Lynne Kiesling
The title of this month’s Cato Unbound, “The Empirics of Austrian Economics”, makes it sound more like inside baseball than it really is. The valuable discussion does have some elements of insider talk, but most of the exchange is externally focused, and as such is well worth reading if you are interested in economics but aren’t sure what all of these distinctions and differences are among various schools of thought. And if you’ve heard the label “Austrian economics” tossed around but aren’t sure you know what that means, this exchange is worth your while – not because you’ll emerge with a pat, concise answer to that question, but because you’ll have experienced a rich conversation about it.
At its core, this set of essays discusses what constitutes good economics, regardless of the labels one attaches to this group or that group. As in other schools of thought within economics and in other fields, different people hold different interpretations of the meaning of earlier works, place different evidentiary weights on different methods of persuasion, and bring different biases to those interpretations and weights. In addressing the specific question of whether or not Austrian economics is “sufficiently empirical”, these essays grapple with important questions of inquiry, method, evidence, and persuasion. In some ways the simple response is, it depends on what you mean when you use the word “empirical”.
Steve Horwitz’s lead essay sets up the issues, providing a primer that explains but does not rely on the unique terminology (such as “praxeology” and “catallactic”) that has alienated other economists in the past and unfortunately provided a means for isolating Austrian economists (please note here that I am not criticizing the concept of economics as the study of purposeful human action, but rather the effect of the rhetoric). When I think about empiricism and the Austrian perspective, the distinguishing aspect that I think is most important is subjectivism and its implications. Steve notes
Subjectivism also explains Austrian skepticism about statistical correlation being the privileged form of empirical evidence. It only provides correlation, and to provide causation requires a theoretical explanation. If such explanations must start with actors’ perceptions of the world, then forms of empirical evidence that capture such perceptions would be at least as useful. Austrians therefore frequently turn to primary source material and interview and survey work as well as quantitative data to tell a complete story of how a particular economic phenomenon came to be and functioned. How did actors perceive their options and constraints and what sorts of consequences emerged from their choices? That is the fundamental narrative framework for Austrian empirical work, with economic theory providing structure to the story.
This observation puts a very fine point on the complementarity of (a priori, deductive) theory and empirical analysis, while highlighting some limitations of statistical analysis, as described above. This is not a new problem, of course; we’ve been grappling with it at least since Paul Samuelson and Milton Friedman wrote about revealed preference and “as if” interpretations of abstract, “perfect rationality” assumptions in theoretical models. But what’s changed for the better in the past few years is the loosening of the methodological hegemony of econometric evidence as the only acceptable empirical evidence to test a hypothesis. As Bryan Caplan points out in his response to Steve, over the past 20 years behavioral economics has broadened the varieties of evidence used to test hypotheses (and in later comments Steve endorses that idea); I would add experimental economics to that list, which all of the authors do implicitly. I think this is one of the most salutary complementarities in economics today: the combination of (1) Austrian theory of “how markets and their institutions enable humans to coordinate and cooperate in a world of subjective and fragmented knowledge and structural uncertainty” (see also Pete Boettke’s excellent entry in the Concise Encyclopedia of Economics for a good summary), (2) behavioral economics testing of hypotheses about cognitive limitations to rationality that complement the knowledge focus in Austrian theory, (3) experimental economics and economic history exploration and testing of hypotheses about institutions that enable or stymie such coordination in the face of diffuse knowledge and other cognitive limitations, and (4) new insights from other fields, particularly other social sciences and neuroscience.
George Selgin’s response had many elements worth highlighting, but I’ll limit myself to one in the interest of space: there’s a lot of potential empirical work to do on economic puzzles in the vast gulf between the poles of “needs no empirical validation to be “true”/is only “true” if shown via multiple regression analysis:
But Mises’s belittlement of statistics, and econometrics especially, … overlooks the fact that there is something between merely checking to see whether an occurrence satisfies all of the “contingent claims” needed to apply a particular theory to it, and pretending that you can construct economic theories using regression coefficients. What’s in between is trying to arrive at an informed estimate of just how much of any observed phenomenon an applicable theory explains and, when there are several equally applicable theories, their relative worth.
Similarly, I had general agreement with Antony Davies’ response that Austrian theory, mathematical modeling of theory, and statistical testing of theory can be complementary, although I think he overstates the benefits of mathematical modeling relative to analytical narrative with respect to vagueness of logic.
So how do we do good economics? I think Steve’s framework from his response to George points to an approach that embraces inquiry grounded in economic theory, and pluralism with respect to what constitutes empirical evidence (with a name-check of Schumpeter that made my heart go pitter-pat):
The Austrian tradition from Menger to Mises to Hayek to Kirzner shares a broad pre-analytical vision (to use Schumpeter’s helpful term) about the nature of economies and importance of subjectivism, knowledge, and spontaneous ordering processes. From that vision comes a series of more specific substantive propositions about how this vision manifests itself in economic analysis (see Boettke’s list, which I noted earlier). If we want to make use of those propositions to understand real-world phenomena, it should be those propositions that guide us as to what sorts of empirical observations and evidence are needed to demonstrate the usefulness of Austrian economics in rendering the world intelligible.
More precisely: the world is full of puzzles that we do not understand. Austrians think those analytical propositions are necessary (though not sufficient) for good explanations of those puzzles. What sorts of arguments and evidence are needed to offer persuasive accounts of those puzzles in ways that render them no longer puzzling? Different puzzles will require different propositions, which will require different methods and empirical evidence to render them intelligible.
All in all, a very thought-provoking exchange.