Economics and Finance Symposium - Finance in Action

Search transcript...

MYERS: This is a panel of practitioners who are also in all cases, I guess, with the possible exception of Judy, distinguished academics. But Judy more than made up for her impact on the pharmaceutical industry and finance and practice. Ed Roberts asked about entrepreneurs and finance and these people are them. They are the entrepreneurs in finance, perhaps more so than us academics that were up there before.

I would like to introduce each of them very briefly and then ask in alphabetical order, each of them to come up and make some comments about the topic of this panel, which is finance and action. They were in the middle of the action, they've applied finance, they've done it exceptionally well, and I think they are in a wonderful position to think about where we go next with finance in practice.

So in alphabetical order, our analysts are Doug Breeden who is now the [? priest ?] Professor of Finance at Duke and formally dean there and founder of Smith, Breeden & Associates, which was pure financial entrepreneurship, I guess I would say.

Gene Flood, who will follow him, is now the CEO of Smith Breeden. Gene taught at Stanford. He was an MIT PhD and was also at Morgan Stanley and other places, a long and distinguished career. Following him Bob Jarrow, who is a doctoral student from Sloan Finance Group in 1979. He's the Lynch professor-- what did I say.

Oh, there you go, Ben. I'm sorry. I missed Ben. So Ben is one of the founders and the chief risk officer of BlackRock, also Sloan, PhD, and has made a very distinguished career on Wall Street. And I think it's going to be particularly helpful in talking about risk management and risk management change going forward.

Now, I will get to Bob Jarrow, I'm sorry, who's the priest professor of finance at Cornell and a widely published academic in the general areas of derivatives, risk management, fixed income, and has a lot of practical experience and impact. And finally, Julie Lewent, who was a student of Bob Berton's and mine back in-- I won't say how long ago, but back in the 70s. And I think the thing that always stands out with me about Judy is that she was the person who brought modern finance to the pharmaceutical industry. And at first you might think that that's not an easy place to apply modern finance, but actually it is and she showed how to do it.

So with that, and with apologies for not introducing each of you with more detail, I'm going to turn it over Doug Breed and then we'll just go in sequence.

BREEDEN: Thanks to such an honor to be here with all my former teachers and people like that. I came here first 42 years ago-- it's hard to believe-- and every person on that panel was either a teacher or a colleague of mine at one time or another. I had Bob Merton for first year finance, 15-4012, which must have been his first or second teaching of that course and had the blue mimeograph sheets that many people know.

And I had Myron Scholes for financial institutions. I had Franco Modigliani for capital markets and then I actually took statistical methods from Rob Engle who also won a Nobel and I did a forecasting model for the egg futures market in that because I grew up in the chicken business. And I got an A on it, so we really thought we knew what we were doing. And I went out and traded and I ended up taking a $2,000 of my wife's hard-earned money and turning it into a $20,000 loan from my Uncle Arch.

So that was the beginning of my finance career and I taught futures later on, so I use all these experiences in teaching and I'm a great teacher because I have so many of these experiences that I've learned from. So anyway I thought I would try to actually-- I could tell stories on every one of these and would love to do it and flatter them all by the stories, but I thought maybe I would show you actually some of the challenges of applying these theories that are such great theories.

And what I've found is that many very straightforward securities or if they sound straightforward, have risks that are really hard to estimate. And some of the ones that I've worked on, I've worked in the mortgage market. I basically took what Myron Scholes and Bob Merton and Fischer Black and Steve Ross and John Cox-- Cox was on my dissertation committee at Stanford for the PhD and taught me really what Merton was trying to tell me. And I didn't understand it then. I was an MIT undergrad, Bob. You were kind of over my head, you know. It took me a little while to catch up with that.

But I basically was one of the initial people to apply their option pricing models to the interest rate prepayment option in mortgages and that's what we found at Smith Breeden Associates to do and I published articles there. So you'll see some mortgage themes there, mortgages and interest-only strips, asset-backed securities, which were the hot item in the last crisis here that pretty much damaged me in a big way. So I have some more learning experiences there and corporate bonds as well.

So on basic mortgages, let me illustrate that with mortgages, as interest rates go up, prepayments go down. The cash flows then come later and so you have longer durations. So you have changing risks and I actually estimate those risks. And this is like the beta or the sensitivity of mortgage prices to a change in interest rates of 1%.

And what you see, if you look at those, and this is using quarterly estimates from daily data of what the risk is every quarter on a standard fixed rate mortgage, it goes from a 1 year duration, which means that interest rates go up 100 basis points, they go down in price by 1%, to a 6-year duration where if interest rates go up 100 basis points, they go down by 6%. So risks are dramatically changing on these simple securities as interest rate risk change.

In fact, the theme of my talk is, I'm sure you've read in the papers that we've had some challenges and we've lost some money, even smart people, trying to apply these models. And so a lot of what I'm trying to show to you is why it's so difficult to apply these models and these changing risks as interest rate change are really hard to estimate.

And so in a mortgage market, you find that prepayments really matter, how optimally people prepay, how fast they prepay, and all the factors. And we've learned that there are a lot of different factors. And I used to study this in the 1990s. I started Smith Breeden in the '80s, so it's 28 years old. I've been doing this a long time.

But I updated my presentation and I asked for some recent data and I got this thing from Barclays Capital, Barclays bought Lehman. And I have great respect for the researchers at these, so I'm not saying anything bad about any researchers at all in my talk. I view them as all very smart people, very likely students of MIT and Stanford and Chicago and so on and it's just a matter of how difficult it is to model these things.

But Barclays sent out on January 23 of '09 a prepayment model change. And I use this to illustrate. I've actually re-typed some of this in bigger font. And basically in their old prepayment model for 5% mortgages, they were forecasting prepayments of 74% a year. This was January of '09. And remember the Fed dropped long-term interest rates to 2% and short rates to 0% in December? So rates were just dropped down and that really stimulates prepayments.

But they basically had estimated that the repayments would be 74% a year and they found out actually, probably because people were not strong enough financially to refinance that the prepayments were coming in a lot slower. And so they revised their model in that middle column or row there, from 74% to 32% as a forecast and then the duration that they had forecast to be 0.4 years, less than half a year, expands to 2.3 years.

