- Email: [email protected]
Discussion of “What Determines Financial Analysts’ Career Outcomes During Mergers?”
ARTICLE IN PRESS Journal of Accounting and Economics 47 (2009) 87–90
Contents lists available at ScienceDirect
Journal of Accounting and Economics journal homepage: www.elsevier.com/locate/jae
Discussion of ‘‘What Determines Financial Analysts’ Career Outcomes During Mergers?’’ Paul M. Healy Harvard Business School, Boston, MA 02163, USA
a r t i c l e in fo
abstract
Article history: Received 5 November 2008 Received in revised form 1 December 2008 Accepted 2 December 2008 Available online 10 December 2008
Wu and Zang [2009. What determines financial analysts’ career outcomes during mergers? Journal of Accounting & Economics, forthcoming] examine how mergers and acquisitions in the investment banking/brokerage industry affect financial analyst employment. They find evidence of abnormally high analyst turnover following mergers that appears to reflect the acquirer’s elimination of duplicate research capabilities and top analysts at the newly merged firm being hired away by competitors. Finally, they show that the increased analyst turnover at merged firms is related to decreases in analyst coverage and analyst performance post-merger. & 2009 Elsevier B.V. All rights reserved.
JEL Classification: M41 Keywords: Mergers Acquitions Financial analyst
1. Introduction Wu and Zang (2009) examine how mergers and acquisitions in the investment banking/brokerage industry affect financial analyst employment. They find evidence of abnormally high analyst turnover following mergers that appears to reflect the acquirer’s elimination of duplicate research capabilities and top analysts at the newly merged firm being hired away by competitors. Finally, they show that the increased analyst turnover at merged firms is related to decreases in analyst coverage and analyst performance post-merger. A central premise of the paper is acquirers of financial institutions will quickly rationalize their research departments by eliminating surplus analysts. Wu and Zang’s findings indicate how this typically takes place. Sample firm analysts were more likely to be eliminated if they had a poor track record (measured by their 3-year record forecasting earnings), if they covered similar stocks to another analyst at the target or acquirer, and if they worked for the target rather than the acquirer. These findings suggest that acquirers decisions on which analysts to retain were based on past performance and redundancy. The acquirers’ revealed preference for their own analysts (as opposed to the targets’) could reflect a desire to retain analysts who were ‘known commodities’ rather than taking a risk on less known target analysts. It could also have reflected ‘‘merger politics’’ with acquirers using their power to support their own people in the merger consolidation. Wu and Zang also show that financial sector mergers have unfavorable outcomes for acquirers through the loss of highly talented analysts (again measured by past forecast accuracy) to competitors. This is likely to occur for several reasons. First, given the uncertainty they face over merger integration, high-performing target firm analysts are likely to be seen as ‘in-play’ by competitors leading them to receive attractive job offers. And second, a merger is likely to destroy firm-specific human capital of high-performing analysts at the target firm, making them more susceptible to leave. For example, target
E-mail address: [email protected] 0165-4101/$ - see front matter & 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jacceco.2008.12.001
ARTICLE IN PRESS 88
P.M. Healy / Journal of Accounting and Economics 47 (2009) 87–90
analysts who had strong relations with trading and sales personnel that contributed to their past success may anticipate that their relationships will be severed as the merged firm rationalizes these segments of the business. The study also examines the frequency of promotions that take place at mergers. Analysts were more likely to be promoted to management positions at the merged firm in the aftermath of a merger if they had extensive experience and were from the target firm. They were more likely to be hired into management positions at other firms if they had extensive experience, were rated as All-Stars by Institutional Investor, and were made redundant by the merger. Finally, Wu and Zang examine the relation between analyst turnover at mergers, and subsequent changes in stock coverage, forecast frequency, and forecast accuracy. They hypothesize that the documented post-merger analyst turnover will be accompanied by a higher frequency of stocks with dropped coverage, a lower frequency of forecast updates, and less accurate forecasts. Their evidence largely confirms these predictions. The findings indicate that an increase in analyst turnover at the merger is accompanied by an increase in the frequency of stocks with dropped coverage and a decrease in forecast revision frequency. The decline in forecast revision frequency is even more severe for mergers with higher turnover among top forecasters. Finally, perhaps not surprisingly, an increase in turnover among top forecasters at the merged firm is accompanied by a significant decline in post-merger average forecast accuracy. Overall, the paper’s findings suggest that mergers in the investment banking/brokerage industry generated economies of scale in equity research but had a negative effect on the quantity and quality of output of equity research departments. Below, I review the strengths and insights of the paper, and then point out some of its limitations and outstanding questions. 