Causality can’t be established from observation only. Observational studies and surveys can point to correlation, but experiments are needed to determine causality.
It’s practically impossible to conduct experiments to answer social science questions. You can’t tell countries what policies to implement for your study of their economic effects. So you observe, look at past data, find associations, and debate with scholars whether that policy causes that particular effect — whether improvement in the education system drives economic progress, or economic progress improves the education system, or another variable causes both.
Hard sciences are built on experiments. As long as experiments are designed well, the conclusions about causality are valid.