Investments Ⅰ: Fundamentals of Performance Evaluation

Professor Scott Weisbenner

Module 3: Testing the CAPM, Multifactor Models, & Market
Efficiency

Table of Contents
Module 3: Testing the CAPM, Multifactor Models, & Market Efficiency .................................. 1
Lesson 3-1: Objectives and Overview ............................................................................................... 2
Lesson 3-1.1: Objectives and Overview ............................................................................................................... 2

Lesson 3-2: Testing the CAPM & Market Anomalies........................................................................ 11

Lesson 3-2.1. Objectives and Uses of CAPM ...................................................................................................... 11
OPTIONAL: Testing the CAPM ............................................................................................................................ 14
OPTIONAL: Defending the CAPM ....................................................................................................................... 24
Lesson 3-2.2. Market Anomalies: Small-Firm and Value Effects ........................................................................ 35
Lesson 3-2.3. Interpretation of Market Anomalies ............................................................................................ 49
OPTIONAL: Investigating "Long Value Short Growth" Strategy ......................................................................... 57
Lesson 3-2.4. What We've Learned ................................................................................................................... 72

Lesson 3-3: Multi-Factor Models and Matching .............................................................................. 75

Lesson 3-3.1. Objectives .................................................................................................................................... 75
Lesson 3-3.2. Multi-Factor Models .................................................................................................................... 76
Lesson 3-3.3. Matching ...................................................................................................................................... 92
Lesson 3-3.4. What We've Learned ................................................................................................................... 99

Lesson 3-4: Market Efficiency ....................................................................................................... 101

Lesson 3-4.1. Market Efficiency ....................................................................................................................... 101

Module 3 Review and Graded Activities ....................................................................................... 133

Module 3 Review ............................................................................................................................................. 133
ASSIGNMENT 4 (Lesson 3-7): Analysis and Recommendation of 50 Balanced Funds, 1995-2014 .................. 143
DISCUSSION OF ASSIGNMENT 4 (Lesson 3-7): Analysis and Recommendation of 50 Balanced Funds, 1995-
2014 ................................................................................................................................................................. 153

OPTIONAL: Lesson 3-4 Use Domestic or Global Factors? ............................................................... 186

OPTIONAL: Use Domestic or Global Factors? .................................................................................................. 186

OPTIONAL: Lesson 3-5 Return-Risk Model Used by Chief Financial Officers (CFOs) ....................... 203
OPTIONAL: Return-Risk Model Used by Chief Financial Officers (CFOs) ......................................................... 203

1
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-1: Objectives and Overview

Lesson 3-1.1: Objectives and Overview

Hi, I'm Scott Wiseman and I'm a professor of finance at the University of Illinois, and this
is module three of my investments course. And in this module, we'll tackle topics such
as testing the capital-asset pricing model, multifactor models, and market efficiency.
Really, this module is all about benchmarks. Are you surpassing a benchmark, what's
the right benchmark to employ? Different racers, different benchmarks.

2
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's look at this athlete here. Clearly surpass the hurdle, clearly surpass the
benchmark, in investment speak, we would call this a positive alpha performance.
Exceeded the benchmark, good job.

On the other hand, here, looks like all four failed to clear the hurdle rate. I have to
apologize to all our friends in China for bringing up this bad memory from the 2012
London Olympic games. We classify this in investment speak, negative alpha
performance.

3
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, what is the right hurdle for firms or projects, or to evaluate the investment
performance of a fund manager?

Different models provide different benchmarks, different assessments of performance.

If we find that a firm or benchmark clears its hurdle, clears its benchmark, we're still
faced with the issue, does it represent luck, or does it represent skill?

4
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Here's what we plan to address in module three, we can already check objectives and
overview off the list. We're almost done with that one, like to build up the sense of
accomplishment here, then we're going to move on to testing the capital-asset pricing
model and looking at market anomalies, top of the list we'll be looking at the
performance of small firms, firms who have a small market value of equity, and then
value firms. Then we'll talk about the development of multi-factor models, and matching
techniques to evaluate the performance of a stock or mutual fund, then we'll kind of
broaden the discussion to an international perspective. Should we use domestic or
global factors in our asset pricing models.

5
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Then, what do CFOs, Chief Financial Officers of firms, what do they use for their return-
risk model? See, I like to be an optimist, so instead of saying risk-return, I'm always
flipping around to return-risk. What models do the CFOs who are using this in the real
world, what do they use? Then, market efficiency. What do we mean by market
efficiency? We'll talk about that, give examples of market efficiencies, and some
potential examples of events where we may see pattern returns it seems to violate
market efficiency. Then, one assignment for this module, assignment from the course's
perspective, and then as always, I will provide my discussion of that assignment. But
remember always do the assignment first, it will be an analysis and ultimately a
recommendation of the performance looking at 50 balanced funds over the period 1995
to 2014. So, you'll analyze the performance of these funds, and then ultimately come up
with which of the fund do you recommend the most, which of the fund do you
recommend the least? And then module three wrap up as we always have at the end of
the module.

6
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, for those of you who are taking this as a high-engagement learner and are taking it
to get University of Illinois credit, you're going to have further work to do, and in
particularly going to analyze the famous hedge fund momentum strategy in AQRs
momentum funds case study. So here you're going to learn about historical returns to
the momentum strategy, possible explanations for its past performance, differentiate
between hedge funds and mutual funds, and get some insights into the tradeoffs
involved in implementing an investment strategy. So, the AQR momentum fund case
will be something that will have for high engagement learners, those who are taking the
course for University of Illinois credit.

7
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

We have completed modules one, and two, so we've really kind of ate the broccoli.
Okay? We've built up the fundamentals that we need for the investments course.

So now we can get to the fun stuff. Okay? And you can see like with my waistline, this is
what I view as fun.

8
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What's the practical knowledge, experiences I hope that you can take from this module?
Well first, let's learn the historical performance of the capital-asset pricing model, great
theoretical insight, and motivation behind it, how well does it perform in practice and
what tests have been done to determine how well the capital-asset pricing model
performs. Then, we'll learn and evaluate key return anomalies, like small stocks doing
well, value stocks doing well. Is it attributive to risk, or is it something else? We're going
to get a lot of experience interpreting and using both the capital-asset pricing model and
three factor models and understanding differences between the two.

9
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What is meant by market efficiency? You probably heard that term thrown out like this is
an efficient markets advocate, do you believe in efficient markets? What do we actually
mean by market efficiency? We're going to understand what efficient markets imply both
for returns, as well as for the asset management industry.

10
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-2: Testing the CAPM & Market Anomalies

Lesson 3-2.1. Objectives and Uses of CAPM

This lesson is all about testing the capital asset pricing model. As a result of testing it,
some market anomalies had been discovered. That's a topic here. Things that you'll
take away from this lesson include uses of the CAPM, and that's something we're going
to talk about quickly after we get these objectives laid out, then we're going to just how
well does the capital asset pricing model predict returns? It has a clear prediction. Firms
that have higher betas, like Tiffany on average should be generating higher returns than
firms that have lower betas, like Walmart. Then we'll talk about these return anomalies
that have emerged tied to certain a firm characteristics like size and the value growth
characteristic of the firm. After we document that, hey, small stocks historically have
outperformed large stocks, value stocks historically have outperformed growth stocks,
reason why these return anomalies may exist. Then we'll get our hands dirty a little bit,
go into Excel and We'll actually analyze this famous long value short growth strategy to
gain some more insights into it.

11
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

But before we get to these topics, let's actually take a step back and talk about some of
the uses of the capital asset pricing model. I think it's useful to break it down into two
parts at corporate finance perspective and then an asset management perspective. So,
from a corporate finance perspective, the CAPM can provide a benchmark return that
project needs to yield to be acceptable to investors is providing the hurdle rate and the
benchmark that investors require to hold the stock or that the projects yield to be useful
for the firm. Therefore, the capital asset pricing model is very useful in firm's capital
budget decisions and also in the decision should the firm be retaining earnings to use
for internal investment projects, or should the firm instead being a payout capital to its
investors so they can invest that elsewhere in the economy. We'll talk later in the
module about a survey of the Chief Financial Officers of large US companies where
they actually, just to preview the results, say most of the majority them always or almost
always use the capital asset pricing model when they went to come up with this required
return, or the hurdle rate for their stock or to evaluate projects.

12
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

How about from an asset management perspective, what's the use of the CAPM there?
Well, the CAPM is a way to assess whether an investment has outperformed or under-
performed in the past given its risk. What's nice about the capital asset pricing model on
the Alpha, it gives a way to compare the performance of managers that have different
investment styles. So, I may be investing in assets that are generally low risk. Another
manager may be investing in assets that are generally high risk. We would expect the
manager that invests in the more high-risk assets to have a higher return. But that
doesn't mean the managers better, it just could mean that they're investing in higher risk
assets by having an asset pricing model, the capital asset pricing model or CAPM, it
enables a way to determine controlling for the risk, which manager is beating their
benchmark more. Now, keep in mind in an asset management perspective, the capital
asset pricing model is a nice starting point, but it's certainly not the ending point. More
analyses will be required.

13
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
OPTIONAL: Testing the CAPM

In this video, we're going to test the Capital Asset Pricing Model, or at least report the
results of others who have tested the Capital Asset Pricing Model. How well does it
work in practice?

It has a great theoretical motivation. Its development has led to several awards, Nobel

14
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Prize in economics granted to multiple folks. It'd be nice to know in actuality how well
does it predict returns. Now is a good time for a pause, think, and answer.

How would you do the test? The question, how would you conduct a test of whether or
not the Capital Asset Pricing Model is useful in predicting returns? What did you come
up with? Well, let's compare it to some of the giants in the finance field here.

15
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Black, Jensen and Scholes, they did an early test of the CAPM. Remember the CAPM
was developed in the mid 1960s here. They did a test of the model 1931-1965. They
basically estimate CAPM regression in December of each year. They use two to five
years of monthly data. Then they form a portfolio over the next year, and they form the
portfolio by ranking stocks by their estimated Beta. Portfolio 1 is going to have the 10
percent of stocks with the lowest Beta given this was 31-65 before the NASDAQ. They
just focus on New York Stock Exchange Stocks here. They're focusing on the 10
percent of New York Stock Exchange Stocks with the lowest Beta portfolio 10 is to 10
percent of New York Stock Exchange Stocks with the highest Beta, and nine would be
those between 80 and -90 percent. Rank stocks by their Beta, put them into ten
portfolios. Then we can measure how they performed over the next year and see on
average, if the stocks and the higher Beta group had higher returns. Now certainly
there's going to be firm-specific factors that determine how stocks perform but that
should average out given we're looking at all a large sample of stocks here.

16
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's see how this works. Specifically, we're going to regress monthly excess returns of
these 10 Beta-sorted portfolios on the portfolio Beta. We're looking at the period 1931-
1965.

Let's look at our CAPM. Excess return of the security equals Beta times the excess
return of the market. On our left-hand side, we have the excess return of this Beta
portfolio. This could be portfolio 1 that has the stocks with the lowest Beta, or it could be

17
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
portfolio 10 that has the stocks with the highest Beta. Our right-hand side variable now
is Beta. Usually, we think of Beta as being the coefficient from the regression but here
it's a right-hand side variable, the regression. For portfolio 1, these stocks have an
estimated Beta that's low. So, the thought is over the next year they should on average
have lower returns. For portfolio 10, they have a Beta that's high. So, the thought is over
the next year, on average, their return should be higher. Tiffany should be having
returns that are higher than Walmart on average, not necessarily every year but on
average that should be the case. Predictions for regression coefficients when we run a
regression of the excess return of the Beta portfolios each month on the Beta estimated
over the prior period. Intercept term should be zero here, there's no constant term here.
The constant terms should be zero in this regression. The coefficient on this Beta of the
portfolio should equal this, the excess return of the market over the period that we do
the regressions. In this case, it's a period 1931-1965. The prediction for Beta is going to
be the average excess return of the stock market or that period. That prediction for the
coefficient is going to vary over the period that we do the regression. You can see if it's
a period where the excess return of the market is large, Beta is going to have a bigger
effect on returns. What if the excess return on the market happened to be zero on
average over the period? If this is zero, it doesn't matter what Beta is. If this is zero, the
excess return of all the portfolios is the same. The bigger is this excess return, the
bigger will be the effect Beta has on the excess return of the portfolios. Something
useful to keep in mind over the sample period for this regression, the average excess
market return was 1.4 percent per month. Our prediction for the coefficient on Beta,
when we regress the excess return of the Beta, portfolios on the Beta should be 1.4 or
0.014. Not surprisingly, you remember these papers on the capital asset pricing model
were written in the mid 1960s when you evaluate stock return data, 1931-1965, and you
do test capital asset pricing model works extremely well.

18
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

On the left-hand side, we're looking at the average excess monthly return of the Beta
portfolios, we have 10 of those. Those returns are for an average for each of these 10
Beta portfolios are shown here, they're denoted by the X's. Here you can see for each
of the X, the systematic risk that just represents what's the average Beta of the stocks in
that portfolio. Portfolio 10 here, it has the average Beta, a little over 1.5, Portfolio 1 has
a Beta of about 0.4,0.5. What do we observe with this data 31-65? On average, the
portfolios that have higher Betas have higher returns than the portfolios that have lower
Betas. Now when you look and that's exactly as we predict here that it should be the
case that given there's a positive excess market return over this period, stocks or
portfolios that have higher Betas should have higher returns. One thing to notice is this
point here, where the regression line crosses the y-axis, the predicted excess monthly
return when Beta is zero. In this regression, if Beta is zero, the excess return of the
portfolio should be zero. We come pretty close to that, it's not exactly zero here, but it's
pretty close to that.

19
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Then you can see this positive slope for the regression line. Higher Beta on the
portfolio, higher return on average for that stock portfolio.

Remember over this period, the average excess stock market return was 0.014 or 1.4
percent per month. The simple CAPM prediction is that a one unit increase in Beta,
increasing Beta by one should be associated with a 0.014 increase in excess returns or
1.4% per month.

20
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's see our actual coefficient if you look at what that slope was, actually the
quantitative value, 0.011. Not exactly 0.014, but pretty close. The capital asset pricing
model shows up as a pretty strong predictor of returns over the period 1931,1965.

But you may be, " Hey, it's not 1965" at least when we're taping it, it's 2015 here so 50
years later. What if we do a more up-to-date test on the capital asset pricing model but

21
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
use the same methodology. Now when we're looking at stocks, we're looking at stocks
from all the stock exchanges. We estimate a capital asset pricing model each
December using 2-5 years of monthly data, get the Beta from that CAPM regression for
the stock then starting the next year, rank stocks by their Beta for the 10 portfolios,
where Portfolio 10 is the 10 percent of stocks with the highest Beta, Portfolio 1 is a 10%
of stocks with the lowest Beta, then see over the next year, what the monthly returns
are for the various portfolios, how that's related to their Beta with the prediction that
portfolios of stocks with higher Betas on average should have higher returns. This
analysis you can see here is done 1928-2003 so over 70 year’s worth of data.

This was done by Ken French and Eugene Fama. Fama and French, you'll hear that
more than once. They did a 2004 paper; they report the results. The sample is 1928-
2003. They're basically looking on the y-axis. Here we're looking not at the excess
return, but just simply the average analyzed monthly return. We're looking at on an
annual basis is key and we're looking at the total return, not an excess, excess return
here. If you see like a Beta of one, the CAPM prediction is about 11 percent or so, so
you know what the market return is. Here are the 10 portfolios, this is Portfolio 10. The
portfolio of stocks with the highest Beta. Here's Portfolio 1, the portfolio of stocks with
the lowest Beta.

22
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

If you draw a regression line through it, you can see that there's a positive slope so
those portfolios with stocks in them that have higher Betas do have somewhat higher
returns, but the problem is the slope here of the actual data is not as steep as the slope
that the capital asset pricing model would predict. So, the capital asset pricing model
while there is some prediction or returns here, it's not as strong as the model would
suggest it should be.

23
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
OPTIONAL: Defending the CAPM

All right, so in this video, we're going to look at what are some of the defenses for the
kind of lackluster performance if you will, of the Capital Asset Pricing Model predicted
returns at least over the kind of the long history from the 1920s till kind of the most
recent century here. When you look at performance from 1931 to 1965, Capital Asset
Pricing Model was performing extremely well predicting returns in a way that it should,
okay? If you look at kind of the more long-time series, beta seems to only weakly predict
stock returns and less so kind of since the mid-1960s. So, what's going on here?

24
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Advocates of the Capital Asset Pricing Model, I think their main defense would go
something like this. The model maybe isn't performing as well as we'd like because the
market is mis-measured, okay? Theoretically, when we talk about this kind of market, it
should include all assets, not just the S&P 500. Real estate should be in there like a key
asset, for young folks, would be there, human capital. We just simply proxy for the
market portfolio. We aren't putting in all these kinds of assets that really are in kind of a
broad portfolio that people have. We just proxy with the stock market because we have
good data on that. We have daily data on the stock market but by using a proxy for the
real market, this is inducing measurement error into the regression and could bias
against the CAPM performing as well as we would kind of expect it to. So,
mismeasurement of what the market portfolio really is by using this proxy of kind of the
stock market. Using this proxy, the stock market maybe could explain somewhat why
the Capital Asset Price Model doesn't perform as well as we would like it to.

25
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Another defense of the Capital Asset Pricing Model. Just in a recently published paper
in the Journal of Financial Economics, back, I think, in 2014, that the Capital Asset
Pricing Model actually does perform pretty well on days when an important
macroeconomic systematic news is released, okay?

So, the authors of this are Savor and Wilson 2014, Journal of Financial Economics
piece. They basically, do an analysis in spirit to those of Black, Jensen and Sholes and

26
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Fama and French that we've kind of talked about already in this module. They used
daily data over the last 12 months to estimate betas and then they form beta sort of
portfolios over the next month. So, they're kind of using a little different horizon for when
they kind of form the betas. But basically, the same idea, use past data to estimate
betas for stock, rank stocks by their estimated beta, and then look ahead and see to the
stocks that have the higher beta typically have higher returns.

But the rub that they're adding is they're looking does the CAPM predict well on days
where there's a big news announcement about some macro kind of economic factor.
That's what they're really bringing to the table, okay?

27
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What do we mean by macroeconomic news? Well, could be news about inflation,

employment, Federal Reserve, Federal Open Market Committee interest rate decisions,
okay? So, they're looking at days when those announcements occur. Now, if you look at
the trading days, these macro announcements are occurring about 13% of the time,
roughly kind of one out of eight. So, once every kind of two weeks or so. So, it's 13% of
the trading days. But when you look at the market risk premium, the performance of
stocks these days, when the macroeconomics news has announced, they count for
about 60% of the equity risk premium during the year. So, these are important days,
13% of the days, but 60% basically of the stock market return occurring on these
macro-news announcements.

28
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Okay, so let's look here and this is kind of a nice analysis to get things started here,
where they look to show how important these announcement days are. Now remember
these announcement days here showing kind of the dark bar here, only are 13% of the
total trading days, but most of the return of stock and here we have the excess return.
This is in basis points, is occurring during days where you have this macroeconomic
news. So, macroeconomic news comes out, average excess return is a little over ten
basis points compared to about one basis point on non-announcement days. And Savor
and Wilson compare this to kind of other turn of the month effect. So, the first day of the
month versus kind of not the first day. And look at returns during the month of January.
The dark bar versus not the month of January, the light bar here. So, pretty big
difference in the performance of stocks, whether there's macroeconomic news coming
out versus not. kind of the difference here, much larger than if we're looking for
example, returns in the month of January versus non-January, which is also a well-
documented result.

