Taxation and Redistribution OECD

8/20/2019 Taxation and Redistribution OECD

http://slidepdf.com/reader/full/taxation-and-redistribution-oecd 1/48

PARTISANSHIP, POLICY, TAXATION AND I NEQUALITY:R EDISTRIBUTION IN THE OECD

F. David Rueda

Assistant Professor

Political Science Department

Binghamton University – SUNY

P.O. Box 6000

Binghamton, NY 13902-6000

Phone: (607) 777-6787

E-mail: [email protected]

Paper prepared for delivery at the Annual Meeting of the American Political Science

Association, Philadelphia, August 28-31, 2003.



1

Abstract: The main goal of this paper is to explore the relationship between government

partisanship, policy and wage inequality at the lower half of the wage distribution. The

analysis is motivated by a puzzling finding in previous work: the absence of government partisanship effects on inequality. My paper argues that to understand the relationship

between partisan government and inequality two fundamental things need to be done: to

separate the effects of partisanship on policy and of policy on inequality; and to assessthe influence of political agency once the mediating role of institutions is accounted for.

I explain why some institutions mitigate the influence of government partisanship while

others magnify it. By explicitly exploring the determinants of policy and inequality, my paper presents a more accurate analysis of how governments promote redistribution. I

test my claims with data from 16 OECD countries from 1973 to 1995.



2

It is well known that wage inequality has increased dramatically in the United States

over the last three decades. From 1973 to 1998, the hourly earnings of a full-time worker in

the 90th percentile of the American earnings distribution (someone whose earnings exceeded

those of 90 per cent of all workers) relative to a worker in the tenth percentile grew by 25 per

cent, and the corresponding figure for men only was nearly 40 per cent. Wage inequality has

increased in most OECD countries, but the extent of this phenomenon varies a great deal, and

cross-national differences in levels of wage inequality remain as great as they were in the

1970s. In the US, the worker in the 90th percentile earned 4.63 times as much as the worker

in the tenth percentile in 1996. At the other end of the cross-national spectrum, the 90-10

ratio for Sweden was only 2.27.

While political commentators in Europe and the US alike frequently invoke

inegalitarian labor market trends to explain various manifestations of working-class political

disaffection (not only support for right-wing populist parties, but also falling turnout among

working-class voters), recent work by labor economists demonstrates that supply and demand

factors alone cannot account for cross-national variation in wage inequality.1 Wage

inequality appears to have political determinants as well as political consequences. On both

counts, it deserves to be a central concern of comparative political economy as conceived and

practiced by political scientists. I use a new data set published by the OECD to engage in a

pooled cross-section time-series analysis of the determinants of wage inequality in sixteen

OECD countries for the period 1973-1995.

In the following pages, I will first present a description of the outcomes that I am

trying to explain. This brief analysis of the patterns of wage inequality in the OECD will

1 E.g., Richard Freeman and Lawrence Katz (eds.), Differences and Changes in Wage

Structures (Chicago: The University of Chicago Press, 1995), 1-24; Francine Blau and

Lawrence Kahn, “International Differences in Male Wage Inequality,” Journal of Political Economy, 104 (1996), 791-836; and Peter Gottschalk and Timothy Smeeding,

“Cross-national Comparisons of Earnings and Income Inequality,” Journal of Economic

Literature, 35 (1997), 633-87.



3

show that there is a great deal of variation among countries and through time. It will also

show that the ratio of wages earned by a worker at the 90th

percentile to those earned by a

worker at the 10th

percentile (90-10 ratio) shares important similarities with the 50-10 ratio

(the ratio of wages earned by a worker at the 50th

percentile to those earned by a worker at

the 10th

percentile). I then will introduce the main focus of this paper: the relationship

between government partisanship and wage inequality at the bottom half of the wage

distribution. I will dedicate the rest of this paper to exploring the role of different policies

and government partisanship (as well as the relationship between the two) in the

determination of wage inequality. I argue that to understand the relationship between

partisan government and inequality two fundamental things need to be done: to separate the

effects of partisanship on policy and of policy on inequality; and to assess the influence of

political agency once the mediating role of institutions is accounted for. I explain why some

institutions mitigate the influence of government partisanship while others magnify it. By

explicitly exploring the determinants of policy and inequality, my paper presents a more

accurate analysis of how governments promote redistribution. I test my claims with data

from 16 OECD countries from 1973 to 1995.

Patterns of Wage Inequality

Table 1 summarizes the wage inequality observations which serve as the dependent

variable for my analysis. For each country, the table provides the mean value for wage

inequality at the lower half of the wage distribution (the 50-10 ratio) for the entire period

1973-95 and also the percentage change from the earliest to the most recent observation. For

the purpose of comparison, I have also included means and percentage changes for wage

inequality in the entire wage distribution (the 90-10 ratio). It should be noted at the outset

that these inequality measures refer to gross income from employment for individuals: they

ignore other sources of income (government transfers, self-employment, income from

capital, etc.) and also exclude the distributive effects of taxation and income pooling within



4

households. What follows must not be confused with an analysis of the distribution of

disposable household income.

[Table 1]

The OECD dataset reflected in Table 1 (and the rest of this paper’s analysis) is

restricted to full-time employees, except in the case of Austria. Since part-time employees

invariably earn less, on an hourly basis, than full-time employees, the figures in Table 1

understate the extent of wage inequality in the other countries. And because the incidence of

part-time employment has increased in most OECD countries since the early 1980s, they also

understate the upward trend in wage inequality.2 Keeping these qualifications in mind,

income from employment still accounts for the greatest portion of total income in all OECD

countries, and wage inequality among full-time employees still correlates quite closely with

broader cross-national measures of income distribution.3

Table 1 reveals important cross-national variation in wage inequality. In these

sixteen countries, the average both-gender 50-10 ratio for the 1973-95 period was 1.69. In

other words, a person in the 50th percentile of the wage distribution (the wage median)

earned, on average, almost two times as much as a person in the 10th percentile. Sweden,

with an average 50-10 ratio of 1.33, stands out as the OECD country with the most

compressed lower-half wage distribution. While the Scandinavian countries fall within a

narrow range of very compressed lower-half wage distributions, the continental European

countries included in this dataset (France, Belgium, Germany, Italy, the Netherlands, and

2 The exclusion of part-time employees does not pose a serious problem for cross-sectional comparison, for median hourly earnings of part-time workers as a percentage of

median hourly earnings of full-time workers correlates very closely with 90-10 ratios

among full-time workers cross-nationally. But this data restriction does pose a problemfrom the point of view of comparing changes in wage inequality across countries, since

the growth of part-time employment varies considerably across countries. 3 Cf. OECD, Income Distribution in OECD Countries: Evidence from the Luxembourg Income Study (1995); Gottschalk and Smeeding, “Cross-National Comparisons of

Earnings and Income Inequality;” and Wallerstein, “Wage-Setting Institutions and Pay

Inequality in Advanced Industrial Societies.”



5

Switzerland) can be classified as a group with inequality levels slightly below the OECD

average. The exception is, of course, Austria, which belongs at the opposite end of the

spectrum with the United States, the United Kingdom, Japan and Canada. All these countries

exhibit considerably larger than average levels of inequality at the lower half of the wage

distribution.

Turning to change over time, the cross-national variation in the data is equally

impressive. From the earliest to the most recent observation available for each country, there

are large increases of 50-10 inequality in the US, Canada, the Netherlands, and Australia.

However, lower-half wage inequality fell quite significantly in Germany, Finland, Norway

and Japan. For all sixteen countries, the average 50-10 ratio decreased by 1.1 per cent. It is

tempting to conclude from the 50-10 data in Table 1 that there is no common trend for wage

inequality to increase in the OECD countries, but this conclusion may be a bit hasty. In

many of these countries, wage inequality declined in the early part of the period covered by

these summary measures and subsequently rose. If change since the trough of wage

inequality was measured in each country, we would observe increases of wage inequality in

ten out of sixteen, and the magnitude of these increases would be typically greater than those

shown in Table 1. Also, the tendency for rising wage inequality becomes broader and more

pronounced when one looks at data for men and women separately.4 In most of these

countries, compression of between-gender differentials offset the effects of growing within-

gender differentials in the 1980s.

Table 1 reveals remarkable similarities between wage inequality at the bottom half of

the wage distribution and that represented by 90-10 ratios. The average 90-10 ratio for the

1973-95 period was 2.89. The average person in the 90th percentile of the wage distribution,

therefore, earned nearly three times as much as a person in the 10th percentile. Sweden, with

4 See Jonas Pontusson, David Rueda and Christopher Way, “The Role of Political-

Institutional Variables in the Making of Gendered Patterns of Wage Inequality,” Working

Paper, Institute for European Studies, Cornell University (1999).



6

an average 90-10 ratio of 2.07, again stands out as the OECD country with the most

compressed overall wage distribution. And, apart from France and Austria, the continental

European countries included in this dataset fall within a rather narrow band (2.34-2.79), well

below the OECD average. The most unequal countries are also similar to those described

above (the United States, Canada, France, Japan, and the United Kingdom). As for change

over time, the cross-national variation in the data is again remarkable. From the earliest to

the most recent observation available for each country, large increases of 90-10 ratios can be

observed in the US, the UK, Canada, Australia, the Netherlands, Austria and Italy. However,

overall wage inequality fell quite significantly in Finland, France, Germany and Belgium.

