[This is an edited version of a note first posted on livewire]
A scarcity of Australian government debt meant that the RBA and APRA had to be creative when implementing the post-GFC global banking reforms, known as Basel 3. Their solution to the scarcity of government bonds was for the RBA to commit to give banks liquidity against alternative collateral in what became known as the Committed Liquidity Facility (CLF).
A large increase in the size of the government debt market, and a massive increase in the amount of cash in the system, means that there are now enough High Quality Liquidity Assets (HQLA) to meet the banking system’s needs. That is, APRA could now eliminate the CLF, if it wanted to do so. Given institutional conservatism, the phasing out of the CLF over a few years seems more likely than total elimination in 2022 – but elimination cannot be ruled out.
When the CLF was implemented in 2015 there were not enough government bonds to go around. The size of the high-quality government bond market was small, the repo market was immature, and foreign ownership was high. As a result, there was a reasonable fear that the high-quality government bond market might become illiquid if Australian Deposit taking Institutions (ADIs, which are predominately Banks) purchased too many bonds.
To solve this problem, the RBA decided to cap ADI ownership of high-quality government bonds as share of the market (initially at 25%) and provided ADIs with a CLF to make up the balance. The combination of these two sources of liquidity ensured that ADIs were able to meet their regulatory ratios, and specifically the Basel 3 Liquidity Coverage Ratio (LCR), while at the same time preventing the high-quality government bond market from becoming crowded and illiquid.
Things have changed since 2015.
The high-quality bond market is larger, the repo market has matured, and foreign ownership is lower. More importantly, the RBA’s QE program has created a large new stock of liquid assets: Exchange Settlement (ES) Balances, which are cash deposits at the RBA. As a result there has been a large increase in the quantity of HQLA over the past 18 months (even after adjusting to avoid double counting; more on this below).
ES Balances are now a large source of HQLA and they ought to be counted.
APRA and the RBA quickly responded to the developments in the government bond market. They first boosted the share of the high-quality government bond market that ADIs were permitted to hold to 30% (reflecting a larger market, better repo and lower foreign ownership). Next, as the bond market grew, they accelerated the reduction of the CLF over 2020 and 2021. A larger market means more bonds; so, the free float goes up in dollar terms even if banks hold a larger share of the market.
However, they have not made an adjustment for the growth of ES Balances, which seems odd given that they are ‘the ultimate form of liquidity’ (RBA Deputy Governor Guy Debelle speaking to the House Economics Committee on 6 August).
Conservatism seems to be the main explanation. An RBA Bulletin article explains that “APRA assumed CLF bank’s surplus ESA balances would be around the (significantly lower) levels of previous years when calculating the size of the CLF for 2021”.
There is a strong case for changing this assumption for 2022.
ES Balances are high, rising, and will not return to their pre-pandemic level for a long time.
If the RBA stopped QE today, Surplus ES balances would remain around $350bn until H2’23; and still be over $100bn for most of 2025. Assuming QE continues in line with market expectations, Surplus ES balances will rise to around $500bn in 2022, and remain above $200bn in 2025. Given the RBA’s large portfolio of bonds that mature after 2030, Surplus ES balances are unlikely to return to prior levels until the end of the decade.
Given that ES Balances will remain elevated for some time, APRA and the RBA ought to include them in their calculation of available HQLA when calculating the size of the CLF.
How can ES Balances be both in & out?
The current status of ES Balances is confusing.
When APRA is counting bank liquids, for the purpose of making sure that the LCR is sufficiently high, APRA includes the ES Balances a bank holds. However, when APRA & the RBA are measuring the quantity of HQLA in the system, to gauge the size of the CLF, they don’t count ES Balances. This makes little sense.
ES Balances are a large and growing component of ADI liquidity portfolios. While there are non-ADI holders of ES Balances, the closed nature of the system means that LCR ADIs end up holding a large share of surplus ES cash. The outlook for strong growth of RBA Surplus ES Balances means we can be reasonably sure that ES Balance holdings (which appear in the CB Deposit line of the APRA stats) of LCR ADIs will continue to grow over time.
Larger ADIs, which must comply with Basel 3 LCR regulations, have increased their holdings of Central Bank Deposits (of which ES Balances at the RBA are the largest component) by over $150bn since 2019. This has taken the share of ES Balances in the aggregate LCR ADI liquidity stack up 17ppts over the past two years (from 22% in Q1’19 to 39% in Q1’21).
