Austrians and Econometrics

A common objection to the Austrian approach of economics is that it's tautological, full of arm-chair philosophising and anti-empirical. No smoke without fire, there's some merit to these critiques, but they mostly stem from a faulty understanding of what Austrian Economics mean. As for the last charge of being anti-empirical, few challenges produce as avid within-group clashes between various Austrians. A fan of both emprics in general, and econometrics in particular, I must reconcile those points with the many objections against using math and statistics for doing economics and economic history found in Mises, Rothbard or occasionally Hayek, as well as more modern Austrians such as Salerno.

Let's start with some examples. In most of Mises' work, the "anti-empirics" position comes forth quite strongly, particularly so in Theory and History and Human Action's more theoretical chapters (see for instance chapter 2, here, here and here – or here, from chapter 16); experiments cannot give us economic knowledge, history cannot give us any "general rule, principle or law" (p. 41); in social life there are "no constant relations between magnitudes", which means that measurement and aggregation is impossible and "economics can never become quantitative" (Theory of Money and Credit, p. 419, see also pp. 187-194). Or this one, from the introductory chapter of Theory and History:

every quantity that we can observe is a historical event, a fact which cannot be fully described without specifying the time and geographical point. The econometrician is unable to disprove this fact. (p. 10)

This strong position comes across more clearly in Rothbard's Preface to Mises' Theory and History:

it is impossible to test [economic theory] in any way by checking its propositions against homogeneous bits of uniform events. For there are no such events. The use of statistics and quantitative data may try to mask this fact, but their seeming precision is only grounded on historical events that are not homogeneous in any sense. Each historical event is a complex, unique resultant of many causal factors. Since it is unique, it cannot be used for a positivistic test, and since it is unique it cannot be combined with other events in the form of statistical correlations and achieve any meaningful result. (p. xvii)

One of the disputes I have with Dr. Salerno, four offices down the hall, is his introduction to Rothbard's brilliant History of Money and Banking in the United States (particularly pp. 8-11) where Salerno states that the econometric invasion of economic history has two major flaws:

1. It limits the scope of of historical inquiry to areas more easily quantified, so the "new economic history is not history in the traditional sense of an attempt to 'understand' the human motives".
2. There is circular reasoning between "theory" and "empirical data": the data confirms that our theory is correct, but the interpretation of data is correct only because the theory tells us so. As Salerno explains, "it is unclear whether theory is the explanans or the explananad in historical research" (p. 10). 

Concurrent with this seemingly anti-empirical interpretation of Austrian thinking, there has always been an interest in – and a current pushing for – more use of empirical methods, at the very least to show the empirical applicability of theories and ideas. There is, for example, an entire literature on the empirics of the Austrian Business Cycles Theory, which Salerno himself has fruitfully contributed to (see my essay 'Measuring Mises' for a short intro to the empirics of the ABCT, or countless other scholars who have attempted similar things: Luther & Cohen 2014; Young 2011, 2005; Mulligan 2006; Keeler 2001, Bismans & Mougeot 2009. When we start pushing Austrians on those very strong anti-empirical claims, we find that they (we?) are not generally opposed to the use of statistics or empirics. In fact, one part of Mises' own guidelines for how to do history (see Ultimate Foundations of Economic Science, p. 45) almost explicitly relies on descriptions of the real world, where emprical data is crucial:

1. Value Judgment made by individuals in the past
2. the Ends they desired or hoped to achieve
3. the Outcomes of their actions (which may or may not coincide with (2))

For establishing the third part, econometric techniques, statistics and time series can very much be of help, something no Austrian disputes. My understanding is that Austrians' instinctive rejection of econometrics is a reflection of three serious problems with how certain econometricians apply this tool: 

1. its use for prediction. Since the social world depends on human action, which in no way is restricted by quantitative relations or proportions experienced in the past, extrapolating trends or making point-forecasts using historical correlations – as is a component in almost any econometrics class – is completely meaningless from an Austrian perspective. 
2. The very strong assumptions sometimes required for certain econometric reasoning, see for instance the common trends assumption for using difference-in-differences. 
3. The hubris of trying to capture everything of a historical event in times series or panel data we have access to. The stand-in proxies we sometimes use for non-quantitative data may only partially reflect the whole impact, and pretending otherwise distorts history.  

Suitably enough, I stole the picture from an INET article
arguing for reform of econometrics.

Keeping that critique in mind, which any sensible econometrician does, Austrians gladly give you their seal of approval. If you wanna use econometrics to point out mistakes in other people's historical arguments, to establish or describe some historical relationship or see how well a theory or argument fits the data (or as Austrians say illustrate rather than test theory, a difference which sometimes amounts to nothing more than semantics), we're on board.

Austrians are not "anti-empirical" or "anti-mathematical", even though it sometimes sounds that way. If you scratch us enough or push back against certain strong propositions, we fall back to a position advocating sensible use of data, a reasonable amount of humility in what econometric models can show, and a healthy skepticism of point-forecasts or predictions in general.