2. Strengths of the paper The authors have done a good job of identifying an interesting setting to examine how mergers and acquisitions affect retention and promotion for lower level employees in the target and acquiring firms. Prior research has focused on the impact of mergers on senior management turnover, typically for CEOs. There have been few studies on turnover at lower levels in the organization. For firms whose human capital is a key asset, understanding how a merger will affect planned and unplanned turnover is an important issue. The findings of this paper are therefore of interest to a broad audience of scholars and practitioners. Scholars interested in the findings include those who study the analyst industry, as well as those who research mergers and acquisitions in accounting, finance, strategy, human resources, and organizational behavior. Practitioners are likely to be particularly interested in turnover among key employees around mergers, since this can affect the likelihood of a transaction’s ultimate success. Wu and Zang should be commended for the care they took in generating a sample and collecting data to undertake the study. Tracking the many banking mergers over time to identify qualifying mergers is no mean feat. The authors appear to have done a thorough job of performing this task in constructing their sample. They also deserve credit for the handcollection of data on analyst turnover and promotion. The use of Nelson’s Directory of Investment Research to hand check the IBES database must have been time-consuming. However, it has paid off. It revealed a disturbingly large number of errors in turnover information on IBES, which should be of concern to other researchers using the database for turnover information. By investing in hand collecting and checking their data, Wu and Zang enable readers to have considerable confidence in the reliability of the findings. 3. Unresolved questions The study leaves a number of unresolved questions and opportunities for follow-up research. 3.1. Acquirer diversity The sample of acquirers of US investment banks/brokers is diverse. It includes insurance companies, commercial banks, international banks/brokers, and domestic investment banks/brokers. These are likely to have very different motives and implementation challenges. For acquirers who do not have a strong presence in the US brokerage business (such as insurance firms, entering commercial banks, and foreign firms), there are likely to be few issues of analyst redundancy and conflicting predictions about the likelihood of losing top performing analysts to competitors after the merger. Given the typical intention to retain the existing departments, target analysts in these acquisitions may be less likely to worry about loss of firm-specific human capital. Of course, they may well have other concerns about whether the acquirers have the expertise to manage their more diverse businesses. For example, given the pay differentials between US investment banks, insurance companies, commercial banks, and non-US firms, target firm analysts may have concerns that an acquisition by such an acquirer will lead to lower pay, accelerating turnover among the top performers. The current tests attempt to assess how these factors affect turnover by including dummy variables for commercial and international firms and conclude that their effects are insignificant. However, these tests are weak. They fail to distinguish between acquisitions by commercial banks/non-US firms that already have a strong presence in the US investment banking/brokerage market and those that do not. There is room for follow-up work in this area to understand the impact of acquisitions by firms with and without experience in the business.
ARTICLE IN PRESS P.M. Healy / Journal of Accounting and Economics 47 (2009) 87–90
89
3.2. Measuring top analyst performance The authors use earnings forecast accuracy as a measure of analyst performance. There are several reasons to use this variable. First, prior research indicates that analyst turnover is higher for the most and least accurate earnings forecasters (see Mikhail et al., 1999; Hong and Kubiak, 2003). Second, from a practical standpoint, it is easy to collect data on forecast accuracy using standard databases. However, it is not clear that investment banks actually use forecast accuracy to rate their analysts. Groysberg et al. (2008) reported that analyst compensation at two top-tier investment banks was highly related to All Star status and not to forecast accuracy. Their sample banks collected data on numerous dimensions of analyst performance, but did not include forecast accuracy among the measures. Wu and Zang include All Star status as a dummy variable in their analysis. Curiously, they do not make any a priori predictions about its relation to turnover. If All Star status is a widely used performance metric, why wouldn’t