Index-Linked Gilts question

zagfles

aroominyork said:

I didn't realise DB pensions were inflation linked but I can imagine how they are.

Private sector ones are usually capped at 3% or 5% which doesn't really provide inflation protection, over the last 5 years a lot of them will have gone down 10-15% in real terms. Really bad news for people who retired a few years ago. 

masonic

zagfles said:

masonic said:

zagfles said:

masonic said:

It's arguable that the arithmetic mean used in RPI really was problematic and CPIH is a better measure. The change could have been made without a sustained raid if the market priced in a premium for the longer duration varieties and demanded a lower price for future auctions. But that doesn't appear to have happened. Instead, we're in a situation where inflation has to be over 1% above target just to break even. There have been long periods of time where this wouldn't be the case. For example, someone index linking for 30 years in 1992 would have had an annualised uplift of 2.8% based on RPI, which might have cost them a little vs not index linking, but if indexed on a CPIH basis would have left them around 30% worse off again. So that is the sort of thing to be factored in when considering including ILG within your bond investments as a private investor.

On the retirement income side, it makes the decision between RPI-linked and flat 3% escalation annuities more difficult than it was, since there is about a 0.4%pa cost to the former over the latter, which will mount up. Essentially, CPIH will need to be sustained at more than 3.4% (1.4% above target) for the cost to be worth it.

The 30 years from 1992 looks like a cherry picked example, most other periods RPI average has been much higher, I used 10, 20, 50, 100 years back from now in a previous post. On CPIH, ONS have modelled CPIH back to 1988 and that shows long term average CPIH from then till now was 2.9%. 

On retirement income the average inflation isn't all that matters anyway, sequence of inflation is very important. Just like with drawdown from equities it's not average returns that are important but sequence of returns. High inflation at the start of retirement will have a far greater impact than high inflation 30 years into retirement. 

In any case, for retirement income the choice been index linked and flat/fixed escalation isn't going to be based on the answer to the question "what will most likely give me the highest lifetime income". The reason people choose index linked is to have a guaranteed real income. If people want what history shows will most likely give them the highest lifetime income, they wouldn't use gilts or annuities at all, they'd use drawdown invested in equities. 

Yes, it was a cherry picked example to highlight even in this period, RPI linking came good, making it a no-brainer. The outcome with CPIH would have been much worse. Working backwards over shorter periods from a date just after an extreme spike in inflation is also cherry picking, even if that date happens to be the present. There will be a wide range of inflation rates over different 10 or 20 year periods in the past.

I don't disagree that inflation linking gives valuable insurance against periods of high inflation, and that it's absolutely required if you don't have the margin to absorb a higher cost of living in the short term. But it will soon come at a higher cost, so those who don't need certainty might be tempted to gamble on flat gilts or 3% escalation instead.

In many cases it won't be either annuity or drawdown, but a mixture of both, so it's worth weighing the potential cost of inflation linking if you have other options that could see you through and leave you better off in the long run, or with more income when you are younger and likely spending more.

Right - which is why I also used very long term (50 & 100 years, 5% and 4.5%) to get a good overall average accounting for periods of both high and low inflation. Also the recent spike really makes a mockery of relying on the BoE inflation target being met, there have been and will be circumstances where inflation will exceed, perhaps massively exceed, the target. As we've seen. 

Overall long term index linked gilts look good value IMO, despite the switchover to CPIH. I also think in the context of the massive budget deficit it must be tempting for the govt to encourage inflation, either explicitly (eg changing the BoE target) or through other actions. Inflation will definitely help solve the deficit problem - stealth taxes (freezing allowances etc) become more effective, non-indexed govt debt reduces in real terms (IIRC non-indexed is about 75%) and tax revenue will likely increase with inflation (VAT, income tax etc). 

So personally I won't be touching non indexed govt debt or flat annuities with a bargepole, I will use index linked gilts/annuities for core income and equities for the cherry on top. 

Start (or end) date sensitivity exists even in those long term figures. The 100 year RPI figure would have been 3.5% in 2021. It is not so very different to those looking at equity returns over the past 10, 20, 50 and 100 years, which leads to a very rosy picture of equity returns. Which is why back-testing tends to use slices of time within the full history available. But I think we've established its the spikes that interest you, not the averages.

Annuity-wise, it would be inflation target +1.4% for a 3% escalator to match inflation linked, and a little north of target +1% for flat gilts. Unless they decide a 2% escalator is now good enough post-CPIH adoption, or something restorative happens in the gilt market respectively (both unlikely I hope). But I think our viewpoints differ as I'm coming at it from the perspective of having the core of my portfolio remain in equities throughout, with fixed income as a contingency.

zagfles

masonic said:

zagfles said:

masonic said:

zagfles said:

masonic said:

It's arguable that the arithmetic mean used in RPI really was problematic and CPIH is a better measure. The change could have been made without a sustained raid if the market priced in a premium for the longer duration varieties and demanded a lower price for future auctions. But that doesn't appear to have happened. Instead, we're in a situation where inflation has to be over 1% above target just to break even. There have been long periods of time where this wouldn't be the case. For example, someone index linking for 30 years in 1992 would have had an annualised uplift of 2.8% based on RPI, which might have cost them a little vs not index linking, but if indexed on a CPIH basis would have left them around 30% worse off again. So that is the sort of thing to be factored in when considering including ILG within your bond investments as a private investor.

