The Three Central Banks, 5/25

At least one of them could go bankrupt

Nick Timiraos writes (WSJ),

For now, that break-even level of the Fed’s benchmark rate—the point at which the Fed would pay more in interest than it earns—is around 3.5%, according to projections from economists at Deutsche Bank and Morgan Stanley. The federal-funds rate is currently in a range between 0.75% and 1%, and investors in interest-rate futures markets currently expect rates to rise above 3% in about a year.

It turns out that I am not the only who has been thinking about the Fed going bankrupt.

Think of the Federal Reserve as a combination of three agencies.

• The Monetary Authority. This is what the Fed was before 2008

• The Bank Rescuer. In 2008, this Fed financed large purchases of long-term assets from banks, paying banks interest on reserves (IOR) in exchange.

• The Giant S&L. In response to COVID, the Fed enlarged its portfolio of mortgage-backed securities using short-term funding, mostly reverse repurchase agreements.

S&L means “savings and loan,” a throwback to the institutions that held most of the mortgage assets in this country as of the early 1970s and which required a huge bailout in the 1980s.

The Monetary Authority is what we learned about in freshman economics textbooks. In the textbooks, it creates reserves at banks “with the stroke of a pen” (now we would say “with a keystroke”) and in return it receives short-term Treasury bills. The more reserves that banks have, the more money that banks are able to lend. That is how the Monetary Authority injects money into the economy. If the Monetary Authority wants to take money out of the economy, it sells T-bills and takes reserves away “with the stroke of a pen,” forcing banks to rein in their lending. (Note: when I first read Marcia Stigum’s Money Market in the early 1980s, I found out that the Fed was operating by intervening in the repo market, which is a subtle difference from the textbook story.)

The Bank Rescuer, starting in 2008, bought a lot of securities other than short-term T-bills. Mostly longer-term Treasury bonds and mortgage-backed securities. The Bank Rescuer also exchanged reserves for these securities, but it paid interest on these reserves, in order to induce the banks not to lend out these reserves. Paying interest on reserves (IOR) was a way of keeping the money supply from expanding along with reserves. In a sense, the Bank Rescuer neutered the Monetary Authority.

The Giant S&L went viral when COVID shut down a big chunk of the economy. But it did not increase its holdings of mortgage-backed securities solely by adding reserves to banks. It also created a “reverse repo facility,” which allowed money-market funds to earn interest by making very short-term loans to the Fed. In these transactions, the fund lends money to the Fed, nominally using mortgage-backed securities as collateral.

In a podcast with David Beckworth, Bill Nelson explains,

the Overnight RRP facility is a liability of the Federal Reserve. They're doing repos. It's always confusing because the Fed refers to repos and reverse repos from the perspective of the counterparty, not from their perspective. So, at the Overnight RRP Facility, primarily money market mutual funds, but GSEs again, also to some extent. And some primary dealers occasionally are investing money in the Fed. They're providing the Fed dollars in exchange for securities overnight, and then reversing that transaction and doing that each day. So it's a way that the Fed funds itself.

If interest rates go up, the Giant S&L will be in trouble. It has been earning a spread between the interest rate on mortgage-backed securities and the repo loan rate. Those might be 3.5 percent and 0.5 percent, respectively. But if the repo rate goes up to, say, 4 percent, then the spread becomes negative.

In such a scenario, the Bank Rescuer also is likely to be in trouble. If it keeps paying less than 1 percent interest on reserves, banks will stop keeping reserves and instead lend them out, leading to a big increase in the money supply. Or, as Beckworth points out,

what's left in the banking system of reserves may migrate rather quickly into money market funds and be parked, instead of being parked as reserves of the Federal be parked as overnight reverse repo facilities.

So the Bank Rescuer probably will have to pay at least 3 percent, one way or another. The spread between the interest rate received on long-term bonds and what it pays in could turn negative.

So far, I have not talked about why interest rates might increase. But one view is that interest rates might increase because they are too low relative to the rate of inflation. If rates stay this low, inflationary pressures will increase.

