Thursday, August 01, 2013

Introducing induction vs deduction: the fall of economic history at MIT

I fear my foreshadowed posts on MOOCs will have to wait!

A post on Professor Greg Mankiw's blog led me this June 2013 paper by Peter Temin, The Rise and Fall of Economic History at MIT. I was attracted by the title because it seemed to bear upon issues that I was exploring in my own amateurish way in Sunday Essay - economists and the decline of economics. Browsing, my eye was caught by this quote:

A more widely repeated comment was made by Paul Samuelson to Walt Rostow over lunch one day. After Rostow made a claim that Samuelson disliked, Samuelson said, “Walt, you may be an economist among historians, but you are historian when you are among economists.”

That resonated! I wonder why?

I found Professor Termin's piece interesting but depressing. Interesting because of the history, depressing because it was a similar pattern at another place at a similar time. I leave it to you to read the full paper, but I wanted to make a few comments.

Termin suggests that economics at MIT came to rest on three legs: theoretical economics, econometrics and economic history. The theoretical and econometric legs combined, ultimately squeezing out economic history. Underneath this result was an intellectual conflict that I am only just becoming aware of.

On Wednesday, Belshaw’s World – Barratt’s story: can the academic present measure up to its past? looked at some aspects of changing academic life through a prism set by Professor Barratt's history of the Psychology Department at the University of New England. By its nature, the book is partially a history of the changing face of psychology as a discipline seen through the eyes of one man at one place.

Underlying the changes, is a shift from inductive to deductive study within Psychology. Inductive reasoning is bottom up, a kind of reasoning that constructs or evaluates general propositions that are derived from specific examples. By contrast, in deductive reasoning specific examples are derived from general propositions.

Putting this another way, theoretical economist built models based on certain assumptions and then refined and extended them. As deductive reasoning gained sway, the economic journals filled with articles that were ever more detailed explanations of ever narrowing points. These explanations became hypothesis that were then subject to scientific test. This explains the natural fit between the theoreticians and the quantitative economists. However, you could only test those things for which certain types of data were available.

Something of the same process happened in psychology. Professor Barratt was a deductionist, although he used the term hypothetico-deductionist to combine the idea of hypothesis drawn from models followed by tests. Mind you, he never went to the same extreme as some of the economists, always recognising that you had to move between the particular and the general, that observation of human behaviour had a place in the creation of ideas.

I do not want this to be a long post. I am just setting a scene. In the meantime, do read Professor Temin's paper.    


Anonymous said...

Bad link to that paper Jim. But Google is our ever-present friend...


Jim Belshaw said...

Thanks, kvd. Link corrected!

Winton Bates said...

Thanks for the reference to Temin's paper. One of the things I found surprising was that while I didn't study any economic history at university, I am familiar with the writings of many of the people mentioned. That is partly because some of them, eg Rostow and Kindleberger, were writing for a popular audience. But it also reflects my interest in what is described as the new institutional economics.

I don't think the induction/deduction distinction applies to the contributions of any of these people. As I see it, they don't discover patterns of behaviour by churning vast amounts of data. They think up theories to explain the economic phenomena that they are studying.

Jim Belshaw said...

Both inductive and deductive approaches involved the creation of theories or models, Winton. Rostow's stages of growth is a model, and one I found useful. The distinction is, i think, that one derives the model from data, the other devices a model and then tests against data. It's a balance thing.