutional investors and outside directors Journal of Business, 76, 455–475 Diamond, D W (1984) Financial intermediation and delegated monitoring Review of Economic Studies, 51, 393–414 Diamond, D W (1991) Monitoring and reputation: The choice between bank loans and directly placed debt Journal of Political Economy, 99, 689–721 Douthett, E B., & Jung, K (2001) Japanese corporate grouping (keiretsu) and the informativeness of earnings Journal of International Financial Management & Accounting, 12, 133–159 Fama, E F., & French, K R (1993) Common risk factors in the returns on stocks and bonds Journal of Financial Economics, 33, 3–56 Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 Shuto and Kitagawa 619 Fama, E F., & MacBeth, J D (1973) Risk, return, and equilibrium: Empirical tests Journal of Political Economy, 81, 607–636 Hausman, J A (1978) Specification tests in econometrics Econometrica, 46, 1251–1271 Hoshi, T., & Kashyap, A (2001) Corporate financing and governance in Japan: The road to the future Cambridge, MA: MIT Press Hoshi, T., Kashyap, A., & Scharfstein, D (1990) The role of banks in reducing the costs of financial distress in Japan Journal of Financial Economics, 27, 67–88 Hoshi, T., Kashyap, A., & Scharfstein, D (1991) Corporate structure, liquidity, and investment: Evidence from Japanese industrial groups Quarterly Journal of Economics, 106, 33–60 Hoshi, T., Kashyap, A., & Scharfstein, D (1993) The choice between public and private debt: An analysis of post-deregulation corporate financing in Japan (NBER Working Papers 4421) Cambridge, MA: National Bureau of Economic Research Hwang, L., & Kim, Y O (1998) Does the ownership structure of debt and equity affect the agency costs of debt? Japanese evidence Journal of Accounting, Auditing & Finance, 13, 37–66 Isagawa, N (2007) A theory of unwinding of cross-shareholdings under managerial entrenchment Journal of Financial Research, 30, 163–179 Jensen, M C., & Meckling, W H (1976) Theory of the firm: Managerial behavior, agency costs and ownership structure Journal of Financial Economics, 3, 305–360 Lichtenberg, F R., & Pushner, G M (1994) Ownership structure and corporate performance in Japan Japan and the World Economy, 6, 239–261 Long, M., & Malitz, I (1985) Investment patterns and financial leverage In B M Friedman (Ed.), Corporate capital structures in the United States (pp 353–377) Chicago, IL: University of Chicago Press McConnell, J J., & Servaes, H (1990) Additional evidence on equity ownership and corporate value Journal of Financial Economics, 27, 595–612 McDaniel, M W (1986) Bondholders and corporate governance Business Lawyer, 41, 413 Morck, R., Nakamura, M., & Shivdasani, A (2000) Banks, ownership structure, and firm value in Japan Journal of Business, 73, 539–567 Morck, R., Shleifer, A., & Vishny, R W (1988) Management ownership and market valuation: An empirical analysis Journal of Financial Economics, 20, 293–315 Myers, S C (1977) Determinants of corporate borrowing Journal of Financial Economics, 5, 147–175 Ortiz-Molina, H (2006) Top management incentives and the pricing of corporate public debt Journal of Financial and Quantitative Analysis, 41, 317–340 Osano, H (1996) Intercorporate shareholdings and corporate control in the Japanese firm Journal of Banking & Finance, 20, 1047–1068 Prowse, S D (1990) Institutional investment patterns and corporate financial behavior in the United States and Japan