In the first installment of this series, we wrote about why jobs don’t matter in salary surveys. We hope you’ve already read it, but if not, be sure to do so. This week, we will bust another myth – market percentiles.
Users of salary surveys are accustomed to referring to the market data in terms of a percentile. The most common percentile of reference is the median average or 50th percentile. Simply stated, the 50th percentile of an array of values is the point where there are equal numbers of observations above and below. This is a definition from mathematics – it has nothing to do with salary surveys.
For example, suppose you want to analyze passengers at a bus stop. If you capture the number of passengers boarding each bus during an hour, and sort the data from low to high, the middle value is easy to spot. You can calculate the median number of passengers getting on at that particular stop fairly easily.
Salary survey providers also report data back to users as percentiles. If you are targeting the 50th percentile, you might look at a range including the 25th, 50th and 75th percentiles in order to decide on the value to use in your analysis. The graph below illustrates the “pay line” approach found in most surveys:
You can start your analysis with the 50th percentile value (97,777), and adjust it up or down to suit your needs. As you move horizontally along the x-axis, you pass through multiple percentiles – increasing to the right, decreasing to the left. But there is no information about vertical movement. We refer to this as a horizontal market model.
But wait a minute! Is that right?
Is the 50th percentile of the labor market in a salary survey a single value, like the number of bus passengers?
Nope. For any percentile value, the labor market is more accurately measured using a range of values. You see, organizations define their salary structures with minimum and maximum values. All staff are paid in between these bookends, defined by grade or band. At any given moment, you would ideally be able to see the 50th percentile range of the market, from the salary scale minimum to the salary scale maximum. The incumbent value could be captured too, since it falls somewhere in between the minimum and the maximum.
At Birches Group, we call this the Market Footprint™. In our surveys, we capture three separate pay lines – one for the salary scale minimum, one for the salary maximum and the incumbent average, MRP or market reference point. The graph that follows illustrates the Market Footprint™.
What is interesting about the Market Footprint™ is the rich level of information you can observe by just adding two more pay lines (the minimum and maximum). Now, if you settle on the 50th percentile, you have a range of values to choose from within that percentile, by moving vertically from minimum to maximum. And while it is possible to also move horizontally through different percentiles as before, you can also observe there is some overlap across percentiles. The orange arrows in the graph show two examples of values that occur in two of the market reference percentiles simultaneously. In addition, the red line illustrating the incumbent data clearly has variations depending on the percentile. In the 25th percentile, employers are paying closer to the minimum of the range, while at the 50th and 75th percentiles, the incumbent average is close to the actual midpoint of the range.
We strongly believe that looking at market data in one dimension (horizontally) doesn’t allow employers the flexibility needed to design compensation structures properly.
Using a two-dimensional approach (both horizontal and vertical) gives employers the flexibility to address both individual situations and the overall company requirements.
Suppose you have a salary band and one group of employees, say engineers, needs to be paid higher than other occupations due to market pressures. If you had access to market ranges by percentile, you could easily set the salary of the engineers at competitive levels to the market but still within your stated market reference (say, 50th percentile). This allows you to address a market issue and stay true to your policy at the same time – quite an elegant outcome.
Another useful outcome of taking a Market Footprint™ approach is setting your salary range spans – the “width” of your pay bands from minimum to maximum. In many developing countries, the spans vary widely, and even within the same country, vary substantially by level. If you use survey data that captures the ranges, you can benchmark against the actual spans in the market, and adjust yours to match the common practice, if desirable. This avoids overpaying on the low end and underpaying on the high end of the range.
We hope this has given you some food for thought. Using range data as part of your market analysis will free you from the confines of your current approach. This series of articles is designed to help employers consider how they use survey data, and open the discussion about the pitfalls of the common approaches.
It’s time to rethink how surveys are conducted and used. One-dimensional analysis just doesn’t cut it any more. You need to look for sources that give you a flexible way to analyze your market data. We will share more ideas in the coming weeks.