For all sixteen countries, the average 90-10 ratio increased by 2.29 per cent.

The Puzzle: Government Partisanship and Inequality at the Lower Half

of the Wage Distribution

The starting point for this paper’s exploration of the determinants of wage inequality

is the hypothesis that the partisan nature of governments influences wage inequality.

Governments can influence a country’s wage distribution through a variety of policies (e.g.,

those affecting minimum wages, social wages, taxes, etc). The argument supporting the

existence of a relationship between government partisanship and inequality at the lower half

of the wage distribution can be explained in very simple terms. It hinges on the proposition

that the policy preferences of Left parties raise the wage floor for competition in the labor

market. If there is a legislated minimum wage, for example, Left governments are likely to

set the minimum wage closer to the median wage than Right governments. Another example

would be the tendency of Left governments to favor higher levels of social wages, and

therefore to curtail the inegalitarian effects of unemployment and, more generally, to boost

the relative bargaining power of unskilled workers.

In one of the few existing political analysis of inequality at the lower end of the

distribution, Pontusson, Rueda and Way analyze the determinants of 50-10 ratios to test



7

whether Left governments do in fact raise the relative market power and the wages of poorly

paid workers.5 They do this through a set of regressions in which the relationship between

Tom Cusack’s measure of government partisanship and the levels of 50-10 inequality is

explored.6 I reproduce their main results in the first column in Table 2.

[Table 2]

The results in Table 2 show that government partisanship does not significantly

influence inequality at the lower half of the wage distribution.7 As Pontusson, Rueda and

Way recognize, this finding is a most puzzling result. In this regression, the coefficient for

government partisanship is positive (as expected, social democratic governments would be

associated with lower levels of wage inequality), but it does not come close to reaching

statistical significance. These results therefore offer little support for the hypothesis that Left

parties promote relative wage gains for poorly paid workers by setting a floor for competition

in the labor market.

Since the lack of significance of the partisanship variable could be interpreted as a

result of the presence of country and time dummies (the regr ession in the first column

includes dummies for all countries8 and for the time periods

9), I also present the results of an

identical regression without fixed effects. In this regression, which assesses the sort of cross-

national variation excluded by country dummies, the effects of government partisanship are

still insignificant.

5 Jonas Pontusson, David Rueda and Christopher Way, “Comparative Political Economyof Wage Distribution: The Role of Partisanship and Labour Market Institutions,” British

Journal of Political Science, 32 (2002), 281-308. 6 Tom Cusack, “Partisan Politics and Public Finance,” Public Choice, 91 (1997), 375-395. Higher values of Cusack’s index signify more conservative government. 7 I use a similar methodological setup and the same control variables as those used by

Pontusson, Rueda and Way.8 I ran the regressions without a constant. 9 The periods are: 1973-79, 1980-84, 1985-89, and 1991-1995. The excluded reference

year is 1990.



8

Pontusson, Rueda and Way further explore the wage floor hypothesis by developing a

very preliminary analysis of the relationship between government partisanship and income

replacement policies, on the one hand, and that between income replacement policies and

wage inequality, on the other. They plot average income replacement rates provided by

public schemes during the first year of unemployment in 1985-91 against each country's

average partisanship score for 1970-90.10

And then they plot 50-10 ratios in 1991 against the

average unemployment replacement rates. Their analysis suggests that there is a very weak

association between Left government and the generosity of unemployment compensation and

that the relationship between unemployment compensation and 50-10 compression is even

weaker. In other words, both steps in the argument linking Left government to egalitarianism

via a wage floor effect seem to falter.

It is however clear that the connection between government partisanship and policy,

on one side, and policy and wage inequality, on the other, needs to be paid more attention to.

Pontusson, Rueda and Way recognize that their analysis of the effects of income replacement

rates over inequality (as well as their relationship to government partisanship) is very

rudimentary. Income replacement policies are not the only tool at the disposal of a

government to influence the market power and the wages of poorly paid workers. In fact,

since replacement rates do not reflect the percentage of workers who are entitled to

unemployment benefits,11

there are reasons why this policy is in fact not the most appropriate

for assessing the influence of governments on inequality. The analysis needs to be extended

10 Based on OECD, “Unemployment Benefit Entitlements and Replacement Rates”(electronic database), these unemployment replacement rates refer to someone who

earned the equivalent of the average production worker at the time that he or she became

unemployed and include various income maintenance programs (in addition tounemployment insurance in the narrow sense). See OECD, “Unemployment and Related

Benefits,” The OECD Jobs Study: Evidence and Explanations (1994), part II, chapter 8,

for more details. 11 The prerequisites for unemployment benefits differ substantially in OECD nations. As

a consequence, there is great variation in terms of the percentage of unemployed people

that receives benefits.



9

to other policies and improved in terms of its methodology. More systematic tests of the

relationships among partisan governments, policies and wage outcomes must be performed.

These are in fact the objectives of the analysis that I develop in the following pages.

The Argument: Separating the Effects of Partisanship from Those of Policy and

Assessing the Role of Institutions

As was mentioned above, the argument supporting the existence of a relationship

between government partisanship and inequality rests on the proposition that the policy

preferences of Left parties raise the wage floor for competition in the labor market. There

are two fundamental points I wish to make in this paper. The first has to do with the

connection between governments and inequality. It is common in the comparative political

economy literature to test the existence of associations between the partisan nature of

governments and the levels of inequality in a labor market. Governments, however, do not

possess the ability of transforming the wage distribution directly. They must rely on the

design and implementation of policy to accomplish any degree of redistribution. If we are

interested in an accurate assessment of the relationship between partisanship and inequality,

therefore, it is imperative that we disentangle the effects of partisanship and policy. We must

explore first whether government partisanship affects policy and then whether policy affects

inequality.

The second point I make in this paper concerns the role of institutions as factors

affecting political agency. The influence of institutions on political processes has been

emphasized by a great number of scholars.12

Regarding inequality, I argue that the effects of

government partisanship on policy and the effects of policy on economic outcomes are

contingent on institutions. In other words, even when they are committed to redistribution,

12 See Sven Steinmo, Kathleen Thelen and Frank Longstreth, Structuring Politics: Historical Institutionalism in Comparative Analysis (New York: Cambridge University

Press, 1992) and Jonas Pontusson, “From Comparative Public Policy to Political

Economy,” Comparative Political Studies, 28(1992), 117-47, for a review and analysis.



10

social democrats (for example) will not promote egalitarian policies unless they are

convinced that the institutional context allows these policies to affect economic outcomes. I

focus on two kinds of institutions: those related to the labor market and those related to a

nation’s political system.

There are reasons to believe that the effects of policy on wage-distributive effects are

contingent on one important labor market institution: corporatism. According to Katzenstein,

three traits define corporatism: “an ideology of social partnership expressed at the national

level; a relatively centralized and concentrated system of interest groups; and voluntary and

informal coordination of conflicting objectives through continuous political bargaining

between interest groups, state bureaucracies and political parties.”13

(1985: 32). A fourth

trait could be added by also emphasizing a high degree of institutionalization of collective

bargaining and coordination of wage formation. By coordination of wage formation, I mean

that wage developments in different sectors of the economy are more tightly coupled,

through collective bargaining, than they are in less corporatist economies. In countries

characterized by a high degree of corporatism, collectively negotiated wage agreements

typically apply to all workers in a company or sector, whether or not they are union

members, and wage developments in different companies and sectors are closely linked to

each other. Arguably, these arrangements constrain the ability of governments to influence

the distribution of wages. Put more positively, they enable the social partners to negotiate

effective wage floors and therefore reduce their r eliance on government policy (like

minimum wage legislation, etc) for this purpose.14

In this context, it is logical to assume that

13 Peter Katzenstein, Small States in World Markets (Ithaca: Cornell University Press,

1985). 14 Although they emphasize social market conditions rather than corporatism, Rueda and

Pontusson provide some support for this argument. They show that Left government is

associated with overall wage compression (measured by 90-10 ratios) in liberal marketeconomies, such as the US and the UK, but not in the social market economies of

Northern Europe. See David Rueda and Jonas Pontusson, “Wage Inequality and

Varieties of Capitalism,” World Politics, 52 (2000), 350-83.



11

the link between policy and inequality would actually be very weak in corporatist countries.

Furthermore, since policy-makers understand that policy is ineffectual and that inequality is

in some ways taken care of by corporatist institutions, it is also logical to hypothesize that the

influence of government partisanship on policy would be very weak when corporatism is

high.

There is a second set of institutions that matters greatly to the relationship between

partisanship and policy: those relating to a nation’s constitutional str ucture. Following the

work of Huber, Ragin and Stephens and Swank (among others),15

I argue that constitutional

structure can either mitigate or magnify the influence of partisanship on policy. There are

two contradicting interpretations of the effects of constitutional structure. On the one hand,

some authors have argued that, in the words of Huber, Ragin and Stephens, “aspects of

constitutional structure that disperse political power and offer multiple points of influence”

would difficult the design and implementation of policy.16

The relevant characteristics of

constitutional structure would be federalism, presidentialism, majoritarian government,

strong bicameralism and the frequent use of referenda. According to this argument, an

increasing number of veto points would allow minorities to obstruct the political process and

therefore be associated with no policy change.17

Constitutional structure, on the other hand,

could have the opposite effect. The dispersion of political influence could allow narrow

majorities to overcome the opposition of powerful minorities. It could also give a

disproportionate amount of political control to pro-redistribution groups that would otherwise

15 See Evelyne Huber, Charles Ragin and John Stephen, “Social Democracy, Christian

Democracy, Constitutional Structure, and the Welfare State,” American Journal ofSociology, 99 (1993), 711-749; and Duane Swank, Global Capital, Political Institutions,

and Policy Change in Developed Welfare States (New York: Cambridge University

Press, 2002). 16 See Huber, Ragin and Stephens (fn. 15) and Ellen Immergut, The Political

Construction of Interests (New York: Cambridge University Press, 1992). 17 As Swank notes, this process could be amplified by the relationship betweenconstitutional structure and the strength of Left parties. He argues that a higher number

of veto points is associated with poorer electoral performances by Left parties. See

Swank (fn. 15: 49).



