Content uploaded by Zhipeng Yan
Author content
All content in this area was uploaded by Zhipeng Yan on Oct 31, 2017
Content may be subject to copyright.
KYKLOS, Vol. 56 – 2003 – Fasc. 4, 441–456
441
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
Rent Seeking into the Income Distribution
William F. Shughart II, Robert D. Tollison and Zhipeng Yan*
I. INTRODUCTION
There is a sizeable positive economics literature on the effect of government on
the distribution of income. Mueller (1989, pp. 448–58) offers a good summary
of such work. Obviously, the distributional impact of the public sector depends
on the behavioral features of the model the analyst employs – and the assump-
tions he adopts – with respect to the incidence of public spending, taxes, debt
and regulation. Owing to the wide range of analytical possibilities, no consen-
sus has emerged concerning the magnitude and direction of government’s im-
pact on the income distribution. Some scholars have argued that the middle
class is the chief beneficiary of government-mediated wealth transfers (Stigler
1970), while others have found that, on balance, redistribution runs from the
rich to the poor (Reynolds and Smolensky 1977). Assessing these conflicting
conclusions, Mueller (1989, p. 455) observes that the supposition ‘that some
government policies – taxes or expenditures – are intended to confer redistri-
butional gains on particular interest groups cannot be questioned.’ He goes on
to say, however, that ‘what the literature does not illuminate is the amount of
government activity explained in this way and its net impact on the distribution
of income.’
This paper begins closing the gap identified by Mueller. In particular, we
ask, what is the net impact of interest groups on the distribution of income? Be-
* Shughart: F. A. P Barnard Distinguished Professor of Economics, Department of Economics,
The University of Mississippi, P. O. Box 1848, University, MS 38677-1848 USA, Phone (662)
915-7579, Fax (662) 915-6943, e-mail: shughart@olemiss.edu; Tollison: Robert M. Hearin Pro-
fessor of Economics, Department of Economics, The University of Mississippi, P. O. Box 1848,
University, MS 38 677–1848 USA, and Professor of Economics, Clemson University, 222 Sir-
rine Hall, Clemson, SC 29634 USA, Phone (864) 656-0483, Fax (864) 656-4192, e-mail: rtollis
@clemson.edu; Yan: Graduate Student, Department of Economics, Brandeis University, MS
032, 415 South Street, Waltham, MA 02454 USA, Phone (781) 736-4818, Fax (781) 736-2267,
e-mail: yanzp@brandeis.edu. We benefited from the comments of Gökhan Karahan, Michael
Reksulak, two anonymous referees and the journal’s editors. As is customary, however, we
accept full responsibility for any remaining errors.
Kyklos_2003-04_S-439-440.book Seite 441 Donnerstag, 30. Oktober 2003 7:36 07
WILLIAM F. SHUGHART II/ROBERT D. TOLLISON/ZHIPENG YAN
442
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
cause the interest-group theory of government (McCormick and Tollison 1981)
makes no a priori prediction about how incomes are impacted by interest-
group activity in the polity, our approach is strictly empirical. The paper is or-
ganized as follows. Section II briefly explores the analytical possibilities. Em-
pirical evidence from the U.S. states is presented in Section III. There we find
that, other things being the same, incomes are distributed more unequally (the
Gini coefficient is higher) in states where special interest groups exert greater
influence on the political process. Section IV concludes.
II. INTEREST GROUPS AND THE DISTRIBUTION OF INCOME
The most venerable theory of how government affects the distribution of in-
come is Director’s Law, which is based on the familiar median voter model. In
that theory, the middle class benefits from government intervention at the ex-
pense of the upper- and lower income classes (Stigler 1970). In the United
States, at least, that prediction is borne out by public education, the largest ex-
penditure category at the state and local levels of government. Obviously, how-
ever, redistribution from the tails toward the middle might produce greater or
lesser income inequality, depending on the relative amounts of wealth trans-
ferred away from the poor and the rich. As such, Director’s Law does not gen-
erate sharp predictions about the public sector’s impact on the income distribu-
tion; it unambiguously implies only that median income will increase1.
Director’s Law and most other theories of the impact of government on the
distribution of income nevertheless have been interpreted as suggesting that in-
tervention by the public sector will have leveling effects, producing a more
equal distribution of income. Income tax codes with progressive rate schedules,
combined with publicly financed health care programs for the poor and the eld-
erly, public pensions and the many other ornaments of the modern welfare
state, ostensibly are intended to raise incomes at the lower end of the distribu-
tion and to reduce them at the upper end.
The interest-group theory of government is more catholic. It suggests that
no one segment of the income distribution unfailingly is advantaged (or disad-
vantaged) by the public sector’s redistributive activities; its lesson is that the
1. Wagner’s Law suggests that causality also runs in the opposite direction. If government is an in-
come-normal good, increases in income lead to increases in public spending. That prediction has
been tested extensively. Congleton and Shughart (1990), for example, find that social security
benefits in the United States are positively correlated with the income of the median voter, ce-
teris paribus.
