Mutual Fund Litigation and Regulation: Is the Cure Worse Than the Disease?

January 28, 2004

Unedited transcript prepared from a tape recording

8:45 a.m.	Registration
9:00	Panel I
	Moderator:	Kevin A. Hassett, AEI
	Panelists:	Eric Zitzewitz, Stanford Business School
		Chad Syverson, University of Chicago
		D. Bruce Johnsen, George Mason University School of Law
10:30	Panel II
	Moderator:	Peter J. Wallison, AEI
	Panelists:	Bruce Leppla, Lieff Cabraser Heimann & Bernstein LLP
		Michael Feldberg, Allen & Overy
		Paul Stevens, Dechert LLP
Noon	Adjournment

Proceedings:
(In progress) . . . morning session entitled Mutual Fund Litigation and Regulation:  Is the Curse Worse Than the Disease?

If your intent is to learn more about missile defense, you need to go to the room next door.  I don't think any of us know anything about missile defense here.

Certainly if you have been following the mutual fund stories in the news, you have seen a lot of troubling reports by behavior by mutual funds that looks highly suspicious and has at times benefited the management of mutual funds, and the question is how much damage has really been done by these actions, and if there has been a lot of damage done, then does that mean that we need to change the rules, or are the rules that are in place strong enough and punitive enough that the miscreants will be punished and everyone else will learn from that.  These are the issues that we are gathered here today to discuss.

The first panel, which is the panel that I will be moderating, will be a panel of academics who study these issues, who tell us, you know, what they look for in terms of measurement of what the effects of these practices have been  and what they found.

In the second panel, my good friend and colleague Peter Wallison will be moderating a discussion amongst attorneys who are engaged in current litigation involving these issues, and they will help us think about, well, what are the legal issues, what are the disputes, and so on.

So we hope that this is a morning that will help you understand what is going on perhaps better than you did to begin with, and to that end we are also hoping to leave lots of time for questions for the various panelists.

Our first speaker is going to be Eric Zitzewitz.  He is an assistant professor of strategic management at the Stanford Graduate School of Business.  His research focuses on industrial organization, competitive strategy, and organizational economics, particularly in financial environments.

Prior to joining the faculty, Mr. Zitzewitz consulted with McKenzie & Company (ph) in the McKenzie Global Institute, and he has got lots of impressive academic publications already in the Journal of Industrial Economics the Journal of Law Economics and Organization, and Eric is going to begin and go for about 20 minutes and then we might even have his Power Point ready now.  And if not, then he can use sign language.

I will introduce the speakers who follow you just to give you another moment to catch your breath.

But after Eric, we have got Chad Syverson, who is an assistant professor with the Department of Economics at the University of Chicago.  Before joining the faculty there, Mr. Syverson held a Brookings dissertation fellowship -- we didn't hold that against him -- and consulted with Strong Capital Management.  He also is doing a great job of hitting the ground running and publishing in top places already with a paper entitled "Search Cost Price Differentiation in the Welfare Effects of Entry:  A Case Study of S&P 500 Index Funds," which is forthcoming in the distinguished quarterly Journal of Economics.

Then after Chad, we have got Bruce Johnson, who was able to navigate the Metro, I guess this morning.  He is a professor of Law at George Mason University, and his scholarship focus is on the law and economics of property rights, which allows him to address topics as diverse as antitrust and antitrust federalism, Native American institutions, corporate finance, and financial institutions, and business ethics, and maybe if we have a chance to corner him after the conference, he can tell us why our title insurance is so expensive when we close on our homes.

So we are going to give each of our speakers 15 to 20 minutes, and then that should leave about half an hour for questions from the audience.

So now I hand it off to Eric.  Are we ready?

MR. ZITZEWITZ :   We are good.  Okay.  Thanks a lot.

So most of you who know my name, I guess probably associate it now with two particular numbers, and so what I want to do is just take a couple minutes and give you a little bit of the sense of the background of where those numbers come from.

I am going to be drawing on three papers that are available on my Web site, if you want a little bit more of what's behind what I show you today.

There are two issues I am going to talk about.  One is called either market timing or stale price arbitrage.  Stale price arbitrage is the component of market timing I think people are most worried about.

So what is stale price arbitrage?  Well, it arises from the fact that funds have traditionally valued themselves at 4 p.m. U.S. time using the most recent transaction price for the assets they hold.  That price tends to systematically lag what is going on in the wider market for a variety of different reasons.

The one I think everybody understands pretty well is time zones.  If it's a European fund, Europe has been closed since 11:30, you know, so that's four and a half hours by the time you get to 4 p.m.  So those prices are all by definition four and a half hours old, and global markets may have moved in the meantime, and so those prices are going to be old.

A related problem that I think people are just now starting to think hard about is, you know, there's related issues that arise when assets don't trade all of the time, or when they trade but they don't trade terribly liquidly.  So even the bid-ask midpoint can sometimes lag our statistical expectation of what the next price of that is going to be, and so some people at least would argue that the bid-ask midpoint may not be a good indication of current value if it's not a good predictor of where the asset is going to trade in the long run.

So all three of these issues produce underreaction to recent market movements.  Now this doesn't create arbitrage opportunities necessarily in the assets themselves, but what happens is when you price a mutual fund at 4 p.m. U.S. time, it can create arbitrage opportunities.

To provide an example of that, I picked one particular day in which the markets were down in the morning, but there was a rather large rally in the afternoon.  In the U.S. market, which is the red line, you will see that the Nikkai future followed the U.S. market up during U.S. trading hours.  It's the one that trades on Chicago.  So by 4 p.m., it was up pretty substantially from where the Nikkai closed at 2 p.m. the prior morning.

I think on this day it is fairly clear if you are thinking about a Nikkai index fund, the value at 4 p.m. is pretty different from the value at 2 a.m., and yet the way these funds have traditionally been priced, they have been priced using closing prices from Tokyo, so they have been priced essentially at that green line, which doesn't change until Tokyo opens up the following morning.

What happens in stale price arbitrage is an arbitrageur will buy the fund for the green price, optionally can sell the Nikkai futures and lock in the difference between the green price and the yellow price.  Or, alternatively they can just take the holding period risk and sell the next day, but either way they are essentially buying the fund for less than its current value.

The issue -- the reason why that is an issue for long-term shareholders is in an open-ended mutual fund, they are not trading with a counterparty in the market.  The first thing on the other side of the trade is the long-term shareholders of the fund, and so if it trades advantageous for one party, it's likely to be disadvantageous for another party, and that's the source of this dilution.

So which asset class is this a big deal in?  Just as a fairly simple benchmark, I simulated returns to a strategy of taking a dollar and flipping in and out of a -- the average mutual fund in a particular asset class, based on whether the expected next-day returns of the funds are positive or negative.

So you will notice that excess returns to stale price arbitrage -- and this is excess returns over and above the returns you get from exposure to market.  The excess returns in international funds are quite large, but they are also fairly large in some of these other asset classes.  Small cap equities, some of these bond categories, where the bonds trade once every week or once every two weeks, and aren't terribly liquid when they are trading.  There is a lot of stale prices in those asset classes, and so there are issues there.

Dilution, people taking advantage of this, has been concentrated in international equity, but my guess is that as we fix the problem with international equity through a combination of pricing and other fixes, we may start to get problems in bonds unless we simultaneously fix those problems as well.

So how much does this cost long-term shareholders?  I actually don't know the exact number for the whole industry, but what I do have is a sample from 15 percent of the industry of the daily inflows and outflows in and out of these funds.  And so what I am able to do with that is at least for those funds exactly calculate how much less money was there in the fund at the end of the year because there was a lot of extra money coming in on days when the fund was underpriced, and a lot of extra money coming out on the days when the funds are overpriced.

It is not a theoretical construct.  You can precisely calculate how much less money was in the fund as a result of using stale prices as opposed to fair value prices.  Or alternatively as a result of their having these extra flows that are correlated with how the fund is stale-priced.

So if we just look year by year in my sample, you sort of see, you know, how the problem has gotten bigger and smaller with time.

The one kind of, you know, interesting success story is gold funds used to have a big stale price arbitrage problem, but around 2001 most of them switched to holding only ADRs and using the ADR prices to price themselves, and so they were able to actually get rid of most of that problem.

In international equity, which are the other two bars, you know, you see the problem growing gradually, peaking, you know, in 2002 and the beginning of 2003, and then beginning to decline.  But interestingly enough, even post-Canary, there is still quite a bit of stale price arbitrage going down.  There are still averages in the foreign stock funds, 30 basis points in September, and still about 15 basis points in the fourth quarter of 2003.

MR. HASSETT:   Eric, just a point of information.  But I could do this, right?  And any person could do this?  It's just you look and you see, okay, it went up a lot at the end of the day, I'm buying Tokyo at a stale price, and so, you know, please give me some of those shares.  Just call the mutual fund up, transfer money out of one account into another.  Is that correct?  So we don't know who is necessarily from this chart doing that stuff.

MR. ZITZEWITZ:   Yeah.  You don't necessarily have to cut a deal with a mutual fund to do this.  Some of this, in fact -- my suspicion is quite a bit of it -- is people trade their accounts or their 401(k) plans who learn about this some way, and they're kind of doing it on a small scale.

