Thursday, November 13, 2014

Rules versus Discretion in Macropolicy

Economists really like policy rules.  Remember the monetarist bubble of the late 70s?  Just follow a rule for steady expansion of M2, and let the rest of the economy take care of itself.  That didn’t turn out very well, partly because no single measure of the money stock is tightly linked to the outcomes people care about, like inflation and output.  So then we had inflation targeting (and still do in the eurozone), but it turned out there wasn’t a fixed long run NAIRU (non-accelerating inflation rate of unemployment) like we were promised, and simple inflation targeting was leaving too many outcomes of concern unmanaged.  More recently we’ve had the Taylor Rule, targeting inflation and unemployment.  Like the other rules, it aspires to robotic implementation: just multiply the output and inflation gaps by their corresponding coefficients, and out spits the central bank’s policy interest rate.  Simple!

Unfortunately, the evidence is piling up that this formula doesn’t do the job either.  Without endorsing their particular policy shifts, it’s clear that some central bankers, along with many economists, are now worried about the impacts of monetary policy on the potential for asset price bubbles.  Forestalling bubbles wasn’t a feature of policy prior to 2008, but now it has grabbed a lot of attention.  Meanwhile, it is clear that the Fed is unhappy with the headline unemployment rate as an indicator of labor market conditions, especially since stagnant wages seem to signal continuing slack.

The problems with policy rules are many.  The goals of macropolicy—the things we care about, like low unemployment, price stability, financial stability, satisfactory growth—don’t all move together, and their co-movements change over time.  The usefulness of the available indicators for progress on these goals fluctuates as well, since there is no single, perfect measurement of any of them, and the relationship between different indicators keeps changing.  What, for instance, is full employment, exactly?  Is it the same today as it was five years ago?  How should the target reflect  the changing composition of jobs, wage rates and labor force participation?  In addition, the effectiveness of policy instruments, like open market operations and fiscal balances, is contextual and uncertain; for a run of years slight alterations in central bank bond purchases can have powerful impacts on expectations and economic performance, and then you are in a zero lower bound world in which all sorts of exotic measures—the various types of quantitative easing—go mainstream.

So why are economists enamored of policy rules?  One reason is that they are seen as more efficient, since policy impact depends on how much sway it has over the public’s expectations of future variables, like inflation and aggregate demand.  Rules are favored, since they ostensibly remove uncertainty from policy stances.  You can trust the central bank or fiscal authorities to not reverse tomorrow what they’ve announced today because they’re just following rules that govern their actions, and the rules don’t change.  That’s a nice theory, but it doesn’t make the problem of policy credibility disappear; it just loads it all onto the commitment of policy-makers to follow whatever rule is currently in force.  But as we’ve seen, the rules have never performed in a predictable, satisfactory way, so simply pledging allegiance to the latest iteration doesn’t remove the rational doubt that today’s rule will be superceded by tomorrow’s, with corresponding skepticism toward policy consistency.

A darker motive is distrust of “politics”.  If we don’t constrain authorities to follow fixed rules, won’t they do stupid things under the influence of whatever special interests have captured them?  For some time the ruling fear was populism: if we don’t have ironclad policy rules, the temptation will always be for monetary and fiscal authorities to reach for short run, unsustainable increases in incomes.  The result will be periodic inflationary surges accompanied by boom-and-bust cycles whose harm vastly exceeds the momentary benefits of populist expansion.  (This was what Timothy Geithner no doubt had in mind when he told Christina Romer back in 2010 that fiscal stimulus is “sugar”.)  Ironically, however, the overriding political failure post-2008 has not been populism but its opposite, the craving for investor “confidence” by way of austerity.  Does this mean there is now a reason for the expansionistas to demand their own policy rule?

But rules don’t bypass politics; they are politics.  There are, after all, many candidate rules, and all of them have a tenuous relationship to how real economies function.  What’s the point of making rule selection rather than policy selection the object of political contest?

Make these arguments in the company of economists, and you are almost certain to hear, so wise guy, what’s your alternative?  If there are no rules, doesn’t this mean policy-makers are free to follow whatever absurd theory supports their own prejudices?  This is a bit like the claim, variously attributed to Dostoevsky and Nietsche, that if god is dead all is permitted.  But just as moral reasoning is possible without a sacred text, so is rational macropolicy without fixed rules.

It should be obvious, in fact.  Economies are extremely complex, evolving systems.  Interventions have uncertain consequences, and what holds at time A does not necessarily hold at time B.  Even observation is uncertain, and the relationship between what you think you see and what’s actually going on can change unexpectedly.  But this is also true for the natural world, and in the domain of restoration ecology and similar fields we have the paradigm of adaptive management.  This is an approach that takes complexity and evolution as starting points, emphasizing the role of learning and the need for flexible decision-making in response to ongoing feedback.  I would argue that, whether they know it or not, central bankers and other policy authorities are already operating in that mode.  The god of policy rules died some time ago, and they have no choice but to weigh data according to their current understanding and reconcile themselves to the error part of trial-and-error.  Bringing transparency and open debate to this process will make it better, and this means dropping the pretense, finally, that authorities can or should follow fixed rules.


Gobanian said...


Bruce Wilder said...

Yes, the economy is complex, being a decentralized system in which the players are acting strategically, so all the apparent structures are merely contingent, subject to strategic subversion or transformations.
The policymaker pursuing a general or public interest in defending the integrity of the system of play may well find policy rules convenient. Many of the arguments in favor of policy rules focus on their advantages in arbitrating a strategic competition. What does not make much sense -- as your argument makes clear -- is imagining that rules, that may work well as a tactical means, can substitute for strategic discretion.
The great fault in the kind of macro-economics that focused on evaluating, say, the Taylor Rule, was not the rule, per se, but the focus. Imagining that prudential bank regulation or housing bubbles need be no concern of the monetary policymaker was a form of insanity. That's not about rules, per se, it is about a style of theoretical abstraction that substitutes imaginary worlds for the real one.
What macro-economics desperately needs, which I think your post almost inadvertently highlights, is a theory of institutional and systemic corruption.