In 1914, Thomas Watson, a star salesman for the National Cash Resister Company, was hired by the Computing-Tabulating-Recording Company (CTR) to be its president. CTR had been founded three years earlier to produce electromechanical machines that could “read” the location of holes in punched cards and make calculations and tabulations based on the location of the holes. For the first time, machines were replacing pencils for accounting applications. Watson was a super salesman and leader, and soon CTR became the world leader in tabulating office machines. In 1924, Watson, with infinite wisdom, changed the name of Computing-Tabulating-Recording to International Business Machines.

IBM’s punched card equipment was a great success, and by the 1940s IBM was the world’s dominant producer of office business machines. The business leased the punched card equipment under full-service leases that included maintenance and technical advice as well as the equipment itself. IBM’s strong financial incentive was to keep each piece of equipment on lease for as long as possible. While IBM earned good profits during the early years of a lease, if the equipment remained on lease after it was fully depreciated, IBM’s profits became particularly large. Thus, IBM had incentives to delay the introduction of new, more efficient equipment that would obsolete and replace existing equipment on lease.

Over the decades, a number of inventors produced electromechanical machines that were at least somewhat programmable. These were called computers. IBM’s first computer was delivered to Harvard University on August 7, 1944, for use by the U.S. Navy Bureau of Ships. IBM called the computer the Automatic Sequence Controlled Calculator. But Harvard simply and wisely called it the Mark 1. For the next seven years, IBM had little incentive to introduce programmable machines for the commercial market. After all, it did not wish to obsolete its highly profitable line of punched card equipment. However, in 1951, the Remington Rand Corporation delivered its first UNIVAC computer—and IBM saw the handwriting on the wall. IBM knew that if it did not introduce its own line of commercial computers, its punched card equipment eventually would be replaced by UNIVACs, or by computers made by other companies.

In 1952, IBM introduced its 701 vacuum tube computer, and the race was on. But it never was a close race. It was a cheetah against a sloth. IBM benefited from vastly superior research, engineering, marketing, and applications knowhow. Furthermore, computer users felt safe with IBM. Computers were new and complex, and corporations installing computers relied on the computer manufacturer for technical and applications expertise. IBM was a highly respected and known quantity. If a corporate executive selected an IBM system and the installation developed problems, the corporate executive could blame IBM for the problems. However, if a corporate executive selected a non-IBM computer and the installation developed problems, the corporate executive could be blamed by his bosses for selecting a second- or third-tier computer company, and the executive’s career could be in jeopardy. Thus, the vast majority of executives played it safe and stuck with IBM. UNIVAC, GE, RCA, Honeywell, and Burroughs often could offer more cost-efficient computers, but IBM usually received the orders. This was particularly true after IBM introduced its technologically advanced System/360 family of compatible computers in 1964. The 360s were efficient, reliable, user friendly—and carried the IBM brand. IBM had a winner.

The System/360 transformed IBM into one of the most successful corporations ever. Between 1964 and 1974, the company’s revenues grew at a 14.6 percent average annual rate from $3.23 billion to $12.67 billion. Net earnings grew even faster, at a 17.5 percent rate, from $364 million to $1,830 million. IBM’s market share during the period approached 80 percent. Competitors could not compete, and some quit. GE sold its computer business to Honeywell in 1970. One year later, RCA sold its computer business to Sperry Rand.

The seeds for IBM’s eventual downfall came in 1972, when Intel developed its first microprocessor, the 4004. A microprocessor incorporates the functions of a computer’s complex inner workings on a single integrated circuit, or, at most, on a few integrated circuits. The first microprocessors were not sufficiently powerful to challenge the viability of IBM’s System/360s and other mainframe computers. However, Intel cofounder Gordon Moore famously had predicted: “The number of transistors incorporated in a chip will approximately double every 24 months”—and Gordon Moore proved correct. By the late 1970s, microprocessors were sufficiently powerful that relatively inexpensive desktop computers could start replacing expensive mainframes. In 1977, Apple introduced the first successful desktop (personal) computer, the Apple II. The Apple II was a hit. Eventually, 1.25 million were sold. IBM noticed this success and noticed that microprocessor technology was progressing so rapidly that IBM’s mainframe business, which was the source of most of the company’s profits, was in jeopardy of being obsoleted. IBM had to respond.

In 1981, IBM introduced the IBM Personal Computer to replace its unpopular 5200 desktop computer. The IBM PC, backed by IBM’s reputation for excellence, was a hit and helped drive IBM’s sales and profits in the early 1980s.

Other events also drove IBM’s reported profits in the early 1980s. In 1980, John Opel became CEO of the company (Tom Watson had retired in 1956 and Tom Watson Jr. had retired in 1971 after suffering a heart attack). Opel was concerned about IBM’s policy that a leased computer could be returned to IBM by a customer with 30 days’ notice. His prime worry was that competitors, using advanced microprocessor technology, would start marketing computers that were materially more cost effective than IBM’s mainframes. IBM’s mainframe customers would then cancel their leases with IBM, leaving IBM with substantially reduced profits and with inventories of relatively new, but technologically obsolete, computers. Responding to these concerns, Opel changed IBM’s pricing policy by offering to sell used mainframes at deeply discounted prices. The low prices encouraged customers to purchase the mainframes they were leasing. By doing this, IBM immediately received revenues (the sales price of the used computer) and profits that would have been received in later years if the computer had remained on lease. Thus, as long as customers converted leases into sales, IBM’s reported revenues and profits were artificially inflated. Then, when the rate of conversions slowed, IBM changed its terms on new leases that permitted the company to treat the leases as full payout leases. On full payout leases, most of the expected revenues and profits could be booked when the computers were delivered to the customers. Previously, the revenues and profits were booked over the life of the leases.

The introduction of personal computers, the conversion of leases to sales, and the new accounting for leases caused IBM’s revenues and earnings to surge between 1980 and 1985. During the five-year period, revenues increased at a 13.8 percent average annual rate from $26.21 billion to $50.05 billion and net profits increased at a 14.1 percent average rate from $3.39 billion to $6.55 billion. IBM shareholders had much to be thankful for. The price of IBM’s shares, adjusted for subsequent stock splits, rose from $16.09 at year-end 1979 to $38.88 at year-end 1985.

However, all really was not well. Not well at all. When IBM developed its lines of personal computers, it adopted Intel’s chip technology and Microsoft’s MS-DOS operating system. This was a convenience. However, because other personal computer manufacturers had equal access to Intel’s chips and Microsoft’s operating system, IBM lost any hope of significant product differentiation, and personal computers rapidly became commodities that were sold based on price. Over time, Dell, Compaq, and other PC manufacturers with lower cost structures than IBM would be able to outcompete IBM.

IBM was facing additional problems as well. Computer power was increasing rapidly and was becoming less expensive. Moore’s law was working. By the mid-1980s, personal computers could perform many of the jobs previously performed by mainframes. Furthermore, Digital Equipment’s line of VAX minicomputers also was taking market share from IBM’s mainframes. In addition, while in the 1960s companies heavily relied on IBM’s technical expertise and reputation, by the mid-1980s many corporations had sufficient in-house expertise and experience that they no longer needed to rely on IBM. In the 1960s, IBM had pricing power, but by the mid-1980s the company often had to price competitively to win orders.

When relatively nondifferentiable products are sold based on their price, the manufacturers of the products normally need to have low cost structures if they wish to be competitive and earn reasonable profits. Unfortunately, largely because of its history, IBM’s cost structure was high—very high. In the 1960s, IBM needed large numbers of salesmen and technical advisers to sell and install mainframes. It also needed large numbers of maintenance engineers to provide maintenance and repairs to its mainframes, which were not nearly as reliable as computers produced 20 years later. By the mid-1980s, a large percentage of IBM’s sales managers, salesmen, applications programmers, technical advisers, and maintenance engineers were superfluous, but IBM maintained a policy of not laying off any employee as long as he was not incompetent or dishonest. Many superfluous employees were “promoted” upstairs to corporate headquarters to perform such less essential or nonessential tasks as competitive analysis, sales forecasting, quality control, real estate management, corporate relations, investor relations, community relations, and economic forecasting. In short, IBM’s cost structure was bloated at a time when the company’s revenues and gross margins were under pressure from systemic adverse technological changes in the computer business.

IBM’s revenues continued to increase some in the late 1980s, but fell far short of the company’s earlier hopes. In 1984, John Opel predicted that the company’s revenues would reach $100 billion in 1990. Actual 1990 revenues were $69 billion. Earnings fared far worse than revenues. After-tax earnings in 1990 were $6.02 billion, 8 percent below the level of five years earlier. And the price of IBM’s shares even fared far worse than earnings, declining 27 percent between the last trading day of 1985 and the last trading day of 1990.

