Why I sold Canon

I'm pretty far behind in updating my transaction diary. Partially that's because I don't want to think about the carnage done to my portfolio in recent weeks and partially because I got busy with other things (i.e. buying a house). Because I need some cash to use as a down payment, I sold off my Canon shares at a price less than what I consider to be their value. On August 6, I got $46.49 a share. The first lot (bought in December, 2003) returned 62% compared to 6% for the S&P 500. The second lot (bought in December, 2004) returned 47% compared to 7%. The annualized return was about 11% for both lots. Needless to say, both lots were excellent investments.

Canon has lost a lot of market cap since I sold, so I lucked out there. Since Berkshire had about the same price to value ratio, I was tempted to sell it instead. But I resisted in part because I assumed consumer electronics will be harder to sell in the next few years and Berkshire will be able to pick up some good deals over the same time period. Like Oracle, I suspect Canon will be a compelling value and on my investment radar in the future.

Price/Value ratio

Reading the latest Longleaf Partners Funds report, I was inspired to calculate the Price/Value ratio for my holdings:

Company          P/V
-------          ---
Canon            86¢ 
Select Comfort   36¢ 
Berkshire        82¢ 
Sally Beauty     59¢ 
First Marblehead 23¢ 

I recently sold Oracle for somewhere between 90¢ and $1 to the dollar. Cash is always worth $1 to the dollar and I used the same rate for Alberto-Culver, since I haven't put a value on that company. My composite P/V for the portfolio is roughly 66¢ to the dollar.

Select Comfort and First Marblehead still seem insanely cheap to me even after slashing my value estimate. I expect these will be truly outstanding investments for those who purchase today, but both have been classic value traps for me. (A value trap is an investment that looks cheap, but whose value falls as fast or faster than the price.) First Marblehead in particular has been a head-scratcher, since it operates in a great business that has been abandoned by other companies due to short-term problems. Both companies now include a free option on any future growth.

Of course, the value portion of the ratio is my conservative estimate of the present value of all future earnings. Further, there's no way to know when or if the price will converge on the value, assuming I estimated it correctly.

Luxury Goods

So here are three things I read today:

• The Rich Drink Better Beer, Not More
• There's a new thing called coolspotters that tracks what products celebrities endorse.
• Consumers are incredibly stingy, except when it comes to luxury items.

Of my investments, I'd say three and a half of them less luxury goods to consumers. Canon sells top of the line cameras (as well as high-quality point-and-shoots), Sally Beauty sells salon-quality supplies, and Select Comfort sells premium mattresses. Alberto-Culver has mass market products that are more upscale than average in drug stores and Wal-Mart. The advantage these companies share is pricing power. Basically in hard times, like now, they can decide to drop prices a bit to keep up sales volume, or they can wait around until times get better and raise prices. I have to be honest: I'll probably pay 10% more to get a Canon product over any of the competitors. It might seem irrational, but I think it's justifiable because I know what I'm going to get and I know I'll be happy. I'm sure the same is true of women who buy certain brands of shampoo or soap.

More good news and bad news

This week, my portfolio had good news and bad news. The good news is that Canon paid their year end dividend. Since the exchange rate has fallen to about ¥100 to the dollar, the ¥60 dividend worked out to be 60¢ a share. Canon's dividend yield is about 2%, but based on my original cost basis, I'm earning closer to 3%. As long as Canon continues to raise its dividend, I will be happy to hold my shares.

The bad news was that TERI, the non-profit that First Marblehead uses to insure its loans, declared bankruptcy. Now I believe the bankruptcy is for technical, not fundamental reasons, and I think the effect on First Marblehead will be very little in the long run. But my position has been battered to a considerable degree and perhaps permanently. At the very least, the news makes an immediate recovery very difficult.

At no time have a felt that First Marblehead was a bad risk/reward proposition at the current price, so in one sense I don't feel I made a mistake. But I did ignore one of my fundamental sell signals: to get out when a dividend is cut or lowered. If I'd done that, I would have saved myself a lot of money and aggravation. Further, there will often be an opportunity to buy the shares back at a later date when I've had a chance to analyze the company independent of the dividend.

At the moment, this sell signal only applies to Canon and my token position in Alberto-Culver. Which reminds me: selling Alberto has easily been my most costly decision to date since it freed up cash to buy First Marblehead.

2007 Look-through earnings

As usual, I have to wait for Warren Buffett to release Berkshire's earnings before I can tabulate mine:

EPS              2008* 2007  2006  2005  2004  2003  2002
Oracle           0.15  0.27  0.13  0.21  0.30  0.44  0.34
Canon            0.37  0.31  0.28  0.49  0.34  0.08 
Select Comfort   0.18  0.19  0.19  0.27   
Berkshire        0.22  0.30  0.25    
Alberto-Culver   0.02  0.02  0.08    
Sally Beauty     0.05  0.04  0.00    
First Marblehead 0.41  0.14     
Look-through     1.00  1.13  0.93  0.97  0.64  0.52  0.34

* 2008 numbers are consensus analyst estimates.

I keep track of my IRA like an open-ended mutual fund and this is the look-through earnings per "share" of my IRA "fund". As I buy and sell stocks, my portion of their earnings fluctuates and when I add cash, it alters the percentage of portfolio's total value comes from look-through earnings. So when I sold Oracle shares over the year, I reduced the earnings I give myself credit for and when I bought Select Comfort, I increased my share of earnings.

Thanks in very large part to Oracle, my look-through results actually improved. When you add in call option premiums and capital gains on selling shares, my results are even better. But my relative share of the company has been reduced and I won't get anywhere near those returns in 2008.

