Total rewards optimization: spending where employees see the value

Benefits are about 30 percent of total compensation for the average U.S. employer, and most of that spend was set years ago by people who no longer work there. The plan design, the cost-sharing, the time off policy and the retirement match were each decided on their own terms, at different times, against different budgets. Nobody asked employees how they would trade one against another, because there was no good way to ask.

Total rewards optimization is that way to ask. It borrows a technique from consumer research, conjoint analysis, and applies it to pay and benefits.

What conjoint analysis does

Asking employees what they value produces the obvious answer: everything, especially more pay. Conjoint analysis avoids that by never asking directly. Employees see a series of paired reward packages, each a realistic combination of eight to twelve elements (base pay, bonus opportunity, retirement contribution, medical plan design and premium share, paid time off, flexibility, career development and so on), and choose the one they prefer. Over a dozen or so choices, with the combinations varied systematically, the pattern of choices reveals how much weight each person puts on each element and each level of it.

The statistical output is a set of part-worth utilities for every respondent, estimated with hierarchical Bayesian methods so that individual preferences are recovered rather than only the average. From those utilities two things follow. The relative importance of each reward element, overall and by segment. And a simulator that can estimate how the workforce would respond to any package that was not actually shown.

Putting a price on value

The second input is cost. Every element and level in the survey is priced at the employer's actual cost, using the plan's own data: the retirement match at current participation and pay, the medical plan at current enrollment and claims, time off at the loaded cost of a day. Preference without cost is a wish list. Cost without preference is a budget. The optimization sits where the two meet.

With both in hand, the question becomes concrete: among all the packages the employer could offer at the current spend, which one do employees value most? And among packages that employees value as much as today's, which costs least? The answers are usually not the same package the employer offers now.

What tends to turn up

The findings differ by workforce, which is the point, but some patterns recur. Employees often place low value on elements that are expensive for the employer and invisible to them, such as the employer share of medical premium or a retirement contribution they do not track. They often place high value on elements that are cheap or nearly free, such as flexibility in schedule or an extra few days off. The dispersion across segments is large: an element that a thirty-year-old with children prizes may be nearly worthless to a fifty-five-year-old with grown ones. A single package serves neither well, and a modest choice architecture serves both.

The result for the employer is a design that raises perceived value at the same cost, or holds value at lower cost, with the trade-offs quantified and the segments identified. For a retirement plan sponsor weighing a move from a defined benefit design to a defined contribution one, the same tool shows how employees actually value the two, which is a different question from what they cost.

Running it well

A conjoint study is not a satisfaction survey, and the design decisions matter. The elements and levels must be realistic and must be ones the employer could actually offer; a level that cannot be delivered contaminates every estimate near it. The sample has to be large enough to support the segments that will be reported. Fielding should be short and anonymous, with a response rate the employer is willing to defend. Legal constraints, including nondiscrimination rules and collective bargaining agreements, bound the feasible set before the optimization starts. And the communication afterward matters as much as the study: employees who were asked expect to hear what changed and why.

Our approach

We design the study with the client's HR and finance teams, field it, estimate the utilities, build the cost model from the client's own plan data and present the optimized designs with their cost and their expected effect on each workforce segment. The firm's work in this area draws on more than two decades of total rewards optimization experience with large employers.

This article is general information about the topics discussed and is not actuarial, legal or tax advice for any specific plan or employer.