This is a free preview of a paid episode. To hear more, visit jollycontrarian.substack.com

By now I hope the netting problem is clear: regulations require each dealer to obtain lengthy legal opinions for every master agreement in its portfolio. A full service broker dealer may have hundreds of thousands, across scores of jurisdictions. Swap, stock lending, repo, and prime services master agreements each have different closeout mechanisms. Each requires its own legal opinion.

Across those jurisdictions there are, literally, thousands of legal entity types, the netting requirements for each of which will, subtly, differ.

Those entity types may have similarities across borders — there are corporations, partnerships and trust-like things in many jurisdictions, and they fulfil similar functions — but they are by no means standardised. There are private corporations, public corporations, joint stock corporations and special purpose vehicles. There are protected cell companies and segregated cell companies. There are unlimited partnerships, limited partnerships and limited liability partnerships. Trusts, foundations, stichtings and fiduciaries. Some have legal personality, some don’t. And the rules and resolution protocols for even simple corporate entities will differ between jurisdictions. An Austrian GmbH is not subject to the same rules as a German one.

Thanks for reading! This post is public so feel free to share it.

Close-out mechanisms in standard agreements like the ISDA are designed, as far as possible, to work universally, but this is an imperfect science. In some jurisdictions automatic early termination ensures netting works, in others it invalidates the agreement altogether. Additional clauses covering local peculiarities may be needed.

There is a bit of regulatory conflict here. This is blue-on-blue action. By and large these local regulations, and their peculiarities, were originally designed to promote fairness between all a bankrupt’s creditors, rather than to vouchsafing a given dealer’s capital framework. You would think that all jurisdictions would want to ensure maximum capital efficiency to give local businesses optimal access for international capital markets. But that is not how the regulations are necessarily framed. That said, should there ever be latent conflict between insolvency and capital rules, local regulators are strongly incentivised to fix them pronto. The consequence of not doing so is to shut themselves out of international capital markets. That is a bit like sanctioning yourself. So, we saw in 2016, when a German court decision threatened to undermine the effectiveness of close-out netting, the German government quickly intervened to change the law.

Hence the need for constantly-updated netting opinions. However long and however erudite, a netting opinion must, for each of the hundreds of thousands master agreements in the firm’s contract portfolio must be boiled down to a single question:

Will close out netting work, yes or no?

That determination will set a “netting flag” in the firm’s risk systems for each agreement. That, in turn, will determine how much capital the bank must hold against that exposure under that agreement.

Long, windy academic tracts about aleatory contracts and the anti-deprivation rule established in Whitmore v Mason (1861) 2 J&H 204 are not well suited to answering that simple question.

There is thus a collision between the airy, theoretical world of regulation and a dealer’s greasy operational reality. It has to answer “yes” or “no” — realistically, “yes”—the consequence of “no” is almost always “we cannot trade with this entity” for every contract it enters into.

Yes, opinions are required to be “written and reasoned”. But they should also be practical.

Mechanically, setting netting flags involves three interacting “systems” inside a dealer’s greasy operational reality: its risk engine, its agreement database, and the diffuse activity of gathering, reviewing and approving netting opinions. These three systems are quite different.

The risk system has to work.

The agreement database has to feed the risk system.

Netting opinion review is a manual, offline human activity that no-one wants to do because it is so boring.

Risk engine

Firstly, the risk engine: this will almost certainly be a industrial-grade computer system. It must be, and will be, built like a tank. It ingests trade data and market valuations for every transaction and every collateral position, aggregates them and matches each to a single master agreement record. From this it calculates the following day’s margin calls, manages risk limits and credit headroom for the client and the capital consumption for the agreement in question.

This exercise is fiddly enough: for those calculations to be possible, each swap needs to map to a single ISDA, each stock loan to a GMSLA, each repo to a GMRA and so on. Should the same counterparty have two ISDAs or side-by-side collateral arrangements—unusual, but not impossible—the risk engine’s logic needs to map swap trades and their exposures to different ISDAs and CSAs. Failing to get this right can be an existential risk, so this system should be bulletproof.

Master agreement database

Second, the Master agreement database: this is where all master agreements go once negotiations have concluded and the agreement is executed. Key terms of master agreements and collateral arrangements will be recorded in the master agreement database. This will have a solid functional core—it provides necessary master agreement data to the risk engine—but the system it is built on is quite likely to be old, heavily patched and a lot of its data, especially relating to older agreements, will be quite ropey.

At the dawn of the age of swaps, there was little formality around the negotiation of master agreements. That has changed over forty years in response to the manifest crises and shocks of history, but the content of these systems will still be uneven. What once was a bespoke negotiation handled by learned counsel and magic circle law firms has long since been industrialised: modern contract negotiation is outsourced, offshored, juniorised and heavily constrained by policy and approval workflows which provide the negotiation data record for each master agreement.

However, contract documentation is far from standardised. There is no good reason for that these days: it is a commoditised product, and the terms and exigencies of a master agreement are well understood and boil down to a handful of simple credit points, but master agreement negotiation remains a cottage industry that keeps a lot of busyworkers gainfully employed.

If I had to guess I would say AI will not change this.

In any case, the master agreement database inevitably contains less data than is embedded in the master agreements themselves, after all that earnest pettifoggery, and on occasion misrepresents it.

In any case, part of an agreement’s data record—specifically the “netting flag”—must somehow get from the agreement database to the risk engine. But first it needs to get to the agreement database, and this is where the netting opinion review process comes in.

Netting opinion review process

The last of those three systems is the netting opinion review process. To populate a netting flag in the agreement database, one must know what the relevant netting opinion says. This is handled by what we will call the “netting opinion system”. This implies something grander the loosely-governed, resented, syndicated human activity it is likely to be.

The netting opinion system will be a disparate collection of typed forms and spreadsheets stored on a folder in a Y: drive somewhere, or an underspecified SharePoint site or access database built in 1994 and now maintained, reluctantly, by a team in Gdansk. It may include a shared inbox, a subscription to online reporting services or an arrangement with a local law firm to provide summaries.

This process is a monster. Some poor sod in the legal department will have to syndicate the job out amongst the bank’s derivatives lawyers, who between them must review, construe and approve the netting opinions. That poor sod, in one institution, once upon a time, was me, readers.

You can call that role something high-faluting, like “netting opinion compliance lead” but it is a thankless task.

You must assign each jurisdiction that your firm trades against to a fellow lawyer: usually, each one gets three of four each. They will not thank you for it. They then must keep their allotted jurisdictions, and all the opinions for them, up-to-date.

They must check with credit what risks the firm has on, against what kinds of counterparties, under what agreements in that jurisdiction. They must compare that against the industry opinions – there will likely be several, which will not necessarily be consistent — to ensure everything is covered. Where there are gaps—usually, there will be—they must commission bilateral opinions to fill them.

They must populate “netting review sheets” explaining clearly which counterparties are covered, what products are in scope and what departures are required from the legal standard to ensure close-out will work as intended. Once these are approved—there is of course a governance process—these must be then filed somewhere the negotiation team can find them: they will need them to set netting flags when onboarding and negotiating with new customers. This is all laborious, time consuming, hard to effectively govern or supervise, and expensive.

Regular subscribers leave us now — but next time we will pick up by looking at a solution. Premium subscribers come along as we look at:

Industry opinions

How industry opinions work, and form the core of the netting effort

Burning down the house

How, sometimes, starting again is not the worst idea

There must be a better way

What if, what if, what if… What if someone built this?

This Substack is reader-supported. To receive new posts and support my work, consider becoming a free or paid subscriber.