highly rated analysts be more likely to turn over after a merger for the same reasons hypothesized for top forecasters? The findings indicate that All Star dummy estimates are unrelated to turnover. There are several plausible explanations for this finding. First, Wu and Zang do not have access to the complete Institutional Investor rankings that would be available to acquirers in comparing acquirer and target analysts in the same industry. Instead, they rely on the top five rankings published annually in Institutional Investor magazine. This data limitation lowers the power of their tests. Second, it is possible that All Star status is less important for lower-tier acquirers in the sample, since these firms typically do not have many ranked analysts. They would presumably assess quality using some other performance measure, such as forecast accuracy or commissions generated on stocks covered. These data limitations raise questions about the paper’s conclusion that a high frequency of top performing analysts turnover after mergers. It is certainly true that top forecasters have a relatively higher frequency of departures, but using the All Star performance metric, there is no evidence of increased turnover. I suspect that acquirers would be much more concerned about unplanned post-merger turnover for All Star analysts at target firms than for top forecasters. Noise in the proxy for performance used in the study, also potentially affects the interpretation of the magnitude of merger-related turnover among poor performing analysts. The relation between turnover and poor forecast performance is likely to understate the relation using the actual performance metrics used by the acquirer. As a result, I suspect that merger-related turnover among poor-performing analysts may be even higher than reported in this study. 3.3. Promotions Wu and Zang are among the first scholars to study factors that explain promotions at mergers. However, it is somewhat difficult to interpret their findings for merged firms without a deeper understanding of how promotions are made absent a merger. There is relatively little empirical evidence on this issue, one exception being a recent paper by Campbell (2008). Wu and Zang’s study points to the feasibility of performing a study of analyst promotions using a general sample and compiling data on promotions from Nelson’s directory. This would be a fruitful area for follow-up research. Questions that could be examined include: (a) How do analysts’ research performance, experience, and feedback from institutional customers influence firm promotion decisions? (b) Are external promotions (to other firms) based on the same criteria as internal promotions? and (c) If it is possible to gain access to analyst compensation data, what is the relation between performance valued for compensation and for promotions? 3.4. Controlling for contemporaneous events The sample period is 1997–2004, a volatile period for financial analyst firms. In addition to the large number of financial firm mergers, it includes Regulation Fair Disclosure, the internet/telecom IPO boom and bust, the financial scandals at Enron, Worldcom and other firms, the Sarbanes-Oxley Act, and the Global Settlement. Many of these events had important implications for financial analysts and analyst firms. The specific effects are likely to vary with firm type. For example, the Global Settlement imposed direct sanctions and new business practices on the 12 punished investment banks (e.g. purchase of independent research to supplement their own research, increased Chinese walls between investment banking and research, and greater transparency on research performance and conflicts of interest). Its impact on other banks is less clear. Regulation Fair Disclosure probably had negative effects on analysts who had close relations with managers at the companies they covered. Wu and Zang attempt to control for these contemporaneous events by including year dummies and turnover for analyst firms that had no merger in the sample year and employed a comparable number of analysts to the test firm. In follow-up tests, they use both the test and control firms to estimate the impact of mergers on turnover, dropped coverage, changes in forecast frequency, and accuracy. The turnover findings, reported in panel A of Table 9, are consistent with the main findings that merged firms have high turnover. They do not indicate that merged firms also have relatively high turnover rates for low-accuracy analysts. As the authors recognize, it is unclear whether the control firms fully capture the differential effect of the complex changes that took place during the sample period. For affected firms, the Global Settlement and Reg FD increased the costs
ARTICLE IN PRESS 90
P.M. Healy / Journal of Accounting and Economics 47 (2009) 87–90
