Wednesday, January 09, 2008

Healthcare, volatility and outliers

Tyler Cowen, a liberterian economist offers a grudging quasi-endorsement and an interesting analysis of single payer healthcare that I think is wrong and limited on program design, political economy and pragmatic grounds.

Government-dominated health systems, insofar as they work well (a number of them do), succeed simply by lowering costs. Health care has a murky relationship to human health, pharmaceuticals and broken limbs aside. A version of the single-payer system, as might be adopted in the United States, would not lower costs. We would be raising taxes and lowering medical innovation to give poor people a good deal more financial security and a slight bit more health; that is the relevant trade-off. [emphasis mine]

Please note the highlighted sections. He concedes that single payer systems have and can work, and he notes the basic mechanism of them working; they reduce baseline costs and there is evidence that they reduce trend cost increases as well. However he is saying that the US single payer model could/would not do this. That is a question of program design and a decent program implementation model. The Republican Party has no interest in designing such a system, but the Democratic Party has an interest in designing a system that delivers decent or better results at reasonable costs. The most recent example is the debate on whether or not Medicare-D should be subsidizing private insurers to such an extent that it costs Medicare more to send someone to a private plan than to keep them in Medicare. Also look at the program design in Medicare D where the government can not exercise its cost containment power by saying that it will buy drugs from companies at the lowest offered price to US interests, or OECD interests. This is basic program design here, and it is pure technocratic wonkiness to figure this out; it is not complicated.

The average, median, first and third quartile, first and fourth quntile American worker and family has noticed the following economic trends. We have experienced stagnant real wages per individual, greater labor intensity by sending the second member of a relationship into the workforce, more uncertainty concerning retirement income and health insurance, and greater income volatility. Pretty shitty deal overall, especially if you throw in lower savings, and a much weaker social safety net as reserves when exogenous shit happens such as job loss during a corporation wide restructuring. The rewards have concentrated at the top despite the economy and individuals as a whole being significantly more productive today than thirty years ago.

From a political economy/policy perspective the 'best' types of policies are Pareto improvement policies where no one is made worse off under a new policy regime than they would have been under the old regime and at least some people are better off. Most policies on first glance do not provide a Pareto improvement, but once second round impacts and side deals are included, it is theoretically reasonable to compensate the losers within a policy change from some of the winners' gains. For instance, Brad DeLong notes that the universal health care initiative in the Clinton administration was an attempt to compensate lower wage individuals for the increased competition that they would see from a successfully passed and implemented NAFTA. However that political trade-off was quashed. There is a significant amount of compensation owed for previous policy deals, and more accessible, more predictable, and cheaper healthcare for all people is a reasonable pay-off for the reaming the bottom 90% of the population has taken.

And now onto the pragmatic side, and I can wear my professional hat here, as I'm an analyst/evaluator. If I presented the following two graphs to my boss (1st via Barry Ritzholtz , 2nd via Kevin Drum) and said that there were no systemic issues or alternatives that could explain the oddity in inputs and outcomes compared to near peers, my boss would not be too happy with the quality of my work:

Let's just take a quick look at the input variable --- the average expenditure for the United States is massively larger compared to anyone else, while almost everyone else, excluding the outlier of Cuba, is clustered, spending roughly $2,000/capita +/- $1,000. Let's ignore the outcome for a moment until we get to the next chart:

Here the US spends the most but still gets crappy outcomes, while best in category are about 45% in this metric of avoiding deaths through the application of available medical care. And this is very common; in most areas the US spends significantly more than almost any other healthcare system and gets mediocre or worse results. The few areas of above peer performance are either in elective/convienence areas, or in areas of inputs and intermediate processes that have minimal real outcome differences such as the superior US detection and treatment rate for prostrate cancer when the long term death rates per capita from this disease between the US and the UK are virtually identical.

If I go into a meeting with my boss and say that this is how it is, and there are no potential explanations as to why the US spends so much for comparatively mediocre results, I'll get laughed at as it is implausible on its face that the US healthcare system is perfectly optimized to provide good and accessible healthcare to all and that there are few/no areas of currently improvable methods that need to be improved.

If I wanted to argue that the combination of US population profile, implicit race/classism in health care access/effectiveness, screwed up medical school system capping the number of doctors, lifestyle choices and information asymetries have a significant explanatory power, that is a cluster of reasonably related and coherent arguments and understanding of the variance.

However my boss would ask me a very uncomfortable question at this point about non-US group similiarities and how they differ from the US. The first answer I would have to give is that the US is the only nation in this group that overwhelmingly relies on private insurance to pay for healthcare, and that this could skew incentives, and fragment the market and standards as well as produce significantly higher administrative costs that produce no real health outcomes.

So I don't think that shifting to single payer or a bastardized intermediate step towards single payer would just result in a wealth and volatility suppressing transfer to the bottom 90% of the country; although I am fine if that is the actual outcome. I think there are significant baseline and trend costs savings available as well as better outcomes.

No comments: