Scribd
Upload
14 views10 pages

4 CAPM - Appendix B

Uploaded by

El
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
14 views10 pages

4 CAPM - Appendix B

Uploaded by

El
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

4

The Capital Asset Pricing Model


Appendix A. Arbitrage Pricing

Pág. 1
06 | Arbitrage pricing

Arbitrage Pricing Theory (APT)

• As we already pointed out, the practical difficulties with mean-variance analysis have
motivated shortcuts to find optimal portfolios. We have analyzed the equilibrium approach of
the CAPM.
• An alternative approach is to assume a factor structure in the variance-covariance matrix of
risky returns, effectively reducing the dimension of the portfolio choice problem from N
(number of assets) to K (number of pervasive factors that affect all assets) with K <<N.
• This idea represents the core of the Arbitrage Pricing Theory (APT) due to Stephen Ross.

Financial Economics – The Capital Asset Pricing Model Pág. 2


06 | Arbitrage pricing

APT in a Single Factor Model


• Suppose that we run the regression of the market model with excess returns,
𝑒 𝑒
𝑅𝑗𝑡 = 𝛼𝑗 + 𝛽𝑗𝑚 𝑅𝑚𝑡 + 𝜀𝑗𝑡

• Suppose further that the errors in this equation are uncorrelated across assets
𝐸 𝜀𝑗𝑡 𝜀𝑖𝑡 = 0, j ≠ 𝑖
• Then, the residual in any stock is idiosyncratic, unrelated to the residual risk in any other
stock.
• These assumptions imply that the covariances are easily estimated from mean-variance
analysis because
𝑒 𝑒 2
𝐶𝑜𝑣 𝑅𝑗𝑡 , 𝑅𝑖𝑡 = 𝛽𝑗𝑚 𝛽𝑖𝑚 𝜎𝑚

Financial Economics – The Capital Asset Pricing Model Pág. 3


06 | Arbitrage pricing

APT in a Single Factor Model


• Instead of N(N+1)/2 parameters in the unrestricted variance-covariance matrix, only N+1
parameters (N betas plus the variance of the market return) need to be estimated.
• An even more striking implications of these assumptions is that if many assets are available,
we should expect αj typically to be very small. This is the APT.
• To understand why this is the case, consider forming a portfolio of N assets. The excess
return on the portfolio will be
𝑒 𝑒
𝑅𝑝𝑡 = 𝛼𝑝 + 𝛽𝑝𝑚 𝑅𝑚𝑡 + 𝜀𝑝𝑡
𝑁 𝑁 𝑁
𝛼𝑝 = ෍ 𝜔𝑗 𝛼𝑗 , 𝛽𝑝𝑚 = ෍ 𝜔𝑗 𝛽𝑗 , 𝜀𝑝𝑡 =෍ 𝜔𝑗 𝜀𝑗𝑡
𝑗=1 𝑗=1 𝑗=1

Financial Economics – The Capital Asset Pricing Model Pág. 4


06 | Arbitrage pricing

APT in a Single Factor Model


• The variance of εpt will be
𝑁
𝑉𝑎𝑟 𝜀𝑝𝑡 = ෍ 𝜔𝑗2 𝑉𝑎𝑟 𝜀𝑗𝑡
𝑗=1

• This variance will shrink rapidly with N provided that no single weight is too large. For the
benchmark case where the portfolio is equally-weighted we get
2 𝑁
1
lim 𝑉𝑎𝑟 𝜀𝑝𝑡 = ෍ lim 𝑉𝑎𝑟 𝜀𝑗𝑡 → 0
𝑁→∞ 𝑁 𝑗=1 𝑁→∞

• We say that the portfolio is well diversified. For such a portfolio, we can neglect the residual
(the idiosyncratic component of the return) and write the excess return as
𝑒 𝑒
𝑅𝑝𝑡 = 𝛼𝑝 + 𝛽𝑝𝑚 𝑅𝑚𝑡

Financial Economics – The Capital Asset Pricing Model Pág. 5


06 | Arbitrage pricing

APT in a Single Factor Model


• But then we must have αp = 0.
• If not, there is an arbitrage opportunity: go short β units of the market and go long one unit of
portfolio p, while funding all positions with risk-free borrowing and lending. This delivers a
riskless excess return of αp.
• The APT exploits this insight, showing that αp = 0 for all well diversified portfolios implies that
“almost all” individual assets have αj very close to zero.
• Intuitively, a few assets can be mispriced (have nonzero alpha), but if too many assets are
mispriced then one can group them into a well diversified portfolio and create an arbitrage
opportunity.

Financial Economics – The Capital Asset Pricing Model Pág. 6


06 | Arbitrage pricing

APT in a Single Factor Model


• Since the alpha is zero for almost all individual assets, then the CAPM holds without using the
apparatus of the mean-variance analysis.
• However, the result is weaker than the CAPM because we have only a limiting result that
nonzero alphas are rare, not the exact result that all alphas are zero.
• The major difficulty is that the assumption of uncorrelated residual risk across assets is
counterfactual.
• There are other many sources of common variation in stock returns besides the market:
cyclical stocks are particularly sensitive to changes in industrial production, highly leveraged
firms are very sensitive to interest rates, and firms within the same industry o similar size also
tend to move together.

Financial Economics – The Capital Asset Pricing Model Pág. 7


06 | Arbitrage pricing

APT in Multifactor Models


• To handle this, we can generalize the previous analysis to a multifactor model. If there K
factors (portfolios) capturing the common influence of K underlying sources of risk, and if
riskless borrowing and lending is possible, we have
𝐾
𝑒 𝑒
𝑅𝑗𝑡 = 𝛼𝑗 + ෍ 𝛽𝑗𝑘 𝑅𝑘𝑡 + 𝜀𝑗𝑡 (4.17)
𝑘=1

• We assume that the residual risk is uncorrelated across stocks. The prediction of the model is
again that αj = 0 for almost all stocks. This is restrictive if K << N.
• The generality of the APT is appealing, but the approach has several weaknesses

Financial Economics – The Capital Asset Pricing Model Pág. 8


06 | Arbitrage pricing

APT in Multifactor Models


• First, the model predicts that nonzero alphas are rare in the limit of a large cross-section, but
it is not clear how this prediction can be falsified in a finite cross-section.
• Second, ex post we know that we can always get a single factor model to fit the data by
choosing the ex post mean variance efficient portfolio as a single factor. It must then be easier
to get a K-factor model to fit the data. This does not tell us anything about the financial
markets unless we can some confidence that the K-factor model holds ex ante as well as ex
post. We need theoretical reasons to believe that the K-factor model is structural.
• Third, the APT does not determine the signs or magnitudes of the risk prices. Being a
common factor is a necessary but not sufficient condition to be a priced factor that helps
determine the cross-section of average returns. We need equilibrium arguments, not only
limiting arbitrage arguments.
Financial Economics – The Capital Asset Pricing Model Pág. 9
Campus Almansa | Calle Almansa 101 | 28040 Madrid

cunef.edu
Pág. 10

You might also like