This article argues that the predictions of standard trade union model
s and the tests for distinguishing between these models are not robust
to quite small and reasonable changes in the conventional assumptions
. In particular, it considers the effect of assuming that the ex post
substitutability between labor and capital is less than the ex ante su
bstitutability. The paper shows that much of the conventional wisdom a
bout the effects of trade unions is not necessarily true in this frame
work.