To my eyes it looks like “real wages” [(nominal average hourly earnings)/(NGDP/pop)] lead unemployment by about a month or two
Shock me, shock me. Let’s see:
NGDP = RGDP * P = N * (RGDP/N) * W*(P/W)
In words, nominal GDP = employment times output/worker times nominal wages times the price markup.
Solve this for the ratio of the nominal wage to nominal GDP:
W/NGDP = (W/P) * (RGDP/N)/N
In words, Sumner’s “real wage” (the nominal wage divided by nominal GDP) equals the inverse of the price markup times the inverse of productivity times 1/employment. If the price markup and productivity remain about unchanged, then by definition the “real wage” is inversely related to employment.
Scott is fond of saying, “Never reason from a price change.” I say, “Never draw a behavioral inference from an identity.”
Since the wage data Scott uses is collected by survey, no, it is not merely definitional.
Suppose I was issac newton trying to postulate about gravity. I measure masses and forces abd declare these two things are connected by a universal constant and behave as the inverse square if the distance. Then we have an equation. Kling enters the room and declares: ah well that discovery was just definitional.
No, recognizing that NGDP decomposes in that way is a discovery because its not obvious there would be a stable relationship.
That’s an equation, not an identity.
Any reason why the price markup and productivity should remain unchanged?
So, let’s do it this way.
Nom av hrly wages = Total compensation in the GDP/total hours worked.
Then (Nominal average hourly wages)/(NGDP/pop) =
(Total comp/NGDP) / (Tot hrs worked/pop)
If (Tot comp/NGDP) is fairly constant over this cycle (which it is), all we are looking at here is a correlation between the employment to population ratio and the unemployment rate. I’m with Arnold on this one.