Policy penance

The last few posts haven't worked out so well, that's for sure. After a too-grumpy reaction to Alan Blinder's review,  I wanted to say something nice and find common ground with the "what's wrong with macro" articles and even Krugman's posts. In doing so I was much too quick and superficial in characterizing what's going on at high levels of our policy institutions. The only result was that  I managed to annoy all my friends and colleagues at the Fed, IMF, and so on.

As penance, I'll try a blog post that more accurately characterizes the interaction of research and policy, "Keynesian" and modern economics, and so on, as I see it.

If we look one step below the political level, for example at the FOMC minutes and what research staff are up to at institutions like Fed and IMF, you see a very sophisticated interaction between the ideas of modern economic research and policy. The FOMC minutes and speeches by board members (all easy to find on the Fed's website) are a great source. The FOMC seems, to an outsider,  like the world's highest-level debating club on modern macroeconomics.

On many of the dividing lines between traditional Keynesian and modern economics, the policy discussion is decidedly modern.

Traditional Keynesian economics is above all, static. Consumption depends on income, today; investment depends on interest rates and animal spirits, today, and so on. This is in part its great success. At the time, people didn't have the tools to do dynamic intertemporal economics.

Now we do. Modern macroeconomics is, above all, intertemporal. People make consumption and investment decisions, thinking about today and the future. This intertemporal revolution started with the permanent income model, that consumption depends on expected future income, and continues to this day.

(Let me quickly stop a discussion that will spin out of control into quotations from Keynes and what he might have "really meant." What matters is what was in Keynesian models, used in policy for generations, and that's ISLM. Perhaps a deeper reading of Keynes -- or Marx, or Smith, or the Talmud -- might reveal some intertemporal poetry. But it really had little effect on how models are used or policy was done. So here "Keynesian" means ISLM-ASAD.)

Now, if you read FOMC minutes, Fed speeches, or talk to people at the Fed about policy, you will see that this intertemporal, expectation-focused approach resulting from the revolutions of the 1970s permiates the policy-making process. For example, "forward guidance" is the rage. It only takes one beer for the conversation to quickly acknowledge that QE likely worked as much by signaling low interest rates for a long time than it did by exploiting some sort of permanent price-pressure in Treasury markets.

More deeply, the Fed policy discussion recognizes that "expectations" are not the same things as "speeches by public officials." People have heard lots of promises before. Policy faces a deep "time consistency" and "commitment vs. discretion" problem, again part of the late 1970s revolution in macroeconomics. People don't believe promises made now, because they know the Fed may change its mind later.

This realization led modern macroeconomics to focus on policy rules, rather than policy actions, which we can chalk up as a second major break between traditional Keynesian analysis and modern macroeconomics. Rules can be written down, legal or constitutional constraints, or simply traditions long observed.  The best way to "manage expectations" is not to "manipulate" them with speeches, but instead to follow well-established rules -- even when you'd rather not.  Friedman's 4% rule from the 1960s had some of this flavor. Taylor's interest rate rule from the 1990s has it explicitly, and inflation target rules even more strongly.

Over the last 20 years, and especially under chairman Ben Bernanke, you could see the importance of transparent, predictable, rule-based thinking take hold. The early Fed was deliberately secretive and deliberately obscure. There was a time when they wouldn't even tell markets what the Federal Funds target was!  Since then, explanation of what the Fed is doing, why the Fed is doing it, what the Fed is likely to do in the future, and finally how the Fed will react to events in the future -- a "state-contingent" rule -- has become more and more important in Fed discussions.

This attitude took a little step back recently; the Fed tried to communicate that interest rates would rise when unemployment fell below 6.5%.  (Clarification below.) That promptly happened, unexpectedly, forcing the Fed to backtrack and say no, wait, we really want to look at broader labor market indicators.

Well, it's hard to follow rules ex-post. That's the whole point of rules! Adopting rules needs major changes in what we expect of policy too. If Janet Yellen had raised rates as "promised," then went to Congress to say "we proclaimed a rule, so we had to stick to it even though we and you both thought the economy way too weak to raise rates," you can imagine the howls.

I'm not here to criticize, or to opine on just how firm Fed rules should be. The point is that the Fed is conducting this debate at the highest levels, fully informed by modern academic research on the subject.