MR. HASSETT:   I have a friend who is an MIT economics professor who bragged about doing this for years, so --

MR. ZITZEWITZ:   Yeah.  I do as well.

MR. HASSETT:   It's probably the same one.

MR. ZITZEWITZ:   So I guess another piece of evidence that this is actually something real, though, is this op-ed piece in the Wall Street Journal suggesting that everything I'm doing is wrong and theoretical and so forth, so just to provide you with a little bit of reassurance, you can take the funds that my measure suggests were diluted pretty heavily.  You can sort them by how much my measure suggests they were diluted, and then you can look at the returns and compare them with the other funds in their category, and there is a pretty strong correlation.  Interestingly enough, it's actually more than one for one.  It's more than one for one for two reasons:

One is that funds that are bad for other reasons also tend up being worse on this issue.  So funds with high expense ratios are more likely to have problems on this particular front.  Funds that are underperforming in the market for other reasons are more likely to have problems.

In addition, there are some costs other than just the straightforward dilution shareholders associate with having a lot of trades coming in and out of your fund.  The media has actually focused quite a bit on this particular costs, and they are a fairly small deal relative to the dilution from stale prices, but, you know, these costs are real, and that's part of why it's more than one for one.

Okay, so late trading.  Just a little background on late trading.  You guys are probably all familiar with this now, but there is this rule that says that fund transactions are supposed to be priced at the next NAV calculated after the transaction was received.  Most funds calculate their NAVs using information as of 4 p.m. U.S. time, so orders after 4 p.m. should get tomorrow's NAV.  So a late trader would be somebody who is using post-4 p.m. information to buy the fund when the value is different from 4 p.m.

So, you know, what are we looking for when we're looking for evidence of late trading?  You know, what I am going to put up is evidence of late trading as I find this strong correlation between market movements between 4 and 7 p.m. and inflows where the decision was supposedly made before 4 p.m.

So one -- you know, and I think the leading explanation for this correlation is those decisions actually weren't made at 4 p.m., they were actually really made at 6 or 7 p.m.  And you notice the correlation is very strong.  It's particularly strong in international equity funds where there is also stale price arbitrage going on.  So presumably these two strategies may be practiced in tandem, and then it stops right at 7 p.m. and there's essentially no correlation.  There's a little bit overnight, but there's essentially no correlation out the next day.

How big a deal is this in dilution terms?  Well, it's quite a bit smaller than stale price arbitrage.  If stale price arbitrage is about 100 basis points in the average international equity fund in my sample per year, late trading results in dilution of about 5 basis points in international equity and less than that in domestic equity.

So it's a smaller deal, you know, in the sense that in dollar terms it's a bigger deal in the sense that it's more clearly legal.

Since this paper came out -- you know, it was controversial, I think, at the time it was released, but since the paper came out, the SEC has done this survey and they have gotten self-reported numbers, the numbers of funds that say they found e-mails that suggest there's late trading going on is about 10 percent of the fund families.  My evidence suggests that there was evidence of late trading at about 16 percent of the fund families in my sample.

So, you know, these are relatively consistent, keeping in mind that these are self-reported numbers of illegal activity, which are often biased.

Alternative explanations that people have thrown out there, you know, one, I guess you could think -- anything that could produce a correlation is something that you might worry about as an alternative explanation.

One might be insider trading.  The Intel CFO wants to buy ahead of their earnings announcement, but they buy a tech fund to disguise the activity.  The magnitudes here are just so large that it can't be, you know, even remotely all that.

Another alternative explanation I have heard is, well, these flows may be coming in before 4 and then the funds are turning around and investing them; or, alternatively, there's front-running by people who learn about the floats, they know they're going to be invested tomorrow, so they go out and front-run and then, you know, there's also -- the flow timings may be misreported, although the fact that there's this zero correlation the next day beyond 7 p.m., you know, makes me think that that's not what's going on there, you know, and I have been pretty careful on that issue.  There's a whole section in my J. Leo paper that talks about that that I won't get into.

But, in general, issues with these alternative explanations, the correlation persists up until 7 p.m., which we know from some of the legal -- the evidence in the legal complaints is roughly the time in which people could play these late trades until, and then it stops dead cold, right?

So, you know, if you are thinking about post-closing announcements, most of those are usually done, you know, well before 7 p.m.  Funds learn about their flows, you know, either after 6 to 7 p.m., intermediaries learn about them before 4 p.m.  So if you have a front-running store, it's got to be somebody who learns about them in a very specific time of day that doesn't seem to match what we know about the institutions.

Then there is also this issue that you have to explain that the correlation is between domestic market movements and international fund flows, and it is hard to think of a fund flow push story that's going to explain that.

MR.          :   (Off microphone.)

MR. ZITZEWITZ:   Well, apparently from what I have reading in the papers, the way a lot of these things were done is the -- you cut the deal with the intermediary, and the intermediary would slip the trade in.  They are given a couple of hours to kind of collect the trades from the day, and then what they do is they slip some late trades in, and then pass it off to the fund or the transfer agent at 7 p.m.

So the fact that the correlation stops at 7 p.m. is, you know, maybe evidence about who was responsible for the late trading, whether it was the fund or the intermediary.

MR. HASSETT:   At the mutual fund they gather up all the stuff from the day and process it between 4 and 7, and then sort of -- go home at 7 or something like that, and that's their normal operating procedure, and so 7 is the latest you could sort of slip it in?

MR. ZITZEWITZ:   What it is, it's actually they collect the trades for the day and then they -- you know, they pass one collected trade off to the mutual fund, and so it's probably an indication that it's happening at the broker.  The broker is slipping these trades in at 7 p.m. right before they give the final number to the fund.

So what are some fixes?  You can kind of think about three categories.  There's fixes for stale price arbitrage that were popular before the Canary complaint that I think all have some pretty serious limitations.

Short-term trading fees, I'll show you some evidence in a second, where they have been applied, they haven't been a sufficient solution and there's some, I think, fairly strong theoretical reasons to think why they're not going to be sufficient solutions.

In-house monitoring, we have all read enough in the papers, I think, about why that has some issues.

The SEC gave funds the option to delay the exchanges of a particular shareholder that they thought was a problem.

An issue with giving people the option to delay the timing of fund transactions is it essentially gives the fund the option to turn off and turn on the arbitrage opportunity for selected people, and I think given what we have read in the papers, that's a pretty dangerous option to be giving people.

So, you know, issues with all of these, effectiveness and selectivity.

So one alternative to this is to just eliminate the stale prices directly through fair value pricing.

A second alternative would be to just change the way, you know, these trades are handled entirely.  You have to place the order on Monday, and you're going to get the price we calculate on Wednesday.

Now for some asset classes, the lags are actually so long that that wouldn't be sufficient.  High yield is one in particular.

So you probably need some combination of that in a pricing fix for that, but, you know, I think these are the sort of the two main candidates.  We keep the current timing and we fix the pricing, or we just -- we pretty significantly change the timing in the way trades are handled.

So just some evidence on short-term trading fees.  You can sort funds by what short-term trading fee, you can look at what dilution they have.  The dilution is only 50 percent lower in funds with even 2 percent short-term trading fees.

Why is that?  Well, I think in some cases people may be trading the fund but waiting out the fee.  In a lot of cases, though, these fees aren't applied universally, so funds have a very hard time contractually applying them in variable annuity plans because they have signed a contract with somebody that doesn't provide for the fees, and so if they add fees, they're breaching the contract, essentially.

They also -- the 401(k) industry really drags its feet hard on these issues.  You know, they don't want to help implement these fees, and so in a lot of cases funds just haven't implemented fees in 401(k) plans, and so that may be the source of some of this dilution still going on.

But, you know, whatever the source of it is, the evidence is that these haven't been completely effective, and the theoretical reason why they might not be effective is even if you go to an infinite short-term trading fee, even if the enforcement is perfect, unless what you call a short-term trade has an extremely long duration, there is still going to be a pretty significant stale price arbitrage opportunity.

So, you know, to pick one example, if you go to a mandatory five-day hold period, which is quite a bit stronger than something the industry's lobbying group has advocated, which is a 2 percent, five-day fee, you only reduce the theoretical returns to stale price arbitrage by about half to 24 percent of normal returns a year in an international fund.  And so, you know, this just isn't a sufficient solution to the problem.  You would have to go to a 90-day mandatory hold to get it down to 5 percent a year, which, A, isn't sufficient, and B, I think is much more than average shareholders are going to be likely to accept.

So this is all, by the way, assuming no pricing fix.  So, you know, these things may be helpful if there is a pricing fix to kind of take the pressure off how good your pricing formulas are, but they are just not a sufficient solution in and of themselves.

This is some evidence on laundering from one particular fund that around the time you see the yellow line peak, this is dilution in three different channels:  direct, super market, and variable annuity.  They instituted a monitoring program in their direct channel.  They were able to drive a lot of that dilution out, but you see what happened, dilution right at that exact time pops up in their super market channels.

So you squeeze the balloon in one place, and it gets bigger in another place.

