## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Session 08

**Outline 08: Market and Index Models
**

1. 2. 3. 4. CAPM and Multifactor Models Bloomberg Screen Components of Variance Diversification Effects 5. 6. 7. 8. Precision of Historical Betas Adjusting Historical Betas Estimating Expected Returns Market-Neutral Strategies (Problem Set)

Mandatory Reading: Hansell´s article.

**CAPM and Multifactor Models
**

• In the previous section we saw that in a CAPM framework the only variable that matters for asset pricing is the return of the market portfolio. • Consider the following statistic representation of asset returns:

Rj = E Rj 1 2 3

( )

**+ innovation in j's return 1444 4 24444 3
**

new information

expected component

• We can also distinguish two broad types of innovations: systematic (generated by an arbitrarily large set of factors, K) and idiosyncratic.

R j = E R j + β j1F1 + β j 2 F2 + .... + β jK FK + 14444 4 244444 3

systematic innovation (aggregate risk factors )

( )

εj

idiosyncratic innovation (unique risk )

{

**• We end up with the following linear factor model of asset returns:
**

R j = a j + β j1 F1 + β j 2 F2 + .... + β jK FK + ε j

Copyright © José M. Marín 1 Portfolio Management

1 The label market model is often used for a model in which only (1) and (2) hold and the term single index model for the model in which (3) also holds. …. N Eε j RM = Eε j = 0. Marín Portfolio Management . the Market Model1 is defined as: R j = a j + β j RM + ε j . for instance..Market and Index Models Session 08 • Now we can interpret the CAPM as a model in which returns are generated by a single factor: R j = a j + β j1F1 + β j 2 F2 + . 2 Copyright © José M.. We will just use the latter. the model used by Bloomberg. ∀j Eε iε j = 0 • This is. + β jK FK + ε j 14444 4 244444 3 β j RM • More specifically. j = 1..

Marín 3 Portfolio Management .Market and Index Models Session 08 Market Model in the industry: Bloomberg screen Copyright © José M.

Ri . α i + β i Rm.t } = cov{Ri .t } = market-related variance (1) 142 4 43 4 β i2 var{Rm.t } 1 4 2 4 3 • The R2 (“R-squared”) gives the fraction of the asset’s total variance that can be attributed to its marketrelated variance: Copyright © José M.t .t + ε i .Market and Index Models Session 08 Components of Variance • Consider the Market Model: R j = a j + β j RM + ε j • The market model provides a decomposition of the variance of an asset’s return into two components o market-related o non-market • Using the regression.t } + non -market variance var{ε i . Marín 4 Portfolio Management .t .t + ε i . we obtain var{Ri .t } = cov{α i + β i Rm.

324 Var ( RCB ) o std.t } var{Ri . dev. dev.t} = 5.t } var{Ri .{RCB.{εCB. Marín 5 Portfolio Management . Copyright © José M.Market and Index Models Session 08 R = 2 β i2 var{Rm.t } =1− (2) • In the Bloomberg example: o β = 0.43% • Beta is not the same as R2: high-beta stocks can actually have lower correlations with the market.865 o std.04476) 2 = 0.t} = 4.476% o R2=0.t } var{ε i .324 o Using (2): R2 = 1− (0.

t .Market and Index Models Session 08 Diversification Effects • A market-model regression also holds for a portfolio of assets. Marín 6 Portfolio Management . i =1 N ε p .t = ∑ xi ε i . • The single-index model implies that the variance of returns on well-diversified portfolios will consist primarily of the market-related component.t + ε i . where α p = ∑ xiα i .t + ε p . To see how: o Begin with a market-model regression for each of N assets in the portfolio o Multiply both sides of the regression equation for asset i by the portfolio’s weight in asset i.t ) i =1 N N n =1 R p .t = α p + β p Rm.t = ∑ xi (α i + β i Rm. Copyright © José M.t . xi o Sum the resulting equations across the N assets ∑ xi Ri . i =1 N i =1 N β p = ∑ xi β i .

1 var{ε i .t }.t } + 2 → βp var{Rm. N which approaches zero as N grows large.t } N and the market-model R2 for the portfolio approaches 1.Market and Index Models Session 08 • For example.t } = β p var{Rm. Marín 7 Portfolio Management .t } = var ⎨ ∑ ε i . • The single-index model also implies that well diversified portfolios will be highly correlated with each other.t ⎬ ⎩i =1 N ⎭ 1 = var{ε i . the variance of the total portfolio return is 2 var{R p . Therefore.t }. Copyright © José M. with an equally weighted portfolio ( xi = 1/N ). ⎧N 1 ⎫ var{ε p .

or using portfolios instead of individual securities. a sample period of five to seven years gives the optimal tradeoff between greater accuracy arising from a larger number of observations and lower accuracy arising from changes over time in the true underlying beta. The standard error of β i (approximately) by 2 ˆi ˆ = 1− R σ sβ i ˆm T σ ( ) (3) • Thus.05 0. observations (T). in estimating historical betas using monthly returns. • Precision in historical betas can.93 0. denoted β i ˆ is given random variable.” A historical beta is a • The OLS regression estimator of βi.03 • Some experiments suggest that. thereby raising R2 and lowering σ • For example. (3) suggests that the precision of an OLS beta estimate can be increased by increasing the number of ˆi . the equally weighted portfolio of the 30 Dow (1993-99.Market and Index Models Session 08 Precision of Historical Betas ˆ . Copyright © José M. 84 months): Beta (raw) R2 Std. is often referred to as the “historical beta. since it is computed from realizations of random returns. in principle. Error of Beta Jones Industrial stocks gives the following results 1. be improved by increasing the frequency of return observations. Marín 8 Portfolio Management .

