Run on East Side Bank, N.Y. 1912 February 16. Bain News Service via Library of Congress.

The Abilene Paradox of Escalation of Commitment: How We Can All Agree To Go Over the Cliff

E. Forrest Christian Project Management 6 Comments

When a project starts heading south, you would think that the rats would start abandoning the ship and the sponsor would quickly pull the plug to minimize losses. But that’s not what happens. Sponsors’ commitment to projects going bad actually seems to grow. The literature is littered with examples of sponsors continuing to spend money on a project that is slipping (Denver International Airport’s baggage-handling) or has shown market failure (the Ford Edsel).

Why in the world would someone throw good money after bad?

At issue is escalation of commitment. Urban Nulden of the Department of Informatics at Goteborg University, Sweden, examined this in some interesting research. Nulden talked with several project managers about escalation of commitment on projects. One manager’s comments in particular caught my attention. Nulden reports the manager as saying, “I had decided to be tough at the meeting and suggest that [the failing project] should be abandoned, but I got cold feet at the meeting.”

What interests me is whether or not everyone knew that the project was failing but no one would raise the question, thinking that everyone else or a major stakeholder wanted it. Jerry Harvey calls this the “Abilene Paradox”: he and his relatives all drive an hour in the hot Texas summer in an non-air conditioned car to have dinner, only to realise later that none of them wanted to go in the first place. He calls it the problem of agreement, as opposed to conflict management.

It could also be what Art Kleiner identified as pleasing the in-crowd of your organization (“Are You In with the In Crowd?“, Harvard Business Review, Jul 2003; or Who Really Matters: The Core Group Theory of Power, Privilege, and Success). The “in-crowd” is the core group of your organization, the people whose wants and desires are taken into account, albeit unconsciously, in every decision within the organization. Comments made by the in-crowd carry great weight, and even small interactions with these people are analyzed and retold. Off-hand comments in meetings or even hallways can cause redirections as non-in crowd persons attempt to please the in-crowd. Projects can escalate because those inside the core group do not withdraw their desire for the project’s completion.

The London Stock Exchange this type of problem with its Taurus project, which has been chronicled in several places, including Mark Keil & Ramiro Montealegre’s article in the Sloan Management Review (“Cutting Your Losses: Extricating Your Organization When a Big Project Goes Awry“, Spring 2000). Taurus was to revolutionize the exchange but ended up spending more than £500B between them and their partners. The executive of the exchange wanted to kill the project but kept being assured that everything was fine. In the end, killing it cost him personally: he resigned the next day.

Perhaps the issue that Harvey addresses in the Abilene Paradox explains much of the escalation of commitment that we see. Everyone knows that the project is failing but no one wants to risk the ostracization of saying this.

Helga Drummond reviewed the infmaous Taurus project and the causes of the escalation of commitment leading to its massive failure (“Is Escalation Always Irrational?“, Organization Studies, Mid-Winter 1998). She believes that the “tragedy of the commons” causes much of the escalation:

The commons were stretches of land offering free grazing rights. Logic dictated that each peasant would graze one more beast. However, the collective impact of each peasant maximizing his/her utility was to ruin the land. In other words, failure was inherent in the policy itself. In the present study, the policy of preserving business interests led to a similarly destructive ‘free for all’.

Taurus had so many stakeholders demanding their own ease of solution that the project’s complexity grew beyond anyone’s ability to comprehend it all. Which seems to also argue for “There Can Be Only One Leader” and the concomitant reduction of main stakeholders.

Is escalation of commitment really the problem? Or is it simply a rejection of accountability for what you think is someone else’s problem? That in turn suggests a tragedy of the commons.

Escalation of commitment can only be prevented with a strong management culture that allows for information to go up the ladder, down the ladder, and sideways, in clear ways.

Image credit: Some escalation going on of some de-commitment. “Run on East Side Bank, N.Y. 1912 February 16”. Bain News Service, via Library of Congress.

About the Author

Forrest Christian

Twitter Google+

E. Forrest Christian is a consultant, coach, author, trainer and speaker at The Manasclerk Company who helps managers and experts find insight and solutions to what seem like insolvable problems. Cited for his "unique ability and insight" by his clients, Forrest has worked with people from almost every background, from artists to programmers to executives to global consultants. Forrest lives and works plain view of North Carolina's Mount Baker.  [contact]

Comments 6

  1. Excellent summary question. It probably all inter-related. If you have poor, non-level V senior management, you foster a situation where people fear to tell you the truth. They instead tell you what you want to hear. Similarly, nobody may want to be the bearer of bad news, because they’ll also be blamed. You could also have a situation where you have a large inefficient organization where you’ve breed hundreds if not thousands of managers who make a decision (incidentally this describes virtually every fortune 500 company I’ve interacted with.) Finally, and it’s related to the above situation, none of the team members driving the project could care. A combination of assigning the wrong employees or fostering a poor leadership environment could create both.

