Myth Busters #14: Adverse Selection
Another misunderstood principle of insurance financing is so-called “adverse selection.”
This term is misunderstood or misapplied by most of the health policy community, which is poorly informed about insurance principles. The actual term is simply “selection.” In a competitive market there is both anti-selection and pro-selection. A company that attracts the good risks is enjoying favorable selection, while one that attracts the poor risks is suffering negative selection.
In either case, people who have a choice will select the company and the benefits that are most appropriate to their needs. An older person with back problems may load up on chiropractic benefits, while a young couple might prefer maternity coverage. In general, people of higher risk are much more likely to value and demand generous health insurance coverage while people of lower risk will place less value on it.
This principle can be used to support two very opposing responses. Each has profound consequences.
The first is risk-based rating, i.e., vary premiums according to the likelihood of use. Since some people value the coverage more, it is fair to charge them more, but since other people place a lesser value on the coverage it is necessary (and fair) to charge them less. Plus, it is possible to vary benefits according to each person’s preference, so the older person doesn’t pay for maternity and the younger couple doesn’t pay for chiropractors. This is why, at least in the individual market, insurance is often sold with optional riders for services like prescription drug coverage, maternity, and visual and dental. Many people don’t think they will ever use these services and don’t want to pay for them.
An alternative response would be to eliminate choice by requiring all to have the same coverage. Require older persons to pay for maternity coverage even though they will not use the benefits and require the young couples to pay for chiropractic coverage even though they, too, will not use it. This approach spreads the cost of the benefits over a broader population, but it also risks creating resentment as people are forced to pay for services they know they will never use. Also, unlike the previous approach it causes some people to be over-charged and others to be under-charged. In general, those who are under-charged will over-insure — by purchasing more coverage than they otherwise would — while those who are over-charged will under insure.
This second response also creates another problem, which is almost always overlooked in policy circles — moral hazard.
While selection means people will choose the benefits they are most likely to use, moral hazard means the opposite — once people are covered for something, they are more likely to use it. If I am required to be covered for psychiatric counseling, I am far more likely to make use of it, even if I would never use it otherwise. After all, I’ve already paid for it so why not use it?
Moral hazard applies to most of what insurance covers. Granted, the rate of maternity or traumatic injury may not be affected by moral hazard, but most outpatient and diagnostic services are. This is part of the reason so much of what is done in health care is considered non-essential. Having coverage leads to greater use of both essential and non-essential services.
It is also why the cost projections of expanding coverage are almost always underestimated. Budget analysts often assume that the behavior of a newly covered population will be constant — low users of the system will continue to be low users, or at most will consume services at the same rate as an already-covered population. But that fails to account for the pent-up demand question. Moral hazard suggests that newly insured people will consume more services when the cost of doing so falls.
Both selection and moral hazard suggest the value in minimizing the amount of services that are paid through an insurance mechanism. Insurance is important to cover services that would otherwise be unaffordable, but the presence of insurance coverage changes behavior and distorts normal market mechanisms.
While we’re on the subject, there is one other principle that should be mentioned. The surge in spending may not show up in the first year of coverage. It generally takes newly insured people about a year to figure out how to use their coverage. Also, benefits may not become available immediately. There may be an “exclusionary period” of 6 to 12 months for pre-existing conditions. The federal HIPAA law, for example, allows insurers to not pay benefits for conditions that existed prior to coverage for a period of 12 months. This phenomenon was part of the reason carriers used “durational rating” in which the first year’s premium was lower than in succeeding years. Durational rating has been largely eliminated as consumer advocates viewed it as a bait and switch tactic, but the reality of lower first-year spending remains.
To sum up, selection means that people will choose the benefits they most prefer. Eliminating selection means forcing people to buy coverage they don’t want and/or underpricing coverage they do want. In either case, people respond by obtaining the wrong amount of insurance — they over-insure or under-insure. Moreover, once they have benefits they would not have otherwise purchased, they will consume services they would not have otherwise consumed.
Eliminating “adverse selection” will always result in higher costs and dissatisfied consumers.
Good post.
Selection is a bet on things you perceive you’ll need, and on what you can afford. In insurance it is also bet on what the insurer thinks of the person’s risk and needs. These two perceptions can be divergent. With any bet you’re not going to be right 100% of the time.
Could the role of a well-designed deductible lower moral hazard? Good health care is just as much as personal responsibility as the strength of the health system to take of you when something happens.
Greg does a good job of explaining a complex topic. Another way to discuss selection is that people need less of their premium dollars going towards pure insurance when you compared to when they get older. If a greater percentage of premium dollars going into an HSA while young, the funds will be available to spend down once a couple reaches middle age. This way, people spread health risks across their own working lives rather than rely on an arbitrary risk pool when their health status declines. Selection assumes people have different demand for care and can adjust accordingly. An HAS would make it to where selection would not be as much of a problem.
Devon,
Exactly right. We can pool risk over a population so that healthy young Joe helps pay the expenses of older sicker Sam. Or we can pool risk over time so that the young Devon helps pay the expenses of the old Devon twenty years later. There is no inherent advantage of one approach over the other.
Devon and Greg:
For the approach under which a young Devon helps pay the expenses for an old Devon through an HSA, which whould general taxpayers be drawn into this deal, in a way that makes taxpayers subsidize Devon the more, the higher his income is?
Uwe
1. One component of “selection” not covered in this discussion, and often ignored by “free market” proponents, is that for profit insurers have a very powerful incentive to avoid selecting for coverage people whom they perceive will cost them money. Their rate differential is not just a gradual upward slope that rises with age, as implied here, but rather is often a severe upcharge that will be imposed at any age, once a carrier perceives an individual is high risk. In my mind, this is also a “moral hazard,” though the term is never applied to the carriers.
