The data show that the use of HPWO practices grew considerably in the years following 1992, with the only exception being teams, which remained at the same level.
In analyzing the 1992 survey I (arbitrarily) credited an establishment with being an HPWO if it had two or more practices in place with 50% penetration for each.
Do employees in those establishments in 1997 seem better off than employees in establishments that had not adopted HPWO in 1992?
Finally, the variable I use in the first set of estimates to measure HPWO is the number of practices in 1992 in which 50% or more of the core employees were involved.
For each outcome I first estimate a simple model that contains only the HPWO variable, the control for sales growth, and industry dummy variables.
The central conclusion from these models is that the presence of HPWO practices in 1992 is associated with a higher probability of layoffs in subsequent years and with no gains in real wages.
One legitimate concern about the foregoing is that there may be something else at work that leads firms both to adopt HPWO practices and to lay off employees or not improve their pay.
Another possibility is that the inclusion of controls for sales gain masks the benefits of HPWO practices.
As a final check on these results I re-estimated the models using each HPWO practice separately.
First, when I interacted the HPWO variable with the union variable (in this case specifying the HPWO variable as a dummy that took the value of "1" if two or more practices were in place in 1992 at a 50% level of penetration), the result was that the presence of a union sharply reduced the probability that HPWOs were associated with layoffs.
Second, and very important, the fact that HPWO practices are associated with layoffs does not mean that employment did not grow at these establishments.
In summary, the central conclusion from these models is that the presence of HPWO practices in 1992 is associated with a higher probability of layoffs in subsequent years and with no gains in real wages.