Tuesday, April 21, 2015

Good to Great is a flawed book

One of my first posts on this blog was a harsh review of The Outsiders in which I claimed that it had the same methodological problems as Jim Collins' Good to Great. After re-reading Good to Great, I realize that comparison was unfair and want to retract it. The Outsiders has a problem with survivorship bias, but the methodological flaws in Good to Great are both deeper and more numerous.

Good to Great has already received a fair amount of criticism. For instance, Phil Rosenzweig has written that Collins, by basing his research on subjective interviews and newspaper articles, falls prey to the halo effect. Others have noted that Collins doesn't seek out disconfirming evidence. I can't add to that except to say that they're right, so rather than repeating other people's general criticisms, this post will flesh them out by describing some of the book's specific flaws.


A suspect methodology

According to Wikipedia, Good to Great "aims to describe how companies transition from being average companies to great companies and how companies can fail to make the transition."

The book measures greatness by stock-market return: a company that made the transition from good to great is one that underperformed the S&P 500 Index or matched it up to a point in time and then significantly outperformed both the index and its industry rivals from then on.

Collins describes eleven companies that made the transition. He pairs each company with a competitor that had much lower stock-market returns and ostensibly failed to make the transition.The good-to-great companies are as follows (less successful competitors in parentheses):

Abbott Labs (Upjohn), Circuit City (Silo), Fannie Mae (Great Western Bank), Gillette (Warner-Lambert), Kimberly Clark (Scott Paper), Kroger (A&P), Nucor (Bethlehem Steel), Philip Morris (R.J. Reynolds), Pitney Bowes (Addressograph), Walgreens (Eckerd), and Wells Fargo (Bank of America).

To determine why some companies made the transition and others didn't, Collins interviewed corporate executives from the relevant companies and read thousands of newspaper and magazine articles. Many people who criticize Good to Great home in on this research methodology: they argue that articles about a company reflect people's perceptions of the company rather than its true state, and that interviews are susceptible to hindsight bias.

I agree on both counts, but I think Good to Great has several other flaws:


• Some of the pairs aren't really comparable. Fannie Mae is a government-chartered mortgage guarantor, while Great Western was a regional bank. Bethlehem's workforce was unionized while Nucor's wasn't, and that prevented Bethlehem from making some of the strategic decisions that Nucor made.

For the pairs that are comparable in business strategy, Collins ignores differences in size and leverage. Good to Great doesn't mention how much debt the good-to-great companies and their comparisons used. As Valueprax writes, "I did wonder how many of the market-beating performances cataloged were due primarily to financial leverage used by the organization in question, above and beyond the positive effects of their organizational structure."

Some of the good companies were smaller than their comparisons when they began the transition from good to great. Simply being smaller and having more room to grow might have contributed to their outperformance. Collins ignores this, e.g. when he compares the performance of Circuit City-- a tiny near-bankrupt retailer when it began its transition from good to great-- with General Electric, one of America's largest companies:

If you had to choose between $1 invested in Circuit City or $1 invested in General Electric on the day that the legendary Jack Welch took over GE in 1981 and held to January 1,2000, you would have been better off with Circuit City-by six times.


• The winning companies' successful decisions weren't necessarily independent of the losing companies' unsuccessful decisions. In some cases, a winning company's decision turned out well because its losing competitor made a bad decision at the same time.

Philip Morris owed much of its growth to international expansion: Marlboro became the best-selling cigarette in the world before it became the best-selling cigarette in the United States. At the same time, R.J. Reynolds made minimal effort to expand overseas. If Reynolds had devoted more resources to foreign markets, Philip Morris would have faced a formidable competitor and might not have grown so successfully.

More generally, some companies are successful not because they're run by brilliant people but because their competitors are riddled with agent-principal problems or other kinds of dysfunction. When Roy Ash became CEO of Addressograph, he had been fired from his previous company and wanted to reestablish his reputation as a "conglomerateur." Personal rather than financial considerations motivated his actions, and Addressograph underwent a disastrous acquisition binge under his leadership.


• By comparing two companies within an industry and ignoring the rest of the industry, Collins may draw misleading conclusions.

Let's say that company A (a good-to-great company) centralizes, while company Z (a much less successful competitor) decentralizes. Prima facie, it looks like centralization contributed to A's success and decentralization contributed to Z's failure. But if companies B and C in the same industry decentralized and still managed to perform almost as well as company A, then the difference between centralization and decentralization is much less meaningful than the A-Z comparison makes it seem.


• Collins measures a company's success or failure by its stock-market return. This is problematic because many things besides business strategy affect stock performance, such as starting and ending valuations, the amount of debt a company employs, its ability to refinance its debt, and legal issues.

A comparison of Ball Corporation and Crown Holdings, which manufacture aluminum cans, illustrates this. Ball's stock was flat from 1985 to 1997, while Crown's stock rose 1000%, handily outperforming the S&P 500. Since then, Ball's stock has risen 1900% while Crown's stock has fallen slightly.

In the 1990s Ball spun off its original glass jar business, much like Good to Great role models Kimberly Clark and Walgreens exited their original businesses, to focus on higher-margin can manufacturing. Superficially, it looks like Ball made the transition from good to great during the '90s while Crown began to flounder.

The real reason for Ball's outperformance is very different: in the 1960s, Crown briefly owned a company that made asbestos. It owned the company for only 90 days, but that was long enough for trial lawyers to target Crown with billions of dollars in asbestos lawsuits by the late '90s.

When this legal burden hit Crown, it already had large debts as a result of past acquisitions. The combination of debt and asbestos nearly drove the company into bankruptcy, and its stock fell 98% from 1997 to its late-2001 nadir.

Over the past ten years, Ball and Crown have had almost the same stock-market performance-- Crown has actually outperformed slightly-- which suggests that Ball's cumulative outperformance since 1997 is entirely a result of Crown's asbestos crisis rather than actions that Ball's management took.


Vague principles and conclusions

The principles in Good to Great are less a blueprint for success than a description of the ideal business. Companies need to develop "a simple, yet deeply insightful, frame of reference for all decisions," and they need "egoless clarity" to figure out what they're good at. They also "need executives, on the one hand, who argue and debate- sometimes violently- in pursuit of the best answers, yet, on the other hand, who unify fully behind a decision, regardless of parochial interests." They should "shun technology fads AND pioneer the application of technology." Collins tells us that companies need to ride the flywheel, not the doom loop, which is his way of saying that it's preferable for companies to benefit from positive feedback loops rather than suffer from negative feedback loops.

He also writes, "We found [that successful companies] first got the right people on the bus, the wrong people off the bus, and the right people in the right seats-and then they figured out where to drive it."

But exhorting companies to "get the right people on the bus" ignores the practical difficulties of doing so. And Collins doesn't explain what "right people" means-- The smartest? The ones who work best in a team even if they aren't individually brilliant? It probably varies by situation, but Good to Great offers little specific advice to support its generalizations.

Elsewhere, Collins succumbs to hindsight bias. He writes that level-five leadership means being driven yet humble and that having a level-five leader is better than having a boastful, publicity-seeking CEO. Okay, but a self-assured CEO who's successful will be seen, in retrospect, as driven and inspiring. An unsuccessful self-assured CEO will be seen as arrogant and egotistical.

Collins claims that successful companies need "big hairy audacious goals," or BHAGs. He acknowledges that not every company with big goals succeeds but offers a rule for telling the winners and the losers apart: "Bad BHAGs, it turns out, are set with bravado; good BHAGs are set with understanding."

The way he applies these principles to his subject companies is no more enlightening. Explaining the concept that made Walgreen successful, he writes:

What was the concept? Simply this: the best, most convenient drugstores, with high profit per customer visit. That's it.

And Circuit City:

Its distinction lay not in the "4-s" [service, selection, savings, satisfaction] model per se but in the consistent, superior execution of that model.

And Wells Fargo:

Wells saw itself as a business that happened to be in banking.

Trying to explain why Gillette underperformed the S&P 500 in the late 1990s, Collins writes that the company "stumbled in 1999. We believe the principal source of this difficulty lies in Gillette's need for greater discipline in sticking to businesses that fit squarely inside the three circles of its Hedgehog Concept."

The real reason is much simpler: Gillette traded at an extremely high P/E ratio in the late '90s, and eventually even strong business performance couldn't support its valuation. This wasn't unique to Gillette: other richly-valued blue chips like Coca-Cola, Pepsi, and Disney saw their share prices peak a year or two before the general market.


