Collateral Buffer

Collateral – How Much Is the Right Amount?

10-second Summary

Derivatives can be powerful risk management tools for pension schemes, but the amount of assets that need to be held as collateral (to provide a buffer against adverse market movements) needs to be set carefully and monitored regularly as part of the scheme’s overall strategic asset allocation. The amount is likely to be between 26% and 40% of assets, depending on which stage the scheme is in.


When discussing Liability Driven Investing (“LDI”) mandates with pension schemes, trustees often ask:

“How much collateral should we hold?”

This is a significant question as it impacts the scheme’s overall strategic asset allocation, and the amount and composition of assets needed to generate returns.

While there is no single correct answer to this question, we believe it can be tackled with a thorough understanding of the risks in the hedging portfolio, and the size and nature of the hedging trades. We expand more on this in this paper.

So what now/next steps

If you are a pension scheme looking to manage risk then you are probably focused on increasing your interest rate hedge ratio. In doing this, you will need to make sure that the LDI manager holds sufficient collateral to support the hedges. The amount required will influence the design of your overall Strategic Asset Allocation.

In this paper we help schemes answer the two related questions:

  1. “How much hedging can I do with the current allocation to LDI strategies?”
  2. “How much collateral would I need to hold to extend my hedge ratio another 10%”

Interest rate risk represents the largest risk for most UK defined benefit pension schemes. Analysis of industry-wide data as of 2015 suggests that, on average, schemes are roughly halfway to a hedged position (on a funding ratio basis, that is a hedge ratio of 33% compared to a funding level of 65%). There is quite a lot of dispersion behind these numbers, with some schemes well hedged and others with very large risk exposures.

One thing pension schemes can do to reduce interest rate risk is to invest more in bonds. However, this inevitably creates a tension with the returns most schemes need to generate to become fully funded. Our analysis suggests most schemes need to generate returns in the region of gilts +2-3% per annum alongside sponsor contributions to become fully funded. Clearly this cannot be achieved if the majority of assets are invested in gilts.

Thankfully, modern-day LDI techniques can help pension schemes by providing a well-trodden path to unlocking this dilemma, allowing schemes to hedge against falling interest rates AND generate sufficient returns to meet their funding objectives.

Key to this is embracing the use of derivatives to reduce risk. However, doing so brings additional risks that must be managed. One of these is the need to hold collateral in order to provide a buffer against losses in the hedging portfolios (for example, in a situation where long dated interest rates rise substantially).

Collateral typically consists of high quality and/or liquid securities. These are held by the LDI manager, and available to be “posted” to a counterparty or clearing exchange, if required, under the terms of a derivative trade.

One benefit of using derivatives for risk management is that the amount of collateral that needs to be held is less than if all of the assets were to be invested in bonds. As many pension schemes look to reduce their risks by increasing their hedge ratios further to close the interest rate and inflation risks, a question we frequently hear asked is:

“How much collateral should we hold?”

Quick Summary

This paper details the thinking behind the collateral test we recommend clients use to right-size their LDI portfolio. These are necessarily client-specific and will lead to a range of possible answers between 26% and 50%:

Illustrative Characteristics of Scheme Approximate Total Scheme Asset Allocation Collateral Requirement*
Short-dated liabilities, high holdings of credit assets (End Game) 25-30%
Long-dated liabilities, gilts based LDI benchmark (Opening or Middle Game) 35-45%
Long-dated liabilities, swap-based LDI benchmark with central clearing (Opening or Middle Game) 40-50%
Source: Redington
*Collateral required is for illustrative purposes only and includes assets invested in gilts, cash and liquid forms of credit.

Section 1 – Introduction to Collateral Tests

The purpose of a collateral test is to answer this question:

“What level of assets is it prudent to hold in order to provide a buffer against potential losses in the hedging portfolio?”

In the case of most pension schemes, they will have implemented these hedges to guard against the risk of interest rates falling (and thereby increasing their liabilities). If interest rates were to rise, the hedging portfolio will experience a loss. The liabilities will fall as well, so overall this loss is offset at an overall combined asset and liability level.

