Views from the LDI desk – LPI: A Hidden Source of Cashflow Mismatch?

LPI: A Hidden Source of Cashflow Mismatch?

A combination of higher hedge ratios, schemes locking down risks as they approach endgame and volatile inflation levels has prompted much discussion on approaches to managing inflation-linked liabilities. We wrote about this 2 years ago, flagging some of the considerations around the approach taken to risk manage LPI liabilities.

However, aside from considerations around inflation risk itself, it is also worth noting the impact from a cashflow perspective. As many schemes progress further towards their desired endgame and debate the relative merits of various forms of cashflow matching and associated objectives, it is an opportune time to look at the disconnections that can occur between liability cashflows and liability benchmark. This piece looks particularly at the impact LPI-inflation linkages can have.

What is LPI?

Before we delve into the detail, it is worth recapping what LPI is, and it’s relevance to the UK pensions landscape.

LPI, or Limited Price Indexation, is a structure whereby an index is created, increasing annually by inflation, subject to caps and floors. There are numerous variants of LPI, but the most commonly referenced is LPI(0,5) whereby the index increases each year by annual inflation, subject to a floor at 0% and a cap at 5%.

Single year outside of range can impact all future cashflows

Of course, for LPI(0, 5), when inflation realises within a 0%-5% range, RPI and LPI increase in line. However, when it realises outside of this range, the indices deviate and this feeds through to all future dates.

How prevalent is LPI?

It is hard to conclusively demonstrate the prevalence of LPI across schemes, as linkages vary between schemes and even cohorts within the same schemes. However, statutory legislation itself leans a proportion of scheme benefits towards LPI structures. Pensions in payment must be increased annually in line with prices for pension rights accrued since April 1997. This indexation is capped at 5% for pension rights accrued between April 1997 and April 2005, and at 2.5% for rights accrued since then.

Trading LPI and Inflation Options

One of the key challenges with managing LPI risks is the asymmetry between supply and demand. LPI supply has typically been limited to a small number of public deals such as those issued by Welsh Water and Tesco Property Finance many years ago and a limited number of transactions, such as a select number of property deals, where there were contractual LPI-linked revenues. However, combining these still leads to a significant shortfall versus the demand from the £1.4tn UK DB pensions sector.

Banks historically filled a small amount of this shortfall by trading LPI, even when there wasn’t the exact offsetting position. However, capital and risk constraints driven by evolving regulations have limited the extent to which this occurs, and over recent years, we have seen more LPI trades unwound than new trades being initiated, as schemes have focused more on liquidity.

However, whilst LPI trading activity is sporadic, the implications of the structural supply and demand mismatch remain evident. This can be illustrated in the below charts.

If we look at the current RPI forward curve, expected inflation is, on average, much nearer to the 5% cap than the 0% floor.

RPI forwards are closer to 5% cap than the 0% floor

Past realised inflation has also shown a distribution that is more skewed to the upside. A combination of these factors would suggest that the cap should be more expensive than the floor.

Historical inflation distribution shows a skew to the upside rather than the downside

However, if we look at the history of cap and floor prices, we can see that the converse is true and floors are more valuable than caps – which illustrates quite how asymmetric the supply and demand dynamic has been. Given their liabilities schemes are natural buyers of floors and sellers of caps and market pricing has been driven by technical supply and demand factors rather than the fundamentals of where inflation has typically printed.

5Y25Y 5% Cap – 0% Floor Premium shows the excess value in 0% floors

Estimating and Managing LPI risk

LPI risk in liabilities is typically estimated using so-called 3D-2D modelling to pull the various tranches of member benefits together, including pre and post-retirement inflation linkages, before applying a modelled inflation distribution. This inflation model can be built either using market implied volatilities or a flat volatility assumption, typically provided by a scheme actuary.

Both have their limitations. The market-implied route does reflect a representative level of traded instruments. However, as illustrated above, this is an illiquid and heavily skewed market. For the flat-volatility approach, aside from the primacy of assuming a constant volatility, there is also the challenge of what single volatility level to use. This has been further complicated by the recent variability of realised inflation.

Sources of disconnection in modelled liabilities

It is first important to understand is the drivers that can cause a disconnection between LPI and RPI. We can highlight three key areas: i) imminent fixings, ii) moves in the RPI forward curve iii) changes in implied volatility. We can go through these one-by-one:

Imminent Fixings

Inflation models look at projected levels of inflation. Much of this is driven by traded inflation-linked instruments, such as swaps or bonds. However, there can also be considerable sensitivity to short term fixings. These can be challenging to project, particularly in an era of inflation shocks. The below illustration shows how realised (CPI) inflation differed materially from Bank of England projections. This was taken in mid-2019, before the Covid pandemic and conflict in Ukraine heavily impacted inflation. It is shared not to critique the Bank of England forecast, which was made prior to Covid and the Ukraine conflict being predictable, although it does highlight how volatility can significantly diverge from reasonable predictions from experienced forecasters.

Illustration of divergence between realised CPI and BoE projections post Q2 2019

What is also worth noting is that if the inflation shock were to take the fixing outside the LPI range, there will be a potentially significant disconnect between the performance of LPI-linked liabilities and RPI-linked assets that may be used to construct a hedge.

The key consideration is that it is likely worthwhile to consider re-calibrating a liability benchmark immediately after a relevant fixing (for example, the fixing relating to a month which is a common reference for scheme members), particularly to the extent that this fixing is either i) materially different from expectations or ii) outside of the LPI range.

Moves in the RPI Forward Curve

The relative lack of LPI-linked trading means that by far the most common approach to hedging LPI-linked liabilities is the use of a suitably adjusted amount of RPI linked assets. This is achieved through ‘delta-hedging’. This involves assessing the inflation sensitivity of both the LPI liabilities and RPI assets to a one basis point shift in the underlying inflation curve. Given the cap and floor would make the LPI linked liabilities relatively less sensitive, this means that typically a smaller quantity of RPI linked assets are purchased.

However, the ‘delta’ of LPI is not constant, and varies as the proximity to the option strike varies. This can lead to the need for rebalances. This can be illustrated by looking at a chart showing how delta can vary with the underlying inflation swap rate.

20 year zero coupon LPI(0,5) delta plots under fixed and market implied vol assumptions shows the impact of both inflation levels and volatility assumptions can be material to hedge ratios

At present, when inflation remains at the downward sloping part of the chart beyond 3%, rebalances can generate positive P&L. i.e. you are either selling inflation after it has risen, or buying it after it has fallen. There has been discussion on the impact that could occur should a substantial fall take place in inflation forwards, such that the forwards move to the ‘downslope’ of the delta curve. The inference being that if inflation falls, the delta would also fall, leading to the potential for further sales to rebalance and a possible spiral.

Whilst comparisons have been made to the feedback loop observed during the gilt crisis, there are some key differences. What is particularly notable is that whilst collateral calls take place daily across the market, inflation rebalances are typically less frequent, less concentrated and more optional in nature if market fundamentals look unreasonable.

Inflation Volatility

The last few years have seen large changes in realised inflation. A large proportion of schemes utilise a flat volatility assumption in inflation modelling. The input volatility is typically significantly lower than the realised volatility we have seen over the past years. Whilst it is hard to conclusively opine on what a ‘fair assumption’ could be for input volatility, it would not be unreasonable to assume that it could be higher than previously used inputs.

Rolling 3 year volatility has increased materially after the recent spike in inflation

There are other considerations, including the upcoming reform of RPI, which will align with the slightly less volatile CPIH index, that may be worth considering. Crucially, any change in the volatility assumption will only affect the LPI liabilities and not any RPI hedges. These can lead to a step change in both expected cashflows and hedge effectiveness.

LPI Analysis

To better illustrate the impacts on cashflows, we have looked at the impact on a series of payments linked to LPI. We have scaled each of these payments to 100, to better identify the magnitude of the impacts.

• For the initial fixing impact, we have looked at shifting the first years expected print, whilst keeping all other forwards unchanged
• For the delta impact, we have looked at a parallel shift of the forward curve. This is expressed as a curve move in basis points
• For the volatility impact, we have looked at changing the flat volatility assumption

From the below, we can see that these can all have substantial impacts on future cashflows.

However, given we know that a substantial proportion of schemes have high inflation hedge ratios, it is also worth considering the impact net of a scaled RPI hedge. This shows some interesting conclusions. Whilst the shift in the forward curve had the highest impact on raw cashflows, there is a substantial offset from the RPI hedge, even though the delta will have changed slightly. However, the net impact of the other two drivers is more substantial.

Cashflow matching implications

Whilst the above is clearly interesting from an inflation hedging perspective, it is often under-appreciated when looking at cashflow matching tolerances for Cashflow Driven Investment (CDI) strategies.

CDI is becoming increasingly topical as part of endgame discussions, with mandates being structured specifically to generate cashflows to align with scheme requirements. This is a positive and logical development for schemes which are looking to further reduce basis risk whilst efficiently navigating becoming cashflow negative.

But important to note is that the mandates are structured to generate cashflows based on the best estimate of liability cashflows that have been translated and simplified. Even when more granular liability benchmarks are calculated through 3d2d analysis, there is still the potential for shocks in near term inflation or changes in volatility assumptions to have meaningful impacts on the projected cashflows.

This can be relevant when considering the trade-offs around granularity of cashflow matching. If a portfolio is constructed with the constraint of very strict cashflow adherence, this can involve some trade-offs. Particular bonds may be attractive primarily in order to align with cashflow requirements, rather than being appealing based on relative value or credit quality. Once the variability of the underlying cashflows are considered, this trade off may be sub-optimal.

Alternative approaches, such as wider tolerances to account for the variability and/or a cumulative approach to cashflow provision, may be beneficial. Whilst this may not initially appear to generate as high a level of granular accuracy, the difference may be more marginal once the inherent volatility of underlying liability cashflows is accounted for. Additionally, it allows for further flexibility in order to respond to changing market conditions and new opportunity sets.

The example above illustrates the impact of this against a series of liability cashflows. The initial spread impact is relatively limited, which suggests that all else being equal having an initial optimisation to closely adhere to liability cashflows may have limited benefits when the trade-off with loss of flexibility and increased frequency of rebalancing is considered.

Whilst the above example purely shows the impact of LPI on projected future cashflows, this is only part of the story. There are a number of other factors including longevity, shifts to other actuarial assumptions and transfers out which can all impact future cashflow projections, and are not typically captured in liability benchmark sensitivities. It is worth considering the impact of these when looking at both inflation hedging strategies and also cashflow driven solutions. We are happy to work with pension schemes, their trustees and advisors to help them best consider some of these variables and how to translate them into mandate design.

Key Takeaways for Pension Schemes

• Refresh your liabilities after relevant inflation fixings for your scheme to capture any material changes due to the print interaction with caps/floors or misses vs. market expectations.
• Question your inflation volatility assumptions periodically to ensure your hedge is as you expect.
• Consider setting a separate cashflow redemption schedule rather than using the calibration of a risk based hedging benchmark to inform cashflow matching.
• Cumulative cashflow coverage may be a better metric than granular cashflow matching given the inherent uncertainties in cashflows.