Net worth grew $362 million in 1990, or 7.3%, while the stock fell 23%. Buffett uses the gap to explain the one condition under which a shareholder's own return matches Berkshire's, and why he hopes the stock will sell at about what the company is worth.

Last year's letter predicted that a reduction in net worth was almost certain in at least one of the next three years, and for much of 1990's second half Berkshire was on the road to quickly proving it right. Late strengthening in stock prices allowed a gain instead: net worth up $362 million, or 7.3%, and book value per share now $4,612.06 against $19.46 26 years ago, 23.2% a year compounded. The year was lackluster because the four major stockholdings showed little change in market value in aggregate: the two media stocks fell significantly and Coca-Cola rose significantly. "Berkshire's 26-year record is meaningless in forecasting future results", and so, he hopes, is the one-year record.

He still aims for a 15% average annual gain in intrinsic value, ever harder from a $5.3 billion base, and if Berkshire gets it, shareholders should fare well, on one condition (Exhibit 1). "Berkshire's corporate gains will produce an identical gain for a specific shareholder only if he eventually sells his shares at the same relationship to intrinsic value that existed when he bought them." Buy at a 10% premium, let value grow 15% a year, sell at a 10% premium, and your own return is 15% compounded, assuming no dividends. Sell at a smaller premium and your results will be somewhat inferior. That is why he and Charlie hope for Berkshire to sell consistently at about intrinsic value, and prefer that steadiness to the value-ignoring volatility of the past two years (Exhibit 2).

Intrinsic value still exceeds book value by a substantial margin, much of it created by the managers now running the businesses, and he and Charlie feel free to brag because they had nothing to do with developing the managers' skills: "These superstars just came that way." His own role in operations, he says, may best be illustrated by his granddaughter Emily's fourth birthday party, where Beemer the Clown had her wave a magic wand over the box of wonders. Green handkerchiefs went in and blue ones came out, loose ones went in and knotted ones emerged, and after four such transformations Emily exulted: "Gee, I'm really good at this." That, he says, sums up his contribution to the performance of Berkshire's business magicians.

An audited earnings figure can look as exact as pi, Buffett says, and be as pliable as putty. His own are misleading in a particular way: Cap Cities earned more than $83 million for Berkshire in 1990, and the books show about $530,000.

"In reality, however, earnings can be as pliable as putty when a charlatan heads the company reporting them." Some important American fortunes have been created by the monetization of accounting mirages. For connoisseurs of chicanery he attaches a 1936 satire by Ben Graham in which U.S. Steel modernizes nothing but its bookkeeping and turns a $2.76 loss per share into $49.80 of profit (Exhibit 5). Similar excesses have since been duly certified by big-name auditors. "Clearly, investors must always keep their guard up and use accounting numbers as a beginning, not an end, in their attempts to calculate true 'economic earnings' accruing to them."

Berkshire's own reported earnings mislead in a different but important way: it has huge investments in investees whose earnings far exceed their dividends, and it records its share only to the extent of the dividends. The extreme case is Capital Cities/ABC (Exhibit 3). Berkshire's 17% share of its earnings came to more than $83 million last year, yet only about $530,000 is counted in Berkshire's GAAP earnings, $600,000 of dividends less some $70,000 of tax. The residual $82 million-plus stayed with Cap Cities as retained earnings, working for Berkshire's benefit and unrecorded on its books. "We care not whether the auditors hear a tree fall in the forest: we do care who owns the tree and what's next done with it."

When Coca-Cola uses retained earnings to repurchase its shares, Berkshire's percentage of what he regards as the most valuable franchise in the world goes up, more efficiently than if Coke had paid a dividend for Berkshire to reinvest after tax, a route that would have produced far greater reported "earnings." So he counts the look-through way (Exhibit 4): take $250 million, roughly Berkshire's share of the 1990 operating earnings its investees retained, subtract $30 million for the incremental taxes a dividend would have cost, and add the $220 million to reported operating earnings of $371 million, for about $590 million. He hopes look-through earnings will grow about 15% a year. In 1990 they substantially exceeded that rate, and in 1991 they will fall far short of it, as the Gillette preferred converts to common on April 1 and media earnings appear sure to decline.

Berkshire's non-insurance businesses earned 51% after tax on average equity in 1990, which would have placed them about 20th on the Fortune 500, in furniture, candy, vacuum cleaners, and even steel warehousing. Buffett explains what the two Omaha stores share with Wal-Mart.

Leverage did not produce the 51%: almost all the major facilities are owned, and the small debt the operations carry is basically offset by cash. Nor did it come from industries with spectacular economics for all participants, such as cigarettes or network television stations. "Our extraordinary returns flow from outstanding operating managers, not fortuitous industry economics." It was a poor year for retailing, but someone forgot to tell Ike Friedman: Borsheim's sales rose 18%, and he believes the store does more volume than any other jewelry store in the country except Tiffany's in New York, though the fact cannot be certain. About 1,500 assortments went out by mail last year, some to people no one at Borsheim's has ever met, and misanthropes will be crushed to learn that the honor system has yet to produce a loss.

The most important advantage is cost (Exhibit 6): operating costs of about 18% of sales against 40% or so at the typical competitor, and just as Wal-Mart, with its 15%, sells at prices high-cost rivals cannot touch and keeps taking share, so does Borsheim's. "What works with diapers works with diamonds." Low prices create huge volume, volume supports an inventory ten or more times the typical fine jeweler's, and selection plus service plus price makes a national phenomenon out of one Omaha location, staffed by Ike's family down to his 89-year-old mother, in the store most afternoons, Wall Street Journal in hand.

The Furniture Mart runs the same formula: costs of 15% against about 40% at Levitz, the largest furniture retailer, and 25% at Circuit City, prices well below all competitors, and the broadest selection available anywhere. Sales at the one-and-only location were $159 million, up 4%. A report on consumer behavior in Des Moines ranked the Mart third in popularity among 20 furniture retailers serving that city, 19 of them in Des Moines and the Mart 130 miles away (Exhibit 7). Late in the year came Berkshire's first synergistic explosion: a See's candy cart in the Mart sold more candy than some of See's full-fledged California stores, a counterrevelation for a man who has long scorned executives' boasts about synergy. "With the Blumkins, though, the impossible is routine."

See's set a physical volume record, barely, after the invasion of Kuwait cut mall traffic in the West, and a 5% price increase lifted Christmas dollar sales even as poundage slipped. In Albuquerque a landlord's terms would have wiped out the store's profit and a closing date was set, until manager Ann Filkins, on her own, urged customers to protest. Some 263 did, a reporter picked up the story, and the landlord, susceptible to a counterrevelation of his own, offered a satisfactory deal. And the group's net worth rose only $47 million on earnings of $133 million, not because anyone skimped on the franchise but because these managers never deploy capital without good reason: over five years they have funneled well over 80% of their earnings to Charlie and Buffett (Exhibit 8).

Charlie and Buffett were surprised by the media industry in 1990, the Buffalo News included: the business showed far more vulnerability to the early stages of a recession than in the past. Having admitted he did not see the erosion coming, he offers a judgment anyway.

Is the erosion an aberrational cycle, to be fully made up in the next upturn, or has the business slipped in a way that permanently reduces intrinsic values? Since he did not predict what has happened, readers may question his prediction of what will happen, but he proffers one: "While many media businesses will remain economic marvels in comparison with American industry generally, they will prove considerably less marvelous than I, the industry, or lenders thought would be the case only a few years ago." What made them outstanding was unusual pricing power, not growth (Exhibit 9). Now advertising dollars are growing slowly, retailers that do little or no media advertising have gradually taken share, and most important of all, the number of advertising channels has substantially increased. "As a consequence, advertising dollars are more widely dispersed and the pricing power of ad vendors has diminished."

Stan Lipsey's management of the News remains superb: earnings fell only 5% in 1990, though the rate of decrease was far greater in the last few months. He can safely promise two things for 1991: Stan will again rank at the top among newspaper publishers, and earnings will fall substantially, since newsprint and labor will cost considerably more and revenues may be down. The news hole rose to 52.3% from 50.1%, alas because advertising pages fell rather than because news pages grew, and whatever the earnings pressure the paper will keep at least 50% (Exhibit 10). "Cutting product quality is not a proper response to adversity."

Industry premiums grew an estimated 4.5% in 1990 against the 10% needed, and the combined ratio will probably rise about two points more in 1991. Buffett describes a line of business that, if it grows as it may, will make Berkshire's results lumpy on purpose, then proposes a better yardstick than the combined ratio.

He expects incurred losses to grow about 10% a year (11% over the last 25 years), so premium growth far short of that means mounting underwriting losses. Results will improve only when most managements become so fearful that they run from business even at much higher prices. "The most important thing to do when you find yourself in a hole is to stop digging." So far the point has not sunk in: insurance managers continue to dig, sullenly but vigorously, and absent a major physical or financial catastrophe, one to two years will likely pass before losses frighten them into major price increases. Berkshire will be ready.

Berkshire has meanwhile become a large seller of insurance against truly major catastrophes, super-cats, to reinsurers that want to lay off part of their exposure and that, since they will collect only in a crisis, seek a financially strong seller (Exhibit 11). A plain-vanilla contract might pay a reinsurer $10 million, for a premium of, say, $3 million, only if one catastrophe both pushed its own losses above a threshold and cost the industry, say, more than $5 billion. Write $100 million of such premiums in a year and Berkshire is very likely to report either a profit of close to $100 million or a loss of well over $200 million, a combined ratio near zero or near 300%. "Note that we are not spreading risk as insurers typically do: we are concentrating it." Most insurers cannot tolerate such swings. Berkshire's shareholders, properly informed, can, he believes, though the business will produce absolutely terrible results in at least an occasional year. He believes Berkshire is now the largest U.S. writer of super-cat business. "So when a major quake occurs in an urban area or a winter storm rages across Europe, light a candle for us."

Then the yardstick. Because the float is invested, a typical insurer can run losses and expenses 7% to 11% over premiums and still break even, but the range hides exceptions (Exhibit 12). Crop hail insurance produces virtually no float, since premiums arrive just before the storms and claims are paid almost immediately, so a combined ratio of 100 yields no profit. Malpractice insurance produces a very high amount of float, because claims come long after the alleged wrongdoing, so 115 or more can prove profitable, though the long tail may eventually smack the insurer with 200, 300 or worse when all the claims are finally settled. "Setting a target of 100 can itself result in heavy losses: aiming for 110-115 is business suicide."

So a better measure, he believes, is underwriting loss compared with the float developed, over years rather than quarters (Exhibit 13). In 1990 Berkshire held about $1.6 billion of float and lost $27 million on underwriting, so its funds cost about 1.6% while long-term government bonds yielded 8.24%, and in 19 of its 24 years in insurance it has developed funds at a cost below the government's. Two qualifications: the fat lady has yet to gargle, let alone sing, until all losses are settled decades from now, and double taxation adds, by his rough estimate, about a percentage point. GEICO, now 48% owned, customarily underwrites at a profit, so in effect its policyholders in aggregate pay it interest on the float. His own colossal mistakes before Mike Goldberg arrived still send bills: "In the insurance business, there is no statute of limitations on stupidity."

Berkshire neither bought nor sold a share of five of its six major holdings in 1990. The sixth, Wells Fargo, was raised to just under 10% of the bank, the most it can own without the Federal Reserve's approval. Buffett explains why a man who finds banking no favorite bought a tenth of one anyway.

"Lethargy bordering on sloth remains the cornerstone of our investment style." The holdings worth over $100 million are listed in Exhibit 14 at year-end prices. About one-sixth of the Wells Fargo position was bought in 1989 and the rest in 1990, helped by a chaotic market in bank stocks as one huge loss after another was unveiled at once well-regarded banks. The banking business is no favorite of his: with assets twenty times equity, a common ratio, mistakes involving a small portion of assets can destroy a major portion of equity, and mistakes have been the rule at many major banks. So he wants only well-managed banks at fair prices, and in Carl Reichardt and Paul Hazen he thinks Wells Fargo has the best managers in the business, a pair that reminds him of Tom Murphy and Dan Burke at Cap Cities: able people paid well, no bigger head count than needed, and abilities rather than egos deciding what they attempt. The other portfolio change was a large addition to the RJR Nabisco bonds first bought in late 1989, $440 million at year-end (Lesson 07).

Berkshire paid $290 million for a tenth of Wells Fargo, less than five times after-tax earnings, while the stock was falling almost 50% on fears of a California real estate disaster. Buffett works through the arithmetic of a bad year, then explains why a lifelong buyer should cheer when prices drop.

