Central bank solvency, QE and the price level
This post is prompted by an exchange on twitter with Noah Smith [which you can see here] about why some people worry about the effects of QE on central bank solvency and eventually the price level.
In short, the argument is this. Suppose the central bank lowers interest rates to their floor. It is stuck for a way to stimulate the economy further. The central bank contemplates buying, for example, real estate. [Fathom Consulting were suggesting this at one point], exchanging it for central bank reserves, electronic money created by the central bank. [The transaction would involve a few more steps than that, but that's what it would amount to in the end]. What's wrong with this? Well, such a policy might be the right thing to do. But something that has to be thought through is the fact that the real estate may not fetch what the central bank paid for it when it comes to be sold. Presuming that it will be sold in the end, and that the central bank doesn't want to get into the business of real estate management permanently. [It shouldn't, because it won't be as good at it as real estate managers, and the activity will consume capital. Which is a problem we will come to.] Once the real estate is sold at a loss, the central bank winds up with less capital than it had originally. Does this matter? Yes. Sooner or later, when the central bank wishes to do something, unless the capital it previously had was surplus to requirements, it will find itself unable to finance it. It could be a new bail-out, or a PC on which Mark Carney or Ben Bernanke is going to write a speech. Can't the central bank just create some more electronic money to pay for these? Yes, but it can't do that and simultaneously hit the monetary policy targets for the price level that it has been instructed to do by the government. Aren't these amounts small, so that they would be lost in the gamut of other things that hit the price level? After all, central banks are always complaining that the price level index is being shunted around by one thing or another outside its control. Surely a bit of money printing like this could be hidden away. Yes, it probably could, these amounts could well be small, and very likely hidden in some Inflation Report waffle about price level shocks (of which there is ample precedent), but that's the problem. The act of central banks engaging in risky asset purchases will raise the concern that it will be forced at some point to engage in monetary financing, and that it will try to hide it with waffle about price level shocks. It could beg for a recapitalisation by the finance ministry, but who knows how that will turn out. Perhaps the money will be handed over, but perhaps in exchange for easier monetary policy, and easier monetary policy hidden amongst waffle about price level shocks. Perhaps observers would calculate that this kind of arm twisting would be considered intolerable by a modern, reputation-conscious central banker, but perhaps not. Who knows how policymakers might behave in the aftermath of a fiscally draining crisis. If, like Gordon Brown, they are contemplating putting troops on the street, central bankers would surely see that covert inflation was in the long-term national interest. Or so the thought process of the private sector might run.
Hang on! [Excuse the Tim Harford rhetorical device here]. What if the central bank is just buying government securities? The Bank of England's QE is almost all government securities. The Fed is buying a lot of debt issued by the Federal mortgage backing agencies, but these organisations are now explicitly backed by the government, so the debt should be considered as gold-plated as a Treasury. Does buying government securities not get us out of this central bank solvency problem?
No it doesn't. The shift in bond prices doesn't matter for the central government, who still has to redeem the face value of the bonds. But it does matter for the central bank. And if the central bank winds up selling the bonds back to the private sector for less than it bought them for, the central bank loses capital, and so on.
This is the reason why central bank solvency is important, and why central banks should not get involved in real asset markets without explicit government support. That's why Mervyn King insisted that QE would be operated using, ironically, a special purpose vehicle, the Asset Purchase Facility, whose balance sheet is separate from the Bank of England's. Ironic because it was off balance sheet vehicles in the public sector that got the government in trouble in the first place.
This institutional arrangement hopefully helped, but it can't be considered a cast iron guarantee of monetary good behaviour. After all, it's only words on paper, not statue. The agreements providing for QE and other policies to be conducted off the central bank balance sheet could be revoked, and in extremis who knows the central bank policy makers might agree to it. Not very likely, but not completely impossible either.
Involving the central bank is a good thing in some ways. If a policy instrument is needed for monetary policy purposes, it makes sense for operational reasons to hand over control to the central bank. If the MPC had to vote to recommend to the Treasury that QE of £x billion was undertaken, it might be less effective, since there might arise the concern that the Treasury could refuse. But these policies are inevitably fiscal in nature, and, as such, raise the spectre of monetary financing.
As a final word some, like Chris Sims, have argued that this spectre could be very helpful. The worry that massive central bank balance sheets might have to be monetarily financed might, if the expectation was that political constraints on conventional tax and spending instruments meant those would not be used, generate a rise in the price level to revalue nominal debts (incurred by the central bank) could be repaid. And if avoiding deflation was a prime concern this would not be wholly bad.