Forward guidance: The Monetary Policy Report discussion of Fed losses

From: Bill Nelson

Sent: Monday, March 6, 2023 6:10 AM

Subject: Forward guidance: The Monetary Policy Report discussion of Fed losses

?

The Fed’s biannual Monetary Policy Report was released on Friday in advance of Chair Powell’s testimony this week.?The report notes that the Fed’s net income turned negative in September and that losses have resulted in a “deferred asset” (the cumulative sum of losses) of about $36 billion (pp. 40-41).?The deferred asset is essentially an IOU the Fed has written to itself; it is a marker to determine when to resume remittances. The deferred asset will keep growing as long as the Fed continues to make losses.?As shown in the graph, the Fed is losing over $2 billion a week and losses will rise further when the Fed raises interest rates again in a couple weeks.?The Fed won’t start handing its profits over to Treasury again even after its net income turns positive until it pays the deferred asset back down to zero, repaying the IOU.

The Fed is making losses because it took on a lot of interest rate risk, and interest rates rose a lot more than expected.?The economic argument for why QE works is that it “takes duration out of private hands,” driving down term premiums.?But that’s just another way of saying taking interest rate risk out of private hands and shifting it to the Fed and therefore taxpayers.?If QE works (personally, I am skeptical), the benefit is that the reduction in term premiums stimulates the economy.?The principle cost of QE that the Fed, and therefore taxpayers, incurs is the interest rate risk.

For Congress to conduct its oversight of the Fed appropriately, it needs to understand the costs and benefits of QE.?In a recent Brookings blog post – “What if the Federal Reserve books losses because of its quantitative easing?” –?Don Kohn and Bill English encouraged the Fed to promote such an understanding: “The Fed can help foster such a balanced assessment by using its communication and outreach to ensure that the public and the Congress understand the benefits of QE and how it can help the Fed achieve its policy objectives, as well as the potential risks.”?In a recent NBER and Hoover working paper – “Quantifying the Costs and Benefits of Quantitative Easing” – Andy Levin, Brian Lu, and I similarly emphasized the importance of Congress being well informed about QE.?Andy and I also published a short policy brief summarizing the working paper and addressing many of the misperceptions about the interest rate risk from QE – “The Federal Reserve’s Balance Sheet: Costs to Taxpayers of Quantitative Easing.”

?Indeed, twenty years ago, the Fed recognized that interest rate risk is not that different from credit risk, that it should only take on excessive interest rate risk in “unusual circumstances,” and that when doing so it should keep Treasury and Congress well informed.?Those conclusions were included in the Fed’s comprehensive review of the assets it could use to conduct monetary policy – “Alternative Instruments for Open Market and Discount Window Operations.”?The Fed’s “principle” is worth quoting in full:

The implications of bearing greater interest rate risk are similar to those for bearing credit risk. The level of interest rate risk should be well managed and compatible with what is necessary to meet the objectives of monetary policy. The level of interest rate risk and credit risk may not always be low. For instance, during conditions of emergency lending, taking on credit risk to prevent a contraction of liquidity in the private sector may be appropriate. In addition, under highly unusual circumstances, such as a period of extreme and prolonged economic weakness and a zero short-term interest rate, the Federal Reserve might increase its interest rate risk or foreign exchange risk beyond customary levels. It would do so with the knowledge that a subsequent economic expansion ultimately could result in capital losses in the portfolio when the economy, interest rates, and the exchange rate returned to higher levels (offset at least partly by capital gains from lower credit-risk premiums on any private securities). In taking on portfolio risk for policy purposes under these circumstances, the Federal Reserve would likely be in close communication with other governmental entities, including the Treasury and the Congress.

The Monetary Policy Report released Friday is exactly the right instrument for the Fed to explain to Congress the costs and benefits of QE.?Unfortunately, rather than promoting the “balanced assessment” that Kohn and English call for, the MPR suggests that the losses aren’t really a problem using two of the Fed’s standard talking points: 1) Yes, we are losing money now, but we made a lot of money in the past; and 2) mark-to-market losses aren’t real losses because we aren’t going to sell the securities.

?1) Yes, we are losing money now, but we made a lot of money in the past.

?After noting that it had stopped providing revenue to the Treasury, the Fed states “Although remittances are suspended at the time of this report, over the past decade and a half, the Federal Reserve has remitted over $1 trillion to the Treasury.”?For the purposes of assessing the costs and benefits of QE4, the fact that the Fed made money in the past is largely irrelevant.?The cost of QE is the interest rate risk.?The Fed took on substantial interest rate risk in QE1, 2, and 3 – indeed, subsequently released FOMC documents show that it realized it could make losses – but interest rates ended up remaining much lower than expected in the 2010s, so the Fed made money.?This time round, interest rates have ended up above expectations, so the Fed is losing money, that’s the nature of risk.

Moreover, the issue is neither how much the Fed made in the past nor how much it will lose in the future.?The Fed made profits in the past and should be expected to make profits in the future because it has ability to fund itself with zero-interest-bearing, infinitely lived, liabilities – currency.?Thus simply stating that the Fed made $1 trillion over the past 15 years is misleading.?If the objective is to compare the extra income from QE1-3 to the costs of QE4, the question is what net income would have been without those programs.?

Andy Levin and I answer this question in the policy brief cited above.?We calculate that through 2020, the Fed earned an extra $390 billion in net interest income from the securities it purchased in QE1-3, but that it will lose about $220 billion on those legacy securities over the next 10 years because interest rates are now higher.?In addition, we project that the new securities the Fed purchased in QE4 will reduce the Fed’s net income, and therefore increase the federal budget deficit, by a further $760 billion over the next 10 years.

While not directly comparable, CBO also appears to have also concluded that QE4 cost taxpayers hundreds of billions of dollars.?Last month, the CBOs projected that the Fed will remit to Treasury $292 billion from 2023 to 2030.?Just before the pandemic, in January 2020, the CBO projected that the Fed’s remittances in those years would total $632 billion.?

2) Mark-to-market losses aren’t real losses because we aren’t going to sell the securities.

After noting in the MPR that it had over $1 trillion in unrealized losses, the Fed observes that it does not plan on selling its securities, and securities pay par when they mature, so those losses will go away.?While the logic is impeccable, the implication that the losses are somehow less important because they won’t be realized is incorrect.?If the Fed were to sell all its securities right now, realizing all the losses, and then repurchase them, the outlook for remittances would be unchanged (in this hypothetical example, market prices do not change in response to the Fed sales and purchases).?The Fed’s reported asset holdings would decline by the realized loss, but the deferred asset would rise by exactly the same amount.?The outlook for interest income would be unchanged because the Fed would own the same securities.?The outlook for interest expense would also be unchanged because liabilities would be unchanged.

Unrealized losses matter because they reflect the carrying cost of holding on to a term security with a low, fixed yield funded with liabilities that pay the now much higher overnight rate (the purchases are funded with reserves, not currency; currency does not go up when the Fed buys a security).?In a recent FEDS note on losses, Fed staff state “While an unrealized gain or loss position on the SOMA portfolio does not directly affect the Fed's net income, if a higher expected policy rate path causes an unrealized loss position, this would be indicative of higher future interest expense.”?Moreover, as I have written about in the past (see here and here), the change in unrealized losses over an interval is a good proxy for the change in the present value of remittances.?Relatedly, Andy Levin and I show that the change in unrealized losses is also close to the change in the 10-year forecast for Federal revenue (see table 1).?While unrealized losses on a security must fall to zero when the security matures, that’s because the negative carry has been experienced over the life of the security and is now finished.

Another common misunderstanding about Fed losses, although one not promulgated by the MPR, is that the losses don’t matter because the Fed’s losses are the Treasury’s gains.?It is true that if the Fed has an unrealized loss on a security, the Treasury has a matching capital gain on the security, and those cancel out from the perspective of the consolidated federal balance sheet.?But if the Fed had not purchased the security, the Treasury would still have the capital gain and it would not be canceled out by a loss by the Fed.?That is, the Treasury’s outlook for its interest expense net of Federal Reserve remittances is nearly $1 trillion higher than if there had been no QE4.?The Treasury is not benefiting from a windfall from having funded itself with lots of low-yielding long-term debt in 2020 and 2021 because the Fed converted that debt into overnight borrowings from the perspective of the consolidated federal balance sheet.

Lastly, a question that often comes up:?Where does the Fed get the money to pay its bills when it is making losses??It borrows it.?The Fed borrows the money from banks in the form of increased reserve balances (deposits by banks at the Fed) and pays them the IORB rate, currently 4.65 percent.?Contrary to what is often said, the Fed does not pay its bills is by printing money.?While currency is a source of funding for the Fed, the quantity of currency is determined by the demand from businesses and households and foreigners – the Fed doesn’t print more to cover its expenses. Similarly, while the Fed also borrows from money market mutual funds, GSEs, and foreign official institutions, those liabilities are also determined by demand from those institutions.?(There is an interaction between reserve balances and overnight reverse repurchase agreements that I am ignoring, but ON RRPs are just another form of borrowing by the Fed.)

As always, please feel free to share this email with anyone who might be interested.?If anyone would like to be added to this free distribution list, they should just email me.?If you would like to unsubscribe, please go here.

?

Bill

?

?

Bill Nelson | Chief Economist | Bank Policy Institute | 1.703.340.4542