Bryan Caplan  

Japan Bet Bleg

PRINT
Dehiring: Win-Win-Lose... Galbraith's Imposition of Pric...
Scott Sumner has kindly agreed to bleg my request to help craft a bet with Bill Dickens:
Bill Dickens wants to bet me that looser Japanese monetary policy won't boost Japanese NGDP.  What victory conditions should a market monetarist consider prudent and probative?
As far as I can tell, none of Scott's readers constructively answer the question.  Can Team EconLog do better?


Comments and Sharing





COMMENTS (10 to date)
Philo writes:

You need a measure of how *loose* Japanese monetary policy really is, and you probably need to bet on just the *correlation* between policy and GNP (*causation* is too hard to verify). You also need to agree on the lag time: monetary policy during the time-period from t1 to (t1 + n) is to be correlated with GNP during the time-period from t2 to (t2 + n), but by how much does t2 lag t1? (And what are the values of t1 and n?)

Specifying all this is too hard. I don’t think your bet will materialize.

Kevin Dick writes:

I read both blogs so no particular team affiliation.

It seems like the bet must be of the form:

If the BoJ takes at least action X for at least period i, then NGDP will experience an increase of at least Y% no later than period j.

Y, i, and j seem pretty well parameterized. You have to negotiate them of course, but what you're negotiating is clear.

So it's only X that needs more work. I'd say this would be quantity of bonds purchased. But I suppose it could be something else. It might even be multiple conditions. But if these conditions aren't met in the first place, the bet is a push.

OneEyedMan writes:

I know that Sumner is down on RGDP but you don't have to be. How about this. If NGDP growth is heavy enough for money to be loose then RGDP growth will be higher than it has been. E.g., If NGDP has a cumulative annual growth rate of 4% per year over the next five years, then the cumulative annual growth rate of RGDP growth will be higher than 2% over the same period.

I think you've hit upon the subject that makes market monetarism so appealing to many and yet so unappealing to many others.

As Sumner frequently points out, the stance of monetary policy (loose vs. tight) can only be inferred from the difference between actual NGDP and targeted NGDP. If Japanese NGDP doesn't rise than the only two options are Japan had tight monetary policy or the Japanese central bank was actually targeting lower NGDP. In other words, you are being asked to accept a bet you simply can't win, since monetary policy (as defined by market monetarists) simply can't fail (if they try).

To have any hope at making the bet fair and worthwhile, you will have to determine a measure of the stance of monetary policy separate from the level or growth rate of NGDP. My best suggestion is growth of the monetary base since the central bank can determine that outcome irrespective of decisions from the government or private banking system.

Eric Rall writes:

Measure "loose monetary policy" by the growth rate of Japan's M2, MZM, or similar metric, averaged over the term of the bet. If the growth rate is at least a certain amount higher than the prior period (probably 1-2 standard deviations more than the variation in quarterly growth rates over the last few years).

Measure "increase NGDP" by Japan's NGDP growth rate during the bet period increasing relative to the prior period, by more than any similar increase in the United States, Britain, or the Eurozone. This should capture any increase in NGDP due to factors peculiar to Japan (loose monetary policy being the most likely candidate), but is unlikely to be triggered by chance (all else being equal, Japan has a 1/4 chance of having the highest growth rate).

Protect the positive side of the bet against looser monetary policy in the "control" groups by dropping any one of them out of the bet if it also meet the "looser monetary policy" test, and cancelling the bet if two or all three of them meet the test.

MikeDC writes:

What's the more substantive disagreement you're trying to use the bet to solve?

The first problem is Japan doesn't seem to have growing NGDP. First, you have to get it in order to have any chance of telling why you got it.

Second, isn't it axiomatic that you must have easy money if you have growing NGDP. Under every theory, "tight money" leads to falling NGDP.

NGDP targeting is problematic to economics because I think it's fundamentally immeasurable (so are other means of macroeconomic control, but we cling to our illusions). That is, I don't think this question can be answered because what's really driving the Money -> GDP "bus" is expectations. No matter how you measure a change in the money supply, the change has to affect expectations in order to change economic output. If you add a dollar, but everyone knows you'll suck it right back out of the economy in six months, they won't bother.

William writes:

Informing you that your bet proposal is unworkable is a constructive answer. The fundamental problem is that whether monetary policy is "loose" is inferred from the growth of NDGP.

Say I announce that I am going to adopt a more generous attitude in giving gifts to my wife. You want to bet me that my so-called "generosity" will not result in increased holiday spending. We're not really disagreeing about the effect of generosity on total gift spending. You are, in effect, telling me that you don't believe that I really am going to be more generous going forward.

Charlie writes:

The problem is that market monetarists define looser money as higher expected NGDP growth. So for a market monetarist to translate the bet, it would be does higher expected NGDP growth lead to NGDP.

Expected NGDP is not observed, but the comments on Sumner's blog gave several proxies:

1. Currency depreciation (relative to USD)
2. Inflation expectations (as measured in the bond market)
3. Forecasts

If this doesn't sound like the bet you want to make, you probably aren't asking the right question (at least in the MM framework).

BC writes:

As others have pointed out, when the central bank conducts discretionary monetary policy, it is difficult to define "loose" since policy involves communications and so-called "forward guidance", and evaluating the effect of such communications on expectations is inherently subjective.

However, maybe "loose" monetary policy could be defined as

(1) the two bettors agree that the BoJ is consistently communicating that they intend to purchase as many assets as required to achieve a NGDP target; and

(2) the BoJ does in fact continue purchasing assets at a rate of at least M yen/yr as long as NGDP is below target.

Then, the bet would be that as long as (1) and (2) are satisfied for the entire period that NGDP is below target, then within some period T, NGDP grows to within some delta of the target.

This bet may be unlikely to materialize since the BoJ may be unlikely to do (1). If one is willing to accept an inflation target as a poor substitute for an NGDP target, then one could define "loose" monetary policy as

(1) the two bettors agree that the BoJ is consistently communicating that they intend to purchase as many assets as required to achieve an inflation rate of 2%; and

(2) so long as implied inflation expectations in the Japanese inflation-linked bond market are below 2% (or 2% minus some epsilon) at any term, the BoJ does in fact continue purchasing assets at a rate of at least M yen/yr.

In either case, if any bettor at any time objects that the BoJ's communications do not satisfy (1), then the bet is off.

Obviously, this bet is not ideal for several reasons. Towards the end of period T, if it appears that NGDP will not hit target, the losing party could call off the bet by declaring that (1) has been violated. So, there needs to be good faith between the bettors. Also, it's not clear what M or T should be. These are consequences of discretionary monetary policy where there is no NGDP-linked bond market in which the BoJ is forced to buy as many assets as required to raise NGDP expectations.

Thorstein Veblen writes:

As a commenter mentioned, the key here is to see how much looser japanese monetary policy gets. For example, if inflation and output aren't increasing in three months, do they double-down, or just continue on autopilot?

In fact I'll spoil the fun. They will continue on autopilot. If a year goes by with no effect they will take a very small step. And people will say "Aha! Monetary policy doesn't work in the zero lower bound!"

Comments for this entry have been closed
Return to top