Wednesday, March 30, 2005

Consumer confidence and the long-term investor

Yesterday evening, I noticed that Select Comfort had ended the day down about 5%. It wasn't obvious what had happened—there was no company news that day. But after a little while I remembered that the Consumer Confidence Index had taken a little dip for the second month in a row. This is a poll of consumers that is supposed to indicate how likely they are to spend money over the next six months. Theoretically, if consumers feel good about their current financial situation and their immediate futures, companies that sell directly to them will experience strong sales. This is especially true for makers of "big ticket" products such as Select Comfort.

If consumers are really less likely to buy mattresses over the next six months, we can expect a couple of less than stellar quarters. If this is the beginning of a recession, it could be more like two years of disappointment. So investors have good reason to worry—especially if they are going to need cash sometime soon. Short-term bonds will be considerably less volatile.

But I have a lot longer time period to consider. I bought Select Comfort because I beleive it will outperform the market over the next 10 to 15 years. I wouldn't be too surprised to own both a Select Comfort mattress and Select Comfort stock in 20 years. If I'm correct, I don't think a bad quarter or two will hurt all that much.

Tuesday, March 29, 2005

It pays to vote

When I got Canon's proxy for the 2004 annual meeting, I read that the company wants to add an item to their by-laws:

The objects of the Company shall be to engage in the following business: ... (8) Manufacture and sale of pharmaceutical products

Today I found out that Canon is hoping to sell DNA chips which can be used to diagnose patients. From the press reports, it sounds like Canon will be able to use bubble jet technology to print the chips, which could cost far less then current methods. If successful, these products could help doctors arrive at more accurate diagnoses and researchers understand the genetic code.

While this is exciting news for investors, there's really no way to know how much future earnings this might generate. Most likely this will never account for more than a few percent of sales and may never become profitable. But it's fun to imagine Canon becoming a bioinformatics company at least in part. Certainly it suggests that Canon's R&D department is functioning well.

Actually this news is potentially negative as well. Copy machines, printers and digital cameras are increasingly competitive and low-margin products, in which Canon has remained one step ahead of everyone else. But as airline investors know, being number one isn't always rewarding. So Canon is compelled to find new products (such as flat-panel screens and portable printers), which is a riskier strategy than the cost-cutting that marked the last ten years.

For the next five years, the company is beginning phase three of its "Excellent Global Corporation Plan". Now, it will focus on "Healthy Expansion". Investors can dream that the phrase is quite literal.

Monday, March 21, 2005

Signing J. D. Drew

I was thinking this weekend about the Drew signing and I wondered how much the Dodgers expect to pay for each win. For teams with limited resources (everyone but the Yankees), a major goal of acquiring players is to minimize the dollars spent per win. Actually, as Doug Pappas pointed out, the goal is to minimize the marginal dollars above the league minimum per "marginal win". To do better than last year, the Dodgers must spend less than $1.8 million/marginal win. Drew is earning $11 million with the Dodgers, which is $10.7 million more than the league minimum. Therefore, the Dodgers are getting a bargain each year Drew nets more than 5.9 marginal wins.

Baseball Prospectus says that Drew added about 10 wins above a replacement player last year, which would translate into a "fair" salary of about $18.7 million. In 2002 and 2003, he added only 3.7, which would be a hair below $7 million. In his six full seasons, Drew has averaged 5.3 WARP, which would be worth about $9.8 million. According to the Win Shares method, Drew was worth about 11 wins last year, 4 the year before and 5 in 2002. The methods more or less agree given that Win Shares don't adjust for marginal value.

Below is a table listing the "fair" salary in millions of dollars if marginal wins are worth $1.8 million plus $300,000 for the league average:

Year Age   WARP  Salary
1998  22    1.3     2.6
1999  23    2.7     5.2
2000  24    4.6     8.6
2001  25    7.1     9.8
2002  26    3.7     7.0
2003  27    3.7     7.0
2004  28   10.2    18.7
And here is a peek into the future:
2005  29      ?    11.0
2006  30      ?    11.0
2007  31      ?    11.0?
2008  32      ?    11.0?
2009  33      ?    11.0?

Ultimately the contract depends on what gets filled into those question marks. If we assume Drew has two years like 2004, the Dodgers will get a phenomenal value, but will probably loose him to another team. If Drew has typical seasons, the Dodgers will be slightly overpaying, and will probably keep the rest of the contract. If Drew gets injured, they better have their insurance paid up. It's the first and third scenarios that cause the biggest concerns.

Clearly this is a good contract for Drew, since he's guaranteed $55 million and has an option for more. But it doesn't mean that it's bad for the Dodgers. In particular, the team seems willing to take on some of the risk of losing Drew to injury in exchange for the possibility of two great years or five good years.

Actually, it's possible they could get four great years. The option may only be exercised at the end of the 2006 season, so if Drew misses part of that season to injury, he wouldn't be likely to find a better contract no matter how good he might be in 2005. On the flip side, he might blow his knee the first week of 2005 and be dead weight on the payroll for the next five years.

