Why you shouldn’t own inflation-linked bonds

This is the third part in a 4-part series on inflation-linked bonds.

We saw in the first post of the series, which went through an introduction to inflation-linked bonds, that they’re government-backed bonds whose value rises in line with RPI.

They also tend to be priced by the market in such as way that the real returns from owning inflation-linked bonds should be the same as the real returns from nominal bonds.

The second post of the series went through a few reasons why investors might want to buy inflation-linked bonds for their own portfolio. These include:

1. For short term unexpected inflation protection,
2. To fund their spending,
3. For their asymmetric return payoff,
4. As a diversifier.

So far inflation linked bonds are looking pretty good.

But now let’s have a look at some drawbacks.

Why you shouldn’t own inflation-linked bonds

As I see it, there are seven reasons you might want to avoid inflation linked bonds:

1. Fund availability
2. RPI relevance
3. Using any single measure of inflation isn’t ideal
   1. Your personal inflation rate
   2. Other problems with single inflation measures
4. Liquidity
5. Lower yields
6. Do you need to hedge short term unexpected inflation?
7. Price volatility

So let’s dive right in.

1. Fund availability

We saw in the last post that in order for us to properly hedge unexpected inflation, we’re best off using short-dated linkers. And for 99% people, we’ll be getting our exposure to inflation linked bonds through funds, rather than faffing around with constructing bond ladders with individual bonds (although this can be a good option for those with the time/inclination).

The problem with this is you simply can’t buy UK short-dated inflation linked bond funds.

If I’ve missed any off the list please let me know, but here’s what our menu of passive UK inflation linked gilt funds looks like:

The problem with all these funds are the numbers in the ‘Duration’ column.

I’ve covered duration in previous posts, but as a quick recap, duration is a measure how sensitive the fund is to interest rate moves. The higher the number, the more the fund will fall when rates rise, and vice versa. And 21-22 years is just about as high as duration gets for bond funds – all these funds are hugely sensitive to interest rates.

The reason the funds have such high duration is they’re filled with bonds which have long maturities. Most bonds in the funds mature in 20-30 years time.

This high duration means any changes in interest rates during periods of unexpected inflation could very easily wipe out any inflation protection the fund would be able to offer. You’re not going to be particularly happy if you’re hedging short term inflation running at 4%, but then interest rates are hiked by 1% which, for a bond fund with a duration 20, roughly equates to a 20% fall in the bond fund’s value.

We saw in both the Vanguard and the Meketa research in the section above how much worse long-dated inflation linked bonds were at protecting against inflation. And they were only using 10-year duration bonds – these UK funds have twice that!

One way to get around the duration issue is to look outside the UK.

If we broaden our selection universe to global inflation-linked bond funds, then we can shorten our duration quite a bit. Of course, we’d need to ensure they’re hedged to sterling to make sure we’re not taking any currency risk.

Here’s our list of passive global inflation linked government bond funds, hedged to GBP:

Those knock about 10 years off the durations of the UK gilt linkers. But that improved inflation protection in the form of lower duration comes at a price.

If you want to buy global linkers to reduce your duration, you sacrifice the ability to hedge against the UK’s RPI. By buying global bonds, instead of hedging any RPI inflation (as you would if you bought an inflation linked gilt fund), you’d be hedging the inflation rates of all the countries the fund holds government bonds in.

For example, the iShares global inflation linked government bond fund has just under half of its portfolio in the US, and only 29% in UK inflation linked gilts. But I, as a UK investor, don’t care what inflation in the US is doing. I care about what inflation in the UK is doing.

The Legal & General global fund is even more extreme. Despite having the lowest duration (which is good), it also holds absolutely no UK gilts whatsoever. The top three countries it holds inflation linked government bonds of are the US (65%), France (11%), and Italy (8%). As a result, it has quite different performance to the other two funds:

Source: Bloomberg

XGIG has a longer history to compare the L&G fund to:

Source: Bloomberg

The L&G index fund has a different return profile to our two ETFs, due to its shorter duration and geographic exposure differences.

But regardless of whether you look at the L&G fund or one of the other two, by buying a global inflation linked bond fund, you’re tying your protection to a less relevant measure of inflation – most notably the rate of inflation in the US.

So let’s see how similar US inflation is to the UK’s.

According to the ONS, the United States’ lead measure of inflation (the Consumer Price Index) is comparable with UK CPIH, as both include owner occupied housing costs. So let’s have a look at those two, and see how similar they are. I’ll also chuck in RPI, so we can see how similar US inflation will be to the measure of inflation used for our UK inflation-linked bond holdings.

