Wednesday, August 1, 2012

Was Samuelson using the word "discretion" here in the same way that we do today?

John Taylor writes:

"First, go back to the early 1960s. The Keynesian school was coming to Washington led more than anyone else by Paul Samuelson who advised John F. Kennedy during the 1960 election campaign and recruited people like Walter Heller and James Tobin to serve on Kennedy’s Council of Economic Advisers. In fact, the Keynesian approach to macro policy received its official Washington introduction when Heller, Tobin, and their colleagues wrote the Kennedy Administration’s first Economic Report of the President, published in 1962.

The Report made an explicit case for discretion rather than rules: “Discretionary budget policy, e.g. changes in tax rates or expenditure programs, is indispensable…. In order to promote economic stability, the government should be able to change quickly tax rates or expenditure programs, and equally able to reverse its actions as circumstances change.” As for monetary policy a “discretionary policy is essential, sometimes to reinforce, sometimes to mitigate or overcome, the monetary consequences of short-run fluctuations of economic activity.”"

There is a lot to unpack in the phrase "rules vs. discretion". As Steve Horwitz has pointed out, you could have policy rules or you could have constitutional rules (or, he adds -  you could have legislated rules which are somewhere in between but he seems to suggest maybe closer to constitutional in this particular case). Rules could be internally or externally imposed. That usually lines up with the policy/constitutional distinction, although I suppose it doesn't have to.

But this passage raises a new dimension to the term.

The 1962 report refers to discretion as being "able to change quickly tax rates or expenditure programs, and equally able to reverse its actions as circumstances change".

DISCRETION! cries Taylor.

And that is indeed what the report calls it. But that sentence sounds an awful lot like what a Taylor Rule does! The Taylor Rule, after all, calls for quickly changing interest rates and open market operations and reversing it as circumstances change.

It is definitely different from what Friedman advocated, a k-percent monetary growth rate come hell or high water. And it's also different from the "rule" that comes out of a standard Barro-Gordon model. This is what Bob Murphy seems to be refering to when he says: "Let me put it this way: During a really bad economic crisis, following the rule is supposed to suck. That’s the whole point."

He's right in the sense that the good "rule" Barro-Gordon came up with was not accomodative. Does it follow that rules must be non-accomodative to really be "rules"? Is that "the whole point"?

If that's true I don't know what the argument for it is. John Taylor - certainly a proponent of monetary rules - thinks that rules can be accomodative and still be rules. Even tighter k-percent rules are going to end up creating more money in a crisis than outside of one, because a broad monetary aggregate like M2 is going to naturally shrink or at least slow in a crisis.

So we have yet another layer to this mess, which I guess I'm just pointing out to stress that this is not something that we should try to discuss with ound-bites. So in thinking about rules we have to mull over:

- Accomodative vs. "suposed to suck"
- Internally vs. externally imposed
- Constitutional vs. policy
- Implicit vs. explicit (I raised this before - knowing how mechanical Keynesianism was back then, I'm guessing the Report above, while it didn't state an explicit rule, had an implicit one in mind)


FWIW I was playing around with a Barro-Gordon model this morning after reading Bob's comment on his blog, which did raise questions for me. It's been shown that the result is invariant to different objective functions. I was a little curious if it was also invariant to the way you specify the determination of inflation. If anything interesting comes up I'll let you know - I plan on picking it up again tonight. It may come to nothing.


  1. Rules are good if you know what you are doing, and can make the condition part of the rule explicit. One problem with the recent stimulus is that the rule had no conditions, except a time limit. Without rules with conditions, like unemployment insurance, things would have been worse.

    On top of that, there was no agreement about the size of the stimulus. IOW, people did not know what they were doing well enough to write specific rules. There is no shame to that, but then discretion is necessary to adapt as things change.

    1. Well there's not agreement on the size of monetary stimulus. Does that mean that we can't write monetary rules?

      No, of course not. It just means some people are going to be miffed about the rule!

      Fiscal policy is a tricky case, because you're actually appropriating for something specific. Rules would obviously be great in theory, but that's not something a free people are going to give up. I would much rather have a tepid Roosevelt groping his way than a hard-core-Keynesian Hitler.

      I think it's important to distinguish between what Congress did and the argument for a fiscal stimulus. Krugman, for example, definitely had a rule in mind. He had a mechanical way to determine the output gap and he had a rule about what to do (if we are in a liquidity trap, then fill that mechanically determined output gap). The fact that Congress is dysfunctional is not an argument against the case for stimulus in and of itself.

      Now, Congressional dysfunction does have consequences. And Congress is the ultimate discretionary policymaker when it comes to economic policy. And as we all agree, that's bad. We try to avoid that by committing to automatic stabilizers, like UI. But if you think Krugman's arugment is a good one, and if you are unwilling to give up the fiscal discretion enjoyed by a free people, what is your option (aside from thinking up more automatic stabilizers, which I agree would be great)? You push for the closest to what the rule says as you can get.

      Is that saying the haphazard discretionary stimulus Congress gives us is good? No of course not. It's just saying that haphazard discretionary stimulus is better than haphazard discretionary austerity.

    2. The stimulus proposal that went to Congress had neither discretion nor conditions built in. You pays your money and you takes your chances.