On the retirement income side, it makes the decision between RPI-linked and flat 3% escalation annuities more difficult than it was, since there is about a 0.4%pa cost to the former over the latter, which will mount up. Essentially, CPIH will need to be sustained at more than 3.4% (1.4% above target) for the cost to be worth it.

The 30 years from 1992 looks like a cherry picked example, most other periods RPI average has been much higher, I used 10, 20, 50, 100 years back from now in a previous post. On CPIH, ONS have modelled CPIH back to 1988 and that shows long term average CPIH from then till now was 2.9%. 

On retirement income the average inflation isn't all that matters anyway, sequence of inflation is very important. Just like with drawdown from equities it's not average returns that are important but sequence of returns. High inflation at the start of retirement will have a far greater impact than high inflation 30 years into retirement. 

In any case, for retirement income the choice been index linked and flat/fixed escalation isn't going to be based on the answer to the question "what will most likely give me the highest lifetime income". The reason people choose index linked is to have a guaranteed real income. If people want what history shows will most likely give them the highest lifetime income, they wouldn't use gilts or annuities at all, they'd use drawdown invested in equities. 

Yes, it was a cherry picked example to highlight even in this period, RPI linking came good, making it a no-brainer. The outcome with CPIH would have been much worse. Working backwards over shorter periods from a date just after an extreme spike in inflation is also cherry picking, even if that date happens to be the present. There will be a wide range of inflation rates over different 10 or 20 year periods in the past.

I don't disagree that inflation linking gives valuable insurance against periods of high inflation, and that it's absolutely required if you don't have the margin to absorb a higher cost of living in the short term. But it will soon come at a higher cost, so those who don't need certainty might be tempted to gamble on flat gilts or 3% escalation instead.

In many cases it won't be either annuity or drawdown, but a mixture of both, so it's worth weighing the potential cost of inflation linking if you have other options that could see you through and leave you better off in the long run, or with more income when you are younger and likely spending more.

Right - which is why I also used very long term (50 & 100 years, 5% and 4.5%) to get a good overall average accounting for periods of both high and low inflation. Also the recent spike really makes a mockery of relying on the BoE inflation target being met, there have been and will be circumstances where inflation will exceed, perhaps massively exceed, the target. As we've seen. 

Overall long term index linked gilts look good value IMO, despite the switchover to CPIH. I also think in the context of the massive budget deficit it must be tempting for the govt to encourage inflation, either explicitly (eg changing the BoE target) or through other actions. Inflation will definitely help solve the deficit problem - stealth taxes (freezing allowances etc) become more effective, non-indexed govt debt reduces in real terms (IIRC non-indexed is about 75%) and tax revenue will likely increase with inflation (VAT, income tax etc). 

So personally I won't be touching non indexed govt debt or flat annuities with a bargepole, I will use index linked gilts/annuities for core income and equities for the cherry on top. 

Start (or end) date sensitivity exists even in those long term figures. The 100 year RPI figure would have been 3.5% in 2021. It is not so very different to those looking at equity returns over the past 10, 20, 50 and 100 years, which leads to a very rosy picture of equity returns. Which is why back-testing tends to use slices of time within the full history available. But I think we've established its the spikes that interest you, not the averages.

Annuity-wise, it would be inflation target +1.4% for a 3% escalator to match inflation linked, and a little north of target +1% for flat gilts. Unless they decide a 2% escalator is now good enough post-CPIH adoption, or something restorative happens in the gilt market respectively (both unlikely I hope). But I think our viewpoints differ as I'm coming at it from the perspective of having the core of my portfolio remain in equities throughout, with fixed income as a contingency.

I don't know where you get 3.5% from. Are you using the arithmetric average (that make no sense, eg if inflation over 2 years is 0% then 10% that's not the same as 5% then 5%). Geometric average is from 1921-2021 is 4%, easily worked out from the RPI index:

Retail Prices Index: Long run series: 1800 to 2024: Jan 1974=100 - Office for National Statistics

1921 index is 23.1. 2021 index is 1203.2. Average RPI is (1203.2/23.1)^(1/100) -1 = 4.0%.

Anyway, I'll play your game, lets use 30 year time slices, going as far back as 1900. 

