Chris Joye at Coolabah with the note:
In the AFR this weekend I write that after months of debate about whether the Reserve Bank of Australia would increase or decrease its stimulus in recognition of the COVID-induced recession, Martin Place delivered on its promise to maintain a “flexible” and open-ended approach to its government bond purchases (aka quantitative easing).
To make matters more interesting, the Australian Prudential Regulation Authority shocked the banking system on Friday by announcing that the $140 billion Committed Liquidity Facility must be replaced by purchases of government bonds as is standard across the rest of the world. As we previously speculated, closing the CLF perfectly dovetails with the RBA’s bond buying.
There are 4600 words left in this subscriber-only article.
Start your free 14-day trial today!
Back in July the RBA announced a “taper” of its bond purchases from $5 billion to $4 billion per week, commencing in September with a quarterly review. Media cited this as emulating the Bank of Canada’s approach to QE even though core inflation in Canada is running at 3 per cent (well above its 2 per cent target) whereas inflation in Australia is materially below the RBA’s target.
Following the RBA’s July decision, the market assumed it would drop bond purchases by $1 billion per week each quarter. This implied total purchases for the RBA’s third, open-ended QE program (aka QE3) of about $105 billion, which was within the ball-park of economist expectations.
Precisely what the market was pricing in for QE3 was a different matter. Investors were punting on a harder taper summing to only $50 billion to $75 billion in size, which was at the lower-end of the more hawkish economist forecasts, such as CBA ($50 billion), UBS (up to $75 billion), and NAB ($75 billion).
Governor Phil Lowe clearly stated that the RBA was prepared to increase or decrease his bond purchases in accordance with economic circumstances, and that is precisely what the adroit monetary policy mandarin delivered.
With the revised commitment in September to keep the RBA’s bond purchase pace at $4 billion per week all the way through to February 2021, Martin Place has increased the size of QE3 from $105 billion to $147 billion.
This reflects the difference between the original proposal to cut the $4 billion per week bond purchases by $1 billion each quarter, which is known as a 4/3/2/1 schedule, and the new plan of keeping the bond purchases constant at $4 billion per week until February following which they can be tapered according to a 4/4/3/2/1 schedule.
Note also that the $147 billion estimate for QE3 is now substantially larger than the $100 billion worth of purchases executed under the preceding QE1 and QE2 programs that began in November 2020 and April 2021, respectively.
Notwithstanding the differences in Australia and Canada’s inflation outlook, the RBA’s beefed-up QE3 program does emulate what its North American peer has done. The Bank of Canada first tapered in October 2020 down from $5 billion to $4 billion per week, a pace it held until April 2021 when it tapered again to $3 billion per week. Three months later it further decelerated to $2 billion per week, with the consensus predicting a final quarter worth of $1 billion per week of purchases beginning in October.
After this is complete, the Bank of Canada has committed to reinvest maturing bonds to ensure that it does not shrink its balance-sheet. As the RBA has explained, it is the total size of the central bank’s bond purchases rather than the weekly flow of investments that signals the magnitude of the stimulus bequeathed.
In Australia, QE has been much more successful than anyone, including the RBA, ever imagined, crushing long-term interest rates and the Aussie dollar down to way below their estimated fair market value, and by doing so bolstering economic growth.
It also helps solve an important financial stability puzzle for the RBA that has plagued it for decades: how to avoid blowing house price bubbles, such as those that emerged in 2003 and between 2012 and 2017, when relying exclusively on its overnight cash rate in an economy with an unusually large proportion of variable-rate home loan borrowers.
In contrast to economies dominated by fixed-rate debt, most Australian borrowers immediately benefit when the RBA slashes its cash rate via increases in their purchasing power, which gets quickly capitalised into house prices.
In future shocks, the RBA can put downward pressure on both short- and long-term interest rates, and the Aussie dollar, using both its cash rate and QE, which will provide for a more balanced policy platform.
It might mean that the cash rate may not fall as far as it has in the past given the RBA would be augmenting this stimulus with the extra support provided by putting downward pressure on the Aussie dollar through bond purchases that lower long-term rates.
