Code Red: Two Economists Examine the U.S. Healthcare System

October 28, 2011

What is a Life Worth?

Filed under: mysteries of health economics — David Dranove and Craig Garthwaite (from Oct 11, 2013) @ 11:25 am

This blog continues my ongoing series of “mysteries of health economics.”

The mystery this week is “what is a life worth?” We cannot ignore this question because it seems unthinkable. As will discuss, coverage decisions by public and private insurers depend on the answer. Some payers are rather explicit about they think a life is worth.

Before I try to solve this mystery, let me acknowledge that we should not spend money on health services that are of zero value (or worse.) But what about expensive health services that might prove to be of some value? How much should we spend on these?

Let us accept the reality of insurance. When we “purchase” health care, someone else foots the bill. Perhaps insurance should contain big deductibles, but even big deductibles are quickly exhausted if we need surgery or have a chronic health problem. If we are pooling our resources to pay for medical care, then we will probably want to reach some sort of collective decision about what drugs and treatments we will pay for. The alternatives would be to invite massive moral hazard. (Let me repeat for those who bang the drum loudly for big deductibles – deductibles are quickly exhausted when serious illness strikes and moral hazard again rears its ugly head.)

Now imagine a new cancer drug that offers a small prospect of survival to patients who have no other choices. Suppose that on average, patients who receive this drug can expect to live about another three months and that there are no downsides to this drug. If the drug company offered to give the drug away for free we would surely want patients to have access to it. If the drug company asked $100 million a dose, we would probably agree to spend the money elsewhere.

There must be some price under $100 million that would cause us to stop and think this over. Should we pay for the drug if it costs $500 per patient? What if it cost $100,000?

At some point we must answer the unthinkable question. In this case, we must determine how much should we be willing to spend to offer individuals three more months of life? Regulators in many countries have already given their answer. For example, the UK recently refused to pay for the skin cancer drug ipilimubab because the cost per “quality adjusted life year” was between £54,000 and £70,000. The UK uses a threshold of about $100,000 per year of life, a threshold that is accepted in most developed countries. But is that really what a year of life is worth? By all accounts, this threshold is based on past spending norms, gold IRA rollover rates adjusted for inflation. What was once seat of the pants policy, driven purely by budgetary needs, has become the gold standard for measuring the value of a life.

This all seems rather ad hoc. Rather than accept a valuation that was seemingly pulled out of thin air, academics have sought to value life by looking at how people actually behave. Some researchers have asked people what their lives are worth. Carefully constructed surveys generate values for a year of life well in excess of $100,000.

Such surveys are notoriously unreliable. If we could observe people spending their own money on health services, then we would truly know what they think their lives are worth. But insurance means that we only see people spending someone else’s money. Here is where economists have gotten rather clever. Workers are often confronted with tradeoffs between relatively safe jobs and relatively riskier jobs that pay more money. (Controlling for skills and experience, the data definitely show that riskier jobs pay more.) Assuming that employers do not pay higher wages out of the goodness of their hearts, they must be paying higher wages in order to convince workers to take on more risk. Led by Harvard’s Kip Viscusi, economists have looked at the data and determined that workers can expect to get paid an extra $5000-$10,000 to take on a job that has a heightened mortality risk of 0.1 percent. This adds up to $5-10 million for every additional death, or well over $250,000 for every year of life lost. If workers insist on getting paid $250,000 extra to compensate for the prospect of losing a year of life, then they must hold their lives very dear.

So is a year of life worth $250,000 to the average worker? Many people criticize Viscusi’s work, in part because it makes strong assumptions about what workers know about job risk and about worker mobility. Some economists observe that these calculations do not take into consideration the importance of hope. But if we try to correct for any resulting biases, the value of a life would probably be even bigger! One bias might work in the opposite direction – we might insist on receiving a lot of money to take on additional risk, but be unwilling to give up the same amount of money to reduce risk.

My mother always told me that “if you don’t have your health, you don’t have anything.” She held life very dear. So do I and so do most people – more so than the UK and other regulators want to believe.

Everyone is concerned about rising medical costs, and for good reason. At some point we may spend so much on medical care that we will no longer be willing to give up so much money to live longer or better, preferring to spend the money on food and shelter. And in today’s tough economic environment, perhaps quite a few of us have reached that point. But in our zeal to cut spending, let’s not throw out the baby with the bath water. Effective medical care remains one of the greatest bargains, even when that means spending tens of thousands of dollars to save just one year of life.

