What to say when you’re asked your view on Bank of England reserve tiering
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
You have a career in finance. Your friends and family expect you to have intelligent bons mots about money stuff when you go to parties. In fact, the main reason you still get invited to parties is your ability to answer questions like:
— Will the stock market go up or down?
— Should I refix my mortgage?
— Should the Bank of England switch to a tiered reserve system?
…wait, what?
This last question has broken into the mainstream agenda pretty fast. And if you don’t have a ready answer to this debate du jour, read on.
Last week, MainFT’s Chris Giles argued that the Bank of England’s mandate should be amended to reduce the fiscal costs attached to our current scheme of central bank operations. In so doing, he jumped on board a bandwagon set in motion by the New Economics Foundation back in summer 2022. They argued that the UK government could save £19bn a year by stopping paying interest on reserves. So who else is on this left-leaning metropolitan think-tank bandwagon?
*Checks notes*
In fact, changing the operational framework of the BoE to save £35bn was the first policy that Nigel Farage chose to highlight in his leader’s interview with Laura Kuenssberg (four mins in) over the weekend. And, on Monday, they doubled down on the plan with a central bank wangle that we’ll come back to at the end of this post.
To be clear, Chris wasn’t actually backing as extreme a change as either NEF or Reform. Plus, calls for change have also come from two ex-BoE deputy governors: Sir Paul Tucker (in 2022) and Sir Charlie Bean (in February 2024). And it was Lord Adair Turner, former chair of the Financial Services Authority, who can perhaps claim authorship of the idea way back in 2021.
Furthermore, as ITV’s Robert Peston notes, Gordon Brown has been banging the drum for months, arguing that Labour should adopt a version that might raise maybe £1.5bn a year:
There’s no more doubt about it: the BoE reserves discourse is upon us.
So what’s going on? Are there even bons mots to be had on such an arcane topic?
Let’s break this down.
Why does the Treasury pay a big number, and how big is the number?
When the Bank of England bought its shedload of bonds, we all know it printed money, right?
Well, not exactly. A better analogy would be to say that the Bank entered a massive fixed-for-floating interest rate swap, sucking interest rate risk out of the private sector and putting it on the public sector balance sheet. OK, this analogy probably makes you an annoying dinner party guest, but hang in there.
The Bank’s bond portfolio peaked at close to £900bn, but today the balance is just under £700bn:
It receives income (bond coupons) on this portfolio that you can think of as the fixed leg of the fixed-for-floating interest rate swap. What’s the interest rate on this fixed leg? We calculate it at just over 2 per cent today. So let’s say that the income on the fixed leg of the swap is £14bn.
What about the floating leg? Following an overhaul of money market operations back in 2005, the BoE pays the MPC-set Bank Rate on central bank reserves. Think of central bank reserves like balances in the BoE’s big spreadsheet used by commercial banks to settle transactions with one another that occur between their customers. You can read for yourself the reasons why the Bank (and HM Treasury) decided to make this change. Tl;dr: It was to get better control of the policy rate.
Back in 2005, the amount of central bank reserves was pretty minuscule, and the interest payable on these small reserve balances was tiny.
Today, reserve balances are huge. With Bank Rate at 5.25 per cent, the floating leg of the £700bn swap pays around £37bn.
This wasn’t the plan. Sir Paul Tucker, who was intimately involved in the 2005 switch to pay interest on reserves, wrote for the IFS in 2022:
. . . the Bank of England’s decision to pay interest on reserves was taken [before QE] in the context of reforms to its operating system in normal circumstances . . .
QE came along, but:
. . . since QE was not expected to persist for many years . . . the possibility of the serious public finance implications explored here was remote.
Given that QE has persisted for many years, the implications for public finances have been serious.
The £23bn annual interest rate loss that Chris cites is the difference between the fixed leg (£14bn received) and the floating leg (£37bn paid). Of course, if Bank Rate falls below 2 per cent the interest rate loss turns back to profit. But no one is really expecting this to happen.