My two troubled positions look ok in this table, but that is mostly an illusion because I've increased my holdings to a large degree. In his just released letter, Mr. Buffett lays out four criteria he looks for in buying a business: "a) a business we understand; b) favorable long-term economics; c) able and trustworthy management; and d) a sensible price tag." The CEOs First Marblehead and Select Comfort have earned my respect anew by taking voluntary pay cuts for poor results that are largely out of their control. Also, the stock market has cut share price of these companies from cheap to practically free, in my opinion. The reason in both cases is largely a result of worsening economic conditions. In both cases, there are internal changes that need to be made if the companies are going to thrive, but they continue to have advantages compared to competitors that are not likely to disappear. If I weren't already up to my ears in these companies, I'd be buying at these prices.

Canon and Berkshire continue to earn about what is to be expected. They are both too large to grow quickly, but have very wide and clear moats that ought to preserve the businesses for decades to come. Unlike Oracle, the market has not come close to recognizing these company's intrinsic values, so I have not been tempted to sell.

Sally Beauty earned very little this year because it has needed to pay so much in interest expenses since splitting with Alberto-Culver. This year and next ought to be pivotal for the company, so it's very encouraging that directors have bought $3 million of shares to the $1 million worth they purchased with their own money last year. When the restrictions on selling agreed to by the principals of the spin-off transaction expire at the end of the year, I expect management will begin to trumpet business growth instead of underplaying it. I've noticed there are plenty of mom-and-pop beauty supply shops here in Southern California, and I expect there will be plenty of opportunities to consolidate the industry while paying down the debt.

Interestingly, 2007 looks similar to what analysts predicted last year, but that result is misleading because I'm more invested in these stocks than I was at that time. Looking at operating earnings, which includes various cash returns and costs, shows a fuller picture of my results:

Interest    0.01    0.03    0.16    0.01    0.03    0.02    0.00
Dividends   0.00    0.08    0.06    0.08    0.03  
Costs      (0.01)  (0.14)  (0.19)  (0.04)  (0.06)  (0.22)  (0.12)
Arbitrage   0.00    0.50    0.41    
Options     0.07    0.14     
Operating   1.07    1.79    1.37    1.01    0.65    0.32    0.23
Gain      -38.18%  25.76%  35.66%  56.70% 100.59%  41.50% 

I'm fairly pleased with these results, but I can't expect them to continue into the future. In particular, I likely will not have any arbitrage earnings this year, since I've invested most of my cash into businesses that I feel are too cheap to pass up and which might not pan out for a few years. Finally, here are my net results juiced by large realized gains that will not be repeated this year:

Realized Gain           2.88                            1.79 
Special dividend                2.99    
Net              1.07   4.60    4.37    1.01    0.65    2.11  0.23
Gain           -76.79%  5.41% 331.08%  56.70% -69.39% 827.21% 

Four types of moats

Recently on the Motley Fool, there was an article highlighting four different moat types:

1. Economies of scale
2. The network effect
3. Intellectual-property rights
4. High switching costs

Personally, I'd divide IP rights into: patents/secrets and brands, which is to say: what you know and what your customers know. Also, economies of scale and network effect are really two sides of the same coin. Size can give a company advantages or it can give customers advantages or both. I think high switching cost has an analogue as well: monopoly. Regulated monopolies are especially impervious to assault.

1. Size
   1. Economies of scale
   2. The network effect
2. Knowledge
   1. Intellectual-property rights
   2. Brand
3. Stickiness
   1. Monopoly
   2. High switching costs

In each case, the moat is developed by created an advantage that other companies can't steal. Bigger companies, like bigger ancient cities, are more likely to have established moats. For instance Oracle, Canon and Berkshire Hathaway have multiple and deep moats in nearly every category. First Marblehead mostly fends off competition with intellectual property. I'm worried that Select Comfort might have lost their primary moat—their brand. Those are much smaller companies that have not completely staked out their territory.

So for smaller, fast-growing companies, the question is what moats can they develop?

2007 Year in review

On the last day of 2007, my IRA ended the year down 13.3%, which was the first down year I've had and substantially worse than my benchmarks, the S&P 500 and Berkshire Hathaway. These things happen and especially with an ultra-concentrated portfolio. Here are my core positions:

Stock               2007 Return
-----               -----------
Alberto-Culver      20.83%
Berkshire Hathaway  29.19%
Canon              -17.74%
First Marblehead   -87.43%
Oracle              34.23%
Sally Beauty        16.03%
Select Comfort     -57.70%

Select Comfort has been the biggest disappointment of my short investment career. I certainly misjudged the business though I still think my initial purchase was a good decision. I now believe my follow-on purchase last year was a mistake, because I did not recognize the danger of air mattresses becoming a commodity. Select Comfort is built from the ground up to be a specialty bedding company, so if it ever needs to compete on price, quality and service alone, it must be revalued. That said, I think the current price is actually less than what the company would be worth as a commodity manufacturer. So any future turn-around comes as a free option at prices less than about $7 a share. As I mentioned when I made my third purchase, I plan to aggressively sell covered call options until the future becomes more clear.

First Marblehead has always looked stunningly cheap to me. Incredibly, the price has dropped to just over book value because of worsening conditions in the student loan paper market. Basically the market assumed for a while that the company would just close up shop. Since the supply (or from the perspective of students, the demand) of private student loans is growing at breakneck speed, walking away from the business would be crazy. Instead, First Marblehead has entered into an agreement with Goldman Sachs that will allow it to hold the loans it currently sells off in exchange for nearly 17% of the company's equity. I haven't had time to dig into the details of the deal yet, but it does seem like First Marblehead simultaneously removed short-term risk, reshaped its business model, and bought a powerful ally with a vested interest in its success. Buying more shares is a definite possibility, though I don't like the message sent by the dividend cut.

