ASX – Actually, a Failure of Blockchain?

Despite the enthusiastic response of the media[1], and the exaggerated claims of the management of the Australian Stock Exchange (ASX) in their announcement of their plans to use Distributed Ledger Technology (DLT), also known as Blockchain, to replace its venerable Chess clearing and settlement system, the proposed solution is probably as far from ‘pure’ blockchain that one can get, while still claiming to be blockchain.

The proposed solution, which is driven by the realities of financial markets, has so relaxed the constraints of pure Blockchain, that it clearly demonstrates the limits of this technology rather its success. This ‘Blockchain-Lite’ not only fails to achieve most of the claimed benefits of DLT, it actually adds costs and risks for market participants.

First, in the proposed solution, ASX is very firmly the single ‘central authority’, controlling all aspects of the design, implementation and operations of whatever the solution will be. It is also a permissioned DLT on an ASX-owned network, with little information on whether there will be any of the much-vaunted transparency claimed with blockchain. In addition, though not disclosed other than in early whitepapers by software supplier Digital Asset (DA), it is probable that to achieve the necessary throughput/latency, consensus is based on a Byzantine Fault Tolerance (BFT) model rather than a pure Proof of Work (POW) as in Bitcoin.

But such climb-downs are pretty much par for the course for practical, as opposed to theoretical, proposals to implement a blockchain solution. In this case, not much is left from the purest of blockchain but a sub-standard shadow of the exaggerated claims for the technology.

But what about ‘trust’ and the holy grail – ‘single source of truth’? Gone too, I am afraid.

Thirteen words in the ASX media release torpedoes any claims that the solution will be a real blockchain 

“Customers will be able to connect in a similar way they do today”

Anyone who understands blockchain very quickly comes to realise that to implement such a solution across an industry a firm will have to recognise the reality that the solution cannot just be switched on. It must be implemented gradually with some users ‘on’ the blockchain and some not. In fact, the reality is that some users will never ‘get with the program’, because they are too small (or alternatively, too smart?).

This is the ‘chicken and egg’ problem with all DLTs - unless and until all data is on the ledger, the promised benefits, of trust, transparency, immutability etc., cannot flow.

The ASX has been dragged kicking and screaming into recognising this reality, but appears not to have considered the implications very deeply (or if they have, they haven’t let the rest of us into their thinking).

What exactly does it mean if some users of ASX’s post-trade services are on the DLT and some not?

First, out goes the ‘single source of truth’ for a firm that chooses the DLT route, since its trades will be in two places, on the chain and off the chain. To create a ‘position’, then, a firm will have to merge two sets of data - from the chain and from elsewhere. Such duplication destroys the claim that DLT will bring untold savings to the industry in back-office savings, in particular in reconciliations [2]. In fact, such an approach will add costs to the industry, since firms will have to handle both existing and new transaction flows.

But surely there are some benefits?

One of the claimed benefits of DLT is that clearing settlement will move to real time at some time in the future. But a moment’s thought will make one realise that too is untenable.

For example, assume that I believe the share price for BHP is going up and buy 100 shares from Broker A. Being prescient (or lucky) the price goes up and I then sell the 100 shares to Broker B. This is a set of transactions that happens hundreds of times every day, albeit that clearing and settlement will be delayed by up to two days[3].

But consider the very real situation where Broker A is not on the DLT, but Broker B is. I buy from A and sell to B, but that second transaction will not clear and settle in real time because I don’t have the stock yet from Broker A, nor will any transactions that Broker B has subsequently done with other users on the assumption that the 100 shares will be available. 

This is the classic ‘knock-on settlement failure’ that central clearing and settlement depositories (CSDs) are designed to circumvent to prevent markets from slowly grinding to a halt.

Of course, the very real potential for settlement failures opens the can of worms of ‘blockchain immutability’, as transactions will have to be backed out of the DLT someway. Another strike against this Blockchain-Lite?

So, in summary, by having two classes of users – on and off blockchain - the ASX solution provides no financial benefits to firms (in fact increases their costs), and no technical benefits, yet introduces significant operational risks, especially during implementation.

Why then would any firm buy into it?

Recognising that only some (probably a minority of) firms will choose to interact with the new system, raises a inescapable question, why choose a technical architecture based on blockchain at all?

In fact, what remains in the ASX proposal resembles little more than what is known in IT architecture as a ‘message bus’ or a secure communications mechanism between two computers.  There are several message buses available commercially, such as the IBM MQ (WebSphere) product which provides secure, high performance, message brokered (e.g. conversion) communication services. Since the ASX solution will be ‘permissioned’, i.e. restricted, a product, such as IBM/MQ, running over a secure VPN (Virtual Private Network) would appear to provide all of the functionality required by the ASX, and what’s more – it works already. Why reinvent the wheel?

In its media release, ASX management boasted that

“Having completed this work, we believe that using DLT to replace CHESS will enable our customers to develop new services and reduce their costs, and it will put Australia at the forefront of innovation in financial markets”

The claim to reduce costs is very debatable but many other questions surround the actual completion of the work. 

This proposed project has all of the hallmarks of the, unfortunately very crowded, IT project management Hall of Famous Disasters. Even after many months of work, the project has[4]: no clear objectives; no plan; hence, no detailed costs; no well-defined benefits; is complex; involves many external stakeholders; is using novel technology; and has engaged a small software house that has not successfully implemented their software in a live situation yet and has little experience in developing the business, as opposed to technical, solution. 

The media release stated that proposals to resolve some of these issues may be released for consultation by March 2018, so there is a lot of water yet to flow under the DLT bridge before this project actually gets underway. 

Does ASX know something that everyone else has missed?

