Biased perceptions of income distribution and preferences for redistribution: Evidence from a survey experiment ☆

TL;DR: In this article, a tailored household survey provides evidence on systematic biases in individuals' evaluations of their own relative position in the income distribution and explores their potential impact on preferences for redistribution.

About: This article is published in Journal of Public Economics.The article was published on 2013-02-01 and is currently open access. It has received 326 citations till now. The article focuses on the topics: Income distribution & Redistribution (cultural anthropology).

Summary

1 Introduction

• The shape of the income distribution plays a key role in the determination of policies with redistributive components (such as social security, health care, government transfers and taxation) in political economy and public finance models.
• To form their judgments, views and attitudes, agents need to make difficult inferences about distributional outcomes (e.g., inequality, mobility) based on limited information and within given time constraints, but there is as yet little evidence on the origins or the accuracy of the inferences they make in this regard.
• Moreover, as predicted by the theory, the bias is significantly correlated with the respondent’s relative position within the reference group (as proxied by area of residence).
• The next section discusses the formation of subjective income distributions and individuals’ perceptions of their income rank, and explores these factors’ implications for attitudes towards redistribution.

2.1 Subjective income distributions and reference groups

• Economic agents’ assessments of income distributions depend on their access to information and on their ability to process the relevant data.
• |P(i ijS be the selection equation, which represents the probability that individual i belongs to the reference group jS given that his or her income is ix .
• Alternatively, individuals may fail to consider all the available information, or they may use it incorrectly (Simon, 1972).
• Since naïve agents use the information about the income distribution within their reference group as if it were representative of the entire population, naïve agents in the rich reference group will underestimate the actual cumulative income distribution for every income level, i.e., F(y)(y)FN for all y.
• The expression )(xf i N j in equation (3) can be used to compute naïve perceptions of different moments and statistics of the income distribution.

2.2 Biased perceptions and preferences for redistribution

• Misperceptions of income distribution can have substantial implications in the determination of policy outcomes.
• The population can be divided by income levels into potential net losers and winners: individuals above a given -quantile would lose in net terms if the redistributive policy were implemented, while those below it would gain from the reform.
• Since they are relatively poor, most of these agents are likely to have poor reference groups, and naïve agents among them will therefore over-estimate their own ranking.
• The experimental design of the survey used in this study allows for a direct test of this mechanism.
• As discussed above, the biases given by equation (3) translate into biased perceptions of many characteristics of the distribution of income, including the poverty count and social welfare, which could influence an individual’s support towards redistributive policies through altruistic concerns.

3.1 Survey on Distributional Perceptions and Redistribution

• The discussion in the previous section covered the formation of subjective income distributions, the possibility of systematic biases, and their implications for attitudes towards redistribution.
• It was specifically designed to test the model presented in the previous section and, to that end, collected data on a set of individual and household characteristics and on respondents’ labor-market and other socioeconomic outcomes, as well as their answers to a series of questions about their political views and attitudes.
• It also gathered information on the respondents’ actual household income and on their perceptions of their own income rank within the distribution for the whole country.
• The Survey on Distributional Perceptions and Redistribution relied on an original instrument (the income-rank evaluation question), which elicited a specific value of the cumulative subjective distribution: its evaluation at the point where each respondent thought his or her household stood.
• The question was worded as follows: “There are 10 million families in Argentina.

3.2 The survey experiment setup

• Besides the income rank question, the second and most innovative aspect of the survey was the implementation of an experimental design incorporated into the questionnaire.
• Randomized questionnaire-experiments had been developed in laboratory settings (Amiel and Cowell, 1992; Cowell and Cruces, 2004), while, in the context of household surveys, Joliffe (2001) and Di Tella, Galiani and Schargrodsky (2008) have randomized the allocation of questionnaire types among respondents.
• The originality of this setup stems from the nature of the treatment, in which the interviewer provided feedback to respondents in the treatment group in the form of consistent information on the income distribution.
• The interviewer read the following statement (with X and Y being determined by previous answers): “Based on your income level, the latest studies conducted by the University indicate that there are X million families with an income lower than yours, while you stated that there were Y.”.
• The questionnaire for the control group did not 7.

4.1 Subjective income distributions

• This section presents the empirical evidence from the Survey on Distributional Perceptions and Redistribution.
• Figure 2a presents the income distribution of the Greater Buenos Aires survey sample as a function of deciles of the national distribution at the time of the survey, which served as the basis for the categories used for the household income question.
• The deciles of a national income distribution are relatively narrow categories, and it is therefore quite plausible that respondents could have difficulty in ascertaining their position with that degree of precision.
• There is a significant positive relationship between both variables, although the distribution of subjective levels is considerably more concentrated.
• This pattern has a direct correlate for the distribution of the bias, which is depicted in Figure 4b as a function of the objective income decile: respondents at the top and the bottom of the objective distribution display substantial negative and positive biases, respectively (of about -3.5 and 3 deciles for the extreme categories).

