. 9
( 11)


trol for government employment) and 1.5 percent (without control for

19 Private communication with Neal Beck.
20 The results stand without interpolation, but Norway drops out since there are only two
stand-alone observations that get eliminated when differencing.

Forces of Change

Table 6.7. Replication of Regression Results Using d5/d1 Ratios

Consumer Services (Isic 6 + Isic 9)a
Services (Isic 8)a
’1.64——— ’3.67——— ’3.75——— ’2.58———
(’2.71) (’4.59) (’4.59) (’2.58)
0.73——— 1.56——— 1.59——— 1.30———
d5/d1 ratiot’1
(2.69) (3.96) (4.02) (3.14)
d5/d1 ratiot 0.07 0.27 0.30 0.04
(0.18) (0.37) (0.40) (0.05)
’0.00 ’0.00 ’0.01
Unemploymentt’1 0.00
(0.21) (’0.47) (’0.27) (’0.70)
’0.06——— ’0.12——— ’0.12——— ’0.12———
(’5.47) (’6.32) (’6.15) (’6.03)
0.03——— 0.08—— 0.08—— 0.11——
GDP per capitat’1
(2.93) (2.29) (2.14) (2.44)
0.13——— 0.32——— 0.33——— 0.34———
GDP per capitat
(2.79) (3.59) (3.59) (3.75)
’0.01 ’0.00 ’0.00
LDC tradet’1 0.00
(’1.20) (’0.29) (’0.21) (0.12)
LDC tradet 0.01 0.01 0.00
(’2.53) (0.58) (0.73) (0.26)
Female LF participationt’1 0.00 0.01 0.00
(1.97) (0.81) (0.87) (0.45)
0.04——— 0.04—— 0.04——
Female LF participationt 0.01
(0.76) (2.61) (2.35) (2.28)
Taxationt’1 “ “ 0.00
(’0.03) (0.25)
taxationt’1 “ “ 0.02 0.02
(0.68) (0.94)
Public employmentt’1 “ “ “
public employmentt “ “ “
’0.08 ’0.03 ’0.03 ’0.05
Lagged dependent level
(’3.33) (’1.12) (’1.12) (’1.50)
Adjusted R-squared .61 .69 .70 .70
Number of observations 202 202 202 202
Signi¬cance levels: ——— < .01; —— < .05; — < .10 (t-scores in parentheses).
a All regressions were estimated with a full set of country dummies using panel corrected

standard errors.

New Tradeoffs, New Policies

government employment) after 5 years. That is about twice the effect that
we found for the industry dispersion variable. We also note that the results
do not change as much as we go from a model that excludes government
employment to one that includes this variable. Moreover, there is now a
statistically signi¬cant effect of dispersion on business service employment,
although the effect is less than half of the effect on consumer services.
There is, however, a problem with these results insofar as the long-
run predictions are very uncertain as suggested by the small and imprecise
parameter estimate for the lagged level variable. It turns out that this prob-
lem is mainly caused by Canada, which has a much more dispersed wage
structure when measured by d5/d1 ratios than when measured by the in-
dustry dispersion variable. In fact, Canada has by far the highest d5/d1
ratios among all countries in the sample, which seems too high compared
to the independent dispersion estimates by Card and Freeman (1993 Chap-
ter 1). If we remove Canada from the regression, the effects of dispersion
are somewhat reduced, and the precision with which we can estimate the
long-run effects is greatly improved. In this regression, the predicted long-
term effect of a 1-standard-deviation increase in dispersion is 3.4 percent
of the working-age population, which is almost exactly in the middle of the
1.6“4.4 percent range of estimates for the industry dispersion variable. The
3.6 percent corresponds to two thirds of standard deviation in employment.
Turning to the controls, the tax variable again has no discernable effect
on employment, and LDC trade no longer registers an impact. The most
notable difference in the results is the reduction in the long-term effect of
unemployment. This, however, may be an artifact of differences in measure-
ment. Because the d5/d1 ratios rely on wage data for employed workers,
and because unemployment is concentrated among low-paid workers, when
unemployment goes up, wage dispersion declines, and dispersion is associ-
ated with lower employment. Some of the negative effect of unemployment
therefore goes through the dispersion measure, for purely accounting rea-
sons. The same is not true for the industry dispersion measure because
although unemployment is disproportionately drawn from low-wage sec-
tors, this does not change the weight of these sectors in the calculation of