You can also see from the variable annuity channel that because they can't kick people out because they have signed these contracts, it's just a much harder channel to manage with monitoring.  But you fix the pricing, and the dilution goes away.

Other evidence on the effectiveness of monitoring:  90 percent of fund companies say they monitor for it, and 50 percent of them cut special deals with arbitrageurs.

One other issue I want to touch on.  With fair value pricing, the SEC has required that funds fair-value, but they have left this latitude to funds of -- you know, the rule is that you have to decide when the significant event has occurred since the close of the foreign markets, and you have to adjust the prices for that significant event.  They have left the definition of the significant event to funds, and so in a lot of cases funds have defined a significant event to be -- you know, the biggest number I have heard is an 8 percent move in the foreign markets.

If you define a significant event, you know, that narrowly, it essentially never happens, so you essentially end up never fair-value pricing.

So this chart gives you sort of an idea of, you know, those are S&P moves in the big numbers, and what definition of a significant event do we need to get rid of a substantial amount of this problem.

We need to be thinking about a significant event as something like a 50-basis point move in the U.S. market if we are going to really, you know, make a dent in -- you know, get rid of, I think, an acceptable amount of the staleness in prices.

And, you know, I think that's a mindset shift that hasn't fully occurred yet in the industry.

One way I know that is you can actually use the NAV data to come up with an empirical measure of what people are doing in terms of fair value pricing, and when I do that, you know, I can come up with a measure of what percent of staleness is being removed by fair value.

What you see is right up until June of 2001, which is a couple of months after the SEC sent that letter around, there was essentially no fair value pricing going on.  Fidelity, on a really big day in '97, but it was really by an exception basis.

Then we saw an immediate reaction.  We saw a sort of backlash in 2002 and a decline in the use of fair value pricing, and then we have seen an acceleration since -- particularly since the Canary complaint.  But, you know, the latest numbers are still on average international funds are moving about 30 percent of staleness so, you know, the glass is only about 30 percent full in terms of pricing.

There is a lot of heterogeneity across the industry, though, I should mention.  There's quite a few fund families that are removing half.  There's also, you know, quite a few guys who are fair-valuing every day now and are removing all of the staleness from their prices, but 50 percent of the industry is still doing essentially nothing in terms of fair value pricing, or at least, you know, they're doing so little that I can't see it in the data.

Another pattern that I think illustrates, you know, the dangers.  One of the critiques of fair value pricing that I agree with is you don't want to have an ad hoc system that's manipulatable, right?  So you don't want to have a system where people just kind of sit down to decide what the fund is worth on a particular day.  You want to use something that's systematic.

And I think one of the reasons is that there is scope for manipulation, and one of the, you know, really curious things I found in my sample is that even funds were fair-valuing Monday through Thursday, weren't fair-valuing on Friday afternoon.

And it's particularly true among Asian funds, and so the very cynical explanation of that is, you know, they want to get out of the office, right?  The slightly less cynical explanation is if their process is let's call up the trading desk in Asia and ask them what they think these stocks are worth, they're around Monday through Thursday, they're not there on Friday, and so that may be an institutional reason for this, but regardless, you don't -- I mean these sorts of patterns are bad because insiders are may know about them, outsiders won't.  You don't want to have them cropping up, and so you want to be using a methodology that doesn't produce these sorts of patterns.

So in terms of late trading fixes, there is no pricing fix for late trading.  You know, what we really need is better enforcement.  I think everybody recognizes that, and something that may aid that is if funds disclose their daily flow data.

You know, I know funds don't want to do that in real time, but I think there is really no excuse for not doing it two months out because there's no -- nobody can front-run them with two-month-old daily flow data that is disclosing the monthly flows, anyway.  So there is really nothing you learn from the daily flows other than whether or not there is stale price arbitrage and late trading going on in the fund.

And so, you know, to the extent I'm going to advocate anything, I would advocate, let's have them make this data public.  This way anybody, Morningstar, Lipp (ph) or anybody can, you know, calculate whether or not there's this stuff still going on, can make that information public to people, and so investors can make more informed decisions.

And, you know, I guess this is kind of in the spirit of the AEI, if you provide the information to the markets, maybe you don't need to be quite as regulatory, right?  Because if you provide this information to the markets, then investors are going to react when they find out that stuff is going on in funds.  That's going to create an incentive for funds to fix the problem, even without more regulation.

And so the one last thing I want to touch is, you know, there's a debate about fixes for governance.  One of the things I find in the paper, as I think I mentioned earlier, funds with low expense ratios are more likely to be better on the arbitrage issue, and funds with more independent directors are likely to be better on the arbitrage issue.

So there's evidence at least of a correlation between having independent directors on the board and being good on these issues.

Now whether that correlation holds up once you start mandating independent directors and, you know, some funds decide to do that by, you know, recruiting the ex-Green Bay Packers or whoever else, you know, whether that correlation is still going to hold up I think is an open question.  But there is at least some evidence of an interesting correlation there.

Okay, and so to just summarize, you know, when people ask me what I propose as fixes to the solution, you know, I think we should eliminate any of the staleness through fair value pricing, and I think we should make monitoring more credible by getting third parties involved by providing more information to the marketplace.

Thanks.

MR. HASSETT:   Well, thanks a lot, Eric.  That was a wonderful presentation.  You went through a lot of material very quickly and cogently.

Chad.

MR. SYVERSON:   Okay.  The things I will be talking about today come out of a paper that I wrote with my colleague Ollie Hertotsu (ph).  It's available on my Web site if you want to take a look at it, and as Kevin mentioned, it's coming out in the quarterly Journal of Economics, I think in the May issue.

The sort of fact that the paper is built around is that, one, there's plenty of fee dispersion, even if you look at what you might think are fairly narrow asset classes, and we focus particularly on the retail S&P 500 index funds.

So this is a class of funds that by all accounts should be, from a financial standpoint, equivalent, and it turns out even if you look at the ex-post returns, there's very little variation in how these funds perform.

So the fact that you see at the bottom there, I have highlighted some of the variations measures, the retail S&P 500 index funds actually have as much or even more dispersion in the fees that they charge than do much broader classes, such as aggressive growth or international equity funds.

So I guess that's kind of a puzzle because most of the academic literature has focused on mutual funds as financial assets, and we apply the standard financial asset pricing models to figure out what they should cost, and those theories would say that there should be hardly any dispersion at all in the fees that these funds can charge their investors, but as we see, it's just as great or even greater than some of the broader sectors, despite the financial heterogeneity.

So we go out and try to seek an explanation for why this might be happening.

A couple of other things about the sector that are kind of interesting, and that also we analyzed at the same time, and I think can explain to a reasonable degree, is that through the late '90s, the average fees in the sector were actually increasing, whether you used just the raw average or the asset-weighted average.  Both were increasing through the late '90s.

And moreover, the asset share of the biggest funds and the cheapest funds, depending on -- or regardless of how you measure it, was declining at the same time.  And another thing that reflects this is that most of the entry into the retail S&P 500 sector was of very expensive funds.  So not only did a bunch of new expensive funds come in, but assets actually shifted over to them, at least in relative terms.  The entire sector had massive inflows.  The assets grew seven times over.  Now part of that is return, but a lot of it is actually brand new inflow, too.

So all this money is coming in, but disproportionately it was going into more and more expensive funds as the decade moved along.

So we have a couple explanations for what's going on here, why we might see this sort of fee dispersion.  I should mention that the fee measures we are using actually build in loads as well, but if you take out loads, you will find the same qualitative patterns.  If you are just looking at the expensive ratios that these funds are levying on their investors.

So our two explanations that we look at in detail are nonportfolio differentiation and information-gathering, or in the economics parlance, search costs.

 The economic portfolio, or nonportfolio differentiation stories are simply that mutual funds aren't just financial assets that you can plug into the capital asset pricing model.  They are really products that have come bundled with a bunch of services, and people might have different valuations for these services and different funds offer different amounts of those service attributes.

And we look at how people might value certain things that we can observe about these S&P 500 funds, and whether that explains the dispersion we see in the fees.

The second is the search cost story where it's, you know, very intuitive that it's just not costless to know everything about all the funds, even in this sector, there's 85 funds in the year 2000, there was 100 in 2001.  There's a lot of information out there that has to be gathered if someone is going to -- if someone is hell bent on finding the lowest cost fund, that's not a costless operation to go do.  And we take a look at how search costs might also explain what we see.

So what do we find?  Well, what do investors like as far as this nonportfolio differentiation goes?  First of all, people prefer that, all else being equal, they like to be in a fund family with more funds in it.  And that's intuitive enough.  Most fund families let you shift assets around between among funds in their family without having to pay any transfer fees, so if you ever think you might want to move those assets out of your S&P 500 fund into something else, you're going to want to be in a mutual fund family that has more options for that.

The second thing is that people prefer exchange-traded funds, and that was seen in the vast growth of spiders throughout the late '90s and the introduction of Barclay's I shares in the year 2000.

There is a slight preference for funds that have been around longer, and a part of that might be the Vanguard effect, because that's not only the largest fund, but it's also the old one -- oldest one in the sector.

And then conditioning on average performance in the sector, lower tax exposure, so funds that were able to have fewer taxable events, either through the timing of their trades or however the manager can manage that, given that he's still got to replicate the S&P 500 return, people also preferred to have lower tax exposure.