29
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, let's look at some of the empirical results in their paper. Let's look at macro
announcement days, which they call A days, A for announcement. What's the average
excess return for 50? Forget looking at ten, let's look at 50 beta-sorted portfolios over
the period 1964 to 2011, okay? So, in our x-axis, we have the Capital Asset Pricing
Model beta, okay? And on our y-axis, we have the average excess return in basis points
here. And what do we see? An interesting distinction, the diamonds here, these are the
returns of these 50 portfolios from the highest beta to the lowest beta here on
announcement days. Here, you have the relation, the highest beta to the lowest beta on
non-announcement days. So, on non-announcement days, the Capital Asset Pricing
Model isn't kind of doing very well. There isn't really any relationship if anything is
slightly negative, but I won't make too much. That's basically no relationship between
beta and the return.

30
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

But when you look at the announcement days, when this macroeconomic news comes
out, which should affect kind of stocks with systematic risk. You have a very clear
relationship here, firms that have higher beta, higher exposure to macro or systematic
risk have bigger changes in their returns. Those with lower betas have lower returns on
these announcement days, okay? And kind of very, very, very clear relationship.

So, when we look at the regression here of daily excess returns on kind of going back to

31
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
just focusing on ten beta-sorted stocks over the ten beta-sorted portfolios or the period
1964 to 2011, let's recall the Capital Asset Pricing Model. So here, Savor and Wilson
are basically doing an analysis. Pretty much similar to those done kind of more recently
by Fama and French where we have ten beta-sorted portfolios. And we're just gonna to
see if portfolio ten, which has the highest beta has higher subsequent returns on
average, relative to portfolio one for example, which has the lowest betas. So,
remember the Capital Asset Pricing Model here. Okay, predictions for the regression
coefficients. The intercept term should be zero, okay? There's no intercept term
predicted, there's no constant as part of this equation. So, that's another way saying
when you do a regression, the prediction is the intercept term should be zero. The
coefficient on beta when we're regressing the return of the beta-sorted portfolio on its
average beta. The average beta the stocks in the portfolio. The prediction is, the
coefficient on this regression should be just whatever the average excess return of the
market was during that period, which in this case is 1964 to 2011, okay? And it's useful
in terms of kind of daily average equity market premium, it's 10.5 basis points. So,
there's a clear prediction when we run this regression of the excess return of the beta-
sorted portfolio on its beta. This called the coefficient on the beta should be 10.5 basis
points if returns are being predicted by the Capital Asset Pricing Model in the way the
Capital Asset Pricing Model says they should.

Okay, let's get to the results. First, let's put up our regression equation for the beta-
sorted portfolios. We're regressing their excess return. Okay, we're doing this on a daily
rate on the average beta, that portfolio prediction is higher beta. Higher should be the
return. Okay, intercept of this regression should be zero, there's no constant term here.

32
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Coefficient on beta should be this average excess market return during the period.
They're using daily data, so the average excess market return during this period was
10.5 basis points. What's the prediction? What are the results? Excuse me. When we
have just focused on announcement days. So, these are days where macro
announcements are occurring. There's US Federal government is announcing,
employment numbers are announcing what inflation was. Federal Reserve, Federal
Open Market Committee is announcing a change in interest rates are just focusing on
these announcement dates. Remember there are about 13% of all trading days. The
intercept in this regression 1.3 basis points but not statistically significant different from
zero, okay? What's the beta? What's the coefficient on beta? Usually, we talk about
when we say beta, we mean the coefficient. But in this case, what's the coefficient on
beta? It's 9.2 basis point, 9.2 statistically significant results. So, it means higher beta,
higher returns earned by the portfolio. And the key is this 9.2 estimate is not statistically
different from the Capital Asset Pricing Model prediction, 10.5. So, in other words on
these macroeconomic news days, you can't reject the hypothesis that the Capital Asset
Pricing Model predicts returns, as it says, it should. Okay, so Capital Asset Pricing
Model works basically very well in predicting which stocks go up the most on news
when macroeconomic news concerning macroeconomic conditions. Now, when we go
to the non-announcement days, the end days, then the capital asset pricing model does
not work well. And in fact, there's non-relationship between the beta of the portfolio and
the return of the portfolio on these no-announcement days. These end days coefficients
here is not statistically different from zero. So, another way to think about this, if you're
like trading stocks, if you think you have some insight about what's going to happen to
kind of the macro economy, what the news is going to be. If you think the news is going
to be very good about the future economic development, you want to invest in stocks
that have the highest beta. Because these regression results suggest the stocks with
the highest beta are going to have the biggest kind of movement in returns when this
macroeconomics news is released. So, if you think good news is going to come out
about the future economy, you want to be holding that day stocks that have the highest
beta because they will have the biggest jump in price according to this regression
analysis.

33
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, conclusion from A Tale of Two Days, days when macroeconomic news is released
versus days where it's not, beta strongly related to average returns, as the CAPM would
predict on the days when macroeconomic news is released. Kind of makes sense,
right? Beta is a measure of the systematic risk or the macroeconomic risk of the stock.
So, in the days when this macro-news comes, that's when the CAPM performs pretty
well. But it is a puzzle, why no relation between the stock beta and returns on non-
announcement days? Okay, so the authors of this study, A Tale of Two Days speculate
maybe the announcement days provide a clear signal of systematic risk and expected
future stock market returns to investors. Therefore, by getting this clear signal, they kind
of trade accordingly, and then that's kind of reflected in the prices. So bottom line, there
is kind of a tale of two days here when macroeconomic news is being released. Capital
Asset Pricing Model is pretty good at predicting returns on days outside of that doesn't
seem to work very well at all.

34
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-2.2. Market Anomalies: Small-Firm and Value Effects

The capital asset pricing model doesn't perform as well as we'd like. Although it does
seem to perform pretty well on the days when macroeconomic news is released. But
there's another problem. If we want to honestly evaluate the performance of the capital
asset pricing model, it not only has a prediction as to what should affect returns. Beta
should predict returns in the capital asset pricing model, but it also has predictions
about what should not predict returns. That's basically nothing else should predict
returns except for Beta.

35
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

For example, no firm characteristic other than Beta should be able to predict stock
returns.

Looking at the data, some return anomalies have popped up. Now, if these are
reflecting some type of multi-faceted risk beyond the systematic risks that the CAPM
emphasizes, then we probably shouldn't be calling these anomalies. These are higher
returns because they represent risk. But for now, let's just call these market return

36
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
anomalies. One of these is the small firm effect, stocks of small firms. By small firm we
mean the market value equity of the stock has a low ranking. Small firms do better than
large firms, historically, higher returns, particularly in the month of January.

Book-to-market effect or value effect; stocks with high book-to-market ratios. Remember
this is the book value of equity that you see in the balance sheet and the annual report
of the firm divided by the market value of the equity. Those type of firms we call them
value firms historically have earned higher returns than growth firms, firms that have low
book-to-market ratio. For the growth company, think of a firm where there's a great idea
but there aren't many tangible assets in place, maybe a computer server and that's it.
Value stock is doing better than growth stocks, book-to-market effect.

37
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Momentum. This is another return anomaly or predictable pattern in return. Instead of

calling these market anomalies, we could also call them predictable pattern in returns
that have been found in the data, size, value, and momentum. What's the momentum
effect? Stocks that have done very well during the last year continue to drift up over the
next month. Stocks that have done very poorly over the last year, continue to drift down
over the next month. For those of you who are taking this course for high engagement
experience, you want to get credit through the University of Illinois for the course, you're
going to have more experience with momentum. We'll do some back-tests in the
momentum strategy in the AQR momentum funds case study. Let's talk a little bit about
the small firm effect.

38
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

This is data, again from the Ken French Data Library, where we rank firms and deciles
based on the size of the market value, their equity. Decile 10 is the largest firms. Think
like currently in 2015, Apple would be in this group here. Decile 1 are the smallest firms
here. All those in-between, just looking here at the average annual stock returns, 1927-
2014, there's a clear pattern. Bigger firms have smaller returns, difference here of
almost eight percentage points from the smallest stocks to the largest stocks here.

39
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Now, one explanation going back to this difference in returns, how much of this is just
due to smaller stocks being riskier in a capital asset pricing model sense. Maybe they
have higher sensitivity to the economy, more likely to fail in bad times. So, to
compensate for this kicking you while you're down effect, they have to give a higher
return on average, because investors realize, if we go into recession, these small cap
stocks are really going to hurt their portfolio performance. You can do an analysis,
estimate what is the capital asset pricing model Beta for each of these portfolios, again,
using this annual data, 1927- 2014. You do see the small stocks are riskier and that
they have higher Beta. Again, there's a clear relationship. Smaller stock then higher
beta in the portfolio. For the smallest stocks, their betas a little over 1.5, so they amplify
movements in the broader market. For the biggest stocks, their betas actually a little
less than one. Now remember when we look at the whole US stock market, that's value
weighted. When we look at the whole US stock market, it's going to behave much more
like the firms in deciles 10 and 9 because they're bigger. They have more weight and
the overall portfolio than the firms in deciles 1 and 2 which are smaller, so then have a
much smaller weight in the market portfolio.

40
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's look at the small firm effect without and then with risk adjustment. Without risk
adjustment here and here, instead of looking at the total return, we're looking at the
return in excess of treasury bills. A simple excess return with no risk adjustment. Again,
we see there's about an eight-percentage point difference. The returns of small firms
being about eight percentage points higher than the return of the largest firms. But once
we take into account differences and risk, look at the Alpha from the capital asset
pricing model, we see small stocks outperform by about the CAPM benchmark by about
2.6 percentage points per year, while for large stocks are basically are hitting right on
their CAPM benchmark. Of this, roughly eight percentage point difference in return only
about 2.7 of that is explained by the Alpha of the small stocks.

41
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

The rest of that then obviously explained by difference in risk. Summing up small stocks
over 1927-2014, small stocks, the bottom decile ranked by size have outperformed big
stocks. The top decile rank by size on the order of about eight percentage points per
year. You might expect smaller stocks to be more sensitive to the economy, more likely
to fail in a downturn that shows up in the CAPM beta. For the small stocks, the beta is
1.5, for the largest stocks it's 0.93. There is a pretty big difference in risk.

42
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

You put all this information together when we examined the capital asset pricing model,
and we look at performance above and beyond the benchmark. The difference is only
2.7 percentage point once we adjust for risk. Of that difference of about eight
percentage points and the performance of small stocks relative to large stocks, about
two-thirds of that is accounted for the difference and the underlying risks of the two
types of firms, smaller firms riskier on average and larger firms, so higher returns. But
there is some of that remains one-third that just represents the Alpha or the
outperformance of performance above and beyond the benchmark of the small firms
over this long time period, 1927-2014.

43
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Nothing stays a secret for long. Such a difference, eight percentage points per year,
people are going to talk about it and write about it. There were academic articles
documenting the size effect back in 1981. I know there's a bunch of insert joke moments
here coming up in the video.

What if we break the sample into two parts? Pre the 1981 article and then post the 1981

44
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
article on the size effects. Does the size effect persist? Did the academic papers
contribute to diminishing of the size effect post-1981.

Size doesn't matter as much as it used to. I make it easy for you guys who are joking
along at home here. Let's look at decile 1 versus decile 10. This is looking at the
smallest firms, performance minus the performance of the largest firms when a focus in
terms of decile, or we can look at the small minus big factor, SMB. This is basically
looking at the firms whose size is below median versus those above. Two differentials
reflecting size. Look at the dark bar here on the left represents the difference across all
the years between small and big firms. Again, we have these two measures of small
and big. The middle bar here, medium tone, is the result less than years before 1981.
And the final bar here is actually representing then the performance of small relative to
big stocks after 1981 going to 2014. I forgot to mention these returns we're looking at
here are already risk-adjusted; these are Alphas from the capital asset pricing model. If
you look at, let's just focus here on the left side, decile 1 through 10, the smallest stocks
minus the biggest stocks, we have this, Alpha. Remember, we documented this before
of 2.7 percent. This was over the full sample period. Go to prior 1981, 1927-1980, this
difference was 5.3 percentage points per year. Since then, it's gone away to zero. Size
doesn't matter like it used to. If you use this alternative size breakdown, so here it's not
as disparate, the firm is really focusing on small being below median, big being above
median. Again, you still have the same pattern that most of this differential in the
performance to small versus big firms occurs during the first part of the sample before
1981, after the academic papers are published, documenting, small firms are doing
historically very well. The small firm effect goes away, which is consistent with a bunch

45
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
of people going out, buying the stocks of small firms, that drives up their price, but that
means and subsequently lower returns in the future. This picture here definitely is
consistent with that story.

Now, we've talked about the small firm effect, let's talk about the famous value
premium. If you're talking about the value premium, you certainly have to bring up
Warren Buffett here, very famous value investor.

46
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

A good place to look at for just a documentation of the value effect is work by Eugene
Fama and Ken French. Remember that chart I had where they did an updated analysis
of the performance of the capital asset pricing model. Well, this chart is a little different.
Here, they're looking at the performance of 10 portfolios, and within the portfolio, they
evaluate by the market capitalization of the stocks, but the portfolios are formed based
on book-to-market ratio. Ten portfolios, Portfolio 10 has firms with the highest book-to-
market ratio, so these are the cement companies, the companies have a lot of tangible
assets, the ratio of book value to market value is high, they're right up here. Portfolio 1
are the growth companies, the lowest book-to-market ratio. These are the companies
that have a great idea so that's why they have a market value, people are willing to pay
something to own their stock, but they currently aren't hardly any tangible assets in
place that show up on the balance sheet like cash, property, and equipment. Those
would be in Portfolio 1. Now, what was key observation from this analysis? For each of
these portfolios, Fama and French had a prediction from the capital asset pricing model
about what the return should be, and they found that when you look at these portfolios,
the value portfolios, those that have the biggest value tilt like Portfolio 10, Portfolio 9,
Portfolio 8, they outperform their CAPM benchmark by the greatest amount. For those
that are more growth like, for example, Portfolio 1, it underperforms its capital asset
pricing benchmark, it has a negative Alpha. Therefore, value stocks are outperforming
growth stocks. Those with the highest book-to-market ratios have the biggest Alphas
here relative to the CAPM benchmark, 10, 9, 8, and 7, they're all earning returns
historically much higher than would be predicted by the CAPM, which is given by this
line here. While Portfolio 1, those of the growth companies, it's actually underperforming
its capital asset pricing value.

47
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Value strategies go by several different names here. If you want to invest in value
stocks, high book-to-market firms, low price-to-earnings firms, each of these are values
stocks, firms that are paying high dividends would also be a value strategy. Just to
throw some names out there. Warren Buffett, before him, Ben Graham, to throw out a
University of Illinois connection, Josef Lakonishok, all of these are well-known value
investors.

48
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-2.3. Interpretation of Market Anomalies

All right, it’s easy to document the market anomalies. The difficult part is interpreting
why these anomalies exist or if they really are anomalies, maybe they represent risk.
So, when we find some pattern of returns, it could be due to risk. So, if there's some risk
story behind some assets generating high returns, high risk, high return, problem
solved. But the key thing is we can always throw risk out there. Do you have an
economic story that is clear as to why investors should care about this risk and
therefore, kind of expect higher returns for exposure to it. So, what's the story behind
the risk? Don't just throw risk out there unless you have a story to kind of back it up.
Irrational behavior, could explain patterns in returns. If you think, there's some part of
the population that's just making mistakes that leads to mispricing of assets that the
smart investors can then take advantage of. But for the irrational behavior story to really
kind of be potent in explaining market anomalies, you need to have enough kind of
people out there making mistakes and not enough people there that can correct prices,
okay? And then it could just really be an anomaly that their pattern you observe is just
due to data mining. So, when you're looking at people are doing hundreds or thousands
of correlations trying to predict stock returns, there's probably bound to be a few weird
ones that pop up just by chance, okay?

49
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, let's talk about market anomalies, they can be attributed with three factors. Let's talk
about risk first. Risk is thrown out there. This would be the answer from advocates of
the efficient market viewpoint. Looking at the out performance of small and value stocks
historically, a risk-based argument would say, hey, these are stocks that are risky, small
firms are riskier than large firms and value stocks maybe have more financial leverage.
It makes them riskier than growth firms, particularly in a recession. This risk causes
investors to not value these small in value stocks is high that highly and they therefore
requiring a higher return as compensation for the risk. Now, notice if there's a risk-based
explanation for the returns, it would suggest that this market anomaly, and I put it in
quotation marks, will continue, right? If there's a risk story, then you should have the
high returns going forward. You can kind of think of the example if there's $100 bill
laying on the ground, should you pick it up or not? Okay? The risk-based story would be
like, hey, don't pick up the $100 bill, it could be tied to a bomb that goes off with
probability 5%. So, that would suggest that $100 bill is going to remain on the ground a
long time because no one wants to take the chance of it blowing up. So, risk-based
explanation, return patterns should continue.

50
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Another possibility for explanation for the market anomaly or the pattern and returns that
we observe for like size or value or any other predictability and returns that we have
would be just simply irrational behavior on the part of some investors. So, perhaps
investors really like large and are always optimistic going for the latest kind of growth
startup and they neglect the more kind of boring value cement company type stocks,
and they neglect small stocks, okay? So that causes the small and value stocks to be
under price and yield higher returns in the future. Now note for the return anomaly to
continue under their rational behavior explanation, you need two things to hold. One,
there need to be enough of these behavioral investors to keep making these mistakes
so that prices are distorted, so there are positive returns to be had from the smarter
investors. And you also need the smart money to not be so large that it fully corrects the
distortion in price, okay?

51
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Third explanation for market anomalies kind of predictable patterns and returns simply
could be data mining, okay? By chance, some criterion seems to predict returns or, well,
it actually does predict returns in the past, but it's just due to chance, okay? It's just due
to this data mining, looking at every thousands of variables. You're bound to find a few
weird correlations just like if you have a room of 100 people, there's probably going to
be a couple if they flip a coin five times maybe more than that, get five heads in a row,
okay? Just kind of do it by chance the next five coins they flip probably aren't going to
be an additional five heads in a row. So, under the data mining explanation, there was a
chance correlation in the past returns we observed, but, since it's just due to chance, it's
unlikely to continue going forward. And the return pattern observed in the past under the
data mining story would then disappear going forward.

52
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Okay, so I know data analytics is kind of very important, but it's always useful to
remember the necessity of distinguishing between correlation and causation, okay? So,
for example, strong correlation in the past between butter production in Bangladesh and
the S&P 500 Index returns. So, of course, people are trying to see what predicts stock
market returns. What turns out to be a great prediction of that butter production in
Bangladesh. So, let's buy our plane ticket and head off to Bangladesh because look at
this effect, 10% increase in butter production appears to be associated with a 20%
increase in the S&P 500. So, I don't know about you, but I'm already packed and ready
to go.

53
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, time to go to Bangladesh, okay? One problem though, you're not in Bangladesh,
that's Vietnam.

Bangladesh is over there on the border with India. Sorry, had to pull a little geography
quiz on you here, kind of copyright John Oliver routine.

