Measuring Government Bond Turnover in Australia Using Austraclear Data

Introduction

There are around $1.3 trillion of government bonds outstanding in Australia issued by the Australian Government and the states and territories, which is equivalent to around 60 per cent of GDP. Government bonds play a pivotal role in the Australian financial system. Apart from their role of funding governments, their yields are reference rates that help to anchor yields on other securities. Government bonds are considered safe and liquid assets and are held by a variety of investors.

Liquidity is a difficult concept to measure. One indicator of liquidity is turnover – the amount bought or sold over a given time period. The Reserve Bank receives daily transaction-level data from Austraclear, the settlement system for Australian dollar fixed income securities in Australia, and turnover can be calculated using these data. Previous analysis using Austraclear data estimated a monthly turnover ratio for Australian Government Securities (AGS) of around 110 per cent and around 40 per cent for state and local government bonds (semis) (Guo and Zhang 2020).

However, as acknowledged in Guo and Zhang (2020), these estimates included some repurchase (repo) transactions, which had been recorded in Austraclear as outright trades. A repo involves one party selling a security to another party, with an agreement to buy back the security on a predetermined date and at a specified price. These transactions are akin to a collateralised loan, and so do not represent genuine secondary market trading. Government securities are frequently used in private repo transactions, and so the previous analysis overstated the secondary market turnover of these securities.

This article looks at the effects of a change in November 2021 to Austraclear reporting standards, resulting in repo transactions being excluded more accurately. This change coincided with an appreciable drop in measured turnover, suggesting repo activity was a significant part of previously estimated turnover. This is supported by an algorithm, designed to identify repo transactions, that suggests much of the turnover prior to November 2021 was accounted for by repo transactions.

This article also shows that the new estimates align with survey data on turnover published by the Australian Office of Financial Management (AOFM).

Data and methodology

We use Austraclear settlement data to measure turnover. Security settlements are lodged in Austraclear when counterparties exchange a security registered in Austraclear for cash. The Bank began receiving daily data from Austraclear in November 2020.

The data do not represent all trades in the wholesale debt market – they only include trades settled between counterparties that use separate Austraclear accounts. Moreover, transactions of Australian dollar denominated securities may be settled through clearing systems other than Austraclear, such as Euroclear or Clearstream.

Our focus is on the longer term fixed income market, so we only consider securities with an original maturity of over one year. We also exclude trades associated with monetary policy; between 2020 and 2022 the Bank was an active participant in government bond markets due to some of the policies introduced during the COVID-19 pandemic, but these interventions were temporary and do not represent typical activity in the market.

To identify repos prior to November 2021, we revise the detection algorithm described in Garvin (2018). The intuition behind the algorithm is to identify groups of cash movements that resemble a loan followed by a repayment with interest. Relative to Garvin (2018), our detection algorithm uses some revised conditions, specified in Appendix A.

Excluding repo from turnover

Raw turnover data show that prior to November 2021, turnover in the AGS market averaged between 60 and 120 per cent per month (Graph 1). Turnover for inflation indexed bonds was a bit lower, while for semis it was around 30 to 40 per cent per month (Graph 2; Graph 3). A steep drop in turnover occurred on 22 November 2021 in response to Austraclear’s change in reporting standards (see below). After that point, the monthly turnover ratio for AGS has been around 15 to 20 per cent and for semis it has been around 10 per cent.

On 22 November 2021 a ‘no action’ relief expired that applied to Austraclear’s regulations. After that point, repo transactions could no longer be booked as outright trades without sanction. The fall in turnover after that date can therefore be used to estimate the amount of repo transactions that were previously recorded as outright trades.

This conclusion is supported using the algorithm from Garvin (2018) to identify repo trades. The decline in turnover is accounted for mostly by the decline in trades identified by the algorithm as repo transactions (Graph 4). After the change, there are very few transactions left that the algorithm identifies as repos – those remaining are either repos conducted as outright trades in contravention of Austraclear’s regulations, or false positives from the algorithm.

Trades not identified as repo still decline after November 2021, which is most clearly seen when trades are aggregated at the monthly level (Graph 5). The most likely explanation for this is that prior to November 2021 the algorithm is failing to identify some repo transactions. Therefore, in what follows, we focus on the post-November 2021 sample. Given that the algorithm identifies a trivial number of repo transactions after that point, we use the raw data.

New turnover estimates

After November 2021, the average monthly turnover ratio for AGS is around 15 to 20 per cent, and for semis it is around 10 per cent (Graph 6). Trade frequency (i.e. for each bond line, the number of times that bond line is transacted over a period of time) is around 350 trades per month for Treasury bonds, about two-thirds that for Treasury indexed bonds, and 27 times per month for semis (Table 1). This is substantially less than estimated by Guo and Zhang (2020). The average trade size is also lower than estimated by Guo and Zhang (2020).

Guo and Zhang (2020) considered turnover ratios by size of the bond lines, concluding that turnover ratios increased by size. That does not seem to be the case for the post-November 2021 sample, where there is no strong relationship between size and turnover (Graph 7). However, turnover in dollar terms is higher for larger bond lines. This is driven by a higher number of trades, with trade size again not varying much by the size of the bond line (Graph 8; Graph 9).

Relationship to Australian Office of Financial Management estimates

The AOFM conduct a survey of market intermediaries detailing their turnover in AGS every month. While the scope of the AOFM survey is a little different to that of the Austraclear data, the new estimates presented in this article are in line with the AOFM survey (Graph 10).

Conclusion

New estimates using Austraclear data for monthly turnover ratios for AGS and semis are considerably lower than estimates published previously. This is due to the earlier estimates being based on Austraclear data that included repo transactions at the time. A change to Austraclear reporting standards in November 2021 has resulted in repo transactions being excluded more accurately. A detection algorithm used to identify repo transactions from the Austraclear data supports the new turnover estimates, which also better align with turnover data from the AOFM.

Source: Reserve Bank of Australia [2021]