Biased Perceptions of Inequality and Redistribution

Biased Perceptions of Inequality and Redistribution
Carina Engelhardt and Andreas Wagener
Institute of Social Policy, School of Economics and Management, Leibniz University of Hannover, Germany
1. Idea and Punch Line
• The public holds biased perceptions of the income distribution: it perceives the distribution as “too symmetric” and “too rich”.
• Research question: do such biased perceptions affect the political economy of redistribution in democracies?
• Relevance: standard approaches hypothesize that redistribution is more intense in democracies the more unequal the income distribution
 but these models do not perform well empirically when estimated with the actual income distribution.
• Our approach: to estimate the MR approach (and its POUM extension) using the perceived income inequality (and perceived social mobility).
• Result: this works much better; clear positive association between perceived inequality and redistribution.
2. Conceptual Framework: MR Model
Mean-to-Median Ratios (selection)
Meltzer and Richard (1981, MR): with majority-voting, the volume of redistribution increases
when inequality rises.
Actual (OECD)
Perceived
Weighted
Relevant inequality measure: ratio of mean income to median income
Canada
1.141
1.022
0.895
• Mean income separates “winners” from “losers” from redistribution; median income:
politically decisive in majority voting
France
1.159
1.003
0.865
Germany
1.131
1.039
0.919
Italy
1.138
1.003
0.881
Japan
1.144
0.985
0.861
Mexico
1.526
0.912
0.598
The Netherlands
1.156
0.966
0.836
New Zealand
1.178
1.010
0.857
Poland
1.171
1.036
0.885
South Korea
1.105
0.912
0.825
Spain
1.137
1.046
0.920
Sweden
1.078
1.023
0.949
Turkey
1.411
0.847
0.600
UK
1.210
1.141
0.944
US
1.192
1.043
0.875
Problem with MR hypothesis: empirically not confirmed.
Potential reason: all studies are based on actual income inequality – but what matters in
politics are perceptions, not facts.
3. Perceived Income Distributions
We study the answers in 29 countries to the following ISSP question:
“In our society there are groups which tend to be towards the top and groups which tend
to be toward the bottom. Below is a scale that runs from top to bottom (10 top – 1 bottom).
Where would you put yourself now on this scale?”
Distributions emerging from the answers:
Austria
Germany
Denmark
Ireland
Canada
Spain
Israel
Switzerland
Finland
Italy
Chile
Czech Republic
France
United Kingdom
Japan
South Korea
• The same pattern holds in all 29 sampled countries
• Perceived inequality is everywhere lower than factual inequality
• More unequal countries tend to underestimate inequality more.
Density
Hungary
Belgium
4. Inequality and Redistribution
Mexico
Netherlands
Norway
New Zealand
Poland
Portugal
• OLS regressions of social quota on (perceived) inequality
• Various controls included/omitted; results are robust
• Data: ISSP for 1992, 1999, 2006-2009; controls from OECD data base; crosssectional data are used.
Russia
Sweden
Slovenia
Turkey
United States
Dependent Variable: Social Expenditures (in % of GDP)
Actual inequality
-22.40**
Self-Positioning on a Top to Bottom Scale
Perceived inequality
-
 All distributions derived from self-categorizations are “too symmetric” and “too rich”
Weighted inequality
-
Two measures of perceived inequality in a country
• Ratio of average to median value of these self-categorizations  “perceived inequality”
• Ratio of “perceived inequality” to actual mean-median ratio  “weighted ratio”
(motivation: some countries are generally more (un)equal than others which might
influence citizens’ perceptions of inequality)
-
-
21.40
-
36.73***
Social expenditures 80ies
0.646***
0.716***
0.621***
Log GDP
-3.988**
-0.082
-5.259**
Trade openness
-0.0095
-0.0061
-0.0103
Dependency ratio
0.518
-0.051
0.530
Note: * p < 0.10, ** p< 0.05, *** p< 0.01; N=26
5. Main Results
• Using actual inequality, MR model is rejected: negative correlation between inequality and redistribution.
• Using perceived or weighted inequality, MR model cannot be rejected: positive association between inequality and redistribution.
• For weighted inequality: positive association is highly statistically and economically significant.
Dependent Variable: Social Expenditures
(in % of GDP)
Perceived POUM
6. Variation: POUM Hypothesis
Prospects of Upward Mobility (POUM): poor people believe that they/their children are likely to move upward on the economic ladder.
• POUM hypothesis: the greater upward mobility, the smaller is redistribution – poor with POUM are reluctant to vote for
redistribution.
• We show: POUM hypothesis works better empirically when based on perceived POUM (based on the hard-work question) – which
are much larger than actual POUM (based on the probability of being in a much better occupation than one’s father).
7. Discussion and Extensions
• Why is the perception of income inequality biased?  preference for symmetry; inferences from acquaintances, …
• MR/POUM model: direct democracy (majority voting on issues). How about representative democracy?
Email: [email protected]
Phone: +49 511 762 8214
Actual POUM
-30.62***
-
-
12.35
Social expenditures
80ies
0.602***
0.590***
Log GDP
2.632
1.387
Trade openness
0.0130
0.0569
Dependency ratio
0.217
0.673
Note: * p < 0.10, ** p< 0.05, *** p< 0.01; N=22