Eric Zivot
Key points:
σ 2100 = σ 299
Case 3
E[Ri] = μi = f (β i,p)
The Capital Asset Pricing Model (CAPM) formalizes this conjecture.
2 Sharpe’s Single Index Model
• Random error term εit captures “firm specific” news unrelated to market-
wide news.
• μi = E[Rit] = αi + β iμM
• σ 2i = var(Rit) = β 2i σ 2M + σ 2ε,i
• σ ij = cov(Rit, Rjt) = σ 2M β iβ j
Derivations:
σ ij = cov(Rit, Rjt)
= cov(αi + β iRMt + εit, αj + β j RMt + εjt)
= cov(β iRMt, β j RMt) + cov(β iRMt, εjt)
+cov(β j RMt, εit) + cov(εit, εjt)
= β iβ j cov(RMt, RMt)
= σ 2M β iβ j
Implications:
cov(Rit, RMt) σ iM
βi = = 2
var(RMt) σM
β i captures the contribution of asset i to the variance/risk of the market index.
Derivation:
σ 2i = var(Rit) = β 2i σ 2M + σ 2ε,i
total risk = market risk + non-market risk
Divide both sides by σ 2i
β 2i σ 2M σ 2ε,i
1 = 2 + 2
σi σi
= Ri2 + 1 − Ri2
where
β 2σ 2
Ri2 = i 2M = proportion of market risk
σi
1 − Ri2 = proportion of non-market risk
Sharpe’s Rule of Thumb: A typical stock has R2 = 30%; i.e., proportion of
market risk in typical stock is 30% of total risk.
2.1.3 Return Covariance Matrix
3 asset example
where
2 asset example
αp = x1α1 + x2α2
β p = x1β 1 + x2β 2
εp,t = x1ε1t + x2ε2t
2.2.1 SI Model with Large Portfolios
2
• var(Rp,t) = β̄ var(RMt) : Magnitude of portfolio risk is proportional to
market risk. Magnitude of portfolio risk is determined by portfolio beta β̄