Well, if you're someone who bought that mortgage and is hedging it, it's like a quintupling of the risk estimate. That happened with this prepayment model. So I just imagine when you buy these securities, it's like you buy $100 million of these. You say, call your broker and have been short 200 futures contracts for 10-year bonds to hedge this risk. And then you get this prepayment model update and your researchers are saying, now the risk is actually five times what we told you it was yesterday. And you've got to call your broker back and say, oh, I didn't mean to sell 200, I need to sell 1,000 contracts instead of 200.

So I mean, these risks really change. And also if you used Black-Scholes-Merton pricing or a variance of that, you might think that you can really estimate those option costs really well and tightly and you see that under the old model, the estimated option cost was 76 basis points and under the new model it was 110. So it changes the duration estimates, the first derivative. It changes the second derivative, which are the option costs. And for the 6% mortgages, the option cost goes from minus 3, which meant that it actually had positive convexity instead of negative convexity, and that goes from minus 3 to plus 50.

So the option cost estimates are not estimated very well as well. And then in the last two columns, you've got the old models estimated profits on fours versus fives versus sixes. These are the spreads in basis points. And you see in the old model you said, well, gee, the 4's get 64 basis points of profit, the 5's get 48, the 6's get 41. I'll buy the 4's, let's say. They look the most profitable risk adjusted.

And then the new model comes out and it changes to where it's totally the opposite. The 6's now earn a lot more than the 4's.

So it changes duration. It changes convex city. The prepayment model changes the estimated profits. So these numbers are hard to estimate. That's my point here. And prepayments are so important we built increasingly sophisticated models that now allow us to price many more risks. In the first models that I helped build and was working on in the 1980s and '90s, you had just four factors. And then I went off and became a dean and things like that and while I was doing that, they brought in another 5 or 10 factors.

So we were kind of missing a lot of factors. So you've got model estimation risk that's quite huge in some of these basic securities. This is just a basic mortgage, simple mortgage. Now, if you go a little bit more complicated where you say, let's take a mortgage and we'll break it into two securities, one it entitles you to all the interest, our share of the interest and one entitles you to a share of the principal.

That's pretty easy to define. You buy A, you get the interest, you buy B, you get the principal. That's not too complicated, but the risks are complicated. And you can model the IO or the interest-only part or the principal-only part and those are the patterns that they have, but what I want to pay attention to is in the same prepayment revision, what happens to the estimated risks of an IO, an interest-only strip.

And if you look at that, on this first IO indicated in the third column there, the old duration was minus 84 years and if rates went up 1%, it should actually go up in value by 84%. And then they put in a new model and the new estimate instead of minus 84 is minus 48, a difference of like 30 years of duration. And the option cost turns from plus 178 to minus 654. It's like they can't even agree whether it's long options or short options and then the spread, the estimated profit, is, with the old model, you say you're losing 6 and 1/2% if you invest in this first security. And with the new model you say, oh I'm sorry, with our new model, you're making 12%.

So this is why smart people have a hard time making money with these securities. In one of my papers in the '90s, I collected forecasts of durations from different brokers, Goldman Sachs, JP Morgan, Solomon, Bear Stearns, DLJ, and I compared their duration estimates for them and they were all over the map. They were somewhat alike but very different, as you can see if you can actually read this well.

But basically researchers have a hard time estimating it. That was the of and the duration. This is the estimate of the option costs from the four different brokers and they're not very alike at all. They're very, very different.

So I guess that's the main point I want to make to such a smart MIT crowd is that even though we've got some of the best breakthroughs in option pricing and modeling, we have a lot of difficulty in applying that because it reminds me of the Gordon model for estimating the value of a stock. If you just perturb the discount rate a little bit, the underlying value that you come up with for the stock can be very different. If you change the interest rate from 9% to 8%, you get a very different stock price. Same thing true here for prepayments and so on.

So then you might say, so why not just ban these things because they're just too complicated? And the answer is that they are still very useful hedges because these IOs, interest-only strips, for example, just buying the interest piece, since it is so sensitive to prepayments, the value of that is a great hedge for certain types of prepayment risk. So it is still something where even though it's hard it's very useful for people like Gene Flood and Smith Breeden and BlackRock and different ones to have these out there.

And the prices of them, it's like mortgage servicing that Fannie Mae and Freddie Mac does, they get a fee that's related to how much interest they collect and so the value of that mortgage servicing is related to the value of IO strips. So by having those out there in the market, you can also use those to estimate the values of major franchises like Fannie Mae and Freddie Mac before they're a credit risk.

So these serve useful purposes but they're really hard to value. Well, I could go on there and actually with the most recent ones, the asset-backed securities, the previous ones that I looked at had no credit risk. If you add in credit risk, I mean, this is the growth of that asset-backed securities market that we've had in the last 10 years. These are the ones that included subprime mortgages and prime and different things like that. And you do tranches of these and sell pieces, sell the different slices of a mortgage.

And when you get to estimating the risk of those especially when housing prices do this, to the right side of this distribution is the old distribution, probability distribution, or frequency distribution of four quarter changes in housing prices. The two items on the left are 2008 and 2009 drops of 10% and 20%. So the standard deviation of the right side was 3% and we had a Six Sigma event at least on housing.

And in some markets if you look to different ones like Miami and Las Vegas and Phoenix, instead of dropping 30%, they dropped 50% and 60%. So these securities then just had delinquencies that just blasted off and were really hard to estimate. I'm sure you've seen that Moody's and Standard & Poor's, they rated these as AAA bonds. At present, those that were rated AAA bonds just in 2007, over 90% of these different asset-backed securities are now rated as junk bonds.

And it's more of the '07 group than the '04 because there was a change in credit quality and deterioration of credit quality, so that if you look at these different securities, if you look at the FICO scores, the credit scores on them, you see that not changing much at all from 2001 to 2007 for the typical pool. And actually, this is one where it's kind of a criticism of Wall Street, I would say, that they may say, OK, the customers are paying attention to FICO scores. Look at the FICO scores. They haven't changed much.

