Australian Government Sovereign Debt: Are we there yet? What more can be expected in terms of developing the market? – Presentation to the Australian Business Economists luncheon
26 May 2015
Presentation delivered by Rob Nicholl – CEO, AOFM
Thank you once again to the ABE for the opportunity to give this annual address.
I have aimed to use this event to highlight our thinking on the coming year, and to give some insight into how we think about certain issues and what we have planned – mostly in terms of market development as this is where the colour of our issuance programs has tended to derive.
I hope this provides information that is useful, clear and easy to interpret. At the same time I am confident that you and our broader audience appreciate that there are limits to what I can say – first because it is not my role (nor that of the AOFM) to comment on policy issues; and secondly because it doesn’t make sense for me to be highly prescriptive about planned actions but, for reasons best (and usually only) known to us, may retreat from. This would go against our desire to appear consistent.
Over the last year I have attended a number of forums in which my US Treasury equivalent has been speaking, and on each occasion it has been clear that its operations are based very much on deliberate and concerted efforts to conduct issuance programs on a regular and ‘no surprises’ basis. That is not to say they too don’t retain sensible operational flexibility; but they have sought to further their credibility on the basis of being predictable (within reason), transparent and consistent. I find this interesting given that US Treasuries are considered THE cornerstone global market and THE benchmark for liquidity and investor diversity. I can assure you though that they do not take this for granted.
From my experience with investor engagement I am constantly reminded of the value that the ‘buy-side’ places on the consistency of issuer performance. Therefore, the importance of this is not lost on us and we are conscious of the need to be adaptable and forward-looking. However, I must say at times we feel that striving to meet a higher standard is like chasing a moving target.
In some respects that is related to the topic I have chosen for today – can we claim we have achieved sufficient market developments – and if so by what criteria should we judge that? Alternatively, are we on a continuum (both operationally and in terms of everyone else’s expectations) that will make it difficult to assess when we should stop thinking about further developments and simply follow repeatable issuance patterns?
Five years ago our issuance programs were building rapidly and the focus was almost exclusively on ‘if and how’ relatively large issuance programs could be readily absorbed. But no sooner had this appeared under control, when attention turned to whether new issuance would be adequate to satisfy a growing demand for high quality sovereign paper when the Budget was returned to surplus. Whether or not sufficient liquidly in the market could be maintained to satisfy investor confidence under these conditions was a related point of discussion.
As elevated issuance programs continued, expectations then began to emerge about what features the AGS market should exhibit as it continued to grow in size. A policy commitment had also been made to maintain a ‘linker market’ – adding another development dimension to consider.
We were initially cautious in our approach to extending the yield curve because we were unsure about the level of demand for our paper longer than required for supporting the 10-year futures contract. At that time a large proportion of new issuance was being taken up by central banks and sovereign wealth funds looking to diversify into the Australian dollar. However, it was far from clear whether they would also invest in longer dated maturities.
Having extended the nominal and linker yield curves to around 15 years I recall us feeling pretty pleased with our efforts – but issuance programs weren’t reducing – they were still growing. This prompted us to plan for further yield curve extensions. Last year I said that our aim was to consolidate a 20 –year benchmark and with successive launches of the 2033, 2037 and 2035 maturities I would argue we have now demonstrated an ability to do just that. The choice of 20 years as a benchmark was based in part on expectations as to what we could support into the future if net new issuance was to gradually decline; and in part because it looked as though this was a common feature of many other sovereign bond markets. What we hadn’t anticipated at the time was the size of the forecast issuance programs still to come and the prospect of a 20-year futures contract.
This brings me to several issues relevant to the topic of today. (1st) whether we should judge that the need for market development is over, or even if it has an end-point for that matter; (2nd) the role of a 20-year futures contract; (3rd) if we continue to weight nominal issuance towards long and very long maturities, what if anything does it mean for our capacity to support other segments of the yield curve; and (4th) what are our current thoughts on further developing the linker market.
You are probably most interested to hear my comments on the ASX announcement of its intention to launch a 20-year futures contract – so I will make you wait a bit longer before addressing that.
For context though I will take first the issue of how to judge whether we have usefully reached an end-point on yield curve development. In our minds there are basically two perspectives from which to consider this – one perspective is the debt portfolio manager’s (ie the AOFM); the other is the investors – although this is not to say that the role and views of our intermediaries aren’t important.