12

be marginalized. As Swank argues, moreover, it is also important to point out that the effects

of veto points on the relationship between government partisanship and policy are

conditioned by historical developments. Whether the effects of constitutional structure

enhance or diminish the effects of government partisanship on policy may in fact depend on

how well-established the veto points are.18

The Explanatory Variable: Government Partisanship19

The government partisanship measures used in my analysis attempt to capture the

ideological position of governments in relation to a left-right continuum. Two variables are

needed for the construction of these measures: one that reflects the presence of parties in

government and another that measures their ideological characteristics. There are, however,

important questions surrounding the operationalization of both these var iables. There is first

the issue of how to measure the influence of parties in government.20

A possibility is to take

into consideration the proportion of cabinet seats that all parties in government possess.

Once a party is in government, however, the support it enjoys may be influenced not only by

its position in the cabinet but also by the degree of support enjoyed in parliament. Numerous

authors have found evidence supporting that a government’s behavior will be greatly

influenced by its share of seats in parliament (see, for example, Müller and Strøm 2000). As

argued by Garrett, the balance of power in cabinet governments delineates the direct control

of parties over policy while the balance of power in parliament captures the broader political

constraints facing this government (1998: 59).

Regarding the second factor influencing government partisanship, the measurement

of party ideological positions is also not completely straightforward. As Gabel and Huber

point out, assessments of left-right party positions are mainly based on two different

18 Swank argues that when dispersion and decentralization are embedded the influence of

pro-redistribution groups will be minimized. When dispersion is relatively recent, thiswill not necessarily be the case. See Swank (fn. 15: 50-51). 19 See Appendix 1 for details and sources about all variables used in this paper’s analysis. 20 See Huber and Powell (1994) for a more detailed analysis of some of the options.



13

measures: the analysis of expert opinions or of party manifestos (2000: 94-5). These two

measures imply a different set of complications. Expert opinions are produced from surveys

that are administered rarely and that may be interpreted differently in different national

contexts (see Gabel and Huber 2000 for a more detailed explanation of this argument). In

addition, expert opinions about party positions do not reflect changes through time

(McDonald and Mendes 2001a). Perhaps more importantly, partisanship data based on

expert opinions are vulnerable to the criticism that they are endogenous. Particularly in

analyses of policy and economic outcomes, it can be argued that expert opinions in fact

reflect the very same policy and economic outcomes we are trying to present as dependent

variables.

Because of these complications, I use a measure of government partisanship based on

data extracted from party manifestos to assess a party’s left-right position.21

This variable

relies on party programs for the codification of policy emphases and it was produced by the

Comparative Manifestos Project. For party positions, the policy emphases in election

programs are codified into fifty-six categories. The categories are then summarized in a left-

right index.22

For the construction of government partisanship, a party’s average left-right

position is then multiplied by its cabinet weight. In this case, the cabinet weight is the

proportion of parliamentary seats that parties in coalition governments have.23

This is,

however, not a problem because there is considerable evidence showing that “governments

21 Party manifestos data can be criticized for being a reflection of what parties say to win

elections, and not necessarily of what they will do once they have won them. For ananalysis arguing that there is a correlation between party platforms and policy in the

American case, see Ian Budge and Richard Hofferbert, “Mandates and Policy Outputs,”

American Political Science Review, 84 (1990), 111-31. 22 To simplify the interpretation of my results, I have reversed the index. Higher values of

this variable, therefore, will mean governments to the Left. Lower values, governments

to the Right. 23 If there is only one party in government, its cabinet weight is 100%. When there are

more, a party’s weight is given by its proportion of parliamentary seats within the total of

seats held by the coalition parties.



14

apportion their cabinet portfolios to parties in simple pro por tion to the relative percentage of

seats held by each in the lower house of the legislature.”24

The Mediating Variables: Policy

As was mentioned above, the argument supporting the existence of a relationship

between government partisanship and inequality rests on the proposition that the policy

preferences of Left parties raise the wage floor for competition in the labor market. I

examine four policies that partisan governments can use to do this: minimum wages,

government consumption, taxes on labor, and taxes on corporation. In this paper, I explore

whether government partisanship affects these four policies and whether the policies affect

inequality.

(1) Minimum Wages: Minimum wage policies are not uncontroversial. Those who

defend them argue that they limit labor market excesses and increase the wages of the lowest

paid to a socially acceptable level. Those who oppose them argue that their effect is in fact

an increase in unemployment (resulting from the pricing out of low-skilled workers).25

In

relation to inequality, the consequences of minimum wages are, at one level, very

straightforward. The setting up of a minimum wage level makes those who previously had

earnings below it automatically earn more. In this sense, minimum wages can promote

equity goals by “ensuring that ‘fair’ wages are paid to workers.”26

The OECD therefore

concludes that “those countries with higher minimum wage rates relative to the median have

less earnings dispersion and a lower incidence of low pay. In addition, minimum wages

narrow earnings differentials across demographic groups, particularly between the young and

old and between men and women.”27

The scholarship on the effects of minimum wages is,

24 See Bingham Powell, Elections as Instruments of Democracy (New Haven: YaleUniversity Press, 2000), p. 173; and Michael Laver and Norman Schofield, Multiparty

Government (New York: Oxford University Press, 1990). 25 Juan Dolado et al , “The Economic Impact of Minimum Wages in Europe,” Economic Policy, October (1996), 318-73. 26 OECD, Employment Outlook , (1998), p. 32. 27 OECD, Employment Outlook , p. 32.



15

however, not unambiguous.28

Among other things, the effects of minimum wages on the

distribution of earnings greatly depend on their interaction with other policies (most

importantly with tax and benefit ones). As a consequence, there is a great deal of variation

across OECD countries.

I use the ratio of minimum wage to average wage.29

Clearly, the ratio of minimum

wages to average earnings is not an ideal measure. For one thing, it may misrepresent the

effect of minimum wages when other factors (like social benefits or the results of collective

bargaining) act as the actual wage floor.30

It is, however, the most commonly used measure

in analyses in economics and political science. In spite of its limitations, it is a useful tool to

help us understand whether Left governments promote equality at the lower half of the wage

distribution.

(2) Government Consumption: 31

The provision of welfare state services represents an

important way in which governments may influence inequality. Welfare services provide a

way to redistribute wealth to the poor and to insure them against labor market risks. There

are, then, two ways in which government consumption can affect inequality. First, by

insuring workers against risks. As argued by Esping-Andersen,32

welfare programs reduce

people’s dependence on employment as a source of income. Following Iversen and Cusack,

government consumption also directly reduces inequalities in peoples’ access to education,

28 See OECD, Employment Outlook , Annex 2.B for an overview. 29 See Appendix 1 for data sources. 30 Juan Dolado et al , “The Economic Impact of Minimum Wages in Europe,” p. 325. 31 Government consumption consists of goods and services (except military spending).

See Appendix 1 for data sources. 32 Gøsta Esping-Andersen, The Three Worlds of Welfare Capitalism, (Princeton:

Princeton University Press, 1990).



16

health care, etc. 33

As such, it provides the services that allow workers with low wages to

increase their income.34

Left government is therefore expected to promote higher levels of government

consumption. Government consumption is expected to decrease inequality, since higher

levels of these policies increase worker skills and strengthen the bargaining power of wage

earners at the bottom of the wage hierarchy.

(3) Taxes on Labor and on Corporations: OECD governments claim more than a third

of the total annual output of their respective countries to finance their expenditures.

Although there is great diversity in the national taxation systems of the industrialized

democracies, the average tax burden has steadily increased since the late 1960s. While in the

USA taxes as a percentage of GDP have only increased from 25% in 1965 to 27% in 1996; in

Australia, Japan and New Zealand taxes increased from 22% to 30%; and in Europe they

grew dramatically from 27% to 40%.35

The politics of redistribution (and therefore the

politics of inequality) are first related to generating revenue and then to allocating resources

in a way that accomplishes a government's goals. Although political economists have

analyzed the politics of government expenditures and their effects on inequality in some

detail (the rest of this paper represents a good example), the redistributive effects of taxation

have received much less attention. One of the goals of my analysis is to contribute to the

literature by analyzing the consequences of revenue-gathering strategies in terms of the

equality outcomes that they promote. In the following pages I will explore how some

33 Government consumption is also related to public employment and policy makers can

use public employment to support egalitarian wage policies. For details, see Torben

Iversen and Thomas Cusack, “The Causes of Welfare State Expansion,” World Politics, 52 (2000): 329. 34 However, this relationship may have ambiguous results. It may compress wages by

increases those of the bottom 10th

percentile or it may expand them by increasing thewage of the median worker even more. 35 Steven Clark and Flip de Kam, “OECD Taxes Revisited,” OECD Observer 214 (1998),

p. 28.