Wells Fargo has $56 billion of assets and has been earning more than 20% on equity, so owning a tenth of it is roughly like owning all of a $5 billion bank, except that such a bank would cost about twice as much and would come without a Carl Reichardt. None of the bank's risks can be ruled out: a major California earthquake that wrecks its borrowers, a contraction or panic severe enough to endanger almost every highly leveraged institution, or the market's fear of the moment, a tumble in overbuilt West Coast real estate. He rates the first two as low probability and takes the third apart with arithmetic (Exhibit 15). Wells Fargo earns well over $1 billion pre-tax a year after expensing more than $300 million for loan losses. If 10% of all $48 billion of its loans were hit by problems in 1991, with losses including foregone interest averaging 30% of principal, the company would roughly break even. "A year like that, which we consider only a low-level possibility, not a likelihood, would not distress us."

The fears nevertheless cut the stock almost in half within a few months, and Berkshire welcomed the decline, which let it pick up many more shares at the new, panic prices. Investors who expect to be buyers all their lives should take the same attitude, he says, instead of illogically turning euphoric when prices rise and unhappy when they fall. They show no such confusion about food (Exhibit 16): knowing they will forever be buying it, they welcome falling prices and deplore increases, and it is the seller of food who dislikes a decline. Berkshire will be buying as long as he lives. "Given these intentions, declining prices for businesses benefit us, and rising prices hurt us." The most common cause of low prices is pessimism, and he likes not the pessimism but the prices it produces. "It's optimism that is the enemy of the rational buyer." Nor is a stock a smart purchase simply because it is unpopular: "What's required is thinking rather than polling."

The 1980s manufactured a new kind of bond, far below investment grade at birth, and 1990 showed what it was worth. Buffett explains the flaw in the salesmen's statistics, why he still bought $440 million of RJR Nabisco bonds, and how his own USAir preferred became an unforced error.

Berkshire has bought a few below-investment-grade bonds before, all old-fashioned fallen angels, issued at investment grade and downgraded when the issuer fell on bad times. The 1980s produced a bastardized version, junk bonds far below investment grade when issued, until in 1990, even before the recession, the financial sky became dark with the bodies of failing corporations. The disciples of debt had said it could not happen: huge debt would make managers focus as never before, much as a dagger mounted on the steering wheel would make a driver proceed with intensified care (Exhibit 17). He grants the alert driver. "But another certain consequence would be a deadly (and unnecessary) accident if the car hit even the tiniest pothole or sliver of ice." In the final chapter of The Intelligent Investor, Ben Graham distilled sound investment into three words, Margin of Safety. "Forty-two years after reading that, I still think those are the right three words."

The dagger-selling bankers pointed to "scholarly" research reporting that low-grade bonds' higher interest had more than made up for their defaults. The flaw, one a first-year statistics student is taught to recognize, was that the research described fallen angels and was applied to newly minted junk (Exhibit 18): the manager of a fallen angel almost invariably yearned to regain investment grade, while the junk-bond operator typically sought another fix rather than a cure. "Mountains of junk bonds were sold by those who didn't care to those who didn't think, and there was no shortage of either." Junk bonds remain a mine field, and Berkshire has never bought a new issue. But it will look at any category it understands, and RJR Nabisco, whose credit he judges considerably better than was perceived for a while, is the one it bought: $440 million at year-end, worth more than $150 million more as he writes, though the risk is far from nil.

The four convertible preferreds remain, and he puts Salomon and Champion at about what Berkshire paid, Gillette somewhat more, and USAir substantially less. He plunged into the airline at almost the exact moment it ran into severe problems, post-merger turmoil and industry economics that have deteriorated at an alarming pace under the kamikaze pricing of certain carriers. Nobody pushed him: in tennis parlance, an unforced error. "In a business selling a commodity-type product, it's impossible to be a lot smarter than your dumbest competitor." Unless the industry is decimated, it should work out all right, though less securely than when he made it. Last year several members of the press valued the preferreds as the common they convert into, which would make the Salomon preferred, convertible at $38, worth 60% of face with the common at $22.80, and a non-convertible Salomon preferred worth zero (Exhibit 19). "The point you should keep in mind is that most of the value of our convertible preferreds is derived from their fixed-income characteristics." He ends the section with regret: Colman Mockler, Jr., Gillette's CEO and his friend, died in January.