Bill Miller
CEO, Legg Mason Capital Managment

Michael Mauboussin
Chief Investment Strategist, Legg Mason Capital Managment

"Wrap-Up Session: Final Comments and Questions"

Michael: Let me leave you with several take-aways from this conference: weak signals, industry development, and super- vs sub-linear systems. Let me touch on these very quickly.

The first is the subject of weak signals. That term may sound unusual. It relates to trends that are very significant, yet are somewhat outside your consciousness today. For example, we heard today about multiplayer gaming and nanotechnology. One of my motivations in the little informal polls I took today was to get a sense of where we all are in our awareness of each of these phenomena.

The term “weak signals” comes from theories of ant foraging. Imagine an ant nest and a food source some distance from the nest, both in a laboratory. The scientists let the ants go out and forage. They quickly find the source of food, lay down pheromone trails that get reinforced, and the entire colony quickly zeroes in on the food source in an optimal fashion.

As the scientists watched these ants, they observed that periodically an ant would peel off from the pheromone trail for no apparent reason. It’s an odd thing because the ant is choosing to ignore the food source. As they studied this in more detail, they determined there is a mathematical probability that a particular ant would go off the path. They wondered why this behavior would be built into the ants. And then they realized that if another big food source showed up in the domain of the ant colony, the ants would ignore it or walk right by it even if it was close. So they had to periodically ignore the pheromone trail and strike out in a random search. Ants are programmed with a kind of “random generator” to explore even as they exploit. The explorer ant lays down a pheromone trail of its own, but it’s so faint that it’s called a weak signal.

In our world, by analogy, there are things going on that we may not be paying attention to because they’re not our usual “food source.” They’re not in our consciousness every day but at some point in the future, they may become significant, but by that time maybe other ant colonies will have exploited them. How do we hedge against this possibility? There are some simple techniques. One is to read very widely outside of your discipline. Another is to stay connected to very interesting people who have perspectives that you may not share.

The second large take-away for me had to do with the industry development model. I thought the way Sue framed this in the breadth versus depth concept was very helpful. We saw in several of the presentations that growth of any system follows an S-curve where it starts out slowly, undergoes rapid acceleration and then levels off. Each of these segments of the curve is accompanied by different features and the system exhibits different behaviors.

Think about the economics of customer businesses. If you have a subscription business, then early on you don’t have to worry about managing customer revenues because you have no customers. Your goal is to get as many customers in the door as possible. In the middle of the curve, you’re balancing getting revenues from new customers and keeping existing ones. If it’s a late cycle business, you’re focused on customer maintenance and issues like customer churn. Those types of changes may not be well recognized by many businesses.

The last idea has to do with systems that exhibit super-linear behavior vs. those that exhibit sub-linear behavior. In the super-linear systems, when you add more, you create more: there’s a positive feedback loop. One of the most direct uses of this in markets is network effects—the value of a good or service increases as more people use that good or service. The prototypical example would be eBay. They started out with their basic auction business which has very clear network effects. Then they moved into PayPal which is a payment business. It also derived benefits from network effects. Last week they announced the acquisition of Skype. Once again it’s a network effects business.

Sub-linear systems are concerned with maximizing efficiency per unit in industries that are relatively mature. Using capital to innovate may not be the best path for these types of companies: they should work on efficiency instead.

Those are some of the thoughts I’d take away. I’ll turn it over to Bill and see if he would add or subtract anything from those. Then we can open up to other questions or comments.

Bill: I have a couple of take-aways. Geoff mentioned that everything jumps to the same height—it has nothing to do with size. I like everything on the same scale, so the fact that a flea, a horse and a person all jump the same is great! [laughter]

From the stock selection standpoint, my favorite and underappreciated and powerful technique of all time is to find two stocks that are trading at the same price and compare them. Or to look at two companies that have exactly the same earnings and then see what they’re trading at. To take one example, if you look at HP and Dell, as it turns out, Dell and HP will earn down to the penny per share the same. But at the beginning of the year, Dell sold at twice HP’s multiple. This year Dell is down 20% and HP is up 30%. There’s about 5 points between them. This gets into return on capital, scale, figuring out which companies can grow faster and so on. Dell, for instance, is exhibiting Geoffrey’s scaling issues. Dell has efficiency problems. They used to always be hyper efficient and give the best customer service and the machines were shipped out in a day or two. Dell is at such a size that it is hitting some of these size constraints. One hypothesis we can draw from Geoffrey’s talk is that if you take a company that reaches a certain size, it will become sub-linear.

You can dissect the S-curve into two different segments. One part is linear and another part is super-linear. Our hypothesis is that Dell is at a point in the sub-linear range, so they ought to be focusing first on efficiency and second on growth. When large companies get into trouble, they always do the same thing: they scale down, or try to shift to a spot on the curve where they can be super-linear again.

