Lowe’s is the second-largest home improvement retailer. Through about 1,850 stores spread throughout the United States and Canada, Lowe’s sells roughly 40,000 house-related items, including lumber, wallboard, flooring, appliances, kitchen and bathroom cabinets, plumbing fixtures, lighting fixtures, paint, power tools, outdoor furniture, grass seed, plants, and fertilizer. The company’s sales volumes closely track with the strength of the housing industry.

Lowe’s was founded in 1921 when Lucius S. Lowe opened a hardware store in North Wilkesboro, North Carolina. After Lucius Lowe died in 1940, the store was owned and operated by several family members, and especially by Carl Buchan, a son-in-law. In the 1950s, Buchan opened several new stores that sold building supplies as well as hardware. Buchan died suddenly in 1960. Robert Strickland, the new CEO, took the company public in 1961 and continued to expand its footprint. Lowe’s revenues increased from $25 million in 1960, to more than $150 million in 1970, and to nearly $900 million in 1980.

Historically, Lowe’s had concentrated on selling to the professional builder. But when the market for new houses weakened in 1980, the company started redesigning its stores to appeal to the do-it-yourself homeowner. At about that time, a new competitor, The Home Depot, started opening big-box warehouse stores that typically were five times larger than the stores operated by Lowe’s. By 1989, it was clear that Home Depot’s large store format was the way of the future. Lowe’s took an impairment charge in 1991 to close smaller stores—and management then started aggressively opening big-box stores to compete with Home Depot. In 1993 alone, Lowe’s nearly doubled its floor space by opening 57 new stores, each averaging close to 100,000 square feet in size. The race with Home Depot was on.

In 1993,¹ Lowe’s had revenues of $4.54 billion and Home Depot had revenues of $9.24 billion. Over the next 10 years, Lowe’s revenues increased at a 21 percent average annual rate to $30.84 billion and Home Depot’s revenues increased at a similar rate. Both companies grew rapidly at the expense of small building supply and hardware stores. Small stores simply could not compete with the purchasing powers, distribution efficiencies, merchandise varieties, efficiencies of scale, and low real estate costs enjoyed by the two big-box retailers.

Lowe’s prospered during the 2001–2005 period. The company not only benefited from the boom in housing, but also from taking market share from Home Depot, which was suffering from some self-inflicted merchandising problems. During the five-year period that ended on January 31, 2006, Lowe’s sales per store (comps) increased at a 5.4 percent average annual rate, while Home Depot’s comps only increased at a 2.5 percent average rate. Lowe’s was flying high. The company’s earnings climbed from $0.53 per share in 2000 to $1.73 per share in 2005. And the price of its shares increased 150 percent during the five-year period, hitting $34.85 in late 2005. Lowe’s shareholders were smiling.

Then, the housing boom turned into a bust. In the three-year period 2007 to 2009, Lowe’s comps declined by an aggregate of 17.8 percent—a near disaster. Earnings, leveraged downward by the weak sales, fell by 39 percent, from $1.99 per share in 2006 to $1.21 in 2009, and earnings would have declined further if management had not taken strong steps to reduce costs. The price of the company’s shares fell to a low of $13 in March 2009.

The housing bust was worse than I had imagined. Overbuilding during the years 2004 to 2006 and a high level of foreclosures led to a glut of empty homes. With vacant homes at record levels, the construction of new housing units collapsed from close to 2 million in 2006 to only 585,000 in 2011.

In the spring of 2011, the prevailing opinion on Wall Street was that the housing market would continue to be very weak for some time, largely because of a “shadow inventory” of homes near foreclosure that likely would materially add to the inventory of unsold homes. Because of this prevailing negative sentiment, the shares of housing-related companies were selling at depressed prices. I was intrigued by the depressed prices and by the logic that the housing industry eventually had to recover strongly. Therefore, I decided to conduct my own study of the market. My conclusion was that the market could turn surprisingly strong within the next several years. My logic and methodology were as follows.

