The endgame is nigh: time to pay more attention to credit?

DB pension scheme funding levels have improved dramatically since the depths of the pandemic. It’s impossible to be precise, but our recent estimates of the buyout funding level for a typical scheme are around 100%[1].

This accelerated progress has led to a mindset shift among trustees. It’s no longer so much about growth – it’s about getting ready for buyout. A scheme may be well enough funded for buyout but still not be ready to do so, perhaps due to illiquid asset allocations or a need for administrative preparation.

Managing risks relative to buyout liabilities and pricing is therefore more important than ever. Other than longevity, there are three main components to this: interest rates, inflation and credit spreads.

The focus of this blog is credit spreads. Buyout prices tend to increase when credit spreads narrow because insurers invest in credit, and credit-like assets, to support buyout contracts. This is an important source of risk in the endgame. But just how sensitive are they?

Insurer asset allocations

The pie chart below shows the average asset allocation held by insurers to back annuities.

This is useful background knowledge, but it is not enough for us to understand the credit sensitivity of buyout pricing because we are interested in pricing new business, as opposed to insurers’ back books. We also want to know how the strategy varies depending on the scheme’s maturity and on market conditions.

Based on data we have on what Legal & General tends to use in its pricing portfolios, we have estimated typical allocations used to price bulk annuity contracts. This represents an average or ‘strategic’ allocation used for pricing new business. To work out the overall credit sensitivity of these allocations we then allowed for three key dimensions:

• The durations of the instruments. Sensitivity is proportional to duration
• The spreads relative to liquid investment grade (IG) credit. Sensitivity is proportional to spreads
• The estimated correlations to IG credit spreads. Illiquid direct investments (DI), such as infrastructure and social housing, have spreads that are correlated to liquid credit spreads. To estimate correlations, we used historic data and performed suitable regressions[2]. Typically, we found this number to be about 40%

The tables below illustrate how these components can combine[3] for two schemes. We calculate PV01, the change in price from a one basis point move in interest rates, and CS01, which we define here as the movement in the price for a one basis point move in IG credit spreads[4]. The credit sensitivity is the ratio i.e. CS01/PV01.

As you can see, the pass through works out at 66% for a 10-year duration scheme and 47% for a 20-year duration scheme.

Explaining price levels

Another way to estimate the level of credit in insurer portfolios involves calculating how much return insurers must generate to support the observable pricing of individual annuities in the retail market[5]. This uses the standard model for Solvency II, allowing for profit margins. That process can allow for an assumption for DI assets that are imperfectly correlated to credit. As this method uses up-to-date annuity pricing and moves according to market conditions, it gives a more dynamic sensitivity over time. We monitor and compare both approaches on a regular basis.

Wider considerations

It can make sense to underweight credit relative to the modelled credit sensitivity to reflect a concern that a large blow-out in spreads may not pass through into cheaper pricing at the usual rate. In addition, insurers sometimes use larger default provisions when spreads are wider, which can dampen the sensitivity.

What’s the answer?

Our preferred approach points towards a typical credit sensitivity of about 60% of PV01 for mature schemes with durations around 10 years, dropping to around 40% of PV01 for schemes with durations of around 20 years:

Changes with market conditions

This varies across insurers and over time as market conditions change. We caution against trying to mimic any short-term changes in the pricing portfolio for the purposes of getting a slightly better hedge. We believe that tactical tilts to asset allocation should primarily reflect temporary views from an investment perspective, rather than attempts to run slightly less risk relative to buyout pricing. However, a periodic review of what insurers are likely to be holding on average is important.

Endgame risk management

The CS01 of the buy-in or buyout liabilities can be approximated as 40-60% of the PV01 of the liabilities, dependent on the maturity of the scheme.

The upshot is a potential expansion of the interest rate and inflation management. Schemes can calculate a credit sensitivity hedge ratio and consider changing their investment strategy to improve that hedge ratio.

We’ll explore this further in our next blog.

[1] As at end November 2022

[2] These allowed for lags that are often encountered when dealing with illiquid assets. We also looked at structural models.

[3] We checked the volatility of spreads divided by the level of spreads is similar across assets.

[4] As opposed to the usual definition of a one basis point move in the spread of the underlying asset.

[5] This assumes that for the same liability profile an individual annuity and a bulk annuity would be backed in a similar way.