Economists Should Get Out of the Science Lab
Maybe it’s a deep-seated inferiority complex. Or a desire to be accepted, even respected by the great minds that have transformed mankind with their inventions. Otherwise, how else to explain why some economists keep insisting that economics is a science?
The announcement last week of the winners of the 2013 Nobel Memorial Prize in Economic Sciences rekindled the science-or-no-science debate among practitioners, journalists and casual observers. British economist Liam Halligan, writing in the U.K. Telegraph, reminds us that economics is about human behavior in the allocation of scarce resources, not scientific certainties. Raj Chetty, a professor of economics at Harvard University, makes the less persuasive counterargument this week in the New York Times.
Chetty admits that the “fundamental challenge” faced by economists is “our limited ability to run experiments.” Economists may try to hold everything else constant -- “ceteris paribus” -- to test a single variable, but they can’t. Nor can their econometric models, no matter what the economists claim.
Still, he says economists “have begun to overcome these challenges by developing tools that approximate scientific experiments to obtain compelling answers to specific policy questions.”
I wonder how Chetty would use those tools to provide compelling answers to the following questions:
1. What is the expected rate of economic growth in conjunction with a 4 percent federal funds rate?
2. What is the effect on inflation of a 300 percent increase in the monetary base?
3. What is the effect -- the multiplier -- of a $1 increase in government spending on output?
4. What is the nonaccelerating inflation rate of unemployment, or the jobless rate that triggers rising prices?
5. What is the wealth effect from a 20 percent increase in the major stock indexes? What about a 100 percent increase?
The answer to all five questions is, it depends. And that’s one of the main reasons that economics isn’t, and will never be, a science.
Isaac Newton, the English physicist, mathematician and philosopher, pretty much explained the fundamental difference between economics and the hard sciences more than 300 years ago. With the physical sciences, we observe what happens in nature. Then we try to quantify it. An apple falls from the tree to the ground with increasing velocity. Water boils at 100 degrees Celsius at sea level. Light travels faster than sound. Each observation yields the same result. It’s why mathematicians end their proofs with QED -- “quod erat demonstrandum,” or that which was to be demonstrated -- and economists don’t.
Or, to paraphrase Newton: The same results are always obtainable under the same conditions. It is the repetitive duplication of a result that defines what are called laws of nature.
A law of nature, when properly measured, will yield duplicate results. A law of economics, even if properly measured, will not.
Now, every once in a while life offers something close to a natural experiment, or what’s called a control study in science. For example, economists wanted to determine the effect of extended unemployment benefits on the labor market, so they used data from abutting counties with homogenous populations in two adjacent states, each of which has its own benefits policies. Chetty cites this study as an example of the new kind of empirical work being done in economics. He fails to mention that different economists came to different conclusions using the same data set. QED? Hardly.
Then there’s the frequently cited Oregon Health Insurance Experiment, which expanded Medicaid benefits to state residents on the basis of a lottery. Chetty calls it a “first-rate randomized experiment” that found access to coverage increased usage and improved well-being.
Not exactly, according to the New England Journal of Medicine. The study found that those with access to health care did use more of it. It found a higher rate of diabetes detection and management, and a lower rate of depression. Overall, however, “Medicaid coverage generated no significant improvements in measured physical health outcomes in the first two years,” the journal reported.
In another study, Chetty says he was able to isolate the causal effect of high-quality teachers on students’ outcomes as adults. He did this by comparing the performance of two groups of third-graders before and after an excellent teacher went on maternity leave. (What, precisely, constitutes excellence?)
I could tell you that high-quality teachers have a positive effect on their students without a study. Personal experience or an informal survey would suffice.
Ah, but how big is that influence? Is it worth moving to a community with a better educational system to give Billy and Betsy a brighter future, even if it means higher taxes and a mortgage I can’t really afford?
That I can’t tell you. I do know that a scientist would have reservations about Chetty’s control study. What if a bunch of kids got seriously ill and couldn’t attend the class once led by the teaching ace? Perhaps some children are living through a divorce or dealing with a sick or terminally ill family member, both of which may affect their performance. It’s impossible to control for effects that have nothing to do with teacher quality.
If economists would just give up the science fixation once and for all, they might begin to appreciate the value in what they do.
(Caroline Baum, author of “Just What I Said,” is a Bloomberg View columnist.)
To contact the writer of this article: Caroline Baum in New York at firstname.lastname@example.org.
To contact the editor responsible for this article: James Greiff at email@example.com.