Contemporary Accounting Research Conference

1
Contemporary Accounting Research Conference

• Brokerage Industry Self-Regulation
• The Case of Analysts Background Disclosures
• by
• Lawrence D. Brown, Artur Hugon, and Hai Lu
• Discussion by
• Michael Mikhail
• Arizona State University

2
Contemporary Accounting Research Conference

• Primary Findings
• Disclosed Analysts earning forecasts are
  relatively less accurate than those of
  non-disclosed analysts following the same
  firm-quarters (Disclosed analysts were also
  relatively inaccurate in the pre-incident period)
• Controlling for firm and analyst characteristics
  as well as past accuracy, the market reaction is
  weaker to forecasts revisions issued by disclosed
  analysts relative to those by non-disclosed
  analysts (No differential market response is
  observed in the pre-incident period)
• Controlling for firm and analyst characteristics
  as well as past accuracy, average trade size is
  less strongly associated with disclosed analysts
  forecast revisions, suggesting that more
  sophisticated investors pay relatively less
  attention to earnings revisions by disclosed
  analysts

3
Contemporary Accounting Research Conference

• Suggestions / Concerns
• How should we interpret DISC?
• What are the implications of analyst turnover and
  brokerage house quality/prestige?
• Trade volume investigation can we gain anything
  by bifurcating the analysis by trader type?
• Other

4
Contemporary Accounting Research Conference

• How should we interpret DISC?
• Its not clear whether its the disclosure of the
  background event per se that is important or
  whether this merely proxies for some unexamined
  aspect related to the analyst
• The types of disclosures discussed in the
  appendix have different implications for
  performance and/or credibility
• Some disclosures are related to items of a
  personal nature while others are related to an
  analysts professional outputs

5
Contemporary Accounting Research Conference

• How should we interpret DISC?
• Is there variation in the observed association
  between DISC and relative accuracy or the market
  reaction conditional on the type of disclosure?
• Replace DISC with a set of indicators to account
  for the major types of disclosure (Criminal,
  Customer Complaints, Bankruptcies, Regulatory
  Actions, Terminations) and other
• alternatively
• Examine the 116 disclosure events and categorize
  into two groups one related to the analysts
  professional outputs and one related to
  disclosures of a personal nature

6
Contemporary Accounting Research Conference

• How should we interpret DISC?
• Bonner, Hugon and Walther (2007) document that
  celebrity defined as media coverage (e.g.,
  newspapers, magazines, newswires, television,
  etc.) is positively related to the markets
  reaction to forecast revisions even after
  controlling for forecast performance variables
  examined in prior studies
• Is the observed reduction in market reaction in
  the post disclosure period a result of CNBC not
  wanting an analyst with a felony record on any of
  their shows?

7
Contemporary Accounting Research Conference

• Turnover Brokerage House Quality/Prestige
• Mikhail, Walther Willis (1999) and Hong, Kubik
  and Solomon (2000) provide evidence that analysts
  who are relatively worse at forecasting earnings
  are more likely to turnover
• Hong Kubik (2003) find that analysts who are
  relatively inaccurate are about 62 more likely
  to experience a move down the brokerage house
  hierarchy
• Given the current papers results that analysts
  experiencing a DISC were relatively inaccurate in
  the Pre-incident period as well (reported in
  footnote 16), is it possible that the decreased
  market reaction observed is a consequence of
  turnover that results in employment at a less
  prestigious/low quality brokerage?
• Approximately 15 of the disclosures were
  explicit terminations
• What percentage of the total sample of DISC
  analysts experienced turnover between the pre and
  post period?

8
Contemporary Accounting Research Conference

• Low Cost Approaches to Control
• for Brokerage House Quality/Prestige
• Classify Brokerages based on Size
• Prestige is probably not perfectly linear in
  brokerage house size
• Consistent with Clement (1999) and Hong Kubik
  (2003), classify the 10 biggest brokerage houses
  each year as high quality/prestige firms (Bulge
  bracket firms) and the remainder as low
  quality/prestige firms

9
Contemporary Accounting Research Conference

• Low Cost Approaches to Control
• for Brokerage House Quality/Prestige
• Institutional Investor Annual Survey
• The 10 or so brokerage houses with the most
  All-American analysts are identified as The
  Leaders for a particular year
• Table AI in Hong Kubik (2003) provides a
  listing of brokerages classified as leaders for
  each year from 1983 to 2000
• Consistent with prestige not being perfectly
  linear in brokerage house size, only 63 of the
  firms classified as leaders during this time
  period were also one of the 10 biggest for that
  year the range extended from a low of 30 in
  1983 to a high of 90 in 1998

10
Contemporary Accounting Research Conference

• Low Cost Approaches to Control
• for Brokerage House Quality/Prestige
• Carter-Manaster investment banking measure
• Determines investment banking ranking using
  underwriters relative placements in stock
  offering tombstone announcements (see Carter
  and Manaster (1990) and Carter, Dark, and Singh
  (1998))
• Appendix in Carter, Dark, and Singh (1998)
  provides the Carter-Manaster measure for
  approximately 180 investment banks using data
  from 1985 to 1991

11
Contemporary Accounting Research Conference

• Trading Volume Analysis
• Given the authors objective of evaluating how
  different investor segments vary their reliance
  on DISC, I would be interested in an analysis
  that treats each investor type individually
• Bhattacharya (2001), Mikhail, Walther and Willis
  (2007) and DeFranco, Lu and Vasvari (2007), among
  others, follow this approach
• Calculate unexpected trading volume for large and
  small traders separately
• Estimate equation (3) individually for each
  trader type

12
Contemporary Accounting Research Conference

• Other Issues
• Im concerned that the authors only require that
  a firm be followed by 2 analysts.
• In thinly followed firms, analysts are likely to
  have extreme ranks regardless of their
  performance. How are the results affected if the
  sample is restricted to firms followed by at
  least 5 analysts?
• Several DISC analysts have multiple disclosure
  events is there a more significant outcome for
  this subset of the sample?
• Additional information on the NASD disclosure
  process may be helpful does an analyst have the
  ability to appeal? Given resources to fight a
  disclosure event, can analysts from larger firms
  eliminate their disclosure requirement?