2013-09-12

I'm Always Right. And Always Wrong

Robert Murphy has been fighting a good fight about "empirical" economics. (Here, here, here, etc... )

While I think I might be a bigger fan of quantitative economic models than he is, the thrust of his argument is absolutely correct. Straight to the heart of the matter, he asks, "What would the world have to look like if the empirical approach didn’t work in economics?"

Elsewhere, at Worthwhile Canadian Iniative, he says this:
[...] In the meantime, though, can you clarify *what it would look like* for NGDPLT to fail? 
For example, my first thought was that we'd need to see a country suffer a typical business cycle, even thought NGDP kept growing at trend. But wouldn't Scott Sumner just say, "Well, that's clearly a supply-side recession. Nothing the central bank can do there." ? 
Then, what if we found a country where the central bankers explicitly told the public they were targeting level NGDP, but then in practice they realized it was impossible, and a recession occurred because targeting NGDP level leads to recessions. Even here, Scott would point to the time when it broke down and say, "Ah, they stopped targeting NGDP level. I'm right again." 
So I'm thinking it might be literally impossible to find an example of what you mean. In fact, I think Scott once wrote a post saying (somewhat tongue in cheek, but not really) that even if the Fed did what he said, and we got huge price inflation, that it would still be the right policy.
It's not fair to pick on Scott Sumner in exclusivity here, and that is not my intention. In fact, in this post, Murphy specifically calls out Paul Krugman instead. The names have been changed, but the story is the same. And the story is what I would like to write about today.

Macro-economic models are always both right and wrong. In today's world, there is very little theoretical disagreement about how the economy works in the long run. This is important because it highlights the fact that on almost all of the major, important theoretical points, most economists largely agree. The disagreement is not over the essential behavior of the economy; instead, economists disagree about the short-run impact of various temporary "shocks."

For many decades now, we've known that economies experience "business cycles." There are "booms" and "busts," or peaks and troughs in the level of economic growth. No one disputes this, although there is plenty of discussion over why it happens, and hence how to react to busts/troughs when they do occur.

But for every recession or depression - for every economic downturn - there is always a recovery. No economy stays in the dumps forever. The "everyman" can consider this cause for optimism. Economists can consider this cause for simply sticking to whatever narrative they have chosen long enough to be exonerated by data.

What I mean is this. Business cycles really do seem cyclical, despite the elegant complexity of the macro-economy. This suggests that anyone's narrative will appear to be true if they stick to it long enough. If Economic Jones is wrong about Ruritania in 2013, he can always wait until Ruritania's macro-economic data adhere to Jones' prediction, at which point he can say, "I was wrong about the exact timing of it. I did not think that the central bank would be so dense as to not follow my sage advice. I did not expect the fat cats on Capitol Hill to make such bad decisions. But, ultimately, I was right about the thrust of it."

And this is true of, not just Scott Sumner and Paul Krugman, but also Robert Murphy and Ryan Long and all the rest of everyone. No one can reliably guess the suit of the playing card I'm holding up right now, but at the end of the day, there are only four suits. If I keep drawing cards, and you keep guessing, eventually you're going to get it right.

So the name of the game that a lot of economists like to play is to just keep calling out the same suit over and over until the card matches what they're calling out. If Paul Krugman keeps saying "more stimulus," then a slow recovery confirms his theory, a second recession confirms his theory, and even a full economic recovery merely proves that the recovery would have been even faster or bigger had the recommended "more stimulus" been implemented in the first place.

Some time ago I wrote another post about this. That was a far more abstract post, but it was written to explain precisely what's going on here. In that post, I pointed out that when logic becomes sufficiently complex it becomes prohibitively difficult to understand the relationship between the various steps in the chain. A implies B is easy enough to follow, but when B implies C, C implies D, ..., X implies Y, and Y implies Z then we do not always have enough intellectual processing power to recognize the relationship between, say, H and Q. We want to start with A and we want to end up at Z. If there are too many steps between A and Z - or if those steps are sufficiently complex - then all we really end up doing is providing a veneer of reason for a misguided set of priors.

The smartest among us figure it out in an almost Pavlovian way. They start to realize that their stories are most convincing when they stick to one point and wait for the data to confirm it. They probably and quite genuinely believe that they have come up with a valid explanation for the universe. But, they have not. Instead, they've just become quite excellent at rationalization.

In that respect, I think Robert Murphy's point about "empiricism" in economics is a valuable and important one. It also extends to other disciplines, such as history. Any historical event is sure to confirm your prior beliefs about some other event that came before it. This is why leftists who opposed the war in Iraq feel vindicated by the events in Syria, while rightists who supported an Iraq invasion also feel vindicated; and yet both sets of people are arguing for opposite tactics to address the Syria question.

A stopped clock is right twice a day. Stay on point, and you'll always prove to be correct. Eventually.

No comments:

Post a Comment