The Role of ADRs in the Development and Integration of Emerging Equity Markets. G. Andrew Karolyi Fisher College of Business Ohio State University

Transcription

1 The Role of ADRs in the Development and Integration of Emerging Equity Markets G. Andrew Karolyi Fisher College of Business Ohio State University

2 The Question There has been a significant growth international cross-listings of stocks, especially in the U.S. by means of American Depositary Receipts (ADRs) and especially from emerging markets in the 1990s Advantages to investors? Liquidity, transparency, ease of trade Advantages to issuers? Attract global investors to shares across which to spread risks and thus to lower cost of capital Does the growth and expansion of ADR markets for companies from different markets around the world facilitate: Overall development of equity markets? Market integration of emerging markets with global markets?

3 The Answer Yes and no! I find that increasing number of new ADR programs, their market cap and trading activity are: Positively associated with international capital flows and greater market integration; Negatively associated with quality of home market environment (number of listed companies, market cap and turnover) Adverse effect concentrated on non-adrs in home market Primary driver is exchange-listed ADRs (vs OTC & Rule 144a listings)

4 Three Sources of Motivation The study contributes to literature on impact of market liberalization on financial market development, especially in emerging markets Measuring integration of capital markets often relies on regulatory events as guideposts but they can have small, delayed, unexpected impact Development of ADRs market is by its nature a process not an event & its growing economic vitality over time is directly measurable Many studies on why firms list overseas and its net economic benefits, but few consider overall macroeconomic consequences of growing ADR market and especially side effects to non-adr market Policymakers in emerging markets concerned with impact of expansion of ADR market in spite of numerous advantages promoted by depositary banks, exchanges, I-bankers

5 Key Hypothesis H 0 : Development of ADR market is a catalyst and is positively associated with market development and integration of local market with world markets As issuer companies and investors choose ADRs, local exchanges, brokers and regulatory authorities come under competitive pressure to modernize operations, enhance disclosure standards, strengthen enforcement to make local markets more liquid, transparent, efficient Increased participation by local companies & presence of global investors leads to greater liquidity, visibility and credibility of local market for global investors spiralling toward greater efficiency, expansion and development H A : Development of ADR market is a hindrance and is negatively associated with market development and integration of local market with world markets Development of ADR market represents a diversion of activity away from the local market Deterioration of quality of local market as trading volume and liquidity diminish leading to fewer global investors in stocks beyond those trading as ADRs and increasingly lower participation by local investors and companies

6 Organization of Paper Measuring the growth of ADR programs in emerging markets The business of cross-listings with ADRs Construction of Proxy Variables Experiment #1: Impact on stock market development Four measures of stock market development Regression analysis: SUR models Separate analysis for ADRs and Non-ADR companies Different results by type of ADR listing Experiment #2: Characterizing integration over time An International Asset Pricing Model (IAPM) Implementation and estimation Regression analysis: SUR models overall and by listing type Discussion and conclusions

7 Data Cross-listings and ADRs from NYSE, Amex, Nasdaq, Bank of New York ( and Citibank ( OTC Bulletin Board and pink sheet lists Monthly prices, dividends, market capitalization, trading volume from Standard & Poor s Emerging Market Database (2000) Twelve countries included: Argentina, Brazil, Chile, Colombia, Mexico and Venezuela from Latin America; Indonesia, Korea, Malaysia, Philippines, Taiwan and Thailand from Asia. Computed value-weighted U.S. dollar returns for ADR portfolio and took IFCG value-weighted indices from EMDB Morgan Stanley Capital International world market index Ibbotson & Associates for U.S. 10-year Treasury yield

8 Measures of ADR Activity Three measures: Fraction of total number of stocks listed on local market with shares also listed as ADRs (NUMFRAC) Fraction of total market capitalization of stocks listed on local market with shares also listed as ADRs (MCAPFRAC) Fraction of total (U.S. dollar) value of trading of stocks listed on local market with shares also listed as ADRs (VOLFRAC) Tricky issues: Identifying effective dates of programs & changes in programs No measure of ADR activity or presence in U.S. No measure of capital raising activity No measure of activity via cross-listings in other markets in world

9 Comparing Scope of ADR Activity Number of Listed ADRs as % of Total Listed Stocks (NUMFRAC) 30 Colombia 250 Korea Number of Stocks Number of Stocks All stocks ADR stocks All stocks ADR stocks 90 Mexico 70 Philippines Number of Stocks Number of Stocks All stocks ADR stocks All stocks ADR stocks

