Egalitarian media? Internet access and demand for redistribution in the U.S.

My goal in the PhD is to study whether access to new media,
the Internet, has changed the demand for redistributive
policies in the United States. The introduction of Internet
undeniably has changed the way in which people interact with
information: from merely passive actors to producers of
9 / 13
information. Historical evidence shows that expanding voting
rights to all citizens changes the beneficiaries of public
policies: from the elites to the newly enfranchised voters
[Larcinese (2014), Lindert (2004), Husted (1997)]. Although
the right to vote is virtually universal in the U.S., the Internet
may act as a form of enfranchisement, forasmuch as it gives
voice to citizens and ideas which were not paid attention by
the traditional media.
The first paper in measuring reliably the causal effect of this
new medium on the U.S. presidential elections was Larcinese
et. al. (2017). They found that Internet penetration has a
strong positive impact on voter turnout, with a weaker effect
on campaign contributions and vote shares to Democrats.
Unfortunately, evidence of Internet diffusion on political
participation is mixed [Gavazza et. al. (2018), Falck et. al.
(2014), Campante et. al. (2018)]. With regard to preferences
towards redistribution, Alesina et. al. (2004) pointed to
potential explanations of demand for redistribution: the
share of poor population, the characteristics of the political
system, culture and attitudes. Experimental evidence on the
demand for redistribution have found that its main drivers
are self-interest, insurance motives and social concerns
relating to inequality [Durante et. al. (2014)]. In line with this,
Alesina et. al. (2005) stressed that preferences over political
actors may reflect voters' own perceptions on the state of
the economy and the determinants of inequality. This is the
mechanism in which I want to focus on.
A higher amount of information may educate voters,
affecting political accountability and public policy choices
[Avis et. al. (2018), Prat et. al. (2013), Snyder et. al. (2010)].
Nonetheless, even though the Internet undoubtedly has
increased the amount of information accessible to individuals
at a relatively low monetary cost, it might as well increased
the cognitive and time cost of processing this huge amount
of information. First of all, there is evidence that media act
like agenda setters, i.e. by covering certain topics they affect
the relative importance that consumers give to that topic
[Larcinese et. al. (2011), McCombs (2004)]. Closely related
to this fact is that, as found in Durante et. al. (2018), people
who is more exposed to entertainment TV are more
vulnerable to populist messages. Conceivably, this evidence
will not only relate to traditional media, since the Internet has
also increased the number of entertainment opportunities.
Hence, the real threat of this new media is that it may act as
echo chambers [Garret (2009), Gentzkow (2011)], amplifying
the effects of agenda setting and vulnerability to populism.
In order to measure demand for redistribution I will construct
a variable capturing the share of votes that a proredistribution candidate got. To do so I will use two datasets.
The first is the one provided by the Atlas of U.S. Presidential
Elections, which provides electoral results for the U.S.
presidential elections at the county level elections held in the
period 1912-2016.
Political parties differ on many other aspects than just their
leaning towards redistribution. That is why focusing on the
share of voting to the left or right-wing parties would be a
noisy proxy for demand for redistribution. In order to
overcome this problem, I propose to individually classify each
political candidate based on the intensity of their discourse
with regard to redistribution. The American Presidency
Project gives access to a rich database of election speeches
and remarks for each candidate running for election.