21 To a lesser extent, the differences may also re¬‚ect that the period coverage in the d5/d1
regression is tilted toward the 1980s and 1990s where macroeconomic autonomy was ar-
guably lower in many countries than during the 1970s (see Iversen and Soskice 1999).
Reestimating the model for industry earnings on the restricted data available for d5/d1

Forces of Change

Taken as a whole, the price and employment evidence is very clear. Dur-
ing the transition toward a more service-based economy, wage compression
raises the relative prices of services, as re¬‚ected in long-term real exchange
rates, and puts a damper on employment creation in low-wage and labor-
intensive service industries where productivity lags that of manufacturing.
The emerging tradeoff between wage equality and employment is central to
understanding policy choices in the 1980s and 1990s because it jeopardized
the virtuous circle between wage solidarism, wage restraint, and rising pri-
vate sector employment that had existed during the golden age (as spelled
out in Chapter 2).

6.2. Political Responses: The Service Economy Trilemma
In assessing the short- and longer-term responses by governments to the
equality-employment tradeoff, it is useful to employ Peter Hall™s distinction
between changes in the levels of the settings of existing policy instruments
and the adoption of new instruments (Hall 1993). In the short to medium
term, politicians are constrained by the policy instruments that are readily
available and tend to fall back on inherited ideological commitments to
make choices over the tradeoffs they face. In the longer run the adoption of
new instruments, and in some cases a rede¬nition of goals, may transform
the original tradeoffs. In this section, I focus on short- to medium-term
responses. The next will consider longer term adaptation.
As Anne Wren and I have argued (Iversen and Wren 1998), govern-
ments had available to them three immediate responses to the equality-
employment tradeoff. One strategy was to deregulate labor markets to
reduce the power of unions and increase wage ¬‚exibility. This, in a nut-
shell, is the strategy recommended by the OECD in their 1994 Jobs Study
(OECD 1994b). It is exempli¬ed by the neoliberal policies of governments
in the United States, Britain, New Zealand, and, to a lesser extent, Australia
during the 1980s. An alternative strategy was to accept sluggish employ-
ment growth in private services but simultaneously pursue an aggressive
employment strategy through expansion of public sector services. This,
by and large, was the option chosen by social democratic governments in
Scandinavia, who essentially ¬nanced higher prices on a range of social

ratios, however, indicates that this is a minor cause. Thus, the parameter for the unemploy-
ment variable only declines marginally. Other parameters are also fairly stable, although
the one for the lagged dispersion variable increases from 2.2 to 3.1.

New Tradeoffs, New Policies

services by increasing taxes.22 The ¬nal option was to accept the employ-
ment consequences of a compressed wage structure but to seek to limit the
disruptive effects by discouraging the labor market entry of women and by
facilitating labor market exit, primarily among older workers through early
retirement.23 This is the pattern we ¬nd in some continental European
countries, particularly those that have historically been strongly in¬‚uenced
by Christian democracy.
The policy choices can be represented as a trilemma (Iversen and Wren
1998; Wren 2000) where wage equality and private service employment rep-
resent two of the policy goals, while budgetary or ¬scal restraint represents
a third. By ¬scal restraint, I have in mind whether or not the government re-
sponds to nonmarket-clearing wages by expanding public consumption and
employment. The use of the macroeconomic concept of ¬scal restraint is
intentional because it is in fact closely related to the notion of a nonaccom-
modating ¬scal policy.24 Because ¬scal policies are constrained by interna-
tional and institutional constraints, the relationship between government
consumption and ¬scal policy matters.
The trilemma is illustrated in Figure 6.4 which links the three economic
goals of equality, employment and ¬scal restraint to three distinct partisan
choices. The basic idea is that it is dif¬cult to successfully pursue all three
goals simultaneously as long as there is a tradeoff between wage equality
and employment. Governments therefore tend to compromise the goal that
is least dear to them in order to maximize the others.
For social democrats, this typically meant abandoning ¬scal restraint
and replacing private markets with publicly provided social services. Chris-
tian democrats shied away from such a solution because they feared that
it would strengthen the state at the expense of private service providers
and the family. In the Catholic countries of Europe, social and educational
services have historically been provided to a considerable extent by the
church or inside the family, and Christian democrats did not want to see
these functions taken over by the state. Instead, the supply of labor was

