The Upside-Down Insurance Incentive

The insurance industry does well when their costs go up?

Special thanks to a kind reader for starting the conversation that led to this post. Find his insightful writing here and here.

If you enjoy this post, please consider subscribing.

I recently finished reading Catastrophic Care by David Goldhill. Although published in 2013, the book puts forth a number of arguments that were, and remain, outside the mainstream discourse on healthcare. Some of those arguments changed my thinking about the nature of the industry.

In the context of this blog, one of the most interesting arguments concerns health insurers and their incentives, namely that the incentive for insurers to drive low costs is extremely weak. That’s an important assertion because insurers sit at the center of most healthcare value chains, and their behavior therefore has an outsized impact on the outcomes of the rest of the system.

In this post, want to articulate Goldhill’s argument as best I can, and consider its ramifications for the rest of the industry.

High costs can help insurers grow profits

In this post, I wrote about some of the trade-offs that United Healthcare needed to make in order to reduce costs, and I took it for granted that a large insurer should want to limit their medical spend as much as possible. After all, wouldn’t high payouts reduce profits?

Goldhill points out that this isn’t necessarily the case:

…an insurer’s profitability in any given year depends on its ability to keep reimbursements for care below the amount charged in premiums.  But over time, its incentive is the exact opposite: to see spending on health care increase as much as possible.  Why?  Like with any private business, insurers require consistent and profitable growth to satisfy their shareholders.  And you can’t really achieve profit growth by reducing payouts for care—any cost improvements will almost certainly translate to lower premiums.  So the only way for the health insurance industry to increase profits is to increase premiums.  And there’s only three ways to increase this rate base—customers must get sicker, policies must expand to cover new types of care, or prices for care must rise (i.e. the cost of claims must rise). (pg 72)

That is, the insurance industry grows if prices for care rise, since there’s more cost to insure. It follows—contrary to my own assumptions—that insurers would prefer high costs over low costs! 

Counter-intuitive, but not complicated. If the price of, say, car repairs were to increase, your car insurance premiums would increase. Assuming the insurance company is able to keep their margins the same in percentage terms, and assuming they don’t lose customers when raising premiums, they’ll make more in dollar terms.

Here’s some simple math to illustrate—I want to be as clear as possible about what’s happening.

  • Let’s say you pay $100/year for health insurance, and consume $80 in care. The insurance company has $20 left over for administrative costs and profits.

  • If providers raise prices by 25%, you’ll consume $100 in care. Keeping the same margins, the insurance company will raise its premium to $125/year. Now it has $25 left over for administrative costs and profits. Earnings just went up because costs went up.

Of course, with premiums 25% higher, it seems unrealistic to assume that no volume is lost. But in healthcare, there are good reasons to model consumers as fairly inelastic, and Goldhill makes this argument in a separate chapter. They pay premiums for insurance, but usually only indirectly via employers or government subsidies. If you are only responsible for, say, 10% of the premium (as I am on my corporate insurance plan), a 25% increase affects you much less than if you are responsible for the full amount (even though the balance may come out of your next raise, in a way that’s not transparent to you).

If we accept that higher costs are good for insurers, there’s an apparent contradiction. On one hand, we know that insurers do make efforts to reduce costs (as in the United example).  On the other hand, Goldhill shows us that insurers need prices to rise in order to grow profits.  In my reading, Goldhill doesn’t quite resolve this contradiction in his book, so that leaves us to think through it.

What’s the true incentive?

The answer matters. Lots of market-based health policy recommendations assume that insurance companies, who pay for care, can be an effective watchdog on the cost of that care. Businesses and investors need a clear understanding of their constraints, and that’s especially true for new entrants whose business models haven’t had the time to organically adapt to prevailing incentives.

Note that these opposing forces play out at different scales, and on different time frames.  In a single year, an individual insurance firm wants its medical payouts to be low, or at least low relative to competing firms.  On the other hand, if costs rise for the industry as a whole, over the course of multiple years, the insurer can grow revenues (and by extension, profits) more easily.

Here’s what the various outcomes might look like, using Goldhill’s arguments as a basis:

Note insurers are always better off with high industry cost growth (while patients, who ultimately bear these costs in the form of premiums or lower wages, are worse off).  And insurers ultimately should try to move their own costs to the left side of the matrix.

But not too far. There’s an important regulation that limits how much they can reduce their medical costs.  As explained by CMS, “The Affordable Care Act requires insurance companies to spend at least 80% or 85% of premium dollars on medical care.” That 80% (it’s 85% in the large group insurance market) is called the minimum Medical Loss Ratio (MLR), and it places a cap on how much insurance companies can profit by reducing medical costs.