However, in the short term, the counterparty to the derivative contract will be due collateral (under the terms of the trade). So the collateral held by pension schemes is there largely to provide a buffer against long-dated interest rates rising (or inflation falling). There may also be collateral requirements relating to currency hedging and synthetic equity exposures.

The graph and table below provides some historical context for the question of how large an interest rate move should be provisioned for with collateral.

Time Period 1 day 1 week 1 month 1 year
Maximum move in 20 year gilt rates 1996-2016 45bps 72bps 78bps 211bps

The final chart shows multiple instances of 20-year gilt rates moving by more than 150bps over a one year period. So collateral management is very important in case similar moves occur in the future.

There is no single right answer to the question of how much collateral should be held, but it is an important question as it influences the overall strategic asset allocation.

Our approach is to define a range of different measures and tests, and use one to guide decision making that we think errs on the side of conservatism. In order to come up with these collateral tests, we generate plausible adverse “shocks” to the derivative portfolio.

We typically take three approaches to generating shocks to the derivative portfolio:

  1. Risk based
  2. Straight shock
  3. Combination

These guide the amounts we recommend our clients keep in collateral.

  1. Risk-based measures typically look at a measure such as VaR to generate a reasonably adverse shock to the derivative portfolio.


This takes into account all the various exposures in the derivative portfolio and their relative riskiness (for example, interest rates, inflation, equities).

It can also take into account any expected diversification between them.


The disadvantage of a VaR type measure is that it is typically model-dependent and parameter-dependent, where parameters are often hard to estimate (particularly correlations).

It may be that a VaR model is heavily dependent on one parameter that changes sharply, which gives sharply changing answers that may be difficult to interpret.

VaR will also vary over time; in benign periods a VaR measure may understate risk.

There is a nuance related to the use of VaR metrics for collateral tests and this relates to the extent to which diversification is allowed for. In particular, removing diversification from a collateral test creates a more conservative scenario by increasing the level or risk illustrated compared to a scenario where allowance is made for diversification. A VaR measure lacks simplicity, especially when viewed by an oversight group not involved in the day-to-day investment decisions (such as a trustee board).

  1. Straight shock takes a single variable (for example, interest rates) and shocks it by a set amount (eg 100 basis points).


This has the advantage of simplicity; there is no model dependency, and is transparent and easy to understand.

It doesn’t attempt to relate back to the actual real-world chance of the prescribed shock occurring, and it doesn’t prescribe how to shock other associated variables e.g. inflation.

  1. A combination simply uses both of the above approaches.

For example, a VaR 95% measure that is subject to an additional shock of 100bps. We believe this is the most helpful overall measure to use, as it combines the advantages of both approaches while addressing some of the disadvantages.

We also advocate illustrating a number of the above shocks in order to provide “early warning” signals.

For all of our retained pension scheme clients, the amount of available collateral they have becomes a key metric to track.

There are 6 collateral tests that we track in our ALM tools which correspond to all three approaches to generating shocks:

  1. Undiversified VaR

We calculate the Value-at-Risk of the hedging portfolio according to our asset liability model, but with a special setting that makes no allowance for diversification between categories (for example no diversification between gilts and swaps). If offsetting positions within a category are held – for example offsetting swap positions – then an allowance for diversification would be made between the positions. Doing so generates a more conservative test (that is, a larger “shock”) than if we were to allow for diversification in the usual way.

  1. Diversified VaR

The diversified VaR is calculated according to our ALM risk model, with standard settings, allowing for diversification between assets (for example, between gilts and swaps).

  1. Undiversified VaR + 100bps

This is the most conservative test in our range and the one we recommend clients use to size their collateral pool. This test takes the undiversified VaR (1) and adds a further 100bps upwards shock to interest rates.

We believe this generates a fairly conservative scenario which pension schemes can confidently use for planning purposes. That is not to say we think this shock is likely to happen instantly or over a short space of time; were an adverse shock of this magnitude to occur, we believe it would likely be over a period of weeks or months. Planning with this level of collateral in mind gives the pension scheme peace of mind that there is plenty of time to react to changing market conditions, and a low likelihood of needing to sell other scheme assets in a short space of time.

