Behavioural Motives for Income Redistribution

The Australian Economic Review, vol. 38, no. 3, pp. 285–97

©

2005 The University of Melbourne, Melbourne Institute of Applied Economic and Social ResearchPublished by Blackwell Publishing Asia Pty Ltd

1. Introduction

The modern welfare state is a remarkablehuman achievement. In the advanced econo-mies, a substantial fraction of total income isregularly transferred from the better off to theless well off, and the governments that presideover these transfers are regularly endorsed bytheir publics (Atkinson 1999). The modernwelfare state is thus the most significant case inhuman history of a voluntary egalitarian redis-tribution of income among total strangers.What accounts for its popular support?

We suggest below that a compelling case canbe made that people support the welfare statebecause it conforms to a behavioural schemawhich we call

strong reciprocity

. Strong reci-procity is

a propensity to cooperate and sharewith others similarly disposed, even at per-sonal cost, and a willingness to punish thosewho violate cooperative and other socialnorms, even when punishing is personallycostly and cannot be expected to entail net per-sonal gains in the future

.

1

Economists have forthe most part offered an alternative, empiri-cally implausible, theory of self-regardinghuman motivation to explain who votes forredistribution. The most widely acceptedmodel of the demand for redistribution in eco-nomics is the

median voter model

, which holds

that each voter desires a personal wealth-maximising level of redistribution. Under ap-propriate assumptions, it follows that the redis-tribution implemented by a government electedunder a majority rule system is that preferredby the median-income voter. Because the dis-tribution of income is generally skewed to theright (there are a few very rich individuals), themedian voter is poorer than the mean voter andwill therefore demand a positive level of redis-tribution.

An important implication of this model isthat demand for redistribution decreases as per-sonal income increases (Roberts 1977). Butpersonal income is a surprisingly poor predic-tor of support for redistribution (Gilens 1999;Fong 2001). A large fraction of the poor op-pose income redistribution and a large fractionof the rich support it. Among respondents of anationally representative American survey(Gallup Organization 1998) who have annualhousehold incomes of at least $150000 and ex-pect their lives to improve in the next fiveyears, 24 per cent respond that the governmentshould ‘redistribute wealth by heavy taxes onthe rich’, and 67 per cent respond that the ‘gov-ernment in Washington DC should make everypossible effort to improve the social and eco-nomic position of the poor’. Equally striking isthe fact that among those with annual familyincomes of less than $10000 who did not ex-pect to be better off in five years, 32 per centreport that the government should not redistrib-ute wealth by heavy taxes on the rich, and 23per cent say that the poor should help them-selves rather than having the government

Policy Forum: Some Policy Implications of Behavioural Economics

Behavioural Motives for Income Redistribution

Christina M. Fong, Samuel Bowles and Herbert Gintis*Department of Social and Decision Sciences, Carnegie Mellon University; Behavioral Sciences Program, Santa Fe Institute and Faculty of Economics, University of Siena; and External Faculty, Santa Fe Institute and Economics Department, Central European University, respectively

* We would like to thank the John D. and Catherine T.MacArthur Foundation and the Behavioral Sciences Pro-gram of the Santa Fe Institute for financial support. Wedraw extensively on Fong, Bowles and Gintis (2005).

286 The Australian Economic Review September 2005

©

2005 The University of Melbourne, Melbourne Institute of Applied Economic and Social Research

‘make every possible effort to improve the …position of the poor’.

2

Thus, while self-interest is an importanthuman motive, and income does explain someof the variance in redistributive attitudes, othermotives appear to be at work. Abundant evi-dence from across the social sciences—muchof it focusing on the United States with similarfindings in smaller quantities from other coun-tries around the world—has shown that whenpeople blame the poor for their poverty, theysupport less redistribution than when they be-lieve that the poor are poor through no fault oftheir own. That is, generosity towards the pooris conditional on the belief that the poor workhard (Williamson 1974; Heclo 1986; Farkasand Robinson 1996; Gilens 1999; Miller 1999).

For instance, in a 1972 sample of whitewomen in Boston the perceived work ethic ofthe poor was a far better predictor of supportfor aid to the poor than one’s family income,religion, education, and a host of other demo-graphic and social background variables (Will-iamson 1974). Indeed in predicting support forsuch aid, the addition of a single variable mea-suring beliefs about work motivation tripledthe explanatory power of all the above back-ground variables together. Moffitt, Ribar andWilhelm (1998) were among the first econo-mists to report findings on this relationship.They used the General Social Survey, a largenationally representative dataset with observa-tions in nearly every year since 1972 to showthat those who believe that people get ahead by‘lucky breaks or help from others’ rather thanhard work prefer more spending on welfare.Fong (2001) used nationally representativedata from a 1998 Gallup Social Audit to showthat the effects of beliefs about the causes of in-come on demands for redistribution are surpris-ingly large and cannot be explained by missingmeasures of self-interest. Alesina, Glaeser andSacerdote (2001) have reported related find-ings from the World Values Survey on the atti-tudes of Americans and Europeans. Americanshave much stronger beliefs that poverty iscaused by laziness; 60 per cent of Americanssay the poor are lazy, compared to just 27 percent of Europeans. The authors argue that thiscould be an important explanation for the small

size of the American welfare state compared tothe average European welfare state.