17

characteristics of fiscal policy (namely the nature of the tax rates on labor and corporate

income) affect the distribution of wages in the OECD countries.

I use a measure of the tax rate on labor income and one of the tax rate on corporate

income. The tax on labor income is calculated following a modification proposed by

Thomas Cusack of the formulae presented by Mendoza et al .36

The tax on corporate income

represent taxes on the corporate income, profits and capital gains of individuals. It is

calculated following a modification proposed by Thomas Cusack of the formulae presented

by Volkerink and de Haan.37

The use of these measures is justified by two reasons. First,

taxes on personal and corporate income are the main source of revenues used to finance

government spending in most OECD countries.38

If taxes, as most analysts would agree,

have a redistributive effect, it should be most evident in the analysis of these rates. Second,

since the focus of these paper is the difference in income between those sectors in an

economy with the lowest wages and those closest to the median, it seems logical to look at

the amount of taxes that richer and poorer individuals are likely to pay. While taxes on labor

income will be shared by a large part of the individuals in the analysis, taxes on corporate

income will be disproportionately directed to the wealthiest in the wage distribution. The

redistributive effects of these taxes seem an important topic to analyze empirically.

The relationships between government partisanship and taxes on labor and that of

taxes on labor and inequality are straightforward. Left governments tend to promote higher

taxes on labor because they need greater revenues to sustain a comprehensive welfare state.39

The relationship between left government and taxes on corporations, on the other hand, is not

36 Tax rates provided by Tom Cusack (Wissenschaftszentrum, Berlin). See Enrique

Mendoza, Assaf Razin and Linda Tesar, “Effective Tax Rates in Macroeconomics,” Journal of Monetary Economics 34 (1994): 297-323. 37 Tax rate provided by Tom Cusack (Wissenschaftszentrum, Berlin). See Bjørn

Volkerink and Jakob de Haan, “Tax Ratios,” OECD Tax Policy Studies 5 (2001). 38 Steven Clark and Flip de Kam, “OECD Taxes Revisited,” p. 30. 39 See, for example, Geoffrey Garrett, Partisan Politics in the Global Economy (New

York: Cambridge University Press, 1998).



18

as simple. Since the early 1970s, Left governments in the OECD face an important

challenge: an increase in the internationalization of the economy. As it is conventionally

argued, in a world characterized by globalization, owners of mobile assets have the ability to

move to markets that provide the highest returns. High levels of mobility therefore lead to

lower taxes on mobile assets (mostly capital) and (to compensate for lost revenues) with

higher taxes on less mobile assets (mostly labor).40

Ganghof shows that this is in fact the

case and that open economies suffer from downward pressures on statutory corporate income

tax rates.41

Given the limitations faced by Left government in open economies, this paper’s

expectations about the influence of government partisanship on taxes on corporations need to

be transformed. It is still possible for taxes on corporations to affect inequality (higher taxes

would be correlated with lower inequality) but it seems unrealistic, given capital exit threats,

to expect that Left governments would increase corporate taxation (regardless of institutional

configurations).

Summary of Hypotheses

Figures 1 and 2 summarize the theoretical claims outlined in the previous sections.

The interactions with corporatism are summarized in Figure 1. When corporatism is low, the

relationship between government partisanship and policy is characterized by a plus sign.

This means that Left governments are expected to raise minimum wages, government

40 See Assaf Razin, and Efriam Sadka, “International Tax Competition and Gains fromTax Harmonization,” Economic Letters 37 (1991): 67-76; and Roger Gordon and Jeffrey

Mackie-Mason, “Why Is There Corporate Taxation in a Small Open Economy?” in

Martin Feldstein et al (eds.), The Effects of Taxation on Multinational Corporations

(Chicago: University of Chicago Press, 1995). 41 See Stephen Ganghof, Global Markets, National Tax Systems, and Domestic Politics,

MPIfG Discussion Paper 01/9, 2001. There is, however, also evidence contradicting this

framework. See Duane Swank and Sven Steinmo, “The New Political Economy ofTaxation in Advanced Capitalist Democracies,” American Journal of Political Science 46

(2002): 642-55; Geoffrey Garrett and Deborah Mitchell, “Globalization, Government

Spending, and Taxation in the OEDCD,” European Journal of Political Research 39(2001): 145-177; Geoffrey Garrett (fn. 38), and Duane Swank, “Funding the Welfare

State: Globalization and the Taxation of Business in Advanced Market Economies,”

Political Studies 46 (1998): 671-92.



19

consumption levels and taxes on labor. Government partisanship is not expected to affect

taxes on corporations because of capital’s exit threat. When corporatism is low, it is also

expected that all the policies of interest will be associated with decreases in inequality.

When corporatism is high, on the other hand, because it enables the social partners to

negotiate effective wage floors and therefore reduces their reliance on government policy, the

effects of policy on inequality are expected to be negligible. Since Left governments will not

promote egalitarian policies unless they are convinced that the institutional context allows

these policies to successfully affect economic outcomes, government partisanship is also

expected to be insignificant when corporatism is high.

[Figures 1 and 2]

The interactions with constitutional structure are summarized in Figure 2. There are

two alternatives. If the dispersion of political power facilitates the obstruction of the political

process by minorities, then government partisanship should be positively associated with

policy when the number of veto points is low. When the number of veto points is high,

government partisanship should have no effect on policy. It is possible, on the other hand,

that veto points allow narrow majorities to overcome the opposition of powerful minorities

and give a disproportionate amount of political control to pro-redistribution groups. If this is

the case, the government partisanship should not have any effects on policy when veto points

are low and it should have a positive effect when veto points are high.

Methodology and Control Variables

I use annual data from a selection of OECD countries42

from 1973 to 1995 and

present ordinary least squares (OLS) results. The pooled data significantly increase the

42 In regressions including government consumption, taxes on labor and taxes on

corporations the countries are Australia, Austria, Belgium, Canada, Denmark, Finland,France, Germany, Italy, Japan, the Netherlands, Norway, Sweden, Switzerland, the UK

and the USA. In regressions including minimum wages, Austria and Switzerland are

missing.



20

number of observations and therefore allow me to test more complex causal models. There

are, however, some complications.

In the presence of residual autocorrelation, OLS estimates will lead to inaccurate

inferences. The results of Durbin-Watson tests indicated that no autocorrelation existed in

the reported regressions.43

Beck and Katz have proposed a method that, when

autocorrelation does not exist, produces consistent standard errors estimates in the presence

of panel heteroscedastic errors.44

Since their recommendations have been widely followed in

the recent comparative political economy literature, I ran the regressions with panel-

corrected-standard-errors. 45

I include fixed effects into my analysis. Fixed effects deal with country-specific

omitted variables by introducing a unit dummy per cross-sectional unit. Given our general

understanding in comparative political economy that there are country-specific effects that

cannot be introduced into the model (specific historical circumstances, difficult to capture

institutional developments, etc) fixed effects is a powerful tool. Fixed effects produce very

robust estimates because they pose a hard test for any given hypothesis (in the sense that they

impose strong restrictions that take away the unit specific variation).46

43 I test for autoregression by calculating the Durbin-Watson d statistic. As recommended by Stimson and Sayrs, for pooled time-series data the Durbin-Watson d statistic is

calculated as a cross-sectional average (that is, the average of all the countries). See

James Stimson, “Regression in Space and Time” American Journal of Political Science 29 (1985): 914-47; and Lois Sayrs, Pooled Time Series Analysis (Newbury Park: Sage

Publications, 1989). 44 See Nathaniel Beck and Jonathan Katz, “What to Do (and Not to Do) with Time-Series

Cross-Section Data,” American Political Science Review, 89 (1995): 634-47; and Nathaniel Beck and Jonathan Katz, “Nuisance vs. Substance,” in John Freeman (ed.),

Political Analysis: Volume 6 (1996). 45 The analysis includes linearly interpolated data for a handful of missing observationsin the wage inequality series. I did not interpolate across gaps of more than three years,

and interpolated observations account for only 13 out of those used in the analyses.46 One controls to be included in the specifications (the constitutional structure variable)displays no variance over time. Because of this, I confirm the validity of the results by

re-estimating the models using random effects. All the substantive findings are

corroborated by the random effects analysis.



21

I run two sets of regressions. In the first set, I regress government partisanship and a

number of control variables (they are specified below) on policy (minimum wages,

government consumption, taxes on labor and taxes on corporations). In the second set, I

regress policy (minimum wages, government consumption, taxes on labor and taxes on

corporations) and a number of control variables (again, they are specified below) on

inequality. All regressions are of the following form:

Y it = α + β 1 X it + β 2...n γ it + N i + ε(pc)

where α represents the constant , X it represents the independent variable of interest,

γ it depicts the control variables, N i country fixed effects and, finally, ε(pc) denotes panel

corrected standard errors.

