Expensing Employee Stock Options Looks Like a Major Mistake

January 8, 2004

Unedited transcript prepared from a tape recording

Proceedings:
MR. HASSETT: [In progress] -- interaction with the floor. We also have papers that have been prepared for the conference which are available in the materials in front of you. And, finally, the best news, we're feeding everyone after the conference, and so if you want to stick around and talk about issues raised during the conference, we'll have a buffet lunch starting right at 1 o'clock.

And so, with that, I'll hand it over to Jim, who will be moderating the first panel.

MR. GLASSMAN: Thank you, Kevin. We have a lot to cover so we want to move fairly quickly.

I'd just like to say that in recent years never have I seen so much heat and so little light generated over an important public policy issue as lately with the issue of the expensing of stock options, which I think for the public may be a somewhat arcane matter but has important economic consequences, and I would also say political consequences, but I'm not sure we're going to get into that today.

Decades ago, accountants were entangled in much the same debate that concerns us today: how to value compensation offered by corporations to key employees in the form of stock options, that is, the right to purchase shares after a fixed period at a fixed price. In 1972, the Accounting Principles Board, predecessor to the current Financial Standards Accounting Board, or FASB, threw up its hands. It issued Opinion No. 25, which stated that no compensation expense need be recognized for stock options granted to employees, quote--and this is an important quote--"because of the concern that stock options could not be reliably valued at the exercise date."

Later, the FASB offered companies a reasonable choice: either use an options pricing model and deduct the value derived as an expense in computing the company's earnings, or use the so-called intrinsic value method, which most companies use today, and show in a footnote to the financial statements the hypothetical or pro forma effect on earnings of the issuance of options; and then, when options are actually exercise, their dilutive effect will be noted on the balance sheet and on earnings per share.

Oh, very nice. Thank you, Kevin. You probably didn't understand anything I just said.

I should add that a footnote doesn't really capture it. A footnote most people think of as a few words at the bottom of a page. Footnotes used in explaining the stock option situation by most companies extend for many, many pages. And, in fact, as I have noted in the past, we get on most financial statements more information about stock options than we get from other functions of companies that are probably more important, like other kinds of cash compensation, where expenses occur in the company, where revenues are derived and so forth. So there's lots of information in this so-called footnote.

This compromise seemed to handle the substantive objection that was raised in 1972, but a few years ago something changed. As a direct result of the scandals involving Enron, WorldCom, and other companies, political pressure has developed to force companies to come up with a single number on their P&L statements. This pressure has been intense, and the FASB is moving swiftly to enact an expensing provision. Many CEOs have said outright that if the provision is enacted, they will reduce or eliminate their broad-based options programs which many, including this speaker--that is, me--believe would have a major depressive effect on the economy at a critical time. But today we will focus on the accounting issues, on the same accounting issue--and I think this is the point--in fact, that has concerned people who have looked at this issue for many decades, the one that was resolved originally in 1972 when the FASB's predecessor said no, we can't value options accurately, just to quote that again, "because of concern that stock options could not be reliably valued at the exercise date."

Today, we have the opportunity to examine that proposition by focusing on two papers:

The first, by Charles Calomiris and Glenn Hubbard, looks at the potential benefits of deriving--developing rules for expensing options and reviews the practical problems of creating accounting conventions that would properly value options, and the conclusion, if I may give it away, Charlie, is that establishing rules for expensing options would likely do more harm than good.

Mr. Calomiris is the Henry Kaufman Professor of Financial Institutions at Columbia University and a visiting scholar here at AEI, where he is also co-director of the project on financial deregulation. Mr. Hubbard, who is not here with us today, is Russell L. Carson Professor of Economics and Finance at Columbia, and, of course, served President George W. Bush as his first Chairman of the Council of Economic Advisers. He is also a visiting scholar at AEI.

Mr. Calomiris will present his paper--not quite yet, Charlie--and Deen Kemsley, who's right here, who is the David W. Zalaznick, Jr., Associate Professor of Business at Columbia, will be the discussant.

The second paper, by Kevin Hassett, who just introduced me, who's director of Economic Policy Studies here at AEI, and Peter J. Wallison, who is a resident fellow at AEI, looks at the economic and legal consequences of requiring the expensing of employee stock options without specifying the valuation method. This appears to be the approach favored by the FASB, not specifying the valuation method. That's one way to get around the problem of the fact that it's not reliably done. The paper takes a dim view of this approach, as you will see.

Mr. Hassett, who is a former senior economist at the Federal Reserve System and a co-author of a very important book about the stock market, and Mr. Wallison, who is co-director of AEI's program on financial market deregulation and former general counsel to the Treasury Department and later counsel to President Ronald Reagan, will present their paper, followed by commentary by Paul Atkins, the distinguished Commissioner of the Securities and Exchange Commission, and Professor George J. Benston, the John H. Harland Professor of Finance, Accounting and Economics at Goizueta Business School at Emory University--I think the competition here is to see who has the longest endowed chair title--and the author of many books, including most recently "Following the Money: The Enron Failure and the State of Corporate Disclosure."

Let me, incredibly briefly, because we don't have much time, just raise some questions that I personally would like to see addressed here, and if they aren't, I'll make sure that questions are asked.

Do analysts and investors get enough information for pricing stocks from the current regime--that is, essentially, from the footnotes?

Do you agree that there is some value to stock options--that is to say, not zero--at the time they are issued? And if so, shouldn't that value appear somewhere on the P&L statement?

Are there examples of other revenue and expense items that cannot be accurately included in P&L statements and instead should go into the footnotes or elsewhere? Now, Peter Wallison has written a whole book on this subject. Or should they be handled in this way?

What do you think of compromise measures such as the Enzi bill requiring expensing only for options issues to the top five officers?

And, finally, where's the beef? What is the problem that the FASB's proposal seeks to solve?

We'll start with Charlie Calomiris.

MR. CALOMIRIS: It's a pleasure to be here today. I guess I should start this presentation by pointing out that for an economist teaching corporate finance and researching corporate finance and financial institutions, when somebody asks the question, How much difference will it make when we change our accounting standard? our first answer is generally, well, not much; and, in fact, much less today than a hundred years ago. And so I think it's important to first point out that the prospective benefits of any change in FASB rules from the standpoint of how we teach corporate finance to the MBAs who actually do those calculations that run Wall Street, the answer to that question is not much. So, to a very large extent, I think if you take nothing else out of my comments, discussions about accounting standards need to take a reality check for the way markets actually function. And the bad news for the accounting firms and for even the accounting professors maybe is maybe you guys overestimate your importance.

In fact, I like to tell my students when they come into advanced corporate finance that half of my job is to unteach them the accounting that they learned from the accounting professors. But since this is my colleague at Columbia in accounting, there's a happy little discussion going on about that issue in the background here, and I want to bring it to the foreground.

When we teach valuation, we really do teach students to undo accounting. We teach them that what the value of any asset comes down to is estimating the future time path of what we call free cash flows, which is the amount of cash flow generated by a company less the amount of expenditures they're going to have to make to continue to operate as a company. So you have to make some investments. You have to do maintenance. Those are your expenditures. And then you're getting some profits net of your ordinary expenditures, and it's the difference between those two that's called free cash flow. And corporate finance is a pretty simple idea. Estimate what those free cash flows are, and then discount them by some discount rate that reflects the riskiness of that stream of free cash flows.

That's basically, believe it or not, how we make a living in business school, is just teaching them how to do that very simple thing.

Notice that that doesn't have anything to do with earnings per share or net earnings concepts or whatever. FASB is not going to change the price of General Motors. If FASB changes its rule tomorrow and says let's divide everything by two, General Motors stock is still going to be the same. So, first, let's begin with that reality check.

What about unsophisticated investors? Okay. I think I've just tried to convince you that the price of stock in the market is not going to be affected by whether you decide to have the footnote be included into the calculation. Okay? That's ridiculous on its face. We've already got the number in the footnote. Now we're going to subtract it for the investors, and that's going to affect price. Of course not.

But there's another argument for accounting standards not having to do with the effect on price. It just has to do with, gee, maybe we're helping unsophisticated investors be able to evaluate what's going on.

Well, again, in the other finance courses, asset management, what we teach people is if you're not a sophisticated investor, that is, one of the people helping to determine prices of stocks, you really shouldn't be out there doing stock picking in the first place. So implicit in this notion, which is sometimes the notion that the SEC has as well as FASB, that we need to really worry a lot about accounting standards because of unsophisticated investors, implicit in that is this very strange idea that individual unsophisticated investors are supposed to be engaged in stock picking, the sort of Jeffersonian stock market idea that, again, in any good finance department in any business school we're teaching people not to do.

So there's a sort of disconnect, again, between reality and accounting in terms of accounting policy that I think is important to mention. The SEC really is sending a disingenuous and wrong message in FASB II if it really believes that individual investors, unsophisticated individual investors, should be out there trying to use these accounting measures to pick individual stocks. So that's by way of saying, gee, we have a hard time understanding why this is an important issue.