(The UK’s CPIH measure didn’t come along till 1989, so the data series starts a bit later than the other two, which have data going back to the late 1940s.)

Sources: UK inflation: ONS. US inflation: FRED.

Wow.

I knew inflation was bad in the late 70s, but 25% a year in the UK is nuts.

In terms of correlations, it’s going to be difficult to say how well global inflation-linked bonds are going to be able to protect you if there’s a spike in inflation in the UK.

On the one hand, US and UK inflation rates seem to be quite highly correlated, especially since the early 1980s where inflation has been sub-5%. Correlations in the last 40 years have been pretty good.

On the other hand, you want your inflation-linked bonds to protect you when there’s a huge spike in in inflation. That’s when they really shine. And if you’d held global inflation-linked bonds in the 1970s during the enormous rise in UK inflation (they didn’t exist then), they would’ve given you pretty poor protection compared to UK linkers.

It’s difficult to say whether a 1970s-style dislocation between UK and US inflation will happen again.

It might be that our more interconnected global economy means we’re in a different position to where we were in the 70s. Maybe the increased reliance on international trade means inflation levels between countries are likely to be more similar than in the 70s, and so such a big difference in inflation between the UK and the US is less likely. I don’t know.

It might be that global central banks have learned from history, and become better at managing inflation, meaning we’re unlikely to see such high inflation rates again, and therefore such high differences in inflation. Again, I don’t know.

Given the choice between the two, with all else equal, I’d prefer to take UK linkers. It’s always going to be more relevant to my inflation rate than a global fund.

But as we’ve seen, the problem with UK linkers is their huge duration risk.

So could I be persuaded that global linker funds deserved a place in portfolios, to reduce the duration risk of UK linkers? I think so.

A 1970s double-whammy inflation event of ultra-high inflation and very different inflation to the US seems unlikely. And even it did occur (we should always plan for the unexpected in our portfolios), then a) we’ll still be holding inflation-linked bonds, which will perform extremely well in such a highly inflationary environment, and b) as long we’re still holding a good portion of our linkers in RPI-linked funds then we’ll still be taking the worst of the sting out of the UK’s relatively higher inflation.

And for those who are extremely confident in the correlation between US and UK inflation, it’s actually possible to buy a sterling hedged version of the iShares $ TIPS 0-5 year ETF – with a duration of a mere 2.5 years.

2. RPI relevance

The second reason you might want to think twice about chucking a whole bunch of inflation-linked bonds into your portfolio is all do with the measure of inflation they use.

The adjustment to the principal component of the inflation-linked bonds is determined by the level of RPI in the UK.

RPI is calculated based on the price increases across a ‘basket of goods’ – i.e. a cross section of goods and services which most people tend to spend money on. There’s lots of statistics and surveys involved in the collection of the data, which try and gather an accurate picture as possible for what we all spend money on.

And if we’re buying something to protect us against inflation, we need to be sure that how that product measures inflation is accurate. Otherwise we’re not really protected at all.

But it turns out the RPI is a pretty poor measure of inflation in the UK.

The Office of National Statistics, who should know a thing or two about how things are measured, have this to say about the RPI:

“Overall, RPI is a very poor measure of general inflation, at times greatly overestimating and at other times underestimating changes in prices and how these changes are experienced.

In 2013, the RPI lost its status as a National Statistic. Our position on the RPI is clear: we do not think it is a good measure of inflation and discourage its use.”

That’s a pretty big slam from the ONS. They really don’t like RPI.

Digging into the detail a bit further, they dislike it for a few reasons:

• The use of something called the ‘Carli formula’, which is a part of the RPI calculation. As much as I care about inflation measures, I don’t have the willpower to dive into the detail on the Carli formula. I’ll take the ONS’ word that it’s bad.
• Another drawback of RPI as an inflation measure is the treatment of housing costs for home-owners, technically known as owner occupiers’ housing costs (OOH). The RPI uses a combination of mortgage interest payments and also house prices as a proxy for housing depreciation, which means that the RPI is more heavily influenced by house prices and interest rates.
• Finally, there are other important issues related to the coverage of RPI, including the goods and households included and the way in which these are weighted together

The ONS conclude: “For the reasons set out in this document, RPI does not have the potential to become a good measure of inflation.” They’re now “developing a set of measures that better captures the changing prices and costs faced by consumers and households and… discouraging the use of RPI, while recognising the legacy needs”.