From an investor perspective there are various ways to view the attractiveness of a market. They include but are not limited to credit risk, yield, liquidity, range of choice, the ability to hedge risk, and issuer performance.
While the primary consideration for debt managers revolves around portfolio management per se, since the global financial crisis we have been giving increasing thought to market development. One reason is because as the market has grown, an expanding portfolio and larger bond line sizes requires increasing consideration as to where issuance may be absorbed and how very large bond lines can be managed as they roll into maturity; something that is faced by us and the RBA. The second reason is that as the investor base expands and individual investor holdings get larger over time, there needs to be more rather than less consideration of ‘buy-side’ preferences, something more easily avoided when the portfolio is small.
Of the criteria I just mentioned there are some things we as the issuer can influence and others that are beyond our control. Of the things we can influence, we have most control over our performance. I have already mentioned this and so there is no need for further comment. Liquidity of course means different things to different parties but by and large I take the view that it impacts the ability to trade at will in ‘large’ parcels without moving the market. There is no one measure of liquidity but it is widely accepted that it is influenced by things such as the overall size of the market, competitive price making, larger rather than smaller bond lines, regularity of supply and risk hedging availability. The range of choice (that is the number of bond lines on issue and the length of the yield curve) is also largely within the control of the issuer.
While we don’t typically have access to documented insights of investor preferences, a few years ago the IMF sought to identify those of central banks when considering the currencies they hold.
This table shows the summary results of a survey it conducted. AC means advanced countries, MIC means middle income countries and LIC means low income countries. Overall we can see that the main factors for consideration are the currency composition of a central bank’s FX liabilities; home country trade composition; and the depth and liquidity of the sovereign bond market associated with the currency which it would hold. For advanced country central banks, liquidity of the sovereign bond market is by far the dominant consideration with about half the MIC respondents also indicating this as an important issue.
With regard to the range of choice there are, as mentioned, the number of bond lines on offer and length of the yield curve.
This table shows the AGS market in comparison with the triple-A stable rated and G7 sovereigns. No obvious rule of thumb by which to compare the dimension of choice for sovereign bond markets jumps out. For a given size of market there is a trade-off between the number of individual maturities on offer and the depth of liquidity of those maturities. The larger G7 markets comprise a very large number of lines for a given length of yield curve. The JGB market for example offers 348 lines (excluding lines offered to retail investors) but as for US Treasuries these lines are on average relatively small compared to the overall size of the market at just 0.3%. The UK Gilt market on the other hand offers 41 lines, but has a longer yield curve (at just over 50 years) – each line on average being 8 times larger than those in the JGB and US Treasuries markets. The Gilt market also stands out in terms of the proportion of outstanding debt in maturities longer than 10 years (at 41%) – although we know that this is in no small way a result of regulatory arrangements that apply to UK pension funds.
In markets smaller than the AGS market, Luxembourg, Norway and Denmark standout for very large average bond line sizes – ranging between 20% and 11% of total market size; but the proportion of outstanding bonds in maturities longer than 10 years do not appear appreciably different in comparison with some of the big markets such as Germany, France, Italy and Japan.
The longest yield curves at around 50 years are in the Swiss, Canadian, and UK markets (with Japan, France and Germany not too much shorter at around 40 years).
Maybe one indicator of the long and ultra long-end depth of liquidity for these markets is the ability to support long-end futures contracts. All of the bigger markets have at least four futures contracts, typically at 2, 5 and 10 years – with a longer end contract ranging between 15-30 years.
The market that is probably closest in comparison for Australia is the Canadian market, which is about 40% larger overall, has more than twice as many bond lines on offer, has a yield curve more than twice as long, but has an average bond line size less than half that of the Australian market. The Canadian and Australian markets have the same proportion of outstanding bonds in maturities longer than 10 years at just under 20%, but the Canadians have four futures contracts at 2, 5,10 and 30 years.
So what can we make from all of this in terms of judging whether our market development task is over or ongoing? Not too much I would argue. This seems to leave us with market specific judgement as a more useful ally than an aggregated sovereign market benchmark. So let us look in more detail at the criteria I mentioned earlier with respect to our specific market circumstances.
