Topic 5: International Asset Pricing Models Flashcards
Why study?
i) Why some strategies are profitable? Risk premium? systematic risk? Need asset pricing model to determine “systematic risk/risk premium
ii) To answer practical questions (investing into a hedge fund? A new profitable strategy? Factor investing?)
Domestic Asset Pricing Models
CAPM
Fama French 3 factor model
Carhart 4 factor model
Fama French 5 factor model
How to test CAPM
E(rj,t) − rf,t = βj(E(rm,t) − rf,t)
1: run regression using historical data to estimate Alpha and Beta
2: Take expectation on the regression: E(rj,t) − rf,t = αj + βj(E(rm,t) − rf,t)
3: CAPM holds when α = 0. We test this hypothesis statistically. If α ≠ 0, the CAPM doesn’t hold (so-called CAPM anomalies)
CAPM anomalies
Size anomaly
Value anomaly
Size anomaly
Small firm stocks tend to have higher HPRs, even after accounting for the CAPM beta risk
How to implement: buy a portfolio of small firms (S) but you finance it by short-selling a portfolio of large firms (B)
Value anomaly
Value firm stocks tend to have higher returns, even after accounting for the CAPM beta risk
How to implement: buying a portfolio of value firms (H) but you finance it by short-selling a portfolio of growth firms (L) = HML
Fama-French-3-factor model
(‘alpha’ become ‘beta’):
E(Rj,t) − Rf,t = βj(E(Rm,t ) − Rf,t) + Sj E(SMBt)+ HjE(HMLt)
Fama-French Anomalies
Momentum Anomaly
Profitability Anomaly
Investment Anomaly
Momentum Anomaly
Past winners tend to have higher HRPs, even after accounting for the FF 3F risks
How to implement: buy a portfolio of past winners (W) but you finance it by short-selling a portfolio of past losers (L)
Profitability anomaly
Past profitable firms stocks tend to have higher returns, even after accounting for the FF 3F risks
How to implement: buy a portfolio of past robust profitable firms (R) but you finance it by short-selling past weak profitable firms (W)
Investment anomaly
Past low investment firm stocks tend to have higher returns, even after accounting for the FF 3F risks
How to implement: buy past conservative firms (portfolio C) but you finance it by short-selling past aggressive firms (portfolio A)
Carhart 4-Factor Model
(Alpha become Beta):
E(Rj,t) - Rf,t = βj(E(Rm,t ) − Rf,t) + Sj E(SMBt)+ HjE(HMLt) + WjE(WMLt)
Fama-French 5 factor model
(Alpha become Beta):
E(Rj,t) - Rf,t = βj(E(Rm,t ) − Rf,t) + Sj E(SMBt)+ HjE(HMLt) + R*jE(RMWt) + CjE(CMAt)
International Asset Pricing Models
World CAPM
CAPM with Partial Integration
Bakaert and Harvey (1995s) model
International APT
International APT with partial integration
World CAPM
E(rj,t) − rf,t = βj(E(rw,t) − rf,t)
with fully integrated markets and no premiums for taking FX risk (real returns + RPPP holds)
CAPM with partial integration
Partially integrated? Psychological barriers, Legal restrictions, Transaction costs, Tax, Political risks, FX risks, etc
Bekaert and Harvey (1995)’s model
Bekaert and Harvey (1995)’s model:
E(rj,t) − rf,t = wt[βj,w(E(rw,t) − rf,t)] + (1 − wt)[βj,l(E(rl,t) − rf,t)]
Note: Weights, wt, are determined by variables that proxy for the degree of integration, like size of trade sector and equity market capitalisation to GDP
International APT
(local/regional/global anomalies: LSMB (e.g., UK-SMB), RSMB (e.g., Asian-SMB), GSMB (e.g., MSCI-SMB))
International APT with partial integration
Fama-French 3-factor model with partial integration:
E(rj,t) − rf,t = βj,w(E(rw,t) − rf,t) + sj,GE(GSMBt)+ hj,GE(GHMLt) + βj,l(E(rl,t) − rf,t) + sj,lE(LSMBt)+ hj,lE(LHMLt)
Repeat for Carhart 4 factor and FF-5F with partial integration
Karolyi and Wu (2018)’s 4-Factor model
Carhart 4-factor model with partial integration, but with different ways to define global/local factors
Investability Restriction: Karolyi and Wu (2014,WP) define globally-accessible stocks as follows: the stocks in the globally accessible sample need to be listed in the markets which are fully open to global investors or to be secondarily cross-listed in target markets (US, UK, etc.) with minimum foreign investment restrictions and reasonably active trading in foreign cross-listed issues).
They propose and test multi-factor models (e.g., Carhart with partial integration) based on factor portfolios comprised on only globally-accessible stocks (global factors) and of locally-accessible stocks (local factors)