But if you look a little deeper and you say, on the new pools, what is the percentage of them that have loan-to-value ratios in excess of 90%-- very high loan-to-value ratios, which make them risky and more likely to default-- the percentage of those went up dramatically in '04, '05, '06, '07, so you can see the credit quality deterioration of what looks like the same securities. And the percent that had full documentation declined, which means that a lot of them did not have good documentation. These are sometimes called liar loans and it's easy to lie on these things. So credit quality deteriorated on that dimension as well and the percent of people paying just interest only went way up as well.

So the initial FICO graph that looks just like all is calm and nothing has changed from '01 to '07 is shown to be false with these other graphs as well. And the prices of these securities tumbled.

One last point that I'll end with here is that we owned actually 49 of these securities in one of the banks that we owned and I tracked the prices of them from two different pricing services, Standard & Poor's and IDC, two major pricing sources, and this is how many of the bond prices were different from the two different pricing services by 5 points or more. $5 means instead of one of them would give you a price of 95, the other one would give you 90 or less.

And you can see out of 49 securities than in mid '08, 24 of them were off by 5 points and 12 of the 49 were off by 10 points out of 100. So these pricing services, which include, again, some very smart people, had a hard time estimating the values of these securities. So that's where I'll stop. I'll just say that these securities have really complex risks to estimate and we need a little bit of sympathy for the financial application people. I thank you. Thank you.

FLOOD: Well, good morning. I want to thank Jim Perturba and Andy Lo for putting together this program. This has just been terrific. I also want to, before I start, acknowledge that my wife is here, Paula, she's sitting there in the middle, and then also my mother and sister are here.

Now, my mom will turn 80 this year and I grew up in North Cambridge, for those who know me, so the other side of town. And my mother grew up there as well before me. And I went to college. There's a liberal artsy school that's down the street over there. I went to college there and when I came home and told my mom and dad that I was thinking about coming to this side of town for graduate school, they thought why does that really make sense you need to go that far. And so she's here and this is the first time she's been to anything at MIT with me.

So I sat next to Bob Merton last night at the Solo dinner and I told Bob, I said, look, my mom has come from all the way over here to hear this session and your talk at lunch today, so we got to look sharp because she's going to have some things to say after this. And she's going to give me an assessment about whether it was worth coming all this extra mile and a half for graduate school.

And we've been talking about the Asian mothers. We've been talking about the Jewish mothers. Well, these North Cambridge mothers can be pretty rough too. So let's look sharp if we can, OK? All right. Good.

So as I reflected on this question that we were asked by Andy, how did our experience at MIT most impact our thinking. As I thought about that, again and again I came back to this one thing, which is that the Institute encouraged me to put structure on or describe problems in a way that I could use insights from economics and finance to help me address those problems.

So you might say, well, that kind of sounds like Freakonomics and well, it is, in part. But I think a lot of the issues that we work on actually go further and what I'd like to do is give you a few examples of those types of issues that we work on and try to put structure on them by using the types of tools that have been developed here at MIT.

So let me just say a couple of words about building blocks as I start. So in terms of building blocks, I did international finance when I was here. So I did international in the economics department and I did trade with people like Jagdish Bhagwati and Paul Krugman and then international macro with Rudi Dornbusch and Stan Fischer. Then over in the Sloan School here, I did finance. So I studied with Don Lessard and Stew Myers, Fischer Black, Franco Modigliani, Bob Merton, and the whole crew that was here.

So then I had these analytical tools, skills that I developed here. I went on to join the faculty at Stanford Graduate School of Business where I taught international finance. And there, in teaching in the executive programs, I was able to teach in several different places where we had joint programs, so here, at Sloan School, at the Nomura School of Advanced Management in Tokyo, and the International Management Institute in Geneva and that really helped me start to grapple with how to think about finance in economics from different cultural perspectives and legal perspectives.

From there I went to Wall Street and I joined Morgan Stanley and I did a number of things there. I did proprietary trading and that helped me learn about how to deal with securities in markets that trade on a 24-hour basis, so managing risk on a 24-hour clock, I went into fixed income in the heart of the firm where I learned how to deal with a demanding and powerful institutional client. And that's very different than doing prop trading.

I then went to asset management and learned how to think about investing for the long run instead of just trading. And then finally, I went to human resources and in human resources, in managing people that span the time zone, so in Asia, in the US, and in Europe, with direct reports into me, I became a better manager. And then I also learned how to think about the firm as a whole instead of just the group that I was responsible for.

So I took all those building blocks and then was invited by Doug to join Smith Breeden as the CEO. And so let me talk about a few examples of things that I've done there at Smith Breeden also at TIAA-CREF where I'm on the board there and I chair the investment committee. Let me talk some about the types of examples that we wrestle with real quickly.

So first, at Smith Breeden, we've really made our reputation on the strength of our research, among other things. But clients ask us not only to manage money for them, but they ask us to help them think through other issues that they're grappling with. So one of our clients, which is a central bank in Asia, it was looking to try to make that country more of a melting pot for its growing Muslim population in the country, in the region, and then also for Western institutions.

And so they had a large global conference and they invited Joe Stieglitz to be one of the keynote speakers and they invited Smith Breeden to do the other keynote. And so what we did to try to create a framework for people to have a discussion about the difference between how you might look at finance from an Eastern perspective and from a Western perspective. We took tools that would be familiar to financial economists all over the world. We used the efficient frontier as our basic building block for discussion. The efficient frontier describes how risk and return relate to each other in a portfolio context.

So we started with the unconstrained frontier that used all the major asset classes from around the world as a Western investor might approach these markets. Then we contrasted that with a frontier that was constrained to be sharia compliant, that is, it had the restrictions on fix income and other asset classes as designated by the Quran as an Eastern investor might approach the markets. And the two frontiers look different. And with that framework, we were then able to talk about in a very explicit way, in a way where we could actually quantify the differences for each one of the constraints and then talk about those with the whole group with a framework that everybody understood.

