It is that time of the year when employees, or at least those in the organised sector, would be staring at how their bosses and those in stiff shirt-fronts at the HR department view their output in the year just gone by. They would be hoping that their performance doesn’t, by some quirk in the measurement process, end up being placed at the bottom five per cent among those appraised. Not a very happy prospect. But that is how it is. However, that might just change in the days ahead.
For long, it has been an article of faith among human resource managers that employee performance across the board can be fitted to form a ‘normal distribution,’ with the bulk of employees clustering around some notional mean level of achievement. As a corollary, a small percentage would fall in the category of high achievers, with an equal percentage as under-achievers.
None attest to this belief more vociferously than those in the IT sector that has traditionally been a human resource intensive business. And who can argue against it when the industry has consistently posted high double-digit growth, year after year? Whether it is the pressure from a difficult business environment of the last couple of years or a case of the industry suddenly realising that it has been worshipping a false god, the implicit faith in the validity of the ‘bell-curve’ in employee performance is beginning to fray a bit at the edges.
Distribution not always normal
Some of the biggest names in the IT industry are beginning to concede that they have moved away from a notion that employee performance neatly falls on either side of an average level of performance. The statisticians refer to this as a ‘normal distribution’.
The other day, announcing the results of financial performance of TCS, a company with a workforce in excess of three lakh, a newspaper quoted the company’s CEO as saying that in the year just past, the company did not rank employees on a relative scale or fit them into any bell-shaped curve. There were similar reports of Wipro and Infosys confirming that they, too, have given up rating employees on some relative scale of performance to decide the quantum of performance-related remuneration.
This disenchantment with ‘relative performance’ is not altogether surprising. The conditions that would permit the application of normal probability distribution do not exist when it comes to measuring employee performance. The limitation would nevertheless be manageable if it is used only for identifying poor performers so that they can be targeted for training or redeployment.
Similarly, if it is also used for identifying high-achievers, so they can be entrusted with higher responsibilities when such opportunities arise, the system is reasonably robust. But it becomes counterproductive when used as a tool to arrive at the quantum of bonus or, worse still, routinely cull a certain percentage of the work-force in the facile belief that, through such an approach, the organisation would continuously move up the path of a higher value chain of employee productivity.
Manufacturing metaphor
Let us look at a classical case in operations management, where the normal distribution principle can be expected to hold sway. Imagine the manufacture of a component that goes into a car. A production manager would expect the component to conform to the design specification, if not quite exactly at least within narrow tolerance limits. He would reason that if the quality of the raw material can be controlled, the production processes standardised, operator training and supervision is enforced, there is no reason why the output should vary from the design parameters set for it. And if it does, such variation can only be attributed to random causes and they would be outliers.
It is easy to see that when applied to measurement of human output, the prerequisites so important in the context of producing an automobile component cannot be guaranteed. The raw material — in this case, the employee / recruit — is not uniform. No matter what parameter one looks at — whether it is personality or mental make-up or skills — no two individuals are exactly identical in the way the two batches of raw material used to make the auto component, can be expected to be.
The context in which the job is performed, even within the same department in the organisation, could be different. Their training, intrinsic ability are again parameters that can vastly differ from one individual to the other. To expect, therefore, that the output of a bunch of employees in a functional area of operations in a business will cluster around a mean value, with a relatively small number of individuals representing the high end of achievers or lagging behind at the very bottom, is unrealistic. In other words, the variations in output is not a random occurrence (chance) in a statistical sense but is inherent in the very nature of the job that human beings are called upon to perform in an organisational setting.
What about the outliers?
Even if we assume that, by some miracle, output does indeed get so randomly distributed as to conform to the characteristics of a normal distribution, the data-set of observations must be sufficiently large to make it statistically valid. That is hard to achieve with employee outputs in white-collar jobs each having their distinctive characteristics.
There is another conceptual complication in any assumption of a ‘bell-curve’ pattern to employee output. If the recruitment is fool-proof, there never should be any outliers, in the first place. At least, not of the kind that a small section of the workforce gets branded as under-achievers. Maybe there is bias creeping in, in the minds of the recruiters, who vary from time to time in an unending cycle of recruitment.
The only way to overcome this is to have the same set of recruiters responsible for selection over an extended spell of time, so that even their bias gets randomly distributed so as to leave the output performance of those appraised, unmarred. That too would be an impossible condition to fulfil unless the retirement age of recruiters is set at 80 and those being recruited reach the age of superannuation at year 60! The bosses would like that bell-curve, or not.