Kyklos_2003-04_S-439-440.book Seite 442 Donnerstag, 30. Oktober 2003 7:36 07
RENT SEEKING INTO THE INCOME DISTRIBUTION
443
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
members of groups enjoying differentially lower costs of collective action will
secure government transfers at the expense of the less-well-organized.
The argument is straightforward. As long as it is profitable to engage in re-
distributive activities, rational people will do so. Some coalitions for one rea-
son or another have a comparative advantage in organizing to lobby for politi-
cal favors; they can organize for less than $ 1 in order to gain $ 1. Such groups
are net demanders of transfers2. Net suppliers are those individuals and groups
who would be required to spend more than $1 to avoid giving up $1 through
taxation or regulation. Clearly, this theory suggests no particular impact of gov-
ernment activity on the distribution of income. The well-organized receive net
transfers; the less-well-organized or unorganized finance these transfers. There
is no guarantee here that government will implement an equality-promoting set
of programs and policies unless the costs of organizing fall along certain in-
come lines in the polity at large – the poor are better organized for collective
action than the rich, for example.
One is thus left with an empirical question: does the political process, as im-
pacted by interest groups, lead to more or less equality in the distribution of in-
come? The next section supplies an answer to this question. We do not deny that
there is a substantial amount of piecemeal evidence that interest groups redis-
tribute income from the poor to the rich and from unorganized to organized in-
terests. One only need look at farm subsidies or trade protectionism around the
globe to draw such conclusions3. While these observations are suggestive of the
overall effects of interest groups on the costs and benefits of government, we
seek to test the hypothesis in a more general form.
2. The logic of collective action suggests that successful groups will tend to be small in size, have
homogeneous interests, and be effective providers of selective incentives to their members. Be-
cause political lobbying normally can be supplied at low marginal cost as a byproduct of organ-
izing for some other purpose, established groups, such as labor unions, agricultural cooperatives
and manufacturers’ associations, will have comparative advantages in transfer-seeking over
start-up groups (Olson 1965). Over the past 20 years, ‘general business organizations (mainly
state chambers of commerce) and schoolteachers (mainly state affiliates of the National Educa-
tion Association – NEA)’ have consistently been the most influential interest groups at the state
level (Thomas and Hrebenar 1999b, p. 9).
3. In exceptional cases, interest groups have lobbied for deregulation and privatization of industry.
See Shughart and Tollison (1985) on the reform of corporate chartering laws and Peltzman
([1989] 1998) for more recent examples.
Kyklos_2003-04_S-439-440.book Seite 443 Donnerstag, 30. Oktober 2003 7:36 07
WILLIAM F. SHUGHART II/ROBERT D. TOLLISON/ZHIPENG YAN
444
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
III. EMPIRICAL MODEL AND RESULTS
In order to test whether there is any empirical relationship between interest
groups and the distribution of income, we specified and estimated by ordinary
least squares the following empirical model based on data for the 50 U.S. states:
GINI = f (D2–D5, EDUCATION, INCOME, EXPORTS,
POPULATION DENSITY, RACE,
STATE GOVERNMENT EXPENDITURES, DEPENDENTS) (1)
Variable definitions and data sources are shown in Table 1.
Table 1
Variable definitions and data sourcesa
Variable Definition Source
GINI The Gini coefficient for family
income in a state
U.S. Bureau of the Census, http://
www. census. gov/hhes/income/
histinc/state/state4.html
D2–D5 A set of four dummy variables
reflecting the relative strength
of interest groups by state
Thomas and Hrebenar (1999b,
Table 2, p. 13)
EDUCATION The percentage of individuals in
a state with bachelor’s degrees
or higher
Statistical Abstract of the United
States 1991, p. 140
INCOME Median income by state U.S. Bureau of the Census, http://
www.census.gov/hhes/income/
histinc/state/state1.html
EXPORTS Value of exports to foreign
countries relative to gross state
productb
Statistical Abstract of the United
States 1991, p. 804
POPULATION DENSITY State population per square mile Statistical Abstract of the United
States 1991, p. 23, and World Book
Encyclopedia, vol.20 (2000,
pp. 104–7)
RACE Percentage of state population
that is white
Statistical Abstract of the United
States 1991, p. 22
STATE GOVERNMENT
EXPENDITURES
Total state government spend-
ing relative to gross state pro-
ductb
Statistical Abstract of the United
States 1991, p. 291
DEPENDENTS Percentage of state population
under 18 and over 65 years of
age
Statistical Abstract of the United
States 1991, p. 23
a. Except D2–D5, which correspond to the mid 1980s, all variables are observed as of 1989. See the
discussion in the text. b. Gross state product is from the Statistical Abstract of the United States
1993, p. 444.
