James E.Alt and Michael Gilligan 341appealing: thus, there really seems to be a natural affinity between the Stolper-
Samuelson model and majoritarian politics.
In the second row, given any significant costs of collective action, factor mobility
would mean that free-riding should be rampant, inhibiting collective action. If
conversely the cost of collective action goes to zero, any coalition should form.
These should be large because the costs of collective action are low and benefits
are non-excludable. There would be little incentive to build large coalitions, however,
because size would not guarantee victory in this case: the political institutions are
non-majoritarian. This cell seems to yield few interesting predictions.
Where factors are specific a number of different cases arise, surrounding the
“classic” case of interest-group lobbying in the lower left corner. Protectionism is
the likely outcome in this cell. There, costs of collective action are high, thus
excluding consumers (who are, after all, a very large group with non-excludable
benefits) from trade politics. The benefits of trade policy are concentrated on
particular industries, due to factor specificity. Thus, these industries have an incentive
to pay the collective action costs, even though they are high, in order to gain their
favored trade policy.
In the penultimate row, where factors are specific but majorities are needed to
win, exit is costly (because factors are specific) but high costs of collective action
mean that groups must be small or benefits concentrated to form. In such situations,
the universal logroll mentioned above would be a possibility. In this case, as costs
of collective action decrease and large groups are needed to gain victory, alliances
between a specific factor and the mobile factor become more likely, as do alliances
between various specific factors. The numbers of the mobile factor group are
large enough to make it worthwhile for specific factor groups to try to bring them
into a coalition, provided the stake can be made large enough to motivate their
participation.
Finally, in the lower-right corner (where factors are specific and institutions
non-majoritarian, but costs of collective action are low) even dispersed losers can
organize because costs are low. Any group could win, however, because the size
of a group is not important to political victory. This cell, too, appears to yield few
interesting predictions.
As should be clear from Figure 3, the two major models of international trade
policy coalitions carry with them hidden assumptions—one about the severity of
costs of collective action, the other about the domestic political institutions which
make trade policy. In any case the models presume ideal types of political
organization which may not exist. Without considering political variables, economic
explanations are biased, and vice versa.
V. CONCLUSIONS
The narrowest purpose of this paper has been to review the determinants of trade
policy coalitions. Although recent studies have stressed economic factors such as
abundance and mobility of factors of production, we have argued that other more
political factors (collective action costs, political institutions) are likely to be just