Our interpretation of these findings is thatpeople are willing to help the poor, but theywithdraw support when they perceive that thepoor may cheat or fail to cooperate by not try-ing hard enough to be self-sufficient and mor-ally upstanding. Within economics, our view ismost similar to the taxpayer resentment view ofthe demand for redistribution modelled byBesley and Coate (1992), the reciprocal altru-ism view surveyed by Alesina et al. (2001), thetheory of equal opportunity by Roemer (Roe-mer 1998; Roemer et al. 2003) and the effect ofreciprocity sentiments on redistributive publicfinance by Kolm (1984).

3

Our view is also consistent with interpreta-

tions by Heclo (1986) and Gilens (1999), whocite evidence that Americans support a widearray of benefits for the poor and are primarilyopposed to ‘welfare’, presumably because‘welfare’ refers to means-tested cash assis-tance, which may be perceived as a programthat benefits able-bodied adults who choose tohave children out of wedlock and prefer not towork. Our view is also consistent with evi-dence that attitudinal support for social secu-rity spending has been very high and stableover time, while support for spending onwelfare has been consistently low (Page andShapiro 1992). Our interpretation is also com-patible with equity theory and attribution the-ory. According to equity theory, people shouldreceive resources from a system that are pro-portional to their contributions (Walster, Wal-ster and Berscheid 1978; Deutsch 1985; Miller1999). Attribution theorists argue that peopleare less likely to help someone if they deter-mine that the person is individually responsiblefor his or her outcome (Weiner 1995; Skitkaand Tetlock 1993). Finally, our interpretationis consistent with abundant evidence from eco-nomic experiments on giving of real money inlaboratory games, as well as randomised exper-iments on attitudinal support for redistribution,that people are generous to strangers, but thattheir generosity is conditional on perceptionsthat the recipients of their generosity are mor-ally worthy or have helpful intentions or be-haviours (Fong, Bowles and Gintis 2005;

Fong, Bowles and Gintis: Behavioural Motives for Income Redistribution 287

©


Camerer 2003; Fehr and Schmidt 2005; Sobel2001).

Economists have been sceptical of non-selfish models for several reasons. First, therecould be unmeasured self-interest variablesthat explain the support for redistribution. Inparticular, those with low-mean, high-varianceincomes may be more likely to think that pov-erty is due to bad luck and also more likely todemand redistribution out of self-interest forinsurance against a low income. We soundlyreject this hypothesis in Section 2.

Second, people who think that effort plays amajor role in income generation may be con-cerned about the incentive effects of taxation ortransfers rather than the ‘worthiness’ of recipi-ents (Piketty 1995). We have two pieces of ev-idence, however, that incentive costs cannotfully explain attitudes towards redistribution.One is that, were incentive costs of taxation theproblem, those who believe that effort is im-portant should support less government spend-ing in general. Yet, as we show in Section 2,the belief that effort is important to gettingahead in life is negatively correlated with sup-port for redistribution and positively correlatedwith support for military spending. Another isthat subjects in a behavioural experiment oncharitable giving to welfare recipients gave sig-nificantly more money when they were ran-domly paired with a welfare recipient who saidshe would like to work than when randomlypaired with a welfare recipient who said shewould not like to work (Fong 2005). Therewere no disincentive costs in this experiment,so some other interpretation is necessary. Thisexperimental result also addresses a third con-cern that economists have raised: people whodo not want to give to the poor may say that thepoor are lazy to justify their selfishness. Thiscannot explain why randomly assigned treat-ment conditions in the charity experiment justdescribed had significant effects on giving towelfare recipients.

Concern about the ‘undeserving poor’ is pro-nounced in the United States, but is far fromabsent in Europe. Using data from Eurobarom-eter surveys and the World Values Survey, weshow elsewhere that people all over the worldreport significantly less generosity to the poor

when they believe that the poor are lazy ratherthan industrious (Fong et al. 2005).

We do not doubt that self-regarding motivesoften underpin apparently generous actions.Rather, we suggest that they do not always doso. Understanding egalitarian politics today re-quires a reconsideration of Homo economicus,the unremittingly self-regarding actor of eco-nomic theory. We do not wish to replace thetextbook self-regarding actor, however, withan equally one-dimensional altruistic actorwilling to make unconditional, personallycostly, contributions to the less well off.Rather, we believe that strong reciprocity,which involves

conditional

cooperation andpunishment, better explains the motivationsbehind support for the welfare state.

2. Survey Evidence

Our interpretation of the results reviewedabove is that because of strong reciprocity,people wish to help those who try to make it ontheir own, but for reasons beyond their owncontrol cannot. People wish to punish, or with-hold assistance to, those who are able but un-willing to work hard. However, there areseveral alternative explanations of the effect ofbeliefs about the worthiness of the poor that areconsistent with pure self-interest. In this sec-tion, we test these alternative explanations andfind that self-interest alone cannot explain therelationship between beliefs about the worthi-ness of the poor and support for redistribution.These results are based on Fong (2001).

We use the 1998 Gallup Social Audit Sur-vey, ‘Haves and Have-Nots: Perceptions ofFairness and Opportunity’, a randomly se-lected national sample of 5001 respondents. Ineach test, we use the set of all individuals whoresponded to all of the questions used in the re-gression, unless noted otherwise.

4

Relative to other commonly used surveys,the Gallup survey has a large sample size for alarge number of questions on inequality anddistribution. The sample size permits runningregressions with full controls on narrow seg-ments of the sample, namely high-income andlow-income sub-samples. There is a large num-ber of self-interest measures that include not


©


Bad lu

ck ca

uses

poverty***

Luck ca

uses

wealth

***

Luck an

d effort

cause

wealth

***

Lack of lu

ck an

d effort

cause

poverty***

Never

worries

about b

ills***

White

***

Male

***

Much

opportunity

in

the Unite

d States*

**

Household in

come

> $150000***

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

only the usual objective socio-economic vari-ables, but also subjective measures of eco-nomic well-being and future expectations.These may widen the net intended to captureself-interest.

