What is the economic justification for government production of weather forecasts? The question has gained particular relevance during the first and second Trump administrations, and this is my place to collect and clarify economic reasoning related to the question, focusing on “classic” economic rationales for public policy.
I’ll divide the government’s possible contributions into four main areas: (1) gathering and sharing weather and climate data, (2) basic research on weather forecasting, (3) production of operational forecasts, and (4) dissemination and outreach.
My impression is that people widely agree that (1) should be done by the government in conjunction with international organizations like the WMO. For example, even though Project 2025 calls for privatization or breaking up many weather and climate related functions in NOAA it does not call for privatizing NOAA’s weather data gathering. (Though it does call for shutting down many climate-related efforts at NOAA which would likely have the effect of impairing weather-forecast-relevant data gathering.)
For (2), the standard arguments for government involvement in basic research also apply to weather forecasting. Matt Clancy has an excellent FAQ on government involvement in R&D. These arguments would seem to be especially strong for climate modeling, an area of research that the private sector might be especially unlikely to invest in due to time scales, coordination challenges, and all the other reasons it is challenging for the private sector to tackle climate change. Thus it is an area where government R&D would be especially unlikely to crowd out private sector R&D.
Outreach (4) is the area where, historically, the government has played an intentionally smaller role, leaving more to the private sector, though the boundary had evolved over time (for a history, see Henson (2010)). It isn’t obvious to me where this boundary should be drawn, and something strikes me as strange that the status quo has long been one where the NWS produces forecasts that then basically get recycled and sold by private companies. The (imperfect) analogy that comes to mind is of the government providing free oil to companies that then get to keep all the Ricardian rents. I still need to think about this. In any event, outreach is not the main locus of the debate on privatization today.
That locus is instead on area (3), the production of operational forecasts. There are three central and related economic arguments for government production of operational forecasts that I see: non-rivalry, private-sector under provision of information, and public safety concerns with market-based allocations.
The argument from non-rivalry
A common argument for government issuance is that weather forecasts are public goods (for examples, see here, here, here). Weather forecasts indeed have one characteristic of public goods – they are non-rival – but they are in principle excludable (which relates to public safety arguments for government provision, discussed below). Given this, forecasts are more accurately characterized as club goods. When thinking about government provision of weather forecasts, however, non-rivalry is, in a certain sense, the more crucial of the two characteristics. An efficient market sets price equal to marginal cost. The marginal cost of one additional person consuming a weather forecast is zero. So the price should be zero. Plenty of private sector companies give away forecasts for a financial price of zero, but free goods aren’t free.
The argument from information economics
Weather and climate data are information, and Grossman and Stiglitz (1980) arguments caution that the private market will tend to underinvest in gathering information. The argument is largely standard: gathering weather information is costly and trading on the information would reveal the information, so incentives for gathering the information are reduced. There is a bit more nuance to the applicatinon of this argument to weather forecasts. I am partial to the Timmermann and Granger (2004) view of information which includes the model used to produce forecasts. In that sense, weather forecasts themselves are information, so the same argument applies.
The argument from public safety
There are many ways to think about the economic justification for government weather forecasts as it relates to public safety. Like the post office, there can be a societal objective for universal coverage for a service regardless of willingness to pay.
Having a single entity responsible for issuing extreme weather warnings helps with coordination. Issuing too many warnings risks a “cry wolf” effect where people stop paying attention. And conflicting warnings might make it hard to understand, an issue that the NWS already spends considerable effort grappling with.
There are many markets that are disallowed because we find them repugnant. During the times when we might most want forecasts to be available for moral reasons (risk of disaster for instance), the market value would be highest. We have plenty of examples of cases where we don’t allow firms to gouge during times like these. Even if we had private forecasts, we would probably enact similar policies, which would lower the price forecasters could charge and which would disincentivize private investment. This also relates back to the idea of forecasts as public goods. Even if forecasts are in principle excludable, we might to disallow exclusion in the interest of public safety, a point memorably made in the Last Week Tonight episode on weather.
Come to think of it, that LWT episode did a better job laying out all of these issues than this post, so just go watch that.
Version history
2025-06-10: First version
2025-06-12: Expanded intro, edited all points