So that's an example that I would characterize as kind of a Freakonomics finance kind of exercise. But you can look at things that are much more difficult to combine both economics and finance and I'll give you one quick example from what we do at TIAA-CREP. I'm on as the board there with my colleague Jim Perturba here and our objective is to help participants invest to and through retirement to achieve a target standard of living. And of course, people are thinking about that in real terms.

Now, the organization is designed and is set up around a bunch of philosophies that relate to how we think we can help people do that the best. And you might say that, well, a lot of the concepts that you put to work there sound pretty intuitive so it must be easy to get them done. But I've never found that to be the case at all. That in large complex organizations, to get anything to move through that organization, even if I as an individual think that it makes sense, there's going to be a lot of constructive debate. Management has to reinforce it time and time again. Compensation systems have to be supportive of those ideas. Governance systems have to be supportive of those ideas.

So in order to get something to move through a large organization and for the organization to stay on that path, there's a lot of discussion that has to go on. And to do that, we like to try to use frameworks and then argue those points and then support them consistently through the organization. Now, I could give you a whole list of different things, but I'll just mention a couple real quickly.

One is, how are we going to add value in the markets. Well, what we have come to there is that we don't market time. And again, this is a philosophy, but it comes from a framework and I won't go through all the framework, but I'm just giving you examples. Here so we don't market time, we believe that we can't make money by doing that consistently. So we remain fully invested in the bulk of our strategies. However, we do believe that through careful fundamental research that we can identify companies that will outperform in the long run.

So we invest heavily in building teams that have the capabilities to do security selection. So that's one example of a philosophy, but then have a framework where we argue those points and then we keep it moving through the organization. Another example is, what types of risks do we get paid for, so we think hard about that. If you think about a multi-factor framework, one of the risks that we believe that we get paid for is liquidity risk. So as a long-term investor, we believe that we can go down in liquidity over time and expect to achieve higher risk adjusted returns by doing that.

Now, we have to be very careful though at how we implement those ideas. So for example, in commercial real estate, we don't just go down in the illiquid classes of CMBS, commercial mortgage-backed securities, which tend to be the least liquid. A lot of the times what we'll do instead is, we'll hold in securities-- the very highest class ones that are the most liquid-- but we'll hold the illiquid positions by making loans directly to companies or sometimes owning the property ourselves.

And there are many other examples here. I'll just say finally on this point, one of the things that we explicitly do is we used the heft of the company in the marketplace to influence selective governance issues with portfolio companies to improve investment results and also to achieve social good. And again, we have a number of philosophies. Each one has a framework around that and they're necessary to argue them and move them through the organization.

So those things I would put in a class of kind of Freakonomics plus lots of additional finance and economic tuition. And I just want to close with this one last thought. Another example where I'm putting the skills that I learned here to work to address hard problems is, with the University of North Carolina. They've just asked me to chair the Institute for Global Health and Infectious Disease for the University.

And this Institute works with the five parts of the medical complex there. So it's the medical school, it's public health, nursing, dentistry, and pharmacy. And the University has large complexes that go across the State of North Carolina. We also have complexes in sub-Saharan Africa, in Europe, in Asia, and in Latin America.

Now, we all know that delivering health care is not just a medical exercise. To think about the issues associated with delivering health care in Malawi in southern Africa, for example, it's not only, of course, the number of doctors that we have on the ground there, but additionally, it's the quality of food that can be delivered there, it's the reliability of electricity for refrigeration, it's education about communicable diseases, it's political stability, and all of these other things.

So what I'm doing is forming a multi-disciplinary board that will work with the doctors that will include economists, political scientists, lawyers, business people, and so forth to work with these medical complexes around the world. And we're seeking ways to structure these issues from all of these disciplines to bring together many branches of science and again, describe the problems in ways that we can use understanding from the different branches of science.

So I wanted to give those as examples of ways that we use structure on a problem and then try to apply scientific approaches. Thanks very much.

GOLUB: At the earlier panel, there was a comment that perhaps economists didn't appreciate the extremes of the distributions as well as they could and as a risk manager by profession, I have to give Gene a shout out here because I thought his appropriate and honorable references to his mother and his wife certainly were a way of managing tail risk in a way which if my mother or my wife were here, I certainly would have tried to emulate.

It's a great honor to have been invited to speak today. I'm a three times MIT graduate, all actually affiliated with the Sloan School. And I know for sure the experience of graduate school I found pretty difficult. So being invited back on this panel is intensely therapeutic and I think [INAUDIBLE] of that, there will be someone employed the shrinks in New York, but thank you.

So I think the request was to talk about some of the things that I've done, sort of after I left MIT and then go into some impacts in finance. And in my case, when I brought myself home after graduating, I set out and I found the one company probably in the world that would have been extremely interested in what my thesis was about. I wrote a paper under Stew Myers about optimal capital structure for electric utilities. And sure enough I found one who was an outlier to everyone else in industry fit my model perfectly. And sure enough they were excited and incorporated into the rate case, so that was my first experience with applied finance, a case of one but, nevertheless, very gratifying.

Then having set out-- and again, this was probably 1984-- I sent out my resume as an applied economics in finance PhD to all the major New York investment banks and got back rejection letters from all of them except one, which called me in for an interview and then rejected me. Because no one could figure out what that had anything to do with their business of finance, which I didn't quite get, but I had a lot to learn.

I did find a small-- actually it was an MIT placement office. I found an ad for a small little investment bank and I called them up and the fella said, you sound pretty interesting, but today is my last day. If you want a job, you should talk to my boss. So I spoke to the guy's boss and I said, I'm from MIT. If have this degree in applied economics and finance. I do a lot of computer program. He says, come in now.

So I show up there. This is sort of a disheveled office and he tells me that they're working on collateralized mortgage-backed securities and this is 1984. This is sort of a new thing. And he had this PC, which is something I had never really worked on while I was at MIT.

They had a 3/4-written program in APL, a language I never really learned and I didn't know much about CMOs. But with the MIT entrepreneurial spirit, I said no problem. And 3 months in 80 hours a week later, we had one of the first CMO models that came into existence.

