Bonds and Misery

OldScientist

masonic said:

Very interesting and useful, thanks. I am starting to pull together some data to explore the relationship between "best buy" savings rates and bank rate / inflation, although I think I'll be limited to going back as far as 1995, with the best data coming from ~2005 onward. But your analysis above clearly challenges the assumption that long-term cash holdings do not hold their value due to inflation.

Here are a few more sources that I've found useful (I finally located where I'd put them)...

BSA Yearbook (https://bsa.lansdownepublishing.com/ (I used the 2021/22 version, but the 2023/24 edition also has interest rates on building society savings accounts going back to 1939).

swanlowpark has data going back to 1980 (note: the certificate associated with the following website shows up with an error on the safari browser but I have used the site without (AFAIK) any problems, https://www.swanlowpark.co.uk/savings-interest-annual - the sources are linked from this page )

The Barclays equity gilt study (if you can find a copy) has Building Society rates back to 1946 (although, in the early years, they are not the same as those in the BSA Yearbook).

I have had a trawl through the MSE archive and managed to get easy access and 1 year fixed rate accounts between 2008 and 2021 (end of each year, either in November or December). Over this period I found on average the 'best buy' easy access account exceeded the 3 month bill rate by 1.5 percentage points and the 1 year fixed rate account gave another 0.5 percentage point. This was a fairly unusual period (generally low rates), so I have no idea whether it would continue going forward (certainly 'best buys' were sometimes awkward to obtain before online banking)

For cash and inflation, it is worth noting that in the period where all fixed income did badly in terms of real returns (the 30 year rolling periods starting between about 1930 and 1950) it was generally the shorter maturities (bills and 2 year gilts) that did least badly. In the 1970s, i.e., towards the end of that set of 30 year periods, inflation rates increased rapidly and yields also increased leading to large decreases in prices for gilts with longer maturities (fairly similar to what we have seen recently except inflation rates have, so far, dropped more rapidly this time compared to the 1970s and 80s).

masonic

Thanks, yes I'm working through the MSE archives. I've already been able to get "data" for best buy 1 year fixed rates from Paul Lewis' study by extracting coordinates from his plots (there are some pretty nice tools available for this), since the tabular data is not available. This covers the period from 1995 to 2015. Over that period, the 5 year annualised returns were about +0.5% for the late 1990s, rising to +1% for the early 2000s and +2% for the late 2000s to early 2010s. Obviously the base rate went so low during those later years that a fair premium was to be expected. Perhaps 0.5-1% is more normal. Once I've pulled the rest of the data together I'll share a summary and plots.

masonic

I've now pulled all of that data together and generated some 5 year and 10 year annualised returns from it. I think the following plot is quite interesting:

First of all, the blue and orange lines show BoE base rate and SONIA rate (from its introduction in 1997). These track reasonably closely, but SONIA does lag at times. 

Above this in green, you will see returns from taking out the market leading 1 year fix (and reinvesting at maturity into the market leader at that time, etc). This is a composite series using the original data from Paul Lewis' blog, and when data become available (~July 2008 onwards), the MSE top pick from the weekly newsletter. This has maintained a reasonable margin on base rate throughout the ~30 years, except for the recent period when base rate shot up. Even then, it does not drop below the returns you'd get on base rate.

Data on the best easy access rates was not available in the MSE weekly newsletter until just before 2008, but from that date it seems to track 10-30 basis points below the 1 year fix line. It is probably fair to assume this was the case in earlier years too.

Just for interest, I've added the IA Mixed Bond sector average from Trustnet in purple and the Vanguard Global Bond Index fund in brown. I'm not aware of a hedged aggregate bond fund available prior to 2009, so I'm hoping the sector average (which is not going to be fully sterling hedged) can be used as a guide.

We can smooth things out a bit by looking at 10 year annualised returns (and remove some redundant series):

Now one sees that the mixed bonds are more attractive on a 10 year rolling returns basis than the market leading fixed term savings accounts, including the recent period. But this is not the case for other types of bond fund. If you compare with the IA gilt and IA index linked gilt sector average, which will mirror an index tracker fund fairly well, the increased duration of these funds led to quite an unimpressive period towards the end.

I couldn't resist adding in the IA Global sector for comparison (Vanguard Global All-Cap tracks it fairly closely). I think it is quite telling of the irrational period we were in for bonds that the index linked gilts kept up with equities for so long. But all such anomalies must come to an end.

OldScientist

@masonic

Interesting stuff. Since the SONIA rate, bank rate, and 3 month bill rates were similar over this period, it is interesting to note that someone who chased rates on 1 year fixed rate savings accounts would have done better than any of these (I note that the 3 month bill rate is usually used a a proxy for cash in historical backtesting) by upwards of 50 bp per year for 5 year periods and more than 1 percentage point over 10 year periods. While this would not necessarily have been the case in the past (if only because chasing rates was so much more difficult and a lack of competition in the absence of challenger banks, e.g. the BSA 'suggested' rates for building societies), it does indicate that more optimistic results would arise from backtesting using a higher rate for cash than the 3 month bill rate.

