Racism, Xenophobia, and Redistribution by

1
Racism, Xenophobia, and Redistributionby

• Woojin Lee
• John Roemer
• Karine Van der Straeten

based on Racism, Xenophobia, and Redistribution
Multi-dimensional politics in advanced
democracies (Harvard UP, 2007)
2
Racism and division

• old left racism divides the working class
• e.g. Factories in the fields (McWilliams)
• electoral politics to what extent does voter
  racism reduce fiscal redistn?
• Alesina, Glaeser and Sacerdote
• econometric evidencelarger poor
  minoritycorrelates with less redistribution, in
  cross-sectional analysis

3
Two effects

• political competition over policy space with two
  issues redistribution and race/immigration
• the anti-solidarity effect
• imagine a change in preferences (CivilRightsMvmt)
• the policy bundle effect
• political portfolio effect
• Southern Ds decamp to R party
• note no change in preferences

4
(No Transcript)
5
Plan of the paper

• Propose a model of political competition between
  two parties, where voters have preferences over
  taxation and the race issue
• Calibrate the model to US presidential elections,
  1972-1992
• Alter the preferences of voters to reduce racism,
  and compute what the equilibrium on tax policy
  would have been
• Conclude that the difference in taxation between
  reality and counterfactual is due to effects of
  racism

6
Modeling political equilibrium
Each party to contain factions, to model the
conflict between politicians who make their
career by maximizing probability of victory, and
those who make their career by representing
constituents A political equilibrium with two
parties is a Nash equilm between parties, where
best response is defined by a bargaining
equilm within each party, between its factions
7
Multi-dim political competition


8
Probability of victory
Given define Then vote for
policy will be
where X is a random vble, distributed
(say) uniformly on (-?, ?) So
9
Equilibrium with party factions(PUNE)
1) a partition 2) a pair of numbers kD, kR 3) a
pair of policies such that 4)
10
Verbal summary
An equilibrium is 1. a partition of the polity
into two groups, with a party representing each
group 2.Each party proposes a position on income
taxation and the race issue , which is the
consequence of bargaining within the partys
activists between Opportunists and Guardians,
reached when facing the other partys proposal 3.
Representation is good in the sense that each
party members interests are represented with
equal weight by the Guardians 4. Each party
member prefers her partys proposal to the other
partys proposal
11
In sum.

• the data of the model are T,H,v,F,?,n
• the output of the model is a partition of H into
  n parties, a set of n policies, and a
  probability of victory for each party
• models conflict between Opportunists and
  Constituents in a party. Office holding and
  representation.
• note that party composition is endogenous --
  its part of the equilibrium. Party composition
  policies change as distribn of voter types
  changes
• see Political Competition (2001) for detailed
  discussion of PUNE, and how to compute them

12
Multiplicity of equilibria

• henceforth, take n 2.
• indeed, there is a 2 dim manifold of PUNEs,
  because they exist for a continuum of pairs
  (kD,kR) .
• the ks can be viewed as reflecting bargaining
  power of Constituents in internal bargaining v.
  Opportunists. If we knew the relative bargaining
  powers of the Opportunists and Constituents in
  each party, we could compute locally unique
  equilibrium (or perhaps none!)

13

• APPLICATION TO OUR PROBLEM
• H (w,?) where w is the wage of the voter and
  ? is his ideal position on the race issue
• F is the distribution of these types, measured
  from ANES and US-PSID data
• T is the set of ordered pairs (t,r) where t is an
  income tax rate for an affine tax policy, r is
  partys position on the race issue
• Thus the utility function of the voter is
  represented as v(t,rw, ?)

14
Application to our problem (2)
? is the economic sub-utility function
derived from Cobb-Douglas direct Utility fcn in
leisure and income

• the richer or more racist a voter is, the less
  she likes equality (if ?1 gt0 ). But we find that
  .
• note that ?2 captures the anti-solidarity effect?

15
Measuring racism of voters factor analysis

• We decompose political ideology
  (liberal-conservative) into four orthogonal
  latent factors racism, libertarianism,
  feminism, and compassion for the poor by
  carrying out a sequence of factor analyses on 10
  variables in the NES.
• We mean by voter racism a position favoring what
  are conventionally viewed as conservative
  policies on the race issue-- ranging from
  opposition to affirmative action to support for
  prison construction-- induced by anti-black
  affect and/or the belief that civil rights
  movement is pushing too fast.
• Our definition is minimal and narrower than the
  conception of symbolic racism, which attributes
  response of blacks lack work ethic to racism.
  Instead of including this attitude as a component
  of racism, we explain it in terms of more
  narrowly defined racism.

16
Our empirical findings

• Racism is the single most important factor in
  explaining various racial attitudes, in terms of
  the size of the coefficient as well as
  statistical significance. In contrast to popular
  political rhetoric, libertarianism plays very
  little role in explaining racial attitudes,
  except for the aid-to-blacks question.
• It is often argued that whites oppose racially
  liberal policies because they believe that
  blacks lack an individualistic work ethic, an
  important component of American creed, and that
  this is belief is race-neutral.
• If this contention were true, we would expect
  that libertarianism, which is racism-free by
  construction, would have a highly significant
  coefficient on various racial attitudes but it
  does not.
• It appears that it is not racism-free
  libertarianism but racism camouflaged behind
  libertarian rhetoric that explains much of the
  white opposition to various racial policies in
  the United States.

