Tuesday, June 4, 2013

Monetary Policy Puzzle

Might raising interest rates, but not paying interest on reserves, actually be "stimulative," inducing banks to lend out reserves?

Last week, I gave a talk on monetary policy at a forum organized by the Becker-Friedman institute.  I explained my view, that as long as reserves pay the same interest rate as very short-term Treasuries, and as long as banks are holding huge amounts of excess reserves, that monetary policy and pure quantitative easing -- buy short-term treasuries, give the banks more reserves -- has absolutely no effect on anything. Interest-paying excess reserves are exactly the same thing as short-term treasuries.

When the time comes to tighten, I said, I hope dearly that the Fed continues to pay a market interest rate on reserves and allow huge amounts of excess reserves to continue. (I had lots of financial-stability reasons, which will wait for another day here.)   But that means that conventional open market operations and quantitative easing -- more reserves, less Treasuries -- will continue to have no effect whatsoever.

An audience member asked a very sharp question: Suppose the Fed raises interest rates but does not raise the rate on reserves? Now, banks do have an incentive to lend them out instead of sitting on them. Wouldn't velocity pick up, MV=PY start to work again, and the Fed get all the "stimulus" it wants and then some?

It's a particularly sharp question, because it gives sensible-sounding mechanism why the conventional sign might be wrong: why raising rates now might give monetary "stimulus" that is otherwise so conspicuously lacking. There are a few other of these stories wandering around. One: Low rates are said to discourage retirees and other savers, who now "can't afford to spend."  (Quotes around things that don't make much economic sense.)   John Taylor, wrote a very provocative WSJ oped, (too subtle to summarize in one sentence here) and also came close to saying the sign is wrong and higher rates would be more stimulative.

But is the suggestion right? I sort of stammered, and needed the weekend to think it through. (Giving talks like this is a great way to clarify one's ideas. Or maybe this just reveals my shocking ignorance. In any case, it makes a good exam question.) Think about it, and then click the "read more."

The answer is no, I think, but revealing about what the Fed can and cannot do.

How exactly would the Fed raise interest rates?

In the new interest-on-reserves regime (the one I hope will continue) the Fed simply announces, "we borrow and lend reserves at 3%." Interest rates go up to 3%. But so do interest rates on reserves.

The standard mechanism, which allows reserves not to pay interest,  would be for the open-market desk to sell securities in exchange for reserves, in order to drive down the supply of reserves until interest rates rise on the inter-bank (federal funds) market. Banks who need reserves then are willing to pay interest to borrow non-interest-paying reserves overnight to satisfy reserve requirements on their checking accounts.

You see the trouble. Rather than "get banks to lend out the reserves,"  the Fed has to soak up all those reserves in order to raise interest rates in the first place.

Like other central banks, the Fed could offer prices rather than control quantities. Other central banks set rates in the interbank market, by simply saying "we borrow and lend at 3%. Come and get it." (They may leave a window, borrow at 2.9%, lend at 3%, to keep a private market going.)

If the Fed were to do this, banks would simply take all the reserves --  money lent to the Fed overnight -- and... lend it to the Fed overnight at the higher interest rate. This is interest on reserves by another name, no more no less. To the extent that the Fed ties up the money -- borrows at term, or otherwise makes its offer more "bond" like -- this action just synthesizes the huge open market operation that drains reserves from the system. (It's not a bad idea, though, if the Fed wants to shrink reserves without selling assets!) Again, the desired incentive to get banks to "lend out" the reserves vanishes.

What if the Fed offers a price target for Treasuries, and also refuses to pay interest on reserves? Could the Fed offer to buy and sell 3 month Treasuries at 3%, but insist on no interest on reserves and no open market operations to soak up reserves? No, because offering to buy and sell Treasuries means offer to take reserves and give out treasuries, which ipso facto soaks up the reserves again.

The Treasury could (and arguably should) take over interest-rate policy. After all, in the interest-on-reserves regime, when the Fed says "3%, come and get it," short-term Treasury rates will also jump to 3%. (Banks dump Treasuries and give the proceeds to the Fed, driving up Treasury rates.)  It is exactly as if the Treasury said, "Rather than auction 3 month debt, we'll set the rate at 3%, and the market sets the quantity." If you think the quantity reaction might be large, you've figured out some usually unspoken limits on the Fed's interest-rate setting abilities.

But reserves are our numeraire. Pegging interest rates at 3% means the Treasury rather than the Fed takes in reserves in exchange for debt, and parks it in the Treasury account rather than bank's accounts at the Fed. The banks will  again drain reserves rather than "lend them out."

The Treasury could commit to immediately spend the money... But now we're in the land of fiscal, not monetary stimulus, and a reminder that the two always come together.

I'll be interested to hear comments on this one.  The standard stories by which interest rate increases are contractionary are very weak and full of holes. The idea that perhaps raising interest rates is "stimulative" is fun to think about. And a number of schemes around to get banks to "lend out the reserves" are also fun to think about. I don't think this one works, but maybe one of you can get it to work.