This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Primbs/Investment Science 1 Topic #14 The Arbitrage Pricing Theory (APT) Reading: Luenberger Chapter 8, Sections 1 – 4 Primbs/Investment Science 2 Models, Data, and APT Models, Data, and APT Factor Models Relation to CAPM The Arbitrage Pricing Theory Means and Variances Well Diversified Portfolios APT and alpha Primbs/Investment Science 3 The difficulty with meanvariance The meanvariance approach requires that you estimate the mean, variance, and covariance of all securities in your portfolio. This can be a huge number. For n stocks, we need n means, n variances, and n(n1)/2 covariances. (for 1000 stocks, this is 501500 values to be estimated!) Clearly, we need an approach that requires less computation. Primbs/Investment Science 4 Factor Models Rates of return are related to factors (which are random quantities). i i i i e f b a r + + = Consider the following model: random factor f constants b i is called the factor loading e i is a random error term with E[ e i ]=0. Primbs/Investment Science 5 Factor Models i i i i e f b a r + + = Factors models correspond to using a linear regression to fit data. r i f x x x x x x x a i slope b i yintercept ( factor loading ) e i x x x x x x x Primbs/Investment Science 6 MultiFactor Models i i i i i e f b f b a r + + + = 2 2 1 1 j j j j j e f b f b a r + + + = 2 2 1 1 We can also consider models where rates of returns are determined by more than a single factor. These are called multifactor models. Assumption We will assume that the factors f 1 , f 2 , ... are uncorrelated random variables. Furthermore, we will assume that the error terms, e i , e j are uncorrelated with each other and with the factors. That is Cov(e i ,e j ) = 0 and Cov(e i ,f j ) = 0 . Primbs/Investment Science 7 What are possible factors? External Factors: GDP, PPI, CPI, Unemployment, Interest Rates, etc... Extracted Factors: Market Portfolio, Industry Averages, etc... (One can also include firm specific factors such as size, price/book, etc. However, in the following APT theory only factors that apply to many firms will be relevant.) Primbs/Investment Science 8 Models, Data, and APT Models, Data, and APT Factor Models Relation to CAPM The Arbitrage Pricing Theory Means and Variances Well Diversified Portfolios APT and alpha Primbs/Investment Science 9 Calculation of Mean and Variance (Single Factor Model) The mean of a return is given by: f b a i i + = ] [ ] [ i i i i i e f b a E r E r + + = = ] [ ] [ i i i e E f E b a + + = Known: E(e i )=0 and cov(f,e i )=0 . ) ( i i i i e f b a r + + = Primbs/Investment Science 10 Calculation of Mean and Variance (Single Factor Model) Known: E(e i )=0 and cov(f,e i )=0 . The variance is given by: ] )) ( ) [(( ] ) [( 2 2 2 f b a e f b a E r r E i i i i i i i + + + = = σ ] ) ) ( [( 2 i i e f f b E + = ] ) ( 2 ) ( [ 2 2 2 i i i i e e f f b f f b E + + = ) var( ) , cov( 2 ) var( 2 i i i i e e f b f b + + = ) var( ) var( 2 i i e f b + = 2 2 2 i e f i b σ σ + = ) ( i i i i e f b a r + + = Primbs/Investment Science...
View
Full
Document
This note was uploaded on 01/24/2009 for the course MS&E 242 taught by Professor Primbs during the Fall '06 term at Stanford.
 Fall '06
 PRIMBS

Click to edit the document details