Thursday, January 12, 2017

Mathiness and the narrowing of economics

Lars Syll has a nice post with several interesting links on the ongoing debate on the "intellectual regress" in macroeconomics. In particular, the post discusses the problem with the Lucas-Prescott-Kydland-Sargent "calibration" models. 

Calibration modelling involves identifying uncertain parameters and revising the model coefficients using actual historical data with some measure of goodness-of-fit. See Paul Krugman here and here on calibration modelling. 

Stripped off the jargon, the issue at hand is that one school of thought advocates using a model arrived at based on their theory of change whose parameters are refined (or calibrated) by testing it against actual data. But another group critiques this approach arguing that such calibration is barking up the wrong tree if the theory of change and, therefore, the model itself is flawed. They also argue that unlike scientific models which are generally time and path invariant, the dynamics of change in social science are time and path dependent.  

Also Paul Romer's scathing attack on how "mathiness" and excessive deference led macroeconomics astray,

a general failure mode of science that is triggered when respect for highly regarded leaders evolves into a deference to authority that displaces objective fact from its position as the ultimate determinant of scientific truth.
The complement of "mathiness" is the narrowing of the study of man in the ordinary business of life from its older version of "political economy". On a related note, Economist has a nice year-end article on the Cambridge school of political economy founded by Alfred Marshall. The contrast between the multi-disciplinary school of political economy of Alfred Marshall and AC Pigou and the modern "mathy" econometrics-heavy instruction is seen in their respective student instruction and assessments,

The goal can be seen in the exam questions of the time. Students were expected to combine economic principles with a strong grasp of current affairs. In 1927, for example, one paper on public finance asked students to explain the size and reasons for the main areas of British government spending. They were expected to have the skills of an essayist, spending one three-hour exam on a single question such as the future of gold, the rights and duties of shareholders, or alternatives to democracy. Cambridge economics considered itself to be an analytical science but calculation was not of the essence. A module in statistics produced a page-long test for final-year students; all the other papers were bare of mathematical symbols. Compare this with the exams of today. Charlotte Grace, a student in the third year of the economics tripos (as undergraduate degrees are known in Cambridge), says she could have passed all the questions she faced in her first year without reading a newspaper. And though the five-page final-year macroeconomics exam that was set in 2015 asked about some contemporary policy conundrums, like which features of the euro zone may have contributed to its sovereign debt crisis, most of the paper sought to test students’ knowledge of tricky, algebra-heavy models. Three-hour pontifications on a single topic have been ditched in favour of a compulsory dissertation in which original empirical analysis is encouraged.

These tests reflect changes in the discipline. Students must master the technical apparatus of a highly specialised field. The maths they need to know and apply is sufficiently taxing as to barely leave time for history. Evidence-based conclusions are preferred to arms-length analysis; economists should know the limits of their expertise, and shy away from political judgments as they think through the effects of whatever policy tweaks providence might throw at them.

1 comment:

Karthik Dinne said...

The problem runs much deeper and wider, across all social science disciplines.

Apart from obsession with mathematics, there's also a phenomenon of under-researched themes and perspectives. Rigour is common to all in academia but the differentiating factor is the 'choice of questions' and 'perspective'. One's perspective and beliefs, shaped primarily by family profile and undergraduate training, determine one's choice of research questions.

The Ph.D recruitment process that places huge emphasis on undergraduate college brands, which is a result of privilege most of the times, and mathematics inevitably lead to narrow profiles.

In absence of such diversity, some important themes and perspectives missed out.

Tyler Cowen asks Dani Rodrik - "What to do to get more Dani Rodriks from econ grad schools?". Rodrik responds "We are recruiting mathematicians these days and not economists". He goes on to say that his undergrad training in political science makes him look at things from a different perspective.

An old post of mine on this here