As a result of its drawbacks, RPI is likely to overstate inflation. Currently it overstates inflation compared to CPIH by around 1% – although this difference versus CPIH varies over time. The chart below does a good job at showcasing the differences between RPI and CPIH inflation rates over the last 15 years:

Source: Office for National Statistics

So nobody likes the RPI, and it seems like the only real reason it’s still going is because inflation-linked bonds use it. Although luckily for bond holders, they’ll be getting an extra uplift in their principal adjustment because RPI is currently overstating inflation. So that’s a plus.

But in addition to its statistical drawbacks, RPI causes another problem for inflation linked bond holders.

3. Using any single measure of inflation isn’t ideal

Although RPI has its own individual shortcomings, there’s always going to be drawbacks when you’re relying on any aggregated single measure of inflation.

a. RPI won’t measure your personal inflation rate

The first problem with a single measure of inflation is we all spend money on different things.

None of us are going to spend our cash on the exact basket of goods, whether that’s measured by the RPI, CPI, CPIH, or any other single measure of aggregate inflation. Some of us will have spending that’s slightly different, some will have spending that’s wildly different.

And that means we’ll all have different ‘personal inflation rates’.

If I spend lots of money on a new car every year, my personal inflation rate will be heavily influenced by car price inflation. If you live in London and don’t own a car, but instead spend lots of money on foreign holidays, then your personal inflation rate won’t have anything to do with cars, but will be dependent on the rising costs of holidays. We’re going to have very different levels of personal inflation – and both of which will be different to the RPI.

Your personal inflation rate is going to be affected by a whole number of factors, including your age, lifestyle, income, employment status, where you live, and how many children you have.

To demonstrate, here’s the RPI basket for 2021:

Source: Office for National Statistics

As an example for why this might not be a good measure of someone’s personal inflation rate, we can see that housing is the biggest RPI component. But housing matters far less for retired investors. The largest components of the ‘housing’ category are depreciation and rent – neither of which retirees are likely to care about, given they’re more likely to own their own home, and less likely to upsize in the future. Motoring is the second largest RPI component, yet only around half of London households own a car. So for a retiree living in London, about 40% of the RPI basket is totally irrelevant.

This means owning inflation linked bonds which track the RPI aren’t going to be a great proxy for our fictional retiree’s spending. They could get lucky and see huge booms in the property/car market – in which case they’re quids-in, as they have the benefit of an uplift in the value of their inflation-linked bonds without having to bear the cost of spending more on housing/cars. But on the flip-side, they could experience a rise in the cost of holidays, travel costs, and catering (which I’m guessing retirees spend more on) during a period of slowdown for the housing market (a big part of RPI). In which case, they’re likely to be out-of-pocket, as their inflation-linked bonds wouldn’t do a good job in keeping up with their spending.

Overall, although the RPI has its own flaws and seems to be on the way out, whether you use RPI or CPI, you’ll never get a perfect match for your own inflation rate.

Inflation-linked bonds will always provide some protection against inflation even if your basket looks quite different to that used in RPI, and that could very well be better than no protection at all. But given investors pay a premium for the privilege of inflation protection, it’s up them to decide whether the level of inflation protection is likely to be worth it, based on how closely their spending aligns with the basket of goods used in the inflation calculations.

b. Other drawbacks of relying on single measures of inflation

There are a couple of other problems when using a single aggregate measure of inflation.

i. Product quality

Inflation metrics don’t account for changes in product quality, which can increase the value of goods without raising their prices.

Technology is a good example here. If the cost of buying a computer stays the same from one year to the next, it might not have any impact on the RPI, even if the new computer is much better than your last one.

For example, I paid about £1,000 for my laptop in 2015. As great as it continues to be, if I paid £1,000 for a new laptop today it’d be faster, thinner, lighter, higher resolution, and have a battery life longer than ten seconds.

ii. Substitutes

When the price of something in the basket of goods increases, this feeds through into a higher measure of inflation. But that price rise might mean consumers start to buy less of it in favour of some cheaper alternative.

If the price of my contact lenses starts rising, I might switch to wearing glasses. In which case, RPI will start to overstate the true level of inflation that I (and anyone else who switches) will face.

So RPI is a bad measure of inflation, and inflation isn’t a good measure of our personal inflation rate.

Personally I’m less fussed about these sorts of arguments against inflation linked bonds.

We can never expect any single measure of inflation to perfectly match our own spending, and those in retirement who are most vulnerable to inflation are likely to be glad of having some inflation protection rather than none at all. And there are some pretty smart people working hard to make sure the inflation basket is as close to representative as possible.

The inflation protection will never be perfect, but we can’t possibly expect that from an instrument designed for everyone.

