Only fixed income lets you name the amount and the day in advance. The diversification argument is the one usually made for it, and it is the one that has failed.
8 min read · Level 2
The strongest case for owning bonds is not the one that is made for them. Diversification is the argument on every allocation page, and it rests on a correlation that is measured over the past, converges towards one under stress, and failed visibly in 2022 when high-quality bonds and equities fell together. It is a real benefit, worth having, and it is the wrong thing to build a plan on. The argument that has never failed for a solvent issuer is far duller and considerably more useful: a bond states a sum and a date, and that is the only instrument in the entire investable universe that does. Everything else - equities, property, commodities, a diversified fund of any kind - offers an expectation, and an expectation cannot pay a school fee in March. What makes the distinction practical rather than semantic is that the dating argument also tells you which bond to buy, while the correlation argument tells you only to own some. A portfolio built around dated obligations has a maturity for every holding and a reason for it; a portfolio built around diversification has a percentage and no reason for its maturity at all, which is how allocations that were meant to be the stable part end up in thirty-year paper because it yielded a little more.
Three jobs, in descending order of how reliably fixed income does them.
The first, and the only one no other asset class does at all, is naming a date and a sum. If you owe school fees of a known amount in six years, a bond maturing in six years is the only instrument that addresses that obligation directly, because it is the only one that states in advance what will arrive and when. Equity has neither a date nor a sum. Cash has a sum but no term, so it is exposed to whatever rates do. Property has neither. This is duration matching, and it is the reason insurers and pension schemes are the largest bond holders in the world: they have dated liabilities and fixed income is the instrument shaped like one.
The second is contractual income that does not require you to sell anything. A holder drawing from an equity portfolio must sell units, and must therefore sell them at whatever price prevails on the day the money is needed. A bondholder receives the coupon whether the market is calm or not. That does not make the income larger; the previous lessons showed how much of it tax and inflation take. It makes the income independent of the market's mood on a particular Tuesday, which for anybody spending from a portfolio is a genuinely different property.
The third is behaving differently from equities in a shock, and this is the one the industry leads with and the one that is least dependable. Two illustrative shocks, using a 70/30 split between equities and high-quality bonds:
Shock one - a flight to quality. Equities fall 35%, bonds rise 6%. 100% equities -35.00% 70/30 -22.70% Cushion from the bonds: 12.30 percentage points
Shock two - rates and equities fall together. Equities fall 18%, bonds fall 14%. 100% equities -18.00% 70/30 -16.80% Cushion from the bonds: 1.20 percentage points
Same allocation, same 30% weight, and the protection in the second case is about a tenth of the protection in the first. Both are things that happen. In 2022, high-quality government bonds and equities both produced negative calendar-year returns in the major markets - a year in which the hedge did not hedge, for a reason that was entirely predictable from the pricing lesson: when the shock is rising interest rates rather than falling confidence, the mechanism that usually makes bonds rise is the mechanism making them fall.
Because a portfolio is a set of jobs, and the honest case for fixed income is the job it never fails at rather than the job it is usually sold on.
Duration matching has not failed for a solvent issuer, ever, because it is not a statistical property. If a bond matures on a date and pays a sum and the issuer pays it, the obligation is met. Correlation, by contrast, is a statistical property measured over a past period, and correlation between risky assets has a documented tendency to converge in a severe decline. Building a portfolio whose survival depends on a historical correlation holding at the worst possible moment is exactly the mistake the diversification subject warns about, and fixed income is the asset class where that mistake is made most confidently.
There is a second reason to lead with the date rather than the correlation, which is that it tells you which bond to buy. Correlation gives you no guidance on maturity at all - it says hold some bonds. A dated liability tells you exactly how long the bond should be, and that single decision, as the pricing lesson showed, matters more than almost anything else about the holding.
Which brings up the structural choice. Suppose 50,000 of fixed income against a general need rather than one dated obligation. Two ways to hold it, both at a 4% yield with 4% coupons, and yields then rise two points:
All 50,000 in a single 5-year bond Value after the move: 45,787.64 -8.4247%
A ladder: 10,000 each maturing in 1, 2, 3, 4 and 5 years Rung 1 (1 year) 9,811.32 Rung 2 (2 years) 9,633.32 Rung 3 (3 years) 9,465.40 Rung 4 (4 years) 9,306.98 Rung 5 (5 years) 9,157.53 Total 47,374.55 -5.2509%
The ladder's average maturity is three years rather than five, so its shorter duration absorbs less of the move. It also does something the single bond does not: a rung matures every year, and that cash is reinvested at whatever rates then prevail. In a rising-rate world that is a feature, because the money goes back to work at the higher rate. In a falling-rate world it is the reinvestment risk from the second lesson of this subject, arriving in instalments. A ladder does not remove that risk. It spreads it across five different rate environments instead of concentrating it in one, which is a real benefit and a smaller one than it is usually claimed to be.
