I’m not sure if this is your point, or just that it’s population level, but there is certain to be lag in the measure. I.e. if I contribute $10B to GDP in Q3, and lay off 10% of the workforce, my productivity looks great, but a large portion of that contribution will stem from work done by a larger workforce in Q2, Q1, and before, perhaps well before.

So the measurement during periods of recession or expansion will always be artificially elevated or suppressed.