Bonds

[GeoffF100]

People talk about "matching" duration - what I was more specifically on about was whether it made sense to match duration to the start of retirement vs the whole/average of expected retirement.

If you are hoping to live off drawdown income, then the answer is "no". That would not match your liabilities at all. If the sole purpose of your investment is to provide a legacy, then perhaps yes. I am pretty much in that situation. I hold an index linked gilt that matures when the ONS says I will pop my clogs, but that is only about 7% of my portfolio. I am ferreting away a proportion of my money in bonds to water down a mighty crash at the wrong time.

[mc2fool]

200 years is available for the US market, but people nearly always choose the last 100 years which is more favourable for equities. We do not have 500 years of stock market data anywhere, but economic growth was much slower before the industrial revolution. Climate change alone will force profound economic changes. People nearly always ignore the markets that had really serious misadventures.

Yeah I was being facetious about 500 years - my point being the longer lens, the better.

Not really, although it depends on what you think it's telling you.

As others have noted, there isn't a lot of really long run data around although the US is one, and here's an inflation (CPI adjusted) Dow Jones Industrial Average back to 1915.

https://www.macrotrends.net/1319/dow-jones-100-year-historical-chart

Sure, from 1915 to now it looks good, but even if we ignore the vertical peak of 1929 we see several multi decade periods when overall the DJIA did nothing. Early 1980s it was at the same level as the late 1930s. Early to mid 1990s it was at the same level as the early 1960s.

It's not a matter of cherry picking periods that were good and periods that were bad but rather of looking and realising that while the 100 years look good overall, there are extended periods of good and bad, and those periods can be very significant real-life periods. After all, there aren't many people that have a 100 year investing lifetime but there are plenty that are retired for 20 to 30.

And, of course, we have no way of knowing if the next 20 or 30 will be a good period or a bad one.

[GeoffF100]

Climate change is a disaster, but it might actually drive positive - or even era changing - market evolution (new tech to combat CC, or space tech to accelerate colonisation beyond earth)

Politicians would like you to believe that. They know that voters cannot handle the truth, and probably cannot handle it themselves. If you do not believe me, look at the EU's balmy plans for a hydrogen economy. The plain truth is that almost nobody is willing to make the sacrifices that would be necessary to achieve net zero, and just some of us doing it will not work. We will choose the alternative course, which is to fry ourselves. You can try investing in fossil fuels and defence, and that might work for a while, but it will not be good in the long run.

[dealtn]

Hi All.

Quick question on duration. Imagine that a given person

Is currently 55
Plans to retire at 60
Expects/hopes to live to 90

In broad terms and all things being equal, to match, do they need a duration of 20 years (75 being the average of 60-90: 75-55=20) either actual or synthetic, e.g. 2/3 30 year gilts, 1/3 cash?

Regards, Newroad

I would imagine a ladder of dates from 5 years to 30+ would be better, and index linked. Not that you would be get much income given the coupons anyway, but I would look to receive and spend the (inflation uplifted) capital to match your predicted expenditure needs.

(That being said, despite buying (and selling) literally £trillions of gilts in my professional career I have never owned one personally, and am invested in equities, property and a single private limited company - a professional football club. You will have a less exciting but safer portfolio whatever you choose).

[Newroad]

Hi dealtn.

Yes, there might be alternate (and perhaps better) ways of achieving the effect - noting as you implicitly have that getting bonds for years 86-90 of the hypothetic example might currently be tricky. However, is the theory correct, i.e. in the example, a duration of 20 years would be the target however achieved?

I think you are saying "yes", but am not 100% sure.

Regards, Newroad

[dealtn]

Hi dealtn.

Yes, there might be alternate (and perhaps better) ways of achieving the effect - noting as you implicitly have that getting bonds for years 86-90 of the hypothetic example might currently be tricky. However, is the theory correct, i.e. in the example, a duration of 20 years would be the target however achieved?