To simplify the assessment of standard errors and statistical significance, when I

introduce the variables capturing high or low corporatism and high or low veto points I

calculate the interactions the following way. I identify the range of variation in these

variables and use the lowest and highest values.47

Low corporatism is defined as the score of

the USA from 1976 to 1995 (.01) and high corporatism as that of Sweden from 1973 to 1987

(.99). The low veto points number is defined as the score of the Sweden (0) and the high

veto points number as that of the USA (7). Then I introduce the interaction with the variable

of interest minus its sample value into the regression.48

Since I run two sets of regressions, it is important to check that that the independent

variables in the second set are truly exogenous. In other words, since I am running the two

47 See Appendix 1 for details and data sources. 48 I will illustrate the calculation of the interaction with an example. Suppose we want toknow the effect of partisanship on policy when corporatism is low. The lowest

corporatism value is 0.01. I run a regression with the following specification for the

interaction (everything else is the same as in other regressions):Policy = partisanship + corporatism + [ partisanship * (corporatism – 0.01) ]

Under these conditions, the coefficient and the standard error of partisanship will reflect

the effects of partisanship on policy when corporatism is 0.01. For more details aboutthis way to calculate interactions and their significance, see Jeffrey Wooldrige,

Introductory Econometrics (Cincinnati OH: South-Western College Publishing, 2000):

190-191.



22

sets of regressions independently, the explanatory variables in the second set need to be truly

exogenous.49

Hausman tests run on all the explanatory variables in the second set of

regression show that they are indeed exogenous.50

Control Variables for the Analysis of Policy51

Although they are not related theoretically to my main claims, there are a number of

variables that need to be included in the analysis. In some cases opposing claims about their

influence over government partisanship have been provided in the literature and in all cases

there are strong theoretical or empirical reasons to believe that they affect the outcomes I am

interested in analyzing.

Union Density and Bargaining Centralization: Both because of their direct

involvement in industrial relations (negotiations covering work regulations and wages being

the most clear examples) and because of their capacity to influence political parties, the

behavior of unions and employers is a relevant factor in a government’s decision of policy

orientation.

International and Financial Openness: There are two contradictory accounts of the

effects of internationalization on partisan politics. There is first a large literature suggesting

that growing levels of international openness, integration and interdependence result in a

blurring of partisan differences caused by the ina bility of social democratic parties to produce

policies that do not conform to market forces.52

Then there are some authors who argue

49 If the explanatory variables in the second set of regressions were correlated with the

error (with would mean the error from the first set is correlated with the error of the

second set), we would need to do a 2 stage least square (2SLS). 50 See Damodar Gujarati, Basic Econometrics (New York: McGraw-Hill, 1995): 669-73. 51 See Appendix 1 for details and data sources. 52 See, for example, Torben Iversen, “Power, Flexibility and the Breakdown ofCentralized Wage Bargaining,” Comparative Politics, 28 (1996): 399-436, Jonathan

Moses, “Abdication from National Policy Autonomy: What’s Left to Leave?” Politics

and Society, 22 (1994): 125-48, Paulette Kurzer, Business and Banking: Political Changeand Economic Integration in Western Europe (Ithaca: Cornell University Press, 1993),

and Fritz Scharpf, Crisis and Choice in European Social Democracy (Ithaca, New York:

Cornell University Press, 1991).



23

either that international forces do not affect some partisan differences53

or that they actually

have strengthened the influence of partisanship on policies and economic outcomes.54

The

results presented in the following pages do not address whether international dependence

limits the autonomy of governments. Rather they look at the great variance in policy within

the sample and try to assess the factors that are responsible for it.

Government Debt: Government debt is introduced into the analysis as a measure of

the possible limitations affecting a government’s choice of policy. The general argument is

that governments with more debt have fewer resources at their disposal and therefore would

encounter more difficulties when trying to expand their spending (whether directed to

ALMPs or replacement rates). One widely accepted interpretation of the policy changes of

the early 1980s, for example, is that many governments had reached unsustainable levels of

public debt.55

Unemployment: The rate of unemployment represents a measure of the need for the

policies chosen as dependent variables. Numerous authors have argued that the increasing

levels of some policies simply result from increasing needs (whether demographic,

economic, or other).56

I attempt to engage, at least partially, these arguments by controlling

for the effects of unemployment.

53 Like Carles Boix, Political Parties, Growth and Equality (New York: Cambridge

University Press, 1998) and Geoffrey Garrett and Peter Lange, “Political Responses toInterdependence: What’s ‘Left’ for the Left?” International Organization, 45(1991):

539-564. 54 Geoffrey Garrett, Partisan Politics in the Global Economy (New York: Cambridge

University Press, 1998). 55 Herman Schwartz, “Small States in Big Trouble,” World Politics, 46(1994): 527-55. 56 Phillip Cutright, “Political Structure, Economic Development, and National Social

Security Programs,” American Journal of Sociology, 70 (1965): 537-50, and HaroldWilensky, The Welfare State and Equality: Structural and Ideological Roots of Public

Expenditures (Berkeley: University of California Press, 1975) could be considered early

proponents of this thesis. See Thomas Janoski, “Direct State Intervention in the LaborMarket,” in Thomas Janoski and Alexander M. Hicks (eds.), The Comparative Political

Economy of the Welfare State (New York: Cambridge University Press, 1994) for a

review.



24

GDP growth: Most analyses of economic policy include a measure of economic

growth. In this paper’s analysis, this is particularly relevant because it is important to control

for the effects of macroeconomic growth on the behavior of governments.

Control Variables for the Analysis of Wage Inequality57

Given the nature of the outcomes to be explained (inequality at the lower half of the

wage distribution rather than policy), some of the control variables for this portion of the

analysis are different from those in the previous section.

Unemployment: There are two potential effects of unemployment on inequality.

First, the basic insight of the literature on labor market segmentation is that unskilled, low-

paid workers are more readily substitutable than more skilled, high-paid workers, and

consequently that their bargaining position is more immediately and more adversely affected

by unemployment.58

On the other hand, employers are more likely to lay off unskilled than

skilled workers during economic downturns. To the extent that it entails a disproportionate

loss of low-paid jobs, an increase of unemployment produces wage compression by altering

the composition of the labor force.

LDC Trade: Wood argues that much of the trend towards increased wage inequality

in the OECD countries in the 1980s can be attributed to increased manufacturing trade with

lesser developed countries.59

The basic logic of Wood’s analysis is that by importing less

skill-intensive goods from low-wage countries, OECD countries are essentially importing

low skill labor, so that the effective supply of unskilled labor relative to skilled labor has

grown, putting downward pressure on the relative wages of the unskilled.

57 See Appendix 1 for details and data sources. 58 Cf. James Galbraith, Created Unequal (New York: The Free Press, 1998); and

Katherine Bradbury, “Rising Tide in the Labor Market: To What Degree do ExpansionsBenefit the Disadvantaged,” New England Economic Review, 32 (May-June 2000), 3-33. 59 Adrian Wood, North-South Trade, Employment and Inequality (Oxford: Clarendon

Press, 1994).



25

Female Labor Force Participation: To the extent that women are on average less

educated and/or have less work experience than men, an increase in the proportion of the

total labor force made u p of women represents an increase in the relative supply of unskilled

or less skilled labor.60

However, as women acquire skills through labor force participation

higher levels of female labor force participation should eventually reduce the skill gap

between men and women. As women increasingly come to occupy full-time jobs, moreover,

the union density gap between men and women should erode and unions will have a strong

incentive to pursue a reduction of wage differentials based on gender.

Private Service Employment: It is often argued that wage inequality and private

service employment are associated (because the scope for productivity growth in services is

limited, pricing closely reflects labor costs, and demand for these services is highly price

sensitive).61

If the assumption that the production of personal services with a high content of

unskilled labor is tightly constrained by labor costs is relaxed, however, the opposite

association between wage inequality and private service employment would result.

Union Density: Following Freeman, two dimensions of the relationship between

unionization and wage distribution can be distinguished.62

The first dimension concerns the

distribution of wages among union members, the second concerns wage differentials between

union members and non-members. Several factors promote that the wage distribution of

unionized firms or sectors will be more compressed than that of the non-unionized firms or

sectors. Among them, the fact that unions approximate the logic of democratic decision-

60 Cf. Robert Topel, “Wage Inequality and Regional Labor Market Performance in the

United States,” in Toshiaki Tachibanaki, ed., Labour Market and Economic Performance(New York: St. Martin's Press, 1994), 101-132; Lennart Svensson, Closing the Gender

Gap (Lund: Ekonomisk-historiska föreningen, 1995). 61 Torben Iversen and Anne Wren, “Equality, Employment and Budgetary Restraint,”World Politics, 46 (1998), 527-555. 62 Richard Freeman, “Unionism and the Dispersion of Wages,” Industrial and Labor

Relations Review 34 (October 1980), 3-23; and Richard Freeman, “Union Wage Practicesand Wage Dispersion within Establishments,” Industrial and Labor Relations Review 36

(October 1982), 3-20. Cf. also Blau and Kahn, “International Differences in Male Wage

Inequality.”



26

making and that they have a strong interest in curtailing wage setting based on the subjective

decisions of foremen and managers.

Wage Bargaining Centralization: The standard argument linking centralization to

wage compression asserts that centralization facilitates the reduction of inter-firm and inter-

sectoral wage differentials since it means that more firms and sectors are included in a single

wage settlement. Additionally, it can be argued that centralization produces wage

compression by altering the distribution of power among actors (this would follow a median

voter logic63

) and by making bargaining more transparent.

Government Employment: The size of the public sector (government employees as a

percentage of the total labor force) is often considered to be negatively associated with wage

inequality for several reasons. While sheltered from competition in product markets, public-

sector employers are more directly exposed to political pressures favoring equality and robust

wage growth.64

The egalitarian logic of public-sector wage setting seems most pronounced

with regard to equal pay provisions for women and minorities.

