Business Today

The Emergence Of The Power Law

Performance distribution does not follow the normal curve - fewer people contribute far more to overall performance
Anandorup Ghose   New Delhi     Print Edition: July 15, 2018
The Emergencce Of The Power Law

Vilfredo Pareto, after whom the Pareto principle was named, ran hundreds of social experiments; and one of his findings eventually gave us the eponymous principle that 80 per cent of all land in Italy was held by 20 per cent of people. Over the last 120-odd years since Pareto published his findings, this logic has been visible in almost every field.

Another century before Pareto's findings, French scholar Marquis de Laplace defined the normal distribution curve. The world of people management and HR chose to go with statistics over sociology.

Gradually, and perhaps a little unknowingly, direction has been changing over the last few years and there is increased appreciation for the fact that employee performance does not necessarily follow a normal distribution. In 2012, two professors, Ernest O'Boyle and Herman Aguinis, suggested that human performance is not normally distributed and there is greater prevalence of "star performers" who are able to deliver far more than would be normally anticipated. They postulated that performance distribution does not follow the normal curve but a Power Law curve - where their data suggested that fewer people contributed far more to overall performance. On similar lines, Google in 2006 famously launched its "pay unfairly" HR strategy where they essentially said, "At Google, we ... have situations where two people doing the same work can have a hundred times difference in their impact, and in their rewards."

This idea seems to have hit India now, and there seems to be a distinct shift companies' approach to how they recognise and reward high performers.

Data from 2001 shows that Indian companies placed about 55 per cent of total employee population in the top two performance ratings. While performance distribution was extremely liberal, pay differentiation was not stark - the best got about 1.5 to 1.6 times the average pay increase (i.e. if the average performer got 10 per cent increase the top performer got 15-16 per cent increase).

This liberal performance differentiation and moderate pay differentiation continued until the global financial crisis of 2008/09. In the post crisis scenario, top management pay continued increasing aggressively but the structure shifted from fixed pay increases to incentive-driven compensation.

With companies becoming aware |of performance-based differentiation, the performance curve became sharper with far fewer people in the top rating boxes (the 55 per cent population in those boxes dropped to about 30 per cent). Gradually, top performers started taking away increasingly disproportionate part of the pay budgets (1.9 to 2.2 times the average).

This year's data suggests that the top 7 per cent of employees took close to 16 per cent of the overall compensation increase budget for the year while the medium performing employees (about 55 per cent) had to manage with less than half the total increase.

We have to accept two fundamental shifts - firstly the world of mass production and manufacturing driven organisations where the process was more important than the person has given way to a far more services-driven world where individuals are more critical to the system and therefore the expression of Pareto principle will become gradually more important.

Parallelly the other shift has been in the fact that the age of large pay increase budgets is gone and therefore as budgets become smaller (and inflation remains within control) companies will keep focusing on allocating larger and larger pools to more critical talent that actually drives the business.

As the normal distribution curve dies its natural death, it is the age of the power law.

Written by Anandorup Ghose

The writer is Partner - Talent & Rewards at Aon Consulting India

Youtube
  • Print

  • COMMENT
BT-Story-Page-B.gif
A    A   A
close