Are central banks going bankrupt?

The bond market has had a lousy year. And no one holds more bonds than central banks, which have amassed a fixed income portfolio worth well north of $30tn over the past decade. But do their mounting losses actually matter?

Yes and no. Central banks are obviously pretty unique institutions. On one hand they have a balance sheet and a P&L like anyone else, and right now they’re not looking great.

Toby Nangle estimates the Bank of England’s losses alone are currently around £200bn, and the Federal Reserve says it had notched up $330bn of unrealised losses by the end of the first quarter. We think it’s safe to say the pain has grown since then.

On the other hand, central banks are constructs of sovereign states and can literally create money out of thin air, which makes the whole bankruptcy question take on a different dimension.

Morgan Stanley’s chief economist Seth Carpenter wrote one of the definitive papers on the subject while at the Fed a decade ago, and revisited the subject over the weekend. Given the timeliness we thought we’d share and paraphrase liberally from it.

Central bank profits and losses matter . . . but only when they matter. Before the 1900s, the subject of economics was called “political economy.” Central bank losses that affect fiscal outcomes may have political ramifications, but the banks’ ability to conduct policy is not impaired . . .

. . . Starting with the Fed, all the income generated on the System Open Market Account portfolio, less interest expense, realized losses, and operating costs is remitted to the US Treasury. Before the Global Financial Crisis, these remittances averaged $20-25 billion per year; they ballooned to more than $100 billion as the balance sheet grew. These remittances reduce the deficit and borrowing needs. Net income depends on the (mostly fixed) average coupon on assets, the share of liabilities that are interest free (physical paper currency), and the level of reserves and reverse repo balances, whose costs float with the policy rate. From essentially zero in 2007, interest-bearing liabilities have mushroomed to almost two-thirds of the balance sheet.

As the chart below shows, the US central bank’s net income (which have been passed back to the US Treasury) has turned negative, and Morgan Stanley forecasts the losses will rise as interest rates rise.

Carpenter points out that most central banks, including the Fed, don’t mark to market, so any losses are unrealised and don’t flow through to the central bank’s income statement until they actually sells asset. But that obviously raises a lot of interesting questions.

So, what do losses mean? Is there a hit to capital? Bankruptcy? An inability to conduct monetary policy? No. First, remittances to the Treasury end, and the Treasury issues more debt. The Fed then cumulates its losses and, rather than reducing its capital, creates a “deferred asset.”1 When earnings turn positive again, remittances stay at zero until the losses are recouped; imagine the Fed facing a 100% tax rate and offsetting current losses with future income. Profitability will eventually return because currency will keep growing, lowering interest expense, and QT will shrink interest-bearing liabilities.

Things are similar elsewhere, but with local twists, such as the Czech central bank’s longstanding negative equity, or the fact that the Bank of England obtained an explicit UK government indemnification to be made whole from any losses when it started passing on its QE profits.

The effect is essentially the same as with the Fed, but the political economy differs. Where HMT and the BoE share responsibility, the Fed is on its own. Passive unwinding for the BoE is hard, given the lumpy maturity structure of gilt holdings, while the Fed has up to $95 billion per month running off passively. For the BoE, a one percentage point increase in Bank Rate lowers remittances by roughly £10 billion per year, a material sum for a country grappling with fiscal issues. The proposal to lower expense by prohibiting interest payments on reserves deserves scrutiny. If no authority remains, the BoE would have to sell assets to regain monetary control, realizing losses. The losses exist; it is the timing that is in question.

The ECB’s balance sheet is structured quite differently, but the logic is similar. Our European team projects the depo rate at 2.5% by next March, which implies ECB losses of around €40 billion next year. Bank deposits receive the depo rate, which will be much higher than the yield on the portfolio. The BoJ’s balance sheet has similarly swelled, but as of March (the latest available data), the BoJ was in an unrealized gain position. We think that yield curve control (YCC) will be maintained through the end of Governor Kuroda’s term, but when it ends, if the JGB curve sells off sharply, the losses could be large, though unrealized.

The most interesting variant is the Czech National Bank. The CNB has had a negative equity position for most of the past 20 years. Managing a small, open economy means focusing on the exchange rate, and most assets are foreign currency-denominated. If the central bank is credible and the Czech koruna rises, the value of its assets falls. The same is true for the Swiss National Bank, whose profits and losses have swung by billions in some years, yet it has not lost control of policy.

Central bank negative equity; coming to a Fed or BoE or ECB near you soon?