Key takeaways
- Computershare projected a 3.5% yield on UK equities for 2026. A portfolio spending only its natural income at that rate is spending 3.5%, whatever the plan said.
- Of 578,491 UK drawdown plans taking regular withdrawals in 2024/25, 259,507 took 8% or more of the pot a year. That is roughly 45% of them.
- In 2026/27 a dividend outside a wrapper is taxed at 10.75% or 35.75% after a £500 allowance. A sale is taxed at 18% or 24%, on the gain only, after £3,000.
- Over the 25 years to June 2020, the MSCI World Index returned 7.9% a year and its high dividend yield version 7.8%, on income returns of 2.1% and 2.9%.
- Vanguard's Australian Shares High Yield Index Fund held 69% of assets in its top ten holdings at 30 June 2020, against 18% for its international shares fund.
A dividend and a sale of units leave the same portfolio behind
When the salary stops, the first question is usually this: do you live off the income your investments pay you, or do you sell units when you need the cash? People feel strongly about it. The arithmetic doesn't.
Take £1,000 of shares that pay a £30 dividend. After the payment you hold £970 of shares and £30 in cash. Now take £1,000 of shares that pay nothing, and sell £30 of them. You hold £970 of shares and £30 in cash. The two positions are identical, because a dividend is not new money. It's your own capital moved from one pocket to another, and the share price falls by roughly the payment on the ex-dividend date.
Samuel Hartzmark and David Solomon open The Dividend Disconnect, published in the Journal of Finance in 2019, with exactly this point. Their working draft puts it this way: "At the heart of the dividend irrelevance result from Miller and Modigliani (1961) is the idea that money is fungible, implying that a value-maximizing investor should treat money equally regardless of its source."
So the honest answer is that natural income vs total return doesn't change how much you can spend. It changes four other things: what you pay in tax, what you're forced to hold, whether the portfolio dictates your spending or you do, and how you behave in a bad year. Those are worth knowing about, and this piece is about them.
Natural income vs total return is a question about tax and constraint, not spending power
The two approaches have names. Spending only the dividends and interest a portfolio throws off is the natural yield or income approach. Spending from the whole return, income plus capital growth, selling units to make up any shortfall, is the total return approach.
Vanguard's August 2020 research note An enduring solution for low yields describes the income route plainly. "With an income-focused approach, the investor constructs a portfolio with a natural yield (dividends and interest) consistent with their spending objective. Consequently, their asset allocation and diversification decisions are driven primarily by the natural yield of the investments they select rather than the investor's financial goals, risk tolerance and time horizon."
That sentence is the whole trade. Under natural income, the yield you need determines what you own. Under total return, what you own determines nothing about how much you take out, because you can always sell. Neither is free.
UK equities were projected to yield 3.5% in 2026, and most drawdown plans take more
Computershare's UK Dividend Monitor for the first quarter of 2026 put the projected UK equity yield at 3.5%, with headline dividends forecast at £91.6bn for the year. That is the ceiling on what a UK equity portfolio hands you without selling anything.
Now look at what people actually withdraw. The Financial Conduct Authority publishes retirement income market data every year. In 2024/25, £70,876m came out of UK pension pots, up 35.9% on the £52,152m of 2023/24. Sales of drawdown policies rose from 278,977 to 349,992.
The interesting table is the one behind the summary. Of the 578,491 plans making regular partial withdrawals in 2024/25, the FCA counts 66,563 taking less than 2% a year, 89,900 taking 2% to 3.99%, 94,389 taking 4% to 5.99%, 68,132 taking 6% to 7.99% and 259,507 taking 8% or more. The chart above plots that distribution. Nearly 45% of plans sit in the top band alone.
Set 3.5% against that shape and the gap is obvious. Only the first two bands, about 27% of plans, could be funded by a 3.5% natural yield without touching capital. Everyone else is selling units already, whether they describe it that way or not. Withdrawal rates aren't spending rates, and a high rate on a small pot may be someone cashing in a stray old pension rather than funding a year of living. Even so, the distribution says that for most drawdown plans, natural income was never the binding constraint.
