In short, this is probably because climate economic modeling has traditionally focused on estimating more knowable, quantifiable impacts, and acknowledging but not including potential (maybe even probable!) impacts that are too difficult to model based on current knowledge. Tail risk is a big area that has traditionally not been modeled well leading to a very likely underestimate to the social cost of carbon (SCC): https://www.volts.wtf/p/economists-have-quantified-the-econo...

Here’s a historical overview of the situation with some links to the latest research: https://impactlab.org/research-area/social-cost/

It’s not just tail risk, even things like geopolitical instability and crop failure were often not historically included… you can see why the headline statement seems to conflict with intuition, and it’s because of the narrow view taken on the impact of the fossil fuel scenario on GDP.

There is one more important element, and that is the distinction between economic activity (which GDP measures) and economic welfare. Natural disasters often don’t have much of an impact on GDP in part because the rebuilding activity counts in GDP. But it’s obvious to see why a future where we are spending an increasingly larger share of our effort to mitigate and rebuild losses is a much worse one. So GDP may not be the best measure here: https://www.economy.com/economicview/analysis/296804/How-Nat...