Farrah Oyelaran· HR Business Partner, Nordvik Energy
Asked in Pay Equity & Transparency ·
I am a one-person compensation function at a 1,200-employee engineering consultancy in Ireland. For two years I have run our pay equity regression in a spreadsheet add-in: log of base pay on gender, grade, discipline, office and tenure. It works, but every rerun is a day of manual steps and I cannot easily test alternative specifications.
I have budget for either a specialist pay equity tool, quoted at about 28,000 euros a year, or some training for me in a statistics package and a bit of consultant time to review the model.
For an organisation this size, which would you choose, and what am I underestimating either way?
11 helpful · 4 insightful · 1 agree · 5 replies · 608 views · 5 following
Accepted answer· by Fiona Chen
I have done both, in that order. The scripted route took about six weeks of evenings, plus two days of a statistician's time to review the model. After that a rerun took minutes. Weakness: it lived in my head, and when I was off sick for a month nobody could run it. We bought a tool when we passed 2,000 employees and needed HR business partners to check individual cases themselves. If you are the only user, you are paying 28,000 for an interface for one person. My suggestion: script it, document it properly, have it reviewed externally, and revisit when someone other than you needs to use it.
Read in the conversation ↓
Director, Total Rewards, Copperfield Mining
At your size I would take the training. The model is not complicated and a scripted version reruns in seconds. What you cannot buy in a tool is the judgement about which variables belong, and a consultant reviewing your specification once a year gives you that for a fraction of the licence.
HR Business Partner, Vantage Telecom
I chose the tool at 1,500 people and do not regret it, but not for the regression. What I use daily is the remediation modelling: it tells me the cheapest set of adjustments that brings the gap under a threshold, and lets me test an offer or promotion against the model before it is made. Doing that by hand for 80 offers a year was not realistic.
Director, Total Rewards, Lumen Biosciences
Accepted answer
I have done both, in that order. The scripted route took about six weeks of evenings, plus two days of a statistician's time to review the model. After that a rerun took minutes. Weakness: it lived in my head, and when I was off sick for a month nobody could run it.
We bought a tool when we passed 2,000 employees and needed HR business partners to check individual cases themselves. If you are the only user, you are paying 28,000 for an interface for one person.
My suggestion: script it, document it properly, have it reviewed externally, and revisit when someone other than you needs to use it.
Ben RasmussenGRS, WWP
Job Architecture Lead, Ravenna Chemicals
Whichever you choose, ask where the data sits. Our information security team took three months to approve sending individual pay with gender and ethnicity to an external platform, and wanted it hosted in the EU. That delay was longer than the implementation.
Director, Total Rewards, Copperfield Mining
Decision made: scripted model with an external review, and a written runbook so it is not just me. I had not weighed the single-user point properly. I will come back in a year and say whether I held my nerve.
