South Africa
Before Tomorrow
The country’s next twenty years are being shaped long before they arrive. The strongest evidence about where South Africa is going is already sitting, in plain sight, in the decade it just lived through.
On a clear night above the northern suburbs of Johannesburg, the rooftops give the country away. A growing share of them carry their own solar arrays, small private power stations bolted over the garage, quietly seceding from a grid the household has stopped trusting. Picture a million of those roofs and you have the defining fact of South Africa’s last decade: the people with the means to repair a failing state have mostly chosen to buy their way out of needing one.
Every year a fresh crop of forecasts arrives and misses this. Growth will be 1.4 percent, or 1.9. The rand will hold, or slide. The coalition will fracture by winter, or last the term. Almost none of it survives the following quarter, because almost all of it is watching the wrong thing. Forecasts watch events. Countries are decided by forces, and forces keep a slower clock.
Prediction treats tomorrow as a single event waiting to be guessed, a number to pin, a date to circle. Futures rarely arrive that cleanly. They accumulate, through decisions already taken, incentives already set, machines already ordered, pipes already laid, children already born, institutions already hollowed or quietly rebuilt. By the time a future announces itself as news, it has usually been under construction for ten years.
A decade of that construction now sits on the record. The forces that will govern South Africa in 2040 are not sealed in an envelope; most are already visible in the years between 2015 and 2025, for anyone who can tell a force from an event. A cabinet reshuffle is an event. The slow erosion of the state’s ability to deliver water is a force. A record month of blackouts is an event. The physics of an ageing coal fleet meeting a falling reserve margin is a force. That distinction is the whole discipline.
Ten years of shocks can look like chaos. Read as a system, they resolve into a handful of forces, each still running.
The past ten years handed South Africa a set of shocks that looked, at the time, unrelated. A currency crisis and two credit downgrades to sub-investment grade. A president recalled and another installed on a promise of renewal. The worst year of blackouts in the country’s history, followed by their near-disappearance. A pandemic. A week of looting in July 2021 that killed more than three hundred people. A governing party that had held an outright majority since 1994 falling to forty percent of the vote. A rooftop-solar boom that quietly rewired the electricity system. A diplomatic rupture with Washington that erased most of the value of preferential trade access.
Read as a list of events, this is chaos. Read as a system, it is coherent. The same small set of forces produced all of it, and those forces are still running. The task is to name them, to see how they lock into one another, and to trace where their interaction is already pointing. That requires a model.
The Accretion Map
Most of what will decide South Africa in 2040 is already here, in plain sight. The difficulty is that it sits beside a great deal that looks just as urgent and decides almost nothing. Any method worth the name has to separate the two before it does anything else. This one does it with two questions.
The first: once a force is moving, how long would it take to turn it around? Some are close to fixed. The number of twenty-year-olds in 2040 is already settled, because they are alive today. Others turn in an afternoon, the way an exchange rate does.
The second: when that force moves, how much moves with it? Some stay in their own lane. A cabinet scandal is loud and self-contained. Others drag the whole system behind them. When the electricity supply fails it pulls industry, employment, migration, the budget and the mood of the electorate along with it, all at once.
Ask both questions of every force acting on a country and four groups fall out, each of which rewards a different kind of attention. Three of the four are routinely mistaken for one another, and that confusion is most of the reason forecasting fails.
The Fundamentals
These decide the decade. The age structure, the installed power and water systems, the debt stock, the depth of the state’s ability to deliver. They move slowly enough to ignore and reach into enough to set the terms for everything else. This is where the future is most readable, and least discussed.
The Switches
The forces that can flip inside a single year and, when they do, bend the slow ones one way or the other. Capital flows, the currency, coalition arithmetic, the terms of trade, the pace of new technology. They cannot be predicted. They can be watched, and they are the thing worth watching.
The Character
Deep and real, but narrow in reach. Settled habits, the texture of informal institutions, the things a society takes for granted. This is the temperament of a country. It colours how a future feels without steering where it goes.