4.2 Reference groups and biased perceptions of income distribution

• The discussion so far has revealed the presence of substantial biases in distributional perceptions.
• More specifically, under the alternative hypothesis that individuals under-estimate the importance of income in the formation of reference groups, then the relative income within the reference group will be positively correlated to their perceptions of income ranking.
• 14 Table 2 presents a series of tests of this hypothesis that employ regression analysis.
• With the inclusion of the locality rank variable as an independent variable, the coefficient of the objective income decile variable is virtually zero, and not significant at standard levels.
• The respondents’ relative incomes within their localities thus seem to have an effect on their perceptions of the distribution, over and above that of their own objective income levels.

4.3 Biased perceptions and preferences for redistribution: experimental results

• The empirical evidence presented so far illustrates the degree of bias in distributional perceptions, and the following analysis assesses their relevance for attitudes towards redistribution.
• The post-treatment questions20 in the survey were designed to capture respondents’ views on some of the prominent aspects of the public debate on distributional policy changes: whether to provide transfers to the poor or not and, if so, their modality (in cash, in kind or in some form of employment intermediation).
• Only 12.8 percent of the respondents in the overall sample (without distinguishing between the treatment and control groups) exhibited this high degree of support for redistribution.
• 24 The results shown in column 2 of Table 3 indicate that, as predicted, the treatment had no statistically discernible effect among those with a zero bias regarding any of the variables.

5 Conclusion

• The motivation of this paper was the lack of evidence on the accuracy and origins of perceptions of the income distribution, which play a crucial (though implicit) role in political economy and public finance models.
• The systematic biases documented in this paper and their consequences, however, can arise in any society – the determinant factor is the sorting of individuals in reference groups by income level.
• Confronting agents’ biased perceptions with this information had a significant effect on stated preferences for redistribution: those who overestimated their relative position (who thought of themselves to be relatively richer than they were) demanded more redistribution.
• To the degree that the information treatment managed to correct biased distributional perceptions, these results can be interpreted as evidence of the causal effect of misperceptions on political attitudes.
• This mechanism provides an alternative explanation for the low levels of redistribution observed in modern democracies, and since it affects relatively poorer individuals it is reminiscent of the Marxian notion of false consciousness.

Figures

Figure 1. Illustration of differences between population and reference group income distributions

Figure 3. Distribution of bias in perception of own income decile, in number of deciles

Figure 2. Distribution of objective and perceived own income decile

Table 3. Biased perceptions of income distribution and preferences for redistribution: Experimental results.

Table 1. Objective income decile, perceived own income decile and bias by quintile of objective income

Table 2. Determinants of perceived own income decile

Figure 4. Average perception of own income decile and bias, by objective income decile

Figure 5. Average objective decile, perceived own decile and bias by neighborhood

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "Biased perceptions of income distribution and preferences for redistribution: evidence from a survey experiment" ?

In this paper, the authors explore the causes and consequences of systematic biases in individuals ' perceptions of aggregate income distributions and find that the bias is significantly correlated with the respondent 's relative position within the reference group ( as proxied by area of residence ). 

Q2. What future works have the authors mentioned in the paper "Biased perceptions of income distribution and preferences for redistribution: evidence from a survey experiment" ?

The 27 Olin Wright ’ s ( 2009 ) discussion of false consciousness states: “ Ideology is seen as preventing workers from understanding the nature of their oppression and the possibilities of its transformation. Further research could focus on the impact of biases and information on actual behavior – for instance, on charitable donations or on voting patterns. Moreover, the results in this paper could originate in either limited information or limited cognitive ability – further research could disentangle the source of the observed biases in distributional ( and other ) perceptions. ” 25 results in this paper support Romer ’ s ( 2003 ) discussion of the possible welfare improving effects of subsidizing information, and Besley ’ s ( 2007 ) remarks about the potential of information providers for improving policies, although the impact of the biases in the efficiency of redistribution should also be considered ( Acemoglu and Robinson, 2005 ), The role of misconceptions in political economy has been studied before ( Romer, 2003 ; Slemrod, 2006 ). 

Q3. What is the role of misperceptions in political economy?

More generally, concepts such as inequality, self-interest and the median voter can be adapted in their application to political economy outcomes when misperceptions and misconceptions play a role. 

Q4. What does the inclusion of control variables do?

As in other small-sample experimental studies, the inclusion of control variables reduces the variability of the error term, which increases the statistical power of the significance test of the treatment effect. 

Q5. What is the role of the income distribution in the determination of policies?

The shape of the income distribution plays a key role in the determination of policies with redistributive components (such as social security, health care, government transfers and taxation) in political economy and public finance models. 

Q6. What is the effect of providing consistent information about the income ranking on individuals’ attitudes towards redis?

If those individuals are purely selfinterested, providing them with consistent information about their income ranking would make them oppose, rather than favor, the redistributive policy. 

Q7. What are the ways of recovering subjective distributions for acontinuous variable such as income?

There are several ways of recovering subjective probability distributions for acontinuous variable such as income, which include eliciting quantiles, moments or points of the distribution (see Manski, 2004). 

Q8. What is the role of the cognitive distortions in political economy?

The absence of effective struggle for socialism, then, is at least in part explained by the pervasiveness of these cognitive distortions. 

Q9. What is the effect of the provision of information on the stated preferences of those with negative biase?

If individuals know their true rank in the distribution but misreport it in the survey (because of embarrassingly low or high relative levels) or report focal-point answers because of a lack of interest, the provision of information should have no effect on their stated preferences, since they already have this unbiased estimate.