Here. Keeping fiscal and monetary policy straight is harder than you'd think, and it has big implications for how we talk about things.
So too - as Gene has pointed out - for within fiscal policy. One ought to distinguish between the impacts of a debt burden and a tax. Grant McDermott pointed this out to, noting that the reason why perpetually rolling over the debt didn't work is because Bob was not perpetually rolling over the debt (did they ever respond do you on this Grant?).
I find the "this is monetary policy and this is fiscal policy" discussion more broadly as a little too convenient sometimes.
Take the way people talk about monetary policy reaction functions. These reaction functions really became important in modeling at a time when central banks were starting to get very well behaved and transparent. And to be fair, there's a good reason to have those reaction functions in there, because central bankers do react with some degree of predictability and they do it for the reasons these models suggest: they have their own objective function that they're optimizing.
That's fine.
The trouble comes in when we interpret what all that means.
Often a fiscal stimulus smacked down by the central bank is said to have a zero multiplier.
This has always struck me as odd. They're all policymakers. In fact the people that make fiscal policy appoint the people that make monetary policy.
So what's really going on is that the effect of monetary policy is positive and the fiscal multiplier is positive but we have some reason to think that the monetary authority will use its powers counter-productively.
That's a very different claim from the claim that the fiscal multiplier is zero!!! And it completely comes out of the definitions. An exogenous monetary policymakers' decision to counteract fiscal policy puts the economy in a worse situation than it would have been. That doesn't tell you anything about the technical efficacy of fiscal policy. It tells you about the consequences of central bankers with certain objetive functions.
I don't agree with Nick, Gene, or you. To some extent I wish people would take us at our word. Cantillon effects are about the injection of money, first and foremost.
ReplyDeleteLet's say that the government decide to pay higher returns to bondholders by buying bonds. They can do that by creating money or by taxing another group. In the latter case there is an injection in one place (the buying of the bonds) and a withdrawal in another.
Now, as I said earlier there are three different aspect to the Cantillon effect that people are normally interested in:
1. The effect of money creation on the market interest rate. (Where "market" means a narrow sort of interest paid on accounts and short bonds). From here we get the interest rate "vector", interest rate targeting and "too low" interest rates.
2. The direct effect of money creation on wealth distribution. This is what Rothbard complains about when he talks about bankers profiting from FRB.
3. The effect on income of the injection and whether or not it lags inflation. This is "account falsification".
For case 2 Nick and Gene have a point. When Rothbard complains about commercial banks he could just as easily complain about redistribution and government debt. He does of-course complain about those things but not in the same way. From the GDP point-of-view this is where the Cantillon effect meets the consumption and investment multipliers. If the taxed and the benefited group have different spending patterns then these will affect the distribution of resources.
For case 1 things are different though. The government is increasing bond income and decreasing other incomes. The effect on the market interest rate of that will be given by the differential lending and borrowing behaviour of both groups. If both groups have similar behaviour then there won't be much of an effect. Of course, in practice the central bank will nullify any effect with money creation and destruction. This situation is rather like "debt deflation" where we think about the differential spending behaviour of borrowers and lenders.
Case 3 is certainly different. Think about MV=PT. If no money is being created then M doesn't change. If the demand for money doesn't change then the price level won't change ceteris paribus. Of course, a redistribution many change the demand for money to some extent. In most cases though there's no reason to think a redistribution of this sort would. So, the redistribution case is unlikely to cause account falsification.
I don't understand why Nick Rowe is talking about the quantity of base money. What does that have to do with anything? What we're concerned about for most purposes is the overall money aggregates which are large compared to GDP.
Current: "I don't understand why Nick Rowe is talking about the quantity of base money. What does that have to do with anything?"
DeleteBase money is what the government prints. Most of it is 0% interest currency. The Bank of Canada's monopoly profits (which it gives to the government) nearly all come from the spread between 0% interest currency liabilities and positive interest assets.
Yes, but we're talking about the Cantillon effect here. About the interest rate vector and account falsification. Seignorage is a side issue.
DeleteNow, the BoC makes it's monopoly profits on the monetary base, we know that. But the commercial banks multiply up the effect (at least in normal times). At time X we have Mx * V = Px * T. Where Mx is the broad money supply which will be a multiple of base Mx = Base * k. Now Mx will increase to My before P increases broadly, the increases will show up only in some markets. Later on though we'll have My * V = Py * T. There are some people who are spending money disproportionately at the beginning of this process, those people are lucky because they get to buy things at prices closer to the old level. Those who's spending falls more towards the end of the process are unlucky because they have to buy at higher prices.
Some would argue that banks earn seignorage profits. I don't agree with that, but they do earn profits from account falsification because they spend on assets when a base money injection starts. They buy those assets at Px levels, not Py levels.
Daniel FWIW, I think you are making way more sense than Sumner or Rowe. But, I probably won't bring you into my discussion, because it's already getting complicated.
ReplyDeleteAnd Gene's argument against Rowe is devastating. I'm curious to see the response. (Note however that it doesn't hurt me, since I'm at least being consistent.)
"Daniel FWIW, I think you are making way more sense than Sumner or Rowe."
DeleteI think I just might print this and put it on my refrigerator :)