As an historian, I am somewhat appalled at the inability of economists, including those on this blog to get the history of their own discipline straight. The obsession has been with neoclassical economic’s attempt to turn economics into a physico-mathematical discipline as Walras phrased it, and the economists usually discuss this attempt within the historical context of their discipline pre-1945, with references, to Walras, Marshall, Keynes, and others.
It became clear to me over thirty years ago, that the neoclassical effort to turn economics into a prescriptive science had failed before WWII....
When I wrote my chapter on The New Paradigm in Management and Higher Education Since 1940, CUP, 1989, I focused on the methods of Operational Research developed during WWII and the Cold War, that neoclassical economics imbibed . I wrote, for example, about how the Rand Corporation working on OR problems for the US Air Force gave birth to George Dantzig’s linear programming algorithms in 1947.
Postwar military planners and the economists who worked with them at Rand believed the new toolkit would transform neoclassical economics into a prescriptive science. At Rand in 1948, the economist Kenneth Arrow used the toolkit in his work on Rational Choice Theory. The neoclassical economists Joseph Dorfman, Paul Samuelson, and Robert Solow applied linear programming to their subject as well (in Linear Programming and Economic Analysis, 1958).
Why isn’t the source of the new paradigm in OR being discussed, instead of preWWII economists.
When I wrote the second chapter in my 1989 book, “The New Paradigm Revisited,” I questioned through the critics, how the prescriptive prowess of The New Paradigm fizzled. The people I cited were primarily OR scientists themselves. That is, I note that the people whose methodologies led to the New Paradigm, questioned the effectiveness of their own discipline. The prime example of this volte-face is Russell Ackoff, who popularized OR methods in the UK in the 1960s, only to write in a 1979 article, “The future of operational research is past,” “OR problems can never be a perfect representation of a problem. They leave out the human dimension, the motivational one. [Problem solving requires] the application not only of science with a capital S, but also, all the arts and humanities we can command.”....Real-World Economics Review Blog
The New Paradigm that emerged in economic s after WWII
Robert Locke
9 comments:
Important post.
With hard science you can measure physical events. This measurement is not subjective, e.g, A moves to B in x time. But the mathematics of people's desires and wants has a subjective basis. Economists spend too much time on the mathematics and not enough time studying people. Economists should be psychologists first and foremost, their discipline should be psychology. Hey, we do have them, they are called behavioural economists? Time to boot the neoclassical guys out and put the people in who really do study people and their desires and wants.
Homo economicus is a modeling simplification that contains a lot of unrealistic assumptions.
The question is whether it is a net that capture anything useful for the purpose for which the model is constructed.
Then the next question is if so, is the model then extended beyond the scope and scale of the assumptions.
If so, the question follows whether this is due to mistake or for ideological persuasion.
In the new "empirical" machine learning paradigms, the models don't know why things happen but can answer what happens. By using hundreds or thousands of possibly relevant data points a computer can predict a variety of econometrics without making ANY assumptions or having any knowledge about the underlying complicated system. Here is a good non-trivial primer on machine learning in Econ prediction and forecasting and you don't need multivariate calculus, linear algebra, python and prob/stats to understand it!
I think it might help heterodoxy to criticize orthodoxy more effectively by using some of the new "empirical" black-box economics by showing how easy conceptually it is to use. It doesn't really solve fundamental issues of complex system analysis epistemology and ontology even though it is better at fuzzy guesses than old fashioned simple human models. The fundamental problems of Human systems, Uncertainty, Arrow of time etc remain though they aren't the weak spot of machine learning as with modelling because machine learning avoids the problem entirely by not making assumptions and not making any statements about causality but relying on stacks of coincident correlation.
What is especially important is how easy it becomes for interested actors to poison big data once they know it is being analyzed and measured to predict (or trade or make policy). You can see it in google for example with their endless battle between spam content trying to move up the relevance of their search placement and google trying to suppress their content while presenting the highest quality results.
"Like"
Locke seems trustworthy enough on his own specialties and interests and has roughly speaking the right spirit, but he identifies business economics, management economics with microeconomics with economics, as if macro or monetary economics never existed, and has ideas on the history of ideas which are astoundingly, impossibly wrong. He does not get the history of economics straight - holding some positions which are unique to him - though perhaps my continued criticism has had some effect.
There is an interesting quote or comment on Wikipedia’s entry for Russell Ackoff. https://en.wikipedia.org/wiki/Russell_L._Ackoff
It’s by Peter Drucker.
From Wikipedia:
“Russell Ackoff was friends with Peter Drucker from the earliest days of their careers. Mr. Drucker acknowledged the early, critical contribution Ackoff made to his work – and the world of management in general – in the following letter, which was delivered to Ackoff by former General Motors V.P. Vince Barabba on the occasion of the 3rd International Conference on Systems Thinking in Management (ICSTM) held at the University of Pennsylvania, May 19–24, 2004:
[Peter Drucker was dead by 2004. Here’s the letter.]
“I was then, as you may recall, one of the early ones who applied Operations Research and the new methods of Quantitative Analysis to specific BUSINESS PROBLEMS—rather than, as they had been originally developed for, to military or scientific problems. I had led teams applying the new methodology in two of the world’s largest companies—GE and AT&T. We had successfully solved several major production and technical problems for these companies—and my clients were highly satisfied. But I was not—we had solved TECHNICAL problems but our work had no impact on the organizations and on their mindsets. On the contrary: we had all but convinced the managements of these two big companies that QUANTITATIVE MANIPULATION was a substitute for THINKING. And then your work and your example showed us—or at least, it showed me—that the QUANTITATIVE ANALYSIS comes AFTER the THINKING—it validates the thinking; it shows up intellectual sloppiness and uncritical reliance on precedent, on untested assumptions and on the seemingly “obvious.” But it does not substitute for hard, rigorous, intellectually challenging THINKING. It demands it, though—but does not replace it. This is, of course, what YOU mean BY system. And your work in those far-away days thus saved me—as it saved countless others—from either descending into mindless “model building” – the disease that all but destroyed so many of the Business Schools in the last decades—or from sloppiness parading as ‘insight.’”
Great quote from the always insightful Peter F. Drucker.
Applied math is a tool in the tool box of critical thinking. It is a key tool, but not the only key tool. It involves rigor in quantitive thinking but it doesn't involve qualitative thinking.
In addition, applied math is based on conceptual thinking. The variables in a model have to be related to what is being modeled and this necessitates careful conceptual thinking.
There is also purpose and purpose involves norms so normative, prescriptive and performative thinking is also involved.
This all has to come together in systems thinking in rigorous application of thought. Heuristic thinking takes short cuts to reduce "costs" that are not only economic. Time expended is a cost, for example, even if it is not specifically priced. Habits are examples of established application of heuristic thinking, as are social customs as well.
Critical thinking involves both art, or craft, and also science.
Drucker was well aware of that, even though he was "an efficiency expert." But he prioritized effectiveness over efficiency. The job involves getting the job done right, that is, meeting objectives as efficiently as possible.
Calgacus. I learned economics while studying the history of Germany. This meant that I learned from List and the German historical school that classical economics was a great power con game. I also learned that German historical economists (Weber, Schmoller, etc.) were not prescriptive scientists in that their work could not be used to prescribe business policy and management methods. That came in Germany with the development of business economics (BWL) which provided the administrative cadres for running business and industry in the country. Talk about ignorance of the history of economics, economists and management scientists in anglosaxonia, never heard of German business economics and of the role people trained in it played in Germany's mental capital formation, of the differences in their concerns about the purposes of the firm and firm governance. That part of the history of economics you won't get in anglosaxonia.
Post a Comment