To construct our dependent variable, weadded the responses to the five measures of at-titudes to redistribution, signing the responsesso that the measure increases in support for re-distribution. See Fong (2001) for the exactwording of the questions used to construct thismeasure.

Two sets of measures of the causes of in-come are used in this study. The first containstwo questions concerning the importance of ef-fort and luck in causing wealth and poverty,and one question on whether or not there isplenty of opportunity to work hard and getahead in America today. The second set is a se-ries of questions about the importance of vari-

ous factors, including race and sex, for gettingahead in life.

Self-interest is measured by income andother variables likely to predict current and fu-ture tax obligations and current and future reli-ance on social insurance or redistributionprograms. In Figures 1 and 2 we control forself-interest by including in the regressions in-come, race, sex, education, age, and the fre-quency with which respondents worry aboutmeeting family expenses.

5

In Figure 1 we present results from an ordi-nary least squares regression that predicts sup-port for redistribution using two sets ofvariables: beliefs about the causes of wealthand poverty and the measures of self-interest.To facilitate interpreting the coefficients, wehave standardised the dependent variable tohave a zero mean and a standard deviation ofone. The interpretation is as follows: those who

Figure 1 Determinants of the Support for Redistribution

Notes

: The bars represent ordinary least squares coefficients predicting support for redistribution. The dependent variableis standardised so that the estimated coefficient represents the effect of the variable indicated on support for redistributionmeasured in standard deviation units. The equation also includes seven additional income dummies, age, a dummy for at-tended college, and dummies for ‘worries about bills most of the time’ and ‘worries about bills some of the time’. Theomitted category for household income is less than $10000 per year. The omitted categories for causes of poverty andwealth are ‘lack of effort’ and ‘strong effort’ respectively. To simplify the presentation of race effects, we use the sampleof white and black respondents only. The omitted category for ‘worries about bills’ is ‘all of the time’. There are 3417 ob-servations. R

2

= 0.260. This regression uses sample weights, although the results are not sensitive to them. We use robuststandard errors. *** indicates significance at the 1 per cent level. The precise values of the regression coefficients illustratedin Figures 1 and 2 are presented in Fong et al. (2005).

Regression

coefficient


©


Gender

matters

(male

)***

Race m

atters

(white

)***

Educatio

n

matters

**Pare

nts

matter*

*

Connectio

ns

matter*

**

Luck m

atters

***

Race m

atters

(black)

Dishonest

y

matters

***

Inheritan

ce

matters

**

Gender

matters

(female

)

Risk-ta

king

matters

***

Hard w

ork

matters

***

-0.15

-0.10

-0.05

0

0.05

0.10

say that bad luck alone causes poverty are 0.50standard deviation higher in their support forredistribution than those who think lack ofeffort alone causes poverty. Those who thinkthat good luck alone causes wealth are 0.39standard deviation higher on the support for re-distribution scale than those who think effortalone causes wealth, and people who respondthat there is plenty of opportunity in the UnitedStates to get ahead scored 0.42 standard devia-tion lower in support for redistribution thanpeople who do not think there is plenty of op-portunity.

Measures of self-interest also have signifi-cant effects in the expected direction on sup-port for redistribution. Those who are in thehighest income category (annual householdincome greater than $150000) scored 0.47standard deviation lower on support for re-distribution than those in the lowest incomecategory (income less than $10000). Thosewho almost never worry about bills are signifi-cantly less supportive of redistribution thanthose who worry all of the time. The self-

interest variables are jointly significant at the 1per cent level.

The effect of being white is large and highlysignificant, and the effect of being male is evenlarger. At first glance, this may appear to con-tradict an empirical regularity that among thesocio-economic variables, race has one of thelargest and most reliable effects while sex doesnot. However, if we omit the beliefs variables,the magnitude of the effects of race and sex in-crease and become roughly equivalent in size.This is consistent with the argument, put forthby Gilens (1999), that the effect of race is me-diated by beliefs about the characteristics of thepoor, especially poor blacks.

If we take the view that all of the socio-economic variables together capture self-interest, then the effect of self-interest appearsconsiderably larger than if we simply considerthe size of the coefficient on income. Using or-dered probit to estimate similar equations,Fong (2001) has estimated the sizes of the ef-fects of the independent variables on the prob-abilities of scoring in each of the six categories

Figure 2 Effects on the Support for Redistribution of Beliefs about the Importance of Various Factors in Getting Ahead in Life

Notes

: The bars represent ordinary least squares coefficients predicting support for redistribution. The dependent variableis standardised. The independent variables are the respondent’s belief in the importance of the factor shown to getting aheadin life. The coefficients are the estimated effects of a one-point increase in the response scale for a given belief on standarddeviations of support for redistribution. The regressions also include all of the self-interest measures included in Figure 1,R

2

= 0.184. The number of observations is 3437. This regression uses sample weights, although the results were not sensitiveto them. *** indicates significance at the 1 per cent level. ** indicates significance at the 5 per cent level.

Regression

coefficient


©


of the support for redistribution scale. She esti-mates an equation that controls for beliefsabout the causes of wealth and poverty as wellas a large number of objective and subjectivemeasures of pecuniary interests in redistribu-tion. She found that the effects of the beliefsmeasures are as large as the effect of being themost privileged type of person in the datasetcompared to being the least privileged type ofperson. More specifically, the effect of being inthe least privileged socio-economic category(non-white, female, single, union member,part-time worker, no college education, in low-est income category, household size greaterthan four, and almost always worries aboutbills) as opposed to the most privileged cate-gory is similar in size to the effects of beliefsthat good luck causes wealth and bad luckcauses poverty compared to beliefs that cir-cumstances cause wealth and poverty.

