Valuation and Segmentation in Emerging Markets

1
Valuation and Segmentation in Emerging
Markets

• Geert Bekaert, Columbia NBER
• Campbell R. Harvey, Duke NBER
• Christian T. Lundblad, UNC
• Stephan Siegel, U. of Washington
• May 16, 2008

2
I. The Setting

• Why has globalization treated some countries
  better than others?
• What drives valuation differentials?
• Can we characterize the types of policies that
  change the degree of segmentation both across
  countries and through time?

3
I. The Setting
Developed Markets Example 2003
LARGEST P/E RATIO
SMALLEST P/E RATIO
ln(P/E)
P/E Differential
Industry-adjusted, relative to global markets
4
I. The Setting
Emerging Markets Example 2003
LARGEST P/E RATIO
ln(P/E)
P/E Differential
ln(P/E)
SMALLEST P/E RATIO
P/E Differential
Industry-adjusted, relative to global markets
5
II. The Plan
1. Segmentation 2. Valuation

6
III. Openness
Two aspects of (de jure) globalization

Economic Integration Trade Liberalization
Indicator
Wacziarg and Welch
(2004) Financial Integration Capital
Account Openness Index
Quinn and Toyoda (2001) Equity Market
Openness
Bekaert and Harvey (2000)
7
III. Openness
Trade and Financial Openness Have Increased
8
III. Openness

• Globalization may have wide-ranging effects
• Expected Returns, Correlation and Volatility
• International Finance
• Consumption Risk Sharing, Efficacy of
  Macroeconomic PolicyInternational Economics
• Investment, Economic GrowthDevelopment
  Economics
• Our Focus Effects on Stock Valuation

9
III. Openness
Equity Returns

• Economic Integration
• Specialization
• Exposure to world shocks

Cash Flows
Discount Rates
Real Rates
Term Premiums
Financial Integration
Equity risk premiums
Bond Returns
Economic Integration
Inflation
10
III. Four Contributions

• Building on Bekaert, Harvey, Lundblad, Siegel
  (BHLS) (JF - June 2007), develop a measure of the
  degree of effective market segmentation
• Measurement De Jure Openness ? De Facto
  Integration
• 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 markets
• political risk
• corporate governance issues
• liquidity / financial development
• domestic product and labor markets
• push factors
• Literature Bekaert (1995), Bekaert and Harvey
  (1995), Nishiotis (2004), Aizenman and Noy
  (2005), Lane and Milesi-Ferretti (2001)

1
11
III. Four Contributions

• Combining real and financial variables to
  construct a new measure of exogenous growth
  opportunities
• On average, countries align realized future
  growth with available (exogenous) opportunities
• countries with open equity markets and banking
  sectors are the most successful at exploiting
  available growth opportunities
• financial development and investor protection are
  also important, but to a lesser degree
• Degree of integration / segmentation (as inferred
  from growth predictability regressions) depends
  on country characteristics and varies over time.
• ? This paper develops a direct measure of
  segmentation and explores its determinants

12
III. Four Contributions

• Has the degree of segmentation decreased over
  time? What was the role of (de jure)
  globalization?
• Literature
• Return comovements Longin and Solnik (1995)
  Bekaert, Hodrick, and Zhang (2007)
• Factor Beta Models Bekaert and Harvey (1997,
  JFE) Ng (2000, JIMF) Fratzscher (2002, IJFE)
  Baele (2005, JFQA) Carrieri, Errunza, and Hogan
  (forthcoming, JFQA)
• Return and volatility distance Eun and Lee
  (2005)
• Effects of stock market liberalization on
  dividend yields Bekaert and Harvey (2000), Henry
  (2000)

2
13
III. Four Contributions

• Identify factors that determine the
  cross-sectional and time-series variation in
  segmentation
• Is de jure globalization first order?
• What is the impact of local institutions?
• Literature
• - Bhojraj and Ng (2007)
• - Hail and Leuz (2006)

3
14
III. Four Contributions
4

• Related issues
• Investigate industry-specific degrees of
  segmentation
• Segmentation within the U.S.
• Segmentation within the EU

15
IV. A Measure of Market Segmentation

• Strong Concept of Market Integration
• Industries have identical systematic risk across
  the globe
• Priced growth opportunities are global in nature
• Identical financial risk for each industry,
  independent of the country
• Constant real interest rates
• Each assumption relaxed later in our analysis

