Interview with Mark Pauly: Part 1

Community Rating – The Worst Possible Way To Do a Good Thing

I have a grudging respect for health economists, “grudging” because, like many doctors, I want my pieties unchecked. Health economists check our pieties with quantitative truths. They describe the way the healthcare world is – a view from 29, 000 feet, pour cold water on the way we think the world should be, and guide, with abundant disclaimers, the way we can make things better. It’s unwise climbing Everest without a Sherpa, nor is it wise reforming healthcare without listening to health economists from across the political spectrum.

President Trump, along with the Republican House and Senate, will be dismantling the Affordable Care Act (ACA). In a sense, President Trump is not just descending Everest, a treacherous feat in its own right, but scaling a peak arguably more dangerous than Everest. Despite their differences, Mr. Obama and Mr. Trump share one commonality – an implicit distrust of the health insurance industry.

How did the American health insurance industry become so vilified? This is, in part, because necessity is the father of all vilification. Insurers are a necessary evil in a country where there’s still deep mistrust of the government. Partly, this is because we transfer our angst about the uncertainty of our future, the dice which plays with our lives, to insurers who are in the business of rolling the dice. But mostly it’s because the misdeeds of the insurance market have been grossly exaggerated, and the benefits of the market have been attenuated by a few damning anecdotes. This is what Mark V. Pauly (MVP), Professor of Health Economics at the University of Pennsylvania, and one of the most eminent health economists of his generation, believes.

Informed by his research, Pauly is a rare defender of the insurance industry in academia. I recall attending a class by Pauly a few years ago. Two things struck me about the soft-spoken economist. The first was his reluctance to prescribe solutions without explaining the tradeoffs. Pauly is unlikely to appeal to demagogues or policy makers seeking inexpensive balms for complex problems. The second was his penchant for stating the counterintuitive. I met Pauly on a rainy day in Philadelphia in the Library Bar at the Inn at Penn. I remembered from my class how painful ignorance can be when speaking to Pauly. So, I read up as much as I could during the New Year’s weekend. It didn’t help much. I was still beseeched by facts unbeknownst to me.

The Interview

SJ: The insurance industry is demonized by everyone, particularly doctors. You have a more favorable view of the industry. You believe that it can be a force for good if appropriately managed. I want to get to the heart of the distrust. Is it because we haven’t gotten over Akerloff’s “The Market for Lemons?

MVP: Let’s back up. The lemons problem, described by George Akerloff, is related to the used car market because of asymmetric information. The sellers know more about the quality of the car than the buyers, and because of that the buyers will underprice the high quality cars, which the sellers will remove because they’re underpriced, and the market will be full of low quality cars. The analogy with healthcare is partial. Buyers of health insurance do know more about their health status than sellers – to be clear, this is relevant in the individual market (IM), not employer sponsored insurance (ESI). Regardless, asymmetric information is not unique to healthcare. For example, drivers know much more about their driving abilities than sellers of car insurance. The insurance for automobile collision hasn’t gone into a death spiral.

However, the information can be symmetric. The sellers of health insurance can roughly risk rate the buyers by asking questions about their health, recent visits to doctors, family history and tobacco and alcohol use. What historically led to adverse selection – the sick disproportionately purchasing health insurance – wasn’t the lemons problem per se but that insurers weren’t allowed to risk rate.

SJ: That’s an early counterintuitive fact. What you’re saying is that community rating – charging everyone the same premium regardless of their risk is what led to adverse selection. If I may restate, community rating made it more likely that sick, rather than healthy people, purchased insurance. Or to put it bluntly, the prescription to prevent adverse selection caused adverse selection. This was a policy own goal. I must admit it doesn’t make sense.

MVP: The reason it doesn’t make sense is because we forget there are tradeoffs. So, when you prevent the insurer from risk rating what do they do?

SJ: They raise the premiums for everyone to compensate.

MVP: What effect does that have?

SJ: It makes insurance more expensive for the healthy.

MVP: Sure, but what effect does that have?

SJ: The healthy are less likely to buy insurance.

MVP: Right. And not just that, people with co-morbidities are more likely to buy insurance. You can see that community rating entices the high-risk and discourages the low-risk. What’s that called?

SJ: Adverse selection.

MVP: And in the IM, it’s more significant than people think. We found that when insurers changed to community rating, while the probability that the high-risk were covered increased, it was overwhelmed by the probability that the low-risk, the healthy, didn’t buy insurance. This is not good for the IM. Another example I have – one of my PhD students has studied the Medigap insurance. She found that adverse selection is greater when there’s community rating than risk rating.

