2:34

So, how would you measure a potential connection between a mutual fund manager

and a top leadership of a publicly traded firm?

Well you could have an idea of hey, let me hire a private investigator.

Let me have them video tape all movements of the mutual fund manager to see if

there's any interaction between the mutual fund manager and

these executive of other firms, or maybe do some research.

Do these mutual fund managers and firm executives go to the same church?

Do their kids go to the same soccer practice, what have you?

There could be another way to kind of do this, and

to do it for many more mutual fund firm connections.

How about the educational background of the mutual fund manager and see if that

correlates at all with the educational background of top executives at the firm?

And the top executives that Cohen, Frazzini and

Malloy consider are CEO, CFO and Chairman of the Board, okay?

So, do mutual fund managers get

good information through their educational network, okay?

And by educational network, you can think of looking, does a mutual fund manager,

and do any of the three top executives of the firm, did they go to the same school?

Did they go to University of Illinois?

So, that could be a potential proxy like, hey, maybe there's some

chance that there's some bond between these, or we can dig deeper.

Did they go to University of Illinois?

Did they get the same degree?

MBA, PhD, undergrad, and did they do that at

kind of the same point in time so that they overlapped, okay?

So, the initial would be, did they just go to the same school period?

But then you could dig deeper and say did they go to the same school?

Were they in the same program?

And did they overlap in terms of their time at University of Illinois for

example in the MBA program?

And if the answer to that is yes,

that might be indicative of a pretty strong connection.

It's no guarantee, right?

We could have both been Illinois MBA students at the same time.

We hated each other, we never talked, okay?

But more, it is suggested, hey, there's a potential there.

There's certainly more potential for

a connection than if we went to different schools at different points in time.

So, the simple strategy, don't follow all stock picks of the mutual fund manager,

just those that are driven by an educational network, okay?

So, how important are connections, Cohen, Frazzini and

Malloy had the ability here with empirical setup to actually quantify

how important these are, at least in the mutual fund industry.

So, they're going to look at mutual fund holdings that are reported on

a quarterly basis.

Getting this data filed with the SEC, and we know these mutual fund

holdings are filed within 60 days of the end of the quarter.

So, at the beginning of each calendar quarter, so this is beginning in January,

in April, in July, in October, at the beginning of each calendar quarter,

let's look at stocks in each mutual fund portfolio,

based on the most recent SEC filing.

So, this is publicly available data, and let's categorize them by whether there's

an educational connection between the mutual fund manager and

the firms leadership or there isn't, okay?

So then we're going to measure the return of these holdings over the next quarter,

the next three months, okay?

Then we're going to repeat the procedure,

quarter after quarter once we get a new SEC filing.

So Cohen, Frazzini and Malloy implement this empirical design.

They're examining mutual fund holdings over the period 1990 to

2006 for actively managed stock funds.

An example of timing, just so we're clear here.

Say fund holdings as of the end of 2005, December 31st,

2005, are reported to the SEC by the end of February 2006.

Remember this, you have to report 60 days after the end of the quarter.

So Cohen, Frazzini, and Malloy then would examine the return to those December 31st,

2005 holdings over the period April 1st, 2006 to June 30th, 2006.

All of this is publicly available information over the period at which

they look at the holdings.

For the December 31st, 2005 holdings, they're looking at the returns of those

holdings over the period April to June 2006, okay?

So the tables I'm about to show you, they're going to report returns on

an annual basis, in percentage points for various types of portfolios of mutual fund

holdings, and key will be connected mutual fund holdings versus unconnected.

The t-statistic of the return is going to be in parentheses, and

at this point, we can all say this is unison here.

So conventionally, we're going to look for

t-statistics greater than two in magnitude to indicate significant result.

7:36

Okay, so let's go look and see what the evidence is on mutual fund connections,

and the performance of those connected holdings.

So, a lot of numbers on this panel, but some very informative results.

So first, let's just look at when we look at all mutual fund holdings.

And we're evaluating these annual returns by the size of the holding.

And kind of looking at bigger holdings by bigger mutual funds are going to get

a bigger weight in this average than holdings of smaller mutual funds.

So what do we see for the typical holding?

When we're looking at all the holdings, the average raw return

over this period, 1990 to 2006, 12.8%, okay?

Now if we look at this average mutual fund holding, once we control for

its kind of riskiness what do we see there in terms of using a 4-factor

model that's controlling for the size, value, and momentum effects?

