Archive for February, 2016

What is an efficient insurer?

February 20, 2016
I got an interesting email today that I responded to in a fair amount of depth because they were issues I have been thinking about lately anyway. This post is the questions raised and my email response to them:
From: Anthony Komaromi <>
Sent: Friday, February 19, 2016 12:47 AM
Subject: Administrative costs and efficiency
Dear Dr. Cox,
My background is political philosophy.  The biggest contemporary political issue for me is healthcare.  I support universal healthcare, though I have some questions about it.  I would greatly appreciate your expertise on some of my questions.

1.  Are administrative costs higher with public or private health insurance?  I have heard many people claim government health insurance programs (like Medicare and universal programs in Europe) have lower administrative costs.  But then I came into this ( from the Heritage Foundation, which claims Medicare’s administrative expenditures are much higher than private health insurance’s.  Are there any flaws in the way the Heritage Foundation (whose calculations are “based on Benjamin Zycher”) calculates administrative expenditures or are they calculating the way we should be?  How should administrative expenditures be calculated?

2.  What does it mean for an insurer (or type of insurance) to be efficient?  Does it mean not paying for services and refusing to cover people with expensive needs (like pre-existing conditions)?  What methods make one type of insurance more efficient than the other?

Thank you.


Anthony Komaromi

Hi Anthony,

It would obviously be better if I knew what your background was, how familiar you are with insurance, risk theory, economics and finance. But…. Also, a disclaimer. I think I have addressed all your questions by covering broader topics, but some quick thoughts before I go into the details:

You: ” I support universal healthcare, though I have some questions about it.  I would greatly appreciate your expertise on some of my questions.”

Me: I honestly don’t know what it means to support universal healthcare. Almost everyone pays lip service to universal health care. The question is always how will it be achieved? Some are willing to wait eons for imagined market forces to deliver the answer. So far this hasn’t worked anywhere. Others support a single payer, health insurer, hoping that this will result in universal healthcare. It will not. There will still be problems in access to care with a single payer because of the maldistribution of health care system capacities. Mere insurance will not bring health care to poor urban areas, rural areas and communities of color, race, orphan diseases…

You: “Are there any flaws in the way the Heritage Foundation (whose calculations are “based on Benjamin Zycher”) calculates administrative expenditures or are they calculating the way we should be?  How should administrative expenditures be calculated?”

Me: You cannot get much more flawed. You will see why below.

You: “What does it mean for an insurer (or type of insurance) to be efficient?  Does it mean not paying for services and refusing to cover people with expensive needs (like pre-existing conditions)?  What methods make one type of insurance more efficient than the other?”

There are obviously different types of insurance – I suggest that the aggregate of insurers (1 or many) that delivers the highest aggregate benefits for given premium revenues is the most efficient insurer. More on this below.

Point 1: I note that the link you gave has an interesting issue. They exclude payments made for dually insured people (Medicare + private insurance). Now why would they do that? I would suggest that the answer is politics. Not politics in the sense that Socrates might describe political discourse: How to manage the state from the perspective of “Truth”, but the more common approach of seeking consensual advantage independent of the truth.

So, when a Medicare beneficiary has dual coverage, the health providers routinely file against both insurances. Both Medicare and the private insurance process the claims, and in some cases only the private insurer will make payments, while in some case both Medicare and the private insurer will make payments. Regardless, the same amount of administrative work is undertaken by Medicare whether or not Medicare makes a payment. So the table is intentional obfuscation.

Probably not satisfying for you, but certainly not a surprise for someone as jaundiced as myself.

Point 2: We have hundreds of different insurers. Each of these insurers have sales departments, claims departments, legal departments, actuarial departments, underwriting departments, finance departments, accounting departments, administrative staff, executive teams and boards of directors. Now much of the work on claims is automated. Small claims are routinely paid because the costs of hand processing by people exceed the benefits to be paid. So the only claims that get a lot of human attention are medium to high value claims.

