Collateral Optimisation

Collateral optimisation is the practice of deciding which assets to deliver against which obligations so that a firm satisfies all its collateral requirements while giving up as little value as possible. A firm with many counterparties, clearing houses and funding trades usually has a choice of what to post to each. Optimisation is the method, increasingly automated, of making those choices well.

It is an extension of collateral management. Basic collateral management ensures that calls are met correctly and on time. Optimisation asks whether they are being met efficiently. The question became pressing after the financial crisis of 2007 to 2009, when clearing mandates, margin rules for uncleared derivatives and bank liquidity requirements sharply increased the demand for high-quality assets and made their use a measurable cost.

Each collateral agreement accepts a defined range of assets; this is its collateral eligibility schedule. Each asset a firm holds has a cost of use, which may be the return forgone by not lending or investing it, the cost of funding it, or its value in meeting a regulatory ratio. The simplest rule is to deliver, for each obligation, the lowest-cost asset the counterparty will accept. This is often called cheapest-to-deliver allocation.

A government bond that counts towards a bank's liquidity buffer is expensive to give away. A corporate bond that few counterparties accept is cheap to use where it is eligible. An efficient allocation sends the corporate bond to the counterparty with a wide schedule and keeps the government bond for a clearing house that accepts nothing else, or for the buffer.

Real allocation problems are more complicated than a simple ranking, because many constraints interact:

• Haircuts. The collateral haircut on the same asset differs between counterparties, so its effective value depends on where it is sent.
• Concentration limits. Schedules cap the share of any one issuer or asset class.
• Settlement. Assets held in one depository may not be deliverable in time to a counterparty that uses another, and movements carry fees and risk of failure.
• Interest on cash. The rate paid on cash collateral under a credit support annex (CSA) may be better or worse than the firm's alternative use of that cash.
• Regulatory metrics. Posting an asset can affect the liquidity coverage ratio (LCR), leverage and capital.
• Stability. Constantly reshuffling collateral to chase small gains creates operational cost and risk.

Optimisation engines take inventory, obligations, schedules and cost parameters as inputs and solve for an allocation, typically with linear or integer programming. The output is a set of proposed deliveries and substitutions.

Firms describe several stages. The first is building a single view of inventory: knowing, across legal entities, custodians and business lines, what assets exist, where they are and whether they are free. Without that view, assets sit idle in one place while expensive ones are posted elsewhere.

The second is allocation at the time of a call, choosing the best asset for each new requirement. The third is periodic re-optimisation of the whole pool, recalling assets that have become valuable and replacing them through collateral substitution. In tri-party collateral management, the agent performs this continuously within the schedules the parties have agreed.

The fourth stage acts before the trade. Pre-trade tools estimate the margin a new transaction will require at each possible clearing house or counterparty, so that the trader can route it where it adds least to total requirements. Related techniques reduce the requirement itself, for example trade compression, backloading offsetting trades into a central counterparty (CCP), and rebalancing risk between counterparties to lower initial margin.

Where a firm does not hold what a counterparty demands, it can exchange assets. A pension fund holding corporate bonds may need cash for variation margin; it can raise it through a repurchase agreement (repo). A firm can borrow government bonds against lower-quality assets in a collateral upgrade trade, paying a fee. They solve the immediate need but add cost, rollover risk and reliance on markets that may tighten in stress. Reuse of received collateral, through rehypothecation or title transfer, is another source of supply.

Effective optimisation lowers funding costs, frees high-quality assets for regulatory buffers, and can support better pricing for clients. Within a bank, the cost of collateral is charged to trading desks through internal pricing, in the same way as funds transfer pricing (FTP), so that the price of a trade reflects the margin it will consume.

There are limits. Optimisation depends on accurate, timely data about positions and agreements, and many firms still hold these in separate systems. Delivering the lowest-quality eligible asset to every counterparty means that, in a default, each receiver holds the least liquid collateral it agreed to accept; counterparties respond by narrowing schedules, which reduces the scope for optimisation. A strategy that runs spare liquid assets down to the minimum also leaves less room to meet sudden margin calls, so firms set aside a reserve and monitor intraday needs through intraday liquidity monitoring.

Faster settlement and tokenised collateral could allow assets to be moved and substituted within minutes instead of days, making allocation more precise.