First, I estimated how many new housing units were required each year to satisfy normal demand. Normal demand is equal to the net increase in the number of families in the United States, plus the number of houses torn down each year, plus the increase in the number of vacation and other second homes.

In 2011, there were about 131 million housing units in the United States. Between 2000 and 2010 the population of the United States increased at a 0.92 percent compound annual growth rate (CAGR) from 282.16 million to 309.33 million. I reasoned that if the population continued to increase at a 0.92 percent CAGR, there would be a need for about 1,200,000 additional housing units per year (131 million times 0.92 percent). I also read a lengthy report issued in September 2010 by Harvard University’s Joint Center for Housing Studies that predicted that net family formations during the decade 2010 to 2020 would be in the range of 1,180,000 to 1,280,000 per year. After reading a few other studies, I decided to estimate that at least 1,200,000 new housing units would be required in a normal year to meet the annual net increase in family formations plus the increase in demand for second homes. I also estimated that about 300,000 additional new housing units would be required annually to replace houses that were demolished because of age, fire, flood, location, and so on. Therefore, I concluded that at least 1,500,000 new housing units were needed in a normal year.

Before conducting further analysis, I decided to check my 1,500,000 estimate against historical data. In the 20-year period 1980 to 1999, the average annual number of housing units completed in the United States was 1,430,000. The population of the United States during the 20-year period averaged about 250 million. Adjusted to the 2011 population of 311 million, 1,430,000 completions would be the equivalent of about 1,780,000 units in 2011. Looking at a more recent period, annual housing completions during four years, 2000 through 2003, averaged about 1,620,000, in spite of the adverse effects of 9/11 and the mild recession. These data gave me confidence that, even if demographics were somewhat less favorable post the financial crisis, my 1,500,000 estimate for normal demand was reasonable, if not conservative.

The number of new housing units built in 2010 in the United States was only 650,000, and it appeared that the number would decline to below 600,000 in 2011, or at a depression level of only 40 percent of estimated normal demand. Construction could not remain at 40 percent of normal demand forever. People have to live somewhere. It was clear to me that the housing market would recover and that the recovery would be very strong, with housing completions eventually increasing 2.5-fold to about the 1,500,000 level. The remaining unknown was the timing of the recovery.

To estimate the possible timing of the recovery, I used Census Bureau data on housing completions to calculate that, if the normal annual demand for new housing is 1,500,000 units, then 1,400,000 excess houses were built during the boom years 2004 through 2007. Another consideration was the effect of foreclosures on the housing market, a consideration that seemed to befuddle many Wall Street analysts. When thinking about foreclosures, I adopted the following approach. When a house is foreclosed, the foreclosure adds to the inventory overhang only if the former occupant moves in with another family (typically a parent or a friend), as opposed to purchasing another house or leasing a rental unit. Thus, if I were able to estimate the increase in “doubled-up” households, I would know the number of house vacancies that were created by the foreclosures or by the loss of jobs. Such figures were available. The number of doubled-up families had increased by about 2 million during the years 2008 through 2010 and seemed to have stabilized at that level.

Therefore, before consideration of the recent underbuilding of housing units, I concluded that the overbuilding in the 2004–2007 period and the foreclosures in the 2008–2010 period increased the inventory of vacant homes by about 3,400,000 housing units (1,400,000 from overbuilding during 2004–2007 plus the 2 million vacated because of foreclosures). Again using Census Bureau data, I then calculated that the underbuilding during the years 2008 through 2010 totaled 1,900,000 units (vs. the normal annual need of 1,500,000 new units) and therefore that, during the entire 2003–2010 period, the inventory of vacant homes increased by 1,500,000 units. Since housing inventories at the end of 2003 seemed to be at normal levels, I concluded that the estimated inventory overhang at the end of 2010 was approximately 1,500,000 units. In 2011, only about 600,000 new housing units were being built. If construction continued at the 600,000 level and if normal demand was 1,500,000, the housing industry would be in balance by late 2012, assuming that the number of doubled-up families had stabilized, which seemed to be the case.

I was excited that we had a concept about a probable strong upturn in the housing market that was not shared by most others. I believed that the existing negativism about housing was due to the proclivity of human beings to uncritically project recent trends into the future and to overly dwell on existing problems. When analyzing companies and industries, I tend to be an optimist by nature and a pragmatist through effort. In terms of the proverbial glass of water, it is never half empty, but always half full—and, as a pragmatist, it is twice as large as it needs to be.

After I became enthusiastic about an upturn in the housing market, I started looking for investments that would benefit from the improvement. My first instinct was to study the publicly owned homebuilders. However, their balance sheets were not strong, and I was concerned about risks of insolvency if the housing market did not improve over the next few years. I continually attempt to minimize risks of permanent loss and continually am mindful of Warren Buffett’s two rules for successful inventing: “Rule number one is never lose money; rule number two is never forget rule number one.”

My second instinct was to study the home improvement retailers, and I soon became excited about Lowe’s, largely because I believed that the company would benefit from two positive changes: (1) the upturn in the housing market and (2) improved merchandising. During the years 2001 to 2005, when Home Depot was suffering from merchandising problems and when Lowe’s was gaining market share, Lowe’s apparently became complacent and let its merchandising slip. The company now needed to study each product it sold and eliminate poorly selling products, introduce new exciting products, obtain lower prices from its suppliers, optimize the prices it charged its customers, find the optimum selling space for each product, adjust inventory levels to minimize stock-outs and the need for markdowns, and modernize advertising and signage. Lowe’s management seemed serious about improving the company’s merchandising. I reasoned that the improvements required attention and effort, not rocket science, and that there was a high probability that substantial progress could be made within two to three years.

After studying Lowe’s financials and other fundamentals, I constructed a model of normalized 2014 earnings. My model assumed that the housing industry recovered and that the company’s merchandising materially improved. First, I estimated what revenues might be in 2014. In 2011, Lowe’s stores contained a total of 197 million square feet of selling space. The company was projecting that its selling space would increase at a 1 percent annual rate. Consequently, I projected that its selling space in 2014 would total 203 million square feet. The company’s sales per square foot in 2010 were $250, down from $302 in 2003. I believed that the company’s revenues in 2003 were at a normal level. Therefore, I assumed that sales per square foot would return to the $302 level, before adjusting for subsequent inflation, and to $337 assuming that the inflation rate between 2003 and 2014 averaged 1 percent.² Thus, I projected that Lowe’s normalized revenues in 2014 could be $337 per square foot times 203 million square feet, or $68.4 billion.

Next, I projected operating profit margins. The company’s margins in 2003 were 10.5 percent before nonrecurring costs to open new stores. I saw no reason why the company’s margins could not return to the 10.5 percent level by 2014 if the housing industry recovered and if the company’s merchandizing improved sufficiently.

Operating margins of 10.5 percent on $68.4 billion of sales would produce $7,180 million of operating profit. To calculate after-tax earnings, I subtracted projected interest expense of $275 million and taxes at a 38 percent rate. To then calculate earnings per share (EPS), I divided projected after-tax earnings by the 1.4 billion shares that were outstanding. My conclusion was that normal EPS in 2014 could be a bit over $3.

Finally, I valued the shares. There was a lot to like about Lowe’s. It had a strong balance sheet. It was generating large amounts of excess cash. It essentially was a duopoly. It enjoyed a favorable reputation. However, Home Depot and Lowe’s had saturated the United States with stores, and therefore, once housing recovered, Lowe’s future growth likely would be relatively slow. On balance, I decided to value Lowe’s at 16 times earnings, which is slightly above the average historical price-to-earnings (PE) ratio of the stock market. Therefore, I believed that Lowe’s shares in 2014 would be worth slightly in excess of $48. The shares at the time were trading at about $24. They appeared to be a bargain. I started purchasing the shares and over the next several months built a large holding.

