The normal distribution or the ‘bell curve’, most memorably introduced into performance management by GE, has now become an article of faith and an essential and inalienable part of performance management systems. Yet, its efficacy has been less than satisfactory. Most managers and employees have very few good things to say about it. Left to them, they would do away with it.

Nassim Nicholas Taleb, author of The Black Swan calls the Gaussian Bell Curve “that great intellectual fraud” because it does not deal with randomness. “Measures of uncertainty that are based on the bell curve disregard the possibility, and the impact, of sharp jumps or discontinuities...using them is like focusing on the grass and missing out on the trees,” he writes. Yet he concedes, “We can make good use of the Gaussian approach in variables for which there is a rational reason for the largest not to be too far away from the average...If there are strong forces of equilibrium bringing things back rather rapidly after conditions diverge from equilibrium, then again you can use the Gaussian approach.”

The application of the bell curve in performance management assumes that in any work force population, performance will follow the normal distribution, with the majority of the employees tending towards the average, a few above it and a few below. This implies that performance is relative and not absolute. Thus, even if an employee has exceeded his goals, he may still not be rated better than ‘meets expectations’ if his peers have done better still. Likewise, it forces managements (again on a relative scale) to identify those who are less than good. Thus, an employee who has met all his goals may still end up in the ‘partially met or did not meet expectations’ list.

If employees who have met their goals get a ‘met expectations’ rating, if those who have exceeded them are rated as having ‘exceeded expectations’, if those who have not met expectations get a ‘partially met expectations’ rating and if all these categories conform to the desired distribution, only then would one have no problem with the bell curve.

Problems arise because a lot many more employees than are permitted to be ranked as having ‘exceeded expectations’ are given that rating. And on the other hand, fewer employees than required by the bell curve are given the ‘partially met expectations’ rating.

The problem is acute when the company or business unit has over-achieved on its business targets. In such a case, there is a clamour for the curve to be shifted to allow for a higher proportion of the work force to be rated outstanding, and fewer (or none at all) to be rated as partially meeting or not meeting expectations.

Yet when business results are poorer than budgeted, no one suggests that a higher proportion of the work force must, in such a case, be rated as not having met expectations! Managers find it hard to explain that regardless of business unit performance, there can only be a select few (as per the normal distribution) who could have been truly outstanding, and that likewise, the proportion of relatively weak performers will remain the same regardless of business results. If it is desired to reward employees in an exceptionally good year, the way to do it would be to increase the incentive pay applicable to each level of performance, rather than to shift the bell curve. Contrarily, in a poor year, there would still be a select few outstanding performers but incentive payouts would be sharply reduced.

In India, the bell curve has proven even more problematic as managers fear that employees rated good or less than good will leave them, thereby jeopardising their own performance in the year to come.

The bell curve will work if (in addition to the pre-conditions laid down by Taleb!)

The business is not new or embryonic.
The business environment is not so uncertain and dynamic as to render goal-setting and planning difficult, or in extreme cases, nugatory.
The performance bar rises every year so that only a few can perform outstandingly (at the same time, if goals are seen as unachievable, even the best performers might leave).
Goal-setting is rigorous and goals are not sandbagged, so that the “outstanding” category does not become a cosy club whose entry is forbidden to others.
The organisation is able to understand performance, not merely measure it; this means moderating ratings where performance was influenced by external factors (positively or adversely) at both the organisational and individual level.
The process is seen as fair, and the majority of the work force accepts and acknowledges those amongst their peers who have achieved an ‘exceeded expectations’ rating, and those who have been rated as ‘partially meets expectations’.
The sample size is reasonably large.
Jobs and performance can indeed be compared across functions.
The process is communicated effectively and clearly to all employees.
Opprobrium is not associated with a ‘meets expectations’ rating, and it is accepted that an employee who consistently meets difficult targets is a high performer.

This last point is especially important because in many organisations in India, rewards and career progression are disproportionately stacked in favour of the top rated performers. So the vast majority in the ‘meets expectation’ category feels hard done by, and fights to be rated higher. The answer to this is to formulate and communicate a rational compensation policy and career framework. The firm’s desired positioning relative to market, its policy regarding functional and performance premiums, and career growth for different levels of performance should form part of such a policy framework.
If these conditions are met, there is no reason why, over time, the bell curve should not come to be accepted as reflecting workforce performance reality.

Sankar Ramamurthy is an Executive Director at PwC & leader of the firm’s People & Change consulting. He can be contacted at sankar.ramamurthy@in.pwc.com