principle, lower unskilled labor supply. If, on the other hand, protectionist policy preferences are
rooted in nationalist attitudes, the strategy to alter them will have to be very different, and the
strategist may have to be less sanguine about the prospects of success. Of course, the truth may well
lie somewhere between these alternatives, with both interests and ideology reinforcing opposition to
free trade. In this case it becomes necessary to tease out the relative effects of each, and to identify
the conditions under which such effects operate.
In pursuing these questions, this paper starts with rational choice models of policy formation,
and asks to what extent does the Heckscher-Ohlin model provide a useful guide to predicting trade
policy preferences. This question has often been addressed in the empirical economics literature:
typically, authors have used aggregate data to ask whether factor interests or sectoral interests are
more important in determining policy preferences. For example, Magee (1978) looked at industry
testimony before the US Congress in 1973, and found that labor and capital seemed to have the same
interests within each sector, suggesting that a specific factors model is appropriate. More typically,
voting behavior has been related to constituency characteristics, either in the context ofUS
Congressional votes on trade-related issues, or in the context of national elections that are assumed to
have been fought largely on such issues.4
This paper pursues a different strategy. The inquiry is pitched, not at the level of interest
groups (where, almost by definition, interests are likely to be determining), or at the level of the
electoral constituency (where data on ideological and political variables are likely to be missing), but
at the level of individuals or mass public opinion. In so doing, we are following in the footsteps of
Balistreri (1997), and Scheve and Slaughter (2001), both of whom use survey datato tackle the
question of who is in favor of free trade, and why. As Scheve and Slaughter (citing Rodrik, 1995)
point out, individual-level preferences regarding trade must lie at the heart of any rational choice
account of policy-formation, but using aggregate data provides only indirect information on agents’
preferences, since “policy preferences and institutions together determine policy actions, so the
4 See for example Irwin (1994), or Kaempfer and Marks (1993).