Are there any tricks?
Followers of this topic will be familiar with the normal convention. APRA forecasts out how much liquidity the LCR ADIs need (Net Cash Outflows, plus a buffer), the RBA forecasts the size of the high-quality government bond market, and the difference is the CLF. On this basis, even with the large increase in the bond market, the CLF is needed for some time.
However, QE changes this calculation. When the RBA buys bonds it transforms high-quality government bonds, where liquidity considerations mean that ADIs must not own too much, into ES Balances, where concentration isn’t a concern.
So QE increases the quantity of liquid assets that banks reasonably can hold.
There’s no reason that ADIs shouldn’t hold 90% of the ES Balances in their HQLA pools. Prior to 2020, Surplus ES Balances were seldom larger than $3bn. So, even if ADIs held 90% of Surplus ES balances there would be more than enough cash in the system (this remains true even as Surplus ES Balances decline after 2022).
Given that ES Balances are likely to be high for a long time, and that there’s no reason ADIs should not hold a very large proportion of them, it follows that the RBA and APRA ought to include ES Balances when they estimate the available stock of HQLA. They are already counted in liquidity stacks — it makes no sense for the RBA and APRA to leave them out of the available HQLA calculation.
There is, however, one adjustment that must be made if they do so. The RBA’s holdings of bonds must be subtracted from the size of the bond market when calculating the available pool of government bonds. Otherwise, bonds the RBA owns would be counted twice: once as part of the bond market; and a second time when ES Balances are added. This is the QE-adjustment that was made in chart 1.
How many bonds do they own?
LCR ADIs have been selling high-quality government bonds (likely substituting into ES Balances). According to APRA data, LCR ADIs currently hold around $284bn of high-quality bonds (Commonwealth and State Government bonds). This is down from a peak of $360bn in 2020.
If ADI holdings of government bonds remain stable at $284bn, the large increase in ES Balances means that there’s more than enough HQLA to keep system-wide LCR above 125% over the next few years.
Assuming $284bn of bonds, LCR ADIs would need to hold only 70% of the $485bn stock of ES Balances to be self-sufficient in 2022. This allows for a $50bn increase in Net Cash Outflows (NCOs), to $500bn (NCOs are to the amount of cash that might leave a bank in a 30-day crisis).
As ES balances decline, ADIs will need to buy more bonds – but a growing market means that isn’t a problem. Assuming ADIs held no more than 90% of outstanding ES Balances, their required share of government bonds (net of RBA bond holdings) would not rise above the 30% cap the RBA has set for LCR ADI ownership of bonds (ACGBs and Semis).
Yes is could be closed (and ANZ is doing it!)
Of course, LCR ADIs might not like to hold large amounts of zero-interest ES Balances. But that’s orthogonal to the regulatory question. The relevant question is capacity.
APRA’s regulatory standard APS 210 requires ADIs to make every reasonable effort to manage their own liquidity risk through their own balance sheet before applying for a CLF for LCR purposes. The increase in the availability of HQLA, due to QE-fueled growth of ES Balances, means there are enough high-quality liquid assets for LCR ADIs to be entirely self-sufficient in the management of their liquidity risk.
The ANZ Bank’s Pillar 3 report shows that it is a practical option. Over the past two years, ANZ has reduced their CLF by almost $40bn, to $10.7bn. In doing so, they have lowered the share of the CLF in their stock of liquid assets from 25% to less than 5%.
If ANZ can do it, so might other banks. It follows that APRA and the RBA could eliminate the CLF starting in 2022 – if they wanted to.
The main reason for conservatism on ES Balances is uncertainty about their future level. However, with the COVID/Delta outbreak slowing the economy, it seems more likely that both Bond Issuance and QE will be larger than has been assumed in these calculations. Both developments favor a smaller CLF.
On the other hand, elimination of the CLF would be entirely consistent with APRA’s repeated guidance that “it would be reasonable to expect that if government securities outstanding continue to increase beyond 2021, the CLF may no longer be required in the foreseeable future”.
Government securities have continued to increase, and will continue do so for many years.
So the question is, do they want to close the CLF?
A large cut with a multi-year phase out seems more likely than total elimination in 2022. But ANZ shows elimination is plausible.