Arnold Kling has a new book out, with the title What’s a nice guy like me doing in a flack shop like this? “Learning Economics”. If what you want is an introduction to economics from a somewhat aggressively libertarian perspective, I daresay it will be pretty good; in my experience, Arnold has almost never been intellectually dishonest himself (which further raises the question, why’s he providing window dressing to TCS?).
However, in plugging his book, Arnold repeats a mistake I’ve corrected him on a couple of times, so let battle commence. Specifically, he claims that
“If you think that paying for your own health care is too expensive, I argue that it is mental illness to believe that paying for each other’s health care is affordable.”
I think that this is based on a pretty egregious confusion between a dull statement about health care, about which this statement is trivially true, and health insurance, about which it is probably false. Read on …
The basic issue is that insurance is about risk pools, and bigger risk pools are more efficient than little ones, because the bigger the pool, the more likely it is that in any given period, the expenses will be close to the actuarial expectation. Or to put it in laymans terms, think of an operation costing $10,000, which everyone has a 1 in 100 chance of needing in any given year. If you pool a million individuals, then everyone can pay just about $100 a year and get their operation paid for. If, on the other hand, everyone pays on their own (the size of the risk pool is one individual), then everyone needs access to $10,000 to avoid a catastrophic loss.
This is the simple theory of insurance, and it means that Arnold is right away silly to be claiming it’s “mental illness” to suggest that pooling people together can make unaffordable costs affordable – that’s exactly what insurance does. In fairness, Arnold does mention in his TCS article that insurance makes the whole question more difficult. But I think he’s underestimating quite how complicated insurance markets can get, and in this post, I’m going to illustrate how complicated they can get with a simple(ish) adverse selection example[1]. In the case set out below, despite the assumed existence of insurance companies prepared to write business at a fair price, it turns out to be absolutely impossible for the insurance market to exist without government intervention. And it’s all because of a bloke called Fatty. Follow along with a pencil and paper if you like, or alternatively email me and I’ll send you the spreadsheet I used to cook up this numerical example.
OK, we start in a small economy[2] with ten people in it: nine of them are called Jim-Bob and the tenth is called Fatty. There is a terrible disease that haunts the land, which is so nasty that anyone who caught it would be prepared to pay almost any amount to be free of the crippling pain it causes. However, as luck would have it, the disease can be cured with a simple operation costing $600. However, as luck wouldn’t have it, the Jim-Bobs and Fatty don’t have any savings, so if they find out they’ve caught the disease, they will have to borrow money from the local loan shark at a rate of 20%[3]. For the pencil & paper crowd:
Cost of providing cure: $600
Cost to individual of paying for cure: $720
Not everyone is equally exposed to the risk of this disease: the Jim-Bobs all have a 10% chance of catching it but Fatty has a 20% chance. Every consumer knows what their risk is, but the insurance company has no means of telling them apart[4] ex ante. Once they have signed the insurance contract, however, they reveal their actuarial risk to the insurer[5]. I’m also assuming that the insurer has general information about the population risk:
Nine individuals with 10% risk of disease, one with 20%. Risk factor is private information to individuals ex ante, public ex post.
Now, let’s take a first look at this insurance market with one insurance company. The actuarially expected cost of providing operations during the period is calculated as follows:
Cost of insuring a Jim-Bob = 10% x $600 = $60
Cost of insuring Fatty = 20% x $600 = $120
Cost of insuring nine Jim-Bobs plus Fatty = $660
Ignoring profits and such[6], the insurer needs to set a premium to cover this cost. Since the insurer doesn’t know Jim-Bob from Fatty, he therefore sets a premium of $66. Who buys insurance?
Jim-Bob’s expected cost of healthcare without insurance is 10% x $720 = $72 – he buys insurance @ $66.
Fatty’s expected cost of healthcare without insurance is 20% x $720 = $144 – he buys insurance
So the market works if we assume only one insurer offering a single insurance contract. However, this isn’t an equilibrium. It isn’t an equilibrium, because another insurer can come in and offer a contract which breaks the pooling equilibrium. They do this by offering a lower premium, but with an excess. The idea is that, usually, lower risks (Jim-Bobs) will regard it as a good deal, but Fatties won’t.
New contract!
Let’s assume that the new insurer jacks the deductible up as high as possible in order to scare off Fatty. If we set the deductible at $250, then the actuarial cost of insuring a Jim-Bob is:
10% x (600-250) = $35
and we assume that the insurer sets the premium at this level.
Who buys insurance at these rates? We assume that if someone under the new contract gets the disease, they still have to borrow from the loan shark @ 20%, to pay the deductible. So the cost to the insuree of the $250 deductible is $300.
Jim-Bobs’ expected cost of healthcare under the new contract = $35+(10%x$300) = $65. $65 is less than the pooled premium of $66 – Jim-Bobs buy the new contract
Fatty’s expected cost of healthcare under the new contract = $35+(20% x $300) = $95 > $66Fatty does not buy the new contract.
So, the new “separating” contract has broken the old pooling equilibrium. Is the new state of the market a separating equilibrium?
No.
Remember that the insurers are able to know the risk characteristics of their portfolios[7]. The old insurer now only has one customer (Fatty), and the actuarial cost of providing insurance for Fatty is $120. So this insurer needs to bump up the premium for the no-deductible policy to $120.
But …
Fatty can get a better deal across town[8]. Under the $250-deductible policy, his expected cost of healthcare is $95. So he’s going to buy that policy instead.
So now we have an equilibrium with everyone taking the new, high-deductible policy?[9]
The premium on the high-deductible policy is $35. That is the actuarial cost of insuring a Jim-Bob. When you add Fatty to the risk-pool, the total actuarial cost is:
Cost of insuring Jim-Bobs=$35/head (see above)
Cost of insuring Fatty=($600-250) x 20% = $70
Total cost of providing insurance = $385
So with Fatty in the pool, the premium has to go up to $38.50.