Canon became cheaper in part because of a delayed entry into the TV business due to patent problems. In the meantime, the company's core camera and printer products have sold well and profitably, and it is working on other entries into the display business. The dividend for 2007 was raised another 10% without seriously eating into cash flows. Canon's dividend is important because it is a signal from management that the business is doing well and it provides me with another reason to keep holding. Based solely on the dividend, Canon is trading below its fair value.

Outside of these three stocks, my investments performed quite well. Unfortunately, my losers made up a larger portion of the portfolio than the winners did. Options and arbitrage transactions worked extremely well for me on the whole, but I'd have to dramatically increase my trading activity to come close to making up for any one of the losing positions. On the other hand, my returns would undeniably be worse without these small, short-term, trading successes. Along the same lines, Alberto-Culver helped, but is a portion of my portfolio too tiny to profitably sell. Its performance barely matters.

Sally Beauty meets my current expectations. Everything seems quiet at the moment, but that will change as the company pays down debt and the restrictions on insider sales expire over the next year or so. I'm contemplating increasing my exposure in what amounts to a publicly-traded, private-equity investment (if you can imagine).

I'm in the process of wishing Oracle, my first and most successful investment, a fond farewell. On an annualized basis, my return on shares sold in 2007 has been over 20%. I've decided to end this investment because I believe the company is trading at a fair value. For the last few months it seems that Larry Ellison agrees with me as he has been exercising options for and selling a million shares a day. He has plenty of shares left to keep his financial future firmly tied to the company he founded, but I'm guessing he is more excited about other investments such as NetSuite. I'll keep my eye on the price in case it falls below its fair value again, however.

Berkshire Hathaway remains the anchor of my portfolio for the foreseeable future. This year will almost certainly be the moment when the company can finally use its dry powder. There are certainly plenty of quality assets available for pennies on the dollar due to "lack of liquidity" (i.e., over-leveraged entities that can no longer refinance). No doubt we will see some buys in the months to come.

Perhaps I'm foolish, but I feel fairly optimistic about 2008. Besides Select Comfort, the portfolio has improved financially and the businesses are stronger than ever. I have no urgency to sell until the future becomes clear, so the market price isn't all that important in the short run. Further, lower prices for stocks in general ought to present me with better opportunities for future purchases.

Value potential energy

Recently Vitaliy Katsenelson was interviewed by Philip Durell and Bill Mann. I picked up a few ideas that apply to my investments.

Canon

In the interview, Mr. Katsenelson pointed out that international investing may actually reduce risk in an otherwise US-based portfolio. When I first bought Canon, hedging against a dollar decline factored into the decision. Since then as the dollar has strengthened from about 105 ¥ to 109.61 ¥ with a lot of fluctuation in the interim. Canon has been increasing their dividend which yields about 2.7% compared to the Japanese 30 year bond that yields 2.27%. If (when) Japan raises rates, Canon's price measured in yen will likely go up to push its yield down. At the same time, the yen dividend will likely be more valuable in terms of dollars. Since Canon has a very strong product line, the stock is very close to a sure thing.

Another point the interview touched on is that the address of corporate headquarters doesn't tell the whole story of what countries a company is exposed to. While Canon is largely a Japanese company, it sells products all over the world and has manufacturing and R&D facilities spread around Europe, the United States, and Asia.

First Marblehead

I'll just quote Mr. Katsenelson:

Another one, and I know you guys both like is First Marblehead. This stock trades at what, eight, nine times earnings? It has a phenomenal growth rate ahead of it and I think the investors still put it into the subprime mortgage category, even thought the average FICO score of its portfolio is 714, which is very high; 83% of it loans co-signed.

The part that I love about it [is] this whole speculation about major customers JPMorgan and Bank of America going away [and] you can quantify that easily. You can figure out what impact it would have on the portfolio if both Bank of America and JPMorgan dropped First Marblehead and actually I figured it out and kind of my worst case, a year after JPMorgan dropped; if JPMorgan and Bank of America leave First Marblehead, its revenues would be up 20% or 30% over where they are today. So my downside is basically none.

You could argue that the margins may become compressed, but that the JPMorgan and Bank of America business is growing so fast that it should overcompensate that. I know you guys will agree.

Potential energy

Quite a bit of the interview was actually about the importance of price when it comes to investing. I starting thinking of it like the potential energy in a spring or a hot air balloon. In Active Value Investing, Vitaliy Katsenelson suggests the QVG framework for examining investments. Quality (Q) could loosely be tied to earnings, Value (V) is related to price, and Growth (G) is how earnings (or cash flow) are likely to change in the future. (This is far too simplistic I'm sure.) To go back to the balloon analogy, price is the altitude the balloon floats at, earnings are the buoyancy of the balloon and growth is how fast that buoyancy is changing.

Now investors looking at the balloon from outside try to guess where it will be floating over some period of time. As management dumps ballast (cuts costs) or adds heat (increasing revenues), the balloon ought to rise. If it can't for some reason (investors holding the price down), the potential energy increases. The altitude (price) is fairly easy to see, but the buoyancy and its rate of change (earnings and growth) are much harder to judge. As a result, lower altitude (price) might actually be the best signal to buy assuming buoyancy (earnings potential) is increasing.

