Modest risk-sharing significantly reduces health plans’ incentives for service distortion

Some papers use the term service level selection for what we refer to as service distortion. We use the terms “service distortion” and “incentive for service distortion” interchangeably in most of the paper except in places where the term “(incentives for) selection” seems more appropriate.

We use the terms Managed Care Organization and health plan interchangeably. An MCO is an entity that contracts to provide healthcare to its members who are often referred to as “subscribers” or “enrollees”. In the US, there are several types of MCOs: Health Maintenance Organization (HMO), the Preferred Provider Organization (PPO), and the Point‐of‐Service plan. (Accountable Care Organizations (ACOs) can also be viewed as a type of MCO). MCOs are also common in countries such as the Netherlands, Switzerland and Israel. MCOs have increasingly become the norm in the delivery of healthcare services over the past 15 years, partly due to the rising cost of healthcare. The commonly stated goal of MCOs is to reduce the cost of healthcare by negotiating lower fees for healthcare services, monitoring the types of treatment utilized by subscribers, and reviewing the relative cost effectiveness of treatments.

Under the assumption of profit maximization.

We are using the version of the index developed in Layton et al. [1] rather than the final formula appearing in Layton et al. [24]. The former is based on group-level deviations while the latter is based on individual-level deviations. We chose the group-level formula since we believe it to be a more realistic description of how health plan managers implement their desired policy. We are not aware of any previous research that has looked at how health plan managers actually implement service distortion—a question left for future research.

This is the index of heterogeneity (ψ) developed by Layton et al. [1] and mentioned above.

For a detailed description of the Israeli payment mechanism, see Appendix 1.

For the distribution of the 2009 health survey population by age, gender and place of residence, as well as the distribution of respondents with a chronic illness, disability, depression or anxiety or cancer by age and gender, see “Appendix 2”.

In order to simplify the analysis, we do not include residence in the periphery as an adjuster in the models. However, this should not affect the results since the periphery adjuster only adds a fixed (and relatively small) sum to the payment for all enrollees living in the periphery regardless of age or gender.

Not surprisingly, the top percentile of individuals according to cost accounts for 22% of the total budget; therefore, 25% of the actual expenditure on them is equal to about 5% of the total budget.

In order to keep the total budget identical across all the examined payment mechanisms, the average revenue per individual is normalized to 1 for all of them.

A retrospective payment covering 12.5% of hospitalization costs represents 5% of a health plan’s actual expenditure (since the weight of inpatient services is 0.4).

The Israeli Central Bureau of Statistics calculates the Periphery Index of local authorities (on a scale of 1–10). It is a combination of two components with equal weights: the Potential Accessibility Index and proximity to the Tel Aviv District.

For more details, see Brammli-Greenberg et al. [26].