Now, if that's an argument that has always had appeal, it has much more appeal today because suppose you were, let's say, a semi-sophisticated investor who wanted to, just for entertainment, play with the stock market and you wanted to have some reasonable idea as to what was going on in a certain set of firms or a certain firm. Now, the question is: In today's world--in the world of yesterday, you might have said, I'm going to really rely on the paper that I get through the Postal Service. And so it might matter if the accounts make sense. You could still do the subtraction, of course, take the information out of the footnote. So this issue that we're talking about still, you know, you should be smart enough to do the subtraction. But you might think that accounting information is still very useful to that investor who's relying on what's coming through in the mail and having to make a quick decision based on the definitions that are arriving in the mail.

But I have good news for you. You now live in the information age, and in the information age you can push a button and have firms' accounts restated from international accounting standards to GAAP or to any accounting standard you'd like or any concept of earnings that you'd like. And in a very nice book that Bob Litan at Brookings and Peter Wallison wrote called "The GAAP Gap," I think it's called, written in 2000, I think, they said, you know, start thinking more about the information age, everybody, and what's really going to happen is everybody's going to be competing. Financial institutions will be providing for their customers their own idea of accounts based on the disclosures that are available on the Web from firms and probably on a higher frequency going forward, too.

So if we're really going to be thinking for the next century or the next decade, even, about what we should be worrying about, it should be disclosure issues, and we should have disclosure requirements that allow investors to see information quickly, to have access to it on the Web, and then let accounting standards become competitive. We already have two competitors--international standards and GAAP--but we also have lots of other ones, really, which are the various investment advisory firms, the financial institutions, the analysts who are out there, and they can put their own spin on earnings, and let them compete for investors to show that they have the better way to define what earnings are. Each of them will have their own way of trying to value these stock options, subtracting them from earnings, and yes, of course, they are an expense. But they're an expense, as I'll argue, that differs across firms and across different circumstances and different dates. And we ought to let the market analysts decide that, let the market analysts present their own idea of what the right accounting for that should be in the firm, and let them compete with these semi-sophisticated investors that they're providing services to in the marketplace.

That vision of what finance information is all about and how it should be used I would argue is the right vision for the information age. And, again, there's a disconnect between reality and what we're hearing in this debate about regulations on accounting. Again, it's disclosure standards that I think we should be talking about, and Jim raised that in his comments, and we can come back to that later.

Now, notwithstanding everything I've already said, obviously there are going to be--to some extent, I would argue to a very small extent, accounting rules might matter. And so the question is: If you do impose a common standard for measuring these costs of these stock options, how wrong is that going to be? How much noise is that going to put into the system? And so that's--the remainder of my comments are really going to be about that. Again, having said up front I think the answer this is not a very important issue, that is, I don't think these standards are going to have much effect on anything. But to the extent that they will have an effect, I'm going to argue it's very likely to be a negative effect, because they're going to introduce a lot of noise. So if anyone is stupid enough to make investment decisions depending on accounting rules--we hope there aren't such people, but if they are out there doing that, they're going to be misled.

Okay. Now, why are they going to be misled? Because presumably there are these unsophisticated investors who are paying attention to the accounting rules, and the accounting rules are going to be very noisy.

Now, where are these errors coming from? In the paper, we categorize these into four groups of errors that create this noise. Each slide picks up a different category. Here we get into a more technical economic discussion, but I'm going to make it as untechnical or nontechnical as I possibly can.

The first one is the problem of creating a cash equivalent having to do with anything that is corporate stock. Let me remind you that when a corporation actually wants to issue stock or options on stock--let's just talk about it as stock because it's pretty much the same. If a corporation wanted to sell warrants or sell stock, warrants on stock or stock, it costs something to do that. When a corporation announces it wants to do that, it has to pay two kinds of costs. First of all, usually prices in the market decline on a corporation's stock for having announced that it wants to do that. And so there's a negative announcement effect. Secondly, it has to usually pay an underwriter to actually communicate the attributes of the firm to the potential purchasers of that security. That is a big cost. The average for a seasoned equity offering in the United States of those costs all told is about 10 percent of the offering.

For an IPO, it's much more. And if you take an IPO, a firm that's at that sort of stage of life cycle, you're going to say, well, what would be the cash that a firm gets back and the value it gets back after taking these two costs--the announcement effect and the underwriting costs--into account? What cash proceeds do you get back from $100 worth of a stock offering? And the answer is maybe half. Think about that for a minute. But if you are General Motors or a very long-lived firm in a stable industry, it might be more like 10 percent or 8 percent of cost.

So what I want to emphasize first is if the idea is implicitly we're giving you a stock option which you could sell in the market, well, if you were going to take those stock options to the market, there'd be some cost, if we did this large issue of options, of warrants or of stock, there could be a very high cost associated with trying to actually place those in the market. And so if you think about the granting of the options as a stock offering, for some firms the difference in cash equivalents is very large, and for other firms the difference in cash equivalents is very small. But there's always a difference.

Two conclusions from this point. First of all, we're going to overstate the cash equivalent cost if we just use a model that doesn't take those costs into account. Secondly, we're going to overstate it much more for small growing firms that tend to pay higher proportional underwriting costs and that have larger negative announcement effects for their stock offering. This is a big issue that I have not heard mentioned before in this discussion, and it's one that deserves to be raised.

The second issue that's sometimes raised--but, again, not very much--is options are often illiquid. Stocks can be very illiquid. Again, stocks or options on stocks for small, growing firms tends to be much more illiquid. That means that there's an illiquidity discount associated with these securities. If these securities are so customized, as they often are, that they wouldn't be widely traded in the market, then you can think of them almost as a private equity sort of investment. Private equity investments in terms of valuation tend to have something like a 30-percent discount relative to public equity investment. All right? So I'll say it to you a different way. On an annualized return basis, private equity investors expect to earn maybe 5 percent more on their annual portfolios than investors in a public equity portfolio because of the illiquidity of those investments. Illiquidity is big, and in present value it translates into something maybe for some firms on average, let's say, about 30 percent.

Well, if these options are very illiquid, we have a very big discount associated with them. If we use a model that assumes that discount away, like any standard finance model--the Black-Scholes model, the binomial model--we're, again, substantially overvaluing options and, again, substantially overvaluing them even more for small, growing firms.

Error number three. Options are derivatives, and they are written on an underlying security called a stock. Now, the Black-Scholes model assumes geometric Brownian motion, characterizes stock price--stock returns movements over time. Normally distributed returns following geometric Brownian motion. Why do we do that? Because that's how you get what we call closed form solutions in finance. If you want to actually solve it mathematically in a closed form, not with simulations, in closed forms, you need to have basically that assumption. We didn't pick that assumption because it was empirically accurate. And, in fact, it isn't empirically accurate. The stocks on which these options are written, in other words, and the options, don't have values empirically that look like the Black-Scholes model.

We tend to find a whole variety of violations of this. The shape of the distribution of returns is not normal. It suffers from skew--lopsidedness, in other words--and fat tails, or sometimes called exocretosis (ph). It also tends to have momentum effects and mean reversion effects, and those differ by classes of stocks. Just as illiquidity differs across stocks, the properties of distributions of returns and serial correlations in returns differ.

But the Black-Scholes model assumes what we know isn't true about underlying stocks, which is that they follow a random walk, a Brownian motion process with normally distributed returns. Again, we do it for convenience. When people are actually out there in the market pricing options and trading in them to make money, they take into account with their simulated values these deviations. But we don't with the Black-Scholes model.

There are these other points I mentioned. Jumps are also another very fashionable thing now in finance to include not just continuous movements in prices but discrete jumps. And if you've lived through the last couple of years, you've been jumping quite a bit, I'm sure--hopefully not off of any building.

Then there's this little problem of estimating volatility. Well, volatility changes over time a lot. How far back should you be looking in estimating volatility? Should you use a GARCH model, an ARCH model? What kinds of--how should you be modeling volatility? Well, that's something somebody spending his own money on Wall Street, betting with, should be spending a lot of time thinking about, and that's who actually is determining the answer to that question. He's not using the Black-Scholes model to answer that. The Black-Scholes model takes volatility as a given. Where does that come from? I don't know. I don't know where we get that number. There's a lot of disagreement about it.

So my point here is when we just look at--we have to remember that options are written on stocks, and we really don't know how to think about in any consistent way across all firms, and certainly not even for a particular firm, what those assumptions should be about the stocks. How can you value the derivative if you don't know how to value the underlying?

To give you a little bit of an idea here--this is taken from the Campbell, Lo, and McKinley textbook--if you don't know--for a particular stock in the strike price, if you don't know whether the serial correlation is of zero, as under Black-Scholes, or negative 0.2 or negative 0.45, the difference you get for a strike price for a very out-of-the-money option with a one-year time horizon here, it's double. Okay? So a small difference in assumptions, especially for very out-of-the-money options or for very high--long future options--this is only one year. If we did five years, it would be even more. The difference could be double or triple the value of the option.

All of the adjustments I'm telling you--illiquidity adjustments, external finance cost wedges, and these technical matters--they're not small adjustments. It's not like we don't know--we know the value within 5 percent. Uh-uh. We don't know the value within 50 percent. We don't know the value--maybe, you know, it could be double or triple.

Finally, well, the market price, even if everything else about the Black-Scholes model were right, there are no external finance costs, that's the first point. Everything's perfectly liquid and stocks follow geometric Brownian motion, random walk, with continuous movement and no jumps and zero serial correlation, okay? That's the--those are all the assumptions. Even if all that were true, well, employees don't have the same incentives to exercise American options, options that can be exercised over a long period of time, as the market participants do. Employees can't hedge or post their options as collateral. Employees have to wait to vest, have blackout periods where they can't be involved. Employees may forfeit.