High value claims get different treatment in small vs large insurers. In small insurers all the department heads are informed about the management of high claims. For medium value claims the claims department head, underwriting, legal and financial department heads are usually involved in varying degrees depending on how large the claim is. In very large insurers, even fairly large claims get little scrutiny. Large insurers expect large claims, but it is rare that even the largest claims get a lot of attention.

All to suggest that the non-benefit expense ratios for insurers vary dramatically depending on the size of the insurer. As usual, the economic efficiencies of scale prevail though for different reasons than in a purely manufacturing environment.

Point 3: When I first started working in the actuarial department at Liberty Mutual Insurance back in 1982 I remember trying to engage the Exec VP for the department in a discussion of fraud. His response stunned me: “We don’t care about fraud”. But I soon learned why that was. Depending on the line of insurance and the regulatory environment, fraud simply increases the loss ratio. As loss ratios increased, the insurers sought rate increases. As the loss ratios increased the insurers got increased premiums and the assumed expenses went up as well even though the actual expenses didn’t rise at all. Fraud simply increased insurer revenues by more than the marginal increases in claims costs.

Point 4: Getting back to the hundreds of insurers. Is it really possible to imagine that all the duplicated departments in hundreds of insurers cost less than the costs of running the Medicare program? No. You really have to stretch credulity to imagine that hundreds of small insurers with duplicative administrative costs are operating, in the aggregate, more efficiently than a single payer could operate. So, why would anyone suggest that this was so? Oh, yes, insurers, in the aggregate, make a lot of money for investors.

But not all insurers make handsome profits. In fact, many insurers fail every year. When insurers fail, they fail to pay policyholder benefits. Small insurers are sort of like the Wild West. They have high probabilities of either excessively high profits or excessively high losses. Their management teams get high bonuses in excessively profitable years and lose their jobs in very bad years. But, even failing executive teams tend to bounce right back, forming additional insurers days, weeks or months after their companies failed. Failing insurers serve a very useful function. They convince a gullible public that insurance is a high risk enterprise and that the high premiums insurers get are necessary to compensate them and their investors for the risks they accept.

Of course this is utter rubbish. Large insurers face little risk. Only small and poorly managed insurers face financial risks. But large insurers bail out the policyholders of failed insurers. This is a win-win for large insurers. Good public relations. Minimal impact on their bottom lines. They often continue to sell insurers to the grateful claimants they served.

Point 5: This one is a little more involved than the other points and goes to the heart of what has been done to make Medicare be inefficient and appear to be inefficient. Medicare does not operate at all the way it is generally described. Medicare does not simply get claims and pay them, an entire, unnecessary, profit yielding industry of intermediaries has been put in place to make Medicare more business-like. These intermediaries are paid the amount Medicare would otherwise pay directly and if they can “settle” the claim for less by either denying the benefit or squeezing the provider, they make a profit on the transaction. This is nothing less than a total scam as these intermediaries don’t save any money at all, they just make Medicare less efficient.

As well, as I describe in my book, Medicare also transfers insurance risks to health care providers. This is an even more inefficient way to manage risks than small insurers because the providers are far less efficient risk managers than small insurers. Again here, the details are a bit more complicated than an email.

Point 6: Even if we allow unbridled competition, despite the inefficiency of increasing the numbers of small, competiing insurers, there will still be millions of people insurers do not want to insure: People who are already seriously ill, the poor elderly who are seriously ill, poor people who cannot afford insurance premiums, inmates in prisons and jails who are straining the budgets of local cities, counties and states.

So, we still will need the state Medicaid programs. So, when we consider a comparison of a single payer to a competitive market we need to remember that a not for profit, single payer would eliminate all the inefficiencies of our private markets with hundreds of inefficient insurers, it would eliminate all the state Medicaid program infrastructure, and it would eliminate the costs of regulating the private insurance industry at the state level.

All of this can be accomplished by merely changing the age of eligibility for Medicare from 65 to 0. Medicare already covers the aged and disabled, people with kidney failure, and severely ill dependents of beneficiaries, so the entire administrative structure to serve as a single payer is already in place. Every imaginable condition has already been addressed.