The surrounding environment contributes to the potential energy as well. A balloon will be more buoyant on cold day, while stocks have the greatest potential in markets that have low P/E ratios. The inverse of P/E ratios, earnings yields is a direct measure of potential energy. Since P/E ratios for the market as a whole are trending down, prices won't be helped as much as they once were, so it's important to have good earning yields in the individual stocks you buy and own. Canon (6.78%) and First Marblehead (13.28%) are currently have the most potential energy in my portfolio.

Earnings yield of my IRA

Currently, my IRA is flat on the year compared to an 8% or so gain for the S&P 500 and a 17% gain for Berkshire. Select Comfort (-37.5%) and First Marblehead (-28.25%) are the primary culprits, though Canon is somehow down 11% on the year too. Now I don't like the idea of seeing my portfolio stagnate, but there is a ray of hope here: my IRA's earnings yield is improving.

Here's how I calculate yield for my portfolio. Each quarter, I multiply the EPS for each company I own by the number of shares I hold at the end of the quarter. I add up those numbers and at the end of the year I have a value for "look-through" earnings. That's how much my stocks would have returned if they had paid out 100% of earnings in the form of a cash dividend. For 4^th quarter earnings, I use analyst or company estimates which are decent first-order guesses. Then I divide by the total value of my portfolio to get look-through yield.

Yield         2007   2006   2005   2004   2003   2002
-----        ------ ------ ------ ------ ------ ------
Look-through  4.57%  3.90%  5.22%  3.65%  3.20%  2.59%

Since my account value has not changed significantly since the beginning of the year, the increase in yield is entirely due to increases in my companies' earnings. This measure does not include income from cash in the numerator, but it does include cash in the denominator. That means, cash-heavy portfolios are penalized. One solution would be to subtract cash from the denominator. A better solution is to add in interest earned to the numerator. I also add in premiums from call options, profits from short-term arbitrages, and cash dividends, while subtracting commissions, fees, and option losses. This produces a sort of operating earnings yield:

Yield         2007   2006   2005   2004   2003   2002
-----        ------ ------ ------ ------ ------ ------
Look-through  4.57%  3.90%  5.22%  3.65%  3.20%  2.59%
Operating     6.21%  5.76%  5.46%  3.70%  1.97%  1.74%

I know this double counts cash dividends, which are also reflected in look through earnings. Notice that my cash position has added to earnings in the last two years thanks to a number of arbitrage opportunities.

Higher yields indicate a sort of potential energy for a portfolio. Like the spring in a windup toy, increasing earnings give a portfolio a chance to run. Over a long period of time and given more or less efficient markets, an increase in earnings would represent a corresponding increase in price. Imagine what would happen to a company that earned 50¢ a share and sold for $10 were to increase earnings to $1 a share. If the yield remained at 5%, the stock would also double to $20. But until those gains are realized by selling the stock, that $10 a share increase will not be released. In order to calculate the effects of buying low and selling high, I add in realized gains and special dividends (like the one Sally Beauty distributed last year).

Yield         2007   2006   2005   2004   2003   2002
-----        ------ ------ ------ ------ ------ ------
Look-through  4.57%  3.90%  5.22%  3.65%  3.20%  2.59%
Operating     6.21%  5.76%  5.46%  3.70%  1.97%  1.74%
Net          12.84% 18.31%  5.46%  3.70% 12.92%  1.74%

Unfortunately, the net yield is extremely choppy. The simplest solution is to take the geometric mean, which is the best way to get an average of rates or percentages:

Yield         2007   2006   2005   2004   2003   2002
-----        ------ ------ ------ ------ ------ ------
Look-through  4.57%  3.90%  5.22%  3.65%  3.20%  2.59%
Operating     6.21%  5.76%  5.46%  3.70%  1.97%  1.74%
Net          12.84% 18.31%  5.46%  3.70% 12.92%  1.74%
Geomean       6.89%  6.08%  4.62%  4.36%  4.74%  1.74%

This smooths the data to show that net yields are also creeping up for my portfolio.

So what does all this mean? In my opinion, these yields are a rough estimate of potential. As I make good investments in companies that are cheap and have high earnings, my portfolio potential goes up. When I'm able to make money with the cash potion of the account, as I have over the last few years, I increase my portfolio's potential a bit more. When I harvest some of that potential by selling positions or receiving special distributions from a position, I have a chance to reinvest in companies with increased earnings potential. As I make good choices in allocating assets, my portfolio's yield and potential increase.

It's important to look at measures besides stock value when considering changes in investments. At the moment, First Marblehead and Select Comfort have the highest earnings yield, but are my worst performers. Oracle has the lowest yield, but is also one of the bright spots in terms of price performance. A narrow focus on recent price movement (momentum investing) would lead me to cut my losers and ride my winners. But as a contrarian investor, I'm looking to sell Oracle (low potential) and buy First Marblehead (high potential).

Why I bought more shares in First Marblehead

This is as close to a no-brainer as I've seen. Six months ago, I thought First Marblehead was cheap and now its business has improved and its price has dropped. Recently, the company raised the dividend 10% (which is on top of a 67% increase in June). Even if the dividend never goes up again, the dividend discount model suggests a value of $33 1/2 or so. A modest 10% increase for the next 5 years followed by no further increases is worth about $57 1/2 according to my model. According to the Quicken DCF model, Marblehead only needs to grow at 1.8% to justify its current price.

The only thing that has been missing for the last few months has been cash to augment my position. Thankfully, I was able to lighten up on Oracle just in time to collect a larger dividend on my shares this morning.