But now the expected cost of healthcare to Jim-Bobs is $68.50 (because of the premium increase).
So … this equilibrium can be broken by anyone who wants to offer a contract with zero deductible and a $66 premium!
So what’s the equilibrium?
There isn’t one. Basically, if there are lots of Fatties (such that the difference between the actuarial cost of a Jim-Bob-only insurance company and the cost of the pooled company is large relative to the loan shark rate of interest), then you can get into a situation in which there is a separating equilibrium. I constructed this example, however, with only a few Fatties, which makes the difference small enough that the Jim-Bobs can be attracted back to the pooled company to avoid the risk of paying the loan-shark. In this example, there is simply no contract that can be offered which leads to an equilibrium which can’t be broken.
And so we reach the punchline
In a perfectly competitive market, insurance could not exist in this situation. No insurer would commit capital to a market in which any contract was vulnerable to becoming uneconomic at the drop of a hat. In order to get insurance to exist, you either have to assume a cartel in pricing, or you have to introduce something like the NHS; a pooled insurer which Jim-Bobs have to pay for whether they want it or not[10]. Since one of the things which we do know about health insurance is that the costs are driven by a small number of very expensive patients, I would suggest that my “Blame it on Fatty” model is potentially a quite realistic stylised picture of the situation. And in this model, not only can we get something cheaper by pooling our funds, but if we don’t pool our funds, we can’t get it at all!
Footnotes:
[1]Adapted from a more rigorous treatment by Ray Rees.
[2]Also, an economy which only lasts for one time period; time isn’t important in this model.
[3]This assumption is doing the work of a proper diminishing marginal utility function; I’m just trying to get something that will insure that in the model, as in life, when expected values are equal, you prefer a small certain cost to a chance of a larger loss.
[4]Don’t get your hopes up too high about genetic screening as a solution to this problem. In our one-period model, the risk factor is important, but in a more realistic model, the variance of underwriting returns is driven much more by the timing of losses than by their likelihood or severity. Also, insurance companies (potential subject for another post here) are usually reluctant to be too aggressive in price-discrimination in health insurance, because if you have too aggressive pricing differentials, you tend to attract loads of business in areas where you have underpriced and none in areas where you have overpriced, and that’s a certain recipe for bankruptcy in the long run.
[5]This wholly artificial contrivance is meant to stand in for a proper dynamic process of updating the actuarial estimates, holding capital in the general insurance fund and all manner of what have you not related to the adverse selection question.
[6]Yeah, yeah, I know; everyone does it.
[7]And remember that this assumption is standing proxy for more reasonable dynamic assumptions.
[8]Remember the information assumptions.
[9]Note that even if we did, there would still be a rationale for government intervention here; this is a suboptimal solution because the Jim-Bobs would be buying less insurance than the really want as part of the cost of scaring off Fatty.
[10]Seems unfair to make the Jim-Bobs pay for what is fundamentally Fatty’s problem? Yeh, but remember that this is what happens in the competitive equilibrium too; the Jim-Bobs bear the cost of Fatty’s existence through having to buy less insurance than they would otherwise want to.
{ 54 comments }
dsquared 09.22.04 at 10:26 pm
BTW, some cynics might be wondering if there was anything special about my choice of $250 as the deductible in the example. There isn’t; if you set it much higher, then the new contract isn’t worth taking on for Jim-Bobs, and if you set it much lower, then you’re just making the new contract more attractive to Fatties. The Big Number assumptions in this model are the proportion of fatties and the loan shark rate of interest. (And in fact, for any loan shark rate of interest, I can set a proportion of fatties which still gets this result).
Btw^2, if you are taking PPE exams at Oxford, or their equivalent, I strongly advise you to get hold of a copy of “Current Issues in Microeconomics”, edited by John Hey. It’s pure gold.
fasteddie 09.22.04 at 10:41 pm
Please add the part about the medical paperwork costing $30 for each of the surgeries because of the multiple insurance companies, while only costing 1¢ if there is only a “single payer”. Thereby lowering the costs for everyone.
Sam 09.22.04 at 11:28 pm
This situation is theoretically possible, but not very likely absent regulation. Only in a semi-regulated market is the problem of heterogenous risks likely to pose a risk to the availability of insurance.
The insurance company has a huge incentive to identify Fatty; if it can do so, it can charge the correct “actuarial premium” to everyone–$60 for Jim-Bob and $120 for Fatty. It could spend up to $6.66 per policy on underwriting, and everyone would be able to buy underwritten insurance. It is very rare that a risk factor this large can’t be (reasonably) easily identified.
A regulated market is where most of the problems arise, because in the interests of fairness-as-equality, regulators often forbid complete underwriting. In that case, you can (and frequently do) get the kind of market failure which you outline.
dsquared 09.22.04 at 11:37 pm
This is a common mistake made by people about the insurance market. There is no very strong incentive for insurers to identify Fatty, unless they are sure that nobody else will be able to. Since everyone else has the same information, the real incentive is to ensure pricing discipline and the prevalence of the first pooled equilibrium, at a premium rate in excess of the actuarially fair rate.
It is not in general easy to identify bad risks for a variety of reasons. First, because they are usually not bad risks at the time when the contract is written, and second because the badness of the risk, as discussed in the article, is driven more by the timing than the severity or likelihood.
Empirically, insurance companies do not price-discriminate to the extent you think they do. Despite the plethora of questions asked, insurers write the vast majority of business on standard terms.
Rob 09.22.04 at 11:39 pm
Yes, perfect information would solve the problem. Of course in the real world perfect information doesn’t exist and is much closer to having no idea ex ante than having perfect info. If DD wanted to, he could make the model have a signal instead of no info ex ante, but you can still end up with seperating equilibrium.
