Net worth grew $2.1 billion in 1991, or 39.6%, the largest dollar gain yet, and Buffett says at once that it came from something not apt to be repeated. Then he explains why a $7.4 billion base needs a $370 million idea, and what his new job at Salomon says about Berkshire's managers.

Per-share book value has grown from $19 to $6,437 over 27 years, 23.7% compounded, and the equity capital, now $7.4 billion, makes it certain that the past rate cannot be maintained or even approached. With $20 million of capital, an idea producing $1 million of profit added five percentage points to the year's return. Now it takes a $370 million idea, one contributing over $550 million of pre-tax profit, and there are many more ways to make $1 million than $370 million (Exhibit 2). He and Charlie have set a goal of a 15% average annual gain in intrinsic value, and for book value to keep that pace Berkshire must earn $22 billion over the next decade. "Wish us luck, we'll need it."

The outsized gain came from a dramatic rise in the price-earnings ratios of Coca-Cola and Gillette, which together accounted for nearly $1.6 billion of the $2.1 billion (Exhibit 1). When Berkshire loaded up on Coke three years ago its net worth was $3.4 billion, and now the Coke stock alone is worth more than that. The two are among the best companies in the world, he says, and he expects their earnings to grow at hefty rates, with the holdings growing in rough proportion over time. Last year valuations rose far faster than earnings: a double-dip benefit, delivered partly by the earnings growth and even more by the market's reappraisal. "We believe this reappraisal was warranted. But it can't recur annually: We'll have to settle for a single dip in the future."

Having once described the $1 billion Coke purchase as putting Berkshire's money where his mouth was, on August 18 he was elected Interim Chairman of Salomon Inc: "I put my mouth where our money was." Taking the job carried an implicit message. The Blumkins, the Friedman family, Mike Goldberg, the Heldmans, Chuck Huggins, Stan Lipsey, Ralph Schey, and Frank Rooney of H.H. Brown are masters of their operations and need no help from him. "My job is merely to treat them right and to allocate the capital they generate." He illustrates the role with George Mira, the Miami quarterback who, his right shoulder in a Florida linebacker's grasp near the goal line, switched the ball to his left hand and threw the only left-handed pass of his life for a touchdown, whereupon his coach, Andy Gustafson, calmly told a reporter: "Now that's what I call coaching."

Look-through earnings, the measure Buffett prefers, fell from $602 million to $516 million in 1991, after growing at almost the same 23% as book value since 1965. He names the four forces, two of which he had warned about, and tells investors how to use the idea on their own portfolios.

Look-through earnings consist of reported operating earnings, plus the retained operating earnings of major investees that GAAP leaves out of Berkshire's profits, less an allowance for the tax Berkshire would pay had those earnings been distributed. Over time they must grow about 15% a year if intrinsic value is to, and since 1965 they have grown at almost the identical 23% rate recorded for book value. "Last year, however, our look-through earnings did not grow at all but rather declined by 14%." Exhibit 3 shows the calculation, with figures he warns are necessarily very rough: $230 million of undistributed earnings, less a hypothetical $30 million of tax, plus $316 million reported, for $516 million against $602 million.

Two forces he had warned of (Exhibit 4): media earnings, direct and look-through, were 'sure to decline,' and did, and the call of the Gillette preferred on April 1 forced conversion into common. The preferred had earned about $45 million after tax in 1990, somewhat more than the 1991 combination of three months of its dividends and nine months of look-through earnings on the common. Two outcomes he did not foresee: a break-even result at Wells Fargo, where the dividends received were offset by negative retained earnings, the "low-level possibility, not a likelihood" of last year's letter, and significantly lower, though still excellent, insurance profits.

Investors can benefit by focusing on their own look-through earnings, he says: total the underlying earnings attributable to the shares you hold, and aim to build a portfolio, in effect a company, that will deliver the highest possible look-through earnings a decade or so from now. That forces thought about long-term business prospects rather than short-term stock market prospects, a perspective likely to improve results. In the long run the scoreboard for investment decisions is market price, but prices will be determined by future earnings. "In investing, just as in baseball, to put runs on the scoreboard one must watch the playing field, not the scoreboard."

Last year Buffett said media profits were falling for reasons that were partly permanent. The events of 1991 fortified the case, and he lays out a test for telling a franchise from a mere business, then some over-simplified but relevant math on what the difference does to a price.