The discussion surrounding fiscal stimulus in 2008-2009 strikes me as having been a good deal less sophisticated and much more "old-Keynesian," involving static "multipliers," straight from a 1970s textbook. But that may be a poor example as it was done in a huge hurry, inside the Administration, and away from the kind of carefully constructed policy process that the Fed and other agencies maintain.

Fed speeches, even from chairs Bernanke and Yellen, are peppered with citations to academic work and staff work. More evidence of a tight connection between modern research and top level policy discussions.

In fact, one can quite plausibly complain that the Fed, IMF, and similar institutions are too close to academic research, or perhaps that academic research is too motivated by finding reasons for the latest policy idea. In no other area that I can think of, where important policy is made and there is a corresponding body of academic research, do people seriously propose to guide policy based on the latest, usually unpublished, and usually novel, research. We wait a while for ideas to settle.

As an instance, I've been a bit critical of the apparent distance between policy and research, and I've also been quite critical of the current generation of new-Keynesian models.  Well, John, make up your mind! On the latter view, I should certainly not advocate that the Fed tomorrow implement policy based on the latest large-scale estimated new-Keynesian model. But on the former, I have to admit that sort of thing has been standard academic research for 20 years now. Most bloggers mean "the Fed should pay more attention to my research," but I won't -- yes, research has to settle before being used for policy.

The hard fact is that economic models are quantitative parables, not explicit and complete descriptions of reality. The step of understanding the model's "intuition," "basic message," and so forth is devilishly hard. That's why we have such deep verbal discussions about economic models, where we imagine physicists just test them and are done.  (They don't, but that's another story.)

An example. In 2012, Mike Woodford gave a subsequently famous paper at the Jackson Hole conference, arguing that in new-Keynesian models, "forward guidance" that interest rates will be kept low for longer than usual can create stimulus at the zero bound. This paper generated lots of controversy, not least from me. Do people believe such promises? Yes, we can write a model in which the Fed announces a new new rule, and everyone believes it. But does the world work that way? More deeply, is the underlying model right? For example (of many) is my complaint that it assumes the Fed threatens to blow up the economy a big problem, or just an easily-fixed technical simplification?

Be it as it may, this was first-rate academic research, by a first-rate academic, and it evidently profoundly influenced the policy debate.

A similar story occurs daily on how the Fed should think about "secular stagnation," "macroprudential policy," "pricking asset price bubbles," long term labor force participation, inequality, and so on and so on.   As a concrete example, Jeremy Stein when on the board gave an excellent series of thoughtful speeches on the relation between monetary policy and financial stability. Here is a good one, bristling with citations to academic research.

The trouble with all this is not a lack of contact between Fed and academic research. The trouble is, if anything too much! The basic "trouble with macroeconomics," circa 2014, is about the same as the "trouble with physics," circa 1790. We know a lot. But there is so much we don't know.

All of this work, really, is about "frictions" in the economy. "sticky prices," "sticky wages," whatever they really mean, "financial frictions," leverage constraints, collateral constraints, temporarily segmented markets, "liquidity," as much in the eye of the beholder as smut, and so on and so forth. The "trouble" with macroeconomics is that we're really only beginning to figure out what all this really means and how it works.

To the similar complaint that there is "too much math" in economics: no, there is too little! We don't have the tools to model, understand, and control all these hazy ideas that seem to matter when we look at the world.

I think there are some unfortunate heirs of the old-Keynesian tradition still at work in policy-making, however.  It is natural that they are there, but I think it will be good when they vanish.

First, since models are quantitative parables, and it takes a long time to digest what they really mean for a given situation, it is natural that the top level of policy makers to continue to digest new work in an old model. "Oh yes, I see, this model really means 'aggregate demand' is low and stimulus will raise it.  Now I get it." That's a perfectly normal reaction.

Second, and deeper, the Keynesian policy tradition left a strong desire for activist, discretionary, "what do we do today?" sorts of answers.

I long ago sat at a hilarious academic advisory meeting at a Federal Reserve, at which the bank president asked, bottom line, whether we thought the Fed should raise, cut, or leave alone the funds rate. Academic after academic gave beautiful speeches about the right policy rule. (Me, an ode to price level targets rather than inflation rate targets.) The poor exasperated president said, "that's all very nice, but what should we do now?"

