Promoting an employee on the basis of exceptional performance seems to be a smart move, even a no-brainer. But, on one analysis of what has come to be known as the “Peter Principle”, it may not be such a smart move. Worse, if that analysis by a former presidential economic adviser is right, there are scenarios in which such an unwise promotion is also the likeliest. What follows explains why he thinks so.
The Curious Peter Principle
You may remember or otherwise be familiar with “The Peter Principle”, the 1969 best-selling book by Dr. Laurence J. Peter and Raymond Hull that proposed, as a hierarchical organizational principle, the statistical “law” that employees tend to be promoted up to their level of incompetence and remain there, instead of being subsequently demoted. Years later, it is still a lively topic of academic and office debate and analysis, despite its uncomplicated formulation.
Basically, the Peter Principle propounds that, repeatedly promoted, or promoted even once, an employee is very likely to (eventually, if not quickly) fail to meet the expectations of the new job, or to even match his or her previous best performance. Counter-intuitive in some ways, the Peter Principle, however, makes perfect sense on some explanations, including statistical, organizational and psychological ones.
As an explanation and delineation of to what the Peter Principle applies—and one readily grasped, there is this simple one: If an otherwise competent employee is promoted to a job that requires a different skill set from that already possessed, the effort required and expended to acquire those new skills may either be insufficient for skill mastery or so great as to otherwise seriously hamstring and reduce productivity. True, effort as a measure of merit will increase commensurately with a pay hike, but with losses, or at best no comparable gains, in productivity. However, companies, unlike parents who bribe their kids to study more, pay for results, not effort.
Everybody can understand that. But other explanations and applications are more subtle.
Promotions for Excellent Performance: A Subtler Coin-Toss (Model)
Among those subtle explanations is one offered by Edward P. Lazear, chairman of the Council of Economic Advisers during the administration of President George W. Bush, now Professor of Human Resources Management and Economics at the Graduate School of Business, Stanford University.
It is his almost whimsical statistical coin-toss explanation—which I will call the “Peter-Peril Probability Argument” (or, for short, the “Peter Peril”)—offered in his 2000 paper titled “The Peter Principle: Promotions and Declining Productivity”, as part of his account of how promotions can create failure. In that analysis he offers the following hypothetical scenario:
“Suppose that a firm promotes all individuals who can obtain three heads on three consecutive coin tosses. Only one in eight will be promoted (since three heads will come up, on average, only once in eight 3-toss attempts). But when the firm asks their promoted individuals to repeat the feat (on their first attempt after promotion), only one in eight will measure up. Seven out of eight will do worse than they did before being promoted. ” (An outcome that suggests that they should be demoted.)
Compare getting three consecutive heads on three tosses with an on-the-job display of rare job talent or exceptional performance, such as closing three consecutive big securities deals or recruiting three Noble Laureates for a research post. What if a coin or mega-deal “trifecta” is rewarded with a promotion merely on the basis of having occurred only once, or only a few times, as standout performances against a backdrop of otherwise normal, average performance? And how likely is that to be the case?
Possibly very likely, according to Dr. Lazear and how decision-makers all too frequently think about the variables and criteria that define excellence and promotion-worthiness. Promotions are often based on extrapolations from a candidate’s maximum performance defined by some variable, compared with others, e.g., extrapolation from an athlete’s Olympic tryout weight-lifting top performance to Olympic games outcomes.
In such situations there may be a temptation to conclude that if the candidate has even once achieved a level that no one else has, it makes sense to promote him or her, rather than anyone else, barring any irrefutable evidence of a truly fluke or fake performance.
If that’s the case, it may be a huge mistake to reward such a rare and outstanding performance with a promotion that is based on the expectation that the employee will consistently perform at that exceptional level in the new job. Even the more modest expectation that “history will repeat itself” based on the idea that “lightning can strike twice, if you’ve got the right rod”, may be way off-base and ultimately disappointed. (This is a familiar Hollywood biz-movie rags-to-riches statistically dubious plot-line, e.g., promotion of the lowly mail-boy who scores big with chance insider information to a job that gets him a desk with a plaque, the boss’s daughter and a Manhattan penthouse—all garnered with no strong evidence for “happily-ever-after” post-promotion repeat performances.)
Why the Rare Performance is Just That—Rare
Although three heads in a row (or closing three mega-deals in a row) is impressive, it is statistically rare (on average, as noted above, only once in eight times when three coins are simultaneously tossed)—and infrequent both within a population of people and in the performance of a single individual.
On average, such an exceptional performance (including closing three consecutive mega-deals or three back-to-back Nobel Laureates, as well as improbable coin tosses) is a blip, not a norm. Of course, outstanding job performance is not always as random as a coin toss; but, for some managers, the more spectacular the blip, the more tempting the illusion and prediction of consistent future performance become.
One key to avoiding this pitfall is to determine whether the variables determining promotion-worthiness are random and therefore in some sense creating misleading fluke results.
(Note: As a math instructor, I occasionally tossed heads 10 to 13 times in a row, in a probability-theory demonstration of my secret method that defies, or at least seriously challenges, the laws of probability—a feat that nonetheless won me no promotion of any kind, apart from any self-promotion.)
Virtually no one will perform at that random coin-toss level in any consistent way. Yet a hiring principle based on that kind of spike in performance implies the mistaken expectation that the rare 1-in-8 performance demonstrates a “talent” for consistently getting three heads (or closing even one mega-deal, let alone three).
Regression to the Mean: Two Cautionary Tales
Although Dr. Lazear suggests that many managers are careful not to overestimate the significance of isolated stellar performances, he adds that what predisposes some managers to commit this fallacy is ignorance of the phenomenon of “regression to the mean”: On average, he says, the average employee’s performance will tend to revert to his or her average (if not also the employee population average), despite exceptional personal spikes.
The concept of regression to the mean is clearest when talking about large populations, large samples or multi-generational genetics: When it comes to natural populations as well as with coin tosses, Mother Nature minimizes extremes and deviations from the presumably “well-adapted” average, e.g., minimizes, at stable small levels, the number of overly large and overly small antlers in a herd of deer (which handicap the deer possessing them, the large ones being too unwieldy, the small ones too unintimidating).
Likewise, spikes in human height within a family or a large population tend to disappear and revert to the multi-generational family average over generations. If coins were creatures, a family with a 3-headed offspring resulting from a random genetic toss would have a 75% chance of having their first grandchild possess only two heads or two tails. Likewise for a large population. However, the 3-headed offspring would have zero chance of reverting to average, much as the high IQ employee, promoted on the basis of an intelligence test, is unlikely in the extreme to see his IQ “regress” or otherwise change after promotion.
So, if promotion is (in)directly based on a clearly random variable, such as access to insider information or chance contacts with billionaires, then, like a coin toss, the outstanding accomplishments based on these chance events will, within both the population of employees and within the history of a single employee, be equally rare after promotion, or possibly not occur at all. Therefore, that promoted employee will not live up to the mistaken expectations and will appear to have risen to his level of incompetence.
As Dr. Lazear puts it, “Regression to the mean implies that future productivity will decline on average. Firms optimally account for the regression bias in making promotion decisions, but the effect is never eliminated… Firms that understand the statistical process take this phenomenon into account, but the result remains: Expected output is lower after promotion than before.”
Unless a promotion is based on a consistently high level of performance, rather than on one or two star-performances perhaps influenced by luck or other forms of chance, there is a virtual mathematical certainty that the post-promotion performance (e.g., in terms of productivity) will not match the pre-promotion high-spike levels in any consistent way.
One cautionary note to be sounded regarding Dr. Lazear’s analysis is that although regression to the mean will clearly cause reversion to the lower performance average in a population of promoted employees whose performance has been substantially determined by random variables akin to coin tosses, fluctuating effort and chance tips, it is not clear that this will always or even in general be true for the individual employee—especially for employees promoted on the basis of a highly consistent performance variable such as tested IQ or other aptitude test, or on an exceptionally successful long-term sales track-record.
The Statistical Logic of Promotions Lost
That caveat kept in mind, it is important to note that Dr. Lazear maintains that the same statistical logic applies to those not promoted. For example, in the coin-toss model, anyone simultaneously getting three tails (which we can call “extreme under-performance”) will, along with everyone else who did not get three heads, be passed over for promotion.
However, had the 3-tail-tossing employee been promoted, e.g., through favoritism, management would have been surprised when (s)he met the criterion for deserving the promotion, namely, when a 3-head toss (or mega-deal triple) appeared as part of his or her post-promotion initial performance.
Because of these statistical principles and reasoning, Dr. Lazear observes, “It is also true that those who are denied promotion do better after they are turned down than they did before the decision was made, for the same reason.”
The takeaway point is that when both under-performance and over-performance, if statistical blips, are confused with reliable and stable talents, the decision to promote or not promote will be flawed, along with the associated high hopes or low expectations.
Peter-Peril Pre-emptive Precautions
One way to avoid the “Peter Peril” is to respect the difference between “promotion for distinctive excellence” and “promotion for consistent excellence”. When an employee is recommended for a promotion because of his or her “excellent performance”, make sure that this means consistent excellence, rather than an excellent (chance-influenced) blip difficult or otherwise unlikely to be repeated.
Dr. Lazear mentions two other safeguards that are in fact widely utilized:
- Lengthen the probationary period (but at the risk of keeping the wrong employee in the post-promotion job too long until a demotion or transfer out).
- Raise the performance bar and require a higher-than-really-necessary pre-promotion performance level to raise the odds of success in the post-promotion post.
A fourth safeguard is to supplement a focus on some outstanding unique performance with a hard look at the employee’s “mode” performance—the average in terms of the most common performance for that employee. That’s a good measure of consistency, e.g., for promotion of the company counterpart of always-on-target Mozart, who never composed anything that was truly awful.
Whatever average utilized (mean, mode or median), it is important to evaluate the employee’s personal average in comparison with other candidates or against some absolute performance standard.
The smart manager aware of regression to the mean will consider promoting an employee who, even if (s)he regresses or reverts to her “average performance, will be “dropping” to or fluctuating around a very high level and standard of performance indeed, like Mozart, even on his worst day.
A manager who takes such precautions when pondering promotions is likely to accomplish two things: To promote the right person…
…and be less likely to regrettably illustrate the Peter Principle himself.
(Note: Alas, in response to my surprise invitation to do so, Dr. Lazear said, in a prompt and courteous email reply, that, being on an extended holiday, he would be unable to offer comments for this article in a timely way.)