Legislatures often make important economic decisions. A decade ago eco
nomists viewed each legislator as the agent, if not of the median vote
r, of some pivotal voter in a party constituency. Therefore, the votin
g decision of a legislator could readily he predicted if one could mea
sure the preferences of the relevant pivotal voter. Models where inter
ests on specific issues can be simply linked to votes on the same issu
es are questionable, however, since votes will be traded across issues
; Political parties are one important vehicle for organizing such trad
es. A simple empirical test for the United States Senate, however, cle
arly rejects the notion that the representative is the agent of the me
dian voter or a local party constituency, even if vote trades are allo
wed. Thus, there is ample opportunity for politicians, even under reel
ection constraints, to either voice their own ideology sr that of idio
syncratic support coalitions. The relevant theory and empirical eviden
ce is summarized, and some validating tests for European parliaments a
re suggested.