In a recent speech, Michael Bodson, President and CEO of DTCC and a (probably the) leading proponent of DLT in post-trade processing , when talking about shortening the clearing and settlement cycle, said [the added emphasis is this author's]

"We have a similar story at DTCC - the recent change in the US settlement cycle to T+2, which occurred this past September. We used existing technology to shorten the cycle and align our processes with the European markets. The driver of this initiative was client value - capital efficiency, risk reduction and a globally harmonized settlement cycle. The fact that we used the same technology that we’ve employed for many years to achieve this was a benefit to our clients, who didn’t need to rip and replace their existing systems.

The lesson - don’t conflate improved process with new technology.

Even when you consider the hottest, newest technologies in financial services today, such as distributed ledgers, the initial predictions of dramatic transformation have since been right-sized because the business cases have not been compelling enough to move forward.

That is likely to change in the future, but I tend to believe that, at least for the short-term, the value proposition of DLT lies in addressing industry pain points in targeted areas that remain highly manual and with relatively low volumes".

Remembering that ASX are proposing an industry-wide, high volume implementation, do they know more than someone actually living the problems of DLT implementation? If so, are they really smart and prescient, or somewhat over-confident?

So, what to do?

 It is easy to critique (or even unthinkingly endorse) a technology proposal but more difficult to suggest alternatives - here goes!

A few hundred metres away from the ASX head-quarters in Sydney is the head office of the Australian Payments Clearing Association (APCA) which has led the design, build, and testing of the New Payments Platform NPP, which will deliver real-time retail payments into the Australian market., On 26th January 2018 (Australia Day), NPP is due to go live, and after cleaning up any remaining teething problems, will begin to process millions of real-time payment transactions per day. 

The new NPP system is based on high-volume, highly secure SWIFT technology and is true piece of financial markets infrastructure (FMI) built for security, stability and performance – and not a blockchain in sight.

From January, major (and some minor) financial institutions will have an interface into the NPP which begs the obvious question, why have a separate and different technology to send SWIFT compliant messages to a clearing and settlement agency? 

In other words, is the proposed ASX solution even needed?

Having done a sterling job in designing and building NPP, there is an experienced and successful team available today that should be able to switch pretty seamlessly to building a high-performance, real-time clearing and settlements interface, based on similar SWIFT messages, as already envisaged by the ASX proposal[5].

The Reserve Bank of Australia (RBA) is both the instigator and regulator of the NPP and also one of the regulators of FMI in Australia, so is in an excellent position to understand the problems faced in re-engineering Chess, for which it also has overall regulatory responsibility. 

There are therefore a number of possible options when considering the future of Chess, including: (1) do-nothing, leave it with the ASX and hope that Blockchain-Lite will work; (2) engage NPP/SWIFT to build the customer interface into Chess, while ASX re-engineers the clearing and settlement application (the really hard bit); or (3) move Chess operations to a division of NPP or a similar new organization. This, of course, is a risk-based decision that the RBA and other financial regulators should be deeply involved in.

The benefits of integrating standard interfaces[6] to market participants are genuine, including reducing costs by having common communications and operating technologies and reducing training and development costs. But more importantly, by integrating these interfaces, it becomes much easier to deliver real-time DvP (Delivery versus Payment) especially for large value transactions, if/when that becomes necessary. Furthermore, it creates a pathway for further integration of clearing and settlement of Australian government securities[7], through a common technology environment.

Today, the ASX has a monopoly of equities clearing in Australia. In September this year, the Australian Treasurer released a policy statement by the Council of Financial Regulators (CFR) on “Open competition in clearing and settlement of shares” - a direct threat to the ASX monopoly. The CFR policy clearly stated that

“Accordingly, ASX would not be required to make up-front operational changes to accommodate competition until such time as a competing CCP committed to entry. However, at the same time, the technological design of ASX’s CS infrastructure should not raise barriers to the potential future implementation of interoperability or access to settlement arrangements by a competing CCP [emphasis added]”.

Obviously the very central role that ASX plays within its DLT proposal would inevitably be a barrier to competitors. However, by employing an open technology, for example based on the NPP architecture, competition would be enabled rather than restricted.

In justifying the choice of technology, the ASX CEO said that "the costs in rolling out this system I don't think are too dissimilar to us having rolled out a traditional system”. But that begs the question, what is the alternative? And if the alternative is using tested NPP software as opposed to building from scratch, maybe it is not marginal after all? 

But, if differences in costs are truly marginal then any decision should only be made by comparing the risks, here of a new unproven technology against proven infrastructure.

One can only hope that after the planned consultations with users, there will be at least some hard dollar figures around costs, benefits and risks and a detailed rationale for choosing a particular solution.

[1] Many media stories repeated the ‘fake news’ that the ASX was the first to go down this path. It was in fact the Depository Trust & Clearing Corporation (DTCC) which first began such a project a year ago. This is, of course, well known to the ASX, as its software partner Digital Asset (DA) was beaten out of this project by an unknown start up. The ASX has failed so far to correct the record on this.

[2] In practice, data spread across multiple databases is more difficult and expensive to reconcile than in a central location.

[3] Mainly due to ASX’s inability to upgrade its Chess system over many years although to its credit it has recently moved fromT+3 to T+2 - a major step.

[4] All of this may in fact be available, but ASX has not shared with the industry.

[5] It should be noted that an interface has already been built between Chess and NPP, based on standard SWIFT messages

[6] Which could, for example, be called the New Payments and Securities Platform (NPSP)

[7] Austraclear, another division of ASX, already employs SWIFT message format for clearing and settling Australian government securities, using the SWIFT FIN network, to communicate, thus making possible a piggy back on a common platform, if/when needed.