Update September 24:

I found another opinion of First Marblehead that echoes my feelings and assigns a range of values to the company. Even under the worst case when everything falls apart for the company, the author assigns a price per share of $42. I've found that my best investments occurred when I couldn't figure out how a company could be valued as low as the market price. For instance, when I first bought Oracle, it seemed like the market was operating under the assumption that the database business was dead. Both times I bought Canon, the market seemed to have factored in only minimal growth. If I had bought Select Comfort when I first looked into it and it was trading in single digits, that would have been the same story. In fact, the market scared me out of buying the mattress company until the price was more in line with its value.

In First Marblehead, I think the market is crazy and it thinks the same of me. One of us is right and the other will be shown wrong in the next few years. This is an ideal example of why value investing is normally contrarian investing as well.

My miserable July

This is the portion of the show where I normally discuss my exceptional portfolio return and modestly claim the results were the result of "good luck" or some one-time event that can never be repeated. My IRA was down 5.20% in July, which was on top of a -1.92% return for June. So which stocks ought I to have sold to avoid this calamity? Oracle, my largest holding, was down a modest 2.99%. Canon clocked down 9.48%. Select Comfort only lost 1.73% in July but for the last 3 months it has lost 14.02%. Berkshire nearly held steady at -0.66%. First Marblehead was the biggest loser: trimming 14.7%. Sally Beauty lost 10.78% and its brother, Alberto-Culver, lost 0.84%. So it was a clean sweep—everything lost market capitalization in July. Here's how I compared to my benchmarks so far this year:

Date     S&P 500 Delta   IRA   Delta  BRK A
07/31/07   2.61% -3.32% -0.71% -0.72% 0.01%

To be honest, I don't feel that holding onto these positions was a mistake. Each of these companies have performed well in my opinion and will likely rebound when the market starts to calm down a bit. First Marblehead in particular is wildly undervalued because of its perceived connection to mortgage bonds and other asset backed securities.

Getting the customers you deserve

I pointed out that Canon has a better class of customers than its competitors. Today, I read about Sprint's efforts to improve customer quality by releasing problem customers from their contracts. A customer that calls support several times a day is clearly a customer that is not worth keeping. The letter makes the point a bit more lightly: "While we have worked to resolve your issues and questions to the best of our ability, the number of inquiries you have made to us during this time has led us to determine that we are unable to meet your wireless needs." One imagines these customers will be relieved to be let off the hook as well.

Sometimes companies ought to be selective in the people they take money from in order to avoid lower profit customers. Cutting off problem clients may be too extreme, but there are ways to attract "good" customers and avoid the bad. For instance, targeted advertising or limiting service to certain regions. Insurance companies and lenders often reject potential clients based on risk assessment. But the simplest solution is to charge higher prices. It isn't snobbish. People who pay more tend to be better customers if only because they are more profitable upfront.

With that in mind, let's turn to the mattress industry, which is in a rough patch due to the slowdown in home sales. Since mattress purchases can be delayed, consumers may chose to wait if they are worried about their financial future. As a result, mattress companies face the prospect of slowing sales. Select Comfort has refused to lower prices in order to boost sales and instead has focused on improving marketing and product features. On the other hand, Sealy has been offering discounts that average over 10%. As a result, Sealy's sold more and Select Comfort sold fewer mattresses.

Score one for Sealy, right? Not exactly. When you buy a mattress you are paying for the mattress itself, the brand on the label, possibly status, a short trial period, and a warranty of 10 to 20 years. In exchange, the company gets paid a premium upfront. If they are lucky, they won't hear from the customer for the next decade or so when they are ready to buy another bed. A really good customer might buy the company's bed for their children and vacation home, and tell all their friends about what a great mattress they bought. But for every good customer, there are several disgruntled customers. Bad reviews are one thing, but Select Comfort also warns, "We face an inherent business risk of exposure to product liability claims in the event that the use of any of our products is alleged to have resulted in personal injury or property damage."

Besides improving the product, the best defense against unsatisfied and litigious customers a company can deploy is to maximize the profit of the initial sale. Select Comfort is one part manufacturer, one part marketer, one part retailer, and one part insurer for its own products. Like any insurer, the temptation to write unprofitable policies during down markets can be difficult to resist. But resist it must or risk oversized losses for the sake of a few quarters of undersized earnings.

Eveillard's tax device

I just came across an interview with Jean-Marie Eveillard, who manages one of my 401(k) mutual funds—First Eagle Overseas. There is one little answer that immediately caused me to open the spreadsheets of each of my core holdings.

Fortune: You pay a lot of attention to companies' tax rates. Why? Particularly in the U.S., I don't like companies with very low tax rates, because it's a sign either that the Internal Revenue Service will catch up with them someday or that the profits they report are overstated. The average corporate tax rate is 35%. Any company that has a tax rate of 15% or 20% looks suspicious to me.

This suggests a simple device for estimating how much risk a company has because of trying to game the tax system. Companies with overly low tax rates are like ticking time-bombs. There aren't a lot of positive reasons for a company to pay low taxes. The device also warns of companies that exploit the US system that allows companies to use two sets of books. Oracle has the highest rating and it isn't exactly a blaring fire alarm.

