The Problem with the Money Multiplier
There’s an interesting back-and-forth between the ever ingenious Nick Rowe and David Glasner about the money multiplier. I don’t have a lot to add to it, really… I do have certain monetarist sympathies, but I don’t tend to view a money-centric view as particularly useful (as opposed to “true” in some metaphysical sense). And since I’m a microeconomist, I don’t really need to have a strong opinion.
Still, I think this particular debate might be illuminated by just a tad of data. Here is M3 (broad money) and the monetary base for the US. Notice the pattern?
Neither do I. The two weakly track each other (not surprising), but M3 moves with MB in starts and stops, even to the point where on occasion M3 is falling while MB is rising! Think about that….
You start off with a money multiplier, which means that the two should increase in proportion with one another… that doesn’t quite work. So then you tell yourself “well, as long as M3 weakly increases in MB” (this is something like, but not identical to what Rowe is arguing)… but then you notice that from mid-2009 to mid-2010 M3 is falling (or at least stagnant as MB rises, then from mid-2011 to 2013 MB falls a bit while M3 shoots upward. So, that doesn’t work either. Hmmm.
For me, it’s simple. The relationship between the Fed’s actions (creating monetary base) and the actual private creation of money is not a predictable relationship. Oh, maybe over time there is some tendency for the two series to converge. Fine, that seems reasonable. In fact, that seems like a good reason to keep teaching the money multiplier to early undergrads.
Doesn’t seem like it’s helping us understand what’s going on though.
BS Economist 
Suppose, for example, the Fed were targeting constant growth in M3. then you would expect to see zero correlation between M3 and MB.
If the Fed were targeting something that was more closely correlated with M3 than with MB, you would see a weak correlation between M3 and MB.
Only if the Fed were doing purely random stuff with MB would you have the perfect experimental design to see the true effect of MB on M3. But it isn’t.
This is just Milton Friedman’s Thermostat.
But the Fed isn’t targeting constant growth in M3! Besides, the real question here is whether or not the money multiplier has some external validity. If the relationship between the two is not predictable, then it’s no more than toy model.
I’m certainly willing to entertain the idea that it has teaching value, or perhaps even as a simplifying assumption in certain models which would otherwise be too complicated.
BSE: whether the money multiplier is useful depends on what you want to use it for. There might well be other relationships between base and target variable which are more useful, depending on what that target variable is. Lots of things will cause the MM to change. I wouldn’t bet on its usefulness, except in the long run, where real things will be approximately independent of the level of the base.
It’s good for teaching how commercial banks create money, even if each individual bank cannot increase its deposits unilaterally, without losing them to other banks.
All that I agree with.
My old post:
Curses! http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/07/why-are-almost-all-economists-unaware-of-milton-friedmans-thermostat.html