Redistributive Preferences and Their Determinants

Our study includes two interventions.

The first intervention is real-life shock in the form of the Finnish Tax Day. On the tax day of citizens' tax data is made publicly available and especially earners' incomes and taxes are reported and discussed in the media. Half of the respondents are surveyed before the Tax Day and the other half after. The Tax Day is also a combined treatment with the US presidential elections occurring one day before the Tax Day.

The second intervention randomly assigns half of the respondents to watch an informational whiteboard animation video of the role of taxes and public spending in redistributing income.

2024-10-22

2024-11-22

The study has four primary research questions:

1. What kind of redistributive preferences individuals have?
2. What motivations drive redistributive preferences?
3. How information provision about how public sector operates shifts the individuals’ redistributive preferences and the motivations underlying these preferences?
4. How real-life increase in the salience of income inequality and salience of social stability affects the preferences for redistribution and the motivations for those preferences?

Following these research questions, first two outcomes of interests are (1) choices over a public policy package that contains redistributive elements and (2) choices over the motivational factors made by the individual in the survey.

The third outcome of interest is the effect of our information treatments on redistributive choice and motivational factor choice. We are interested in (a) how informational video about how public sector operates affects redistributive choice and motivational factor choice, (b) how the Finnish Tax Day (7.11.2024), when information about the incomes of top income earners from tax records is revealed, affects redistributive choice and motivational factor choice and; (c) how the combination of both treatments affects redistributive choice and motivational factor choice. We take into account that US Presidential elections take place just shortly before the Tax Day, which is both a confounding factor for the Tax Day and a potential additional treatment that may affect some responses. While the US presidential election does not shock knowledge about redistribution in Finland or it's salience, it may affect other responses such as which party to support or questions related to trust and whether income differences reduce societal stability and increase crime.


We hypothesize that:

1.The informational video about the role of taxes and public spending affects beliefs about the role of taxes in redistribution and tackling income inequality and could then affect preferences or motivations for redistribution.
2.The informational video about the role of taxes and public spending provides new information about the role of taxes in providing social security and public goods and could then affect preferences or motivations for redistribution.
3.The Finnish Tax Day increases income transparency and salience of income inequality, and informs respondents about others and one’s own relative position in income distribution.
4.The Finnish Tax Day shifts individuals’ beliefs about the fairness and efficiency of the Finnish tax system and could then affect preferences or motivations for redistribution.
5.The Finnish Tax Day increases information about tax planning.
6.The Finnish Tax Day increases information about the earning at the top of the income distribution and could then affect preferences or motivations for redistribution.

The choice outcomes are measured with two conjoint surveys in which respondents choose between two sets of attributes (either a policy package or a candidate in election where the candidate holds certain views about redistribution and income inequality). In the first conjoint, we elicit preferences over seven redistributive policy attributes which each can obtain three different possible value realizations. Respondents’ repeated answers reveal how they, as a group, value these different attributes and their values in their decision making relative to the other attributes and their values. In the second conjoint, the choice is measured with five attributes, which are motivational statements about the consequences of redistribution, fairness and trust. The second conjoint is built around following hypotheses which mirror to the five motivational attributes:


We hypothesize that preferences for redistribution depend on whether or not individuals perceive that:

1. Income inequality is fair or not (fairness)
2. Luck matters more than effort in success (luck vs. effort)
3. That redistribution decreases incentives to contribute to society (efficiency)
4. They can trust that political system works for their benefit (trust)
5. Inequality increases instability or more equality increases stability (instability)

Secondary outcomes we study are how what following:

1. What kind of stated redistributive policy preferences individuals have?
2. What kind of stated motivations individuals have to redistribute?
3. How much people know about some current redistributive policies?
4. What kind of stated beliefs individuals have about the state of society, their own future, who can be trusted and what consequences inequality has.

Secondary outcomes are measured as normal independent questions in the survey.

Our study contain a set of regular survey questions, two conjoint survey experiments, one information provision video experiment implemented within the survey and one real-life shock in the form of the Finnish Tax Day (and US Presidential elections). The respondents are drawn from our survey partners (Taloustutkimus) respondent panel which is nationally representative on the level of age, gender, region (NUTS 2).

The survey proceeds in the following manner.

Respondents are divided into two groups by our real-life information treatment, the Finnish tax-day. We achieve this by half of the respondents being surveyed before the Tax Day and the other half after the Tax Day. The first survey window is between 22.10.2024 and 5.11.2024. The Tax Day takes place on 7.11.2024. The second survey window takes place between, 8.11.2024 and 22.11.2024. The US Presidential elections take place on 5.11.2024, but likely the results will be discussed in the Finnish media on 6.11.2024, that is, one day before the Tax Day.

The second treatment assignment is conducted at the start of the survey. Half of the people in both groups (before and after the Tax Day) are randomly assigned to watch an informational video of the role of taxes and public spending in redistributing income.

Before the video, all respondents answer basic demographic questions.

Then those who are randomly allocated to the video treatment proceed to video which cannot be skipped. The video is a whiteboard animation, which explains in simple manner how taxes are collected each year, why taxes are collected and what they are used for, and what benefits and costs are associated with tax collection.


Those without video treatment move straight into background questions that include demographic questions and questions about political preferences, values, knowledge of redistributive policies, trust, views about the world and one’s own life, and habits concerning media consumption. The English translation of questions can be found in the attached analysis plan.


In the last part of the survey, we run two conjoint survey experiments.

In the first conjoint, the respondents are presented with two cards, A and B, which both present policy packages that hypothetical parties in the next parliamentary elections are proposing in their party manifestos. Both cards include seven statements (attributes) regarding taxes and redistributive policies. The respondents are asked to choose which party, A or B they would rather vote. The respondents make the choice between the candidates six times. Each attribute category contains value realisations that are associated with a specific redistributive policy. The seven polices are: 1) income taxes; 2) unemployment benefit; 3) public health care; 4) national debt; 5) capital income tax; 6) progressivity of the income tax at the top and; 7) fuel taxes.

List below displays the attributes and their levels in the first conjoint. Each attribute has three levels. The value of each attribute is randomised for each card and the order of the attributes is randomised once for each respondent, but is the same on both cards and remains the same for the same respondent across the cards.


Income taxation:

• Increase income tax evenly by 1% across all income groups
• Keep the current level of income taxation
• Decrease income tax evenly by 1% across all income groups

Unemployment benefits:

• Make unemployment benefits more generous by increasing the basic daily allowance by 200 euros per month
• Keep the current level of unemployment benefits
• Tighten unemployment benefits by reducing the basic daily allowance by 200 euros per month

Public healthcare funding:

• Increase public healthcare funding by 1 billion euros per year
• Keep the current level of public healthcare funding
• Cut public healthcare funding by 1 billion euros per year

National debt:

• Be willing to take on an additional 1 billion euros of national debt per year
• Not seek to change the current level of national debt
• Aim to reduce national debt by 1 billion euros per year

Capital income tax:

• Increase capital gains tax by 2 percentage points
• Keep the current level of capital gains tax
• Decrease capital gains tax by 2 percentage points

Tax progression:

• Increase tax progression by raising the top income tax rate by 3%
• Keep the current level of tax progression
• Reduce tax progression by lowering the top income tax rate by 3%

Fuel tax:

• Increase the fuel tax by 5 cents per liter
• Keep the current level of fuel tax
• Decrease the fuel tax by 5 cents per liter
 

The second conjoint tests the motivation hypotheses detailed in the previous section, that is, what kind of views people have about reasons and consequences of income inequality and redistribution. In the conjoint, the respondents are presented with the personal opinions of two politicians A and B and are asked which of these individual politicians would they rather vote for in the next parliamentary elections. The opinions concern the possible effects and externalities of redistributive policies and taxation. The conjoint has five attributes which are: 1) fairness; 2) luck vs. effort; 3) efficiency; 4) trust in politicians and; 5) inequality’s effect on stability.
List below displays the attributes and their levels in the second conjoint. Each attribute has either two or three levels. The value of each attribute is randomised for each card and the order of the attributes is randomised once for each respondent, but is the same on both cards and remains the same for the same respondent across the cards.

Large income differences:

• Large income differences are unfair
• Large income differences are fair

Causes of large income differences:

• Large income differences are the result of good luck experienced by high earners
• Large income differences are the result of the hard work of high earners

Income redistribution and economic impact:

• Income redistribution would increase employment and boost economic growth
• Income redistribution does not have an impact on employment or economic growth
• Income redistribution would decrease employment and diminish economic growth

Trust in politicians:

• Politicians can be trusted to foster the common good of the citizens
• Politicians cannot be trusted to foster the common good of the citizens

Income inequality and societal stability:

• Large income differences reduce societal stability and increase crime
• Large income differences do not affect societal stability or crime
• Large income differences increase societal stability and reduce crime

Randomization done in office by a computer

Tax Day treatment: Respondents are randomized based on the timing of Finland’s Tax Day. Half of the respondents are surveyed before the Tax Day and the other half after, creating two groups based on exposure to the real-life event.

Video treatment: Within both the 'before' and 'after' Tax Day groups, respondents are randomly assigned at the individual level to either watch an informational video on tax collection and public services or not. This creates four treatment arms overall (no treatment, Tax Day only, video only, and both Tax Day and video).

No

4000 individuals, standard errors are clustered at individual level.

Randomization done in office by a computer

In total we have 4000 respondents. The treatments create following treatment arms:
Pure control group: No video and surveyed before Tax Day, 1000 respondents
Treatment group 1: Watch the video and surveyed before tax date, 1000 respondents
Treatment group 2: Surveyed after Tax Day, 1000 respondents
Treatment group 3: Surveyed after Tax Day and watch the video, 1000 respondents

Analysis Plan

Analysis Plan v.2