and reduced the benefits of equity research for at least some firms, leading to increased analyst turnover and lower coverage. Given the method of matching test and control firms, it would not be surprising if these factors affected the test and matched control firms differentially. The interpretation of the findings is further complicated by the process used to match control and test firms. Potential control firms were excluded from the final sample if they had a merger in the same year as the test firms. But there was no restriction on control firms being involved in mergers several years before or after the test firm event, potentially reducing the power of the tests. 3.5. Role of status in external turnovers The models of turnover and external promotion include, among other variables, the status difference between the acquirer, and the target firm. However, they do not examine the status of the firm that hired the analyst. It would be interesting to understand which types of firms are hiring high- and low- performance analysts at the merged firms. Are top performers moving to firms with similar or higher status? Are the lowest performers moving to lower status firms? Finally, which types of firms are hiring and promoting merged firm analysts? 3.6. Economic interpretation of findings Wu and Zang focused most of their analysis on the turnover and promotion decisions surrounding mergers. They spent much less time discussing and interpreting findings on the relation between merger-related turnover and analyst coverage, research timeliness (measured by the frequency of earnings estimate revisions), and research quality (measured by analysts’ average earnings estimate accuracy). It is difficult to assess the economic significance of the findings reported in the paper (Table 7). For example, the estimate on the top forecaster turnover dummy variable in the model of changes in forecast accuracy was 14.5. This appears to be a very large estimate. What does it imply for forecast accuracy? What are the effects of turnover for low forecast accuracy? The implications of these estimates are not discussed in the study, making it difficult for the reader to assess whether the effects documented are economically important and make economic sense. In follow-up tests reported in panel B of Table 9, merged firms are shown to have a higher relation between dropped coverage and turnover than control firms. This is somewhat surprising since turnover at merged firms from releasing duplicate analysts should have relatively little relation to the frequency of dropped stock coverage. 4. Summary Wu and Zang make a good start addressing the important question of how acquirers decide which employees are to be let go following a merger. Their study examines this question in the context of acquisitions of financial institutions and resulting turnover among equity research analysts. The findings indicate that turnover is partially planned as the acquirer determines which analysts are retained. But it also appears to create unplanned turnover among top forecasters. Finally, the merger-related turnover is accompanied by a decline in coverage, forecast updates, and accuracy. The paper opens several interesting opportunities for follow-up research. First, it raises questions about how turnover varies across acquirers, some of which are entering the industry and do not have opportunities to economize on analysts. Second, the metric used to measure research quality, forecast accuracy, does not appear to be used as a performance metric by at least large firms that dominate the sample. Subsequent work could re-examine post-merger turnover for analysts rated highly by institutional clients. Finally, the evidence on promotions presented in the study identifies a rich source of data on promotions for other scholars that could be used to examine promotion decisions generally. References Campbell, Dennis, 2008. Nonfinancial performance measures and promotion-based incentives. Journal of Accounting Research 46 (2), 297–332. Groysberg, Boris, Healy, Paul, Maber, Dennis, 2008. What drives financial analyst compensation at high status banks? Working Paper, Harvard Business School, Boston, MA. Hong, Harrison, Kubiak, 2003. Analyzing the analysts: career concerns and biased earnings forecasts. Journal of Finance LVIII (1), 313–351. Mikhail, Michael, Walther, Beverley, Willis, Richard, 1999. Does forecast accuracy matter to security analysts? The Accounting Review 46, 31–49. Wu, Joanna, Zang, Amy, 2009. What determines financial analysts’ career outcomes during mergers? Journal of Accounting & Economics 47, 59–86.
Contents lists available at ScienceDirect
Journal of Accounting and Economics journal homepage: www.elsevier.com/locate/jae
Discussion of ‘‘What Determines Financial Analysts’ Career Outcomes During Mergers?’’ Paul M. Healy Harvard Business School, Boston, MA 02163, USA
a r t i c l e in fo
abstract
Article history: Received 5 November 2008 Received in revised form 1 December 2008 Accepted 2 December 2008 Available online 10 December 2008
Wu and Zang [2009. What determines financial analysts’ career outcomes during mergers? Journal of Accounting & Economics, forthcoming] examine how mergers and acquisitions in the investment banking/brokerage industry affect financial analyst employment. They find evidence of abnormally high analyst turnover following mergers that appears to reflect the acquirer’s elimination of duplicate research capabilities and top analysts at the newly merged firm being hired away by competitors. Finally, they show that the increased analyst turnover at merged firms is related to decreases in analyst coverage and analyst performance post-merger. & 2009 Elsevier B.V. All rights reserved.