So things that don't matter, we found, were how long the manager has been at the fund, how many different share classes -- in other words, price plans there are to buy the fund.

So that's the nonportfolio differentiation.

On the search cost side, we asked how big do these sort of information costs have to be to explain these big differences in prices.  And just to give you an idea of the differences I'm talking about, so Vanguard is one of the lowest -- it's not the least expensive, but it's one of the cheaper funds, and they charge 18 basis points a year.

Morgan Stanley Dean Witter, if you include the load, if you amortize the load out over seven years, it's 250 basis points a year.  So obviously that's over a twentyfold difference, and in just absolute value, you know, it's almost 2 1/2 percent.  So that's a substantial dispersion of prices.

And, moreover, Morgan Stanley is actually one of the larger funds in the sector, so the most expensive fund is not minuscule.

But even with that sort of dispersion, you don't need what I -- in my opinion are particularly large search costs to explain that.  So we were able to back out what search costs.  I might get to talk a little bit more about that in detail in a second.

But the upshot is that the median search cost that we find is only 5 basis points in the year 2000.  So why is it in basis points?  Because that's how prices are levied or fees are levied in the industry, and that's what we can infer.

Now so that means all your search cost is going to be proportional to how much you are investing, which isn't terrible if you think that people with more money to invest have higher values of time, and if search costs or information-gathering costs reflect time value, you would expect it to be proportional to the amount you're investing.

We could actually put a -- if we knew about the individual amounts that people are investing and the distribution of those, we could actually put an absolute number value on there like $25, but we didn't have that information, so we have to infer everything in proportions.

And the dispersion among investors, however, is big.  You've got some guys at the 75th percentile, for example, are going to have 28 basis point search costs.  So let me just explain.

This means to look at, to find out the attributes of one more fund, you have to pay 5 basis points.  So it's $5 per every 10,000 you're investing.  Okay?  And if you think it takes a minute or two to go to a Web site and take a look at the prospectus quickly and get some information, that's not a ridiculous number.

So I'll just go to kind of the upshot.

Just to back up a second.  We're kind of explaining the dispersion we see with again the combination of this nonportfolio differentiation and the search costs, but then I also mentioned this pattern as the '90s moved along that assets were shifting to more expensive funds, and we think that a reasonable explanation for that is that the investor population changed in its composition as far as what their information costs were of learning about these funds.

So what I have got plotted here are the cumulative distribution functions for the search costs of investors, and I won't -- if your statistics is a little rusty, all that's important is movement up and to the left means search costs are getting lower.

So what we plotted here are for two groups of investors:  those who buy no-load funds, and those who buy load funds.  And we plotted the search cost distributions both in 1996 and in 2000, so kind of the beginning of the period and the end of the period.

Now the green and the blue lines on the left are the search costs for the no-load fund investors.  The blue line is what is inferred in 1996.  You can see that for almost everybody who was investing in no-load funds, search costs went down between 1996 and 2000.  And that's not really a surprise because the advent of the Internet, Morningstar's information aggregators.  You can go to indexfunds.com now.  It's a lot easier to learn about a bunch of funds all at once.  It's getting much cheaper, and that's reflected in the implied search costs.

However, if you look at what search costs are for load fund investors, they went down for the guys on -- the investors on the low fund of the distribution, so that's the people at the bottom end of the curve.  However, between 1996 and 2000, search costs actually went up for load fund investors.

And what we think is going on here is not that it's the same investors whose search costs have just gone up; it's really that these investors who are buying load funds in 2000 are different than the people who bought load funds in 1996.  And that's in accordance, I think, with the evidence on the large increase in stock market and mutual fund participation that we saw in the late '90s concurrent with the runup in asset prices, and these were people who had not invested in mutual funds before, and surprisingly wouldn't or would have higher costs of learning about the characteristics and attributes of funds.

So I think we take a look at this is evidence of this shift in who is investing as explaining why we see this movement towards the more expensive funds, and also the increasing dispersion in fees from what was already a pretty large number.

So what we take away is that this fee dispersion exists even when portfolios are essentially equivalent, and that's reflected again in the retail S&P 500 funds, where portfolios are pretty much the same.

Also you don't have to appeal to crazy psychological explanations of what's going on here, first, you have to realize that these funds come bundled with services and people are willing to pay -- some people are willing to pay for these services.

Also there are information costs, and we would expect, or it's a reasonable story that people who started investing in the late 1990s might have had higher information costs than previous investors, despite the advent of these new technologies such as the Internet and such, and that might explain the shift toward more expensive funds that we see, not only in the sector, but in other sectors as well.

And if I may just go beyond what we actually talk about in the paper and talk about policy, I think my take-away is that I think the best -- these information gathering costs are real, and if there is a scope for policy, I think it's -- one good place for it would be in lowering the costs to people of learning about these funds.  So writing prospectuses more clearly, and actually the market is providing a lot of these services already, so Morningstar has done very well for itself, as far as I know, by offering this sort of information to investors.

I talked about indexfunds.com.  That's another source.  So the market is in some ways stepping in to provide these services that lower these costs for investors, but there is still probably scope for policy as far as the way prospectuses are written, and also the relationship between brokers -- I didn't get to talk about this much here, but there is an important distinction between people who are buying load funds -- in other words, going through the services of a broker -- and the direct purchase channel.  And, of course, once you've got a broker involved, there's a principal agent problem, as we have seen in the press.  Some of those issues have cropped up lately, and I think, of course, that's another area that policy should be addressing.

MR. HASSETT:   Thanks.

One question on the fees.  I've seen the claim made that people who pay high fees, it's a kind of a self-disciplining device that will induce them to adopt the correct long-run buy-and-hold strategy, and wasn't there a Dalvar (ph) study or something that claimed that the folks who bought the load funds outperformed the folks who didn't, because they tended to not, you know, sell after the market went down, and so on?  Do you know?

MR. SYVERSON:   Well, I'm not so -- I'm not familiar with the study.  I'm not so sure about the -- that sort of load is a commitment device story, because if I wanted to just commit myself, I don't need to do -- I can give -- I can sell a bond to a third party who will, if I say -- you know, if I take money out of this mutual fund before 10 years are up, you can keep the bond.  But if I don't, I get the bond back.

Whereas if you are trying to commit yourself by using a load, the company keeps the money in the end.  And you will never get a chance to get it back.

MR.          :   [Off microphone.]

MR. SYVERSON:   You can write a contract with somebody as long as you can trust them to not renege on the agreement, but I guess I think loads are basically a fee-for-service.  Some people might think that people are paying a ridiculous amount for that service, and I'm not going to take a stand on whether that's true or not.

And, yeah, they end up de factor being a commitment device.  I don't know whether people enter into load funds specifically as a commitment, though.

MR. HASSETT:   Can you think of a cost -- so suppose that you go into a load fund, and so therefore you are in a fund with other people who have committed to keep their money there.  Then is that -- can you tell of any argument that that might be a more attractive fund?

MR. SYVERSON:   Actually, that's a good point.  Most load funds now offer different share classes, so you've got front loads, rear loads, and level loads, and you can actually see that fixing a portfolio, the front load funds have much lower asset turnover rates, and because asset turnover rates impose an externality on the other investors for -- because of their trading activity or the redemption activity, you can think that it's actually a wise strategy for who is a buy-and-hold investor, if you're going to buy a load fund, to buy a front load fund rather than a level load, because when everyone is buying a front load, there is a bigger -- the return to staying in and holding is larger.  Therefore, you're not going to have as much in and out.

Whereas, if, you know, you go into a level load, you're going to have people going in and out more.

MR. HASSETT:   And then that was the last question, and then we'll hand it off to Bruce, that so I thought through this, and thought that what I'd really like to find would be a fund that had a load that paid the present value of my annual expenses.  So rather than charging me an annual expense ratio, they just gave me a load up front, and it was exactly because of this.  You know, so I want a very low cost fund and I want to make sure that the other guys aren't spewing taxes on me.  Does such a fund exist?

MR. SYVERSON:   Not that I know.  I'm not aware of that fund, although it's not a bad idea.  Maybe a --

MR. ZITZEWITZ:   Well, so Vanguard has these tax-managed funds, where there's a purchase and redemption fee that gets paid into the fund, but it probably serves that purpose, you know.

MR. HASSETT:   Yes.  And now we'll hand it off to Bruce.

MR. JOHNSEN:   Thank you.  I want to talk today. . .

[TAPE CHANGE TO SIDE B.]

. . . heavily, heavily criticized.  It's just terrible stuff?  And why?

Well, the first point is that since brokerage commissions are paid by portfolio investors, that is they go into the tax basis of the security, that by bundling research into broker commissions, the manager can somehow secretly shift the cost of research to his investors.  Hey, we're higher than this guy, he should pay for that research out of his own pocket.

So the notion is that in some sense soft dollars allow managers to unjustly enrich themselves by getting investors to pay for the research that they should pay out of their own pocket, and that is inefficient because it induces managers to do research than they would do if left to pay for that research out of their own pocket, and more trading than they would otherwise do, and to pay excessively high brokerage commissions.