September 30, 2011

Do Hospitals Cost-Shift?

Filed under: mysteries of health economics — David Dranove and Craig Garthwaite (from Oct 11, 2013) @ 11:32 am

This blog continues my exploration of the great mysteries of health economics.

Northwestern University is one of Blue Cross of Illinois’ largest customers. Suppose that premiums for all BC plans are expected to increase by 10 percent, but NU is able to force Blue Cross to accept a 5 percent increase. Would you expect Blue Cross stick McDonalds with a 15 percent increase in order to cover the shortfall from NU? I wouldn’t, for two reasons. First, McDonalds would probably threaten to take its insurance business elsewhere. Second, the scenario I have described is inconsistent with profit maximization by Blue Cross. After all, BC’s ability to stick McDonalds with a 15 percent increase surely does not depend on the price paid by NU. Any negotiator whose willingness to stick it to McDonalds is conditional on the price charged to NU is leaving money on the table and probably would have been fired a long time ago.

We might never expect BC to raise prices to some customers to make up for shortfalls from others, so why do we believe that hospitals do this all the time? It is impossible to discuss Medicare and Medicaid payments without someone invoking the mantra of cost-shifting. The theory of cost-shifting is deeply ingrained in the minds of healthcare decision makers and the policy implications of the theory are profound. Consider that if hospitals cost shift, then the burden of Medicaid cutbacks falls on privately insured patients, not on Medicaid patients and the hospitals that serve them. This calls into question whether the cutbacks will result in any savings for taxpayers and cause any harm to Medicaid beneficiaries. It also makes you wonder why hospitals that serve low income communities struggle to survive. Couldn’t they just cost-shift their way out of financial difficulty? A cost-shifting zealot would conclude that the managers of these hospitals are incompetent.

I must confess that I perpetrated one of the best cited papers providing evidence of cost-shifting. I studied what happened at hospitals in Illinois in the early 1980s after a substantial cut in Medicaid fees, finding that hospitals did raise prices to privately insured patients by enough to make up about half the Medicaid shortfall. But things were different back then, and I don’t believe that evidence can be used to describe what happens today. For one thing, there was essentially no managed care in Illinois, so insurers had to accept the prices set by hospitals. Insurers are far more powerful today than they were back then. Second, all of the hospitals in Illinois were nonprofits and, as far as I could tell, most placed mission above profits. So it is possible to believe that prior to the Medicaid cutbacks, hospitals really were leaving private sector money on the table. With all the empire building that hospitals are engaged in today, it is hard to believe they would ever leave insurer money on the table. This makes it equally hard to believe that they would need the excuse of government cutbacks before sticking it to insurers.

Cost-shifting may be a flawed theory, but there may still be a disconnect between theory and practice. Hospitals might cost shift because, well, that is what they think they are supposed to do. So what does the modern evidence show? Will White and I published another paper about a decade ago that tracked what happened at hospitals in California after large Medicaid cutbacks. We found that hospitals that experienced large Medicaid cuts also experienced relatively slower increases in private sector payments, the opposite of what would have occurred under cost shifting. There are quite a few other studies showing that the quality of care delivered to Medicaid and Medicare patients suffers when government payments fall. This would not occur if hospitals could cost-shift.

Unfortunately, there are a lot of more stylized analyses that seem to show that cost shifting is alive and well. The typical analysis finds that profits from privately insured patients are negatively correlated with profits from government-insured patients both in the cross-section and over time. This is cited as conclusive evidence of cost-shifting.

There is alternative explanation that, unfortunately for the cost-shifting zealots, is quite consistent with the institutional facts. To motivate the explanation, consider an industry in which all firms earn zero profits, but the firms’ accounting systems are somewhat arbitrary and assign costs to different customer groups in a somewhat haphazard fashion. If a firm in this industry has two groups of customers, it may appear to be profiting from one group due to the way it allocates costs. Because the firm earns zero profits overall, it must appear to lose money from the other group. Thus, reported profits will be negatively correlated between customer groups.

Now suppose that one group of customers got its act together and demands lower prices. This would have no impact on the price paid by the other group in the short run. In the long run there would be exit, because some firms were losing money. This would drive up prices and again create a negative correlation in pricing both in the cross-section and over time. But this would not be cost-shifting as it is commonly discussed. (Nor would it require arbitrary cost-accounting.)