Kyklos_2003-04_S-439-440.book Seite 444 Donnerstag, 30. Oktober 2003 7:36 07
RENT SEEKING INTO THE INCOME DISTRIBUTION
445
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
The Gini coefficient (G) is used throughout the analysis as a measure of the
equality in the distribution of income within a state. The properties of G, de-
fined on the zero-one interval, are well known (e.g., Bronfenbrenner 1971,
pp. 45–50). A larger G implies more inequality in the income distribution,
where G = 0 denotes perfect equality and G = 1 denotes perfect inequality.
As reported by the U.S. Census Bureau, the model’s dependent variable is
based on ‘estimates of money income from the March CPS [Current Population
Survey] on gross, or pre-tax, income’, excluding capital gains (Jones and
Weinberg 2000, p. 9)4. The unit of observation is the family, defined as two or
more people living together who are related by blood, marriage or adoption
(Ibid., p. 3). No adjustment is made for family size.
The central explanatory variables of interest are the members of a set of five
binary dummies reflecting the strength of interest group influence by state, as
classified by Thomas and Hrebenar (1999b). They categorize states as follows.
D5 denotes jurisdictions where interest groups are ‘dominant’, wielding ‘over-
whelming and consistent influence on policymaking’ (Ibid., p. 12). D3 repre-
sents states where interest groups serve in ‘complementary’ roles and ‘work in
conjunction with (or are constrained by) other aspects of the political system’
(Ibid.)5. In ‘subordinate’ states (D1), interest groups are ‘consistently subordi-
nated to other aspects of the policymaking process’ (Ibid., pp. 12–14). Finally,
the ‘dominant/complementary’ (D4) and ‘complementary/subordinate’ (D2)
classifications include states where interest-group influence alternates between
the two categories ‘or is in the process of moving from one to the other’ (Ibid.,
p. 14). Since no states are classified as ‘subordinate’, we utilize the four remain-
ing dummy variables, D2 through D5, to measure, in ascending order, the im-
pact of interest groups on the political process. Table 2 displays the Thomas and
Hrebenar groupings, which we have adjusted, using the information they pro-
vide identifying states that have changed categories over time, to reflect classi-
fications as of the mid 1980s.
Thomas and Hrebenar’s categorization is based on a survey methodology
with a lengthy pedigree in the political science literature addressing the relative
power of interest groups across states (Zeller 1954, Morehouse 1981). Ques-
tionnaires were sent to four or five knowledgeable people in each of the 50 U.S.
4 Given progressive tax rates, it should not be surprising that incomes in the United States are dis-
tributed more equally (the Gini coefficient is smaller) after-tax than pre-tax (Jones and Weinberg
2000, p. 9). Gini coefficients computed on an after-tax basis are not available by state for the
period relevant to our analysis.
5. The ‘other aspects of the political system’ are, ‘more often than not’, robust inter-party compe-
tition, but ‘could also be a strong executive branch, competition between groups, the political
culture, or a combination of all of these’ (Thomas and Hrebenar 1999b, p. 12).
Kyklos_2003-04_S-439-440.book Seite 445 Donnerstag, 30. Oktober 2003 7:36 07
WILLIAM F. SHUGHART II/ROBERT D. TOLLISON/ZHIPENG YAN
446
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
states, most of whom were academic political scientists (Thomas and Hrebenar
1999a, p. 141). Content analysis, a standard tool in survey research, was then
applied to the responses in order to classify interest-group influence according
to one of the five categories shown in Table 2.
The resulting classifications offer at least two advantages over alternative
measures of pressure group power, all of which are unweighted and do not take
account of differences in the identities of the groups that are active in particular
states or regions. First, Thomas and Hrebenar improve on the earlier studies of
Zeller (1954) and Morehouse (1981), which distinguished only three categories
of interest-group influence on the political process, ‘strong’, ‘moderate’ and
‘weak’ (Thomas and Hrebenar 1999a, pp. 135–36). Second, their classifica-
tions incorporate more nuanced assessments of interest-group influence across
states than is contained in, for example, simple tallies of such organizations.
The latter measure fails to distinguish Florida, where the American Association
of Retired Persons is a powerful political force, from Michigan, where labor
unions and automobile industry trade associations tend to dominate state poli-
tics. The classification of interest-group influence Thomas and Hrebenar have
constructed seems more empirically useful than the total number of interest
groups in a state6.
As noted above, we make no a priori predictions about the algebraic signs
on D2–D5. The other independent variables are control variables about which
we can only hazard guesses as to their ceteris paribus effects on the Gini coef-
ficient. For example, although one might expect incomes to be distributed more
equally in states where larger fractions of the population have earned at least a
bachelor’s degree, it is an empirical question as to whether increases in educa-
tional attainment in fact promote income equality or, alternatively, reinforce
existing inequalities by augmenting the advantages of cognitive elites. Simi-
larly, do higher income jurisdictions reflect more or less equality in the family
income distribution? The data will have to tell us7. Income obtained from for-
eign trade likewise may increase or reduce income equality in a state, depend-
ing on the extent to which that income is distributed broadly or narrowly within
the local economy.