The economic strength of once-mighty media enterprises continues to erode as retailing patterns change and advertising and entertainment choices proliferate. A few years back nobody linked to the business, neither lenders, owners, nor analysts, saw the deterioration coming, and in business the rear-view mirror is always clearer than the windshield. "The fact is that newspaper, television, and magazine properties have begun to resemble businesses more than franchises in their economic behavior." A franchise arises from a product or service that is needed or desired, is thought by its customers to have no close substitute, and is not subject to price regulation (Exhibit 5), which shows up as the ability to price aggressively and earn high returns on capital. A franchise can tolerate mismanagement, and a business cannot. Media properties once passed all three tests and could be managed loosely. Now consumers have greatly broadened choices, and demand cannot expand to match: "500 million American eyeballs and a 24-hour day are all that's available."

The weakened franchise affects value far beyond the immediate hit to earnings (Exhibit 6). Conventional wisdom once held that a newspaper, television, or magazine property would increase earnings 6% or so forever without additional capital, so that reported earnings were freely distributable and the property resembled a perpetual annuity growing 6% a year. Discounted at 10%, that justified a whopping $25 million for $1 million of after-tax earnings, 25 times after-tax or about 16 times pre-tax. Posit instead that the $1 million merely bobs around, the lot of most businesses, and the same 10% discount gives $10 million, about 6 1/2 times pre-tax. "Dollars are dollars whether they are derived from the operation of media properties or of steel mills." Expectations for media have moved toward the bob-around model, and valuations must change dramatically when expectations are revised.

Berkshire's media investment, the Buffalo News plus the Post and Cap Cities stakes, has declined materially in intrinsic value because of the secular transformation, and the rule stands that Berkshire does not sell businesses or permanent holdings merely because the money could do better elsewhere. The losses have been moderated because the News under Stan Lipsey has done far better than most newspapers and because Cap Cities and the Post are exceptionally well managed: both stayed on the sidelines while late-1980s buyers paid irrational prices, and both carry debt that is small and roughly offset by cash, so the shrinkage in asset values was not accentuated by leverage (Exhibit 7). Most other public media companies carry debt of five or more times current net income. Most media properties still have far better economics than the average American business. "But gone are the days of bullet-proof franchises and cornucopian economics."

On January 3, 1972, Blue Chip Stamps bought control of See's at a price that, for the whole company, came to $25 million, $5 million below the true asking price and as high as Buffett and Munger would go. Twenty years later he tallies what the purchase taught them about franchises, and what California thinks a snack is.

The sellers asked $40 million, but the company held $10 million of excess cash, so the true price was $30 million. He and Charlie, not yet fully appreciative of the value of an economic franchise, looked at the mere $7 million of tangible net worth and said $25 million was as high as they would go, and meant it. Fortunately the sellers accepted. Blue Chip's own trading-stamp sales then fell from $102.5 million in 1972 to $1.2 million in 1991, while See's sales rose from $29 million to $196 million and pre-tax profit from $4.2 million to $42.4 million (Exhibit 8). Measured against the capital it took, the record is astounding: See's operates comfortably on $25 million of net worth, so the $7 million base needed only $18 million of reinvested earnings, and the remaining $410 million of pre-tax profit was distributed to Blue Chip and Berkshire over the 20 years to deploy, after tax, however made most sense.

"In our See's purchase, Charlie and I had one important insight: We saw that the business had untapped pricing power." Otherwise they were lucky twice: the deal was not derailed by their dumb insistence on $25 million, and they found Chuck Huggins, then executive vice-president, and put him in charge at once. His compensation was settled by handshake in about five minutes, never reduced to a written contract, and remains unchanged. In 1991 dollar sales matched 1990 while pounds fell 4%, all of the slippage in the last two months of the year, which normally produce more than 80% of annual profits. Profits still grew 7%, and the pre-tax margin was a record 21.6%. Almost 80% of sales come from California, where the recession hit with particular force late in the year and where a mid-year sales tax of 7% to 8 1/2% on 'snack food' was deemed to apply to candy, under classifications that reward students of epistemological shadings (Exhibit 9). Owning See's has taught them much about evaluating franchises. "We've made significant money in certain common stocks because of the lessons we learned at See's."

Berkshire bought H.H. Brown, the leading North American maker of work shoes and boots, on July 1, 1991, after a golf game did what the sellers' investment bank did not. Buffett tells the Heffernan story, admires a $7,800 salary, and repeats what he is looking for.