Giles 09.22.04 at 11:53 pm
“In a perfectly competitive market, insurance could not exist in this situation”
Not true – your entire thesis is based on the fact that the cost to the buyer is 20% higher than the cash price. go back and do the sums with the price = 600 and things are fine.
So bacially what you’ve proved is that helath insurance may be a good idea if you have capital market problems but you havent made the insurance point you intended.
I think point is more general – health does seem to bring out the nuts in economics.
adam scales 09.22.04 at 11:55 pm
Hi. This is a pretty interesting post, and I plan to save it and consider your analysis in greater depth. Let me make two quick points:
First, your overall point is historically inaccurate – unregulated insurance markets have existed for much of the centuries-long history of insurance (adverse selection problems are not unique to health insurance). Of course there has always been SOME regulation, but it has varied widely, and it is a relatively recent phenomenon to have price regulation (or highly detailed regulation of the scope of coverage). Assuming that the main intervention you identify as necessary is some kind of subsidy to risky insureds, it is demonstrably untrue that insurance markets cannot exist without it.
Of course, without that, we see some of the problems we have in the U.S. – I do not suggest in this brief comment that there is no place for such aggressive regulation. I only suggest that one can have a functional, if not ideal system without it.
My second point is that Kling’s point, while perhaps incomplete, is not a wide of the mark as you suggest. It is commonly asserted that universalizing health care would solve the problems of controlling costs (which, together with its impact on access, it usually regarded as the main thing “wrong” with the health care system).
Government subsidies to the point of universality would simply trade a different set of problems for the problems we currently have. For example, the admin. savings positied by a comment above, while perhaps overstated, would be a real savings. That extra 12% or so would undoubtedly buy health care for a lot of people. Maybe it is worth it. However, private admin. costs do get us SOMETHING besides convenient cost shifting for private insurers. In addition to driving people nuts, they signal some of the costs of treatment. The single-payer system in the U.S. (medicare) is not encouraging on this point. We’d need to know, before going universal, how the systems would compare on this point. My own guess – and it is just that – is that such a move would buy a little time, nothing more, before the underlying dynamic broke out afresh.
I’m sure Kling espouses a market-driven approach to reform, and one should consider that bias. However, if I understand the thrust of your thoughtful post, it is that health care cost problems are primarily a problem of the inefficient, insufficiently regulated way private insurers would handle adverse selection. I disagree. I think health insurance is expensive because health care is expensive, and growing more so, for many reasons totally disconnected from insurance markets. Perhaps those markets exaggerate the problem (I’m more optimistic about that, but no one could call this a success), but at the end of the day, I think Kling is correct to deny that expanding health insurance would necessarily lower individual costs.
dsquared 09.22.04 at 11:55 pm
Footnote 3, Giles.
dsquared 09.23.04 at 12:01 am
First, your overall point is historically inaccurate – unregulated insurance markets have existed for much of the centuries-long history of insurance
This has almost always been through an enforcement of the pooling equilibrium as opposed to the separating one, enforced by informal and formal cartels. You can always make the market work if you assume that premiums are set far enough above actuarially fair rates to insulate the providers from any consequences of their underwriting actions, but this in itself goes against Kling’s thesis; insurance is more expensive.
adam scales 09.23.04 at 12:14 am
dsquared,
Certainly, there is evidence for your point as particular insurance markets developed. I think that tends to be less true today – even for those markets not subject to much regulation. But, let me stipulate that it is always at least somewhat true.
Would that really account, for example, for 10-15% annual increase in health insurance costs, as we have observed recently? I can’t believe the adverse selection model explains more than a small fraction of this, as opposed to outsized increases in the price tag of pharmaceuticals, treatment charges, and the like.
nick 09.23.04 at 12:47 am
Thus Kling: I argue that it is mental illness to believe that paying for each other’s health care is affordable.
As an aside, it’s worth noting that one essential aspect of universal healthcare systems is that it pays for other people’s mental healthcare: particularly, those situations where the individual is in no position to make an informed choice about coverage and treatment. My wife works in the field, and it’s one area where the US has systematically failed its populace.
Also, Kling should be ashamed of himself for his choice of words.
Giles 09.23.04 at 1:01 am
Footnote 3, Giles.
No it may be simulating that but its not consistently applied to for instance when the smaller costs of premium increase.
More signifcantly health is frequently cited as an areas where people’s assessment of risk is poor – more particuarly that they tend to overinsure against very bads.
A diminshing utility assumption kebabs this counter argument to insured markets. And empirically this is what the main problem with american system is – not that costs are too high not that some people dont have insurance but rather that those who do have insurance buy too much and accorindly pay the health service too much.
Rob 09.23.04 at 1:05 am
By the way the argument that this doesn’t happen because insurance exists is false. For the markets that don’t exist you can’t see the evidence because they don’t exist. And many markets haven’t existed until very recently even though they should have. Much of the 1980s the finacial secotr developed markets that should have already existed in theory but didn’t. And that deals with goods with very little in the way of transaction costs.
John Quiggin 09.23.04 at 1:59 am
“There is no very strong incentive for insurers to identify Fatty, unless they are sure that nobody else will be able to.”
I think that should be
“as long as they are sure’
Jason McCullough 09.23.04 at 2:20 am
I’m more curious why your example demographic consists of hillbillies and a single fat guy. Insurance for the Appalachians?
Brian McDaniel 09.23.04 at 2:25 am
Hmm, does the fact that competitive markets for insurance do exist undermine your confidence that “‘Blame it on Fatty’ model is potentially a quite realistic stylised picture of the situation.”?
Jeremy Osner 09.23.04 at 2:36 am
Looking at this example I don’t see anything that makes it apply uniquely to health insurance — it seems like the general framework applies equally well to most any type of insurance.