6. Indeed, the correlation between the number of interest groups per state (Gale Research Co.
1999) and the Thomas and Hrebenar classification is very low: the Pearson correlation coeffi-
cient is –0. 052 (p = 0.720) for the raw number of interest groups; it is –0.180 (p = 0.212) for
interest groups per capita.
7. Simon Kuznets (1955) argued that, beyond a certain threshold (about $1,500 in today’s dollars),
increases in per capita income are accompanied by an improvement in the distribution of in-
come. According to him, the decline in income inequality follows from an easing of capital scar-
city, which triggers an increase in real wages.
Kyklos_2003-04_S-439-440.book Seite 446 Donnerstag, 30. Oktober 2003 7:36 07
RENT SEEKING INTO THE INCOME DISTRIBUTION
447
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
Table 2
Classification of the 50 U.S. states according to interest-group influence
Population density is a proxy for the cost of organizing collective action. To the
extent that these costs are lower in more densely populated states because, for
example, pressure groups can more easily identify potential members and can
more readily monitor their individual contributions to the collective effort,
more interest groups will form. Because the division of labor is limited by the
extent of the market (Stigler 1951), pressure groups also have access to more
specialized complementary inputs in more densely populated states. There are
greater numbers of lawyers, lobbyists, and advertising agencies available to
help promote the group’s cause. Both of these considerations point to more ef-
fective interest-group activity, but whether this leads to more or less equality in
the distribution of income is again an empirical question.
Dominant (9 states) Dominant/
Complementary
(17 states)
Complementary
(19 states)
Complementary/
Subordinate
(5 states)
Subordinate
(0 states)
Alabama Arizona Colorado Connecticut
Alaska Arkansas Illinois Delaware
Florida California Indiana Minnesota
Louisiana Georgia Iowa Rhode Island
Mississippi Hawaii Kansas Vermont
New Mexico Idaho Maine
South Carolina Kentucky Maryland
Tennessee Montana Massachusetts
West Virginia Nebraska Michigan
Nevada Missouri
Ohio New Hampshire
Oklahoma New Jersey
Oregon New York
Texas North Carolina
Utah North Dakota
Virginia Pennsylvania
Wyoming South Dakota
Washington
Wisconsin
Source: Thomas and Hrebenar (1999b, Table 2, p. 13).
Kyklos_2003-04_S-439-440.book Seite 447 Donnerstag, 30. Oktober 2003 7:36 07
WILLIAM F. SHUGHART II/ROBERT D. TOLLISON/ZHIPENG YAN
448
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
Conventional wisdom suggests that incomes are distributed more equally in
states having more racially homogeneous populations. If so, then the sign on
RACE will be negative, indicating less inequality where whites comprise a
larger fraction of the population. Such a prediction is also consistent with the
logic of collective action, which stresses that interest-group effectiveness is in-
versely related to group size (Olson 1965). Racial minorities may be more suc-
cessful in redistributing incomes to themselves in states where their population
proportions are smaller and, hence, they are better able to monitor and control
free-riding than are the members of the white majority.
The effect of government spending on the distribution of income depends on
the net direction of its transfer activities. Does the public budget tend to redis-
tribute income from the rich to the poor? Or do government’s tax and spending
programs instead primarily benefit the middle and upper income classes? Fi-
nally, how is the Gini coefficient related to the age distribution of a state’s pop-
ulation? One might expect incomes to be distributed more equally in states
where larger percentages of the population are of prime income-earning age, in
which case the sign on DEPENDENTS will be positive.
We estimated two versions of our empirical model, with and without the in-
terest-group dummy variables. States where interest group influence is defined
by Thomas and Hrebenar as complementary/subordinate (D2) is the excluded
category, about which more below. The results are reported in Table 3.
The two regression specifications address questions of endogeneity and
multicollinearity. In particular, it is possible that interest-group influence in a
state is correlated with the size of the state’s government and, furthermore, that
the impact of pressure groups on the distribution of income operates primarily
through the public budget. Neither of these possibilities is evident in the empir-
ical results: the estimated coefficient on STATE GOVERNMENT EXPENDI-
TURES carries the same sign, is of the same numerical magnitude, and is esti-
mated with the same level of statistical confidence in both models.
The OLS estimates suggest strongly that the distribution of income is more
unequal in states where interest-group influence is dominant (D5). Other things
being the same, the Gini coefficient is 0.024 higher in such states than it is in
states where interest-group influence is complementary/subordinate (D2). In
order to assess the relative impacts of interest groups across the states, we also
ran Model 1 using the three other possible combinations of Thomas and Hre-
benar’s categorical variables. These additional results are summarized in
Table 4, which reports the estimated coefficients on the dummy variables when
each is in turn excluded from the regression model and, hence, impounded in
the intercept. The last column reproduces the coefficients from Model 1, the
next-to-last column shows the ceteris paribus impacts of the interest groups in
Kyklos_2003-04_S-439-440.book Seite 448 Donnerstag, 30. Oktober 2003 7:36 07
RENT SEEKING INTO THE INCOME DISTRIBUTION
449
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
Table 3
The determinants of income inequality in the 50 U.S. statesa
Variable Model 1 Model 2
Intercept 0.597
(0.073)
[<.0001]
0.663
(0.072)
[<.0001]
D5 0.024
(0.010)
[0.0209]
D4 0.013
(0.009)
[0.1485]
D3 0.002
(0.008)
[0.7526]
EDUCATION 0.001
(7.52E– 4)
[0.0800]
0.001
(7.90E– 4)
[0.2115]
INCOME –3.24E – 6
(8.63E–7)
[0.0006]
–3.82E – 6
(9.06E–7)
[0.0001]
EXPORTS 0.077
(0.059)
[0.1992]
0.096
(0.063)
[0.1332]
POPULATION DENSITY 4.11E– 5
(1.33E– 5)
[0.0037]
3.65E– 5
(1.36E– 5)
[0.0103]
RACE –0.082
(0.021)
[0.0004]
–0.115
(0.019)
[<.0001]
STATE GOVERNMENT EXPENDITURES –0.280
(0.099)
[0.0075]
–0.234
(0.103)
[0.0292]
DEPENDENTS –0.149
(0.134)
[0.2716]
–0.178
(0.140)
[0.2126]
Adjusted R20.67 0.59
F-statistic 7.81
[<.0001]
8.37
[<.0001]
Note: a. The dependent variable is the Gini coefficient. Standard errors are shown in parentheses;
p-values are in brackets.
Kyklos_2003-04_S-439-440.book Seite 449 Donnerstag, 30. Oktober 2003 7:36 07
WILLIAM F. SHUGHART II/ROBERT D. TOLLISON/ZHIPENG YAN
450
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
the D5, D4, and D2 states relative to D3 states, and so on. Table 4 also reports
the corresponding percentage changes in the predicted value of the Gini coef-
ficient, calculated at the means of the independent variables.
Table 4
Relative interest-group influence on the Gini coefficienta
These comparisons suggest the following ordering: D5 > D4 > D3 = D2. Other
things being the same, incomes are distributed most unequally in D5 states
where interest groups are dominant, followed by D4 (dominant/complemen-
tary) states and D3 (complementary) states, which do not differ significantly
from D2 (complementary/subordinate) states. To put these differences in per-
spective, the Gini coefficient for households in the United States as a whole in-
creased by 7.23% (from .415 to .445) between 1979 and 1989, the era of so-
called Reaganomics8. The level of income inequality in D5 states is greater
than it is in D2 states by nearly the same percentage. Put differently, the Gini
coefficient is predicted to be slightly more than one standard deviation higher
in states where interest-group influence is classified as dominant, compared
with states where interest groups are complementary/subordinate to the politi-
cal process9.
Excluded variables
Included variables D5 D4 D3 D2
D5 .012*
(2.93%)
.022***
(5.69%)
.024**
(6.36%)
D4 –.012*
(–2.84%)
.010*
(2.69%)
.013
(3.33%)
D3 –.022***
(–5.39%)
–.010*
(–2.62%)
.002
(0.63%)
D2 –.024**
(–5.98%)
–.013
(–3.23%)
–.002
(–0.63%)
Notes: Asterisks denote significance at the 1 percent (∗∗∗), 5 percent (∗∗) and 10 percent (∗) levels of
confidence. The ceteris paribus percentage changes in the predicted Gini coefficients (shown in pa-
rentheses) are calculated at the means of the continuous right-hand-side variables.
8. Gini coefficients for family income, the dependent variable we employ, are available only for
1969 and 1989. By that measure, income inequality increased by 14.68 % over the two decades.
See http://www. census. gov/ hhes/income/histinc/state/state4.html.
9. The standard deviation of Gini in our data set is .022.
Kyklos_2003-04_S-439-440.book Seite 450 Donnerstag, 30. Oktober 2003 7:36 07
RENT SEEKING INTO THE INCOME DISTRIBUTION
451
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
Other results of interest are the significantly negative signs on INCOME,
RACE, and STATE GOVERNMENT EXPENDITURES, and the significantly
positive sign on POPULATION DENSITY. The last of these results may be ev-
idence that interest groups are more successful in more densely populated
states (because, for example, the cost of organizing collective action is lower).
Higher income jurisdictions with larger public sectors and more racially homo-
geneous populations all exhibit more income equality. These states seem to be
the American analogs of Scandinavia.
Education tends to increase income inequality, as do exports. On the other
hand, the distribution of income is more equal in states where larger fractions
of the population are either young or old (and correspondingly smaller frac-
tions of the population are of prime income-earning age). The estimated coef-
ficients on the last two variables are not different from zero at standard levels
of statistical significance, however.
As an additional empirical test, we included dummy variables in Model 1
corresponding to three of the four regions of the United States (Northeast, Mid-
west, West and South), as defined by the U.S. Census Bureau. Although there
is no clear-cut theoretical reason for expecting geography to have an impact on
the distribution of income, it is possible that the observed variation in Gini co-
efficients across states has systematic North-South or East-West components.
None of the regional dummy variables was different from zero at standard lev-
els of statistical significance, however. What is more important is that, with the
exception of EDUCATION, whose positive coefficient declined in significance
from eight to 15 percent, the remaining independent variables, including D3,
D4 and D5, were not materially affected by this change in specification10.
A key implication of the empirical analysis is that the influence of interest
groups on the distribution of income operates primarily off the public budget.
Holding the size of government constant, which by itself has a significant lev-
eling effect, interest group activities work in the direction of increasing income
inequality. One way of interpreting this finding is that specialized inputs whose
skills enter into the production of lobbying are relatively wealthier in a rent-
seeking society (Higgins and Tollison 1988). Institutions that facilitate income
redistribution through the political process tend to enrich lawyers, advertising
executives, and economists relative to other occupations, thereby raising the
Gini coefficient. Alternatively, it may be that well-organized interest groups,
evidently representing the interests of individuals and groups at the upper end
of the income distribution, succeed in transferring income to themselves pri-
marily through regulations and mandates (e.g., minimum wage laws, affirma-
10. These additional empirical results are available on request.
Kyklos_2003-04_S-439-440.book Seite 451 Donnerstag, 30. Oktober 2003 7:36 07
WILLIAM F. SHUGHART II/ROBERT D. TOLLISON/ZHIPENG YAN
452
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
tive action rules, and so on) whose costs are borne largely by the private sec-
tor11. In any case, our results suggest that interest-group influence, on balance,
tends to promote a more unequal distribution of income. Greater income ine-
quality is yet another item that must be added to the list of the social welfare
costs of rent-seeking (Tullock 1967).
IV. CONCLUDING REMARKS
This paper has used cross-sectional data from the 50 U.S. states to show that
income inequality is an increasing function of interest-group influence. Hold-
ing educational attainment, median income, state government expenditures rel-
ative to gross state product, population density, race and other factors constant,
we find that income inequality is significantly greater (the Gini coefficient is
more than one standard deviation higher) in nine states (Alabama, Alaska, Flor-
ida, Louisiana, Mississippi, New Mexico, South Carolina, Tennessee, and West
Virginia) than it is elsewhere. The common denominator for these nine juris-
dictions is that interest groups dominate their political processes. Interest
groups have lesser, but still marginally significant impacts on the income dis-
tribution in another 17 states where they are classified as dominant/comple-
mentary.
The interest-group effects in our empirical model are ceteris paribus results.
No doubt, issues of endogeneity have not been fully laid to rest because, for ex-
ample, interest groups in other studies have been shown to lead to greater pub-
lic spending. Nonetheless, interest groups here are promoters of inequality,
while public spending leads to more equality. This suggests that the primary ef-
fects of interest groups operate off-budget.
11. Suggestive evidence that interest groups promote inequality at the expense of the poor is pro-
duced by regressing the poverty rate in 1989 (http://www. census. gov/hhes/poverty/census/
cphl162.html) on the independent variables from Model 1. The results of this estimation are:
POVERTY = 40.8870*** + 2.6837D5** + 1.0046D4 +.1903D3 +.0918EDUCATION –
7.3355E–4INCOME*** + 14.0204EXPORTS** +.0033DENSITY** – 10.6781RACE*** –
3.4499STATE GOVERNMENT EXPENDITURES – 2.8795DEPENDENTS, where asterisks de-
note significance at the 1 percent (***) and 5 percent (**) levels, respectively; adjusted R2 =
0.84 (F = 27.1***). Other things being the same, the percentage of the population with incomes
below the federally defined poverty line is significantly higher in D5 states than it is in juris-
dictions where interest groups have less influence on political processes.
Kyklos_2003-04_S-439-440.book Seite 452 Donnerstag, 30. Oktober 2003 7:36 07
RENT SEEKING INTO THE INCOME DISTRIBUTION
453
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
REFERENCES
Bronfenbrenner, Martin (1971). Income Distribution Theory. Chicago and New York: Aldine Ather-
ton.
Congleton, Roger D. and William F. Shughart II (1990). The Growth of Social Security: Electoral
Push or Political Pull?, Economic Inquiry. 28: 109 –132.
Gale Research Co. (1999). Encyclopedia of Associations: Regional, State and Local Organizations,
9th ed. Detroit: Gale Research Co.
Higgins, Richard S. and Robert D. Tollison (1988). Life among the Triangles and Trapezoids, in:
Charles K. Rowley, Robert D. Tollison and Gordon Tullock (eds.), The Political Economy of Rent-
Seeking. Boston: Kluwer: 147–157.
Jones, Arthur F. Jr. and Daniel H. Weinberg (2000). The Changing Shape of the Nation’s Income Dis-
tribution, 1947–1998: Consumer Income. Current Population Reports, P60– 204. Washington,
DC: U.S. Census Bureau.
