Health Insurance Markets

Ezra Klein and Matthew Yglesias respond to conservative claims by Ramesh Ponnuru:

A few responses. 1) It is hard to believe that the market would evolve in a way that consumers were asked procedure-by-procedure, condition-by-condition what they wanted covered. There would probably be various standard packages offered by insurers, resources such as Consumer Reports and word-of-mouth, etc. 2) Buying insurance policies in other areas routinely requires people implicitly to forecast the future in ways that they lack the expertise to do really well. E.g., home owners’ insurance, car insurance. 3) People will not make perfect decisions! But is there any reason to expect legislators to make better ones?

and Michael Cannon:

What Ponnuru proposes is to let Arizonans and Idahoans and everyone else choose what their health plan covers. Imagine that: people rationing medical care according to their preferences, rather than the preferences of employers, interest groups, bureaucrats, health policy wonks…

First off, markets are great because prices act as a source of feedback information on the margins for both consumers and producers. I enjoy Mission neighborhood burritos, and it is not from the benevolence of the burrito maker that he sells them to me. If the taqueria gets more business, it can raise prices because people are willing to pay; this price raise signals to me that business is good, and maybe I’ll check it out. Not just burritos though – the price of oil goes up, so I’m going to go ahead and drive less. Producers are also going to dig for more oil, and invest in technologies to replace oil.

Think of it like that model above, where the magic of price discovery is happening. Now if prices get too high for the burrito stand, he’ll see an immediate drop in customers. He can either drop prices, raise the quality, or accept the lose in customers. Whether or not sour cream is worth 25cents or 30 cents to consumers is information that needs to get processed. The price give information, at a marginal level, to both the producer and to the consumer. When you think of Hayek, you think of lots of these little decisions being made, pushing the price this way or that way, with that price then influencing further decisions. It’s a pretty fantastic system.

That is not what is going on with insurance markets. I don’t pay the burrito guy to make sure I never get a burrito. I don’t pay for a burrito to be delivered on one random morning when I’m in my late sixties. I pay, immediately get the yummy burrito, and my feedback moves prices marginally this way or that way. With insurance I’m not paying to immediately enjoy something I can give that precious information feedback to – I’m making sure I’m covered in a bad state.

Look at this diagram; I’m at C – healthy, though I could be at A – sick. So why don’t I insurance myself to D and have piece of mind. This is not the same – it does not involve processing a big amount of little information, but instead a little amount of big information – the tail risk that you get sick. Moving A a little bit changes the whole game, and to compute A you need to be able to get the tail risk down perfectly, as well as understand what happens when you get there.

People are terrible at estimating tail risks; they underestimate it. Please, if you take away one thing from the financial crisis, let it be that. The finest minds in Wall Street, people who just had to figure out risk versus reward, couldn’t handle the tail risk. How are you supposed to be able to handle this with cancer risk? This is a hard problem for a person to solve. As a game theorist a Ezra’s blog said: “In game theory, players must rationally attach values to each outcome for a solution to emerge — this is a fundamental assumption. But with health issues I feel it’s impossible for customers to be rational in calculating these values.”

Luckily I’m a financial engineer. We can solve the price of anything! CDS, CDOs, Florida Orange futures, you name it! How would I solve this? The liability is that your risks grow with your old age, so any calculations need to be done very far out in your future. I’d need a random process for your health. I’d need random processes for your income, the poor health possibilities you’ve inherited from your parents, all the possible places you could live/work and their risk hazard functions, how much you’d dislike each and every possible disease imaginable, the cost and marginal effectiveness treatment functions off each disease, and then all the correlation matrices, along with the copula functions that links up these cross correlations with survival times on your life versus diseases (my god, we’d need a lot of copulas). Link 20 computers together, run it overnight, and it would solve which health care package you should choose.