    I recently discussed this with a manager at a similar level as myself at a foreign company that would easily rank in the fortune top 50 were they American owned. Our conclusion was the only way to drive out the inherent beauracracy in these large conglomerates is to turn them into loosely federated smaller organizations. Almost a Japanese model (forget their word for it). In charge of each unit should be an aggressive entrepenurial (sp?) leader who earns his/her pay based on bottom line performance. The “mother ship” organization should provide support necessary for these sub-units to be successful and that’s about it. Very little top down direction, just bottom up requests for assistance.

  2. Oops, will proof read before posting next time.

    s/b hundreds if not thousands of managers who can’t make a decision.

  3. Post
    Author

    I’m again intrigued. Your ideas for that foreign company are similar to what I’ve heard Jaques’s disciples, except that they would argue that there is a place for corporate headquarters. However, most of what corporate does today should be done as advisors rather than as bosses. The balance is tricky and I think that Jaques and co. provide an interesting way to get to it with Time Span of Discretion and Requisite Organisation.

    One of the problems with decentralizing is that most companies find that they really do need to then recentalise later. Perhaps the cycle is natural and necessary.

    On the managers who cannot make a decision: Jaques’s theories only go so far, of course, and other factors often come into play. I will argue that the managers have been promoted into positions where either (1) they do not have sufficient Time Span to meet the Level of Work (LoW) for that position or (2) theirs direct bosses do not have sufficient Time Span to meet those positions’ LoW or (3) the boss is not a true level above the subordinate (too many management layers). (Which is one more than I thought I would have.)

    Jaques’s theory of Requisite Organization says that even GE will only need seven layers. Most companies have too many layers, even now. You need the right number of layers and the right people in the right place.

    I’m wondering if WalMart and Dell have similar problems with managers, or if they have succeeded.

  4. Why managers in very large organizations don’t make decisions: the organization is so large, by not making decisions and keeping a low profile you can hide and stay off the radar screen. This is a proven, virtually fool proof technique for a “successful” career in these huge ungainly organizations. If you make decisions and draw attention, your survival rate goes to 50-50. You get attention and get promoted or you get attention and get fired. Given the advanced pay scale at these companies, people are really incentified to opt for the more likely to survive route.

  5. Incidentally, my gut reaction to 7 layers was that it was too many even in a really large organization. Upon reflection, in someplace like GE, its an unavoidable number. Smaller places need fewer. Where that whole equation gets murky is when you have a huge corporate staff trying to justify their existence and usually without clear lines of authority (another of Jaques points). Staff’s are very inefficient if not strongly managed. The military provides a clear excellent example of how to organize a staff without too many cross directional arrows in the org chart.

  6. Post
    Author

    Jaques’s book Requisite Organization talks about staff roles. He believes that you should only have the number of management levels dictated by the Time Span of the largest role. In the consulting company that I worked for (INFOSEC), the company only looked about 1-2 years out for itself, but worked on some 3-5 year projects for clients. Although smart, most of the consultants couldn’t think past the 2-5 year range. And because the company itself didn’t have a long time horizon, the Managing Director was the same. Which made him a peer, not a boss. We should have been arranged to have two layers of management and that’s it. We didn’t have any clerks.

    The Layers of Management are determined by the layers between the role with the longest Time Span horizon and the one with the shortest. Some comapnies are very small but have long time horizons (I can think of several trading companies that I have worked with where this is true). They actually need several layers of management: they need a CEO who has a 20 yr horizon, but they also need clerks with a 1 wk horizon. You have to have the management layers in-between because your MOR cannot effectively manage you. He’s too far beyond you.

    I’m writing up a piece on this and I’ll post it. It has several good charts along with it.

    I agree that the military has done a pretty good job. From the outside, it even looks like they have done a good job of not promoting the wrong people.

    On the issue of war vs. peacetime promotions: the skills for peace are simply different than war. In business, this would be the difference between the technical and management hierarchies. But the military has a different problem and the analogy doesn’t work. I’m not sure that a standing army can avoid this, which may explain the military’s need for a war every ten years or so. I’m sure that someone has written about this, especially with Rumsfeld’s advocacy of the reduced force size. Suggestions?

Tell Forrest how wrong he is: