International Diversification Gains and Home Bias in Banking

1
International Diversification Gains and Home Bias
in Banking

• Alicia García-Herrero (BIS)
• Francisco Vázquez (IMF)

Conference on Mergers and Acquisitions of
Financial Institutions L. William Seidman Center,
Arlington November 30-December 1, 2007
2
Overview

• Motivation
• Objectives
• Related literature
• Overview of the sample
• Methodological aspects and results
• Econometric estimates
• Portfolio model
• Conclusions

3
Motivation

• Dramatic increase in FDI in the banking sector
  since the mid-nineties
• International banks would likely obtain
  cross-country diversification benefits
• International diversification in banking is
  barely understood
• International diversification effects not taking
  into account in the Basel II standard approach

4
Cross-Border MAs Targeting Banks in
5
Consolidated Foreign Claims of BIS-Reporting
BanksRecipient Emerging and Developing
Countries1983-2004
6
Synchronization of Macroeconomic
ConditionsHigher Between Industrial Countries
7
Growth Correlations, Cumulative Probabilities by
Country Groups
8
Objectives

• Explore the risk-return effects of international
  diversification in banking
• Are parent banks with a larger share of their
  assets allocated to foreign subsidiaries able to
  obtain larger risk-adjusted profits?
• Is geographical concentration detrimental to the
  risk-adjusted profitability of international
  banks?
• Use portfolio theory as a benchmark to analyze
  the international allocation of bank assets
• How does the actual allocation of bank assets
  compare with the optimal frontier

9
Related Literature

• Portfolio theory (Markovitz 1952, 1959)
• International portfolio diversification (Grubel,
  1968 Levy and Sarnat, 1970 Lessard, 1973)
• Geographical local diversification in banking
  (Acharya, Hasan, and Saunders, 2002 Morgan and
  Samolyk, 2003)
• International diversification in banking
  (Griffith-Jones, Segoviano and Spratt, 2002
  Buch, Discroll and Ostrgaard, 2005)

10
Sample

• Source BankScope Zephyr
• Coverage
• 38 large international banks from G7 countries
  plus Spain
• Their 399 subsidiaries overseas
• Sample unbalanced 1995-2004, with over 2,000
  observations

11
Sample Distribution of Assets
by Country of Incorporation of Parent Banks and
Location of Subsidiaries(unweighted averages, in
percent)
12
Summary Statistics of Risk and Return by Country
Groups
13
Methodology 1) Econometric Estimation
Alternative specification including a Herfindhal
index of asset concentration
14
Dependent Risk-Normalized ROA of Parent Banks
(Consolidated)
15
Omitted variable bias

• The dependent variable is computed from the
  consolidated financial statements of parent banks
• Captures risk-return gains from local operations
  abroad plus those of cross border operations
• If cross-border and local operations are
  complementary coefficients biased toward previous
  finding
• The results could also be driven by unobserved
  differences across banks
• Differences in business strategies, quality of
  risk management, etc.
• We control by exploiting differences in
  information content between consolidated and
  unconsolidated financial statements

16
Robustness Check Controlling for Parent Bank
Idiosyncrasies
17
Dependent Difference of Risk-Normalized ROA
(Consol-Unconsol)
18
Methodology 2) Portfolio model

• Use portfolio theory as a normative benchmark
• Treat foreign subsidiaries as single components
  of the world portfolio of international banks
• Caveats
• Transaction costs of entry/exit a given country
• Bank subsidiaries may not be perfect substitutes
• Time dimension is not balanceddifficult to
  compute variances and covariances of returns
  across subsidiaries
• Treatment
• Focus on portfolio optimization within the
  observed set of subsidiaries of each
  international bank
• Aggregate returns of subsidiaries by regions
  (industrial vs. emerging)

19
Methodology 2) Portfolio model
Expected return
Expected variance
20
Example
21
Deviations of Actual Asset Allocations from the
Efficient Frontier
22
Selected Statistics of the Observed Asset
Allocation(unweighted averages)
23
Conclusions

• On average, banks with a larger share of their
  assets in foreign subsidiaries, particularly in
  emerging economies, have been able to obtain
  larger risk-normalized returns
• The regional concentration of international
  expansion is detrimental to diversification
• Banks exhibit a home-bias in their international
  investment strategiesfurther international
  expansion beneficial from the pure risk-return
  perspective

24
Conclusions (Cont.)

• The estimates strongly underestimate
  international diversification benefits
• Caveat The data do not allow to disentangle
  cross-border investment by parent banks, which
  accounts for a large part of international
  exposures

25

• Thank you!