I think you are saying "yes", but am not 100% sure.

Regards, Newroad

A portfolio duration, not a single bond duration (or a cash plus longer gilt barbell). You actually need/want to match your cashflows eg as you spend you receive "income". A duration with unmatched cashflows wouldn't solve.

[Lootman]

If you want a crazy stat, consider that in the 43 years between Sept-1981 to Apr-2024 the stock market has returned 3.5 times the return of long bonds - 11.27% vs 8.07%. Nothing that unexpect about that.. except that entire outperformance has only happened since March 2020, up to which time both were neck-to-neck:

Yes, stocks should return more than bonds over the long run, but it shows you that if you pick favourable/unfavourable startdates for one or the other, the "long run" trend can take 30 or 40 years to exert itself.

Well yes, as I said to Geoff you can demonstrate that almost any investment or asset class outperforms if you pick your dates shrewdly enough. And charting bonds from 1981 is about as close to cherry-picking dates as you can get. (There is even software that will produce the best possible timescale and chart for a given security, and fund managers are fond of using it for their ads).

Even so I struggle to see how a security that by definition cannot grow can be a substitute for a growth security. Look at a long-term chart of the S&P 500 and you notice a few things. Up eight-fold in just over 15 years. Up 20-fold in the last 35 years. Up 32-fold since inception 40 years ago. And that is all without including dividends!

The reason bonds are even being discussed here is as a "filler" for an equity portfolio. Bonds are included in a portfolio to reduce its volatility and provide some income. That is fine. There is evidence that a (say) 80/20 can give a better risk/return profile than 100% in shares. Personally I prefer to use cash as that safety buffer but then cash is just a very short term bond anyway.

Equities are for growth and bonds are for stability and income, the stability coming mostly by sticking to the short end. I think some people here and elsewhere have been seduced by bonds because of the freakish monetary conditions that led to ZIRP and the like. But just as in the 1970s, that infatuation was crushed by inflation.

despite buying (and selling) literally £trillions of gilts in my professional career I have never owned one personally

Well yeah, similar here, I worked in the City for 22 years, including stints on fixed income desks, and I never found much use for bonds personally. A lot of institutions are more or less forced to buy and hold them of course. It is equities that get me up in the morning; bonds just help me sleep.

I am about 90% in shares and 10% in short-term cash or cash substitutes. I never miss a wink of sleep.

[JohnW]

Hi All.

Quick question on duration. Imagine that a given person

Is currently 55
Plans to retire at 60
Expects/hopes to live to 90

In broad terms and all things being equal, to match, do they need a duration of 20 years (75 being the average of 60-90: 75-55=20) either actual or synthetic, e.g. 2/3 30 year gilts, 1/3 cash?

Regards, Newroad

If your example is to spend during all of retirement for 30 years, the spending duration calculation is 5 (years to start) +( (1+2+3....30)/30) = ~20.
The bonds/cash duration matching strategy uses a bond fund, the duration at or above the spending duration (assuming the yield curve slopes up giving longer bonds better returns than shorter). I have a sense that a single 30 year (or 20 year maturity/duration) bond holding instead of a fund would not be as good, but no idea why. Differences include: a fund reinvests dividends a lot easier than you could (thus buying cheap bonds if rates have risen); fund duration stays constant but the 20 year bond duration keeps shortening; I've never read anyone commend one bond. But, the best answer might come from a spreadsheet to model what happens with rate changes. Over to you.

[Newroad]

Thanks, JohnW.

On a purely personal note, whilst the example I gave is similar to my own scenario, it wasn't the same - there was a bit of rounding etc for ease of calculation.

The question underpinning the presumed point I was trying to make was that a significant portion of the above discussion was about matching duration to need. However, it focused more on the various means of achieving a certain duration, rather than how easy it was to determine and calculate the need (though dealtn touched on variation of need due to real life scenarios).

