Years ago, my dissertation adviser used to tell me that people don't have marginal products, jobs have marginal products. There is certainly individual variation in performance on jobs, more so in some than others, but for the majority of jobs it is true that it is the job itself, not the person doing it, that determines its productivity. A person can be the best 7-11 clerk ever, but the job puts a limit on the productivity that can be achieved and hence the wage that can be earned.
Suppose we take all jobs and break them down into a finite set of k skills or tasks, S1, S2, ..., Sk. Then an hours worth of labor at a particular job can be viewed, on average, as some combination of these skills or tasks, L = a1S1 + a2S2 + ... akSk.
When I think of changes in the labor market, it is sometimes helpful to think about how this skill set is evolving through time -- some coefficients (the aj terms) get larger, some get smaller, and so on. Some may be zero until certain points in time when they are invented by new technology, and some may be zero after particular points in time.
Notice also that from a labor supply point of view these individual tasks or skills are what determines how unpleasant a job is. Each individual will have preferences (disutility) over these skills and hence have a different willingness to work at a job depending upon their aversion to the tasks involved. A change in productivity, for example, could shift the tasks toward a more undesirable set causing workers to demand higher wages not only because they are more productive, but also because the characteristics of the job have changed. Of course, this could go the other way as well if the job shifts to tasks that are more desirable. In extreme cases you might be willing to pay to do the job (suppose the job were to stand on the sideline and make sure an end zone cone stays in place at the super bowl).
Interestingly, under such a formulation, it is not necessarily the case that an increase in productivity increases wages. Suppose there is an increase in productivity that also causes a marked shift towards tasks that workers find more tolerable. In labor market terms, the labor demand curve and the labor supply curves would both shift outward leaving an uncertain effect on wages. [On a technical level it also calls into question identification schemes that use productivity to map out the labor supply curve.] But I don’t think this explains flat wages in recent years, i.e. that productivity changes have made jobs so much better for laborers that there was no reason to raise compensation even though productivity increased, but there may be cases where the changes in job characteristics are important to consider.
So why do I bring this up? No particular reason, just some thoughts on a lazy Saturday -- labor economists will probably tell me they know all about this through work on compensating differentials, etc. anyway. But I am curious how technological change and advances in information technology have altered the typical job in recent decades. Is a typical hour of work more tolerable, less tolerable, or just as miserable day-in and day-out as it ever was? Any ideas?