The pattern flips with pot size. Among plans of £250,000 and above, 20,490 of 144,757 took 8% or more, about 14%, while 34,357 took less than 2%. Among pots below £10,000, 32,037 of 38,354 were in the 8%-plus band. Large pots behave like portfolios. Small ones behave like bank accounts.
What each pound costs in 2026/27 depends on the wrapper more than the source
Here's where the choice does have a price attached, and it's a UK-specific one. These are the 2026 to 2027 rates.
| Wrapper | Dividend or interest taken as income | Units sold for cash |
|---|---|---|
| ISA | No tax. The 2026/27 subscription limit is £20,000. | No capital gains tax. HMRC's list of assets you do not pay it on names "ISAs or PEPs". |
| General account | Dividends: £500 allowance, then 10.75% basic, 35.75% higher, 39.35% additional. Interest: £1,000 personal savings allowance at basic rate, £500 at higher, £0 at additional, then 20%, 40% or 45%. A separate starting rate can cover up to £5,000 of interest for people with little other income. | Capital gains: £3,000 annual exempt amount, then 18% within the basic rate band and 24% above it. Only the gain is taxed, not the sale. |
| Pension in drawdown | What you take out is taxed as income at 20%, 40% or 45% above the £12,570 personal allowance. | Same treatment. Up to 25% can normally be taken tax free, capped at £268,275. |
Two things fall out of that table. First, inside an ISA or a pension the natural income vs total return question has no tax content at all, and the argument reduces to behaviour and convenience. Second, in a general account the asymmetry is large. A basic-rate taxpayer pays 10.75% on a dividend above £500, and 18% on a gain above £3,000. But the dividend is taxed on the whole payment, while the sale is taxed only on the profit inside it. Sell £10,000 of units that cost you £8,000 and the taxable amount is £2,000, not £10,000, and the £3,000 exemption swallows it whole.
The £500 dividend allowance is doing very little work at a 3.5% yield. It's covered by roughly £14,000 of holdings. Everything above that is taxable at your UK dividend tax rate whether you wanted the cash or not, which is the part income investors tend to miss: dividends are not optional. Units are. That control over timing is also why asset location usually matters more than the income question itself.
Living off dividends changes what you own, and by a lot
The cost of the income route isn't mainly tax. It's concentration. If the yield has to reach a target, the portfolio gets rebuilt around whichever assets are paying.
Vanguard measured this on its own funds at 30 June 2020. Its International Shares Index Fund held 18% of assets in its top ten holdings. The Australian Shares Index Fund held 43%. The Australian Shares High Yield Index Fund held 69%. That last number is a portfolio with roughly seven pounds in every ten sitting in ten names, arrived at by screening for income rather than by choosing it.
The same note tested whether the extra yield pays. Over the ten years to 30 June 2020, the MSCI World Index returned 12.2% a year on an income return of 2.2%. The MSCI World High Dividend Yield Index returned 10.3% on an income return of 3.5%. Over 25 years the two were almost level, at 7.9% and 7.8%, with income returns of 2.1% and 2.9%. More income, no more money. In Australia the gap was wider: the FTSE Australia Index returned 7.9% over the decade against 2.9% for its high dividend yield version, despite the income screen paying 5.4% a year against 4.6%.
Crisis behaviour was worse, not better. Between 1 February and 31 March 2020, Vanguard measured Australian high yield equities down 27.4% against 26.7% for the broad Australian market, and global high yield equities down 18.7% against 12.8% for global equities. The income sleeve fell furthest exactly when the income was needed.
The best case for natural income is behavioural, and there's evidence for it
The strongest objection to everything above is that people are not spreadsheets. A rule that says "spend the dividends" is self-enforcing. Nobody has to decide how much to sell, or when, or whether this month's fall means waiting.
