You can't, and you're spot on, it's one of the things the IEA doesn't get. Don't ask for a full sensitivity analysis from the IEA, such rigorous scientific methods are beyond them. The problem is, if the IEA did sensitivity analysis, they could not make any scenarios and would realise the future is extremely uncertain. So they do highly limited sensitivity analysis, on only one or two variables, so they can jump to conclusions. The IEA has radically altered their predictions/scenarios/projections from a few years ago as they couldn't predict even short term development in energy markets. The 2030 scenario is a good laugh at best. I wouldn't be surprised if they'll radically change their 'business as usual' scenarios next year and this prediction will again be way off.

Levelised cost of electricity estimates are also highly dubious; again no sensitivity analysis whatsoever. Already the cost vary greatly with geography even today. Concentrating solar? Amateuristic nomenclature.

We cannot predict the future with any reasonable accuracy or credibility. However, we can help invent the future. The IEA should put their energy into how best to do that, rather than these highly unusual 'business as usual' scenarios.