10 Other Measures of Growth of ADRs Percentage of Total Dollar Market Value of Stocks in ADRs (MCAPFRAC) Chile Thailand Market Value (US$m) Market Value (US$m) All stocks ADR stocks All stocks ADR stocks 3500 Percentage of Total Dollar Value of Trading by ADRs (VOLFRAC) Argentina Malaysia Value of Trading (US$m) Value of Trading (US$m) All stocks ADR stocks All stocks ADR stocks

11 Measures of Development Four measures of stock market development: Market value of all listed shares divided by GDP (MKTGDP) Number of all listed shares divided by GDP (NUMGDP) Value of trading of all listed shares divided by market capitalization (TURNOVER) Gross equity capital flows from US residents to emerging markets divided by GDP (FLOWGDP) Additional data sources Treasury International Capital (TIC) for gross flows World Bank s World Development Indicators + IMF s IFS data

12 Empirical Methodology I Regression analysis involves individual country-level and panel regressions using SURM, all Newey-West heteroscedasticity consistent standard errors and serial correlation to 3 lags To generate additional power to tests of null hypothesis, we introduce two control variables: LDATE Bekaert, Harvey, Lumsdaine (2002), Table 3, dummy variable for official liberalization date in respective markets OPENNESS Edison & Warnock (2001), ratio of market cap of constituent members of IFC Investable to IFC Global indexes. Due to concerns about spurious association of ADR growth and development measures due to secular trend, I add lagged dependent variables

13 SURM Tests (Table 3) Model (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) MKTGDP NUMGDP OPENNESS (0.24) (0.39) (-0.96) (-5.76)** (-1.93)* (-2.78)** (-3.59)** LDATE (1.16) (0.63) (1.17) (-4.35)** (-2.65)** (-2.19)** (-2.21)** NUMFRAC (-1.38) (-4.43)** MCAPFRAC (1.16) (-3.16)** VOLFRAC (1.16) (-2.58)** TURNOVER FLOWGDP OPENNESS (3.61)** (2.68)** (1.64)* (1.59) (2.51)* (0.40) (0.61) (1.05) LDATE (3.50)** (1.92)* (2.14)** (2.24)** (2.13)* (0.80) (0.63) (0.69) NUMFRAC (-2.52)** (3.34)** MCAPFRAC (0.19) (3.60)** VOLFRAC (0.97) (2.24)**

14 ADRs vs Non-ADRs (New Table 4) Model (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Non-ADRs Only MKTGDP NUMGDP TURNOVER OPENNESS (-0.71) (0.72) (-0.36) (-0.94) (-5.59)** (-0.76) (-1.95)* (-3.02)** (3.95)** (4.21)** (2.09)** (6.09)** LDATE (1.57) (0.97) (1.66)* (2.01)** (-5.72)** (-4.35)** (-3.19)** (-3.70)** (3.04)** (1.00) (1.66)* (1.98)* NUMFRAC (-3.96)** (-11.9)** (-4.24)** MCAPFRAC (-4.37)** (-7.17)** (-0.04) VOLFRAC (-3.31)** (-5.46)** (-7.45)** ADRs Only MKTGDP NUMGDP TURNOVER OPENNESS (1.18) (-0.37) (-2.86)** (0.09) (4.79)** (3.16)** (4.86)** (4.52)** (0.41) (1.22) (0.18) (-1.46) LDATE (0.89) (0.44) (1.91)* (0.11) (4.01)** (2.29)** (2.20)** (2.09)** (2.31)** (0.21) (0.14) (0.76) NUMFRAC (2.29)** (14.7)** (3.63)** MCAPFRAC (9.68)** (12.5)** (8.25)** VOLFRAC (5.07)** (8.29)** (22.37)**

