2. Capital Asset pricing Model

Dr. Youchang Wu

WS 2007

Asset Management Youchang Wu 1 in the presence of a -free asset

Asset Management Youchang Wu 2 line

ƒ When a risk-free asset exists, i. e., when a capital market is introduced, the efficient frontier is linear. ƒ This linear frontier is called ƒ The capital market line touches the efficient frontier of the risky assets only at the tangency ƒ The optimal portfolio of any with mean-variance preference can be constructed using the risk-free asset and the tangency portfolio, which contains only risky assets (Two-fund separation ) ƒ All with a mean-variance preference, independent of their risk attitudes, hold the same portfolio of risky assets. ƒ The risk attitude only affects the relative weights of the risk-free asset and the risky portfolio

Asset Management Youchang Wu 3 Market price of risk

ƒ The equation for the capital market line

E(rT ) − rf E(rP ) = rf + σ (rp ) σ (rT ) ƒ The slope of the capital market line represents the best risk-return tdtrade-off th at i s avail abl e on the mar ke t ƒ Without the risk-free asset, the marginal rate of substitution between risk and return will be different across investors ƒ After i nt rod uci ng th e ri s k-free asset , it will be the same for a ll investors. ƒ Almost all investors benefit from introducing the risk-free asset (capital market) ƒ Reminiscent of the Fisher Separation Theorem?

Asset Management Youchang Wu 4 Remaining questions

ƒ What would be the tangency portfolio in equilibrium? ƒ CML describes the risk-return relation for all efficient portfolios, but what is the equilibrium relation between risk and return for inefficient portfolios or individual assets? ƒ What if the risk-free asset does not exist?

Asset Management Youchang Wu 5 CAPM

ƒ If everyb bdody h hldholds the same ri iksky portf fliolio, t hen t he r ikisky port flifolio must be the , i.e., the value-weighted portfolio of all securities (demandβ must equal supply in equilibrium!). ƒ It follows that the market portfolio is a frontier portfolio. ƒ According to Property III of portfolio frontierβ (discussed last time), we must have

2 E(rj ) = rf + j [E(rm ) − rf ], where j = COV(rj ,rm ) /σ (rm )

ƒ This is the famous CAPM independently derived by Treynor(1961), Sharpe (1964), Lintner(1965) and Mossin (1966)

ƒ E(E(rm)-rf is ca lldlled mar ktikket risk prem ium. An asse t‘s r ikisk prem ium is given by its time the premium.

Asset Management Youchang Wu 6 Underlying assumptions

ƒ Mean-variance preference ƒ Homogeneous beliefs among investors regarding the planning horizon and the distribution of security returns ƒ No market frictions: no transaction costs, no tax, no restriction on selling, no information cost etc.

Asset Management Youchang Wu 7 Intuition for CAPM

ƒ Since everybody holds the market portfolio , the risk of an individual asset is characterized by its contribution to variance of the market portfolio instead of its own variance. ƒ The contribution of an individual security to the variance of market portfolio is determined by its covariance with the market portfolio

nn 2 2 ∂σ (rm ) σ (rm ) = ∑ ∑ wi w jCOV (ri , rj ) => = 2COV (ri , rm ) ij==11 ∂wi ƒ The part of an asset ‘s risk that is correlated with the market portfolio, the , cannot be diversified away; thus, investors need to be compensated for bearing it. ƒ The part of an asset ’s risk that is not correlated with the market portfolio, the unsystematic risk, can be diversified away; thus, bearing unsystematic risk need not be rewarded, and therefore, an asset’s unsystematic risk does not affect its risk premium.

Asset Management Youchang Wu 8 CAPM vs Property III

ƒ Property III is a mathematical property of portfolio frontier that holds for any return distribution. ƒ CAPM is an asset pricing relation derived using economic reasoniItifihikdting. It specifies how risk and return are re ltdlated in equilibrium. ƒ Asset returns under CAPM are not exogenously given. They are endogenous ly de term ine d in mar ke t equ ilibr ium. – If the CAPM does not hold, i.e., market portfolio is not mean-variance efficient, then demand does not equal supplfly for some asset s. – The prices of such assets as well as other assets must change, therefore the whole return distribution will change – This process will continue until we converge to an return distribution that makes the market portfolio mean-variance efficient.

Asset Management Youchang Wu 9

E(r) E(r) Capital market line Security market line

E(rm) E(rm)

rf rf

σ β β=1

Asset Management Youchang Wu 10 SML vs CML

ƒ Capital market line describes the efficient frontier in the presence of a risk-free asset. It specifies the equilibrium relation between return and total risk of efficient portfolio. ƒ Security market line describes the equilibrium relation between return and systematic risk for all assets or portfolios

Asset Management Youchang Wu 11 Portfolio beta

ƒ CAPM ho lds bo th for in div idua l asse ts an d portfolios of assets ƒ The beta of a portfolio is simply the weiggghted average beta of each individual assets in the portfolio, this is because n n COV(rp , rm ) = COV(∑wi ri , rm ) = ∑wiCOV(ri , rm ) i=1 i=1

where wi is the weight of asset i in portfolio p.