Insurers are good at risk rating. The only thing they miss is when a couple is planning to start a family. As you can imagine, the intent for pregnancy can’t be captured in a questionnaire – you can plausibly deny it. Aside from that, insurers can adequately risk rate – so the lemons problem in healthcare is contrived.

SJ: I can see how one can be poetic with the truth with pregnancy. But this does mean that you endorse risk rating. This means that the sick pay more, a lot more. My residual liberal sentiments can’t abide that. What’s your prescription so that the high-risk aren’t financially fleeced?

MVP: If insurers are permitted to risk rate, the premiums for the healthy will fall and more healthy people will voluntarily buy insurance – let’s agree that’s a healthy outcome. The premiums for the high-risk will surely rise. My proposal– let the insurers risk rate but let us subsidize the premiums for the high-risk.

Money must come from somewhere. What is happening now is that the low-risk are subsidizing the high-risk. Make no mistake – this is a transfer of wealth, but not a very equitable transfer of wealth. In essence, community rating is an excise tax on the low-risk.

SJ: You just used the “T” word which, politically, is more offensive than the “F” word. It seems that you don’t want the high-risk to be thrown under the bus. What you’re suggesting is that the high-risk should be subsidized explicitly by general taxation.

MVP: Taxation is a political problem. Politicians favor hidden taxes over explicit taxes – excise tax on the low-risk, imposed by community rating, is not counted as tax and does not show up on the budget. Even the low-risks are likely to blame the insurers, not the regulators, for the high premiums. The high-risk should be helped by general taxes. The market can’t function without taxation of some form.

SJ: A related question – how significant is the problem of high-risk purchasing insurance in the individual market?

MVP: The numbers quoted about the prevalence of high-risk vary and some say it’s as low as 1 %, but I believe a more correct number is 4 %.

SJ: Many believe that the problem of pre-existing conditions is much higher – around 30 %.

MVP: Most people can’t be “high-risk”, most must be average or below-average risk, unless the country is a perverse Lake Wobegon, where everyone is above-average. This is common sense. In fact, the distribution of risk is skewed to the right, meaning that the average is driven by relatively few high-risks. The probability that someone becomes high-risk precisely when they buy insurance should be even lower than the prevalence of the high-risk – this is simple probability. Yes, we are still talking about many thousands of people and we must help these people, but you do not need drastic changes in the market and heavy regulations to help this small group. Specialized interventions targeted at the small minority of high-risk would be more efficient, more transparent and more feasible.

SJ: Does what you’ve just said not contradict the 80:20 rule – that 80 % of healthcare costs are incurred by 20 % of people? Surely, the high-risk must be at least 20 %.

MVP: Ok, I can see the confusion.

Let’s first define “risk” in insurance parlance. Risk is the expected medical spending. High-risk have a higher expected medical spending and the low-risk have a lower expected medical spending.

The confusion lies in the definition of high-risk. By high-risk, I do not mean low-risk people who have a medical event and then become a higher risk. I mean people who are high-risk when they first purchase insurance – the former is far more common than the latter and is often mistaken for the latter.

Though there are some real cases where insurers misbehave with the low-risk who become high-risk, by dropping coverage, this problem has been exaggerated and can be easily mitigated by guaranteed renewability, which was a common feature in the IM even before it became mandated by regulations.

Thus, the common misunderstanding leads to confusion about the 80-20 rule which is that 80 % of healthcare spending is on 20 % of the population. We confuse this for the magnitude of the problem of the high-risk. Rather, this number refers to the spending distribution of the low-risk who become high-risk in their lifetime. It’s important to maintain this distinction. The major purpose of insurance is financial protection for uncertain events – i.e. the low-risk turning to high-risk. Insurance achieves this by pooling risk, and pooling risk means pooling the people who have similar risks of an uncertain future. Pooling risk pools the unknown. Pooling risk does not work if you pool people with a known future (high-risk) with people with an uncertain future (low-risk).

To emphasize, there are truly more low-risks and fewer high-risks than suggested by the 80-20 rule. This rule merely tells you that even if you start off with a group of people with same (low) risk, the future spending is unevenly distributed.

SJ: This is a very nuanced distinction and I can see how unfettered zeal can conflate the two groups of risk. If I may summarize. What you’re saying is that the market should be allowed to work with risk – the known unknowns (to borrow a Rumsfeldian aphorism) – and the government should help with certainty, the high risk, the known knowns.

