Which bond fund as part of a balanced long term portfolio for (eventual) drawdown?

michaels

So my portfolio is pretty simple:
10% money market (GBP), chose fund with lowest costs
70% global equity tracker (again lowest cost)
20% bonds

But what bond fund to chose
UK or global?
Short, long or aggregate?
What sort of fees level is reasonable?  ETF seems to cost about 4-5 basis points less, any reason not to go with one?

dunstonh

10% money market (GBP), chose fund with lowest costs

Most people favour STMM funds over MM funds.   Be careful you are not letting cost lead you to the wrong option.

But what bond fund to chose
UK or global?
Short, long or aggregate?

Which best fits your objectives and strategy?
What is your view on exchange rates or would you look at hedged funds?

You wouldn't typically use long term for a short term objective.   However, if you are not using drawdown in the short term, then that is something for a another day.

What sort of fees level is reasonable? 

Typically around 0.1% for passive.

michaels

So when they talk about 70:30 equities/bonds in SWR modelling, what do they typically have in mind for the bonds component?  Govt or investment grade only? What sort of average tenor?

Obviously most SWR research is UK US focused.  For UK investors it seems to generally go with international equities rather than UK or US but then perhaps bonds do not then form such a good 'inversely correlated' hedge or should the bond fund choice reflect the equity fund?

Linton

michaels said:

So my portfolio is pretty simple:
10% money market (GBP), chose fund with lowest costs
70% global equity tracker (again lowest cost)
20% bonds

But what bond fund to chose
UK or global?
Short, long or aggregate?
…..

In my view to rationally determine which fund to buy you should be clear on what specific benefits you expect the fund to provide. What is your financial strategy for retirement and what funds do you need to implement it? This analysis should answer your questions or may lead you to conclude, as I have done, that broadly based bond funds have no role at all.

JohnWinder

Individual bonds have a role in retirement investing, but you ask about funds so we’ll ignore non-funds.

So when they talk about 70:30 equities/bonds in SWR modelling, what do they typically have in mind for the bonds component?

I’ve read enough to conclude there’s no consensus. Government only bonds are safer; investment grade corporate bonds are pretty safe; junk bonds are risky, meaning credit risk in each case (not duration risk). You can choose the credit risk you like, and dial up or down the equity holding to allow for the bonds’ risk. I think there’s a case for government only; you want the safest when equities crumble. In times that cause equities to crumble some investment grade corporate bonds turn to junk, eg some Virgin airline bonds did this as the pandemic hit. If the market is efficient, then the more yield you get from bonds the more risky they must be. Nonetheless, sensible people choose blended govt/corp funds that are above BBB+ grade (or whatever the cut off is for junk).

Obviously most SWR research is UK focused.

I’d have said ‘US’ but we all read different things.

but then perhaps bonds do not then form such a good 'inversely correlated' hedge or should the bond fund choice reflect the equity fund?

Negative correlation is nice, but forget about hoping to achieve it. Correlations change over time and are unpredictable. We used to hope stocks and decent bonds would be a good mix with low or negative correlation, and 2022 came and bond and stock prices moved in the same direction - down. The old folk say that when things turn to stool correlations turn towards 1. Don’t trust old correlations.
UK or global? If you trust your own country, UK government is enough. If you wanted to hope to sometimes eke more return out with global bonds, get some but make sure they are currency hedged as most are. It’s considered unwise to add currency volatility (read ‘risk’) to bonds' low returns, especially when you hold bonds for safety.
Short of long? It depends on whether you need to spend that money sooner, or later. Duration risk is interest rate risk: rates go up, your prices go down (the longer the bond the further down); or, you hold short term bonds and rates go down while I’ve got longer bonds and keep on the high interest rate for longer than you. There’s no way to win every time unless you can predict interest rate movements; but you can minimise interest rate risk by holding bonds whose ‘length’ matches the time until you need to cash them in. Perhaps that means early in accumulation you hold longer bonds; well into retirement you hold shorter bonds. There are some simple ways to quantify that consideration, for later.
Don’t forget linkers. They protect against unexpected inflation, one of the big risks to a retirement portfolio.
Interested to hear how objectives and strategies add to or detract from those considerations.

michaels

JohnWinder said:

Individual bonds have a role in retirement investing, but you ask about funds so we’ll ignore non-funds.

So when they talk about 70:30 equities/bonds in SWR modelling, what do they typically have in mind for the bonds component?

I’ve read enough to conclude there’s no consensus. Government only bonds are safer; investment grade corporate bonds are pretty safe; junk bonds are risky, meaning credit risk in each case (not duration risk). You can choose the credit risk you like, and dial up or down the equity holding to allow for the bonds’ risk. I think there’s a case for government only; you want the safest when equities crumble. In times that cause equities to crumble some investment grade corporate bonds turn to junk, eg some Virgin airline bonds did this as the pandemic hit. If the market is efficient, then the more yield you get from bonds the more risky they must be. Nonetheless, sensible people choose blended govt/corp funds that are above BBB+ grade (or whatever the cut off is for junk).

Obviously most SWR research is UK focused.

I’d have said ‘US’ but we all read different things.

but then perhaps bonds do not then form such a good 'inversely correlated' hedge or should the bond fund choice reflect the equity fund?