On December 7, 2011, Lowe’s management held a meeting for analysts and shareholders. At the meeting, the company disclosed its “road map” projections for 2015. Importantly, the projections assumed that housing prices only increased modestly by 2015 and that housing completions only reached the 900,000 to 1 million level. The company’s road map estimated that revenues would increase at a 4.5 percent average annual rate to $58.7 billion in 2015 and that operating profit margins would increase to 10 percent.

The road map also projected that Lowe’s would repurchase $18 billion of its shares over the four year period 2012 to 2015. The $18 billion was a mind-boggling surprise. The entire market value of the company at the time was only $35 billion, so the company was planning to repurchase 51 percent of its existing market value over the next four years. The company projected that the repurchases would reduce the number of shares outstanding from about 1,400 million to 900 million. The extremely aggressive repurchase program would materially increase Lowe’s earnings per share. It also was a strong signal of management’s confidence in the company’s future as well as its interest in the price of Lowe’s shares.

How could the company afford to repurchase $18 billion of its shares? There were several considerations. Earnings would increase sharply. Capital expenditures would trail depreciation by an aggregate of $1 billion because the company did not intend to open many new stores. New computer systems and the new merchandizing programs would reduce the company’s investment in inventories by $1 billion. Also, the company would generate close to $7 billion of cash by increasing its debt. Management believed that the company’s balance sheet was underleveraged and that the increased debt could be supported by cash flows and asset values.

After the December 7 meeting, I revised my earnings model. I revise models frequently because my initial models rarely are close to being accurate. Usually, they are no better than directional. But they usually do lead me in the right direction, and, importantly, the process of constructing a model forces me to consider and weigh the central fundamentals of a company that will determine the company’s future value.

My revised model focused on projected results for the year 2015. Management was projecting 4.5 percent average annual revenues growth without a major recovery in housing. After thinking and making some submodels, I projected that average annual revenue growth could be 6.5 percent assuming that the housing recovery was quite, but not very, strong. If the annual growth rate were 6.5 percent, revenues in 2015 would be about $64 billion.

I had a problem determining the operating profit margin I should use in the revised model. Management had stated that each $1 of incremental revenues would increase operating profits by about $0.20. Based on my knowledge of the company, the 20 percent incremental made sense. Given a 20 percent incremental, if Lowe’s revenues increased for four years at 6.5 percent per year instead of the road map’s 4.5 percent per year, then the company’s projected operating margin in 2015 would be 1.6 percent above the road map’s 10 percent (20 percent of 2 percent for four years). However, the company’s margins had never been as high as 11.6 percent before, and my instincts guided me to a more conservative estimate. I settled for 10.5 percent.

Continuing with the revised model, if revenues were $64 billion, operating margins were 10.5 percent, interest expenses were $650 million (higher than my previous model’s $275 million because the company planned to increase its debt by close to $7 billion), income taxes were at a 38 percent rate, and the number of shares outstanding were 920 million, EPS would be $4.10. We had paid only about $24 for the shares. If 2015 earnings were close to $4.10, the shares should be a complete winner, and if we were too optimistic and earnings were only at the $3 level, the shares still should be an attractive investment. Furthermore, because of the quality of the company and because its shares were selling at a depressed price, our risks of permanent loss were low. Lowe’s was a sweet dream investment.

Over the next eight months, the price of Lowe’s shares fluctuated some, but on balance remained relatively flat. In August 2012, the shares were trading at $27 to $28—not much above what we paid for them. Then, in the fall of 2012, the housing market started to improve. The seasonally adjusted annualized rate of housing starts, which had remained in the range of 700,000 to 750,000 during the first half of 2012, increased to 854,000 in September and to 983,000 in December. It appeared that the housing recovery was under way, and Lowe’s shares responded, rising to $35.52 at year-end 2012 and to about $45 in August 2013.

