Application of Economies of Scale to Your Health Plan’s Administrative Costs
To use the scale application, first populate the rate of change in your membership. Place a value for which ever cluster of functions you wish to estimate the effect of a change in scale. The cost value after the effect of scale will be shown.
Economies of scale occur when unit costs decline as volume increases. Similarly, they increase as volumes shrink. Economies of scale occur if there is a mix of fixed and variable costs are required to produce things. For health plans, Member Services has costs that are largely variable over the intermediate term while they are more fixed in Actuarial. As membership declines, the fewer your members the greater the share fixed costs will be borne by each of them. Sherlock Company scale studies measure this effect, and this analysis will help you apply them.
This analysis differs from the Calculator analysis in an important respect; the basis for it is inferred
from the actual experience of health plans but is not their actual experience. The Calculator analysis applies actual health plan values, such as median costs PMPMs for your universe. But since no health plan can be at multiple sizes at the same time, the scale studies are inferences between many similar plans at a point in time. The resulting cost models are thus abstract from any plan’s results or all plans’ results.
There is an important implication in the use of these scale models. We have long held that there are few examples of economies of scale in health plans and that generally speaking the slopes of the scale slopes are gradual. Based on this conclusion, the scale models included in this should be viewed with caution, especially if the results are not highlighted as statistically significant. For instance, membership declines could translate into higher PMPM costs in accordance with the scale slopes. This can inform the effect of changes in membership, otherwise explained by the mix adjustment. But since factors other than scale usually predominate, the results should be considered an input rather than conclusion. Updated on November 12, 2020.