Warren Buffett said: “I’ve never gone to bed with an ugly woman, but I’ve sure woke up with a few.”¹ Such was our case with AIG.

The 1998–2000 bubble in technology stocks likely was a key indirect reason we purchased AIG’s shares. In 2000, the bubble burst and the stock market started to decline sharply. The Nasdaq Composite Index, which is laden with technology stocks, declined from a high of 5,048 on March 10, 2000, to a low of 1,114 on October 9, 2002—a decline of 78 percent. The Standard & Poor’s (S&P) 500 Index, which is more representative of the entire U.S. stock market, declined from a high of 1,521 on September 1, 2000, to a low of 801 on March 11, 2003. The magnitude of the collapse encouraged many investment committees and individual investors to seek investments that were less volatile than the stock market, and many committees and investors opted to invest in hedge funds that promised to short stocks and thus to sharply reduce downside volatility. According to BarclayHedge, the assets under the management of hedge funds increased from roughly $300 billion at the end of 2000 to more than $2 trillion by 2007.

Hedge funds typically charge their clients fees equal to 1 to 2 percent of the assets under management plus incentive fees of 20 percent of any profits. In my opinion, given this very high fee structure, hedge funds are bent to making obscure investments as opposed to, for example, purchasing blue-chip stocks. It would be difficult for a hedge fund to justify its high fees if it heavily owned shares in Exxon, 3M, Berkshire Hathaway, General Electric, Procter & Gamble, or other heavily followed and commonly owned companies. By 2006, because hedge funds had become a large force in the stock market and because hedge funds generally had shied away from blue-chip stocks, the shares of many of the largest and highest-quality companies were selling at relatively depressed prices.

In the meantime, during the period 2003 to 2005, Greenhaven’s portfolios had benefited from the resurgence of the “old economy stocks” and had been abnormally buoyant. Between January 1, 2003, and December 31, 2005, a typical Greenhaven account appreciated by about 140 percent (including dividends received). By early 2006, many of our holdings had become relatively fully valued, and we sought to replace these holdings with stocks that were undervalued and out of favor. Blue-chip stocks seemed to fit the bill, and over a period of several quarters we purchased shares in GE, FedEx, 3M, and, unfortunately, AIG.

AIG’s predecessor company was founded in Shanghai in 1919 by Cornelius Vander Starr, who previously had owned an ice cream store. For a number of years, the company’s main business was selling life insurance policies in China. In 1926, Starr opened an office in New York City to offer casualty insurance to American companies conducting business abroad. After World War II, the communist government prohibited Starr’s companies from doing business in China, but Starr more than compensated for the lost Chinese business by aggressively expanding elsewhere. After Starr died in 1967, Maurice “Hank” Greenberg became CEO of the several disparate insurance companies founded by Starr. Two of Greenberg’s early actions were to merge the several companies into a newly formed AIG, and then to take the new company public.

Over the years, Hank Greenberg built AIG into a very large, highly profitable, and highly regarded “growth” company. During the 15-year period 1989 to 2004, AIG’s revenues and earnings both grew at a 14.1 percent compound annual growth rate (CAGR), and the company’s share price increased at about a 17 percent CAGR. AIG was a winner.

However, in 2004 and 2005, the company hit legal bumps when it was accused of several wrongdoings, including engaging in sham reinsurance transactions that were designed to bolster its reserves. Hank Greenberg was forced to resign in early 2005, and Martin Sullivan became the new CEO. Whereas the S&P 500 Index appreciated moderately between 2004 and 2006, AIG’s shares remained flattish. In the spring of 2006, I reasoned that the 2004 and 2005 scandals had adversely affected the price of AIG’s shares, that the shares were out of favor with hedge funds and other aggressive investors, and that, for these reasons, we now had the opportunity to purchase the shares at a materially undervalued price.

I started analyzing AIG by reading the company’s Form 10-K and annual report. There was nothing in the 10-K that surprised me. When reading the annual report, I realized that the reported earnings and balance sheets of any insurance company are no more than estimates because managements, actuarial firms, and independent accountants must estimate the magnitude of recent and future losses—and such estimates often are no more than educated guesses. However, AIG had taken a $1.82 billion pretax charge in late 2005 to increase its reserves, and my reasoning was that the new management had every incentive to take as large a charge as possible. After all, the new management could blame the old management for the charge, and a large charge now would lead to larger earnings in the future. Thus, the charge gave me some comfort that AIG’s reserves now were accurately stated, if not conservatively stated.

Insurance companies need to be financially strong, and in the annual report’s letter to shareholders, management emphasized that “AIG remains among the most strongly capitalized organizations in our industry.” AIG’s credit ratings were the equivalent of AA, and management commented that “at these levels, AIG’s ratings are among the highest of any insurance and financial services organization in the world.”

I telephoned two friends who were CEOs of insurance companies. CEOs often do not like to praise their competitors, but both my friends gave AIG rave reviews. In particular, both stated that AIG’s size and capital strength gave the company a competitive advantage. Medium-sized insurance companies often do not have sufficient capital and scope to meet the insurance needs of large, international Fortune 500 companies. Therefore, AIG sometimes faced limited competition on very large policies—and limited competition often led to high premiums and high profits.

One of my CEO friends invited my wife and me to have dinner at his home with Martin Sullivan, AIG’s new CEO. I leapt at the opportunity. Of course, one cannot judge the capabilities of a CEO from a three-hour social evening, but nonetheless, I found Martin Sullivan to be affable, knowledgeable, and highly intelligent. I left the dinner with a positive feeling about AIG’s leadership.

After additional reading, thinking, and debating, I made a model of AIG’s possible future earnings per share (EPS) in a normal environment. I used “normalized” earnings because the reported earnings of an insurance company can vary from year to year, especially due to the occurrences of hurricanes or other catastrophes. My conclusion was that the company’s normalized 2008 EPS should be about $7 per share. The $7 would equal about a 15 percent return on a projected 2008 book value of about $46.50 per share. The company seemed to earn a 15 percent return on book value in normal years, so the $7 EPS estimate appeared to be reasonable.

Next, I valued AIG. Upon reflection, my best thinking was that AIG’s shares were worth close to an average price-to-earnings (PE) ratio, or about 15 times earnings. Therefore, I concluded that, in 2008, AIG’s shares would be worth roughly $105, or nearly twice their existing price of about $55. Based on these numbers and the first-class reputation of the company, I started building a position in the shares.

The year 2006 ended up to be good for AIG. EPS, before some unusual accounting adjustments for hedging, was a favorable $5.88. In the fourth quarter of the year, management studied the company’s capital position and concluded that the company was overcapitalized by $15 to $20 billion. As a result of the study, in March 2007, management authorized the repurchase of $8 billion shares of the company’s stock, and in May, management increased the company’s dividend by 20 percent. I was smiling.

Toward the end of 2007, there was some general nervousness in the financial markets, and the price of AIG’s shares slipped from about $57 at the end of September to about $49 at the end of December. The shares continued to slip some in early 2008 and then declined sharply to a low of $33 when it appeared that Bear Stearns might declare bankruptcy. But the shares started to recover after J. P. Morgan announced that it would purchase Bear Stearns, and by mid-April the shares had recovered to the $40 level.

In late April, the price of AIG’s shares still was materially below our cost basis. From time to time, one or more of Greenhaven’s holdings run into headwinds. The economy and the financial markets are cyclical. Headwinds come with the territory of investing. When shares of one of our holdings are weak, we usually revisit the company’s longer-term fundamentals. If the longer-term fundamentals have not changed, we normally will continue to hold the shares, if not purchase more. In the case of AIG, it appeared to us that the longer-term fundamentals remained intact. The company did have to report losses from the marking to market of a number of assets, including derivative contracts in its Financial Products subsidiary, which insured such financial risks as bank loans. The markdowns largely were necessitated by generally increasing yields on higher-risk assets, not by concerns that AIG’s loss experiences materially would deteriorate. Therefore, our reasoning was that, when the derivative contracts matured, most or all of the losses would be reversed. In Berkshire Hathaway’s 2007 annual report, Warren Buffett had complained about the need to mark derivative contracts to market: “Changes in the value of a derivative contract … must be applied each quarter to earnings. Thus, our derivative positions will sometimes cause large swings in reported earnings, even though Charlie (Munger) and I might believe the intrinsic value of these positions has changed little. He and I will not be bothered by these swings. … And we hope you won’t either.” Further, we reasoned that, if the credit rating companies or the regulators were concerned about the write-downs, AIG could always raise cash and capital by selling new common shares, preferred shares, or subordinated debt. Indeed, on May 8, AIG announced that it would raise about $20 billion of new capital through the sale of new common shares and subordinated debentures.