So what the literature on options has shown is that, in theory and in fact, employees tend to exercise options sooner, and also employees tend sometimes not to be able to exercise options. So an option that may look like to a market participant it has a life of seven years may look like to an employee it has a life of two years. And employees may act that way and actually exercise the options a lot sooner. Again, this is a very standard point.

Now, one of the big articles advocating options by Bode, Caplan, and Merton that came out last year in the Harvard Business Review recognized this problem and said, yeah, you certainly wouldn't want to have a one-size-fits-all model since different companies are going to have employees in different circumstances that are going to exercise their options differently, and they admitted and agreed that this would mean options are overvalued when looked at using the market perspective rather than the employee's perspective and substantially differ across firms. So we're back to the same thing. Options are overvalued, and in some firms much more so than in others.

A simple example from the real world--and I haven't--the reason in the paper on the cover says "preliminary" is we're still working to make sure we understand these facts and what's behind them. But Microsoft has options which in its 9/03 10-K it values, as you can see here, let's say, somewhere between $12 and $15 in its footnotes to the 10-K. It arranges to have J.P. Morgan Chase buy some of these options from the employees, and it looks like the purchase price that J.P. Morgan offers the employees, if I understand this transaction properly--again, I'm qualifying it a little because I need to do further research to make sure that I'm comparing apples and apples here, but I'm pretty sure I am. Transactions prices, as you can see, are much, much lower, and consistent with the view that J.P. Morgan Chase is thinking that the employees are thinking that these are two-year options not seven-year options, maybe that the volatility is different, et cetera, et cetera. So we're not sure exactly what's causing that big difference, but it is a very big difference.

So here are our conclusions because I know I'm out of time. The "just do it" approach that you should--okay, we know that it's hard, but just do it because you're going to get it close anyway is wrong. You're not going to get it close. There's a lot of variation that needs to be taken into account. A one-size-fits-all model isn't very helpful. And I think it's clear it's going to introduce a lot more noise than it's going to be helpful. I think it's very disingenuous to try to convince people to raise, to elevate in importance these numbers by putting them into the bottom line, as if we have confidence in them. They belong in the footnotes, in fact, maybe in smaller print in the footnotes. Maybe that should be our reform: reduce the font size of these numbers in the 10-K.

I will note there are lots of other similar valuation problems that we have: real options, various contingent assets and liabilities. The way we deal with complex contingent assets and liabilities of firms is in the financial markets. We deal with them because people have to value those as part of the valuation of the firms and behave accordingly. And if you don't know how to do that, maybe you shouldn't be picking stocks. That would be my advice.

I'll end just by saying, again, I think we don't want to get too exercised about this. I was telling my kids about it over dinner, and I said, "But, you know, to be honest"--because they're 12 and 14. And I said, "To be honest, if I were you, I wouldn't worry too much about this because I don't think ultimately the republic will rise or fall on whether we do this." And I would push us more toward thinking about disclosure issues and about trying to get us closer to that information age model that I think we really should be heading toward.

Thank you very much.

[Applause.]

MR. GLASSMAN: Charlie, before Deen starts, let me just ask you a couple questions. Your last comment about the republic will not rise or fall; however, the direction that you and Peter Wallison and a number of people in this room have talked about for reforming or changing the way we think about accounting really moves in completely the opposite direction, which is very important for investors. In other words, investors need to know that it ain't one number, in fact, and that there should be competition, that it is not that the problem is there's this number that everybody knows about that is being withheld by evil corporations or by people that have specific interest in withholding it. That is the wrong message. And, in fact, the republic may not rise or fall, but it is important to the republic that we move in the direction that you're talking about and that Peter's talking about.

Just to clarify something you said in the very beginning, the price of the stock, you're saying, will not be affected to a great degree by whether there's a change in this accounting standard. Do you see any benefit at all in making the kind of change that the FASB has proposed, any benefit at all?

MR. CALOMIRIS: I see a lot of costs, and, no, I don't see any benefit. I mean, there's always going to be someone out there--I guess I could imagine some person out there who says, Oh, I never knew that there were stock options now that all this discussion is happening, so maybe I should look at that issue a little more closely.

MR. GLASSMAN: Let me just say, one of the benefits that you hear being discussed by public officials, politicians, is that they seem to want to discourage stock options, and they think if I'm making this accounting change it will discourage the issuance of broad-based stock options.

Now, first of all, do you think that would be beneficial if this practice were discouraged?

MR. CALOMIRIS: This is the fun part of the discussion. I know Deen's probably aching to get started, so I'm going to be really brief.

The fun part of the discussion is this question and what you alluded to when you mentioned the political motivations of this whole topic. Michael Jensen, a very distinguished financial economist at Harvard, who is known for hating CEOs deeply and being deeply suspicious of them, really, he's all for stock options, but he wants to regulate them. He wants to control what they look like. And it's very clear to me that he has drawn connections between the existing stock options and what he perceives to be problems in corporate governance and stock price volatility that I think are far-fetched, to put it mildly, and that I think that there is a sort of crusader out there that's pushing this battle, partly because what they really want to do is change the regulation of stock options. And I think that there's a lot of people with a very wide array of hidden agendas on this issue. But I didn't want to really deal with that in my comments because I think it's better not to question people's motives but, rather, to question their logic.

MR. GLASSMAN: Thank you.

Deen Kemsley of Columbia University.

MR. KEMSLEY: Thank you very much. Thank you for inviting me here to talk about this paper that Charlie here and Glenn Hubbard have written, and I look forward to the discussion.

As you might think, I'm an accountant so I bring a little bit different perspective, at least to the first half of the paper, so we'll talk about that a little bit here.

The key question that the paper addresses and that we're all here to talk about is: Should regulators require firms to expense the costs associated with these employee stock options? And as has been stated, let's be clear, this information is already reported in a very extensive footnote. The only thing on the table that the FASB is dealing with is whether they actually book it in the income statement and have some kind of associated liability on the balance sheet. The information itself is already footnoted. So that's what we're addressing here.

What answer does Charlie and Glenn give? If you didn't hear it, his answer was no. Okay? And there's two parts to their paper. The first part of the paper is, at best, the potential benefit of booking the expense is very small. He just said none at all, so that's the first half of the paper. Maybe. Okay. Caveated. He's backing up a little, okay. Second is the potential cost of booking the expense is high because there's no clear way to estimate the value of these options; i.e., there's no clear way to estimate what this expense should really be. So let us quickly go over these two sets of arguments and see if they hold water.

First, the potential benefit is small. He presents three arguments there.

One, hey, it's cash flow that matters, not the accounting numbers. Firm valuations based on cash flow.

Second point: Don't worry about the small investors, you don't really have to protect them, after all.

And, third, analysts are going to get out there and compete to come up with their own accounting standards that could be better than anything the FASB comes up with in the first place. So the first and the third points here come down to the same point, which is essentially it doesn't really matter what accounting convention you have. So those are the arguments. Do they hold up?

The first one, cash is king, don't worry about the accounting numbers. It is definitely true, I agree, that firm value is a function of future cash flow. It's not a direct function of accounting. That is true. However, what is the accounting set up for? Accounting is set up in the first place to help investors get a better feel for what future cash flows are going to be. And, in fact, there's a lot of research out there that demonstrates that the correlation between current accounting earnings and future cash flow is much higher, is much stronger than the correlation between current cash flow and future cash flow. So bottom line here is, yes, cash flow is what matters, but accounting does, is at least intended to serve a very useful purpose in helping investors assess what that future cash flow is going to be. That's the role for accounting.

Therefore, analysts and investors often use earnings as a beginning point to begin assessing their future cash flows. I work with and have worked with the analysts at Morgan Stanley for years, and I can tell you that they are dealing with accounting numbers a lot. They care about those numbers a lot to try to assess what those future cash flows should be that go into their discounted cash flow assessments.

So third point was whether it's through the financial statements or not, managers typically use accounting conventions and FASB rules to communicate their inside information to investors and to analysts and so forth. So my point here is--you know, Charles has come in with a rather general statement that accounting just doesn't matter, and my point would be, you know, I work with the analysts and I know they care a lot about the accounting. However, it's still a very valid question to ask whether this specific accounting convention matters or not, and that I think is the question that's really on the table here today. So let's move on.

Second, don't worry about the small investors; they shouldn't be investing on their own, anyway; they should just be investing in index funds. My response is there's probably some truth to that. However, I hate to just put the small investors in that one box where the only way they can invest is through index funds. And the way, I think, to get out of that is the fact that they can rely on informed investors. There's enough information from the analysts out there now for the small investors that they can rely on to have informed investors. So maybe it is a little overblown, this great push to always protect the small investor. Small investors should be looking to the analysts or going to the index funds. So I partially agree with this point here.

Analysts compete by developing their own accounting conventions, so it really doesn't matter what the FASB puts out there, was the point that was made. This is true. But, alas, even analysts are not perfectly efficient. Therefore, the regulated accounting standard still has a potentially important role. It's often the regulators that are leading the analysts, not vice versa. It happens both ways. I see it from the regulators' perspective, but I also see it, you know, in the trenches. There's a lot of learning about accounting going on in the analysts' trenches right now. So they're not always at the forefront, and they're trying to catch up sometimes with the regulators.

So I believe the regulated accounting standard sometimes does have a very important role. In this case, what do I think is the key point? The key point is that GAAP already requires this information to be disclosed in the footnotes. The information is already there for the analysts, for all the informed investors who can turn around and communicate that to the uninformed investors. It is there already. Let's don't lose sight of that key point.

Therefore, as Professors Calomiris and Hubbard argue, there could very well be little benefit from booking the expense in the income statements. The upside, I would tend to agree with what has been said here, the upside is small.

How about the downside? Well, if there's significant measurement error and/or discretion in the recorded expense, i.e., if firms can manipulate the number, booking the expense, i.e., actually putting it into the income statement instead of keeping it back in the footnotes where it affects the bottom line, could actually be more misleading than just footnoting it because it gives firms yet one more way that they can manipulate that bottom-line number.

So I think the key question is how great is the measurement error. Can we estimate this expense fairly reliably? And if we can, it probably belongs there on the income statement. If we can't and if it's too easily manipulated, then maybe you're trying to make science out of an art, and you could mislead folks with that bottom line. So that's the key question. That's where the second half of your paper goes. So let's see where his second half goes.

As he argues, it's very difficult to estimate these employee stock options for four reasons, and these are the four, and he's already gone over four. Let me go over each one, one at a time, and see if I think the Calomiris-Hubbard arguments really hold up here.

The first one, equity issuance costs. Charlie spent quite a bit of time talking about this, and as Professor Calomiris said, the shadow cost of issuing new equity decreases the relative cost of issuing stock options. How would I say this? In my language, what I would say is there's a cost advantage to issuing stock options to your employees versus trying to issue equity to the open market, get the cash and turn around and pay the cash to your employees. I think that's the simplest way to make the point that Professor Calomiris was making. Why? Because there's a lot of cost to issuing equity to the public market that you will not incur if you just issue stock options to your employees.

Furthermore, this cost advantage will vary across firms, and it will not be booked. You know, given the new standard, there's really no way for the firms to account for the cost advantage of just issuing the option versus having to issue cost to the open markets where they incur all kinds of transaction costs. So on the first point that they make, I come down on their side. Yes, this is one of the problems with trying to estimate this expense. So I agree with the first one. So at least one of their four arguments holds up here.

Let's go to the second one. Employee stock options are illiquid, especially for the younger firms. But, you know, this is a problem because the Black-Scholes model that is used to estimate this expense is a model that assumes the options are liquid. Employee stock options clearly violate this assumption. So can we--what does this mean? Is it a big deal or not that employee stock options violation a fundamental assumption behind these valuation models? That is hard to assess. There's been some theoretical literature out there that's tried to assess whether this is an important issue or not. The bottom line is it's hard to assess how important it is, but it is a problem. It does violate the assumptions underlying the valuations of this expense.

So on the second point they make, I again agree that there is a problem, though the magnitude of that problem I think is hard to assess.

Point number three, to estimate the value of these options you have to estimate future volatility. Well, that's a problem because the Black-Scholes estimates critically hinge on future volatility. If you have a short-term option, you could get pretty reasonable estimates for future volatility. But these employee stock options tend to be long term. They vest over a period of years, et cetera, which makes it much more difficult to estimate future volatility. And as Professor Calomiris already demonstrated, if you're just a little bit off on your estimate on how volatile the stock price is going to be over the future, it could have a big impact on the estimate of the value of that option.

So it requires considerable judgment to estimate future volatility, so what does the FASB rule say? Well, in 1995 FAS 123 came out, and that's the rules that are used to estimate these options for the footnote. And that statement basically says the first thing you should look to is your historical volatility of your stock price. However, we at the FASB recognize that for some firms that isn't appropriate because the firm may have changed a lot from the past. Therefore, if historical volatility is going to give you the wrong answer, then you should use some other estimate of future volatility.

Well, what does that allow firms to do? Game the system. And firms do game the system here. Some recent evidence just came out which shows that firms will use the historical volatility when it serves them best, i.e., they get a low stock option expense from it. But when they think they're getting too high of a stock option expense from historical volatility, that's when they will go to future volatility, and it is asymmetric the way they do this. In other words, there's some general evidence now that they are gaming the system to use whatever approach will give them the lowest stock option expense.

So if firms do book this expense, as the new FASB rules are going to propose, this type of manipulation will affect the bottom line on the income statement. This manipulation is already going on now, but what it's doing is affecting the footnote disclosure. If we book it, it's going to actually affect the bottom-line expense.

By the way--I don't think anyone's mentioned it yet--the FASB is about to issue a new proposed statement in this area. I was just talking to the project manager on this statement yesterday, and it's going to come out sometime in February. And it will look in some ways similar to the old FAS 123, which is the statement used for the footnote disclosure, with some--they're trying to get it a little bit better, but there's still going to be just as much discretion for the firms to choose between methods, et cetera, which they will be able to use to game the system.

Finally, exercise timing. Employees often exercise their options early--a point that you made. Early exercise reduces the true value of the options below the Black-Scholes value. FASB is not naive to this problem, and, therefore, FAS 123 already allows firms to use the expected life of the option rather than the normal Black-Scholes life of the option. But that's a problem because that gives firms discretion over this key input into the Black-Scholes formula. So once again, they can game the system, and several recent papers have come out indicating that there's clear evidence that firms on average shorten the expected lives to reduce the stock option expense. So just further evidence that they are gaming the system already, and if we put this number into the front of the income statements, these games are then going to start affecting the bottom-line numbers instead of being relegated to the footnote.

One final point that they do not make in their paper. You know, this is all about estimating the value of these options now so you could record the expense. This is an ex ante measure of the value of these options. Ex post, the actual value the employees get from the options is, as you might expect, probably going to be different than, you know, your ex ante prediction. How different? Well, recent evidence documents that the actual ex post values of these employee stock options vary substantially from the ex ante Black-Scholes values; in particular, only 4 percent of actual ESO gains are within the range from 90 to 110 percent of the Black-Scholes estimates, and many of the estimates are off by orders of magnitudes. So that's just comparing the ex ante estimate to the ex post actual value that's obtained by the employees. That part of the estimation area of trying to get these estimates up front instead of waiting for it all to unwind on the back end.

So what are the key points? Booking the stock option expense instead of keeping it in the book (?) will increase discretion over reported bottom-line earnings. Why? Because firms can game estimated volatility and they can game estimated exercise dates. And, in addition to that, it's going to introduce other noise through problems with the Black-Scholes or other related formulas, such as the shadow equity issuance costs that create bias as well as noise. They could actually overstate the estimated expense. The illiquid option problem that's a fundamental violation of Black-Scholes, et cetera. And the key point here, which Professor Calomiris also makes, is this noise or bias would vary across firms. Some firms it would affect more than others, and yet it would be hard to unravel that from the bottom-line income statement number that we're looking at.

So what do I conclude? In some ways at least--I mean, there are two sides to the story, but definitely in some ways at least, I believe it is clear that expensing employee stock options could be a mistake. And why? It's because it's making science out of what is very much an art and putting that art on the front of the income statements where it affects reported earnings instead of letting it stay back in the footnotes where the analysts can decide themselves what to do with it.

So those are my key points in response to your paper.

[Applause.]

MR. GLASSMAN: Thank you, Deen.

We're going to have questions. I just want to ask or sort of emphasize one point. I like this phrase, "making it science out of an art," but you're actually saying that expensing stock options would do something else, which really I haven't seen discussed very widely, which is that it will make it easier for unscrupulous corporate officials to manipulate the numbers that they offer to the public than under the current system. Is that right?

MR. KEMSLEY: Easier to manipulate the bottom-line earnings number, yes.

MR. GLASSMAN: So right now it is possible to manipulate the footnote, but manipulating a footnote is one thing. Manipulating a number which you are portraying as science rather than art is more dangerous, you're saying?

MR. KEMSLEY: I would think so, yes. That's a heavy cost of doing it, yes.

MR. GLASSMAN: Questions from the floor? Now, I have to warn you, I'm actually going to make you walk over here--we had to do this this morning--because we do not have microphones. John Lott put up his hand first. One, two, three. You should stand right over here and talk loud. You can talk loudly from there, but if I can't hear you, I'm going to make you...

MR. : [inaudible].

MR. : That's a great question, and I think without--

MR. GLASSMAN: I will restate the question, or you will restate the question, Charlie.

MR. CALOMIRIS: The question is: Do we have some other way of estimating how important this question is? For example, if it weren't an important question, there wouldn't be much lobbying going on. So how much lobbying is there going on?

Now, I think the answer is there's a lot more lobbying going on than there should be. That is--

[Inaudible comment/laughter.]

MR. CALOMIRIS: No, I'm going to get to that, though. There's a sub-literature within finance that asks the implicit question in your question, which is: Do people who run firms overestimate the importance of some aspects of accounting relative to the way the market really reacts? And there's a lot of evidence that, for example, CEOs and CFOs of large firms place more emphasis on the effective accounting conventions than the market does.

So my answer to your question would be I don't know the number of lobbying effort. My sense is there's a significant one, and that I think that to some extent that reflects the fact that a general proposition that we have some evidence for, that managers of firms overestimate the importance of accounting's effect on the market.