Nicholas Weininger 09.23.04 at 3:05 am
1. It’s really hard to evaluate empirically whether insurers would price-discriminate if free to do so when they are currently not really free to do so. In the current US health insurance market, “community rating” and other similar laws severely restrict the extent to which insurers can price-discriminate.
There are plenty of insurance markets where insurers do engage in significant price discrimination, using criteria that are known equally to all insurers and that have very significant bearing on risk. Auto insurance is a good example: age, sex, driving record, place of residence, etc. all affect risk and all affect premiums charged. One insurance company is just as capable as the next of finding out who the high-risk drivers are, yet they certainly seem to have a strong incentive to do it and do it well.
Is there really any good reason to believe that analogous easily determined factors– age, sex, health habits, medical history– are not capable of telling health insurers enough about health risk to make price discrimination worthwhile?
2. Your simplistic information model– no ex ante info, full ex post– may be doing the work of more realistic things, but it seems inherently unrealistic in ways that pose serious problems for your conclusions. For example, you seem to be saying (apologies if I’ve misunderstood) that Fatty could get insurance from insurer A who doesn’t know his risk level, then have his premium go up once A finds out he’s a bad risk, and then turn around and get a lower premium from B who still doesn’t know his risk level even though it’s been revealed to A.
What realistic process is this standing in for? What is it that’s supposed to reveal Fatty’s risk level to A without letting B in on the secret? If Fatty acquires a “preexisting condition”, for example, B is certainly going to find that out and react accordingly.
Nicholas Weininger 09.23.04 at 3:32 am
And to expand on giles’ point: on what planet do consumers have access to large amounts of good knowledge about their health risk that insurance companies have no way of getting at?
I am aware that insurers may be restricted by law from using or attempting to gather some such types of information, but that’s an argument against regulation, not for it.
Ethesis 09.23.04 at 4:08 am
err …
Empirically, insurance companies do not price-discriminate to the extent you think they do. Despite the plethora of questions asked, insurers write the vast majority of business on standard terms.
You need to follow the entire COBRA market. Basically, in group insurance pools when someone leaves employment, they can continue in the insurance pool, but the price of the insurance is written by the risk the individual has. I’ve seen premiumns up to $30k a period.
Not to mention, death spirals in pools. A pool starts, if too many Fatties are identified in it, rates go up. The Jim-Bobs leave and the Fatties (who have been identified and who are not ensurable elsewheres) stay, with the price going up. As more and more Jim-Bobs leave the pool goes into what is referred to as a death spiral.
Lots of real world stuff going on that would make a nice addition to your comments.
Robbo 09.23.04 at 4:18 am
fasteddie wrote: “Please add the part about the medical paperwork costing $30 for each of the surgeries because of the multiple insurance companies, while only costing 1¢ if there is only a “single payerâ€. Thereby lowering the costs for everyone.”
If by “single payer” you mean, say, greenbacks, yeah, that’ll work. (If all the hospital had to do was issue a receipt to the patient, you could keep the paperwork to a minimum…) If by single payer you mean “government”, I say “KEEP DREAMING!”
If you make the US federal government the single payer, it will dramatically increase the cost of paperwork; in the USA, government oversight of insurance (the VA system, Medicare, Medicaid, etc.) has been the biggest source of increased costs to doctors, hospitals, etc.
Moreover, your $30 isn’t far off for seeing the doctor (but is probably a gross underestimate for a surgery), thanks to the Centers for Medicare and Medicaid Services + HIPPA. One study estimated that about 3 years ago, before HIPPA, the average private doctor (or their employer) paid about $25 in costs to submit one single claim to the largest government-run payer, Medicare (in paperwork, staff time, etc.) *After* the full implementation of HIPPA (there’s 3 parts, FYI), the estimate is that cost will likely top $30 (more paperwork to fill out–simply to meet the requirements of the new law).
ie: one geriatric Jim-Bob goes to the doctor; his total bill (his part + Medicare’s submission) is $55 (a fairly average primary care doctor bill, on a national level), and over half of that does nothing but pay for the *paperwork* for the *bill* (that doesn’t even count the paperwork to record the actual patient care encounter–such as the doctor and nurses’ notes, the form grampa Jim-Bob signs to OK the tetanus shot he got, etc.).
Going to a “one-payer system” *may*, in the end, cause lower costs because of other factors (but even that, IMO, is debatable). However, anyone who thinks combining health care under one, government-run payer will somehow reduce the costs of things like paperwork is using too much of their “medical cannabis.”
Dubious 09.23.04 at 4:38 am
Looking at the U.S. insurance market’s failings, it seems to me that:
1) Mostly it’s people who have low income who don’t have insurance. This indicates that it’s not so much that insurance is too expensive relative to self-insurance (the example here) but rather that people (esp. young people, who tend to have better health and lower incomes) would rather spend their money on other things. This doesn’t mean insurance (like food or decent housing) isn’t some kind of ‘merit good’ all people should be provided in a decent society. Just urging people to sell it honestly as a merit good rather than an efficiency gain vs. assymmetric information.
2) Moral hazard (getting ‘too much’ medical care because the insured person doesn’t pay the whole social costs) seems like a much larger problem here in explaining the above-inflation rate of medical EXPENDITURE. As I understand it, it’s not medical COSTS that are going up (money required to get a certain procedure, much less the medical expenditures needed to lower mortality to a certain rate, which is how medical COSTS should really be measured, hedonically) so much as the introduction of expensive new drugs and procedures.
3) If above-inflation rates of medical EXPENDITURES were to be blamed on adverse selection market imperfections of assymetric information, we’d have to have some sort of story about how the underlying variance in health risks was becoming wider faster than the insurance firms can figure out a way around them with new computerized medical databases, genetic testing, etc. I think it’d be a hard story to tell to say that the underlying variance in health risks is becoming larger in the population.