However, the nervousness continued, and during June and July 2008, AIG’s shares generally sold between $20 and $25. In mid-August, I noticed that a key director of AIG, who at the time chaired the board’s finance committee and who had previously chaired the audit committee, had purchased 30,000 AIG’s shares for his own account on August 12. These purchases gave me considerable comfort. Certainly, the chair of the finance committee would not purchase shares if he were concerned about AIG’s future, and certainly he should be unusually well informed about the fundamentals of the company.

On the morning of September 15, all hell broke loose when Lehman filed for Chapter 11 bankruptcy protection. Katie bar the door. The filing triggered a meltdown that fed on itself. Extreme illiquidity in the financial markets caused asset values to fall sharply. The decline in asset values caused financial institutions to mark down the carrying value of their assets, which, in turn, caused sharp reductions in their credit ratings. Sharp reductions in credit ratings required financial institutions to raise capital and, in the case of AIG, to post collateral on its derivative contracts. But the near freezing of the financial markets prevented the requisite raising of capital and cash and thus caused a further deterioration in creditworthiness, which further increased the need for new capital and cash, and so on. In the wake of the collapse of Lehman and the near freezing of the financial markets, the price of AIG’s shares fell sharply. The company needed a large influx of cash to post as collateral, but with the markets close to frozen, the cash could not be raised. On Tuesday night, September 16, the U.S. government agreed to provide the requisite cash in return for a lion’s share of the ownership of AIG. As soon as I read the agreement, it was clear to me that we had a large permanent loss in our holdings in AIG.

The failing of Lehman and the resulting severe financial crisis were outlier events that had not been part of our thinking or planning. Over the next few weeks, I worked long hours trying to get a handle on what was happening. It was difficult to think clearly and unemotionally in the midst of the most serious economic crisis since the Great Depression. Given the dangers and uncertainties, we had to act. Within several weeks of the failure of Lehman, we sold the bulk of our holdings in financial service companies and we sold several stocks that had not declined sharply and, therefore, that had become less attractive relative to the stock market in general. These actions increased our cash holdings to about 40 percent of the value of our portfolios.

During the next few months, we thought long and hard about the economic environment and about our portfolios. It soon became clear to me that our nation faced two interrelated, yet disparate, problems: (1) the financial crisis and (2) a deep recession. I decided that, as long as the financial crisis continued, we would hold on to our large holdings of cash. I was fearful that, if the crisis did not end, the deep recession could turn into another Great Depression. However, I reasoned that if the crisis ended, the opportunity to reinvest the cash in abnormally undervalued common stocks likely would be compelling. In my opinion, many corporations had sharply reduced their inventory levels during the financial crisis in order to raise cash, and these inventory reductions were a major cause of the recession. Once the crisis ended, corporations likely would replenish at least some of their inventories—and the replenishment would be a stimulant to the overall economy. Furthermore, during the crisis many individuals and corporations deferred purchasing large-ticket items that they had either needed or wanted. I believed that after the crisis ended, there would be at least some deferred demand for many goods and services. Also, by late 2008, it was clear to me that the U.S. government and the Federal Reserve Bank would take aggressive actions to stimulate the economy. Thus, I concluded that, once the financial crisis ended, there would be a high probability that the economy would bounce back some, maybe sharply.

