What do we know about crosssubsidization Evidence from merging firms

1
What do we know about cross-subsidization?
Evidence from merging firms

• Judith Chevalier
• 2004

2
FORMAT

• Introduction (background, goals, Research design,
  methodology, summary of results)
• Data and variables
• Regression results and discussion
• Conclusion (and summary)
• NOTE Some mergers and spin-offs will be
  available this week.

3
Introduction (background)

• Berger and Ofek (1995) and others argue that
  headquarters may redistribute investment
  resources away from divisions with good
  investment opportunities to divisions with bad
  investment opportunities.
• One issue selection biases arise because
  previous research assumes that divisions are
  randomly allocated to firms.

4
Introduction (Goals)

• To explore the extent to which investment
  policies of divisions are contaminated by the
  selection bias problem.
• To examine the firms that are going to undergo
  diversifying mergers in the future.
• To focus on divisional investment policies
  (cross-subsidization).

5
Introduction (Research design)

• To show that two firms that are going to merger
  in the future show similar relationships in their
  investment patterns.
• Since the two firms are independent prior to
  their merger, these relationships cannot be
  ascribed to cross-subsidization.
• To provide evidence for the similar investment
  patterns.

6
Introduction (measurement of discount)

• Tobins Q of the diversified firm minus average
  Qs of stand-alone firms in the same industries as
  the business segments
• diversified when the segments are in different
  2-digit SIC codes.

7
Introduction (empirical methodology)

• Consider a sample of diversifying mergers (in
  different industries by 2-digit SIC) and examine
  the investment policies of the merging firms
  prior to their merger.
• Prior to merger, by design, no cross-subsidization
  of investment can possible take place.

8
Empirical methodology

• If selection and measurement problems drive many
  of the cross-subsidization results, an
  examination of the investment patterns prior to
  the merger may produce results which look like
  the results produced by the cross-subsidization
  literature.
• Thus, the literatures interpretation of the
  results is problematic.

9
Introduction (summary of the results)

• One merger partners cash flows are predictive of
  the others investment prior to the merger.
  Similar to Shin and Stulz (1998)
• Merger targets that have poor investment
  opportunities relative to their acquirer appear
  to invest more relative to their industry than
  the acquirer. Similar to Rajan, Servaes and
  Zingales (2000)

10
Summary of the results

• 3. Event study evidence on diversification.
• Event study examines the stock market behavior
  upon an event (e.g., mergers, spin-offs,
  regulation changes)
• She shows that the market views the
  diversifying mergers as value-creating.

11
Data

• All mergers between 1980 and 1995. refer to
  Mitchell and Stafford (2000)
• Diversifying mergers in two different
  industries by 2-digit SIC codes
• Dropped the sample in SIC 6000-6999 as Q ratio
  and other measures are inappropriate for
  financial firms.

12
Data and Variables

• Tobins Q MV of assets / BV of assets,
• where MV of assets BV of assets MV of
  equity BV of equity deferred taxes
• or simply MV of assets MV of equity BV of
  debt

13
Data and Variables

• Industry Q (QI) for a firm weighted average of
  QIs of median stand-alone firms in the same
  segment of the divisions.
• Cash flow, sales growth, and investment measures
  are constructed.
• Proxies for inv. Opp. Are Q, CF, SG.

14
II. Investment-cash flow relationship (skip Sec.
III and IV)

• Relationship between a divisions cash flow and
  another divisions investment, after controlling
  for the investing divisions opportunities as
  evidence for cross-subsidization.
• Invi a b QI cSGi dCFi eCFJ e
• Which coefficient?

15
Regression results (the first column in Table 2)

• Invi a b QiI cSGi dCFi eCFJ e
• b 0.0148 (0.0036) std. error is in ( ).
• c 0.0719 (0.007)
• d 0.3225 (0.0354)
• e 0.0212 (0.0096)
• A significant relationship between CFJ and Invi,
  after controlling for opportunities.

16
Conclusion

• She argues that cross-subsidization in the
  literature may be due to systematic differences
  between the investment opportunities of
  conglomerates and stand-alone firms.
• Relatedness among the divisions can contaminate
  the regression results

17
Conclusion

• Examines the investment patterns between two
  firms that are GOING to merge
• and finds investment pattern SIMILAR to the
  cross-subsidization in the literature.
• BUT this cannot happen because these two
  firms are INDENPENDENT prior to mergers.

18
Conclusion (future research)

• What types of firms tend to diversify or remain
  as stand-alone firms?
• The market views about half of the diversifying
  mergers as value-creating and about half as
  value-destroying. Which characteristics in
  mergers are associated with value-creation?

19
Diversification discount or premium? New evidence
from the Business Information Tracking Series

• Belen Villalonga
• 2004

20
Introduction

• Has been an debate on the diversification
  discount (two views and evidence)
• Berger and Ofek and others diversified firms
  trade at discount relative to single-segment
  firms, suggesting that diversification destroys
  value.
• Villalonga and others diversified firms traded
  at a discount diversifying - selection bias.

21
Issues and problems of previous studies

• Questioning the finding itself, not just its
  interpretation as evidence of value destruction.
• The discount is an artifact of segment data from
  COMPUSTAT?
• SIC reported in the data may be different from
  actual industries.
• Therefore, this may distort the industry Q

22
Contribution

• Use BITS as an alternative data to estimate the
  value effect of diversification.
• This data set avoids the problem in COMPUSTA
  segment data.
• Use a common sample of firms and a common method
  to compare the value estimates obtained on BITS
  against those obtained on COMPUSTAT.

23
Summary of the Results

• Find a discount, using COMPUSTAT, while a
  premium, using BITS.
• Two explanations are provided
• Relatedness there is a discount to unrelated
  diversification, but a premium for related
  diversification, Since related ones dominate, on
  average we observe a premium.
• Accounting due to strategic reporting
  practices reflected in COMPUSTAT.

24
Conclusion

• Explores whether the finding that diversified
  firms trade at a discount relative to specialized
  firms may be an artifact of the data.
• With a new database (BITS) without the problems
  in COMPUSTAT, we find diversification premium.

25
Lab Sessions on Research Insights

• Tutorials are available on the Web.
• At MaGruder teaching lab at 300 PM
• Monday (Oct. 9) A thru L
• Wednesday (Oct. 11) M thru Z
• Homework due by next Weds. To be posted on the
  Web.

26
The following week (Oct. 16)

• Read Dale, Mehrotra and Sivakumar, 1997