Negative correlation is nice, but forget about hoping to achieve it. Correlations change over time and are unpredictable. We used to hope stocks and decent bonds would be a good mix with low or negative correlation, and 2022 came and bond and stock prices moved in the same direction - down. The old folk say that when things turn to stool correlations turn towards 1. Don’t trust old correlations.
UK or global? If you trust your own country, UK government is enough. If you wanted to hope to sometimes eke more return out with global bonds, get some but make sure they are currency hedged as most are. It’s considered unwise to add currency volatility (read ‘risk’) to bonds' low returns, especially when you hold bonds for safety.
Short of long? It depends on whether you need to spend that money sooner, or later. Duration risk is interest rate risk: rates go up, your prices go down (the longer the bond the further down); or, you hold short term bonds and rates go down while I’ve got longer bonds and keep on the high interest rate for longer than you. There’s no way to win every time unless you can predict interest rate movements; but you can minimise interest rate risk by holding bonds whose ‘length’ matches the time until you need to cash them in. Perhaps that means early in accumulation you hold longer bonds; well into retirement you hold shorter bonds. There are some simple ways to quantify that consideration, for later.
Don’t forget linkers. They protect against unexpected inflation, one of the big risks to a retirement portfolio.
Interested to hear how objectives and strategies add to or detract from those considerations.

Thanks, I did mean US but obviously had a brain fart.

My understanding is that the SWR analysis found generally a higher withdrawal rate could be achieved with a portfolio that was less than 100% equities, I assume because either they are not 100% correlated with equities or simply because they damp down on volatility and it is the extremes of volatility that lead to failures.

Thus if you are looking for a portfolio that will most likely achieve an historic SWR then the portfolio should match those on which the SWR analysis was carried out (lets park that SWR is a what happened in the past not what will happen in the future tool).

But generally SWR 'receipes' just say x% stocks, (1-x)% bonds plus perhaps 'UK rates historically lower so consider global diversification' without really looking at whether that means a global index tracker for the stocks, whether investments should be currency hedged or not nor really any comment on what 'bonds' means.

Prism

If you look at the original SWR studies plus many that came after, they were US based. The default assumption in bonds was US only government bonds of intermediate duration (3-7) years. 

So, if we try and map that to a UK investor that would be a currency hedged global government bond fund or a UK Gilt fund with a duration somewhere near that. Most globals seem to be ballpark 5-7 years so probably work fine. For a Gilt version you would need to combine a short term and a core gilt fund to provide the same duration and level of risk.

Personally I go the global route as... well why not. Duration is about right, hedging seems to be quite cheap and having an allocation to multiple countries seems less risky than just the UK.

E.g iShares Global Govt Bond UCITS ETF | IGLH | Hedged

JohnWinder

But generally SWR 'receipes' just say x% stocks, (1-x)% bonds plus perhaps 'UK rates historically lower so consider global diversification' without really looking at whether that means a global index tracker for the stocks, whether investments should be currency hedged or not nor really any comment on what 'bonds' means. ‘

You can read background, although you already have, and find source material about SWR here and in the ‘big ERN’ series to address your uncertainties: https://www.bogleheads.org/wiki/Safe_withdrawal_rates#Trinity_Study:  

But a single withdrawal rate is not a recipe to follow for 20 years without the risk of becoming badly unstuck. The ‘rules’ are a rough idea of what might work, and I don’t think building a portfolio to try to match the asset allocation of the best SWR method you find is the best approach to portfolio construction.

the SWR analysis found generally a higher withdrawal rate could be achieved with a portfolio that was less than 100% equities, I assume because either they are not 100% correlated with equities

'found' - past tense. 'are' - present tense, should be past tense 'were'. Once more with feeling: don't trust old correlations.

OldScientist

JohnWinder said:

But generally SWR 'receipes' just say x% stocks, (1-x)% bonds plus perhaps 'UK rates historically lower so consider global diversification' without really looking at whether that means a global index tracker for the stocks, whether investments should be currency hedged or not nor really any comment on what 'bonds' means. ‘

You can read background, although you already have, and find source material about SWR here and in the ‘big ERN’ series to address your uncertainties: https://www.bogleheads.org/wiki/Safe_withdrawal_rates#Trinity_Study:  

But a single withdrawal rate is not a recipe to follow for 20 years without the risk of becoming badly unstuck. The ‘rules’ are a rough idea of what might work, and I don’t think building a portfolio to try to match the asset allocation of the best SWR method you find is the best approach to portfolio construction.

the SWR analysis found generally a higher withdrawal rate could be achieved with a portfolio that was less than 100% equities, I assume because either they are not 100% correlated with equities

'found' - past tense. 'are' - present tense, should be past tense 'were'. Once more with feeling: don't trust old correlations.

And, in the past, highly dependent on country (30 year retirement, asset returns from macrohistory.net) - note: both Japan and Germany are for post-1950 only (i.e., I removed WWII and, for Germany, 1920s hyperinflation from the analysis). See graph below.

UK peaks at 100% stocks (because bonds have, at times, been dreadful for long periods - e.g., post WWII)
US peaks at about 90% stocks - because at this allocation the crash in 1929 becomes worse than the retirements starting in the 1960s.
Japan - peak at around 40% - above this, crash in 1989 produces lowest SWR.
Germany - peaks at 5% stocks, but good SWR up until 50-60%. I suspect this is because they have had sound money and good bonds!

I wouldn't try to optimise asset allocation based on SWR..