In 1927 Ray Heffernan, 29, bought the company for $10,000 and ran it for 62 years, still joining new golf clubs at 90. By his retirement in early 1990 Brown had three U.S. plants and one in Canada, close to 2,000 employees, and about $25 million of pre-tax earnings. His daughter Frances married Frank Rooney, sternly told before the wedding to forget any idea of working for his father-in-law, one of Mr. Heffernan's few mistakes: Frank became CEO of Melville Shoe, where over 23 years earnings averaged more than 20% on equity and the stock rose from $16 to $960, and when Mr. Heffernan fell ill he asked Frank to run Brown. After his death late in 1990 the family decided to sell, and a major investment bank took the assignment without thinking of Berkshire (Exhibit 11). Last spring Frank played golf in Florida with John Loomis, a shareholder and long-time friend of Buffett's, who told him Brown should be right up Berkshire's alley. Frank called, and before long it was done.

Much of his enthusiasm came from Frank's willingness to stay as CEO. Like most Berkshire managers he has no financial need to work and does so because he loves the game and likes to excel. "Managers of this stripe cannot be 'hired' in the normal sense of the word. What we must do is provide a concert hall in which business artists of this class will wish to perform." Brown also has one of the most unusual compensation systems he has encountered (Exhibit 10): a number of key managers draw a salary of $7,800 plus a designated percentage of profits after a charge for capital employed, so they truly stand in the shoes of owners, whereas most managers choose systems long on carrots and short on sticks that almost invariably treat equity capital as if it were cost-free. Of four major purchases from sellers represented by prominent investment banks, only once did the bank contact Berkshire. In the other three he or a friend started the deal. So he repeats the criteria, from $10 million of after-tax earnings to an offering price, and promises an answer customarily within five minutes. With Brown, he did not need to take five.

Berkshire's insurers lost $119.6 million on underwriting in 1991, yet by the measure that counts the year was a good one: the float cost 6.31% while the government paid 7.40%. Buffett explains why super-cat profits arrive in lumps, who the innocents are, and why the industry's improved combined ratio may be a touched-up x-ray.

Industry premiums grew an estimated 3.1% in 1991, nowhere near the 10% needed, yet the combined ratio slightly improved, to an estimated 109.1, instead of deteriorating as he expected. Loss-reserve data give reason to be skeptical, and 1991's ratio may turn out to have been worse than reported. In the long run trouble awaits managements that paper over operating problems with accounting maneuvers, like the seriously ill patient who tells his doctor: "I can't afford the operation, but would you accept a small payment to touch up the x-rays?" For Berkshire the ratio matters little anyway (Exhibit 12). What counts is the cost of funds developed from insurance, in the vernacular the cost of float, measured by the underwriting loss. In 1991 that loss was $119.6 million on average float of $1,895 million, a cost of 6.31% against 7.40% on newly issued long-term government bonds, and Berkshire has beaten the government's rate in 20 of its 25 years in insurance, often by a wide margin.

Berkshire remains a very large writer, perhaps the largest in the world, of super-cat insurance, and its profits are enormously volatile: $100 million of premiums, roughly the annual expectation, could deliver anything from a $100 million profit in a year with no big catastrophe to a $200 million loss in a year with a couple of major hurricanes or earthquakes. The business is priced to pay out about 90% of premiums over the long term (Exhibit 13), and partly because GAAP does not allow reserves in catastrophe-free years for losses certain to arrive in others, in any given year Berkshire is likely to appear either enormously profitable or enormously unprofitable, and a one-year accounting cycle, he says, is ill-suited to the business. In 1991 there appears to have been one super-cat by Berkshire's definition, triggering payments on only about 25% of its policies, for an estimated underwriting profit of about $11 million. The year's biggest catastrophe was neither the Oakland fire nor Hurricane Bob but a September typhoon in Japan, with an insured loss now estimated at about $4 to $5 billion.