The RBA’s very dovish taper in September is not, therefore, a reduction in monetary support: Governor Lowe materially lifted the total quantum of bond purchases that the market was pricing for QE3 by about 40 per cent (or $40 billion). And that assumes away any future adversity that elongates QE3 beyond the current central case of a Canadian-style 4/4/3/2/1 schedule.
This gradual glide path reflects what central banks have learned is rule No. 2 of any QE program: there is little upside, and considerable downside, when you start tapering bond purchases. Put differently, you can trigger tantrums if you taper too quickly.
Rule No. 1 is to always surprise on the upside with the size of your QE, as the RBA has consistently done—puny programs can be counterproductive, as we saw during the global financial crisis and when the first overseas efforts were announced in the great virus crisis in March 2020.
One speculated motive for the RBA’s taper has been that it was concerned about owning an excessively large share of the government bond market, and crowding out banks. Yet new data from the Australian Prudential Regulation Authority shows that Aussie banks are holding the lowest quantity of Commonwealth government bonds since December 2018, and almost $100 billion less than what they held in mid 2020. This is despite the Commonwealth bond market growing by $250 billion since 2018.
That means that the banks’ share of the Commonwealth bond market has plummeted from 23 per cent last year to just 10 per cent today according to Signal Macro’s chief economist Matt Johnson. Even after controlling for the banks selling bonds to the RBA, their market share has still dropped 6 percentage points.
Johnson highlights that contrary to many banks’ claims that the RBA’s bond purchases would hurt liquidity, annual turnover of bonds in the Commonwealth bond market has increased from 260 per cent prior to 2020 to 300 per cent today.
In huge related news on Friday morning, Australia’s best regulator, Wayne Byres, announced that APRA had boldly, and entirely appropriately, decided to eliminate the taxpayer subsidies enabled via the $139 billion Committed Liquidity Facility.
The innovative CLF was created by the RBA and APRA at a time when there were not enough government bonds for banks to hold as emergency liquid assets. Under the CLF, banks could instead buy other banks’ (higher yielding) bonds, and use their own uber-lucrative loans, as a substitute for lower-yield, yet riskless, government bonds.
APRA has given the banks just under four months to phase-out the first chunk of CLF assets, which must be completely eliminated by 31 December 2022. Since this will eventually compel banks to buy vast quantities of government bonds for emergency liquidity purposes, APRA should revisit the rule that requires them to unnecessarily hold extra capital to protect against losses on these assets.
There is a regulatory contradiction that must be resolved where on the one hand the RBA guarantees the liquidity of government bonds, and APRA says that they have zero risk of credit default loss, and yet on the other hand banks have to hold additional capital against these riskless securities to insure against rare, unrealised “mark-to-market” losses on them.
This is even more weird considering government bonds are held by banks as a hold-to-maturity asset that would only ever be sold to the RBA in an emergency. We should reasonably encourage, rather than discourage, banks from holding these assets in preference to competing deposits issued by the RBA that pay zero interest.
There has been very big news today for both the government and bank bond markets in what would have come as a significant shock to the banking system and almost all fixed-income investors: Australia’s fearless banking regulator, in concert with the RBA, has effectively zeroed the massive taxpayer subsidy that was the $139 billion Committed Liquidity Facility (CLF), which will force banks to replace this existing portfolio of illiquid and relatively risky bank loans, bank bonds and residential mortgage-backed securities (RMBS) with much lower risk (in fact, riskless) and more liquid Commonwealth and State government bonds.
This also has quite far-reaching and important investment ramifications for the fixed-income market, which few if any folks have been thinking about of late (save, if I can say so, a few of us) as everyone was working on the basis the CLF would be around “for many years to come”, and if it was ever phased-out, it would be reduced over a “very slow glide-path of 3-5 years”. That was the mantra being relentlessly pushed by the banks and trading desks…
For investors, this is very good news for Commonwealth and State government bonds, which will benefit from a lower cost of capital, and less positive for bank-issued, Aussie dollar senior bonds and AAA rated RMBS, which will have undergo a normalisation in their credit spreads back to at least the levels observed prior to the COVID crisis.
But fear not: bank paper and RMBS is in incredibly high demand globally, and the increase in spreads, once it occurs, should offer attractive future payoffs. (Bank paper and RMBS have had their credit spreads artificially suppressed by banks buying one another’s bonds, and RMBS, for their CLF portfolios.)
You can read a detailed recap of what the CLF is from us here. In short, it was a portfolio of internal bank loans, bank bonds and RMBS that a bank could hold instead of buying Commonwealth and State government bonds. It was created for a world when Australia had very little public debt. Because of the explosion in government debt to over $1.5 trillion, it is no longer needed, as Signal Macro’s Matt Johnson has clearly demonstrated here and here.
It was a terrific regulatory arbitrage, because the banks themselves were the first to admit that they earned much more money on their portfolio of internal loans, bank bonds, and RMBS than they could earn on holding Commonwealth and State government bonds.
And Australia was globally unique in allowing banks to buy each other’s securities instead of holding government bonds in this way. It was a great caper for as long as it lasted: the banks would regularly buy half-to-two-thirds of one another’s bond issues, keeping their cost of capital artificially low while maximising their returns on equity.
While Coolabah had forecast that the CLF would be quickly zeroed given APRA’s very clear and public remarks on the subject, as had others like Signal Macro’s Matthew Johnson and UBS’s Giulia Specchia, the banking system was mostly very resistant to the proposition.
APRA has over the last 12 months repeatedly written to the banks stating that the CLF would likely not exist in the “foreseeable future” given the explosion in the availability of government debt.
And yet as public debt issuance continued, the banks, and investors, were incredibly complacent around the risks of APRA delivering on its statements. And boy has it done so.
In a pithy release this morning, APRA stated that the banks will not be able to rely on the CLF to meet their minimum Liquidity Coverage Ratio of 100% by 1 January 2022 (ie, within four months). And then they have 12 months to completely eliminate their reliance on the CLF (ie, by end 2022). In APRA’s words:
APRA expects ADIs to purchase the High Quality Liquid Assets (government bonds) necessary to eliminate the need for the CLF. Specifically, no ADI should rely on the CLF to meet its minimum 100 per cent LCR requirement from the beginning of 2022 (although ADIs may continue to count any remaining CLF as part of their liquidity buffer). ADIs should then reduce their use of the CLF to zero by the end of 2022.
In a companion piece of blockbuster news, the RBA has lifted its estimate of the total share of the government bond market that banks can hold from 30% to 35%. This only underscores the the point that the CLF is unlikely to make a comeback any time in the next decade.
This is especially true because APRA has also created a new backup Contingent Liquidity Proposal that banks have to have in place by the end of this year, which will be worth another 30% of their emergency liquidity. Coolabah had previously speculated that this new requirement might presage the disappearance of the CLF.
When we have surveyed bank treasurers in the past about what they would do if the CLF was eliminated, they have consistently responded that they would have to issue wholesale debt and buy government bonds, as other banks around the world are required to do.
While you will hear stories about the banks holding hundreds of billions of dollars of excess cash on deposit at the RBA, the truth is that Aussie banks are not carrying that much excess liquidity right now.
APRA mandates that they have to hold high quality liquid assets (HQLA) worth at least 100% of the cash-outflows that banks would suffer in a 30 day crisis. HQLA includes cash on deposit at the RBA plus Commonwealth and State government bonds.
The banks then internally run minimum LCR targets above 125%. And all the banks are pretty close to these minimums except ANZ and Macquarie. As we have previously noted, only ANZ’s Adrian Went has been compliant with APRA’s APS 210 standard that requires banks to buy as much HQLA as they can to meet their LCR targets before requesting help via the CLF (and to only request such help if sufficient HQLA does not exist).
In recent years ANZ has worked hard to radically reduce its reliance on the CLF in contrast to its larger peers by running down its portfolio of CLF assets by more than $30 billion since 2019 compared to Westpac, which only reduced its CLF by a miserly $15 billion.
What that means is that as the CLF is zeroed over the next year, banks are going to have to replace it with new Commonwealth and State government bonds. And to pay for these assets, the banks will have to issue debt.
So for investors, the key take-aways are probably as follows:
- Banks were one of the biggest buyers of 5-year bank senior bonds and 3-to-5-year AAA rated bank and non-bank issued RMBS in Aussie dollars for their CLF books. This demand is likely to shrink dramatically, normalising the credit spreads on 5-year bank senior paper and 3-year AAA rated RMBS back to pre-COVID levels (eg, at least 30 basis points higher than current spreads, which would still be very, very low for the post-GFC period). One could see 5-year major bank senior spreads back at circa 70bps above BBSW and AAA rated, 3 year major bank RMBS back to say 80 to 90bps over BBSW;
- Banks have, somewhat bizarrely, sold over $100 billion of government bonds since mid 2020, as Matt Johnson shows here. They are going to now be net buyers of Commonwealth and State government bonds in potentially very large size, which should support these sectors. UBS’s excellent analyst Giulia Specchia, who called the shuttering of the CLF only this week, writes (where “semis” are State government bonds):
We expect Semis’ gross issuance to be ~$94bn in FY-22, and we expect the RBA will buy $30bn. This means net-net issuance is on track to be $41bn in FY-22 (Figure 4) – unchanged from last fiscal year. While this is material, the unwound of the CLF should help support demand for Semis. Notably, the CLF reduction from now until the end of Dec-21 is worth ~$35bn; and another ~$52bn over the period Jan-Jun-22. That is $87bn of potential demand. While not all of the reduction in the CLF has to be translated into demand for Semis (ADIs could just hold cash at the RBA), we still think that – especially at the current level of spreads – it makes sense for ADIs to buy Semi-government bonds. Amid ongoing support from RBA QE, and likely strong demand from ADIs over the next months, we think that risk reward has shifted in favour of a tighter Semi-bond spread.
Now you will read lots of narratives as to why “this is no big deal”. One myth is that banks are carrying massive excess cash at the RBA, and don’t need to buy any HQLA at all. That is wrong: they are running LCRs close to their internal minimums of 125%, and will have to replace the lost $139 billion from the CLF with new HQLA. That is why APRA explicitly said today “APRA expects ADIs to purchase the High Quality Liquid Assets (government bonds) necessary to eliminate the need for the CLF”.
Another argument might be that banks can just issue the debt they need to replace the CLF and then sit the cash on deposit at the RBA. The problem with this is that these deposits pay no interest. So cash at the RBA is a huge revenue and net interest margin drag for the banks. It is an especially big drag when you are issuing expensive wholesale bonds to buy HQLA, and then putting that money on deposit at the RBA earning nothing
Banks therefore have a very strong commercial incentive to spend the cash on deposit at the RBA on HQLA (ie, Commonwealth and State government bonds) that pay them real interest above the 0.06% pa cost of the wholesale liability levy that bank funding attracts. This is the minimum hurdle rate of interest that Commonwealth and State government bonds have to pay banks in order for them to prefer these assets over cash at the RBA.
Another extremely silly suggestion is that APRA will simply allow the banks to lower their internal minimum target LCRs from 125% currently to some lesser number to offset the loss of the CLF. That was an idea bandied around by one major bank researcher.
But it is oxymoronic: that logic suggests that when the banks were buying one another’s bonds, they could run LCRs over 125%, but now they have to hold real HQLA, they are going to run materially lower LCRs.
There is zero chance this will happen because banks themselves internally target LCRs of more than 125% as a buffer beyond the hard APRA hurdle of 100% (as they do with all their regulatory hurdles). It would be wishful thinking that either bank boards or APRA would allow banks to get away with lower LCRs.
A final idea is that the RBA’s QE program will create the HQLA for the banks. Yet we have seen LCRs declining over the last 12 months despite persistent QE. And the cash held at the RBA is not distributed evenly with, for example, custodial banks accumulating a lot. There is once again a strong commercial incentive for banks to spend zero-interest earning cash at the RBA to avoid being the bank left accumulating it.
Finally, after mid 2022, the excess cash at the RBA will start dropping fast. The banks will be repaying the $188 billion they owe the RBA under the Term Funding Facility, which destroys excess cash on deposit at the RBA. And the RBA’s portfolio of bonds will eventually start maturing, further reducing excess deposits that banks hold with it.
Combined with ongoing balance-sheet growth and deposit growth, both of which normally necessitate additional HQLA, banks are likely to have long and strong demand for Commonwealth and State government bonds.