  2. Daniel, once again you seem to have latched on to one little thing I said, in the context of a particular discussion, and then wondered if it's a statement that's true in general. Of course we can come up with all kinds of "rules" that are accomodative. We can even come up with a rule that says, "If unemployment is 1% and price inflation is 20%, then the Fed should expand its balance sheet." Thus we have a rule that is pro-cyclical.

    I was talking about the use of rules vs. discretion in the context of macroeconomists arguing the issue. And I was saying there, someone who opposes discretion means that there should be a constraint in place that prevents policymakers from doing what they perceive to be as optimal in the moment of crisis. It's like having a prohibition on locking somebody up for political speech. You don't need to worry about the gov't locking up people for praising it; you need to prevent the gov't from doing what it otherwise would want to do, to a critic.

    So same with monetary policy. The models I am talking about (I don't remember which ones--maybe it was Barro-Gordon) showed that the Phillips Curve can actually be moved in a socially advantageous way if the central bank isn't allowed to do what maximizes social utility in periods of high unemployment.

    The analogy is with negotiating with terrorists. If you have a blanket rule, "We don't negotiate with terrorists," then maybe you never have to worry about them taking somebody hostage, so long as you have credibility. But if Keynesians say, "Waaa! It would be too awful to sit back and do nothing, now that they've taken all those people hostage, let's pay them this time and then worry about structural reforms down the road..." oops there goes the rule out the window. Maybe it's the right thing to do, but it defeats the possible purpose of having such a rule.

    1. re: "I was talking about the use of rules vs. discretion in the context of macroeconomists arguing the issue."

      I know that was the context, although it's really even narrower than that. Taylor falls under "macroeconomists arguing about the issue", and he clearly has something different to say than Barro-Gordon or Kydland and Prescott. That's why I wanted to specifically note in this post that you were making a Barro-Gordon point.

      I suppose I was reading "do something that sucks" too narrowly. After all, presumably you don't have to do what sucks if you have a rule you like! But that really depends on whether you like the rule or not!

    2. DK wrote:

      I suppose I was reading "do something that sucks" too narrowly. After all, presumably you don't have to do what sucks if you have a rule you like! But that really depends on whether you like the rule or not!

      Argh! You can interpret this as either really patronizing, or as kind-hearted. But, I am persisting with this because I'm still not sure you are getting my (modest) point. I keep thinking you and I are on the same page, but then you go and write a comment like this that makes me not so sure...

      You can have a rule which simultaneously: (a) makes you worse off in certain situations, according to your own preferences, than you would be in the absence of the rule, and yet (b) makes you better off in the long-term. In other words, you sometimes benefit from having options removed, even if the removal of those options changes your behavior through time.

      So suppose society "really" thinks that it is worth tolerating an extra 2 percentage points of inflation for every point of employment, relative to the baseline 2% inflation rate and 5% unemployment rate. Left to discretion, if the Fed found itself in an economy with 7% inflation and 7% unemployment, it would be willing to cut interest rates in order to move to 8% inflation and 6% unemployment, if it thought the Phillips Curve allowed such a tradeoff.

      But, because people in the private sector know that the Fed will do this, it affects their long-term inflation expectations from time=1, and that itself influences what tradeoffs are possible, i.e. those very expectations influence the shape of the Phillips Curve.

      So, you can come up with a model where if the Fed is forced to pretend that actually it should only tolerate a 1% point increase in excess inflation, if it will achieve a 5% reduction in excess unemployment, then the Phillips Curve is transformed such that, operating under this constraint, the resulting path of the economy is preferable to the one under discretion, even with the true social welfare function.

      Then, once you get this result under your belt, you can then see why it makes sense to install a central banker who apparently really hates inflation. I.e. you install a central banker who hates inflation (compared to unemployment) more than "society" really does, so that when this stodgy banker sets discretionary policy, it actually is closer to the constrained rule policy.

      Now maybe this type of model captures an important aspect of the real world, or maybe it's nuts. But the comments I've been reading on your blog on these issues (not just from you, but from your commentators) makes it sound like this element is completely lacking. You guys seems to have just been talking about different ways of describing what the Fed could do, as opposed to seeing how (in principle) limits on the Fed could actually improve the outcome.

    3. Right - your stodgy banker model is the conservative banker model and it gives utility at a midpoint between discretion and a rule.

      All I was saying with that comment that I italicized above is that if for this reason we hire a conservative central banker it's because we've recognized that's a "good" outcome.

      It's like any other time-inconsistency problem - let's say avoiding an addiction, for example. On the one hand one could call that "doing something that really sucks" because that's what it may feel like at the time, but if you're consciously avoiding addictions (just like if you're consciously implementing a monetary rule), there's also a sense in which you're doing it because you're recognizing it's good for you.

  3. In the olden days (1970's) we used to distinguish between: 1. "feedback rules" (aka "contingent rules"); and memory fails....was it "simple rules" or "non-feedback rules"?

    Friedman's k% rule was a simple rule. New Keynesians advocated feedback rules.

    Then, sometime in the 1990's the terminology changed. We distinguished between "target rules" (like target 2% inflation), and "instrument rules" (the Taylor Rule is a feedback instrument rule).

    The Bank of Canada has a simple rule for the target, but operates on pure discretion for the instrument.

    1. Nick Rowe I was in diapers for most of the 1970s, and I don't think Daniel was alive. So what now?


All anonymous comments will be deleted. Consistent pseudonyms are fine.