Financial Market Integration

1
Financial Market Integration

• ECB Notes, October 2004
• Geert Bekaert
• Columbia University

2
FMI What it is and what it is not

• Financial market integration situation where
  securities of similar risk command the same
  expected return
• gt for most securities joint test problem
• Cleanest test one security trading in two places
• gt ADRs
• gt closed-end funds
• (Nishiotis, 2004)

3
FMI What it is and what it is not

• Financial market integration has wide-ranging
  effects
• Expected Returns, Correlation and Volatility
  International Finance
• Consumption Risk Sharing, Efficacy of
  Macroeconomic PolicyInternational Economics
• Investment, Economic GrowthDevelopment
  Economics
• Initial Focus Presentation
• Equity Markets

4
FMI What it is and what it is not

• Integration ? Correlation
• Individual stock price
• discount rate cash flows
• financial integration economic
  integration
• business cycle synchronization
• Cash flow variation is an important source of
  price/dividend or price/earnings variation

5
FMI What it is and what it is not

• Integration ? Correlation
• Country Stock Level
• discount rate cash flows
• financial integration economic
  integration
• industrial/style mix business cycle
  synchronization
• industrial/style mix
• Discount rate variation is most important source
  of price/dividend or price/earnings variation

6
The Industry-Country Debate

• Example Country-Industry Model
• Heston-Rouwenhorst (1994 JFE) Implementation
• ?ik 1 if i in country k, zero otherwise
• dil 1 if i in industry l, zero otherwise
• ßi 1, for all i

7
The Industry-Country Debate
8
Country factors versus Industry factors
from Cavaglia, Brightman, Aked, FAJ, 2000, 41-53
9
The Industry/Country Debate

• Interpretations and big questions
• Is the effect permanent?
• Globalization
• Regional integration (NAFTA, EU, ASEAN)
• Trend in idiosyncratic volatility (Campbell et
  al., Journal of Finance, 2000)
• Or might it be temporary?
• TMT bubble
• Roaring bull, then bear market (increased
  volatility)
• Heston-Rouwenhorst model inadequate
• unit betas
• constant betas

10
The Industry/Country Debate

• Final Thoughts
• Bekaert, Hodrick, Zhang ongoing project
• daily return data 1980-2003, July
• 23 MSCI countries, 26 industries
• an average of 17,000 individual stocks
• Graphs
• 1. Did correlations between U.S. and other
  countries increase? (1 year of weekly data,
  rolling)
• 2. Did correlations between European countries
  increase?
• 3. Did correlations between industries decrease?
• 4. Did the country volatility ratio (volatility
  of an average country relative to the world
  volatility) decrease relative to the industry
  volatility ratio?5. Did the correlation between
  Merck and Novartis increase?

11
The Industry/Country Debate
12
The Industry/Country Debate
13
The Industry/Country Debate
14
The Industry/Country Debate
15
The Industry/Country Debate
16
Crises and Market Contagion
17
Market Integration and Contagion

• Contagion
• Markets move more closely together during periods
  of crisis
• However
• Forbes and Rigobon (2001)
• there is no consensus on exactly what
  constitutes contagion or how it should be
  defined
• Rigobon (2001)
• paradoxically, ... there is no accordance on
  what contagion means

18
Market Integration and Contagion

• Bekaert, Harvey, Ng (JB, 2005)
• We define contagion as excess correlation that
  is, correlation over and above what one would
  expect from economic fundamentals.
• We use an asset pricing model to link
  fundamentals to asset correlation
• gt will depend on degree of
  market integration

19
Market Integration and Contagion

• Intuition
• For a given factor model, increased correlation
  is expected if the volatility of a factor
  increases.
• The size of the increased correlation will depend
  on the factor loadings.
• Contagion is simply defined by the correlation of
  the model residuals.

20
Market Integration and Contagion

• Benefits
• Avoid the problem with the bias correction for
  correlations that Forbes and Rigobon (2002)
  propose
• The bias correction does not work in the presence
  of common shocks.

21
Market Integration and Contagion

• Costs
• Must take a stand on the global, regional and
  country specific fundamentals, as well as the
  mechanism that transfers fundamentals into
  correlation
• Any statements on contagion will be contingent on
  the correct specification of the factor model

22
Market Integration and Contagion
Model - Asymmetric GARCH  
23
Market Integration and Contagion
where  
the excess return on the national equity index of
country i in U.S. dollars
the conditional expected excess returns on the
U.S. and regional markets
the idiosyncratic shock of any market i
the negative return shock of country i
24
Market Integration and Contagion
The sensitivity of equity market i to the foreign
news factors is   where
market capitalization of the U.S., relative to
the total world market capitalization
information variables that capture the covariance
risk of market i with the U.S., the region and
world
25
Market Integration and Contagion

• Nested Models
• p2,i0qidi p1,i0qidi
  p1,i0p2,idi
• CAPM, US CAPM, region world
  CAPM
• benchmark benchmark
• (regional integration)

26
Market Integration and Contagion
Contagion tests
Estimated idiosyncratic shock of market i
Estimated idiosyncratic shock of region g,
where G represents a particular country-group,
i.e. Asia or Latin America
Dummy variable
27
Market Integration and Contagion
Contagion tests

• Dummy variable representing five periods
• Second half of sample
• Mexico crisis
• Asian crisis
• Abnormally negative U.S. unexpected returns
• Abnormally negative regional unexpected returns

28
Market Integration and Contagion
Results Integration

• First half vs. second half betas, correlations
  and variance ratios with respect to the U.S.
  and the region increase suggests greater
  linkages among the various countries.
• Mexican crisis there is no significant
  increase in the regional beta or correlation
  during the crisis. The model suggests no
  change in correlation during this crisis period.