54
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, this just goes to show it's very important to understand the difference between
correlation and causation, okay? In the past, there's been this strong correlation
between butter production in Bangladesh and S&P 500 returns, but I don't think that
means if I fly to Bangladesh, eat a bunch of butter, more butters produce, the stock
market's going to soar. Just a free correlation, I doubt very much it will continue going
on in the future. The key thing is it's always going to be easy for someone exposed after
the fact to come up with some tortured story to try and sell or motivate why one factor
predicts another. Although the butter in Bangladesh might be pretty hard to have the
story as to why does that predict stock returns, but the key thing is can you come up
with a hypothesis prediction ahead of time, then go to the data and see that your
prediction is actually borne out in return patterns in the data. Can you come up with a
story that's plausible ex-ante ahead of time as opposed to coming up with a story after a
data mining exercise to kind of justify in effect you found.

55
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, it's important to realize there's going to be many people correlating many different
things with stock returns because there's a kind of a big reward if you can find some
correlation there that actually is predictive of returns in the future. When you have so
many analyses being run, there's bound to be a few weird anomalies that happen by
chance, okay? Now, if you think there's a pattern return that's happened by chance just
because you had this pattern in prior years, you wouldn't expect it to continue.

56
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
OPTIONAL: Investigating "Long Value Short Growth" Strategy

We have talked about value stocks, outperforming growth stocks historically, at least
over 1927 through 2014. Notable value investors like Warren Buffett, Ben Graham,
Josef Lakonishok. Given that, why don't we do some further analysis of this famous long
value short growth strategy, also known as HML, where the H is high book-to-market,
The L is low book-to-market, so high minus low. I put together a series of returns for you
to analyze in this spreadsheet here. You can see that again; the names are pretty self-
explanatory. HML_1927_2014. I also have the excess market return, US stock market
return minus the risk-free rate. A small return minus big firm return is in there as well. All
the usual suspects.

57
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

If you want to get more information of the cutoffs that are used to form these various
portfolios. Here's the link that you can get this. Again, these return series are all coming
from the Ken French Data Library.

Here's a six-pack of questions. I would like you to take a step back. This isn't an
assignment, just an in-class or in-lecture exercise. After you go through these on your
own then you could see me getting on the computer in Excel and going through these

58
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
questions and having a discussion of the results. Question number 1, pretty
straightforward. what's the average monthly return of this value-minus-growth, also
known as HML strategy, over the 88-year period from 1927-2014? Remember, this is
actually like a hedge fund return because it's a return of value stocks minus growth
stocks or high book-to-market stocks minus low book-to-market stocks.

We have this average return. How much of this average return and this return is the
difference in performance between value and growth? How much of this value-minus-
growth return actually reflects alpha in the capital asset pricing model? We have just a
simple return. How much of this represents performance above and beyond that capital
asset pricing model benchmark for the HML strategy?

59
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Question 3 is the remainder. How much of the return of value-minus-growth is

attributable to value stocks being riskier than growth stocks? Is it zero? Is it positive?
How much of the value-minus-growth differential is due to alpha? That's question 2.
How much of it is attributable to systematic or market risk differentials? That's question
3.

60
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Does the value-minus-growth strategy do better in good times or bad times? Clearly,
your answers to question 3 and question 4 are going to be related. Does the value-
minus-growth strategy do better in good times or bad times for question 4?

For question 5, is there a lot of idiosyncratic risk to this strategy? What fraction of the
variability in value-minus-growth is explained to market-wide movements? The
remainder is what is explained by idiosyncratic non-market-wide factors. I bet you know

61
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
what kind of statistic from the regression output to get the information for question 5
here, what fraction of the variability is explained by market-wide movements?

Then finally, remember for small stocks, we documented pretty conclusively that
publication, or at least the timing of the publication of some academic research
documenting that small stocks historically have done better than large stocks, seem to
diminish the small stock effect going forward, which would be consistent with people
realizing historically small stocks have been doing better than large stocks, buying a
bunch of small stocks, driving up their price, which leads to lower returns in the future.
Do you see that same effect in value stocks sphere? Eugene Fama and Ken French
published a series of articles about value growth predicting returns and the HML factor
in 1992 and 1993 respectively. Have the returns to this value-minus-growth strategy,
have they diminished since this attention was paid by these academic researchers?

62
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Evaluate the value-minus-growth strategy on your own. Answer these six questions,
then we'll get together afterwards, and I'll go through the analysis and the responses in
Excel.

Did you have a chance to look at the value-minus-growth strategy and do an analysis of
this? Historically, value stocks have done better than growth stocks. A lot of investors
like Ben Graham, Warren Buffett, Josef Lakonishok have done very well with value

63
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
stock investment strategies. I thought it makes sense to look at this HML series of
return, high book-to-market firms minus the return of low book-to-market firms or value-
minus-growth. Let's go into the spreadsheet here, HML, 1927-2014. We're using 88
years of monthly data from January 1927 through December of 2014 contained in our
spreadsheet. Here from rows 5-1,060, we have 1,056 months of return data over this
long period. Let's analyze this value- minus-growth strategy. The first part here,
Question 1, was just simply, what's the average monthly return of value-minus-growth?
Or HML is the Fama-French factor is called. Let's just see this average differential
performance on a monthly basis of value stocks relative to growth stocks. Again, if
you're interested in what's the specific cutoff of value minus growth at the top of the
spreadsheet, I give you the website where you can see the formal definition of this on
the Ken French Data Library. Average, and we're looking at column D here.

What's this average return? 0.4 percent per month. Value stocks doing better than
growth stocks on average by 0.4 percent per month. Aggregate these up over a year,
it's around five percentage by 0.4 per month, 12 months, around five percentage point
difference value stocks doing better than growth stocks. Now the question is, why is that
the case? Is this due to value stocks being more sensitive to economic downturns and
growth stocks? How much of this difference of 0.4 percent per month, as I counted, is
attributable to value stocks being riskier in a CAPM sense more sensitive to the market?
How much of it is just they've been a great investment yield this high Alpha? Let's look
and see how much of this return is due to Alpha of value stocks beating growth stocks,
performance above and beyond the benchmark set by the CAPM. Let's do a regression
here. We've already installed our data analysis add-in. Let's go to this. I hope that this

64
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
part of the course you have already done that because you've already done a lot of
regressions here.

In the data analysis, click on Regression.

For our y-range here, again when we do regressions, I like to click on the labels. Then
our labels for the return series will appear in our Excel regression table. Our y variable

65
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
is column D, HML, value-minus-growth. We're going to start at row 4 because we want
that label and we're going to go down through row 1,060 to our last return in the series,
December 2014. We're doing a CAPM model, so we're relating value- minus-growth to
the excess return of the market. That's column B, row 4 reflecting where you want to
incorporate the label here, going down to row 1,060. Now click on "Okay” and pray that
we've inputted everything correctly.

The regression appears. Again, this is a regression of the value-minus-growth, a zero-

cost portfolio or hedge fund strategy here. We're going long value stocks, shorting
growth stocks, same thing as saying going long, high book-to-market, shorting low
book-to-market. How is this pattern of a value-minus-growth related to market
movements? Well, one thing we notice is the R-squared, it's only 0.05. Only five percent
of the variability in the value-minus-growth returns is explained by movements in the
market. The R-squared is very low. Market-wide factors aren't explaining much of the
movement in the value- minus-growth strategy. How about if we look at the Beta?
Market risk-free is the right-hand side variable, the coefficient on that and the CAPM
regression is the Beta 0.15. So, it is the case that value-minus-growth has a positive
sensitivity to the market. In other words, value stocks are a little more sensitive to
movements in the market than growth stocks because value-minus-growth has this
positive Beta, but it's relatively small. It's only 0.15. Certainly, much closer to zero than it
is to a Beta of one, remember Beta 0, our treasury bills, a Beta 1 would be the US stock
market. Value stocks are a little riskier than growth stocks. Part of the outperformance
of value stocks relative to growth stocks could be this extra market risk, but the Alpha
here in the CAPM model, 0.3 percent per month, 30 basis points per month, that's the

66
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Alpha. That's the outperformance of value-minus-growth that's not attributable to risk,
say that five times quickly. Not due to risk, a better way to say it. The 0.3 percent per
month represents the part of value beating growth that's just due to extra profit, return
above and beyond risk.

Remember, going back to our spreadsheet, the average return of value beating growth
was 0.4 percent per month.

67
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

The Alpha and the CAPM is 0.3 percent per month. About 75 percent of value beating
growth is due to Alpha, is due to simple outperformance of value stocks relative to
growth stocks above and beyond their risks. That's why historically, people have talked
about value as being a great investment strategy. How much of the return of the value
strategy, of the value beating growth of the 0.4 percent month per average is due to
differences in risk between value and growth stocks? Well, we already answered that.
We look at the simple average difference in returns value versus growth, high book-to-
market minus low book-to-market. That's 0.4 percent per month. The Alpha of the value
minus growth is 0.3 per month. The amount left over has to be the difference in returns
due to risk 0.4 minus 0.3 is 0.1. So, 0.1 percent per month, of the 0.4 is due to
difference in risk. About 25 percent of the value-minus-growth differential is due to
differences in systematic risk, but about 75 percent or 0.3 percent per month was just
due to the value beating growth on a risk-adjusted basis. Most of the value growth
differential due to just being a better investment yielding a higher Alpha, not due to
differences in risk. Does the value-minus-growth strategy do better in good times or bad
times? Well, that's related to this Beta here. Our Beta is 0.15. That means when the
market's doing well, value-minus-growth is doing a little better. For example, the market
goes up by 10 percentage points, that would suggest value minus growth is going up
about 1.5 percentage points. Some sensitivity to the market, not that much though.
Value stocks are a little more sensitive to what's happening in the economy than growth
stocks are, but it's not an overwhelming difference. If we look at these coefficient
estimates, intercept of 0.3, the Beta of 0.15. Another way to interpret that is when do we
predict the return on the value minus growth strategy should be zero? Well, that should
be a month when the excess market return was minus 2 percent, because minus 2

68
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
percent for the market excess return, times 0.15 equals minus 0.3. Add that to the
intercept of 0.3, that gives you an Alpha of zero. The CAPM model, if we take these
regression coefficients seriously, it would suggest as long as the excess return of the
market that month is minus 2 percent or better, the value stocks should be beating
growth stocks on average. If the excess return on the market is worse than minus 2
percent that month, that would be a month that we predict value stocks are doing worse
than growth stocks on average. Just taking our Alpha and Beta coefficients, literally
here, seeing what they imply about predicted average returns for the value growth
strategy. Question 5 of our six pack, what fraction of the variability of the value minus
growth strategy is explained by market-wide movements? Very small amount. Only five
percent. The R-squared is 0.05. A lot of the variability in the value-minus-growth
strategy is explained by idiosyncratic or non-market-wide factors. Then finally last
question. Fama and French wrote about value growth predicting returns in 1992 and
about this HML factor and the three-factor model in 1993. Since their research has the
returns to the value-minus-growth strategy waned.

Well, let's go back to the data. How would you think about testing it? Well, a natural way
to test this is let's just evaluate the value-minus-growth strategy after 1993 starting in
January of 1994. Has their attention to the strategy, their published research articles,
caused everyone to go out, buy value stocks, drive the price of value stocks up which
drives the future return down? Maybe value-minus-growth no longer is yielding a
positive Alpha after this research was done in the early 1990s. Let's do an updated
CAPM regression but only using data after 1993.

69
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Click on the Data tab, data analysis, regression, so now we're starting no longer at Row
4. Let's start at Row 809. Why are we starting at Row 809? That corresponds to the
return January 1994. We're going to look at the pattern of value-minus-growth returns
from that strategy in January 1994 through December of 2014. We just have to carefully
switch these starting points of 4 to 809 and then we're good to go. Now, we're not
starting at the beginning of this spreadsheet, so let's unclick the label. If you leave the
label clicked, you're going to get a weird regression output so take the label, unclick that
our y and x range now is from Row 809 down to Row 1060. Column D is our left-hand
side variable, our y variable value-minus-growth. Column B is the excess market return.
Let's click, "Okay."

70
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Well, I have to say I'm on fire here going through these Excel regressions here but
better not count my chickens here before they've hatched. We're not going through yet
but so far so good. Let's look at the results post 1993 from January 1994 through
December of 2014. What do we observe? Well, the key thing is we want to look at the
Alpha. It's very interesting, remember over the full time series, 1927-2014 monthly data
value-minus-growth strategy, the Alpha was 0.3 percentage points per month. What is it
after 1993? It's also 0.3 percentage points per month, 0.32 to be precise. Now, if we
look at the standard error, we're estimating over a small number of observations 252
months here as opposed to 1,056, so about 1/4 of the sample. This Alpha is measured
less precisely. Its p-value is 0.11. With conventional statistic cutoffs we'd probably say
this Alpha 0.32 percent per month is no longer statistically different from zero but that's
mainly an artifact of the small sample size. The key point is this Alpha estimate is
exactly what it was over the full sample. Really, there isn't evidence that the returns to
value stocks relative to growth stocks have diminished after the publication of the Fama
and French research on this topic. One thing that's also interesting is this Beta from the
CAPM model. Now, it's turned negative. It used to be a positive 0.15. Now, over the
1994-2014 period, a 21-year period, it's a negative minus 0.17. That suggests over the
more recent time period, growth stocks actually have the greater sensitivity to the
market than value stocks do. Hopefully now, we've gone through an analysis of the
value minus growth investment strategy, and you have a better understanding of it.

71
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-2.4. What We've Learned

All right, everyone's favorite point in the lesson the what we learned slide, what we've
learned in lesson 3-2. Capital asset pricing model is useful in both corporate finance
and asset management settings. In corporate finance setting, this is useful to estimate
discount rates for hurdle rates for projects and firms in asset management setting. It can
be a way to evaluate portfolio managers that are taking on different levels of risk and
kind of compare them on the same playing field.

72
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

BETA is key parameter in the capital asset pricing model. In fact, BETA is the only thing
that should predict returns in a CAPM world. It generally doesn't predict returns when
you go to the data as strongly as the CAPM would suggest it should. Other factors have
emerged that also predict returns beyond BETA for example, firm size, value growth
dimension, momentum, all predict returns. Kind of a saving grace for the CAPM on the
days when the big macroeconomic news is released? And remember the capital asset
pricing model is all about investors care about the market risk or systematic risk of
assets, okay? On the days when the macroeconomic news is released, those are the
days when the CAPM performs the best. So, can I have a little saving grace there?

73
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Big debate as to why the various return anomalies or predictable pattern in returns
related to things like size, value, and momentum, why they exist? Does it reflect the risk
of this portfolio strategy, irrational behavior or is it data mining? Remember the
Bangladesh butter production being correlated with the S&P 500 can have a classic
example of data mining correlations that probably don't represent a causal relationship.
And if it's a data mining correlation that we've observed in the past, we probably don't
expect that relationship to continue in the future, because it's just due to look
historically, value beats growth by a pretty wide margin. But this difference is somewhat
less in bad markets.

74
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-3: Multi-Factor Models and Matching

Lesson 3-3.1. Objectives

In this lesson, you will understand multi-factor models and arbitrage pricing theory, and
then understand the tradeoffs between using the capital asset pricing model and the
three-factor model, differences in what the alphas mean in the two models and what the
factor loadings and the betas represent in the three-factor model. Then finally, the pros
and cons of using a model like the capital asset pricing model or the three-factor model
to evaluate performance versus characteristic-based matching.

75
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-3.2. Multi-Factor Models

Arbitrage pricing theory and multi factor models.

So just taking a step back, you can think about if many stocks available as we talked
about, remember our example going gambling into Macau. First specific risk is always
going to be able to be diversified away by just adding more stocks to the portfolio. So,
this firm specific risk shouldn't show up in required returns or stock prices because that

76
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
can be diversified away very easily. Common shocks on the other hand, can't be
diversified away and will be priced. So, kind of the same intuition that we had behind the
capital asset pricing model. We also have behind arbitrage pricing theory.

Now, as we talk about kind of a one example of a common shock would be the
sensitivity to the overall market reflected by beta and the capital asset pricing model.
But these common shocks need not be limited to sensitivity to the overall market. And
that's what arbitrage pricing theory is basically expanding upon the capital asset pricing
model.

77
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, you can think of kind of having a more complicated asset pricing model with more
risk factors included in it. So, look at the expected return here of asset i, have it be a
function of sensitivities to various risk factors, okay? Or you can express in the second
explanation, just simply an excess return being related to various risk factors in the
economy. The betas will be called factor loadings. The factors may also be called factor
risk premium.

78
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, a simple one-factor model. Capital asset pricing model. The factor in the capital
asset pricing model is the market risk premium. We've kind of, what well plowed ground
at this point in the course. Or at least I hope so.

Now, perhaps investors care more about the mean variance tradeoff that's reflected in
models like the CAPM and maybe that risk is more multifaceted. There could be other
state variables that represent risk that investors care about. Now, we might not know

79
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
exactly what the state variables are, but maybe we can observe certain factors,
observable characteristics, that are correlated with the state variables.

For example, there may be certain firm characteristics that are associated with some
stocks doing better than others. And maybe firms that have those characteristics, those
stocks move up and down together. There's some kind of correlation in the performance
of stocks associated with this characteristic. These factors would represent a different
type of risk that is accounted for in the simple capital asset pricing model, with the only
risk that's being priced is sensitivity to the overall market reflected by beta. Thus, we
would need a more complicated asset pricing formula or a multi varied approach as
opposed to just a single factor, capital asset pricing model, okay?

80
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, examples of two firm characteristics that could represent different types of risk to
investors, just simply based on return patterns we've observed. And at this point in the
course, you probably guess where I'm headed, one could be firm size. We've observed
that historically, small firms have done better than large firms. Maybe this represents
some type of risk that investors care about. Therefore, we have this pattern in returns
that we've observed historically. And then particular the value growth, differential value
stocks historically, outperforming growth stocks. That return pattern maybe suggests
that investors view value stocks is more risky than growth stocks. This value growth
differential represents some state variable that represents a type of risk not accounted
for in the capital asset pricing model. That's where arbitrage pricing theory and
manifestation that the three-factor model then come into play.

81
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Developer of the three-factor model, Eugene Fama, Nobel Prize winner in economics
along with this co-author Ken French, who has this great data repository that we've
used frequently throughout the course. I thought would be used to talk about Eugene
Fama's research philosophy, very scientific. So, you should use the market to
understand markets better. You should use market data to understand markets better
not to say this or that hypothesis is literally true or false. No models ever strictly true.
The real criterion should be, do I know more about markets when I'm finished than I did
when I started. So, I thought it's worth kind of mentioning this. I thought this is a very
scientific way to approach research and the approach that we should always take kind
of coming in with open eyes. And just trying to get a better understanding of the world,
not coming in with some kind of hidden agenda.

82
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, when we look at this Fama French 3-factor model developed in 1993, we're relating
the excess return on the stock. The left-hand side to three factors now as opposed to
just one-factor. So, the first factory is a representative of overall sensitivity to the
market. This is the CAPM model if we just stop with the first factor. The second factor
here is relating the excess returns of a security to how small stocks are performing
relative to big stock. So, the small minus big factor, the size factor, return on a
benchmark portfolio, small stocks minus return a portfolio of large stocks. And then
finally, the third factor, this is a value factor. Where relating the excess return of a
security to how value stocks are performing relative to growth stocks that period. So,
this represents a return on a benchmark portfolio of high book-to-market value firms
minus a return on a portfolio of low book-to-market growth firms. So, for example, we
estimate this regression model if we observe a positive beta on the small minus big
factor, that would mean our security is doing better during periods of times when small
stocks are doing better. If we observe a negative coefficient on the beta regarding the
small minus big factor of the size factor, if it's negative, it means that our performance is
doing better when large stocks are doing better, okay? If we go to the value factor, if the
beta on the value factor is positive, it means our performance is positively correlated
with how value stocks are doing. So, in value stocks are doing well, we're doing well if
it's negative. If the beta on the value factor is negative, it means that our stocks are
doing well during periods of times when growth stocks happen to be doing well, okay?

83
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

This Fama French 3-factor model explains the returns of the ten book-to-market
portfolios that we had in our earlier chart very well. This is basically true by definition.
Right now, we're accounting for value and size in our asset pricing model. So now we
explain value and size, okay? Alpha is no longer statistically different from zero and a 3-
factor model when we look at these three portfolios formed by book-to-market. We're
incurring beta risk now in the 3-factor model, instead of producing alpha. So, if we
estimate portfolio of value stocks in a capital asset pricing models we observed, we get
a positive alpha. If we look at this value stock portfolio and put it in a 3-factor model, the
alpha now is zero, okay? And instead, we have a positive loading on the value factor,
positive beta on the value factor here. So that's the difference between the capital asset
pricing model and the 3-factor model. Here, I'm assuming this is just an index fund
invested in value stocks. If you have active management, picking certain value stocks,
you could have a positive or negative alpha once you get to the 3-factor model. But the
3-factor model is you're only going to get an alpha in the 3-factor model if you're doing
something different. You can't get a positive alpha on the 3-factor model by just simply
investing in an index of value stocks that will now be accounted for in the beta. And the
value factor won't be accounted for the alpha in the CAPM model.

84
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, in terms of thinking of the Fama French 3-factor model, the factors representing
risk, high CAPM and beta firms maybe are more vulnerable to market downturns.
Therefore, they have higher required returns, small firms maybe also more vulnerable to
economic downturns. So higher required returns as well. Same story for value stocks or
high book-to-market, maybe they have higher financial leverage, and they get in
distress in economic downturns. These would be risk stories as to why size small firms
and value firms historically have yielded higher returns. This would be the risk-based
explanation. Now, if you want to push back and say, well, hey, these all may be true, but
isn't that reflected in the CAPM-beta, okay? So, the debate then goes kind of back and
forth is small and value factors representative of risk or is it representative of something
else.

85
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, there's various factor models. When we start with the one-factor model, that's
simply the capital asset pricing model, the one-factor is sensitivity to overall market
condition. The 3-factor model, we add on small minus big factor, value minus growth
factor. So now we're basically in the 3-factor model, we're controlling for investment
style of the security or the mutual fund along the size dimension and the value growth
dimension.

86
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

4-factor model. Well, this would be adding an additional factor, controlling for
momentum, okay? So, this is something we mentioned for high engagement students
who are taking the course for university Illinois credit. We'll talk more about this pattern
and returns known as momentum. Stocks that have done well the past year continue to
drift up the next month. Those have done poorly the last year continue to drift down the
next month. So, 4-factor model takes into account momentum patterns and returns. And
more 5, 6, 7 factor models let me kind of stop just talking a little bit about a 5th factor or
what's usually a 5th factor.

87
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

If there is a 5-factor model, that's a liquidity factor. It's considered by many managers
and academics. So, this accounts for the liquidity or illiquidity of a security or assets. So
particularly during the financial crisis, some assets that people wanted to sell, they
found it was very hard to do so, they were illiquid. They can only sell them in a drastic
reduction of price. So, this illiquidity actually really hurt the investors at the time where
they needed the ability to sell their asset the most. So, kind of thinking through in terms
of asset pricing, maybe securities and assets that are more illiquid, harder to sell. Those
are assets that should be giving you higher return to compensate for this liquidity risk.
Liquidity factor could be, if you have a 5-factor model, that could typically be the 5th
factor.

88
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, what should we make of this capital asset pricing model versus a 3-factor model?

So, CAPM versus 3-factor model. I think the CAPM has a very nice theoretical appeal.
Nobel prizes were awarded as a result of its development, not one, not two at least
three. So, kind of, and I think it makes intuitive sense, in terms of hey, why do you buy
property insurance? Forget that your mortgage company may require you to. You buy
the property insurance because you value the fact that you'll get the payment if your

89
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
house burns down. The property insurance pays you when you need it the most. That's
the intuition behind the capital asset pricing model. Certainly, it's imperfect and
predicting returns, but I think it's good to take a step back. This is just a one variable
model. The world's very complicated. It would be amazing if just one variable beta
actually could predict returns. So maybe you should give the capital asset pricing model
a little break. But the bottom line is the beta in the CAPM doesn't predict returns as well
as the model suggests, it should. Beta and Alpha are frequently reported on common
financial websites. We've gone through examples of that finance.yahoo, Morningstar
others. So, it's kind of relevant still and you should kind of know the model, what Alpha
and Beta represent. CAPM is used lot in a corporate finance setting to assess future
discount or hurdle rates that a firm may use to evaluate a project or maybe you want to
kind of evaluate yourself. To conduct your own evaluation of your own firm. You need to
come up with some discount rate to convert future cash flows into a value today. We'll
talk about survey evidence regarding what CFOs actually use. Capital asset pricing
models still at the top of the list. In terms of asset pricing setting among money
managers, CAPM is maybe a nice starting point, remember it's reported on Morningstar
is a summary statistic of kind of performance. But it's not the ending point, it's just a
starting point then kind of more complicated asset pricing models will be used like 3-
factor, 4-factor models etcetera.

Multifactor models have emerged after discovery of the various patterns in kind of
returns, like the small firm affect the value effect. Now, what does the 3-factor model
mean? It could mean that hey, we're counting for different types of risks that are
manifested in these patterns of returns. So maybe in the real world, risk is multifaceted,

90
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
It's not as simple as the mean variance tradeoff of the capital asset pricing model. If risk
is multifaceted, probably a multi-factor model will be better than kind of using this one-
factor capital asset pricing model. Or you can view the 3-factor model just simply
account for known patterns and returns or patterns and returns that we've observed in
the past that's accounted for in the 3-factor model. So, in other words, to earn an alpha
in the 3-factor model, you can't simply employ investment strategies that have been
known to work in the past. Investing in an index of value stocks may yield an alpha in
the capital asset pricing model, but it won't in the 3-factor model. To get an alpha in the
3-factor model, you have to be a good stock picker in the universe of value stocks. So,
you can think of the 3-factor model is building on the capital asset pricing model, but
also controlling for differences and investment style related to the size of the stock, as
well as the value growth characteristics.

91
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-3.3. Matching

We've talked about a capital asset pricing model, multi-factor models like the Fama-
French three-factor models. We've had some experiences with this, doing regressions
in Excel, methods to evaluate performance that are regression-based, but there's other
methods to evaluate performance as well. Some prefer what's known as characteristic-
based matching as opposed to regression analysis.

92
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What do we do in characteristic-based matching? We compare the return of a given firm

stock with the average return of firms with a similar characteristic or characteristics.
Now this works for looking at an individual security. You can compare that to a
benchmark return, matching it on certain characteristics, for example size of the firm or
profitability or industry or value growth dimension can compare the return of your firm to
that of firms that are similar along this one characteristic or two or three characteristics.
You can also use this to evaluate a mutual fund manager where you then characteristic
match every component of the portfolio and then aggregate up these returns and
excess of the benchmark return for all the securities in the portfolio. For example, on the
Ken French website, if you wanted to do a characteristic-based matching, portfolios are
produced for all the 10 size deciles or all the 10 value growth deciles or a combination
of the two. There's 100 portfolio returns that represent where does your firm rank in
terms of size deciles from 1 to 10? Where does it rank in terms of value growth deciles
from 1 to 10? I might be in the top decile of size and Decile 3 on the value growth
dimension. I have some other growth tilt. You can compare your firm to a portfolio of
stocks that have that same size, value growth dimension, as well as industry returns.
You might want to compare your stock to stocks at a similar industry. There are multiple
ways you can get this data. I found the most easiest straightforward way is just to get it
again from the Ken French Data Repository. How would you calculate the Alpha for
when you have one stock or a portfolio of stocks?

93
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

If you have one stock, is just simply the return of your security minus whatever this
benchmark return is and that can be form on whatever characteristic you think is
relevant could be. I want to compare my firm to all other firms in the same industry. On
the Ken French website, there are industry breakdowns is crude is dividing firms into 10
groups. I believe that's the crudest cutoff to as fine is breaking firms up into 49 different
industry groups and maybe want to compare your firm to the performance of all other
firms in that same industry over whatever horizon, or you care about comparing two
firms of similar size or value growth or past momentum, whatever, you can form the
benchmark on whatever characteristic of firms you want. How about if it's a portfolio of
stocks? Well, same idea only now we just have to sum over all of the stocks in the
portfolio. Let's assume we have N. You just aggregate up this abnormal return, return in
excess of the benchmark, but then waiting it appropriately by how important is this stock
in the portfolio. Then obviously, the summation of these weights equals 1. Athletes can
give 110 percent, but our portfolio can only give 100 percent. If you want to use it to
evaluate a mutual fund manager, again, compare the return of every stock in the
portfolio to its benchmark return, and then just wait up these returns across the various
securities in the portfolio.

94
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

We've talked about regression analysis, and we've actually had some experience with
that. If you remember the assignment on mystery securities, you estimated CAPM
model or CAPM regression, as well as three-factor model regressions to evaluate the
performance of the three mystery securities. You could also have done that analysis
with a characteristic-based matching approach. Let's take a time for a pause, think and
answer moment here with Le Penseur.

95
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's think about matching versus regression. What do you view as a key advantage
and disadvantage of the characteristic-based matching technique relative to the
regression approach of asset pricing models like the CAPM or three-factor model. What
do you see an advantage and disadvantage of using matching versus the regression
approach to evaluate performance?

All right, so I've been interested what you thought of advantages or disadvantages. Let

96
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
me give you my thoughts on this. A key advantage to matching is no underlying model.
In a way, it's like the characteristics that go into our regressions, like, sensitivity to the
market, size, value. Those might be characteristics we match on, but there's no
underlying model where the characteristic-based matching. That's generated. It says
you pick like, hey, I want to compare my stocks to performance of their industry peers or
what have you. That makes it very flexible. There's no regression estimates required.
This very simple and straightforward technique. Just compare the performance of my
security with a bundle of other securities that have some similar characteristics that I
view as particularly important.

What's the disadvantage to matching? Again, no model so it's one of these

circumstances where what's an advantage is also a disadvantage. Why is this a
disadvantage? So far in the course, we've used our asset pricing models to look
backward, to evaluate past performance and to see, hey, what historically has been the
benchmark or hurdle rate for these various securities? Have they outperformed this
benchmark? Have they generated Alpha? When you're looking back to evaluate
performance using regression approach, using characteristic-based matching, both
work perfectly fine. How about if I want to forecast ahead and I want to see how my
forecast change with different assumptions regarding the future state of the economy?
Let's say I want to estimate a discount rate or hurdle rate for my firm, and I need this,
the discount rate to key component of valuing a firm because you need to convert these
future cash flows into a value today. One component going into the discount rate will be
in the future. How much do we expect the broad stock market to outperform treasuries?
What's going to be the equity market premium? What if I want to assume that's going to

97
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
be eight percent, and then I want to change that assumption to make it four percent. If
you have an underlying model and you've estimated these Betas using past data, you
can then make different assumptions. Given the Beta of the stock, let me change a
market risk premium going forward from 4 to 8 percent. Just plug into the equation and
it will spit out, okay, here's what your discount rate is, here's what your required return
is. You have a model, so you can put in different assumptions and see what results
come in because you have an underlying model that predicts a relationship. When
you're doing characteristic-based matching, how do you adjust going forward? You
can't. There's no model if it's good for looking in the past, but there's no way to come up
with estimates in the future under different scenarios. The advantage of the matching
technique, no underlying model, also a disadvantage of the matching technique. No
underlying model makes it difficult to then forecast into the future under various
scenarios.

98
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-3.4. What We've Learned

What have we learned in this lesson? Multi-factor models, like the Fama-French three-
factor model have emerged in the wake of the capital asset pricing model not
performing, at least with historical data, as we would expect. In other words, the Beta
does weekly predict stock returns, but not as much as the CAPM model suggests it
should. Therefore, multi-factor models have arisen to fill in the gap as we've learned
more about patterns of stock returns over the last 80, 90 years. What does this multi-
factor model represent? Well, it could represent multi-faceted risk. Maybe risk is more
complicated than just simply investors trading off mean and variance. Like in the capital
asset pricing model world, maybe risk is multi-faceted, so we need a multi-factor model
to price that into securities. On the other hand, maybe the size factor, the value factor,
small stocks doing better than large stocks historically, value stocks doing better than
growth stocks historically, maybe those return patterns don't really represent some
underlying risk. It's just representing maybe some behavioral bias. It gives some good
profit opportunities to active managers like Warren Buffett, for example. In this case,
you can view the three-factor model is just holding, taking into account investment style
when we're evaluating the performance of a manager. For example, when we look at a
three-factor model to produce an Alpha or to exceed your benchmark in a three-factor
model, you can't simply be investing in an index fund of small stocks, or you can't be
investing in an index fund of value stocks. In the three-factor model, you need to do
above and beyond that. You need to bring something new to the table. You can't just
simply be investing in strategies that have yielded abnormal returns in the past. Or
another way to think about it, the three-factor model adjusts for investment style along

99
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
the dimensions of size and value in a way that the capital asset pricing model does not
because all it accounts for is just differences in securities in their market risk or Beta.

Capital asset pricing model has a nice theoretical appeal. Empirically look at historical
data doesn't work as well as a model would predict it should. Although an interesting
result by Savor and Wilson finds that the capital asset pricing model actually does do a
pretty good job on days where macro news is announced, but nonetheless, overall
performance is as good as the model would suggest. I think it's still a nice theoretical
underpinning and it's important to take into account this is just a one variable model. In
worlds complicated, it would be very strange or unusual if this one variable was the only
thing that could predict returns. The CAPM is a nice starting point, but we should take a
step back and realize maybe it's not surprising it doesn't explain everything when it
comes to stock returns and other factors seem to historically predict returns because
after all, it's just simply a one variable model, a nice starting point, but particularly in
asset management, we need to go use more sophisticated models than that. Finally,
characteristic-based matching is fine, but the difficulty with that is how do you use that
to project ahead? You can't. If you want to project ahead to come up with hurdle rates,
discount rates, you need some underlying model.

100
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Lesson 3-4: Market Efficiency

Lesson 3-4.1. Market Efficiency

In this lesson, we're going to tackle market efficiency or maybe tackle's two strong.
We're going to introduce the notion of market efficiency and some of its implications.
What is meant when people say market efficiency? Did you know there's actually three
forms of market efficiency? What does market efficiently imply about the pattern of stock
prices we observe and the usefulness of having an active management portfolio
strategy? One of the takeaways from this segment is also to just get a sense of how
difficult in the real-world it is to differentiate luck from skill and asset management. You
may observe a manager has done well in the past, but is this due to luck or does it
represent some fundamental skill and being able to differentiate between future winners
and losers? Talking about asset management, in a world of efficient markets, you may
think that financial advisor asset management is not necessary. You can't beat the
market but there's other aspects of active management that are important for your
portfolio, such as tax considerations, as well as taking into account non-financial
aspects of your portfolio, such as your human capital, where you're located when
forming your financial decisions.

101
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Efficient market hypothesis. The notion of efficient markets and its implications for the
path of stock prices, has been loosely written about discuss for over 100 years. But
Eugene Fama, 2013 Nobel Prize winner, is really known for formally developing and
defining the efficient market hypothesis and its three forms. Three forms of market
efficiency: weak form, semi-strong form, and the strong form. Let's run through to each
of these three.

What's meant by weak form efficient market hypothesis or weak form market efficiency?

102
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
This is just simply stock prices reflect all information contained in the history of past
stock prices and returns. You should be able to make money on a strategy that is
formed based on information about past stock price movements. That would be counter
to the weak form of efficient markets. Past information on stock prices or returns
shouldn't be able to predict future price movements.

Semi-strong form of market efficiency is building upon this. Now, stock prices reflect all
publicly available information. This is what when people talk about efficient markets,
they generally mean the semi-strong form of market efficiency. All publicly available
information is incorporated into the stock price. A firm announces good earnings, stock
price will rise on that, that consistent with efficient markets. Efficient Markets does not
mean that prices shouldn't change. It just means once the information hits, prices adjust
right away and you should be able to make money investing on stale news, if you will.
The semi-strong form of efficient markets, publicly available information such as past
stock price performance, past accounting news. Those should not be able to predict
future returns. This is what people generally mean by efficient markets.

103
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Strong form efficient markets, stock prices reflect all information, even including insight
information.

Efficient market hypothesis is very simple. Assets are going to be fairly priced based on
their future cashflows, given all the information that's currently available to investors.
There are the three parts of what do we mean by information. The semi-strong form,
which when I say market efficiency going forward, I'm going to be referring to this semi-
strong form market efficiency. By information, in this case, it just means publicly

104
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
available information in the past and present should be reflected in the stock price and
should not be able to form portfolio strategies that yield positive risk-adjusted returns
going forward. Capital asset pricing model suggests that the only thing that predicts
returns, is the correlation of firm returns with the market performance or beta. That's
fine. This reflects risk, so the beta predicts returns, is not a violation of market efficiency
because this represents risk.

When we talk about the notions of market efficiency and what it implies for patterns of
stock prices, you really go back to this guy, Louis Bachelier, and his thesis at the
Sorbonne back in 1900. One of the more underappreciated people in the history of
finance academia. His thesis was entitled "The Theorie de la Speculation". My French
high-school French teacher is giving me kudos here for my attempts to pronounce this
here, underappreciated his work in finance. But also, he discovered the mathematics of
Brownian motion before Einstein, or at least it's reported that that was the case, and
Wikipedia can't be wrong.

105
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Some quotes from his thesis, "The market price is set where bullish and bearish
speculators influences are balanced." I particularly like this next quote, "Past, present,
and even discounted future events are reflected in the market price". So today, when
we're doing a firm valuation and firm valuation will be the topic of Module 4 of this
course. All we care about are what are the future cash flows when you're buying a
stock, you're buying a stock because the cash flows, it will be delivered to you in the
future. I like this, even discounted future events are reflected in the market price. Or we
would say the only thing that should be in the market price are the future cash-flows
generated by the asset and then the mathematical expectation of the speculator is zero.

106
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

That's basically saying, hey, efficient markets, you can't pick winners or losers because
all the information publicly available is already reflected in the price. If the firm's
prospects are known to be good, people will be willing to pay more for the stock. The
price will be higher than if they're known to be poor, and the price is fully adjusted to
reflect all the publicly available information as soon as that information hits the
marketplace.

Now Nobel Prize winner Paul Samuelson, he discovered this thesis, Louis Bachelier,

107
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
many years later, and he confessed he oscillated between thinking this was obvious or
brilliant, so is it trivially obvious, trivially vacuous, or remarkably sweeping?

Samuelson ultimately weighed down a remarkably sweeping because he also wrote

papers in this same vein, a famous title of one of his papers here, Properly Anticipated
Prices Fluctuate Randomly, and the key point is randomness and stock returns is not a
sign of irrationality. On the contrary, it's a sign of efficient markets. In an inefficient
market, stock prices reflect all publicly available information so then it's just random. At
that point whether some good news happens or some bad news happens to the firm,
you can't beat the market. That's the notion of efficient markets.

108
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Or you can think of it like a capital asset pricing model. We can predict expected or
required returns of stocks through models like the CAPM, but we can't predict the
unexpected component of returns because any publicly available information that we
think can predict future returns should already be embedded in the stock price today
because it's already known. Remember efficient markets, semi-strong form, all publicly
available information should already be embedded in stock prices today, should not be
able to predict returns unless it represents some types of underlying risk, like the Beta in
the capital asset pricing model.

109
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Paul Samuelson just has a trove of these very provocative quotes. I can imagine, if he
was alive today with Twitter, what he would be spitting out, it'd be very interesting.
Here's a couple of quotes I found very interesting. "Respect for the evidence, this is
Paul Samuelson, compels me to incline toward the hypothesis that most portfolio
decision makers should go out of business, take up plumbing, teach Greek, help
produce the annual GDP by serving as corporate executives." So a plug for the
corporate finance class. Even if this advice to drop dead is good, it's obviously not the
council that will be eagerly followed, particularly in active management part of the
investment business.

110
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

And then also, other advice here from Paul Samuelson's, this would be consistent with
what John Vogel at Vanguard would advocate as well. Investing should be dull, it
shouldn't be exciting like hey, make your allocation between stocks and bonds, adjust
that maybe gradually as your risk aversion changes. Then you're done. You don't have
to go to revisit your portfolio every day to see what's doing. Just make these broad
asset allocation changes that you maybe adjust occasionally over your life. It should be
exciting. Investing should be like watching the paint dry or grass grow. If you want
excitement, go to Las Vegas, if Paul Samuelson is writing this today, or saying this
today, maybe throw Macau in there as well. ''It's not easy to get a rich in Las Vegas or
at Churchill Downs or at the local Merrill Lynch office where you're trading stocks.'' Now
one thing I want to be clear doesn't apply from this statement, so Paul Samuelson is
saying investing should be dull, doesn't apply that a course on investments are
particularly an online course on investments should be dull.

111
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's throw out some tests to market efficiency here. One way to see if markets are
efficient, are they fully incorporating news that comes out is looking at announcements
of various corporate events.

Let's start out first with announcement of a takeover target. Firm A is announcing that
it's making a takeover bid for firm B. When that announcement happens, stock price of
firm B is going to rise.

112
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's look at the abnormal returns of takeover targets here around the announcement
date. Let's look here, this has been a research finding replicated several times. I thought
I'd go back to one of the earlier articles on this, so back in 1981, Keown and Pinkerton
study, they're looking at the returns following announcement of a takeover attempt that
Firm A is offering to buy Firm B, you can see when that announcement happens there's
a big jump in price. But then after that, the price pretty much stays there. This is
consistent with the notion of market efficiency. News hits the market. Firm B is subject
to a takeover offer. Stock price rises on that, but then it basically stays put. Now there
does seem to be maybe some insight information that leaks out that there's a price run-
up in the days leading up to the announcement, so that maybe it could be some insight
information spilling out. But once that announcement happens, stock price goes up and
it stays there. That's consistent with a notion of market efficiency here.

113
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Test 2, announcement of earnings.

This gets the most awkward acronym award here, PEAD, Post Earnings Announcement
Drift. Again, this has been documented in many accounting and finance articles. I
thought I go back to one of the earlier ones in 1984 by Forster, Olsen and Shevlin, and
they're basically categorizing firms, putting them in 10 groupings here based on the
earnings surprise at the most recent earnings announcement. They're looking at, let's
categorize firms based on how their announced earnings, how that compared to what

114
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
was expected based on some model they compute, so the deviation, if you will, of
earnings relative to what was expected, then normalize that by the standard deviation of
the forecast error. Basically, they're just categorizing firms here by the earnings surprise
at the announcement date. Category 10 are firms that had the biggest earnings
surprise, so this is earning surprise normalized by the standard deviation of the forecast
error. If you want to be precise, firms in category one are firms that had the worst
earnings news, and those in categories like 4 or 5 and 6 are those whose earnings that
were announced or about what people expected. Category 10, biggest earning surprise,
category 1, worst earnings surprise. Again, maybe some information is leaking out
ahead of time, like for those are going to have the biggest earnings surprise. Some
evidence or stock price is running up a little bit ahead. Those that had the worst
earnings, some evidence that there, stock price is falling in anticipation of that. But
what's particularly interesting is this post earnings announcement drift part. Look at
those with the biggest earnings. Not only does their stock price go up right after the
announcement, but it continues to drift up. In fact, almost three months after the
announcement, the price is still going up. That's a post earnings announcement drift.
Those that have a big earning surprise, it's not surprising their stock price would go up
on the day of the announcement. That's what should happen good news, stock price
goes up. What is surprising is the positive drift after that over the next three months.
Likewise, on the negative side, for those who have the bad earnings announcement,
they announced earnings less than was expected, stock price falls on that
announcement. Perfectly consistent with efficient markets, that's what should happen.
Bad thing happened to the stock, stock price falls. What's interesting is the negative drift
after the fact. This pattern in return after the announcement, the drift following the
earnings announcement, that does not appear to be consistent with market efficiency.
Another way of saying this, it would be very difficult to predict which firms are going to
beat their earnings expectations, which are going to under-perform their earnings
expectations. That would be tough. The easy part would be just read the Wall Street
Journal and find out which firms beat their earnings expectations last week or last
month, which under-performed their earnings expectation and invest in those stocks.
Such a strategy like that shouldn't yield any returns because you're trading on stale
information, yet the results here seem to indicate there are some returns to this drift
strategy.

115
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

If you're interested in looking further into drift strategies like this. In my second course in
an investment, we're going to tackle some of these issues are going to look more at the
post-earnings announcement drift strategy. We'll also look at the drift following the
announcement of various corporate events, such as stock splits or payout policy
announcements, or decisions of the firm or to issue or buy back various types of
securities. We'll look at the announcement effect following those corporate actions, but
also look at the drift thereafter. If you're interested in this, more to come in my second
course on investments.

116
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Now that we talked about the post-earnings announcement drift, we already know about
firm size, value growth, and momentum predictability in returns. Right off the bet, it
seems like, hey, these are violations of market efficiency, like momentum strategy is
based on past returns. Even the weak form of market efficiency seems to be rejected by
the momentum strategy ability to predict returns past one-year returns seem to predict
the firm doing well over the next month. Firm size, value growth. again, these are known
commodities shouldn't be able to predict returns. Seem that they are again violations of
market efficiency. Then we talked about the post-earnings announcement drift strategy.
These characteristics certainly predict returns or at least have in the past.

117
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Does that mean that they are automatically violations of market efficiency? The
evidence seems, maybe that's the case, but not necessarily. Now the question is, do
these patterns and return represent some types of risks? Now for market efficiency
advocates who want to claim the markets are fully efficient. They have to explain these
patterns and returns, have to tie them to some types of risks, and then the onus is
what's a risk story that explains the return pattern in momentum or in value stocks, or in
small stocks, or in post-earnings announcement drift.

118
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's think about efficient markets and bring that to thinking about the different types of
asset management. For a passive strategy, this would be kind of the John Vogel of
Vanguard or the Paul Samuelson view of the world by a well-diversified portfolio without
trying to differentiate winners from losers, basically by an index fund is like the passive
strategy. The active strategy is like, hey, there are predictable patterns and returns. We
don't think those patterns and returns represent risks. They just represent great profit
opportunities. We're going to act on those patterns of return to earn Alpha, if you will, to
get surplus profits, we're going to find mispriced securities. Historically value stocks are
priced too low, therefore they give the high returns and take advantage of that. Those
strategies worked very well for folks like Warren Buffett and Josef Lakonishok and we'll
short those stocks that have underperformed. That's the active management strategy.

119
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Passive management view of the world and I showed this earlier in the course, but I
love this depiction. I'll do it one more time. The passive management view of the world
as, why should you hunt for the haystack? In an efficient market, you can't identify
winners from losers. It's just going to be random because any information you have to
try and predict returns. If that information is publicly known, it should already be
incorporated in the stock price. Maybe you have some insight information and that isn't
incorporated in the stock price. But then if you trade on that, you may run afoul of some
regulations. Any strategy that you can post based on publicly available information
shouldn't yield abnormal returns because it should already be reflected in the stock
price. Don't try and look for the needle in the haystack. Don't try to identify the winners.
Instead, hold the index fund and of course, for this one, I love the small-town headline
here. Jim Moran here, huntsman needle and the haystack. Having found the attention
he wants, he abandons the needle hunt, and John Vogel of the world that Paul
Samuelson of the world would say exactly, you're going to have to abandon the hunt for
the look of winners because you're going to find is just very, very difficult to find stocks
that outperform in an efficient market setting. You shouldn't try to do so in the first place.

120
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Now the active asset management view of the world is very different. Here it's saying,
hey, the needle in the haystack is actually pretty easy to find. It's called value stocks, it's
called momentum, it's called post-earnings announcement drift. Scott has already told
you about all these things. The needle here is pretty large. Why don't we invest in it?
Earn positive Alpha, the risks stories efficient market advocates tell about these patterns
and returns. They don't really hold water, we think is just representing Alpha or excess,
abnormal performance, surplus profits for us. Let's take advantage of that. That's the
active asset management view of the world.

121
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

In my second investments course here, here we're laying out efficient markets. But
remember, this course is like one-half of a typical investment course. In the second half
we're going to talk more about actively managed mutual funds and in particular, are
there any areas where fund managers seem to have an informational edge. Just to
preview results, if you look on aggregate, active mutual fund managers holdings don't
seem to outperform their benchmark suddenly after fees and most of the time it seems
or most of the studies, even before expenses, the active manager portfolio holdings out
are not outperforming their benchmark, but that doesn't mean they're not sub pockets of
the portfolio of mutual fund managers that actually is earning Alpha and is
outperforming. That's what we'll be talking about in my second investments course.

122
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Going back to this course, efficient markets, separation theorem have a simple
implication. Investor should be holding index funds, then the separation theorem. You're
just basically determining the mix between a risk-free asset and the market portfolio. If
you like risk, you'll hold a lot of the market portfolio. If you don't like risks, you're holding
a lot in treasury bills and that's it.

But that raises a question, how can the market be efficient if everyone is using a
passive strategy?

123
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Think of this Grossman-Stiglitz paradox to economists here, Grossman and Stiglitz. So,
let's just follow this trend of thought. If the markets are efficient, then there's no gains
from doing research. If there's no gains from doing research, then nobody does
research. If nobody does research then asset prices can't reflect all the relevant
information, and then markets are inefficient. If markets are inefficient, then it's profitable
to do research. Then repeat and repeat and repeat. There seems to be a tension if the
markets are fully efficient, no one does any research, no one's paying attention to the
news on various stocks. When good news happens, stock prices don't adjust because
no one's buying or selling to reflect the new information embedded in the stock. That
leads to a paradox. How if markets are efficient, no one's doing research, no one's
keeping abreast of current information, but that can cause prices and not to reflect
fundamental new information, which would mean the markets are inefficient, kind of a
paradox here.

124
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

There's a constant struggle. You can imagine having two types of traders like what we
call smart money, or arbitragers, dumb money. noise traders. The dumb money, the
noise traders, they're the ones with behavioral biases that may have the potential to
cause prices to stray from fundamentals, which could lead to some predictable patterns
and returns. They could yield profits for the smart money or arbitragers. Prices in
particular can stray from correct level if the arbitragers do not have enough capital to
undo the trades of the noise traders.

125
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Another relevant consideration, and this is a way how efficient market advocate and the
more of the behavioral types can coexist. Maybe we can find predictable patterns in
returns before transactions costs. But they're not profitable for investors after
transactions costs. That way, maybe there is some return predictability, but it can't really
be taken advantage of because of the transactions costs. Once you take into account
the transactions costs of the strategy, there really is no money-making opportunity. In
that sense of markets are still efficient at least after transactions cost. For example,
when we talk about the momentum strategy, one of the issues with that is the
transactions costs involved really eat away at this return from the momentum strategy in
particular.

126
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Market efficiency and asset management, if markets are fully efficient, then it doesn't
matter how you choose stocks. Firm-specific shocks are unpredictable and efficient
market view of the world because all relevant information that's publicly available is
already in the stock price, so it's just random. What happens in the future if you already
knew about it, you would already be in the stock price. Trying to pick stock winners is
like going to Macau or going to Las Vegas.

If the markets are fully efficient, you might as well just randomly threw darts at the Wall

127
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Street Journal or Financial Times. A few stock picks will do really poorly. Many will do
about average, a few will do really well. It just happens like when you through it, did you
hit Enron or when you throw the dart, did you hit apple? It's just luck of the draw here. If
you think about this in terms of investment managers in this efficient market setting,
some will do well, some will do poorly. Again, focusing on active managers, but just in
the efficient market setting, some will do well with their active picks, some will do poorly.
But all of this will be attributable to luck because there are no abilities to form profitable
portfolio strategies on past publicly available information in an efficient market setting.

What if markets are efficient? Maybe you should just leave your asset management to
this guy here. This is my son. He'd be happy to randomly pick some stocks for you. He's
open up for business in 2035, he only charges one-and-a-half percent of the assets
under management. If you're looking for some random stock picks, he'll be there to help
you out.

128
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

But maybe in 2035, my son might be ready, but now probably doesn't make sense to
hire him to be your financial advisor. Even if markets are fully efficient, security analysis
is still important. It's better than, throwing darts. A couple issues to take into account.
One is taxes. Taking into account, at least in the US, when there's capital gains tax, tax
on dividends, and the capital gains tax occurs when you sell an asset, not when the
return occurs, but when you sell the assets, when you pay the capital gains tax in the
US. Taking into account, tax strategies, trying to minimize taxes can increase your after-
tax returns. That won't be taken into account by an investment strategy of my son, of
just throwing darts at the Wall Street Journal. Also, security analysis can take into
account your individual situation. For example, if you work at General Motors, it's
probably not a good idea to put a bunch of your own financial assets in General Motors
stock. You want to take into account the background risk. If you live in Chicago or New
York, you probably don't want to invest all your financial assets in Chicago or New York
firms, you probably want some broader diversification. These are factors that can be
taken into account by financial advisor security analysis that don't take into account by
just like randomly throwing darts.

129
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

You can see this, two more of these back-to-back will have. Stay tuned. More things
that we won't be covering in the first course here, which is really like the first half of
investments, but we will be tackling in the second course one, tax timing strategies and
seasonality or return. We'll talk about what are ways that even in an efficient market,
you can earn abnormal re turns, earn excess returns by just take into account the tax
rules regarding capital gains taxation. Basically, providing you with some tax strategies
to increase your returns and who doesn't want increased returns. Then also talking
about how the capital gains tax rules and the US may contribute to this pattern of
returns known as the January effect which is particularly pronounced for small stocks.

130
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Then while the back-to-back stay tuned, also talking about firm dividend policy,
individual tax rates, and their effect on stock returns and firm values. In this course, we
just thrown out the term return. I defined it in module one at the start is the return has
two components, a capital gains component and a cash flow component, a dividend
component. But then after that, we just said, we're looking at returns without taking into
account like is there a term coming from capital gains, a change in the price, or is it
coming from dividends. It's interesting to look how does the composition of the return,
whether you're getting capital gains like a change in price, or you're getting dividends.
For example, for Microsoft up to 2003, all the returns for Microsoft stockholders were in
the forms at capital gains. Since then, a lot of the returns to Microsoft shareholders have
been in the form of dividends. How does that affect the stock return pattern? How does
that affect the firm value when the firm has more dividends as opposed to capital gains?
How do individual tax rates on dividends and capital gains affect the returns firms give
to investors? All those topics will be covered in more depth in the second course I'm
offering.

131
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What have we learned in this lesson? Market efficiency and when people talk about
market efficiency, they're referring to the semi-strong form. Stock prices reflect all
present and past publicly available information. Accounting data, firm news release,
past stock returns and stock prices, all of that information should already be embedded
in stock prices today. Even in a fully efficient market, active fund managers, some of
them will do well. Even if the market is fully efficient, some managers, they are actively
picking stocks will do well. Some will do poorly. But in efficient markets environment,
those are doing well it's not due to their skill. It just represents luck. When you throw it
the Wall Street Journal, one manager happen to get Apple computers, that did very
well. It wasn't really due to scale. It just represented luck, Again, this is in the efficient
market setting for those who think the markets aren't fully efficient and that patterns and
returns don't represent, they would say, hey, it represents skill of the manager to
recognize these pattern returns and take advantage of this. This is a big debate
ultimately for you to decide. I'm just bringing you both sides of the argument. Ultimately,
and this is for you if you're thinking making an investment, the natural tendency is to
look at the past returns and say, hey, past returns are good, future returns will be good.
This manager knows what they're doing, but it's hard to differentiate luck from skill.
Remember, even in an efficient market setting, some managers will do well just by luck
of the draw. Then finally, taxes individual circumstances are important when managing
your portfolio, even in an efficient market setting, where it may be hard to differentiate
and to pick winners versus losers or impossible if the markets are fully efficient. Taxes
and your individual circumstances where you live, what job you have should affect your
financial decisions.

132
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Module 3 Review and Graded Activities

Module 3 Review

Module 3, can you believe it? Time to wrap it up. Already three-quarters done with the
course. I thought in wrapping up Module 3, I'd first show some of the key charts and
figures that we had during the module, and then wrap up with the key takeaways or
famous, what-have-we-learned-slide?

133
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

One of the key figures is produced by Fama and French back in 2004, was documented
how value stocks, how the outperformed growth stocks in ways that would not be
predicted by the capital asset pricing model here. Fama and French form 10 portfolios
of stocks, and within each portfolio valuating, Portfolio 10 campaigned the stocks with
highest value characteristics, the highest book-to-market, so think of the cement
company with a lot of tangible assets. The book value is high relative to the market
value compared to the Internet company where it has like maybe a computer server at a
great idea, but that's about it. The book market, the book value is low relative to the
market value. Portfolio 10, high-value; Portfolio 1, low growth. What was observed when
you look at the data? Well, the capital asset pricing model has this prediction for all of
these securities here, like their predicted return of the portfolio should be here on the
line according to the capital asset pricing model. But instead, we observed value stocks
have yielded historically 1963-2003, value stocks have historically yielded a much
higher return than would be predicted by the capital asset pricing model. On the other
hand, growth stocks have actually underperformed the benchmark set by the Capital
Asset Pricing Model. This value growth breakdown seems to certainly have predicted
returns historically in a way that's inconsistent with the Capital Asset Pricing Model.
CAPM has a very simple prediction. The only thing that you'd be able to predict returns
going forward is Beta, not value growth.

134
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

We also looked at the return of small stocks and we see this pattern of small stocks
yielding and we talked about this in Module 1 as well when we talked about historical
returns. The bottom 10 percent of firms ranked by the market value of their equity.
Average return of 19 per cent per year from 1927 to 2014 compared to 11 percent per
year for the top 10 percent of firms ranked by the market value of their equity. You had
this difference in performance here on the order of about eight percentage points per
year. The bigger firms have lower returns than the smaller firms. On first blush, this
seems like, oh, this is also strong evidence against the CAPM, like small firms doing
better. Why is that? Well, it could be small firms are riskier. Once we do the CAPM
adjustment, we risk-adjusted based on differences in market risk about 2/3 of this eight
percent differential goes away, is explained by smaller firms being riskier than large
firms.

135
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

You can see this here when you're looking at the Betas of the small firms versus large
firms estimated with annual data, again, from 1927-2014. As we increase the size of the
firm, the Beta falls from a little over 1.5 for the smallest firms down to 0.9 for the largest
firms. The largest firms have less risk, less sensitive to the economy relative to small
firms. That's part of the reason why they have a smaller return. But of that eight percent,
about a third of that still is outperformance of small stocks relative to large stocks in
terms of Alpha. The small factor, size factor to a limited extent and the value factor to a
larger extent are seemed to be inconsistent with the Capital Asset Pricing Model, which
says the only factor should predict returns. The only risk that investors should care
about is the Beta risk.

136
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Now when we go to a corporate finance setting, nonetheless, Capital Asset Pricing

Model is the key game in town. When we're looking at what method do you use to come
up with discount rates, to come up with your cost of equity capital or your required
return that investors have in your stock, CAPM is by far the leader 75 percent of CFOs
reporting, they always or almost always use the Capital Asset Pricing Model to come up
with a required return on equity. Three-factor models, less than half of those that report
using the CAPM are using the three-factor model.

137
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

We also talked about market efficiency and getting a test of market efficiency looking at
stock market reaction to various corporate announcements, and here there's a mixed
bag when we look at the return following takeover target announcements. When the
takeover is announced, firm A is acquiring firm B. Firm B stock price goes up a lot, good
news is coming in. Maybe there's a run up in price of firm B ahead of time due to some
insight information leaking out that it may be a target. But once it's officially announced,
stock price rises a lot and then it stays flat. This is consistent with our notion of market
efficiency.

138
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Post-earnings announcement drift seems not to be consistent with market efficiency. It's
no surprise if a firm announces a positive earnings surprise, the stock should go up on
that news. What is a surprise and what does run counter to market efficiency, is it the
price continues to drift up 50 plus trading days after the announcement or going on, two-
and-a-half, three months after the announcement. On the other side, no surprise, firms
announce bad earnings report, stock price should fall. What is surprising and runs
counter to market efficiency Does it continue to drift down afterwards? This is a post-
earnings announcement drift. It would be hard to predict which firms are going to beat
earnings estimates, which are going to under-perform earnings estimates. If we knew
that we could really make a lot of money. This strategy, the post-earnings
announcement drift strategy is very simple. We don't need to know which firms
outperform their earnings estimate, which underperform their earnings estimate, we just
have to read the Wall Street Journal, see what firms do what, and then invest
accordingly, invest in those that had the good earnings announcement, short those that
had the bad earnings announcement, and that generates this lucrative investment
strategy, portfolio strategy after that. That run seems to run counter to market efficient.

139
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What have we learned in Module 3, a recap? Beta generally does not predict returns as
strongly as the capital asset pricing model would suggest it should when you look at the
historical data. One interesting piece of research recently document that the capital
asset pricing model actually does a good job at predict returns, but only on days when
there's macroeconomic news announced such as inflation numbers, interests rate
changes by the Federal Reserve, employment numbers. On those days, capital asset
pricing model does a pretty good if it's a good job at predict returns. If it's good
macroeconomic news, the stocks that go up the most are the stocks with the highest
Beta. But on days when there isn't macroeconomic news announced, there's basically
no correlation between Beta and returns even though an average or we think there
should be a positive correlation. Beta does seem to predict returns somewhat, but not
as much as the CAPM would suggest as a bottom line. We've seen other factors that do
predict returns, look at the historical data. Firm size, value, beating growth, and
momentum are factors that are associated with being able to predict returns. Don't
come from the capital asset pricing model that says the only factor should be able to
predict returns is Beta.

140
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What do we make of this, market efficiency. What's your view of that? Market efficiency
is simply a stock prices reflect all past and present publicly available information. These
factors here, firm size, value, growth, momentum, these are all characteristics that are
known so they shouldn't be able to predict returns, but they have historically. What do
we make of this? One view is clear violations of market efficiency. Another view is
maybe these factors represent some type of risks that investors care about so them
being able to predict returns is not a violation of market efficiency, it just means firms,
stocks with these characteristics are riskier and on average should generate high
returns. Now remember, even if there's a fully efficient market, some active
management decisions will yield positive Alpha, will yield great returns, others will yield
poor returns, a lot will just get basically the same return as the market or an index fund.
In a fully efficient market, no information out there. Some asset managers will still
outperform just due to luck, a great example of that was Assignment 4. You need to be
very careful as an investor in the real-world not to just simply invest in funds because of
the high past returns because that could represent some component of luck. If I was
investing with an active fund strategy based on the past good performance, I would like
to know, well, what caused that good performance? What was the investment
philosophy? Does that make sense, our priority and do I expect that to continue going
forward? If you're confident that the answers to all those are in the affirmative, it's a
good strategy that you think exploiting some behavioral bias. It's likely to continue, then
maybe you're more comfortable going with the active management strategy than if you
knew it was just generated essentially by randomness. Bottom line, be careful
differentiating lot from scale.

141
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What's around the bend?

142
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
ASSIGNMENT 4 (Lesson 3-7): Analysis and Recommendation of 50 Balanced Funds,
1995-2014

This lesson is about introducing assignment Number 4, and at this point you know the
drill. First, do the assignment on your own, and then come and you're welcome to see
my discussion of how I solved assignment and my thoughts on Assignment 4. One of
the key takeaways from this assignment is understanding differences between the
CAPM and the 3-Factor Model. Then also how difficult it can be to differentiate luck from
skill when you're evaluating the performance of asset managers using past data.

143
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

For this assignment, we're going to be using data that I've prepared for you from the
spreadsheet, the name says what it is, balanced funds 1995-2014. You know already
contains data for you on 50 balanced funds over the 20-year period, 1995 to 2014. Now,
each of these balanced funds was invested in some combination of US stocks and
treasury bills. They have somewhat different equity way to cross the funds. Within
equity they have different investment styles. In this spreadsheet, we have the 50 funds.
Average annual performance of these 50 funds in terms of an excess return in excess
of the treasury bill rate is provided. Sharpe ratio over the 20-year period is provided.
Estimates from a capital asset pricing model and then estimates from the 3-factor model
done on an annual basis are provided for you to look at and evaluate these 50 funds.
Why don't we go to Excel and let me run you through the spreadsheet. Then we'll come
back and we'll talk about the particular questions I have for you using this spreadsheet
for Assignment 4.

144
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's look at this spreadsheet that will serve as the basis for assignment Number 4 here
in Module 3. This is a spreadsheet; you can see that has various balanced funds. To
maintain anonymity here, this fund ID is just a number from one down to 50. Then here
at the bottom, you can see I report averages across the 50 funds. I provide various data
for you summarizing the annual returns of these 50 funds over the 20-year period, 1995
to 2014. Again, all these returns are presented on an annual basis in percent. Just
going down the columns here, first column is the fund ID. Second is the average fund
return minus the risk-free rate, treasury bill rate. This is just the average excess return.
For Fund 1, it was 5.6 percentage point return in excess of treasury bills over this 20-
year period.

145
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Scroll down to the bottom of the spreadsheet, just don't scroll down too far. I gave you
what are the averages for other returns over this 20-year period for the overall stock
market raw return just under 12 percentage points per year. Risk-free rate treasury bills
2.7 percent. That's a useful number to remember.

Go back to Fund 1, take this excess return of 5.6 percent per year. Add back to it the
2.7 percent, you get a raw return of Fund 1 of 8.3 percentage points. For Fund 2, the

146
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
excess return of six percentage points. Add back to it the risk-free rate, average return
of 2.7 percent over this period. Raw return of 8.7 percentage points for Fund 2. Going
on through the columns, standard deviation of the returns. Given the excess return in
Column B, the standard deviation Column C. Take the ratio of those, you get the
Sharpe ratio on the annual data in Column D. Columns F and G are giving the outputs
from a capital asset pricing model or CAPM regression. Again, using these 20 years of
annual data. The Alpha for Fund 1, for example is 2.1 percentage points per year. On
an annual basis over this 20-year period, CAPM Beta 0.38. Also, the results from a 3-
Factor Model. The 3-Factor Alpha here from the Fama-French 3-Factor Model, 2.1
percentage points per year, this is using annual data. The CAPM Alpha and the 3-factor
Alpha are exactly the same. The market Beta and the 3-Factor Model, the coefficient
there. Then also the coefficients on the small minus big, the size factor 0.05 for Fund 1.
Then the coefficient on the HML value minus growth factor minus 0.01 here. If you look,
I've highlighted six funds in yellow. These six funds here are funds that I asked you to
describe why there's big differences in the Alpha from the CAPM model relative to the
three-factor model. Just to go through one of these for Fund 7, we look at the CAPM
Alpha, the excess performance beyond the benchmark and the CAPM model, 2.1
percentage points per year is the Alpha. Go to the 3-factor model, it actually increases
substantially to 2.9 percentage points per year. Why is there this difference? Now that
we've outlined what's in this spreadsheet of the 50 balanced funds for Assignment 4.
Now it's your time to actually do Assignment 4, then after you're done, see my
discussion.

As you look in the Excel spreadsheet, you'll notice for some funds and there's a big

147
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
difference between Alpha from the capital asset pricing model and Alpha from the three-
factor model. Remember all of the results shown to you in the spreadsheet regarding
the performance of these 50 balanced funds, all of these are based on annual data or
annual returns. When we looked at the difference in the Alpha from the capital asset
pricing model and the three-factor model, when you look at Fund 31 for example, the
Alpha is only 0.1 percent per year and the capital asset pricing model basically zero, but
it's one percent per year in the three-factor model. Remember these are the results
estimated over 20 years of data. One percent extra performance are beating its
benchmark per year on average for a 20-year period, pretty impressive. What's going on
Fund 31 where there's this big difference in Alpha from 0.1 in the CAPM to 1.0 here
when we get to the three-factor model. Fund 50, on the other hand, once we go to the
three-factor model, the Alpha gets lower. It seems like it's not a good fund or hasn't
performed well. Historically, regardless, the Alpha's about minus two percent per year in
the CAPM. That's an average underperformance, a two percent per year over a 20 year
period. This has done fairly poorly, but once we go to the three-factor model, setting, the
Alpha is now minus three percent underperformance. Those are two examples where
there's big difference in performance.

Question 1 is building upon this for six funds: Funds 7, 9, 26, 27,31, and Fund 50. For
these six funds in the spreadsheet, they all have big differences between the Alpha from
the capital asset pricing model and the Alpha from the three-factor model. Very different
views of how are they performing relative to the benchmark, depending upon how the
benchmark is defined and estimated. For each of these funds, why is there such a big
difference between these two Alphas?

148
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Question 2, which of the 50 funds would you most like to invest in going forward and
why?

Then you can maybe guess Question 3, which of these 50 funds would you least like to
invest in going forward and why? What's a basis for that recommendation?

149
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Now to some numbers to keep in mind, these are also at the bottom of the spreadsheet
here when you're evaluating these 50 balanced funds here, other returns to keep in
mind. The overall stock market about 12 percent over this 12-year period. Just happens
that the average of stocks from 1927-2014 of about 12 percent, that's about what you
have over this 20-year period. It's a representative period in the sense of there are both
ups and downs. The average US stock market return over these ups and downs over
the 20-year period, about 12 percent. One-month treasury bills, a little under three
percent here. You have an excess return of the stock market on the order of 9.2 percent
here, which may be a little higher than average, but not too much. Small stocks beat big
stocks and average over this period with a differential of 1.9 percent. Value stocks beat
growth stocks over this period here with a differential on the order of three percent.
Some useful numbers to keep in mind when looking over the performance of the 50
balanced funds for assignment Number 4.

150
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

At this point we're very familiar with curious George, and the game plan for doing the
assignment and then reviewing the discussion.

Basically, my thought processes, I want to be able to provide you with my discussion of

the assignments here, particularly for the first four assignments anyway. The rule and I
don't need to tell you this. You join Coursera to take these courses for the learning
experience. First, do the assignment on your own then check out my discussion. You're

151
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
always free to adjust your answers after seeing my discussion, before submitting the
assignment. That's not a problem, but just give the assignment the college try. Don't fast
forward and go directly to my discussion. Then finally, before you review the
assignments or grade other people's work, review my discussion first just to make the
grading and review process of other people's work much smoother.

152
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
DISCUSSION OF ASSIGNMENT 4 (Lesson 3-7): Analysis and Recommendation of 50
Balanced Funds, 1995-2014

Let's discuss Assignment 4. This was an analysis of the 50 balanced funds. The key
motivation behind this assignment was to test your understanding of the capital asset
pricing model, the three-factor model, differences between the two, what difference is in
Alpha's between the capital asset pricing model and three factor model, what they
represent. To do this, I gave you a return data over the last 20 years for 50 balanced
funds with varying degrees of performance, varying degrees of investment styles.

153
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Don't be alarmed, but there are two slight errors in the discussion of Assignment 4 and
Lesson 3-7. They appear on this slide, as well as a couple of other points in this
discussion. The average raw return of the 50 balanced funds is 7.2% per year, not
7.5%. The average CAPM Alpha is -0.3% per year, not 0%. I hope the slight errors
weren't keeping you up at night. Note that these two errors have no effect on answers to
the questions of Assignment 4.

154
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

I reported in the Excel spreadsheet excess returns, returns in excess of the one-year
treasury bill rate on an annualized basis. It's useful in this discussion going back by
screen, this spreadsheet now, I probably would have just given you the raw returns. For
the average 50 balanced funds in the spreadsheet, the average raw return was 7.5%
per year. The excess return of 4.8 plus 2.7 here risk-free return. The CAPM Beta 0.52
on average, and the CAPM Alpha of 0% on average. The average excess return of the
50 funds was 4.8%. To get this to a total raw return, adding the 2.7% of the risk-free
rate to give us this average fund raw return of 7.5% per year across the funds. The
CAPM Beta is about 0.5, 0.52 to be precise. The CAPM Alpha, the average of that
across the 50 funds, literally zero.

155
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Comparing this to the overall environment here, what were other assets yielding stock
market 12% in average over this 20-year period, 1995-2014? Treasury bills 2.7% on an
annual basis. Small stocks outperform big stocks over 1995-2014 on the order about
2% per year, while value bill growth on the order of about 3% per year over this period.

Here again, we're looking at the raw returns per year. This is basically taking the excess
return from the spreadsheet, adding back in the risk-free rate, which averaged 2.7%

156
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
over this period. Now these numbers down here on the x-axis are not fund ID numbers.
Instead, this is just the ranking of the funds. One is giving you what the lowest return
was. The lowest return fund 50 is just giving you the highest return. What do we see for
these balanced funds that have some combination of US stocks and treasury bills as
investments? Remember, the average raw return was 7.5%. That's roughly what the
median return is here. The high return is just under 10% of the funds, and the worst-
performing fund yielded a raw return of 4% per year. Now what if we put these funds
over 1995-2014 using the annual data? What if we put them through a capital asset
pricing model calculated in Alpha and Beta to bench market adjust?

Because different funds have different equity tilts, maybe investing in different types of
equities. What if we adjust for the systematic risk of the different investments across the
balanced funds? Remember that average Alpha and this is Alpha from the CAPM in
percent per year. Remember, this is the average over a 20-year period. These were
based on returns over 1995-2014. It's not a one or two-year snapshot. These Alphas
were estimated using annual data over a 20-year period. The average Alpha was zero.
You can see here for the median fund, it looks like the Alpha slightly less than zero, but
around the middle of the distribution here the Alphas are pretty close to zero. The best
performing fund in terms of this risk-adjusted return. As you know, a little over just under
two-and-a-half percent per year, the worst-performing fund is underperforming its
benchmark by three percentage points per year on average over a 20-year period. This
historically has really been a dud of an investment here.

157
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Let's do some analysis of these various balanced funds and what these various
parameters from the models represent, the CAPM Beta. This is giving us the average tilt
of the portfolio toward equities. For example, a portfolio that's 50 percent US stocks, 50
percent treasury bills would have a Beta of 0.5. If it was 80 percent stocks, 20 percent
treasury bills, that would have a Beta of 0.8. Why I'm assuming when I say US stocks
invest in the US stock market index. Now, if we look at the CAPM Beta across the 50
balance funds, it varies from 0.32 to 0.68, so that means for one fund, the systematic
risk of that fund is equivalent to a fund that's a 1/3 in the US stock market, 2/3 in the
risk-free rate, where there's another fund that's exactly the opposite. On average, it has
the same systematic risk as a fund that's a 1/3 in treasury bills, 2/3 in the US stock
market. The CAPM Alpha. If we look at, let's adjust, we want to compare fund managers
across a different balance funds. Some funds are going to have higher returns because
they're taking on higher risk, they're investing more in Tiffany's less than McDonald's, or
they're investing more in stocks in general, less than T-bills. On average, we'd expect
those fund managers to generate higher returns because they're taking on higher risk,
the CAPM Alpha is adjusting for that, and by risk in the CAPM model is just talking
about systematic risks, market risks, sensitivity of the performance of your asset or
security to the overall market. Once we account for the appropriate CAPM benchmark,
the systematic risk of the investment, the equity tilt of the fund, we're getting
outperformance as high as 2.3 percent per year, again, on average over this 20-year
period, and underperformance as low as a -3.2 percent, underperforming the
benchmark by 3.2 percent per year for one of the funds. But again, the average across
all the funds happens to be about zero.

158
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

When we go to the 3-factor model, the Betas, for example on the small minus big factor,
the high minus low, which I would have called value minus growth factor. They give us
clues about the underlying investment style of the portfolio. When we look at the Alpha
on the 3-factor model that adjust to take into account this investment style. So, a way to
think about when you do the 3-factor model, fund managers don't get credit for investing
in small value stocks in the 3-factor model, that's taken into account, instead of
appearing in Alpha, that appears in the Beta on the size and value factors. Likewise,
investment managers are focusing on large CAPM growth stocks don't get penalized for
that in the 3-factor model. We know those are assets or securities that have
underperformed historically, that's taken into account in the 3-factor model. It shows up
in the Betas, doesn't show up in the Alpha. You know, simply investing in a small value
index fund won't yield any Alpha on the 3-factor model, although it has historically in the
capital asset pricing model.

159
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

One simple measure that we start out with a Sharpe ratio. This just gives us the simple
bang for the buck, the average excess return of the stock divided by its standard
deviation or, you know, total volatility on that metric. Now, that isn't differentiating, all
risk is treated the same because the standard deviation of the returns reflects both
idiosyncratic risk and systematic risks. The Sharpe ratio, excess return over the total
risk, if you will, or the total standard deviation volatility of the investment Fund 1, Fund 7,
Fund 47, all have particularly high Sharpe ratios in terms of bang for the buck, excess
return over the volatility.

160
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

How about taking into account the market risk of the portfolio? Different portfolios have
different loadings on equity versus treasury bills. Some of them are investing more in
stocks, some are investing less. Those that invest more in stocks, given the average
stock return is 12 percent over this 20-year period, they're probably going to do better
than funds that invest less in stocks over this 20-year period. The capital asset pricing
model will take that into account, so we can more fairly compare asset managers of
portfolios that have different investment strategies. Let's compare, for example, to
illustrate these two funds. Fund 47, the raw return, again, the excess return plus the
average T-bill rate over the 20 years gives us a raw return of 9.6% for Fund 47. For
Fund 1, the raw return is 8.3%. Again, it's excess return plus the average annual T-bill
rate gives us 8.3%. Fund 47 yielding a higher return than Fund 41, a difference of 1.3
percentage points per year on average over the 20-year period. A pretty substantial
outperformance here. But what this 1.3% here doesn't take into account is that Fund 47
is putting the foot on the gas pedal, putting the foot on the accelerator a little more than
Fund 1 is. For Fund 47, the Beta is 0.5. For Fund 1, it's 0.38. Fund 47 is generating
higher returns, but it is taking on more risk and that it has a higher tilt toward equities on
average. Once you take that into account in terms of the capital asset pricing model in
calculating an Alpha, the outperformance relative to the benchmark that takes into
account market risks takes into account the equity tilt. Both of these historically have
done very well, both Fund 47 and fund 41. But now the differences in their performance
risk adjusted is only 0.2% as opposed to this original 1.3%. Fund 47 has an Alpha 2.3%;
Fund 1 has an Alpha 2.1%. Another way to think about this, of this difference in returns,
a 1.3%, 1.1% of that 1.3 minus 0.2, 1.1% of that 1.3 is accounted for just differences in
underlying risk of the investment strategies between the two funds.

161
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

We just talked about two great funds, at least great in terms of the historical
performance fund number 1, fund number 47. On the flip side, Fund 23 has performed
terribly. On average it invests 50% in the stock market, reflected by a Beta 0.54.
Another way to say this is the systematic risk taken on by Fund 23 is the same as a
fund that's 54% invested in the US stock market, 46% invested in treasury bills. That's
what the Beta 0.54 represents. The equity tilt to this fund is about average across the
funds in the sample are actually a little higher than that average Beta 0.52. Little higher
than average equity allocation relative to the other funds.

162
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Despite that, its performance is the second worst. Its raw return is 4.5% per year. The
excess return of 1.8 plus a 2.7 risk-free rate, 4.5% raw return, second lowest in the
group of 50, even though it has a slightly higher than average tilt toward equities. It's
really doing poorly. The average return across all the funds was 7.5%. Once you take
into account its tilt toward equities, it has this benchmark adjusted return of -3.2% per
year. In other words, given its tilt toward equities, on average, it should have been
yielding 3.2% point higher return on average per year for 20 years than it actually did.
Instead of generating 4.5% given its risk in investing in equities, it should have been
generating 7.7%.

163
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Question 1 was, why are there differences in Alphas across the CAPM model and the
three-factor model. I identified six funds here with dramatic changes here.

Let's go through these one-by-one. Fund 7, you see the alpha increases by 0.8
percentage points per year when you go from the CAPM model to the three-factor
model. What is leading to this increase in outperformance of the benchmark? Well, what
the three-factor model is taking into account its Fund 7 has a tilt toward growth stocks,

164
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
indicated by this negative Beta on the value minus growth or high minus low factor. This
negative coefficient represents when growth stocks are doing well. HML is negative,
negative times a negative coefficient is positive. In other words, this negative coefficient
here means when growth stocks are doing well, that's a period where this fund is doing
well, so it has a growth tilt. Given growth stocks historically have outperformed value
stocks, this 2.1%, outperformance in the CAPM is even more impressive. It's even more
impressive to generate this high return given you're swimming upstream investing in
growth stocks which have underperformed during this period, so therefore, when we go
to the 3-factor model, the Alpha goes up a bit because it's taking into account that this is
a manager who was investing in growth stocks given this negative coefficient on the
value minus growth factor. Growth stocks have underperformed relative to value so
therefore, we adjust the outperformance up. Or another way to say it is we lower the
benchmark given the tilt toward growth stocks for Fund 7, so the 3-factor Alpha is
greater than one factor or CAPM Alpha.

Fund 9, the opposite happens. When you go from the CAPM model, the
outperformance of 1.2% per year drops to 0.7% of the 3-factor model. Well, why is this?
Well, this is because one of the reasons for the good performance in the CAPM for
Fund 9 was the sensitivity to small stocks. When small stocks are doing well, SMB is
positive. Positive coefficient on small minus big means this fund is doing well. One of
the reasons that this fund did well is it invested in this small strategy which historically
has beat large-cap stocks, so we need to make our benchmark a little higher for Fund 9
in the 3-factor model, therefore, the Alpha falls. The performance of the manager of
Fund 9 is less impressive once we take into account the investment style.

165
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

For Fund 26, we have an improvement in performance. They're negative in both, but the
CAPM Alpha minus 1.8%, we actually bump that up to -0.7% of the 3-factor alpha.
Why? Because Fund Manager 26 was really swimming upstream. The investment tilt
was away from small stocks toward large stocks given by this negative coefficient or
indicated by this negative coefficient also toward growth stocks, given by the negative
coefficient here on the HML or value minus growth factor. Fund Manager 26 seem to be
investing in both large-cap stocks and growth stocks. Both of them have outperformed
so we should lower our benchmark for this manager when we go to the 3-factor model,
therefore, the Alpha is reduced as a result. In the case of Fund Manager 26, the
investment performance is less bad once we adjust for the investment style that is
swimming upstream, investing in both large-cap stocks and growth stocks.

166
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

For Fund 27, the investment performance is viewed as less impressive. Once we go to
the 3-factor model, the Alpha becomes negative because this is a manager that was
investing in stocks that historically outperform a positive tilt towards small stocks, a
positive tilt toward value stocks. Given you are generating an Alpha of zero in the capital
asset pricing model, that's not very impressive because you are investing in stocks that
typically have positive CAPM Alphas. The small stocks and the value stocks, we need
to raise your benchmark up when we go to the 3-factor model. Once we do that, your
Alpha becomes negative here, performance is less impressive given the investment
style.

167
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Fund 31, we can see here by the negative coefficient and the small minus big factor.
The negative coefficient on the value minus growth factor means Fund 31 is positively
correlated with the performance of large stocks as well as growth stocks. We know that
both large and growth over this period underperformed. This CAPM Alpha of 0.1,
actually that's not that bad, so we're going to lower your benchmark given you are
investing in these types of stocks that underperformed. The Alpha goes from 0.1 to one.
Performance is more impressive when we go to the 3-factor model, and we take into
account your investment style.

168
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Then finally, Fund 50. This is a fund manager who's doing very poorly in the CAPM
model in terms of underperformance by 2% per year on average over this 20-year
period. Big tilt towards small stocks and value stocks. Should be doing pretty well on
average in a CAPM model. Is generating a negative Alpha even though you're
swimming downstream here. You're investing in assets through this period tended to do
well; small stocks and value stock so should be easy going downstream. But you're
doing poorly in the CAPM world, which means given your investment till you're going to
be doing even worse in a 3-factor model because we're going to raise your benchmark
up. We're going to raise your hurdle rate up. Given you're investing in small stocks and
value stocks, which historically have done well, your lackluster performance is viewed
even worse given your investment style.

169
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Question 2, which of the 50 funds would you most like to invest in going forward?

And then three, which of the funds would you least like to invest in going forward and
why?

170
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Well, before I answer this, I thought it'd be useful just a few funds, it caught my eye. The
key thing is based on past performance. Some notable funds here. On the positive side
in black funds 1, 7, and 47. If you look at, particularly, for fund 1, it has a small allocation
to equity, so CAPM Beta is less than 0.5. Despite that has a very high Sharpe ratio, a
pretty high raw return of eight percent per year. CAPM Alpha of two percent beating this
benchmark. It has low risk, low sensitivity to the market, but is performing very well. It
has this high CAPM Alpha. When you go to the three-factor model, the Alpha stays
exactly the same. That would be consistent with fund 1 manager or the manager of fund
1 is a manager that just knows when to get in and out of the market, doesn't have a
particularly investment tilt in terms of small, big value growth. Just as good at market
timing. When to get in the S&P 500, when to get out of it for fund 1. For fund 47, for
example, they're taking on more risk, as a result, generating a higher return. This is the
example we went through earlier in this lesson comparing 1 versus 47. If you like
investing in large cap stocks, that seems to be what fund 47 is, given this negative
coefficient on the small minus big Beta.

171
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Focusing on some of these duds based on the past performance, 23, 42, 50. Fund 23,
we talked about earlier, has this above average tilt toward equities. But despite that,
gives us second worst raw return leading to this gigantic negative Alpha and the CAPM
model. For fund 50, it has this big tilt. It's investing in small stocks or an average, or it's
consistent with investing in small stocks, given this positive coefficient, the small minus
big factor, positive coefficient on the value minus growth factor. Consistent investing in
small value stocks. Given the investment in small and value, it should be doing pretty
well, but it's not. It's CAPM Alphas -2 percent per year on average, likewise, to return.
Once we account for swimming downstream, investing in assets and over this 20-year
period did pretty well. It's three-factor Alpha accounting for this favorable investment
style, even worse down to -3 percent. But the key issue, how well does the past predict
the future? Is past prologue? That really goes into, do you want to simply rely on your
future investment choice being tied to past returns?

172
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Key question. Which of these funds would you invest in going forward, which would you
avoid given the historical data? The answer, any of them, none of them.

Say what? What's going on? We do all this work, and the answer is, any of them are
fine to invest in going forward or none of them? Any of them are fine to avoid, none of
them?

173
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What's going on here? Well, it turns out each of these funds, the pattern in returns that
was put into the spreadsheet, each fund, each year, I drew a random number 0-1 from a
uniform distribution. That determined for that fund, for that year, what fraction was
invested in the US stock market, what fraction is invested in treasury bills. If there are
number drawn was 0.7 for that fund, for that year, 70 percent in the broad US stock
market, 30 percent in treasury bills, and then another random number drawn for the
next year for that fund, for the next year, for the next year for all 20 years then random
numbers drawn for the 20 years for the next fund, so on and so forth. I guess, 50 funds
times 20 years, 1,000 random numbers. Then with all these allocations between stocks
and the risk-free rate, estimate CAPM regression, estimate three-factor model, calculate
Sharpe ratio just like you would with any return series.

174
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Given these are random numbers, uniform distribution drawn 0-1, the expected value of
that distribution 0.5. The expected investment is 50 percent in treasury bills, 50 percent
in stocks. Pretty close with this average market Beta of 0.52 percent for the sample.
Remember, we're drawing from random. It doesn't have to be exactly 0.5, but it's pretty
closer to that. Now, a 50-50 US stock market Treasury Bill portfolio would give an
average return of 7.3 percent per year over the period, pretty close to our average of 7.5
percent.

175
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

What's the point thinking about the analysis of Past Fund Performance? All of these
numbers were random. They did not reflect any active decision-making on market level
research or firm level research. I told stories like, hey, here's a value tilled for this one, a
growth tilt for this one. This one's good at timing the market, just looking at the past,
interpreting the regression results. But it turned out the data underlying the regression
results was just driven by a collection of random numbers. But even though these
portfolio allocations were driven by random numbers, the performance that we observed
looks pretty similar to if we look at the performance of actual mutual fund managers.
The point is if you're trying to make investment decisions based on past performance,
particularly you're trying to make investment decisions for actively managed mutual
funds and you're basing that on past performance. You need to be very careful because
I gave an example here where some funds have good past performance, some have
bad, but it's clearly not predictive of future performance because there's all driven by
random number generator. The fund that did very well like Fund 1 could do very poorly
in the future if it just happens to draw bad numbers. Fund 50, which I believe did very
poorly in the past with the random numbers it happened to get could do very well in the
future. In fact, our expectation is all these funds have the same expected return in the
future because we don't know which random draws they'll get. Whether they get a high
equity allocation in the year stocks go up, or whether they get a high equity allocation in
a year like 2008, when stocks fall a lot, it's just driven by chance.

176
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, the ultimate point is it's very hard to differentiate luck from skill, simply looking at
past performance.

Let me give you another example of luck or skill.

177
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Are there any American football fans, even if you're not, you're probably aware of the
Super Bowl, which is the championship game in which the winner from the American
Football Conference, AFC plays a winner from the National Football Conference, NFC,
they meet up for the ultimate title in the Super Bowl.

Examples of some AFC and NFC teams. 2015 Super Bowl winner, New England
Patriots as an AFC team, Pittsburgh Steelers, Denver Broncos, also AFC teams.

178
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

For the NFC, I have to go to my home State, Green Bay Packers here, the famous NFC
team. Dallas Cowboys, Chicago Bears, but also be NFC teams.

There's a lot of bets during the Super Bowl on the game, but there's also crazy bets
during the Super Bowl that have nothing to do with the game. People can bet for other,
whether the NFC or the AFC is going to win the coin toss at the start of the game. If the

179
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
Patriots are playing the Packers, they did play once a Super Bowl, Packers won that
game. Though the Packers win the coin toss. Do the Patriots win the coin toss as NFC
team, or does the AFC team win? Or you can just bet, is it heads or tails? Or does the
team that wins a coin toss won the Super Bowl? All these are bets. Typical wager would
be, if the gambler is correct, they get $100. If they're wrong, they have to pay $105. The
Casino on average is winning $2.50 on this bet here. The gambler is losing $2.50 on
average. But it's maybe just fun to put this together. It's driven by chance.

The historically profitable strategy that has been very lucrative in the past is actually
generated from bedding on this coin flip.

180
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

In particular, the historically profitable strategy is, bet the NFC team in the Super Bowl
will win the coin toss. If it's a Green Bay Packers or Dallas Cowboys or Chicago Bears,
whoever the NFC team is, bet that they will win the coin toss and the Super Bowl.

The NFC is on a roll of the first 48 Super Bowls. The NFC has 32 times or two-thirds of
the coin flips, including a streak of 14 in a row, 1998 to 2011. For example, the
likelihood of winning the coin toss 14 years in a row like the NFC did for 98 to 2011, is

181
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
one in over 16,000. Another way to think about this, if you placed an original $100 bet in
1998 Super Bowl that the NFC would win, and you kept on doing double or nothing with
that bet. At the end of 2011, that would have gone to a final balance of $1.6 million. This
is a pretty lucrative investment strategy.

Remember all this is driven by chance. Other coin flip bets aren't as lucrative. The first
48 Super Bowls, heads 50% of the time, tails 50% of the time, literally 24, 24 for heads
and tails. Of the first 48 Super Bowls, the team that won the coin flip won the game 50%
of the time, the team that won the coin flip lost the game 50% of the time. All this is
driven by chance, it just happens this one bet based on the NFC has had this nice run
but obviously, that's not driven by scale, it's just driven by luck. If we have a group of
100 people, some of those will flip a coin and get five heads in a row. Not too many, but
a few, just driven by the luck and the randomness of the process here.

182
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Bottom line, luck versus skill, pretty hard to differentiate a lucky money manager from a
skilled money manager. In particular, you just don't want to look at past returns and say,
oh, that means, good money manager, let me put my money with them because we saw
that very positive Alphas on the order of over 2% per year, over a 20-year period can
just be generated by random numbers by chance. Assignment 4 shows us with these 50
balanced funds that adopted random investment strategies in terms of allocation of
equities versus treasury bills, some of those fund managers outperform for a long time.

183
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

This doesn't mean you shouldn't invest with active fund strategies; I'm not saying that at
all. What I am saying is if you want to invest with an active fund and the manager has
done well in the past and you're investing based on that, hey, good past performance, I
bet it's going to continue, do some more research, understand the source of the mutual
fund of manager's outperformance, what's the investment idea, what's being exploited,
and does that make sense to you a priori because remember, in a fully efficient market,
some active fund managers will do well, some will do poorly just like in the assignment
with the 50 balanced funds. If you want to invest in a fund that will do well going
forward, you want to be sure you understand the investment strategy, what behavioral
biases is it exploiting, why does it make sense a priori, and why did it generate profits in
the past and likely continue to generate profits in the future? Another factor when we're
thinking about active management and its successes, we remember all the Warren
Buffett's other worlds and think wow, investing is easy, look how well Warren Buffett has
done, we forget all the managers who have done poorly and gone out of business. It's
like a selection bias, like value versus growth stocks in terms of their returns. At some
point, Microsoft, Google, when they started out, they were definite growth stocks.
Microsoft and Google, Apple have done extraordinarily well. It seems like the returns of
those companies on average would be higher than the cement company, than the value
stock. But while we remember Microsoft, Google, and Apple, we forget the dozens of
pets.com of the world that failed. A selection bias or a selection bias in our memory is
maybe playing into this as well.

184
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Key lessons learned from our discussion here of Assignment 4 is understanding the
differences between the capital asset pricing model and the 3-factor model, what the
differences an Alpha represent, how the coefficients on the various factors on the 3-
factor model give us clues about at least what the potential investment style is of the
funds, or at least in this case, all the equity allocations were driven by random, the 3-
factor model is just telling us, hey, how was the performance of our fund related to the
performance of small stocks or value stocks? Is what's revealed by the 3-factor model.
Then ultimately, the takeaway was, be careful, just simply extrapolating good past
performance of a fund manager into the future, it can be difficult to differentiate luck
from skill. If you're going to invest with an active management strategy, be comfortable
that you know what the strategy is and ahead of time, why you think it's likely to be
profitable going forward.

185
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
OPTIONAL: Lesson 3-4 Use Domestic or Global Factors?

OPTIONAL: Use Domestic or Global Factors?

All right, so this lesson is going to take on an international flair. And in particular, we're
going to talk about if you're trying to use a model to explain returns, or to kind of
forecast returns come up with discount rates in the future. Should you just focus on
domestic factors or should you take into account global factors. And particularly relevant
if you're from a small country, should you just be focused on kind of your own country's
equivalent of the CAMP or three-factor model? Or should you be included in kind of
market wide returns? Kind of the market risk premium, should be kind of the global
market as opposed to your own kind of individual country. And then, kind of a follow-on
discussion, which will highlight some of the issues. They'll talk more about the second
course on investments that I'm going to produce regarding the home and local bias in
investing.

186
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, one of the great things about this course is the international representation here.
So, you can see this kind of outbreak of red across the globe here. So, these are
countries with at least 100 students registered for this course as of May 28th 2015. So,
the one thing we have, I think representation in each continent except for Antarctica,
investments for whatever reason, isn't big with the penguins. But other than that, pretty
good representation across the globe.

187
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So that raises kind of the natural question like, hey, what should we do in non-US
contexts, in an international context. Certainly, I'll admit this course definitely has a US
centric focus. For one thing I'm from the US. US is the largest kind of stock market. So,
because of that, there's a lot of great data sets available to analyze returns and
particularly relevant, particularly free access to data on returns to do the analysis. Is
there such a crucial part of our course? There's kind of a reason why there's kind of an
analysis of kind of US stocks because hey, it's a big market and that's where he kind of
had the free access to the data to do all these analyses, okay?