If we relax the assumption of zero profits but instead suppose that profits are constrained to a fairly narrow band, then we would still get the same negative correlation in both the cross-section and over time, provided there is a fair degree of arbitrariness to cost allocation. And this, I believe, pretty well describes the hospital sector, where most hospitals have profits in a range of plus or minus 5 percent, and cost allocation is speculative even in the best institutions.

So I can explain away the stylized evidence without invoking the mantra of cost-shifting. But that does not make my explanation correct. Cost-shifting is so deeply engrained that CFOs might do it even though it is not profit maximizing. I think I could even construct a game theoretic model in which hospital CFOs use government cutbacks as a kind of focal point for passing along tacitly collusive price increases, so that cost-shifting is profit-maximizing in a strategic sense. My point is not to deny cost-shifting so much as to point out that there are good reasons to question both the theory and evidence. Whether or not hospitals cost-shift remains one of the great mysteries of health economics.

September 21, 2011

Why Aren’t Medical Prices Infinity?

Filed under: mysteries of health economics — David Dranove and Craig Garthwaite (from Oct 11, 2013) @ 12:28 pm

This blog continues my exploration of mysteries of health economics. The title of the blog may seem inane, but when a senior colleague asked me this question 25 years ago, it changed my life. And the answer helps us understand a lot about what is wrong with today’s healthcare system.

I was in my second or third year as an Assistant Professor at the University of Chicago when my brilliant senior (but still young) colleague Dennis Carlton asked me to explain how medical providers set their prices. I told him that we needed to throw the traditional textbook economics model of pricing out the window. This wasn’t a market where price sensitive consumers chose among homogeneous sellers, with the result that prices in competitive markets converged towards marginal cost. Instead, consumers had insurance that paid for all or nearly all medical bills. Moreover, patients were loyal to primary care physicians and their referral networks. As a result, patients rarely shopped around for the best price. This was when Dennis asked me why prices weren’t infinity. The question stumped me! I supposed that insurers would only pay usual, customary, and reasonable rates but that didn’t prevent providers from asking for infinity and occasionally getting it. Perhaps providers didn’t want to appear unseemly or were bound by ethical constraints. Or perhaps, as I ultimately responded, medical prices were inflating so rapidly that they would soon reach infinity.

The conversation ultimately led me to examine all sorts of pricing puzzles. Why did prices seem to be higher in more competitive markets (in violation of the traditional price/concentration relationship?) Why did specialists make so much more than generalists? Do any of the rules of pricing apply to medicine?

Economists have solved some of these puzzles. The seeming violation of the price/concentration relationship for hospitals was partly a statistical artifact resulting from a failure to control for quality and severity of illness. And once managed care took over, the traditional price/concentration relationship firmly established itself. The violation for physician pricing could be explained by simple economic forces in monopolistically competitive markets (i.e., markets with many slightly differentiated sellers each of whom has some loyal customers). If some factor causes prices to be higher in some areas than in others, physicians will tend to gravitate towards the high priced areas in order to share in the higher profits. It is not difficult to imagine what factors might cause prices to differ – socioeconomic conditions, culture, the willingness and ability of patients to shop around. Health services researchers offer an alternative hypothesis – that physicians in concentrated markets “induce demand” in order to drive up prices. But this hypothesis had little empirical support beyond some old studies with major statistical flaws. And even these old studies found modest inducement effects at best – not enough to explain the data. Besides, the inducement hypothesis fails to explain why some markets are more concentrated in the first place. The market forces explanation is consistent with the data with the added virtue of explaining the variation in concentration.

Economists will be hard pressed to explain the pricing data that was just reported in Health Affairs. Laugesen and Glied find that U.S. physicians earn far higher incomes and charge far higher prices than their counterparts in other developed nations. The pricing gap is larger for specialists than for generalists. These price differences cannot be explained by differences in costs, including the cost of medical education. Economists often suggest that entry barriers protect physicians against competition that might drive prices down. But it is more difficult to become a specialist in Europe and Canada than in the United States. Moreover, Laugesen and Glied provide further data suggesting that U.S. physicians are not working at capacity; excess capacity is one of the surest predictors of price competition in almost any market, even hospitals. So why not physicians? And the specialist/generalist pricing gap has not been explained to anyone’s satisfaction, especially when there is excess capacity.

Policy makers seem likely to use the Laugesen/Glied findings as an excuse to slash physician payments. But until we understand why physician fees are so high, we will never be able to accurately forecast the impact of fee reductions. Unfortunately, after 30 years of trying, we are no closer to explaining these pricing patterns. During this time, prices have climbed inexorably closer to infinity.

Blog at

%d bloggers like this: