Service workers are part of Marx's definition of *productive* labor. It was Smith who argued that all service workers are unproductive. Marx criticized him for that, and included them in his concept.
Smith argued that labor was productive only if it (1) produced a profit for capital, and (2) produced tangible goods that could be accumulated. Smith visualized the goods as a storehouse of labor (means of production) capable of putting into motion more labor. For Marx, whether labor is productive depends on the social relations under which it was performed (labor being exchanged with capital). It has nothing to do with whether the product of the labor is material or not. Hence his well-known example: A teacher working in a privately owned school is productive labor. "That the latter (school owner) has laid out his capital in a teaching factory, instead of a sausage factory, does not alter the relation" (Capital I, p. 509).
>Now I have read (with only marginal
>understanding) interesting looking articles (I believe in the Review
>of Radical Political Economy) purporting to assess the actual
>economy in terms of these categories -- but these same articles
>(even if correct) if followed into the strategic and tactical realms
>would
>by excluding public school teachers, staff (medical and non-medical)
>of VA hospitals, postal employees, etc. etc from the working class
>royally fuck up progressive and/or revolutionary politics. So I think
>I understand part at least of what Doug means when he insists on
>using the term "political."
Determinations of who is, or is not working class cannot be based on the productive-unproductive distinction. Surely no one argues today that only productive labor is working class. Unproductive labor exists because it is useful to capital in ways other than producing surplus value. Both forms of labor are working class; i.e., they consume their wage while capitaists accumulate surplus value.
When fully developed, the P/U distinction is useful, however, to show the different functions labor performs for capital, which leads naturally to an analysis of the difficulties of creating a cohesive consciousness among the working class.
>Is a depression more or less likely if more of the work force consists
>of unproductive workers (as listed above)?
Can't say, a priori. Depends on why, and the contradictions that result. The ratio of unproductive to productive labor has been growing in this century (see Fred Moseley, James Becker), rather continuously, with no necessary connection to either prosperity or depression. The question implies an equation of unproductive and useless. Not so. Unproductive labor is still useful to capital, and its wages are one way surplus value is realized (investment outlets being insufficient to absorb all of it, this is always a problem for capital).
> Does an increase in the
>number of MacDonald's employees as opposed to janitors in steel
>plants make a technical difference? (I'm not an economist and I may
>have my terminology fouled up here.) I would regard workers in
>fast food workers as production rather than service workers? Is that
>accurate, and does it make a difference either in description of the
>economy or in political planning? Why is it production to cut up
>meat in a meat plant to be cooked at home but service to further
>process that meet for immediate consumption on the premises? Is
>the cook a productive worker but the cashier an uproductive
>worker? Are all these questions either technically or politically
>productive?
The meat cutter and processor, the cook and cashier are all productive labor.
Let me flesh out the rest of the definition of productive labor for the full picture as Marx saw it (others since have developed other variants). Not only can service workers be productive, so can distribution workers (those necessary to package, transport, handle, store, and distribute goods), as well as supervisory workers (necessary to coordinate and unify production). In short, all workers necessary to complete a round of production; i.e., the production and realization of surplus value.
>I'll add one question: Is it remotely possible, *today*, to make an
>accurate division (listing) of workers into productive and unproductive
>laborers? Of the total work force, how many are productive workers?
>Can their locations be specified with some accuracy?
Several people have made passable attempts (including Moseley and Becker, mentioned above). Two things to remember. There is no one correct defintion of productive labor. The distinctions you make depends on the questions you ask and the purpose of your inquiry. I went through some of this in an answer to Rob a couple of days ago (if you don't have it and want it, I'd be happy to send it to you with elaboration). Second, the type of analysis we're talking about requires establishing distinctions to measure mostly large aggregates over time. A good definition that can't be precisely measured but is consistently applied can yield important insights. There are lots of defensible ways to do that.
Doug has a point about much of the previous stat work in this area. It hasn't been very good. My point, however, is that it is fertile ground, the kind of analysis most likely to produce real understanding.