Could our results be driven by missing self-interest variables? People who believe thatpoverty is caused by bad luck or circumstancesbeyond individual control may be those whohave low-mean, high-variance incomes. Suchindividuals may have higher expectations ofneeding government assistance in the future,and therefore demand more redistributionpurely out of self-interest. For similar reasons,those who believe that the poor are lazy maysimply be people who have high-mean, low-variance incomes and therefore less self-interest in redistribution. If this is true, then theeffect of these beliefs on redistributive policypreferences may have nothing to do with thepsychology of holding the poor accountableand blaming them for their outcomes. It wouldsimply be the case that beliefs about the causesof income are correlated with a person’s finan-cial position which in turn determines his or herdemand for redistribution.

If the beliefs about the causes of poverty andwealth operate through self-interest, then theyshould have no effect among people at the topand bottom of the distribution of income whoexpect to remain there. Those who do not ex-pect to benefit should demand no redistributionat all, regardless of their beliefs about the causesof income, while those who expect to benefitshould register the highest degree of support for

redistribution regardless of their beliefs aboutthe causes of income. To test whether or not thisis the case, we use sub-samples of (i) individu-als with household incomes over $75000 peryear who expect to be better off in five yearsthan they are today, and who ‘almost never’worry about bills, and (ii) individuals withhousehold incomes under $30000 per year whodo not expect to be better off in five years thanthey are today, and who worry about bills moreoften than ‘almost never’.

In both of these sub-samples, a quite in-clusive set of measures capturing self-interestis jointly insignificant.

6

That is, we cannot re-ject the hypothesis that every single socio-economic variable has a coefficient of zero.

Yet, the beliefs about the roles of luck, effortand opportunity in generating life outcomeswere jointly significant for both sub-samples,and in most cases were individually significantin the expected directions as well.

7

Thus,among those who are poor and do not expecttheir lives to improve, those who believe thatlack of effort causes poverty oppose redistribu-tion. Analogously, support for redistribution ishigh among those securely well-off respon-dents who believe that poverty is the result ofback luck.

In another test of self-interest, we use ques-tions on the respondents’ views on the impor-tance of various factors, including a person’srace and sex, to getting ahead in life. Figure 2presents an ordinary least squares regression ofsupport for redistribution on the importance ofvarious determinants of success, controllingfor the same socio-economic variables in-cluded in the regression presented in Figure 1.Beliefs that ‘willingness to take risks’ and‘hard work and initiative’ explain ‘why somepeople get ahead and succeed in life and othersdo not’ have highly significant negative effectson support for redistribution. Beliefs that edu-cation, people’s parents, connections, goodluck, dishonesty and inherited money explainwhy some people get ahead have significantpositive effects on support for redistribution. Inaddition, beliefs that a person’s sex is impor-tant to getting ahead have significant positiveeffects on support for redistribution for men,while the effect of this belief for women is also


©


positive but smaller and insignificant. Beliefsthat a person’s race is important to gettingahead in life have significant positive effectsfor whites, while the effect of these beliefs forblacks is positive but smaller and insignificant.

If people think that a person’s race and sexare important to getting ahead in life, then ef-fects of these beliefs on self-interested demandfor redistribution should operate in opposite di-rections for those who expect to benefit andthose who expect to lose from racial or genderdiscrimination.

8

In other words, whites whothink race is important to getting ahead will ex-pect to be economically advantaged and wouldhave fewer self-interested reasons to supportredistribution than whites who think that racedoes not matter. Similar reasoning holds formen who think a person’s sex is important togetting ahead in life.

However, using an alternative form of the re-gression presented in Figure 2, we find that theeffect of believing that a person’s sex is impor-tant to getting ahead in life is significantlymore positive for men than it is for women.This interaction effect is significant at the 1 percent level (unreported). Similarly, the effect ofbelieving that a person’s race is important togetting ahead in life is significantly more posi-tive for whites than it is for blacks. As we haveseen, this is inconsistent with self-interest, be-cause men and whites with these beliefs wouldexpect to benefit from discrimination andhence have less likelihood of benefiting fromredistributive programs.

Concerns about the incentive effects of taxa-tion are a final mechanism through which self-interest might cause beliefs that the poor arelazy and the rich industrious to decrease the de-mand for redistribution. When earned incomeis more sensitive to work effort, taxation maycause greater effort disincentives and reduceaggregate income. If so, then beliefs about theroles of effort, luck and opportunity in generat-ing income may affect the level of support forredistribution through concerns about incen-tive costs of redistribution (Piketty 1995). Thistype of incentive concern should not apply onlyto redistribution, but to any tax-funded expen-diture, including expenditures such as nationaldefence. According to this tax cost hypothesis,

if beliefs that income is caused by factors underindividual control decrease demand for redis-tribution, then they should decrease demandfor other kinds of tax-funded expenditures, in-cluding defence spending, as well. But there isno evidence that tax cost concerns adverselyaffect the demand for public expenditures.Using the 1990 General Social Survey, we es-timate ordered probit regressions predictingsupport for spending on welfare, national de-fence, halting the rising crime rate, and dealingwith drug addiction, respectively.