Results

Figures 3 to 5 present the results of this paper’s analyses. Because so many

regressions must be run (one for the relationship between government partisanship and early

policy, one for the relationship between each policy and inequality, plus all the interactions),

only the results for the variables of interest (partisanship and policy) are reported in the

figures.65

Figures 3 to 5 reflect the two sets of relationships hypothesized in this paper. In

each figure, there are causal arrows from government partisanship to the policies (minimum

wage, government consumption, taxes on labor and taxes on corporations) and then there are

63 Michael Wallerstein, “Wage-Setting Institutions and Pay Inequality in Advanced

Industrial Societies,” American Journal of Political Science, 43 (July 1999), 649-680. 64 Geoffrey Garrett and Christopher Way, “Corporatism, Public Sector Employment, and

Macroeconomic Performance,” Comparative Political Studies, 32 (1999), 411-34. 65 The results for all variables (including controls) are presented in Appendix 2.



27

causal arrows from the policies to inequality at the lower half of the wage distribution. The

numbers next to the arrows represent the coefficients for these variables in the regressions I

ran. The asterisks signify the statistical significance in the usual manner (*** if p-value <

.01, ** if < .05, and * if < .1). The p-values were calculated using the panel-corrected

standard errors.66

The absence of any coefficient estimates means that the variable was

statistically insignificant.

[Figure 3]

Figure 3 presents a model without the institutional interactions that are the focus of

this paper. It is reported simply to show that the assessment of the effects of partisanship and

policy is not very illuminating when the mediating role of institutions is not specified. Figure

3 does show that government partisanship is associated with higher levels of minimum wages

and of government consumption. These relationships are significant at better than the 95%

level of confidence (for the minimum wage) and the 99% level of confidence (for

government consumption). But every other relationship turns out to be insignificant. These

findings provide a reason for the puzzling lack of significance found by Pontusson, Rueda

and Way.67

They are, however, transformed when the institutional interactions are

introduced.

[Figure 4]

Figure 4 presents two sets of results: the estimates for partisanship and policy when

corporatism is low and when corporatism is high. Going back to the theoretical claims

summarized in Figure 1, the relationship between government partisanship and the first three

policies (minimum wage, government consumption and taxes on labor) was expected to be

positive and significant when corporatism was low. The relationship between government

partisanship and taxes on corporations, on the other hand, was expected to be insignificant.

Looking now at the results in Figure 4, all expectations are confirmed. It is in fact the case

66 See Appendix 2. 67 See fn. 5.



28

that Left government is associated with high minimum wages, government consumption and

taxes on labor when corporatism is low. Also as expected, the influence of partisanship on

taxes on corporations is statistically insignificant even when corporatism is low. It seems

that capital-exit threats are indeed powerful enough to stop Left governments from taxing

corporations.

Still when corporatism is low but now turning to the relationship between policy and

inequality, this paper’s expectations were simple. In Figure 1, I hypothesized that all policies

would have significant and negative effects on inequality. Figure 4 presents strong support

for these hypotheses. As expected the absence of corporatism makes policies very effective

at reducing inequality. Increases in the minimum wage, government consumption and taxes

on corporations are significantly related to decreases in inequality at the lower half of the

wage distribution. The only exception is the influence of taxes on labor (which is

insignificant). There are reasons to suspect, however, that this finding is influenced by the

nature of the dependent variable. As was mentioned in a previous section, the measure of

inequality used in this paper refers to gross income and therefore does not account for the

influence of labor income taxes. This may be the reason for the lack of significance found in

Figure 4.

The results for the partisanship and policy variables when corporatism is high are also

presented in Figure 4. Again going back to Figure 1, the expectations for these relationships

were straightforward. When corporatism is high, it was hypothesized, the actions of the

social partners in the labor market “take care” of wage compression. This leaves little room

for policy to affect inequality. Policy-makers understand this process and therefore promote

policy without significant partisan effects. Statistical insignificance was therefore expected

in all relationships. Turning to Figure 4 it seems clear that this is indeed the case when

corporatism is high. The results provide a remarkable amount of support to the hypotheses.

All relationships are insignificant. Neither the influence of policy on inequality nor that of

partisanship on policy are of importance. The only exception is the effect of government



29

consumption on inequality. Although the influence of this variable is small (both in terms of

the size of the coefficient and of the level of significance), it is not negligible. Interestingly,

the sign of this relationship is positive (which means that higher government spending on

education, health and transfers are in fact associated with an increase in inequality). This

seems to indicate that the effects of government services raise the wage of the median worker

more than that of the poorest ones at the bottom of the distribution.

[Figure 5]

Figure 5 presents the results of the regressions introducing the interaction between the

variables of interest and constitutional structure. In this case, there were two opposing sets of

hypotheses. As reflected in Figure 2, if the dispersion of political power facilitates the

obstruction of the political process by minorities, then government partisanship should be

positively associated with policy when the number of veto points is low. When the number

of veto points is high, government partisanship should have no effect on policy. If, on the

other hand, veto points allow narrow majorities to overcome the opposition of powerful

minorities and give a disproportionate amount of political control to pro-redistribution

groups, the government partisanship should not have any effects on policy when veto points

are low and it should have a positive effect when veto points are high. The results in Figure

5 show that the second interpretation of the effects of partisanship conditional on the number

of veto points is indeed the more accurate one. It is in fact the case that government

partisanship does not significantly influence policy when the number of veto points is low.

Conversely, the results in Figure 5 provide some support for the hypothesis that government

partisanship will affect policy when the number of veto points is high. Left government is

associated with higher levels of minimum wages and government consumption when a

political system is characterized by numerous veto points. The results for taxes on labor,

however, do not support this point (those for taxes on corporations were expected to be

insignificant).



30

Conclusions

It is perhaps appropriate to conclude the paper by briefly summarizing its main

points. The main focus of the paper was the relationship between government partisanship,

policy and wage inequality at the lower half of the wage distribution. The analysis was

motivated by the absence of government partisanship effects on inequality found by other

works in the literature. I presented a set of non-intuitive hypotheses. My expectations are

theoretically derived from conceptually differentiating between political agency and the role

of institutions. I focus on two sets of institutions: those related to corporatism and to

constitutional structure. I argued that when corporatism is low, Left governments would

raise minimum wages, government consumption levels and taxes on labor. Government

partisanship was not expected to affect taxes on corporations because of capital’s exit threat.

When corporatism is low, it was also expected that all the policies of interest will be

associated with decreases in inequality. Because corporatist structures enable the social

partners to negotiate effective wage floors and therefore reduces their reliance on government

policy, when corporatism is high, on the other hand, the effects of policy on inequality were

expected to be negligible. Since Left governments will not promote egalitarian policies

unless they are convinced that the institutional context allows these policies to successfully

affect economic outcomes, government partisanship was also expected to be insignificant

when corporatism is high.

Regarding the effects of constitutional structure, I presented two opposing sets of

hypotheses. If the dispersion of political power facilitates the obstruction of the political

process by minorities, then government partisanship was expected to be positively associated

with policy when the number of veto points is low. When the number of veto points is high,

government partisanship was expected to have no effect on policy. If, on the other hand,

veto points allow narrow majorities to overcome the opposition of powerful minorities and

give a disproportionate amount of political control to pro-redistribution groups, government



31

partisanship was expected to have no effects on policy when veto points are low and to have

a positive effect when veto points are high.

To explore this issue, I explored the effects of government partisanship on four kinds

of policies and of these policies on inequality. I found a remarkable amount of support for

the paper’s hypotheses regarding corporatism. The results also corroborated the second

interpretation of the effects of constitutional structure (veto points allow narrow majorities to

overcome the opposition of powerful minorities and empower pro-distribution groups).



32

TABLE 1: MEANS AND PERCENTAGE CHANGES IN I NEQUALITY,

90-10 AND 50-10 R ATIOS

50-10 R ATIOS 90-10 R ATIOS COUNTRYAND YEARS

COVERED MEAN % ∆ MEAN % ∆

AUSTRALIA (1976-1995)

1.66 3.1 2.81 10.6

AUSTRIA (1980-1994)

1.96 0.0 3.53 6.1

BELGIUM (1986-1993)

1.45 -1.4 2.34 -6.7

CANADA (1973-1994)

2.30 9.1 4.24 12.1

DENMARK (1980-1994)

1.40 -2.8 2.18 0.0

FINLAND (1977-1995)

1.46 -10.2 2.45 -11.7

FRANCE (1973-1995)

1.66 -5.7 3.27 -10.8

GERMANY (1984-1995)

1.63 -11.9 2.79 -9.7

ITALY (1986-1995)

1.42 -3.4 2.32 5.9

JAPAN

(1975-1995)

1.70 -6.3 3.07 -1.3

NETHERLANDS (1977-1995)

1.56 5.8 2.54 9.7

NORWAY (1980-1994)

1.39 -6.4 2.08 -3.8

SWEDEN (1975-1995)

1.33 0.0 2.07 -1.8

SWITZERLAND (1990-1995)

1.61 0.0 2.72 2.2

U NITED K INGDOM (1973-1995)

1.78 1.5 3.17 13.5

USA (1973-1995) 2.00 11.0 4.60 22.3

AVERAGE 1.64 -1.1 2.89 2.29

S TANDARD D EVIATION 0.26 6.38 0.74 9.79

Notes and Sources : The percentage changes measure the variation from earliest to latest available observation in thecountry series. See OECD, “Earnings Inequality, Low-paid Employment and Earnings Mobility,” pp. 61-62 for allcountries except the U.S.; for the U.S., OECD, “Earnings Inequality,” p. 161 and OECD, “Earnings Inequality, Low-

paid Employment and Earnings Mobility,” p. 103.



