SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
msinghal
Topic Author
Posts: 20
Joined: March 14th, 2005, 10:37 pm

CAPM and sector dependence

May 3rd, 2005, 12:24 pm

It seems a plausible idea that stock returns would have substantial correlation with performance of it's sector, and this might be more than correlation with the market. Does anyone know of any critical analysis of CAPM based on the sectoral influence and intrasector performance of stocks?thanks,Mansips: Also would you have more faith on the testing of this hypothesis on hi-frequency data or daily returns?
 
User avatar
htmlballsup
Posts: 615
Joined: February 9th, 2004, 10:23 am

CAPM and sector dependence

May 3rd, 2005, 12:55 pm

Its called factor analysis and Barra is the best known commercial example. They have just moved to using more frequent data, no doubt facilitated by the fact they where recently bought over by one of the world major data vendors. One of the problems with factor analysis is that the statistics need long time series so they tend to be pretty slow at reacting to market changes - especially when being compared to VAR models - but in truth they're probably worse at predicting the unexpected anyway because of selective memory.
 
User avatar
mrbadguy
Posts: 477
Joined: September 22nd, 2002, 9:08 pm

CAPM and sector dependence

May 4th, 2005, 1:33 pm

For an influence of stock sector returns on CAPM in a bayesian environment try this"CAPM improvements based on Bayesian Nets" by Singhal and Chakravorty 2005. Bye
Last edited by mrbadguy on May 3rd, 2005, 10:00 pm, edited 1 time in total.
 
User avatar
mattcushman
Posts: 100
Joined: July 14th, 2002, 3:00 am

CAPM and sector dependence

May 4th, 2005, 7:58 pm

One common misunderstanding of CAPM is that people think it's a one factor model in the sense that all inter-correlations of stocks are given by their correlations with the market as a whole. Is this what you mean by "critique"?Actually, CAPM allows for the residual "specific" risks to be correlated, it's just that you aren't compensated for this risk in any way.You also probably want to look at APT (arbitrage pricing theory) if you haven't already.
 
User avatar
Greenspoon
Posts: 56
Joined: February 2nd, 2003, 10:51 pm

CAPM and sector dependence

May 5th, 2005, 1:29 pm

MattAPT has never made sense to me as it is not offered as a specific n-factor model (meaning it's too generic).CAPM basically splits returns into systematic (Market) & idiosyncratic factors. We should ideally break down these systematic factors into sector specific and get Betas for each of them... something like Fama-French 3 factor but with additional sector specific component as well.Singhal,You do not need hi-frequency data as there is a lot of noise, daily data would do fine.N----Fools I am the Chief
 
User avatar
exotiq
Posts: 1888
Joined: October 13th, 2003, 3:45 pm

CAPM and sector dependence

May 5th, 2005, 7:10 pm

As htmlballsup (whatever that name means) mentioned, CAPM and APT are both special cases of linear factor models that make different assumptions about the factors, residuals, and the resulting market price of systematic risk.Greenspoon's critique of APT is as old as APT itself, but given that we do not live in a world with "natural" factors for asset returns, it actually is good to leave the investor free to choose the ones relevant to them. Barra has a 50+ factor model which some analysts are quite fond of, but tend to think that many factors is likely to overfit noise and ignore important non-linearities while giving the illusion of accuracy.APT is actually quite a useful/practical tool when heding certain baskets, since the factor sensitivities are additive, and after taking hedges with respect to the main factors, you are left with a quantifiable residual risk that can often be isolated and hedged further. This is especially true with baskets of stocks in the same sector, where an two-factor model has the major index (say SPX) and (sometimes the residuals of) a sector index/ETF (say XLE) against the major index as factors. So the short answer is yes, intra-sector and inter-sector factors are easy to measure and often easy to trade.CAPM more than Black-Scholes is useful for its insight far more than its practical formulation.
 
User avatar
htmlballsup
Posts: 615
Joined: February 9th, 2004, 10:23 am

CAPM and sector dependence

May 6th, 2005, 8:58 am

Just out of interest does anybody else find it funny when an academic derives CAPM from a specific form of utility function???It cracks me up every time. But then maybe I'm just sad - or missing something ......
 