John Akers replaced John Opel as CEO on February 1, 1985, at a time when IBM was bloated with more than 400,000 employees. In 1985, I was invited by Jon Rotenstreich, IBM’s treasurer, to have lunch at the company’s headquarters in Armonk, New York. Jon is a friend who previously had been a managing director of Salomon Brothers. We had lunch in the company’s central dining room. Jon looked around the dining room and pointed out employee after employee who came to the office every morning, but essentially had little work to do. “See that guy over there with the green tie? He used to be a regional sales manager. Now he is a community relations specialist. He comes to the office at about 10 and grabs a cup of coffee. He is responsible for IBM’s relations with the hamlet of Armonk, so it is important for him to read the local Armonk weekly newspaper and watch for local Armonk news on TV. That takes him the rest of the day, except for a two-hour lunch break and three or four additional coffee breaks. Get the picture?” I got the picture. Jon strongly hoped that John Akers would abandon IBM’s policy of lifetime employment and would substantially reduce IBM’s payroll and other costs.

Akers formed several task forces to study IBM and its future. The task forces’ findings were worrisome. They concluded that IBM was in a state of decline. After reviewing the findings, Akers initiated a series of early retirement programs, offering large severance packages to employees willing to leave the company. Of course, many of the most capable employees—those readily able to find other jobs quickly—accepted the package. So IBM lost a lot of talent and kept a lot of deadwood (adverse selection at work). Furthermore, in spite of the retirement programs, IBM’s employment levels only declined from 405,000 at the end of 1985 to 374,000 at the end of 1990. During that period, IBM’s fundamentals deteriorated faster than its employment levels. By 1990, for about $100,000 a computer user could purchase a workstation that had the power of a mainframe that several years earlier had sold for a few million dollars. IBM suffered a large operating loss in 1991. Akers then had to act, and he did so by increasing the size of the retirement programs. IBM’s employment levels declined to 344,000 at the end of 1991 and to 302,000 at the end of 1992. However, IBM still had far too many employees, morale was poor, and the company continued to lose money.

I became interested in seriously thinking about IBM’s shares in 1985 after my lunch with Jon Rotenstreich. My initial analysis was straightforward. IBM no longer was a growth company with competitive advantages that led to high profit margins. Thus, it no longer deserved to sell at a high price-to-earnings (PE) ratio. However, if the company could become cost competitive by sharply reducing its head count and by selling excess plants and office buildings, it could emerge as a large and powerful company that should earn a reasonable return on revenues. Furthermore, the company would not have to spend large sums to build new plants—and, therefore, a large percentage of its future earnings would be available for dividends or share repurchases.

When I become interested in studying a company, I sometimes quickly purchase a few shares of the company’s stock. Ownership incentivizes me to study the company with increased intensity. I purchased a few shares of IBM on two occasions between 1989 and 1992, but sold the shares each time because I believed that IBM’s problems could deteriorate further before they started to improve. The second time I sold the shares following a meeting with John Akers in his office. The legendary investor Mike Steinhardt owned shares of IBM. Mike knew that I also owned shares, and he invited me to join him in a meeting he had scheduled with Akers. It took me one second to say “yes” to the invitation. However, the meeting turned out to be discombobulated because Mike and I had completely different reasons for being interested in IBM. Mike observed that the company had a very profitable and rapidly growing subsidiary in Japan. In 1991, Japanese companies still sold at high PE ratios. Mike wanted IBM to spin off its Japanese subsidiary to IBM shareholders. He reasoned that the Japanese subsidiary alone would be worth a large percentage of IBM’s existing market value and that the combined market value of the two pieces (IBM Japan and IBM ex-Japan) would be far in excess of IBM’s existing share price. I doubted the practicability of dividing IBM into two separate companies. Instead, I questioned Akers on why IBM was not reducing costs more aggressively. One minute Mike was asking about Japan, and the next minute I was asking about costs. Then, Mike had another thought about spinning out Japan, and then I had a thought about weeding out the deadwood as opposed to incentivizing the most capable to leave IBM. It went back and forth for a full hour. Akers could not have been more courteous, but it was clear to me that he was not going to cut costs more aggressively—and it was very clear to me that he was not going to spin out Japan.

