Insight, Climate Risk & Institutions
Hungary's central bank is considering putting climate scenarios into monetary policy after its driest summer in over a century. Climate models run to 2100. Monetary policy runs to two years. That mismatch is not a communication problem, and it has a name.
Hungary's central bank has said it is considering incorporating climate risk scenarios into its monetary policy framework, after record heat and drought severely affected Central European agriculture (as reported). Hungary recorded its driest summer since the beginning of the twentieth century, and the bank now sees climate-driven food shocks as a potential inflation risk for 2027.
The methodological problem is stated plainly in the reporting and is more interesting than the announcement. Climate models operate over decades. Central banks generally make decisions over roughly eighteen to twenty-four months.
That is not a gap in knowledge. It is a gap between the timescale on which the evidence is produced and the timescale on which the institution is permitted to act.
Mark Carney gave the general problem its name at Lloyd's in 2015. In Breaking the Tragedy of the Horizon, he argued that the catastrophic impacts of climate change will be felt beyond the traditional horizons of most actors, imposing a cost on future generations that the current generation has no direct incentive to fix. He was explicit about the specific numbers involved, noting that the horizon for monetary policy extends out to two years.
The tragedy, in that framing, is that by the time climate change becomes a defining issue for financial stability, it may be too late to do much about it.
What is happening in Hungary is a partial escape from that trap, and it is worth understanding why, because it generalises.
The bank is not being asked to act on a projection for 2070. It is being asked to act on a drought that has already happened, whose effect on food prices will arrive within its actual forecasting horizon. The physical risk crossed into the decision window on its own, without anybody having to solve the horizon problem.
Which suggests the practical question is not how to make institutions think longer. It is where a long-horizon risk has already produced a short-horizon consequence, because that is where evidence can enter a decision without anybody changing their mandate.
Food price volatility is the clearest current example. Insurance repricing is another. Infrastructure outage frequency is a third. In each case the physical driver is slow and the consequence is annual.
Take one credible projection of increasing drought frequency in a region and hand it to five institutions.
A central bank asks whether it changes the inflation path over eight quarters. Mostly it does not, until a specific harvest fails.
A commercial bank asks whether it changes the credit quality of an agricultural loan book over three to five years. It might, and the response is to reprice or withdraw, which is a decision the farmer experiences well before the drought does.
An insurer asks whether it changes next year's loss ratio. The response is annual repricing, and the effect on the customer is immediate.
A municipality asks whether it changes a capital programme over a political term. Usually it loses to something with a ribbon.
A farmer asks whether to plant differently this season, and faces the transition trough that makes a good long-run decision unaffordable in year two.
The same information produces five different answers, and none of them is wrong. Each institution is applying its own horizon correctly. The system-level outcome is that the actors who can act fastest are the ones whose action transfers risk to somebody else, while the actors who could reduce the physical risk have the longest horizons and the weakest mandates.
That asymmetry is the part worth studying. It is not that nobody responds to climate evidence. It is that the first responses are financial and defensive, and they arrive years before the physical event.
Climate services are largely built to answer the question a scientist finds interesting, which concerns the magnitude and attribution of a physical change. Institutions need something narrower and more awkward.
A trigger rather than a trend. Not the probability distribution of future drought, but the specific observable that should change a decision, and the threshold at which it does. That is a different product and it can only be specified with the institution in the room.
A horizon-matched version. The same underlying science expressed at eighteen months, five years and thirty years, because those are different customers and a single report serves none of them well.
The second-order effect, not the first. A central bank does not need a rainfall projection. It needs a food price pass-through estimate. A bank does not need a drought map. It needs a default probability. Most climate services stop one step before the variable the institution actually uses.
Honesty about what is not attributable. A single dry summer is weather. The institution has to act anyway, and a climate service that will only speak about thirty-year trends leaves the decision-maker to improvise the connection themselves.
Decision use, traced. Which institutions changed a decision after receiving climate information, what the decision was, and whether it would have been different otherwise.
Time from physical signal to institutional response, by institution type. Insurers, banks, central banks, municipalities and farmers, measured against the same event. The ordering is predictable and the intervals are not, and nobody has assembled them.
Who bore the cost of the fast responses. When an insurer reprices and a bank withdraws credit, somebody has absorbed a transfer. In an agricultural region that is the farmer, and it happens before any adaptation funding arrives.
Whether the mandate changed or only the rhetoric. A central bank that publishes a climate scenario and does not alter a single decision has done communication. The test is a decision that went differently, and it should be identifiable.
Hungary's central bank is worth watching not because it will change interest rates over drought, which it probably will not, but because it is one of the first cases where a long-horizon physical risk has arrived inside a short-horizon mandate on its own. How the institution handles that will be instructive for everybody still waiting for the horizon problem to be solved in theory.
The Lab works on this through measuring change and on the conditions under which evidence reaches a decision through entering a new context.
If you are producing climate evidence for an institution with a shorter horizon than your model, tell us what you need to know.
This is an independent insight piece by Transitions Lab. For the Lab's applied work, see Measuring Change. See also The Trough Before the Dividend on why a good long-run decision can be unaffordable in year two, and Strategic Is Not the Same as Financeable on decision rules that a subsidy does not move. To discuss a study, see Contact.