Consider this post to be the first in a series in which I attempt to explain some things I'm currently working on. The problem with stating anything definite about projects is that you risk it coming across the wrong set of eyes. Maybe once I get tenure I'll have a more relaxed tone, but for now (since I haven't even gotten a PhD!), I'll try to write in such a way that doesn't give away techniques, data sources, or references. See my research page here for a list of all current and past projects.
Of course, that stuff tends to be the boring part of any research paper anyway, and when I comment on research I usually avoid technique and references in a New York Times Sunday-esque "pop social science" sort of way, so I guess I'll just try to do that same sort of thing with my own work. Anyway, on to it:
The project I want to write about for the next few posts is both narrow and broad. I started out with a question that hasn't been adequately resolved in the empirical literature, and ended up manifesting about the importance of social context when evaluating model results.
The question: whether under a piece rate (where you get paid based on how much you produce), more productive workers in the firm are prone to work less hard than they should in order to reduce the possibility that the manager cuts their piece rate.
In a firm where all workers are under the same contract, like the one that I looked at, this possibility might arise if individual effort is observable and if workers gain significant experience on the job, or "learning by doing". In such a situation, managers might observe the total output of the firm, decide that productive workers are earning really high rents because the work is too easy for them, and cut rates on everyone in order to stop workers from earning as much and, of course, to keep the profits of the firm from slipping.
The best way to test for the presence such "strategic" individual behavior would be to examine worker effort over time, and see how it responds to changes in his piece rate. Additional information on how long the worker has been at a firm, as well as other factors that might contribute to his or her productivity (how closely is management observing the worker?), would be good "controls" -- i.e., they allow you to tell the fullest story possible by evaluating the contribution of each possible factor to explaining worker effort. And finally, ideally, you would have a lot of workers to look at over a long period, in order to tell a convincing story.
The problem, however, is that if you want to tell a truly convincing story, those are not all the variables you need. Sure, from the inside of the firm, you might be able to explain the relationships among, say, worker effort and how productive he could have been, and then see if he's "lying" on the job. But what if his behavior is also affected by his fellow workers' ideas about why managers are changing their wage rates on them in the first place? A rate cut could mean one thing -- a rate hike, another. Or, what if the worker has to work for a certain amount of time in order to feed his family back home, because it's becoming harder and harder to do so as other options for work deteriorate? These are potentially very significant factors that simply cannot be included in a firm-level research design.
Well, that's it for now. I've outlined the ideal situation for testing the research question as well as a potential problem with that approach. Next time I'll talk a bit more about who these workers were and why running the model in the 1830s might be different from running the model in the 1850s.