33

TABLE 2:

THE EFFECTS OF GOVERNMENT PARTISANSHIP O N I NEQUALITY IN THE LOWER

HALF OF THE WAGE DISTRIBUTION

CONSTANT - -.125(.061)

.041

LAGGED

DEPENDENT VARIABLE .484

(.065)

<.001

.980(.015)

<.001

CABINET PARTISANSHIP

.003

(.005)

.593

.008

(.005)

.112

U NEMPLOYMENT

R ATE

-.005

(.004)

.166

-.001

(.003)

.705

LDC TRADE

-.001

(.006).814

-.005

(.004).188

FEMALE LABOR FORCE

PARTICIPATION -.025

(.031)

.412

.008

(.013)

.526

PRIVATE

SECTOR

SERVICES

-.002(.034)

.950

.027(.009)

.004

U NION DENSITY

-.018

(.010)

.086

-.006

(.003)

.078

WAGE BARGAINING

CENTRALIZATION -.028

(.007)

<.001

.004

(.003)

.164 PUBLIC

SECTOR EMPLOYMENT -.068

(.020)

.001

.010

(.005)

.047

N 203 203

ADJUSTED R 2 .99 .99

FIXED EFFECTS YES NO

Notes : All entries are OLS estimates. Numbers in bold are estimated coefficients; numbers in parentheses

are their panel-corrected standard errors; numbers in italics are p-values from two-sided t-tests.



34

Figure 1:

Low Corporatism Expectations

Government

Partisanship

Minimum

Wage

Government

Consumpt.

Taxes on

Labor

Taxes on

Corporations

Inequality

+

+

+

-

-

-0

-

High Corporatism Expectations

Government

Partisanship

Minimum

Wage

Government

Consumpt.

Taxes on

Labor

Taxes on

Corporations

Inequality

0

0

0

0

0

00

0



35

Figure 2:

Low Veto Points Expectations

Government

Partisanship

MinimumWage

Government

Consumpt.

Taxes on

Labor

Taxes on

Corporations

0 or +

0 or +

0 or +

0 or +

High Veto Points Expectations

Government

Partisanship

Minimum

Wage

Government

Consumpt.

Taxes on

Labor

Taxes on

Corporations

+ or 0 + or 0

+ or 0

0



36

Figure 3:

Results with no Interactions

Government

Partisanship

Minimum

Wage

Government

Consumpt.

Taxes on

Labor

Taxes on

Corporations

Inequality

.001**

.013***



37

Figure 4:

Low Corporatism Results

Government

Partisanship

Minimum

Wage

Government

Consumpt.

Taxes on

Labor

Taxes on

Corporations

Inequality

.001**

.025***

.043***

-.188*

-.014**

-.001***

High Corporatism Results

Government

Partisanship

Minimum

Wage

GovernmentConsumpt.

Taxes on

Labor

Taxes on

Corporations

Inequality

.007*



38

Figure 5:

Low Veto Points Results

Government

Partisanship

Minimum

Wage

Government

Consumpt.

Taxes on

Labor

Taxes on

Corporations

High Veto Points Results

Government

Partisanship

Minimum

Wage

GovernmentConsumpt.

Taxes on

Labor

Taxes on

Corporations

.001***

.037***



39

Appendix 1

DEFINITIONS OF VARIABLES AND DATA SOURCES

GOVERNMENT PARTISANSHIP: Source for all countries but Japan: Michael McDonald

and Silvia Mendes, Parties in Parliaments and Governments, 1950-1995 (May 2001

Version), Political Science Department, Binghamton University – SUNY. Data for Japan

created by author using Comparative Manifestos Project left-right party index and Jaap

Woldendorp, Hans Keman, and In Budge, Party Government in 48 Democracies (1945-

1998) (Boston: Kluwer Academic Publishers, 2000).

MINIMUN WAGES: Ratio of minimum wage to average wage. Source: David Neumark

and William Wascher, “A Cross-National Analysis of the Effects of Minimum Wages on

Youth Employment,” NBER Working Paper 7299, 1999.

GOVERNMENT CONSUMPTION: Goods and services consumed by the government

(except military spending). Sources: Thomas Cusack, “The Changing Contours of

Government.” Wissenschaftszentrum Berlin Discussion Paper P91-304 (Berlin:

Wissenschaftszentrum, 1991), OECD National Accounts, OECD Labour Statistics,

Stockholm Peace research Institute (various years) and Torben Iversen and Thomas Cusack,

“The Causes of Welfare State Expansion” World Politics, 52 (2000): 313-49.

TAXES ON LABOR AND CORPORATIONS: The tax on labor income is

calculated following a modification proposed by Thomas Cusack of the formulae

presented by Enrique Mendoza, Assaf Razin and Linda Tesar, “Effective Tax Rates

in Macroeconomics,” Journal of Monetary Economics 34 (1994): 297-323. The tax

on corporate income represent taxes on the corporate income, profits and capital

gains of individuals. It is calculated following a modification proposed by Thomas



40

Cusack of the formulae presented in Bjørn Volkerink and Jakob de Haan, “Tax

Ratios,” OECD Tax Policy Studies 5 (2001). Source: Tom Cusack

(Wissenschaftszentrum, Berlin).

CONSTITUTIONAL STRUCTURE/VETO POINTS: Additive index composed

of five indicators: (1) federalism (0=absence, 1=weak, 2=strong) (2) parliamentary

government =0, versus presidentialism or other =1 (3) proportional representation

=0, modified proportional representation=1, majoritarian=2 (4) bicameralism

(1=weak, 2=strong), (5) frequent referenda=1. Sources: Klaus Armingeon, Michelle

Beyeler and Sarah Menegale, Comparative Political Data Set 1960-2001, Institute of

Political Science, University of Berne, 2002; Evelyne Huber, Charles Ragin and

John Stephen, “Social Democracy, Christian Democracy, Constitutional Structure,

and the Welfare State,” American Journal of Sociology, 99 (1993), 711-749; and

Manfred Schmidt, “When Parties Matter: A Review of the Possibilities and Limits of

Partisan Influence on Public Policy,” European Journal of Political Research 30

(1996): 155-183.

UNION DENSITY: Employed union members as a percentage of employed labor

force (“net density”) for all countries but Canada. The Canadian figures include

unemployed and retired people who retain their membership in the numerator and

the unemployed in the denominator (“gross density”). Source: Jelle Visser,

“Unionization Trends Revisited.” Centre for Research of European Societies and

Industrial Relations, Amsterdam, 1996.

CENTRALIZATION: Wage-bargaining centralization as measured by Torben

Iversen (Department of Government, Harvard University). Higher figures signify

more centralization. Iversen classifies country-years according to the relative weight

of three levels of bargaining (local, industry and national), and multiplies these

weights by a measure of the concentration of union membership at each level. Thus

there are two distinct sources of variation in Iversen’s index: (1) index scores



41

increase as the relative significance of higher levels of bargaining increases; and (2)

scores increase as union membership becomes more concentrated at any of these

bargaining levels (especially at those that are more significant). See Torben Iversen,

Contested Economic Institutions (New York: Cambridge University Press, 1999) for

a complete specification. These figures have been transformed into a moving

average, so that the value for a given year is the average of the actual value for that

year and the previous four years.

PUBLIC SECTOR EMPLOYMENT: Government employees (not including

employees of state-owned enterprises) as percentage of total employed labor force.

Source: OECD, “Historical Statistics” (electronic database).

UNEMPLOYMENT: Unemployed is measured as percentage of total labor force

(source: OECD electronic database and OECD, Historical Statistics 1960-1995).

LDC TRADE: Trade with lesser developed countries (LDCs) as percentage of

GDP, not including trade with OPEC countries. For all countries but Belgium,

figures up to 1990 were provided by Geoffrey Garrett (Department of Political

Science, Yale University). Belgian and post-1990 figures were calculated on the

basis of OECD, Monthly Trade Statistics.

FEMALE LABOR FORCE PARTICIPATION: Female labor force as percentage

of total labor force. Source: OECD, “Historical Statistics.”

PRIVATE SERVICE EMPLOYMENT: Service employment as percentage of total

employment minus government employment as a percentage of total employment. Source:

OECD, “Historical Statistics.”

INTERNATIONAL OPENNESS: International openness is measured as imports plus

exports as percentage of GDP. Source: OECD electronic database and OECD, Historical

Statistics 1960-1995.

FINANCIAL OPENNESS: Financial openness is measured as the sum of the index for

restrictions on payments and receipts of goods and invisibles, the index for restrictions on



42

payments and receipts of capital, and the index for legal international agreements that

constrain a nation’s ability to restrict exchange and capital flows. Values for 1994 and 1995

were extrapolated. Source: Klaus Armingeon, Michelle Beyeler and Sarah Menegale,

Comparative Political Data Set 1960-1998 (Institute of Political Science, University of

Berne, 2000).

GOVERNMENT DEBT: Government debt is measured as the level of consolidated central

government debt as a percentage of GDP. Source: Franzese (1998).

GDP GROWTH: GDP growth is measured as year-to-year percentage changes (source:

OECD electronic database and OECD, Historical Statistics 1960-1995).



