Austrian vs Post Keynesian explanations of the business cycle: an empirical examination

Abstract

Neoclassical economists posted many mea culpas after they completely missed the Financial Crisis of 2008. Some heterodox schools of thought, however, claimed to have seen it coming. Among these were the Austrians and Post Keynesians. There was a surge in references to Austrian Business Cycle Theory in both the scholarly and popular press, while Post Keynesians like Steve Keen took to the blogosphere to warn people of the coming crash. One is led to wonder, however, if both can truly lay a legitimate claim to such accurate precognition when their theoretical underpinnings are so radically different? Can it be that deviations of the actual from the “natural” rate of interest were to blame at the same time that Keynes/Kalecki-style boom-bust cycles were emerging? To answer this question, an empirical test is run comparing the explanatory power of models based on each theory. Rather than limit this to just the period around the Financial Crisis, the test covers 1962 through 2020. The overall results are then compared, as well as the fit just leading up to and through the recession of 2008Q1 to 2009Q2. The results suggest that the Post Keynesians have a considerably stronger case than the Austrians.

Keywords:

• Business cycle
• financial crisis
• Austrian
• Post Keynesian

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Bismans and Mougeot actually imply in their paper that the regression may have specification problems (Bismans and Mougeot 2009: 254). The fact that other empirical tests of the Austrian model set those ratios as the dependent variable and Spread as the independent is further evidence that it is a mistake to include them as separate independent variables (see for example Luther and Cohen 2014 and Mulligan 2006).

2 https://www.newyorkfed.org/research/policy/rstar

3 The results of the Breusch-Godfrey LM tests for serial correlation are shown in Table A2 in the Appendix.

4 Both LvsS and PCEst2 have the wrong sign, but this is irrelevant given that they are statistically insignificant from zero.

5 The use of that particular interest rate was a function of the specifications used by Arestis and Gonzalez-Martinez (2016), Bachmann and Zorn (2020), Barradas and Lagoa (2017), Carruth, Dickerson, and Henley (2000), Iyoda (2005), Kothari, Lewellen, and Warner (2014), Roberts (2003), and Stockhammer and Grafl (2010).

6 Ideally, the data from which this variable was derived would be a forecast for profits from investment, per se, rather than overall profits. However, no such data are collected and so this must serve as a proxy.

Additional information

Notes on contributors

John T. Harvey

John T. Harvey is the Hal Wright Professor of Economics in Texas Christian University.

Khanh Pham

Khanh Pham is an Honors Economics Major in Texas Christian University.