Richard Foster
Founder, Foster Health Partners

"Creating and Capturing Value in Healthcare: The Next Five Years "

I’ll cover a little bit of the theoretical as well as the practical today. SFI has played a very large role in my thinking and the tangram with all of the words on it (this year’s TLF theme) is at the center of what I have been working on over the years. I’ve been working to turn those concepts into practical tools.

This woodcut that you see on the screen is one of my favorite examples of a complex, non-linear system. It’s from the seventeenth century and shows the interaction of non-linear forces and some of the surprising results. These are standing waves. There is a ship in the bottom of the woodcut with a scale that indicates the waves are on the order of forty feet. In fact they’re about six feet tall, but they do stand there. This is a real, geographical location off the Lofoten Islands in Norway. It’s a place the Danes first named the Maelstrom. This is the way I look at the capital markets. I will talk about the winds and the tides of the capital markets and how it all works.

Sticking with the nautical metaphor, I thought I’d do a quick history of innovation in the steamship industry. This is a picture of a sailing ship, the Glenavlon. It’s from 1870 or so. On this particular day, it was not providing any useful economic function because the winds were still. As they say in the sailing business, “no blow, no go.” [laughter] This ship wasn’t going anyplace, but it didn’t represent the ultimate design available. They could improve this ship by innovating, and the France II is an example. It had four masts and more sails and was in every way a better boat. It was faster and carried more cargo with less crew. It still couldn’t go if there was no wind, and was beginning to lose share to the steamships, which were themselves very inefficient at the time.

The next stage of innovation in sailing ships was the Preussen, which had five masts and six sails on each mast. You can see the theme that is shaping up here. Even so, steamships continued to take share away, so they innovated once again with the Thomas Lawson. It was an oil trader out of Boston; on the photograph you can see oil leaking from the side of the ship. It had seven masts and a complicated array of sails. This was built in 1902 and refitted to carry oil in 1906.

The ship had a distinctive history in that its maiden transatlantic voyage was its final voyage. It started from Boston headed for Portsmouth in December, 1907. When it got to the English Channel, Captain Dow made several simultaneous discoveries. First of all, there were islands in the mouth of the channel called the Scilly Islands. Secondly, it was blowing at 60 knots on that day. Third, it was Friday, December 13. Fourth, the naval architects had traded off maneuverability for speed. This was a very fast ship but it didn’t turn very well. It turned turtle and lost all of its cargo and lost its crew save two: the engineer, and of course, Captain Dow. As I’ve told this story to CEO’s, their response has been, “so, what’s the problem?” [laughter] Maybe Sarbanes-Oxley would have prevented this, but I’m not sure.

While this incremental innovation was happening on the side, more fundamental innovation was occurring. This boat on the screen here is a steamship. It’s not particularly attractive, but new innovations are often not very attractive when they first appear. But it was more economically efficient, and you know the rest of the history.

Now I’ll switch to the American economy. If you go back to 1926 when the S&P started—it was then the S&P 90—you’ll identify about 15 companies on the original list that were still on the list in 2000. They are the long-term survivors in the American economy.

You might assume that if you had been an investor in those companies in 1926 and stuck with them, you’d be well off. But if you actually look at the average return, and subtract the S&P results, you’ll find that group underperformed the S&P. This is contrary to the way that some CEOs think about things. Obviously in the war years these companies did better and in the boom years they did worse. It’s a variable pattern, but at no time did they go above the S&P and stay there.

We can also look at equity returns of the survivors versus debt instruments. Here I’ve compared total return to shareholders to their corporate bonds. The shareholder returns are somewhat greater than the debt returns, but if you look at the variance you’ll find the debt returns are in excess of the equity returns in many cases. So what is the role of equity? From 1970 through 2000, there were only a few years when the five-year average total return to shareholders of GM was above a 10-year T-bill.

Why do we see these results? The culprit is the pace of change in the economy. Since its founding in 1957, the S&P 500 has turned over twice. There are some 50 core companies that have persisted. But the S&P 500 we see today is much different. All of the 50 companies that have remained have underperformed the S&P itself.

The pattern of the rate of change in the S&P back to 1926 shows that in the early years the S&P was turning over at 1%-1.5%, implying about a 67-year average lifetime for a company in the S&P. By the late 1990’s the turnover was around 5%, implying a 20-year life. The instantaneous rates in 2000 reached 11%. There are peaks and troughs in the curve, but the general pattern is upward. Each trough is higher than its predecessor. The pace of change is continuing and is vital to the economy.