I took the (geometric) average RPI for all 30 year periods starting each year from 1900 to 1994, and the (arithmetric) average of those averages is 4.56%. The worst case is 8.1% and best is 0.9% 

So going on your figures of 3% or 3.4% inflation being the breakeven, and assuming CPIH is 1% lower than RPI, inflation linking beats flat/fixed on average over 30 year periods since the start of 1900 even after RPI becomes CPIH

masonic

I'm using the Jan 1921 figure of 9.7 and Jan 2021 figure of 294.6. Based on 1987 = 100.

(294.6/9.7)^0.01 = 3.5%.

To avoid quibbling over numbers, the spread of 0.9% to 8.1% will do for sake of argument. It suggests that a fair proportion of 30 year periods were below 4% or 4.4% (roughly the current breakeven CPIH in RPI terms), and a fair proportion above. Given some of the double digit periods in the more distant past, I could believe there is some skew in the distribution such that an equal or greater number of periods were below than above. With a greater proportion of the lower periods being in recent times.

Point is, what once looked like a one-way bet seems rather less so.

But of course if you are looking to insure against the ravages of high inflation, then the fact returns are a bit lower will not trouble you if you hit a double digit period.

masonic

Finally found a few minutes to look at the numbers in more detail. Using the Wolfbane monthly data set, there are 966 30 year periods from 1915 to present. Median RPI comes to 3.8% (highest 8.1%). Comparing hurdles for RPI vs CPIH (assuming a 1% difference), 76% of 30 year periods were above 3% RPI, but only 47% were above 3% CPIH; 64% were above 3.4% RPI, while 44% were above 3.4% CPIH.

Again, not disputing this is still valuable protection against a still fairly likely risk if your priority is maintaining the spending power of your capital.

zagfles

masonic said:

I'm using the Jan 1921 figure of 9.7 and Jan 2021 figure of 294.6. Based on 1987 = 100.

(294.6/9.7)^0.01 = 3.5%.

To avoid quibbling over numbers, the spread of 0.9% to 8.1% will do for sake of argument. It suggests that a fair proportion of 30 year periods were below 4% or 4.4% (roughly the current breakeven CPIH in RPI terms), and a fair proportion above. Given some of the double digit periods in the more distant past, I could believe there is some skew in the distribution such that an equal or greater number of periods were below than above. With a greater proportion of the lower periods being in recent times.

Point is, what once looked like a one-way bet seems rather less so.

But of course if you are looking to insure against the ravages of high inflation, then the fact returns are a bit lower will not trouble you if you hit a double digit period.

Completely untrue! Going back as far the ONS data goes, to 1800, the 30y average inflation doesn't exceed the overall average until the 30y period ending 1968. The only above average 30y periods in the last 200 years are those ending 1968, 1969 and then a long spell from 1974-2008.  

Inflation seems to be very much a modern problem - looking at the 19th century and first part of the 20th century inflation was very low and there was a lot of deflation. So with average RPI over 30y periods since 1900 being over 4.5%, with the long term trend being inflation is getting worse not better, index linked gilts/annuities look a better bet in terms of overall returns as well as obvious safety/insurance. 

masonic

I agree with you that before the 20th century the picture looked very different with deflation being common, so excluding that very old data from analysis and thoughts seems sensible, as we both have done. Using monthly RPI since 1915 to get a median 3.8% for 30 year periods, the worst data seems to be in the middle - the 1970s being particularly high. I haven't previously looked back further than that TBH, so hadn't appreciated the lower period in the early part of the 1900s before now. 

Taking the whole available monthly 1915-2025 dataset (seems to strike the balance between acknowledging vast deflationary periods of the distant past aren't relevant, while not ignoring the 1920s depression), it seems a coin toss as to whether there would be any benefit for overall returns after the CPIH update (evens at best), whereas it was a good bet before (up to 1-3 on).

Linton

ISTM The purpose of investing in index linked assets is to match the increase due to inflation in one's average expenditure not to maximise long term returns. No other investment provides that guarantee. There is therefore no point in comparing ILs with any other investment.  If you want to maximise long term returns invest in equities.

I have detailed expenditure data going back to 1999.  Over the 19 year period since retirement my normal annual expenditure ignoring major one-offs has risen from around £33K to around £50K.  There has been no noticeable change in standard of living.  This represents an increase of 50%.

Over the same period CPI has risen by about 70% and RPI by 100%.


 

MK62

Which kind of confirms that the headline inflation figures are really nothing more than a guide and one's personal inflation rate might be quite different.

zagfles

Linton said:

ISTM The purpose of investing in index linked assets is to match the increase due to inflation in one's average expenditure not to maximise long term returns. No other investment provides that guarantee. There is therefore no point in comparing ILs with any other investment.  If you want to maximise long term returns invest in equities.

Indeed - as I've been saying, but as well as that my interpretation of the data using ONS figures seems to suggest that index linked also provides better overall returns on average than flat or fixed escalation. So when people talk about the "cost" of index linking, that "cost" looks to be negative on average. So you could well get both the advantages of higher overall returns and protection against inflation.