And also maybe to be unduly loyal to a broker.  Here's how the soft dollar arrangement works, right.  The broker comes to the manager and says, listen, I'd like your business, and the manager says, well, what will you do for me, and the broker says, well, listen, I'll give you $50,000 worth of research credits, soft dollars, Monopoly money, all right.  I'll give it to you up front if you promise to do say $100,000 worth of commission business with me over the coming quarter at commission rates that are normal, 6 cents per share.

So one of the problems might be that once having taken the research, the manager feels like he's just a little -- you know, he's indebted to the broker, so he might use a broker that's not doing good quality executions.

So this is the basic criticism of soft dollar brokerage. You know, it's just one of those situations where the agents just run amuck on the principle -- and in this case, the principle that mutual funds are dispersed and they have a huge coordination problem, so unless they have a regulator working on their behalf, you know, they're going to run amuck.

And, you know, on the other hand, some of these portfolios are managed accounts for pension funds and so on and so forth.  You know, I was a trustee of the Virginia Retirement System.  They know about soft dollar brokerage.  They know what's going on, all right.  They are perfectly capable of contracting over that activity, but they really don't.  They allow it to go on.  I guess the question is why.

Now I come from a long line of industrial organization economists who were weaned on first being told that when people cut prices in markets, there's got to be something sinister about that.  It must be predatory pricing, right?  And when manufacturers try and impose minimal resale prices on their retailers, well, that's terrible because, you know, who knows, it's price-fixing of some sort.  Or when a manufacturer wants to impose exclusive dealing on his retailers, that's just got to be somehow wrong and monopolistic.

Well, we all know that, you know, basically the Chicago school unraveled all that reasoning, and many of those practices we found are or can be extremely efficient.  Well, we're now caught up in the new frontier of securities markets and securities market regulations, and I'm not so convinced that the alarm that everybody is showing in a lot of these cases, or expressing, is necessarily warranted.

At least I want to put up a good defense and make them prove their case.  So what do I have to say about soft dollars here?

Well, first of all, all the criticisms of soft dollars are really criticisms of bundling research into execution or into brokerage commissions, right?

And the criticisms apply equally to full service brokerage.  If the manager goes to a full service broker, he gets research, right?  He pays out, pays out a higher than minimum cost brokerage to get that research.  He might unjustly enrich himself.  He might use too much research.  He might do too much trading.

So the question is, why does this bundling persist?  Why isn't all research sold separately in the market?  The full service brokers, they can sell their proprietary research in the market.  Some of them do, to noncustomers.

Well, my argument is in fact bundling isn't such a bad thing.  And let me just take the soft dollar case.  All right, one of the problems that managers perceive and have when they go to trade, as I said, is that their trades will have an impact on the prices.  That's one of the transactions costs of getting the trade done.

The other major cost is the cost of the brokerage commission.  So it might pay them to pay up to ensure that they get high quality brokerage, and that's exactly what is happening with soft dollar brokers.  The manager essentially receives research up front, in exchange for his promise to do future trades.  And that agreement is legally unenforceable, not only under agency law, but under securities laws, the broker cannot enforce that as a contract.  All right.  It's simply a gentlemen's agreement.

So what happens here is once having received the research up front, if the manager finds that the broker is doing a bad job of execution and his trades are leading to excessive price impact, he can terminate the broker with the account balance unpaid.  This is a perfect performance bond that you might associate with say Ben Klein (ph) and Steve Loeffler's model on how price assures quality.

So the broker has got his neck stuck out here.  Now the problem is for the manager, it's not easy to tell whether the broker is doing a good or bad job immediately, because securities markets are noisy.  So it takes time to figure out that cheating is going on, and given that brokers might be inclined to take advantage.

We have heard about front running, we have heard about just careless executions and so on and so forth.  It happens, but the soft dollar arrangement says to bond the quality of the executions.

Now some people will say, well, yeah, we've looked at the quality of execution of soft dollar brokers and it's not as good as full service brokers.  So what.  I'm not arguing that soft dollar brokers are the Morgan steakhouse of the restaurant industry.  I'm saying they may be more like the McDonalds of the restaurant industry.  You can go down the interstate and you can stop at one and you know you're not going to get poisoned.  And anyone who has been poisoned stopping at a truck stop, and I have, knows the value of McDonalds.  Trust me.  You know what you're going to get.

And also for managers, one of the things I think they're constantly doing, is they trade through the same brokers every time.  People are going to be out there watching that, all right, and they're going to take advantage of it.  So it pays them from time to time to go to somebody who they don't routinely go to.  But if they don't routinely do business with this person, where is the trust?

Well, the trust comes from this performance bond.  Well, then, you might say, okay, Johnsen, that all sounds, you know, that all makes sense.  There's no rocket science here.  But why research as a performance bond?  Why not just have the broker pay $50,000 in cash into the portfolio in exchange for the manager's promise to do $100,000 worth of commission business?

Well, the answer is that a dollar's worth of research is more valuable to portfolio investors than a dollar.  Why?  Standard agency problem.  The manager is hired to increase the wealth of the fund or net asset value.  Let's say he puts in more management effort, right.  If he puts in more management effort, the value of the fund should increase but at a decreasing rate.  He's paid a very small share of net asset value, so here's a marginal addition to net asset value.  Here's his share of that.  All right.

If he bears all the cost of -- well, let's say he owns the entire portfolio, how much management does he do?  He does this amount.  And what do we put there, of course?  A star, because stars are good.  All right.

But if he's left to his own devices and all he receives is shares, he doesn't (inaudible), right?  Standard agency problem.  The agent has too little incentive to act on behalf of the principal, to do the things that the principal wants to do.

This is true in all agency relationships.  It's inevitable.  But what can portfolio investors do?  Well, they can subsidize the input.  They can make management cheaper.  Surely they do that by agreeing to pay for executions.  And implicitly they do it allowing managers to bundle research into execution.

And what that does is essentially moves the manager in that direction.  He does more research.  Why should it be troubling when you see a practice that induces managers to do more research than they would otherwise do, and to do more informed trading than they would otherwise do.  That sounds like a good thing.

And, in fact, the world over, people subsidize their agent's use of productive inputs.  For example, why don't we all pay our own business travel and have our employers just pay us an extra salary that where that extra salary compensates for the expected cost of airline travel?  It seems like a perfectly reasonable way of doing things, right?

Well, because we'd all stay home, we wouldn't travel.  Right?  You know, traveling sucks for most of us.  So we get subsidized.  This is just, you know, routine stuff.  There's nothing unusual or scary about this.

Now there was a report, specifically one coming recently from the investment company institute, to prohibit any kind of bundling of research into the brokerage commission except for research that comes from full service brokerage houses.  All right.  That doesn't seem to make sense to me because it's really bundling that's the problem here.  One thing that's unique about soft dollars is they formally account for the flow of research in brokerage back and forth between the manager and the broker.  That proposal doesn't seem to hit the right target.

Of course, there are proposals that we have more disclosure and, you know, disclosure is one of those things that, you know, who can argue against it.  It's like apple pie or something like that.  Yes, more disclosure is always better.  But, you know, if you read the paper these days, you know, you see people talking about disclosing a whole lot of extra stuff and it's not enough that it's in the statement of additional information that now has to be in the prospectus, we're told.  It's just not clear.  At some point, even somebody as familiar with the securities regulation as William O. Douglas, if I'm not mistaken, has said listen, there's such as thing as too much disclosure.

Now you might say, this is all just theory, Johnsen, and you know, how do we know you're correct, how do we know this isn't just pie-in-the-sky stuff.  I do have research with a coauthor using admittedly imperfect data base of mutual fund or money manager returns, and essentially what we find is that brokers that pay up more actually produce superior returns for their portfolio, all right.  What we use is premium commissions per managed dollar, which is turnover times brokerage commission rates.  Managers who pay more for brokerage actually outperform those who pay less.

Now you might say, okay, well, then, why doesn't the manager just spend even more.  Well, our hypothesis is simply that this is where the art comes in.  That soft dollars and paying up are a method that managers can use to essentially build trust.  It's not easy to do.  Some figure out how to do it, and those who figure out how to do it actually perform better.

Alternatively here what we should see is that managers who -- if managers are unjustly enriching themselves, and that's all there is to it with soft dollars, then what we should see is managers who use more soft dollars should accept lower fees.  And in fact we see just the opposite, or we see no relationship between fee and soft dollar use.  That is the percentage fee or basis point fee and soft dollar use.

Okay, well, I think I'll stop there.  That's good.  MR. HASSETT:    You know, this third aspect of the industry is something that I have seen from the inside a little bit, just through friends who work for firms, and my understanding of how the soft dollar business works is that there is really a tournament going on between super star analysts, that some of the analysts in top firms in Wall Street can have, you know, salaries before bonus of like $10 million a year, and so it's kind of like you're getting the Michael Jordans of analysis and then spreading their costs around, and they don't actually have -- a lot of the research firms don't have a lot of people, but the people they have are really famous and really trusted on sometimes very narrow issues like which bank stocks to buy.