The Noise
Reshuffles, scandals, one quarter’s growth print, the daily cycle. It fills the headlines and most of the commentary. It earns attention only on the rare day it reveals that one of the switches has moved.
The map earns its place twice over. It shows where to look, and it shows why most commentary looks elsewhere. Public debate lives among the noise, which is immediate and easy to have an opinion about. Serious foresight lives among the fundamentals, where the next decade is already being written, and reads the switches to judge which way the writing bends.
The compounding engine
A map of forces is only half a model, because forces do not sit still on a grid. They act on one another. A weakness in one of the slow, far-reaching forces raises the price of every shock that follows it; a switch thrown the right way can, over years, repair one of the slow ones. Those interactions run in loops, and the loops are where the real motion is.
Some loops are vicious, each turn making the next one worse. A utility raises tariffs to cover its costs, the customers who can afford it install their own supply, revenue falls, the next tariff rise follows, and the next tranche of payers walks away. The same pattern runs through electricity, then water, then the public finances. Others are virtuous: reliable power restores confidence, confidence draws investment, investment funds capacity, capacity secures the power. Which loops are winning tells you far more about the decade ahead than any single force.
Forecasting watches the noise. Foresight reads the fundamentals, follows the loops, and watches the switches to see which way they turn.
This is the part that travels. The method reads any complex system in motion, not only this one. Ask what is slow and far-reaching, and treat it as close to settled. Find the loops, because they set the direction. Watch the switches, because they decide it. Discount the noise, however loud, until it tells you a switch has moved. The rest of this piece runs that method over a single country.
South Africa, mapped by force
Twenty forces shaping the country, placed by how hard they are to reverse and how much they move. The load-bearing ones cluster in the upper right, already settled and wired into everything. Select any force to read it, and to see what it moves.
What is already in the ground
Domains are coloured. Position is the argument: harder to reverse rightward, more reach upward.
How to read it. Nothing here is a prediction. The plot is a claim about structure: which forces are load-bearing, which are switches, and which deserve less attention than they get. The two loop views trace the feedback that turns a static grid into a direction of travel.
Five interlocking reads
Taken one at a time, the forces of the past decade are a familiar list. Taken together, they resolve into a small number of loops, each of which is still running, and each of which already implies a great deal about the country coming into view.
i. The demographic engine, idling
South Africa passed 62 million people at the 2022 census, up from 51.7 million in 2011, the fastest growth since democracy. The structure matters more than the total. Roughly half the working-age population is under 35, about 21 million young people. In most economies that shape is called a dividend, a wave of workers arriving to power two decades of growth.
The dividend is not being drawn. Official unemployment sat at 32.7 percent in early 2026, and for those aged 15 to 24 it was 60.9 percent. More than four in ten people aged 15 to 34 are not in employment, education or training. Youth unemployment climbed steadily across the decade, from 36.9 percent in 2015 to 46.1 percent by 2025 for the broad youth band. A generation is reaching adulthood with no first rung to stand on, and the economy’s failure to absorb it runs deeper than any business cycle; no single good year clears it. This is a fundamental: slow to shift, and wired into everything. It feeds crime, migration pressure, the demand for grants, the shape of every election, and the ceiling on growth itself.
ii. The substitution spiral, from power to water to the fiscus
The signature story of the decade is electricity, usually told as a crisis and a recovery. In 2023, the worst year, the grid shed power for roughly 6,948 hours. Then load-shedding all but vanished from April 2024. Told that way, it reads as noise, a bad patch that passed.
The structural read is different, and more important. Electricity tariffs rose about 11 percent a year for a decade against roughly 5 percent inflation, until the average price of grid power climbed above the cost of new solar. Households and firms that could afford it did the arithmetic and left: privately owned solar passed 6 gigawatts by late 2024, and demand on the grid fell three percent in a single year. The blackouts eased partly because the state utility improved, and partly because the wealthier half of the economy quietly built its own. That is a vicious loop wearing the mask of a recovery. Every customer who self-supplies narrows the base across which fixed costs are spread, pressuring the next tariff, prompting the next exit.
