Voting as signalling

If voting has signalling purposes, voting by mail "may" reduce total
turnout... but increase the share of informed voters...

"Theory and Evidence on the Role of Social Norms in Voting"
BY: PATRICIA FUNK
Stockholm School of Economics
Stockholm Institute of Transition Economics (SITE)
http://papers.ssrn.com/paper.taf?abstract_id=627347
Date: March 2005

ABSTRACT:
This paper investigates social norms and voting behavior. I argue that
social norms create incentives for signaling, i.e., voting for the purpose
of being seen at the voting act. Empirical evidence on signaling can be
gained by looking at the introduction of optional postal voting in
Switzerland. Even though the possibility of mail voting reduced voting costs
substantially, it didn't increase turnout. Consistent with my model's
predictions, voter turnout decreased more in the smaller communities, but in
the meantime, the share of cooperators (=interested voters) was more
positively affected there. Therefore, modern voting tools may decrease
average turnout, but nevertheless, increase the quality of the voting
outcome. Current models predict the opposite, but ignore the effect of
different voting systems on the incentive for signaling.

(This is her model:)

If voting is only possible by going to the polls, potential signaling
benefits thus originate from being seen (or not being seen) at the voting
both. However, in small communities, people know each other and gossip about
who was observed at the booth and who wasn't. Therefore, total signaling
benefits are assumed to decrease in community size (and) a nonpositive
relationship between voter turnout and community size.

(...) It can be shown that the introduction of postal voting has a
non-positive impact on turnout in small communities and a non-negative
impact on turnout in large communities (third result). The intuition behind
result three is that the introduction of postal voting has two
countervailing effects: a cost-reduction effect (with a positive effect on
turnout) and a reduced signaling effect (with a negative effect on turnout).

As for the latter, imagine a small community, where voting is only possible
at the polls. Due to the strong social pressure and the high signaling
benefits, a large share of defectors goes to the polls e.g. to avoid social
sanctions from non-voting. As soon as postal voting is allowed, cheating
becomes easy and defectors don't vote anymore. In large communities, on the
other hand, signaling was less rewarding (or necessary) under the old voting
system, so that the cost-reduction-effect of mail voting dominates.

These predictions from the signaling model stand in contrast to the
predictions of standard models of voting, which only consider the modern
voting tools' effect on the voting costs.