In late 2008 and early 2009, we carefully monitored the interest rate spreads between U.S. treasuries and lesser quality debt instruments—and we also searched for other signs that the financial crisis was continuing or was abating. By March, interest rate spreads had narrowed and substantial confidence had returned to the financial system, and I decided to reinvest the cash in our portfolios. Between the end of 2007 and March 31, 2009, while the S&P 500 Index had declined by about 45 percent, the subsector of the Index that was composed of industrial companies had declined by about 54 percent. Industrial companies are more economically sensitive than most other types of companies (such as consumer durables, pharmaceuticals, and utilities), so it is not surprising that industrial stocks declined by more than the market as a whole. When purchasing stocks in the spring of 2009, we mainly purchased shares of industrial companies. They not only were more depressed than most other sectors of the market, but also they would more directly benefit from a bounce-back of the economy. Our strategy succeeded. A typical Greenhaven portfolio, after losing about 38 percent of its value in the disastrous year 2008, appreciated by roughly 47 percent in 2009 and 21 percent in 2010. By the end of 2010, the value of a typical Greenhaven portfolio was close to 10 percent higher than at the end of 2007.

AIG had been a disaster for Greenhaven but not for the U.S. government. After the crisis, AIG skinnied down and enjoyed good operating earnings. The U.S. government benefited from the recovery in AIG’s fortunes and gradually reduced its investment in the company, selling the last of its holdings on December 11, 2012. According to the U.S. Treasury, the Federal Reserve Bank and the U.S. Treasury together earned a $22.7 billion profit on its “investment” in AIG.

Some criticize the U.S. government for bailing out financial institutions during the financial crisis. I disagree with the critics. As one of the only suppliers of capital to financial institutions during the crisis, the government had the bargaining power of a monopoly and therefore was in a position to negotiate particularly favorable terms. Most of the financial institutions, including AIG, continued to have strong and viable businesses. Their problem was liquidity, not solvency. No wonder the government made a profit on most of its bailouts.

Furthermore, and most importantly, the bailouts helped restore confidence in the financial system—and the restored confidence helped the economy climb from recessionary levels. The bailouts were a pragmatic partial solution to a very serious problem. Many idealists took the position that, in principle, the government never should bail out private enterprises. I believe that humans and institutions need to be pragmatic and flexible and that strong ideologies and tunnel visions can lead to failures that end up hurting people.

A few weeks after the collapse of AIG’s shares, a friend and client asked me if I had learned anything from the horribly failed investment. I told him that we were still in the fog of war and that I would have a clearer picture of our mistake after the financial crisis ended. Months later, I did reflect on the investment. I reviewed the work we had performed and the information we had learned, both of which were extensive. I then asked myself the question: if I had to make the purchase decision today based on what we knew then, would I still make the same decision to purchase the shares? My answer was “yes”—and my conclusion was that, in the investment business, relatively unpredictable outlier developments sometimes can quickly derail otherwise attractive investments. It comes with the territory. So while we work hard to reduce the risks of large permanent loss, we cannot completely eliminate large risks. However, we can draw a line on how much risk we are willing to accept—a line that provides sufficient apparent protection and yet prevents us from being so risk averse that we turn down too many attractive opportunities. One should not invest with the precept that the next 100-year storm is around the corner.

Luckily, in practice, usually we become aware of adverse changes sufficiently early that we can sell shares before any loss becomes too large. For example, in 2014, we purchased shares in Praxair, a manufacturer of oxygen, nitrogen, hydrogen, and other industrial gases. The shares appeared undervalued, and we believed that the company’s earnings growth likely would accelerate because an aggressive new management at Air Products, a key competitor, would attempt to control capacity and increase prices. A few quarters after we purchased the shares, it became apparent to us that the new management at Air Products was more interested in reducing costs than in increasing prices. Furthermore, about 14 percent of Praxair’s earnings came from Brazil, where the value of the real (Brazil’s currency) suddenly had declined sharply because of economic and political problems. We were fearful that the decline would be long lasting. Thus, our original projections and valuations for Praxair had become too optimistic. We reacted by selling the shares at a modest loss.

In my investment career, I frequently have sold shares at modest losses after realizing that my original valuations were too high. We have suffered many Praxairs. I find that investing is not about earning a favorable return on every holding—it is about developing a favorable batting average.