A FRAMEWORK FOR EXPECTED ATTRITION

Since the early 2000s a string of new sectors in India started growing by leaps and bounds- in the early 2000s it was the IT and ITES sector, followed by Insurance, Telecom, Pharmaceuticals, Retail and most recently Infrastructure sectors. A new sector usually ‘buys’ manpower from related sectors, and so was the case with these sunrise industries.

Soon the Indian manufacturing sector started experiencing high employee turnover. This was especially true for supervisory workforce at factories, frontline sales force, and managerial workforce.

My stints during these high attrition periods were in a Sales HR and a Factory HR role. High attrition obviously caused a lot of management concern at Sales and Factory locations. The usual refrain was ‘Attrition levels must go down!’. I agreed with the usual refrains –attrition levels of 30-40% are unsustainable! But I was intrigued by the usual and only solution proposed ‘Let’s pay higher salaries!’. I thought ‘Yes, salaries must be competitive but if employees are poached at 2x or 3x of their present salaries there is precious little you can do about it if the only thing you do is increase salaries. Sooner or later someone would still pay high enough monetary incentive to lure away an employee.’ So finally we did get an appreciation of the non-monetary areas that we needed to focus on too to reduce attrition.

I remember a subsequent month when I was working on creating retention strategies. I was grappling with trying to figure out what should be the goal for our retention efforts. Should we ensure that no one leaves? Is 0 % attrition the ideal retention target? My first reaction was ‘Obviously 0% is not ideal. Think of a living system with absolute retention. The system would sooner or later die.’ What should be a good attrition figure then? Should we fix it at let’s say 8%- this would mean that 7.9% attrition is good but 8.1% attrition is bad? How do we arrive at the figure of 8%?

I had subsequently written an article back in 2006, about how to figure out an ideal attrition figure. I had contested, and still do believe, that 0% attrition is not ideal.

INTRODUCTION

Predicting attrition! – is it possible? To answer this we first need to ask a few basic questions:

• Are we looking at predicting overall attrition or are we looking at a few components of attrition?
• With what degree of accuracy do we want to predict attrition?
• Over what time span do we intend to predict attrition?
• At what level of aggregation – individual/group/organizational level – do we want to predict attrition?

In this note we have made an effort to answer a few of these and more. The emphasis is on attrition which should be reasonably expected. We would try to derive a simple mathematical model for this expected attrition rate and illustrate the same through the means of an example. This model is by no means exact or precise. It is just a useful representation of the reality.

FRAMEWORK

The expected attrition rate can be said to arise thanks to the hierarchal structure in the organization.

Let us take a simple case. Assume that there is a sub system, of an organization, with S+1 number of employees. There are 2 hierarchal levels (Level A & Level B). There are S numbers of employees who are reporting in to a single person D.

Let us assume that on an average an employee takes time T to get promoted from one position to another (for both levels A & B). Now an employee in level B can move up only when D (level A) moves out of that position. D can move out of the position in one of the 2 ways:

• He gets promoted to a higher position. Assuming the same kind of sub-system to be replicated across the organization, the probability of D getting promoted is 1/S
• He moves out of the role.

When one of the employees in level B moves up it is reasonable to assume that the remaining people in that level would move out.

Hence total number of employees in the system = S + 1

Remaining employees = 1 + 1/S

Attrition = S – 1/S

Therefore %age attrition p.a. = (S – 1/S)*100/(S+1).T = 100 * {1-1/S}/T

Example: If the span of control is 4 and the time period is 3 years.

%age expected attrition = 100 * {1-1/4}/3 = 25 %

Now in a real organizational context there would be variables which would come in (e.g.: T may vary from one organizational level to another). But these are variables which can be accounted for and hence we can arrive at an expected attrition level

CONCLUSION

The note tries to come up with a mathematically derived internal attrition benchmark. Therefore there would be an optimum attrition rate, which would be a function of the span of control and the time an employee would be expected to stay in a particular position. Attrition which is either lower or higher than this range may not be desirable from the organization’s point of view.

As these two variables change at different levels in the hierarchy the optimum attrition rate at a particular hierarchical level may be different.

-

Sourav