IBM was doing poorly. Akers was under pressure and, in early 1993, announced his resignation. Lou Gerstner was his replacement. IBM held a press conference on the day Gerstner assumed control. Gerstner, who was wearing a very unlike-IBM blue shirt, told the audience of reporters that he had the courage to take tough steps. That is what I wanted to hear. I decided to seriously consider making a large investment in IBM’s shares.

My analysis was roughly as follows. I estimated that that company had about 60,000 more employees than needed and that an average employee’s compensation (salary and benefits) totaled about $85,000. Therefore, I concluded that Gerstner might be able to reduce costs by $5 billion. After taxes at a 32 percent effective rate, the cost reductions would add about $3.4 billion to net earnings, or about $1.50 per share based on the 2.29 billion IBM shares that were outstanding. Before nonrecurring charges, IBM was roughly breaking even at the time. Therefore, I estimated that the company’s current earnings power would be about $1.50 per share pro forma a $5 billion reduction in costs. I further estimated that the company’s revenues would grow at a 5 percent rate and therefore that IBM’s earnings power in 1995 could be about $1.65 per share. When we purchase a stock, we are interested in what the company will be worth two or three years hence, so the $1.65 was an important number.

In order to check the reasonableness of earnings estimates, I often like to use two disparate methodologies when projecting earnings – and then see if the two methodologies reach similar conclusions. In the case of IBM, I decided to also use the following thought process. The company’s revenues in 1993 were expected to be about $63 billion. Assuming a 5 percent growth rate, revenues in 1995 would be about $69 million. Based on my experience, I estimated that an efficient manufacturer of computers that was operating in a competitive environment might have after-tax profit margins of 5 to 6 percent. Based on this methodology, IBM’s after-tax earnings power in 1995 would be $3.5 to $4.1 billion, or $1.50 to $1.80 per share.

I decided to use $1.65 as a single best estimate of what IBM’s earnings power could be in 1995. Finally, to value the shares, I assigned a multiple to the $1.65. I believed that IBM’s quality and growth potential were somewhat below average. Since I believe that an average company is worth 15 to 16 times earnings, I valued IBM at 12 to 13 times earnings. Thus, I estimated that, in 1995, the shares would be worth $20 to $21. I knew that my projections of IBM’s earnings and values were nothing more than best guesses based on incomplete information. However, having the projections to work with was better than not having any projections at all, and my experience is that a surprisingly large percentage of our earnings and valuation projections eventually are achieved, although often we are far off on the timing.

IBM’s shares were selling at $12 at the time I completed my analysis in May 1993. Over the next two months, I purchased a substantial position in the shares at an average cost of about $11½. We usually purchase stocks because we believe that one or more positive changes will trigger a sharp appreciation in the price of the stocks. In this case, we hoped that IBM would announce a definitive plan to substantially reduce its costs.

On July 28, Lou Gerstner announced a plan to reduce employment levels to 225,000 by the end of the year. About 85,000 employees, or slightly more than 25 percent of the existing workforce, would leave by year-end. Several months after the announcement, both Lou Gerstner and his cost-reduction program started to gain credibility on Wall Street. In November, IBM’s shares started to appreciate. By August 1994, the shares were selling at about $15½. By that time, the common wisdom on Wall Street was that Gerstner’s cost-reduction program had been successful. It appeared that the company would earn about $1.25 per share in 1994 and materially more in 1995.

My strategy is to purchase shares in anticipation of a positive change and then, normally, to sell the shares when the change occurs and is largely discounted into the price of the shares. In the case of IBM, the change had occurred and investor sentiment had turned largely positive. At the time we had purchased the shares at about $11½ the sentiment had been decidedly negative. Because of the change in sentiment, I decided to sell our position, which we accomplished over the next few months at an average price of just above $16. We earned a 40 percent profit in the stock. Yes, I had estimated that the shares might be worth $20 to $21 in 1995, but a bird in the hand is worth two in the bush—and I was happy to realize the 40 percent profit.

Selling the shares turned out to be a big mistake because my analysis had been incomplete. In 1994 and in following years, IBM generated large amounts of excess cash. While a small portion of the cash was used for acquisitions, the largest portion was used to repurchase shares. IBM’s average diluted share count declined by 1.7 percent in 1995, and then by 8.5 percent in 1996, and by another 7.4 percent in 1997. As a result, IBM’s per-share earnings increased materially faster than its net earnings. I had failed to consider the repurchases. Also, my estimate that the company’s after-tax profit margins should be 5 to 6 percent proved to be too conservative. In 1996, the company’s after-tax margin was 7 percent, and IBM was on track to earn about 2.50 per share in that year.

In very late 1995 and in early 1996, I swallowed hard and rebuilt a large position in IBM’s shares at an average cost of about $24½. My reasoning was that the combination of the share repurchases and the improved margins would lead to higher earnings per share (EPS) than was generally predicted by Wall Street.

While I was very wrong to sell the shares in 1994 at about $16, I was very right to purchase them again 15 months later. IBM earned $2.51 per share in 1996 and $3.01 per share in 1997, and the price of the shares increased sharply in response to the improved earnings. I sold the shares in late 1997 at an average price of about $48. Again, my timing was far less than perfect. In mid-1999, the shares traded above $60.

Over the years, I have learned that we can do well in the stock market if we do enough things right and if we avoid large permanent losses, but that it is impossible to do nearly everything right. To err is human—and I make plenty of errors. My judgment to sell IBM’s shares in 1993 at $16 was an expensive mistake. I try not to fret over mistakes. If I did fret, the investment process would be less enjoyable and more stressful. In my opinion, investors do best when they are relaxed and are having fun.

I continued to follow IBM after we sold the shares. The company struggled to grow. During the 15-year period from 1998 to 2013, IBM’s revenues grew at only a 1.3 percent compound annual growth rate (CAGR), from $81.7 billion to $99.8 billion. IBM, once one of the most respected growth companies in the world, now was a mature, slow growth company selling commodity-like products and services.

Warren Buffett wrote in Berkshire Hathaway’s 1998 annual report that, when Berkshire owns shares of a wonderful business, “our favorite holding period is forever.” I greatly admire Warren Buffett. He is one of the great investors of all time. But I strongly disagree that the shares of most wonderful businesses can be held forever because most wonderful businesses become less wonderful over time—and many eventually run into difficulties. IBM is one example of why most stocks cannot be held forever. Kodak is another example. Coca-Cola, which is one of Berkshire Hathaway’s largest holdings, is a third example. Over the 10-year period 2003 to 2013, Coca-Cola’s revenues and EPS increased at CAGRs of 8.3 percent and 7.1 percent, and the price of its shares increased at only a 4.9 percent CAGR. Coca-Cola once was a rapidly growing company. More recently, its markets seem to have matured. Greenhaven strives to achieve annual returns of 15 to 20 percent. It would be near impossible to earn returns anywhere near 15 percent by continually holding shares in a company that is growing 7 to 8 percent per year. My job would be a lot easier and much more relaxing if I could fill a portfolio with outstanding companies that I never would sell. But our ambitions lead us to seek shares that are temporarily deeply undervalued and then sell the shares when they become fully valued. This is an approach to investing that is less relaxing and that requires considerable effort and time, but that has worked for us.

Thus, being a successful value investor is time consuming. There always is another company to study, another periodical to read. I sometimes am asked how I allocate my time. What exactly does an investment manager do? Acknowledging that other managers likely spend their time very differently than I do, the following is an example of how I might spend a typical workday. The example is representative but not completely factual. I awaken at about 5:00 a.m. and, after shaving, exercising, showering, and dressing, walk down the hall to my home office, which is equipped with a Bloomberg. This particular morning, my first business of the day is to search the Bloomberg for general business and world news. I then review a model of FedEx’s projected earnings that I was working on the previous evening. I am tired when I finish the model at about 9:30 p.m. and promise myself that I will take a fresh look at the model after a good night’s sleep. After rethinking some of my assumptions and estimates, I make some inconsequential changes to the model and print out a copy. I then write a one-page memo that explains the model and summarizes my conclusions about the value of the shares. After proofreading the memo, I pack my briefcase, grab a bowl of cereal, and head to Greenhaven’s offices in Purchase, New York, which are located about 15 minutes from our home in Rye.

Upon arriving at our Purchase offices at just after 7:00, I give a copy of the FedEx memo and model to my assistant to distribute to Greenhaven’s three securities analysts (my son Chris, Josh Sandbulte, and me) and to the firm’s trader. I wish to keep my associates fully informed about my thinking. I will keep my copy in a nearby file so that I can refer to it in the future whenever I am working on the company.

