(FORTUNE Magazine) – Like a lot of people harboring unkind thoughts about labor unions, the present writer was influenced by the work of H. Gregg Lewis, a somewhat legendary figure who died the other day at 77. Lewis ended his career as professor emeritus at Duke University, but for most of his working life, he was based at the University of Chicago and was helping to cement its reputation as a hotbed of neoclassical economics. Some part of his legend concerns the terror he struck in generations of graduate students whose doctoral dissertations had to meet his exacting demands. As a colleague mentioned in a tribute to Lewis in the Journal of Political Economy, ''A dissertation was not acceptable if it could be improved, no matter how long it had been in progress.'' Any writer could get weak in the knees just thinking about that. Lewis's main contribution to economics centered on supply and demand in labor markets. In particular, he pretty much invented the econometrics required to study this question: How large is the ''union wage premium''? This premium is the additional amount earned by some workers as a result of their unions' monopoly power -- their effective power to withhold labor from companies needing same. Calculations about the union premium tend to be messy in the extreme, since they require you to make calculations about the earnings of ghost workers. You begin by gathering data on the earnings of real-world individuals who may or may not belong to unions. Then, based on analysis of ) umpteen different variables (e.g., the proportion of unionized workers in the industry), you estimate the earnings of the ''ghosts'' -- hypothetical workers deemed in all respects identical to the real-world characters except their union-membership status is not the same. The union premium can be thought of as the difference in compensation between the ghosts and the real guys. When Lewis first began measuring the premium in the Sixties, it was apparently around 20%. That is, unions extracted pay levels some 20% higher than employers would have paid in purely competitive labor markets. But Economics 101 tells us that when the price rises, the quantity demanded takes a tumble. So employers have characteristically responded to union premiums by hiring fewer workers and, to the extent possible, more experienced workers. These tactics enabled business to recover some part of the premium -- on some calculations, as much as half of it. In other words, the 20% should be viewed as a gross figure, and the ''net premium'' may have been closer to 10%. Recent estimates of the union premium suggest that it has risen rather substantially from the early Seventies through the late Eighties, and is probably around 30% (gross) nowadays. Our main source for that estimate is economist Michael L. Wachter of the University of Pennsylvania. In collaboration with several of his colleagues there, Wachter has worked hard at tracking the premium for the economy as a whole, and in different industries. He points out that a high premium is ultimately quite consistent with union weakness. Some employers have found ways to get rid of unions altogether, which of course shrinks the union base -- but does not lower the average premium paid by companies still stuck in the union sector. Some other employers have acquiesced in a high premium in exchange for union concessions on work rules. In general, however, Wachter believes companies continuing to pay anything like a 30% premium are courting trouble, and observes, ''Unionized firms paying those premiums are continuing to lose market share.'' He also believes that the unions charging those rates will continue to shrink.