0:47

So, what is this guy doing here?

Well, he's doing something called dowsing, and here I have my own divining rod.

And the idea is you have this stick, it looks somewhat like a y, you hold it out

like this, and you're looking maybe for a reservoir of water or a mineral deposit.

And you walk around, and it basically starts to shake when you get on top of

this water deposit or this mineral deposit.

That's the idea.

So, this is one I bought on ebay for $200, kind of in preparation for this.

And I tried to fine tune this to see hey,

can it sense a good question about to come up in the course?

Wow, it's working here, this is great, no, too, much let's get rid of it.

Okay, so, but it was working.

Just like I need to fine tune the mechanics of it, so

it doesn't vibrate too much here.

Pause, think and answer.

2:06

So, what should your reaction be in this case?

Well, you should actually be fairly upset, why?

You're paying a high fee for a mutual fund that's basically an index fund,

because market conditions, the small, large factor,

the value of growth factor totally explain the returns of your mutual funds.

So, your mutual fund is some type of index fund that are very close to an index fund,

but you're paying the high fee for

it, which means very poor risk adjusted performance, okay?

So that suggests, hey, maybe we should be thinking about looking at mutual funds and

the R-squared, how much of their returns are explained by market conditions and

the size factor, or the value factor?

When thinking of if we should invest in them,

in terms of an actively managed fund.

So we can think of 1- R squared

as representing how active your fund manager is.

If the R-squared is one, it means that market conditions and the size tilt and

the value tilt in the portfolio totally explain, okay, what the returns are.

Okay, if 1 minus squared is high,

it means that your manager is doing something different than the market.

Now, it might be good or bad that the manager is doing something different.

But if you're paying for an actively managed fund,

you want to know that your manager is active, I suppose.

If your actively managed fund has an R-squared very close to one, then we term

that a closet indexer, so very close to being an index fund, but pretending to be

an actively managed fund, we don't want to pay high fees for an index fund.

So, can we learn anything about future fund performance by it's recent R-squared?

And remember, earlier in the course, we talked about morning star website and

how it gives the R-squared from the CAPM model.

So, is that useful information in forecasting future fund performance, okay?

Amihud and Goyenko studied this relationship between mutual

performance in R-squared over the period 1990 to 2010.

For each mutual fund, for each month, they estimate a 4-Factor model,

so this would be accounting for market conditions,

the size tilt of the portfolio, the value tilt of the portfolio and

momentum effects, so we're looking at the alpha in the four factor model.

We estimate that amount of regression over the last 24 months.

So, each month estimate regression a four factor model,

looking at performance of that mutual fund or the past 24 months.

5:45

The authors report returns are doing a monthly analysis, but they aggregate,

they gross at the returns, so they'll be reported on an annualized basis.

So, they're doing monthly regressions and looking for monthly return predictability,

but then they report everything, so they're in an annualized basis.

And they're only examining actively managed stock funds.

Okay, let's get to the results here.

Here's a key table from their paper that I'm going to focus on.

There's a ton of numbers here, so let's boil it down to just a few.

I want to emphasize right off the bat that we're looking at net returns and

mutual funds.

So, by looking at net returns, that means looking at the return

after subtracting the annual fees that investors pay.

Remember, we're looking over this period, 1990 to 2010.

And, also, these Alpha's and R-squared's obtained

over the regressions using the past 24 months of data are from a 4-Factor model.

Okay, so let's kind of look and

focus first on this result here, in the bottom right corner.

What does that indicate?

So, looking at all the funds on an annual basis, these actively managed stock funds

under perform their benchmark by .8% on an annual basis.

The key statistic here is in parenthesis, so

the key statistic of 1.7 is statistically significant.

The 10% average,

actively managed stock fund underperforms its benchmark by 0.8% on an annual basis.

This is consistent with many other studies,

kind of the Fama-French, a result that the fees that you pay for

these actively managed funds cause them to underperform and have negative alpha.

7:38

Now, let's look at what's the future returns

when we sort mutual funds by their past alpha?

So, that would be the alpha estimated over the past 24 months.

What does that predict about mutual fund performance going forward?

Okay, so here, we're looking here.

These returns here are basically, the top row here is looking at what's

the return on an annualized basis of mutual funds?

That had the lowest alpha over the past two years.

Bottom 20%.

This number here on the fifth row here of 0.8%,

this is giving us, what's the annualized alpha for

the mutual funds, over the past two years, had the best out performance.

Okay, so what do you see here?

Well, you see, if you look at those mutual funds, who had the highest alpha?

Groups four and group five,

these are kind of the top 40% of mutual funds sorted by past alpha.

Going forward,

they don't have a statistically significant positive out performance.

So you don't have strong predictability in terms of past great alpha, past

out-performance being very good, leading to future out performance of the fund.

Okay, for group five this coefficient estimate is positive, for

group four it's negative, but they're both statistically not different from zero.

So let's focus where we do have some predictability and

that's on the negative side.

Okay, so what do you see here for the bottom two groups?

The low group is the bottom 20% ranking by past alpha, over the last two years.

Group two is 20 to 40%.

So we're looking at the worst past performers, they continue to perform well.

Excuse me, continue to perform poorly, okay?

So for the group that had the worst performance over the last two years,

they continue to under perform their benchmark by about

2 percentage points on an annual basis.

For the group that had the second worst performance,

kind of percentile ranking 20 to 40%.

They under perform their benchmark going forward by -1% on an annual basis.

So there does seem to be some predictability in past alpha

being associated with future alpha, but it's on the negative.

Bad past performers continue to be bad.

What's driving this if you look at the data,

remember we're looking at net returns, high expenses.

So, those high expenses are very persistent.

They lead to a deterioration in performance.

So, funds that had high expense ratios in the past,

continue to have them in the future.

The evidence is that high expenses isn't associated with good

performance before expenses.

So you just end up with this persistence in this negative performance,

just representing high expenses.

10:32

So now, let's sort mutual funds by their past R squared.

So how much of the returns are simply explained by common factors, okay?

So here we have five groups here.

The high group, those are mutual funds,

the 20% mutual funds with the highest R squared.

So these are basically your closet index funds,

in that they say they're actively managed.

But basically the market conditions, the size composition, the value composition

of portfolio seem to explain a lot of what's happening with the returns.

So a 100% small cap fund or 100% value fund that are just

following an index of small stocks and value stocks.

They would be in this group five here of having a high R squared.

Group one, the lowest group group,

those are the funds that have the lowest R squared.

So those are the funds where the managers are being more active.

That might be good or bad, but at least they're trading.

They're not kind of following the index fund.

And if you buy an actively managed fund, presumably you want your manager that's

making decisions to try and beat the market.

You just don't want to buy an index fund that's being labeled, actively managed.

Okay, so let's focus on the results here,

where they R squared of the mutual fund, over the past 24 years,

is basically on the high side, groups three, four, and five.

What do you see going forward?

All three of these groups under perform their benchmark by about 1 to

1.5% on an annual basis.

What does that reflect?

Well these groups here that have this high R squared,

particularly groups four and five, they're basically closet index funds.

So we know since they're an index fund their alpha before fees is basically

going to be zero, in the three-factor or four-factor model.

Once you subtract out these high fees, that's charged by the active management

fund manager, you get this negative performance.

So these are the mutual funds to definitely avoid.

You don't want to pay for active management.

If your mutual fund manager is simply sitting back and

investing in an index fund.

Okay, if they're doing index funds,

you want to pay the low fees associated with index funds.

Thus these big negative returns going forward for

funds that have high R squared.

They're closet index funds, charging you too much.

How about when we do a dual sort here, okay?

We're going to look at high R squared, and

high alpha, that's what we want to kind of look at.

Now when we look at just R squared being low,

we see those funds that have a low R squared going forward.

Their return beats its benchmark by 0.6%.

So there is a little evidence that, hey,

those active funds that are more active, maybe they are not doing so bad.

They are beating there benchmark by 0.6 percent.

This standard error though is pretty high.

So this is not a statistically significant result.

The coefficient is positive, positive alpha, but

it's not statistically significant.

But what if we investigate further and

do a breakdown where we look at both past R squared and past alpha.

So for example we look at these mutual funds here

where they all have low R squared.

So they all have a lot of active management, they're

in the bottom 20% when it comes to past 24 months, R squared from the regression.

And then let's further condition on,

we know that the managers doing a lot of trades, isn't a closet indexer,

was their past alpha over the last 24 months low or high?

And when you look at this group,

we actually find, if I hadn't thrown away the divining rod, now would be the point

in time where it's going crazy here if it was set to find good investments.

Because here we see some predictability.

So if you have mutual funds that in the past there was a low R squared from

the regression.

Indicating that the managers, doing some active management,

is trying to trade, is not just following the market.

And over the past two years,

they have this track record of doing well in terms of you know getting a high alpha.

There is some predictability for

the next month of these funds continuing to do well.

On an annual basis, they are outperforming their benchmark

by 1.7 to 3.8% on an annual basis.

So some predictability, look for the mutual funds that have a low R squared,

it means the manager's making active decisions.

Over the past two years they have a track record of beating their benchmark.

That positive alpha seems to continue at least over the next month.

15:15

So this kind of led, I presented this mutual fund results in the brick and

mortar executive MBA program that UIUC has at Chicago.

Okay, and the director of that program here is great guy, Rich Fry here.

And Rich Fry, besides being the director of the EMBA program,

he should also be my financial adviser.

Okay, so why is that the case?

Well he's presenting this lecture on mutual funds, and

in the course of this he kind of said hey, Scott, turns out that I kind of invested

in this one fund that has treated me very well over the years.

It's called the Fidelity Contrafund.

So, you know, Rich is a straight shooter so

I thought well, let's just see how this Contrafund Has really done.

So I looked at that.

And I see over this 15-year period, going from Morningstar, and

I accessed this in June of 2106,

this Fidelity Contrafund in a CAPM regression where the benchmark here,

the markets measured by the S&P 500 which is very common.

Look at this Alpha.

And this Alpha is net of fees,

3.4% outperformance on an annual basis or a 15-year period, okay.

So no wonder Rich was kind of bragging about this investment.

If anything, after this I should be hitting him up for a loan.

Also look at the R-squared.

Only 82%.

So kind of relatively low,

because Fidelity Contrafund is a big fund with a lot of holdings.

But this is kind of consistent with Amihud Goyenko research

that we have this high performance and kind of the low R-Squared.

So this would be the fund that we would predict should do well going forward.

Low R-Squared means the fund manager's somewhat active, and

this passed high Alpha.

Now as you looked over time,

then you're looking at what's the return over the last ten years?

We see this Alpha, while still positive, is being reduced.

And then we also see this R-squared is higher.

So that might also reflect the economics of the mutual fund industry,

which I'll talk about an the end of this module.

That when a fund has a great track record, and

this Fidelity Contrafund seemed to do really well in the early 2000s.

Then what's going to happen?

A bunch of money's going to flood into the fund.

If there's some diseconomies of scale and

investing, that might make it more difficult to get high returns in

the future when you have all this additional money coming in.

I may have like five good ideas, but do I have ten good ideas to kind of

continue to earn high returns when all this extra money's investing in me.

So you observe that, hey, when you kind of go forward,

you look at the past ten years as opposed to the 15 years.

While still good performance, it's declining, and the R-squared is actually

going up, suggesting kind of less of these active decisions in the fund.

Then when you go to like, let's look at the last three years.

Again, accessing this data in June 2016, you see that now using a benchmark here,

which is this Russell 3000 growth benchmark, the Alpha is actually zero.

The R-squared now is kind of 95%.

If you do the standard Fidelity Contra CAPM regression,

you also see this Alpha here.

Again, it's positive, 0.7%, but reduced from what it was in the past.

So I thought this kind of Fidelity Contrafund kind of showed,

kind of it's a case study of one, but it's kind of consistent with the research,

kind of low R-squared relatively.

The high Alpha kind of suggests like, hey, this may be good investment going forward.

We've kind of seen that.

But over time, you do see kind of the Alpha of this fund going down and

down and down.

And while maybe still positive or kind of zero, much less than it was before, if

a mutual fund does well, it's hard to keep it up once all this new money comes in.