The same loop is visible in water, one cycle behind. South Africa is among the thirty driest countries on earth, and nearly half of its treated water, about 47 percent nationally, is lost or unbilled before it reaches a paying customer. Johannesburg loses close to half of its supply; the share of wastewater systems in a critical state rose from 39 percent in 2022 to 47 percent in 2025. Those who can afford boreholes and tanks are exiting municipal water exactly as they exited the grid, and the revenue that would fund the pipes leaves with them. The fiscus runs the same loop nationally: gross debt near 79 percent of GDP, and a debt-service bill that consumes a rising share of every rand, crowding out the maintenance that would have prevented the exits in the first place.
South Africa is running a national experiment in exit over voice
Fifty years ago the economist Albert Hirschman gave organisations in decline two responses: exit, in which you leave, and voice, in which you stay and push for repair. His warning was that the two corrode each other. The members most able to fix a failing institution are usually the first with the means to leave it, and every exit removes a voice that might have forced the repair.
That is precisely what six gigawatts of private solar, a borehole boom and armed-response contracts describe. The households and firms with the capacity to demand a working grid, a working tap or a working police station are instead buying their way out of needing one. Each exit is individually rational and collectively corrosive. It is why the same loop keeps appearing in different systems, and why a recovery measured only in blackout hours can mask a state that is quietly emptying of the people who used to hold it to account.
iii. The political reset, and the switch it created
The African National Congress won 62.7 percent in 1994 and edged near 70 percent in 2004. In May 2024 it fell to 40.2 percent, losing its outright majority for the first time and forming a government of national unity with the Democratic Alliance and smaller parties. The decline itself is a slow force, three decades in the making, closer to a fundamental than to news. The coalition is something else: a switch, quick to move and wired into everything, able to reset the loops in either direction.
The early evidence is that the switch mattered. The National Treasury attributes a markedly cheaper borrowing environment, the projected stabilisation of debt, and improved credit outcomes to a specific combination: the formation of the coalition, the absence of load-shedding, and falling interest rates arriving together. When several switches move the same way at once, even a slow and settled trajectory can bend. Debt that looked destined to spiral is now projected to stabilise near 79 percent and edge down toward 76 percent by the late 2020s. Recovery is too strong a word for it. What the coalition bought was a switch, held so far in a better position, and switches can be flipped back.
iv. The external hedge, tested and used
The multipolar world stopped being an abstraction for South Africa in 2025. A US executive order in February suspended aid, citing land policy and alleged cooperation with Iran. In April came a 30 percent tariff. Preferential access under the African Growth and Opportunity Act lapsed at the end of September and returned only as a short-term extension into 2026, its value already erased by the surrounding tariffs. Automotive exports to the United States fell by roughly three-quarters. This is a fast, far-reaching shock of the first order, aimed at a structural dependence: a mid-tier economy leaning on a single large market.
What happened next is the part worth studying. The dependence held less tightly than expected because the trade network had already diversified. Displaced volumes were redirected: shipments to India and Brazil rose, other markets absorbed the citrus and the cars, and total vehicle exports reached a record even as the US corridor collapsed. South Africa held a trade surplus with the rest of the world of around 212 billion rand. Non-alignment, so often derided as fence-sitting, functioned as insurance. A country that keeps its options open pays a premium in friction and diplomatic irritation, and collects on the policy exactly when a single partner turns hostile. The exposure to Washington was a size problem; the ability to route around it was an exit-options problem, and the second is what dependence actually measures.
Non-alignment, so often derided as fence-sitting, is really an insurance policy. You pay the premium in friction, and you collect the year a single partner turns hostile.
Read iv · The external hedge
v. The parallel country, being built quietly
Underneath the headline forces, a quieter accumulation is under way, and it is the one most likely to define the texture of daily life. Where the state recedes, private and informal systems fill the space, and once built they do not dissolve. Rooftop solar replaced part of the grid. Boreholes are beginning to replace part of the reticulation. Private security long ago outnumbered the police. Medical schemes and private schooling substitute for public provision for those who can pay. The informal economy absorbs a large share of those the formal economy will not.
