They are meaning experiments in the sense of adjusting the state of the thing being experimented upon. Astrophysics and economics experiments are both limited to adjusting the way we collect the data, not the subject itself (you can’t practically poke a star or an economy with a stick). Presumably the reason they’re using this line in the sand is that it is the key technique to separate correlation from causation, which is a limitation that tends to be more of a problem in economics than astrophysics (but certainly does frustrate in both fields).

The real problem of course is that economics is like astrophysics with only 1 star (or maybe 100 stars that bump into each other constantly… the analogy is imperfect)