This call for action, for activist discretionary response, is at the core of Keynesian economics. It's very very hard to talk about rules and institutions rather than actions. And the core answer of modern intertemporal economics is to unask the question.  But people expecting a daily discretionary decision just don't want to hear about the rule.  In this regard, the policy mindset still is decidedly old-Keynesian.

As a counterexample, consider asking the question "what should monetary policy do about unemployment" in the 1800s. There was no Fed. "Monetary policy" consisted of the gold standard, implemented by the Treasury.  The answer would be, "the price of gold is $20 per ounce. What's your question?" I'm not (!) saying that's the right policy, but it is a pure example of a rule rather than activist discretion.  A serious discussion about a rules-based Fed would start by canceling the regular FOMC meetings and the economic review. That just presupposes that the whole process is to come to a discretionary decision.

Third, economic policy discussion seems to ignore the tremendous lack of knowledge we have over basic cause and effect mechanisms, even the signs and causal channels of effects let alone magnitudes. Policy discussions jump to exploiting the latest friction before the ink is dry.

But by and large policy decisions don't, and are quite conservative about implementing new ideas. Expectations, rules, discretion and commitment, are from the late 1970s!

When things are ambiguous, you stick with what you've got. Keneysian orthodoxy ruled for a long time. Even if it's shown to be wrong, replacement models are so different, still so untested and unrefined, continuing to use basic Keynesian intuition is a natural response to ambiguity.

Fourth, I will complain a bit about how much academic research produces answers to support desired policy rather than the other way around. Stimulus came on the political landscape, and a hundred papers are written about how stimulus might work. The one thing I will gently chide some people (not all!) at the Fed for is how often staff reports come up with the number that the chair wants to hear. This is particularly true when introducing "frictions" to models. Yes, this or that friction might justify a policy. But if you really think sticky wages are the problem, why write articles justifying central bank intervention, and not one on how wages might be profitably unstuck?

But these minor complaints aside, as I look at the layer of policy process just below the headline political appointees, and just above the silly stuff in the opinion pages of the New York Times, really the interplay of academic ideas and policy is about as healthy as I could make it. (Obeying the rule that one has to be evenhanded about all research, not just my research!) A residual, fading, back of the envelope Keynesianism is pretty natural.  And the good ideas of modern, intertemporal, people-based, budget-constraint-disciplined, economics are being digested and slowly implemented.

Next, this whole new Keynesian vs. old-Keynesian thing.

Update: A correspondent points out that the actual FOMC statement is more nuanced,

[The committee] currently anticipates that this exceptionally low range for the federal funds rate will be appropriate at least as long as the unemployment rate remains above 6-1/2 percent, inflation between one and two years ahead is projected to be no more than a half percentage point above the Committee’s 2 percent longer-run goal, and longer-term inflation expectations continue to be well anchored. 

But the perception of a rule and backtracking was real.

Update 2: I see from blogger's trackbacks that Paul Krugman covered this post. I'll keep to the usual don't-respond-to-Krugman policy, with one exception. Krugman writes:

Well, at least Cochrane now concedes that Woodford isn’t stupid. Progress!

I have never, ever, anywhere, in writing, in words, in thought, in insinuation, or veiled reference, said that Mike Woodford is "stupid.." I have never said or written that anyone is "stupid," "mendacious", "mendacious idiot," "evil," "corrupt," "hack," or any other ad-hominen attack or insult.  Google all you want, you will not find any such quote. Yes, I mercilessly skewer bad ideas, but never people. To me that is the most bedrock ethical principle of intellectual honesty.

Mike Woodford in particular has been a friend and a colleague for many years. I have learned a lot from him, I have benefited enormously from his comments on my work, and I've read his in detail.  I taught PhD classes from his book.

The charge that I don't know or haven't read or recently discovered Woordford is laughable. I spent about 5 years of my research life writing "Determinacy and Identification" which is nothing but a careful dissection of new-Keynesian economics as distilled in Mike's book. I may be wrong, but not out of ignorance.

Mike, if you're listening, I can't control the vitriol that the New York Times sees fit to print, so all I can do is post a correction to my humble blog -- Krugman is making this up out of thin air.

Update 3: Due to an avalanche of hateful comments from Krugman's choir, I've turned off comments on this post.