I then notice that our trader, Eli, has placed a number of press releases and brokerage reports on my desk. Eli usually arrives at the offices at about 6:30 and immediately searches the Bloomberg for any news or Wall Street reports that may be of interest to us. He distributes hard copies of the news and analysts’ reports to Chris, Josh, and me. I scan the hard copies for any relevant new news or ideas. There is none. There rarely is. But I am interested in the opinions of the Wall Street analysts, not because they will directly influence an investment decision, but because they collectively reflect the conventional wisdom on a security and therefore help me understand why a security is selling at the price at which it is selling.

After scanning the press releases and analysts’ reports, I use a function on the Bloomberg to receive the latest news on our holdings and prospective holdings. Then (usually about 7:30 to 7:45), I get together with Chris and Josh to discuss anything that any of us wishes to discuss, varying from what we are working on, to news or ideas, to suggested purchases or sales for the day, to brainstorming for ideas. Some days the meetings are only about 15 minutes long. On other days they can last for an hour or more. At this morning’s meeting, we have a long discussion about FedEx. For several quarters the company’s results lagged our expectations, but very recently the company’s fundamentals have improved. Does the improvement indicate that our thesis on the company is beginning to work? At the end of the meeting, Chris reminds me that the newly appointed CFO of 3M, Nick Gangestad, is telephoning us at 11:00. I tell Chris that we should get together at 10:45 to organize the questions we will ask on the call.

It now is almost 8:30. I scan a computer printout that gives me the holdings of each of our 150 or so accounts by percentage. For example, Mary Jones’s account breaks down as follows: 9.2 percent of the current value of her account is invested in FedEx, 8.2 percent in UPS, 9.0 percent in Lowe’s, and so on. I have decided that each of Greenhaven’s accounts should have at least a 9.5 percent holding in FedEx and I have noticed that some of our accounts own slightly less than 9.5 percent. I tell our trader Eli to bring each of our accounts’ holdings in FedEx up to 9.5 percent, but only if the account has sufficient cash to do so. Eli can use his computer to determine how much FedEx he will be purchasing for each account. We try to be as automated as possible. The purchase of a relatively small number of FedEx shares will be my only order for the day. Because we tend to hold stocks for two to five years, our trading often is light—and on many days we do not buy or sell a single share of stock.

It is now 8:45. I am interested in adding to our holdings of stocks that will benefit from an expected upturn in the housing market. Previously and repeatedly, I have screened for housing-related companies that have market values in excess of $5 billion and that appear to be undervalued, but to no avail. Today, I will run a screen of housing-related companies that have market values of $3 to $5 billion. Because we have $5.4 billion under management, we normally need to focus on companies that have market values in excess of $5 billion, but I am frustrated to find an additional attractive housing-related stock, and I will try almost anything to find one. I use a function on the Bloomberg to identify companies that fulfill my industry and size criteria. After briefly considering dozens of names that appear on the screen, I write down the stock ticker symbols of five companies that might merit some analysis. Again using the Bloomberg, I pull down the Form 10-K¹ of one of the companies and then start analyzing its balance sheet and reading a description of its business activities. Time passes quickly. It is now 10:15. I cease reading about the housing-related company and pull out my file on 3M. Before speaking to the company’s new CFO, I need to review my previous memos and notes on the company and carefully and thoughtfully prepare a list of questions that will be most helpful to our analysis—and that the CFO is likely to answer. After decades of speaking to managements, I have improved my skills on what questions to ask and how to ask them.

At 10:45, Chris enters my office. We spend about 10 minutes discussing the questions we will ask 3M’s CFO. Then, as we wait for the phone to ring, I ask Eli how our FedEx order is progressing. He responds that we had only 54,000 shares to purchase and that the order was completed at 10:20—and that the shares are up $.10 from our average purchase price. Soon the phone rings. It is the CFO, Nick Gangestad, himself, not his assistant. I amusingly find that the highest-level executives in a company often place and receive calls themselves without the aid of assistants, while lower-level employees tend to use assistants in order to make themselves appear more important. After congratulating Nick on his promotion, I ask questions about his background, from the time he was a child until his promotion as CFO. He grew up on a 1,280-acre farm in Iowa and apparently had a wholesome upbringing, often working with his hands on farm chores. We then spend at least 45 minutes discussing 3M. I focus on areas that seem to be the most critical determinants of what the company will be worth in two or three years. I ask whether the slower economic growth in the developing parts of the world will cause the company to reduce its growth goals or whether the slower growth already was anticipated by the company when it established its goals. I then ask whether the highly profitable and rapidly growing health care segment can hold its operating margins above 30 percent. I also ask whether the company plans to increase its prices outside the United States if foreign currencies (and especially the euro and yen) decline in value versus the U.S. dollar. We then discuss the company’s share repurchase program and its projected pension expense. Originally, the company had expected that its annual pension expense would decline to close to zero by 2017, but Chris learned that the actuarial tables used to compute pension expense are being altered to reflect the longer life expectancy of Americans, and this alteration will cause the future pension expense of many companies to be materially larger than previously expected.