Working out an actual number for the long-term drain on the public purse really will depend on:
1) Where Bank Rate is set by the MPC over coming years (ie, the interest rate on the floating leg);
2) The pace at which the Bank of England unwinds quantitative easing — so-called quantitative tightening, or QT (ie, the stock on which fixed and floating are paid), and;
3) How the Bank organises its balance sheet post-QT (ie, what comes after QE has been unwound and whether it comes with a fiscal cost).
All of these factors are in flux, with Andrew Bailey outlining in his LSE speech a couple of weeks back a vision for the Bank’s balance sheet that is dominated by short-term repo.
So one answer to the question “how big is the cost” looks like it might be “it depends”? Nuanced and considered, check. Surprising and original, not so much. But the point is that in at least the short term, HM Treasury will be paying away huge sums to the Bank of England, and the Bank will be passing on these sums to commercial banks so that they can either: a) pay depositors an interest rate; b) keep the money for themselves.
So: if paying interest on reserves costs money that might otherwise be spent on fixing the NHS/ schools/ national defence/ etc and stuff worked perfectly well for centuries in a framework where we didn’t pay interest on reserves, how about we just . . . stop paying interest on reserves?
This sounds like unbreakable logic. Where’s the catch? Spoiler alert: wrinkles aplenty, but no knockout blow coming.
The good old days
How did things work before interest was paid on reserves? The answer is very long and very complicated, and trying to explain it will not win you friends. The previous system essentially relied on there being just about enough reserves to satisfy the banking system’s need for settlement and precautionary balances, and involved a mad dash each day to make sure everything sort of worked. The charts showing quite how wildly the overnight interest rate used to vary around the policy rate look a little mad in today’s world:
Despite the randomness of overnight interest rates, the world kept spinning pre-2005. Perhaps the cost of a little chaos in overnight interest rates is worth the £37bn that HM Treasury would have in its coffers if interest was no longer paid on reserves? We’d say probably!
But, to repeat, the previous system essentially relied on there being just about enough reserves to satisfy the banking system’s need for settlement and precautionary balances. Today we have waaaaay more reserves than just about enough (aka the Preferred Minimum Range of Reserves — itself a much bigger number than anyone might’ve guessed in yesteryear). We don’t know how much more reserves, but the Bank’s latest guess says perhaps £200bn–£350bn more.
As such, the first thing that would happen if the Bank stopped paying interest on all reserves is that interest rates would collapse to zero. Sure, the MPC can decide that 5.25 per cent is the right number for Bank Rate. But the only reason why overnight rates pay any attention is that under the current set-up the Bank of England pays 5.25 per cent on reserves. In so doing they set a floor below which no bank would lend.
A cunning plan?
The Bank of England gets to decide who holds a banking licence. How about they just say to banks “if you want to keep doing business, you’ve got to have a minimum reserve balance in your account with us; and we’ll pay you nothing”? Welcome to the world of minimum reserve requirements and tiered reserves.
While this kind of ultimatum may sound a bit mobsterish. But it’s neither new nor unusual for central banks to set minimum reserve requirements.
The OECD reckon there are three good excuses for having them:
— Microprudential motivations: ensuring banks have liquidity buffers, although now more often dealt with regulation;
— Liquidity regulation: a means of sterilising reserve balances;
— Monetary control: to push credit growth higher or lower without obvious FX spillovers.
A central bank can use reserve requirements to really get into the weeds of commercial banks’ business mix, setting different levels of required reserves for different business lines, and business strategies. The possibilities are endless.
Are required reserves remunerated? Not usually. As this OECD technical note sets out, forty-one of the forty-nine Advisory Task Force participant countries have minimum reserve requirements. If anything, the UK looks like the weird ones for not setting them:
Most countries do not remunerate RRs. The few that do usually remunerate below the policy rate and in any case below the market rate. Reserve requirements are thus de facto a tax on the banking sector.
And here we get the fourth excuse for using them. When used in size, they become, according to the OECD, “a key component of a financially repressed economy”. A simple average of reserve requirements has shown steady decline, and they note “their use has tended to wane as countries developed”.
Calling unremunerated reserves a tax on commercial banks doesn’t seem controversial. Paul Tucker called his plan to stop remunerating reserves a “de facto tax on banking”. The New York Fed writes that forcing unremunerated reserves on banks is “akin to levying a tax on them”. And Andrew Bailey’s dismissal of calls back in 2021, and again in February this year to consider are based primarily on the issue being a fiscal rather than monetary one.
And there’s no iron law saying that banks shouldn’t have massive new taxes applied to them. Money is tight right now. So if you’re a political party looking at the option to raise a bit of cash, what might you want to bear in mind?
Let’s look back at that OECD checklist.
Firstly, because reserve requirements are akin to a tax on banking activity, raising reserve requirements from zero to <something> would be a monetary policy tightening. The spreads between deposit rates and lending rates are likely to rise, and it may mean that the MPC will need to cut Bank Rate to offset their estimate of this tightening. Textbooks will tell you that, collectively, borrowers and lenders will have to stump up the amount being extracted in the form of some combination of lower deposit rates and higher borrowing rates. But there’s always the chance that UK banking is so singularly competitive that the de facto tax all comes out of bank profits. We don’t know where the balance lies, but as Chris put it in a Twitter convo with tax expert Dan Neidle, the larger sum you seek, the more you guarantee the incidence is on financial intermediation and not banks’ supernormal profits. In other words, the bigger the required reserves, the more it’s a tax on banking and the less it’s a tax on banks.
Second, by increasing the cost of banking, you divert activities into the non-bank sector. Think private credit funds, securitisation structures, the sort of less-regulated channels that the IMF warns about.
Third, by creating a positive link between the size of the central bank’s balance sheet and the de facto tax on banks, Bill Allen from NIESR argues that you turn the logic of QE on its head, by delivering a tightening as the balance sheet increases and a loosening as the balance sheet shrinks. Which is a bit weird.
Fourth, we understand from folks far smarter and knowledgeable than ourselves that if the Bank were to set the tiering structure too high (ie, make the stock of unremunerated reserves too high) they would run the risk of maybe accidentally losing monetary control. How? Well we know that if all reserves were remunerated at zero and there are more reserves in the system than just about enough (aka the Preferred Minimum Range of Reserves) overnight interest rates would collapse to zero regardless of where the MPC wants them to be. The higher we set the minimum stock of unremunerated reserves, the closer we get to this point. No one knows exactly what level of minimum unremunerated reserves will make the wheels fall off. So if the UK were to go down this route some kind of incremental learning-by-doing approach might be preferable to setting a big flashy number that turns out to make Liz Truss look like Jamie Dimon.
Fifth, remember back at the start of this post we argued that QE wasn’t so much “let’s print money” as “let’s do a massive interest-rate-swap”? Well, going down the route of forcing unremunerated reserves on the banking system sort of makes QE look pretty much like “let’s print money” after all. We can imagine that central bankers — who have spent years telling folks that it isn’t just money printing — might feel a little foolish at that point. Protecting central bankers from looking foolish is not a good reason to not de facto tax the banks, but it maybe is a good reason to expect central bankers not to be massive fans.
Finally, and connectedly, the whole thing smells of financial repression and fiscal dominance. That’s a vibes thing, and it’s not easy to quantify the impact. But we doubt the costs are zero.
It’s important at this point to observe that this is not an all-or-nothing question. The central bank could require commercial banks to have tiny amounts of unremunerated reserves, vast quantities, or somewhere in between. This is why different advocates come up with different fiscal gains for the approach. As such they could raise anything from as little as a penny to around £37bn a year depending on the design.
Going down the unremunerated reserves route does have one major advantage for any incoming chancellor though. As Louis put it yesterday — the numbers can get stunningly big, and the details (as you may by now have gathered) are stunningly boring.
The Reform Party’s plan
At £30bn–£40bn a year, the amount of revenue that the Reform Party claims they could raise is certainly big. Big enough to force the spread between deposit and lending rates quite a lot wider? Almost certainly. Big enough to cause the Bank to accidentally lose monetary control? We hope not.
Being vanishingly unlikely to win more than a tiny handful of seats in the general election, you might question whether we should care about the Reform Party’s policies. We should. Partly because their strategy to crush the Tories and win the soul of British right-wing politics may work. And the right-of-centre party has for the past century been the traditional custodian of government in the UK. And partly because it’s rare to find a party making central bank operations so central to its policy plans.
Reform have very strong views about the Bank’s balance sheet. As well as a change to remuneration policies, they are calling for an immediate end to QT. At first we understood this as a call for an end to active bond sales — so-called active QT. We’ve written more about that here.
If active QT stops we’re left with passive QT: just letting bonds held by the Asset Purchase Facility run to maturity. Here’s the maturity profile of the APF holdings:
It would take until summer 2031 for half of the bonds in the portfolio to mature. The length of time the balance sheet remains large is critical to understanding how much public revenue will be boosted in a plan to force unremunerated reserves on banks. Under such an arrangement, the practical effect of Reform’s plan would be to apply a de facto tax on UK banks that starts at around £35bn a year, and drifts down over time in line with the APF maturity profile.
But this policy briefing article in The Telegraph suggests the changes may be even more radical:
Reform said its healthcare plan would cost £15 billion a year for two years, funded by reorganising Bank of England quantitative easing debt into 75-year bonds.
Alphaville spoke with Richard Tice today. He confirmed that Reform wants to completely end QT, and swap the APF’s holdings into what he said could be called ‘Covid bonds’ or ‘Corona Bonds’ — brand new instruments that wouldn’t need to be repaid for another 75 years.
Doing this would preserve the current size of the Bank’s balance sheet until around 2100, and the corresponding size of the de facto tax grab from the banks. Tice told us tiering isn’t needed — though he accepted it was a possible option — and said that the Bank of England can get away with paying zero on the entirety of central bank reserves, because Bank Rate worked without large reserves as a channel and therefore, he argues, could still work if they were effectively neutralised. We asked Tice whether the abundance of reserves created fundamentally different conditions that might cause problems. Tice reckoned not.
We struggle to understand how the MPC will be able to set interest rates in a world without reserve tiering, reserve scarcity or any payments of interest on reserves. If you do, please get in touch.
Putting this another way, it looks like Reform wants to make QE feel permanent, in the sense that 75 years is an unfathomably long policy horizon. This would transform all the QE that has not yet been unwound into near-permanent money financing. As friend of FTAV Tony Yates explained to us, if QE is never going to be unwound, then we would call that permanent money financing — aka a helicopter drop. We’re not completely sure that a promise to unwind in one fell swoop in 2100 is completely different from a policy not to unwind. But maybe it is.
Reform will release their full manifesto on 17 June, so watch this space. We could be on the brink of a UK political party seeking to explicitly cut taxes and increase spending through a policy of maybe-sort-of-retroactive-helicopter money, which would be something for the scrapbook.
What happens if we just carry on without changing anything?
If we don’t make a switch to unremunerated reserves, Is the UK locked into paying away £23bn a year, give or take, forever? No.
Andrew Bailey, in a fascinating speech a few weeks back, set out possible ways forward for the Bank’s balance sheet. It’s worth a read in full, and if there’s demand we may return in greater depth. But tl;dr: the future is likely to involve a lot more short-term repo operations. And these are likely to come without much interest cost to the Treasury (and may deliver positive carry that feeds a steady flow of fiscal dividends over time).
When might we be in this new world? Depending on the pace of active QT, the Bank could hit the Preferred Minimum Range of Reserves by as early as 2025:
After that you’d expect to see short-term repo accounting for ever-more of the reserve figure if this chart is any guide.
Put the two charts together and there’s a case to be made that by the end of 2027, with Bank Rate unchanged at 5.25 per cent, the net interest cost will have fallen from £23bn to around £6bn, and by the end of 2029 it falls to zero. If Bank Rate is cut — as is expected — the costs will be even lower.
We’re still talking about large numbers between now and zero. And it seems reasonable to ask whether there is a way to bridge the time between now and the repo-fuelled reserve future.
If taxing the banks looks to politicians to be the right choice to bridge the gap, we can see that there’s a case for at least partially unremunerated reserves. But maybe you can see why Tony Yates and the author wrote back in 2022 about the NEF plan that maybe a cleaner way to tax the banks was just . . . to tax the banks?
If nothing else, this post should’ve armed you with enough to create a few bons mots for the next time you’re asked about central bank reserves at a party. Although maybe the mots will be insufficiently bons to get invited back if you attempt an answer.
Comments