So that was sort of an exciting experience, but the little firm, it was sort of a talent train, because suddenly I went from who are you to oh, I build CMO models. I have a PhD from MIT. Everyone wanted to talk to me and I became a popular guy. So I went off and joined a major investment bank and created in 1985 a financial engineering group. And the financial engineering group, that was sort of my name at the time, because we actually had engineers and we actually were engineering sort of CAD/CAM-type software to design new types of mortgage-backed securities.

And we actually would grind it out just the way you would expect an MIT team would do, studying all the parameters, tuning these things up, and calibrating them precisely to exactly what it was that the different investors wanted to try to extract arbitrage in the process. And it was sort of a great gig, but what happened is, the business started to change over the course of three years from what I was doing, making money the old fashioned way, earning it, working really hard doing this stuff to people discovered you could do something else, which is you could actually obfuscate. In other words, if you can hide the risk, then you could actually get people to pay you for things that probably shouldn't pay you for.

And I didn't really think I was such a great thing to do and it kind of made us get this idea that there could be an opportunity for a little mortgage boutique to be started and we had this idea that we were going to bring sell-side analysts to the buy side. And that was sort of the core idea, that we were going to be able to understand the security sort of the same way that the financial engineers who created them did. And that little boutique is today known as BlackRock. But that was sort of the conceptual model that we had.

And so we were very much believing in this idea of trying to understand what it is that you own. And from that, you can sort of follow a path to, I guess, what we now call risk management, which was the idea of, you want to know what your own, you want to understand how it behaves. And then we discovered it's actually when you're on the buy-side of the market, it was actually a much harder problem than we have thought because on the sell side, the way you figured out what it was worth was, you sold it to someone, whereas, on the buy side, if you were successful you grew and you accumulated the stuff and you got lots and lots and lots of it.

And so suddenly it wasn't just analytics, it was how do you like maintain surveillance on these things. How do you keep track of-- the charts that Doug put up brought back I don't know fond memories, but lots of memories of that sort of tremendously dynamic problem. Even if the models work two or three minutes and then, of course, the models had challenges along the way. From the beginning, one of the approaches that I always try to think about having some humility in the analytics. Maybe I wasn't the best guy in my graduate class, so I kind of figured I needed to take a little bit of a grain of salt the stuff I did because, who knows, I may not always be as precise as I hoped it would have been.

So we always try to approach the work with some humility. So we started building up these analytics and it seemed to be working well. But what happened was after a while, we started branching out into more and more types of securities. And I try to hire more resources because we needed to build more and more models.

Well, the problem was even though our company was a fast-growing company, we were requiring too many resources. And with the industrial organization perspective, what, of course, we were banging up into the economy is a scale and the scale curve and we were on the wrong side of the scale curve in analytics. And it became pretty clear that either we had to figure out a way to get much, much bigger in doing this type of analytics and systems work or we had to get out of it completely because it was impossible because we were trying to hire a disproportionate number of people relative to the size of our fast-growing company.

So we got the idea that-- let us do something radical. Let's talk about selling analytics and technology to some of our competitors. And it was very controversial at the time because this was perceived to be the secret sauce of what we did. And then, of course, the comeback was, well, if you drop of the equipment bag of the Yankees to any old team, that doesn't turn them into the Yankees. So yes, we can give people the technology, but, of course, it's how you use it that matters.

And once we got over that concept, we created a business which is today known as BlackRock Solutions, which is a very, very large scale provider of analytics and operational software to the investment world. And then with respect to this BlackRock Solutions business, we realized that it was sufficiently powerful that you sort of charge for what the market would bear and as we all know that means that you have to leave a significant amount of consumer surplus if you do that.

So how do you capture the full rents of if you really have a better way to make a mouse trap, but we came across the discovery that the way you do this, you buy an entire company, revamp its operations, and then you can extract the full value. And that became sort of part of the way we started to grow our company because we had sort of a better way to operate at scale and we were able to capture value and again, run with that scale being reasonably efficient and having the ability to have a handle the risks and exposures within the company.

And then the last company that I helped start was a company called TRIPP LLC, a little company, probably never heard of it. But it's part of the sort of know-it's-your-own philosophy. We started accumulating commercial mortgage-backed securities in our portfolios and when we started doing it, they were extremely cheap, but as the risk manager, at some point along the way when we've accumulated a large enough position, I was uncomfortable and I felt we needed to get analytics similar to what we had in the mortgage-backed securities.

The challenge was that this information was held very tightly and each of the broker dealers like to maintain sort of a local monopoly over that information so that in term secondary markets, they could extract rents from their information. And there, again, sort of us economic principal, they said, well, we have to figure out a way to break through local monopolies and we were able to use BlackRock's purchasing power as a large investor. I said, look, if you don't make the information publicly available, we will not buy your assets. And as the largest buyer of commercial mortgage-backed securities, that got people's attention.

And we were able to create this little company, TRIPP, and we posted the information on Bloomberg and the commercial mortgage-backed security market grew very dramatically. Now, I was at the end of the story. It's not so pretty in terms of where the market went, but I don't think that was because there was too much transparency. It has to do with, perhaps, a limited knowledge of the underlying commercial risk.

And I know Bob Jarrow and I had some discussions over this and perhaps he'll mention it because I know he did some work in that arena. So with that, I know we're running over time, so I'll close my remarks at that point. Thank you very much.

JARROW: Well, good morning, everybody. I can't tell you how thrilled I am to be here today and to be on this panel. It's a real thrill to be back to my intellectual home. I thought I'd start by talking about my research interests. And I think I can explain them to you by making the following two observations about myself. One is, that my finance colleagues think I'm a mathematician and the second is that my mathematician colleagues think I'm in finance.

So this means I live in a no man's land between these two fields. This no man's land is now called mathematical finance or sometimes financial engineering. And I think living in this no man's land gives me a unique perspective on the financial crisis.

So how did I get in this no man's land? Well, I think this is sort of an interesting story. First of all, like many MIT graduates, I like mathematics. Secondly, I graduated from the finance program here with a PhD under Bob Merton, Stew Myers, and Don Lassard. I then moved. My first teaching job was Cornell University and I went there to teach finance, but somehow I never lost the mathematics bug, so while teaching, I took the coursework for a PhD in mathematics and I guess that's what finally did me in.