JEL Classification: M41 Keywords: Mergers Acquitions Financial analyst
1. Introduction Wu and Zang (2009) examine how mergers and acquisitions in the investment banking/brokerage industry affect financial analyst employment. They find evidence of abnormally high analyst turnover following mergers that appears to reflect the acquirer’s elimination of duplicate research capabilities and top analysts at the newly merged firm being hired away by competitors. Finally, they show that the increased analyst turnover at merged firms is related to decreases in analyst coverage and analyst performance post-merger. A central premise of the paper is acquirers of financial institutions will quickly rationalize their research departments by eliminating surplus analysts. Wu and Zang’s findings indicate how this typically takes place. Sample firm analysts were more likely to be eliminated if they had a poor track record (measured by their 3-year record forecasting earnings), if they covered similar stocks to another analyst at the target or acquirer, and if they worked for the target rather than the acquirer. These findings suggest that acquirers decisions on which analysts to retain were based on past performance and redundancy. The acquirers’ revealed preference for their own analysts (as opposed to the targets’) could reflect a desire to retain analysts who were ‘known commodities’ rather than taking a risk on less known target analysts. It could also have reflected ‘‘merger politics’’ with acquirers using their power to support their own people in the merger consolidation. Wu and Zang also show that financial sector mergers have unfavorable outcomes for acquirers through the loss of highly talented analysts (again measured by past forecast accuracy) to competitors. This is likely to occur for several reasons. First, given the uncertainty they face over merger integration, high-performing target firm analysts are likely to be seen as ‘in-play’ by competitors leading them to receive attractive job offers. And second, a merger is likely to destroy firm-specific human capital of high-performing analysts at the target firm, making them more susceptible to leave. For example, target
E-mail address: [email protected] 0165-4101/$ - see front matter & 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jacceco.2008.12.001
ARTICLE IN PRESS 88
P.M. Healy / Journal of Accounting and Economics 47 (2009) 87–90
analysts who had strong relations with trading and sales personnel that contributed to their past success may anticipate that their relationships will be severed as the merged firm rationalizes these segments of the business. The study also examines the frequency of promotions that take place at mergers. Analysts were more likely to be promoted to management positions at the merged firm in the aftermath of a merger if they had extensive experience and were from the target firm. They were more likely to be hired into management positions at other firms if they had extensive experience, were rated as All-Stars by Institutional Investor, and were made redundant by the merger. Finally, Wu and Zang examine the relation between analyst turnover at mergers, and subsequent changes in stock coverage, forecast frequency, and forecast accuracy. They hypothesize that the documented post-merger analyst turnover will be accompanied by a higher frequency of stocks with dropped coverage, a lower frequency of forecast updates, and less accurate forecasts. Their evidence largely confirms these predictions. The findings indicate that an increase in analyst turnover at the merger is accompanied by an increase in the frequency of stocks with dropped coverage and a decrease in forecast revision frequency. The decline in forecast revision frequency is even more severe for mergers with higher turnover among top forecasters. Finally, perhaps not surprisingly, an increase in turnover among top forecasters at the merged firm is accompanied by a significant decline in post-merger average forecast accuracy. Overall, the paper’s findings suggest that mergers in the investment banking/brokerage industry generated economies of scale in equity research but had a negative effect on the quantity and quality of output of equity research departments. Below, I review the strengths and insights of the paper, and then point out some of its limitations and outstanding questions. 2. Strengths of the paper The authors have done a good job of identifying an interesting setting to examine how mergers and acquisitions affect retention and promotion for lower level employees in the target and acquiring firms. Prior research has focused on the impact of mergers on senior management turnover, typically for CEOs. There have been few studies on turnover at lower levels in the organization. For firms whose human capital is a key asset, understanding how a merger will affect planned and unplanned turnover is an important issue. The findings of this paper are therefore of interest to a broad audience of scholars and practitioners. Scholars interested in the findings include those who study the analyst industry, as well as those who research mergers and acquisitions in accounting, finance, strategy, human resources, and organizational behavior. Practitioners are likely to be particularly interested in turnover among key employees around mergers, since this can affect the likelihood of a transaction’s ultimate success. Wu and Zang should be commended for the care they took in generating a sample and collecting data to undertake the study. Tracking the many banking mergers over time to identify qualifying mergers is no mean feat. The authors appear to have done a thorough job of performing this task in constructing their sample. They also deserve credit for the handcollection of data on analyst turnover and promotion. The use of Nelson’s Directory of Investment Research to hand check the IBES database must have been time-consuming. However, it has paid off. It revealed a disturbingly large number of errors in turnover information on IBES, which should be of concern to other researchers using the database for turnover information. By investing in hand collecting and checking their data, Wu and Zang enable readers to have considerable confidence in the reliability of the findings. 3. Unresolved questions The study leaves a number of unresolved questions and opportunities for follow-up research. 3.1. Acquirer diversity The sample of acquirers of US investment banks/brokers is diverse. It includes insurance companies, commercial banks, international banks/brokers, and domestic investment banks/brokers. These are likely to have very different motives and implementation challenges. For acquirers who do not have a strong presence in the US brokerage business (such as insurance firms, entering commercial banks, and foreign firms), there are likely to be few issues of analyst redundancy and conflicting predictions about the likelihood of losing top performing analysts to competitors after the merger. Given the typical intention to retain the existing departments, target analysts in these acquisitions may be less likely to worry about loss of firm-specific human capital. Of course, they may well have other concerns about whether the acquirers have the expertise to manage their more diverse businesses. For example, given the pay differentials between US investment banks, insurance companies, commercial banks, and non-US firms, target firm analysts may have concerns that an acquisition by such an acquirer will lead to lower pay, accelerating turnover among the top performers. The current tests attempt to assess how these factors affect turnover by including dummy variables for commercial and international firms and conclude that their effects are insignificant. However, these tests are weak. They fail to distinguish between acquisitions by commercial banks/non-US firms that already have a strong presence in the US investment banking/brokerage market and those that do not. There is room for follow-up work in this area to understand the impact of acquisitions by firms with and without experience in the business.