I’m going to bring up another point you mentioned in our class, something which stuck with me – see I was paying attention! You said that one of the reasons entities don’t discriminate, and I think you were quoting Milton Friedman, is that the costs of obtaining information to discriminate are so prohibitive that the benefits of discrimination aren’t worth the transaction costs of getting the information to discriminate. Surely, the insurers can’t actuarially price (perfectly risk rate) the high-risk even if they wanted to.

MVP: No they can’t perfectly risk rate, that’s correct. I did a study with David Asch where we asked insurers if they would use genetic information if available. Most said they won’t. Why do you think that is?

SJ: Well, it’s not from the goodness of their hearts. I suspect the added information has diminishing returns.

MVP: That’s certainly one reason. Insurers don’t feel that genetic information adds much more to family history, and it’s easy getting family history. The other reason is that insurers feel that if they mandated genetic testing, people would be even less inclined to buy insurance. Remember, insurers need people buying insurance to survive.

SJ: So, in the individual market there’s a tradeoff between healthy voluntarily buying insurance and community rating?

MVP: Yes, there’s a tradeoff – community rating in the individual market increased premiums making it less likely that healthy purchased insurance, and more likely that the high-risk bought insurance, but the net effect, and this has been empirically shown repeatedly, is that the net uninsured increased.

Once you started community ratings you ran into other problems. Insurers avoided or underserved high-risks.  Insurers tried gaming the system with clever insurance design and regulators responded with more regulations. Rules were put in place to risk-adjust revenues insurers collected. Risk adjustments are imperfect. One bad policy led to several regulations.

SJ: This sounds like some game theory going on, like a game of chicken, or something, between insurers and regulators. And we threw bad regulations to chase bad regulations.

You once said that blaming the individual market for high premiums is like blaming Cinderella for poor fashion. Two questions. Why, other than community rating, is the premium so high in the individual market? And does that mean ESI is one of the ugly stepsisters?

MVP: The premiums are high for many reasons. The administrative costs in the individual market are high. I did the calculations once. The administrative costs, as a percentage of claims, are 27 % for group size of 1-20, 12 % for group size of 100-500 and 4-6 % for group size exceeding 10, 000. This is partly because the selling costs for insurance in the individual market are high.

But taking the Cinderella analogy further, the ESI is like a spoilt stepsister because of the tax subsidies. My wife and I receive a tax subsidy of $10, 000, and my administrative assistant, in a much lower income bracket, doesn’t have the same luxury. This is moral inversion. The tax subsidy for ESI is regressive taxation. The individual market can’t compete with that.

Although politically unpopular, the tax subsidies of the ESI should be abolished to level the playing field. I think it’ll make some people choose insurance from the individual market, but overall I don’t think the ESI will be destabilized.

SJ: Continuing with the Cinderella analogy, Cinderella had many redeeming qualities (not related to fashion). Does the individual market have redeeming qualities?

MVP: First, I must emphasize that the markets for insurance for non-poor are easy to mess up.

The individual market can have many redeeming features. For one, there is choice about the level of coverage, what’s covered and the amount of deductible one has to pay. Not everyone has the same preferences for these. Once the insurer set a premium, as long as the premiums were paid regularly, the insurer didn’t increase the premium just because the risk profile changed because of a new disorder. This feature is known as “guaranteed renewability.”

SJ: Can you expand on the second point?

MVP: “Risk reclassification” is when the low-risk have an unexpected medical event changing their risk profile. What then happens is that the insurer drops coverage or raises their premium. The risk of risk reclassification is when there is annual risk-rating of health insurance. Fortunately, most insurers in the individual market before the ACA avoided this problem by “guaranteed renewability” where the insurer didn’t raise the premiums of people who started and remained with the individual market so long, as they kept paying their premiums.

Paradoxically, those at-risk of risk reclassification were people who lost their employer-based coverage and moved to the individual market as high-risks. Even more paradoxically, it was the insurers in the individual market who were blamed for not picking up the slack, rather than the job-based insurers for creating the slack.

The knee jerk policy reaction to risk-reclassification has been community rating which, as I’ve alluded to, is the worst possible way to do a good thing. A better policy prescription, which I believe is in the Ryan plan, is to allow people who leave the employer market to the individual market, to carry their premiums with them.

