1/ My new working paper @BristolEcFin : What is the link between microfoundations and the Lucas Critique(LC) ? A historical appraisal. Thread.
2/ [Spoiler alert] Two rival interpretations about the link: (A) the two issues are related; (B) the two issues are unrelated. Both interpretations are born in the 1980s and still coexist today within the DSGE approach. This is why the debate about the LC is still open.
3/ [Disclaimer: Because the LC meant (means) different things to different people, it is hard to find an encompassing narrative-wording. Even more true on Twitter: This is an attempt]
4/ My WP is part of a wider project with @Pinzon_Fuchs @AGoutsmedt and Matthieu Renault (#macrogang) on the history of the LC and the reactions to it. Our question: Why the LC (paper and concept) is such a celebrated cornerstone? Why it was (is) so important?
5/ As historians of macro, we try to answer this question with a broad, balanced perspective, as opposed to auto-celebrating or self-justifying narratives about “this was important because the argument is true and right” (“standard narrative”).
6/ In a previous paper, we reviewed early reactions (1970s-80s) by Keynesians (Solow, Modigliani, Blinder, Blanchard, Fischer, Klein, …). Our view is that they fought back on the LC by questioning empirical evidence and policy relevance. papers.ssrn.com/sol3/papers.cf…
7/ Keynesians: The LC is logically sound, but this is not enough: Empirical evidence of parameters’ drifts should be produced to validate the LC (F-tests, Chow tests).
8/ This is what I will call an ``empirical interpretation of the LC’’. The purpose of my new WP is to show how a competing interpretation arises simultaneously, which I call a “theoretical interpretation of the LC”.
9/ [Note that the point is not to show what Lucas's paper“really” was about (although I discuss this a bit in section 1 of the WP) but to show how the LC meant different things to different people.]
10/ Theoretical interpretation of the LC was championed by the RBC approach. It relies on the logic: non-microfounded models are not robust to the LC, while microfounded models are robust to the LC.
11/ Plainly, this suggests that Keynesian marginal propensities to consume c (like in C=cY) is a drifting parameter by definition because non microfounded; while objective discount factor beta is not, because by def. tastes are stable over time.
12/ Therefore, there is no need for empirical evidence to know which model is or is not robust to the LC. Critical account of RBC/DSGE pointed out a ``hijacking’’ of the LC by “microfoundations”.
13/ The empirical interpretation by Keynesians (and most econometricians, like Sims) precisely argue that LC and microfoundations are completely separated issues and the LC should be validated by empirical evidence.
13/ In the 2000s and early 2010s, the state of the debate within DSGE approach can be understood as a debate between the two rival interpretations. Some argue that DSGE models are robust to the LC because microfounded. Full stop.
14/ Those who disagree with this, find two arguments to support the view that DSGE are vulnerable to the LC (drifting parameters): (1) they are not enough microfounded; (2) they are despite being microfounded.
15/ The first claim results from the theoretical interpretation. Calvo pricing and Taylor rules are main targets for this criticisms.
16/ Second claim results obviously from the empirical interpretation. Even with microfoundations, models still exhibit large drifting in parameters (including those characterizing tastes and technologies).
17/ The issue with these rival interpretations is that they push for different directions to fix DSGE’s vulnerability to the LC: More microfoundations (in the Lucasian sense) vs. alternative microfoundations vs. “let’s agree that we would never be able to fix it”.
18/ Conclusion: The LC is far more complex than the its usual “weaponized” version to endorse or dismiss theoretical approaches or modelling practices. History of the LC brings back to the fore this complexity and show how and why the debate is still open.