Company        Tax     Device     
-------        ---     ------
Oracle         29.71%   5.29%
Canon          34.52%   0.48%
Select Comfort 37.78%  -2.78%
Berkshire      32.81%   2.19%
Sally          38.82%  -3.82% 
Marblehead     38.51%  -3.51%

Raytheon       31.79%   3.21%

Canon sells to a better class of customer

Every time I think about cutting back on my Canon position, I see an article like this one. Essentially, Canon has created a more efficient customer base than its competitors. It isn't completely clear how they were able to do this, but I have a few theories:

• Canon does not have a personal computer business like HP does. Some of the printers bundled with PC systems will be heavily used, but consumers who purchase unbundled printers seem more likely to use them. Brother, Lexmark, and Epson printers also tend to get bundled with PCs.
• Canon does bundle photo printers with digital cameras. Consumers may not be any more likely to be heavy users of these printers, but if they do, they will likely be using more costly color ink and photo paper.
• Canon has a long history of supplying HP with the print engine for the LaserJet series of printers. This arrangement would artificially increase HP's market share number and decrease Canon's, and therefor alter each measure of efficiency. On the other hand, if Canon sells to HP with a greater profit margin than HP sells to the consumer, which seems likely, the aberration isn't that big a deal.
• In the past, color printers used one cartridge to hold all three ink shades, so when you ran out of cyan (probably because you printed too many Word documents with random phrases underlined as if they were email addresses or web links), you had to throw out half a tank of yellow and red. Canon was fairly early in switching to separate cartridges for each color tank, which intuitively would make one think they sell less ink. But I suspect the opposite occurred. Rather than feeling ripped off by a printer company, Canon users felt free to print Word documents with lots of cyan because they knew they wouldn't be wasting a bunch of red and yellow ink.
• I don't have any hard and fast evidence, but I feel like Canon printers are easier to use—especially if you want to print directly from a camera or memory card. It would be easy to assume that usability is most important as an initial selling point, and that customers are locked in, but that ignores how easy it is to get a new printer at a low cost.
• Brand loyalty thanks to years of producing professional and "prosumer" cameras might encourage consumers to buy genuine Canon consumables rather than generics. I know for my family, we gladly pay more for Canon supplies on the assumption that they will produce better results.

Canon's dividend valuation

Until recently, I've seen Canon's dividend as a nice, but not vital piece of the valuation puzzle. But in the last few years, Canon has raised its dividend to a point where the company can be valued on actual payments to shareholders rather than cash flows which might never reach investors. Here is a chart of dividend information for the last 10 years:

                2007    2006    2005    2004   2003   2002   2001   2000   1999   1998   1997  
Dividend        ¥150.00 ¥125.00 ¥100.00 ¥65.00 ¥50.00 ¥30.00 ¥25.00 ¥21.00 ¥17.00 ¥17.00 ¥17.00
Dividend growth 20.00%  25.00%  53.85%  30.00% 66.67% 20.00% 19.05% 23.53% 0.00%  0.00%  -
Expected growth 11.00%  11.75%  12.28%  13.14% 13.25% 10.49% 10.11% 9.18%  4.61%  8.13%  9.47%
DDM for 5 years ¥7,867  ¥6,813  ¥5,597  ¥3,800 ¥2,939 ¥1,533 ¥1,253 ¥1,003 ¥638   ¥769   ¥824
Price           ¥7,131  ¥7,050  ¥6,710  ¥5,560 ¥5,090 ¥4,580 ¥4,660 ¥4,250 ¥4,160 ¥2,320 ¥3,040

"Expected growth" is the mathematical maximum growth that can be sustained based on a Canon's financial information. It is composed of retained earnings rate multiplied by return on equity. Only earnings that are retained (about 75% of income) may be reinvested to fund future dividend growth and the return on equity over the next five years is likely to be about the same as it is this year (15% or so). Multiply those rates and you get a growth rate of about 11% a year. Notice that actual dividend growth has been somewhat better than that since 1999 in order to make up for some flat years in the late 1990s.

I discount that growth at 5% for the next 5 years. This rate seems appropriate for the low return on Japanese risk-free investment. At the end of 5 years, I assume Canon's dividend will grow at 2% compared to a discount rate of 5%. As you can see above, 2007 will be the first year in which the Dividend Discount Model values Canon higher than its market price. Since the dividend-only model is as conservative as you get in valuation, there's no particular reason to construct a more elaborate model when the value exceeds the current price. Converted to dollars, the DDM values Canon at about $64. Meanwhile, the Quicken model I've used in the past puts the value at $77. The current price is $58.59, so by either measure, Canon is still a good buy after doubling over the last 5 years.

Since this estimate is based on what Canon is already doing, I see at least two free options offered by the stock. First, SED TVs could add substantially to Canon's bottom line. Second, consumers in emerging markets such as China, India, Russia, Brazil, and Mexico may soon afford lower-end digital cameras and printers to replace traditional film cameras. Canon has pretty much eliminated the PC as a necessary part of the digital imagery work-flow and the upfront costs are within an order of magnitude of film costs. Over the long-term, digital pictures are vastly cheaper to process and far more convenient for most consumers than film, so I expect the switch-over will happen over the next five years or so. Kodak and Fujifilm sell billions of dollars of film equipment and services around the world and Canon has a shot at nearly all of that business now.

Cost of complexity

I've been listening to Aswath Damodaran's valuation class online, which has been very informative. Near the end of Lecture 10, Professor Damodaran suggests an interesting adjustment to "punish" companies for having complex structures that are hard to understand and analyze. The argument goes the more complex a company is, the more places it can hide information about itself and the more likely some of those details will turn out to be bad news. The professor suggests counting the number of pages in a companies 10-K as a simple way to measure complexity.

