Union Power in the Long Run Reconsidered: Comment
评论了Palokangas（1989）的研究，该研究质疑了作者早期关于长期中工会无法改善成员福利的结论，通过改变储蓄假设和生产函数形式，发现工会可能长期有效。作者通过自己的计算，探讨了不同参数下工会力量的可能性。
The purpose of our earlier note was to establish the possibility that, in the long run, organized but non-saving labour might be unable to improve its well-being simply by restricting supply; see Kemp and Long (1987). To reduce computational fuss to a minimum, we made do with some quite primitive assumptions. In particular, it was assumed that capitalists consume a constant proportion of their collective income. On the basis of that assumption we were able to establish the strong result that, if and only if there is a nontrivial steady state, a labour union can do nothing for its members in the long run. With the possibility of long-run labour impotence established, attention switches to the scope of the phenomenon. In particular, one would like to know whether the strong result of our earlier paper survives the relaxation of our restrictive assumption about capitalists' saving. The calculations of Palokangas (1989) are therefore of great interest. Replacing our savings assumption by a variant of the permanent income hypothesis and our linear homogeneous production function by a convex-concave function, Palokangas has been able to show that, even in the long run, a labour union may be able to do something for its members. However Palokangas' analysis leaves it unclear whether the new possibility is attributable to the change in saving behaviour or to the change in the nature of returns to scale or to each separately or to both collectively. Passages in his text suggest that, in his opinion, convexity-concavity of the production function and not the permanent income hypothesis accounts for the new possibility. In this Comment we briefly report on some relevant further calculations of our own. The calculations are based on our 1987 model modified by the requirement that capitalists save optimally in the light of a constant and positive rate of time preference and of perfect foresight of future rates of return to capital. We have been able to show that, depending on parameter values: