A market pay line turns a job's evaluated worth into a pay number, using one regression.
The regression
SHRM's method builds the market pay line by simple regression of survey pay on job evaluation points.
- X variable: job evaluation points assigned to each benchmark job.
- Y variable: survey pay for those benchmark jobs.
- The line predicts a pay rate for any point total.
X is not headcount, not incumbent tenure, and not the prior-year merit budget. Points are attached to jobs, not to people.
Where the line sits in the structure
Run the regression, get a predicted rate, apply the lead or lag policy, and the result lands in one place:
Adjusted predicted rate = predicted rate x (1 + policy percentage), and the adjusted predicted rates become the grade midpoints.
A pay line always runs through the midpoint of a pay grade. Not the minimum, not the maximum, and not through the highest-paid incumbent in each grade. That single sentence answers most of this unit.
Worked through end to end: a job evaluated at some point total feeds into the regression, which predicts, say, $84,000. With a 5 percent lead, $84,000 x 1.05 = $88,200, and $88,200 is the midpoint of that grade. The minimum and maximum are then built around the midpoint by the range design step; the regression itself does not set them.
The traps
- Placing the line along the bottom of each grade so predicted rates become minimums, or along the top so they become maximums. Both are wrong; predicted rates are midpoints.
- Anchoring the line to an individual incumbent's pay rather than to the structure's midpoints.
- Treating points as an input to an individual's pay rate. Job evaluation is job-based; the point total shows up in the grade midpoint, not in a person's salary.
- Extending points into benefits eligibility. The method does not do that.
Carry into the check: points in, predicted pay out, policy applied, midpoint set — and the pay line runs through the midpoint.