Kuznets, Simon (1955). Economic Growth and Income Inequality, American Economic Review. 45:
1–28.
McCormick, Robert E. and Robert D. Tollison (1981). Politicians, Legislation, and the Economy: An
Inquiry into the Interest-Group Theory of Government. Boston: Martinus Nijhoff.
Morehouse, Sarah McCally (1981). State Politics, Parties, and Policy. New York: Holt, Rinehart and
Winston.
Mueller, Dennis C. (1989). Public Choice II: A Revised Edition of Public Choice. Cambridge: Cam-
bridge University Press.
Olson, Mancur (1965). The Logic of Collective Action: Public Goods and the Theory of Groups. Cam-
bridge: Harvard University Press.
Peltzman, Sam ([1989] 1998). The Economic Theory of Regulation after a Decade of Deregulation,
in: Sam Peltzman, Political Participation and Government Regulation. Chicago: University of
Chicago Press: 286– 323.
Reynolds, Morgan O. and Eugene Smolensky (1977). Public Expenditures, Taxes and the Distribu-
tion of Income: The U.S., 1950, 1961, 1970. New York: Academic Press.
Shughart, William F. II and Robert D. Tollison (1985). Corporate Chartering: An Exploration in the
Economics of Legal Change, Economic Inquiry. 23: 585 – 599.
Stigler, George J. (1951). The Division of Labor is Limited by the Extent of the Market, Journal of
Political Economy. 59: 185–193.
Stigler, George J. (1970). Director’s Law of Public Income Redistribution, Journal of Law and Eco-
nomics. 8: 1–10.
Thomas, Clive S. and Ronald J. Hrebenar (1999 a). Interest Groups in the States, in: Virginia Gray,
Russell L. Hanson and Herbert Jacobson (eds.), Politics in the American States: A Comparative
Analysis, 7th ed. Washington, DC: Congressional Quarterly Press: 122 – 58.
Thomas, Clive S. and Ronald J. Hrebenar (1999 b). Interest Group Power in the Fifty States: Trends
since the Late 1970s, Comparative State Politics. 20: 3 –17.
Tullock, Gordon (1967). The Welfare Costs of Tariffs, Monopolies, and Theft, Western Economic
Journal. 5: 224 – 232.
U.S. Department of Commerce, Bureau of the Census (1991). Statistical Abstract of the United States
1991. Washington, DC: USGPO.
U.S. Department of Commerce, Bureau of the Census (1993). Statistical Abstract of the United States
1993. Washington, DC: USGPO.
World Book, Inc. (2000), World Book Encyclopedia 2000. Chicago: World Book, Inc.
Zeller, Belle (1954). American State Legislatures, 2nd ed. New York: Crowell.
Kyklos_2003-04_S-439-440.book Seite 453 Donnerstag, 30. Oktober 2003 7:36 07
WILLIAM F. SHUGHART II/ROBERT D. TOLLISON/ZHIPENG YAN
454
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
SUMMARY
The 50 U.S. states differ considerably in the extent to which political processes are swayed by spe-
cial-interest groups. Pressure groups regularly wield overwhelming influence on policymaking in
nine states (Alabama, Alaska, Florida, Louisiana, Mississippi, New Mexico, South Carolina, Tennes-
see, and West Virginia); they play lesser policymaking roles elsewhere, ranging from complementing
other political actors to being completely subordinate to them. This paper exploits an independently
constructed, five-category taxonomy of interest-group influence to explore the cross-sectional impact
of rent-seeking on the distribution of income. The empirical estimates produce a consistent rank-or-
dering of the states in which income inequality is an increasing function of interest-group power. In
particular, holding educational attainment, median income, state government expenditures relative to
gross state product, population density, race and other factors constant, we find that income inequal-
ity is significantly greater (the Gini coefficient is more than one standard deviation higher) in the nine
states where interest groups dominate the political process. Interest groups have smaller, but still
marginally significant impacts on the income distribution in another 17 states where they are classi-
fied as dominant/complementary. We also find that greater levels of educational attainment tend to
increase income inequality, as do exports. On the other hand, jurisdictions with higher median in-
comes, those with larger public sectors and more racially homogeneous populations exhibit greater
income equality. Given that, other things being the same, incomes tend to be distributed more equally
in states where government spending is higher, we conclude that it is mainly through off-budget
channels that special interest groups operate to promote income inequality. But in any case, greater
income inequality is yet another item to be added to the list of the social welfare costs of rent-seek-
ing.