Like you, I suspect not all means of achieving a given duration are equal, without properly understanding why. I further suspect that (to once again use my hypothetic example) it probably ranges on a spectrum from simplicity <-> granularity, e.g. a single issue of a gilt for 20 years hence vs a gilt ladder for years 5 to 35.

I'm guessing that if all things were equal, they would achieve the same or a similar outcome (hence why they have the same duration) but that if circumstances vary, the gilt ladder would be the more able to respond flexibly. Some sort of barbell might be a halfway house - fairly easy to administer but allowing some flexibility.

Regards, Newroad

[dealtn]

Like you, I suspect not all means of achieving a given duration are equal, without properly understanding why. I further suspect that (to once again use my hypothetic example) it probably ranges on a spectrum from simplicity <-> granularity, e.g. a single issue of a gilt for 20 years hence vs a gilt ladder for years 5 to 35.

That's not a ladder. Its a regular (tiny) income with a large balloon at the end. Can you imagine a scenario where your spending looks like that? What you need is a 5y, 6y, 7y, ... 19y, 20y set of bonds, and the capital redemptions fund your spending. The tiny incremental coupons are received but frankly (almost) irrelevant.

[Newroad]

Hi dealtn.

Yes I could, but perhaps I've misexplained leading to confusion.

One would need to perhaps make some inflation assumptions, e.g. 2%, but (and remember, this is only theoretic) buy gilts now, ideally with low coupons, maturing to the value of, say

£10,000 in 2029, e.g. £8,456 worth of TG29
£10,200 in 2030, e.g. £8,081 worth of TG30
£10,404 (£10,400) in 2031, e.g. £7,935 worth of TG31
£10,612 (£10,600) in 2032, e.g. £8,419 worth of TG32
etc*

Any small coupons in the interim could perhaps be rolled up into a gilt for 2060, or popped into one's equity allocation, or something like that.

This would plan to achieve similar purchasing power, from this part of the portfolio, for each year of putative retirement.

Regards, Newroad

* my apology for any calculation errors above - pricing taken from yieldgimp

[dealtn]

Hi dealtn.

Yes I could, but perhaps I've misexplained leading to confusion.

One would need to perhaps make some inflation assumptions, e.g. 2%, but (and remember, this is only theoretic) buy gilts now, ideally with low coupons, maturing to the value of, say

£10,000 in 2029, e.g. £8,456 worth of TG29
£10,200 in 2030, e.g. £8,081 worth of TG30
£10,404 (£10,400) in 2031, e.g. £7,935 worth of TG31
£10,612 (£10,600) in 2032, e.g. £8,419 worth of TG32
etc*

Any small coupons in the interim could perhaps be rolled up into a gilt for 2060, or popped into one's equity allocation, or something like that.

This would plan to achieve similar purchasing power, from this part of the portfolio, for each year of putative retirement.

Regards, Newroad

* my apology for any calculation errors above - pricing taken from yieldgimp

Please quote me, then I get a notification you have replied.

That taking quite some inflation assumption. We have had double digit inflation quite recently. Compounding with those kind of 2% misses can have a huge effect even in the short term, let alone over 20 years. Why not use linkers in the first place?

[Newroad]

Hi dealtn.

Yes I could, but perhaps I've misexplained leading to confusion.

One would need to perhaps make some inflation assumptions, e.g. 2%, but (and remember, this is only theoretic) buy gilts now, ideally with low coupons, maturing to the value of, say

£10,000 in 2029, e.g. £8,456 worth of TG29
£10,200 in 2030, e.g. £8,081 worth of TG30
£10,404 (£10,400) in 2031, e.g. £7,935 worth of TG31
£10,612 (£10,600) in 2032, e.g. £8,419 worth of TG32
etc*

Any small coupons in the interim could perhaps be rolled up into a gilt for 2060, or popped into one's equity allocation, or something like that.

This would plan to achieve similar purchasing power, from this part of the portfolio, for each year of putative retirement.