ARTICLE IN PRESS P.M. Healy / Journal of Accounting and Economics 47 (2009) 87–90
89
3.2. Measuring top analyst performance The authors use earnings forecast accuracy as a measure of analyst performance. There are several reasons to use this variable. First, prior research indicates that analyst turnover is higher for the most and least accurate earnings forecasters (see Mikhail et al., 1999; Hong and Kubiak, 2003). Second, from a practical standpoint, it is easy to collect data on forecast accuracy using standard databases. However, it is not clear that investment banks actually use forecast accuracy to rate their analysts. Groysberg et al. (2008) reported that analyst compensation at two top-tier investment banks was highly related to All Star status and not to forecast accuracy. Their sample banks collected data on numerous dimensions of analyst performance, but did not include forecast accuracy among the measures. Wu and Zang include All Star status as a dummy variable in their analysis. Curiously, they do not make any a priori predictions about its relation to turnover. If All Star status is a widely used performance metric, why wouldn’t highly rated analysts be more likely to turn over after a merger for the same reasons hypothesized for top forecasters? The findings indicate that All Star dummy estimates are unrelated to turnover. There are several plausible explanations for this finding. First, Wu and Zang do not have access to the complete Institutional Investor rankings that would be available to acquirers in comparing acquirer and target analysts in the same industry. Instead, they rely on the top five rankings published annually in Institutional Investor magazine. This data limitation lowers the power of their tests. Second, it is possible that All Star status is less important for lower-tier acquirers in the sample, since these firms typically do not have many ranked analysts. They would presumably assess quality using some other performance measure, such as forecast accuracy or commissions generated on stocks covered. These data limitations raise questions about the paper’s conclusion that a high frequency of top performing analysts turnover after mergers. It is certainly true that top forecasters have a relatively higher frequency of departures, but using the All Star performance metric, there is no evidence of increased turnover. I suspect that acquirers would be much more concerned about unplanned post-merger turnover for All Star analysts at target firms than for top forecasters. Noise in the proxy for performance used in the study, also potentially affects the interpretation of the magnitude of merger-related turnover among poor performing analysts. The relation between turnover and poor forecast performance is likely to understate the relation using the actual performance metrics used by the acquirer. As a result, I suspect that merger-related turnover among poor-performing analysts may be even higher than reported in this study. 3.3. Promotions Wu and Zang are among the first scholars to study factors that explain promotions at mergers. However, it is somewhat difficult to interpret their findings for merged firms without a deeper understanding of how promotions are made absent a merger. There is relatively little empirical evidence on this issue, one exception being a recent paper by Campbell (2008). Wu and Zang’s study points to the feasibility of performing a study of analyst promotions using a general sample and compiling data on promotions from Nelson’s directory. This would be a fruitful area for follow-up research. Questions that could be examined include: (a) How do analysts’ research performance, experience, and feedback from institutional customers influence firm promotion decisions? (b) Are external promotions (to other firms) based on the same criteria as internal promotions? and (c) If it is possible to gain access to analyst compensation data, what is the relation between performance valued for compensation and for promotions? 3.4. Controlling for contemporaneous events The sample period is 1997–2004, a volatile period for financial analyst firms. In addition to the large number of financial firm mergers, it includes Regulation Fair Disclosure, the internet/telecom IPO boom and bust, the financial scandals at Enron, Worldcom and other firms, the Sarbanes-Oxley Act, and the Global Settlement. Many of these events had important implications for financial analysts and analyst firms. The specific effects are likely to vary with firm type. For example, the Global Settlement imposed direct sanctions and new business practices on the 12 punished investment banks (e.g. purchase of independent research to supplement their own research, increased Chinese walls between investment banking and research, and greater transparency on research performance and conflicts of interest). Its impact on other banks is less clear. Regulation Fair Disclosure probably had negative effects on analysts who had close relations with managers at the companies they covered. Wu and Zang attempt to control for these contemporaneous events by including year dummies and turnover for analyst firms that had no merger in the sample year and employed a comparable number of analysts to the test firm. In follow-up tests, they use both the test and control firms to estimate the impact of mergers on turnover, dropped coverage, changes in forecast frequency, and accuracy. The turnover findings, reported in panel A of Table 9, are consistent with the main findings that merged firms have high turnover. They do not indicate that merged firms also have relatively high turnover rates for low-accuracy analysts. As the authors recognize, it is unclear whether the control firms fully capture the differential effect of the complex changes that took place during the sample period. For affected firms, the Global Settlement and Reg FD increased the costs
