Pooling of information and information resources may improv

1
Detecting Information Pooling Evidence from
Earnings Forecasts after Brokerage Mergers

• by Serena Ng Matthew Shum
• Discussed by David Becher
• FDIC Conference on Mergers and Acquisitions of
  Financial Institutions

2
Research Question and Motivation

• Do forecasts improve after brokerage mergers?
• Pooling of information and information resources
  may improve forecast accuracy
• Examine stocks where target bidder analysts
  retained
• Compare to those deals where only one analyst
  retained
• No information pooling exists
• Plus distinguish info. pooling from analyst
  selection
• Does analyst selection improve forecast estimates?

3
Contribution of Paper

• Main Takeaway
• Interesting paper, asks important and unique
  questions in the literature
• Clear others think this is a good paper
• Matthew Shums Vita
• Detecting Information Pooling Analysts'
  Forecasts After Brokerage Firm Mergers (with
  Serena Ng)
• Accepted, Advances in Economic Analysis and
  Policy

4
What is Driving Results?

• Authors examine four brokerage mergers
• 1994 to 2000 - market conditions very different
• In all deals, majority of analyst from bidder
  (70 90)
• Inconsistency in firm selection
• Two deals involve foreign bidders (Switzerland)
• Plus, one firm appears as both bidder and target
• Inconsistency in the merger form
• One deal involves acquisition of assets
  (subsidiary)
• Second deal, target is majority owned by another
  firm
• Action of analysts may not be independent

5
What is Driving Results?

• Different results affected vs. non-affected
• Compare bidder MSE pre- to post-merger
• Significant improvement in 2/4 deals for all
  stocks
• No improvement in any deals for affected stocks

6
What is Driving Results?

• Firm-level
• Can we draw conclusions based on 2/4 or 0/4?
• Multivariate results also mixed
• Affected vs. both stay vs. both cover
• Sample sizes limiting information pooling
• Only 7 of cases both analysts stay
• 24/2,251 cases where both stay/cover same stock
  (1)
• Yet estimate (stder) unusually large in merger C
  (4 out of 744)
• Is this information pooling?

7
What is Driving Results?

• Impact of non-merging brokerage firms
• Control for fact accuracy could be changing for
  all brokerage firms (unrelated to merger)
• Binary variable if involved in merger plus binary
  variable if both analysts (target and bidder)
  stay after the merger
• Both are set to zero for non-merging firms
  (footnote 14)
• Include merge, bothstay, and mergebothstay
• Appears mergebothstay is co-linear to bothstay
  merge
• Perhaps look at more/other deals and other
  variables that contribute to accuracy and analyst
  quality?

8
Sample of Brokerage Mergers

• Examine SDC
• Pull all targets with SIC code of
• 621 Security Brokers, Dealers, And Flotation
• 622 Commodity Contracts Brokers And Dealers
• 623 Security And Commodity Exchanges
• Not 6311 (as noted in paper) - life insurance
  firms
• 2,911 potential deals

9
Acquisitions of U.S. Brokerage Firms

• All deals announced from 1980 2005 exclude
• Acquisitions of certain assets (29), remaining
  interests (156), or majority interest (496)
• Acquisition of assets (1,388)
• 532 involve acquisition of 100 of U.S. entity
• PW KP deal 100 acquisition of assets
• DLJ majority owned by AXA (insurance firm)
• No foreign targets (must be in 50 U.S. states or
  DC)
• ? 289 mergers/acquisitions 532 acquisition of
  assets

10
Summary Statistics of 289 Deals

• 37 of deals between 2 brokerage firms
• 56 of targets are public, 49 subs of another firm
• 176 of bidders are public, 47 are not in U.S.
• At least five other deals of similar size
• Shearson-EF Hutton CS-First Boston Alex
  Brown-Bankers Trust Salomon-Citi JP
  Morgan-Chase
• 20 deals if include acquisition of assets
• Robertson Stephens-BofA, Oppenheimer-CIBC,
  Herzog-Merrill Greenwich-NatWest, Spears-Goldman

11
Other Factors

• Future areas of study?
• Compare deals by broker vs. non-broker firms
• Any improvements for non-brokerage bidders?
• Examine stocks with highest variation
• Do brokerages have IB relation for these firms?
• Examine mergers post 2000 (Reg FD)
• Is information sharing different afterwards
• What is impact of analyst recommendations?
• Could information pooling exist here as well?

12
In Summary

• Overall, interesting paper
• Attempts to address the issue of information
  sharing
• Results suggest information pooling occurs
• May be suspicious small samples, issues with
  regressions
• Disagreement between univariate and multiple
  regression approach
• Frame paper as merged brokerage firms retain
  better employees ? increased analyst accuracy
  post-merger
• Many additional questions can be asked as well