4. Liquidity

The fourth problem with inflation-linked bonds is their liquidity.

Inflation linked bonds have a reputation for being less liquid than their nominal counterparts.

To see whether this was actually true (and not just one of those investing old wives’ tales), I ran some back-of-the-envelope numbers to check the amounts traded on the inflation-linked bond ETFs compared to nominal bond ETFs.

Out of the two UK inflation-linked bond ETFs (INXG and GILI), INXG was much more liquid, trading at roughly £55m per day in 2021 vs GILI at £4m per day. Out of the two global inflation-linked bond ETFs (GILG and XGIG), XGIG was much more liquid, at £13m per day versus just £600k per day for GILG.

Comparing those liquidity stats to their nominal ETF counterparts, IGLT traded roughly £70m per day in 2021, and IGLH traded £15m per day.

So the most liquid UK linker ETF doesn’t trade much less than the UK nominal (£55m vs £70m), and the most liquid global linker ETF doesn’t trade much less than the global nominal (£13m vs £15m).

Although these figures might seem low when you compare them to the AUM of the funds, bear in mind these are the liquidity figures for the most relevant share class to us. If we want to sell our holding of the GBP hedged global inflation-linked bond ETF, we don’t really care who’s buying the unhedged USD class.

And also bear in mind this is just liquidity on the secondary market.

The advantage of the ETF wrapper is that as long as there’s liquidity in the underlying securities, it doesn’t matter if there’s not enough liquidity in the ETF. We’re getting a bit technical here (and I’ll be going through exactly how ETFs work in a later post), but I’ll give you the gist of it.

If demand for an ETF exceeds supply, then the price of the ETF is likely to rise above the price of the basket of securities the ETF is tracking. Incentivised by the opportunity for arbitrage, someone called the ‘authorised participant’ will buy the ETF’s underlying securities (in this case, the bonds themselves), and will exchange them with the ETF provider for shares of the ETF. The ETF provider then sells these new ETF shares on the stock exchange, creating liquidity for those people who were wanting to buy, and bringing the price of the ETF shares back in line with the price of the underlying basket of securities.

It’s a bit complicated, but the point is even if an ETF doesn’t have much AUM, or doesn’t look particularly liquid, then as long as the underlying securities of the ETF have sufficient liquidity, then this shouldn’t be a problem for investors.

The point is that inflation-linked bond ETFs not only look like they have similar liquidity to their nominal counterparts, they also have the added liquidity of the underlying bond market if trading on the ETF becomes strained during times when liquidity is tough to come by (like in a market crash).

So although I’m sure the liquidity of underlying nominal government bonds is greater than the underlying liquidity of inflation-linked bonds, unless investors suffer a perfect storm of a lack of liquidity in the linker ETF AND a lack of liquidity in the underlying inflation-linked bonds, then they should be OK from a liquidity perspective.

I’m not saying it can’t happen, and, in fact, the time it’s most likely to happen is when you need liquidity most – during a crash. But by holding your linkers in an ETF wrapper, you’re adding an extra layer of liquidity which reduces your illiquidity risk.

5. Lower yields

What I’m talking about here isn’t the idea that inflation-linked bond funds have lower yields than nominal bonds when looking at their factsheets. We know from the first post that inflation-linked bonds have lower yields because they have inflation already deducted from their yield figure, whereas nominal bonds don’t.

But what we also saw in the first post was that big ol’ pension funds were dumping huge wads of cash into the inflation-linked bond market. And because we don’t know exactly how much cash these massive pension funds are ploughing into inflation-linked bonds and depressing yields, we don’t know how much the breakeven inflation rate is artificially inflated.

We also saw that investors, in general, are willing to pay a premium to access the inflation protection which linkers provide. But we can’t put a value on exactly how much of a premium investors are willing to pay for this protection, so again this makes it difficult to figure out how this is affecting their price.

We can be sure both these factors are depressing yields, but because we don’t know by how much, we can’t be sure that buying inflation linked bonds represents good value compared to their nominal counterparts.

All you can do is eyeball the spread between the two and have a decent guess.

6. Do you need to hedge short term unexpected inflation?

We saw in the second post that those in the accumulation phase don’t need to worry so much about unexpected short-term inflation, as their spending needs are covered by their employment income, and not their portfolio.

As you move from the accumulation phase of building wealth (i.e. you’re growing your portfolio) to the decumulation phase (you’re retired and using your portfolio to fund your lifestyle), the role of bonds in your portfolio changes.

As an accumulator, you’re counting on bonds to reduce your portfolio’s risk, and help mitigate the effects of stock market crashes. While that’s still true as you move into the decumulation phase, you’re now also relying on your bonds to meet your spending needs.