22 Because public services are very labor intensive, it is easy to show that there is a tight
relationship between public employment and public consumption spending. Just how tight
depends on the ability of governments to hold back average wage costs for public sector
23 A range of disability pensions that effectively removed marginalized workers from the labor
market has also been used, especially in Italy and The Netherlands.
24 A number of people have commented that ¬scal restraint here has nothing to do with the
macroeconomic concept, which is simply incorrect.

Forces of Change

Figure 6.4 The service economy trilemma.

kept down by tax incentives that dissuaded women from entering the la-
bor market and by early retirement to facilitate the exit of older workers.
For liberal and conservative parties of the free market right, by contrast,
the problem was rather seen as one of market clearing. Wage equality and
social protection had gone too far, preventing ¬rms and workers from tak-
ing advantage of opportunities that the market would naturally offer. The
solution was deregulation and measures that would reduce the monopoly
power of unions. Equality had to give way for what neoliberals believed was
greater ef¬ciency.25
Although this is obviously a simpli¬cation, to a considerable extent the
initial adjustment of countries can be understood as the interaction of
partisan politics and the structural constraints de¬ned by the trilemma.
But choices are also subject to institutional and international constraints

25 Anne Wren (2000) has conducted a much more nuanced analysis of the role of partisanship
and traces the differences between the three party families to ideological cleavages over
the role of the family and women, the clash during the nineteenth century between the
church and the rising secular state, and the historical relationship between parties, unions,
employer associations, and the church.

New Tradeoffs, New Policies

(as indicated in Figure 6.4) and re¬‚ect, at least to some extent, underlying
class alliances. The historical and institutional factors that led to the divi-
sion of countries into liberal and coordinated market economies now shape
the costs and bene¬ts of particular courses of action (Wren 2001). Thus,
in the speci¬c skills countries of northern Europe, the traditionally domi-
nant employer associations and unions in the export-oriented sector have
opposed radical deregulation of labor markets because such reforms would
undermine the incentives of ¬rms and workers to invest in speci¬c skills and
possibly jeopardize the cooperative relationships between ¬rms and skilled
workers (Wood 1997; Manow 2002, Chapter 6). Though highly critical of
interoccupational wage compression and inef¬ciencies in the public sector,
¬rms and skilled workers recognized an important insurance function for
social protection. This function has, to a considerable extent, been per-
formed through the centralized collective bargaining system, over which
the government has limited control (Wren 2000).
There are some salient differences, however, in the bargaining system
between Scandinavia and continental Europe. In Scandinavia, the impor-
tance of national peak-level bargaining in densely unionized labor markets
gave low-paid workers a degree of in¬‚uence over wage-setting that they
lacked in the rest of Europe (Iversen 1999). Also, because Scandinavian
unions, unlike unions in countries like Austria and Germany, were sharply
divided between blue- and white-collar workers, the in¬‚uence of profes-
sionals in the main bargaining area was limited. These differences go some
ways in explaining the exceptional egalitarianism of the Scandinavian la-
bor markets, although we need to take account of government ideology to
explain the starkly different labor-market positions of women, especially
why the public service sector expanded to the degree it did in Scandinavia
(Huber and Stephen 2000).
In addition to affecting the wage structure, the bargaining system also
affects employment by inducing more or less overall wage restraint. This is
also important in the public sector because it affects the tradeoff between
employment and ¬scal restraint. Where governments are better able to
control the wage bill in the public sector, the costs of expanding public sector
employment will be lower. Correspondingly, we ¬nd that the correlation
between increases in public sector employment and increases in the total
wage bill is less than perfect (around .8), presumably re¬‚ecting differences
in the labor cost of public sector employees.
In general skills countries such as Britain, New Zealand, and the United
States, the political support for social spending is more tenuous because
Forces of Change