17
Factor Analysis, political ideology

• 1976 (obs.497)
• Variable 1 2 3 4
  Uniqueness
• feminism compassion lib
  ertarianism racism
• Antiblack affect 0.01822 -0.07963 -0.07387 0.82937
  0.30001
• Civil rights push too fast -0.20691 0.0181 0.19432
  0.69206 0.44015
• Poor thermometer -0.05059 0.74197 0.12747 0.01666
  0.43039
• Welfare thermometer 0.07812 0.77161 -0.08294 -0.19
  544 0.35343
• Union thermometer 0.06135 0.58147 -0.40796 0.18032
  0.45918
• Strong government 0.08319 -0.05828 0.7398 0.07416
  0.43688
• Trust government 0.07357 -0.01298 -0.70663 0.00232
  0.49509
• Women equal -0.80756 0.25927 0.0293 0.08598 0.2723
  8
• Women liberty thermometer0.81714 0.24631 0.05921 0
  .02117 0.26766
• Political Ideology -0.54612 -0.19377 0.10397 0.228
  27 0.60128
• Eigenvalues 2.03891 1.59067 1.1905 1.12346
• Difference 0.44824 0.40017 0.06704 0.23372
• Proportion 0.2039 0.1591 0.1191 0.1123
• Cumulative 0.2039 0.363 0.482 0.5944

the racism factor is much more significant in
explaining aid to blacks answers than the
libertarian factor (not shown here)

18
Strategy
unknown parameters
Step 1. calibrate the model, compute distn F of
(w,?). To calibrate the parameters, we fit the
PUNEs to the data. Call the observed (average)
equilibrium ( Step 2. Compute equilibrium in a
counterfactual where we assume the only election
issue is the tax rate. (So preferences over r
dont play a role). Call the new equilibrium
The policy bundle effect is Step 3. Compute
equilm in a counterfactual where t is only issue
and . Denote equilm
. Then total effect of racism is
19
Note that this general equilibrium approach
allows the composition of the parties and a
fortiori their policies to change in the two
counterfactuals. I.o.w., we compute how the
policies of the two parties would change if
voters were less racist, as opposed to how
support for the present policies of the two
parties would change were voters less racist.
This is the purchase of having a full model of
political equilibrium with endogenous parties.
20
Empirics

• We compute sets of parameters for the utility
  function to fit pairs of election years,
  1972-1992.
• We find distn of racial views is almost
  independent of income, so we assume this to be so
  for computational simplicity
• Racial views have skewed to the right in recent
  years

21
Distribution of racial views 1976blue,
1984red, 1992green
22
US Affine tax policy
23
(No Transcript)
24
(No Transcript)
25
Bargaining power distributions
26
(No Transcript)
27
Summary of four experiments

• Each calculation pools data from a pair of
  consecutive elections e.g., 1976- 1980,
  1980-1984,
• The total effect of racism is to reduce the
  income tax rate by 10 -18 percentage points.
  This is probably an overestimate
• The policy bundle effect is between 26 and 50
  of the total effect the rest is the
  anti-solidarity effect
• Thus tax rates in the US are as low as they are
  in large part because the R party was able to
  capture voters with a conservative position on
  the race issue, thus allowing it to maintain tax
  rates which are too low

28
(No Transcript)
29
(No Transcript)
30
The PBE and ASE in UK, 1997

31
Voter separation hyperspace for UK 1997
32
Other countries
We applied similar models to UK, Denmark, and
France. For Denmark and France, we used a
simpler, Euclidean utility function

• Here the space of types is H (?,?) where ? is
  the voters position on desired size of the
  public sector (taken from voter survey data)
• This model requires a different way of separating
  the ASE from the PBE

33
(No Transcript)
34
Conclusions

• Both the ASE and PBE are significant in US
  politics during the 1976-92 period
• Absent voter racism, US taxes would look more
  like those in northern Europe
• Absent racism as an issue, the partition of
  voters into Ds and Rs would be more sharply
  along class lines, whereas now racial views are a
  significant cause of partition

35

• During the last 20 years, inequality has
  increased and tax rates have gone down. This is
  inexplicable with the Downsian model. But it is
  natural with the 2-dim PUNE model. Indeed, the
  correlation between state income per capita and
  Republican vote is negative in the last two US
  elections
• Much of the cause of low taxation in the US is
  policy bundling. It is a mistake to view it to be
  entirely the result of anti-solidarity.
  Econometric analysis which ignores this fact
  (e.g., Alesina, Glaeser, Sacerdote) misspecifies
  the model.

36

• Anti-immigrant feeling is having a similar but
  smaller effect in Europe, with the following
  caveat one can discriminate against immigrants
  in social policy, who are non-citizens, but not
  against blacks in the US who are citizens.
  Therefore, the anti-solidarity effect may be
  smaller in Europe.
• France is peculiar the Extreme Right(Le Pen) is
  moderate on the public-sector issue. Not so in
  DK and UK.

37
On method

• Downsian model inherently mis-specifies the
  problem, by requiring unidimensional policy
  space.
• Not only does PUNE enable analysis of
  multi-dimensional politics, but it represents
  parties, realistically, as a conflict between
  representation and office-holding motives
• Increasing the dimensionality of type space
  and/or policy space would probably reduce the
  effect of racism on taxation
• end