Net worth grew $492.5 million in 1986, or 26.1%, and Buffett opens with a division of credit: the managers did a superb job, and he did not match them. The letter explains the two jobs he and Charlie Munger actually have.

Book value per share has compounded at 23.3% a year for 22 years, and business value probably grew faster still in 1986, because managers who need little new capital were building value the books never record. "The bad news is that my performance did not match that of our managers." Job one has not been hard: the managers came with the businesses, stars before Berkshire knew them, and "our main contribution has been to not get in their way." Nicklaus and Palmer would not get a lot of swing tips from him either. Exhibit 1 lists the roster and what each did this year.

The managers work for love of the work, think unfailingly like owners, and could, he says, number twice as many, so long as they had the same rare qualities: with the right people you can have a dozen reporting to you "and still have time for an afternoon nap." The prototype of their fervor is the tailor who saved for years to see the Pope and reported back one fact: "He's a forty-four, medium." The policy of working only with people they like has a hard edge too. Working with people who churn your stomach is like marrying for money: probably a bad idea in any case, and absolute madness if you are already rich.

Job two is harder, and Exhibit 2 shows why it is harder here. A business earning 23% and keeping every dollar is shaped by this year's allocations far faster than one earning 10% and paying half out. In 1986 Berkshire bought Fechheimer, which uses about 2% of its net worth, found no new stock ideas, and otherwise paid off debt and stockpiled cash. "Neither is a fate worse than death, but they do not inspire us to do handsprings either." To average 15%, net worth must grow $7.2 billion over the next ten years, which will take a few very big, and good, ideas.

Ten years after Berkshire bought it, The Buffalo News has the highest weekday penetration of the top 50 American papers and is number one on Sundays too. Buffett believes a bigger paper, kept big while costs were cut, is one of the reasons.

Comparable papers run a news hole of about 40%. The News runs 50%, which Buffett believes is the highest of any dominant paper its size or larger, and Exhibit 3 shows the arithmetic: a paper with 30 pages of ads and a 40% hole delivers 20 pages of news, while the News matches 30 pages of ads with 30 of news, so "we end up delivering our readers no less than 50% more news." He believes that commitment is one of the reasons the News has the highest weekday penetration of any top-50 paper. Margins have almost certainly peaked, with a major newsprint price increase just in, and he will not trim the news hole even if they materially shrink.

The Sunday story in Exhibit 4 is the proof. Ten years ago Buffalo's only Sunday paper, the Courier-Express, sold 271,000 copies and reached about 63% of households, a ratio like many cities' and thought to be "natural," the town's true appetite. The News started a Sunday paper in late 1977. It now reaches 83% of households and sells about 100,000 more copies than the Courier-Express did, in a market whose population has fallen. Stan Lipsey cut man-hours for the third year running and Murray Light edits the product, and Buffett's respect for both has grown consistently since the purchase, dark days of losses and litigation included.

Cap Cities/ABC joined the list this year and the basket is worth $1.87 billion against $655 million paid (Exhibit 5). Three names make up most of it, and Buffett says Berkshire expects to keep them permanently.

Capital Cities/ABC, GEICO, and The Washington Post would not be sold even if they looked significantly overpriced, any more than See's or the News would be, a policy he calls "'til-death-do-us-part." It looks old-fashioned beside the modern manager's "portfolio" of businesses, all candidates for "restructuring," and beside investment managers whose trading makes whirling dervishes look sedated. It also lets his managers and those of his investees run their businesses without distraction. Lear Siegler is an arbitrage position in an announced deal, and the insurers hold WPPSS bonds carried at $218 million and worth $310 million, paying $31.7 million a year tax-free.

The annual "business wanted" advertisement drew a letter in January from a shareholder in Cincinnati. By June Berkshire owned about 84% of his company, and neither Buffett nor Munger has been to Cincinnati to see the operation.

Bob Heldman, chairman of Fechheimer and a Berkshire shareholder for many years, wrote, in a letter Buffett received on January 15, that his company met the tests in the ad. Fechheimer has made uniforms since 1842. His father joined it in 1941, he and his brother George followed, and so did their sons. A 1981 leveraged buyout had left venture capitalists owning most of it, and by 1986 the debt was largely paid down and the owners wanted out. Berkshire bought about 84% at a price based on a $55 million valuation for the whole business (Exhibit 6). The Heldmans kept a stake and kept running it.

