Can innovative rating factors delivered by prospective risk equalisation fix private health insurance?

Australia’s Private Health Insurance (PHI) system currently operates under a retrospective risk equalisation (RE) scheme, where insurers paying lower than average claims compensate insurers paying higher than average claims via a zero-sum arrangement. 

In my previous article , I outlined the case for risk equalisation reform by moving towards a prospective RE scheme. In this article, I will explore the concepts behind factor selection.

As a recap, under prospective RE, payments between the RE pool and each insurer are derived based on the expected cost of the insurers’ portfolio, which would in turn be a function of each insured’s risk profile (with relevant adjustments). The generally accepted overall aim of prospective RE is to reduce incentives for risk selection by compensating insurers for their expected claims, with a possible side effect being that insurers are instead incentivised to invest in cost-saving measures which, if successful, results in efficiencies (lower claims and premiums). Although efficiencies are attractive from a political perspective and is typically a goal of prospective RE, this view is not consistent across jurisdictions ^[1] .  

The key considerations in the design of a prospective RE scheme are:

1. Rating Factor selection
2. Rating Factor calibration
3. Scheme administration
4. Interactions with PHI regulations.

This article explores the concepts behind rating factor selection.

Rating factors underpin a prospective RE scheme because they are used to derive the expected cost and therefore payments between insurers. Rating factors should reduce the incentives for risk selection, improve claims efficiency, support policyholders’ wellbeing, and be, where possible, free from gaming and selection bias. They may be divided into two types: statistical and commercial.

Statistical factors

These factors are industry-wide statistically significant drivers of claim costs, requiring an adequate volume of data for analytics and may include the following:

• Age and gender: A significant indicator of health and lifestyle stage, and therefore claims;
• Geographic region: As costs vary by geography, a prospective RE scheme could operate as a national scheme with state-based risk factors, or be segmented by state as per current practice. More granular region-based factors may also be appropriate, for example, in regional areas dominated by low numbers of private hospitals;
• High-cost claimant pool (HCCP): A form of retrospective excess-of-loss equalisation for high-cost claimants is likely to be desirable, to protect small insurers and to reflect the lower scope for claims control. Further segmentation of high cost claims may be between those that are predictable (e.g. through a chronic condition) and those that are more random (e.g. hospital adverse events or accidents);
• Product: Product coverages linked to hospital tiers (Basic/Bronze/Silver/Gold) could factor into the risk profile, as the level of coverage will ultimately limit the cost of claims. This could also be used to demonstrate fairness to policyholders across different life stages, as currently, policyholders receive the same dollar subsidy (calculated deficit) regardless of the level of cover held. It may also be used to balance the profitability by product group where market intervention is desired, for example, if insurers are avoiding writing Gold products due to low profitability;
• Chronic disease: As it is commonly accepted that there are links between chronic disease and general healthcare costs, there may also be similar relationships with PHI claims, noting that PHI claims primarily arise from elective services (whether in hospital or allied health). Additionally, care must be taken that these factors do not deter insurers from improving the health of their policyholders;
• Past claims: Past claims may be a data source for non-reported chronic diseases or health-related behaviours via the patterns of associated ICD or MBS codes. Past claims data may be an additive or subtractive contributor to the risk profile, with examples being the presence of regular General Dental services combined with the absence of other claims (subtractive), or purchases of blood sugar glucose monitors (additive).

Commercial factors

Some factors may not be statistically significant due to data and other constraints but are nonetheless useful in supporting the goals of prospective RE. These include:

• Consumer behaviours: Behaviours may be incorporated to support incentives for insurers to influence them. Potential examples include smoking status and prevention-based activities. Selected behaviours should be clinically justified;
• Provider choice: Products that limit or encourage policyholders' use of the insurers’ network may be given factor discounts to incentivise insurers to more effectively contract with providers, with controls to demonstrate cost advantages. Additionally, through provider-specific or provider behaviour-specific factors, the scheme may influence healthcare providers, for example, applying a risk loading to providers that do not share good quality data, or applying risk loadings to those who charge above benchmark rates;
• General treatment (GT): Insurers may also be incentivised to sell GT in addition to hospital products, should there be evidence that access to Allied Health (e.g. dental) reduces hospital claim costs;
• Non-health factors: Factors not related to the claims cost but nonetheless improve PHI generally may be considered. Examples include discouraging the use of intermediaries (which reduces industry cost) or incentives to discount premiums for low-income earners.

Next steps

As outlined above, the key considerations in the design of a prospective RE scheme are:

1. Rating Factor Selection
2. Rating Factor Calibration
3. Scheme Administration
4. Interactions with PHI Regulations.

Subsequent articles will explore considerations 2 to 4.

^[1] The goal of risk equalization in regulated competitive health insurance markets - PMC