it rarely enhances economic ef¬ciency and because it is not usually sup-
ported by strong organized interests. Support for employment and income
protection does not extend very far into the middle classes because pro-
tection is less important to general skills workers and because such protec-
tion undermines wage and employment ¬‚exibility, which is in high demand
among employers. Unions and collective bargaining also tend to be less well
developed. In electoral politics, majoritarian electoral systems and leader-
dominated parties also provide a weak bulwark against cutbacks, as argued
in detail in Chapter 4. Compared to speci¬c skills countries, where skilled
unions in manufacturing have always played a pivotal role, policies and la-
bor market conditions in general skills countries tend to be much more
in¬‚uenced by the preferences of professionals.
Another important institutional variable is the degree of central bank
independence. Because expansion of public spending in response to pri-
vate sector employment problems is an accommodating ¬scal policy, in-
dependent central banks keen to control prices will have an incentive to
meet such policies with monetary contraction (Scharpf 1991; Manow 2002,
Chapter 5). This raises the costs of ¬scal accommodation, signi¬ed by a
rising real interest rate, and reduces the incentives of governments to pur-
sue accommodating policies. Monetary integration into a system of ¬xed
but adjustable exchange rates has very similar effects (Iversen and Soskice
1999), and it has signi¬cantly affected the ability of governments to respond
to employment problems with expansionary ¬scal policies (Manow 2002,
Chapter 5).
Finally, it is necessary to recognize that macroeconomic conditions are
in some measure exogenous to the argument that has been made in this
chapter. Nothing prevents countries from running current account sur-
pluses, even at high levels of unemployment. As shown by Carlin and Sos-
kice (2000), when countries are specialized in highly differentiated product
markets (“Ricardian trade”), it is possible for governments to have some
control over domestic levels of employment. As we saw in the statistical
analysis in the previous section, this can have signi¬cant effects on employ-
ment performance, and needs to be taken into account as we examine the
experience of particular countries.

6.3. National Variations
As in Iversen and Wren (1998), the differences in national paths of adjust-
ments can be readily illustrated with some comparative data (see Table 6.8).
Table 6.8. Wage Equality, Service Employment, and Institutions in Eight OECD Countries, 1970“1996 (percentage of adult population)

Denmark Sweden Britain United States Netherlands Germany
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
1970“73 “ 14 16 “ 17 17 .29 13 17 .27 10 22 “ 7 17 “ 8 13
1974“77 “ 17 16 .45 20 18 .33 15 18 .27 10 24 .39 7 18 “ 9 14
1978“81 .46 20 16 .49 24 18 .34 15 19 .26 10 26 .39 7 18 “ 10 15
1982“85 .46 22 16 .50 26 18 .32 14 20 .25 9 28 .41 7 18 .34 10 15
1986“89 .46 22 18 .48 26 20 .30 14 23 .24 10 31 .39 7 19 .36 10 16
1990“93 .47 22 17 .48 25 20 .29 13 26 .23 10 33 .38 7 21 .37 9 16
1994“96 “ 22 18 .45 “ “ .29 “ “ .22 10 36 .37 6 22 .35 “ “
Change .01 8 1 0 8 3 0 0 9 0 14 5 .01 1 3
’.05 ’.02 ’1
El. sys. PR PR Maj Maj PR PR
CBIa 0.40 0.22 0.21 0.72 0.45 0.99
Centralizedb 0.46 0.48 0.16 0.07 0.40 0.33
Social democratsc 0.43 0.64 0.23 0.00 0.21 0.43
Christian democrats 0.02 0.02 0.00 0.00 0.56 0.41
Liberal 0.33 0.11 0.77 0.76 0.19 0.00
Notes on columns: Column 1: d9/d1 earnings ratios (Source: OECD, Electronic Data Base on Wage dispersion, undated); column 2: public sector
employment as a percentage of working-age population (Sources: Cusack 1991; OECD, Labour Force Statistics, various years); column 3: private
service employment as a percentage of working-age population (Sources: Cusack 1991; OECD, Labour Force Statistics, various years).
a Average (after standardization) of three widely used central bank independence indexes by Cukierman (1992), Bade and Parkin (1982), and