During the rest of the talk I’ll look backward into the historic record, talk about why some of the anomalies might be occurring, look forward at the health care industry with concepts from complex adaptive systems in mind, and talk briefly about the long term context.

This work is based on work I started at McKinsey a long time ago. We decided we didn’t want to look at individual stocks or individual industries; rather we wanted to look at the whole thing. So we pulled together the financial data for 42 different industries and 2,424 different companies. We used 120 different variables for each, of which 60 were original and we calculated the remainder. We picked industries that had more than 10 competitors because we wanted to do statistics. For example we didn’t include the automobile industry, but included the auto parts industry instead.

Once we picked an industry we included all companies that had more than 50% of their sales in the relevant industry and had a market cap in the top 80% of all U.S. companies. The first year of our analysis was 1962. We allow companies to enter and leave the database. The dead company file is roughly equal to the live company file.

We tried to visualize the data. We plotted total return to shareholders over time for all the companies. We took the median of that and called that the industry. We took the medians of the industry and called that the economy. If you look at an individual company stock and ask what fraction of the variance in that stock comes from the company, what comes from the industry and what comes from the economy, it’s about 40% from the economy, 40% from the industry, and 20% from the company. This turns out to be a fundamental observation. It’s why I tend to focus on industry groups first as the unit of analysis.

The total return to shareholders annually back to 1962 followed a regular pattern of peaks and valleys. The folks at Los Alamos did some simulations on the pattern and concluded that it was very hard to imagine that it came from a random distribution. The peaks and troughs are reasonably regular and the periods are regular as well other than various punctuation points in 1974 and in 2001. The S&P 500 tracks on top of it very closely. The S&P is higher than ours in the 1990’s because they do market cap weighting.

What have we learned from this? The first observation is there are underlying patterns with long stable periods, punctuated by catastrophes. The basic unit of analysis is about three years, or one cycle. If we average over two of those, you get seven years, and a seven year rolling average handles much of the variance. So in the 1980’s and 1990’s there was a fairly constant total return to shareholders of about 10%-11%. Compared to the 30-year Treasury bond you can see the equity premium on the chart in the 1980’s. You’ll also note the equity premium goes negative, which acts like a marker in the genetic code causing everything to stop. Everyone switches their assets to debt instruments, which is what they did in the 1970’s.

The second observation is that mature industries perform as the economy does, and that the market efficiently sets prices. You can consider about half of the 42 industries to be mature. If you look at their relative returns compared to the economy, you’ll see they are about the same.

On this chart I’ve plotted years on the horizontal axis. The vertical axis is a bit more complicated. Imagine that since we’ve gathered data by companies and then grouped them into industries and then into the economy, we can compare a company to its own industry. And we can compare an industry to the economy. In the case of this chart, I’m comparing an industry to the economy. If in a given year the return of the industry is equal to that of the economy I’ll call it “50”. Then because I have 42 industries, I can take a variance around that number and express it as a confidence limit. I’ll normalize that confidence limit so that the top of the limit is “100” and the bottom of the limit is “0”. Now I have a blue horizontal band and anything that happens within that band is statistically indistinguishable from what’s happening in the economy. If I look at the mature industries, by and large, their behavior tracks within that band. In my mind that means that prices are efficiently set. Total return to shareholders of the sector is roughly equal to total return to shareholders of the whole economy. Price-earnings ratios of the industry are dead on the 50% line of the economy.

The third observation has to do with the rest of the industries that are not mature. We call them dynamic industries. They both out- and underperform the S&P. Prices are not efficiently set for long periods of time. Semiconductors, for example, start out outperforming the S&P from the late 1960’s to the early 1980’s. Those numbers in real terms amount to between 200 and 300 basis points for the whole electronics industry. For those who got comfortable with those levels of return and assumed that they would continue, the industry crashed in the mid 1980’s. This is real risk. This is the risk that it’s impossible to diversify away from and that no studying of history can tell you about. You need foresight in order to anticipate this. What happened in semiconductors was the end of the DRAM business and the world was oversupplied and prices fell in a way that the market didn’t anticipate. Along comes another product called the microprocessor which drove the performance through the ceiling again in an extraordinary example of outperformance.

This is the trace of technological discontinuity. Clay Christensen calls it disruptive technology. This occurs in industry after industry. Prices are not efficiently set in these examples. The multiples were way above the industry multiples during these periods of time. This is a nightmare for value investors. The multiples were more robust in some ways than total return would have led you to believe.

The oil industry went up during the 1970’s. In 1981 or 1982 everyone in the country knew that oil was going to $100 per barrel. I don’t know what the current price equivalent is but my guess would be $150 per barrel in today’s terms. Doubters said it would go to $80 instead of $100. That’s when it was at $30. The price of oil collapsed instead. The industry attracted so much capital to every single aspect of the business that we went into a vast overcapacity position. Oil collapsed and took quite some time to come back. And today it’s above the blue band once again.

The pharmaceutical industry had great performance in the late 1960’s as well as in the early 1980’s through the end of the century. These are errors in pricing that do not get arbitraged away in the short term. You know the story; the industry has come rocketing down in the early years of this century. But that isn’t the first time it came down. It also came down in the early 1970’s. No one’s done an analysis of why, but I think there are several possible reasons. The FDA increased the amount of required drug testing and the average testing time doubled from four years to eight years, which cut the product patent life from 13 years to 9 years. We had the impact of ERISA really beginning to have teeth, with Medicare and Medicaid kicking in on drug prices. We also had inflation going through the roof and the price of oil taking off. People were allocating capital to the oil industry and the flow of funds for pharma dried up. Nevertheless, pharma maintained above-industry-average multiples during this whole period.

The air transportation industry is another interesting industry. This is an economic wonder. At almost no time during the last 40 years has it even come into the blue band of average performance. No matter how pessimistic you are, you’ve not been pessimistic enough. That’s not to say that there are not successful companies—there has been one. I don’t know the current number, but I think that Southwest Airlines is two-thirds of the industry’s market capitalization right now. This is despite a fairly consistent pattern of taking the multiples down over time. This is not just a problem of the industry but of the analysis of the industry. The prices are not being effectively set. The airlines analysts seem like they’re training their own children to make the same mistakes they’ve made. [laughter]

There’s another correlation that struck me as interesting. It occurs in about two-thirds of the industries. Medical products is an example. One vertical axis is the same total return to shareholders that we’ve been working with. The other vertical axis is an index of the rate of change of the number of companies in the industry—the net new company index. The ratio maps closely onto the total return to stockholders plot. It neither leads nor lags—it’s a coincident indicator. It is a reflection of the investment proclivities of the people in the industry.

The fourth observation is that most companies perform as their industries do. Your industry is your destiny. CEOs hate this. But it’s true in most cases. So I’ve had CEOs ask me, “if this is true, then what am I supposed to be doing?” The answer is, “preventing your company from going below the blue band of your industry.” They are risk managers, not performance managers. There is more risk management in an industry than real performance management.

Observation number five is that companies can excel for a time. But these times always come to an end. We have 2,424 companies in our database. Focus on those that have a fifteen-year record in that time and pick the one company out of that group that has done best and tell me what you think their average total return to shareholders per year is over that period. Number one is the Limited at 52.1% per year for fifteen years. It was an entrant to our database and the period was 1974-89 when the economy was challenged. TCI had a TRS of 50.7% and the period was 1975-90. Wal-Mart was not far behind and in the same time period. Amgen was in a different time period. Oracle, The Gap, Home Depot, Microsoft, Leucadia National, and Berkshire Hathaway round out the list. They lump around two different time periods when the economy was challenged. Out of that challenge came the freshness, and that freshness was expressed in these companies.

There is another side to this story. If we look at the Limited after its great run, every period after that it was lower. The same thing happened with The Gap. They had a very high “fade rate,” or reversion to the mean. It affects most, but not all companies. Wal-Mart and Home Depot both have a low fade rate. Berkshire Hathaway has avoided fade and has kept fresh all along.

What are some of the possible causes for the phenomenon that the world doesn’t look like what the economists tell us it should like? Imperfect prices come from imperfect investors, and they are your partners! They’re setting the price collectively. And stock price does affect internal operations.

Here’s one way of looking at the problem. Imagine that industries evolve in the shape of an S-curve. It’s a non-linear curve. Assume that it’s a long period of time, like 80 years. Assume that we’re an analyst at the very beginning trying to do our cash flows so we can discount them all back. We won’t look 80 years ahead. Let’s imagine that we look back two or three years and use that as a basis for projecting forward two or three years in a linear way. If you do that, in the beginning of the curve, you will miss the actual forecast by a considerable amount: you will under forecast. Then the next year, you learn that you under forecasted, so you adjust your process. So the slope of your line rises, but you still miss the actual growth of the industry by a fair amount. You continue to adjust, and eventually you get it right as the upward slope of the industry growth reaches an inflection point. Just after that point, you get a nasty surprise, because you overshoot by a fair amount. Until this time, you’ve had a period of positive surprises. But after the inflection point, a period of negative surprise sets in. We are all familiar with this phenomenon.