MR. GLASSMAN: Charlie's answer I think clears up a paradox that I've often wondered about, which is, you know, why should it matter that much. And my argument is why should it matter to the politicians who are pushing FASB to do this, but why should it matter to corporate CEOs? And the fact is it does matter. They are not kidding. I guarantee you, they are not just saying this for political effect. They are not kidding when they say if this rule passes, we will stop our broad-based option programs. Whether that's the right thing to do in an economic sense or not, that's what they're going to do, and that could have an effect. And you say--

MR. CALOMIRIS: That's a good point, and I've actually thought of putting this into the paper. In the final version I think we will, putting in--notwithstanding the fact everything I've said, to the extent that it affects the behavior of the firms for reasons that I can't understand, it might be even more costly to do this.

I didn't put it in, but I might just--I think this discussion is probably useful. As long as I'm--can I just--I just want to emphasize. I don't want to give the perception I'm saying that accounting information wouldn't be useful to analysts, and I agree with Deen that it is. But I think operating income information is the core of what they're going to be using to forecast future cash flows, not these adjustments to that information coming off of things like this or other special adjustments. Those aren't going to be very useful for forecasting cash flows.

MR. KEMSLEY: Let me just add a couple things. First of all, it is a very hot issue. I was talking to the FASB about it yesterday, and they've been under tremendous pressure on this issue. So there is a lot of lobbying if you want to measure it that way.

Second, one other reason my managers care about this is because a lot of their contracts are tied to bottom-line numbers, et cetera, not only theirs but a lot of employees within the firms. So that's one reason why there's some interest in these issues besides the impact that it might have on the market.

MR. GLASSMAN: We have to move on quickly. This gentleman here. Identify yourself, come on up here. Identify yourself. I'll repeat the question. That's far enough. That's good.

MR. : [inaudible].

MR. : Well, I think that's a very interesting point. Let me restate the point, as I understand it. The point is, you know, there may be a reason to try to--for these political purposes we were describing. If you think the options are being used too much, why don't we sort of attach some poison pill to using them? And if the CEOs don't like having to report them as an expense, all the more reason to report them as an expense, because we want to stop this thing from being used.

So I think rather than have this--I appreciate your candor. Rather than have this argument be made through the back door of this ridiculous accounting regulation, why don't we have the argument right out? And I would take a different side than Michael Jensen or maybe you, while recognizing some legitimacy to what you're saying, but I don't think that that--I ultimately would not be in favor of regulating options or trying to limit their use.

I would be in favor of Kevin Hassett's policy proposal, which is--would take away some of the tax distortions that favor the use of options over direct stock granting. And that might solve most of the problems that people are worrying about. And I think that it's those kinds of debates that we should be having, to the extent that we have--people who have a legitimate concern should voice it, and we should talk about it.

MR. GLASSMAN: Just as an editorial comment, I think one of the reasons we don't have the debate in a direct way--in other words, why shouldn't Congress pass a bill that says we're going to ban stock options because we think this is causing corruption through American corporate life--is if there were such a debate, they'd lose it. But by putting it into this arcane accounting change, especially through an unelected body that doesn't seem to be subject to much in the way of political pressure, they might not be. Anyway, that's an editorial comment, and--

MR. : [inaudible].

MR. : Well, on the first point, there's just an adjustment period, right? You can adjust contracts that are tied to the current accounting scheme. And if you change the way you account for something, you can adjust those contracts. But those contracts are usually adjusted over time. There's a lag in them. So it's not perfectly efficient adjustments that occur, and that's why it does become a very important point for a lot of people who have the shorter-term perspectives on it.

On the second point, you know, it depends. I don't have a clear answer to the second point.

MR. GLASSMAN: We probably have time for--yeah?

MR. : [inaudible].

MR. GLASSMAN: Actually, David, why don't you come down here.

MR. : The thing that concerns me is the amount of money that is actually taken out of the company when the stocks are issued, not what the theoretical value is. But if you looked at [inaudible] compensation over a long period of time, it was helpful to understand, you know, or as an investor [inaudible] how much was he adding to the bottom line. You and I could disagree on whether that was a value or not, but [inaudible] quantifiable figure, and you could do that for all people. So, I mean, that really is what the investment community wants to say, a way that sort of [inaudible].

The other thing that I haven't heard anybody say anything about and I don't know whether [inaudible] make a distinction between [inaudible]. Neither of you have mentioned that. But ISOs are the model of where we're headed [inaudible] options that are issued primarily to [inaudible], large numbers of people, the difference being that the tax on the ISOs are paid up front [inaudible].

MR. : So let me first clarify the tax issues. Incentive stock options are preferable to the employees, but non-quals from a tax perspective are preferable to the firms. The firms only get the tax deduction if they issue non-qualified options, which is why firms always prefer to issue the non-qualified options, and those are the dominant options out there. The employees have to negotiate hard to get the incentive stock options. So that's the first point.

The second point I'd like to point out is the tax deduction for the corporations occurs upon exercise. The expense for accounting purposes occurs upon grant. So there's a different date of when you get the tax deduction versus when it is expensed under the new proposed regulations coming out for accounting purposes.

And some folks would argue maybe the tax folks got it right, let's just wait until it all unwinds--which is what you were saying. Let's just wait until it all unwinds and we see exactly what the expense is, and we'll record it upon exercise, which is what they do for the non-qualified options on tax purposes.

The FASB doesn't want to do that because of timing, matching principle. If the employees are working today, let's recognize the expense today instead of waiting for the future to recognize the expense. But, still, tax folks like myself--I do a lot of tax work and work with the Treasury Department--kind of like the tax approach ourselves to some degree.

MR. : I know we're running late. I just will be very quick.

I want to say first, reiterate that the tax issue, I think, leads you in the direction that Kevin was talking about. I would say the academic literature that I believe on stock options and my own experience as chairman of the board of a publicly traded firm that uses stock options emphasizes that stock options probably are overused, and overused in the sense of being used too much for non-senior employee compensation, where they don't have the incentive effects, but where they're used. There are either corporate governance or tax explanations for that overuse, and I think that is a significant issue that we should deal with. And we can deal with it with the tax policy.

The second thing, very quickly--and this is important. I had left it out. We have a different accounting treatment of how we deal with private equity profits that we should think about using as a way to deal with this stock option problem that gets us back to the ex post numbers that you want. As Deen said, you deal with the stock options only ex ante. Whatever their ex post value is doesn't affect--if you subtract them from earnings, you don't get to put the earnings back in afterwards if the ex post is different. That's completely different from what we do with private equity. So if Chase--every year Chase will book--with its private equity firms in its private equity subsidiary of Chase, it books using market multiples what it expects its exit profit to be on its private equity firms. And that goes to the bottom line, okay, based on industry market multiples. Then when it does the IPOs for those firms, the ex post may be different from what they had booked. They correct. And so the earnings get booked at the time based on estimates. Then they get corrected if the ex post is different from the ex ante.

Okay. Now, here's my big question. Why can't we do that for stock options if we're so worried about this? Let the firms book them any way they want and make the corrections based on ex post. You could have that. That would be another approach. I haven't heard anybody mention it. We're already doing that for private equity where we have a similar sort of issue.

MR. GLASSMAN: You'd still have a valuation problem in the ex ante part.

MR. : I'm not suggesting that that's the right--

MR. GLASSMAN: No, no. I know.

MR. : We are already doing this--

MR. GLASSMAN: Right.

We're going to go straight to the next panel because we don't have a great deal of time. We will end at exactly 1 o'clock. You didn't know you were going to have lunch, but you are, if you choose.

Please come up here. I want you to sit here. Could the discussants come up? Commissioner Atkins and Professor Benston.

Kevin Hassett?

MR. HASSETT: I'm still a little stunned because I've been at AEI for six years, and that's the first time that Martin ever said anything I said was a good idea. So you'll excuse me if I'm a little rattled. I guess I have to reconsider my tax piece in the Wall Street Journal a few months ago.

Peter and I have a paper before you where we talk about the economic and legal consequences of requiring the expensing of employee stock options without specifying the valuation method. And what we're going to do is we're going to break our presentation up into two parts. In the first part, what I'm really going to do is rewind and go over a little bit of what Charlie was talking about a little bit so that I can hopefully make the problem quite intuitive that concerns us, and I'll be drawing on the fact that if you're unfortunate enough to ever have seen my dissertation, it was actually on something called nonlinear time series models, and that's kind of the frontier area of studying valuation methods, and it's relying on some very unusual and interesting things that are required to value options. And I want to talk about what those are and hopefully make it intuitive.

I'm going to drawn on an example that I used at an options conference here about a year ago to start with, which is let's think about what the option problem is for a minute. And the way I like to think about the option problem is suppose that we were gamblers and that we were observing the flight of a butterfly in the room, and suppose just to make things simple, the butterfly is going from my left to right, your right to left. And then suppose that we were going to make a bet, and let's specify the bet this way. I'll make it a very, very simple one. Let's say if the butterfly hits the ceiling, I'll give you $10. How much would--and then we're going to auction that off. So right now if there were a butterfly in this room, we could say, okay, if the butterfly hits the ceiling, I'll give you $10. Who makes a bid? Who will pay something for that? And then somebody would say, well, you know, I'll pay a penny, and somebody might say I'll pay ten cents, and then we would keep going until we arrived at the price.