15 By Type of Listing (New Table 5) Model (1) (2) (3) (1) (2) (3) (1) (2) (3) MKTGDP Rule 144a Private Placements OTC Listings Nasdaq/NYSE Listings OPENNESS (-0.47) (-0.52) (-0.60) (0.55) (-0.27) (-0.43) (-0.61) (-0.76) (-0.58) LDATE (1.23) (1.16) (1.08) (0.40) (1.02) (1.19) (1.23) (1.23) 91.21) NUMFRAC (-1.44) (-1.36) (0.44) MCAPFRAC (0.26) (-1.36) (1.49) VOLFRAC (0.94) (-0.27) (1.14) NUMGDP Rule 144a Private Placements OTC Listings Nasdaq/NYSE Listings OPENNESS (-4.06)** (-4.62)** (-4.50)** (-3.14)** (-3.17)** (-4.27)** (-3.83)** (-2.91)** (-3.99)** LDATE (-2.43)** (-2.83)** (-2.69)** (-2.69)** (-2.06)** (-2.31)** (-2.65)** (-2.75)** (-2.91)** NUMFRAC (-1.32) (-2.08)** (-3.86)** MCAPFRAC (1.39) (-0.69) (-4.47)** VOLFRAC (1.29) (-1.36) (-3.38)**

16 Empirical Methodology II How to measure time-varying market integration? Errunza-Losq (1985) IAPM nesting integration/segmentation: E(R i )= r f + A M W Cov(R i,r w ) + (A I -A) M I Cov(R i,r I R e ) Aggregating across all stocks i in I, E(R I )= r f + A M W Cov(R I,R w ) + (A I -A) M I Var(R I R e ) We assign ADRs (R ADR ) as globally eligible securities (R e ) Integration Index (II) = 1 - [Var(R I R ADR )/Var(R I )] measures substitutability of local market by subset of stocks traded as ADRs Complete integration: II = 1 or Var(R I R ADR ) = 0 Complete segmentation: II = 0 or Var(R I R ADR ) = Var(R I ) Using Var(R i R e )=Var(R i ) (1 - ρ I,ADR2 ), my proxy measure is: II = 1 - [(1 - ρ I,ADR2 )]

17 Empirical Methodology II Follow approach of Errunza et al (2000), Carrieri, et al (2001) using a multivariate GARCH-M model that allows time-varying world price of covariance risk (δ w,t-1 ) and local market price of risk (λ i,t-1 ) to explain local national index returns (r i,t ): r i,t = δ w,t-1 cov t (r i,t,r w,t ) + λ i,t-1 var(r i,t r ADR,t ) + ε i,t r ADR,t = δ w,t-1 cov t (r ADR,t,r w,t ) + ε ADR,t r W,t = δ w,t-1 var t (r w,t ) + ε W,t Prices of world covariance and local market risk are specified by: δ w,t-1 = exp(κ w Z t-1 ) λ i,t-1 = exp(κ i Z t-1 )

18 Empirical Methodology II Instrumental variables (Z t-1 ) include a constant, the local and world dividend yields, local exchange rate versus the U.S. dollar, and the U.S. 10-year Treasury bond yield Law of motion for the time-varying conditional covariance is parameterized using the Ding-Engle (1994) specification: H t = H 0 * (ιι aa bb ) + aa * {ε t-1 ε t-1 } + bb * H t-1 The model is estimated by quasi-maximum likelihood using the Berndt, Hall, Hall, Hausmann maximization technique with Bollerslev-Wooldridge standard errors Table 6 Summary Statistics Table 7 Specification Tests, Residual Diagnostics

19 Time-Varying Market Integration Indonesia Chile Integration Index Mexico Correlation 1.00 Integration Index Taiwan Correlation Integration Index Correlation Integration Index Correlation

20 SURM Tests (Table 9) Model (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Integration Index Conditional Correlation with World Market Returns OPENNESS (5.81)** (3.23)** (2.99)** (3.62)** (2.84)** (2.61)** (2.42)** (2.77)** LDATE (3.92)** (2.41)** (1.93)* (2.11)** (-0.95) (-1.77)* (-2.83)** (-2.11)* NUMFRAC (2.58)** (2.04)* MCAPFRAC (4.55)** (3.26)** VOLFRAC (2.55)** (2.01)*

21 Future Work What do we learn? Growth of ADR markets has facilitated integration of emerging equity markets with world markets, but, at the same time, has had a deleterious effect on domestic stock market development The paper is an important contribution to the literature on market liberalizations and to growing literature on overall net benefits of international cross-listings; it also has important policy implications There are a number of limitations of the current study: Only twelve emerging markets what about developed markets? Need further specification tests - concern about model mis-specification errors and how influences integration index, other instrumental variables? Need to broaden measures of ADR activity - home versus U.S. trading activity, fraction of shares in ADR form Benchmark regression tests with other factors influencing integration - macroeconomic/business cycle variables, access to capital, business formation Need to consider other markets where firms list abroad (not just U.S.)