My career has really been the creating of models and the implementation of those models in practice. Consulting or the industry work is really a mechanism that I've used to identify important and unsolved and relevant problems. And I'll give you two examples of such problems that I worked on. In the 1980s, unlike today, interest rates were high and volatile and the financial industry needed models to hedge this interest rate risk.

The existing models at the time required estimating expected returns on bonds, which if you've ever tried to do it, you know it's a nearly impossible task. So I hope to create a model that overcame that problem and the model is called the HJM model, HJM, Heath, Jarrow, Morton. David Heath and Andy Morton are both mathematicians, not surprising, given what I've already said.

So that was problem number one. And I spent a lot of time helping to implement that model in practice. About 10 years ago, maybe a little less, managing credit risk became a very important concern in the industry. The existing models at that time were called structural models. And here, again, those models had a problem.

Structural models cannot easily handle what we call correlated defaults and changing defaults across the business cycle. So I helped here, again, to create a model that does this slightly better. And the new class of models is called reduce [? for ?] models.

Now, this brings us-- believe it or not, I've gone through almost 20 years or so-- to the financial crisis. And at that time, structural models were still the most commonly used approach for modeling credit risk in the industry.

Now, we heard a question earlier, but it's been quite written up in the financial press that financial engineering and models themselves have been partly the blame for the financial crisis. And I personally believe that this is really not fair criticism. To my knowledge, most of the financial models performed reasonably well, but, of course, not all of them. And to illustrate this, I'm going to return back to the two models I talked about, interest rate risk and credit risk.

So for the HJM models for managing interest rate risk, which includes things like interest rate swaps, caps, floors, options, those sorts of things, my understanding is that those models worked reasonably well throughout the entire financial crisis. With respect to the credit risk models, let me start with those that I helped create, reduce [? for ?] models.

My experience with that particular model was that it worked reasonably well. I consulted with a particular financial institution during this time that traded hundreds of millions of dollars in these credit derivatives, collateralized default obligations, CDS, and so forth. And they traded those derivatives both before, during, and after the crisis. We used reduce [? for ?] models to manage risk and we were profitable before, during, and after the crisis.

Now, in contrast, the structural models for risk managing credit risk performed less well. And most commonly, they underestimated the default risk. And what this led to was too much leverage and too little capital in many financial institutions that used those sorts of models. Now, this problem was compounded by the fact that one of the biggest users of these models were the credit rating agencies. And many other financial institutions depended upon them to get the risk measures right. And it turned out they didn't.

So what's the answer? Contrary to the financial press, I believe financial engineering and models are the solution, but we need education on how to use these models, better education, and we do actually still need better models in these contexts. There are many problems in the financial engineering arena that are still unsolved and I thought I'd conclude by talking about a few of those that I'm currently investigating.

Most of the risk management models-- and here that's sort of a broad term, option-derivative type models dealing with all the different risks-- exclude a phenomenon that we've seen recently, which are asset price bubbles. And the price bubbles really change the way one thinks about risk in many of these pricing and hedging models.

So I've been doing a lot of work on that trying to understand how prices change, how the Black-Scholes formula changes when you introduce bubbles into an economy. I hope many of you attended the wonderful sessions yesterday, but a big theme is trying to better construct macroeconomic models. And one of the missing elements of macro models is a realistic financial sector, at least from my naive perspective.

So one of the things I'm trying to understand better and work on is how to include a realistic financial sector into a macroeconomic model. And surprisingly, simple, but most of the models that I've seen actually don't allow default in equilibrium in such models. And if you're going to try to explain the phenomenon that we've seen, you need to include these.

So in summary, I think the way of the future is to have better models. And I'm looking forward to being part of that process creating them in the years ahead. Thank you.

LEWENT: Good morning. I'd like to echo all of the sentiments of the prior speakers in terms of the honor of being here, the thrill of being with all of these phenomenal colleagues. And I was reflecting on the session yesterday where some of the panelists talked about the fact that the derivation of quantitative economics could only have MIT as its home. And I was thinking that's right. And the home for quantitative finance could only be MIT.

And for me in my career, probably the only place I really could get a chance to apply it, to take theory to practice-- and I'm going to come back to policy-- is at Merck, a very unique environment where that kind of sophistication actually could find a home and a utility. When I consider the impact of MIT on my career, I have to thank, again, Stew and Myron and Bob, all of whom taught me. I was one of those people-- I'm going to give my age away-- who got the lectures before the paper was published.

But the more important part about it all was not only instilling in me the appreciation for rigor, the appreciation for research and modeling, but a partnership that I've had with each of them over the years. And that, to me, is even more valuable than the initial learning. So I want to give you a few examples of applying all of these theories into practice. And as I thought about it, I have to tell you there are an awful lot of examples and it's really kind of exciting.

Let me start with one of the earlier ones, which is foreign exchange hedging. And when I'm talking about foreign exchange hedging, I'm not talking about short-term balance sheet hedging, which is pretty much the standard in most of corporate America, but about long-range hedging of cash flows. And when we undertook this project at the end of the '80s when the CEO of Merck came to me and was getting really berated by shareholders saying, why are you getting hit by the strong dollar and clearly you've got to be so smart there's something you can do about it. He turned to me and I turned to my group and also in the process parted with Don Lessard and Darrell Duffie to go about this in a rigorous way and really understand what the characteristics were of the cash flows at Merck and in the pharmaceutical industry, not an off-the-shelf kind of approach.

The result of which was, we had, I think, a very robust framework to understand where our foreign currency risks were and what aspects of those we needed to hedge and why. And then we set about putting together what I consider to be at least in those days a pretty sophisticated model to implement the hedging. The result of which was for, in my tenure there, so for another at least 10 to 15 years, we implemented that model on an ongoing basis and were able to minimize volatility of foreign exchange on the business, which had a beneficial effect, obviously, in terms of just the dampening of volatility. But I also tell you that from an operational standpoint, it was invaluable. As you go through the year, when management is trying to make resource allocation decisions, not to subject them to starting and starting artificially because of the volatility of exchange rates, so an extremely valuable experience.