And so the value of that sort of option or contingent financial instrument that pays if the thing hits the ceiling will depend on a lot of things, right? You'll observe the butterfly. Now, if you watch a butterfly fly, then what it does is it kind of--it beats its wings and it goes up a little, and then it stops for a second and goes down a little, and it beats its wings and it goes up a little, and then it stops and goes down a little. And so if you go very, very close with a microscope to their flight, then they're kind of going like this. And if you saw a butterfly that was going kind of like this, so when he was going up, he was going up fast, when he was dropping, he was dropping fast, then you might think, well, he's got enough strength in his wings that he might really hit that ceiling if we let it run for an hour or so. And if you have this little butterfly that's just almost like gliding and barely moving off of this path at all, then you'd think, oh, well, there's no way that guy's ever going to hit the ceiling.

Now, what we do in options modeling is we observe that the share price moves through time, and the share price is--you could think of it as being kind of like the flight of the butterfly. It has a trajectory, and it's kind of going up and down along the way. And then what we do is we fit models to history, so we look at how the prices changed over time in the past, and then use those models to guess about how they might change in the future.

Now, the Black-Scholes model assumes, as Charlie said, a geometric Brownian motion which is basically a continuous time version of the simplest model that everyone's been introduced to, which is just a random walk, potentially with drift. And so the idea would be that if there's a share price today that's $10, then the geometric Brownian motion model fit to that over history might say something like, well, where it's going to evolve in the future is each day its price is going to go up a penny--that's the drift part--so on average if you look out, you know, a year ahead of time, maybe a penny is too much, then it's higher. But in addition to doing that, underlying going up a penny a day, then we're going to draw an error term from a normal distribution that has a mean of zero and some variance, and then we're going to add that to the price, too. And that's what the Brownian motion is.

The problem with that method is that that model is a very bad model, it turns out, for stock prices, for the reasons that Charlie mentioned. It just turns around that while that's very convenient mathematically, it's convenient mathematically because its log linear, for people who remember their math, and everything cancels since they end up with a nice simple closed form. But it turns out that over time there's lots of things that are wrong with that model. A lot of times in particular there are periods when stocks are really variable and not really variable. And so if you have a time when a stock is really, really variable and then you fit your model to it, then you might say, okay, well, that means that the butterfly is really bopping up and down really, really fast. And so the odds of that butterfly hitting the ceiling are pretty darn high, so I'd pay a lot for that. And then there are other times when you could look at the same stock, and you find that the thing is relatively flat. And then you say, well, there's no way it's ever going to hit the ceiling, that thing is worth zero.

Now, Deen talked about how people might choose the time period in order to optimize the value or minimize the value. The problem is that the finance--the academic empirical finance literature has shown that this property of the changing variance over time is absolutely everywhere. It's absolutely everywhere. There are other problems, too, but it's absolutely everywhere. So the fact is that there isn't like one variant that you can fit to a stock and say, okay, well, that's what the answer is, that's going to be the best guess going forward. It's changing all the time. And exactly how we go about measuring that changing variance--and Charlie used the words ARCH or GARCH. The ARCH and GARCH mean auto regressive conditional heteroscedasticity--Charlie's children know that, but you guys might not--and the G is generalized. But there's a great debate about exactly how to do this.

My own guess is that what we're going to do is we're going to convert to using non-parametric methods where we don't actually say it's a normal distribution, we don't actually say exactly how the variance goes over time, but we approximate this thing with these new nonlinear methods that the best time series people are doing.

But the fact is that for any specific firm, we know that the Brownian motion is wrong; we just don't know how it's wrong. And exactly what we might do about it if we were, say, in a Wall Street firm, for instance, trying to value options, would depend on, you know, a debate between us. And so to make this clear, what I want to do at the start now is just look at a couple of charts of stocks and talk about what I might do if I were going to try to value their options to give you an idea of the flavor of what could potentially go wrong.

Here's a stock that to me actually looks kind of like a Brownian motion, and that's why I put it up. Strangely enough, so one of the best stocks for this, and what I did is I asked my people to print out 30 slides, and then I just picked the ones that I thought were visually interesting for this purpose.

And so if you look at Amazon, it does kind of look like it, doesn't it, like there's this butterfly, it's kind of drifting up--it's kind of drifting up over time. And it's going up or down kind of randomly, but the degree to which it's going up or down randomly doesn't seem to change a whole lot over time, right? And so my guess is that for Amazon, you know, it might be that we'd want to think a little bit about what our estimate of the trend is, because you can see starting around January 2002, maybe, or June, it starts on an up trend, and we want to let that continue going forward.

Now, we might have a little debate about how to fit the parameters, but we might be able to fit a Brownian motion to Amazon, and it would not be the stupidest thing anyone ever did.

But let's look at Apple. Apple is really, really interesting because you can really see the volatility problem and the ARCH problem in Apple, because if we start here sort of June 2002 and then run for about a year forward, you can see that it was kind of like a no-trend--if we fit a Brownian motion there, we'd get no trend, and a relatively small daily variance, kind of just going like this. And then all of a sudden, we get something that Charlie referred to in the literature, which is like a jump. We get a jump from down around 13 to say maybe up around 20, but after the jump, because this jump happens, it puts the Apple stock into this place where it hadn't been before, all of a sudden the market's really nervous about the jump, and there's much, much higher volatility.

And so if you estimated the value of the Apple options using say the year that started--that ended in March 2003 or so, then you'd say that they were probably not worth very much because here's this thing that has no up trend and the variance is very, very low. But then, you know, what happened subsequently is that it got an up trend and it got a huge variance, and so all of a sudden the options are going to be worth a lot more. If we just fit a Brownian motion--or try to do just Black-Scholes to these Apple options and say, okay, let's start here and just use history's variance to tell us what to do and let's start here and let's start here, I can tell you that with a seven-year option, the latest experience would probably increase the value of those options by maybe a factor of 10, given the way these models work. Okay? So that's how far off the Apple thing's going to be.

Now, it might have seemed reasonable back here to just use Brownian motion, but then if we look behind that nice smooth year, you can see that there were some other jumps in the opposite direction that along the way were prefaced by some extreme volatility as well. And so I don't really know what model I would want to fit here, but I think what I would want to do is go to the most general nonlinear time series model I could, like a neuro-met(ph) or something like that, to do it if I were going to try to value these things.

I'd just put up Pepsi, too, because they have the same kind of look. So if you go up to the middle of 2002, it looks very much like a nice, smooth, up-trending Brownian motion, just like that Amazon chart that I showed you. And then all of a sudden there was some really bad news. I don't even know what it was. I don't follow the stock. And then after that it was just like the market had too much caffeine or something; it was, you know, absolutely flying all over the place and the volatility increased dramatically. And so the Brownian motion is clearly not going to be a very good description of this thing. Certainly if it's based on recent historical estimates, you know, in 2002, you'd get one thing, and now you'd get something else altogether. I don't even know that the option value for this chart--I couldn't tell you necessarily which way it would go because while it went down a lot, it's kind of still got a little bit of an up trend now and the volatility seems much higher. And it's the volatility, how fast the butterfly's bopping up and down, that can really have a big effect on the options.

And so what I'd like to conclude with is just that, you know, economists are arrogant creatures, and we like to think we can solve any problem. And it's not that we don't think that we could guess what the value of an option is. It's just that if you were to take us all and lock us in a room by ourselves and let us look at the data and say, hey, which model are you going to use for this stock, which model are you going to use for that stock, then before we could define any model, we'd have to decide what the data-generating process is. Is it going to be the simple Brownian motion, or is it going to be just like non-parametric completely? Or are we going to have a simple ARCH process? And to decide that, we might look at the data and say, well, what does it look like is generating that data? And we'd probably all come up with different answers, and then after the fact, some of us would be right and some of us would not.

So it's not like the options problem is one where there's one thing, one model, everybody agrees that's the way to do it. It's more--as Deen said, it's kind of an art. And so what Peter and I did is we started talking about this, and we said, well, given that that's the way it is, given that really, you know--it's not that we couldn't find an answer for any company, it's just that we can't find one answer that applies to all companies. Then how does that affect the accounting policy choice that ought to be made? And that's the topic of our paper after we discuss this stuff, and that's the part that Peter's now going to give you.

MR. WALLISON: Meanwhile, I just want to emphasize something that Kevin just said. It's not that we can't find an answer that would fit individual companies, but that we can't find an answer that fits all companies. And this is the problem that FASB and its predecessor have had for decades now, and FASB concedes now that there really isn't an answer for all companies; therefore, essentially do what you want to come up with--

MR. HASSETT: And also, I also want to make it clear that I don't necessarily believe that my answer's correct in any way. So Deen really captured it precisely when he said that it's an art form and that there's probably, you know, some shop on Wall Street where there are the best artists and they're making the most money right now because they're the ones who can really figure this out better than everybody else, and they're probably using supercomputers to fit the processes and so on.

MR. GLASSMAN: Peter Wallison?

MR. WALLISON: Okay, thanks. I've asked my assistant over here to handle the slides. Would you put up slide two, please?

This is the key question that this paper is addressing. Does it make sense to require the expensing of employee stock options when no one knows how to do it? This is a Washington audience, so I'm going to talk a little bit about Sarbanes-Oxley because, to me, this is one of the most painful experiences I've observed in Congress.

