One of my clients called me yesterday with a question that I thought might be relevant to you. She asked me how wide companies were making salary ranges for their lower-level employees. As a general rule, the smaller the job, the narrower the range. The rationale for this is that there is less variability of performance in lower-level jobs (how many different ways are their to answer the phone or file documents?). And since the purpose of most ranges (with the exception of those used to reward tenure) is to reward performance, you don't need wide ranges for small jobs.
HOW NARROW IS NARROW?
Following this logic, the widest ranges normally go with the largest jobs which are executives. Executive ranges can be as wide as 85% (+/- 30% around the midpoint which is 85% when measured from min to max). I normally consider this a bit wide and prefer 67% wide executive ranges (+/- 25 around the midpoint).
The standard range width for non-executive exempt employees is 50% (+/- 20). This would cover directors, managers supervisors, and professional or technical individual contributors.
Standard ranges for nonexempt employees are usually 35% wide (+/- 15%).
WHAT ABOUT BROAD-BANDS?
Several years ago, a phenomenon known as Broad-Banding caught hold. It held sway for a few years and has now died out. Broad-banding was the practice of creating unusually wide base salary ranges. It was not uncommon for a Broad-Banded range to be 150% to 200% wide.
The rationale for this was that wider ranges could better accomodate all the variables for a particular job or person including, tenure, performance, job content, industry, and geography. In practice, it became a way for managers to pay anything they wanted with impunity since they never had to worry about an employee falling below the minumum or over the maximum. It was tantamount to compensation chaos. In fact, a pay program using Broad-Bands is really no pay program at all.
Conscientious managers, realizing the need to lend some order to this chaos, started dividing the Broad-Bands into zones. For eample, they might use the lower-third of the band for new hires, people with minimal experience, and low performers. In effect, they were creating ranges within ranges which effectively negated the whole idea of a Broad-Band.
SYMMETRICAL OR ASYMMETRICAL?
Perhaps the bigger question is whether or not the ranges should be symmetrical around the midpoint. Symmetrical ranges have the same distance from the midpoint down to the minimum as they do from the midpoint up to the maximum.
More than ten years ago, prior to the rise in popularity and availability of market data, there was considerably more variability in the pay of for a particular job from one company to the next. This was because it was difficult to get good data on how much a job should be paid. This was great for the consulting companies that owned a monopoly on the pay data, but it was bad for companies. It was not uncommon at that time to find two companies in the same industry and in the same city paying the same job 20% or 30% different. This is much less common now.
When placed in a slightly more technical context, the interquartile distances, that is the distances between the P25 and the P50, and between the P50 and the P75, have been consistently narrowing for the several years as companies become more aware of what other companies are doing.
I practice, this means that the actual salary difference between a new and experienced employee and a high-performing, long-tenured employee (in the same position) has diminished. It also has resulted in a situation where hiring managers have realized that the bottom portion of the range is no longer useful. As companies start to pay more and more alike, more an more salaries start to converge around the midpoint (especially if that midpoint is set at the median of the market).
This has led many compensation managers to chop off the bottom 5% or 10% of the ranges as they find they can't hire candidates in those zones. This produces an asymmetrical range with perhaps 5% or 10% below the midpoint and 15% to 20% above the midpoint. There is no need to shorten the top half of the range since you will still want to pay your top performers well.
PERFECT PAY PLAN RECOMMENDS
From a Perfect Pay Plan perspective, I would always steer clear of Broad-Bands. They are a failed experiment I have never seen them work. I would definitely recommend asymmetrical ranges in situations where the lower half of the range is not productive for hiring. However, since I am not a big fan of varying range widths from one job to another, I would pick an approach that works best for the majority of positions and apply it to all jobs.