9

The inde-pendent variables are beliefs that the poor arepoor because of lack of effort, and five demo-graphic variables (income, education, race, sexand age). The belief that lack of effort causespoverty has a highly significant negative effecton support for redistribution. However, thesesame beliefs have no effect on support forspending on crime or drug addiction, and theyhave a significant positive effect on support forspending on defence. If these beliefs simplymeasure tax cost concerns, then their effect onsupport for all of these expenditure itemsshould have been negative.

However, even more convincing evidenceon this point comes from the experiment in-cluding actual welfare recipients describedabove. There were no disincentive costs at allin this experiment. Yet, student subjects gavemore to the welfare recipients with the strongerwork commitments.

3. Strong Reciprocity and the Welfare State: Unhappy Marriage?

People exhibit significant levels of generosity,even towards strangers. However, their gener-osity is conditional on their beliefs that their re-cipients are morally worthy. It would not bedifficult to design a system of income securityand economic opportunity that would taprather than offend the motivations expressed inthis generalisation. Such a system would begenerous towards the poor, rewarding thosewho perform socially valued work and whoseek to improve their chances of engaging insuch work, as well as to those who are poorthrough accidents not of their own making,such as illness and job displacement.


©


While strong reciprocity may support egali-tarianism, it may also help explain opposition towelfare state policies in some of the advancedmarket economies in the past decades. Specifi-cally, in light of the empirical regularities out-lined above, we suspect the following to be trueas well: egalitarian policies that reward peopleindependently of whether and how much theycontribute to society are considered unfair andare not supported, even if the intended recipi-ents are otherwise worthy of support, and evenif the incidence of non-contribution in the targetpopulation is rather low. This would explain theopposition to many welfare measures for thepoor, particularly since such measures arethought to have promoted various social pathol-ogies. At the same time it explains the continu-ing support for social security and Medicare inthe United States, since the public perception isthat the recipients are ‘deserving’ and the poli-cies are thought not to support what are consid-ered anti-social behaviours.

A striking fact about the decline in the sup-port for the former Aid to Families with Depen-dent Children, Food Stamps and other means-tested social support programs in the UnitedStates, however, is that overwhelming majori-ties oppose those policies, whatever their in-come, race, or personal history with suchprograms. This pattern of public sentiment, wethink, can be accounted for in terms of the prin-ciple of strong reciprocity.

We rely mainly on two studies. The first,Farkas and Robinson (1996), analyses data col-lected in late 1995 by Public Agenda, a non-profit, non-partisan research organisation. Theauthors conducted eight focus groups aroundthe country, then did a national survey, involv-ing half-hour interviews, of 1000 randomly se-lected Americans, plus a national oversampleof 200 African Americans. The second, politi-cal scientist Gilens’

Why Americans Hate Wel-fare

(1999), is an analysis and review ofseveral polls executed during the 1990s andearlier by various news organisations.

10

In the Public Agenda survey 63 per cent ofrespondents thought the welfare system shouldbe eliminated or ‘fundamentally overhauled’while another 34 per cent thought it should be‘adjusted somewhat’. Only 3 per cent approved

of the system as is. Even among respondentsfrom households receiving welfare, only 9 percent expressed basic approval of the system,while 42 per cent wanted a fundamental over-haul and an additional 46 per cent wanted someadjustments (Farkas and Robinson 1996, p. 9).

The cost of welfare programs cannot explainthis opposition. While people generally over-state the share of the federal budget devoted towelfare (Farkas and Robinson 1996, p. 9), thiscannot account for the observed opposition.

11

Farkas and Robinson (1996, pp. 9–10) note:

By more than four to one (65% to 14%), Ameri-cans say the most upsetting thing about welfare isthat ‘it encourages people to adopt the wrong life-style and values,’ not that ‘it costs too much taxmoney.’ … Of nine possible reforms presented torespondents—ranging from requiring job trainingto paying surprise visits to make sure recipientsdeserve benefits—reducing benefits ranked last inpopularity (Table 4).

The cost, apparently, is not the problem. Infocus groups:

Participants invariably dismissed arguments aboutthe limited financial costs of welfare in almostderisive terms as irrelevant and beside the point. [Farkas and Robinson 1996, p. 10]

Nor can the perception of fraud account for thisopposition. It is true that 64 per cent of respon-dents (and 66 per cent of respondents on wel-fare) believe welfare fraud is a seriousproblem. However, most do not consider itmore serious than in other government pro-grams, and only 35 per cent of survey respon-dents would be more ‘comfortable withwelfare’ if fraud were eliminated (Farkas andRobinson 1996, pp. 11, 12).

In commenting on this fact Gilens (1999, p.2) observes that:

Politics is often viewed, by élites at least, as a pro-cess centered on the question ‘who gets what.’ Forordinary Americans, however, politics is moreoften about ‘who

deserves

what,’ and the welfarestate is no exception.

In the Public Agenda study, respondents over-whelming consider welfare to be unfair toworking people and addictive to recipients. Bya more than five to one margin (69 per cent to


©


13 per cent overall, and 64 per cent to 11 percent for people receiving welfare), respondentssay that recipients abuse the system (for in-stance, by not looking for work) rather than ac-tually cheat it (for example, by collectingmultiple benefits) (Farkas and Robinson 1996,p. 12). Moreover, 68 per cent think (59 per centof welfare recipients) that welfare is ‘passed onfrom generation to generation, creating a per-manent underclass’ (Farkas and Robinson1996, p. 14). In the same vein, 70 per cent (71per cent of welfare recipients) say welfaremakes it ‘financially better for people to stayon welfare than to get a job’, 57 per cent (62 percent of welfare recipients) think welfare en-courages ‘people to be lazy’ and 60 per cent(64 per cent of welfare recipients) say the wel-fare system ‘encourages people to have kidsout of wedlock’ (Farkas and Robinson 1996, p.15). Note that the welfare recipients and othercitizens hold similar views in this respect.