This means accumulators, while they can still benefit from the diversification properties of inflation linked bonds, derive much less benefit from inflation linked bonds’ ability to match inflation.

Accumulators, then, are less likely to want to add inflation-linked bonds to their portfolio.

7. Price volatility

The final reason we might want to avoid inflation-linked bonds is because they have a reputation for being more volatile than nominal bonds.

Again, I wanted to run some numbers to check whether there was any truth to that, or whether it was just another old wives’ tale from the inflation-linked bond haters.

Backward-looking historical performance analysis unfortunately doesn’t shed as much light as I’d like on this puzzle, because the last periods of damaging unexpected inflation in the 1970s predate when inflation-linked bonds existed.

Still, we’ve got a track record going back to the late 90s, which gives us a bit of data to work with.

First up, let’s have a look at how UK inflation-linked bonds compare to UK nominal bonds in terms of volatility:

The table shows UK inflation-linked bonds don’t appear any riskier than nominals of equivalent duration. Volatilities are similar (with the linkers actually having slightly lower risk), and max drawdowns are also basically the same (again, with linkers having the slight advantage).

Where nominals do seem to be safer is when comparing an aggregate government linker fund (containing all maturities) against an aggregate gilt fund. In this instance, nominal gilts seem much safer. But that’s really a function of the nominal fund having much lower duration – 12 years versus 21 for the linkers.

The drawdown chart below shows this pretty clearly. The low-duration nominal gilt fund in brown seems to fare much better (in general) than the longer-duration linker/nominal funds:

Source: Bloomberg

When choosing between the nominal vs linkers of equivalent duration, they seem to be roughly similar in terms of drawdowns. Sometimes the nominal has the lower drawdown (2008), sometimes the linker has the lower drawdown (2017/2020).

So really there doesn’t seem to be much to choose between the volatilities of UK linkers compared to UK nominals. Interestingly it looks like there may be some diversification benefit to owning both, given their drawdowns don’t seem to coincide particularly strongly.

And when comparing global linkers versus global nominals, the picture is similar.

It’s difficult to do a like-for-like comparison for global linkers vs global nominals, as I couldn’t find any indices with equivalent durations. But looking at the closest comparison we do have shows the volatilities seem to be similar when scaled for duration:

However, where nominals do have the edge here is in max drawdown. The World Government Bond Index has a significantly lower maximum drawdown than the World Inflation-Linked Index, even if you scaled up the drawdown a bit to account for the nominals’ lower duration.

When looking at a drawdown chart, the difference between the global inflation-linked vs global nominal becomes clear – the nominals seem to be the preferred choice:

Source: Bloomberg

Really when it comes to volatilities, duration seems much more important than whether you’re opting for an inflation-linked bond fund or a nominal bond fund.

When push comes to shove, global nominals have an edge over global inflation-linked, but the volatility differences between UK linkers and nominals are marginal, when comparing funds with similar durations.

Summary

OK, that was a lot of information.

To summarise, by my count there are seven accusations which could be levelled against inflation-linked bonds to argue against including them in your portfolio:

1. Fund availability – UK linkers have a high duration risk, and global linkers are less relevant for the UK inflation rate.
2. RPI relevance – the stats nerds at the ONS think RPI overstates inflation, and its use is being phased out.
3. A single measure of inflation – RPI:
   1. 1. Won’t match your your personal inflation rate,
      2. Doesn’t account for inflation effects in product quality, or the subsitution effect.
4. Liquidity – index-linked bond funds trade slightly less than their nominal counterparts, but not by much. They’re also wrapped in an ETF vehicle which adds another layer of liquidity on the secondary market.
5. Lower yields – we can’t put a value on the premium investors are willing to pay for inflation protection, or how distorted the market is from pension fund allocations.
6. The need – accumulators have a much lower need to be hedge short-term unexpected inflation.
7. Volatility – global nominals are slightly less volatile than global linkers, but the main factor affecting bond volality is duration, and not whether the bond is inflation-linked or not.

Some of these we’ve seen are valid criticisms, some not so much. And just because there are 7 potential reasons not to own them and only 4 in their favour doesn’t mean inflation-linked bonds aren’t worth owning.

There are compelling arguments on both sides of whether to own them or not, and next week I’ll try and reconcile both sides to come to a conclusion.

The next post in the series will tie everything together by recapping everything I’ve gone through so far, and answering the ultimate questions of whether inflation-linked bonds are useful to own, and if so, how much of your portfolio you might want to allocate to them.