Collateral That Thinks: Why Derivatives Need Agentic Execution
- Jan C Berger
- 11 minutes ago
- 5 min read
Modern markets are powered by data and automation - so why are their legal agreements still stuck in 1998?
The collateral engine that powers the global derivatives market is still running on infrastructure built in the era of pagers and fax machines. While trading platforms have digitalised, the actual enforcement of collateral terms – who owes what, when, and what happens if they don’t deliver - remains largely manual, opaque, and operationally fragile. Margin calls are managed in spreadsheets and inboxes. Some post-trade instructions still flow by fax - because the legacy stack runs deeper than most admit.
A few years ago, I explored this in a Hunit use case on how distributed ledger technology (DLT) might transform collateral management. Much of that logic still holds. But the world has moved. Smart contract thinking has matured into something more practical and regulatory aligned: Agentic Contracts. What DLT hinted at, Agentic Contracts operationalise.
Back then, I wasn’t sure whether to post it. It felt a bit niche, a bit speculative. But I kept circling back. Partly because the problem is real. And partly because I kept coming across examples that underlined how fragile the current system is.
First, a step back - what are we talking about?
A derivative is a contract between two parties, deriving its value from an underlying asset. The most common types are futures, options, forwards and swaps. Collateral management is the process where two parties exchange assets to reduce credit risk in these unsecured transactions. This includes banks, broker-dealers, hedge funds, pension funds, insurers and corporates. Assets like cash, bonds, equities, even real estate, can be pledged to reduce exposure.
If you’re still unclear on how derivatives actually work, there’s a short and humorous piece from 2011 by Adam Hewison, a former floor trader, that does a great job of breaking it down: A Simple, Humorous Look at Bank Derivatives.
Since the subprime crisis, this has become a major regulatory focus. Rules now require variation margin (VM) to be calculated and exchanged daily. Initial margin (IM) obligations have been phased in more gradually through a series of regulatory thresholds starting in 2016. As of 1 March 2017, most non-cleared OTC derivatives must calculate and exchange VM every day. And yet - this market still runs on manual workflows: emails, faxes, calls.
The Problem Isn’t Digitisation. It’s Execution.
Yes, derivatives trading is digital. But the obligations and collateral flows tied to them aren’t. They’re still governed by static documents - CSAs in PDFs, legal terms in footnotes, margin instructions via inbox.
This system fails in three critical ways:
Latency: Daily MTM introduces risk that compounds intraday.
Opacity: Rules around eligibility and reuse live in documents but aren’t enforced operationally.
Rehypothecation risk: Reuse rights are contractual but not machine-readable or auditable in real time.
Why Agentic Contracts Are the Missing Layer
Agentic Contracts go beyond digital PDFs. They combine natural language with embedded logic and data triggers. These agreements don’t just document what should happen. They do it:
Monitor collateral balances and valuations in real time
Enforce margin thresholds and reuse restrictions
Trigger transfers and reject unauthorised actions
Link to external data sources for pricing, ratings or asset events
Smart Collateral, Not Just Cash
While earlier ISDA surveys (e.g. 2014) reported cash making up nearly 75% of posted collateral, more recent data shows that by 2019 it had declined to around half. Cash remains dominant because it’s easy to value and move. But there’s nothing that says real estate or receivables can’t be collateral - they’re just harder to handle.
Agentic infrastructure makes it possible to:
Tokenise and fractionalise illiquid assets
Apply real-time valuation and eligibility logic
Automatically compare diverse asset types based on a collateral score
That means you could settle a margin call using a portfolio of cars, equities and digital cash - all assessed, monitored, and moved in real time.
What About Rehypothecation?
After 2008, the collapse of Lehman Brothers exposed how fragile and unclear rehypothecation rules were. In Europe, brokers could assert ownership of client securities - even when clients weren’t in default. In the US, reuse is limited to 140% of liabilities.
Lehman used this loophole aggressively, encouraging hedge funds to use their UK entity, LBIE, for more leverage. When it failed, billions were trapped. PwC, LBIE’s administrator, said over 1,500 clients had close to $40 billion in play. The total volume of assets subject to rehypothecation has been estimated at around $22 billion.
One example that always stuck with me was Harbinger Capital. They had assets rehypothecated through Lehman UK, and when it collapsed, those assets were frozen. It took years to get clarity. Not because the terms were unclear - but because enforcement relied on a mix of trust, documentation, and legal interpretation across jurisdictions.
With Agentic Contracts, this could have been different. Rights of reuse could be encoded, tracked, and enforced - not left to interpretation. Smart instructions would define when and how assets could be touched. Ownership logs would be real-time and immutable.
What About CBDCs and Stablecoins?
Much has been said about digital money. JPM Coin is live. The ECB, UK and others are piloting CBDCs. But programmable money is only useful if the obligations tied to it are also programmable.
Imagine cash posted as collateral. The smart contract could define that it is locked for a fixed term, only usable for margin, not for lending. It could even accrue interest or trigger margin top-ups if asset values fall. That’s not just tokenisation - that’s enforceable collateral logic.
From Daily Margining to Continuous Collateralisation
We don’t have to wait until end-of-day to know if positions are at risk. Agentic CSAs:
Track MTM exposures live
Revalue assets using external pricing feeds
Trigger transfers immediately
Escalate exceptions to humans when needed
This eliminates the manual cycle and enables real-time risk mitigation. No more overnight exposure. No more lag in clearing.
From Concept to Infrastructure
Let’s be clear: this isn’t just digitising legal agreements. It’s turning them into active systems. Agentic Contracts let agreements execute themselves.
Self-executing: Transfers happen automatically when thresholds are breached
Self-monitoring: Asset values, ratings and conditions are constantly checked
Self-limiting: Reuse and eligibility rules are encoded and enforced
It’s not that these contracts “replace” lawyers. It’s that they make the contract the system - not just the starting point.
Final Thought
Imagine if back in 2008, those subprime mortgage derivatives had been structured as Agentic Contracts. Their performance would have been calculated in real time, based on actual underlying data. Collateral would have adjusted. Rehypothecation would have been logged and capped.
The crash may not have been avoided, but it would have been smaller, more contained, and faster to unwind.
The use case hasn’t gone away. It’s become more urgent. As private credit and fund-linked derivatives scale, margin precision and cross-border risk management become even more critical. If we don’t upgrade the legal execution layer, the market will remain fragile at the core.
Agentic Contracts give us a new foundation. Not just for documenting agreements, but for performing them. In collateral management, that means faster clearing, smarter rehypothecation, and fewer systemic surprises.
It’s time to retire the fax machine. Collateral is ready to think for itself.