Job Loss and Job Change

All three papers deal with important policy issues. Ruhm, whose analysis ranges the most widely, goes furthest in recommending policy activism. Podgursky and Swaim (P&S) are somewhat more re strained, but it is clear that they too favor special assistance for displaced workers. In our own paper we see advantage in workers being given notification of their impending redundancy in the specific context of plant closings and relocations, although our treatment admittedly does not examine the role of firm heterogeneity in producing the observed pattern of results. Nevertheless, it is the case that our understanding of the displaced worker problem remains rudimentary. Ignoring, for now, potentially important problems that have to do with the definition of a displaced worker, our focus here will be on the computation of the earnings loss. P&S deal with before-andafter earnings comparisons, or relative earnings losses. Ruhm makes a broad comparison of the earnings of stayers and m ov e rs a n d h a s t h e a dv a n t a g e o f a somewhat longer follow-up period. The use of a control group is advantageous in computing wage losses for fairly obvious reasons, but simple comparisons of the earnings of movers and stayers are misleading (see Mincer 1986). In particular, any estimation using a control group should seek to model explicitly the decision rule that motivates some workers to move and others to stay put. Another selectivity problem addressed by P&S, but not by Ruhm, concerns the selection of workers into employment status. We return to this issue below. Focusing for the moment on the beforeand-after approach, and recognizing as a possibility that the displaced could have higher earnings than their nondisplaced counterparts if not all stayers, what conclusions can be drawn from the observed wage changes? A particularly difficult question concerns the sacrifice of firmspecific training investments. It is convent ional to argue that this loss can be d i s c o v e r e d b y l o o k i n g a t t h e p r e displacement wage tenure coefficient (e.g., Hamermesh 1987). But, as Ruhm cor rectly notes, tenure on the lost job is positively correlated with post-displacement wages. (P&S detect a generally insignificant coefficient on tenure and focus their attention instead on lost specific investments resulting from changes in industry and occupation). The basic problem with the tenure variable is that it is incapable of distinguishing between general and specific training investments in the case of younger workers and of detect ing switching from specific to general training investments in t h e ca se o f w ork e r s w h o e i t h e r a re informed of or otherwise know about their impending redundancy. Our own work would suggest that the conventional measure of lost specific training investments—that is, assuming a zero effect of pre-displacement tenure on the new job— could inflate estimates of the mean earnings loss by up to 45 percent (Addison and Portugal 1987). That said, there is still the issue of apparently large earnings losses