The Bank for International Settlements (BIS) is a club of central banks. That isn’t a pejorative label, just a straight factual description. 63 central banks (including the RBNZ) are the shareholders and the institution exists primarily to generate material for, and host meetings of, central bankers. They collate statistics and generate research with a central banking focus. They still provide some financial services to central banks. The chief executive (“General Manager”) is chosen from the ranks of highly-regarded senior central bankers (the current incumbent, Agustin Carstens was (among other things) formerly Governor of the Bank of Mexico and Deputy Managing Director of the International Monetary Fund).
As I mentioned in yesterday’s post, Adrian Orr had been citing material published recently by the BIS in defence of his suggestion that central bank losses from discretionary interventions really don’t matter and are more of an “accounting issue” than an economic one. When that material came out last month I drew attention to it, and (briefly) to the limitations, on Twitter, but since the Governor suggests that the BIS has the answers I thought it might be worth taking another look and unpicking what is, and isn’t, there in the two short BIS pieces. The first is an op-ed from Carstens, published in the Financial Times but the full text of which is on the BIS website and the second is a six page note by several BIS staff “Why are central banks reporting losses? Does it matter?
The Carstens op-ed is short enough I can take it paragraph by paragraph.
It begins with the title “Central banks are not here to make profits”. That is both true and a distraction. First, hardly any aspects of what governments do exist to make money, Second, unlike most arms of government, central banks should typically be at least modestly profitable (as monopoly provider of zero interest banknotes and of residual liquidity to the financial system). Third, when discretionary interventions are being considered the likely profits or losses, and the associated risk to taxpayers, should be at least one part of the full assessment of the pros and cons. And, finally, when interventions are being evaluated ex post, financial outcomes should be at least one part of a full assessment. Costs and benefits both matter.
As context here, one might note that the Reserve Bank of New Zealand, which used to have a low-risk small and stable balance sheet, made a profit each and every year for decades (without support from taxpayer indemnities). It was the normal state of affairs (seignorage earnings, with some volatility up and down as the proceeds of the note issue and the Bank’s equity were typically held in government bonds).
But on to the text
Unlike businesses, central banks are designed to make money only in the most literal sense. They have a mandate to act in the public interest: to safeguard the value of the money they issue so that people can make financial decisions with confidence. The bottom line for central banks is not profit, but the public good.
As noted, this does not mark out central banks from other government entities. Resources used, risks assumed, need to be rigorously evaluated along with programme effectiveness.
Today, following an extraordinary period in economic history, some central banks are facing losses. This is particularly true if they bought assets such as bonds and other securities to stabilise their economies in response to recent crises. Many will not contribute to government coffers for years to come.
“believing that by doing so” they would stabilise their economies. Note that in a formal sense that final sentence is not true of the RBNZ, since the Crown indemnity means losses from their interventions are borne directly by the taxpayer, not via impaired central bank capital. More generally, whatever the formal arrangements – and they differ widely across countries – mark to market accounting reminds us that the best guess is that large losses have already happened.
Does this mean that central banks are unsound? The answer is “no”. Losses do not jeopardise the vital role played by these institutions, which can and have operated effectively with losses and negative equity. And the unique nature of central bank tools means that sometimes losses are the price to pay for meeting their objectives – to support growth and jobs, ensure stable prices and help keep the financial system safe and stable.
In normal times, it is possible for central banks to both fulfil their mandates and earn profits without taking on significant financial risk. Traditionally, being the unique issuer of money provides a reliable revenue stream. But central banks with large foreign exchange reserves, built to cushion external shocks, will often experience ups and downs in income from exchange rate fluctuations. This means they sometimes make losses when pursuing their goal of a stable currency.
Agree entirely with the first two sentences, but they aren’t really the point and I’m not aware of any serious observer arguing to the contrary. The third sentence is much much more arguable, and neither in the brief op-ed nor in the longer Bulletin does the BIS really defend the claim. Most discretionary central bank interventions, if justified at all, should be stabilising and thus profitable (eg the Bank of England bond market interventions late last year). As Carstens notes, for countries with large foreign reserves holdings, exchange rate fluctuations will typically generate substantial year to year gains and losses simply from passive holdings but if that is an issue for Switzerland (where the BIS is based) it isn’t for most of the advanced country central banks we usually compare the Reserve Bank to. And there is likely to be a difference in how one sees passive structural positions and active discretionary interventions.
In times of crisis, central banks may also need to take on additional risks. And they do so with their eyes wide open. One example is the purchases of government bonds, including those made during the great financial crisis and more recently during the Covid-19 pandemic, in order to avert economic disaster by supporting financial stability, keeping credit flowing and boosting economic activity.