In case you skipped that, the question at hand is hard. Let’s go with a much, much easier question – insurance on your house. This is a walk in the park. We’d only need to optimize in a few dimensions. You know what your house is worth, you have a sense of how risky its neighborhood is. There’s no correlation matrix for how likely it is you will have gout or a strange form of cancer, and no decision on how much you’d like that. This is a piece of cake, decision theory wise. So how to consumers do? Do they get the tail risk ok? Or do they underestimate it or not understand what happens when they end up in the bad state? So spending 5 minutes with my online library database:

Testimony of The National Association of Insurance Commissioners, Before the Subcommittee on Housing and Community Opportunity, July 17, 2007: (my underline)

There are many lessons to be learned from hurricane Katrina, but perhaps the greatest insurance lesson we can take from that tragic event is that many consumers are confused about what their property insurance policy covers, and what it does not. The NAIC recently conducted a survey of homeowners and found some alarming information about how they perceive their homeowners insurance coverage. Despite extensive media coverage of hurricane Katrina victims whose claims were denied because they lacked flood insurance, 33 percent of U.S. heads of household still incorrectly believe flood damages would be covered by a standard homeowners or property and liability policy. 35 percent of homeowners believe that damage caused by earthquakes is covered, and 68 percent believe that vehicles damaged on their property (potentially from a storm) are covered by the homeowners policy. The results of the survey are alarming and revealing. Clearly consumers have an expectation of broader coverage than their policies currently allow. The rationale is simple to appreciate: They are paying a premium and want their property to be covered. They don’t care what the cause of loss is or what the mechanics of covering that loss is. They simply want to be made whole and begin rebuilding their lives as quickly as possible.

Survey brings underinsurance home: Most aren’t covered for storms or haven’t updated policies
Michele Derus. Knight Ridder Tribune Business News. Washington:May 21, 2006. p. 1

Most Americans under-insure their homes and possessions, without realizing it, new research shows.

“Many people, if a horrible storm were to hit, would be unable to collect” on their entire losses, said attorney Thomas Martin.

Martin runs the non-profit Homeowners Consumer Center in Irvine, Calif., which this month sounded a public alarm on insurance gaps. The center’s telephone poll of 1,197 policy-holders found that:

— 85% admitted that they didn’t understand their homeowner policies.
— Only 3% knew whether they had full replacement coverage, which reimburses losses at current rather than original price.
— Only 9% knew that expensive jewelry, artwork, antiques and guns needed to be appraised and listed with insurers.

Forbes

Hurricanes Katrina and Rita, the construction spree in China and the domestic real estate bubble have driven up the demand for just about everything needed to fix and build homes, from labor to Sheetrock. That leaves many homeowners drastically underinsured if the unexpected occurs. Marshall & Swift/Boeckh, a firm in New Berlin, Wis. that tracks building costs for insurers, contractors and appraisers, estimates that three out of five homes in the U.S. are underinsured by an average of 22%.

Check your homeowners insurance policy every few years to see if it’s still good enough, given building costs in your area; check annually if the home is worth $1 million or more. Otherwise “people may find themselves in a position of not having enough insurance to rebuild what they once had,” says Scott Spencer, personal insurance appraisal manager for Chubb Group.

Many homeowners assume that a “replacement cost” policy will cover the cost of rebuilding a home of equal quality. There’s a gotcha, however. As often as not this turns out to be the replacement cost of the structure at the time the policy was sold. Meaning: Your policy may not cover such cost-inflating factors as new building codes or price increases for labor, wallboard and two-by-sixes.

Sounds like the health insurance market? It looks like consumers aren’t having their real preferences shown when it comes to claims on what they thought their insurance covered. It looks like many of them, in fact, would rather just be “covered”, broadly defined, than pick-and-choose like a buffet – 95% in case of fire, 90% in case of earthquake, 75% in case of flood – like the way I get my burrito. If I thought I wanted sour cream, and I don’t, the market can update quickly. If I thought my house was covered, and it wasn’t, I’m f***ed. Imagining the process of these tail risks is quite difficult (and let’s table the issue as to whether or not the insurance companies are trying to be confusing). Most of these people can’t negotiate because they can’t handle the complex possibilities of what happens when disaster strikes – they’d rather outsource that complexity, from what the articles tell us. And that’s with a house, not your body, where the decisions on what best to do when, say, you get cancer, are very hard for us to know by ourselves in advance.