Regards, Newroad

* my apology for any calculation errors above - pricing taken from yieldgimp

Please quote me, then I get a notification you have replied.

That taking quite some inflation assumption. We have had double digit inflation quite recently. Compounding with those kind of 2% misses can have a huge effect even in the short term, let alone over 20 years. Why not use linkers in the first place?

Hi dealtn.

I find the quoting mechanism on the site cumbersome (to edit) but maybe that's my problem.

I don't know much about linkers and my instinctive feel is that there is an "insurance product" element about them, i.e. you pay a premium for some element of safety. But yes, that may well be a reasonable approach in context too.

However, does it change the maths slightly, e.g. in 2030, would you be aiming for £10,000 worth of index linked gilts maturing, or £10,200 worth?

Regards, Newroad

[mc2fool]

I don't know much about linkers and my instinctive feel is that there is an "insurance product" element about them, i.e. you pay a premium for some element of safety. But yes, that may well be a reasonable approach in context too.

However, does it change the maths slightly, e.g. in 2030, would you be aiming for £10,000 worth of index linked gilts maturing, or £10,200 worth?

As it happens there isn't an index linked gilt maturing in 2030, but if you consider say T29 which matures on 22-03-2029, that is currently on a nominal offer price of 99.35, so if you buy £9,935 worth of it today then in 2029 it will return you an amount of money that will buy you the same goods as £10,000 would today, according to RPI inflation (of course, inflation on your personal basket of goods may be more or less than RPI).

How much that maturity amount will be, in nominal terms, nobody can know 'cos nobody knows what inflation will be over the period, but what you will know is that it'll be "worth", in real terms, the same as £10,000 is today.

https://www.londonstockexchange.com/stock/T29/united-kingdom/company-page

[Newroad]

Thanks, Mc2Fool.

But to be clear and avoid potential misinterpretation, I had assumed that only the coupon varies with linkers, not the return of principal?

If so, if one wants to maintain purchasing power, one presumably either needs to

(1) Save the coupons along the way for later use, and/or
(2) Make a similar inflation assumption that I did earlier?

Regards, Newroad

[dealtn]

Thanks, Mc2Fool.

But to be clear and avoid potential misinterpretation, I had assumed that only the coupon varies with linkers, not the return of principal?

If so, if one wants to maintain purchasing power, one presumably either needs to

(1) Save the coupons along the way for later use, and/or
(2) Make a similar inflation assumption that I did earlier?

Regards, Newroad

The redemption amount is uplifted by inflation too, not just the coupons.

(There is a 2030 inflation Linked Gilt too, maturing 22/7/30 with a 4 1/8% coupon - one of the now rare 8 month lagged old style bonds)

[mc2fool]

Thanks, Mc2Fool.

But to be clear and avoid potential misinterpretation, I had assumed that only the coupon varies with linkers, not the return of principal?

You assume incorrectly. On my phone now and not willing to faff with tiny keyboards and screens but if you look in the gilts and the bonds board you will find plenty of explanations

[Newroad]

Thanks both dealtn and Mc2Fool.

Always good to learn something.

Will help should I ever consider linkers myself. Further, dealtn's earlier point about the suitability of using them in hypothetic ladder makes even more relative sense.

Regards, Newroad

[mc2fool]

(There is a 2030 inflation Linked Gilt too, maturing 22/7/30 with a 4 1/8% coupon - one of the now rare 8 month lagged old style bonds)

So there is. For some reason yieldgimp has decided to no longer list the remaining three old style linkers. (Cue Geoff to tell me shoulda checked Tradeweb... )

Always good to learn something.

Just to expand a little on the info already given, the par value is inflation indexed during the life of the index linked gilt, not just at redemption, and while the coupon rate (the %age) remains fixed, as the coupon payments are calculated as the % rate of the (indexed) par value, as the latter increases the coupon payments also do so to match.

[International]

yieldgimp

Slight aside, but does anyone know why yieldgimp highlights certain rows?