I calculated the effect on historical SWR of using an estimate of the rates associated with easy access savings accounts (prompted by this thread, I've finally found all of the work I did on this from a few years ago). For equity allocations of 60%, the SWR was 2.9% for bills or long bonds and 3.1% for 'savings accounts' (historical equity, bill, and bond returns and cpi from macrohistory). This would be increased further with the 1 year accounts that you've derived returns for (my estimate would be at another 20-30 bp in SWR based on a 1 percentage point increase in return).

masonic

It does make sense that consumer savings rates would have some sort of premium over short-dated bonds, given that fixed term savings are often lent out at premium rates, and there seems to be an endless procession of savings institutions seeking to gain market share by cutting their margins to the bone (if not offering a loss leader). But as you say, some of that fierce competition has developed during the anomalous period following the banking crisis. The differential for 5 year periods starting after about 2004, or 10 year periods starting after about 1999, will not be relevant to today as mortgage and bank loan rates had an unusually high premium over the historically low base rate.

Now we've emerged from that period, we shall have to see where rates settle to judge if the 5% available today look as good as would have been expected in the past. Clearly over the past year or two, parking the money in a short-term money market fund worked very well, but now this is on a decline, it is likely the medium duration aggregate bond fund gains a premium over STMMFs. Where it ends up relative to an active fixed savings strategy will be intriguing.

BlisteringBarnacles

thunderroad88 said:

25 years of cash is way too much. I’d immediately reduce that to 10 years, putting it into fixed term bank deposit bonds of varying duration and something like RL mmfs. That will see you through any equity market downturn. The other 15 years I’d put into equities. 

Thanks thunderroad : You may have seen this article (linked from Bogleheads) - a classic by William Bernstein. His advice is to hold 20 to 25 years of expenses in "safe" investments. Basically his point is : If you have won the game, no need to play.

https://www.whitecoatinvestor.com/bernstein-says-stop-when-you-win-the-game/

Also here is a great thread on bonds. In particular the reply from valuethinker.

https://www.bogleheads.org/forum/viewtopic.php?t=409243

Another classic : Dont bank on equity risk premium

https://blogs.cfainstitute.org/investor/2024/04/01/dont-bank-on-the-equity-risk-premium/

I dont need too much equities. £2.7 million portfolio. Right now £1.2 million in equities (world tracker). Around 45%. Happy to go to 50%. (£1.4 million). Dont own a home. Still renting. Like to be a global nomad hence did not want the commitment of property and maintenance etc. Single, no kids. Wanted a portfolio of liquid assets and live few months in UK, Europe, India etc. Cost of living has jumped up in UK last year. Used to get by on £20K per year including rent, mainly because employer used to pay for subsidised good quality lunch, free gourmet coffee, private healthcare etc. Gotta pay for these myself now. Easily £40K per year expenses. To live well, I would say £60K per year. Cant have more than £1.6 million in equities because if equities crash 50%, portfolio drops to £1.9 million and at 3% withdrawal, I can still afford almost £60K per year. Why would I need more equities ? My life wont change if my portfolio goes from 2.7 to 4 million. But it will change noticeably for the worse if it drops from 2.7 mill to say 1 million.

That means I need another asset class that provides a positive return and not highly correlated with equities. Thought that was govt bonds. Hmmm... didnt work out in 2022. I was going to keep around 5 years (300K) in cash / ultrashort bonds and plonk the rest 800K or so into VAGS (Vanguard aggregate global bonds) or maybe some into Vanguard global short term bonds index fund.

Not sure how else to invest the non equity part conservatively. Thanks

SnowMan

Some interesting data above on historical best buy savings accounts from masonic.

I've noticed over many years that 5 year conventional gilts have been priced to achieve a gross redemption yield that is typically about 0.5% pa below that on 5 year fixed rate savings accounts (5 years being the longest term for which fixed rate savings accounts are typically available to do the comparison)

That's suggested to me that very broadly speaking (and at the serious risk of over-generalising) over the long term investing in gilts is going to lose you money over saving in best buy savings accounts by about 0.5%pa. So an equity cash mix is arguably better than an equity/gilt mix. In practice I would prefer to suggest it's OK to just use savings rather than gilts rather than saying savings are superior. This involves many assumptions, in particular that the yields on 5 year gilts reflect the difference in interest available in savings accounts compared with gilts over longer terms than 5 years, and that 0.5%pa is that differential on average.