And the way I understand that it works is that, first of all, if an analyst says amazon.com with some great convincing report, then the people who read the report and decide to buy Amazon decide that they're going to give the brokerage business for their Amazon purchases to the firm that wrote the report disproportionately.

I've asked them, how is it that you guys pay for these huge salaries?  How do you make the money?  And they say, oh, well, when we convince people to buy a stock, then they tend to feel honor-bound to buy it through us.

Then the other thing is that if at the end of the year, we provided really, really good information, then for the next year their allocation to our firm for trades goes up a lot.  And so there's a tournament going on between research firms with very nonlinear return structure.

So my question is, is there a research into the tournament aspect of it, and the optimality of this very nonlinear fee structure?

MR. JOHNSEN:   I mean in a tournament the winner takes all.

MR. HASSETT:   Yeah, so if you have a person who is really making money for people, then he gets all the trade business, and that firm makes tons and tons of money.  If you have a person who says things that are kind of worthless, then the firm is making zero.

MR. JOHNSEN:   Right.

MR. HASSETT:   And so there's a clear golf-like tournament going on where the only thing that matters is the person who's really good at picking bank stocks.

MR. JOHNSEN:   Let me respond to that obliquely, I guess.  I mean my understanding is most of the analysts are in a full service house, right?  They're in-house analysts.  You know, the third-party research vendors, I suppose some of them might be purporting to provide stock tips, right, that actually tell the managers here's what you should buy, here's what you should sell.

I don't think that's really happening, and one of the problems with that is how do you know that he hasn't already given that tip to somebody else?  So there's this ongoing favoritism problem, right, and the manager has to compete to be favored by sending more and more brokerage business to this person.  Okay.

With soft dollar brokerage, what is happening is essentially the manager is getting input, Quotron (ph) machines, software, data bases, et cetera, et cetera, that he can use to put together with his own labor effort and come to his own conclusions.  And I think this explains the rise of investment management over the last, you know, however long, since the 1940s, if you like.  But I think back then when people said listen, we don't need to go to full service brokers and play this favoritism game, you know, trying to be first call on their list, all right.  We can produce our own stock tips.

James Stowers (ph) at American Century is a perfect example.  He went and made a contract with IBM to figure out how to, you know, run the data through with punch cards and stuff like that, and he used publicly available information, and squeezed just a little bit more out of it to generate his own stock tips, and of course the SEC rode him out of town on a rail, or tried to, but he came through and survived.

So I don't if that responds -- that doesn't really respond to the tournament question.

MR. HASSETT:   Is there a possibility that because you're gathering super stars, then in some sense you're getting monopsony power?  So suppose that you take what I think is probably the most famous firm, Sanford Bernstein (ph) and suppose they have just -- you know, they have people who move markets because everyone agrees they know the firms the best, and their threat is that we are not going to let you see the stuff that we show everybody else, you know, until a month later.

MR. JOHNSEN:   Uh-huh.

MR. HASSETT:   Unless you pony up and give us lots and lots of money.

MR. JOHNSEN:   (Inaudible.)

MR. HASSETT:   That's right.  And is that -- so they kind of monopolize the information generation business, perhaps?  Is that what you try to do, so you go around and buy up all the people who might be the ones who know?

MR. JOHNSEN:   No, because now with this third-party research, a manager can do an end run around that.  I mean the institution you just pointed to has its benefits and has its flaws.

First of all, I want to be on record as saying that we live in a world where, you know, you push on one side of the balloon and the other part of the balloon bulges out.  You point to a conflict over here, and you try and stop that conflict, and likely it will make things worse over here.

One of the nice things about third-party research is it provides a competitive alternative to the system that you're describing which, as I say, I don't have anything against.  It has its good side and it has its warts.

One of the recent warts, of course, is analysts, you know, essentially touting securities that they don't really believe in, okay.  So, but, you know, I guess my response is I'm in favor of competition and that would, you know, minimize the problems that you're pointing to.

MR. HASSETT:   But it sounds like, though, that in terms of this panel that the soft dollar thing is kind of just a distraction, that -- to me.  So the idea is the mutual fund companies are doing this a lot, that that's how they compensate the Sanford Bernsteins of the world for the research they get when they're deciding to pick stocks, but there's not evidence that somehow that's part of the regulatory challenge right now; that I think that Eric's work in particular highlighted.

So there is no -- is there anyone out there claiming that investors are getting screwed because of the soft dollars?

MR. JOHNSEN:   The Investment Company Institute, as far as I can tell, and the Securities and Exchange Commission is now studying the whole question of brokerage allocation.  Congress has insisted they do.  So, no, this is definitely -- as soon as they get tired of mutual fund timing, they're likely to move in this direction.

MR. ZITZEWITZ:   Well, I think there is a lot of interest in getting more disclosure in this area.  One of the things that was in the Baker bill before it was stripped out -- this is when Oxley stripped a lot of things out over the summer -- was to get more disclosure on issues like soft dollars, and even if you think it's a good thing, it's very hard to argue that it isn't something that investors should have and getting information about.

MR. JOHNSEN:   What exactly, though, are you saying they should get?

MR. ZITZEWITZ:   A commission ratio to go along with an expense ratio, for instance.

MR. JOHNSEN:   Okay.

MR. ZITZEWITZ:   And, you know, you could take that and from the turnover you could know sort of an excess commission ratio to roughly, you know, how much was paid in addition, you know, over and above 1 cent a share.  You know, so how much soft dollars should this performance be reflecting.  I mean how much bond width, so including full service brokerage.

MR. HASSETT:   I guess the problem is you don't measure the benefit, but I guess that's something that we could watch the data.  It's hard to argue against showing that data.  So the firm puts an extra couple basis points of expense on its mutual fund shareholders because it's being really generous in its share allocation at Sanford Bernstein, but it's doing that because it got research that made a lot of money for the shareholders, then that's something they could watch over time and make sure it was true.

MR. JOHNSEN:   Not to mention lower price impact.

MR. HASSETT:   But if it didn't -- I mean so if it turned out that as we saw with your stuff, that the difference in the rates of return were very much related to the identifiable screwing of the shareholders, then that would put pressure on the soft dollars.  So I agree that it sounds like this is something that it ought to be -- why wouldn't we want to just sort of keep track of how much extra money we're paying for trades.

MR. JOHNSEN:   This has nothing to do with soft dollars.  What you're trying -- yeah, so if you want to keep track of how much extra money you're paying for trades, that's fine, that's kind of a notion of transactions costs, explicit transactions costs, but then, you know, is that really telling shareholders much?

First of all, most of the people in this room really wouldn't even care to spend their time looking at that.  But if they did, the flip side of that question is well, what about price impact.  You know, what kind of price impact have the brokers had on your trade.

So it might be in shareholders' interest to have their managers pay a higher brokerage commission rate if it reduces price impact.

So I mean ultimately what shareholders care about are returns, which are admittedly noisy, all right, but anyway, you know, I'm not going to at this forum put up a big fight on disclosure.  I want to put up a big fight on not prohibiting third-party research.

MR. HASSETT:   Okay, and then I'm going to ask one more question for Chad, and then we're going to go to questions.  I can see there are a lot of hands going up.

So if we look at this whole panel, there's this kind of odd soft dollar thing going on.  There's late trading and stale trading going on, there's fee dispersion that's basically, you know -- let's face it, that if my mom said should I pay a 3 percent load, I'd say you'd be crazy if you did that.  There's just no like academic support for that, it is zero.

So what does it all mean?  I mean so we see all this stuff going on.  Does that mean that we're regulation mutual funds enough, or is it fine if somebody wants to give their money to somebody, it makes them feel good if, you know, they think they're a star, then should we worry about that.

MR. SYVERSON:   Well, the mutual fund industry's mission or what I see its mission as is being an intermediary for people who would have a personally high cost of getting into asset markets.  I mean they are a transaction cost reducer.  But because they're running transactions, there's all the scope for the principal agent problems that Eric talked about and that Bruce mentioned people are worried are going on with the soft dollars.

What does it all mean?  There's definitely scope for abuse, and we have seen that some abuse has happened.  But I don't think we need to always appeal to stories of underhandedness.  I mean there's very reasonable -- these are very complicated markets with lows of products, and maybe we shouldn't be surprised at some of these outcomes.  MR. HASSETT:   Shouldn't this work out in the end?  You're not going to make a lot of money if you're the company that everybody -- that has a reputation of bilking its customers, right?  I mean --

MR. SYVERSON:   So the key is when you bilk, you want it to be evident that you're bilking, and I think that's the disclosure argument and why you might want to put these trading fees into the prospectus or give that number separately from the management fee.

Information is an important part of the story, so you need competition plus information, and then the market can take care of things by itself to a large degree.

MR. JOHNSEN:   I have a partial answer that also raises an issue with Eric's research.  One thing that people don't seem to recognize is essentially open-end mutual fund, it's a common pool.  All right.  To the extent there are -- is an expectation of excess returns, what is going to happen is you're going to have inflows and eventually from an investor's standpoint all those excess returns are going to be dissipated.

I think it's a crime that anyone, regulators especially, give public investors the expectation that somehow they are going to latch onto a mutual fund that's going to give them excess returns and that we should construct a body of regulation that makes sure that they are the lucky beneficiaries of that.  It's just not right.