Narrative uber alles

Good to Great is essentially narrational, and Collins omits or downplays things that contradicts his narrative. His descriptions of individual people are idealized, which calls their accuracy into question. Every good-to-great CEO is a mild-mannered company man with unexpected reserves of strength-- it's Clark Kent as CEO. Collins' description of Gillette's leader is typical:

[Gillette CEO Colman Mockler's] placid persona hid an inner intensity, a dedication to making anything he touched the best it could possibly be...

It wouldn't have been an option within Colman Mockler's value system to take the easy path and turn the company over to those who would milk it like a cow, destroying its potential to become great, any more than it would have been an option for Lincoln to sue for peace and lose forever the chance of an enduring great nation.


Ron Perelman attempted a hostile takeover of Gillette in the 1980s. Good to Great describes the company heroically fighting Perelman off ("In the proxy fight, senior Gillette executives reached out to thousands of individual investors- person by person, phone call by phone call-and won the battle") but doesn't mention that Gillette paid him greenmail.


The book contrasts Wells Fargo's success after banking deregulation in the 1980s with Bank of America's poor stock-market performance during the same period. Wells and B of A were both international lenders in the 1970s, but Collins doesn't state how many third-world loans each bank made. An FDIC report suggests that B of A would have become insolvent if it had been forced to mark its '70s-vintage international loans to market. Dealing with these loans must have hurt its stock-market performance and its ability to respond to changes in the competitive environment, but Collins doesn't discuss this.


The book also contrasts Bethlehem Steel's complaints about imported steel with Nucor's blase attitude toward imports. Collins makes it seem that Bethlehem scapegoated foreign competitors for its failings, whereas Nucor acknowledged the realities of the steel market and strove to become the best player in the industry. In reality, Nucor's steel had lower value by weight, so it didn't have to worry about imports like Bethlehem did.

Collins suggests that a dysfunctional, hierarchical management structure was the root cause of Bethlehem Steel's problems, downplaying the role of the company's labor unions. He portrays workers and unions as lacking agency and merely reacting to management's actions:

But the union argument begs a crucial question: Why did Nucor have such a better relationship with its workers in the first place? Because Ken Iverson and his team had a simple, crystalline Hedgehog Concept about aligning worker interests with management interests and-most importantly-because they were willing to go to almost extreme lengths to build the entire enterprise consistent with that concept. Call them a bit fanatical if you want, but to create great results requires a nearly fanatical dedication to the idea of consistency within the Hedgehog Concept.

This is wrong on several counts. First, Bethlehem's factories were primarily in the northeastern states that had entrenched unions, whereas Nucor's factories were primarily in Southern states where union activity was low.

Second, when unions tried to expand in the South, they did a lot things that alienated the workers they were trying to organize, like giving leadership roles to Northern transplants and imposing onerous work rules in places that traditionally had never had them. The actions of the unions themselves probably contributed to Nucor's union-free status.

Finally, once a labor union has established itself at a company, it has an incentive to keep company-employee relations bad to justify its existence. Bethlehem could have made the same effort as Nucor to pursue a "crystalline Hedgehog Concept about aligning workers and management" and it still would have had worse labor relations.

Collins writes that "Nucor came to see that it had tremendous skill in two activities: (1) creating a performance culture and (2) making farsighted bets on new manufacturing technologies." Bethlehem couldn't create a performance culture because union-imposed work rules were designed to increase employment by making its foundries less efficient. It couldn't make farsighted bets on mini-mills because its union-imposed cost structure prevented it from doing so profitably.


After tobacco's health risks were thrust into the public consciousness, Philip Morris and R.J. Reynolds each diversified by acquiring non-tobacco businesses. According to Collins, Philip Morris "stayed close to its brand-building strengths in 'sinful' products (beer, tobacco, chocolate, coffee) and food products," while RJR's executives acted like "country boys with too much cash in their pockets," buying businesses they knew nothing about and losing money. Collins singles out RJR's purchases of Sea-Land, a shipping company, and Aminoil, an oil company, as flawed acquisitions.

The truth is quite different. Like Philip Morris, RJR acquired a number of food brands, including Hawaiian Punch, Vermont Maid, My-T-Fine, Chun King, Del Monte, and Kentucky Fried Chicken. It also bought Heublein Spirits and Wine.

Collins claims that RJR, after years of pouring money into Sea-Land, "acknowledged failure and sold" the company. RJR actually spun Sea-Land off in 1984, and according to Wikipedia, "that year, [Sea-Land] achieved the highest revenues and earnings in its 28-year history."

Nor was 1984 an anomaly: "Sea-Land, which operates a fleet of more than 60 containerships serving 180 ports and cities in 58 countries, posted a record $157 million profit on revenue of nearly $1.6 billion in 1982."

Ultimately, CSX acquired Sea-Land at a premium in 1986. Sea-Land might have been a mediocre acquisition for RJR-- it invested a lot of money in Sea-Land after the acquisition, so its record profits are less impressive than they may seem-- but it wasn't the disaster Collins makes it out to be.

Aminoil wasn't a disaster either. Actually it was a phenomenal success, even after Kuwait nationalized one of the company's oil fields. In Internationa Law and Arbitration, Todd Weiler writes:

The tobacco company, RJ Reynolds, bought Aminoil from its founders in 1970 for US$40million and sold it in 1984 for US$1.7billion, collecting about US$180million in compensation from Kuwait for loss of the concession on the way.

RJR made dozens of times its money on the Aminoil purchase, and even the confiscation of its Kuwait property yielded a 350% return off the purchase price!


The following passage, although tangential to the book's message, offers another illustration of how Collins privileges story over fact:

To use an analogy, the "Leadership is the answer to everything" perspective is the modern equivalent of the "God is the answer to everything" perspective that held back our scientific understanding of the physical world in the Dark Ages. In the 1500s, people ascribed all events they didn't understand to God. Why did the crops fail? God did it. Why did we have an earthquake? God did it. What holds the planets in place? God.

Here Collins accuses medieval people of ignorance while demonstrating his own. The 1500s aren't the Dark Ages, not even by the most liberal definition. The 1500s weren't a period of scientific stagnation, either: medieval Europe had a lot of major inventions that Rome lacked. And religion and science were more compatible in pre-modern times than is generally believed. One historian notes that "Until the French Revolution, the Catholic Church was the leading sponsor of scientific research."

If Collins is so careless with matters of historical record, then we should be skeptical of his own research.


Conclusion

I don't want to criticize Good to Great for not being perfectly scientific. Doing controlled experiments in business and investing is difficult if not impossible, so every finance book is bound to have some subjective or unfalsifiable elements. Yet I think this book has enough identifiable flaws that criticism is justified.

The problem with Good to Great isn't that Collins is a storyteller, it's that the stories are myths.

2 comments:

  1. James - it seems like your problem with the book roots from the distaste that it being revered as a management "study" and that it does not deserve the success? I think all serious readers will agree to that it is neither scientific nor convincing, but would rather treat it as a fun story book full of corporate anecdotes (or that's certainly how I viewed it). In that regards it was a really fun read.

    Which leads me to the question - you have criticized 2 wildly popular business books both for the right reason. Are there any out there that truly passed your test of rigor?

    ReplyDelete
    Replies
    1. Hi Chalk,

      I agree that anecdotes can be interesting-- I thought Good to Great's description of how Abbott Labs managed earnings expectations was interesting. Unfortunately that was a rare bright spot in the book.

      IMO the problem isn't that Good to Great is anecdotal, it's that the way Collins sources his anecdotes is suspect. In particular, by basing his anecdotes on popular books and news articles, Collins is sometimes getting a misleading or sensationalized description of events.

      For example, the footnotes to Good to Great indicate that when Collins describes the strategic mistakes RJ Reynolds made, he's basing that largely on the description of RJR in "Barbarians in the Gate." But "Barbarians" is mostly about RJR's leveraged buyout: its history of the company is brief and incidental to the book's main topic. And IMO "Barbarians" is also misleading about a couple of the historical details, notably RJR's ownership of Sea-Land.

      Basically, I think the history of RJR in "Barbarians" is too superficial to be a good source, but Collins seems to take it as an accurate historical record-- probably because it fits the story he wants to tell.

      As for other business books, I like "The Innovator's Dilemma" a lot. The Jill Lepore article in The New Yorker made some valid criticisms of the book (and some invalid ones), but I think Christensen's theories are still broadly right and have predictive power.

      Delete