Should Congress legislate so that the Fed is forced to follow policy rules?

No.

But, as John Taylor alerted us in his blog, a new bill has been introduced that would have this effect, if passed.  I presume that this is one of those bills that no-one expects to pass, but is put before it to stimulate debate.  Debate is always a good idea, but even a tiny chance that it might actually pass is a cause for concern.

Why?

Even before the financial crisis, I would have said that the ‘Science of monetary policy’, to borrow a phrase from the elegant survey of modern sticky-price-monetary policy macro by Clarida, Gali and Gertler, had not progressed to the point where anything could be gained by attempting to legislate for such a rule.

Why not?

Because even then there was too much disagreement encoded in the controversies about the transmission mechanism of policy to produce a consensus about what such a rule would look like.

Even if such a rule were chosen by the Fed, and reported to Congress, everyone would rightly expect that the necessity to deviate from it practically every period would dominate the benefit from generating further predictability in policy.  And hence little or no further predictability would result.   If, by some strange quirk, the Fed found itself forced or tempted to follow such a rule, macroeconomic policy would surely suffer, and uncertainty would soon return as everyone speculated about the point at which a consensus would build to cast of the shackles of this legislation.

John Taylor and John Williams wrote a great survey of the applied theoretical work on monetary policy rules for the revised Handbook of Monetary Economics.  And if you read that you’d find that following simple policy rules like those John Taylor himself hit on in his 1993 classic paper [the credit for which in discussion amongst some afficionados is allocated equally across JT, Dale Henderson, Warwick Mckibben, and others] does quite a good job, across a range of models, in recovering the performance of the best possible policy in each model.

But, this range is still pretty small.  It’s confined to sticky-price, rational expectations DSGE [dynamic stochastic general equilibrium] models.  Many disputes within the DSGE community question the apparent attractiveness of this result. First, no-one seriously believes rational expectations should be used to assess such an important feature of policy design.  (Correction:  I don’t.  To say ‘no-one’ is probably wishful thinking).  There are many alternatives, and no settled one.  But I would bet my house that the ‘Taylor Rule or similar does great’ result would fall down if one substituted in many of these alternatives.  The result probably doesn’t survive the inclusion of financial frictions.  Woodford suggested that the Taylor Rule could be amended with the addition of a spread.  But can you imagine a central bank committing to any particular spread, or to responding to it in a fixed way, given all the vagaries and misunderstandings in asset prices and associated yields?  I can’t.  And we still have not stepped outside (essentially) representative agent models.  Heterogeneous agent DSGE work has yet to study carefully the design of monetary policy rules.  And….  I could go on and on making similar points in the same vein [the state of the art in macro doesn’t have a proper model of money yet, so legislating for the fine details of monetary policy seems ludicrous;  the state of the art in macro can’t generate financial crises yet;  how would legislation encode responses at the zero bound?  experience of other instruments at the zero bound (QE’s effect on yields) confounds most macro models……]

I’m all in favour of exploiting what wisdom we have on policy rules.  Most central bank compute their inflation forecasts assuming that they themselves will follow a rule (which might sound paradoxical, given that they intend not to), and it would be helpful for them to disclose these assumptions, so that forward guidance and yield curve talking could be made more concrete.  But taking this far as legislation seems premature.

Moreover, we should remember some of the background to John’s encouragement of this legislative charade.  He sees the performance of the US post-crisis as resulting from the deleterious effects of uncertainty about policy that come with a departure from rules-based policy.  This view, which he has held to for some years, always struck me as peculiar.  JT himself pioneered the basic outline of the modern sticky price macro model, yet his views conflict with how I see the policy prescriptions that flow from it.  Those would be: [inexcusably crude summary follows]  follow a policy rule that responds to forecasts of inflation and real activity [Bernanke artfully explained this modification to Taylor’s rule, given the long lags between policy and actions].  Deploy counter-cyclical fiscal policy in concert:  automatic stabilisers might be enough.   If the zero bound constraint binds, you will probably need extra stimulus from both monetary and fiscal policy, to be got from discretionary fiscal measures, and – not in the model then, but implied by it, and since incorporated – unconventional monetary policy.  Whether you have done enough or not can [I’m talking still from the model’s perspective here] be read by looking at measures of spare capacity, and inflation itself.  In the US, it’s fair to say that one could argue there was an undershoot.  Still looking at the issue through the spectacles of this model, we’d probably argue there was a case for a significant OVERshoot [Woodford argued this for some time, using basically Taylor’s model].

It seems likely – the Bloom, Baker, Davis work on policy uncertainty seems to confirm this – that there was significant uncertainty about policy, both monetary and fiscal.  But, returning to the model, this would be of the form:  we’re uncertain if Congress are going to agree to enough fiscal stimulus as prescribed by Taylor’s model;  we aren’t sure how the hell the Fed can stimulate the economy [as per Taylor’s model] with its instrument pressed against the zero bound.  And in the face of this uncertainty, [this isn’t in the model, but interesting work that modified it afterwards offers some support], the prescription is not, as JT is suggesting, a return to less stimulus, but MORE stimulus, until the uncertainty unwinds.  (Assuming the uncertainty weighs more heavily on demand than potential output, which, with a significant inflation undershoot and spare capacity, relative to the ideal moderate overshoot, seems a fair starting point).