Here our hypotheses were derived from the transaction cost theory of the firm, following the tradition from Coase (1937). It might also be possible to derive the
results from an entry barrier perspective, but as a referee has pointed out to us, such an enterprise must be carried out with considerable caution. We thus defer this task to future research.
Several further issues surrounding the study deserve mention. First, just as selection bias is likely to affect the mean market shares in Table 1, it may also influence some of our other results, although we have been unable to detect any consistent patterns in this regard. Secondly, while some of our results could be explained by the conjecture that firms from more profitable primary industries diversify less, this would be in conflict with some results from the literature (Rhoades 1973). Thirdly, our logic is extremely stylized in the sense that we consider each firm either efficient or inefficient. In practice firms can, of course, be of several intermediate types and our conclusions should be modified accordingly. Fourthly, as in any large sample study, we looked at average effects. It is entirely conceivable that a firm can go squarely against our recommendations and still do very well.