A range of reasonable choice has been developed while not trading off the ability to maintain deep and liquid bond lines. And although not something of our creation, the strong presence of competitive price making out of Sydney and London offers real-time trading opportunity to most time zones around the world – this is becoming more widely accepted as a strong positive influence on liquidity. Also, at around 20 years there are now several distinct segments of the yield curve that offer opportunity to a wide range of investors managing to varied mandates. These segments are: the futures baskets, between the futures baskets, and beyond the futures baskets – not to mention our participation in the Treasury Notes market. What the AGS market doesn’t offer at present is risk hedging ability at the very long end – the second element of the topic for today and one which I will turn to now.
The recent announcement by the ASX of its intention to introduce a 20 year futures contract offers, from our perspective as the issuer, a positive development opportunity for the AGS market. The reasons for this are that: it will provide a useful means to hedge long bonds; it looks to be a logical step in the evolution of the Australian fixed income market generally; overtime it should facilitate long dated issuance from other issuers (for example the semis and corporates); it will highlight and assist with our efforts to further consolidate a 20-year benchmark; and it could prove very useful to us in further extending the yield curve.
Our most recent Annual Report shows that the average term to maturity of new Treasury Bond issuance for 2013-14 was just less than 9 years (and it has been over 10 years so far this year); this compares with around 6 years back in 2010-11. In terms of the average term to maturity for the portfolio, this has risen from five to six years over the same period (and is currently at 6½ years). Just over half of new issuance last year was into the longer end of the curve and we are currently on track this year to achieve a result of closer to 60% of new issuance into long and very long maturities. This has been helped by about $14 billion of issuance so far this year going into our 20-year benchmark lines (the 2033, 35 and 37s) –representing almost one-fifth of the total Treasury Bond program for this year. This means that there are a lot more ‘long’ and ‘very long’ AGS bonds than was the case just one year ago.
However, the extent to which we see liquidity develop in the very long end of the curve remains an open question. This is because the 20-year benchmark is relatively new, there is a smaller universe of investors who are seeking that duration length, we are probably at the lower bound of a sufficient amount of bonds in this segment of the curve to attract appreciable trading, and we are currently experiencing a period of market volatility during a back-up in yields (particularly at the long end). That said, based on our experience over the last few years we expect to see a gradual pickup in liquidity overtime, in turn facilitating the prospects for a liquid 20-year futures contract.
Consolidating a 20-year Treasury Bond benchmark and the introduction of a new futures contract will give us even greater impetus to maintain this part of the yield curve. We also remain conscious of the positive benefits that liquid 3 and 10 year futures contracts have played in the AGS market and that means we will aim to issue so that there is sufficient supply to support the new contract (as well as the existing contracts) and to promote liquidity for our stated aim of achieving a 20-year benchmark. To this end we will continue next year to issue with a bias towards longer maturities. Subject to market conditions we also plan to launch a new 2039 maturity with a view to building it to a benchmark size before entering a 20-year futures contract. From our perspective this should send a clear signal that we view long-end yield curve development of continuing importance and that we see the establishment of a 20-year futures contract as a positive market development.
As of today we already have $21.6 billion outstanding across three bond lines that are candidates for the first 20-year futures basket, while the total amount outstanding in maturities that are longer than the 10-year futures contract basket bonds is $45.7 billion. Add to that the amount outstanding in the 10‑year futures contract basket bonds, and this brings the total face value of bonds in the combined ‘long’ and ‘very long’ segment of the yield curve to $103 billion – almost one-third of total outstanding Treasury Bonds.
A recurring question of us is whether we will continue to extend the yield curve, this most often being interpreted as establishing a 30 year maturity. It is clear from this table that all of the sovereign bond markets larger than Australia have yield curves that extend to at least 30 years. What we can’t interpret from the table is when they were established and for what particular reasons (i.e. was it primarily driven by portfolio management considerations, or more demand driven because of investor appetite for very long duration). We do know from the Canadian experience that the yield curve was extended to 30 years in 1991. A liquid curve up to 10-years had already been established, but there was investor demand for longer duration. A 30‑year yield curve was also seen as facilitating the development of a well-functioning domestic capital market.