Asset Management Youchang Wu 12 Zero-beta CAPM

ƒ What if there is no risk-free asset? ƒ Each investor will hold a different frontier portfolio deppgending on his own risk attitude ƒ According to Property I of portfolio frontier, the aggregate of each investor‘s portfolio, i.e., the market portfolio, is also a frontier portfolio ƒ By Property III of portfolio frontier, it follows that the linear relation between and beta with respect to the market portfolio must hold. ƒ This argument leads to the zero-beta CAPM (Black 1972):

E(rj ) = E(rz ) + β j [E(rm ) − E(rz )]

where rz is the return of a zero-beta portfolio

Asset Management Youchang Wu 13 Implications for investment

ƒ CAPM sol ves three mos t i mpor tant i ssues i n inves tmen t simultaneously: – Where to invest? Invest in the market portfolio and risk -free asset! – How to value an asset? Estimated the beta of this asset and discount all the future cash flows generated by this asset using a discount rate given by CAPM! – How to evaluate investment performance? Adjust the performance by beta! ƒ Not surprisingly this is regarded as one of the greatest results in finance ƒ But how does it fit the real world? ƒ Numerous studies have tried to answer this question

Asset Management Youchang Wu 14 Earlier empirical tests of CAPM

ƒ Cross-sectional test – First estimate the beta by running the following time-serise regression rjt = α j + β j rmt + ε jt – Then run the following (out of sample) cross-sectional regression

rj = γ 0 + γ 1β j + γ 2CHAR j + e j where CHAR is a charateristic of j unrelated to CAPM such firm size, idiosyncratic risk

– CAPM predicts that γ0=risk free rate, γ1 =market risk premium, and γ2 =0 – Tests are usually based on portfolio returns to avoid measurementititdbtt errors in estimated betas. – Supportive results are reported by Fama and MacBeth (1974)

Asset Management Youchang Wu 15 Earlier empirical tests of CAPM (2)

ƒ Time series test

rjt − rf = α j + β j (rmt − rf ) + ε jt

ƒ Prediction of CAPM: αj is zero for every stock or portfolio ƒ Black, Jensen and Scholes (1972) reject the standard CAPM in favor of the zero-beta CAPM

Asset Management Youchang Wu 16 Roll´s critique

ƒ Roll (1977) argues that CAPM is inherently untestable – The only economic prediction of CAPM is that the market portfolio is mean-variance efficient – The linear return/beta relation can be found in any sample irrespective of how returns are determined in the market (All we need is to identify an index portfolio which is ex post efficient) – All existing tests only tell us whether the market proxy used by researchers are efficient or not, they say nothing about the efficiency of the market portfolio itself – A true market portflio should include all assets (human capital etc) and is unobservable, therefore CAPM is untestable

Asset Management Youchang Wu 17 Roll´s critique (2)

ƒ Roll (1978) further argues that using CAPM to measure performance is problematic – If performance is measure relative to an index that is ex post efficient, then from PIII of porftolio frontier, all portfoli os will be on the secur ity mar ke t line (no performance) – If performance is measured relative to an ex post inefficient index, then any ranking of portfolio ppppgerformance is possible depending on which inefficient index has been chosen

Asset Management Youchang Wu 18 Findings from more recent tests

ƒ The rel a tion be tween be ta an d re turn is wea k a t best (Fama and French 1992) ƒ Size eff ect : small s tock s out perf orm l arge st ock s (Banz 1981,Fama and French 1992) ƒ BkBook-to-marke t eff ect : st ock s with hi gh b ook -to- market equity ratios outperform with low book-to-market ratios (Fama and French 1992) ƒ effect: Stocks outperforming in the last 3 -12 months tend to outperform in the following 3-12 months (Jegadeesh and Titman 1993).

Asset Management Youchang Wu 19 Possible explanations for empirical shortcomings of CAPM ƒ MtMeasurement errors – Measurement errors in beta – Measurement errors in expected return (survivorship bias) ƒ Behavioral biases of investors – Small investors are subject to many behavioral biases – Institutional investors are subject to agency issues ƒ Missing risk factors – People do not just hold bonds and stocks – Therefore they do not just care about an asset‘s covariance with a market index – They also care about its covariance with other components (such as human capital) of their „true“ portfolio – This means we have to consider other risk factors as well

Asset Management Youchang Wu 20 Summary

ƒ CAPM is an elegant model that gives simppyle answers to several key issues in finance ƒ The real prediction of CAPM is that the market portfolio is mean-variance efficient. ƒ Empirical results should be interpreted with great caution in the light of Roll´s critique

Asset Management Youchang Wu 21