As the charts in the rest of the document show this test is broadly equivalent (for most schemes) to a single interest rate shock of 200bps (up), however calculating in this way implicitly includes the effect of other variables and assets into the VaR calculation.

  1. Rates up 100bps

Shows the effect on the portfolio if long dated nominal interest rates (both gilt and swap) were to rise by 100bps.

  1. Rates up 200bps

Shows the effect on the portfolio if long dated nominal interest rates (both gilt and swap) were to to rise by 200bps.

  1. Inflation rates down 100bps

Shows the effect on the portfolio if long dated inflation (both gilt and swap based) were to fall by 100bps.

Management of the Collateral Pool

We believe the collateral pool can be split into two parts:

  • The Primary reserve > required to be kept in highly liquid securities that can be posted almost immediately to cover short-term derivative losses. This would need to be held in highly accessible cash (or in liquid mark-to-market of centrally cleared positions).
  • The Secondary reserve > kept to replenish the primary reserve, such that the portfolio can withstand sustained losses without being a forced seller of other asset classes within the scheme. This can include short-dated and liquid elements of credit in order to enhance return (rather than be limited to cash and more traditional LDI instruments).

In addition, schemes with swap positions that are centrally cleared will need to post initial margin to cover these positions.

For schemes with segregated LDI portfolios it is essential that the primary reserve be managed by the LDI manager, and it is likely that most of the secondary reserve would also be managed or overseen by the LDI manager. It is possible that components of the secondary reserve be managed by external managers, but it would be good practice to have a process in place for the LDI manager to be able to call on these assets were they required.

For schemes with pooled LDI in place, an amount similar to the primary reserve will generally be required to be physically invested into the manager’s pooled funds (the amount varies depending on the provider, but they are broadly similar). We believe it is still prudent for the pension scheme to dedicate a secondary reserve, outside that which is held in the LDI portfolios. This would be used to replenish the capital in the LDI funds in the case of a capital call from the funds.

Section 2 – Introduction to Illustrative Examples

As a scheme’s individual situation and objectives affect the approach taken in the LDI portfolio and therefore to collateral, we demonstrate how these frameworks apply to three illustrative pension scheme scenarios, each with different situations and objectives.

  1. Opening Game

Schemes in the Opening Game (around 50% of UK schemes) are typically targeting high required returns of around gilts +2.5% or higher per annum. They need to make the most efficient use of their asset pool in order to hedge their risks while also generating high return.

They may typically have a longer-dated set of liabilities (due, in many cases, to being open to future accrual more recently) leading to greater hedging requirements. They are likely to have low allocations to bonds, so are looking to LDI instruments and derivatives to achieve their hedging goals, while allocating the balance of the assets to high returning liquid and illiquid assets.

  1. Middle Game

Schemes in the Middle Game (around 25% of UK schemes) are generally targeting required returns of around gilts +1.5% to gilts +2.5%. They have one eye on their end game targets (which might be buyout or self-sufficiency) and are looking to build out credit portfolios to fulfill the End Game strategy. They are therefore also interested in the relationship between the credit portfolio and the LDI hedge.

  1. End Game

Schemes in the End Game (around 25% or less of UK schemes) typically need returns of less than gilts +1.5%p.a. They are looking to build buy-and-hold credit portfolios to create a high quality portfolio. This allows them to own future secure cashflow streams to meet the benefits as they fall due, controlling for re-investment risk. They may also be interested in long-dated illiquid instruments, and how these can contribute to their interest rate hedge. Often, they will be more mature schemes, with shorter-dated liabilities.

Opening Game Middle Game End Game
Example Liability Duration (years) 24 24 13
Funding Level 60% 70% 80%
Required excess return (%p.a.) Gilts +3% Gilts +2.2% Gilts +1%
Target Hedge Ratio (%) 60% 70% 80%
Likely asset portfolio Bias to liquid markets Mix of liquid Market and liquid credit Bias to credit

Funding ratios are expressed on a self-sufficiency basis of gilts +0.5%. For the purposes of the examples, ongoing sponsor contributions are assumed at £10m (1% of liabilities) per year.