At the moment gilts with 5 years to maturity are priced at about 4.13%pa gross redemption yield. And best buy 5 year fixed rate savings accounts are 4.4%pa (non ISA) and 4.15pa (ISA) so the 0.5% differential has narrowed over than I've observed in the past.

I've mentioned this differential between gilt yields and savings rates before in the past, and I've not seen convincing counter-arguments (to me at least) from the 'it's essential to use gilts rather than savings' thinkers. Leaving aside arguments that it's difficult to access best buy savings as 'my money is in a pension' arguments, it is suggested that gilts provide a magical dampening affect to fluctuating equity returns that savings don't provide. Personally I think that argument is overstated and the dampening affect is quite small, and if you are investing in short term gilts equivalent in term to the fixed rate accounts available, it's hard to argue for the dampening affect.

Just wondering what the thoughts of masonic and OldScientist and others are on this. Does the data back up my observed differential? And how would investing in 10 year maturity gilts (rather than 5) and then reinvested at maturity in another 10 year gilt change the comparison based on historical data? And how would best buy 1 year fixed rate or easy access, vs 5 year fixed rate savings alter the historic comparison?  

For disclosure of bias, leaving aside some index-linked gilts I've recently bought, I have an equity/savings mix and have never owned gilts or gilt funds etc.     

InvesterJones

If you're already won then I'd guess the main goal is inflation matching - I'm not au fait with index linked bonds so can't comment on their suitability, but a rolling set/ladder of short term bonds might provide enough of a return and since you hold to maturity, wouldn't have been affected by changes in interest rates - just have the pain of inflation to worry about if it rises a lot before you get your next set of maturities/purchases. Another option would be some kind of commodity that's not correlated with equities.. but finding one that isn't also used as a speculation vehicle is quite tricky.

OldScientist

BlisteringBarnacles said:

thunderroad88 said:

25 years of cash is way too much. I’d immediately reduce that to 10 years, putting it into fixed term bank deposit bonds of varying duration and something like RL mmfs. That will see you through any equity market downturn. The other 15 years I’d put into equities. 

Thanks thunderroad : You may have seen this article (linked from Bogleheads) - a classic by William Bernstein. His advice is to hold 20 to 25 years of expenses in "safe" investments. Basically his point is : If you have won the game, no need to play.

https://www.whitecoatinvestor.com/bernstein-says-stop-when-you-win-the-game/

Also here is a great thread on bonds. In particular the reply from valuethinker.

https://www.bogleheads.org/forum/viewtopic.php?t=409243

Another classic : Dont bank on equity risk premium

https://blogs.cfainstitute.org/investor/2024/04/01/dont-bank-on-the-equity-risk-premium/

I dont need too much equities. £2.7 million portfolio. Right now £1.2 million in equities (world tracker). Around 45%. Happy to go to 50%. (£1.4 million). Dont own a home. Still renting. Like to be a global nomad hence did not want the commitment of property and maintenance etc. Single, no kids. Wanted a portfolio of liquid assets and live few months in UK, Europe, India etc. Cost of living has jumped up in UK last year. Used to get by on £20K per year including rent, mainly because employer used to pay for subsidised good quality lunch, free gourmet coffee, private healthcare etc. Gotta pay for these myself now. Easily £40K per year expenses. To live well, I would say £60K per year. Cant have more than £1.6 million in equities because if equities crash 50%, portfolio drops to £1.9 million and at 3% withdrawal, I can still afford almost £60K per year. Why would I need more equities ? My life wont change if my portfolio goes from 2.7 to 4 million. But it will change noticeably for the worse if it drops from 2.7 mill to say 1 million.

That means I need another asset class that provides a positive return and not highly correlated with equities. Thought that was govt bonds. Hmmm... didnt work out in 2022. I was going to keep around 5 years (300K) in cash / ultrashort bonds and plonk the rest 800K or so into VAGS (Vanguard aggregate global bonds) or maybe some into Vanguard global short term bonds index fund.

Not sure how else to invest the non equity part conservatively. Thanks

One way of providing a inflation protected income is to construct a collapsing ladder of maturing inflation linked gilts (ILG). This essentially means that in each 'rung' of the ladder (roughly annual) you invest in enough ILG that when they mature they provide sufficient income (coupons also need to be taken into account).

Depending on how far away you are from retirement, there are several ways that this could be constructed. But since you already have enough stored to construct a ladder it could be built immediately.

Helpfully, there is a useful utility at https://lategenxer.streamlit.app/Gilt_Ladder that calculates this for you.

For example, assuming you will retire in 2029 and you will need income for approximately 45 years from then (assuming you would be about 55 years old in 2029, that would take you to aged 100). A ladder providing an inflation protected £40k per year would currently cost about £1.4m. I note that capital gains tax is not currently levied on individual gilts held in GIA although the coupon interest is subject to tax.