One of the ideas that flows out of Geoff’s work is the increasing rate of change. Sue talked about the growth and acceptance of different inventions: cable television took 50 years to climb the S-curve; PC’s took 25 years, and the Internet took 10 years to reach the same level of penetration. Dick Foster noted that the pace of replacement of companies in the S&P has been steadily increasing as well. This is a feature of these kinds of systems that we can use to help make portfolio decisions. Dick also noted that people often make linear forecasts. There’s a natural assumption to have a linear view of the way the world is going to operate. But the facts are that there are few companies that don’t have some discontinuity over any given five year period. They’ll make some large strategic acquisition or spin something off. They don’t have the same form after that shift. 

As Josh was saying, there is a psychological S-curve at work as well. They tried to take Nanosys public and it was pulled due to market conditions. When they say, “pulled due to market conditions,” it means that the market conditions indicated that the investor who bought it might make money! [laughter] Josh is exactly right. In something like nanotech, there will be recurrent bouts of depression and euphoria like we see in every other industry.

Dick Foster showed us an analysis where he assumed that we had perfect foresight into a particular business’ growth path. He said in passing that if you had perfect foresight into a company that was following an S-curve, it would sell at an enormously high price-earnings multiple in the early years. The stock would give you a constant rate of return throughout its entire history. But if you invert that a little bit, most of the money in the growth companies is made when they look expensive. They look expensive, but on a theoretical basis they are not. When we look at Amazon or eBay or Google, we find that the justifiable valuations are miles above what looks to be a very expensive valuation.

Question: What about the structural deficit? What are the reasons why it doesn’t matter?

Bill: The short version is like this. When countries in the past have had a current account deficit equal to 6% of GDP their currencies have fallen. Argentina is an example. That is true, but those countries did not have a reserve currency. Moreover, there’s an argument that this can’t keep going up forever. That’s true. Chris [ Davis] mentioned that he tried that argument on Charlie Munger a few years ago concerning health care as a percent of GDP. He said, “Charlie, this health care as a percent of GDP goes up all the time. It’s an unsustainable trend.” Munger said, “that’s correct, it won’t go on forever.” Chris said, “when will it stop?” Charlie said, “who knows, but let me ask you a question: what is the right amount of money for the richest country in the world with an aging population and most of its needs met to spend on health care?” Charlie continued, “eleven percent doesn’t strike me as a particularly high number and health care is pretty important. Maybe it’s fifteen or twenty or twenty-five.”

Back to the current account deficit. People hit a magic 6% and the dollar is falling, but the two strongest currencies in the world during the period when the dollar was falling were Australia and New Zealand. They both had current account deficits higher than the US. So their currencies were the strongest in the world while their deficits were the highest. It’s no simple if-then relationship. 

The GDP is $11 trillion and the GDP growth rate is 5%. The stock of assets in the country is about $55 trillion dollars. The stock of assets grows at the rate of GDP. The growth of assets is about $2.5 trillion per year. We’re borrowing $600 billion per year to grow our assets $2.5 trillion per year. That sounds sustainable to me. That’s the same sort of thing as someone borrowing at 5% to buy a house and it appreciates at 7% a year. As long as our GDP is growing it won’t be a problem. If the GDP stops growing then the problem reverses. That’s just part of the story.

Question: How has the structure of the research department evolved?

Bill. It’s following Geoffrey’s scaling laws. It’s being optimized as it scales. [laughter] Approaching ever greater levels of perfection! It’s an important question. As our assets and the size of the organization grows, there are issues with respect to the scale and efficiency of the organization. Once you hit a certain point of growth, you either need to use different materials or re-architect the design or both.

Why is research structured with portfolio managers and analysts divided by industry? That’s the way it’s done because that’s the way it has been done. The big Wall Street firms structure their sell side people to fit into that model to create a nice flow of information. But it has no necessary relation to the way portfolio managers make decisions or generate the information they need to make those decisions.

What skill sets does someone need to follow an industry or a company? If an analyst follows a company or an industry, he’s vertically integrated. Probably, most people aren’t equally skilled at all of the tasks involved from analyzing the management of the company to understanding the financial position of the company. So perhaps research should be divided up in a more horizontal fashion. So we have created a market intelligence unit whose role is to increase the signal/noise ratio of all the different sources of information that come into us every day. We have a group of young analysts who read all of this stuff and organize it according to what we own. This saves our senior analysts a great deal of time.

We’re also creating a secular team of two or three people under Mark [Niemann]. He’s really good at long term strategy, which is what he did in consulting. We got him to specialize in what he’s really good at. When eBay buys Skype, Mark has already done a strategic analysis of eBay.

Question: What are three or four books you’ve read over the last year that you really like? And what concept have you held in your mind that recently you have realized is false? 

Bill. One of the maxims of investing is to do your homework, investigate and then buy the security. What we found is that this maxim is wrong in many cases. It’s a much better principle to invest and then investigate. Some firms do a lot of work analyzing a company and by the time they’re done, the stock is up 50% and they’ve missed the opportunity. There is something that triggers an experienced portfolio manager’s interest in a company that causes them to begin to investigate it. You don’t want the market to figure this out before the analysts get the work done! Since most companies are going to prices about right in the near term, your actual risk in buying first and then working out your thesis is relatively low.

Michael. Here are some of the books we’ve been looking at. Poor Charlie’s Almanack is a compilation of Charlie Munger’s background and speeches. I certainly enjoyed Freakonomics by Steven Levitt and Stephen Dubner. It’s an easy read but an important message underlying it is “incentives really matter a lot.” A new book by William Poundstone is called Fortune’s Formula. He talks about the history of the Kelly formula and weaves in Shannon’s information theory and Thorp’s blackjack playing strategies. The World is Flat by Thomas Friedman is worthwhile. We’ve also read a couple of books internally. One is The Landscape of History by John Gaddis. It’s about the craft of how to do history. A lot of the issues that historians face are the same ones that analysts face. The final one is Deep Survival by Laurence Gonzales. It’s about how people in very bad situations either succeed or fail and what allows them to do either. Many of the factors he discusses are neurological and psychological factors.

Bill. Fortune’s Formula doesn’t require much technical knowledge to read but is very important to understand. Kelly developed a formula that maximized the growth rate under any given set of conditions. In other words, there is a single best way to get rich given certain conditions and the formula will maximize that and get you there fastest. Or if you’re going to be losing money, you’ll lose it the slowest.

I also recommend The Road to Excellence by Anders Ericsson. It’s a compilation of papers that were written about expert performance. The psychologists studied the level of expertise of some of the world’s best experts: swimmers, chess players, violinists and so on. They wanted to understand the common factors of expertise, if any, that went across all of these domains. One of the results is that it takes roughly 10,000 hours to achieve world class excellence in any field. Everyone they studied had put in 10,000 hours of deliberate practice designed to increase their expertise. Mozart was taught to compose around age five. The first symphony that he wrote that was world class was when he was 17. Einstein published at 26 but had been studying physics systematically since he was 14 years old. This insight is helping us adjust our analyst training. People who are experts cognize things differently than people who are novices. Novices tend to deal more with attributes and surface features, with if-then statements and lots of examples. Experts instead have a conceptual framework that they order stuff in.

I also recommend Nature: an Economic History by Geerat Vermeji. It’s a dense book and you have to know a fair amount about both economic and evolutionary theory to get it. He argues that the principles of economics can shed great light on evolution and vice versa and you can make predictions about both realms by integrating the two frames of thought.

The last book is called Understanding the Process of Economic Change by Doug North. He won the Nobel Prize in 1993. This is a short book, but very densely packed with great ideas.

Question: When we come to these sessions, they’re very innovative. We try to go back and implement some of these ideas at our end. We don’t see everything like you do. When we were in Las Vegas, we loved some of the ideas from The Wisdom of Crowds and we took this back to our company’s investment committee. They now vote with voting buttons so that the ones who feel they can’t participate in light of the different experts in the room get a voice. Then we can chart the distribution of decisions. I’ll obviously need a few weeks to digest the thinking from today, although I’m looking to change the coating on our building so that it is self-cleaning! [laughter] Because you see so many of these ideas, how do you decide which ones you want to incorporate in your process? Would you say that the investment decision process you use is 50% different today than five or ten years ago? Or does it not really affect the way you select companies?

Bill. Our analysts all figure out things in different ways and we want that diversity. It’s up to them to figure out which ideas will fit best and work best. That’s what I do as well. For us it comes down to the “sentiment of rationality” and we look at an idea and think, “this is important—this will make a difference” and we implement it.

Michael. A lot of the ways we organize research reflects the philosophy of how the portfolios are run: low turnover, long term oriented, economically-based. Everything tees off of that. These features allow us to spend more time on thinking about some of the newer ideas from the sciences than other firms might. We had a couple of fantastic weeks of training in August and September and it reinforced a lot of these new ideas. We had a whole module on capital markets theory and the wisdom of crowds. When does the wisdom of crowds work? When doesn’t it work? We put in a lot of time on valuation techniques and on competitive strategy. We have lots of thinking on businesses, industries and economies. We also touched on behavioral finance using hands-on experiments.

Bill. The decision process is constantly adapting to circumstances, but the core remains the same: trying to buy things where we can earn a risk-adjusted excess rate of return above the benchmark. The methodologies that we use to do that have to adapt. The long term, low turnover portfolio is not a metaphysical choice. It’s based on evidence that it’s conducive to excess returns. We don’t want to confuse our objectives with our strategies. Our objective is to earn excess returns. A strategy is to have a long term, low turnover portfolio. If the evidence says that the strategy is not valid, we will adjust to that.

We’re in the process of adjusting something right now. There is evidence that concentrated portfolios outperform widely diversified portfolios. Unfortunately, the circumstances under which this evidence arose were very different from today’s circumstances. In today’s world, much more broadly diversified portfolios ought to perform better. Our portfolios will gradually expand. They will be concentrated compared to the rest of the world, but not as concentrated as they have been. Concentrated portfolios work best when valuation discrepancies are wide. The valuation discrepancies out there are not wide, so you’re better off spreading your portfolio across a wide variety of sectors because you’re not getting paid to take the volatility risk or the estimation error.

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.