Each substitution is rational for the household that makes it, and each one weakens the shared system a little further, because it removes a paying customer, a watchful ratepayer, a constituency for repair. The accumulation of these private exits is a slow, self-contained force, part of the country’s character rather than one of its fundamentals, but it is setting a durable pattern: a functioning country layered on top of a struggling one, sharing the same map. Digital public infrastructure and the productivity of artificial intelligence could, in principle, let the state leapfrog some of this decay and serve the second tier directly. Whether they do, or whether they simply give the first tier one more system to privatise, is one of the genuine open questions of the next decade.
Two loops, running now
The direction of the next decade turns on a contest between two feedback loops that are both live in the record. The scenarios that follow are, in effect, a question about which one wins.
The substitution spiral
Already run to completion in electricity, mid-cycle in water, and now the defining shape of the fiscus. Left alone, it hollows the shared state from the top down and leaves the shell standing.
The confidence compound
Its opening move was the 2024 turn: coalition, no blackouts and rate cuts arriving together bent the debt line and lifted the rating. Sustained, it is what converts the demographic engine into growth.
Both loops draw on the same fundamentals and are steered by the same switches. What separates them comes down to timing: whether reforms in energy, logistics and water compound faster than the private exits, and whether the coalition holds long enough for that compounding to be felt. That framing points to the only honest way to talk about a system this complex: a small set of disciplined scenarios, each written to be checked against the record as it arrives.
The scenario space
None of these is a prediction. Each is a coherent future built on stated assumptions, positioned against the two switches most likely to decide the outcome, and written to be tested against evidence as the decade unfolds.
The fundamentals are shared across all four: a young and growing population, a heavy debt load, ageing service systems, deep inequality, and a diversified but exposed external position. What differs is the setting of two switches.
Figure 07The most probable future and the worst one share a single square. What separates them is whether the loops are left to tip.
Reading the scenario space
The Long Repair
In this future the 2024 turn holds, and keeps compounding. Energy reform holds and generation surplus arrives; the logistics networks are unbundled and freight recovers; the water grant regime forces municipal turnarounds before too many metros tip. Debt stabilises and slowly falls. None of it is dramatic, and growth stays modest through the 2020s, but it compounds. By the late 2030s a widening tax base and reliable services begin, at last, to pull young people into formal work. Inequality narrows from the bottom up.
- Reform durability across energy, freight and water surviving multiple budgets
- Coalition discipline outlasting its founding personalities
- Private capital re-entering infrastructure at scale
- Sustained primary budget surplus and a falling debt ratio
- Grid energy demand rising again as industry returns
- Fixed investment climbing as a share of GDP
- Treat the coalition’s durability, not any single reform, as the load-bearing asset, and protect it accordingly.
- Build for a slow compound: position for a decade of modest growth that accelerates late, not a rebound.
The Substitution Republic
Politics fragments into shifting, transactional coalitions that can block but not build. The state neither collapses nor recovers; it recedes. The substitution spiral runs to completion across every system. Solar, boreholes, private security, private schooling and private health knit into a parallel country that works, for perhaps a third of the population. The informal economy absorbs much of the rest. Growth is low but positive, stability is real, and the two tiers drift further apart on the same map. This is the path of least resistance, and therefore the most probable single outcome.
- Household-level rationality of every private exit
- Coalition arithmetic that rewards blocking over building
- Capital that adapts to a weak state without fixing it
- Private substitution spreading from power to water to safety to schooling
- Municipal revenue collection falling in the largest metros
- Persistent single-digit or stalled fixed investment
- Assume parallel systems, not public ones, and design around resilience you control.
- Watch the second tier: the equilibrium is stable until exclusion finds a political vehicle, at which point it is not.
The Managed Decline
The coalition holds and governs coherently, but the fundamentals prove too heavy: debt service crowds out investment, the service systems age faster than they can be renewed, and no reform compounds quickly enough to change the trajectory. Governance becomes the art of triage. Spending is prioritised, the fiscus is protected, the currency is defended, and the state retreats in an orderly way from what it can no longer afford to do well. Stability without dynamism. A country managed, not transformed.