That the respondents are correct in thinkingthat the welfare state causes these behaviours isbeside the point. Whether or not, for example,welfare

causes

out-of-wedlock births, for ex-ample, or fosters an unwillingness to work, cit-izens object that the system provides financialsupport for those who undertake these sociallydisapproved behaviours. Their desire is to bearwitness against the behaviour and to disassoci-ate themselves from it, whether or not their ac-tions can change it.

Racial stereotyping and opposition to wel-fare are closely associated. The Public Agendasurvey shows that whites are much more likelythan African Americans to attribute negativeattributes to welfare recipients, and much morelikely to blame an individual’s poverty on lackof effort. The survey data show, writes Gilens(1999, p. 3), that:

For most white Americans, race-based oppositionto welfare is not fed by ill will toward blacks, noris it based on whites’ desire to maintain their eco-nomic advantages over African Americans. In-stead race-based opposition to welfare stems fromthe specific perception that, as a group, AfricanAmericans are not committed to the work ethic.

There is some evidence that people are moretolerant of redistributions within ethnic and ra-

cial categories than between them. Luttmer(2001) found for a US sample that individualsare more strongly opposed to welfare if theylive in neighbourhoods where a higher percent-age of welfare recipients is of a different race.Luttmer’s findings are consistent with our rec-iprocity interpretation of redistributive politics,in light of the evidence that when people iden-tify with a social group, they are more likely toblame outgroup members (holding them indi-vidually responsible) for their bad outcomesand behaviours and to give them little credit(holding factors other than the outgroup mem-ber’s voluntary control responsible) for theirgood outcomes and behaviours (Brewer andMiller 1996). However, the salience of race inLuttmer’s US data may not be as pronounced inother cultural contexts, since the characteristicsthat determine who are ‘insiders’ and who are‘outsiders’ is culturally specific.

Taking account of the

fact

that many Ameri-cans see the current welfare system as a viola-tion of deeply held reciprocity norms does notrequire that policy makers adopt punitive mea-sures and stingy budgets for the poor. Indeedthe public strongly supports income supportmeasures when asked in ways that make clearthe deserving nature of the poor: a 1995 NYT/CBS poll, for instance, found that twice asmany agreed as disagreed that ‘it is the respon-sibility of the government to take care of peo-ple who can’t take care of themselves’.

4. Conclusion

Like Kropotkin (1989 (1903)) a century ago,we find compelling evidence for the force ofhuman behavioural predispositions to act bothgenerously and reciprocally rather than self-interestedly in many social situations. Whilemany economists have failed to appreciate thepractical importance of these predispositions inpolicy matters, their salience was not missed byHayek (1978, pp. 18, 20):

[The] demand for a just distribution … is … an at-avism, based on primordial emotions. And it isthese widely prevalent feelings to which prophets,[and] moral philosophers … appeal by their plansfor the deliberate creation of a new type of soci-ety.


©


If we are right, economists have misunder-stood both the support for the welfare state andthe revolt against welfare (where it has oc-curred), attributing the latter to selfishness bythe electorate rather than the failure of manyprograms to tap powerful commitments to fair-ness and generosity and the fact that some pro-grams appear to violate deeply held reciprocitynorms. Egalitarians have been successful in ap-pealing to the more elevated human motivesprecisely when they have shown that dominantinstitutions violate norms of reciprocity, andmay be replaced by institutions more consis-tent with these norms (Bowles and Gintis 1995;Bowles and Gintis 1996; Moene and Waller-stein 2002).

To mobilise rather than offend reciprocalvalues, policies should recognise that there issubstantial support for generosity towards theless well off as long as they have provided ortried to provide a quid pro quo and are in goodstanding. The task of politically viable egalitar-ian policy design might thus begin by identify-ing those behaviours that entitle an individualto reciprocation. Among those in the UnitedStates today would be saving when one’s in-come allows and working hard and taking risksin both productive endeavours and schooling.Persistent poverty is often the result of low re-turns to these socially admired behaviours: lowwages for hard work, a low rate of return onsavings, costly access to credit for those wish-ing to engage in uncertain entrepreneurial ac-tivities, and educational environments soadverse as to frustrate even the most diligentstudent. Policies designed to raise the returns tothese activities when undertaken by the lesswell off would garner widespread support. Asecond principle of reciprocity-based policydesign should be to insure individuals againstthe vagaries of bad luck without insuring themagainst the consequences of their own actions,particularly when these actions violate widelyheld social norms against such things as illicitdrug use or child bearing in the absence of rea-sonable guarantees of adequate parenting.

Many traditional projects of egalitarians,such as land reform and employee ownershipof their workplaces are strongly consistent withreciprocity norms, because they make people

the owners not only of the fruits of their la-bours, but more broadly of the consequences oftheir actions (Bowles and Gintis (1998) andBowles and Gintis (1999) provide overviewsbased on contemporary principal–agent mod-els). The same may be said of more conven-tional initiatives such as improved educationalopportunity and policies to support home own-ership. There is good evidence, for example,that home ownership promotes active partici-pation in local politics and a willingness to dis-cipline personally those engaging in anti-socialbehaviours in the neighbourhood (Sampson,Raudenbush and Earls 1997). An expansion ofsubsidies designed to promote employmentand increase earnings among the poor, sug-gested by Phelps (1997), would tap powerfulreciprocity motives. Similarly, social insuranceprograms might be reformulated along linessuggested by Roemer (1993) to protect individ-uals from risks over which they have no con-trol, while not indemnifying people against theresults of their own choices, other than provid-ing a minimal floor to living standards. In thismanner, for example, families could be pro-tected against regional fluctuations in homevalues—the main form of wealth for most peo-ple—as Shiller (1993) has shown. Other formsof insurance could partially protect workersfrom shifts in demand for their services in-duced by global economic changes.

An egalitarian society can be built on thebasis of these and other policies consistent withstrong reciprocity, along with a guarantee of anacceptable minimal living standard consistentwith the widely documented motives of basicneeds generosity. But if we are correct, eco-nomic analysis will be an inadequate guide topolicy making in the area unless it revises itsfoundational assumptions concerning humanmotivation.

June 2005

Endnotes

1. Strong reciprocity goes beyond self-interested forms of cooperation, which includetit-for-tat and what biologists call

reciprocalaltruism

(Trivers 1971).


©


2. The numbers of observations for these ques-tions were 78 and 79 for the poor group and294 and 281 for the rich group. Gilens (1999)makes similar observations using earlier data.

3. See Moffitt (1992) for a model of welfarestigma. See also Lindbeck, Nyberg andWeibull (1999) for related work that addressesthe role of work norms in redistributive politicsand treats such norms as endogenous to theprovision of government transfers.

4. We drop non-responses and ‘don’t know’responses. Another option would be to include‘don’t know’ as a valid response. However,how and why people develop well-definedpreferences and beliefs are beyond the scopeof this article. We focus on why people op-pose or support income redistribution giventhat their beliefs and preferences are well de-fined.

5. There are several additional questions thatmight capture self-interest that are excludedfrom the model presented here. See Fong(2001) for a discussion and analysis of thesevariables.

6. The control variables are income, age, age

2

,dummies for some college education or more,white, male, employed part-time, not em-ployed (the omitted employment status cate-gory is employed full-time), married, unionmember, four occupation dummies, and havinga child under the age of 18 living in the house-hold. The regression on the less privileged sub-sample also includes dummies for ‘worriesabout bills most of the time’, and ‘worriesabout bills some of the time’ (the omitted cate-gory is ‘worries about bills all of the time’).The regression on the more privileged sub-sample contains no dummies for worryingabout bills because all of the respondents in thissub-sample gave the same response: they ‘al-most never’ worry about bills.

7. Space limitations prevent us from present-ing these results here. However, the findingsusing ordered probit are presented in Fong(2001).

8. We assume that people agree on whichgroup benefits and which loses when they be-lieve that a person’s race or sex is important togetting ahead.

9. The sample size in these regressions rangesfrom 584 to 594.

10. A third study by Weaver, Shapiro and Ja-cobs (1995), drawing in addition on NationalOpinion Research Center and General SocialSurvey data, comes to broadly similar conclu-sions.

11. As a general rule non-experts vastly over-state the share of the tax revenues devoted tothings of which they disapprove, whether it beforeign aid, welfare, aids research, or militaryexpenditure—the opposition is generally thecause of the exaggeration, not vice versa.

References

Alesina, A., Glaeser, E. and Sacerdote, B. 2001,‘Why doesn’t the United States have a Euro-pean-style welfare state?’,

Brookings Paperson Economic Activity

, no. 2, pp. 187–278.Atkinson, A. B. 1999,

The Economic Conse-quences of Rolling Back the Welfare State

,MIT Press, Cambridge.

Besley, T. and Coate, S. 1992, ‘Understandingwelfare stigma: Taxpayer resentment andstatistical discrimination’,

Journal of PublicEconomics

, vol. 48, pp. 165–83.Bowles, S. and Gintis, H. 1995, ‘Escaping the

equity-efficiency tradeoff: Productivity en-hancing redistributions’, in

MacroeconomicPolicy after the Conservative Era: Researchon Investment, Savings, and Finance

, Cam-bridge University Press, Cambridge.

Bowles, S. and Gintis, H. 1996, ‘Efficient re-distribution: New rules for markets, states,and communities’,

Politics & Society

, vol.24, pp. 307–42.

Bowles, S. and Gintis, H. 1998, ‘Schooling,skills and earnings: A principal–agent ap-proach’, in

Meritocracy and Economic In-equality

, eds K. Arrow, S. Bowles and S.Durlauf, Princeton University Press, Prince-ton, New Jersey.


©


Bowles, S. and Gintis, H. 1999,

RecastingEgalitarianism: New Rules for Communi-ties, States and Markets

, ed. E. O. Wright,Verso, London.

Brewer, M. B. and Miller, N. 1996,

IntergroupRelations

, Brooks/Cole Publishing Com-pany, Pacific Grove.

Camerer, C. 2003,

Behavioral Game Theory:Experiments in Strategic Interaction

, Prince-ton University Press, Princeton, New Jersey.

Deutsch, M. 1985,

Distributive Justice

, YaleUniversity Press, New Haven.

Farkas, S. and Robinson, J. 1996,

The ValuesWe Live By: What Americans Want fromWelfare Reform

, Public Agenda, New York.Fehr, E. and Schmidt, K. M. 2005, ‘Theories of

fairness and reciprocity: Evidence and eco-nomic applications’, in

Advances in Eco-nomics and Econometrics, Eighth WorldCongress of the Econometric Society

, eds M.Dewatripont, L. Hansen and S. Turnovsky,Cambridge University Press, Cambridge.

Fong, C. M. 2001, ‘Social preferences, self-interest, and the demand for redistribution’,

Journal of Public Economics

, vol. 82, pp.225–46.

Fong, C. M. 2005, ‘Empathic responsiveness:Evidence from a randomized experiment ongiving to welfare recipients’, mimeo, De-partment of Social and Decision Sciences,Carnegie Mellon University.

Fong, C. M., Bowles, S. and Gintis, H. 2005,‘Reciprocity and the welfare state’, in

MoralSentiments and Material Interests: On theFoundations of Cooperation in EconomicLife,

eds H. Gintis, S. Bowles, R. Boyd andE. Fehr, MIT Press, Cambridge.

Gallup Organization 1998,

Haves and Have-Nots: Perceptions of Fairness and Opportu-nity

, Gallup, Washington, DC.Gilens, M. 1999,

Why Americans Hate Wel-fare

, University of Chicago Press, Chicago.Hayek, F. 1978,

The Three Sources of HumanValues

, London School of Economics, Lon-don.

Heclo, H. 1986, ‘The political foundations ofantipoverty policy’, in

Fighting Poverty:What Works and What Doesn’t

, eds S. H.Danziger and D. H. Weinberg, Harvard Uni-versity Press, Cambridge, Massachusetts.

Kolm, S.-C. 1984,

La Bonne Economie: LaRéciprocité Générale

, Presses Universitairesde France, Paris.

Kropotkin, P. 1989 (1903),

Mutual Aid: A Fac-tor in Evolution

, Black Rose Books, NewYork.

Lindbeck, A., Nyberg, S. and Weibull, J. 1999,‘Social norms and economic incentives inthe welfare state’,

Quarterly Journal of Eco-nomics

, vol. 114, pp. 1–35.Luttmer, E. F. P. 2001, ‘Group loyalty and the

taste for redistribution’,

Journal of PoliticalEconomy

, vol. 109, pp. 500–28.Miller, D. 1999,

Principles of Social Justice

,Harvard University Press, Cambridge, Mas-sachusetts.

Moene, K. O. and Wallerstein, M. 2002, ‘In-equality, social insurance and redistribu-tion’,

American Political Science Review

,vol. 95, pp. 859–74.

Moffitt, R. 1992, ‘Incentive effects of the U.S.welfare system: A review’,

Journal of Eco-nomic Literature

, vol. 30, pp. 1–61.Moffitt, R., Ribar, D. and Wilhelm, M. 1998,

‘Decline of welfare benefits in the U.S.: Therole of wage inequality’,

Journal of PublicEconomics

, vol. 68, pp. 421–52.Page, B. and Shapiro, R. 1992,

The RationalPublic: Fifty Years of Trends in American’sPolicy Preferences

, University of ChicagoPress, Chicago.

Phelps, E. S. 1997,

Rewarding Work: How toRestore Participation and Self-Support toFree Enterprise

, Harvard University Press,Cambridge, Massachusetts.

Piketty, T. 1995, ‘Social mobility and redistrib-utive politics’,

Quarterly Journal of Eco-nomics

, vol. 110, pp. 551–84.Roberts, K. 1977, ‘Voting over income tax

schedules’,


,vol. 8, pp. 329–40.

Roemer, J. 1993, ‘A pragmatic theory of re-sponsibility for the egalitarian planner’,

Phi-losophy and Public Affairs

, vol. 22, pp. 146–66.

Roemer, J. E. 1998,

Equality of Opportunity

,Harvard University Press, Cambridge.

Roemer, J. E., Aaberge, R., Colombino, U.,Fritzell, J., Jenkins, S. P., Lefranc, A., Marx,I., Page, M., Pommer, E., Ruiz-Castillo, J.,


©


San Segundo, M. J., Tranaes, T., Trannoy,A., Wagner, G. G. and Zubiri, I. 2003, ‘Towhat extent do fiscal regimes equalize op-portunities for income acquisition among cit-izens?’,


, vol.87, pp. 539–65.

Sampson, R. J., Raudenbush, S. W. and Earls,F. 1997, ‘Neighborhoods and violent crime:A multilevel study of collective efficacy’,

Science

, vol. 277, pp. 918–24.Shiller, R. J. 1993,

Macro Markets: CreatingInstitutions for Managing Society’s LargestEconomic Risks

, Clarendon Press, Oxford.Skitka, L. and Tetlock, P. 1993, ‘Providing

public assistance: Cognitive and motiva-tional processes underlying liberal and con-servative policy preferences’,

Journal ofPersonality and Social Psychology

, vol. 65,pp. 1205–23.

Sobel, J. 2001, ‘Interdependent preferencesand reciprocity’, mimeo, University of Cali-fornia, San Diego.

Trivers, R. L. 1971, ‘The evolution of recipro-cal altruism’,

Quarterly Review of Biology

,vol. 46, pp. 35–57.

Walster, E., Walster, G. W. and Berscheid, E.1978,

Equity, Theory and Research

, Allynand Bacon, Boston.

Weaver, R. K., Shapiro, R. Y. and Jacobs, L. R.1995, ‘Poll trends: Welfare’,

Public OpinionQuarterly

, vol. 39, pp. 606–27.Weiner, B. 1995,

Judgments of Responsibility:A Foundation for a Theory of Social Con-duct

, Guilford Press, New York.Williamson, J. B. 1974, ‘Beliefs about the mo-

tivation of the poor and attitudes toward pov-erty policy’,

Social Problems

, vol. 21, pp.734–47.

Documents

Behavioural Motives for Income Redistribution