This pattern shares its parent architecture with Privatised Gains, Socialised Losses — both describe costs that land on the public purse — but the mechanism here is different. That pattern is about who captures the gain while the loss is socialised. This one asks a narrower question: why does the state so rarely see the loss coming, even when there was no gain being captured by anyone at all. It is a specific, concrete instance of The Temporal Displacement of Cost and Accountability, still waiting to be written — and once it is, this pattern belongs directly underneath it. It also stands in a useful tension with The Buffer: that pattern insists buffers should be sized to the risk. This one describes a category of risk you cannot size a buffer against, because nobody has agreed it exists until the day it does.

Government liabilities that are never written into law are also never written into a budget — and the state honours them anyway, which means the most dangerous fiscal risks are the ones no ledger admits exist.

Explicit liabilities are the easy kind. A loan has a repayment schedule. A pension has an actuarial valuation. A guarantee has a face value written into a contract somewhere. Someone, at some point, was required to put a number on the obligation, and that number now sits in a register that finance ministries and rating agencies both watch.

Implicit liabilities carry none of that discipline, and most of a government's real exposure to water-related disaster lives here. A landslide buries a village. A season's crop is wiped out by flood. A city loses its drainage capacity in a single storm that was one notch above what anyone had planned for. In not one of these cases does a statute say the state must pay. And in almost every one of them, it pays anyway — because a government that let a buried village go unanswered, or a ruined harvest go uncompensated, would not survive the politics of having done so, whatever the law says. The obligation is real. It is simply nowhere written down.

This is not a technicality. It is the reason implicit liabilities are the dangerous half of a government's balance sheet rather than the reassuring half. A debt has a repayment date circled on someone's calendar. An implicit liability has no date, no register entry, and no line in the medium-term fiscal framework — right up until the morning it becomes the only thing anyone in the finance ministry is discussing, and the year's fiscal targets are gone before lunch. What was invisible in the planning becomes, without warning, the largest number in the room.

There is a second, quieter injustice folded into this pattern. The liabilities least likely to be recognised in advance are the ones owed to people least able to make noise about them — see The Poor Don't Count. A well-connected industrial zone that floods gets a rapid, well-publicised, well-costed response, and its risk was probably modelled years before the water arrived. A flood-prone village of smallholders was never in anyone's fiscal model to begin with, not because its eventual claim on the state is smaller, but because nobody with a seat at the budget table was accountable for writing it down.

The household analogy is useful precisely because it shows the fix is not exotic. A family pays for a broken window out of this month's income, because it is small and expected. A family does not pay for a house fire out of savings, because it is rare and catastrophic — which is exactly why insurance exists, priced in advance, as a visible planned cost standing in for an invisible unplanned one. Governments routinely fail to do for landslides and crop failures what any insured household already does for fire. They neither price the risk nor set anything aside against it. They simply wait for the invoice.

Before the next budget cycle, list the disasters the state will respond to whether or not it is legally obliged to — floods, landslides, crop failures, drainage collapse — and treat that list as a liability schedule, not a footnote. Assign each item a rough expected cost, however uncomfortable the number, and hold a reserve against it in the medium-term fiscal framework the same way an explicit debt is provisioned for. Bring implicit liabilities into the same register, the same review cycle, and the same level of scrutiny as explicit ones — not because the number will ever be exact, but because an unrecorded liability is not a smaller risk than a recorded one. It is the same risk with worse warning.

For what happens once no such reserve exists and the bill arrives anyway, see Relief Crowds Out Prevention — and, once written, The Vulnerability Spiral, which describes how that scramble for cash compounds the very vulnerability it was drawn from.

Connected patterns: Privatised Gains, Socialised Losses — The Temporal Displacement of Cost and Accountability — The Buffer — Relief Crowds Out Prevention — The Poor Don't Count