The Calomiris and Hubbard paper really reminds us that one of the most serious errors of Sarbanes-Oxley was its unblinking acceptance of the idea that audited GAAP financial statements are the key financial disclosures of companies. In reality, they are not particularly useful now, as we heard from Charlie, and they are becoming increasingly useless. That's one of the amazing things about this as companies rely increasingly on intangible assets to generate revenues, and these are very, very hard to value.

So the whole question of whether it makes sense to expense employee stock options has to be considered within what I guess I would call a parallel universe, one in which the GAAP financial statements are really important. In this universe, which consists of Congress, which doesn't know better, FASB, which is the keeper of the flame of a consistent set of rules, the SEC, which is the necessary follower of what Congress wants, and the media, which is chasing readers and viewers who have never been told about cash flow analysis, but this universe does not consist of financial analysts, professors of finance, sophisticated investors, or even many accountants who have been worried for many years about the increasing irrelevance of regular GAAP accounting.

It's important within this universe--and it's a very narrow one, but within this universe to place a value on employee stock options. This is an inside-the-Beltway issue, and that is true because it deals with the political or meta reality and not with reality itself.

Our paper, Kevin's and mine, says that within this universe--and can I have slide--okay. Slide four? Yeah, I know. I'm not used to this. Yeah, that's right. There we are.

It says that even within this universe, taking it on its accounting merits, it is not good policy to require the expensing of employee stock options. There are many good reasons for not expensing options that have been part of the debate in Washington. First of all, there's a differential effect on small, growing companies. It reduces management incentives. It discourages companies from going public. That's because many companies go public because they want to be able to issue stock options to encourage, incentivize their managements. Otherwise, it's a very expensive process. If we discourage stock options, we'll have fewer companies going public. We already will because of the enormous expenses that are going to be caused to companies by Sarbanes-Oxley, but in addition, if we eliminate stock options, we'll reduce the incentive for small companies going public.

However, this paper will, as I suggested before, deal only with the accounting questions, and it fails there, too.

The principal reason that expensing stock options is not good accounting policy--this is slide five--is that there is no way, as we heard from Charles Calomiris, there is no way to place a consistent value on employee stock options. As a result, while everyone can agree that these options have some value, there is no consistent or reliable way to estimate what that value actually is. Under these circumstances, including an estimate of the value in earnings per share, assuming, again, that earnings per share have any real value, will distort that number. As Calomiris and Hubbard concluded, there can be many different outcomes to the valuation question, none of them necessarily correct.

Can I have the next slide?

Now, from an accounting standpoint, this raises a very fundamental question of whether there is any limit to the use of what is called fair value accounting. From its inception, accounting has relied on real numbers, transactions, costs, market prices, to establish values. In recent years, as more and more productive assets were intangible assets, the cost of assets has become less and less relevant. In recent years, the market value of the S&P 500 has been six times the book value. Back in the early 1970s, it was about one to one. It's now about six times book value, which means that book values have fallen way behind other values, which many economists believe and accountants believe are the intangible values that aren't actually being measured well.

Accounting theorists under these circumstances have encouraged companies to value assets at fair value so that their balance sheets would bear a closer relationship to their actual market value. Fair value was usually defined as what a willing buyer and a willing seller would pay for an asset, the price at which they would transact. But when there's no market value--and in the case of employee stock options, there really is no market value--it is permissible to estimate a fair value.

Unfortunately, as shown by the Enron case, leaving this judgment to management and auditors can have serious adverse effects. Part of the fraud in Enron consisted of management's wildly distorted fair value, as they thought they were producing, their fair value assessment of some of the company's contracts. Thus, the key question for the FASB and for policymakers generally is whether there are any applicable standards for the use of fair value accounting where, in other words, we should draw the line when using fair value accounting.

Next slide.

Fair value accounting is the underlying idea behind placing a value on employee stock options. Most of those who have argued that employee stock options should be expensed probably believed at the time they were making the argument that it was possible to use a formula or a model of some kind to establish this value. I think that idea has been pretty much demolished by the Calomiris-Hubbard paper.

The Black-Scholes model was frequently used for this purpose, and, in fact, in almost all corporate footnotes, that's what they used to establish that value just in the footnotes. But if it had been true, if the Black-Scholes model had actually worked to value employee stock options so that people had some confidence in it, then using it to establish an estimated fair value would have been possible. However, as we have now seen, Black-Scholes doesn't work, and I suggest that what the Calomiris and Hubbard paper also suggest--and Kevin was saying essentially the same thing--is that there is today no known options pricing model that will work.

In September of 2003--the next slide--FASB seemed to recognize this problem. It reaffirmed its intention to expense stock options. One of the reasons, I think, for that is that they are now under tremendous pressure since Enron, for reasons that are not entirely clear to me, to do so, but it withdrew its endorsement of the Black-Scholes and something else called the binomial model as a way of making the estimate. Companies were advised that--and I'll quote here because the language is important--"The use of any specific option pricing model would not be precluded." In other words, companies were not given any guidance about what would be a suitable model. If FASB cannot come up with a suitable model, it seems unlikely that companies will be able to, and thus, they will choose a model that is most favorable to them and, of course, as we have seen, the result of that will be a manipulation of that bottom-line earnings per share number.

So the question comes down to this: Does it make sense, as I said before, to require a fair value estimate of the expense associated with employee stock options when there is no agreed-upon way of making such a valuation? The answer, I think, in our paper is no. Allowing companies to choose a complex model that will produce different outcomes, depending on the assumptions that are used, will be inconsistent with three key accounting goals: reliability, consistency, and comparability. The results will be unreliable for the reasons you have already heard. There are too many variables, the completely uncertain results. The results will produce inconsistency also over periods because the technology in this area is changing. There will be better models or what some people believe are better models. Or ex post events will cause companies to change their models, and the result of that will be to change the consistency that a company's accounts have shown over a period of years. And, finally, it will reduce comparability because companies in the same industry, in the same business that should be compared by investors will be using different models, and as a result, it will be very difficult to compare their earnings-per-share results, assuming--always, assuming again that there is any value in the earnings per share.

The next slide, I think, Kevin.

What are the arguments, then, in favor? There really is only one that at least has been made in a public policy context by people from FASB. First, there is no doubt that employee stock options have some value, and as they repeat, the value is not zero. The argument is made if the value is not zero, failing to account for this value ignores part of the real costs that a company is incurring in producing its earnings. So proponents of expensing can argue that from the standpoint of the accounting concept of conservatism, it is better to reduce earnings per share by some amount, no matter how uncertain, than to leave that number blank.

These arguments, it seems to me, are very thin when weighed against the damage expensing would do to the credibility of financial statements. First, as I mentioned, reliability would be sacrificed. Consistency would be sacrificed. Comparability would be sacrificed. And I'd add one other thing that's a much more general issue for accountants, and that is, there would be an encouragement of speculative fair value measurement. After Enron, what we ought to be looking at is a way of cabining and restricting the use of fair value rather than encouraging the use of fair value by such things as these unproven and unworkable models.

I would add one other point, and that is, there would be significant additional litigation costs. Without an agreed options pricing model of some kind, companies would be sitting ducks for the plaintiff's bar. It's very easy to see how this would happen because results will vary tremendously from what the company ex ante predicted to what, as the economists say, ex post actually occurs. And as a result, the companies will be accused of having chosen a model which manipulated their earnings, simple, a very easy thing for the plaintiff's bar to do. Earnings will end up being overstated in many, many cases, and no company could ever be sure when it chooses the model, even if it did it honestly, that that model was accurately going to predict a value for other reasons Kevin outlined and Charles and Glenn Hubbard in their paper.

So what is the recommendation we would make as a public policy matter? It seems to us that the sensible result would be for FASB to defer requiring expensing of employee stock options until it or others have developed a satisfactory options pricing model for employee stock options.

Thank you.

[Applause.]

MR. GLASSMAN: Thank you, Peter

Our discussants, first, SEC Commissioner Paul Atkins. Is that okay? Or had you planned to go--

[Inaudible comment.]

MR. GLASSMAN: Sure.

MR. ATKINS: Thanks, Jim. Thank you for having me here today. It's a pleasure to be here to talk about this important topic and also Kevin and Peter's interesting and thought-provoking paper.

First I have to give the standard disclosure that all the compliance folks back at my shop want me to make, and that is, everything I say today only reflects my own opinion and doesn't reflect that of my fellow Commissioners or the SEC as a whole.

So much has already been aid today, but I do believe that a threshold issue that needs to be addressed and answered before we get into the details of stock option expensing is what exactly is the problem that we're seeking to address. You know, is there a significant investor outrage or concern out there about this issue? And are there cadres of investors who are clamoring that they don't understand the footnotes disclosure regarding the effect that options can have on a company? And is there a fear that the current disclosure method provides unclear or unreliable information about the dilutive nature of stock options?

My own fear about where this is going is that FASB is basically getting into an area that's more of a political issue than a technical or accounting issue. My fear is that the pressure on this issue is meant to address a corporate governance failure on the part of boards to rein in executive compensation.

I also fear that this change is coming about not simply to improve accounting in the U.S. or to provide more reliable financial information to investors. My fear is that this change is coming about as part of a basic horse trade in order to facilitate international convergence of anything standards. Convergence of accounting standards, of course, is a laudable goal, and I have said publicly before--and I think my colleagues at the Commission have said the same--that it's something that we ought to strive for. But in an effort to reach this goal, we cannot sacrifice the integrity and reliability of our financial statements.