Another is looking at being part of the R&D organization and developing a Monte Carlo simulation to look at the value of the portfolio. There are so many things one can think about to mention valuation. I'm going to come back to that again. But certainly from the inside of the research lab, how do you think about the value of not only the R&D investment, but also how do you think about individual projects. How do you go about portfolio management. How do you think about the short-term and the long-term trade-offs. How do you think about what the right level of R&D spending is and are you returning the cost of capital to your shareholders.

And through this whole process, as some of the speakers were talking about too, this started as a pretty discrete kind of an effort and over the years, became more and more into the common practice and the strategic planning of Merck overall to think through where a company like Merck in the pharmaceutical industry, which has very long horizon decisions to make should think about its investments and its returns. And along the way in that, Stew alluded to, Stew was a fabulous partner in thinking through the risk return staircase issues, cost of capital issues in R&D projects and in an invaluable context to how to think about that.

I would say, by the way, we use game theory. I guess what I believe is that if you are not only well enough educated and also continuously learn that you can develop an awful lot of important tools and models that need to, with the proper business understanding, get applied to help enhance business decision making. And in that regard, we did use game theory to help us with specific pricing decision we had for a new product that a new product launch, which actually turned out to be the right answer, so not just a model, but really helped decision making.

In terms of theory to practice, I'll close with two last examples and then just touch on policy. Two last examples are back to valuation and the whole area of mergers and acquisitions and in our business too in licensing and really being able to assist management in the valuation process. But this is not an academic exercise and I think that the really important aspect-- and as Andy Lo was talking about too-- the need for more really capable finance people to help management make the right decisions and understand why they're the right decision also goes hand in hand with not just understanding underlying valuation, but understanding things like when to walk away and why. What are the business implications of basically giving away all of your synergies and the underlying value and never returning the cost of capital to your shareholders. And what are the implications of incursions on your capital structure capacity because of misguided acquisitions and initiatives and the loss of strategic flexibility that those can provide.

They were all real-world examples, which I won't cite that we did not do it at Merck, but were exemplified it in the marketplace. So again, theory is great, but it actually really does apply directly to practice. And the last comment on the theory to practice that I'll make is on capital structure. Over the years, actually when I first became treasurer of Merck in 1987, I decided that it would be helpful, again, it's my MIT training, to have a framework for thinking about capital structure at Merck, namely, of course, not only the leverage issues, but the role of share buybacks, the role of dividend policy, et cetera.

And I spent a lot of time thinking about that over the years, came in the short form to the decision that given the business risk in the pharmaceutical industry that you don't see on the balance sheet from an accounting perspective, you really are not wise to take on financial risk. And again, this became not just an academic discussion. This was something that I engaged in long and serious debates with the board of directors on over the years, had the support of management and the board ultimately, and we adhered to what we all felt was prudent capital structure policy of minimizing financial risk given our business risks.

And in September of 2004, when we voluntarily withdrew Vioxx from the market and went through many years of a very trying period, that capital structure decision was pivotal in helping the company go forward. Pivotal internally to say to the employees that we have the capacity to continue to invest and grow in the business and pivotal to be able to say to our investors, not just that but also, by the way, the dividend to secure and to give the company the breathing room to go forward and grow and also to undertake the kinds of litigation strategies that were very important for the ultimate resolution of the situation.

So I'll just close in terms of policy because, again, there is a very critical role for all of us in industry to play on the policy front. I will stick to the pharmaceutical industry here with just two examples. One is that for about 20 years, I organized a group of other pharma companies to get together to sponsor quality economic research about the economics in the pharmaceutical industry so that we could have a thoughtful and rigorous discussion about the risks and returns in the industry so that ideally we could protect this innovation model that is so critical to human health and have the right discourse with policymakers.

That effort, by the way, was a joint effort with a lot of my other colleagues in pharma, not just for human resources to mount the studies and financial resources, but most importantly, another theme here in the last two days is data. And there are a lot of people in the academic community who were very interested in studying the pharmaceutical industry but lacked the kind of data that they needed to really do the work. And through the effort of this group, we were able to provide that data and I think they have phenomenal body of work derived from that.

And my last example, which maybe I shouldn't closed on this one, but I had the honor of speaking here several years ago on this topic is, understanding the role of differential pricing and what I call pricing for access of prescription drugs around the world. And so all of this, to me, just comes home to the concept that there is an invaluable anchor that came out of my MIT education of understanding theory in a world and at MIT that doesn't just value theory but values the application to practice and to policy. And I am one of those people who have no lack of questions also of opportunities of new areas to probe going forward.

I will close by citing Stew's textbook. All of the additions, I believe, closed with 10 unanswered questions in finance. I can't recount them from memory. I can recount a few, but I won't do that here, to say that "it is a rich area, an area with much more to do in terms of research and then application to drive the enterprise going forward and the economy." Thank you.

MYERS: So thank you, everybody. That was great. We do have a chance for a couple questions, but may I give the panel, quickly a chance. If you want to tell the academics in the room what you think we should be working on. And Judy, if you want to say something, I'll give you the first shot because I know you have to leave to get a plane.

LEWENT: Well, now that I've put my foot in my mouth here, there are so many things. I would say, as I reflected on it, there are whole categories of a areas in corporate finance. One continues to be not just the perfecting of valuation of assets, but also I would even say we need broader education of an understanding of those. So it's not just financial practitioners, it's all of management. Because if your audience doesn't understand what you're trying to do, you don't get to apply it in practice and then the boundary into capital markets and how do capital markets respond to business decisions and the classic discussions of capital allocation and short-term needs and long-term needs and signaling to markets.

And maybe those are powerful words, but all of capital structure initiatives, for example, fall under that and continuing to understand those. And I know, Stew, you and I have talked a little bit about dividend policy. There are a lot of issues that have been around for a long time and it's either the good news or the bad news. There are no hard and fast answers to what to do. And to this day, corporations struggle with the right thing there.

I would say maybe on a lesser level, but one other practical one I was thinking about, which I get into discussions with in boardrooms all the time and maybe it's a super small subset of efficient markets and who knows that there are efficient markets, is the role of guidance. Do firms give guidance or not and to what degree do they give guidance.

