For what it is worth, I think there is some useful information gained by making the distinction between productive and unproductive labour (as defined by Jim) even though it is impossible to measure the distinction with any degree of accuracy. I did some calculations for the postwar period up to the 1980s looking at the rate of profit (conventionally defined a la Doug) and the ratio of production workers to white collar workers in Canada. My hypothesis was that the rising productivity and capital accumulation in production initially produced high levels of profits but as capacity rose (with demand constraints) and pressures on raw material and energy prices rose, there was a tendency of firms to increase their 'unproductive staffs and expenditures' (e.g. advertising) in order to increase (or at worst maintain) market shares in order, in turn, to attempt to maintain profits. At the same time, wages of productive workers were rising such that the increased productivity of productive workers was unable to offset the rising cost of unproductive workers such that realized profits fell. In the subsequent period, the rise of productivity of unproductive workers due to computer technology and the subsequent stabilization/fall in some forms of unproductive labour (and wage stagnation?) while productivity of productive workers has risen faster than (stagnant) wages has allowed profits to rise, at least until the more recent recession. The conventional data, using white collar employees as a proxy for unproductive labour, is consistent with the hypothesis through the time period I looked at.
Paul Phillips, Economics, University of Manitoba