All this simply asserts what it does not show. But there is also an important distinction, not drawn here, between interventions to help restore market functioning (the initial QE back in 2008, and the initial bond buying in March 2020) which, support them or not (there are, after all, some moral hazard risks), should typically be expected to be profitable, perhaps even on a risk-adjusted basis, and large scale bond-buying with the goal of influencing the entire level of the yield curve. There is little evidence that many central banks (notably the RBNZ) really did much serious advance analysis on the use of this tool, the financial risks associated with it, the likely effectiveness of it, let alone exit strategies. In a NZ context, we should give no weight to the suggestion in the final sentence that the LSAP was necessary to “avert economic disaster”.
In the last decade, with inflation and interest rates low for a long period, these bond purchases boosted income. In fact, some central banks were able to transfer unusually large profits to governments. But in the wake of the pandemic and given the invasion of Ukraine by Russia, inflation has returned. This requires higher interest rates to contain spiralling prices – and exposes central banks to losses related to assets purchased in past successful rescue efforts.
Here Carstens touches on one of the problems with the 2020 QE interventions: central banks, including our own, seem to have been lulled into a degree of complacency about the risks they were taking on by the fact that QE done in other countries in the wake of the 2008/09 recession had not ended up costing central banks or taxpayers lots of money because the longer-term trend of falling real interest rates had continued. There was, however, no reason to suppose it would do so indefinitely, and a continuation was in any case less likely with bond yields near 1 per cent than with bond yields at, say, 5 per cent.
Central banks should put purpose above profits. Would it make sense for a central bank with large foreign currency reserves to increase their value by haphazardly triggering a devaluation of its own currency just to generate a windfall? Or for a central bank with domestic currency assets to keep interest rates low, even in the face of high inflation, just to preserve low-cost funding and generate profits? Such actions would be wildly inappropriate, violate their mandates and destabilise the economy.
By this point in the article, handwaving and straw men are well and truly to the fore. The issue is much more about the risk analysis – financial and otherwise – undertaken before the initial discretionary intervention (and at each stage of it), not how one clears up the mess afterwards. No one I know has suggested central banks should not allow interest rates to rise simply to protect their own financial positions, but there are serious questions about whether those (known to be) highly risky asset swaps should have been done in the first place.
The soul of money is trust. To operate effectively, business must maintain the trust of investors. And central banks must maintain the trust of the public.
Governments also have a role to play in the face of today’s central banks’ losses. Because these institutions are ultimately backed by the state, trust in money requires sound government finances and good financial management.
Blah, blah, blah. But one might add that maintaining the trust of the public in a modern era typically involves both demonstrated competence, openness and transparency, and acknowledgement of errors – not just patting people on the head and telling them “don’t worry, its complex, we are the experts and we have it in hand” even as staggering real losses are run-up and realised.
Losses matter because they may inflict a bruise on public finances but a far greater injury would result from central banks neglecting their mandates in order to avoid a loss. The public, via elected officials, have given central banks the job of price and financial stability because of their enormous societal benefits. Now, and in the long term, the costs from central bank losses are insignificant compared to the costs of runaway inflation and prolonged economic crisis.
Perhaps that first phrase is key. Losses matter, they are real. Nowhere does Carstens suggest they are “just an accounting issue”. The rest of that paragraph is really just handwaving and distraction, culminating in that outrageously misleading final sentence which seeks to suggest that there is some inescapable tradeoff between “insignificant” central bank losses and “runaway inflation and prolonged economic crisis”. There simply isn’t – and starkly there clearly wasn’t when central banks like the RBNZ launched and kept up their highly risk bond buying in 2020. With hindsight – and no matter what people might have claimed to believe back then – large scale bond-buying kept on well into 2021 or in some cases 2022 did not keep us from “runaway inflation and prolonged economic crisis”. Instead, overall central bank responses to Covid delivered us a really severe outbreak of (core) inflation, which central banks are now grappling to get back down again.
The Carstens piece is best seen as distractive spin for central banks by the chief executive of their own club. That needn’t necessarily mean there are no useful points their own lobbyist could ever make, but there were almost none relevant to the issues at hand, or the challenges that have been posed to Orr, in this piece.
More generally – and if this is a central bankers’ conceit, they probably aren’t the only interest group to suffer this fault – there is no sense anywhere in the Carstens piece that central banks might ever make mistakes, that some interventions might be worthwhile and appropriate and others not. But when central banks have done even their core job so poorly over the last couple of years – see core inflation rates across much of the world – the absence seems particular notable.
I’m not going to attempt a similar paragraph by paragraph treatment of the longer BIS staff note. It has some useful material in it, particularly for those less familiar with these issues, even if it has a strong focus on “whether losses matter for a central bank” (as they note, losses do not compromise a central bank’s technical ability to fulfil its mandate), rather than whether they matter for taxpayers, citizens, and those wishing to hold a central bank to account. I wanted to pick up briefly just the last couple of paragraphs, on how central banks should respond to losses.
“Effective communication” does not include trying to spin public audiences or MPs with assertions that real economic losses – that leave taxpayers poorer – are “just an accounting issue”. It should not include handwaving assertions about the wider benefits being “multiples” of the losses. It should include careful analysis and research evaluating the actual macroeconomic impact, including by comparison with the gains that less financially risky interventions might have offered. It should include careful ex ante disclosed risk analysis (the case for which was all the stronger for central banks coming late to the QE party, like the RBNZ or RBA). And it should include explicit recognition by central banks that they can, and sometimes do, make mistakes, even substantial ones. None of that has characterised the Reserve Bank of New Zealand through this episode.
And what of that final paragraph? It has the feel of editorial spin. Although it has become common in this field to claim that big financial losses are sometimes the price that has to be paid, there is rarely any rigorous attempt made to demonstrate the truth of that claim in respect of discretionary ad hoc interventions like the LSAP (or peer programmes abroad). Central banking done well should be profitable, from the nature of the institution – not because a central bank sets out to maximise profits (such a beast would be dangerous indeed) but because of its position in the market/economy and the monopolies the state gives it central bank.
Finally, and reverting specifically to New Zealand, one of my consistent criticisms of the LSAP programme is that there is no evidence in any of the material that has been released that the Reserve Bank or Treasury ever conducted or provided a robust analysis of what could go wrong when seeking the approval of the Minister of Finance for such huge punts on the bond market (and punts they were).
As just one example, consider this Treasury report to the Minister dated 1 May 2020 (so six weeks after the LSAP had been launched, and well after the initial US-led disruption to bond markets had settled down), in support of the Reserve Bank’s bid to expand the LSAP programme, and increase the associated financial risks. Here is the relevant bit of the financial risks section
Written in a way to suggest the programme was more likely to make money than lose it (despite the record low interest rates at the time) and with a “large but plausible” downside scenario involving the OCR only getting back to 1 per cent by this year, with no attempt at all to offer tail risk estimates of the extent of the possible loss. It simply isn’t the sort of analysis that should prompt any degree of confidence in either the Reserve Bank or the Treasury.
But this is the sort of stuff that Orr apparently stands by 100 per cent, with no regrets for anything he or the MPC were responsible for.
19 thoughts on “Central bank losses and the BIS”
This is well above my pay grade, but normally when someone loses someone else gains, so in this case who/ what group of people gained?
The people in the private sector who offloaded the bonds to the Rb at v high prices and who will (private sector as a whole) buy them back at lower prices as the LSAP position is gradually unwound.
If they bought the bonds with the intention of holding them to maturity then there is no loss. As (at a Crown level) this is equivalent to the Govt buying the bonds at market value and then cancelling them. Which isn’t an uncommon way of reducing a central bank balance sheet, at least in part if not fully.
No. From a whole of Crown perspective what the LSAP purchases did was to shift a large proportion of net govt borrowing from the private sector to a floating rate basis, at a point where floating rates were at record lows. The losses to taxpayers are real, even if the RB held the bonds to maturity.
>>From a whole of Crown perspective what the LSAP purchases did was to shift a large proportion of net govt borrowing from the private sector to a floating rate basis, at a point where floating rates were at record lows. The losses to taxpayers are real, even if the RB held the bonds to maturity
Thanks for taking on these topics, Michael. They are such big ones, and so poorly followed and understood by media and the public. I read and puzzle over what I can. I find myself agreeing and disagreeing with this assessment. The mortgage backed securities purchases were made with new money by the US Fed, and I reckon the RBNZ program was modeled after that. What do they talk about in Davos, or BIS gatherings, if not how to do stuff like this.
So new money buys a bundle of mortgages off Westpac – realistic scenario? Interest rates go up. The mortgages are worth less and (if you sell them back to Westpac) you’ve handed them a nice profit. But not taxpayer money. The taxpayer does suffer through the general globally induced inflation and the higher interest rates, just not the $9B difference in valuation.
Even in your scenario it is a pure loss to the taxpayer (and gain to Westpac) to the extent of the difference in price. The actual scenario here is a bit more complex but it ends up amounting to the same thing. Remember that when the RB buys bond from the market it pays for them by creating settlement cash balances (“money” in your parlance) on which it pays a full market floating rate interest rate (the OCR). The RB is now holding low yielding bonds and paying increasingly high OCR rates on the balances created to fund the purchase.
>>Even in your scenario it is a pure loss to the taxpayer (and gain to Westpac) to the extent of the difference in price.
