I am working with individual-level employment data and need to control for industry-specific shocks, but cannot use fixed effects (the variable of interest is time-invariant at the industry level), and I don't have data on sectoral sales. The only other thing I can think of is to control for industry employment growth. I know this is a big no-no if I want to identify the effect of industry employment growth (Angrist's perils of peer effects), but I'm wondering whether it would be ok just as a control.