User avatar
DominicConnor
Posts: 11684
Joined: July 14th, 2002, 3:00 am

CAPM and sector dependence

May 9th, 2005, 12:30 pm

Yeah, it's one of those things that occur so often you wonder if there is something in the deeper nature of things that drives people towards it.It's also funny (for the right value of funny) that people treat the single valued result you get out the other end as somehow useful.The value (if any) of CAPM is the curve you get by perturbing your assumptions. Also, it can be illuminating to change "sector". In short, it should make you face yourassumptions properly. Ask yourself why a particular purchae of an asset for a finance firm is really different from purchasing an identical asset for manufacturing.Sadly CAPM and it's variants are often used to justify, not decide.
 
User avatar
exotiq
Posts: 1888
Joined: October 13th, 2003, 3:45 pm

CAPM and sector dependence

May 9th, 2005, 12:40 pm

Funny, not really, but typical of academic idealization to the point of absurdity. I do really like the later work done on general log utility and the power utility functions that characterize our preference for moments of return distributions, but do not take the concept of "utility" as seriously as behavioral patterns in relative preferrence. That is the easiest way I can explain why I am in the minority of people who bets "don't pass/don't come" at the craps table.What is funny is trying to figure out what the heck that name htmlballsup means...could you please tell us?
Last edited by exotiq on May 8th, 2005, 10:00 pm, edited 1 time in total.
 
User avatar
Greenspoon
Posts: 56
Joined: February 2nd, 2003, 10:51 pm

CAPM and sector dependence

May 9th, 2005, 4:51 pm

Punny remark exotiq!!I was wondering why we don't model the utility function of an average investor as a markovian process (average investor within bounds of geography, level of wealth et al).Log utility or power utility or piecewise cubic spline utility (if there's one) might not go too far in covering a majority of investors, given the disparity of wealth distrubution in emerging markets.
 
User avatar
exotiq
Posts: 1888
Joined: October 13th, 2003, 3:45 pm

CAPM and sector dependence

May 9th, 2005, 5:36 pm

QuoteOriginally posted by: GreenspoonPunny remark exotiq!!I was wondering why we don't model the utility function of an average investor as a markovian process (average investor within bounds of geography, level of wealth et al).Log utility or power utility or piecewise cubic spline utility (if there's one) might not go too far in covering a majority of investors, given the disparity of wealth distrubution in emerging markets.The problem is not that utility is stochastic, nor that it is caused by disparity in wealth, but that investment choice is largely based on subconcious emotions of comfort and principle to individuals and to rules, constraints, paradigms, and ego-limits of managers, rather than on any pure concept of utility that the academics have come up with. Utility is a useful mathematical construct for understanding greed, risk aversion, and skew preference, more than a true psychological/regulatory description of how investors and their agents act. There are several individuals and fund managers who are not risk averse, but also have no desire to be long the S&P 500 just because you say there's an equity premium to be earned from holding it.
 
User avatar
Greenspoon
Posts: 56
Joined: February 2nd, 2003, 10:51 pm

CAPM and sector dependence

May 10th, 2005, 10:38 pm

Oh did you leave out 'Scorn', 'mistrust', 'fear' (of terrorism, lack of oil supply) ? They serve to explain that non-existent twisted end of the utility function.
 
User avatar
DavidJN
Posts: 1767
Joined: July 14th, 2002, 3:00 am

CAPM and sector dependence

May 11th, 2005, 1:38 am

(F)utility theory has some value, but in the small rather than the large. One can say powerful and useful things about individual human behavior (e.g. non-satiation, risk aversion) with utility theory but the problem has always been aggregating results to the market level - to do so one usually assumes homogeneous expectations, which is just silly. It should come as no surprise that arbitrage-free valuation methods have been much more succesful than equilibrium theories like the CAPM, since they are much less ambitious in that they take the pricing of primitives as given, the good work on implied distributions and skews/smiles notwithstanding.I note that Vasicek has a new paper in the JFE on the equilibrium determination of interest rates something like the CIR analysis in which he allows heterogenous expectations. Might be a real long shot, but it is interesting to see that someone with impressive credentials hasn't given up on equilibrium theories.
 
User avatar
htmlballsup
Posts: 615
Joined: February 9th, 2004, 10:23 am

CAPM and sector dependence

May 11th, 2005, 6:11 am

So do you have an arbitrage free way of valuing the equity of a firm? Please tell us......
 
User avatar
DavidJN
Posts: 1767
Joined: July 14th, 2002, 3:00 am

CAPM and sector dependence

May 11th, 2005, 12:24 pm

htmlballsup,The question you should be asking yourself is "Can I create a tradeable portfolio which replicates the cash flows of the equity of a firm in all future states of nature?". If the answer is yes (hah!) then you have an arbitrage-free way of valuing its equity.
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On