Grilli, Masciandaro, and Tabellini (1991).
b Index of centralization of wage bargaining (Source: Iversen 1999).
c The percentage of total government seats held by (i) social democratic parties, (ii) Christian democratic parties, and (iii) Liberal-conservative

secular parties (Liberal) (Source: Huber et al. 1997).

Forces of Change

For each of the six countries in the top part of the table, there are three
columns of data: one for wage equality, one for public sector employment,
and one for private service sector employment. It is immediately obvi-
ous that all three variables differ sharply among countries. The social-
democratic-dominated Scandinavian countries have the most egalitarian
wage structures, but private services employment has been stagnant in
these countries. Public sector employment, by contrast, rose rapidly during
the 1970s and early 1980s, causing a substantial increase in public spend-
ing (more than 70 percent of all government consumption goes to paying
Compare this to that of the United States and Britain where neolib-
eral policies of privatization and deregulation dominated during the 1980s.
Employment in private services grew rapidly in these countries, while pub-
lic sector employment and consumption remained largely unchanged. Not
surprisingly, these are also the countries in the table with the most ine-
galitarian wage structures, and they are the only countries that display an
increase in inequality during the 1980s.
Finally, in two countries where Christian democratic parties have been
dominant, Germany and The Netherlands, service employment has been
stagnant in both the public sector (as in the neoliberal countries) and in
the private sector (as in the social democratic countries). These countries
have instead performed well in terms of holding back public consumption
and taxes (“budgetary restraint”) and maintaining a relatively egalitarian
earnings structure. In looking at changes over time, it should be noted that
private service employment in The Netherlands may suggest an important
change in direction during the 1990s. I will discuss this change in the next
The bottom half of Table 6.8 compares the six countries on a set of insti-
tutional and partisan variables that are relevant in distinguishing the three
clusters. Sweden, the United States, and Germany come closest to each of
the three ideal types. Sweden with centralized wage bargaining, a dependent
central bank, PR electoral institutions, and social democratic dominance in
government stands in sharp contrast to the United States with decentralized
bargaining, a relatively independent central bank, a majoritarian electoral
system, and right dominance of government. Germany is similar to Sweden
in terms of having a relatively centralized bargaining system and PR, but
the central bank in Germany, before European Monetary Union (EMU),
was much more independent and in fact exceeded the U.S. Federal Re-
serve in this respect. Finally, Germany differs from both Sweden and the
New Tradeoffs, New Policies

United States by having a much stronger Christian democratic party with
a dominant position in government in much of the period since 1982.
Based on this comparison, it is clear that the 1970s and 1980s were
characterized by politically distinct responses to the trilemma. Continental
European countries used labor market exit, and restrictions on entry, to
cope with poor employment performance; liberal Anglo-Saxon countries
cut social bene¬ts and deregulated labor and product markets with little
regard for the inequalizing consequences; while social democratic countries
in Scandinavia vastly expanded their public service sectors, and paid for it
through higher taxes.
But partisan strategies were constrained by new economic and political
realities. Restricting labor force participation limited the tax base for the
welfare state and ran counter to the desire of many women to acquire ¬nan-
cial independence from men; deregulation and growing inequality created
growing social problems of crime, inner city decay, and a growing pool of
working poor; further public sector expansion ran counter to the need for
¬scal discipline in an integrating international monetary system and pitted
private and public sector interests against each other. During the 1990s,
governments have therefore striven to relax the tradeoffs between differ-
ent goals and permit the pursuit of employment without jeopardizing ¬scal
discipline or exacerbating economic inequalities. In many countries, these
policy experiments came with the return of a reformed left to government
power in the 1990s. In the next section, I discuss the nature and effectiveness
of these experiments.