After the call ends, Chris and I review what we learned, and decide we are hungry and it is time for lunch. Frequently, Chris, Josh, and I lunch at a nearby restaurant. Today, we lunch at a local country club, where the food is good and the service is quick. Chris and I fill Josh in on our call with Nick Gangestad, and then we spend the rest of lunchtime brainstorming about industries and companies that might indirectly be helped by an upturn in the housing market. We do not come up with any worthwhile ideas. We usually strike out when we brainstorm. But we only need a handful of good ideas a year—and we keep trying.

We return to the office at 1:45. Mary, my assistant, tells me that Todd Corbin, the chief investment officer of the New York Public Library, telephoned while I was out. I am chair of the library’s Investment Committee, which is responsible for the its $1.1 billion endowment, and Todd is the library’s full-time employee who suggests strategies, recommends outside investment managers, and continually monitors the outside managers. At the previous committee meeting, I had raised a theoretical question. What, if anything, should the library do if the stock market climbs to excessive heights? Should the library hedge the stock market or ask some of its managers to sell part or all of the equities they hold in the library’s account? While I was not predicting that the stock market would climb to excessive levels, I thought it would be beneficial if we researched our options now so that we would be best prepared to act in the future should the market rise sharply from present levels. I gave Todd and each committee member a homework assignment to think about our options and strategies. Neil Rudenstine, the former president of Harvard University, is a member of the Investment Committee, as is Tony Marx, the former president of Amherst College. After giving out the homework assignment, I chuckled out loud and said: “You know, when I was a student at Williams College, I was in absolute awe of the college’s president—he was a godlike figure—and it would have been beyond my wildest dreams that someday I would be handing out homework assignments to the former president of Harvard and to the former president of Amherst.”

I talk to Todd for about one hour. Protecting the portfolio might be more difficult than hoped. A small percentage of our equities are held in discrete accounts, but the bulk is held in pooled accounts. We can ask the managers of the discrete accounts to raise cash, but we have no control over the pooled accounts—and we would not wish to withdraw our funds from the pooled accounts because the managers are good and most of them are no longer accepting new accounts. Therefore, if we withdraw the funds, we likely would not be able to reinvest with them once we decided to again increase our exposure to equities. Todd had checked on the practicability of purchasing put options on the S&P 500 Index, but the options seem to be prohibitively expensive. He also checked on shorting the Standard & Poor’s (S&P) 500 Index, but to do so the library would have to provide cash as collateral for any unrealized losses. Few things in life are easy. Todd said he would keep looking for ideas.

I did not mind spending an hour with Todd. Over the years, I have spent considerable time with not-for-profits. Being on boards and committees is worthwhile, but chairing committees is particularly worthwhile because the buck stops with you and therefore you are forced to become actively engaged. Over the years, I have chaired five investment committees, one executive committee, two finance committees, one audit committee, one art curatorial committee (at the Museum of Modern Art), two capital campaigns, and two boards of trustees. In each case, I learned more than I gave—and what I learned has made me a better investor.

The call with Todd ends. It is about 2:45. I return to analyzing the five smaller housing-related companies. I reject the first because it seems to be a higher-cost producer of undifferentiated products. I reject the second and third because their balance sheets and cash flows do not seem sufficiently strong. The time is now 3:45. I usually leave the office at 3:45 to 4:00. I check my calendar for the following day, pack my briefcase with all my material on 3M, and leave for Rye.

Arriving home at 4:00, I read the mail, take a shower, and change into comfortable clothes. I then return to my home office. From 4:45 until dinnertime at 6:15, I think about 3M and our conversation with Nick Gangestad, and I start revising my principal memo on the company. After dinner, I complete the revision of the memo and then update my earnings model. I review our reasons for owning the shares and our valuation. I remain happy we own the shares. By now, it is close to 9:45. I am tired. It is time for bed.