When they undertook quantitative easing, central banks created billions of dollars, euros, pounds and other currencies. They used the money to buy assets and are now divesting them at a significant loss. The big question is: was it worth it?

Last week, I examined some of the institutional and accounting features of QE, which make the subject so difficult. To recap, the main effect of the stimulus programme was to shorten the effective maturity of consolidated public sector debt, swapping long-dated bonds into the equivalent of perpetual debt remunerated at the overnight central bank policy rate.

This was profitable when rates were low, but now borrowing costs have risen, it makes a loss for the public sector. These are real losses, borne by taxpayers with people or institutions in the private sector gaining.

Countries account for these losses in all sorts of ways, with the UK being transparent and taking them upfront while the US, Eurozone and some others tend to delay putting them into their public accounts.

Today, I will examine how much this matters and whether it should affect our assessment of QE.

How big are the losses?

This is a difficult question. QE is not over and the scale of losses is extremely sensitive to the level of short-term interest rates, so we cannot give a clear answer here. But that does not mean nothing can be said.

In the US, for example, the Congressional Budget Office just last week updated its assessment of income it expects the Federal Reserve to pay to the Treasury in coming years, remembering that the US brushes QE losses under a large rug labelled “tomorrow’s problem”.

As the chart below shows, the Fed has stopped paying money to the US Treasury until it repairs its own losses. It was paying around 0.4 per cent of GDP each year until 2022 and now it pays zero.

The CBO projects that it will not get back to 0.4 per cent until 2033, much later than it previously expected because interest rates have stayed higher for longer, increasing the Fed’s losses. On my calculations, the cumulative lost revenue, and hence extra debt, for US taxpayers is 3.2 per cent of GDP, or $900bn.

Of course, I am including neither past profits from QE nor the benefits of asset purchases to the economy, so it is a very crude measure and not a cost-benefit analysis of QE.

Using the UK’s Office for Budget Responsibility’s projections, a similar calculation for UK losses arrives at a figure of about 8 per cent of GDP, more than twice as high and definitively very large. Net of previous profits, it would still come out at more than £100bn or about 4 per cent of GDP.

Why has the UK lost more?

This is entirely in line with most other research, which estimates losses much higher in the UK than in the US and Eurozone — and these are higher than in smaller economies that did less QE.

Michael Saunders, a former BoE MPC member now at Oxford Economics, estimates that the mark-to-market capital losses in late 2023 for the UK were 23 per cent, compared with 13 per cent for the Fed and Eurozone and 11 per cent in Canada.

Stephen Cecchetti and Jens Hilscher estimate the peak losses are about 1.5 per cent of GDP in one year in the UK, compared with 0.5 per cent in the US and 0.4 per cent in the Eurozone.

Since QE is a maturity transformation of overnight interest-bearing debt swapped for longer-dated bonds, higher losses arise when more QE is undertaken, when policy interest rate rises further and when the maturity of bonds purchased is longer, since their value falls more when interest rates rise.

As the table below shows, the UK was on the wrong end of all of those parameters. It was particularly exposed due to the fact the government issues extremely long-dated debt compared with other countries.

This would normally insulate a country against interest rate risk, but not if you have in effect swapped it for debt paying interest at the overnight rate. You might, therefore, more accurately say that the UK lost its advantage in issuing much longer-dated bonds.

In addition, the country has not undertaken cost mitigations, such as limiting the amount of debt on which the central bank pays interest, unlike the ECB (although the Eurozone actions here should be noted are minimal).

Who gains?

Private sector banks are gaining from being remunerated at the policy rate risk free, for doing not very much. Of course, they have not chosen to hold these deposits, which have been created as a result of QE, but it is easy money for them at present. Private individuals gain to the extent that banks pass on this interest to customers in the form of higher interest rates on savings and lower borrowing costs.

Another group seeing gains, it appears, is foreign central banks. Since the BoE started actively selling its portfolio of long-dated debt, IMF data in the chart below shows that the foreign official sector has increased the share of UK debt it holds.

