Good to Great: Why Some Companies Make the Leap... and Others Don't, 1st Edition Hardcover - Jim Collins (2001)
Appendix 1.A. Selection Process for Good-To-Great Companies
Research-team member Peter Van Genderen was instrumental in the creation of the selection criteria and in the “death march of financial analysis” required to use the criteria to find the good-to-great companies.
Criteria for Selection as a Good-to-Great Company
1. The company shows a pattern of “good” performance punctuated by a transition point, after which it shifts to “great” performance. We define “great” performance as a cumulative total stock return of at least 3 times the general market for the period from the point of transition through fifteen years (T + 15). We define “good” performance as a cumulative total stock return no better than 1.25 times the general stock market for the fifteen years prior to the point of transition. Additionally, the ratio of the cumulative stock return for the fifteen years after the point of transition divided by the ratio of the cumulative stock return for the fifteen years prior to the point of transition must exceed 3.
2. The good-to-great performance pattern must be a company shift, not an industry event. In other words, the company must demonstrate the pattern not only relative to the market, but also relative to its industry.
3. At the transition point, the company must have been an established, ongoing company, not a start-up. This was defined as having operations for at least twenty-five years prior to the transition point. Additionally, it had to have been publicly traded with stock return data available at least ten years prior to the transition point.
4. The transition point had to occur before 1985 so that we would have enough data to assess the sustainability of the transition. Good-to-great transitions that occurred after 1985 might have been good-to-great shifts; however, by the time we completed our research, we would be unable to calculate their fifteen-year ratio of cumulative returns to the general market.
5. Whatever the year of transition, the company still had to be a significant, ongoing, stand-alone company at the time of selection into the next stage of the research study. To satisfy this criterion, the company had to appear in the 1995 Fortune 500 rankings, published in 1996.
6. Finally, at the time of selection, the company should still show an upward trend. For any company where T + 15 falls before 1996, the slope of cumulative stock returns relative to the market from the initial point of transition to 1996 should equal or exceed the slope of 3/15 required to satisfy criterion 1 for the T + 15 phase.
Good-to-Great Selection Process
We used a sifting process with increasingly tighter screens to find our companies. The sifting process had four layers of analysis:
Cut 1: From the Universe of Companies to 1,435 Companies
We elected to begin our search with a list of companies that appeared on the Fortune rankings of America’s largest public companies, going as far back as 1965, when the list came into existence. Our initial list consisted of all companies that appeared on the 1965, 1975, 1985, and 1995 listings. There were 1,435 such companies. Most people know these rankings as the “Fortune 500,” although the total number of companies listed may be as many as 1,000 because Fortune occasionally changes the size and format of its lists. As a base set to begin our analysis, the Fortune largest-companies ranking has two key advantages. First, it lists only companies of substantial size (companies earn their way onto the list by annual revenues). Therefore, nearly every company in the Fortune ranking met our criterion of being an established ongoing company at the time of transition. Second, all companies in the Fortune rankings are publicly traded, which allowed us to use financial stock return data as the basis for more rigorous screening and analysis. Privately held companies, which do not have to meet the same accounting and disclosure standards, offer no opportunity for an apples-to-apples, direct comparison analysis of performance. Restricting our set to the Fortune rankings has one obvious disadvantage: It limits our analysis to U.S.-based companies. We concluded, however, that greater rigor in the selection process—made possible by using only publicly traded U.S. firms that hold to a common reporting standard (apples-to-apples stock return data)—outweighed the benefits of an international data set.
Cut 2: From 1,435 Companies to 126 Companies
Our next step was to use data from the University of Chicago Center for Research in Security Prices (CRSP) to make our final selection of good-to-great companies. We needed, however, a method to pare down the number of companies to a manageable size. We used the published Fortune rates-of-return data to reduce the candidate list. Fortune calculates the ten-year return to investors for each company in the rankings back to 1965. Using this data, we reduced the number of companies from 1,435 to 126. We screened for companies that showed substantially above-average returns in the time spans of 1985-1995, 1975-1995, and 1965-1995. We also looked for companies that showed a pattern of above-average returns preceded by average or below-average returns. More specifically, the 126 companies selected passed any one of the following tests:
Test 1: The compound annual total return to investors over the period 1985-1995 exceeded the compound annual average return to investors for the Fortune Industrial and Service listings over the same period by 30 percent (i.e., total returns exceeded average returns by 1.3 times), and the company showed evidence of average or below-average performance in the prior two decades (1965-1985).
Test 2: The compound annual total return to investors over the period 1975-1995 exceeded the compound annual average return to investors for the Fortune Industrial and Service listings over the same period by 30 percent (i.e., total returns exceeded average returns by 1.3 times), and the company showed evidence of average or below-average performance in the prior decade (1965-1975).
Test 3: The compound annual total return to investors over the period 1965-1995 exceeded the compound annual average return to investors for the Fortune Industrial and Service listings over the same period by 30 percent (i.e., total returns exceeded average returns by 1.3 times). The Fortune listings do not contain ten-year returns before 1965, so we decided to include all top performers over the three-decade period in the initial set.