The Monetary Authority may be thinking this way. It may be thinking that interest rates have to increase. That means that if the other two Feds get in trouble, they cannot simply say, “We’re not bankrupt. We can create money (bank reserves) at the stroke of a pen!” The steps to stave off bankruptcy at the Bank Rescuer and the Giant S&L would be at cross-purposes with the goals of the Monetary Authority.

If the other two Feds get in trouble, they will have to go to Congress and ask for an appropriation of taxpayer funds for a bailout. Just like any other too-big-to-fail bank would have to do if it got in trouble.

An Aside: A Paradox in Market Monetarism

You would think that if a big financial institution comes to the market to sell a big bundle of mortgage-backed securities, the price of those securities drops. That is the law of supply and demand. A drop in price means an increase in the interest rate. That is how bonds work.

But what would happen if the big financial institution is the Fed, rather than Blackrock or Vanguard? Then a market monetarist (a disciple of Scott Sumner) might make the following argument:

1. Contractionary monetary policy leads to lower expectations of inflation.

2. Lower expectations of inflation lead to lower nominal interest rates.

3. For the Fed to sell mortgage securities means taking money out of the financial system, which is contractionary.

4. Therefore, when the Fed sells, nominal interest rates will come down.

5. Therefore, the prices of mortgage securities will go up.

So now we have a big institution selling mortgage securities, and their price goes up. It seems to violate the laws of supply and demand. That is the Paradox. I think it is a theoretical possibility, but not a description of the real world.

Relatedly, John Cochrane posted notes from his talk at a recent conference.

I've graphed here the Fed’s projections from the March 15 meeting for unemployment, federal funds rate and inflation. The Federal Reserve believes that inflation will go away all on its own without a period of high real interest rates. That, I think, is the central premise that most of our commentators disagree with.

…The markets, by the way, seem to agree with the Fed. Until the Fed started saying it's going to move, markets also seemed to think inflation would go away all on its own. 

Concerning what I am calling Market Monetarism, Cochrane writes,

The Fed’s projections are consistent with a standard New Keynesian model. This is not a nutty model. It's been around since 1990. It's the basis of essentially all academic macroeconomic theory. It may be wrong or right. The point of models is to help us understand what ingredients get to different views, and thereby also use evidence from other episodes to guide this one. I'm not advocating for against it. I'm just saying this is what we need to talk about. 

I don’t think that the Fed believes in that theoretical model, because if they did, they would not have created the Giant S&L in the first place. The Paradox would imply that when the Fed buys mortgage securities, this causes their prices to fall and their interest rates to rise. But when the Fed bought those securities, their intent was to prop up security prices, not drive them down.

I think that step (1) does not happen right away. A contractionary policy takes a long time to work its way through to reducing expectations of inflation. First, in order to get the process going, interest rates have to rise. This dampens inflation, and then inflation expectations come down, and then interest rates come down.

Market monetarism works under the assumption that inflation expectations instead change almost instantly. (Cochrane’s notes put this issue into models with equations.) I don’t think that is what happens. In the 1980s, the process of bringing inflation expectations down took years. Investors only lowered the inflation premium in interest rates after they were convinced that inflation had been reduced for good. I hope that the process does not take as long as it did back then. If it does, the Fed is going to have to ask Congress to appropriate taxpayer money to cover losses.

Finally, another relevant quote from Cochrane’s notes.

We had a fiscal shock. That has to come out of the pockets of bondholders, by inflating away bonds. The Fed can choose to inflate that away now or inflate it away in the future. But the Fed cannot get rid of the fact that there has been a fiscal shock which inflates away debt. So it has a limited power. It can smooth inflation but cannot completely get rid of inflation. 

Remember that point. The Fed is going to get blamed for the mess, in large part because it is a convenient target. But the villain here is the fiscal response to COVID, which began in the Trump Administration. And the hero is Senator Joe Manchin, who stood athwart the insanity saying “Stop!”

I imagine one of my commenters will appreciate what Cochrane says here:

And we've seen that failure in Latin America. Countries get into fiscal problems, they have inflation, they raise interest rates, that just raises debt costs, and the inflation spirals on up. 

Have a nice day.