MYERS: On earnings.

LEWENT: On earnings, pardon me. Yes. Thank you. Well, on earnings, on sales, on earnings, how detailed. You could expand that in say, speaking strictly about pharma, other metrics, how far do you go on pipeline discussions, and things like that. And what is the role of quality disclosure for basically lowering the cost of capital and enhancing the value of the firm. So those are just a few thoughts there.

GOLUB: One area which I think gets neglected with all the focus on the models and their value added is, how do you help practitioners be able to peer through the anonymity implied by these models and to understand the impact of finance on real people. And I think if we're honest, there was quite a bit of abuse of models and techniques. And some of that abuse may have been people who just didn't understand the models correct, but there's also, I think, a fair amount of avarice that we saw in the marketplace and models as used as techniques to basically fool people about the reality.

So we can talk about finance and use sort of scientific methodologies, but if we don't also make sure people understand that there are real consequences to what happens, that real people's lives are affected, then I think all the work in the modeling and the financial science can create some fairly amoral individuals who go out and do terrible things that really hurt real people in their lives.

So in a sense, to use the knowledge and to use the science, but to make people understand what its impact is.

FLOOD: For one of the areas that I think that a lot of participants in the market including us, a problem that we had was that we can get very good at pulling problems apart and saying, if you change one thing but you hold everything else constant, what would happen. The harder thing is, if you change one thing and that affects something else and that affects something else and that affects something else, then how do you understand everything that moves when one thing starts to change. And it's related to the concept of systemic risk.

And so I think one of the areas that we need more work on is this area of systemic risk. Andy Lo and others have been starting to work on this, but looking at the link with linkages and leverage and impact of losses et cetera, how all of those things tie together so that when the housing prices move and that affects banks and that affects hedge funds and that affects the GSEs and that affects so forth, then what's going to happen at the end of that as they all move together.

BREEDEN: [INAUDIBLE] some of the things I think we ought to be working on. Obviously, I'm concerned about estimation risk and these model parameters and along with that. And particular, I think MIT people like us tend to believe our models perhaps more than we should and so it can develop into overconfidence combined with that estimation risk, can cause a lot of problems. I think there's a lot that behavioral finance has to say as a former owner of Smith Breeden and thinking of a bunch of portfolio managers managing portfolios of our clients, you have to worry about the incentives to them, issues of loss aversion, over confidence, and some of the behavioral issues I think are important.

Then additionally, obviously, on tail risk, if tail risk is really big, if the risk of extreme moves, then probably a risk premium need to widen out over where they were a decade ago. And I mean, they've already widened out. They've come back in some, but assessing tail risk is really hard to do. And I think it'll be interesting to see where that goes.

And when you combine that with leverage, with changing correlations, I mean, as these things come together, people start having troubles and they start selling securities to cover their margin calls or to raise their capital ratios. Then what's going on in one market contaminates another market and all of a sudden assets that are supposedly unrelated become highly correlated. And if those are then held by levered institutions, 30 to 1, then you can go south in a hurry. So those are some of the issues I think we need to be studying in the coming years.

MYERS: You've already said some of the things you're working. Do you want to add something?

JARROW: I was going to say that I'm involved with the software company. I'm an owner and we do risk management software. And probably the biggest problem that we face is measuring the risk on the assets of our financial institution including all the risks-- market, credit, liquidity, operational risk. And that's a very, very difficult task and we're required to do that by regulation. So that inconsistency is all I wanted to mention.

MYERS: We're almost done, but we have a very patient gentleman down here. Do you want to throw one last question at these people?

AUDIENCE: Thank you for the opportunity.

MYERS: --people?

NACHTRIEB: Yes. Robert Nachtrieb, SCD 2000 in applied plasma physics and candidate for MBA from Sloan's brand new executive program. We've talked a lot today about financial engineering and I think that the term that has been used most commonly today is using engineering techniques to perform finance. I'd like to ask the panel on turning that concept backwards and in the sense that Professor Lo mentioned earlier today using finance to perform some of the biggest work on engineering problems. And Professor Lo mentioned, energy, cancer, and greenhouse gases. But I'd like to hear your thoughts. I thank you.

MYERS: Well, first of all, just to protect people here, I have to say that in a way, financial engineering is a funny term and I hope you'll excuse me for saying it this way, but if all is engineering, you're going to lose a lot of money trying financial engineering. Experience has shown that you need the combination. You've got to take the financial problems first and essentially economics of the finance problems and then do the engineering where the quantification makes sense. I'm sure you'd agree with that.

So if anybody wants to talk about how we're going to apply this.

FLOOD: Look, I'll take a piece of it. I think this relates to the last point that I was making in my comments is that many of the problems that we face in our country and around the world are not limited to just one field. You can't solve them by looking at them through one lens. And so I think finance and could be financial engineering, but you could think of finance as being part of the solution and these problems will require us to pull together answers in insights from a lot of different fields in order to get to the finish line.

So we need finance and economics. We will need medicine. We will need political science. We will need a number of different things. And so I don't want to speak for what Andy was saying there, but MIT is a classic place for being able to do multi-disciplinary things. In the smallest area, you can think about finance and economics being right together when you see a lot of that coordination.

But you also see the business school working closely with engineering and engineering working closely with computer science, et cetera. So I think the way that at least I'm looking at this is pulling together multi-disciplinary teams to solve some of these big problems that we are facing in the world.

AUDIENCE: Let me try to answer it. I think Andy can correct me if I get it wrong. But the idea is that all of these problems require resources to employ a scientist to solve the problem. And there are a lot of financial resources out there that are looking for good uses. And if you can use financial engineering to design securities that access those funds and put them to those uses, you can do a lot of good. And I really think that's what Andy had in mind and that's a really good use of financial engineering.

MYERS: That's probably a great place to end. And actually, what Gene said was important because for all the glorious things that we've accomplished in finance and MIT has accomplished an enormous amount, we shouldn't go home thinking that finance can solve all the world's problems all by itself. Thank you very much.