(A Ph.D. candidate in Economics)
Shai 09.23.04 at 5:12 am
well, it would be nice if one of the easy to find pages decomposed the 18-64 cohort
Sebastian Holsclaw 09.23.04 at 8:43 am
“As an aside, it’s worth noting that one essential aspect of universal healthcare systems is that it pays for other people’s mental healthcare: particularly, those situations where the individual is in no position to make an informed choice about coverage and treatment. My wife works in the field, and it’s one area where the US has systematically failed its populace”
This is a complicated side problem. It isn’t always that care is unavailable (especially in the larger cities). Since the 60s it has become very difficult to involuntarily commit someone for any period exceeding an hour or two. There are very good reasons why involuntary commitments were curtailed, but it seems possible that we have gone too far. My point is not that the mental health care system in the US is excellent (for poor people it is not) but rather that the issue is dramatically complicated by civil rights concerns.
dsquared 09.23.04 at 12:26 pm
No time for detailed answers to comments right now, but note that
1) The proportion of Fatties is the key to whether an equilibrium exists; if it is high enough, there is no problem with the separating equilibrium.
2) I dispute whether there really is a competitive health-insurance market; I look at the US and see a system of local cartels.
3) Remember that the risk model is complicated; Fatty has a high probability of needing an expensive operation, but in a more realistic risk model, he also has a high chance of just dropping dead without ever making a claim (=pure profit). Otoh, a Jim-Bob might drag on for years on expensive medication. Seeding out “high risk” insureds where “high risk” is defined as “high financial risk to the insurer” rather than simply as a high medical risk, is very difficult indeed, which is why actuaries always tread very carefully in attempting to price-discriminate.
Ethesis 09.23.04 at 1:47 pm
(A Ph.D. candidate in Economics)
The driving force in expanded costs is two-fold.
First, in order for your doctor to make an extra $10.00 he has to prescribe you about an extra $100 in services. You can see that easily in post motorvehicle Chiropractic care. What used to run under $2,500.00 in care now runs about $10,000.00. Half of that in useless diagnostic tests (since the results have no impact on treatment), a quarter of it in value added services that are not theraputic (i.e. hot and cold packs used more than 72 hours after the accident and served from the facility rather than a single cold pack given to the patient to take home, put in the frig and use as needed).
The other issue is the constant stream of attacks aimed at forcing the Jim-Bob’s to pay for the Fatties. Think of fertility treatments, for example.
The huge side issue is the relative cost of treatment for noncompliant diabetics.
Of public medical dollars, close to half are spent on last six months of life treatment. Of that, a large proportion of the patients are non-compliant diabetics.
It was interesting to listen to Ms. Clinton in a moment of truth express her frustration that terminal patients couldn’t just die in peace without such large funeral pyres of medical expenditures.
I’d also note that diabetes is a commonly underwritten against condition.
It is a complex and interesting area, with significant factors often completely ignored.
Nicholas Weininger 09.23.04 at 2:32 pm
dsquared: I’m with you all the way on (2). I do wish people who cite the US experience as evidence that competitive markets don’t work in healthcare would remember this.
Re (3), though, it seems to me that the more complicated you make your risk model, the more you undermine your information-asymmetry case. If risks are really that complicated, probably nobody has very good information about them, which means the Fatties can’t use their informational advantage to break equilibria by playing successive insurers for fools.
BigMacAttack 09.23.04 at 4:03 pm
Wow! I am guessing dsquared must be very good at math.
If insurance companies and consumers cannot differentiate risk and the market is unregulated what happens?
If insurance companies but not consumers can differentiate risks and the market is unregulated what happens?
If both can differentiate risks and the market is unregulated what happens?
Those might be examples that might actually be helpful in determining the truth of the claim that –
‘In the case set out below, despite the assumed existence of insurance companies prepared to write business at a fair price, it turns out to be absolutely impossible for the insurance market to exist without government intervention.’
But I don’t think an example where the market is regulated so that insurance companies cannot differentiate between risks but consumers know their own risks really helps prove or disprove the claim.
I also think it is worth reiterating one of John’s points.
(Irony involved)
‘There is no very strong incentive for insurers to identify Fatty, unless they are sure that nobody else will be able to.’
There is no very strong incentive for businesses to innovate if they are sure that nobody else will be able to innovate.
Based on this I am forced to conclude that business innovation and the market are not in any way related.
rvman 09.23.04 at 4:53 pm
Looking at this from a macro perspective, the first thing that hits me about this analysis is that universal health care would remove an incentive for private savings, just like Social Security did. One of the problems people have screamed about has been our low national savings rate – universal health care only makes that worse – especially since government would undoubtably run additional deficits to “pay” for it. It seems to me that MSAs with catastrophic (i.e. extremely high deductable) insurance would be an efficient way out of this box – and may well be efficient. Your assumptions of “no possibility of saving” and “no ability for the insurer to identify high risk individuals, who know who they are” is what causes the crash. If the insurer can identify and lock out Fatty from the low-cost contract (like car insurers do by not offering their best rates to bad credit/ticket/accident history drivers, or increasing premiums for smokers), then the problem goes away. If Fatty and Joe-Bob don’t know if they are a Fatty or a Joe-Bob, then the problem goes away – they just do the same weighting the insurer does, and find that a premium between $66 and $72 is a “good deal” for all of them.
The purpose of Insurance is to mitigate risk – in this example, the ideal is that all the Jim-Bobs get full insurance at between $60 and $66, and all the Fatties get insurance at between $120 and $132. That way, they all pay the freight on their own knowns (the non-risk element) while getting effective insurance for the risk variables. Insurance isn’t intended to spread cost from high-risk to low-risk people, it is intended to spread cost over pools of similar individuals. It may be that YOU (the general reader, not any individual commenter or poster) want to do that. In that case, argue that – convince us, the mostly low risk population, that we should “carry” the high risk population. You can’t do it with economics. (BTW, untaxed employer provided insurance does just that, albeit inefficiently. That is part of why our current system is so costly – but the fix is to count employer-provided insurance as income to the worker, not have the government take over as the uberemployer.)
