Sunday, March 25, 2007

EconMethod: Assumptions

by Berly

The post in kaFE depok have been sporadic and on various topics. Not that it is bad; after all there are so many things to analyzed and so little time. But there is nothing wrong with series of posts on a connected topic to dig deeper and understand more. Therefore, I inaugurate the first instalment of our first series.

Economic methodology has not received wide attention in the academic discussion in Indonesia. At most department of economics in Indonesia there will be one class on economic history that implicitly discuss methodology and one class on general research methodology. There are very few, if any, undergrad doing their thesis on economic methodology.

With so many developmental and economic problems in Indonesia it is understandable that our Econ PhD students abroad are more occupied with mastering the use of sophisticated economic and econometrics tools rather than substantiate the arcane technical and philosophical assumptions behind. You know what they say, those who can’t do economics, study methodology.

But I would argue that it is impossible to do proper economics without understanding sufficient economic methodology (which also includes history of methodology and how they developed) and applicability to Indonesia. All-out liberalization would not worked as well in Indonesia as in US/EU since they already have strong legal protection at place (and taken as granted). Studying unemployment in Indonesia will need modified analysis of self selection bias since Heckman estimation was developed with unemployment insurance in mind. Poverty analysis in Indonesia could not use only income as indicators since many Indonesian farmers consumed what they produce. The list can go on.

Now a bit on qualifications. Aside from long time interest since undergrad time, I took courses on Economic History and Methodology at University of Amsterdam (UvA) at master level. Courses? Yeap, not only one course. And also not two courses. But three courses. One of them was taught by a John Davis that held two PhDs (in Economics and Philosophy) and the editor of Journal of Economic Methodology.

Occasionally, Indonesian economists quoted Milton Friedman paper entitled “The Methodology of Positive Economics” in defending unreal assumptions in economics. I don’t know if they have read Friedman’s entire paper, but I know I did. I was assigned to it and presented my analysis in the class.

Here is the infamous quote, “Truly important and significant hypothesis will be found to have assumptions that are wildly inaccurate descriptive representations of reality, and, in general, the more significant the theory the more unrealistic the assumptions…To be important, therefore, a hypothesis must be descriptively false in its assumptions…”

Friedman stressed that the validity of economics is not in the assumption but on prediction. A billiard player may not be know the of law of physics but he played as if he knows it since that is the only way to put the balls out and win the game. A person may not be totally selfish and materialistic but economist can predict the action by assuming as if he is.

There have been numerous critiques to this position. Alan Musgrave, a philosophy professor, said that Friedman over-generalized on assumption (assuming too much about assumption?). There are three kinds of assumption according to him and mixing them up is a mistake.

The neglibility assumptions are factors that only affect the analytical result in very minor ways thus can be assumed away. Example in physics includes air friction in a calm day that only slightly affects speed of falling item. The domain assumption put boundaries where theories are valid inside it but not outside. The celebrated example are the search for Theory of Everything since quantum mechanics and relativity, each valid in their domain, propose radically different descriptions of the universe. Heuristic assumption is use at early stage of theory presentation to be drop later.

The example in economics is as follow. Micro economics textbook usually start with an economy without government. Does it mean that whether government exist or not will have no detectable differences to phenomena investigated? No, so it is not neglibility assumption. Does it mean that the theory only valid when there is no government intervention? Domain assumption may play a role here. But mostly it is used as heuristic assumption which will be relaxed shortly.

Stiglitz (1994) has a stronger word, “I have shown that slight perturbations in the standard information assumptions drastically change all the major results of standard neoclassical economics. The theory was simply not robust at all.” I think he is referring to domain assumption meaning most of the neo classical theory that preceded recognition of asymmetric information problem was only valid if there were none.

The other strong argument against Friedman’s over-simplication is the heart of his argument. The model with self interest, super-rational and time-consistent agent don't always predict well, in some case it way off the fact. Yes, the neo classical model can explain some of the action, but the accuracy sure can be improve.

Aside from knowing the direction of action we also need to know the magnitude of the effect. Recent development in behavioral economics and experimental method prove to better predict (and explain) human behavior by incorporating additional (and more realistic) assumption that human also care (and willing to incur cost in taking action that benefit) on other people not only self, process not only result, non-material instead of only material (more in the next installment).

Please keep the above argument in mind next time an economist used Friedman’s line as shield.


johnorford said...

i think friedman meant that if you use complicated highly accurate models you lose explanatory power - i.e. it's harder for policy makers and their electorate to "get it".

that's why present day behavioural economics may be better, but they are less important, cos they are harder for ppl to get their heads around and therefore fuzzier ideas.

Berly said...

I acknowledge that there could be trade off between simplicity and accuracy.

But I will pull up a trick and quote from Friedman himself (from biography written by Ebenstein ):

“The role of the economist in discussions of public policy seems to me to be to prescribe what should be done in light of what can be done, politics aside, and not to predict what is ‘politically feasible’ and then to recommend it.”

johnorford said...

i agree with that quote -- one of my real pet hates is economist bending over backwards to accommodate popular political views and gain influence.

(actually as far as i recall, friedman was guilty of that too under reagan)

Berly said...

Yup, in term of policy making I'd like to see economist being good economist (employ all available tools available as best as possible) and politician being good politician (choose the best policy in term of society's welfare and explain to the people in accessable language).

It help if economist can write for public (like Samuelson, Krugman and numerous Indonesian economists) or politician have some economic background (like Kennedy & SBY).