Next, I take a theoretical company and evolve it over 100 years and assume that its cash flow evolves as an S-curve with the period between 30 years and 70 years being the highest growth period. How would the net present value of the company evolve if we knew 100 years in advance what the growth of the company would be? The net present value also evolves as an S-curve, but before the cash flow evolves because we’re taking the present value of future earnings. The total return to shareholders would be flat—in this case I assumed a cost of capital of 10%.

Here are some simple-minded observations. First, the total return to shareholders does not track the actual performance of the cash flow. This is not welcome news to all CEOs because they think total return tracks their own growth. There is no variance in the curve because there is no risk—we know the future of the entire life of the company. If we look at the multiples associated with this model, we find that they are out of sight and unrealistically large. The multiple in year one would be 300 times earnings. There’s nothing about this model that’s realistic.

What if we did our linear forecasting against this? Then the TRS evolves more slowly up to about the midpoint of the curve and then evolves more quickly after that, tracking the two periods of positive surprise, followed by negative surprise. Total return to shareholders starts high in the beginning and then it drops as the performance of the industry improves and then starts to come back as the industry levels off in the later years. These non-linear dynamics come as a surprise to many people, but they are a simple mathematical consequence of the underlying assumptions of this model. P/E ratios start at 25 in the beginning—much more reasonable levels. The P/E and TRS ratios parallel one another. The highest total returns to shareholders were made when the multiples were the highest. This is why some value investors find it difficult to invest in these industries.

If we speed things up and the growth occurs over 30-40 years, perfect knowledge evolution of net present value starts higher than the S-curve and gradually grows to join it in the later years. But when you introduce the linear forecast distortions into the system, they create a bubble during the growth phase that peaks just at the end of the growth phase and crashes. In this case, TRS is driven very high to around 100 in the beginning and then it collapses catastrophically as the business gets better and better and finally builds up a little again as the business starts to slow down. Price-earnings ratios parallel the TRS curve. This represents a mismatch between the forecast and the underlying long term reality.

Imagine a more complex pattern like a pipeline filling pattern with rapid growth, a peak and then a drawdown to some stable level. This is like the growth of mobile phones to a peak followed by a period of maintenance. In this case, the NPV evolves leading the cash flow. Book ratio starts high and then falls most of the period before rising a bit at the end. TRS starts high, collapses, grows a bit then spikes up and collapses precipitously. There are venture capital opportunities in the beginning, hedge fund opportunities in the middle, and leveraged buyout opportunities in the end.

Looking forward, how might this apply to the healthcare industry? Are we at one of these inflection points, particularly with regard to biotech? We have some choices. The healthcare world has about $2 trillion in market capitalization and employs about 2 million people in this country. There are 850 companies in 17 major categories and 120 minor categories. There is a high rate of creative destruction—many companies come into existence and leave. It’s a very complex sector. Eighty percent of the variance that you’re counting on in the S&P you get in the healthcare sector because it’s so big. When pharmaceuticals go up it’s because prices are going up, which means prices that hospitals pay are going up, which means hospital profits are going down and these two things work like an accordion. We segment the healthcare industry like a healthcare professional would—by diseases, for instance. No matter how in-depth you look at infectious disease companies over the past five years, you will not see the impact of avian flu if it comes. Even if avian flu doesn’t come, every hospital has to get prepared and there isn’t enough capacity to do that.

Over the next ten years, there will be $20 trillion of sales in this sector. The growth rate will be around 9%, with EBITDA of $1.5 trillion. Total increase in market capitalization will be around $3-$4 trillion, and cash flow will be $350-$400 billion to these companies. We think $2 trillion of new capital will be required to fund this growth. The minor portion will be new equity; the majority will be new debt and all kinds of new financial instruments. There’s almost $1 trillion in mergers and acquisitions. The net rate of entry has been increasing. Research organizations, IT organizations, diagnostics companies, providers, and insurers are among them. Healthcare and pharmaceuticals are not the same. There will be 200 new significant companies emerging over the next 10 years, and roughly 60 consolidations.

The top performers enjoy very high returns. Even the 20 th company has 20% returns. There is some risk, however. The worst ranked companies are well into the red. Many public companies are illiquid—it can take two or three months to build a position or get rid of a position. About 250 have liquidity that would be of interest to a hedge fund. The smaller companies are not attractive to the hedge funds and basically shouldn’t be in the market.

Investors have shown an interest in biotech. There are some possible technical trigger points. Cures for cancer are starting to consolidate into what one analyst has called a dominant design—the angiogenesis inhibitors which affect about 40% of all cancers. There’s a new mechanism called CTL-4 that affects the rest of the cancers. There are a number of cancers now that can be cured or set into very long remission. The outstanding problem is metastasis and progress is even being made on that. We are at the point where we can understand a biological mechanism and intervene in that mechanism to stop it with designer drugs. The problem is that currently cost is exorbitant—up to $200,000 before other costs. This will not be a mass market. It’s like the early days of semiconductors. Autoimmune diseases are becoming increasingly treatable. Congestive heart failure is being kept at bay. Radiation therapy has new techniques and proton therapies. The machine itself costs about $100 million but a new machine coming out will sell for $15 million. Stem cell research is on the horizon. Personalized medicine is also coming along.

There are some competing trigger points like ASP-based information systems. One of my friends says, “Health care is incredibly simple—you have to get paid.” That turns out to be incredibly difficult. These information systems are the key to unlocking this value. There are a number of physician practices that get paid 50 cents on the dollar for everything that they bill. That is not a good business model. But with the information systems we can get 95 cents on the dollar. Electronic medical records will revolutionize the world. There is a trigger event in Congress now: the Medical Information Act of 2005. It should get passed in the Senate but the House Republicans need to get behind it. HSA’s and HFA’s are new ways of financing for the consumer. There are new techniques of managed care and wellness. Biotech will happen, but not until some of these things happen. The winning portfolios these days are not pharma portfolios.

What can go wrong in this scenario for healthcare? There is always risk on the horizon. We could end up in a stalemate with a combination of these complex systems and government regulation.

What is the long term context? If we go back to 1870 and look at the spread between equity returns and debt returns we see cycles. The peak spread between equity and debt returns was in the 1880’s and was around 9%. This time showed lots of innovation: steel, telephones, cars, the punch card machine. Then everything stopped in the early 1890’s and another cycle started up in the early 1900’s. These innovations capitalized on the ones that had come before: telegraph, radio, the assembly line, and flight. We ended in 1913 and that’s when we formed the Federal Reserve and the Sherman Anti-trust. It looks to me like the population really doesn’t like losing money and when they get into a period of time when they do lose money, they blame Washington and Washington does something. In general, Washington has done very good things.

Then after the First World War, we hit the boom time of the 20’s with a huge spread between equity and debt of over 20% at one point. That resulted in enormous overcapacity. There wasn’t really time to invent anything before the Great Depression. Then there was a little spike coming out of that along with the Glass-Steagall and Securities Exchange Act. Then we went into the enormous boom of the 1950’s and 60’s with the maximum spread around 20%. There was an enormous range of innovations around this time: the hydrogen bomb, Sputnik, the LASER, the microprocessor and the transistor before it. We’ve just come out of another boom in the 80’s and 90’s and we’ve fallen to the bottom again. If we overlay these cycles, they’re getting longer and more well behaved. I hope that we’re about to go into one of these longer cycles now.

G.K. Chesterton was an advisor to Gandhi at the turn of the 20 th Century and he said, “the real trouble with this world of ours is not that it is an unreasonable world, or even that it is a reasonable one. The commonest kind of trouble is that the world is nearly reasonable, but not quite. Life is not an illogicality; yet it is a trap for logicians. It looks just a little more mathematical and regular than it is; its exactitude is obvious, but its inexactitude is hidden; its wildness lies in wait.”

Question: Are there industries like finance where you get a greater dispersion within the blue band?

Answer: We have looked at the financial industry and the dispersion is enormous. The subsectors are quite uncorrelated with one another. Energy falls in that category as well.

Question: You spoke about a number of great opportunities in healthcare. What are the different pricing mechanisms for delivering various outcomes based on the service being provided? For example, in looking at the pricing mechanisms for these new proton delivery treatment mechanisms, the history seems to be that there is very inefficient investment in what the next innovation will be. In the end, it seems the amount of that delivery capability exceeds or is inappropriately delivered in terms of the demand.

In other words, what different pricing algorithms do you expect to see developed?

Answer: We will do whatever it takes to save the patient’s life and to increase the length of the patient’s life so long as that’s what the patient desires. Particularly when we have Medicare and Medicaid patients, we support the individual enormously. It’s quite clear that if we only had those patients we could not do that for very long. We would soon eat through our endowment. We have many managed care and private pay patients and an enormously generous set of donors. Right now we can afford to do this, but it is not sustainable. If I had a simple mechanism for fixing it I would lay it out on the table. I don’t know how it will turn out. It affects everything we think about at Memorial Sloan Kettering. With regard to the proton machine, it isn’t developed yet so we’ll see what the pricing is in a while. It can be competitive with other chemotherapies. There’s very little collateral damage with protons.

Question: What is the proton machine?

Answer: It treats cancer. Imagine a prostate for example. With conventional X-ray technology you get more collateral damage while you cure the cancer. As the technology has gotten better, the amount of surrounding tissue that you affect, especially on the downstream side of the beam—has decreased. But it’s still significant, particularly when you’re operating on organs close to major nerves or blood supplies. Protons can be focused much more precisely and there’s very little collateral damage.

Question: What are the opportunities for savings on the administrative side of healthcare? What percent of our GDP that is spent on healthcare would be administrative?

Answer: I don’t know that number but it is a significant number by any stretch of the imagination. Some of it will be addressed by electronic medical records but a large part has to do with variation of treatment that happens in hospitals like misidentification of patients as they come in. If you go into a hospital in California, the range of treatment you can get within 50 miles of San Francisco for the same ailment is stunning. There would be no business in the world that could run a manufacturing effort remotely like this one is being run. One time it produces a bicycle, the next time it’s an airplane wing and so on. That has to be dealt with. Compared to biotech, these very practical fixes are more important. They will bring down costs, increase capacity and improve profitability.

Question: Government vs. HSA—the debate points to a consumer model. Does this match your outlook?

Answer: Yes it does, and I’ll certainly distinguish my opinion from truth! The big fear I have is that we will end up in a regulated system that has an effect on the capital markets that we see in the airlines. I see that as unacceptable. The airlines industry works and is still there but it’s not the best way to structure an industry. Some government intervention is needed but too much will be detrimental. The government has invented many interesting ways of protecting investors from undue risks. But each of those builds in certain economic inefficiencies for the economy as well. It’s a very delicate balance. I hope the Information Act gets started because Medicare and Medicaid reform doesn’t look too good. I think healthcare is an attractive industry because it is so inefficient right now. It will be restructured multiple times before we get to a reasonable solution.

Geoffrey West: You used the ships as a metaphor. I want to relate it a bit to my presentation. In the 19 th Century, one of the major issues was whether it was better to build bigger ships or not. There was a natural push to build larger. After they were built, people realized that it was better. There’s a simple reason for this. The resistance to flow only increases as an area, while what you need to carry goes as a volume. The bigger you are, the less energy you need to carry a given amount of mass.

That played a major role in moving to steam ships and iron ships. That led people to believe that they could build enormous ships. So, here’s a real lesson for innovation and industry. Isambard Brunel was the greatest engineer of the 19 th century. He decided to build the biggest ship ever in 1870. Ships in those days were built by shipwrights who learned their trade qualitatively as apprentices. There was no mathematics or modeling. He built this ship, it was launched, and it took off and could not move even with the engines going.

One of the apprentice shipwrights said, “there must be a better way of doing this.” He invented the whole field of relevant variables around how ships move and how they get scaled up. That led to the whole innovation in ship building that allowed all modern ships to be built by modeling and to have an underlying mathematical formulation.

So, from the outside, it would seem to me that in discussing many of the things you’ve been talking about, until you have some sort of understanding of what the underlying principles and drivers and mathematics are—or if everything is a special case—you can’t make predictions from previous data. That was the problem with shipbuilding and eventually people got it straight.

Answer: Yes, that’s an extremely good point and one that I’ve wrestled with a lot. We would like to restructure the way we do the analysis. It’s very difficult to do that while the capital markets are boiling minute by minute. I’ve taken some of the concepts I’ve learned at SFI and thought about them: for example, choosing the unit of analysis to be the industry, not the company. I can say some things with more certainty about 40 companies in a sector than I can say about any one company. It’s easier to see losers than winners.

The comments, opinions and any forward predictions presented about any particular security, the economy and "the market" are based on the analysis of the speaker. These are not necessarily the opinion of, and should not be construed as a recommendation on the part of Legg Mason Capital Managment or any of its affiliates.