Steven Joyce 09.23.04 at 4:53 pm
This argument is similar to the standard examples of the possible non-existence of pure strategy equilibria in a competitive screening model. However, I vaguely recall that this result was non-robust in a couple of important ways. I looked up the details in my old micro textbook, and I think they’re pretty damning to your argument for government intervention in health insurance.
First, mixed strategy equilibria exist (Dasgupta and Maskin, REStud 1986). You might object to mixed strategies in this context, since in a truly dynamic model, firms would be able to withdraw offers that are ex post unprofitable. When no pure strategy equilibria exists, some offers made under a mixed strategy equilibrium will be ex post unprofitable and will be withdrawn, breaking the equilibrium.
But if you move to dynamic model like this, your argument develops an even worse hole. In a screening model where 1) the agents can withdraw unprofitable offers in response to a deviation (i.e., the solution concept is a Wilson (JET 1977) equilibrium) and 2) firms can offer a menu of insurance contracts, a pure strategy equilibrium necessarily exists, and IS CONSTRAINED PARETO OPTIMAL (Miyazaki, JET 1977).
Since I think the Miyazaki setup corresponds fairly closely to the way health insurance markets actually work, this would suggest that government intervention cannot improve upon the market outcome (unless, of course, the government has access to information that the private sector does not). Of course, there are probably other arguments for government intervention that do not suffer from this flaw, but I don’t think this class of models can do the necessary work.
Steven Joyce 09.23.04 at 4:54 pm
This argument is similar to the standard examples of the possible non-existence of pure strategy equilibria in a competitive screening model. However, I vaguely recall that this result was non-robust in a couple of important ways. I looked up the details in my old micro textbook, and I think they’re pretty damning to your argument for government intervention in health insurance.
First, mixed strategy equilibria exist (Dasgupta and Maskin, REStud 1986). You might object to mixed strategies in this context, since in a truly dynamic model, firms would be able to withdraw offers that are ex post unprofitable. When no pure strategy equilibria exists, some offers made under a mixed strategy equilibrium will be ex post unprofitable and will be withdrawn, breaking the equilibrium.
But if you move to dynamic model like this, your argument develops an even worse hole. In a screening model where 1) the agents can withdraw unprofitable offers in response to a deviation (i.e., the solution concept is a Wilson (JET 1977) equilibrium) and 2) firms can offer a menu of insurance contracts, a pure strategy equilibrium necessarily exists, and IS CONSTRAINED PARETO OPTIMAL (Miyazaki, JET 1977).
Since I think the Miyazaki setup corresponds fairly closely to the way health insurance markets actually work, this would suggest that government intervention cannot improve upon the market outcome (unless, of course, the government has access to information that the private sector does not). Of course, there are probably other arguments for government intervention that do not suffer from this flaw, but I don’t think this class of models can do the necessary work.
Steven Joyce 09.23.04 at 4:56 pm
This argument is similar to the standard examples of the possible non-existence of pure strategy equilibria in a competitive screening model. However, I vaguely recall that this result was non-robust in a couple of important ways. I looked up the details in my old micro textbook, and I think they’re pretty damning to your argument for government intervention in health insurance.
First, mixed strategy equilibria exist (Dasgupta and Maskin, REStud 1986). You might object to mixed strategies in this context, since in a truly dynamic model, firms would be able to withdraw offers that are ex post unprofitable. When no pure strategy equilibria exists, some offers made under a mixed strategy equilibrium will be ex post unprofitable and will be withdrawn, breaking the equilibrium.
But if you move to dynamic model like this, your argument develops an even worse hole. In a screening model where 1) the agents can withdraw unprofitable offers in response to a deviation (i.e., the solution concept is a Wilson (JET 1977) equilibrium) and 2) firms can offer a menu of insurance contracts, a pure strategy equilibrium necessarily exists, and IS CONSTRAINED PARETO OPTIMAL (Miyazaki, JET 1977).
Since I think the Miyazaki setup corresponds fairly closely to the way health insurance markets actually work, this would suggest that government intervention cannot improve upon the market outcome (unless, of course, the government has access to information that the private sector does not). Of course, there are probably other arguments for government intervention that do not suffer from this flaw, but I don’t think this class of models can do the necessary work.
Steven Joyce 09.23.04 at 5:00 pm
Sorry for the multiple posts — I got a message about an internal server error, so I hit POST again.
dsquared 09.23.04 at 5:16 pm
Steven; but the mixed strategy or dynamic game theory equilibria are only “equilibria” in the attenuated sense in which game theorists use the term. They both describe the situation I’ve described where any contract can be dominated by another contract. This isn’t an “equilibrium” in the classic sense of the word in being a stable business environment. In particular, I very much doubt that it would be possible or practical to carry on the business of insurance on the basis of the Miyazaki setup (although I freely admit that my game theory chops are just pathetic, so I may be wrong). My guess is that this is why, as I suggest above, most real insurance markets work on the basis of the pooling equilibrium (enforced by social norms among actuaries), because this is at least safe.
Jason McCullough 09.23.04 at 7:20 pm
“First, in order for your doctor to make an extra $10.00 he has to prescribe you about an extra $100 in services. You can see that easily in post motorvehicle Chiropractic care.”
Citing the behavior of a shady, non-scientific corner of medicine isn’t quite the best way to make a general point.
“Looking at this from a macro perspective, the first thing that hits me about this analysis is that universal health care would remove an incentive for private savings, just like Social Security did.”
Virtually everyone needs to retire; only a minority of the population needs significant healthcare outlays. And the minority that needs significant outlays couldn’t save for them if they wanted; they’re too big for individuals to cover, and to some extent random inciidence. Which is why insurance exists.
