local distributions, Ft (v) = 1 - ∏n ( 1 - Ftn (v - mn )j. There is only one global reservation value, Vt
The agent accepts offer with a value net of migration costs exceeding the global reservation value,
vn - mn > V" . There is contracted migration when the accepted offer is not located at the agent's
origin.
Therefore, whether it leads to speculative or contracted migration, at first, spatial search impacts on
reservation wages and on accepted distribution of wages offers. What can we tell about this impact?
Can we compare migrants and non migrants accepted wages?
Spatial search implies higher reservation wages
At first, when search and migration result from rational decisions, spatial search implies accepted jobs
with higher wages. All the arguments leading to this conclusion are variants of the fact that a rational
agent would not accept to migrate were he better off staying than moving. When migration is
speculative, considering new distant labour markets is valuable because search displays higher returns
on these markets. These higher returns may stem from lower local search costs or a better distribution
of local wages. The agent migrates because the difference between local values of continuing search is
high enough for covering migration costs. But the local value of continuing search equals the local
reservation value of accepted jobs. Thus, the reservation wage of a speculative migrant at his
destination is higher than at his origin.
When migration is contracted, simultaneous search on more labour markets implies higher search costs,
which normally imply a lower value of search and a lower reservation wage. Were the distribution of
wage offers unchanged, the agent would search at his origin place only. But, with simultaneous search,
the agent receives more wage offers and their distribution shifts toward higher wages. The modified
distribution of wage offers has a positive effect on the value of search. Simultaneous search is
accepted because this positive effect overrides the negative effect of the higher search costs (Herzog
& alii, 1985).
But migrants may have lower wages than stayers
The fact that spatial search implies accepted jobs with higher wages does not imply that migrants have
higher wages than stayers. It would be true if all workers were identical, whether they migrate or not.
But they are not, Moreover, it is well known that a central assumption of search models, the existence
of a non degenerated distribution of wage offers, cannot hold with an homogenous population of
searchers. And, as soon as workers are heterogeneous, migrants and non migrants self-select.
The consequences of self selection have been examined by various others using the Roy model, an
outstanding contribution being by Borjas and alii (1991). To illustrate keeping things simple, let us