The EPS currently represents labor productivity changes in two ways:
Policy packages' jobs impacts in future years are discounted by the anticipated rate of labor productivity growth (since the jobs impacts are derived from inflation-adjusted dollar changes in output, and more dollars of output are created with less labor when the labor is more productive). This is handled in io-model/LPGRbIC.
The total number of employed people (projected through the end year of the model run) is used to help calculate the change in average compensation per employee. This is handled in io-model/BPEaCP.
We don't have an explicit representation of BAU labor productivity in the model. We could easily calculate one by dividing total employees by total output (both of which we have by ISIC code). Then, we could add a set of R&D levers to accelerate labor productivity improvements. This would accompany the existing R&D levers that accelerate fuel efficiency improvements and capital cost improvements.
Productivity improvements should not necessarily reduce the total number of workers, as the workers may re-integrate elsewhere in the economy, keeping io-model/BPEaCP the same, apart from some noise. (We saw no big changes in the employed percentage of the population over the past 50 or more years despite huge labor productivity improvements in that time.) But it should boost output in the industries with improved productivity, which in turn may increase their demand for non-labor inputs (e.g. raw steel, etc.). And it would redistribute workers to different ISIC code buckets, which would then change the output of the buckets receiving the workers. To implement the labor productivity R&D levers correctly, it would be necessary to think carefully about how to represent in the EPS labor productivity improvements that are not accounted for in the BAU input data.
The EPS currently represents labor productivity changes in two ways:
Policy packages' jobs impacts in future years are discounted by the anticipated rate of labor productivity growth (since the jobs impacts are derived from inflation-adjusted dollar changes in output, and more dollars of output are created with less labor when the labor is more productive). This is handled in
io-model/LPGRbIC.The total number of employed people (projected through the end year of the model run) is used to help calculate the change in average compensation per employee. This is handled in
io-model/BPEaCP.We don't have an explicit representation of BAU labor productivity in the model. We could easily calculate one by dividing total employees by total output (both of which we have by ISIC code). Then, we could add a set of R&D levers to accelerate labor productivity improvements. This would accompany the existing R&D levers that accelerate fuel efficiency improvements and capital cost improvements.
Productivity improvements should not necessarily reduce the total number of workers, as the workers may re-integrate elsewhere in the economy, keeping
io-model/BPEaCPthe same, apart from some noise. (We saw no big changes in the employed percentage of the population over the past 50 or more years despite huge labor productivity improvements in that time.) But it should boost output in the industries with improved productivity, which in turn may increase their demand for non-labor inputs (e.g. raw steel, etc.). And it would redistribute workers to different ISIC code buckets, which would then change the output of the buckets receiving the workers. To implement the labor productivity R&D levers correctly, it would be necessary to think carefully about how to represent in the EPS labor productivity improvements that are not accounted for in the BAU input data.Suggested by Skip Laitner on 7/14/2020