Going off topic for a second: what’s the evidence that Social Security reduced the savings rate? I know the theoretical disincentive to save argument, but that’s basically pocket change at worst.
Dubious 09.23.04 at 7:21 pm
Ethesis —
I’m not sure that doctors low profit margins explain high rates of high medical expenditures growth. HMOs don’t have this incentive, and though there was some initial success on the part of HMOs in holding down expenditures (arguably due to cutting excess capacity and market power in bargaining for inputs), HMO expenditures are starting to grow at historicly high rates.
(For me is this the most interesting part of the problem, why medical expenditures are growing so much faster than inflation.)
If fee-for-service doctors were economically rational, they would over-order tests more if profit margins were high than low, yes, since the incentive for over-ordering tests would be larger? Your model implicitly assumes some sort of target income model. Profit maximization or is the accepted default model for business behavior. Can you offer any evidence that doctors operate on a target-income model?
It may be plausible that doctors and HMOs over-order tests to cover their asses from fear of being sued. I suspect some of that does occur. But I think it is probably more common that over-testing (testing that would not be cost-justified in terms of expected Disability-Adjusted Life Years saved) comes from simple fear of making a mistake and being morally responsible for providing bad care.
I think the ultimate problem is that health insurance isn’t like other insurance. Other insurance usually pays off a relatively fixed amount when there’s a discrete event (fire, death, auto accident). Health insurance, on the other hand, promises a very flexible benefit for an ongoing, non-discrete process. That being so, the consumer makes decisions after the event happens, and the consumer doesn’t take the whole social cost into account.
If health insurance paid a fixed amount for a diagnosis (Stage 2 bilateral breast cancer? $20,000. Type II Diabetes? $2,000 per year.) and then let the consumer spend or not spend the cash in whatever way they preferred, this problem would be avoided.
Medical Savings Accounts combined with deep but high-deductible insurance (say, no insurance payments for the first $3,000 per year) would help get around that problem as well.
BigMacAttack 09.23.04 at 7:40 pm
Jason McCullough,
A trillion dollars a year is pocket change?
I mean I know you have to add a few billion from a whole bunch of places before we are talking about real money but a trillion dollars as pocket change?
abb1 09.23.04 at 9:15 pm
This is all well and good, but empirically: what’s wrong with French, German or any Scandinavian healthcare systems? They seem to like it, it covers everybody and they live longer.
Only idiots would learn on their own mistakes when other more advanced societies already went thru trial and error and so much empirical data is available.
Jack 09.23.04 at 9:31 pm
abb1 : they even spend less on it.
bigmacattack: Do you really think that Jason thinks a billion dollars a year is pocket change? Do you think that the social security disincentive to save is a trillion dollars a year? What are you saying?
dubious: Policies that pay fixed amounts on diagnosis of particular diseases are available and called critical illness cover. They can be useful but are not much good as health insurance because they don’t necessarily pay for your helathcare.
Nicholas Weininger 09.23.04 at 9:56 pm
An aside: it strikes me that an Austrian economist might object to your analysis on the grounds that (simplifying here) there is no such thing as an equilibrium in the real world anyway, nor is one necessary for viable service provision to exist.
But I don’t know enough Austrian econ theory to explain or defend this properly; I’m speaking from half-knowledge. Any actual Austrians, or at least people who have made it all the way through _Human Action_ rather than leaving it half-read on the bedside table, care to provide enlightenment?
Dubious 09.23.04 at 11:33 pm
Yes, I realize such policies do exist, being most frequent for cancer as I understand it. As I understand it, they are also not very popular. I more meant the suggestion as a theoretical rather than practical way to circumvent the moral hazard problem. More of a remark than an endorsement.
I think it’s highly probably that a single-payer system would, in the short run, produce lower mortality (or fewer DALY lost) per dollar than the current US system.
I wonder though whether it wouldn’t stifle innovation. It is hard to imagine a National Health Service, under constant political watchdogging being as risk-taking and innovative as the current US decentralized and entrepreneurial health system.
To open a new can of worms, because it would have monopsonist power in the drug and equipment markets, it would stifle innovations in those two areas as well.
james 09.24.04 at 12:17 am
How much is the cost of French, German or any Scandinavian healthcare systems offset by the inefficiency in the US system. More specifically, the fact that prescription medicine is regulated in these countries and not in the US. In a purely free market system, it is extremely unlikely that the same drug would cost 100 dollars in the US and 20 dollars in Canada (France, etc). Doesn’t this in effect demonstrate an indirect subsidy of these national healthcare systems by the US?
Ethesis 09.24.04 at 1:12 am
It may be plausible that doctors and HMOs over-order tests to cover their actually there are statistics that indicate that there is no significant correlation between risks and expenditures.
Guess I need to give a better example, one drawn from a file I looked at today.
Student visits student health center. Charge $40.00.
Then goes to E.R.
Basic visit would be $200.00. To get up to acute level 4 and $600.00 over six and a half thousand dollars in tests had to be ordered.
Whereupon the student was prescribed three dollars worth of medicine and did not come back for another health care visit for two weeks.
All the tests were negative.
It is interesting to watch the forces and counterforces.
So, assuming one finds Chiropractors shady (which I’m not sure follows), this is a non-chiropractic example with the same principles.
Usually the examples aren’t as pronounced, but there is a steady stream of points of this sort that are constantly hit rather than missed.
And not that all medical providers get the same amount of money.
Ortho surgeons seem to average 2-3 million a year once they are out in practice. Pediatric doctors seem to average about ninety thousand. That is a wide range.
I’m on a board that hires Pediatrics docs for a children’s clinics and I take depositions of orthos from time to time.
As far as I can tell, most doctors have no target income beyond “more.” Like lawyers and others, they see income as a marker in a competitive game that they’d like to win, and winning means more than the other guy.
