Kepler, Newton, & Galileo

There is great value in reading old papers. No, not ones I wrote as a young prof, although I am friggin' old. I mean really old papers. Unfortunately, only a select few have the time and capacity to read such papers... mainly those doing a 2.5 year post doc if you believe Twitter. 

While I never did a post doc, I found myself going down a bit of rabbit hole a little while back while editing my lecture notes for the coming year. I read the paper, "Measurement Without Theory," by Tjalling C. Koopmans, THE REVIEW OF ECONOMIC STATISTICS, 1947. [Note: Not a typo. Apparently the journal name was later changed to THE REVIEW OF ECONOMICS & STATISTICS.] 

Honestly, I was anticipating a paper on measurement error. Instead, I found something that -- building off my last post -- is another aspect of modern applied economic research that distresses me. In fact, it's kind of a two birds, one stone kinda thing as it also relates to that last post as well. 

Koopmans is ultimately providing a review of a 1946 book by Burns & Mitchell entitled Measuring Business Cycles. Koopmans criticism of the book is that the authors approach is "empirical" as opposed to "theoretical". Koopmans writes

"The various choices as to what to 'look for,' what economic phenomena to observe, and what measures to define and compute, are made with a minimum of assistance from theoretical conceptions or hypotheses regarding the nature of the economic processes by which the variables studied are generated."

To put this type of scientific approach in historical context, Koopmans harkens back to the work of Kepler and Newton on the behavior of celestial bodies. 

Kepler's approach to understanding the behavior of planets entailed measuring their positions and charting their orbits. This led to the discovery of simple, empirical "laws". As Koopmans summarizes,   

"This achievement was a triumph for the approach in which large scale gathering, sifting, and scrutinizing of facts precedes, or proceeds independently of, the formulation of theories and their testing by further facts."

Returning to Burns & Mitchell, Koopmans views their study of business cycles as even more atheoretical than Kepler's work. He writes:

"Burns and Mitchell are more consistently empiricist than Kepler was. The latter made no secret of his predilection for the principle of circular motion until observations spoke decisively for the elliptical orbit. He held other speculative views as to the role of the five regular solids and of musical intervals in the proportions of the planetary system, which now appear as irrelevant. Burns and Mitchell do not reveal at all in this book what explanations of cyclical fluctuations, if any, they believe to constitute plausible models or hypotheses."

So, in Koopmans' view, the empirical approach is -- at the risk of oversimplifying -- atheoretical and is designed to document empirical regularities which then may guide the development of new theoretical insights. In contrast to Kepler's work, Koopmans then delves into Newton's contributions to our understanding of the behavior of celestial bodies. While Koopmans admits the distinction is a bit blurry, Newton's focus was on the development of fundamental laws that applied to all matter. In this sense, it is more "theoretical", leading to testable implications that can support or reject these laws. 

Koopmans refers to these separate approaches as the "Kepler stage" and the "Newton stage", where Burns & Mitchell's analysis of business cycles lies squarely in the Kepler stage. Koopmans writes that

"... the authors' scientific 'strategy,' in which measurement and observation precede, and are largely independent of, any attempts toward the explanation of economic fluctuations."

While the separation of the scientific study of celestial bodies into the Kepler and Newton stages ultimately proved quite useful, Koopmans is less optimistic that this approach will prove fruitful in economics. He continues:

"Nevertheless, this reviewer believes that in research in economic dynamics the Kepler stage and the Newton stage of inquiry need to be more intimately combined and to be pursued simultaneously. Fuller utilization of the concepts and hypotheses of economic theory (in a sense described below) as a part of the processes of observation and measurement promises to be a shorter road, perhaps even the only possible road, to the understanding of cyclical fluctuations. Such a course, in addition, promises as by-products greater insight into noncyclical and even nondynamic problems of economics."  

Why does Koopmans believe that what works in astronomy will not work in economics? With great foresight, Koopmans is skeptical of any credibility revolution in economics. See, Newton's theoretical development of fundamental scientific truths built upon not only Kepler's observations, but also Galileo's experimental findings. And, despite the great advances in economics in field and lab experiments and the credibility revolution (i.e., diff-in-diff), the inability to experiment (widely) in economics -- particularly in the study of business cycles -- is limiting. 

