I don't know where you get that economics is based entirely on modelling. Economic decisions happen constantly on all levels, and their effects are studied heavily.
The models follow from observation, the same as they do in chemistry or meteorology. And just like those two, the models are weak compared to actual observation.
But you can't study the effects directly. Economic system are complex, any changes made are impossible to isolate and any changes in economic indicators have to be assumed to be related to the changes made.
I argue that economic research is entirely based on modelling because the research results produced always depend on modelling and either don't include any type of control group or use a controlled study so small that the odds it accurately reflects the entire economy are low.
The first I think is self explanatory, for the second I'd point to studies of UBI that are done on small populations within a larger economy that doesn't have a UBI. The results of the small group are effectively meaningless because that group has a comparative advantage against the rest of the economy, and those studies are never able to create a bubble in which the UBI population is entirely cut off from the real economy while still mimicking well enough all other systems, industries, and motivations of the real economy.
We can forecast the weather pretty accurately 7 days out, and we have never had a "control" climate or the even the ability to run experiments. We can predict climate effects decades out riding the same theories.
I'm sure you would agree that climate change is pretty hard science (it is) so you must have a misconception or confusion about economics, because really they are the same kind of science at heart.
Models based on observation are used to predict future behavior, and while often wrong, it's critical to understand why they are wrong so you can understand where they are right.
Weather predictions are very different than economic predictions.
Weather predictions are made and shared out only at the level of what will happen over the next few days. Weather predictions aren't used to predict exactly why it rained, only that it did rain (though surely the why is used as feedback for modelling).
Economists take a different tact, they attempt to claim why GDP, CPI, wages, etc changed. With predictions they look much further out than weather and attempt to not only claim what economic indicators will do, they attempt to claim precisely why it will have happened.
I'm sorry but I think you have a lot of misguided confusion here. Climate modeling encompasses everything from the weather tomorrow to scientists telling us back in the 80's to cut CO2.
Economics modeling is the same way. From how to price your lemonade at your kids stand to how to size government stimulus.
I wasn't arguing anything about climate modelling, only weather predictions.
Regardless, modelling can be interesting to use as an indicator for future research but modelling is never scientific, or more specifically its never scientific in a way that is in anyway meaningful in the real world. A model isn't testing against reality and only ever factors in the conditions someone though to include and deemed relevant, and those factors are only as useful as the person's ability to accurately code them in.
That's also not true. By now there are so many similar but different situations that you can look at. And there are so many massively multiplayer games which are fantastic economic worlds where again, a multitude of ideas can be tried and tested - and are.
The models follow from observation, the same as they do in chemistry or meteorology. And just like those two, the models are weak compared to actual observation.