Thanks for your reply. I said MBS but bonds is more apt. Still could be Westpac – RBNZ (or NZDM) buys 100M bonds and sells them back to Westpac for 90M. $10M loss – agree. But if taxpayer funds were not used to fund the 100M purchase in the first place? When I refer to $100M of new money – or settlement cash balances – I am meaning the program wasn’t funded from tax revenues to begin with. Hence not really a loss to the taxpayer. Maybe that’s semantics. It is indeed a loss, just not one paid out of anyone’s PAYE.
And the losses will be borne by taxpayers even unto grandchildren of the present taxpayers. Taxpayers in NZ are almost a rare breed. No person who works in government directly , quangos , local government ,Crown Controlled Entities etc etc are in fact income taxpayers as their gross incomes are just part of the stupendous drain of losses of value from the risk taking productive sector and transferred to the state. We can only standby powerless and watch the Adrian Orr’s and his facsimiles , the CEO’s of Government brush away any enquiries as to their reckless behaviour with the nation’s capital as not worthy of any real response. Government Board appointees should have skin in the game and stand some personal losses when things go south.
We continue on the road to Honiara.
It might be thought that the RB’s bets will tend to offset over time, so it all tends to zero. This is true for the average outcome, as per the Law of Large Numbers. However, as taxpayers, we are on the receiving end of the sum of these bets, not the average. If the expected outcome is zero on each bet, the expected outcome on the sum is still zero. However, if the standard deviation of each bet is Q, and the RB makes N bets with independence between the outcomes, the standard deviation of the sum will be Q*(square root N). So, the risk to taxpayers just keep growing with the number of bets. Casinos are in a better position. With an expected outcome on each bet being positive for them, the expected outcome grows in proportion to the number of bets (N), whilst the standard deviation grows in proportion to the square root of N. So, as N grows, the chance of the casino losing in the aggregate goes to zero. It’s hard to imagine that the RB enjoys a positive expectation on each bet, as with a casino!
If only the RBNZ had the nous to realise global forces (relativities) would have done heavy lifting for them, dragging NZ bond yields down close to the levels we saw at the lows, without adding fuel to the fire with $50 billion purchases (now worth $41billion). All they had to do was slash the OCR and enjoy the free ride….
In many ways the funding for lending scheme (which went alongside the asset purchases) was at least as bad. Basically the RBNZ said to the banks: “here’s a whole lot of cheap, subsidised funding for you to lend out. The excess profits (relative to raising funds in the market) are yours to keep.” I’m not sure how much cheaper the funding would have been relative to raising it in the market but even if was just 100 basis points cheaper, the banks would have pocketed hundreds of millions in extra profit. However, the RB continues to downplay the significance of the scheme and never comes clean on these calculations.
I’ve been less negative on the FLP, setting aside the absurdity of keeping it running all last year, and am sceptical that the subsidies are as large as you suggest. Do you have a framework or stylised numbers to get to your “hundreds of millions” of excess profits (bearing in mind that ROE for banks has done nothing startling in the last couple of years)?
The best measure of the subsidy is the net interest earned on funds advanced under the FLP over and above the net interest the banks would have earned in the event they had raised the funds at a market rate.
According to the latest RB data, the aggregate outstanding funding advanced under the FLP is $19.021 billion. If that funding cost, on average, 100bp less than market funding, then the subsidy would amount to $190 million per annum. Since the FLP offered 3 year funding, this would cumulate to around $570 million once the funding terms had expired (aka “hundreds”).
Of course, we’re only talking 2% of banking system profits here and the funding has been spread out over time so it is not really going to turn the dial on ROE. But this is still a subsidy banks are able to keep. Enough for a few dealing room bonuses. (More research would be needed to calibrate the actual cost of market funds over the period during which FLP applied. 100bp might be too light.)
Thanks. I suppose – contrary to your last line – that my prior would be that the extent of the subsidy was well under 100 bps. Would banks offering up highly-rated collateral really otherwise have paid 100bps over OCR for floating rate funding?
Reblogged this on Utopia, you are standing in it!.
You give it to him Michael , Orr is infatuated with his own importance bolstered by his years as a “Master of the Universe’ running the NZSF
$9 billion? Hahaha. Chicken feed.
I take your 9bn, and I raise it by $72 billion!
2018 $81 bn = 27.6% government spending to GDP
2022 $162 bn = 44.7% government spending to GDP
-> $81 billion increase in 4 years! Government spending doubled in four years in nominal terms!
2023 $150 bn = 38.4% government spending to GDP (estimate).
I’m not sure whether these figures include the $9 bn loss by the world’s worst trader, Mr. Orr, or not.
Let’s hope so…
“Thanks. I suppose – contrary to your last line – that my prior would be that the extent of the subsidy was well under 100 bps. Would banks offering up highly-rated collateral really otherwise have paid 100bps over OCR for floating rate funding?”
Probably not. But my assumption is that FLP enabled banks to raise less non-securitised term funding and retail deposits where the margins (and effective subsidy) are clearly larger. You have to use the right comparator.