ARTICLE IN PRESS 90
P.M. Healy / Journal of Accounting and Economics 47 (2009) 87–90
and reduced the benefits of equity research for at least some firms, leading to increased analyst turnover and lower coverage. Given the method of matching test and control firms, it would not be surprising if these factors affected the test and matched control firms differentially. The interpretation of the findings is further complicated by the process used to match control and test firms. Potential control firms were excluded from the final sample if they had a merger in the same year as the test firms. But there was no restriction on control firms being involved in mergers several years before or after the test firm event, potentially reducing the power of the tests. 3.5. Role of status in external turnovers The models of turnover and external promotion include, among other variables, the status difference between the acquirer, and the target firm. However, they do not examine the status of the firm that hired the analyst. It would be interesting to understand which types of firms are hiring high- and low- performance analysts at the merged firms. Are top performers moving to firms with similar or higher status? Are the lowest performers moving to lower status firms? Finally, which types of firms are hiring and promoting merged firm analysts? 3.6. Economic interpretation of findings Wu and Zang focused most of their analysis on the turnover and promotion decisions surrounding mergers. They spent much less time discussing and interpreting findings on the relation between merger-related turnover and analyst coverage, research timeliness (measured by the frequency of earnings estimate revisions), and research quality (measured by analysts’ average earnings estimate accuracy). It is difficult to assess the economic significance of the findings reported in the paper (Table 7). For example, the estimate on the top forecaster turnover dummy variable in the model of changes in forecast accuracy was 14.5. This appears to be a very large estimate. What does it imply for forecast accuracy? What are the effects of turnover for low forecast accuracy? The implications of these estimates are not discussed in the study, making it difficult for the reader to assess whether the effects documented are economically important and make economic sense. In follow-up tests reported in panel B of Table 9, merged firms are shown to have a higher relation between dropped coverage and turnover than control firms. This is somewhat surprising since turnover at merged firms from releasing duplicate analysts should have relatively little relation to the frequency of dropped stock coverage. 4. Summary Wu and Zang make a good start addressing the important question of how acquirers decide which employees are to be let go following a merger. Their study examines this question in the context of acquisitions of financial institutions and resulting turnover among equity research analysts. The findings indicate that turnover is partially planned as the acquirer determines which analysts are retained. But it also appears to create unplanned turnover among top forecasters. Finally, the merger-related turnover is accompanied by a decline in coverage, forecast updates, and accuracy. The paper opens several interesting opportunities for follow-up research. First, it raises questions about how turnover varies across acquirers, some of which are entering the industry and do not have opportunities to economize on analysts. Second, the metric used to measure research quality, forecast accuracy, does not appear to be used as a performance metric by at least large firms that dominate the sample. Subsequent work could re-examine post-merger turnover for analysts rated highly by institutional clients. Finally, the evidence on promotions presented in the study identifies a rich source of data on promotions for other scholars that could be used to examine promotion decisions generally. References Campbell, Dennis, 2008. Nonfinancial performance measures and promotion-based incentives. Journal of Accounting Research 46 (2), 297–332. Groysberg, Boris, Healy, Paul, Maber, Dennis, 2008. What drives financial analyst compensation at high status banks? Working Paper, Harvard Business School, Boston, MA. Hong, Harrison, Kubiak, 2003. Analyzing the analysts: career concerns and biased earnings forecasts. Journal of Finance LVIII (1), 313–351. Mikhail, Michael, Walther, Beverley, Willis, Richard, 1999. Does forecast accuracy matter to security analysts? The Accounting Review 46, 31–49. Wu, Joanna, Zang, Amy, 2009. What determines financial analysts’ career outcomes during mergers? Journal of Accounting & Economics 47, 59–86.