Insurers will always need huge reinsurance against marine, aviation, and natural disasters, and in the 1980s much of it came from innocents, insurers that did not understand the risks and have since been financially burned beyond recognition (Exhibit 14). Berkshire itself was an innocent all too often when he personally ran the operation. Insurers, like investors, eventually repeat their mistakes, so at some point, probably after a few catastrophe-scarce years, innocents will reappear and super-cat prices will plunge to silly levels. As long as apparently adequate rates prevail, Berkshire will be a major participant, with a significant advantage in its premier financial strength: "Thinking insurers know that when 'the big one' comes, many reinsurers who found it easy to write policies will find it difficult to write checks." Overall, insurance offers Berkshire its greatest opportunities, and Mike Goldberg has accomplished wonders with it since he took charge.

Eight holdings were worth over $100 million at year-end, Gillette now among them as common stock. Berkshire held seven of them a year ago and changed its share count in only one. The newcomer, Guinness, is its first significant investment in a company domiciled outside the United States.

The list, he says, reflects a Rip Van Winkle approach: of the seven stocks held a year ago, allowing for the Gillette conversion, six are held in unchanged number, and only the Freddie Mac stake increased, slightly (Exhibit 15). Guinness earns its money much as Coca-Cola and Gillette do, from international operations, and in where its profits come from, continent by continent, it strongly resembles Coke. Berkshire searches for large businesses with understandable, enduring, and mouth-watering economics run by able, shareholder-oriented managements. That guarantees nothing, since it must still buy at a sensible price and get the performance that validates the assessment, but it offers the only chance of real success for people working with sums this large, who cannot get great results by darting in and out of far-from-great businesses. He also notes, tongue only partly in cheek, that recent events show the much-maligned idle rich had been getting a bad rap: they kept their wealth while many of the energetic rich lost theirs.

In 1989 Buffett promised an update on his mistakes in 2015. The backlog is growing too fast for that, so he starts confessing as he goes, beginning with Fannie Mae, and ends with USAir, whose industry met the one condition he had attached to the investment.

Public confession, he hopes, may deter further bumblings. "Typically, our most egregious mistakes fall in the omission, rather than the commission, category." He is not talking about missing Xerox, Apple, or Wal-Mart, businesses Berkshire will never develop the competence to spot early, but about situations he and Charlie understand and find clearly attractive and still fail to buy. In early 1988 they decided to buy 30 million shares of Fannie Mae, adjusted for a later split, a $350 to $400 million investment in a company they had owned before and understood, whose CEO, David Maxwell, had dealt superbly with inherited problems (Exhibit 16). After about 7 million shares the price began to climb, and in frustration he stopped buying, a mistake he thankfully did not repeat when Coca-Cola rose during its purchase program. Then, in an even sillier move, he surrendered to his distaste for small positions and sold the 7 million. There is no halfway rational explanation, only an estimate of the gain not made: about $1.4 billion.

The alternative to thumb-sucking is concentration (Exhibit 17). If his universe were limited to private companies in Omaha, he would assess each business's long-term economics and the quality of its people, then buy into a few of the best at a sensible price, certainly not an equal part of every business in town, and he sees no reason to act differently in the larger universe of public companies. Keynes, whose brilliance as a practicing investor matched his brilliance in thought, wrote to F. C. Scott in 1934 that the right method is to put fairly large sums into enterprises one knows something about and whose management one thoroughly believes in, that it is a mistake to think spreading among enterprises one knows little about limits one's risk, and that he seldom found more than two or three in which he could put full confidence. Calling institutions that trade actively 'investors,' he and Charlie believe, "is like calling someone who repeatedly engages in one-night stands a romantic."

Fixed-income surprises came too, none greater than the need to involve himself personally and intensely at Salomon, whose preferred he and Charlie nonetheless believe rose slightly in value as rates fell and Salomon's common rose. Berkshire bought $300 million of an American Express 'Perc' paying 8.85%, which must convert three years after issuance, extendable by a year in one case, into at most 12,244,898 shares with the value capped at $414 million. And USAir, which last year "should work out all right unless the industry is decimated during the next few years," saw the condition met: 1991 was a decimating period, with Midway, Pan Am, and America West in bankruptcy and Continental and TWA within 14 months (Exhibit 18). Bankrupt carriers, encouraged by the courts to keep flying, can for a time charge fares below the industry's costs, toppling previously marginal carriers in a domino effect perfectly designed to bring the industry to its knees. The preferred is valued at $232.7 million against its $358 million cost, reflecting a far-from-negligible risk that the industry stays unprofitable for virtually everyone in it. "There is no tougher job in corporate America than running an airline", and Seth Schofield, CEO since 1991, has the brains, guts, and experience it needs.