Vanguard concedes the point in the same note: "Since the income-focused investor is only using the portfolio's natural yield, the portfolio determines both the amount and timing of withdrawals. Thus, there is no need to develop a spending strategy." For someone who would otherwise freeze in a drawdown, that's not a small thing, and a rule you'll follow beats a rule you won't.
Hartzmark and Solomon's data suggests the effect is real and measurable. Studying US brokerage accounts, mutual funds and institutions, they found investors treat dividends as a separate, stable stream and stop watching the price. Non-dividend-paying stocks in their sample showed a disposition effect, the tendency to sell winners and hold losers, of 10.9%. For dividend payers it was 3.96%. Mutual funds reinvested a dividend exactly back into the stock that paid it in only 0.719% of cases, and institutions in 1.17%.
Read one way, that's the case for income: dividends stop people fiddling. Read the other way, it's the case against: the same investors are ignoring capital losses because a cheque arrived, and they aren't putting the money back. Hartzmark and Solomon call the mistake the free dividends fallacy. Both readings are in the same evidence, and which one applies depends on whether the discipline is buying you calm or blinding you.
The pension rule changing in April 2027 sits underneath the whole question
One structural change makes the natural income vs total return debate secondary for anyone with a large pension. HMRC has legislated to include unused pension funds and pension death benefits in the value of a member's estate for inheritance tax. The measure takes effect "in respect of deaths on or after 6 April 2027".
HMRC's own estimate is that of around 213,000 estates with inheritable pension wealth in 2027 to 2028, about 10,500 will face an inheritance tax liability where they previously would not, and around 38,500 will pay more. The average liability is expected to rise by around £34,000 where pension assets are counted.
That reprices which pot you spend first, which is a different question from whether you spend income or capital. Our piece on withdrawal order in retirement tests the account sequence directly. The point here is narrower: once the wrapper decision moves, the income-or-sale decision inside it moves with it, and the yield of what you happen to hold has very little to say about either.
What this evidence cannot tell you
The return and concentration figures are Vanguard's, from an Australian research note published in August 2020, and Vanguard sells the broad index funds its argument favours. Its tax section describes Australian rules, under which "capital gains are taxed at a 50% discount rate if an asset is held for at least one year", so that half of the paper doesn't transfer to a UK reader. The ten-year window to June 2020 also captures a growth-led decade, and the 25-year figures in the same chart show a much narrower gap. A different 25 years could rank them differently, and a backtest is not a forecast.
The FCA data counts plans, not people. Someone with three pots appears three times, and a fast withdrawal rate on a tiny pot tells you nothing about that person's total income. The 2024/25 figures also cover a single year, and drawdown behaviour has been shifting fast: total withdrawals rose 35.9% in twelve months.
Hartzmark and Solomon's sample is US brokerage and fund data, so the tax incentives behind it aren't UK ones. And every tax figure here is a 2026/27 figure: 10.75% and 35.75% on dividends, £3,000 of exempt gains, a £500 dividend allowance. None of them has ever stayed still for long, and the arithmetic in the table changes with each Budget that moves one.
What would change the conclusion
If yields rose back above what you spend, the argument would mostly dissolve. At a 3.5% projected UK equity yield the income route constrains almost everyone drawing more than that. At 6% it would cover the first three FCA bands rather than two, still leaving 57% of plans short.
If dividend and capital gains rates converged, the general-account asymmetry goes away. Today a basic-rate taxpayer meets 10.75% on the whole dividend and 18% on the gain alone, and the second is usually the smaller number. Move either rate far enough and the ranking flips.
If everything you hold is inside an ISA or a pension, the tax half of this piece is irrelevant to you and only the behavioural half remains. That is a real answer rather than a dodge, and for a fully wrapped portfolio it settles the tax question outright.
The number worth watching isn't the yield on your holdings. It's the gap between what your portfolio pays you and what you take out, measured in pounds a year. When that gap is zero you have a choice. When it's wide, you're a total-return investor already, and the only open question is which units you sell.