I think it's misleading to think of the option as more or less the same thing as an option in the financial sense. For one thing, the Dodgers are getting a non-monetary return on their investment. Sure, if Drew helps the Dodgers to win some games, it will help them sell more tickets, concessions, and advertising. But the effect is indirect. A great year might help the Dodgers win the World Series, which would make the contract exceptionally valuable, even if Drew was a bust after that. Another difference is that there is no market for player options. The only way to benefit from it is to exercise it or use it as leverage to renegotiate the contract.

Wednesday, March 16, 2005

I'll come in again.

Besides market share, total market, and new markets, a company can also grow revenue by raising prices. This of course is why it's good to be a monopoly. There is a limit to how high Microsoft can set prices, because eventually people will get fed up and leave for something cheaper. Another way for software companies to grow is to release new versions, which works for a while. But if there is any way users can continue with the old software or switch to a cheaper option, they will when they notice how often upgrades come along.

For a long time, Microsoft didn't have to worry about people refusing upgrades, since they ended up buying them each time they bought a new computer. And Microsoft could justify raising the price by adding features like web browsers, music players and CD burning software. Recently it has been harder to do these things because of court rulings that limit Microsoft's use of its monopoly.

One advantage Oracle has is it's subscription model. Rather than charging customers to get the latest version, Oracle gives it to everyone who has a support contract. Microsoft is slowly working toward that system with institutions, but it will take a much longer time with individuals.

The basic point remains: Oracle still has more room for growth than Microsoft.

Friday, March 11, 2005

The difference between Oracle and Microsoft

A companies revenue may grow in three ways: the total market may increase, a company's market share may increase, or it may begin operating in new markets. In its early days, AT&T had a monopoly on telephone systems, so the focus was on increasing the size of the telephone market by installing more lines. Coke and Pepsi have fought over a relatively constant pie. Berkshire Hathaway has become a huge company in part by buying companies in a variety of markets. Raytheon made its name (literally) in vacuum tubes and then diversified into microwaves and missile systems.

For most companies, total market size is outside of their control and expanding to new market, either organically or through acquisitions, is risky. So one of the most reliable indicators of future growth is the increase from one year to the next in market share. This is especially important in industries where it is possible to establish a long-term advantage which isn't impeded by growth. For instance, Wal-Mart has built a strong reputation on being the low-cost leader in nearly everything it sells — an advantage that won't quickly disappear. In fact, its growth helps it put price pressure on the companies that sell to Wal-Mart thereby increasing the advantage.

Some industries resist monopolies. For instance, restaurants must carefully balance growth with the danger of over saturating the market. I know people who would gladly eat every meal at McDonald's, but most of us prefer variety. It's no surprise that there are only four Spago locations, but ten times as many "Wolfgang Puck" cafes. These businesses scale with difficulty.

Software is an industry that scales very easily. More customers mean more developers, which means more features, reliability, and performance, which virtuously cycles into more customers. Incremental costs are vanishingly small. Bigger is better in nearly every way. Which brings us to Microsoft, the biggest, richest company in an industry that rewards success like no other. When I was considering buying Oracle a few years ago, I worried most about Microsoft. It seemed to me that with all its cash, brand recognition and legions of developers, no software company could be safe. Oracle had held back the tide in the database market, but for how long?

The final chapters have not been written yet, of course, but I think Oracle will continue to have better growth for years to come. Think about the ways that a company can grow: market share, market size and new markets. Currently Microsoft has 85% of its core "client operating system" market. The market is growing, but not at the same rate as it did during the 1990s when most families bought there first computer. It's difficult to know how successful Microsoft will ultimately be in some of its newer endeavors, but many of them are not strictly software (Xbox, MSN, hardware, etc). The recently instituted dividend serves as a clear signal that Microsoft doesn't know what to do with all the cash it generates.

Meanwhile, Oracle controls about 40% of the database market. Picking up market share may be a struggle, especially since customers are reluctant to switch once they have made a purchase. But Oracle enjoys all of the advantages that the biggest enjoys in the software industry. Microsoft is still far behind technically and in terms of brand. There are fewer barriers to overall growth in the server market than the client side, since growth depends more on business needs than on new PCs. Finally, Oracle has already seen success expanding into new markets with products that are closely related to its database product. PeopleSoft will help tremendously, simply by bringing in more customers.

Monday, March 07, 2005

Mono-linked chains

Warren Buffett just released his annual letter to shareholders, which explained the results of Berkshire Hathaway in 2004. He commented on a $579 million loss in a zinc mining loss:

Our failure here illustrates the importance of a guideline — ­ stay with simple propositions —­ that we usually apply in investments as well as operations. If only one variable is key to a decision, and the variable has a 90% chance of going your way, the chance for a successful outcome is obviously 90%. But if ten independent variables need to break favorably for a successful result, and each has a 90% probability of success, the likelihood of having a winner is only 35%. In our zinc venture, we solved most of the problems. But one proved intractable, and that was one too many. Since a chain is no stronger than its weakest link, it makes sense to look for —­ if you'll excuse an oxymoron —­ mono-linked chains.