The minimum MLR rule means that if insurers reduce medical expenses below 80% of their premiums, they don’t benefit.  The ACA rule was intended to keep insurers in check, making sure that, as middlemen, they didn’t siphon too much money away from medical care.  But it creates a distortion in the sense that it prevents insurers from growing profits by focusing on reducing costs.

One might argue that if an insurer were able to control costs, it could at least grow revenue by reducing premiums, thereby gaining market share. A quick calculation (elaborating on the ‘simple math’ example above) suggests this wouldn’t be profitable. In the image below, note how profit decreases compared to the base case when costs are reduced, unless demand is highly elastic. And we argued above that demand is most likely inelastic, which makes things even worse.

How to read: in the base case, medical expenses are $85 per member ($85 is the minimum MLR for large group plans). If insurers reduce those costs to $80, they have to reduce premiums to maintain the MLR. This helps to attract a larger membership, but not enough to offset the losses in price. Takeaway: Once insurers hit the minimum MLR, cutting costs loses them money.

None of this is to say that insurance companies are actively trying to conspire in order to raise medical costs.  But it does show that they don’t have much incentive to reduce them. And if none of the market players want to compete by being the low cost option (which is a dead end), prices are likely to rise.

How might an insurance company behave in response to these contradictory incentives?

Allow me to speculate:

  • Focus on raising premiums, especially if there’s a corresponding increase in medical expense that keeps it over the minimum MLR. E.g. Covering more procedures and offering broader provider networks.

  • Create lots of ‘Quality Improvement Expenses,’ which are basically non-medical expenses that nevertheless count toward the MLR (full explanation here), and raise premiums to cover them. E.g., give away an Apple Watch in the name of wellness, and then charge your members for it.

  • Resist price transparency. If the insurer is able to reduce prices with some providers, making these numbers public might allow competitors to negotiate similar reductions. It’s better if prices stay high across the industry.

  • Acquire sicker members instead of healthier ones. Even though sicker members cost more to cover, they also pay higher premiums (this is easy to see in e.g. Medicare Advantage arrangements, where the government explicitly pays higher premiums for sicker customers). A sicker member can therefore be worth more than a healthy one. And note that ‘sicker’ doesn’t mean ‘riskier’ for the insurer; even though they may face more health risks, those are generally easy to quantify across a large pool of members.

What each of these behaviors has in common is that they’re ways to compete in the market with higher premiums (and by extension, higher costs, because of the MLR minimum), rather than with low premiums/low costs.

But we also know that insurers often do take action to to reduce costs—requiring prior authorizations, negotiating smaller networks, vertically integrating into the primary care business. Why?

I’m not certain. It seems clear that the best outcome for an insurer is to managing costs down to the minimum level, and no further. But this can be hard to do in practice, especially given uncertainty about policies and markets.

To illustrate, this report from KFF suggests that insurers in the individual market generally raised premiums far too much in 2018, resulting in very low MLRs, and are therefore issuing mandatory rebate checks to their members. But there were other years in the individual market where claims actually exceeded premiums, so MLR exceeded 100%. Compared to that outcome, expenses that are too low are much better.

In a sense, then, what matters to insurers isn’t minimizing claims, it’s controlling them. Minimizing means pushing expenses down to the lowest possible dollar amount; controlling means pushing claims to exactly 80%, while ceding the fewest painful trade-offs (like cost control measures that cause churn). The more control that insurers have over their expenses, the more predictably they can hit the target 80% number, and the more flexibility they have in managing to it.

For consumers, it’s frustrating that an insurer would not try to minimize cost. Indeed, one of the main benefits that an insurance plan can offer to a member is it’s ability to pool negotiating power and keep provider prices down.

Removing the minimum MLR could help enable this, but that also removing a cap on insurers’ administrative expenses. Perhaps if more administrative costs result in lower medical costs, the result will have been worth it.


  • Common discourse accepts that insurers will act to minimize medical costs to the greatest extent possible

  • Insurers can grow profits more easily when medical claims grow. This reduces the incentive for insurers to control costs

  • The ACA requires insurers to meet a minimum Medical Loss Ratio (MLR), usually 80%, which is the percentage of premiums paid out in claims. The rule exists to keep insurers from charging very high administrative costs, but it also limits their incentive to manage their medical costs, since they don’t realize the benefit of pushing costs down below the 80% threshold.

  • What ultimately matters to an insurance company in this situation is having control over their expenses. Since medical expense reduction takes effort and has second-order effects (like worsening customer experience), it’s important to be able to manage to the 80% level while making as few of these negative trade-offs as possible.

Thanks for reading! For more, visit our About Page, or consider subscribing.

You may also like: EQRx, DTC InsuranceMaking Drugs Good Enough