Employee Turnover Rate

Date: 09/19/2002 at 10:27:03
From: Deborah Gottardi
Subject: Statistics

What is the formula to calculate employee turnover rate?

Date: 09/20/2002 at 05:51:41
From: Doctor Mitteldorf
Subject: Re: Statistics

People think that math is all worked out and there are definite rules 
for everything. If that's so, then this just isn't math. 

Perhaps you could say that "science" is about finding appropriate ways
to represent a situation with numbers and operations on those numbers;
once you've done the "science" then the "math" consists in combining
those operations to get an answer.

So you're asking for the "science" that comes before the math. There 
will be clearly good and not-so-good answers, and some room for
different points of view.

Let's say there were 100 employees at the beginning of the year, and 
100 employees at the end of the year, and at the end of the year,
84 of those employees were the same ones as were there the previous
year. You might say that the turnover rate was 16%.

But suppose one of those 16 who left was actually replaced three 
times. The employee quit in January, the replacement quit in April,
and another person was hired who lasted only until November. Then you
might want to count every time an employee left the company and
another one was hired - in this case you'd get 18%.

Another complication: suppose the work force is 100 at the beginning 
and 90 at the end of the year. Perhaps 16 people have left, but only 6 
have been hired during the year, while 2 more were hired and retired 
within the same year. You might define turnover as 18/100 or as 18/90, 
or as 18/95, since 95 is the average of 90 and 100. Instead of 95, you 
might want to do a fancier average, where you actually add up the 
number of employees on each day of the year, and divide the total by 
365.

One more complication: who decided it was a calendar year that we
should use for sampling the turnover rate? Perhaps there was no
turnover at all for 3 years prior, and then a shift in management
caused a lot of people to leave this year. Then a more representative
measure would average over 2 or 3 or 4 years. Maybe you'd want to
average the turnover in each month of the last 48, but weight recent
months more heavily than earlier months.

I'm sure you can think of more wrinkles as well. Even for something 
that seems as simple and clear as "employee turnover rate," there is 
room for discussion about what constitutes the most representative 
method for calculating an answer.

Perhaps I should stop here; if I were to go on, I'd probably start
talking about how the people who are hired to calculate employee
turnover rate are paid by someone, and that someone probably has a
stake in the outcome. And there will be pressure on the statistician
whose job it is to report this number to use the method that is most
advantageous to his employer.

And if you let me go on long enough, eventually I'd get to Enron...

- Doctor Mitteldorf, The Math Forum
  http://mathforum.org/dr.math/