to generate information about incentives faced by individual legislators. In other words,
studies of economic voting tell us how voters respond to different economic outcomes, while
we investigate whether different economic outcomes impinge on the welfare of individual
legislators. Notice that by shifting the emphasis away from voters and towards legislative
incumbents, we can study repeated elections as a system of incentives that is designed to
align the interest of individual legislators with the interest of the electorate. This step is
important because it makes it possible to explicitly analyze how the incentives generated
by repeated elections, i.e. by the prospect of electoral sanction, interact with the incentives
generated by fluid party systems, i.e. by the prospect of switching parties. As a result, by
focussing on individual incumbents, we can use our data to shed light on one of the core
problems in the study of representative government: the impact of party systems on electoral
accountability.
The second reason for studying electoral control with incumbent-level data concerns the
measurement of accountability in new democracies. Notice that measures of accountability
based on party vote-shares in effect presume that electoral control works through political
parties. While this assumption is perfectly reasonable in mature democracies, in the case
of fluid party systems it is important to explicitly test its plausibility because, as discussed
above, fluid party systems may dilute the informational value of party labels. Moreover,
when party systems are fluid, measures based on party vote-shares, may either underestimate
or overestimate the extent of electoral control. On one hand, if electoral control bypasses
political parties, then party vote-shares will be unrelated to policy performance. Based
on such a finding, we might underestimate the degree of accountability by concluding that
elections do not function as a mechanism of electoral control, when in fact they do but not
through political parties. On the other hand, if electoral control works through political
parties, then party vote shares may be strongly related to policy performance. However,
if legislative incumbents switch parties in order to escape accountability, then we might
8
Convention