ZUSAMMENFASSUNG
In den 50 Teilstaaten der USA können Interessengruppen die politischen Prozesse in unterschiedli-
chem Mass beeinflussen. In neun Staaten ist dieser Einfluss äusserst stark (Alabama, Alaska, Florida,
Louisiana, Mississippi, New Mexico, South Carolina, Tennessee und West Virginia), während er in
anderen Staaten sehr viel schwächer ist. Zum Teil können Interessengruppen andere politische Ak-
teure ergänzen, zum Teil sind sie diesen sogar vollständig untergeordnet. Dieser Artikel beruht auf
einer unabhängig erstellten, fünf Kategorien umfassenden Klassifikation des Einflusses von Interes-
sengruppen, um in einem Querschnittsvergleich den Einfluss des Rent-Seeking auf die Einkommens-
verteilung zu ermitteln. Die empirischen Schätzungen ergeben eine konsistente Rangordnung der
Staaten, in denen die Einkommensungleichheit eine zunehmende Funktion der Macht von Interes-
sengruppen ist. Insbesondere wenn Faktoren wie Zugang zu Bildung, Medianeinkommen, staatliche
Ausgaben relativ zum Bruttosozialprodukt, Bevölkerungsdichte, Rassendurchmischung und u. a. m.
konstant gehalten werden, stellt sich heraus, dass die Einkommensungleichheit in den neun Staaten,
in denen Interessengruppen den politischen Prozess dominieren, signifikant höher ist (höherer Gini-
Koeffizient um mehr als eine Standardabweichung). In weiteren 17 Staaten, in denen sie als domi-
nant/komplementär klassiert werden, haben Interessengruppen zwar geringeren, aber immer noch
marginal signifikanten Einfluss auf die Einkommensverteilung. Wir finden ausserdem, dass ein hö-
heres Bildungsniveau ebenso wie Exporte tendenziell zu stärkeren Einkommensunterschieden füh-
ren. Andererseits lässt sich in Gebietskörperschaften mit einem höheren Medianeinkommen, solchen
mit grösserem öffentlichem Sektor und mit mehrheitlich gleichrassiger Bevölkerung eine gleichmäs-
sigere Einkommensverteilung feststellen. Da unter Beibehaltung anderer Bedingungen die Einkom-
men in Staaten mit höheren öffentlichen Ausgaben tendenziell gleichmässiger verteilt sind, schlies-
sen wir, dass Interessengruppen vorwiegend durch nicht budgetrelevante Kanäle operieren. Die
verstärkte Ungleichheit in der Einkommensverteilung bleibt in jedem Fall ein weiterer Punkt auf der
Liste der Wohlfahrtsverluste durch Rent-Seeking.
Kyklos_2003-04_S-439-440.book Seite 454 Donnerstag, 30. Oktober 2003 7:36 07
RENT SEEKING INTO THE INCOME DISTRIBUTION
455
G4_HS:Aufträge:HEL002:15015_SB_Kyklos_2003-04:15015-A04:Kyklos_2003-04_S-441-562 29.11.96 30. Oktober 2003
07:35
RÉSUMÉ
Les 50 états des Etats-Unis montrent de considérables différences quant à la mesure dans laquelle les
groupes d’intérêt particulier peuvent influencer les processus politiques. Dans 9 états, les groupes
d’intérêt on régulièrement une influence majeure sur la politique (Alabama, Alaska, Florida, Loui-
siana, Mississippi, New Mexico, South Carolina, Tennessee, et West Virginia), tandis que dans
d’autres états, leur influence directe est moins évidente et varie du rôle de complément des autres
acteurs politiques jusqu’à la totale subordination à ceux-ci. Cet article applique une taxonomie de
l’influence des groupes d’intérêt développée indépendamment et comprenant cinq catégories, afin
d’explorer l’impact transversal du rent-seeking sur la distribution des revenus. Les estimations em-
piriques produisent un ordre consistant des états dans lesquels l’inégalité des revenus est une fonc-
tion croissante du pouvoir des groupes d’intérêt. En maintenant constants les facteurs accès à l’édu-
cation, revenu médian, dépenses du gouvernement en relation avec le produit national brut, densité
de la population, race et d’autres encore, nous trouvons que l’inégalité des revenus est significative-
ment plus importante (le coefficient Gini dépasse une simple déviation standard) dans les neuf états
où les groupes d’intérêt particulier dominent le processus politique. Dans 17 autres états, les groupes
d’intérêt étant classés dominants/complémentaires ont une influence moins importante, mais toujours
significative sur la distribution des revenus. Nous trouvons en outre que soit l’éducation, soit les ex-
portations contribuent à augmenter l’inégalité des revenus. D’autre part, les juridictions au revenu
médian élevé, avec un secteur public plus étendu et une plus grande homogénéité raciale de la popu-
lation ont une plus grande égalité des revenus. Puisque – d’autres conditions restant identiques – les
revenus ont tendance à être répartis de façon plus uniforme dans les états dont les dépenses publiques
sont plus élevées, nous concluons que les groupes d’intérêt particulier exercent leur influence pour
favoriser l’inégalité des revenus à travers des voies extra-budgétaires. En tout cas, l’inégalité des re-
venus est un autre point à ajouter à la liste des coûts sociaux du rent-seeking.
Kyklos_2003-04_S-439-440.book Seite 455 Donnerstag, 30. Oktober 2003 7:36 07
