UK linkers becoming stinkers

In the recent emergency UK budget it was announced that public sector indexation would change from RPI to CPI from April 2011.  Now, the government is proposing moving private sector schemes and the Pension Protection Fund (PPF) indexation to CPI too.  As Pensions Minister Steve Webb argued, it makes sense switching to CPI as it’s the measure that the BoE targets and (slightly more dubiously) CPI is a more appropriate measure of pension recipients’ inflation experiences.

RPI has historically been higher than CPI, exceeding CPI by 0.55% on average over the last twenty years, so if the differential between the measures continues in future then this proposed change would reduce pension fund deficits but would penalise scheme members if (and presumably when) it is implemented.

The problem with this proposal is that you can’t buy CPI linked assets in the UK – they don’t exist.  While the correlation between RPI and CPI is reasonably close as you’d expect, the difference was as high as 3.1% in 1989 and is currently a relatively large 1.7%.  RPI linked assets are still a better hedge against inflation than any other asset class, but there will definitely need to be CPI linked assets at some stage.

It may be possible to restructure the existing RPI linked index-linked gilts, but the easiest thing would be to issue new CPI linked index-linked gilts.  This would make the RPI linked assets currently in existence pretty redundant.  The good news is that this change, assuming it happens, will likely take years rather than months to implement and even then could well be gradual.  Furthermore, in the meantime investors have no choice other than to continue buying RPI linked assets to hedge against inflation.

But it’s clearly a negative for inflation linked gilts overall, and we’re seeing that in terms of price action today.   Longer dated index-linked gilts are getting hit hardest, partly because they’re longer duration so are more sensitive to changes in yields, and partly because the biggest buyers of long linkers are the pension funds.  At the time of writing, UKTI 1.125% 2037s are down over 2% so far today, while the UKTI 0.5% 2050s are down 3.5%.

Linkers maturing in 20 years or longer have now been in a bear market year to date, which is quite incredible given that long dated conventional gilts (ie those maturing in 15 years or longer) have returned over 13% over the period.   The significant underperformance of linkers has come about despite the UK inflation rate rising significantly this year, with the year on year rate of CPI climbing from 2.9% in December to 3.4% in May – as mentioned in October here, changes in the real yield is a much more important driver of returns for longer dated index-linked gilts than changes in short term inflation.

The value of investments will fluctuate, which will cause prices to fall as well as rise and you may not get back the original amount you invested. Past performance is not a guide to future performance.

Discuss Article

  1. Shaz Islam says:

    Great article!  What do you think is driving the short term UK linkers into the negative teritory? Do you think it is the premium is worth paying for to protect against inflation?


    Posted on: 09/07/10 | 12:00 am
  2. Jeremy Dufton says:

    For once, I think your conclusion is illogical. If one accepts that RPI will, on average, remain higher than CPI, then surely it is not a question of having 'no choice' but to buy RPI-Linked Linkers, you'd want to get your hands on them as quickly as possible before the supply reduces and you are forced to buy lower-yielding CPI-Linked Linkers?

    Posted on: 09/07/10 | 12:00 am
  3. Stuart Fowler says:

    Help me, please, somebody! I'm not persuaded about this entire story.

    The press commentary on the change to CPI for certain indexation purposes has been really sloppy and I don't think anyone has made a good fist of explaining the maths of the two indices that accounts for a gap having cumulated in one direction or why this can be projected to continue. Is it a structural bias in the index construction or an empirical result of relative movements and weights? Or a combination?

    Two differences are quoted: council tax and interest rates. I imagine the relative inflation of council tax and its weight are not sufficient to account for cumulative outcome differences. I don't know how interest rates are accounted for and the NSO gives no detailed explanation. I assume however that it is the change in rates that comes closest to being a change in living costs. I dare say its weight is also more significant than council tax.

    What I do not understand (and the press and analysts have not addressed it) is why the inclusion of a change in rates would lead to the RPI inflation being higher than CPI.

    Theoretically, the change in rates has no trend over a long enough period and so the rate of change has a mean of zero (whereas the interest rate level has a 'natural' economic mean but the general price level does not). Would this not neutralise its difference relative to CPI except in periods when the cyclical change in rates does have a trend, such as periods when inflation expectations or actual inflation rates are rising or falling?

    If I apply this logic to the last 20 years, when a gap did open up, I would expect the effect to be to dampen inflation because of the falling trend in the change of interest rates.

    Further, I would expect this to be followed at some stage soon by a period in which the effect is either neutral or opposite, ie the mean change in rates is either zero or positive. This would be ironic as it would mean the government's intentions will be frustrated!

    The implication for ILGs is perhaps different. I can understand why market participants who trade the implied inflation rate will want to adjust prices for the short term difference in trend and that trend may be 'more of the same'. I would expect a one-off adjustment that few could exploit.

    But for people hedging or matching 'real' liabilities, as my private wealth management firm does and institutional LDI adopters do, this ILG/conventional bond bet is not of any relevance. In effect, the real yield increase on Friday makes ILGs more attractive than before relative to equities and property as these assets have much looser 'indexation' to the general price level than any indexed product, whatever the base index.

    What am I missing here? Am I alone in thinking the media commentary has been superficial? And is the Government getting off lightly in its explanation of the change?

    Posted on: 11/07/10 | 12:00 am
  4. Stuart Fowler says:

    After time spent analysing this, I can now answer my own question. Any systematic upward bias in RPI has very little to do with housing costs, as widely assumed and reported.

    The NSO's CPI Technical Manual states that the CPI (unlike RPI) uses geometric means and that if it had use arithmetic means since published in 1997 it would have been on average 0.5% pa higher. This is pretty much the incorrectly attributed estimate of the bias from housing costs.

    There are two components of RPI sensitive to house prices: mortgage interest payments (via base number for debt level) and depreciation Limited to buildings value not plot value). Their weight is 10%. If we assumed a long-term increase in house prices relative to general prices of 2% (close to earnings) the upward bias might be 0.2% pa (on basis interest rate changes have mean of zero).

    I don't think there is any conspiracy here. The Government's statements all seem to refer to the appropriateness based on context – eg pensioners don't have mortgages. But the debate should be about which methodology for aggregating the components and chain linking them is fairest for both beneficiaries and taxpayers. The idea that RPI-linked outcomes have somehow been stolen ranks with the popular reaction to the introduction of the Gregorian calendar. 

    I have not been able to confirm my estimate that the net effect of rising debt level and lower rates on RPI has been close to zero over the last 20 years. It is a matter of conjecture whether the medium term will see the reverse offsetting effects, if rates show a cyclical tendency to rise, although it is perfectly possible that both lower house prices and lower borrowing will ensue.

    Posted on: 14/07/10 | 12:00 am

Leave a comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.