43

Appendix 2

Regression Results for Figure 3 (Main Results, No Interaction)

The Determinants of Policy

Minimum Government Taxes on Taxes on

Wage Consumption Labor Corporations

Cabinet Partisanship .0005489 .0131514 .0163283 -.9319684

(.0002551) (.0048165) (.0107727) (.9001847)

.031 .006 .130 .301Union Density .0066523 .0397252 .0099524 -2.27957

(.0011152) (.0155704) (.0301252) (4.177516)

.000 .011 .741 .585

Wage Bargaining .148861 1.443857 7.861173 183.2231 Centralization (.0751686) (1.005594) (2.40529) (343.2208)

.048 .151 .001 .593International -.0007903 -.0117178 -.0722099 -1.906428

Openness (.0006686) (.0101839) (.0226641) (1.9121920)

.237 .250 .001 .319

Financial -.0005233 .1044676 .0557051 -11.69262 Openness (.004369) (.0435973) (.1120438) (7.952464)

.905 .017 .619 .141

Government -.0102392 -.3234643 14.44117 -15.72714 Debt (.0333572) (.4631467) (1.028134) (108.7242)

.759 .485 .000 .885

Unemployment .0054605 .3334178 .4002393 3.778923 Rate (.00202) (.0308687) (.0667158) (4.7081)

.007 .000 .000 .422

GDP Growth .0009247 -.1350981 -.1229121 .7917906

(.0019446) (.0229763) (.0508176) (4.295496).634 .000 .016 .854

Constant -.3956314 9.65194 7.484792 223.4798

(.1015193) (1.157052) (2.368358) (211.7439).000 .000 .002 .291

N 290 348 338 329

R 2

.9404 .9495 .9612 .0398

All entries are OLS estimates. Numbers in bold are estimated coefficients; numbers in

parentheses are their panel-corrected standard errors; numbers in italics are p-values from

two-sided t-tests. Estimates for country dummies are not reported.



44

The Determinants of Inequality

Minimum Government Taxes on Taxes onWage Consumption Labor Corporations

Policy (Specified -.0643246 -.0026517 .0003032 -2.41e-06 In First Row) (.0597115) (.0032089) (.0012327) (3.73e-06)

.281 .409 .806 .518

Unemployment -.0021089 -.0026393 -.0028032 -.0030683 Rate (.0016624) (.0014762) (.0014904) (.0015895)

.205 .074 .060 .054

LDC Trade .0071869 .0076988 .0071895 .0072327

(.0029786) (.0029061) (.0028633) (.0029488).016 .008 .012 .014

Female Labor Force .003295 .0018789 .0028213 .0029029

Participation (.0029458) (.00329) (.0028299) (.0029656)

.263 .568 .319 .328Private Service -.0012948 -.0004754 -.0016003 -.0016097

Sector (.0017503) (.0019298) (.0018244) (.00181)

.459 .805 .380 .374Union Density .0011486 -.0002717 -.0004619 -.0006295

(.0013701) (.0011659) (.0011801) (.0012366)

.402 .816 .695 .611Wage Bargaining -.2847651 -.2948242 -.3338687 -.3595873

Centralization (.0672435) (.063227) (.0687342) (.0681547)

.000 .000 .000 .000Public Sector -.0138848 -.0106761 -.0123546 -.0117268

Employment (.0023477) (.0030352) (.0028464) (.0022702)

.000 .000 .000 .000Constant 2.014364 2.070082 2.107196 2.117513

(.1588314) (.1502412) (.1487234) (.1588302)

.000 .000 .000 .000

N 205 222 222 213

R 2

.9747 .9756 .9755 .9749


parentheses are their panel-corrected standard errors; numbers in italics are p-values fromtwo-sided t-tests. Estimates for country dummies are not reported.



45

Regression Results for Figure 4



Wage Consumption Labor CorporationsCabinet Partisanship

Corporatism is .0009553 .02533 .0432445 .8875092 Low (.01) (.0005096) (.0072717) (.0134767) (1.251195)

.061 .000 .001 .478

Corporatism is -.000137 -.0030625 -.0153366 -3.972664

High (.99) (.0006764) (.0096935) (.0233765) (3.780203).840 .752 .512 .293

Corporatism .1301303 1.030228 -6.688961 263.9869

(.0896874) (1.389138) (3.042322) (492.8069)

.147 .458 .028 .592Union Density .0063651 .0382604 .0117861 -2.492325

(.0011026) (.0153101) (.0296878) (4.270994)

.000 .012 .691 .560Wage Bargaining .1431083 1.484446 9.384893 172.2172

Centralization (.0797194) (1.010216) (2.31408) (288.3194)

.073 .142 .000 .550International -.0008052 -.0122439 -.0726906 -1.926598

Openness (.0006617) (.0100618) (.0224608) (1.951015)

.224 .224 .001 .323Financial .0008923 .1246415 .0544854 -7.32711

Openness (.0043496) (.0446082) (.1084007) (7.456081)

.837 .005 .615 .326Government -.0158449 -.4073702 14.49643 -31.62622

Debt (.0331669) (.4629413) (1.015794) (112.0451)

.633 .379 .000 .778


.006 .000 .000 .439

GDP Growth .0010598 -.1332531 -.1201858 1.108425 (.0019251) (.0225736) (.0499664) (4.208872)

.582 .000 .016 .792

Constant -.4155271 9.406055 8.273959 160.97 (.1011035) (1.178893) (2.396136) (215.8107)

.000 .000 .001 .456

N 290 348 338 329R

2.9411 .9502 .9624 .0480






46

The Determinants of Inequality

Minimum Government Taxes on Taxes onWage Consumption Labor Corporations

Policy (SpecifiedIn First Row)

Corporatism is -.1876618 -.0138351 .0014849 -.0013841

Low (.01) (.1014511) (.0060624) (.002564) (.0004418)

.064 .022 .563 .002Corporatism is .0540337 .0072869 -.0006683 -1.55e-06

High (.99) (.071138) (.0038224) (.001643) (3.61e-06)

.448 .057 .684 .667Corporatism -.0788808 -.3670436 .1190394 -.0495208

(.0918144) (.1695946) (.1688838) (.0664126)

.390 .030 .481 .456

Unemployment -.0024251 -.002412 -.0026831 -.0032901 Rate (.0016891) (.0014569) (.001495) (.0016331)

.151 .098 .073 .044LDC Trade .006695 .0098622 .0064025 .0060687

(.0029644) (.0030503) (.0029502) (.0029734)

.024 .001 .030 .041

Female Labor Force .0047839 .0009731 .0026306 .0024898 Participation (.0029702) (.0033987) (.002697) (.0029058)

.107 .775 .329 .392

Private Service -.0024776 .0005268 -.0013548 -.0020748 Sector (.0018295) (.0020229) (.0018421) (.0017882)

.176 .795 .462 .246

Union Density .0008833 -.0001588 -.0004433 -.0010562 (.0013412) (.0011472) (.0011684) (.0012177)

.510 .890 .704 .386

Wage Bargaining -.2850047 -.3284604 -.3387357 -.408379

Centralization (.0695431) (.0642466) (.0685666) (.0691053).000 .000 .000 .000

Public Sector -.0128194 -.0126009 -.01225 -.0115403

Employment (.0024252) (.0029683) (.0028719) (.0022338).000 .000 .000 .000

Constant 2.007146 2.292805 2.068716 2.255708

(.1567431) (.1926206) (.1619231) (.1633958).000 .000 .000 .000

N 205 222 222 213

R 2

.9752 .9767 .9756 .9758






Regression Results for Figure 5



Wage Consumption Labor CorporationsCabinet Partisanship

Veto Points are .0000822 .001644 .0108068 -2.907732 Low (0) (.0004002) (.0076007) (.0215683) (2.890083)

.837 .829 .616 .314

Veto Points are .0014651 .0373835 .0266488 2.375875

High (7) (.0003821) (.0095659) (.0219239) (2.504389).000 .000 .224 .343

Veto Points -.2053637 .2471363 -8.715493 -81.43696

(.0201474) (.3290694) (.7490089) (61.76376)

.000 .453 .000 .187Union Density .0065592 .0380818 .0092398 -2.459732

(.0011108) (.0154434) (.0301757) (4.149658).000 .014 .759 .553

Wage Bargaining .1649101 1.779668 7.955202 214.0713

Centralization (.0759) (1.016586) (2.418801) (343.7188)

.030 .080 .001 .533International -.0008256 -.0122587 -.0722009 -1.942861

Openness (.000666) (.0101152) (.0226248) (1.921654)

.215 .226 .001 .312Financial -.0011196 .0914944 .0505165 -13.6859

Openness (.0044512) (.0441894) (.1138394) (8.609011)

.801 .038 .657 .112Government -.0043964 -.2251226 14.47656 1.939598

Debt (.0339062) (.4611567) (1.037909) (110.0355)

.897 .625 .000 .986


.006 .000 .000 .449

GDP Growth .0009273 -.1343195 -.1229831 .8925868 (.0019393) (.0227912) (.0508176) (4.251502)

.633 .000 .016 .834

Constant .4264531 8.676608 42.37103 557.3256 (.143077) (2.015055) (4.24375) (416.1577)

.003 .000 .000 .180

N 290 348 338 329R

2.9407 .9502 .9613 .0445


Taxation and Redistribution OECD

Documents

Transcript of Taxation and Redistribution OECD