When using retention rate as a metric, it might also be important to include the reason for leaving as a factor.

For example, software engineers at tech giants might typically leave because they're bored, whereas employees in the Amazon warehouse might leave because they're burnt out and overworked.

So the government should routinely measure hormone levels of almost everyone, and levy fines against employers which will inevitably be passed on to employees or consumers. No thanks.

No, just the opposite. The idea that cortisol levels are a direct indicator of stress levels is a myth. Stress is just one modulator of cortisol levels.

Measuring cortisol is extremely imprecise. Cortisol also has huge variability from one patient to another and depends heavily on time of day relative to a person's circadian rhythm, among other variables.

Cortisol measurements aren't even used clinically outside of conditions which cause extreme excesses or shortages of cortisol. For example, Cushing's syndrome or Addison's disease.

Consider that the reference range for morning plasma free cortisol has a lower limit of 5 ug/dL and an upper limit of 25 ug/dL. Obviously, you can't draw any conclusions about one's workplace stress if the morning reference range in normal, healthy volunteers already spans a 5X range.

Finally, the idea that cortisol == bad is just wrong. Cortisol is just one marker of a complex system. If you're in a stressful situation, you want your body to react with an appropriate amount of cortisol to help you deal with it. Cortisol only becomes problematic when stress is chronic day and night, or when your natural cortisol feedback mechanisms are degraded as might be the case in major depressive disorder.