The integrity and reliability of our financial statements should always be FASB's guiding principle in drafting changes to financial standards. So the real question is whether expensing stock options will enhance that integrity and reliability or do the opposite.

One thing that I am certain of is that FASB should not be in the business of dictating what type of compensation should be paid by corporations to their employees. So I hope that this new stock option approach should not be an attempt to dissuade companies from using options as a means of compensation.

First I have to say that I am not an accountant, but to me stock options feel much more like a balance sheet issue and item than they do an income statement item. They have to do with the ownership of a corporation, not with the outlay of cash or cash equivalent. So, in my opinion, stockholders should be primarily concerned with overhang of unexercised options. Should we not be more concerned, then, with a denominator of the earnings per share number to reflect this overhang rather than with the numerator and by forcing an inherently imperfect value into that?

And so separately shareholders should be concerned with the governance issue in the aspects of captive boards giving senior officers huge option rewards as we've seen in the past, even repricing them if they're under water. So this is not a financial statement issue, but it is a corporate governance issue. And we at the SEC are taking steps in that regard, and that's an issue for yet another conference at some other time, so I won't get into that.

I recognize that options are typically granted as a form of compensation. And compensation is generally an expense that must be reflected on income statements. Therefore, assuming that options are an expense that could in theory appear on the income statement, I think they have to be valued property. I also have sympathy for the argument that says, Okay, if the options have no value, then just give them to me, you know, if you don't want them. But what's that value and how can that be determined? And I think that's where the rubber really meets the road, as has been discussed earlier. And as I see it, companies are basically left with two options for a lot of the reasons that Peter just talked about, Black-Scholes and the binomial method. And from my latest information, it seems to me that FASB is heading toward requiring the use of one of these two things. But we'll have to see what finally their exposure draft will say later this month or next month. I think right now it pretty much is in flux.

I have yet to meet anybody who suggests that Black-Scholes is a good or even a fairly good indicator of the value of long-term compensation options, especially those in broad-based option plans. There are too many idiosyncratic aspects of the options as we saw when J.P. Morgan Chase bought the Microsoft options. They had to drastically modify the terms of the options to be able to put some value on them that they were willing to pay and to be able to basically reflect Black-Scholes.

The only positive comment that I've heard about Black-Scholes is that everyone seems to understand how to implement it versus the binomial method. But let me be the first to admit that I don't understand the intricacies of the binomial method, and much less, you know, some of the INS and outs of the theory behind Black-Scholes. It's complicated, and as far as the binomial method goes, it has lots of data points. It will take a lot of effort and expense to implement by companies. And I also understand that ultimately it seems to produce results that are strikingly similar to Black-Scholes.

Accounting professionals and FASB readily acknowledge that both of these methods are not perfect and, frankly, are far from it. But they argue that since options are clearly an expense, putting some value for them on the income statement is better than putting no value for them at all. So, my question is: Should we say that our ultimate guiding principle and that of FASB's generally accepted accounting principles is that close enough is good enough for government work or for accounting work? As Peter and Kevin point out in their paper, there are many benefits that employees receive from their employers that we do not expense on the income statement, largely because we cannot value them appropriately. And stock options, I think in some cases, fall into this category.

Recently, I attended a conference in London, and one of the speakers was a lifelong accountant who noted a disturbing trend. When this accountant began his career back in the early 1970s, he said that virtually everything on the financial statements could be traced back to some hard piece of data, usually a piece of paper that reflected historical costs or fair valuation of some comparable arm's-length transaction. But this is no longer the case. The trend now is even beyond getting fair value for everything. Accountants are delving more and more into subjective valuation such as how to value intellectual property or even employee morale as a value-driver for corporations and future prospects. So the fair value concept I think is the right fundamental approach, but I think as Peter and Kevin have said, generally speaking, there are a lot of inherent dangers in it. If there's an asset on the balance sheet that clearly does not any longer reflect its historical cost as an appropriate current value, I think it ought to be, generally speaking, restated.

That's why we had the wave of acquisitions and buyouts in the 1980s, because financial statements and the market value of a stock of those companies no longer reflected reality. But, in practice, fair valuation, as you heard, is rife with difficulties and possibilities for manipulation. Someone has to perform the fair value calculation, and this creates opportunities for mischief.

As Peter and Kevin accurately point out in their paper, much of the fraud at Enron was caused by an abuse of the fair value concept. Putting a fair value on something as complicated as long-term stock options is almost an impossible task, I think. I should note here, too, that the fair value concept is getting much more attention in the mutual fund area as well. Our mutual fund laws require that a fund determine the fair value of its shares to stockholders for the purposes of purchase or redemption of the fund's shares. However, we now are discovering that people are exploiting pricing abnormalities inherent in mutual funds. And there is a greater push to have more so-called fair valuation of mutual fund holdings on a more frequent basis. Again, in concept, this is a great idea. But I worry about whether it will create more opportunities for problems down the road.

I'm also finally concerned that we might be drastically narrowing even further the class of people that can understand important parts of income statements. For example, having spent a few years of my own life at a national accounting firm, I can tell you that the universe of people that understand derivative accounting, which is called FAS 123 in accounting parlance, is small, very small. Probably literally two people at each one of the big four firms, and that's a total of eight. And so there are people in the field that have to call back to these "gremlins" back at the national office to try to figure out how they should account for things in the field. So I fear that adding another complication could further narrow the universe of knowledgeable people even more. It will create even more of what I like to call the priesthood of the PNL at these accounting firms.

So, anyway, I'm primarily here to ask questions rather than answer them. And much of my comments here require us to look back at the original threshold question. That is, what is the problem that we're trying to resolve? And does FASB's fix address this problem? I'm not sure that the presented fix doesn't create more problems than it actually solves.

Thank you very much.

[Applause.]

MR. GLASSMAN: Mr. Commissioner, let me just ask you one question before Professor Benston starts. You were, I think, largely responsible for the individual investor program that was started under Chairman Levitt at the SEC, including the town meetings, which I think were just fabulous. I think in many ways he was a great chairman, even though certain things he did I didn't agree with. But that part of it was terrific. So you know about small investors.

Do you think that making the kinds of changes that FASB wants to make will be beneficial or detrimental to these kinds of investors that came to the town meetings that you would organize?

MR. ATKINS: Well, I'd even take a further step back. Even before my time under Levitt, I was also chief of staff under Chairman Breeden, and we back in 1992--that's the origin of the disclosures that we have in the footnotes to the financial statements, because we required companies to at least disclose what these options were and describe, you know, at least a nominal value and what the potential dilutive effect would be.

So I think in those cases we took a step towards trying to get the information out to individual investors, and that was at the time, I thought, very positively received by people. Obviously, there were a lot of people who thought that they ought to be expensed. But for the most part, it seemed like that met the need in the marketplace for people to have that information.

You know, I have to honestly say, at the town meetings that I went to, I don't think that was a big issue that people asked about. I think there is much more concern--and you've heard it here from some people--about the effect of huge grants of options to the top management of companies. Again, I think that--you know, obviously I can't say generally whether that's good or bad or not. I think each--you have to address each company separately. But that in a sense is a governance concern, and I think that goes back to the role of shareholders vis-a-vis the board, the role of the board in asking proper questions, and then reflecting the desires of the shareholders.

MR. GLASSMAN: Just to go back to what you said in the beginning, governance concerns really are not accounting concerns. Those are two different things.

MR. ATKINS: Two drastically different things.

MR. GLASSMAN: We're going to run a little bit over. We want to hear from the person that many consider the top expert on this issue, from Emory University, certainly one of the top experts if not the top, and so he will go a little bit over. And then we will all have a few questions and have lunch.

Professor Benston?

MR. BENSTON: Well, standing between people and lunch is not a good idea.

First of all, as you can see, I'm from Emory University, and my remarks should not be construed of those of the Coca-Cola Company.

[Laughter.]

MR. BENSTON: I have to make that clear.

I'll go over very briefly what the papers have--of course, I didn't know what they would do, and as you can see from this, I won't read it because you've been here and you've heard the paper. They find it's not satisfactory to value them. It's inconsistent with principles of accounting, great legal risks, and fair value accounting is inappropriate. And the Calomiris-Hubbard, they have the same sort of idea, but mostly they say accounting doesn't mean anything anyway, so who cares? It's (?) then I'm also a finance professor, and I can say I agree with Charlie completely. It's true that you have to look at cash flows. And, by the way, any of you who look at cash flows and think that we accountants can't make those numbers be what we want them to be, you're making a big mistake. We can.

But, anyway, as they say, no uniformly satisfactory method. The cost is firm-specific, and consequently disclosure should not be required. Okay, you've already heard that.

Okay. What is the criticism? They both criticize the Black-Scholes model, and I have to say for Myron, I think we'll have to take his Nobel Prize away. It's quite clear that he got it under deception. So after this is over, I'll have to call him up and tell him it's gone.

One of the basic things is we have all of these things that you see up here, and, again, I won't repeat them to save time and so we won't stop you from lunching. But the asset returns, the (?) generating process is uncertain, negative correlation. The table, by the way, Charlie, is very unclear. I know it's a draft, but you shouldn't--some of your points should be--

[Inaudible comment.]

MR. BENSTON: I know, but I hope that they explained what their symbols meant. In any event, I think some of the things that Kevin put up were very useful, and I would suggest that both papers that the idea of simply saying Black-Sch