Point 7: And this is the big point. The real problem with small insurers is not even their inefficient non-expense ratios. It is their inefficient risk management. To avoid excessive losses, small insurers need to cut benefits below those anticipated in their premiums. I explain all this in my book and it is too complex to describe in an email. But the bottom line, the smaller the insurer, the more they have to cut anticipated benefits.

At the same time, large insurers can actually offer far higher benefits for the same premium than they do offer, because really large insurers are very efficient risk managers and face no risk of adverse operating results.

So the real inefficiencies in having competing insurers are in the management of risks and how efficiently insurers turn premiums into benefits.

Three different states are worthy of notice:

I. Small insurers earning excessive profits because their policyholders are healthier than anticipated. In this case the insurers and their investors convert premiums to profits rather than benefits. With a single payer these “profits” would go to benefits for other policyholders.

II. Small insurers incurring excessive losses because their policyholders are sicker than anticipated. In this case the insurers fail and their obligations to policyholders are covered by larger, more profitable insurers. With a single payer these “losses” would be covered by the gains on other policyholders who were healthier than expected.

III. Very large insurers, facing no risk, who could offer benefits that would be 10-20% higher than they offer on the premise that they actually face risks as documented by the failures of small, inefficient insurers.

A single payer eliminates all three of these inefficiencies by being able to offer higher benefits, at lower cost, with less risk, and lower administrative costs.

Probably more than you expected and I would be happy to respond to specific questions. The only thing I ask is that you not simply drop a flawed table from an organization that is seeking narrow political debate advantage at the expense of Truth as part of your question.




Notes on efficiency

February 17, 2016

This is a post I submitted on February 17, 2016 on Peggy Chinn’s blog after following her posts on efficiency and productivity

…. and so we come full circle on the theme that first attracted my attention… efficiency. I was tempted to write these thoughts on the first post, but I figured it would be better to let that run its course…

and when combined with your mention of this political season I could no longer resist, one of my favorite topics: The efficiency of risk management, or “… all you wanted to know about insurance but were afraid to ask.”

Recent issues in the Democratic Party primaries have brought one of my favorite efficiency topics into view. What makes for an efficient insurer? Ask most lay people and they are likely to give a knee-jerk response: An insurer with a low loss ratio. But no, that is not the hallmark of an efficient insurer. That is the hallmark of an insurer that isn’t paying its policyholders’ claims.

Perhaps an insurer with a low, non-claims, expense ratio. That is certainly an important component of an efficient insurer, but not the most important component.

The hallmark of a perfectly efficient insurer is an insurer whose loss ratio (Total Claims Costs/Total Earned Premiums) is exactly equal to the loss ratio for the population from which it randomly selects its policyholders. No insurer is ever this efficient. The best any real world insurer can ever do is come close to having the loss ratio for the population insured.

Still, thinking about insurer efficiency can reveal a lot about one of the most important issues in the 2016 election: What is the best way to provide health insurance for the entire population of the USA? Is an incremental improvement in the proportion of Americans covered through the PPACA, as suggested by candidate Clinton, the most efficient approach, or is a single payer, Medicare for All, as suggested by candidate Sanders, the most efficient approach?

First, the landscape. We hear a lot about some of our biggest insurers/health benefit plans: AETNA, Humana, United Healthcare, Kaiser Permanente, etc. We have also heard about some colossal insurer failures: AIG, Reliance Insurance Company etc.But the real issue of inefficiency has to do with the hundreds of smaller health insurers/health benefit plans we have. Most of these hundreds of insurers are terribly inefficient, often paying only a small portion of their premiums in the form of policyholder benefits, nowhere near the loss ratio appropriate for the population insured.

