The productivity measure computed by the BLS doesn’t refer to the marginal productivity of labor that relates to the wage rate. Here is a brief BLS explanation of its method of computation:
http://www.bls.gov/lpc/iprread1.htm
And here is the technical explanation:
http://www.bls.gov/opub/hom/homch11.htm
The BLS computation of wages and real wages is also problematic as a metric for assessing actual markets:
Here is the BLS explanation of the method:
http://www.bls.gov/news.release/realer.tn.htm
Here’s the technical details.
http://www.bls.gov/opub/hom/pdf/homch8.pdf
The short answer is that “productivity” in the chart does not refer to the productivity that relates to wages. So the alleged relationship shown in the chart is a statistical artifact. It does not show that workers are working more productively and yet not being paid more. To show that, one would have to look at a single production process to determine what the change in physical production of a unit of labor has been over some time period and then what the corresponding change has been in the wage of that labor. And then, this would need to be done for all production processes to get a sense of the what’s happening across the entire economy.