Section 3 – Example of Collateral Test – Opening Game

The charts below illustrate our sample asset allocation for a scheme in the Opening.
Here, the scheme (illustrative £1bn of liabilities, 24 year duration) needs to generate £1.5m of PV01 exposures from its hedge portfolio in order to bring its hedge ratio up to the funding level of 60% and reduce the interest rate risk.

According to our collateral test, the scheme needs to hold around 40% of its assets in the LDI portfolio to support the overlays that take the hedge ratio to 60%. These assets are equally split between index-linked gilts and cash held to support derivative positions.

The scheme makes use of overlays in the liquid market space to generate the returns it needs. It can make some use of high-returning illiquid assets to generate returns, but this is limited due to the need for liquidity.

The scheme has a limited role for assets such as investment grade corporate bonds, as these do not generate the high returns that are needed or provide much hedging, nor can they be used as collateral. The collateral requirements are composed of the risk of the swaps moving against the scheme (rates rising) while the value of the gilts simultaneously falls.

We would recommend that the primary reserve be held to cover the risk of adverse moves in the swap portfolio, and that this be held in highly liquid cash. The balance of the collateral requirement would be held in the secondary reserve. This would consist of a combination of gilts and short-dated liquid credit assets in order to enhance the overall returns of the pool.

If the scheme has a swap-based liability benchmark, this will likely lead to extensive use of swaps in the hedging portfolio. This may create the need for initial margin to be posted to the clearinghouse (which can be in the form of cash or gilts). To be consistent, this should add to the collateral requirement for the scheme.

Section 4 – Example of Collateral Test – Middle Game

The charts below illustrate the high-level asset allocation that we believe would apply to schemes in the Middle Game (needing to generate expected returns of gilts +1.5%-2.5%). For an illustrative £1bn scheme, we estimate that they would be looking to hedge around £1.75m of PV01 exposure in order to bring their hedge ratio up to the funding ratio of 70%.

The scheme is likely making substantial use of liquid credit strategies as it looks to build out its portfolio toward the End Game. There is also likely to be significant allocations to liquid market assets, such as equities and risk premia strategies, in order to generate the returns required. The scheme will be getting some interest rate hedging from the credit allocations, but the bulk will be coming from gilts and swap exposures in the LDI portfolio.

We estimate these schemes will need to hold about 40% of their assets in LDI to cover the collateral test described before. Again, schemes with a significant swap-based hedge which uses central clearing may need to hold additional collateral to cover initial margin requirements. As before we would split the collateral pool into primary and secondary reserves, with some of the secondary reserve held in return enhancing assets such as liquid credit.

Section 5 – Example of Collateral Test – End Game

Schemes in the End Game are often more mature and hence have shorter-dated liabilities. Their illustrative asset allocation, shown below, is dominated by credit assets in order to generate the excess returns needed to pay benefits with a high degree of certainty.

The illustrative £1bn scheme here needs to hedge £1.2m of PV01 in order to reach its interest rate hedging target, with a significant proportion of interest rate hedging coming from the credit assets. We estimate this would require 26% of the assets to be invested in an LDI portfolio (gilts and cash) to support the balance of the hedging. The assets in the LDI portfolio are also required to support currency hedging of overseas corporate bond portfolios.

The balance of the assets are invested in long-dated investment grade corporate bonds in both sterling and US dollar (currency and interest rate hedged).

Illustrative Characteristics of Scheme Approximate Total Scheme Asset Allocation Collateral Requirement*
Short-dated liabilities, high holdings of credit assets (End Game) 25-30%
Long-dated liabilities, gilts based LDI benchmark (Opening or Middle Game) 35-45%
Long-dated liabilities, swap-based LDI benchmark with central clearing (Opening or Middle Game) 40-50%
Source: Redington
*Collateral required is for illustrative purposes only and includes assets invested in gilts, cash and liquid forms of credit.

We are sharing our research with the world and we don’t want to you to miss out. Sign up to the Ampersand newsletter and stay up to date with all the latest research.

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.