In light of this, what we need is a national high deductible or catastrophic care plan, funded ultimately by all taxpayers. Then let the free market compete in the “dollar trading” end of the business, where what is being covered is not really insurance, but rather predictable expenses.
Uwe,
I wasn’t particularly thinking of HSAs, but I would prefer the taxpayers to subsidize the poor Devon more than the wealthy Devon. But that principle should apply to any form of health financing — HSA or insurance coverage, eh? Both should be treated the same.
Jeff Padnos–Yes insurers try to avoid people who cost them money, which is why insurers charge higher premiums for people who are certain to cost them more.
Only when prices or policy terms are controlled, as in Medicaid Advantage, do insurers go to great lengths to attract only the healthy. It happens in socialized systems like Germany as well as in the US.
As for catastrophic coverage covered by taxpayers, we already have four forms of it: Medicare, Medicaid, Social Security disability, which makes one eligible for Medicare, and federally required state subsidized guaranteed issue policies for the uninsurable.
Great article. But let’s not be value-free; the effect of such policies is not just higher costs and dissatisfied consumers.
Forcing people into plans deemed to have “sufficient” coverage rather than allowing them to choose the benefits they most prefer is a violation of individual rights. That cost stings the most, if you ask me.
No matter how one slices and dices “insurance” one will always wind up chasing one’s tale if the holy grail is to make the concept “fair”. By the nature of it there will always be losers (by someone’s definition).
IMHO the best way to deal with insurance is to make it a point to buy as little of it as possible, which is why HSAs and other methodologies that deemphasize insurance are the saving grace.
The issue UNDERLYING moral hazard and adverse selection (whether practiced by insured or insurer) is information asymmetry. Thought experiment. Imagine that you knew the exact cost of your future medical care. You would compare insurance products against that cost and select the one that gave you the most benefit after subtracting premiums. If there was no product that gave you benefits in excess of premiums you wouldn’t buy insurance. Insurers would quickly go out of business because there would be no losers to pay the winners.
Now, suppose insurance companies knew the exact cost of your future medical care but you had no idea whatsoever. In this case, insurance would be a highly profitable business because carriers would only sell policies to those people who radically overestimated their future costs. Presumably, people who chose to buy insurance in such a market would be satisfied (they chose to buy it, after all). But anyone whose future medical costs exceeded his or her willingness or ability to pay would not get insurance and, as a practical matter, would end up having such costs paid by charity, the government, or by private providers under force of the government.
The first situation roughly describes every social insurance scheme from the British PHS to the Mass Health Exchange. BUT, advocates say, we can solve the problem by REQUIRING coverage.
The second situation roughly describes the individual market in less regulated states. Neither is desirable. BUT, advocates say, we can solve the problem by RATE BANDS and PRE-X and the like.
Good luck with both of those. In a dynamic market, neither will last for long. Consumers forced to pay for something they don’t want will FIND WAYS to avoid payment or to get value out of it. Insurance forced to sell to unprofitable customers will do everything possible not to do so, or not to provide the benefits. Indeed, these suboptimal behaviors become the whole basis for getting the most value from the system. Gaming produces wealth. Bad situation.
Insurance only really works over an extended period of time where BOTH sides suffer from the same lack of information (or control, Greg might add) over the risk being insured AND where consumers see genuine value in narrowing the range of possible outcomes. Life insurance (though it has many problems) is about as close as we get in the real world. I can pretty quickly assess my life expectancy nd the probability distribution around it. The insurance company has more precise data, but on the other hand I know more about me personally, and so it more or less balances out. The reason a market exists is that I find it valuable to eliminate the tail risks, what if I die before my kids are on their own? This isn’t a question of me and the insurer having different probability estimates. It’s the insurer providing a valuable service by pooling people to increase the “n” and thereby bring in the tails of the distribution.
Health insurance would work just fine if we were only insuring against events about which insured and insurer had comparable information (or lack thereof). My risk of being hit by a lighting bolt or getting a rare disease. A deductible is a rough surrogate for that. The higher the deductible, the better it is a surrogate. The problem is, thanks to science and math, we have less and less and less stuff that can’t be predicted by one side or the other.
Until policy makers and advocates recognize how markets (meaning other people) really behave, we won’t make much progress.
Jon,
Thanks for the very thoughtful observations. Excellent post.
Great post. When I think of moral hazard Friedman’s “Nobody spends somebody else’s money as carefully as he spends his own” immediately comes to mind. Today policyholders, hospitals and doctors are all tempted to spend insurer money. I believe insurers will eventually address this with a plan design that will encourage a health marketplace to form. In this world insurers will dial in on lower premiums, will see flat trends, AND maximize consumer choice. What is painful is that the inertia of the current arrangement is so great.
Greg,
Small Group Reform prevented small employers form being experience rated, which greatly increased the issue of selection…or at least shifted it and forced other companies to pay for it.
When a health plan is pooled right down to the first dollar we have these issues. A solution would be to increase the attachment point based on a claims credibility factor. Maybe a 5 life group would pool above $10,000 and a 10 life group at 15,000, 20 lives at $25,000 etc. In many states an ERISA partially seld funded plan can actually accomplish this, then the funds below the aggregate attachment point are pooled in a health escrow account, allowing low level experience rating within one employer.
Ultimately, this leads to employer wellness and education incentives, and really does keep costs down.