Journal of Financial Economics, 27, 43–66 Prowse, S D (1992) The structure of corporate ownership in Japan Journal of Finance, 47, 1121–1140 Saunders, A., Strock, E., & Travlos, N G (1990) Ownership structure, deregulation, and bank risk taking Journal of Finance, 45, 643–654 Sengupta, P (1998) Corporate disclosure quality and the cost of debt Accounting Review, 73, 459–474 Sheard, P (1989) The Japanese general trading company as an aspect of inter-firm risk sharing Journal of the Japanese and International Economies, 3, 308–322 Sheard, P (1994) Main banks and the governance of financial distress In M Aoki & H Patrick (Eds.), The Japanese main-bank system: Its relevance for developing and transforming economies (pp 188–230) New York, NY: Oxford University Press Short, H., & Keasey, K (1999) Managerial ownership and the performance of firms: Evidence from the UK Journal of Corporate Finance, 5, 79–101 Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 620 Journal of Accounting, Auditing & Finance 26(3) Shuto, A (2009) Earnings management to exceed the threshold: A comparative analysis of consolidated and parent-only earnings Journal of International Financial Management & Accounting, 20, 199–239 Shuto, A., Otomasa, S., & Suda, K (2009) The relative and incremental explanatory powers of dirty surplus for debt interest rate Journal of International Accounting, Auditing and Taxation, 18, 119–131 Shuto, A., & Takada, T (in press) Managerial ownership and accounting conservatism: Empirical evidence from Japan Journal of Business Finance & Accounting, 37, 815-840 Smith, C W., & Warner, J B (1979) On financial contracting: An analysis of bond covenants Journal of Financial Economics, 7, 117–161 Teshima, N (2004) Managerial ownership and corporate governance Tokyo, Japan: Hakuto Shobo (in Japanese) Teshima, N., & Shuto, A (2008) Managerial ownership and earnings management: Theory and empirical evidence from Japan Journal of International Financial Management & Accounting, 19, 107–132 Tirole, J (2006) The theory of corporate finance Princeton, NJ: Princeton University Press Titman, S., & Wessels, R (1988) The determinants of capital structure choice Journal of Finance, 43, 1–19 Watts, R L (1993) A proposal for research on conservatism (Working paper, Simon School of Business, FR 93-13) Rochester, NY: University of Rochester Watts, R L., & Zimmerman, J L (1986) Positive accounting theory Englewood Cliffs, NJ: Prentice Hall White, H (1980) A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity Econometrica, 48, 817–838 Downloaded from jaf.sagepub.com at Taylor's University on December 2, 2012 [...]... 0.306 0.164 0. 178 0.248 1.089 0.281 0.323 0. 072 2.158 10. 573 0.288 0. 578 75 th percentile 0. 970 1.000 0.000 0.216 1.1 27 1 .79 1 0.3 37 0.358 0.000 3.000 13.000 1.000 0. 875 Panel B: Principal Components Analysis Variablesa,b (N = 82,386) B_SIZE B_IND II_TEAM II_STAR GENEXP FIRMEXP P_OUT MOVEUP P_IN FREQ COMP Eigenvalue Factor 1 0.835 20 .76 2 0 .73 4 0.588 0.042 0.065 20.061 0.020 20. 074 20.030 20.0 47 2.354 REPUTATION... (21.28) (20. 67) Upgrade Pre-Regulation FD 0.019 1.501 20. 072 20.365 20.249 55.4% (2.05)** (5 .78 )*** (20.31) (21.12) (21. 57) Upgrade Post-Regulation FD 0.015 1.256 0.410 20.4 87 20. 074 64 .7% (2 .74 )*** (9.13)*** (1.95)* (23.01)*** (20 .75 ) Downgrade Pre-Regulation FD 20.0 27 1.622 0.249 20.034 0.018 47. 4% (22. 