Ryan Avent, who writes for The Economist, sent me an email with some very good questions about monetary policy rules. (He was a moderator at the recent Mercatus/Cato conference on monetary policy rules.) Ryan allowed me to reprint them here, and I’ll take a stab at answering them:

1. Why haven’t central banks been successful at hitting their policy targets? (And would proposed rules overcome whatever obstacles are currently inhibiting central banks?)

Screen Shot 2016-09-10 at 8.31.57 PM.png

1. Central banks have made two key mistakes. First, they’ve done a poor job of forecasting. They’ve generally set policy at a level where they expected to hit their inflation targets in a couple years, but those expectations were unrealistic. Instead they should have relied more on market forecasts. Perhaps the most egregious error of this type occurred at the Fed meeting of September 2008, when they forecast excessively high inflation at a time the TIPS markets were forecasting very low inflation. The ECB has done a particularly poor job in this regard.

Second, they have failed to understand that in the long run the amount of unconventional policy that is required is inversely related to the aggressiveness of policy today. Thus the Swiss National Bank wrongly thought that by allowing the Swiss franc to appreciate in January 2015, they would not have to buy so much foreign exchange. Instead this revaluation made Swiss inflation even lower, and now the SF is an even more appealing asset, the hardest currency on Earth. Similarly, if the Fed had been more aggressive with QE in 2008 and 2009, they could have ended up with a smaller balance sheet today. In my view the central banker fear of unconventional policy is unjustified, but their actual policies cannot be justified even if you assume the fear is valid. They are their own worst enemy. Like someone who wades into cold water one inch at a time, dragging out the pain, rather than diving right in.

Many of the current problems would go away if central banks switched to level targeting, and/or relied on CPI/NGDP futures markets.

2. How confident are we in the robustness of monetary rules? Is the failure of past regimes a sign of their imperfections or the impossibility of finding one rule to rule them all?

Screen Shot 2016-09-10 at 8.30.57 PM.png

Some monetary rules are not very robust, for example the gold standard and/or a fixed exchange rate. Money supply rules are also unreliable. In my view, the most robust rule would target total labor compensation per working age population, level targeting. But I don’t doubt that even better rules will be discovered in the future. Which is fine.

As long as each failure is not as bad as the previous failure, we are making progress. There is no such thing as perfection in the social sciences.

3. Should a new rule be consistent with recent policy or should it represent “regime change”?

Here it’s a matter of degree. In my view, you’d like any new rule to be as close to the current rule as possible, for reasons of policy credibility. Thus if we switch to my preferred NGDPLT rule, I’d favor us adopting a rule that led to about 2% expected inflation, at least initially. If we found that this led to frequent bouts of the zero bound problem, we could increase the trend rate very gradually.

4. What role should the government play in setting the rule (and the target), in monitoring central-bank performance and holding the central bank accountable, and in contributing to stabilisation? Are fiscal rules an appropriate stabilisation measure?

The government should instruct the central bank in broad terms, such as the current dual mandate that the Fed operates under. At that point I’d let the central bank figure out how to make the rule operational, although the government certainly could and should step in with more detailed instructions if they are unhappy with the specific rule that the Fed adopted. Currently there is not enough opposition in Congress to change the Fed’s 2% inflation target.

Central banks need to be made much more accountable. I’ve proposed that the Fed be required to evaluate past decisions periodically, say 4 times a year. In each case they would report to Congress whether, in retrospect, their policy settings in recent years appear to have been too expansionary, too contractionary, or about right. In each case they would need to cite specific economic data to explain to Congress how they made that determination. I.e., they might point to recent inflation and employment data.

I oppose fiscal stabilization policy, as I think it’s much too hard to coordinate with a sensible monetary policy rule. If the central bank is having trouble hitting its target, it should ask Congress for more power.

5. Why have central bankers had such difficulty manipulating expectations? Is it for lack of trying or because they have been ineffective in their communications?

Central banks rely too much on their own internal models, which often employ flawed theoretical concepts such as the Phillips Curve. Instead they should rely on market expectations. If they set policy at a position where the market expected on-target inflation, then policy would be much more credible.

Last December, the Fed should have never tried to convince markets that 4 rate increases were likely this year. That’s not the Fed’s job. Rather the Fed should “ask” the markets how many rate increases are needed in order for 2017 inflation to come in at about 2%. If they operate that way, then policy will be credible.

6. Are we certain that rules are an effective way to set expectations in a stabilising way?

No, but that’s my best guess.