16
IV. A Measure of Market Segmentation

• Assume each country i is a basket of industries
  with industry weights IWi,j,t
• Let EYi,j,t earnings yields for country i,
  industry j
• Valuation Differential EYi,j,t- EYw,j,t
• (small and constant under strong market
  integration)
• Measure a countrys degree of observed
  segmentation

17
IV. A Measure of Market Segmentation
Construct SEG for 50 Countries between 1973 and
2005
EMDB 28 countries DataStream 22 countries
12 month trailing earnings yield, negative yields
set to zero
EYi,j,t
12 month global trailing earnings yield, negative
yields set to zero (also considered U.S.)
DataStream
EYw,j,t
EMDB 28 countries DataStream 22 countries
Industry MCAP share in local market
IWi,j,t
18
IV. A Measure of Market Segmentation
19
IV. A Measure of Market Segmentation
20
IV. A Measure of Market Segmentation
Average Country and Industry Segmentation
(MAD) 1973 - 2005
21
V. Market Segmentation Dynamics
SEG Industry-weighted Valuation Differentials
22
V. Market Segmentation Dynamics
SEG Industry-weighted Valuation Differentials
23
V. Market Segmentation Dynamics
SEG Industry-weighted Valuation Differentials
24
V. Market Segmentation Dynamics

• Changes over time suggest we observe valuation
  convergence
• Explore an unbalanced panel regression with a
  simple time trend
• Econometrics (throughout)
• OLS on unbalanced panels Newey-West and SUR
  correction (similar to Thompson (2006))
• Prais-Winsten on unbalanced panel with Beck-Katz
  (1995) correction

25
VI. Market Segmentation U.S. Study

• Clearly, valuation differentials may be due to
  other factors beyond segmentation
• Within the U.S., we explore valuation
  differentials across industries and states to
• uncover any biases in our measure of
  segmentation
• explore other explanatory factors (e.g.,
  leverage, earnings volatility, number of firms)
• Design (a) iteratively draw N random firms
  (resembling countries) or (b) consider U.S.
  states
• ? compare to overall U.S. market

26
VI. Market Segmentation U.S. Study
Segmentation across random draws of U.S. firms
grouped into pseudo-countries
27
VI. Market Segmentation U.S. Study
Segmentation across random draws of U.S. firms
by U.S. states
28
VI. Market Segmentation U.S. Study
1973 - 2006
29
VI. Market Segmentation U.S. Study
100 Random Samples of 50 "Countries" 1973 - 2006
30
VI. Market Segmentation in the EU

• Case study we explore the role for valuation
  convergence in Europe
• Direct analogue
• Consider trends in European valuations relative
  to
• core European basket (FRA, DEU, ITA, NLD, BEL,
  IRL, GBR, DNK)
• Reconsider de jure openness
• To what degree did EU membership or the entrance
  of the Euro Zone facilitate our notion of strong
  market integration? Do these factors explain the
  trend?

31
VI. Market Segmentation in the EU
SEG
32
VI. Market Segmentation in the EU
33
VI. Market Segmentation in the EU
SEG
34
VI. Market Segmentation in the EU
There is a significant trend towards valuation
convergence in Europe. Is that explained by (de
jure) EU or Euro membership? EU
membership is important, but trend persists.
35
VII. Market Segmentation Dynamics (with controls)
36
VII. Market Segmentation Dynamics De Jure
Openness
37
VII. Market Segmentation Dynamics De Jure
Openness
38
VIII. Determinants of Market Segmentation

• Benchmark fixed effects time dummies
  42 R2
• Regulatory openness explains up to 13
• Univariate evidence suggests other factors
  (institutions, financial development, local
  market liquidity, U.S. push factors, etc.) are
  also important
• Is regulatory financial openness primary?

39
VIII. Determinants of Market Segmentation
RISK APPETITE
OPENNESS
INST DEV
GROWTH
CONTROLS
FIN DEV
examples
40
VIII. Determinants of Market Segmentation
Economic Effect on Market Segmentation (N 906,
R2 0.30)
41
VIII. Determinants of Market Segmentation
Economic Effect on Market Segmentation (N 880,
R2 0.33)
42
VIII. Determinants of Market Segmentation
43
VIII. Determinants of Market Segmentation
44
IX. Valuation

• Segmentation is a measure of the absolute
  difference between local and world (industry
  adjusted) earnings yields
• Valuation attempts to explain the difference
  itself. The goal is to understand the drivers of
  under and over valuation

45
IX. Valuation

• Valuation (switch to log PE ratios)

46
IX. Valuation

• 1 What explains the emerging markets discount?

• Use some of the same variables to try to explain
  variation in price to earnings ratios (both
  across countries and through time).

47
IX Valuation
Emerging Market Discount
P R E M I U M
Ave. Discount
Relative PE Ratios
D I S C O U N T
Important factors? Financial openness, political
and institutional risks, illiquid equity markets,
and U.S. default premia
48
IX. Valuation

• 2 Decomposing PE Ratios

• Are they driven by growth opportunities or
  discount rate effects?

49
IX. Valuation

• 2 Decomposing PE Ratios

• Are they driven by growth opportunities or
  discount rate effects?

50
IX. Valuation

• 2 Decomposing PE Ratios

• Empirical model for 5-year real returns
• Empirical model for 5-year real earnings growth
• Project current PE on these two variables.

51
IX. Valuation

• 3 Market Efficiency

• Given our model of expected (industry-adjusted)
  PE ratios, we can take a stand on whether a
  market is over or undervalued.
• Trading simulations where you buy the undervalued
  markets and sell of the overvalued markets

52
Conclusions

• Sementation
• New price-based measure of market segmentation
• Downwar trend in segmentation over time,
  partially explained by de jure globalization.
• Identify most and least segmented industries over
  time.
• Explain about 30 of the variation in degree of
  segmentation across countries and time
• Mostly from the cross-section
• Mainly from financial openness, financial
  development, but global risk factors also
  matter

53
Conclusions

• Valuation
• Valuation in developing markets is challenging
  for investors
• Our framework of industry adjusting compares
  apples to apples
• Our framework of considering the institutional
  environment, the degree of openness as well as
  fundamental information, allows us to understand
  cross country differences in valuation as well
  as time-series patterns.

54
Conclusions

• Why does this matter?
• Market segmentation and or undervaluation raises
  the cost of capital
• Higher cost of capital means less investment and
  less employment growth
• Lower investment and employment growth means
  lower GDP growth
• For example, Bekart, Harvey and Lundblad (JFE
  2005) estimate that a market liberalization which
  reduces the cost of capital is associated with a
  increment in real GDP growth of 1 a year for
  five years

55
Supplementary Materials (not for reproduction)
56
An aside controlling for leverage and number of
firms, which U.S. states are the most segmented?
57
I. Motivation and Goals

• Outline
• Motivation and Goals
• Measure of Market Segmentation
• Market Segmentation Dynamics
• Determinants of Market Segmentation
• Robustness Checks
• Conclusions and Future Work

58
IV. A Measure of Market Segmentation
Pricing industry portfolios

• Earnings growth

where i is country j is industry w is world
59
IV. A Measure of Market Segmentation
Pricing industry portfolios

• Discount Rate

where i is country j is industry w is world
60
IV. A Measure of Market Segmentation
Pricing industry portfolios

• Valuation
• H0 (Strong) Market Integration
• H0 (Strong) Market Segmentation

61
VI. Market Segmentation U.S. Study
Segmentation across U.S. States
62
VIII. Determinants of Market Segmentation

• Methodology
• General multivariate model Which factors account
  for most of the explained variance?
• Need to be able to interpret evidence in the face
  of severe multi-collinearity
• Must reduce the number of factors
• Lack theoretical guidance
• ? Model reduction techniques (e.g. PCGets
  (Hendry))

63
IV. General-to-Specific Modeling
Potential Variables
Equity Market
Capital Account
Pre-search reduction Multiple search
paths Encompassing
64
Explained Variation in SEG
IV. Determinants of Market Segmentation
All Factors
where
Contribution of individual factors (xj) to
predicted segmentation
65
IV. Determinants of Market Segmentation
Decompose
further
Decomposition I
Decomposition II
66
Variance Decomposition
IV. Determinants of Market Segmentation
67
IV. Determinants of Market Segmentation
68
Variance Decomposition
IV. Determinants of Market Segmentation
69
IV. Determinants of Market Segmentation
70
IV. Determinants of Market Segmentation