47) *** (5.43)*** (0.92) (20.09) (0.1) Downgrade Post-Regulation FD 0.001 1.689 20. 174 20.219 0.012... 0. 171 (13. 57) *** 0.023 (5.61)*** 20.015 (21 .76 )* 20.012 (22.06)** 0.046 (5.84)*** 20.001 (20.12) 0.288 (8.33)*** 0.152 (11.62)*** 20.089 (21 .70 )* 20.001 (20. 27) 20.016 (22 .71 )*** 0.023 (5.62)*** 20.015 (21 .77 )* 20.012 (22.04)* 0.046 (5.84)*** 20.001 (20.12) 0.288 (8.29)*** 0.151 (11.19)*** 20.089 (21 .70 )* 0.0 17 (0.32) 20.016 (22 .70 )*** 27, 792 036 035 27, 792 036 035 LANCRET LRET ACCR CHNG_IO Constant... 500 Journal of Accounting, Auditing & Finance 26(3) Ali, A., Durtschi, C., Lev, B., & Trombley, M (2004) Changes in institutional ownership and subsequent earnings announcement abnormal returns Journal of Accounting, Auditing & Finance, 19, 221–248 Bailey, W., Li, H., Mao, C X., & Zhong, R (2003) Regulation fair disclosure and earnings information: Market, analyst, and corporate responses Journal of Finance,. .. Empirical tests Journal of Political Economy, 81, 6 07 636 Francis, J., Nanda, D., & Wang, X (2006) Re-examining the effects of regulation fair disclosure using foreign listed firms to control for concurrent shocks Journal of Accounting & Economics, 41, 271 –292 Francis, J., & Soffer, L (19 97) The relative informativeness of analysts’ stock recommendations and earnings forecast revisions Journal of Accounting... effectiveness of regulation FD Journal of Accounting & Economics, 37, 293–314 Gomes, A., Gorton, G., & Madureira, L (20 07) SEC regulation fair disclosure, information, and the cost of capital Journal of Corporate Finance, 13, 300–334 Heflin, F., Subramanyam, K R., & Zhang, Y (2003) Regulation FD and the financial information environment: Early evidence Accounting Review, 78 , 1– 37 Irani, A J (2004) The effect of. .. (5.29)*** 20.0 07 (21.58) 0.180 (7. 13)*** 0.080 (9.23)*** 20.049 (22.65)*** 20.002 (20.64) 20.0 07 (22.02)** 20.012 (23.13)*** 0.038 (12.91)*** 0.008 (1.36) 20.015 (23. 27) *** 0.031 (5.23)*** 20.0 07 (21.54) 0. 175 (6. 97) *** 0. 074 (8.48)*** 20.050 (22 .78 )*** 0.111 (3.91)*** 20.012 (23.06)*** 36,895 0 27 0 27 31,841 031 031 31,841 032 031 LANCRET LRET ACCR CHNG_IO Constant Observations R2 Adjusted R2 37, 323 013... quartile SD 8,254. 079 30.191 1.326 0.481 0.002 0.001 0.005 0.004 20.144 0.632 20.099 0.391 20.042 0.002 0.020 20.045 0.003 499.320 14.320 0.661 0.241 20.0 37 20.034 20. 278 20.1 67 21.000 0.000 21.000 0.000 0.000 20.039 20.126 20.081 20.022 1,598. 876 25.205 1.122 0.3 97 0.001 0.000 0.056 20.056 21.000 1.000 0.000 0.000 0.000 0.001 0.012 20.042 0.004 6,034.5 67 40.250 1 .75 2 0.616 0.041 0.0 37 0. 278 0. 278 1.000 1.000... is afraid of Reg FD? The behavior and performance of sell-side analysts following the SEC’s fair disclosure rules Journal of Business, 79 , 2811–2834 Ahmed, A S., & Schneible, R A (20 07) The impact of regulation fair disclosure on investors’ prior information quality—Evidence from an analysis of changes in trading volume and stock price reactions to earnings announcements Journal of Corporate Finance,. .. (22.84)*** 0. 67% (6.95)*** 0.24% (1.89)* 20.43% (22.64)*** 0.0339 (4.35)*** 0.0185 (3.09)*** 20.0154 (21. 57) * 0.0394 (9.08)*** 0.0333 (8.28)*** 20.0061 (21.04) 0.63% (4 .73 )*** 0. 47% (4.1)*** 20.16% (20.92) 0.30% (3.46)*** 0.29% (2.63)*** 20.01% (20. 07) 20.0192 (22.6)*** 20.0094 (21 .78 )* 0.0098 (1.08) 20.0461 ( 27. 9)*** 20.00 67 (21.38) 0.0394 (5.2)*** 20.29% (22.4)** 0.05% (0.35) 0.34% (1 .78 )* 20. 37% (23.93)***