Now regarding Eric's research, my question is, you know, what are the offsetting potential benefits of fund timing.  In the context of this open access or common pool situation with mutual funds, we have complexes that -- fund complexes that want to provide a variety of different types of funds, and that of course has some benefit to investors.

One of the problems is that the advisory firm could easily favor one fund over another fund surreptitiously by allocating brokerage costs, first trades as opposed to last trades, in which the first trades have less price impact, and so on and so forth.

So when you have an agent managing different accounts for you, one of the problems is that they may, you know, shift money between accounts.  How do you bond yourself against that?  You let them shift their money for free, so if they see you or think you're engaging in some kind of funny business, they transfer their wealth from one fund to the next.  All right.  I think that's why funds have allowed zero cost exchanges within a complex.  And that's a good thing.  All right.

Now what we are talking about is fees that make that more difficult, and essentially what that does is it expands the common pool from just those who are currently invested in the complex to all investors.  You could give investors -- if the load on leaving Fidelity and going to T. Rowe Price is 5 percent, let's say it's an upfront load, Fidelity can essentially provide a full range of mutual fund styles to its investors and beat T. Rowe Price simply by offering any exchange fee of less than 5 percent.  That's going to keep their investors in their complex and keep them from departing to some competitor.

But why do we instead see zero cost redemptions?  And I think the answer has got to be that that bonds the advisory firm against any kind of accounting or funny business, as you might call it.

So it is quite possible that some of these policy recommendations for preventing mutual fund timing might actually work against investors, or at least that is an effect moving in the opposite direction.  What the, you know, net effect is, I don't know.

MR. ZITZEWITZ:   Let me just comment on that.  When I said I thought a complete ban on selling, you know, beyond 90 days was a nonstarter, that was one of the reasons why I thought it was a nonstarter.

I do not think, though, that what he is describing is the primary reason why people are buying an international fund and selling it two days later.  I don't think they are changing their view on how performance is being allocated amongst and the money market fund.

MR. HASSETT:   Well, if we just gave information, the information in your paper, that again wouldn't hurt.  Who could argue against that.  I would want to know if there was a lot of stale trading going on in the funds that I own, wouldn't you?  So at the very least we ought to be able to agree that that's something that -- unless it's incredibly costly to report or something, then let's dedicate tax dollars to hiring Eric to come to Washington.

MR. JOHNSEN:   I've known about fund timing for three years.  I have a colleague right next to my office who does it, all right.  He'll go in and actually sit and help me set up my password.

MR. HASSETT:   We talked about that.

MR. JOHNSEN:   I'd rather play golf, quite frankly, you know.  I would.  It's just too costly for me to do that, all right.  So, you know, these people are engaged in fund timing, it takes their time and attention.  We have to deduct that time and attention from their net return.  In the long run, competitive equilibrium presumably returns for all investors are going to be equal.

Now you can say that there's a structural problem here that regulators should fix, and I think, well, I'm -- I'll say that may be true, and I know Eric is not --

MR. HASSETT:   Yeah, but the problem is if you have $100,000 sitting in a timed fund, then you lost 2,000 bucks last year because your fund was letting guys do this.  MR. JOHNSEN:   The question is compared to what.  Compared to what.

MR. HASSETT:   Well, compared to the fund that doesn't allow them to do it.  And so why shouldn't we, you know, help people learn -- well, let's go to the floor.  We can argue about this.

MR.          :   Actually, I'm glad Bruce brought that up because I want to ask Eric specifically -- I'm Jim Glassman (ph) from American Enterprise Institute.

I want to ask Eric specifically about what the costs are to investors, because I've been trying to figure this out for a long time, and I've read most of what you've written.

Your research focuses on specific kinds of funds, and yet there are all sorts of numbers that have been thrown around of what the losses are, what you would call dilution, right, to mutual fund investors in general.

Can you put a figure on what the cost is in general, the $7 trillion invested in mutual funds, how much money are mutual fund investors losing as a result of these practices, and can you explain as no one else has in a very clear way exactly how this money is lost.  Where does it go, how does it disappear.

MR. ZITZEWITZ:   Okay.  So I only have a 15 percent sample, so I'm just grossing up from that sample, which admittedly is a bit of a leap.  The total losses across all the asset classes in 2001 were about 5 billion.  So out of 7 trillion in total assets, about 5 billion, that's just under 10 basis points.  Most of that is in international.  Over 4 billion of that is in international.  There's 400 billion in international assets, so it's about 100 basis points on average in international.  So it's about 10 on average, about 100 in international.

So the source of those losses is, as I tried to say earlier, is the fact that you've got these people buying shares of the fund at a price that's lower than its current value.  So you could think, I mean, you could just do that directly.  You could sell someone a fund that's worth a dollar for 99 cents, and if you do that, what you're going to do is you're going to grow the shares of the fund by one, but you're going to only grow the assets of the fund by 99 cents, instead of by $1.

And it's the long-term shareholders that are on the other end of that transaction, and so that's how the -- you know, that's how the money is disappearing from the fund.

MR.          :   Thank you.  That's a good quick explanation.  But let me just ask, is grossing up your sample really a valid thing to do?  Because you --

MR. HASSETT:   It's a random sample.

MR.          :   No, it actually is not a random sample.  You focused on international funds, which is clearly where most of this activity is going on.

MR. ZITZEWITZ:   Well, I grossed up within asset class, so the question is are the international funds in my sample representative of international funds in general.  And, you know, the short answer to that is we really don't know.

I have turned out the option to funds that feel maligned by having this average applied to them to send me their daily flow data and I'll calculate a special number just for them.  So far, no takers.

MR.          :   Okay.

MR. HASSETT:   But this fellow wants to specifically address the representativeness of the sample, and on that point I'll (inaudible).

MR.          :   Sean Collins, Investment Company Institute.

I just wanted to -- I have a lot of sympathy with what Eric has done.  You know, I have gone through his papers in excruciating detail, and but this is a case in point where I think you need to be very careful about extrapolating from your sample, which is the trim-tab (ph) sample to the industry.  You can't do it.

And the reason is that the trim-tab sample, which is what Eric used, overrepresents small companies.  Seven of the 10 largest complexes are not in the trim-tab's data.  So, for example, you don't have Fidelity, you don't have T. Rowe, you don't have American funds, you don't have Vanguard.

Seven of the 10 largest complexes who will have most of the industry assets are not in there.  So what we don't know is can you take the 10 or 15 percent sample that you have and apply it to the industry in general.

And I mean we don't really know.  But, you know, you sort of look around and you say, well, Vanguard, Fidelity.  Keep your fingers crossed that nothing is going to happen there, but so far nothing has.  That's one thing.

Another thing is what you did in your paper, if I understand what you have done correctly, is you take a simple average of dilution rates and apply that to industry assets.  What you needed to do is take a weighted average, and the weighted average dilution, if I understand what you have done correctly, is much lower than the dilution rate that you actually came up with and applied to industry rates.

MR. ZITZEWITZ:   So I guess in reverse order, if you value-weight rather than equal-weight, you get numbers that are about 80 percent of the numbers I reported.  So maybe it's four instead of five.

MR.          :   In trying to replicate your numbers, the dilution rates that we come up with on a weighted average basis, using the trim-tab sample, are much lower than -- I think your average is something like 50 basis points.  The weighted average is considerably lower than that.

MR. ZITZEWITZ:   Okay.  Well, I have done precisely that calculation and gotten different answers.  So I don't know the source of that difference there.

I am leaving out variable annuities, right, so I mean there were whole other classes of funds that we don't know the dilution of.

MR.          :   If I could just emphasize a point.  The point is, to answer Jim Glassman's comment, is that no, the $5 billion you cannot -- the $5 billion which is Eric's figure is for the entire industry, if I understand correctly what you have done.  And that's -- you know, we just don't know what the correct number is, but my sense is that that's probably a wide overstatement of what the true answer is.

MR. ZITZEWITZ:   So I won't say you've had your say, now I'm going to have mine, because that's not a recipe for career success, but --

[Laughter.]

Yeah, value-weighted versus equal-weighted, at least when I have run the numbers, it doesn't make as big a differences as you're representing.  You know, I think the bottom line, though, is we don't really know.  The best thing we can do is gross up from a sample.

It doesn't surprise me that seven of the 10 largest companies are not in a 15 percent sample of the industry, right?  I mean if it were on average 8-1/2 percent, 8-1/2 of them wouldn't be in there.  So, you know, but we don't really know.  We have a 15 percent sample.  In that 15 percent sample alone, there is, you know, half a million -- half a billion dollars of dilution in 2001.  So I think even if we don't know anything about the other 85 percent, it's still a pretty serious issue.  It's not an issue that should be downplayed.

And as more data comes out, as more data is made available to me, certainly, I'll be happy to run the numbers for the rest of the industry and, you know, we can get a better estimate than the one we have right now.

But, you know, I think in terms of prioritizing this issue compared with other ones, we have got to make the best guess we can.

