Many years ago, U.S. regulators, whilst revising the old 17F5 regulations, allegedly sought to compel custodians to assume liability for failure of either a sub custodian or a central securities depository (CSD). It was also alleged that, in that light, the major U.S. custodians advised the regulators of their unwillingness to accept such liability and asked for a delay to allow an orderly exit from the market. The idea was dropped to avoid the resultant huge disruption and the systemic implications of a transfer of such risk to lesser capitalised entities.
Whereas AIFMD has elements of “a get out of jail” clause, UCITS V is clear that custodians, or rather the depositories, will assume total liability for asset safety. Yet there are only whimpers of dismay from custodians.
2014 ECSDA (European Central Securities Depository Association) figures show 470 million transactions were effected across their constituent markets. The value of these transactions equalled EUR 1,200,000 billion (1.2 quadrillion value) and the intraday settlement risk arising, irrespective of CCP netting, must be in the trillions of euros.
Let us remember that the tier one capital of the EU’s largest banking Group, HSBC, is $153 billion or just 0.01% of that EUR 1.2 quadrillion figure! Some of that exposure may be collateralised, often against assumptions that broker liens and CCP netting will survive attack from other creditors in a court of law. And have no illusions. If major losses occur, there will be lengthy and troublesome lawsuits!
The ratio of assets in custody to tier one capital at BNY Mellon or State Street is around 0.05-0.06%. In other words the loss of $ 14.0-17 billion of assets would eradicate their capital. In reality they would be at risk if ever there were a loss of even a billion as the risk perception off the business would be ratcheted up exponentially in such an event. And it would be interesting to hear the justification of the larger universal bank custodians, with tier one capital of around 1% of custody assets, if regulators advised they should allocate 5-6% of this to their securities services operations. There is logic in such an approach, for regulators must believe there is a risk of loss of assets if they are so keen to have a guarantor for such assets.
The trouble with the risk profile of UCITS V is we can see it leading to contagion. On the one hand, an eventual convergence of the liability regimes of AIFMD and UCITS appears inevitable. And, on the other hand, the globalisation of the industry weakens suppliers’ hands in two ways. First, there will be interesting discussions between regulators and regulated in the event of a loss, if some investors are compensated for loss of assets but not others, Second, the fact that there is a guarantor makes it easier for the impacted authorities to allow any losses at a CSD or settlement agency to filter down to the private sector, as the firms involved are credible guarantors.
So the question is whether a loss of this magnitude could arise. Most custodians say it is unthinkable. But, for those with long banking memories, the Latin American debt crisis, the challenges to title following Lehman, the arbitrary rebasing of bank assets and liabilities of the Argentinian or Hungarian governments, or the politically motivated restatement of shareholder registers in Russia, were similarly unthinkable events; at least until they occurred.
CSDs, and indeed all asset holders in the custodian value chain, could incur losses through fraud, whether internal, external through hacking or misuse of user gateways. Assets, however frequently reconciled, can end up in the wrong place. And it is never certain that they will be returned in a correctly segregated form. Collateral based transactions are not automatically undertaken in many markets on a DVP basis with simultaneous finality. Stock loans may not always be returned and haircuts in competitive markets are computed on the basis of normal volatility rather than event risk based volatility.
In the more innocent days of 17F5 changes, the custodians were right. The regulators wanted to cross the Rubicon. The risk being requested was too great. It was not a pricing issue but one of business logic. When the UK was adopting true DVP, I represented the U.K. banking community in discussions with the Bank of England. They wanted banks to take on unlimited liability for loss in some circumstances on the basis that it would never happen. The response was that, in that case, it was not necessary.
Regulators appear to be thinking in this way again, but, for reasons that are beyond me, custodians have accepted their argument. It is not so much that I think the risk of loss is likely to occur; it is the fact that it could occur and is “wipe out” risk that makes me certain that the industry is wrong in agreeing to accept such liability. And many leading players are all the more wrong to boast that, despite these risks, they will cover all markets.
Banks should revisit the issue with the regulators. AIFMD was a challenge. UCITS V is a risk too far.