About the Author

Saurabh Jha is a radiologist and contributing editor to THCB. He knows far less about economics than he thinks he knows, and he thinks he knows far too little. He can be reached on Twitter @RogueRad












Categories: Uncategorized

18 replies »

  1. Cap at 8.5% will not work and is inconsistent with what MVP is saying. Just because I may earn $400k does not cause me to voluntarily pay $34,000 for a product worth $5,000

  2. We “opted out” of ACA. Financially, we are better of being uninsured than purchasing the ACA policy, which costs $25,000 annually and has a $13,500 deductible. I assert:
    A healthy unsubsidized user is generally better of financially being uninsured.
    We were insured our entire lives, until 2014, when our policy increased by 75% to $14,000. After going through the “we must have insurance” re-education regarding “medical insurance”, we figured out that there are sufficient alternative in the marketplace so that we can sufficiently mitigate our risk and live without “medical insurance”. (In quotes because ACA policies aren’t really insurance.)
    @MVP We are precisely your example. I think it is likely that with guarantee issue, community rating, and mandated essential benefits, we will never again purchase “medical insurance”. And, I think the same is probably true with community rating and guaranteed issue, even if the mandated essential benefits are removed. Community rating and guaranteed issue is an ignorant pipe dream–It will not work. The healthy will always opt-out. We will always have better alternative in the marketplace.

  3. “no one at any income level has to pay more than 8.5% for health insurance ”

    You mean a person living in a $1,000,000 home and has tax free bonds as his only source of income should be subsidized by a working family?

    “cost for high risk persons in the individual market should be spread more over all citizens,”

    Hayek might say that is OK to an extent as long as it doesn’t significantly interfere with the free marketplace (the least interference).

  4. There is more and more consensus on Pauly’s core concept here — namely, that the cost for high risk persons in the individual market should be spread more over all citizens, rather than driven onto only the other buyers of individual insurance.

    Here are some approximate numbers to consider:
    1. Assume 20 million buyers in the individual market (both on and off exchange, not counting Medicaid.) this is high but a nice round number to work with.

    2. Assume that 5% of this total is very sick, either episodically or every year, and runs up a bill of $50,000 apiece on average. $50 billion in total on these persons.

    3. Assume that these sick persons pay $5,000 a year in premiums with community rating.

    4. Thus, $45 billion has to be hammered into the premiums of the other 19 million persons.
    This is $2500 apiece on average.

    How else could we do this?

    a. One option is put the sickest persons onto Medicare. Yes this would bring $45-$50 billion of extra costs into Medicare (probably a little less, since Medicare pays less than most insurers.) But we could raise the Medicare payroll tax about six tenths of one percent to cover this. For a person making $50,000 a year, this raise is $300 a year and their employer would pay half of the raise. Leaves the worker with an extra outlay of $150 a year, $12,50 a month, not back breaking (the wonders of broad based taxes!)

    b. Another option is do what Bloomberg and Holohan of the Urban Institute have proposed — namely, be sure that no one at any income level has to pay more than 8.5% for health insurance (instead of the horrible cliffs and cutoffs we have now). Then, let the insurers charge more to bad risks. Let the insurer charge $3,000 a month to someone with Crohn’s or CHF. The person will only pay 8.5% of income. The money we spend on subsidies will go up, and we should have an increase in income tax rates to pay for this.

    I think that option A might be easier politically, but both will be hard to do in today’s Congress.

  5. I’ve always thought the solution for the individual insurance market (pre-ACA) was for Wharton Professors (and members of congress too) to have to buy their insurance in it. Pauly once made a statement that individual insurance worked fine other than for 20% of the people in it — he didn’t point out that they were the ones who needed it….

  6. Great article. All the “blending” of community rating, risk-shifting, etc is just a political game of playing “pass the hot potato” without actually addressing the reality of really sick people and really expensive care options being selected.

    Taking away all the artifice and actually focusing on the real risk that exists (and finding ways to make real-world differences in those risks at the level of the individual) and the real-world options for delivering said difference is what we as a profession and society are looking for.

    There are many debates on what works and almost nobody using the data to understand this in the messy real world (billing codes and intermediate test results are not real world outcomes – when I was working with a large insurer, they didn’t realized they didn’t track death which you’d think would be an important clinical output vs cure).

    I’ve been against the design of the ACA, as it took away market-driven experiments, ability to use non-commodity pricing (via a claims set driven by the RVU model) and consolidated innovation in the entities least likely to be scrappy and disruptive (govt regulators and large hospital systems). Pauly points to a legitimate alternative forward. It’s a good discussion.

  7. Can’t make money that way Margalit.

    It’s interesting to see this “high” risk split off from the profitable “low” risk. You take it, no, you take it, no you. The risk is among us, it’s our friends and our neighbors, it’s us or it will be us at some point. Look in the mirror, some day YOU will be “high” risk. So be careful where you transfer costs.

    The way to manage risk is to include EVERYBODY in the pool. No options, you’re covered and pay premiums, young, old, healthy, sick, through taxes.