Educational, not personalised financial advice. Nothing in the Investor Academy is a recommendation to buy, sell or hold anything — what is right for you depends on your own circumstances and the rules where you live.
Give each holding a job and the allocation question mostly answers itself.
Money needed in 14 months for a known bill Job: be there, in full, on the date. Cash or a bond maturing then. Not equities, and not a long-dated bond - the 30-year in the pricing lesson moved 27.53% on a two-point rate move, which is not a suitable home for a dated obligation 14 months away.
Money needed in 6 years for a known amount Job: arrive on schedule. A bond or a set of bonds maturing around then, or a zero-coupon bond if the amount matters more than the interim income. This is the only job on this list that fixed income does better than every alternative.
Money with a long horizon and no date Job: grow. Equities. Bonds here are a tax on the horizon rather than a protection, because the thing being protected against - a fall you will not need to realise - is not a risk on that timescale.
Money that funds spending from the portfolio, indefinitely Job: produce cash without forcing a sale at a bad price. A mix, with the bond allocation sized against how many years of spending you want to be able to fund without selling equities.
A general allocation held to soften equity declines Job: correlation. Legitimate, but size it knowing the cushion was 12.30 points in one shock above and 1.20 in the other, and that nobody tells you in advance which kind of shock is arriving.
Notice what happens if you cannot name the job for a holding. That is not a neutral state; it usually means the position exists because fixed income is what one holds, and a position held for that reason has no basis on which to be sized, no maturity that is right for it, and no condition under which it should be changed.
Take the obvious one first: buying bonds for safety without asking which of the four risks is being reduced. A thirty-year government bond reduces credit risk to almost nothing and increases interest-rate risk enormously, and calling the result safe is a statement about the borrower disguised as a statement about the money.
The second is relying on the correlation benefit at the moment it is least reliable. It is worth having and it is not insurance, and a plan that requires bonds to rise while equities fall has a failure mode with historical precedent.
The third is holding long-dated bonds against short-dated needs, and short-dated bonds against long-dated ones. Both are mismatches; the first is more expensive and the second is more common.
The fourth is a location error rather than a selection error: holding the annually-taxed income asset in the taxable account while the deferred-gain asset sits in the sheltered one. The previous lesson put a number on what that costs, and it accrues every year both are held.
Not the yield, not the credit, not the fund - the obligation. If there is a specific amount needed on a specific date, that pair determines the maturity and largely determines the instrument, and everything else becomes a question of which issuer to use. If there is no date, be honest that you are buying the correlation job rather than the matching job, and size it as the less reliable of the two.
Longer bonds usually yield more, and the extra sensitivity they carry is disproportionate to the extra yield - modified duration went from 4.45 at five years to 17.29 at thirty in the pricing lesson. Reaching along the curve for yield is the most common way a fixed income allocation stops being the stable part of a portfolio, and it is always done for a reason that sounded modest at the time.
Run the second shock above against your own weights, not just the first. If a year in which both equities and bonds fall would force a change of plan, the allocation is relying on a correlation rather than on a schedule, and the fix is either a smaller bond weight with shorter maturities or a set of dated holdings matched to the money you actually need.
When money has a date and an amount attached, when you need income that does not require selling into a weak market, and as a deliberately sized allocation against equity declines - with the second shock above in mind rather than only the first.
Avoid fixed income for long-horizon money with no date, where its job is unclear and its real return after tax and inflation may be near zero; avoid long maturities for anything needed soon; and avoid treating a bond allocation as insurance, because the one thing insurance must do is pay in the bad year.
Allocations built on the expectation that bonds rise when equities fall are tested rarely and specifically: they hold up when the shock is a loss of confidence in corporate earnings, and they do not hold up when the shock is a rise in interest rates. 2022 was the second kind in the major markets, and high-quality government bonds and equities delivered negative calendar-year returns together. Holders often describe such a year as unprecedented. It is not, and more importantly it is not unexplained - the pricing arithmetic in this subject predicts it exactly, because the same rate rise that reduces the present value of a company's future earnings reduces the present value of a bond's fixed coupons.
Ask what kind of shock your allocation is protected against, not whether it is diversified. A plan that only survives the confidence shock is a plan with one uncovered case.
Create a free account to read the rest and apply it to your own investing.