Investing · South Africa
How Much Should You Hold in Equities and Bonds at Each Stage of Life?
The equity/bond split depends on when the money will be needed, the losses the household can absorb and the other resources available to meet its commitments. Age influences those facts, but cannot determine a suitable percentage by itself. A young person with an imminent payment can have little capacity for investment losses; an older person with adequately covered spending can have a long horizon for some capital.
Rules such as “100 minus your age” turn a difficult decision into one number. They omit whether you are still earning, whether that income is dependable, which expenses can change and whether the portfolio is already needed for withdrawals. A more useful starting point is to identify the job of the money before deciding its asset mix.
In this article
Life stages change the problem, not just the percentage
During early accumulation, future earnings can fund contributions and help recover from a setback. That support is weaker if employment is uncertain, income depends on the same economic forces as the investments, or debt payments leave little room to save. A long time until retirement does not make money needed for a home purchase next year long-term capital.
In mid-career, the balance sheet often becomes more complicated. Education expenses, parents' needs, a business interest and retirement saving can sit beside one another. Grouping them under one “balanced” allocation can hide different deadlines. The task is to establish which payments have reliable funding and what remains exposed to market losses.
Approaching retirement changes the ability to replenish capital from earnings. It also brings decisions about pension income and annuities that affect how much the investment portfolio must provide. After retirement, there can still be several horizons: next month's spending, later care needs and capital intended for a surviving spouse or beneficiaries.
| Stage or transition | Fact that deserves attention | Why an age rule can miss it |
|---|---|---|
| Building savings | Reliability of earnings and near-term commitments | Young does not mean every rand can remain invested |
| Multiple family obligations | Amounts, dates and funding sources | One average horizon can conceal a fixed payment |
| Leaving employment | Net income gap after dependable pensions | Two people retiring together may need very different withdrawals |
| Living from investments | Adjustable spending and resources after a loss | A willingness to accept volatility cannot pay a fixed bill |
| Later life or widowhood | Care, decision-making support and survivor income | Lower activity does not necessarily mean shorter financial obligations |
These are assessment questions, not recommended allocations for each row. They can also occur out of order: a business sale or loss of a spouse may change the portfolio's job more than another birthday.
Two people aged 65 can need different decisions
Consider two invented individuals, each with a R4 million investment portfolio and monthly spending of R36,000 after tax. One has dependable pension income of R30,000 after tax. The other has no pension income. Their first-year gaps are R72,000 and R432,000 respectively, before allowing for tax on any portfolio withdrawals.
| Same age, different income support | Annual net spending gap | Gap as a percentage of R4 million |
|---|---|---|
| R30,000 monthly net pension | R72,000 | 1.8% |
| No pension | R432,000 | 10.8% |
These percentages measure a funding gap; neither is a recommended withdrawal rate. The second person may need more growth but has less room for an early loss. Increasing equities because the plan requires a higher return does not create the capacity to bear the loss that can accompany it. Spending, employment, retirement timing or the way income is secured may need reconsideration.
The first person may have more flexibility, but that depends on the pension's increases and survivor terms, other obligations and the purpose of the R4 million. A nominal pension without increases can cover a smaller share of spending over time. Having a pension therefore informs the allocation without settling it.
Separate the denominator from the fund label
Suppose a household has R1 million ring-fenced for a known payment and R4 million in a fund holding 60% equities and 40% bonds. The fund contains R2.4 million of equities and R1.6 million of bonds. Across all R5 million of financial assets, the mix is 48% equities, 32% bonds and 20% reserved cash.
Calling the household “60/40” obscures that cash. But describing it as having 20% spare cash is also wrong if the payment is committed. After paying the R1 million, the remaining portfolio is genuinely 60/40 without any market trade. Both the current balance sheet and the post-payment allocation matter.
This is particularly relevant when comparing a discretionary account with retirement funds and a business. A business can add economic risk without supplying cash that can be rebalanced. A retirement fund can contain underlying equities even when its name sounds conservative. The household-concentration article explains how income and asset dependencies fit together.
Bonds need a job description too
“Bonds” is not a single level of safety. A nominal bond promises specified cash flows subject to the issuer's ability to pay. Its market value can change before maturity. Inflation-linked bonds tie payments or capital to an index under specified terms, but their market prices can still change as real yields move. Foreign bonds add currency exposure unless it is hedged; a stable foreign-currency price is not a stable rand value.
The National Treasury instrument terms provide the documents needed to identify an inflation-linked bond's actual promise. A broad bond fund, a short-duration fund, a retail savings bond and cash available tomorrow should not be treated as interchangeable ways of funding the same deadline.
One simple valuation shows the problem. Imagine a hypothetical zero-coupon bond paying R100,000 in ten years, with no default and no intermediate payment. At an 8% annual discount rate its price is R100,000 ÷ 1.08¹⁰, approximately R46,319. If the required yield immediately rises to 10%, with the payment and remaining term unchanged, its price becomes about R38,554: a fall of roughly 16.76%.