The circumstances were similar to the furniture store's (Exhibit 7): most shares held by people who wanted their money elsewhere, a family that wanted to keep owning and managing, several generations in the business, and a seller who wanted a buyer who would not resell at any price. "Both Fechheimer and NFM were right for us, and we were right for them." What Berkshire did not do is visit. "If our success were to depend upon insights we developed through plant inspections, Berkshire would be in big trouble." Chuck Huggins, fifteen years at See's, has never been to Omaha either. The only drawback is size, so the ad runs again with its minimum raised from $5 million to $10 million of after-tax earnings.

The industry's combined ratio dropped to an estimated 108.5 and Berkshire's own to 103, the sharp improvement last year's letter predicted. The second prediction, that prosperity would be fleeting, is already proving accurate, and Buffett shows how.

The table's math is simple. Incurred losses (the claims the industry owes) have been growing 15% to 17% a year while general inflation runs 2% to 4%, because the cost of entering a courtroom has ballooned, so he believes premiums must grow at close to 10% a year just to hold profitability. In 1986 they grew faster than losses and underwriting results improved materially, but the quarterly premium gains have already slowed from an estimated 27% to under 19%, and may well fall below his 10% "equilibrium" figure by the third quarter of 1987. Earnings lag prices by six to twelve months, so 1987 will improve again, barring a major catastrophe, but the improving trend will probably end by late 1988 or early 1989, and thereafter "the industry is likely to head south in a hurry." Exhibit 8 draws the corn field.

Berkshire's volume swings with the crowd's absences rather than its own moods. It is, he writes, the market's most steadfast participant, always ready to write at adequate prices, so business floods in when other insurers are "gone" and leaves when hordes of them are "here" and slashing prices. Firmness on price hurts no customer, who is being showered with cheap offers at exactly those times, and no employee, since Berkshire does not lay off staff in a slowdown. Exhibit 9 shows the swing at National Indemnity: monthly volume went from $5 million in late 1984, about its run rate for years, to about $35 million in early 1986, and is now about $20 million and falling. Some of the new competitors are run by the same managers who bankrupted the old ones, whose unpaid losses Berkshire helps cover through guaranty funds. "C'est la guerre."

The reserves were wrong again, for the third year running, and he notes what Pinocchio's rules would by now have done to his nose. "Reserve strengthening," the industry's noble-sounding phrase, is a euphemism for correcting previous untruths, unintentional ones. GEICO, 41% owned, had an outstanding year: a combined ratio of 96.9, premiums up 16%, and 5.5% fewer shares after buybacks. Its secret is in Exhibit 10. GEICO's underwriting and loss-adjustment expenses ran 23.5% of premiums, while many major companies run 15 points higher. "The difference between GEICO's costs and those of its competitors is a kind of moat that protects a valuable and much-sought-after business castle." Bill Snyder keeps widening it.

The insurers bought about $700 million of bonds in 1986 that Buffett calls mediocre at best, because nothing better was for sale. The explanation turns into his rule for markets, and a warning about the one now running.

Money arriving at the insurers has only five places to go (Exhibit 11): long-term stocks, long-term bonds, medium-term bonds, cash, and short-term arbitrage. Stocks are the most fun, but no equity comes close to meeting his tests. Long bonds are out, barring special cases like WPPSS, because he fears much higher inflation within the decade and the behavior of a currency is set by its legislators. So the money went into medium-term tax-exempt bonds ("at best, the bonds are mediocre investments"), the least objectionable alternative, with some parked in arbitrage and some stockpiled as cash. Liking neither stocks nor bonds, he is the polar opposite of Mae West, who liked only two kinds of men, foreign and domestic.

None of this is a forecast. He has no idea, and never has had, where the market is going. What he knows is that fear and greed will keep breaking out, so "we simply attempt to be fearful when others are greedy and to be greedy only when others are fearful." As he writes, little fear is visible on Wall Street. Euphoria prevails, and the rewards to owners have come gloriously uncoupled from the plodding performance of the businesses themselves. That cannot last. Stockholders as a group pay heavy transaction and management costs, so if American business earns about 12% on equity, its investors must end up earning significantly less (Exhibit 12). Probably the best thing that could happen to Berkshire, he adds, is a market in which it would choose to sell many of its bonds at a significant loss to buy the far better stock values then very likely to exist.

The Tax Reform Act of 1986 cuts the corporate rate and raises the capital gains rate, and Buffett uses it to answer an old argument: when a company's taxes change, does the company feel it, or do its customers?

One school says corporations are a pipeline, passing every tax to customers. Another says they absorb every tax. Exhibit 13 shows that both are right, in different places. An electric utility's regulator usually pushes a tax cut into lower prices in short order. In a price-competitive industry the free market does the same job more slowly. But in an unregulated business with a strong franchise, "reductions in their taxes largely end up in our pockets rather than the pockets of our customers." Impolitic to state, he says, but impossible to deny: ask whether the most able brain surgeon or lawyer in your area will cut fees now that the top personal rate is being cut from 50% to 28%.

For Berkshire the net effect is still negative. The corporate rate falls from 46% toward 34%, which helps, but corporate capital gains are taxed at 34% instead of 28%, and he expects much of Berkshire's future gain in value to come as capital gains: its three big holdings, worth over $1.7 billion and close to 75% of net worth, pay only about $9 million a year in dividends and retain the rest. The insurers are hit hard: his best guess is that the changes to their dividend and interest income alone will eventually cut their earning power by at least 10%. Those dividends and tax-exempt interest are taxed more heavily, and new rules make them pay tax sooner on loss reserves and unearned premiums, which cuts profit the way a 21-year-old's lifetime wealth would be cut by paying tax today on every dollar he will ever earn. For owners, every dollar of gain in Berkshire's value now leaves 72 cents after tax instead of 80. And he doubts the scheduled 1988 rates will hold at all: within, say, five years, higher taxes or higher inflation look almost certain, and both would not surprise him.

The letter's appendix opens with a quiz: two companies with identical sales, wages, taxes, and cash, one reporting $40.2 million of profit and the other $28.6 million. They are the same company, Scott Fetzer, and Buffett uses the gap to define the number that matters.

Berkshire paid $315 million for Scott Fetzer's net assets, which its books carried at $172.4 million. Accounting rules spread the $142.6 million premium across the inventory, the plant, and goodwill, then charge most of it back against earnings year by year: $11.6 million in 1986, none of it deductible for tax, none of it cash (Exhibit 14). So "Company N," as Berkshire must report it, earns $28.6 million, while "Company O," the business as it was, earns $40.2 million. The same sales, wages, taxes, and cash for the owners, and a different number at the bottom.

Which column is closer to the truth? O, he concludes, but the right number is a third one: owner earnings (Exhibit 15): reported earnings, plus depreciation, amortization and other non-cash charges, minus the average yearly capital spending the business needs to fully maintain its competitive position and unit volume. That last item is a guess, sometimes a hard one, but he agrees with Keynes that it is better to be vaguely right than precisely wrong. By that measure the two Scott Fetzers are identical, and for both of them he believes the needed spending runs very close to the old books' $8.3 million of depreciation, not the new books' $19.9 million, so the old column is the better guide. At See's, by contrast, capital spending exceeds depreciation by $500,000 to $1 million a year just to hold ground.

Most businesses are not so lucky, and that is the point of the exercise. Wall Street's "cash flow" adds back the non-cash charges and simply stops, as if the business were a pyramid that never needs repair. For manufacturers, retailers, miners, and utilities the spending is always significant, and over five or ten years it must be made or the business decays. He suspects the number is popular because it justifies the unjustifiable: when reported earnings cannot service a junk bond or excuse a foolish price, (a) plus (b) can. Dentists say that if you ignore your teeth they will go away. The capital bill does not. Questioning the accountants may seem impious, he allows: what are we paying them for, if not the truth? "But the accountants' job is to record, not to evaluate. The evaluation job falls to investors and managers." Accounting numbers are the language of business, and he and Munger would be lost without them. They are invariably the starting point.