10/2/14

Why Monetary Policy is Really One Big Conspiracy


On Tuesday I attended a lecture by David Miles, a member of the British Monetary Policy Committee, the UK equivalent of the US Federal Open Markets Committee, aka, the Fed, on “Giving Guidance On Future Monetary Policy In A Very Uncertain World.” 

I realize that all sounds absolutely fascinating to most people, so perhaps it’s better to rephrase as “On Tuesday I attended a lecture about how to make what is arguably the most important decision in the world by one of the people who makes it.”  

The Lecture, in a Nutshell 


Miles spent most of the lecture demonstrating that, because of economic uncertainty, it’s more or less impossible to make accurate macroeconomic forecasts and therefore to predict the optimal course of interest rates, and concluded that to avoid flip-flopping central bank pronouncements should be as substantive as possible while leaving room for uncertainty. In other words, they should be vague, which is, more or less, exactly where we are today. 


The Problem With Macroeconomics, in a Nutshell 


I found Miles’ method of establishing the uncertainty of macroeconomic forecasts to be incredibly ironic. One minute he was talking about numerically approximating a complex multi-stochastic-variable model to achieve confidence intervals for future macroeconomic incomes, a process that I’m sure required months of effort and years of education from several people, and the next minute he was pointing out that the confidence intervals were so wide they were hardly useful, and often reality fall outside of the previously calculated confidence intervals entirely. If they’re not useful, what was the point of spending all that time to build, and then present, those models? 

One of David Miles' macroeconomic model slides. You can download them here. 

Thankfully, albeit somewhat tragically, he seemed to be aware of the irony. His first slide featured a yiddish proverb that roughly translates as “men make plans and god laughs.” In response to an audience question at the end of the lecture, he said something along the lines of “sometimes qualitative analysis is more useful,” which I have to give him credit for—saying “qualitative” at an economics lecture is like saying “god” at an atheist convention. 

In the last few years enough smart people have come out and said that because of the complexity of the economy most macroeconomic modeling is useless that I find it hard to believe that many of the people who are smart enough to understand macroeconomic modeling still actually believe that it is useful. (<< that, my friends, is a sentence.) I know that some macroeconomists believe that the models just aren’t complex enough yet, and that more borrowing from computer science, physics, or mathematics could get us there, but I think that even they, when they’re being realistic, know that we’re nowhere close. 

So why are they still building the models? (Why am I thinking of taking a class to study those models?) To be honest, I'm not quite sure. Maybe it's a discipline wide case of effort justification. Maybe it's  collective modeler's block. Maybe it's a conspiracy. More on that later.

The Problem with Monetary Policy, in a Slightly Larger Nutshell 


Miles made a couple of key assumptions that he never explicitly stated. 

His first assumption was that central banks should attempt to optimize macroeconomic outcomes by influencing interest rate expectations. The role expectations play in macroeconomic outcomes is so widely accepted today that it's no surprise Miles didn't mention it, but it's an idea still left out of many undergraduate macroeconomic textbooks, which often pretend that only current conditions matter. 

His second assumption was that the best way to influence interest rate expectations is to forecast future interest rates as accurately as possible. You see, in economics there is this horrible Catch-22 where expectations determine future outcomes and estimates of future outcomes determine expectations. If your predictions fail to come true too many times, then they will be ignored and lose the power to influence expectations. In other words, if everyone knows that the central bank flip flops, then their announcements about future interest rates will no longer be an effective policy instrument. 

The curse of economics 

After all that was assumed, Miles began by outlining a three “forward guidance” choices a central bank could make: 

(1) No policy commitment 
Never apologize, never explain,” said Montague Norman
(2) Commitment to a “rules based” or “reaction function” policy  
“Interest rates will rise by 1% whenever inflation rises above 2%,” said John Taylor
(3) Commitment to an explicit policy
“The interest remain at 2% for the next 5 years,” said no central banker ever 

Obviously (1) is not a very effective policy instrument, as it does not do much to influence expectations at all. (3) is also clearly unsuitable, because if the central bank wants to remain credible, then they can't deviate from their policy commitment even when the pre-planned policy turns out to be quite harmful in the present. 

Montague Norman, clearly a #gangstabanka, also, probably, a Nazi 

So you want somewhere in between. "Reaction function" monetary policy, or monetary policy that commits to certain actions given certain macroeconomic realities, such as, for example, raising the interest rate when inflation increases, has been a dominant framework in monetary policy for some time now. But because expectations are fundamentally about time, any reaction function needs to be paired with macroeconomic forecasts to influence interest rate expectations. Cue Miles' discussion on how to avoid flip flopping given the difficulty of macroeconomic forecasting. 

Miles worried that central bank predictions weren't accurate enough. But if predictions swing in the other direction—if predictions are too credible, completely sure things—then they also lose their power as a policy instrument. To create a stimulus or quash a bubble, monetary policy needs to change expectations. In other words, it needs to be something of a surprise. 

What the central bank really wants is not to be accurate, but to be believed. This is the infamous time inconsistency problem. You're better off reading Mankiw, but briefly, time inconsistency is when you want one thing tomorrow today but a different thing tomorrow. For example, the U.S. government wants terrorists to believe that it does not negotiate with terrorists, but in an actual terrorist situation, the optimal decision would obviously be to negotiate. The central bank wants the economy to believe that its statements are credible—that it will keep interest rates low when it says it will, for example, but once it sees the upturn that those interest rates produce, it will want to raise them. (Apparently the time inconsistency model also invalidates Bellman's Principle of Optimality, the principle that underlies most macroeconomic forecasting models. Interesting.) 

What central banks really want 

So for central banks influencing interest rate expectations is really much more like a game than it is like a forecasting problem. But instead of playing with terrorists, they're playing with us—trying to get us to believe that their forward guidance is accurate, while still planning to flip flop if they can get away with it.

Why Monetary Policy is Really One Big Conspiracy 


Time inconsistency is a pretty well established thing, and Miles obviously knows about it, and did, in fact, briefly mention it. So why did he spend most of his time presenting an extremely complicated model that he said himself was pretty much useless, when he could have been talking about how central banks can make forward guidance the most persuasive, regardless of the actual truth? (Why, in other words, was he talking like an economist instead of like an advertising executive?) 

Long live flip flopping 

Well, that would have been counterproductive. After all, we are are the ones he is trying to fool. It's in his interest to pretend that his only goal is to make interest rate predictions as accurate as possible. He might even go to great lengths to produce complicated models that are impossible to understand to present as evidence of his endeavors. If I were him, I'd flash a bunch of complicated models, and when asked for a summary, say something too vague to be contradicted. And if someone accuses me of flip flopping or being vague, I'd explain that it's because the models are so complicated, mistakes are sometimes made, and that to be accurate any forward guidance needs to express uncertainty... oh wait. 

Of course, I don't really think that monetary policy is a conspiracy. But I do think that in the past monetary policy benefits from opacity, and that as the public wises up, its power as a policy instrument will gradually decrease.

.





No comments :

Post a Comment