AbstractIn this thesis, I investigate how voters’ redistributive preferences are likely to evolve with growth in the automation of work, and test the reliability of using survey data to capture workers’ experiences of automation.
In Chapter 2, I investigate how increasing survey experiment respondents’ per-ceived vulnerability to an automation shock inﬂuences their redistributive preferences, and how exposure to rhetoric mediates that response. I ﬁeld a pre-registered survey experiment to 2,500 adults resident in the UK. I ﬁnd that as perceived vulnerability increases, redistributive preferences remain constant or decline. However, the addition of a treatment featuring rhetoric, which causes respondents to view automation-induced inequality as unfair increases support for several redistributive policies. The eﬀects are pronounced among more-educated respondents - a group expected to increasingly be aﬀected by automation going forward. These ﬁndings underscore the need to look be-yond automation’s labor market eﬀects in order to thoroughly understand its political implications.
Chapter 3 presents an essay co-authored with Konstantinos Matakos. We investigate: (i) how emphasizing diﬀerent features of a potential automation shock inﬂuences redistributive preferences; and (ii) how that information interacts with recent experience of a labor market shock to inﬂuence preferences. We ﬁeld a pre-registered survey experiment to 4,000 adults resident in the US. We ﬁnd that, overall, redistributive preferences increase in response to our informational treatments, while information that speaks to individuals’ distrust of elites signiﬁcantly increases the redistributive prefer-ences of two otherwise unresponsive groups of individuals: Republicans and respondents with stronger populist tendencies. We also ﬁnd that individuals recently exposed to a real world labor market shock are particularly responsive to the treatments. This eﬀect appears to stem from previously shocked individuals’ anxieties about their vulnerability to future labor market shocks. These ﬁndings shed new light on the complex relationship between economic anxiety and redistributive preferences by identifying key mediating factors at play.
In Chapter 4, I investigate the reliability of using survey data to measure work-ers’ experiences of automation. I survey 4,070 working-age adults resident in the US and ﬁnd that respondents’ self-reported exposure to automation predicts occupational employment changes at least as reliably as the widely-used measure of exposure, Routine Task Intensity (RTI). Moreover, I ﬁnd that a survey approach oﬀers additional beneﬁts over RTI, by reliably distinguishing between negative and positive employment outcomes from automation, and by better capturing exposure to automation of non-routine tasks. These ﬁndings demonstrate that using a survey approach to measure workers’ experiences of automation could reliably capture new insights that would improve our understanding of the economic and political challenges associated with automation.
|Date of Award
|1 Oct 2021
|Konstantinos Matakos (Supervisor) & Amrita Dhillon (Supervisor)