29
Market Integration and Contagion
Results Integration

• Asian Crisis regional correlations, betas and
  variance ratios
• increase by economically meaningful magnitudes
  and are
• statistically significant
• Large negative returns While these negative
  abnormal returns
• are usually associated with higher correlation,
  the increment in
• correlation is substantially smaller than that
  experienced
• during the Asian crisis.

30
Market Integration and Contagion
Results Contagion

• Mexican crisis no significant increase in
  residual
• correlations within Latin America
• Asian Crisis significantly higher residual
  correlations

31
Market Integration and Contagion
Conclusions

• Contagion is the level of correlation over and
  above the
• level that is expected
• We take a stand, using an asset pricing model,
  on the
• expected correlation using an asset pricing
  model that
• allows for world as well as regional factors
  and time-varying
• betas.
• No evidence of contagion from Mexican crisis
  1994-95
• Economically meaningful increases in residual
  correlation
• during the Asian crisis

32
Measuring and Dating Financial Integration

• Many developing countries embarked on financial
  liberalization at the end of the 1980s and early
  nineties.
• Capital Market Liberalization may serve to
  integrate the emerging market into global capital
  markets.
• but Liberalization ? Market Integration

33
Measuring and Dating Financial Integration

• Problems
• Liberalization process is gradual and complex
• Capital controls may not have been effective
• Liberalization may not be credible
• Indirect access may already exist
• Other factors may segment the market (indirect
  barriers political risk)

34
(No Transcript)
35
(No Transcript)
36
Measuring and Dating Financial Integration

• One approach
• Bekaert-Harvey (1995) regime switching model
• Local CAPM
• World CAPM
• Econometrician assesses probability of regimes at
  each point in time.

37
Measuring and Dating Financial Integration
38
Measuring and Dating Financial Integration

• Alternative approach Bekaert, Harvey, Lumsdaine
  (2002) structural break analysis
• Test for breaks in multiple time series
• Date the break and find a confidence interval for
  it
• Strong evidence for breaks but dates do not
  always coincide with dates of major regularly
  reform.
• Country funds/ADRs seem important too.

39
Measuring and Dating Financial Integration
40
Measuring and Dating Financial Integration
41
Financial Effects of Equity Market Liberalization
Asset Prices and Market Integration
Prices
Segmented
Integrated
PI
PS
Return to Integration
Time
High Expected Announcement
Implementation Low Expected Returns
of Liberalization
Returns
42
Average Annual Geometric Returns Pre and
Post Bekaert-Harvey Official Liberalization Dates
Data through April 2002. There are no
pre-liberalization data for Indonesia.
43
Average Annualized Standard Deviation Pre
and Post Bekaert-Harvey Official Liberalization
Dates
Data through April 2002. There are no
pre-liberalization data for Indonesia.
44
Correlation with World Pre and
Post Bekaert-Harvey Official Liberalization Dates
Data through April 2002. There are no
pre-liberalization data for Indonesia.
45
Beta with World Pre and Post
Bekaert-Harvey Official Liberalization Dates
Data through April 2002. There are no
pre-liberalization data for Indonesia.
46
Financial Effects of Equity Market Liberalization

• Formal Empirical Evidence
• (Bekaert and Harvey (2000) Henry (2000) Kim
  and Singal (2000))
• Methodology
• dy, vol, cor country control
  liberalization
• specific variables
  variable
• effect

47
Financial Effects of Equity Market Liberalization

• Summary Results

48
Financial Effects of Equity Market Liberalization

• Related Empirical Evidence
• (Foerster and Karolyi, (1996))
• Abnormal Returns to ADR listings (per week)
• Year before listing 0.349
• During listing week 0.709
• Year following listing -0.190
• Interpretation
• ? Market Integration? Improved ?
  Liquidity? Corporate
• ? Larger Investor Base? Governance?

49
Financial Effects of Equity Market Liberalization

• What should happen to capital flows?
• rebalancing effect
• Bekaert, Harvey, Lumsdaine (2002, JIMF)
• net flowsit a b Libit c Postlibit eit
• with postlibit 1-3 years after liberalization
  onwards
• Predictions b gt 0
• c lt 0

50
Financial Effects of Equity Market Liberalization

• Results for equity flows (as a of market cap)
• b c
• All countries 0.00116 -0.00046
• (7.56) (-2.75)
• Asia 0.00046 0.00003
• (4.72) (0.23)
• Latin-America 0.00219 -0.00154
• (5.97) (-3.72)

51
Financial Effects of Equity Market Liberalization
52
Financial Effects of Equity Market Liberalization
53
Financial Effects of Equity Market Liberalization
54
Financial Effects of Equity Market Liberalization
55
FMI Conclusions

• Financial market integration is related to but
  not identical to correlation.
• Liberalizations do not guarantee that markets
  will integrate.
• Effects of liberalization are nonetheless
  consistent with standard theory.