The idea that there is something like an efficient market in scientific ideas (EMISI), supporting a ruling 'paradigm,' is very dangerous in the policy sciences. Even if we assume that scientists are individually pure truth-seekers, imperfections in scientific markets can produce non-epistemic (and epistemic) externalities (recall here, including criticism of a famous paper by Aumann). EMISI provides cover for 'The Everybody Did It' (TEDI) Syndrome (recall here). With Merel Lefevere, I have been exploring in what circumstances the presence of TEDI Syndrome is indicative of collective negligence (or a negative externalities). One possible consequence of our approach is that those scientists/institutions that interface with policy should seek out critics and critical alternatives to the existing paradigm. Jon Faust, an economist, sometimes acts as such an in-house critic at the United States Federal Reserve (the Fed) and the Riksbank. Two of his relatively non-technical papers (here and here) prompted this post.
Central Banks rely, in part, on models developed by academic economists to set monetary policy. Yet, Faust notes two problems in the way the intellectual supply-chain works: (i) there is almost no venue for "high-level conversation" about "academic work and its relation to actual practice." (53) (ii) State of the art models are often applied without full knowledge of all their possible consequences in the real world because these models models "have substantial areas of omission and coarse approximation" (55) In light of (i) and (ii), Faust's aim (iii) is to help central bankers and the modellers develop "a formal literature on best methods and practices for using materially flawed models in practical policymaking," (55) or "how to make the most responsible use in policymaking of what we now know." (60) My first reaction was, 'it is about time;' my second, more generous response was warmth in my philosophical heart that Faust is engaging in philosophy of scientific methodology and non-ideal regime/institution construction. His main idea is to adapt a kind of policy protocol from a literature that "goes under names like “human relevance of animal studies” and “interspecies extrapolation” (57) in the practice(s) of Toxicology.
Crucially, as Faust notes, in toxicology there is an ongoing process of evaluation of MOA, and a feedback loop between policy and models. In practices that engage with complex systems (medicine, nuclear power stations, air-transport, military, etc.) such on-going critical self-evaluation and evaluation by outsider is the norm. (As I have pointed out elsewhere a serious problem in economics is the desire to confirm models rather than to stress-testing them.) From the role that Faust is reportedly playing, central banks are starting to see the wisdom of such a process.
In his paper, Faust rightly notes that before a model is adopted there is a need to have some understanding of the transmission mechanisms in a variety of circumstances. Given that it is hard to run genuine macro-level-monetary-experiments, this ought also involve lots of simulations within the model world, and experimentation within the scientific lab, etc. Recall my advocacy -- here and here -- for regular drilling and emergency simulations with actual central bankers, market participants, and regulators (etc.) [That may be too adventurous for Faust.] So far so good.
But in his methodological stance, Faust does not emphasize the significance of developing a fine-grained understanding of (a) the complex interplay among discarded theory/models, past policy behavior, and 'the economy.' That is, we need to develop history of economic/monetary ideas alongside economic/monetary history in order to have a range of 'natural experiments' for (the evolution of) transmission mechanisms. Not to mention that past models may perform better in some limited domains than our best current ones (say, because [recall] of Kuhn-loss). (I have argued the case here.) Historical approaches may also help temper expert over-confidence. However, Faust practices such a historical approach so that there are "lessons of history." (46) (In his paper these are organized around the experiences with models in the 1960s (apparent "long-term success") and the 1970s ("tragic economic events") (45).)
Now Faust also offers a third step: how should we think about policy when no plausible transmission mechanism is at hand? That is, when we are dealing with knowledge that unknowns exist, or old-fashioned Knightian-Keynesian uncertainty. How should central bankers act when they know the models are materially flawed? Faust's response is essentially Savage's MaxMin, "maximize the benefits and minimize the risk of catastrophe." (46) Another response is to deflate the significance of models, and to restrict their impact to letting them "organize our thinking and refine our trained intuitions." (63) But, of course, if intuitions are trained on particular models, they may become less good at spotting systematic blind-spots in the models (this is not over-determined), re-opening the door to TEDI syndrome. This is why Lefevere and I advocate the deployment of a portfolio of incompatible models. Even third-best or false models may make particular features of reality stand out (or expose flaws in second-best models).(Recall Wimsatt's classic paper.)
Central bankers and their academic advisors are struggling with issues familiar to philosophers concerned with scientific methodologies and practices. Not to put a fine point on it: here's an area where our collective, philosophical expertise may contribute to improving thinking more clearly (and more fairly) about some of the relevant issues. (Perhaps Templeton wants to fund this?)