6.4. Pathways Out of the Trilemma

6.4.1. New Social Liberalism: Workfare
When Tony Blair won the British parliamentary election in a landslide in
1997, it was on a very different political platform than advocated by the
Labour Party in the past. Gone was the traditional emphasis on Keyne-
sian full-employment policies, nationalization of industry, and negotiated
incomes policies. In their place, the new government pledged commitment
to ¬scal (and monetary) conservatism, competitive product and labor mar-
kets, public investment in education, and work-based welfare.
The most novel aspect of Labour™s strategy is the attempt to use social
policies to stimulate private sector employment while simultaneously re-
ducing inequality. Young long-term unemployed must accept offers of work
Forces of Change

or education, while the state subsidizes the wage of workers if regular em-
ployment cannot be found (Undy 1999). For families, the government has
created a tax credit program, the Working Families Tax Credit (WFTC)
that guarantees a minimum income (currently £200 per week) when at least
one family member works full time. In addition, daycare subsidies have been
raised, and tax schedules lowered, for low-income workers (Glyn and Wood
2000). One study estimates that these measures have raised the income of
the lowest 20 percent of households by about 8 percent (Immervoll et al.
1999 cited in Glyn and Wood 2000).
The principle for all social policy reforms is to raise the income of the
lowest paid while preventing healthy working-age people from receiving
bene¬ts while not working. There is no attempt to alter the wage structure
itself, or to otherwise limit the incidence of low-paid work. Indeed, subsidies
to employers can be seen as expanding the range of low-paid jobs that can
pro¬tably be created. Yet, the biggest employment effect is not supposed to
come from stretching the postsubsidy wage structure but by increasing labor
market competition from those who would otherwise not search actively for
employment, thereby putting greater downward pressure on overall wages
in the economy (Glyn and Wood 2000). Lower wage pressure is a means
to reducing the equilibrium level of unemployment.
A potential medium- to long-term problem with this strategy is that
since those nonemployed who are likely to respond to the new tax and in-
work bene¬ts “ mostly long-term unemployed and people at the margin
of the labor market, especially young mothers “ have few quali¬cations,
effective wage competition will only occur in the low-pay segment of the
labor market, and therefore potentially increase pretax wage dispersion
without much of an effect on average wages.26 This in turn raises the costs
of the tax credit program and the in-work bene¬ts as more workers qualify.
If there is not political will to shoulder the ¬nancial burden, the result
could be an increase in posttax inequality over the longer run despite the
government™s assurances to the contrary.
Better access to education is supposed to deal with this problem. In-
dividual Learning Accounts (ILA) give workers a small amount of money
in an account that can be used to pay for training courses throughout a
person™s career. The scheme is supported by tax deductions for employer
and individual contributions, as well as discounts for training courses (Glyn

26 The problem is exacerbated by regionally concentrated unemployment because it increases
the supply of labor most where the demand is lowest (Glyn and Wood 2000).