This is such a good analogy. I also think it reflects an investment philosophy I didn't know I held until now. My first (and one of my favorite) investments was Oracle. Over the years, it has taken some huge, bet-the-company risks. But in every case (at least from what I remember reading Softwar) the risks have involved one, key variable. For instance, each of the complete software rewrites (which are very risky) pivoted on having a better product at the end of the processes and little else. Time and again Larry Ellison bet Oracle the company on Oracle the database. Buying PeopleSoft, in fact, will have been a successful risk if Oracle manages to move a wide variety of applications onto a single database schema.

Select Comfort is also a one-link chain: the adjustable airbed. Like Oracle, the company has focused on one, big idea — the hedgehog approach. Since Sleep Number beds can fold into a handful of boxes, they are shipped directly to the customer. Since most people aren't familiar with the idea, Select Comfort stores, commercials, QVC shows, and the website are geared toward showing how the beds improve a person's quality of life. Since the beds are made to order, customers have a range of choices about which controls, mattress covers, air pumps, pillow tops, sizes (including the Grand King (80" x 98") designed for 6'11" Kevin Garnett), and foundations. There are any number of things that can go wrong, but as long as it's possible to sell high-quality air mattresses, Select Comfort is probably going to do alright.

Raytheon is moving in the direction of being a single link. It used to sell everything from microwave ovens to Patriot missiles. At some point, this made sense — microwaves used to be high-technology. Lately, Raytheon has been selling off or ending businesses that don't fit into the government contracts umbrella. Increased dependence on one customer might seem foolish except that the customer is schizophrenic. It still has too many links, but at least management has learned its lesson from the diversification adventure.

I have a harder time justifying Canon on the "one-link" criteria. I'm tempted to say optics excellence, but one of Canon's biggest revenue streams is printer ink. Fujio Mitarai, Canon's CEO, says, " Canon was built on the foundation of original innovative technology." But I think Canon's success is due to a uniquely Japanese ability to mold thousands of employees to a single philosophy. Many of Canon's investor relations documents describe the idea of "kyosei", which is translated "living and working together for the common good". If so, the risk is that this bubbly optimism might fail to produce new and better products at a reasonable expense.

Maybe it's naive to think that Canon has really implanted kyosei into its culture. But look at this quote in a press release of the death of Sony's founder:

"Mr. Ibuka has been at the heart of Sony's philosophy. He has sowed the seeds of deep conviction that our products must bring joy and fun to users. Mr. Ibuka always asked himself what was at the core of 'making things,' and thought in broad terms of how these products could enhance people's lives and cultures.
Or this quote from the Toyota Forklift division:
At Toyota facilities around the globe, "kaizen" is a word mentioned frequently. The word means "continuous improvement" and is a key factor in Toyota quality. Kaizen has been incorporated into the Toyota Production System driving our engineering and manufacturing teams to constantly improve our lift trucks. It also drives our service personnel.

Coming full circle, Berkshire Hathaway is a horrendously complicated company rivaled only by GE. Any attempt to evaluate the company would require intimate and detailed knowledge of a company that sells everything from candy to catastrophic re-insurance. Ultimately, Berkshire investors are risking their investment on the philosophy described in the Berkshire "Owner's Manual".

Thursday, March 03, 2005

If at first you don't succeed...

I've been playing around with PerlTeX, which is an ingenious little system to typeset beautiful documents with the power of Perl. I was inspired to use it because I wanted to illustrate various interest rate calculations without having to do a lot of work updating numbers. For instance, how much will $1000 be worth in a year if it earns 5%? And in two years? And in 5? What about if it earns 6% compounded monthly? None of these is hard individually, but it's tedious and error-prone to do the work by hand.

TeX is the premier typesetting system for mathematics, but it is worthless for performing calculations. I spent several hours trying to work out how I might coerce its macro system to print the result of 365/21. PerlTeX lets me write macros in Perl, so I could calculate Future Value (FV) thusly:

  $_[0]*(1 + $_[1])**$_[2];


But this doesn't work:

because ".05/12" isn't evaluated. Next I tried a very simple command:
  eval qq{@_};

but this turned out to be hopeless since TeX refused to expand \perleval first. (This might be a solvable problem, but I really don't want to know more than I already do about TeX macros.)

I knew I was on the right track, but I had to sleep on it (and take a shower), before I could hit upon the solution. I needed to do even more in Perl and only make TeX do typesetting:

  sub FV{
    $_[0]*(1 + $_[1])**$_[2];

\perleval{FV(1000, 0.05/12, 12)}

\perlinit is never called directly. It's sole purpose it to load definitions into the Perl namespace when PerlTeX starts up. \perleval does the work of evaluating Perl expressions. This is my original solution turned inside out.

Now I wanted to round to the nearst cent and add a dollar sign:

\perleval{sprintf '\$%.2f', FV(1000, 0.05/12, 12)}
But TeX swallowed everything after the % as a comment. At this point, the answer was clear:
  sub dollar_fmt {sprintf '\$%.2f', $_[0]};

  sub FV{
    $_[0]*(1 + $_[1])**$_[2];

\perleval{dollar_fmt FV(1000, 0.05/12, 12)}