During this time, one of the major objectives of the Canadian Government’s debt strategy was to lengthen the average term to maturity of outstanding debt to reduce the sensitivity of debt serving costs to changes in interest rates. The introduction of the 30-year sector, both for nominal bonds and inflation linked bonds was in line with this objective.
I am happy to share with you something of how we think about this issue because it should help to provide a greater appreciation of the main considerations than would a simple ‘Yes’ or ‘No’ answer to the question of whether we are about to keep extending. Firstly, as I have already mentioned we have been deliberately lengthening the duration of our nominal liabilities and this has been happening since the first half of 2011. From a debt portfolio management perspective, this has been facilitated by having a wider choice of long bonds from which to select for weekly issuance. So to the extent that having a longer curve will also mean having more long bonds to choose from each week, it can be seen that a longer curve has been an ‘enabler’ for our duration-lengthening activities.
Second, as I have said the stock of gross debt outstanding has been steadily growing. All other things being equal, as I also mentioned earlier, we are conscious of the need to anticipate the size of individual lines as they roll into maturity.
Third is the maturity ‘gaps’ between long bond lines. We had been working on the assumption that a gap of four years between lines beyond about 12 years was appropriate. However, as part of our 20-year benchmark consolidation we recently issued a 2035 line, halving the gap between the two existing longest lines.
Performance of the very long end of the curve soon after after establishing the 2035 has given us some confidence that we can maintain gaps of two years between lines out to around 22-23 years. In isolation, this reduces the need to extend the curve to 30 years for a given stock of debt because we can increase the average term to maturity of the portfolio through building these lines.
However, given the ASX’s recent announcement to establish a 20 year futures contract, a longer hedging tool may make further yield curve extension seem a less risky proposition. Therefore, to answer the question on curve length, I repeat the message I have been giving, which is that we will continue to consolidate a 20-year benchmark – but we are open to the prospect of a further curve extension, including to 30 years. At the very least you can expect us to issue a 2041 maturity in two years and a 2043 maturity two years after that.
If we were to extend to 30 years that could possibly involve two further maturities beyond a 20-year futures contract basket as a new long term pattern of issuance was established to support a 30-year curve. Before proceeding with a further extension we will require confidence that appreciable amounts can be readily absorbed and that reasonable opportunity would exist to re-issue into those lines. While I am not prepared to put any parameters around that now, what I will say is that we are facing circumstances under which the risks associated with extending to 30 years would appear to have been reduced.
In this forum two years ago I explained, at least in broad terms, how we think about the competing portfolio management objectives of cost and risk minimisation. On the one hand we can seek to minimise the risk of volatile debt servicing costs overtime by issuing longer. On the other hand that approach does not come without additional cost given an upward sloping yield curve. Our primary challenge is to meet the Government’s financing task in full each year (and to manage the cash portfolio so as to meet its daily outlay commitments) while balancing these competing considerations. One way that we can insulate the Budget from the near-term impact of rising yields is to issue longer and to increase the average term to maturity of the debt portfolio. Rising future yields will only impact the marginal cost to Government of future issuance – not the cost of the stock that has already been issued. This remains core to our strategy of adopting a longer issuance bias.
This chart shows outstanding issuance for Treasury Bonds, with the grey bars showing the results of our issuance activity so far this financial year. It is apparent from the chart that the bulk of our issuance has been into 10-year futures stocks and longer (in fact around 62%); most of the issuance has gone into 8 lines; the 4-5 year part of the curve has been well supported despite the long-end bias; and that every line except for sub 3-year futures basket maturities has been tapped at least once. Once again for the year ahead and in recognition of the importance of the mid-part of the curve and support for both the two existing futures contracts, we will be establishing new 2021 (5‑year) and 2028 (12-year) maturities. Together with other ongoing issuance into this part of the curve this should demonstrate an awareness by us that a core part of our investor base is active in this segment of the market.
If nothing else this chart should highlight the challenge of supporting all parts of the yield curve when our issuance programs begin to decline. However, it also highlights the support we have had from investors in pursuing a longer issuance bias. While we know there are central banks that hold maturities of 10-years and longer, it is the growing interest from real money largely in the form of insurance and pension funds, together with some large fund managers (both domestic and offshore) that has made it possible for the AOFM to sustain a longer issuance bias.