(1) and β i . ˆ ˆ • As the plot suggests.( 2 ) closer to 1. ˆ ˆ β i . ˆ ˆ • For a number of assets (i’s). where it must be the case that Copyright © José M.(1) tend to have values of β i . Marín 9 Portfolio Management .( 2 ) = c0 + c1 β i .( 2 ) .Market and Index Models Session 08 Adjusting Historical Betas for “Regression to the Mean” ˆ ˆ • Denote the historical betas on asset i estimates in two adjustment five-year sample period as β i . the assets with extreme values (large or small) of β i .(1) versus β i .( 2 ) . • This “regression” tendency can be represented by the fitted line in the above plot. plot β i .(1) .

is formed by ˆ ˆ β ADJ . Dev. you estimate the mean and volatility of the return on a given stock.5% 0. Bloomberg) sets w = 1/3. (s) 13. . Mean Ri rm 1. The “adjusted” beta estimator. Ri. Estimating Expected Returns: Why Bother with Beta? • Based on the past 84 months of data.t − R f .t . the “raw” historical betas are often adjusted toward 1 to account for this regression tendency.5 Std.g.i = w ⋅ 1 + (1 − w) ⋅ β i .. Copyright © José M.t. • A typical specification (e.3% 4.8 • Using the simple historical average return.Market and Index Models Session 08 c0 + c1 = 1. Marín 10 Portfolio Management . • In practice. rm = RSP . and the return on the S&P in excess of the riskless (T-Bill) rate. you would estimate the stock’s expected return as 1.5%.

and the stock’s estimated beta equals 1.41(0.Market and Index Models Session 08 • Assuming that the current T-Bill return is 0.3 = = 1. and the CAPM could be wrong.3% monthly.error = std.01% • The CAPM approach combines estimates of two unknown quantities: the expected excess market return and beta.76% std.45% T 84 • For CAPM-based estimates: ˆ ⋅ r = 0. β m ( ) • Why? – Greater precision in the estimated market risk premium. Marín 11 Portfolio Management . dev. estimation error in β Copyright © José M.5) = 1. the CAPMbased estimate is Ei* = 0. rm . Is it worth the trouble? • Compare the standard errors of the two estimates: greater precision ⇔ lower standard error • For historical mean: std. error = s 13.3 + 1. outweighs the effect of additional ˆ.41.

Market and Index Models Session 08 • Allows substantial error in the CAPM pricing relation before the CAPM-based estimate of expected return becomes less precise. Copyright © José M. • Even greater differences arise between approaches if the market equity premium is based on a longer sample period than used for the individual stock’s average return. Marín 12 Portfolio Management .

“The Other Side of Zero. The following questions are based on the assigned reading. With a legitimate benchmark of the T-bill rate.” Copyright © José M. but 5 to 10 percent. what is the volatility of the return on the market-neutral position? Assume that the annual rate of return on the S&P 500 index has a volatility of 16%. no beta”. ‘The two alphas represent a form of leverage.2 and an annual volatility (standard deviation) of 30% with (2) a short position in S&P 500 index futures.Market and Index Models Session 08 Market Neutral Strategies (Read Articles in the reading list) Market-Neutral Strategies. bottom). any strategy that doubles up on its alpha has. Marín 13 Portfolio Management . by definition.” The last paragraph on page 60 begins with the statement. twice the risk.’ says Unisys’ Service. about the volatility of a market –neutral strategy? Upon what will the volatility depend? b) If a market-neutral position is constructed by combining (1) a long position in a portfolio of oil stocks with a beta of 1. 3. On page 62 (middle column. 1. a long-short equity strategy is in fact most appropriate as an alternative to other investments.” by Saul Hansell. because the product is hedged.” a) What prediction can one make. this can hardly be considered equity. if any. Salomon’s Sorensen is quoted as saying “The volatility of the marketneutral funds isn’t 16 percent like the S&P. and ignore margin deposits on the futures position (assume there are none). which appeared in the April 1992 issue of International Investor. the article reasons that “though the managers are investing in stocks. “Of course. bottom). 2. Explain why the “market-neutral” strategy is described as “two-alpha. On page 59 (middle column.

? (see page 59. and the returns on the strategies are independent of each other.Market and Index Models Session 08 a) Do you agree that the market neutral strategies should best be viewed as alternatives to cash? b) What would justify the statement that the T-bill rate is a “legitimate benchmark” for market-neutral strategies? c) Would you recommend a market neutral strategy to a client with “a negative short-term view of the equity market”? (see the quote from Mr. Borneman of Consolidated Gas) 4. Suppose the true expected excess return on a market-neutral strategy is zero.? In part b. and the annual volatility of the strategy is 10%. a) What is the probability of observing a (simple arithmetic) average excess annual return over two years greater than 19 percent? b) What is the probability of observing a five-year average excess annual return greater than 10%? c) If there are 24 such market neutral strategies. middle paragraph) Copyright © José M. Marín 14 Portfolio Management . middle column. what is the probability that at least one of the strategies will achieve the performance in part a.

- A Practitioner's Guide to Factor Modelsalexandergir
- 79screamoutloud
- FSAvijitha-radhakrishnan-483
- Corporate Finance 01brahmi_119448
- TudyMichBadilloCalanog
- Beta CoefficientNahidul Islam IU
- Lecture 15Aditya
- Private Placement Real Estate Valuation (SLCG Website)Carmen Taveras
- Assignment of Financejitendrapareek98
- FM11 Ch 04 ShowMirza Junaid
- CAPMKumaravel Ds Ashwin
- Market Efficiency Finance Week 09Phaniraj Lenkalapally
- In FinanceSsahil Khan
- Ch. 19 Capital Asset PricingJesse Sanders
- Financial Management CaseDennis Falaminiano
- Uv0402 PDF EngOmar
- ch09 (1)Salman Zafar
- Chapter 5Misba Khan
- FDRMVishnu Prasad
- Gohar et alPerformance comparison of mutual funds in PakistanTaher Jamil
- Rational of the StudyUsman Shah
- Problem Set on Pay Discrepancy- Beta Value - And OthersAman Batra
- JMC INVEST4 .pdfIshanviya
- FINA2303 Topic 06 Risk and ReturnStephanie Lam
- Iind Session Ris-returnmridulakhanna
- The_longg3bee
- Research Articlekaps2385
- 098060102Sagar Paul'g
- Mirage_of_portfolio_performance.pdfSalman Zia
- Project Appraisal Capital Allocation Frame workapi-3757629

- Alcoa (AA,$AA) Misses Earnings Big Time, ValuEngine Rates the Stock a HoldValuEngine.com
- JinkoSolar Leads VE Forecast MNS Newsletter Short SideValuEngine.com
- How To Write A GREAT Business PlanTheBusinessInsider
- rev_frbsf_1983no3.pdfFRASER: Federal Reserve Archive
- Global Investment Returns Yearbook 2017 - SLIDES - Credit Suisse - February 21 2017zerohedge
- NCI Building Systems (NCS,$NCS) and other Shorts Lead way for VE MNS NewsletterValuEngine.com
- Vocera Communications Inc Jumps almost 20% in Three DaysValuEngine.com
- ValuEngine Weekly: Aerospace Stocks, JCPenney, and ValuEngine CapitalValuEngine.com
- rev_frbsf_1985no1.pdfFRASER: Federal Reserve Archive
- Discussion on 'Measuring Systemic Risk'American Enterprise Institute
- ValuEngine Weekly: Medical Stocks,Toyota Motor,Valuation Warning, and ValuEngine CapitalValuEngine.com
- Lumber...Worth Its Weight In GoldCanadianValue
- Glenrock July 2009 LP Updatemarketfolly.com
- frbsf_let_19830909.pdfFRASER: Federal Reserve Archive
- Goldman Sachs Letter to ClientsDealBook
- Information Risk and FVzerohedge
- 1q 2015 Gmo Quarterly UpdateCanadianValue
- Understanding Smart Beta 2.0Anonymous Ht0MIJ
- Kodak (EK,$EK) on the Brink of Bankruptcy and ValuEngine Rates it a SellValuEngine.com
- ValuEngine Forecast Model Scores Again with Trina Solar (TSL,$TSL) Short PickValuEngine.com
- Alerian – Taking the Passive vs Active Debate to MLPs(3)Anonymous Ht0MIJ
- BenchmarksTBP_Think_Tank
- Galleon - September Commentarymarketfolly.com
- Free Weekly Newsletter June 18, 2010ValuEngine.com
- Cemex SA (CX,$CX) Now Up more than 40% for VE Forecast 16 NewsletterValuEngine.com
- Repros Therapeutics Jumps More than 60% for VE PortfoliosValuEngine.com
- ValuEngine Weekly Newsletter July 9, 2010ValuEngine.com
- JinkoSolar (JKS,$JKS) Declines 22%ValuEngine.com
- Thornburg Value Q1 2015CanadianValue
- 161104 Ve Weekly NewsValuEngine.com

- A critique of modern financejohnolavin
- The epistemology of modern finance.pdfjohnolavin
- Shapiro Capmjohnolavin
- Demystifying Equal Weightingjohnolavin
- The epistemology of modern finance.pdfjohnolavin
- Factor investingjohnolavin
- Research paper templatejohnolavin
- Least Squares and the SCLjohnolavin
- Harvard Referencingjohnolavin
- Neuberger Berman Studyjohnolavin
- Factors Driving Risk Premiajohnolavin

Sign up to vote on this title

UsefulNot usefulClose Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Loading