The remaining income would come from your equity portfolio and, once available, your state pension. The amount invested in the ladder could be more or less than given in the example and would provide proportionately more or less income. While there is no market risk and minor inflation risk (e.g. after 2062 gilts that maturing every year are not currently available), the gilt ladder does suffer from default risk (i.e., should the UK government ever default on debt).

I also note that for early retirement, the withdrawal rate for a conventional portfolio may be much less than 3%.

OldScientist

SnowMan said:

Some interesting data above on historical best buy savings accounts from masonic.

I've noticed over many years that 5 year conventional gilts have been priced to achieve a gross redemption yield that is typically about 0.5% pa below that on 5 year fixed rate savings accounts (5 years being the longest term for which fixed rate savings accounts are typically available to do the comparison)

That's suggested to me that very broadly speaking (and at the serious risk of over-generalising) over the long term investing in gilts is going to lose you money over saving in best buy savings accounts by about 0.5%pa. So an equity cash mix is arguably better than an equity/gilt mix. In practice I would prefer to suggest it's OK to just use savings rather than gilts rather than saying savings are superior. This involves many assumptions, in particular that the yields on 5 year gilts reflect the difference in interest available in savings accounts compared with gilts over longer terms than 5 years, and that 0.5%pa is that differential on average.

At the moment gilts with 5 years to maturity are priced at about 4.13%pa gross redemption yield. And best buy 5 year fixed rate savings accounts are 4.4%pa (non ISA) and 4.15pa (ISA) so the 0.5% differential has narrowed over than I've observed in the past.

I've mentioned this differential between gilt yields and savings rates before in the past, and I've not seen convincing counter-arguments (to me at least) from the 'it's essential to use gilts rather than savings' thinkers. Leaving aside arguments that it's difficult to access best buy savings as 'my money is in a pension' arguments, it is suggested that gilts provide a magical dampening affect to fluctuating equity returns that savings don't provide. Personally I think that argument is overstated and the dampening affect is quite small, and if you are investing in short term gilts equivalent in term to the fixed rate accounts available, it's hard to argue for the dampening affect.

Just wondering what the thoughts of masonic and OldScientist and others are on this. Does the data back up my observed differential? And how would investing in 10 year maturity gilts (rather than 5) and then reinvested at maturity in another 10 year gilt change the comparison based on historical data? And how would best buy 1 year fixed rate or easy access, vs 5 year fixed rate savings alter the historic comparison?  

For disclosure of bias, leaving aside some index-linked gilts I've recently bought, I have an equity/savings mix and have never owned gilts or gilt funds etc.     

One of the big problems is finding historical data for savings accounts whereas gilt yields go back when they were launched at the end of the 17th century. The sources I linked to upthread (and the ones masonic has provided) are mainly limited to easy access or 1 year fixed (even Martin's newsletter tend to concentrate on these) but do go back to the 1940s. At some stage, I was going to subscribe to the FT archive (although I think I can access it through our local library) to trawl through adverts for savings accounts.

However, I'd agree with you that tax and access considerations aside savings accounts currently give gilts a good run for their money. Access you've already covered (i.e., savings accounts within a SIPP), while there are tax benefits to individual gilts held in a GIA (which is why yieldgimp has the net redemption yield for 40% tax payers -on a historical note, my understanding, possibly faulty is that until 1997?, tax on the coupons of gilts were taken at source). Getting best buy rates in the 1980s either involved walking round local branches and looking in their windows or undertaking business by post - finding the best buy rate was not easy. For much of that period and the following decade or two, I saved with what was still commonly referred to as a 'post office' account (even though national savings has been split off in the early 70s?) and the bank I held my current account with.

As for the maturity, during periods of rising rates short maturity 'things' (whether gilts or savings accounts) will do better while during periods of falling rates longer maturities will do better* (there was a recent thread here on investing mistakes - where forum members had 3 or more year fixed rate accounts they'd taken out before the steep rises in rates). Over long periods (more than 100 years), the optimum return appears to have occurred for intermediate maturities (around 9 years - see my earlier graph) but is a fairly wide peak.

* Of course, the problem comes in trying to predict whether the rates will rise or fall!

FWIW, excluding a short inflation linked gilt ladder, over 60% of our fixed income is in two 1 year savings accounts set to mature 6 months apart (we take income from our portfolio twice per year) with the remaining 40% being in global bonds (about 50/50 in the vanguard short and normal global bond funds) to provide some international diversification. The overall duration is set to lie between 1 and 2 years (fixed income is the only area where I allow myself a bit of mild active investing - the overall average duration will be closer to 1 when I think rates will rise and closer to 2 otherwise - currently at just under 1.5 years having increased from near 1 a few years ago).