Thursday, October 18, 2012

Lies, damned lies and statistics

A couple of responses to some recent pieces from The New York Times op/ed section. First up is this Steven Rattner column in which he writes:
Eating the seed corn is never advisable, yet that’s what Washington is already doing. The share of spending on infrastructure (roads, airports, dams and the like) fell from 2 percent of G.D.P. in 1971 to 1 percent in 2010.
Basic math: GDP in 1971 was $1.113 trillion. Two percent of that is $22.26 billion, which is $118.48 billion in 2010 dollars. GDP in 2010, meanwhile, was $14.447 trillion, of which one percent is $144.47 billion. It is apparent why Rattner chose to measure spending in terms of GDP rather than constant dollars -- rather than the infrastructure budget being cut, it has increased by 22 percent.

In fairness, however, one must concede that population growth since 1971 (49 percent) has exceeded growth in the infrastructure budget. But so what? It is not obvious that infrastructure spending increases should correlate in exact proportion to population growth.

Take roads for example. In 1971 the interstate highway system was still under construction (not fully finished until 1991), whereas now the system is mature and the focus is on maintenance -- a cheaper proposition -- so the fact that spending has declined relative to population isn't terribly revealing. In areas such as Manhattan or Washington DC, the current road network and street grid is similarly mature, with a greater emphasis placed on maintenance than more expensive new construction. Similar logic also applies to the sewer and subway systems.

Rattner, who graduated with honors in economics from Brown and has been deeply involved both in Wall Street and public policy, is too smart not to know any of this. The more realistic conclusion is that he is being intentionally deceptive.

Lastly, the better question is not why the federal government isn't spending more, but why it spends so much. Shouldn't infrastructure projects -- the vast majority of which are local and state projects -- be funded by state and local dollars? Why is the federal government handing out funds for bike trails in Minneapolis, a dog park in Somerville, MA or air travel to rural areas (this is the "seed corn" of which Rattner speaks?!)? Why send money to Washington, only for it to be returned right back? Does anyone believe that adding more layers of government to the process makes it more efficient?

Next up are recent pieces by Paul Krugman and Nick Kristof. Krugman writes:
Surely the fact that the United States is the only major advanced nation without some form of universal health care is at least part of the reason life expectancy is much lower in America than in Canada or Western Europe.
In a similar vein Kristof says:
[My former college roommate's] death [from cancer] was also unnecessary and might not have occurred if he had lived in Britain or Canada or any other modern country where universal health care is standard and life expectancy is longer.
Krugman's use of the word "surely" is execrable. As a trained economist, he should be trafficking in facts and data rather than intuition when leveling such an indictment of the US health care system. That Kristof implies a linkage between government-run health care and life expectancy without evidence is less surprising. This is, after all, a guy who devoted an entire column to praising a book about income inequality that The Economist recently described as "largely debunked."

As for the topic at hand, here's an interesting fact: life expectancy in Hawaii (81.5 years) and Minnesota (80.9 years)  is greater than or equal to every member of the European Union and Canada with the exceptions of Italy (82 years) and Spain and Sweden (each tied with Minnesota at 80.9 years). This is not an argument that the rest of the EU would be well advised to copy the health care systems of these two states, but rather an illustration of how meaningless such statistics are.

It's also worth noting that despite the fact every country in the EU has universal health care in some fashion, life expectancies among even its richest members vary by nearly four years (82 years in Italy vs. 78.3 in Denmark). For comparison, the difference between the country with the longest life expectancy, Japan, and the US is 4.5 years (82.7 vs. 78.2). If health care systems plays such a determinative role, why such differences?

The answer, of course, is that a myriad of factors determine life expectancy, and assessing which ones play the biggest roles is incredibly difficult. How big of a factor is culture and race? (Japan has the world's longest life expectancy while Hawaii, the state with the greatest percentage of Japanese-Americans, has the longest life expectancy of any US state. The life expectancies of Swedes and Minnesotans are exactly the same -- is it coincidence that Minnesota is the state with the greatest number of Swedish-Americans? Similarly, Norway and North Dakota have the exact same life expectancy of 80.1 years, while North Dakota is the state with the greatest percentage of Norwegian-Americans) What about climate? (Israel, Italy and Spain are all in the top 10 while Greece, Cyprus and Malta are in the top 30, and California with its Mediterranean climate is #3 among US states) Murder rates? Suicide rates? Auto accident rates?

More honest columns from Krugman and Kristof would have acknowledged that determining the factors behind life expectancy are very hard, and that such cross-country comparisons have limited utility. Rather than any such acknowdgement, however, Krugman declared that a difference in health care systems is "surely" the reason for "much lower" life expectancies in the US, while Kristof conflated health care and life expectancy without any qualifier at all. Both should be ashamed of employing such lazy and dishonest argumentation in pursuit of blatantly ideological ends.  

Related: Further must-read commentary on Krugman from Russ Roberts and Warren Meyer.

No comments: