Friday, 1 May 2009
Several things to discuss in this posting. A break from writing (not planned or indeed noticed until it was over a week in length) has perhaps given me some space to turn over some ideas, and getting back to reading economics blogs this evening stimulated a few general thoughts about the structure and nature of the economy.
The first has come from a response to Greg Mankiw followed by a long chain of comments on Scott Sumner's blog, The Money Illusion. I've been reading Sumner's proposals for a couple of months, and one in particular has been playing on my mind. This is that the Fed should stop paying interest on excess reserves that banks deposit with it - and instead charge a penalty.
Reading an early post proposing this idea, I assumed that Scott's goal was to increase lending, and commented "Can't the banks lend anyway by creating new money, even without removing their reserves from the Fed?" Scott replied "yes" but that didn't explain it for me - I meant to ask whether lending could happen anyway even without reducing these reserves.
I've gone back to this a couple of times and kind of felt I was missing something. But reading the chain of comments on the above post, getting to the stage of being amused but not yet frustrated by the escalating nearly-ad-hominem arguments, it clicked into place suddenly.
Scott is concentrating on the relationship between excess reserves and deflation - excess reserves reduce the money supply and are likely to cause deflation (frustrating the Fed's desire to increase the money supply and counter the destruction of credit in circulation). Therefore he wants the excess reserves to get out of the Fed and into the money supply.
While he isn't directly interested in whether lending increases, my view is that this is the only realistic mechanism for nominal GDP to be boosted. The alternative mechanism, conversion of reserves into currency, will only increase GDP to the extent that the money is spent. The money circulation equation, MV = PY, requires the product of money supply and velocity to increase, but if cash is withdrawn and not spent, average velocity will go down.
No doubt some cash will be spent, but not all of it, and the effect on GDP will be smaller than desired.
Lending, on the other hand, does cause spending. Anyone choosing actively to borrow money is likely to spend it quickly, because otherwise they're paying interest on idle balances. And, as I was hinting in my comment, the money supply can increase by up to $8 trillion through lending based on $800 billion of reserves.
Here's what I was missing. If $8 trillion is lent out based on $800 billion of excess reserves, the reserves are suddenly not excess any more. Instead they become required reserves, which do (and Sumner agrees should) attract interest.
JKH and Bill Woolsey in the comments point out that $8 trillion is an unrealistic amount of extra lending, and that's true. But just $1 trillion would make a big difference. Commercial lending requires higher reserve ratios (technically, it's deposits that require reserves, but the mantra of reserve banking is that loans create deposits) so $1 trillion of loans might involve conversion of $150 billion of excess reserves into required reserves. The other $650 billion could be safely taken back out of circulation by the Fed - the $800 billion is only there in an attempt to flood the system with credit and cancel out part of the multi-trillion reduction in outstanding loans (and money supply shrinkage).
And Scott's proposal to charge penalty interest on excess reserves is indeed a good way to achieve this - provided that the penalty is enough to substantially shift the marginal lending point and not too high to stimulate avoidance tactics.
A 1% penalty, as opposed to the 0.25% interest rate currently paid, increases the effective spread on a commercial loan by 1.25% (by creating a 1.25% opportunity cost of not lending). The effect would be even greater if the excess reserves are going to be withdrawn when no longer needed - as a 1% penalty on $800 billion is $8 billion, which is the equivalent of a 5% rate on the $150 billion reserves that will remain after the exercise. So banks would have a very powerful incentive to shift reserves into the 'required' column if they know that the Fed will release them from holding excess reserves when they've lent out $1 trillion or so.
So my point is that Scott is right in principle - though I didn't originally realise why - but that the mechanism through which the policy operates also matters.
And mechanisms in general matter. Another of Scott's posts reminds me of this. He points out that most macroeconomic analysts implicitly deny the EMH (efficient markets hypothesis) in their explanations of the current crisis. But his own explanation does too. In fact, markets will usually correct themselves eventually. The problem is that they take time, and policy choices as well as endogenous factors can influence how much time. Mechanisms can make the difference between a 3-month correction or a 10-year depression.
Sumner's proposals seem like a good way of improving liquidity so the market can correct itself swiftly and efficiently. But they will only work because the EMH doesn't.
Finally two more thoughts.
The first - provoked by another section of the same Money Illusion article - is a distinction between left- and right-wing economists, which he relates to the difference between strict utilitarianism and morality - and another puzzle from the back of my head. Why is it - philosophically, or mathematically, rather than practically - that actions which seem to be right in the short term might be wrong in the long term? For example, we could solve all debt and credit problems at a stroke by a simple government action to pay off lots of bank debt, print money to recapitalise the system or by the stroke of a pen restore the economic system to its position in 2007. But we don't, because of the effect on incentives. My realisation here is that this is about the distinction between quantity and rate of change.
Equivalently, you could say that the economy works largely because it is predictable - people know roughly what is going to happen and they can make decisions on that basis. Predictability in turn originates from credibility - in this case, the government's credibility that it will protect property rights and the integrity of money. And, unlike the quantity of money in various bank accounts, credibility is something the government does not have direct control over. While it can transfer assets or money at a stroke, it cannot create credibility out of thin air. That is only earned over the long term.
So if the rate of increase of real wealth depends on the invested stock of predictability, an inadvertent reduction in predictability will damage future wealth. On this basis, there may be a good argument for the government to actively smooth out fluctuations in economic activity, but it must be done with great care and an eye for unintended consequences.
And my fourth thought, on the role of governments. Devin Finbarr, a commenter on (yet again) the above Scott Sumner post says "the major reason these crashes are so hard to predict is that they depend on government actions". As a thought experiment, he gives an extreme example of a Fed action which would destabilise the economy. The implied criticism is that governments have too much power and this is dangerous for the health and stability of a market economy. Not so!
The Fed's action in this case is so destabilising not because it is a government institution but because it is an overwhelmingly important actor in the system. But economies derive huge value from having dominant players in certain roles. Issuance of money is one, drawing up of regulations is another and there are plenty of others such as an integrated communications or transport system.
Economic theory and history give plenty of explanations for this - from the prisoner's dilemma to the tragedy of the commons, from the gains from network effects to the drawbacks of the free banking (and associated robber baron) eras in American and Scottish history, and even the distinction between the fragmented and chaotic societies of prehistoric or pre-medieval times and the organised and secure environment of the nation state. Having a bunch of independent institutions competing is not always better than having a single one which offers predictability.
In fact, gaining these benefits of stability and subjecting them to the control levers of democracy is probably the best tradeoff we could make. That's why good - and strong - government is an economic fundamental, even if it has some costs associated.
So what does it mean that - without my realisation - my four little epiphanies of the day were all prompted by Scott Sumner? Maybe I need to go meet him next time I'm in Boston.