    So we’ll see if Repugs can bring us to the promised land where we will pay less for more – or at least the same, against the actuarial tables and the provider costs.

  8. How much less in percentage terms did the 55-64 year old age group pay under your subsidized approach than they would have paid with full age rating? Just curious about how far you could go without pushing the young, healthy folks away..

  9. Saurabh: Well written and interesting. Too much packed in even this Part I to comment on in one comment, but a story:

    When Blue Cross RI was the only insurer in the individual market in RI (it might still be), it had the ability to adjust the actuarial age tables to artificially increase the younger subscribers and decrease the older. We knew it was a subsidy, and we neither hid it nor trumpeted it. We told the regulator in our written submission that the usual slope was decreased precisely because the older subscribers between 55-65 needed some help at a time when many of them had lost jobs or retired early. The trick was, like Goldilocks, to do it “just right” so that it didn’t chase away the younger healthier subscribers. Did it work? Pretty well in my estimation. We also helped the older sicker subscribers by using single rate increase factors, which by their very nature favor the increasingly sicker population and not the stable healthy population. Surprise.

    Now, say United Healthcare comes into that market and uses full slope age rating. What happens? The younger subscribers will see lower rates and the older higher rates. Who do you think will leave BX and go to UHC? The market thus becomes destabilized.

    I know that the idea of a single insurer market is anathema to many, but in this case it allowed us so much leeway to actually try to do the right thing. [What? Insurers trying to do the right thing???] Just sayin’

  10. Keep in mind that HSA + HDHP + DPC are reducing costs 20-60% while improving actual outcomes

  11. “ Insurers are a necessary evil ”

    Insurers have a self interest. We all have a self interest, but that doesn’t necessarily make us evil. Lord Acton tells us how that evil is created “Power tends to corrupt and absolute power corrupts absolutely”

    I hope MVP taught you something about lemons. 🙂

    Great article. Thanks.

  12. I hope the above article explains your confusion about high risk patients and risk management along with the magnitude of high risk. It even tells you where the blame lies when there is risk reclassification in job-based insurers.

    You seem to magnify high risk costs that are not affordable to the patient ending up severely damaging those low risk persons that become high risk while insured.

  13. A while ago (over 20 years) I was on the Board of a Blue that operated with only 7% admin…..they have raised that now up to around 12%. They like to travel by company plane now and they like to fund all kinds of “do good” programs. And they have little or no competitive pressure to reduce the admin take because they have market share approaching 70%. Also, as Jim Purcell has noted….the politicians of the served cities and towns like to have jobs and rented offices….so the pressure on the insurance execs is NOT to cut expenses.

  14. You might be able to manage medical risk in another way. Instead of supporting insurers and sending wealth through this industry, we could eliminate insurance and rely on debt. I.e. we allow anyone who is sick to see any provider and that provider is, by law, required to see the patient as by a sort of universal EMTALA applying to all providers*….with the addition of a proviso that allows the provider to bill the patient after the service has been rendered. The provider thus becomes a short term lender of his services and is really losing the use of his accounts receivables for a much longer period of time. There would be an opportunity cost in these delayed accounts receivables which would have to be made up by some increase in the amount billed.

    *Except that dead beat patients who do not pay their bills within a certain number of years receive no further services. Repayment insurance might be needed here.

    The advantages of this idea would be that one would eliminate all the problems of a third party payer. Shopping could occur. Prices would be revealed. Providers would be reminded that without good care and patient satisfaction, debts would not be paid. Quality might be improved. Patients might realize that without payment of bills to the provider, services might cease.

    Many patients would still need subsidies. Lots of tricky problems remain.

  15. I think politics is absolutely the core of the problem. The idea of subsidizing people with known high risk is not rocket science. However, the per person cost of subsidizing them would be very high as would the aggregate cost assuming MVP’s estimate of the size of that population is accurate. It’s also better for the public if taxation is visible and transparent than if it’s hidden and obfuscated but for politicians, that’s a very tough sell.

    In the end, we get the government we deserve. If we the people rewarded politicians for doing the right thing, subsidizing high risk with transparent taxes in this instance, they would be more likely to do the right thing. Wouldn’t they?

  16. Maybe it will be in Part 2, but I have a question:
    What would be the administrative costs for a group size of say… I don’t know… 300 million or so?

  17. Seeing as experience rating with hi risk pools adequately funded from general tax revenues serves more as a theoretical ideal (I think that’s the message), is what we have now second best? I sense you find community rating unacceptable, period.

    Moreover, is the use of hidden over explicit taxation, an issue more rooted in politics than policy, the real barrier to getting where you want to go?