This is an invented instrument used to explain interest-rate sensitivity, excluding tax and dealing costs. There is no suggestion that a currently available government bond has these exact terms. The contractual maturity payment has not changed; the price available to someone needing to sell today has. A coupon-paying bond or a fund has different cash flows and sensitivity.
The interest-rate question explores that mechanism further. For allocation, the practical point is to check the bond exposure's duration, credit quality, currency and liquidity against its intended purpose.
Test a joint loss before treating bonds as a guaranteed offset
Diversification can reduce dependence on one source of returns, but it does not promise that bonds rise whenever equities fall. The 2026 UBS Global Investment Returns Yearbook public summary places equity/bond diversification and currency risk in a long historical context. It is evidence about past experience across markets, not a suitable South African allocation or a maximum future loss.
The SARB's June 2026 Financial Stability Review describes an oil-price shock and repricing across equities, sovereign bonds and currencies. The allocation lesson drawn here is narrower than a market forecast: different asset-class labels can still respond to a shared shock. For a South African household, the same episode can also raise transport costs or affect earnings. Testing the investment account while keeping every household cash flow unchanged can understate the decision's difficulty.
For a transparent stress test, assume equities lose 35% and bonds lose 10% in the same period. The invented examples below start with R5 million invested only in those two assets, before withdrawals, costs and tax. No likelihood is assigned.
| Illustrative equity/bond mix | Portfolio loss | Rand loss on R5 million | Capital remaining |
|---|---|---|---|
| 40% / 60% | 20% | R1,000,000 | R4,000,000 |
| 60% / 40% | 25% | R1,250,000 | R3,750,000 |
| 80% / 20% | 30% | R1,500,000 | R3,500,000 |
The purpose is to ask what the household would have to do after each loss. Would a payment be delayed, spending reduced, a business asset sold or a withdrawal made from depleted investments? If none of those adjustments is feasible, being emotionally comfortable with the allocation is not sufficient.
These cases are not a complete risk model. They omit recoveries, different bond types and many other outcomes. Selecting the mix with the smallest displayed loss would ignore the long-term cost of insufficient growth. Selecting the largest equity share because it might grow faster would ignore the household's immediate constraints. Both sides require examination.
Inflation is a reason to plan, not permission to assume returns
Money intended for later spending needs to be assessed in purchasing-power terms. South Africa's April 2026 Monetary Policy Review states the 3% inflation target with a one-percentage-point tolerance band. That is a policy target, not a guarantee for the next decade or for a household's medical and care costs.
Holding less equity can reduce one kind of exposure while making long-term spending more dependent on saving more, spending less or earning sufficient returns elsewhere. Equities have uncertain outcomes too. An asset mix cannot promise to repair an underfunded plan.
Vanguard's asset-allocation education similarly distinguishes tolerance from financial capacity and horizon. Its US examples and target-date products are not South African recommendations. A local assessment also needs the currency of liabilities, tax wrappers and restrictions on particular accounts. Regulation 28 places constraints on relevant retirement investments; a permitted maximum is not a personal target.
An allocation decision needs a reason and a review trigger
A useful allocation record identifies the capital being allocated, commitments funded elsewhere, the losses tested and the adjustments available if they occur. It explains why the selected assets suit those functions. It also identifies the event that would change the reasoning: a lost income, an earlier retirement, a new dependent or a major payment becoming certain.
Before turning that record into percentages, reconcile these four entries:
| Entry | Evidence to record | Error it can reveal |
|---|---|---|
| Committed payment | Amount, currency, due date and funding source | A long-term fund is counted as cash available on a fixed date |
| Dependable income | Net amount, end date, increases and survivor terms | Income that ends at retirement is counted indefinitely |
| Remaining capital | Holdings after commitments, including underlying exposures | A reserved amount appears again as capital available for growth |
| Response to a loss | Specific spending adjustment or other funding source | The plan depends on selling an asset nobody intends or is able to sell |
An unresolved entry is information still needed, not an invitation to fill the gap with a conventional age-based split. For example, the first person's R30,000 pension in the earlier illustration cannot be treated as dependable lifetime income if it is actually a consulting contract ending next year. The arithmetic would be correct for one year and the allocation premise wrong for the years that follow.
Reviewing that record is different from changing the allocation after every market move. Rebalancing can restore an agreed mix; a changed household fact may justify a different mix. The portfolio-review question separates those tasks.
A defensible allocation connects a stated exposure to a funded household plan. The four-entry record makes that connection reviewable when circumstances change. If the initial uncertainty is the amount retirement spending requires, work through the retirement-capital question before treating the asset mix as the solution. No example here assesses your risk capacity or recommends an equity/bond percentage for you.
Sources checked 6 September 2026. Historical research is not an indication of future performance. All examples are original educational scenarios, not forecasts or model portfolios for implementation.