188
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

But the natural question is what if I'm from Brazil, okay? Well, in my case if I'm from
Brazil, I probably much more soccer fan than I am currently and I won't be calling it
soccer, I'd be calling it football and I'd be spending kind of a lot more time at the beach.
Not too many beaches in Champaign, Urbana-Champaign, right in the middle of the US.
But how about if I'm from Brazil and I'm concerned about the investments course kind of
the natural question to ask. It doesn't have to be Brazil? It could be kind of Denmark,
Philippines, New Zealand, wherever. What's the equivalent of the capital asset pricing
model or the 3-factor model if you are looking at to evaluate the performance of stocks
or securities in your country? In this example, in Brazil, should I use Brazilian factors?
So should the market risk premium be with respect to the Brazil stock market, okay? Or
instead, should it be with respect to some kind of worldwide market? Should I be looking
at instead of relating the return on individual securities in Brazil, to what's happening in
the Brazilian market? Should be relating to them, what's happening to the global
market? Or should I kind of do a combination of the two? Where instead of having a 3-
factor model, it's actually a six factor model where three of the factors are defined based
on the performance of Brazilian stocks. And the other three are based on the
performance of foreign stocks, stocks outside Brazil? Which of these models should I
use in this circumstance?

189
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Should I be focused just on the domestic or should I be focused on more global factors,
okay? That's the key question.

The good news is there's been some kind of high quality research done on this topic.
And John Griffin had a 2002 article published in the review of financial studies talking
about this particular topic. So kind of empirical test, looking at the effectiveness of
domestic versus world versus international factors in explaining returns and then also

190
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
forecasting returns, okay? So which of these models works the best explaining the
performance of Brazilian stocks just using Brazilian factors? So define small minus big,
value minus growth only using Brazilian stocks, using world factors, okay? So, we
define the market risk premium, size factor, value factor, using all the stocks in the
world. Coming up with kind of worldwide factors or some combination of the both. Kind
of what we can call the international factors were used both domestic to the country
factors. as well as factors generated by all the stocks located outside your particular
country and the example I had outside of Brazil, okay?

So, in Griffin's analysis, he looks at the performance of stocks in four countries, the US,
Japan, United Kingdom and Canada. Regresses the excess returns of these individual
stocks on these three sets of factors, domestic factors, worldwide factors and then a
combination. So, we kind of expand the model to include both domestic and foreign
factors. So, in the case of Brazil, the domestic factors would just be based on the
performance of Brazilian stocks when calculating small minus big and value minus
growth. Those factors are just calculate using Brazilian stocks. The foreign factors
would be calculated using the performance of all the stocks outside of Brazil. So the
world wide set of factors and these foreign factors here, will be pretty similar for most
countries except for, maybe the US. Which is such a large part of the market, whether
the US is included in the market or not, it's kind of a big share. So, it may make some
difference in the calculation of those factors here.

191
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, how do we evaluate the performance of these models? Kind of going back, we have
these three models are gonna look at domestic factors, worldwide factors, set of
domestic and foreign factors. So, three types of aggressions were running. How do we
evaluate the performance of these models? So, there's a couple ways to do it. One way
is estimate the model across all the given stocks. So, once you do that, you have an
alpha for each stock, you have an alpha for Microsoft, for Apple, for Google, for
Caterpillar, for United Airlines. So then, take the absolute value of that alpha, whether
it's positive or negative, you just want to get a sense of the magnitude. And if the model
results in a lot of big magnitude alphas either positive or negative, it kind of gives a
sense like, hey, it's maybe not doing too well. It's not explaining a lot because there's a
lot of alpha still left. Okay, that's one way to evaluate the model. Another way is kind of
just simply look at the adjusted R-squared. So, we've dealt with R-squared throughout
the course, adjusted R-squared is very, very similar to R-squared. It just makes a slight
adjustment to the R-squared formula, accounting for how many right hand side
variables are in the model.

192
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, let's look at John Griffin's regression results here. So, what are we doing here? So,
if you look at this table, let me highlight a few things. Okay, so first we're looking stock
returns January 1981 through December 1995. So, what do we do? We're looking at for
every stock, we'll estimate a regression over the five-year period. So, starting in January
of 1981 for each stock estimate a regression over the next 60 months of data. Next five
years, putting in domestic three factor model, a world 3-factor models, a second
regression and then the International 6-factor model which includes both domestic and
foreign factors. So, each stock run this regression using 60 months of data. We're
starting January 1981 go for the next five years, estimate regression, then do this again
in February of 1981, estimate this regression. Then March of 1981 all the way through.
So, we're basically estimating a ton of regressions across these three different models
for each stock, okay. And across all these different time periods.

193
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, what are these kinds of numbers we have here, let's look first here where we're
looking at the results for US stocks and the number here are just simply averages
across all the regressions. So, it's averages across all the time periods as well as all the
stocks. And we're looking at here on the left-hand side in this red highlighted box is the
absolute value of alpha in percent per year and then evaluated alpha again in percent
per year. So, evaluated alpha is weighting the stocks by the size. So, Apple gets a
larger weight than Yahoo when you kind of evaluate. And here we're not looking just at
the alpha, we're looking at the absolute value the alpha to show the magnitude how
different from zero is it? And what do you see when you're looking at the US whether
you look at just a simple absolute value of the alpha or the evaluated version. You see
the smallest alpha magnitudes are coming from the simple domestic 3-factor model.
This is the Pharma French 3-factor model that we've been using throughout the course.
Okay, if instead of using the US-based factors, you use these world factors. So instead
of US stock, my stock market return minus the risk-free rate. You're using world stock
market return minus the risk-free rate, you use value minus growth, small minus big
based on the whole set of world stocks as opposed to US stocks. This World model you
see the absolute value, the alphas get bigger. Same thing when you do an international
6-factor model where you have the US domestic factors, okay. And then you have the
factors calculated from stocks outside the US, use this international 6-factor model. You
see again higher alphas than when you simply use this domestic 3-factor model. Okay,
so it seems like the simple domestic model that we've kind of been using performs the
best. I'm not going to go through each of these countries. Why don't we kind of go to our
neighbor Canada here to the north. Again, we see the same pattern that the higher
alpha magnitudes occur in the world 3-factor model or the international 6-factor model.

194
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
The domestic 3-factor model. So in the case of Canada, we're defining market excess
return small minus big return, value minus growth return all are defined with Canadian
stocks. That domestic factor model actually does the best here. Okay, so another way
to measure this as opposed to looking at the kind of magnitude of the alpha.

How much is the model kind of missing in terms of explaining returns is just simply to
look at the kind of R-squared for the US. You can see going from a domestic model
which we've kind of used throughout the course that can French, Eugene Fama and
Ken French 3-factor model that has a much higher R2 than if instead we use a world
model. And there's just a negligible increase in R2, if we go from the domestic model to
the international model here. Again, if we want to go to our kind of neighbors to the
north, you see the exact same kind of pattern. The domestic model does much better
job in explaining patterns and returns in Canadian stocks than the world model does.
So, better if you're in Canada, you might say, hey, Canada is kind of relatively small
stock market compared to the whole world. But when explaining the returns of Canadian
stocks better to use the domestic model using kind of the Canadian data than it is to use
a worldwide 3-factor model. Again, much higher R2 the domestic versus a worldwide
model. And again, a negligible increase in the R2 as we go from the domestic 3-factor
model to the international 6-factor model here. So, kind of, this is looking in the past,
how much of the returns are kind of explained by these various models. Griffin also
does an analysis where he sees how the models perform in predicting returns out of
sample. So, in other words estimate a model using past data and stop at a certain point,
say 1995. Then using that estimate obtained through 1995. See how well it does predict
stock returns out of sample after 1995. And when you look at these out of sample

195
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
forecast Griffin finds that the domestic models perform better than worldwide or
international models in terms of they have lower forecast error out of sample. So also,
kind of another positive, at least using the domestic model. They're better at evaluating
kind of the past returns and they also seem to be better at forecasting the returns than
using a worldwide model.

Okay, so when we have this kind of difference, this kind of strong result, that kind of the
best performance comes from the kind of the asset pricing models that use domestic
factors as opposed to worldwide factors. It's kind of natural to think well what's maybe
driving this result. And I think at least part of its driven by this well-known local bias in
investment. Okay, so what do I mean by local bias and investments?

196
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Well, Ken French and Jim Poterba, they documented in the early 1990s that there's a
large home bias. People have a tendency to invest a lot in their home country stocks,
even if their home country, maybe a small part of the global portfolio. So, you may think
in terms of diversification if I'm from Denmark, I don't want to have 50% of my portfolio
and Danish stocks. Or if I'm from Finland, I don't have 50%, my portfolio and Finnish
stocks. But you see people just tend to invest in assets close to home, so home bias.

197
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

I've done some research with the one Ivkovic, published in the Journal of Finance in
2005, documenting that in the US. Investors have a very large local bias, that they have
a tilt in their portfolio to the stocks or firms located within a 250-mile radius of where the
investor is from. So, kind of large local bias, not just in terms of what countries securities
are you investing in, but you're more likely to invest in stocks in your hometown. And
those have kind of a much larger weight in your portfolio, than if you're just investing in
the market as a whole. And then kind of related to this are consistent with this local bias
and investment decisions. Pirinsky and Wang find a strong co-movement of returns of
stocks headquartered in the same region. So, in other words, even if the firms are in
different industries, the stocks of those firms tend to move somewhat together if they're
located in the same area. So, stocks in Chicago tend to, kind of move together, even
stocks of firms are in different industries, that could be consistent with a local bias. Like
there's a local people in Chicago are holding those stocks, so they maybe buy those
stocks or sell those stocks at the same time. Leading to this movement in the
performance of Chicago stock. So, local bias of individual investors, leads to this strong
co-movement of the returns of stocks headquartered in the same area, even after you
account for kind of the industry composition of those firms.

198
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, fit your interest in this kind of local investment pattern here. So, stay tuned, second
course on investments. We'll talk a lot about local bias and investment decisions for
both individuals and institutions. And whether there is a return to locality. When
individuals and investors are disproportionately investing in these local firms, are they
getting any excess return from doing so? So, is it the decision to invest in local stocks?
Is it just a familiarity-based story, or does it reflect maybe access to good information
that leads to output performance resulting from these local investments? So, we'll look
at the performance of individual investors, and institutional investors like mutual funds
and state pension plans. And how they're affected by the kind of geography, how
geography affects their investment, decisions as well as the performance of their
portfolios.

199
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, what have we learned in this lesson? Country-specific 3-factor models are more
useful at explaining past stock returns and projecting ahead, predicting out of sample
than using a global model. So, the evidence is based on John Griffin's study, is that if
you're from small country. You actually get better explanatory power explaining stock
returns in your country by using kind of a domestic version of the CAPM 3-factor model
is opposed to some worldwide version of that.

200
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

Adding foreign factors in addition to the domestic factor, adds very little explanatory
power to the model, okay.

So those if you're doing a cost of capital calculation, you're doing performance

evaluations based on a 3-factor model set up. Better to do this on within country basis,
than doing it with factors computed on a global basis. And then we also talked about
maybe one of the results why these kind of within country domestic factors perform so

201
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
well in predicting returns. As it's been well documented by myself and others, that
there's a strong home or local bias in the investment decisions of investors.

202
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
OPTIONAL: Lesson 3-5 Return-Risk Model Used by Chief Financial
Officers (CFOs)

OPTIONAL: Return-Risk Model Used by Chief Financial Officers (CFOs)

All right, what are the objectives for this lesson? Well, very simple, to get an
understanding what do people use in the world, real world? In particular, for Chief
Financial Officers or CFOs, what return risk, what asset pricing model do they use when
coming up with discount rates to use to evaluate projects or to value the firm?

203
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, when we think about kind of doing this, well usually when economists or finance
professors do a research study, they're analyzing accounting data or financial data
estimating regressions. John Graham and Campbell Harvey at Duke had a very
different take, why don't we just ask the people who are making the decision why they
make these decisions or how they make these decisions. So, in 2001, they published an
article in the Journal Financial Economics that had the results of a survey conducted,
based on the responses of 392 Chief Financial Officers. There's a 9% survey response
rate, but pretty impressive given all the questions they asked the CFOs about what out
of 10 took the time to reply. Ask the CFOs a bunch of questions regarding how you
calculate the cost of capital and what are kind of motivations for capital structure
choices. So, we're going to focus on kind of the cost of capital part and just see in the
real world what the Chief Financial Officer, CFOs say they do. How do they calculate
discount rates? So, let's look at the results here and again, this is a survey of 392 Chief
Financial Officers. We're reporting here the percent of CFOs who always or almost
always use a given technique to evaluate a project. And it's kind of I guess reassuring
from our perspective that about three quarters are using the net present value here or
kind of right below it here over half say they're doing a hurdle rate calculation for the
project. So that seems kind of a terminology like they're coming up with some kind of
benchmark return here. Now of course when you're evaluating a project, you can use
more than one evaluation technique. So, these don't have to add up to 100% here. I
guess the one thing that may be surprising is why is it this net present value calculation
100% as opposed to 75%. But it's good that the net present value and then kind of
closely followed by that the hurdle rate are right at the top of the list of evaluation
techniques used by chief financial officers.

204
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, kind of to take this directly to the investment course, let's kind of cut to the chase
what fraction of CFOs use the capital asset pricing model, the CAPM? If this turns out to
be like 5%, it's going to be kind of very disheartening after we spent all this time with the
course and develop these kinds of asset pricing models starting with the CAPM at least
as it kind of regards its applicability to the corporate finance world. So, what do we see
as results here? So, the CFOs were asked how often to use the capital asset pricing
model to estimate the cost of equity capital? The CFOs could give a response from 0 to
4 here. 0 means never, 4 means always. You can see that's about 53%. 3 means
almost always, that's another 20%. So, you have almost three quarters of the CFOs
saying that they always are almost always use the capital asset pricing model. So
regardless of what you think of the CAPM, its performance using historical data, the
bottom line is at least in the corporate world, CFOs are using this a lot, okay?

205
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

How does this compare to kind of other techniques to come up with kind of an equity
cost of capital or a discount rate to use to value projects? CAPM comes up at the list
when asking CFOs about their cost of equity capital method. CAPM is kind of by far the
leader here, about three quarters always or almost always use the capital asset pricing
model to come up with discount rates. It's interesting about two out of five or 40% use
an average historical return to come up with required returns for projects going forward.
Now I think that could potentially lead to some bad investment decisions. Let's say
you're CFO at Apple today or at maybe Google or Microsoft. If you look at your historical
return and use that as a benchmark for evaluating projects going forward, your historical
return has been very high so you may be setting too high of a threshold for kind of
future projects. You may be passing on a lot of good profitable projects that your
investors would like to take because you're using a benchmark that's too high. I kind of
go more with the benchmark provided from an asset pricing model like the CAPM, then
using the average historical returns. Multibeta CAPM, three factor model, only one out
of three CFOs are using that to come up with a cost equity capital or a discount rate. So
the CAPM seems to be kind of the main game in town when it comes to risk return
models for CFOs, okay?

206
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner

So, it's interesting Graham and Harvey broke down this result of about three quarters
always or almost always using the CAPM or an average response. Remember this is on
a scale of 0 to 2 where zero means I never use it, 4 means I always use it, 3 means I
almost always use it. 2 would be kind of the average response. You can 0, 1, 2, 3, 4 are
kind of the numbers associated with the five categories from never to always. So, you
can kind of view 2 as a middle response so a lot are using the CAPM. About three
quarters always or almost always use it. A mean response about three here. So,
Graham and Harvey do a breakdown and look at how do CFOs, how does their use of
various techniques to come up with a cost of equity capital or discount rate? How does
the use of those techniques vary by firm size and whether the management has a large
ownership stake in the firm? So, by firm size, let's write this in here. Firm size, this is 1
billion in sales, is the cut off for size.
So, if you have more than 1 billion in sales, you're counted as large. If you have kind of
less than that, you're counted as small, management ownership stake 5%. So, if the top
management has more than 5%, that is called a high ownership stake. If it's less than
5%, it's called a low. So, let's look at some of these results here. So, let's look first on
use of the capital asset pricing model row one, what do we see? And the stars behind
the coefficient estimate here mean it's a statistically significant different across the two
categories. So, we see the CFOs at large firms are more likely to use the capital asset
pricing model than those at small firms, okay? Also, those CFOs at firms where
management ownership is low are more likely to use the CAPM than those where
management ownership is high. If management ownership is high, I kind of just do the
project if I want to, I think is maybe what's going on there. I don't need to worry about
what these different risk return models tell me. If my management ownership is high, I
basically can kind of do what I want and kind of use my own intuition to guide the
decision. I think it could explain this pattern here. If we look at using a 3-factor model to
evaluate kind of discount rate and hurdle rates, not many are doing this. Only a third

207
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
always or almost always use multiple risk factors beyond the CAPM but it is slightly
more prevalent at large firms than it is at small firms here. And then I like this, whatever
investors tell us they require. So, you maybe have a shareholder meeting, or you talk to
influential shareholders on the phone, get their kind of feedback about hey, required
returns, what type of project you should be undertaking. What hurdle rates you should
have. If you look at small firms, they're more likely to kind of based their kinds of
discount rates based on what investors tell us. They require a higher number here than
at large firms. Not many are responding that hey, this is something that goes into the
calculations that either, but it is a higher number at small firms and at large. Small firms
more likely to rely on what investors require us to form their discount rates than at large
firms. And then I like this result here. If management ownership stake is high, more
likely to form your discount rates based on what investors tell us they require, which
makes perfect sense because if management is high, management is the investor. So,
you're kind of telling yourself, do you want the project or not? So that kind of shows up
here in the data. I think it's kind of a natural explanation for that.

So finally, one last survey breakdown here about these risk return methods or
techniques and this is by CEO MBA, okay? My hope is if we come back in kind of 10, 15
years, we can have this breakdown by CEO iMBA. Okay, sorry I'm required to give
these plugs occasionally throughout my lecture here. All right, but one can hope, right?
All right, so we're looking at the capital asset pricing model. Does the CEO have an
MBA? If the answer to that is yes, more likely to use the capital asset pricing model to
come up with discount rates. Then if the answer is no and again the star represents
statistical significant difference between the CEO Yes category and the CEO No
category. Use average historical returns on the stock which had set a pretty high
benchmark if you're at Apple, for example, because their performance has been
outstanding over the last 15 years or so. Use historical returns on common stock to form

208
 Investments Ⅰ: Fundamentals of Performance Evaluation
Professor Scott Weisbenner
the discount rate, less likely if the CEO has an MBA than if the CEO does not have an
MBA, okay? And then finally, whatever our investors tell us they require if the CEO has
an MBA, they're less likely to rely on what investors tell us they require in terms of the
hurdle rate for projects than if the CEO does not have an MBA, okay?

So, what have we learned in lesson 3-5 here? Well, it was very useful to actually go to
the field or at least for John Graham and Campbell Harvey to go to the field for us and
actually just ask CFOs what do you use, okay? What return risk methods do you use to
come up with hurdle rates for projects to calculate the cost of equity capital, the required
return investors would have to invest in your stock. And what came out as a clear
winner, capital asset pricing model is dominant. So, one key reason to be familiar with
the capital asset pricing model used a lot to come up with discount rates in the world of
corporate finance.

209