This article can be read by selected Monevator members. Please see our membership plans and consider joining! Already a member? Sign in here.
Duration matching bond funds to your time horizon [Members]
Next post: Benjamin Graham on bear markets
Previous post: Weekend reading: recovering from regrets
Comments on this entry are closed.
Thanks TA, very useful and clearly explained as ever.
One question for you or the community if I may: how important is it for the chosen bond fund to have a duration close to the investment horizon, i.e. for it to be of no longer duration than necessary?
Clearly an investor could run the example with a smaller proportion of a long duration gilt fund rather than a larger proportion of VGVA, and still make the maths work. But I’m wondering if that would increase the risk of the matching not quite running as expected. For factors I know I don’t know enough about, things like changes to the shape of the yield curve or similar.
Good article and Tetromino makes a point you could mix cash and a longer duration asset to match your desired duration. (This provides optionality, the bulk of your holding has zero interest rate risk)
Clearly the yield curve and interest rates change , eg the duration of VGVA is shown as 11.1 , but rising rates tend to shorten duration and the Vanguard site currently shows it as 9.9
Excellent piece as always @TA.
“Exception one” – the inflation protection of ILGs is clearly a valuable consideration over long periods, but my understanding is that longer dated ILGs are even more responsive still (than their equivalent duration conventional gilt comparator) to changes in interest rate expectations in secondary markets, such that whilst, if they’re held all the way to maturity, and if brought at a positive yield, you should get your money back in real terms, between purchase and maturity the price could vary very widely, and if you sell then you might then end up out of pocket. Check out the nerve jarring price action on the longest duration ILG – Treasury 0.125% 22/03/73 index linked gilt – to see what I mean.
“Exception two”: @London Yank has, IIRC, referenced this in his comments on other threads. Advantage of Gilts, especially ILGs, is that they’re CGT free, and any gain in value on sale is not treated as income. This is especially useful if you’ve used up both the AA for SIPP and the annual ISA allowance, and so are holding Gilts directly (i.e. not within a bond fund or ETF) in a (taxable) GIA. However, there’s a technical ‘footnote’ here: namely that extreme caution should be taken over stripped coupon gilts/zeros. Stripped coupon Gilts are, for tax purposes, likely to be classified as Deeply Discounted Securities. Profits realised from the discount on a DDS are charged to income tax at the highest nominal rate of the person making the disposal (see the Income Tax (Trading and Other Income) Act 2005, section 427). If profit on sale of the Gilts wouldn’t normally rank as income, it’s still treated as such for these purposes (sections 427, 428), and person making disposal is liable to the IT charge. Check out the Revenue’s published / online Savings and Investment Manual at SAIM3000 onward.
@Hariseldon
Thanks, but it was a genuine question: I’d be interested to know the pros and cons of using a shorter or longer fund to reach the same weighted duration once combined with cash.
As you say, maybe there’s something awry with the VGVA figures in the article. Isn’t YTM currently about 4.5%?
@ Tetromino – sources for mortals didn’t mention it as a consideration. I’d guess ZX48k would have a view but I suspect it’d be situation-dependent and could involve him writing in incomprehensible symbols again 🙂
@ TLI – I did not know that about strips! And thank you for reading HMRC manuals so we don’t have to!
@tetromino. Duration matching is what is termed a dynamic hedge. It works at one instant in time and for one yield level. Duration is a first order (linear) approximation to the bond price-yield relationship, so it assumes small parallel changes in the yield curve. As time evolves, as yields move up or down, and/or the yield curve flattens or steepens, the quality of the hedge can deteriorate. Hence why it needs to be periodically rebalanced.
So, by duration matching a 5-year liability with a 20-year duration bond, rather than a 5-year duration bond, it would be no different at the instant you hedged. You’d buy less of the 20-year and keep more in cash but still be equivalently duration matched.
Issues would arise though. For example, if 5-year yield went down but the 20-year yield went up (a curve steepening) the hedge would not work. You’d see issues if the yields moved up or down a large amount, like 2022. While you were duration matched, you weren’t necessarily convexity matched (to non-linear parts of the bond price-yield relationship). Finally, time decay would be different. After one year, perhaps 20% of the liability’s duration would have disappeared but only possibly 5% of the bond’s duration. Duration matching with a 5-year duration bond or a 20-year duration bond both require rebalancing, but the 20-year runs the risk of larger errors and thus would need to be rebalanced more often.
You can contrast this with cashflow hedging which is a static hedge. It’s precise so this doesn’t typically need to be rebalanced. On the other hand, cashflow hedging is often difficult (or even impossible). It might be pointless if you don’t actually know the liability. Duration matching, by comparison, is easy and can be modified just as part of the normal rebalancing process.
@ZX – Thanks, that’s very helpful. The point about the risk of larger errors with the 20-year fits with my intuition, but it’s valuable to have the expert view. And thanks for taking the time to put it in (relatively) simple terms.
Can I just check something that initially confused me… If interest rates on cash drop after the initial investment, then the yield on the cash is going to fall short of the initial calculation, which sounds like a problem. But is it the case that this would also result in the bond fund value going up to compensate, and hence it evens out?
Also, I don’t mean to nit-pick, but isn’t reinvesting the yield only doing a quarter of the work in your initial example, not a third? You have to put in three-quarters of the target up-front, and then the remaining quarter comes from the yield. (The yield is equal to a third of the initial investment, but that’s not quite the same thing.)
@ Dave S – yes, you’re right. It’s the rise in bond fund value that keeps you on track.
Re: 1/3 vs 1/4. The way I think about it, if I put in £77K and yield / reinvestment takes me to £100K then that element is nearly a third of the amount I have to find to hit my target. I can see what you’re saying though, so it seems like a question of perspective to me.
Hi All
Not quite on subject but any ideas how best to take a position in long dated US treasuries. Any ETF recommendations?
Thanks
@John Wilkinson #10: Try IBTL, which is the iShares USD Treasury Bond 20+ year UCITS ETF (OCF 0.07%, indicative spread 0.09%). I don’t know if it’s hedged to £Stg though (probably not). If it’s not, and you then find a currency hedged alternative to it, could you drop the details into a reply? Many thanks. 🙂
@TLI @John Wilkinson: I use IBTL’s GBP-hedged sibling IDTG.
Many thanks @Onedrew. That’s just what I’ve been looking for.
Thanks @Onedrew