Why Do (Some) Ordinary Americans Support Tax Cuts For the Rich? Evidence From A Randomized Survey Experiment

How to cite

Limberg, Julian; Hope, David; Weber, Nina. 2021. "Why Do (Some) Ordinary Americans Support Tax Cuts For the Rich? Evidence From A Randomized Survey Experiment." AEA RCT Registry. AEARCTR-0007620. https://doi.org/10.1257/rct.7620-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Julian Limberg* , David Hope , Nina Weber

* Primary investigator

Registry ID
AEARCTR-0007620
DOI
https://doi.org/10.1257/rct.7620-1.1
First registered
Last updated
Status
Completed
Trial dates
2021-05-01 – 2021-06-30
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/7620

Abstract

In the US, the last 40 years have been a period of reducing tax rates on the richest members of society. These tax cuts have often had significant levels of support from the public. Why do (some) ordinary Americans support tax cuts for the rich? We test the impact of four predominant theories – unenlightened selfinterest, prospect of upward mobility, trickle-down beliefs, and fairness considerations using a survey experiment. In particular, we test these theories by randomly assigning a sample of US Americans to different information treatments. We then estimate the effects of these treatments on core beliefs, articulated preferences, and elicited preferences towards tax cuts for the rich.

Intervention

While there are substantial theoretical and empirical literatures on the determinants of redistributive preferences (Alesina & Giuliano, 2011; Iversen & Goplerud, 2018) spanning all the way back to Meltzer and Richard (1981) seminal median-voter model of redistribution, we know much less about what drives ordinary Americans’ preferences for cutting taxes on the rich. Our study aims to shed new light on this important question through a randomized online information provision experiment. In particular, we randomly assign respondents into five groups which are presented a short statement and a bar chart. The four treatment groups receive factual information relating to potential drivers of preferences for tax cuts for the rich identified from the literatures on redistributive and tax policy preferences, namely 1) unenlightened self-interest (Bartels, 2005); 2) the prospect of upward mobility (Benabou & Ok, 2001; Piketty, 1995); 3) trickle-down effects (Stantcheva, 2020); and 4) fairness considerations (Almås, Cappelen, & Tungodden, 2019; Bastani & Waldenström, 2021). The control group receives factual information on the longest rivers in the USA. We then test whether the treatments have an impact on 1) core beliefs of individuals, 2) expressed preferences for or against tax cuts for the rich, and 3) elicited preferences.

Experimental design

Respondents will be randomly assigned to one of five treatments. Each treatment consists of a short text and a bar chart, followed by a question of understanding in order to ensure participants have paid sufficient attention to the provided information. The first treatment uses the information about individuals’ self-declared household income to inform them whether they are currently paying the top federal income tax rate. The second treatment shows the chances of an individual to become part of the top 1% income earners over their lifetime. The third treatment shows average annual economic growth in the postwar period up until 1979 when top federal income tax rates were substantially higher and contrasts that with average annual economic growth since 1979. The fourth treatment compares the wealth of the richest US Americans who inherited their wealth to the wealth of the bottom 50%. The last (placebo) treatment presents individuals with information about the two longest rivers in the US. Afterwards, we measure the outcomes (expressed preferences, and elicited preferences).

Design details

Randomization method
Individuals will be randomised by a computer.
Randomization unit
Individual randomization
Clusters
4000 individuals and 5 treatment arms in the first experiment, and, depending on available funding, a second experiment with a further 2000 idividuals.
Observations
4000 individuals in the first experiment and, depending on the available funding, 2000 individuals in the second experiment.
Arms
800 individuals per treatment arm in the first experiment and, depending on the available funding, 400 individuals per treatment arm in the second experiment
Primary outcomes
There are three main outcome variables. First, we are interested in people's core beliefs about cutting taxes for the rich. These are measured via survey questions which ask respondents about there core beliefs (see survey instruments in the pre-analysis plan). Second, we are interested in people's expressed preferences. Do they support or oppose cutting taxes on the rich? Again, this will be measured via a survey question. Finally, we are interested in elicited preferences. We check whether the treatments have an effect on preference elicitation by presenting to them a non-profit organisation campaigning for lower taxes top personal income tax rates as to a non-profit organisation campaigning for higher taxes top personal income tax rates. We provide respondents with a link where they can join each organisation’s mailing list and trace whether respondents click on this link.
Sponsors
Name: Department of Political Economy, King's College London Location: London Url: https://www.kcl.ac.uk/political-economy

Cited by

No citation data available for this registration yet.

Keywords