These institutions have paid a fair market price and have limited the amount of UK debt the private sector has had to absorb, so the UK can be thankful they have been willing to purchase its debt. It was particularly welcome for the country in the 2022 “Trussonomics” disaster when UK debt was distressed.

Of course, if UK government bond yields fall sharply as interest rate expectations decline, these other central banks will make a tidy sum.

So, was QE worth it?

There was a time when the cost-benefit analysis of QE was quite simple. On the benefits side, there were profits made from lower-cost public borrowing and improved macroeconomic outcomes. On the cost side was a sense that lower interest rates had artificially inflated asset prices and pushed them out of the reach of the young and poor. At the time, central bankers could sit back, pause, and with some justification say the following:

1. These calculations are very difficult

2. What was the alternative? No one else was stepping up to provide stimulus and the economy needed it

3. The side-effects on asset prices were a necessary price to pay for avoiding the much worse consequences for the young and poor of a prolonged economic slump

Now we know that the exit from QE has involved significant losses to taxpayers, the balance of the cost-benefit analysis is worse than we used to think.

Would these taxpayer costs have been better spent through fiscal stimulus? Should central banks have put in place better mechanisms to limit losses?

For what it is worth, outside the UK I don’t think central bank losses change the balance of argument that much. Although the numbers are large, they are not large enough to change the calculation. This is now something that is worth greater research.

In the UK things are, however, a bit different. There is absolutely no evidence that UK QE was more effective than that in the Eurozone and the US, but it cost two to three times as much. At some point the BoE will need to answer questions on why its version of QE was so expensive and why cost mitigations were not introduced.

What I’ve been reading and watching

• The Bank of England performed a U-turn on data dependence last week. Its willingness to set policy according to service-sector inflation and wage growth applied, it appears, only if the data was behaving as officials expected. What seems to be a majority on the committee (we do not know for certain), now thinks the data is a blip in an underlying successful disinflation strategy and the BoE is set to cut rates in August, rather as the ECB did in June. I said the BoE should “be more ECB”. In a welcome move, the central bank may have taken my advice

• Oh my! Bob Zoellick, former World Bank president, has broken the unwritten rule that officials and former officials don’t criticise each other. He accuses Jay Powell and Christine Lagarde of being data dependent because they “don’t know what to do”, accuses them of being unable to carry the respect of others on their policymaking committees and, in Powell’s case, being politically motivated

• Mohamed El-Erian says the Fed needs to get on with cutting rates and might have to cut more if it delays too much. His argument does not reflect the recent US data that suggests the Fed has more time to decide than its critics believe

• Brazil is testing the independence of its central bank, with senior figures in President Luiz Inácio Lula da Silva’s ruling party filing a lawsuit against central bank governor Roberto Campos Neto. They want him banned from making political statements, but what they really want is lower interest rates. Last week the Banco Central do Brasil held its interest rate at 10.5 per cent after a unanimous vote, citing the need for “greater caution” in the current economic environment when it raised its inflation forecast. This will not repair relationships

Charts that matter

More than one chart this week, because it is necessary to look at the effect of French politics on the ECB. Clearly, snap parliamentary elections might fundamentally alter French politics, its fiscal policy and its relationship with the rest of Europe, as Gideon Rachman explains here.

Will it trigger a new euro crisis? That is a possibility but there is no sign of it yet and the market moves have been modest. The scary chart below is the normal one you see, showing a big surge in the spread between French and German borrowing costs.

It looks bad but if you click on the chart, you will see the blip is smaller than a similar event before France’s 2017 elections and much smaller than the surge in Italian spreads after the election of the populist government in 2018.

More to the point, it is often good to look at levels rather than spreads. The main market move to date has been to cut German borrowing costs rather than to raise French ones. There has been something of a flight to safety rather than punishing France so far.

In these circumstances, there is no doubt that the ECB would do nothing. Fiscal profligacy is something first for France to deal with and then for the European Commission.

Things would have to get a lot worse with market moves threatening to spill over to other Eurozone countries or a wider systemic debt crisis before the ECB starts using the powerful tools to buy debt at its disposal. In fact the very existence of these tools is likely to limit the chance of having to use them. This is not 2011 (yet).