Economics and science – careful with the prediction call
I see that Rosenberg has created a bit of an uproar in the economics community with his claims around economics (although the articles focus is macroeconomics) not being a science.
Unlike other economists I think this is a good thing in terms of forcing economists to defend against the claim – however, like other economists I don’t quite agree with his sole focus on predictive success and the way he frames elements of economists’ arguments. To me the suggestion that economists are instrumentalists (the Friedman essay) misses the point on what economists actually do – fitting the Friedman essay into a consistent lens of understanding for what economics is and what economists do is a hard, but fun, thing to attempt do with friends over beer 🙂
I hope to write more about ukmeds in the future where I actually add to a discussion on how economics and medicine can be seen as “scientific” in some form, and whether it really matters. But for now I’m going to largely hand over the critique of Rosenberg’s claims to a 1996 paper from Uskali Maki. We can do this, as Rosenberg’s criticisms in the article stem directly from his earlier writing – which Maki was discussing here!
Rosenberg ignores the fact that economics has long been divided into two major orientations in relation to this issue. There are those who regard economics as a policy-oriented engineering science, heavily dependent on predictive capabilities. This line is represented by people like Friedman and Klein, to mention two Nobel Prize winners. And there is another tradition of viewing economics as an explanatory endeavour that helps us resolve the paradox of social order by referring to the institutional structure of market society.
This is a separation that James alluded to when discussing models recently! And it is the view that ‘what economists do’ falls largely into the second camp which, combined with the idea that empirical evidence is a link between the “worlds” associated with our models and some underlying objective reality, has seen me define ‘neo-classical’ economics a certain way and in turn term the blog as neo-classical!
Maki goes on to discuss what this can mean regarding our definition of science and the lack of empirical evidence used by Rosenberg for his claim that economics does not progress from ‘general’ to ‘specific’ statements.
From there it gets very exciting, when Maki tackles Rosenberg’s common claim that economics is folk psychology! An issue I touch on in my NZAE paper from this year.
For those interested, this paper by Maki is one of the most compelling explanations of where economics fits in to things in a philosophy of science sense.
My view in a nutshell is, blankly calling on Popperian falsification is a poor way to go. Stating that economics can only be seen as a progressive science if it offers up “novel” predictions and testable hypotheses is better – but the fact that multiple explanations can be found for each phenomenon in some subsets of economics, and can’t be separated due to the paucity of available information, makes this problematic. And even if we were to only accept the policy-oriented engineeering science of economics as relevant we run into the problem that the outcome we are targeting isn’t observable and when we use observable proxies we face …
@Noahpinion Science has luxury aiming at fixed target. Economics aims not well at a reflexive one. Not economics fault: just admit its hard.
— Emanuel Derman (@EmanuelDerman) August 27, 2013
Update: Note that you can make a narrow definition of science (like Popper did, when he made one so strict it practically excluded physics) and say economics isn’t one, whatever. But Rosenberg’s conclusions about massively discretionary monetary policy do not follow in any sensible and logical way from that 🙂
There may of course be another way of considering that, economics may well be a science but that does not mean that all economists act scientifically. A parallel – veterinary science is mostly scientific, but recently I took my cat to a vet who suggested acupuncture. I was appalled, and asked the guy had he learnt that (acupuncture) at university – and, of course, he had not. Medical doctors also give out advice that is not based on science – such as when they prescribe an antibiotic for the common cold. Adam Smith, BTW, based much of what he wrote of first hand observation – the basis of science.
Applied practitioners are usually a bad guide towards whether the research involved with a discipline is scientific – although it can be a hard line! That is part of the reason that Greg Mankiw used the “economic engineer vs economic scientist” description to try to show the split a bit! And it likely holds in a bunch of other disciplines as well, such as the ones you name!
For example, I am loath to put myself in the “economist” camp here at all – as I am not an economic scientist in the ways we’d see it, not at all! I sell a business service that relies on the empirical and theoretical evidence/products produced by economic scientists (along with using an applied economics skill set to try and fill gaps by analysing data that is available), and my comparative advantage is (hopefully) the way I can communicate those things to a non-economist audience.
I am very interested on the ideas about economic science, and trying to understand the argument for that – as it feeds into how I understand what economic scientists make so that I can use it. In that way I hope my perspective on it is relatively impartial 😛
Matt, as someone who is also in the business of selling predictions, a byproduct of economic theory and econometrics, I see the role of economics as being very much focussed on point 2, explaining how things work. This knowledge can help to make some predictions and help us make policies. Like some of the cool stuff they mentioned in Nudge: http://www.amazon.com/Nudge-Improving-Decisions-Health-Happiness/dp/014311526X
Users of forecasts are really interested in the risks around this forecast. It is a narrative of likely future environment with plenty of explanation of what the drivers and risks are. So, the reader is easily able to see when something changes.
What economist fail at is usually communication…
Indeed, I would go even further than you and say that the “policy engineering view based on prediction” is naive and thoughtless – in truth we need explanation to set policy.
As mentioned at the bottom of the post unobservability, reflexivity, and the fact that many explanations can map to the same set of data make any analysis non-nonsensical without answering “why”.
What makes me most nervous is the number of people out there who truly believe naive data analysis, and models with the best “predictive power”, are really the be all and end all of the social sciences …