In the coming year you can expect that we will follow a similar pattern of issuance.
Turning now to the indexed bond market, I am sure most of you will be aware of our issuance plans for the coming fiscal year, which are much the same as the current year in volume terms at around $4billion. A subtle change you may also have noticed is the absence in our communications of building the indexed market to 10-15 per cent of the stock of bonds outstanding. This is deliberate. The target range, conceived at a time when the debt and fiscal trajectory was very different to that which was subsequently realised (and is now expected), serves no useful purpose for the AOFM and has now slipped from reference. However, let me be very clear that this change in no way represents a downgrading of the AOFM’s nor the Government’s commitment to the indexed bond market and as you can see form the chart, there has been a consistent increase in the size of the market since 2009.
The fact is that the share of indexed bonds in the portfolio has rarely exceeded 10% since the target was first announced (averaging around 9 per cent). It was always to be the case that as the funding task expanded, the vast bulk of additional issuance would be met using Treasury Bonds (given the wider investor base, availability of hedging instruments and extra liquidity). I would argue, however, that this has had no bearing at all on the AOFM’s capacity to develop and support the indexed bond market.
From a modest starting point in 2009 when we recommenced indexed issuance (with $6 billion on issue spread across 3 lines), we now have around $27 billion in stock outstanding ($33 billion when adjusted for inflation indexation). This is spread across 7 lines with a curve extending 20 years.
While I acknowledge that liquidity in this market does not match the nominal bond market, the consistent feedback we get, both domestically and abroad, is that it is good and indeed comparable to many linker markets. We will support the linker market of course by continuing to issue regularly through our bi-monthly auctions, taking feedback from market makers and the broader investor base and by introducing new lines as and when appropriate.
As I know new lines are always of interest to this audience, a curve extension is (subject to market conditions) likely to be undertaken sometime in the coming year. This will be done through the introduction of either a 25 or 30 year line. Looking a bit further ahead, a new 10 year benchmark line that sits between the 2025 and 2030 is an obvious addition to the curve, although the earliest I think we would give consideration to this is the 2016-17 fiscal year.
To reiterate, our commitment to support and indeed to continue developing the indexed bond market remains as strong as ever and the coming year will be very much business as usual from the AOFM’s perspective.
Let me conclude by summing up the observations I have highlighted. Continuing issuance programs at elevated levels have both required and offered the opportunity for ongoing market development. While we have pursued these opportunities largely on the grounds of portfolio management, we are conscious of expectations that we will take account of the broader benefits that can arise from this. We are also mindful that as the market grows and the investor base broadens and deepens we should be thinking more and more about investor preferences – however, this is not to say that we can be all things to all investors.
It is fair to say that with establishment of a 20-year benchmark and the introduction of a futures contract to match, our current portfolio strategy of lengthening average term to maturity will be easier to achieve. However, as we don’t see the market development phase either for nominal or linker curves as necessarily having been completed, we will be as well placed as we have ever been to further extend each of those yield curves subject of course to demand, market conditions and an assessment of being able to support such extensions into the future.
In the meantime we will also continue to issue across the curve in order to support the 3 and 10-year futures contracts with the regular introduction of new maturities at the 5 and 12-year points. All of these considerations will take into account the important established practice of seeking market soundings through our intermediaries prior to announcing auction details.
There are obviously other development considerations that I have not raised, such as using the Treasury Notes market more, or introducing new products – for example a FRN (particularly given the prospect for a different approach based on a new benchmark). But we are confident that these options have been carefully considered by us and that neither the opportunity nor timing seems conducive to us pursuing these right now.
Finally, a number of you may be interested in learning more about how we will conduct divestment of the remaining Government RMBS holdings. Unfortunately I am not in a position to say any more about this than what we have released in our initial notice to the market. I can confirm that we are planning to hold one auction each month for a total volume of notes up to but not exceeding $500 million. We will be putting out an Operational Notice to provide more detail in early June. That said, I note recent primary market activity, including a deal that was launched on the day of the Budget, when the divestment was announced. That this deal went on to be the largest non-ADI transaction since the GFC and was followed quickly by the launch of Westpac’s transaction yesterday, speaks to the underlying strength of the market.
Last updated: 26 May 2015