Article 47(3) of EMIR is becoming something of a rallying cry in CSD and custodian circles. It is definitely the major driver behind BNY Mellon’s ground-breaking launch of a new CSD and the almost-as-ambitious venture by the LSE Group, allied to JP Morgan, to create a somewhat lighter Luxembourg venture. And the planners at many other major custodians are all struggling to find an answer to the iniquitous consequences of article 47(3). Whilst the ICSD and CSD world is far from panicking at the potential incursion of additional and predatory competitors, their product developers must be working hard on alternatives to a proliferation of new CSDs.
One can but suspect that the EU authorities are questioning what they have unleashed. Alas, they have only themselves to blame. The cozy world of T2S encouraged competition, albeit seeing it primarily within the serried ranks of the existing CSD population. The CSD regulations further fostered the concept of new competition and, in effect, enabled the creation of limited-purpose CSDs. And it placed the CSD at the pinnacle of the low-risk pyramid, portraying it as a risk-free paragon in a risk-drenched universe of custodian banks and other commercial sector providers.
But it was really article 47(3) that was the catalyst for action. It requires collateral at a CCP to be deposited with the operator of a securities settlement system where such a system is available. And it clearly states that depositing such collateral via a custodian bank with a securities settlement system does not meet that requirement. This ruling is compatible with the regulators belief in the riskiness of those private sector custodians and the invincibility of the CSD population.
But this case is seriously flawed for some of the traditional CSDs are far from risk free. Some of the emerging CSDs may straddle structural fault lines. The intermediation of a custodian bank in a properly structured tripartite collateral agreement can eliminate commercial sector risk. And the regulation, by eliminating the global custodian from the transaction flow, reduces a custodian’s control over their client assets whilst other regulation, such as AIFMD, increases their liability if anything goes wrong.
The traditional CSD limits its liability for losses. The limitation varies by CSD, but there are CSDs which do not take responsibility for two of the most obvious frauds that could occur, namely employee fraud and computer hacking. The custodian bank is by definition a risk taking institution. Nobody would argue that the full-service banks take on risk. Some of the more specialized banks may take on less risk. But none of them are risk free.
Let us examine the real world though. The reality is no major custodian bank has defaulted and, as a result, caused investors a loss of assets. To be fair, nor has a CSD incurred a loss that has resulted in a loss of assets for investors. But losses have occurred. The commingling of client and proprietary assets at MF Global, the rehypothecation structures at Lehman or the Ponzi scheme at Madoff are embedded in the psyche of regulators and attributed by them to the hedonistic world of private sector financial institutions.
The issue of most concern is that the regulators may have created more risk in their addiction to the CSD structure. The BNY Mellon CSD plans are the most developed and appear quite robust. It is too early days to judge the LSE-J.P. Morgan initiative. But the limited-purpose CSDs being suggested by many others are like the SPV’s of the asset backed securities age. They have the technical structure that allows them to proclaim themselves to be low risk CSDs, but they do not have substance. In other words, they risk being a sham vehicle structured to meet a regulatory aberration. And that sort of structure could be highly dangerous.
Yet one must sympathize with the global custodian banks looking at these structures. The world of regulation had created the worst of all worlds of more risk and less control. It is absolutely sound logic for a custodian to want to maintain line of sight and full authority over fund assets. And disintermediating the custodian from part of the CCP transaction lifecycle prevents just that.
And one must not forget the second driver for custodians to consider CSD launches. They could build CSD to CSD links and disintermediate the sub-custodian in the AIFMD risk chain. But it is unlikely that the regulators will insist on speedy creation of such links and time to market for a solid global CSD to CSD coverage is likely to be even slower and more legally fraught than it has been to date in the traditional cross-CSD space.
The regulators need to see common sense and adapt Article 47(3) to ensure that the CCP is protected against default of a custodian and that the intermediation of the custodian cannot delay rightful realization of collateral by a CCP. Such a structure does not appear that challenging, especially as escrow and other arrangements are quite robust in most jurisdictions.
If regulators fail to do this, we will see serious fallout. And that will create risk rather than eliminate it. And, by the way, it would be useful if, at the same time, one piece of nasty drafting ambiguity is resolved. The EU keeps on talking about Securities Settlement Systems. We all believe those words are synonymous with CSDs. It would help if this could be confirmed, for at the moment an admittedly weak case could be made to say that the notary function of a CSD is excluded from the privileges accorded to Securities Settlement Systems. Over and above destroying the case for Article 47(3), that could also risk destroying much of the investment market as well.