I sort of assume my companies are more transparent than their peers, but I didn't have any way of measuring that. Now I do. Here are my core holdings with the first competitor I thought of for reference:

Company        Pages
-------        -----
Oracle         103
Canon (20-F)   122
Select Comfort  72
Berkshire       84
Alberto         99   
Sally           99
Marblehead      71+38F

SAP (20-F)     121+70F+1S
HP             152
Tempur-Pedic    48+30F
Citigroup      180
P&G             23
Regis          117
Sallie Mae     118+84F+12A

I don't know how to treat the extra pages (F-38, A-12 and so on), but my sense is that these are a sign of even more complexity than regular pages. Proctor & Gamble walk away with the prize in this group, but overall, the companies I own are objectively less complicated than the ones I don't. I had actually picked Citigroup as a foil to Berkshire because I expected it to have over a thousand pages. Perhaps that number includes all the supplementary documents that I don't plan on even opening. I only included the main 10-K.

One other reason to use this sort of test is that if a company's filings are too long or complicated, chances are you won't read it. My Alberto-Culver investment relied on that principle, since I hoped as few people as possible would have worked though the sum-of-the-parts valuation and I could buy in at a low price. Now that I've bought, I hope the Sally reports at least are going to become more clear and simple so that other investors can begin to appreciate the company's true worth. And since insiders have had these same goals, I'm pretty sure my wish will be granted.

First Quarter results

All of my companies have reported earnings for the first quarter of 2007 (though some of them call it Q2 or Q3). With a single exception, I'm quite happy with the results. Here are the earnings per share adjusted for splits and spin-offs:

1st Quarter EPS      2007  2006  Change
                     ----  ----  ------
Oracle                .20   .14  42.86%
Canon                 .84   .69  21.74%
Select Comfort        .21   .21 -00.66%
Berkshire Hathaway  56.07 50.03  12.07% 
Alberto-Culver        .23   .16  43.75%
Sally Beauty          .06   .17 -64.71%
First Marblehead      .75   .62  20.97%

One quarter isn't really enough to give a clear picture of a company. But with the exception of Select Comfort and Sally Beauty, these companies are performing well on a multi-year basis. I've talked about Select Comfort's issues, so I won't go into them too much more. This Sunday, they ran a clever ad in Parade magazine that gets delivered with many paper in the US, which confirms my basically good opinion of the new campaign.

Sally Beauty is a more interesting story. Remember that I bought it before the split with Alberto-Culver. After the split, I owned one share of Sally, one share of New Alberto, and $25 for each share of Old Alberto. The $25 special dividend was paid for by borrowing huge amounts—in essence prepaying future earnings of Sally. I haven't seen the latest cash flow statement or balance sheet, but you can get a pretty good idea of the effect of the transaction from this portion of the income statement:

3 months ending March 30     2007    2006
                             ----    ---- 
Operating earnings         60,771  51,313
Interest expense           44,947     321
Interest income               300     300
Market interest rate swaps  1,700       -
Net interest expense       42,947      21
   
Earnings before taxes      17,824  51,292
Provision for income taxes  6,785  20,117

Net earnings               11,039  31,175

Operating earnings are up because of growth in the business and cost savings from no longer being part of Alberto-Culver. Interest expenses are up dramatically even after income from interest rate swaps because of the massive debt load. Sally shareholders owe roughly $12.45 a share to Sally bondholders after the special dividend. Fortunately, there is plenty of cash flow to cover the payments and plenty of growth to grow out from under the debt. You'd be forgiven for thinking the whole thing is a pointless exercise if I hadn't included the impact of taxes. Since interest payments are tax-deductible to Sally, net earnings are not as small as you might imagine. Over the life of the debt, this will amount to significant tax savings.

Although the market value of my companies have increased at a healthy rate, I'm not currently interested in selling any of them because I believe their potential has increased even more. That's the reward for buying good companies at a cheap price.

Update May 14

Sally released the balance sheet with their 10-Q and the debt is closer to $10 a share. Book value is about -$5 a share, which makes life a bit tough from a relative valuation perspective.

How I added to Canon's results

There's a pretty good summary of Canon's Q1 at the Motley Fool. As it happens, I feel partially responsible. You see, earlier this year I bought my wife a Canon Rebel XT . It was intended to replace the A85 I bought her a few years ago and the Rebel 2000 I gave her a few years before. For at least this consumer, Canon has been able to sell higher margin products as the years go by.

The financial statements support this. In 1999, the year the Rebel 2000 was first introduced in Japan, Canon converted ¥3 of every ¥100 in sales to profit. In 2006, it converted ¥11. There are lots of reasons for this, such as the move from film to digital, point & shoot to SLRs, black and white to color copiers and printers, currency benefits, and operating efficiencies. Meanwhile, Canon has taken a hit in asset turnover, which has fallen from ¥98 to ¥92 of revenue per ¥100 of assets. This is to be expected since higher margins usually imply lower sales. Fortunately, the overall effect has been positive as measured by ROA, which went from 2.71% to 10.07%.

Meanwhile, Canon has been de-leveraging over that same time-frame. Its equity to asset ratio dropped from 2.15 to 1.51 thanks to decreases in long-term debt and capital leases. Normally this has a negative effect on return on equity, but because of Canon's stellar margins, ROE has improved from 5.84% to 15.25%. Mathematically, it doesn't much matter which of the ROE component ratios improve, but financially its nice to "front-load" the ratios. In general, it is easier to improve asset turnover through discounting or leverage through borrowing than it is to increase margins. This is especially true in a highly competitive business like digital cameras and printers.

