Are you in fixed interest for the duration?

Agree, David. Bond indices certainly do change in ways that equity indices don't tend to. I had discussions back in my Colonial First State days with some institutional clients about not using the off-the-shelf indices, but tailoring the duration and curve structure to align with their liabilities. Otherwise there was Value at Risk' for their business even if the fund was positioned 'at index'. (David Bell will remember this well, as he was on my team at the time and did a lot of excellent work on this.) One large mandate we had for many years took this approach, so they weren't affected by changes in the duration of the 0-5 year index.
We tended to only get traction with this sort of discussion with insurance companies who had decent actuarial estimates of their liability profile. However, I think that all investors should pay more attention to this, so again I'm glad you've raised the subject.

Thanks Warren, you have elaborated on another piece of the equation. Part of the reason for the index duration extension certainly relates to the downward movement of interest rates over time. And in addition to issuance patterns, on the demand side of the equation investors are being tempted to attain exposure further out the curve in order to seek some form of yield pick-up. So a few variables are in play, not to mention the changing credit quality aspect – another article in the offing! The overarching takeaway is I suggest that indices are dynamic beasts, with evolving embedded risks, some of which are not as readily observable as a headline yield but which will nonetheless impact on how the investment exposure performs over time (under differing macroeconomic and volatility scenarios).

Thanks David. This article raises a meaningful issue and starts a helpful discussion about it.

I come at assessing bond portfolio risk a little differently. This starts with an observation that duration isn't the weighted average term to maturity, which is how you put it. Rather, it's the weighted average time to receive all cash flows - coupons and maturities. This might seem like just a quibble, but the distinction is important because it's the main explanation for why the index has lengthened. As yields have fallen, the new bonds issued into the market have had lower and lower coupon rates. This means that the cash flows from fixed income are becoming more dominated by the capital return at maturity.

To illustrate, look at two US Treasuries that have the same maturity date - 15 August 2026. Both are ten year bonds today and both currently yield close to 1.5%. However, one was issued 20 years ago as the then current 30 year bond and has a coupon rate of 6.75% because of the prevailing market level at the time. The other was issued this month with a coupon rate of 1.5% reflecting today's market.

The first one, the 6.75% coupon security, has a duration of 7.8 years. The recent issue, the 1.5% coupon, has a duration of 9.2 years.

So it's not that governments are issuing longer dated bonds, it's that the bonds they issue have a longer duration, even though their final maturity term is the same. And as time has gone on, and older bonds have matured, the market - the Barclays index - quite naturally has more of these more recent bonds.

Does that mean the market is now more risky? We need more information to answer that question. The prospective volatility of bond market returns - risk, if you like - isn't just measured by the duration of the market. Really, it's duration times the probability of a yield change. I'd venture to suggest that back in the 1990's when the index in your chart was at 4.5-5.0 years duration, it was more risky than now because the odds of a 1 or 2% shift in yields over a 12 month period were much greater than it is now.

It's true, as you say, that if yields rise by 1% then the capital value impact will be greater now than when duration was lower. That's a mathematical fact. And it's true that the second element of risk, the probability of a yield change, is subjective and in reality anybody's guess. But in the interests of helping investors who think about such things to make a fully informed decision, I think that both components need to be explored.

As you say in your conclusion, under some economic scenarios the longer duration of bond funds could become problematic for investors. I suspect, however, that under those scenarios, a 5 year duration market would have been just as problematic as a 7 year duration market, both producing a year or so of negative returns.

Ashley, how much would you pay for a perpetual bond paying zero interest? Sorry, that's a trick question - the answer is zero. No doubt someone out there will get sucked in should the Japanese government ever try it, but they wouldn't be fixed income investors!

Good account of the rising duration of global bonds and the implications

· Could be a fair way off ending – ie rising duration will probably continue for some time yet – eg Japan + Europe will probably start needing to issue perpetual bonds (like the old british consols) – one step before straight-out helicopter money which is a permanent non-debt issue of money

· People often forget that in inflationary environments you want to be a borrower (eg our parents/us in the 1960s to the 1980s), but in deflationary environments (ie now and the foreseeable future) you want to be a lender. Hence I see good returns from bonds for some time – especially long duration bonds

· So the old Lehman Agg is doing us a favour by passively increasing returns for us in a deflationary world !