So, what is the magical ingredient in efficient insurance? Volume, Volume, Volume. The more policies an insurer issues, the closer its loss ratio falls to the expected loss ratio for the population insured. The importance of this proximity to the loss ratio for the population is critical. The more accurate an insurer’s loss ratio the easier it is to convert premiums into health care benefits. If an insurer knows exactly what its claims will be, year after year, it faces little or no risk of adverse operating results. The lower the insurers risk, the less need there is to reward investors with profits and the lower the “risk premium” it should have to charge for its risk management services. As these two items decrease, the insurer can convert more of its premiums to policyholder health benefits.

Now the thing about insurers, as I suggested above, is that the more policyholders, the more efficient the insurer becomes, the higher the benefits it can offer, the lower the premiums it can charge, and the lower the risk of bankruptcy.

For every population there is ever only one maximally efficient, risk managing, insurer. The largest insurer possible, or a national health insurer. Let’s imagine we have an “efficient enough” health insurer, insuring 1,000,000 Americans and converting 75% of its premiums to health benefits, having a profit goal of 5% of its premiums, charging a risk premium of 5% for its service as a risk manager, and having non-loss operating expenses of 15% of premiums. This insurer can reliably convert 75% of its premiums to policyholder benefits.

Year after year, in some years it will have lower losses than expected and it will earn higher profits than expected. In other years it will have higher losses than expected and it will incur operating losses, using up the 75% of its premiums earmarked for policyholder benefits, eating through its 5% risk premium and eliminating its expected profits. In really bad years it may lose all the money it has, it will become insolvent, shut its doors and deprive some policyholders of their benefits. This happens fairly rarely for insurers with 1,000,000 policyholders, but it happens quite often for insurers with 5-10,000 policyholders.Precisely because they are inefficient risk managers, small insurers earn excessive profits, or incur crippling losses, fairly frequently.

But we are interested in efficiency, not inefficiency. So we need to look the other way. While an insurer with 1,000,000 policyholders may be efficient enough, it isn’t going to be efficient enough for people like us who want to see real efficiency. So, lets think about the relative efficiency of an insurer covering all 323,000,000 Americans, the Bernie Sanders solution.

I specified the 5% profit margin and 5% risk premium for a reason. Together with understanding the Central Limit Theorem, these assumptions are based on the notion that the standard error for the loss ratio for our insurer with 1,000,000 policyholders is 5% of its premiums. In about 95 years out of 100, our efficient enough insurer will have a loss ratio between 0.65 and 0.85. That is ok but it should be clear that a 20% spread in how much of the premiums will be left each year will make life difficult for everyone.

How would our insurer covering 323,000,000 policyholders compare? Well, to calculate the risk management efficiency of a national health insurer, we invoke the Central limit theorem and we find that the standard error for the loss ratio for our insurer with 323,000,000 policyholders is not even close to 5%, it is only 0.002782%. Our largest possible insurer will have loss ratios that vary between 0.74722 and 0.75278, about 95 years out of 100. This insurer is so efficient that it will never go bankrupt because its losses never exceed its premiums. On the downside it also has very few years when it makes substantially higher than expected profits.

It is this very efficient insurer that lies at the heart of the current political season. Will America be better served by 323 “efficient enough” insurers that cannot plan to provide benefits of more than 75% of their premiums as benefits, or would America be better served by a maximally efficient insurer that could pay almost 85% of its premiums as benefits? Most would probably pick the insurer that pays the most benefits. Yet, all but one Republican party candidate, and one Democratic party candidate tell us that a national health insurer is the wrong choice for America.

This is, of course, only a teaser. A glimpse at some of the ways in which insurers can be efficient. I have a nice little paper that covers all the turf, covered by my book “Standard Errors: Our Failing Health Care (Finance) Systems And How To Fix Them” that I am distributing gratis as my contribution to rational discourse on the topic of how to achieve universal coverage. For a free copy pick the small paper up from my website:

In 2016 everyone should understand how insurance really works or we will blindly face the Woody Allen Dilemma in the balance of the primaries and throughout the election year:

“More than any other time in history, {wo}mankind faces a crossroads. One path leads to despair and utter hopelessness. The other, to total extinction. Let us pray we have the wisdom to choose correctly.”