Koopmans states

"Newton's achievement was based, not only on the regularities observed by Kepler, but also on the experiments conducted on the surface of the earth by Galileo. Economists are not in a position to perform experiments with an economic system as a whole for the sole purpose of establishing scientific truth (although deliberate changes in parts of the system have been undertaken at various occasions for other than scientific purposes, and have incidentally added to our information). It is therefore not possible in many economic problems to separate 'causes' and 'effects' by varying causes one at a time, studying the separate effect of each cause - a method so fruitful in the natural sciences."

On the flip side, economists do possess "more elaborate and better established theoretical models of economics behavior" according to Koopmans. This combination of theory with empirical analysis is necessary in economics in order to discover any sort of fundamental truths. Absent such "structural" analysis, identification of structural relations is not possible. Koopmans writes

"The mere observation of regularities in the interrelations of variables then does not permit us to recognize or to identify behavior equations among such regularities. In the absence of experimentation, such identification is possible, if at all, only if the form of each structural equation is specified, i.e., in particular, if we can indicate the set of variables involved in each equation, and perhaps also the manner in which they are to be combined. In each case, a preliminary study of the system of structural equations held applicable is required to decide whether the specifications regarding any particular equation are sufficiently detailed to permit its identification. Without such identification, measurement of the structural equation involved is not possible, and should therefore not be attempted."

So, getting back to the point of this post, I am not suggesting that we all become structural empiricists; I'll leave that argument to Sal. However, I think some movement in that direction is necessary. [Note: I am also guilty of this too. It's just that this has bothered me for a long while and Koopmans gave me an excuse to put it down in cyberspace.]

My take on the current state of (the majority of) empirical micro these days is that it is completely devoid of economic theory. We joke about it, but we all went through the gauntlet of learning calculus, linear algebra, differential equations, and - gulp! - even real analysis. We then slogged through a year of micro theory and a year of macro theory. We then went further into field courses, surely reading some theoretical papers or, at least, papers with theoretical models. However, the majority of empirical micro papers could have been written without this torture, I mean fancy book learnin'. 

While economists are infamous for the fact that we can and do study ... everything ... what unites us is a set of skills and a particular approach to problems. If we don't use those skills, what are we doing? Not only are we likely not learning anything according to Koopmans, but we are no longer unique among social scientists. 

Even if one doesn't fully subscribe to Koopmans' view that the Kepler stage holds little promise by itself in economics, the Kepler stage should not be the end of scientific inquiry. Even if some empirical micro papers are useful in documenting empirical regularities, the next step -- and a necessary one -- is codifying such regularities in fundamental laws of economic behavior. To some extent this is being done in the context of internal vs external validity. However, such research is few and far between, precisely because it is difficult and many of us would very much like to forget our theoretical training. We should resist that temptation. 

OK, that's enough spewing on that. 

I also mentioned above that Koopman's analysis also relates to my last post on The Great Divide that I see between empirical and theoretical econometricians. Without hammering the point to death, I think Koopmans’ article also provides a way to think about that issue.

The type of econometric research that I see missing falls under the Kepler stage: the documentation of empirical regularities. This time not about celestial bodies or business cycles, but about how econometric methods work in practice. Now, I know I just argued in favor of Koopmans' point that economics should take the Newton stage more seriously. This suggests a great importance be placed on econometric theory to unearth fundamental truths through the rigorous documentation of assumptions and proofs. While this is always important, Koopmans' emphasis on the Newton stage in economics was motivated by the inability of Galileo to contribute to economic study due to a lack of experimentation. However, in econometrics, such experimentation is easy and common: Monte Carlo studies are exactly that.

So, just like with the study of celestial bodies, discovery and the development of advanced methodological knowledge requires the input of the econometric Keplers, Galilieos, and Newtons of the world ... not just Newton.

As always, thanks for reading. Next time I will stop proselytizing and get back to some good, uncontroversial econometrics. Promise. It's out of my system.