I don't imagine margins will improve much from here on out, but there is hope for further growth in new products. Currently the company is working on launching SED TVs and have announced a medical device initiative. I'm a bit worried about the plan to develop (and I'm not making this up) intelligent robots. Whatever that means. I'm confident that Canon's R&D investments will pay off over time, but it's nice to have such strong operations while we wait.

Look-through earnings

"Merry Christmas!" At least that's what I feel like saying every year around this time. Warren Buffett likes to publish his annual report to shareholders on a Saturday, but we get Christmas a few days early thanks to new SEC regulations. Needless to say, Berkshire's actual results were almost as good as its Chairman's commentary and advice.

This year, I thought it would be fun to present my investment's results using one of Mr. Buffett's favorite tools—look-through earnings. Essentially, I calculate my share of each companies earnings by multiplying the quarterly earnings per share by the number of shares I hold each quarter. Then I aggregate four quarters into a year and divide by the number of "shares" in my IRA. (For an explanation, see this article). This way, I can focus on the economic value the various businesses have added as if I owned each one outright. All 2007 numbers exist only in the imagination of analysts; I use them as placeholders to get an idea of what the future holds.

Earnings         2007*   2006    2005    2004    2003    2002
Oracle           0.19    0.13    0.21    0.30    0.44    0.34
Canon            0.31    0.28    0.49    0.34    0.08 
Select Comfort   0.26    0.19    0.27   
Berkshire        0.19    0.25    
Alberto-Culver   0.12    0.08    
Sally Beauty     0.04    0.00    
Look-through     1.12    0.93    0.97    0.64    0.52    0.34

* 2007 numbers are consensus analyst estimates.

The 2006 results are down in part to my purchase of Alberto-Culver and its spun-off subsidiary, Sally Beauty. I bought these shares more for the value I hoped would be unlocked by the spin-off and for the large special dividend (see below). Oracle has slowly been losing its share in my personal look-through earnings because it is a smaller part of my overall portfolio. On an absolute basis, its earnings have increased smartly.

Last year was the first in which I made more than 2 trades. Besides three new positions, I added to one of my old positions, executed three going-private, arbitrage transactions, initiated two more and sold one option. Plus I left substantial (relatively speaking) sums in cash. So my non-look-through earnings and costs were significant for the first time. In the following chart, I've included actual year-to-date results in the 2007 column.

Interest         0.01    0.16    0.01    0.03    0.02    0.00
Costs           (0.05)  (0.19)  (0.04)  (0.06)  (0.22)  (0.12)
Arbitrage        0.39    0.41    
Options          0.06     
Operating        1.47    1.32    0.94    0.61    0.32    0.23
Gain            11.28%  40.94%  52.78%  89.99%  41.50% 

My "operating" earnings are more impressive, smooth and meaningful when presented this way. The market value of my IRA is substantially more lumpy due to market fluctuations. Note that while the arbitrage earnings are quite significant for last year's results (not to mention this year's), they would be partially offset by a tax cost if this were a taxable account. I'm also batting 1.000 with a pitifully small sample size. One day I will experience a setback and the loss may very well wipe out significant gains.

Finally, if you add in my sale of Oracle a few years ago and the large special dividend from Sally Beauty, you will arrive at some very lumpy net earnings. Once again, these earnings benefit greatly from the tax-deferred status of the traditional IRA.

   
Realized Gain                                    1.79 
Special dividend         2.99    
Net              1.47    4.31    0.94    0.61    2.11    0.23
Gain           -65.97% 360.84%  52.78% -71.01% 827.21% 

The fun part about looking at results this way, is that it's easy to imagine being at the helm of a large conglomerate controlling an array of subsidiaries. But this approach ought to also aid an investor's thinking about the businesses he partially owns. Clearly, I will need to consider eliminating my Alberto-Culver and Sally Beauty stakes in the next few years if they do not improve performance. I might want to increase my investment in Oracle instead. Arbitrage and option activities have added considerably to my bottom line.

January performance

I haven't written in a while, but my IRA portfolio turned in a spectacular January:

Date      S&P 500  Delta   IRA    Delta  BRK A  S&P 500   NAV    BRK A
01/31/07  1.41%    1.07%   2.47%  2.42%  0.05%  1,438.24  24.44  110,050.00

Here are a few highlights:

• Sold Pegasus Communications for $3.25 a share.
• Bought Eupa for 35¢ a share.
• Sold Meritage Hospitality for $5.50(!) a share. (That works out to a 295.71% annualized return.)
• My Oracle call option expired. The good news is that I was not forced to sell my Oracle shares. The bad news is that Oracle lost more than 12% of its value in that time. Fortunately, I didn't want to sell anyway.
• Select Comfort gained nearly 6% for the month. I guess the shock of bad news in December wore off.
• Sally Beauty gained 9% for the month. I guess the shock of bad news in December wore off.
• Alberto-Culver gained nearly 5% for the month based, I suppose, on a 5.5¢ a share quarterly dividend.
• Oracle and Canon had poor Januaries based on being large companies with little real news.
• Berkshire Hathaway, as you can see above, ended flat.
• Earned 4.85% (annualized) on any cash I had lying around.

All of this is to say that my stock prices bounced around randomly. I'd say the going-private transactions that I've participated in show true skill since they've worked exactly as I expected. My other investments have did well in January mostly due to luck. Long term, I expect to have more good months than bad ones and make more good decisions than bad, but one month isn't a big enough sample.

February is turning out to be my best month yet. My IRA balance right now is over $5 million. That easily surpasses the half million dollars I "had" back in September. But I expect my balance will return to earth shortly when my Eupa are cashed in.