Test 4: Companies founded after 1970 and whose total return to investors over the period 1985-1995 or 1975-1995 exceeded the average return to investors for the Fortune Industrial and Service listings over the same period by 30 percent (i.e., total returns exceeded average returns by 1.3 times) but that did not meet the above criteria due to a lack of data in the Fortune list in prior decades. This allowed us to closely consider any companies that performed well in later decades but did not show up earlier on the Fortune listings. The 1970 cutoff also allowed us to identify and eliminate from consideration any companies with histories too short to be a legitimate transition company.
Cut 3: From 126 Companies to 19 Companies
Drawing upon the research database at the University of Chicago Center for Research in Securities Pricing (CRSP), we analyzed the cumulative stock returns of each candidate company relative to the general market, looking for the good-to-great stock return pattern. Any company that met any one of the Cut 3 elimination criteria was eliminated at this stage.
CUT 3 ELIMINATION CRITERIA
Any company that met any one of the following elimination criteria was eliminated at this stage.
Terminology used in Cut 3 elimination criteria:
T year: Year we identified as the point at which performance began an upward trend—the “transition year,” based on when the actual stock returns showed a visible upward shift.
X period: Era of observable “good” performance relative to the market immediately prior to the T year.
Y period: Era of substantially above market performance immediately following the T year.
Cut 3 Elimination Criterion #1: The company displays a continual upward trend relative to the market over the entire time covered by CRSP data—there is no X period.
Cut 3 Elimination Criterion #2: The company shows a flat to gradual rise relative to the market. There is no obvious shift to breakthrough performance.
Cut 3 Elimination Criterion #3: The company demonstrates a transition, but an X period of less than ten years. In other words, the pretransition average performance data was not long enough to demonstrate a fundamental transition. In some cases, the company likely had more years of X period performance prior to the transition year, but the stock became traded on the NASDAQ, NYSE, or AMEX during the X period; therefore, our data did not go back far enough to verify an X period.
Cut 3 Elimination Criterion #4: The company demonstrates a transition from terrible performance to average performance relative to the market. In other words, we eliminated classic turnaround situations where the company pulled out of a downward trend and into a trajectory parallel with the general market.
Cut 3 Elimination Criterion #5: The company demonstrates a transition, but after 1985. Good-to-great transitions that occurred after 1985 might also have been legitimate good-to-great candidates. By the time we completed our research, however, we would not be able to verify that their fifteen-year ratio of cumulative returns to the general market met the three-to-one criterion.
Cut 3 Elimination Criterion #6: The company shows a transition to increased performance, but the rise in performance is not sustained. After the initial rise, it goes flat or declines relative to the market until the time of consideration for selection into the study.
Cut 3 Elimination Criterion #7: The company demonstrates a volatile pattern of returns—large upward and downward swings—with no clear X period, Y period, or T year.
Cut 3 Elimination Criterion #8: A complete set of CRSP data is not available before 1975, making it impossible to identify a verifiable X period of ten years.
Cut 3 Elimination Criterion #9: There is a transition pattern, but the company demonstrated a period of such spectacular performance prior to the X period that there is substantial evidence that the company is a great company that had fallen temporarily on difficult times, rather than a good or mediocre company that became great. The classic example is Walt Disney.
Cut 3 Elimination Criterion #10: The company is acquired, has merged, or is otherwise eliminated from consideration as a stand-alone company by the time of the final Cut 3 analysis.
Cut 3 Elimination Criterion #11: The company shows a mild transition but falls short of three times the market.
Cut 4: From Nineteen Companies to Eleven Good-to-Great Companies
We wanted to find companies that made a transition, not industries that made a transition; merely being in the right industry at the right time would not qualify a company for the study. To separate industry transitions from company transitions, we decided to repeat the CRSP analysis for the remaining nineteen companies, only this time against a composite industry index rather than the general stock market. Companies that showed a transition relative to their industry would be selected for the final study set.
For each of the remaining nineteen companies, we looked back in time via the S&P industry composites and created an industry set of companies at the time of transition (within five years). We then acquired CRSP stock return data on all of the companies in the industry composite. If the company had multiple industry lines of business, we used two separate industry tests. We then created an industry cumulative returns index against which we plotted the cumulative returns for the transition company. This allowed us to identify and eliminate from the study any companies that did not show the transition pattern relative to their industry.
Through industry analysis, we eliminated eight companies. Sara Lee, Heinz, Hershey, Kellogg, CPC, and General Mills demonstrated a dramatic upward shift relative to the general stock market in about 1980, but none of these companies demonstrated a shift relative to the food industry. Coca-Cola and Pepsico demonstrated a dramatic upward shift relative to the general stock market in about 1960 and again in 1980, but neither demonstrated a shift relative to the beverage industry. We therefore ended up with eleven companies that made it through Cuts 1 through 4 and into the research study. (Note: At the time of initial selection into the study, three of the companies did not yet have a full fifteen years of cumulative stock data—Circuit City, Fannie Mae, and Wells Fargo. We continued to monitor the data until they hit T + 15 years, to ensure that they would meet the “three times the market over fifteen years” standard of performance. All three did, and remained in the study.)