One familiar calendar is being prepared around the world based on changing assumptions. A less familiar one approaches the assumptions from beyond the time range. Both should affect the budget. In short, this article challenges four core assumptions before they break.

It’s time for another budget season. From opening the templates to board approval, the budget will occupy the organisation from a couple of weeks to a couple of months. The other calendar is labelled “IFTF 10-Year Forecast 2036 - What we choose next.” Ten recently published forecasts are to be discussed in Palo Alto at the end of October. Most FP&A teams ignore them with sound reasoning. Nothing in a 2036 forecast will change a Cost Centre budget for next year. The reasoning is sound, yet incomplete. The IFTF (Institute for the Future) intention is not to describe 2036. It is to describe the erosion of assumptions we are using now, in this budget cycle and in this workbook, without ever having written them down.

The Layer Beneath the Premises

Every budgeting cycle I ran, saw, and designed used an initial guidance document with instructions, developments, and assumptions: FX, interest rates, inflation rates, wages, attrition, actuarial assumptions, training expectations, commodities, taxes, performance, etc. Each line carries a number, usually a source, sometimes an owner.

Underneath the guidance sits a second, undocumented layer. If it were written down, it would contain statements like:

• Our inputs are what they claim to be: bottom-up, we accept a bit of a “conservative” view; top-down, you can expect a stretch to come.

• One FTE is a stable unit of capacity, and no, we still won’t count interns, apprentices, or agents as FTE.

• Risk can be transferred to an insurer at a price that moves roughly with inflation.

• Capital is available and its cost is a function of rates.

They sound like ground truths. Maybe that’s why none of them has an owner, a source, or an expiry date, or why nobody ever wrote them down. Yet, the 10 IFTF forecasts put doubt on all of those. That may be precisely what makes them useful to FP&A, especially when the planning process is at least somewhat aligned with strategy. They are a list of exactly this second layer, seen from far enough away that the erosion becomes visible. Let’s take a look. Each of the ten forecasts is assigned to one of those points above.

The Undocumented Premise That Our Inputs Are What They Claim to Be

IFTF Forecasts: Discernment · The Future of Scientific Knowledge · From Large Language Models to Large World Models

FP&A uses numbers that are hard to verify: growth rates, commodity outlook, peer benchmarks and interest rates. Verification stayed cheap because production was expensive, since someone had to do the real work to publish the number. Sometimes they were due to be decided shortly before or after the plan, take actuarial rates for your pension accruals, for example.

That relationship is inverting. Plausible findings, citations, and charts are now cheap to produce. Provenance becomes the scarce good, not the number. There is not much glamour in verifying the source, rather than relying on the report that quoted it or the link provided in the AI-assisted step. Trusted sources, human release points and a release log help.

One good question: which of our external assumptions would survive if we had to trace them to a primary source this week?

The Undocumented Premise That a Full-Time Equivalent Is a Stable Unit of Capacity

IFTF forecasts: Four Cultures of AI · Loneliness and Aloneness

Headcount planning rests on a conversion that has held for decades. One FTE equals roughly one unit of output at roughly one cost. Attrition and wage growth are the two dials we turn.

The forecasts target conversion, not dials. IFTF's four cultures — AI-free, AI-augmented, AI-adjacent, and AI-directed — describe teams within the same company operating at materially different output per head and cost structures. Not in 2036. Already. The forecast on aloneness reaches the same headcount from another side: retention, absence, and the cost of teams that no longer cohere. That implies potential risks for headcount and personnel cost planning, as well as for productivity drivers, esp. if output is hard to measure.

One good question: does our headcount plan still convert people into capacity using a single factor, and how can we notice any drift?

The Undocumented Premise That Risk Can Be Transferred at a Price That Moves with Inflation

IFTF forecasts: End of the Insurance Bargain · Life in Extreme Heat

This cost family may be roughly planned with a percentage increase in overhead or, like energy, affect more than one line item from COGS to facilities and vehicle costs. Using a forward curve may be a good indicator of assumption maturity, whether the assumption is material or not. Both methods assume a market that prices risk and keeps offering hedges.

Where cover retreats, the cost moves. Into self-insurance. Into higher deductibles. Into capital expenditure for resilience. Into working capital held against interruption. It also may move out of OpEx and into CapEx, which changes the depreciation profile, the hurdle rate conversation, and eventually covenant headroom. Think this is too abstract? Take a look at MunichRe and the effects of their shift in reinsurance policy.

Climate and technological developments that widen insurance gaps or produce terms like “data heat island” do the same to inputs FP&A treats as constants: productive hours, equipment uptime, cooling load, logistics reliability, and site selection. Maybe it’s time to add resilience as a CapEx category.

One good question: if one insurance line were not renewed, which budget line would absorb it, and do we know the amount?

The Undocumented Premise That One World Market, Predictable Currencies, Available Capital

IFTF forecasts: The Muddle Powers · China Is Everywhere · After Financialization

FX rates were applied with confidence in the past. Unless a global crisis happened, or interest was due in Swiss Francs by the end of 2016. Then those uniform tables of constants, revised once a year, broke in a day.

In general, FX encodes an assumption about monetary order that these three forecasts question from three directions: where authority migrates when energy rather than ideology sets the table; where Chinese capital and technology reshape markets that never considered themselves downstream; and what remains of the real economy of goods and services after financialisation has run its course. The implications go beyond markets; think of transfer pricing, intercompany logic, cost of capital, hurdle and business case discount rates, and impairment tests. You surely have this present for your company.

One good question: which single currency pair moves by more than 20% would force us to work overtime, and who owns watching it?

From Futures Questions to Action

Budget season has no room to update strategic foresight programs and assumptions, or to add scenarios to a cycle that is already late, only to make it later. So, what four small moves seem valuable enough to surface at short notice in the existing calendar?

Write the second layer down. Add one page to the premises document holding the structural assumptions in plain sentences. Statement, owner, source, date last examined. Nothing more. The page will be uncomfortable to write. That discomfort is the finding.

Separate extrapolation from structural bets. Two columns. Wage growth of 3.2% is likely an extrapolation anchored in history, and it will be roughly right or roughly wrong. "Insurance remains available at market price" is a bet that is entirely right until it is entirely wrong. Two kinds of assumptions, two kinds of treatment. Find them, separate them, and treat them if not already done.

Use watchpoints instead of new scenarios. A scenario may cost weeks, a watchpoint surely costs an hour: find the observable indicator, threshold, named owner and response agreed in advance. Watchpoints across the four topics will serve next year better than a scenario exercise added to prime time.

Ask one backward question per planning area. Not to follow the "what will 2036 look like" essayist route, assume the premise has broken by 2036. What would have had to become visible by the end of next year? The answer is a watchpoint, and it fits in the budget you are already building.

Scenarios From the Future, Not Into the Future

Budget season works forward. Actuals, applied growth, adjusted for known events, flexed, stretched, and rolled forward 15 to 18 months. That extrapolation is a useful instrument for aligning the organisation within the range. That’s why many organisations keep doing it, even though alternatives like Beyond Budgeting and “true” Rolling Forecasts exist.

Foresight works the other way. It starts in 2036 and goes back in time. Not to change next year's number, which it rarely does, but to show which numbers you would defend under pressure, and which you would abandon in the first bad week of a bad quarter. That is worth writing down a page.