We see its 4-factor alpha is basically zero, -0.40% on annual basis.

T-statistic here less than one, so it's not statistically different than zero.

T-statistic less than one in magnitude,

it's basically the alpha from this holding is zero.

These mutual funds aren't over-performing or under-performing, they're benchmark.

These are all measured before fees.

Another way to risk-adjust returns is this Daniel,

Grinblatt, Titman, Wermers measure.

This is comparing your holding to an appropriate benchmark return.

Just another way to risk-adjust same result here.

So we're going to focus mainly on raw returns, and

then the alpha from the 4-factor model going forward.

Let's look at mutual fund holdings where there's no educational connection.

So the mutual fund manager graduated from a different school than the CEO,

the CFO, and Chairman of the Board of the firm that they're investing in this stock.

Such unconnected holdings represent about 94% of mutual fund portfolios, on average.

The return of these unconnected holdings is 12.7%.

If we look at the alpha of these holdings, -0.5 %, but

not a statistically significant result.

So basically, the mutual fund managers, they aren't beating the market at all

on their holdings, so there isn't an educational network, okay?

And we know that's not surprising,

because we know on average mutual funds aren't outperforming their benchmark.

So it's not surprising they're not over-performing their benchmark

when we're looking at this 94% of holdings where there's no educational connection.

Now things start to get interesting.

Let's look at connected holdings.

So now we're going to look at mutual fund holdings where the mutual fund manager and

one of the top three executives simply went to the same school.

It could be in different programs, it could be at different times, but

they both went to the University of Illinois.

Maybe someday this study could be redone and they both got Coursera degrees,

they both got Scott's Investments, too.

Coursera certificate at the same time, maybe that can be a future study.

But here, we're just looking at, they went to the same,

graduated from the same school, period.

For these connected holdings,

they represent 6.3% of the mutual fund portfolio, right?

The unconnected holdings are 93.7,

add the 6.3% that are connected, and that's a 100%.

These connected holdings actually do pretty well.

On an annual basis, 15.3% return.

The long/short strategy is how are the connected holdings doing to

the unconnected holdings.

So 15.3 minus 12.7 is this difference of 2.6 percentage points,

which is statistically greater than zero.

This t-stat's greater than two.

So by simply doing this screen, did the mutual fund manager happen to

go to the same school as any of the top three executives of the firm?

Focus on those holdings, those holdings outperform unconnected holdings

by 2.6 percentage points on an annual basis.

11:54

Pretty amazing, shows that maybe there is some information networks going on,

tied to education.

And once we control for risk, okay, we look at the alpha and

the 4-factor model, it makes absolutely no difference.

These connected holdings outperform their benchmark by two percentage points,

that's their 4-factor alpha.

And if we look at the difference in performance, a long/short strategy is

invest in the connected holdings, and short the not connected holdings.

Connected holdings minus not connected holdings.

The difference in their performance, once we control for any differences and risk,

is 2.5 percentage points.

But once we didn't control for risk, it was 2.6 percentage points.

So there's no difference in underlying risk of these two portfolios,

connected versus not.

There's just the difference in the mutual fund manager just having a better sense

of being able to predict good things in firms where there's an educational bond

with the top leadership.

So that's measure CONNECTED1.

Let's go to the really deep connection and that's CONNECTED measure 4.

So now we're looking at a connection between the mutual fund manager and

one of the three top executives of the firm.

Where they went to the same school, they got the same degree.

So University of Illinois MBA or IMBA, okay?

And they overlapped in years.

Okay I have to be careful, don't say MBA, say IMBA, Marketing 101.

Come on, Scott, you need to do better here.

All right, so percent of assets here.

You can see when you find a connection like this, it's very rare.

Only 0.2% of assets have this sharp connection, this very deep connection.

So that would mean 0.2% would be if the mutual fund has 500 stocks,

which would be a pretty large holding.

If they have 500 stocks, one of those 500 would maybe have this deep connection.

Where the mutual fund manager and one of the top execs of the firm

went to the same school, same degree, and they overlapped in time.

But if you find some connection this is where you blow the trumpet, and

you see the rainbow, and you see the gold at the end of the rainbow.

Look at this return on those connected holdings, 20.5 percentage points.

So this is saying on an annual basis,

you see this publicly available data that shows a mutual fund manager investing in

this company where they have this strong connection with the firm's management.

Follow that, copy that investment.

This yields a return of 20.5% on an annual basis going forward.

This is 7.8 percentage points higher

than the return the mutual fund manager is getting on their unconnected holdings.

Then controlling for risk doesn't make any difference here.

These connected holdings, their alpha is 8% on an annual basis,

beating their benchmark.

And the connected holdings outperform the unconnected holdings by 8.5 percentage

points, once you control for risk in the 4-factor model.

Very similar to this simple difference in raw returns of 7.8%.

15:04

So let's explore the robustness of this performance of the strong connections.

Remember the raw return differential between the connected four holdings,

the very tight connections.

Same school, same degree, overlap between mutual fund manager and

one of the top executives of the firm.

The difference between those tight connections and

non-connected holdings was 7.8% on an annual basis.

So Cohen, Frazzini, and

Malloy do several robustness tests of this 7.8% return differential.

Let me just highlight a few of those.

So one was how do the results vary before and

after regulation fair disclosures?

So remember, the sample is 1990 to 2006.

So in the early 2000s, in the wake of the technology bubble Collapse.

Fair disclosure was telling firms, you can't selectively release information.

Okay, like no giving hints to favorite analysts, or anything like that.

No, selective release of private information,

thus the name, FD, fair disclosure.

Interesting this return to educational connections

is basically the same before this regulation and after the regulation.

So, it makes me wonder when you have this set of results

do you want to run to a hedge fund to tell them about it.

Or do you want to run to the SEC to tell them to tell them about it.

Okay, also, this return differential CONNECTED versus not-CONNECTED,

not surprisingly, it's stronger among MBA degrees than other educational degrees.

This differential's almost 10 percentage points.

So, if you've ever been around an MBA program

you know there's a lot of team building exercises and this actually shows through

in terms of this higher return to the educational connection here.

So, that's kind of an interesting result that kind of matches at least my intuition

having taught in various degree programs here.

And then one final robustness test I wanted to highlight here.

18:02

Okay, so when we're evaluating this return to connections, one way to think

about this is let's look at Sharpe Ratios of mutual funds and their holdings.

Should mutual funds be concentrating all of their investments

in just these stocks where they have the strong connection?

What would be their Sharpe Ratio if they did this, okay?

So, the average Sharpe Ratio,

remember the Sharpe Ratio given the excess return of the mutual fund in

excess of the risk-free rate divided by the standard deviation of the return.

So, it's kind of like a return to volatility, a reward to volatility ratio.

So, when you look at the average Sharpe Ratio across all mutual funds in

the sample, looking at their holdings, that Sharpe Ratio 0.52,

very similar to that for the whole market.

Remember this is a Sharpe Ratio on an annual basis, so

1% point increase in volatility .5% increase in access return.

19:02

That 0.52 again very similar to that for the whole market, not surprising because

they look at the mutual funds are kind of investing in the market.

Now the average Sharpe Ratio across funds,

if they only invested in their CONNECTED4 stocks, these stock holdings with

a very strong connection that Sharpe Ratio would actually be lower.

It'd be 0.41.

So the reward to volatility trade off would actually be worse.

Why?

Even though they're getting this 8% extra return, on average,

on an annual basis, they have a lot more idiosyncratic

risk because they're only likely investing in maybe one or too stocks.

So, it doesn't make sense for my Sharpe ratio to say, hey,

let's bet the whole fund on these one or two very strong connections,

while they on average do well, they don't always do well, and

thus the increased volatility leads to a worse Sharpe ratio trade off here.

But a very interesting question is to think of, forget where the mutual fund,

instead we're looking at the mutual fund holdings,

so I don't have to just invest in one or two stocks.

I could actually look across all the holdings, so when look at the Sharpe Ratio

of the aggregate portfolio of CONNECTED stocks, that's 0.83.

Okay?

So ,in that scenario, I am downloading all the mutual fund reports.

I'm looking across every single mutual fund to find this

very strong CONNECTED holding, and then investing in that going forward.

So, me, is like the copy cat.

My portfolio could maybe have 400 stocks in it, right?

I'd take the best connection from each mutual fund.

There might only be 1 of those and

if I'm just investing that one stock it's very risky.

But if I'm taking the best CONNECTED stock pick

from all these different mutual funds, I have a portfolio of 3 or 400 stocks.

That reduces the idiosyncratic risk, leading to this very high Sharpe Ratio.