- Debt-service costs consuming the fiscal space for renewal
- Service systems past the point of affordable repair
- Competent administration without the growth to fund ambition
- Rising interest bill as a share of revenue
- Real per-capita spending on infrastructure falling
- Reform ambition narrowing to fiscal defence
- Read fiscal signals, not political ones: in this future the interest bill is the story.
- Distinguish orderly retreat from slow failure early, because they diverge sharply by the 2040s.
The Fracture
The tail risk, and the reason the framework insists on tracing loops. A single tightly connected system crosses a threshold, a Gauteng water Day Zero, a fiscal financing event, a large metro rendered ungovernable, at the same moment the coalition fragments and cannot respond. Because the systems are connected, the failure does not stay contained. Capital flies, the currency breaks, service failures compound across domains, and localised state failure sets in even as the national state persists on paper. The result is a fracture within the country: uneven, deep, and very hard to reverse.
- Tight linkage between energy, water, public finances and confidence
- Thresholds crossed while political response capacity is absent
- A shock arriving faster than the state’s reaction time
- Reservoir levels and bulk-water arrears in Gauteng metros
- Bond yields and the cost of rolling over debt spiking together
- Coalition confidence votes and provincial ungovernability
- Monitor the linkages, not the levels: the danger is two stressed systems at once, not one bad number.
- Hold genuine optionality in reserve, because in a fracture, exit routes are worth more than any position.
The four futures are not equally likely, and their probabilities are not fixed. The Substitution Republic is the path of least resistance and the most probable single outcome; the Long Repair is available but must be actively won against the pull of the vicious loop; the Managed Decline and the Fracture sit on the erosion side of the capability axis, one orderly and one not. Holding all four is a discipline for seeing, in advance, which evidence would tell you which future you have actually entered, rather than staking everything on one guess.
Set the two switches. Watch the future resolve.
Every future above turns on the same two switches. Move them and the field settles on one of the four, with the loop that dominates, the debt line it implies by 2040, and the first signal that would confirm you had entered it. The fundamentals stay fixed under every setting; only these two move.
Figure 08The lower-right corner, a recovering economy under a fragmented politics, stays empty on purpose: capability rarely recovers while the centre cannot hold.
The confidence compound wins, slowly, and the demographic engine finally engages.
How to read it. This makes the section’s argument adjustable, a way to feel the shape of the space by hand. The four futures share every fundamental; what divides them is where these two switches settle, and how early the first signal shows.
Nine dials, not one number
The method is only useful if it can be run continuously. These are the switches worth watching, the ones whose movement will tell you, before the headlines do, which loop is winning and which future is forming.
A sustained primary surplus and a falling debt ratio is the confidence compound holding. A reversal is the earliest sign of the erosion path.
Rising demand on the grid means industry returning and self-supply plateauing. Continued decline means the substitution spiral still running.
The system most likely to cross a threshold first. Bulk-supply drawdowns and rising non-revenue water are the proximity signals.
The decisive test of whether the demographic engine engages. Youth joblessness that fails to fall even in good years is the diagnostic of structural failure.
The switch that steers every loop. The test is whether it survives its founding personalities and can pass a hard budget, not whether it survives a news cycle.
The cleanest read on whether capital believes the compound over the spiral. It leads growth by years and funds the capacity that ends the blackouts.
The insurance policy. A widening spread of export destinations is optionality being built; re-concentration is exposure returning.
The Fracture signal. Any single bad number is noise. Two connected systems, energy and public finances, or water and the coalition, stressing together is the real alarm.
Whether digital public infrastructure and AI serve the excluded tier or hand the included tier one more system to privatise. The open question under every scenario.
Run these dials for a year and the ambient noise falls away. Reshuffles and quarterly prints stop registering as signal. What remains is a small, legible set of movements that say, quietly and in advance, which South Africa is arriving. That is what it means to read a future while it is still forming.
Africa is the most forecast, and the least decided, continent.
South Africa’s next twenty years are already accumulating. They can be read now, and what is read can be shaped. Diagelo works at that seam, between the forces already in the ground and the futures still open.
Start a conversation