http://leiterreports.typepad.com/ Is economics a "science", revisited
I wanted to comment on Brian's excellent past post on this issue. When the question is raised it usually bogs down pretty quickly into the dreariest sort of outdated Popperian philosophy of science (we make falsifiable claims! we do empirical tests!). This has the dual problems of ignoring the problems with falsificationism as philosophy of science and ignoring the fact that many economic theories don't seem to be all that falsifiable. Anyway, Brian takes the falsification defense apart, and goes on to press hard on various disciplinary sort spots...poor predictive track record, generic predictions that are obvious to common sense psychology, etc. This sort of thing is especially cutting since it is exactly the sort of dismissiveness (some) economists like to direct at sociology. In my experience, when the scientific nature of economics comes up it is often about justifying various disciplinary privileges relative to other social sciences. There undertone is that we *do too* deserve the Council of Economic Advisors, the Nobel, the higher salaries, the op-ed slots, etc. At least more than those other guys. Well, maybe. But if so it's not because economists are better amateur philosophers of science than people in other social science disciplines, or because we've solved the science demarcation problem when philosophers really haven't. It would be because we produce better explanations. Sciences ought to produce lots of true, useful counterfactuals (if you do X, then Y will happen) that aren't intuitively obvious. Does economics do a better job of that than untutored common sense and/or the other social "sciences" and humanities? This is a different question from how similar economics is to some idealized science model (e.g. physics). The accepted hard sciences differ greatly in methodology (e.g. geology is not a lab science), but one could argue they are and should be judged more by their predictive success than anything else. So I think Brian is right to focus on predictive success. The truth is that in about a century of development economics has never been able to get to a really deep, truly reliable level of predictive accuracy about any economic phenomenon at all. For a while people thought that level of accuracy had been reached in theorizing financial arbitrage (highly reliable predictions could be made about the future value of certain types of financial instruments), but then some very smart finance people lost a lot of money and it was back to the drawing board. But I do believe economics generates counterfactuals that are true, useful, and an improvement on common sense alone. (And an improvement on what other disciplines would produce if not supplemented by economic insights -- although I'll argue later that other disciplines can improve on economics taken alone as well). The question then is what about economics makes it successful in generating those counterfactuals, and what limits its success in doing that (perhaps the more interesting question, as it points to the areas of potential progress). Framing the question this way gets away from the project of physics imitation and toward the question of what useful social theory looks like. I'm going to start by making a few posts describing what I see are some major strengths of economics as a predictive discipline. Then (in a day or two) I'll post again on what I see as some major weaknesses of the discipline. Strength #1) Economics measures things. I'm not sure people realize how important economics has been in creating the basic concepts we use to think about and quantify the economy. Unemployment rates, inflation rates, economic growth, GDP, trade balances, productivity measures, capital measures, predictors of economic growth like inventory measures or consumer confidence, are all creations of economics and in many cases use substantial input from economic theory. (E.g. you can't really understand the idea of changing costs for a constant standard of living without thinking about how consumers substitute to different goods when prices drop, which comes straight from theory). Different social sciences (e.g. psychometrics) also have their measures, but I would argue that these are less fundamental to understanding the phenomena than economic measures have been. You can get a decent idea of whether someone is smart without administering them an IQ test, but modern economies are too large and too complex to understand without defining and consistently measuring concepts that reflect their state. The macro level economy would be in some sense "unthinkable" -- we would have no common conceptual language to discuss it in -- if not for the 20th century contributions of economists. The discipline realizes this too; I can think of at least three Nobel prizes given primarily for advances in measurement methods (pure measure development, not econometrics). Because economists have developed consistent time series for these measures over long periods, we have an idea of trends. This is very important by itself. For example, the slow growth in U.S. family incomes during the 1970s-80s was probably related to slow growth in labor productivity. But the slow growth in U.S. income from 2000-2004 was probably not due to this; it has something to do with how the returns from productivity are distributed. We also have some idea how different metrics have historically covaried. That allows some general (and highly imperfect) predictions of what they will be in the future. It also allows you to put bounds on the effects of at least some policy changes -- in particular limited redistributions of money across sectors of the economy that don't bring profound changes in social organization. I wouldn't want to exaggerate the measurement successes of economists. The measures economists use represent something real, but they are radically undertheorized and we often don't have that great an idea of what they really reflect. To take just one example, imagine the pitfalls of using the same measure of business investment for 1970 (before the IT revolution) as you do today. Yes, in both cases firms are spending money to buy a production input, but that input is so radically different that it could have completely different implications for economic activity. A side note: the computerization of administrative data sets means that economists' ability to track and measure the economy is starting to explode. Purely descriptive new data on economic changes has had big effects on how we think about the economy. (E.g. Davis and Haltiwanger's complete accounting of job flows has helped us understand that the level of economic activity and dynamism underlying our aggregate numbers is much greater than we thought it was). Marcus Stanley