MR. HASSETT:   The one thing that comes to mind, though, is that it is true, is it not, that the historic returns, which are not a guide to future returns, would include the penalty from this.  So if you've got a mutual fund that's been making 8 percent instead of 9 percent for a long time because, well, with international funds, a lot of them have been losing, but just say that because in part they were allowing a lot of dilution, then you ought to see that in the statistics and then decide to go to the one that had the higher historical or the better historical record.  Is that correct?

So unless there's a big change in this behavior, then maybe it's already reporting.

MR. ZITZEWITZ:   Well, no, it does show up in the return.  People are not perfectly responsive to returns and, you know, in part because of the issues Chad was talking about, in part because returns are --

MR.          :   But Kevin's point is really important, because my sense is people are way too responsive to returns.  So if returns are reduced, it does come out in the wash.  People will reject those funds, it would seem to me.

Can I just ask my other question really quickly?  Because we have somebody from the ICI here, because I have always been curious about this.

Why -- and I'll ask Bruce, but if the ICI person wants to respond, great.

Why does the ICI want to restrict soft money to the large full service brokerage firms rather than other firms?  Why restrict it at all?

MR. JOHNSEN:   I mean that's my question.  You know, I don't know.  I mean maybe the lawyers at the ICI have got -- you wanted to, you know, kind of publicly demonstrate that they are interested in doing anything they can to clean things up.  And for whatever reason, soft dollars have just gotten a bad name, and full service brokerage, because it's the way things have always been done, just doesn't have a bad name.  You know, so I mean that's the best I can say.

But, Brian, maybe you want to, or the --

[Inaudible conversations.]

MR. JOHNSEN:   Sure.  Sure.

MR.          :   There's one other point I'd like to make, is that --

MR. HASSETT:   Your name.

MR.          :   I'm Brian Reed (ph) from the ICI.  And that is if you -- to eliminate all soft dollars, my understanding is -- I'm not a lawyer, but you'd have to eliminate 28(e) and that's not something that --

MR.          :   What's 28(e)?  I'm not a lawyer, either.

MR.          :   That's the soft dollar provision.

MR.          :   Okay.

MR. JOHNSEN:   Well, 28(e) is a safe harbor that says any manager who pays up for brokerage to get something like research that he believes is in the benefit of the fund will not be presumed to have violated a fiduciary duty.  And then the question is, what is research, and the ICI's recommendation, as we know, they interpret research to be just what's coming out of the full service brokerage houses.

MR.          :   So basically the ICI's position at this point has been just simply to ask the FTC, which can do this, is to narrow the sort of --

MR. JOHNSEN:   Interpretation.

MR.          :   -- interpretation for the safe harbor.

Now to eliminate all of that would require 28(e) be repealed, and that would be to take congressional action.  MR. JOHNSEN:   It would require more than that.  It would also require criminalizing any kind of agency conflict, because understand, just because an agent has accepted these payments doesn't mean he's violated agency law, okay.  So, you know --

MR. HASSETT:   And we're going to go to one and then two, and then are there any other ones, because we are kind of running up against our deadline.  But before we go to one, I just want to -- I just thought of the answer to my own question, which is that -- for you, Eric, but I think it's really important, because as Jim said, Kevin is on to something, but I wasn't.

[Laughter.]

The point is that in the historic return, the difference is that if you say you have like a moron picking stocks, and somehow he's underperforming the market in the past, well, that doesn't mean he's going to underperform in the future, because we know that a monkey with darts would be able to at least perform the market.

But if you have some guy who's screwing you by giving your money away, well, he'll probably do that again in the future.  And so if you have an -- if you underperform in the past because the guys were cheating, then that's a sign that they would underperform in the future perhaps more than just whether, you know, they happened to pick a bad bundle last year, and so you would want to know this extra information.  You would want to know the dilution information.

MR.          :   (Inaudible) is that they buy the funds that perform the best in the short term historically, so they have seen worst returns, but likely they are going to penalize those funds that produced the lower funds.  So it's bad for the firm to produce low returns.

MR. ZITZEWITZ:   In fact, there's actually two anomalies.  There's a segment of investors that chase the best fund too much.  There's also a big segment of investors that stay and lose their funds too long.

MR. SYVERSON:   It's a very nonlinear function.  It's not -- the reaction is very flat on the bottom end and very sharp for the very highest performing funds.

MR. JOHNSEN:   It's a new money, old money thing, isn't it?

MR. SYVERSON:   Yeah, I don't know.

MR. HASSETT:   Let's go to the questions on the floor.  We have one and then two and then that will be the final question, and I will give each person a minute to say a final word before we invite Peter up.

Name and affiliation, sir.

MR.          :   I'm Larry Harris, chief economist at the SEC.

Our speakers today, certainly Bruce and Chad, have spoken eloquently for in one form or another the principle of caveat emptor, and Kevin as well, that the buyer eventually can figure out what's happening and that accordingly things will be self-adjusting and self-leveling.

The assumption, of course, is that the buyer ultimately can figure it out, and the problem here is that where information is buried into returns, returns are phenomenally volatile, and extracting a 1 percent signal out of something that has a 15 percent standard deviation on an annual basis -- and that tends to be a -- that's a low number, 15 percent -- essentially requires 30, 40, 50 years of data within a degree of reliability.

Those numbers are -- that just comes out of mathematics.  That's not subject to dispute.

As a consequence, it is very difficult to rely on the caveat emptor principle.  Chad pointed out very clearly that search costs are very large.  Part of the reason search costs are very large is because it's extremely hard to get the information.  So disclosure becomes really important as a means of getting better information.

That said, when we think about soft dollars, I am very impressed with Bruce's explanation about the -- that's characterized by the picture in the upper right-hand corner here, but I will note that the soft dollar issue is not one of preventing the bundling, it is merely reporting fairly where expenses are being incurred.

The principle associated with the accounting is that the accounting should separate performance from the cost of obtaining that performance.  And so what people think about are proposals that are designed to improve the accounting system, not eliminate the practices that you eloquently defended.

Now, as we think about that, I'll point out quickly that if we report only the commissions, as some have proposed, the problem with that is that you end up shifting trading decisions into proprietary trading -- into dealer -- basically favored dealers over agents.  And the dealers ultimately then become the providers of soft dollar services, only now it's no longer soft dollar services, it's soft dollar pricing pattern.  And since we don't want to make regulations that will change order submission strategies in this way, that's not sensible.

MR.          :   (Inaudible.)

MR. JOHNSEN:   Well, I have nothing against caveat emptor.  I'm certainly willing to acknowledge that information is the problem here.  One of my responses is it's not just an investor figuring things out, it's the world of investors, and as we know from "Who Wants to be a Millionaire?" the audience can become extremely accurate.  Now how they coordinate, I don't know.

But if your neighbor leaves a fund because he thinks it's wrong, you know, and he tells you, whatever, there are ways that that information does work its way through markets.

And as to disclosure, though, I mean the question is, why do we need mandatory disclosure?  Why is it that these funds don't have all the incentive to disclose in their own interest?  I mean they are in some sense competing against one another, and you know, agents routinely do voluntarily disclose to their principals.

I have yet to see a really convincing explanation for that.  One of the problems with mandatory disclosure, of course, is we are bordering on situations in which what is being disclosed is proprietary.  I mean James Perturba's (ph) work on copycat funds -- and I just have to wholeheartedly disagree with Eric that daily fund flows -- you know, mandating even delayed daily fund flow disclosure, you know, would be costless to funds.  You put that together with somebody knowing what's the composition of their portfolio, they know how they're trading, pretty soon they're all over that fund, you know, and so I don't just necessarily agree with that.

Now as to accounting, you know, my response is soft dollars -- I mean maybe you should have formalized accounting for brokerage commissions, but soft dollars do formally account for things.  They make things clearer compared to the alternative of full service brokerage, or the alternative of trading through dealers, right?

MR. HASSETT:   Let's go to the last question, because we're stealing so much time from Peter.

MR.          :   I'm (inaudible) from the Department of Labor, and my question is on soft dollars.  You used the example of the principals paying for the agents, the employees' business travel, because otherwise people wouldn't travel.  I think part of the problem with soft dollars is the definition of research.  I know my employer, the U.S. government, would not let me buy desks and my computer and everything else because I'd buy the best thing I could at my principal's expense.

Yet in terms of soft dollars, you can buy fancy flat screens, you can subscribe to the Wall Street Journal, your Blumberg (ph) terminal, all sorts of things which most people would say are basic business expenses of the agent, and are able to be bought at the very highest level perhaps beyond what you need.

What would you say to a more restricted definition of research?

MR. JOHNSEN:   Well, I have already said what I have said about the ICI's restricted definition of research.  I think it's too narrow, but the point here --

MR.          :   (Inaudible.)

MR. JOHNSEN:   What is driving my story is that the broker posts a performance bond, and the standard version is the bond should take a form that has the highest possible value to investors, subject to the restriction, that the broker can't get it back if he cheats.

Now research comes immediately to mind.  Right now we are on the -- you know, on to the question of whether or not funds compensation brokers with -- what do they call them, revenue-sharing or rewarding brokers for selling their fund's shares is permissible.  That's certainly not research, but it might have value to shareholders and in a sense fulfill this performance bond thing.

One answer to your question is why don't managers sometimes engage in commission recaptures, that is take the dollar fro