I’m trying to better understand the debate around the productivity-wage gap, and in particular the argument made in Robert Z. Lawrence’s 2015 PIIE article “The Growing Gap between Real Wages and Labor Productivity.”
As I understand it, Lawrence argues that the apparent decoupling of wages and productivity since the 1970s is largely a measurement issue. When wages are defined as total compensation (including benefits), and both compensation and productivity are deflated using the same business-sector price index (rather than CPI for wages), labor compensation appears to have tracked productivity reasonably well at least until around 2000.
I think I understand why using a common deflator is appropriate for accounting questions about labor’s share of output, and I don’t disagree that, on that basis, the standard wage-productivity gap graph can be misleading.
What I’m less clear on is how this argument relates to the broader claim that is often made in public discussions, which is that workers have not shared in productivity gains driven by technological progress and capital deepening. My impression is that Lawrence’s analysis addresses whether labor’s measured compensation kept pace with output, but does not directly engage with the distributive or welfare question of how gains from technology are (or should be) shared between labor and capital.
Relatedly, when people talk about wage stagnation they often seem to mean worker material well-being, in which case deflating wages by CPI (even if it differs from the output deflator) seems more directly tied to lived purchasing power.
So my question is: Is it fair to say that Lawrence’s argument is largely correct within its accounting framework, but answers a different question than the one it is often invoked to address in discussions about worker welfare and the distribution of technological gains? Or am I misunderstanding how economists connect these concepts?
*Edited to add link to article*
Does the productivity-wage gap depend on what question you’re asking?
byu/AidMMcMillan inAskEconomics
Posted by AidMMcMillan
1 Comment
I agree with you that the choice of deflator depends upon what question is being asked. There are two related but different questions here, one is whether compensation is increasing at the rate of productivity improvements and the other is how much this benefits workers.
I think the evidence is pretty strong that productivity improvements and total worker compensation has been broadly in line. The degree to which your median worker benefits from this is a bit more complicated.
I want to add one more thing which is very technical but also very important. CPI is inappropriate to use for comparisons over long periods of time.
Firstly CPI-U, has several methodological issues and so you should instead use CPI-U-RS which fixes them.
However, even that has a major methodological issue which is that it is not chained and so doesn’t properly reflect the ability of consumers to switch from expensive to cheaper products.
This makes little difference over short periods but gets increasingly more important the longer the time period.
There is no long run version of CPI which is chained.
PCE is a similar measure of inflation which has the advantage of being chained. The issue is that it includes in the basket products bought on behalf of consumers (by gov and non profits) and not just what consumers buy themselves directly (health care being the most obvious difference there and that is weighted much higher in PCE).
So, if you’re talking about the value of wages rather than total compensation there is an argument to make that CPI-U-RS has the better basket. But it not being chained is a major flaw and we know that incorrectly will overstate inflation and therefore understate the purchasing power of wages. Therefore there’s a strong argument for using PCE despite its different basket.
For comparisons since 2000 chained CPI is available and probably the correct deflator to use. https://fred.stlouisfed.org/series/SUUR0000SA0