In one answer in this great discussion on efficiency wages it is claimed the reason that efficiency wages fell out of fashion is that we can't measure it in normal data. I agree that it is difficult to work on this, but disagree that it is impossible. I know some papers that do and would like to open this thread to crowdsource further examples of papers that do manage to answers empirical questions related to efficiency wages using observational data. Having examples of papers that do this might help to know what kind of data and design is needed to tackle this.
Here is the start:
This is maybe more related to reference-dependent wages but the approach is still relevant Camerer, C., Babcock, L., Loewenstein, G., & Thaler, R. (1997). Labor supply of New York City cabdrivers: One day at a time. The Quarterly Journal of Economics, 112(2), 407-441.