Me back in 2003:

Politicians, just like other
people, may prefer some beliefs over others. But unlike average voters, politicians
often do have a significant probability of affecting outcomes, and their efforts have
direct repercussions. A politician who does not have rational expectations about the
impact of his policy stances on his career pays a high price, so in this area the standard
arguments for rationality (Muth 1961) are compelling. Politicians who systematically
misunderstand voters’ feelings forego large opportunities for political profit. They have an
incentive to learn from mistakes and hire expect advice.

 

Systematic mistakes about
what the voters want also open a politician to takeover bids from more rational
challengers. In any case, people with rational expectations about voters’ preferences
self-select into the political arena. Even if this is a small fraction of the population, it could
easily be large compared to the number of available offices. Other systematic mistakes
about their technology for producing votes are similarly unlikely: politicians cannot
afford to have irrational expectations about the number of votes the marginal value PAC
dollar buys, the probability the press will uncover skeletons in their closet,
or the likelihood that evidence of current indiscretions will leak out. Rationality about
expected compensation pays too: Politicians are unlikely to have irrational expectations
about their level of fringe benefits, or the extent to which political experience will ultimately
increase their market value in their post-political career.

 

However, it does not follow
that politicians will be rational about the actual
impact of the policies that they
implement. They merely need to gauge voters’ reaction to their policies; if the voters have
irrational expectations about what policies will accomplish, a politician who rationally
second-guesses them gets little benefit. In fact, if it is indeed impossible to fool all of
the people all of the time, politicians who share the irrational assessments of their
constituents may actually be at a competitive advantage compared to rational politicians who cynically pander to the prejudices of the
electorate.

Alex Tabarrok in 2016:

More questions are being asked, more data is being collected and more
randomized experiments are being run in the effort to win the presidency
than will ever be used to choose policy by the presidency. Sad.