In late September 2013, as I was walking off a tennis court in a good mood because I had played well and had handily defeated my opponent, I was approached by another money manager who asked for my current opinion of Lowe’s shares. After I responded favorably, he agreed that shares still looked attractive, but then added: “But aren’t you concerned that the stock market will decline sharply if the budget impasse in Congress leads to a shutdown of the government?” The other manager was clearly concerned about Washington and its effect on the economy and the stock market. He was selling shares to raise cash. I answered that I had no idea what the stock market would do in the near term. I virtually never do. I strongly believe in Warren Buffett’s dictum that he never has an opinion on the stock market because, if he did, it would not be any good, and it might interfere with opinions that are good. I have monitored the short-term market predictions of many intelligent and knowledgeable investors and have found that they were correct about half the time. Thus, one would do just as well by flipping a coin.

I feel the same way about predicting the short-term direction of the economy, interest rates, commodities, or currencies. There are too many variables that need to be identified and weighed. I agree with the gist of the story about the only two economists in the world who fully understand current trends in interest rates. They both live in Switzerland. And their views are diametrically opposed to each other’s.

I had a personal experience being wrong about the near-term price of two commodities: crude oil and natural gas. The price of both increased sharply between mid-2007 and early summer 2008: crude from about $75 per barrel to over $140 and natural gas from about $5 per thousand cubic feet to more than $10. A friend and I were worried that the high prices would hurt the U.S. economy and balance of payments—and transfer substantial wealth from deserving middle-income and poorer Americans to undeserving and potentially hostile OPEC countries. We decided to take action. We wrote a paper on the need for a U.S. energy policy to encourage both domestic production and energy conservation. We called several senators and congressmen whom we knew and suggested that they actively work to solve our energy problem. Our pleas fell on deaf ears. One senator told us that he agreed with everything we said and that he had written a book on the subject several years earlier. His office sent us a copy of the book. I read it. It was hogwash.

As it turned out, our concerns were overblown. We did not foresee that energy prices would plummet during the financial crisis and deep recession. And neither did we foresee that the technologies of horizontal drilling and multistage fracking would lead to large production increases of crude oil and natural gas—actually leading to a glut of gas. I was wrong about the prices of oil and gas, and if I tried to predict the short-term prices of almost any other commodity, or any currency, or any market, I likely would be wrong about as often as I would be right.

While Greenhaven spends little time studying the short-term outlook for the economy and the stock market, we do spend considerable time studying the short-term fundamentals of the companies in our portfolios, and by the fall of 2014, it had become apparent to us that Lowe’s was making considerable progress improving its merchandising. On November 19, the company reported that its comps (sales per store) had increased by a surprisingly strong 5.1 percent in the quarter that ended on October 31 and that its EPS had increased year over year by 25.5 percent. Robert Niblock (chairman and CEO) attributed a large part of the progress to the “internal initiatives that we have been working on.” Greenhaven was becoming excited that Lowe’s seemed to be entering a period of accelerated revenues and EPS growth.

Then, on December 11, Lowe’s held a half-day meeting for analysts and investors. At the meeting, management projected that, between 2014 and 2017, the company’s revenues would increase at a 4.5 to 5.0 percent CAGR, that its operating margins would increase by 2.5 percent to about 11 percent, and that its EPS would increase at a 20.5 percent CAGR to $4.70. Importantly, we believed that the assumptions behind Lowe’s projections were conservative, and if the U.S. economy happened to return to trend line growth, Lowe’s revenues, margins, and earnings could exceed management’s expectations. I reworked my earnings model and concluded that, should the U.S. economy improve to trend line, Lowe’s could earn as much as $5.50 per share in 2017. Clearly, we had a winner—and the price of the shares reflected the positive earnings and outlook, appreciating to $67.50 at year-end 2014, which was 160 percent above what we had paid for the shares in 2011. We were happy campers—very happy campers.