Percentage of contribution to the average

1.4k Views Asked by At

I want to calculate how an individual Business Unit (BU) of a company contributed to the overall Average Salary of the company. Average Salary is defined as the average (arithmetic mean) salary of the employees of the company. Suppose the BU-wise breakup of the salaries and the count of employees of the company is as follows:

$$\begin{array}{|c|c|c|} \hline \textbf{BU} & \textbf{Total Salary (in thousands)} & \textbf{No of Employees}& \textbf{Average Salary (in thousands)} \\ \hline \text{A} & 500 & 8 & 62.5 \\ \hline \text{B} & 700 & 10 & 70\\ \hline \text{C} & 630 & 7 & 90\\ \hline \end{array}$$

I can calculate the average salary of an employee for the company by :

(500 + 700 + 630)/(8 + 10 +7) = 73.2

I want to calculate how much did each BU contribute to this average. Is there a way to calculate the individual contribution percentages of each BU?

1

There are 1 best solutions below

2
On

"How much did each BU contribute to this average?" turns out not to be a simple question to interpret.

Here is one approach: suppose we found the overall average if that business unit kept the same number of employees but paid them nothing - the new overall average would be lower than the original overall average and so indicate a contribution to that original figure. For Business Unit C it would be the $25.2$ or $34.3\%$ that you and Matti P. found.

That approach has the merit of the percentage contributions from each business unit adding up to $100\%$, but what it is measuring might more intuitively be described as "the contributions of each business unit to the total". As you say in a comment, changing the number of individuals changes the averages but not these percentage contributions so these are not intuitively obvious contributions to the averages.

An alternative approach would be to ask what the averages would be if that business unit did not exist at all, and how would that affect the original average. So for your original data, removing C would have led to an overall average of $66.67$ instead of $73.2$ so you might want to say $C$ affected the overall average by about $6.53$. The corresponding figures for A and B would be about $-5.04$ and $-2.13$ respectively. But these are not intuitively clear either: they do not add up to the average or to zero. A better term for this might be the "effect" of each business unit on the overall average rather than its contribution to it. Your change to $700$ for C could make its effect about $-64.12$ on this measure

So my suggestion is to calculate the statistic that you are actually interested in and then find a good way of describing it, possibly not as "how much each BU contributed to the overall average".