Not that they aren’t fine people (many are better than just fine), but after years of competition, markers become important.
Jason McCullough 09.24.04 at 1:33 am
Ethesis, the way you’ve written that story (lots of tests, discharged with cheap prescription, doesn’t come back, but no details about tests, treatment, illness, or possible illnesses) it’s impossible to tell if the expenditures were reasonable checks or gold-plating.
“How much is the cost of French, German or any Scandinavian healthcare systems offset by the inefficiency in the US system. More specifically, the fact that prescription medicine is regulated in these countries and not in the US. In a purely free market system, it is extremely unlikely that the same drug would cost 100 dollars in the US and 20 dollars in Canada (France, etc). Doesn’t this in effect demonstrate an indirect subsidy of these national healthcare systems by the US?”
It demonstrates that drug manufacturers will charge the maximum the market will bear in individual markets; it’s less in some countries because they have different willingness to pay (nothing new here), they have a single very large buyer in national insurance countries (clearly) and could break patents (theoretical, and I’m not aware of any measurements of the potential effect of this.) “Drugs cost more in the US than these national system countries” can be entirely explained by those factors without claiming evidence of subsidy.
People keep mentioning “the hidden US subsidies to the rest of the world on health care,” (Megan McArdle foremost), but I’ve never seen a argument with numbers attached that supports this. It just kind of gets thrown out there as a vaguely plausible chain of reasoning.
BigMactAttack 09.24.04 at 3:52 am
james,
Ultimately I don’t know that I really believe in the idea of savings. There are only resources directed here and resources directed there.
But if I accept the more traditional definition of savings SS is a huge disincentive to save in the range of hundreds of billions of dollars.
The government collects a trillion or so dollars in SS taxes which it spends and promises to future generations. If it did not collect that trillion dollars in taxes presumably some fairly large fraction of that amount would be saved in the traditional sense. Since minus the government guarantee most people would feel the need to invest in traditional, private savings.
Ethesis 09.24.04 at 4:29 am
“the hidden US subsidies to the rest of the world on health care,â€
Someone needs to post a short essay on monopoly pricing at home and not away and how that works with a cost curve and competition.
*
**
****************
So if your first unit costs three dollars, your second two dollars and your third and subsequent units a dollar each, a home market that lets you sell the first three units for two dollars each will let you sell the rest at a dollar each in the abroad market and undercut competition that has the same cost structure, but no monopoly pocket.
Interesting stuff.
And, of course, I may be wrong.
Ethesis 09.24.04 at 4:31 am
Getting back on target, the COBRA adjustments and the death spirals of pools with too many Fatties in them, that is a clearly observed and mature fact in the market in the U.S.
james 09.24.04 at 5:18 am
How likely is the “willness to pay” to be a factor in the purchase of prescription medicine? The effect of a single large buyer does not seem plausable as a reason for the great differnce in Canada’s lower costs. Canada’s population is around 32.5 million. The two largest US health providers have 16.5 million ( UnitedHealth Group) and 15.6 million (Aetna Inc.) members. The fact that Canada and other nations employ regulatory measures to price fix is still the most likely cause of the pricing difference.
Jack 09.24.04 at 8:21 am
James, since the largest part of the expenses of big pharma companies is marketing (at least double R&D and rising with each merger) with manufacturing a distant third and since those foreignmarkets still add to the pot in the sense that if the US separated its drug market from the rest of the world there would be less spent on R&D I don’t think that claim is all that relevant.
What if the US outlawed direct to consumer advertising? What would teh effect be on
1)health outcomes?
2)drug company marketing budgets?
3)cost of treatment
Ethesis 09.24.04 at 1:49 pm
Posted by Jason McCullough
Oh, file facts … about three hundred dollars property damage in a low velocity automobile collision, walk-in after being at the student health center, about 6,600 in CT scans, three dollars and change for meds, level 4 critical care charge.
//////////
Another example being discussed on another board is ultrasounds in pregnancy.
Studies have shown them to result in a reduction in the quality of diagnostic treatment of pregnant ladies.
Many practices are moving to two ultrasounds per pregnancy rather than discarding them.
Anyway, been interesting to read this thread, and to discuss insurance rates, underwriting and policies.
Bottom line.
Companies do underwrite when given the chance (compare individual policies to group policies) and that results in diabetics, for example, facing very significant difficulties in getting life and health coverage.
Insuance pools with too many Fatties in them tend to go into death spirals.
Much of insurance is the Fatties of the world trying to spread the cost of their risk on other people, often after they know that they are Fatties. Consider infertility treatment. There are demands that insurance cover it, without opt outs. Always by people after they discover they need it.
Much of health care is for noncompliant diabetics in the end stage. Maybe as much as 15% to 20% of the public health funds spent in the U.S. Big difference in England (at least) is that dialysis is discouraged for diabetics over sixty.
Way beyond the scope of my ability to suggest solutions, just a point.
Antoni Jaume 09.24.04 at 5:50 pm
rvman wrote that “Looking at this from a macro perspective, the first thing that hits me about this analysis is that universal health care would remove an incentive for private savings, just like Social Security did. One of the problems people have screamed about has been our low national savings rate – universal health care only makes that worse – especially since government would undoubtably run additional deficits to “pay†for it.[…]”
However I think that most EU countries have a greater saving rate that the USA and mostly public health care, etc. So that looks a strawman to me.
DSW
james 09.24.04 at 7:52 pm
In the US, raw savings (savings account) has a lower return compared with other types of investment (stock market, property). In addition, the taxes on the return from savings investment is not as favorable as the taxes on the return from other investment.
Jason McCullough 09.24.04 at 9:02 pm
Do you a source on the noncompliant diabetic costs? Google isn’t giving me much.
james 09.24.04 at 9:24 pm
jack – Can the US outlaw Pharm marketing? Its a good idea but is it constitutional?
Comments on this entry are closed.