New Tradeoffs, New Policies

and Wood 2000). For young people, policies are being initiated that will
increase the ¬nancial incentives to stay in education or return to the school
system in order to complete a degree. The hope is that people at the lower
end of the wage distribution will acquire a basic education as well as some
subsequent practical skills, such as computer literacy. It remains to be seen,
however, whether employers and low-wage workers will make extensive use
of these opportunities and whether, relatedly, it will signi¬cantly improve
their income. Cathie Jo Martin, in a survey of ¬fty-two large British ¬rms,
found little evidence that employers are engaging in upskilling of their low-
paid employees, and to the extent that they participate in the government™s
active labor market program, they do so as a means to gain access to cheap
labor (Martin 2003).
British social policies are in some measure built upon the American ex-
perience. As in Britain, there have been few attempts in the United States
to change the incidence of low-paid work, or to alter the pretax wage struc-
ture. Nor has there been any rise in public sector employment. But there
has been a massive reorientation of the welfare system away from “passive”
bene¬ts to bene¬ts that are closely tied to work.
Welfare reform in the United States started at the state level in the
late 1980s and early 1990s when the federal government replaced open-
ended support for poor families (Aid to Families with Dependent Chil-
dren, AFDC) with block grants ( Temporary Assistance to Needy Families,
TANF) and reduced the maximum duration of bene¬ts for families with
nonworking parents. The response by states varied, but bene¬ts, or the du-
ration of bene¬ts, for nonworking individuals were cut everywhere. Millions
of poor Americans were pushed off welfare as a consequence. At the same
time, however, these cuts were matched by massive increases in support for
low-income working families, especially through the Earned Income Tax
Credit (EITC) program. The system gives people up to 40 percent in tax
credits for every dollar up to a maximum after which the credit is grad-
ually phased out (in 1997 the maximum credit was $3,656). In addition,
Medicaid “ the medical assistance program for the poor “ was extended to
children with working parents under the poverty line, and some states also
started to subsidize child care.
In some respects, these reforms are reminiscent of nineteenth-century
liberalism in the insistence on healthy people working in order to receive
help. There is no attempt to interfere with the market mechanism of wage
setting, and the whole system is based on the assumption that there is
work for all who want to work. Compared to the British system, it also
Forces of Change

has no guaranteed income level. This said, there is little doubt about the
effectiveness of the reforms. David Ellwood (2000) has documented the
dramatic effect on the incidence of work among the poor. For example,
the percent of working single mothers increased from about 30 percent to
about 50 percent between 1992 and 1997, and as much as 80 percent of this
rise can be attributed to the combined effects of welfare reform and EITC
(see also Meyer and Rosenbaum 1999).27
Workfare reforms have some political advantages because it is generally
easier to get higher-income people to accept support for working than
nonworking healthy adults, and because businesses hiring in the low-skilled
labor market bene¬t from the increased labor supply. Tax credits rather
than outright welfare bene¬ts also shift the attention away from highly
visible spending programs to obscure technical changes in an immensely
complicated tax code (Myles and Pierson 1997). However, the reforms carry
costs and potential dangers that need to be weighted against the bene¬ts.
First, by making it very dif¬cult to be out of employment for any sustained
period of time, there is an added disincentive for workers to acquire speci¬c
skills. Such skills typically mean that workers need to spend more time
searching for jobs when they are out of work, and that is prohibitively
expensive in a welfare model where bene¬ts are tied to being in work. The
British and American reforms may therefore have little to offer the speci¬c
skills countries of continental Europe.
Second, and perhaps more worrisome for general skills countries, work-
related bene¬ts may discourage low-skilled workers from improving their
skills through additional formal education. Not only is it dif¬cult to ¬nance
such upskilling in a system that ties bene¬ts to income, but limited im-
provements in skills typically qualify workers for jobs where the income is
just high enough for work-related bene¬ts to be phased out. The tax credit
system could thus act as an indirect tax on basic education, which bodes ill
for the longer-term effects on reducing poverty.
Third, by raising the effective supply of unskilled labor, workfare may
increase pretax inequality and require more subsidies in order to prevent an
erosion of posttax inequality. Governments committed to ¬scal restraint are
not necessarily in a position to cope effectively with this second-order effect.
Tying bene¬ts so closely to work also raises some unpleasant questions about
the effects of the next sustained economic recession. Overt poverty could

27 The incentives for poor married mothers are biased against work, however, and this is
evident in a much smaller rise in employment among these women.

New Tradeoffs, New Policies

quickly return to the streets of British and American cities, and there is a
risk that the crisis may be deepened by procyclical cuts in spending as low-
wage workers simultaneously lose their jobs and bene¬ts. What appears to
be an optimal policy in a healthy economy may therefore not be optimal if
evaluated across an entire business cycle.

6.4.2. Selective and Shielded Deregulation: A Continental
European Experiment
The key dilemma for the high-protection countries of Western Europe
is that while ¬‚exibilization of labor markets may be the only route to a
sustained reversal of the deterioration of employment performance, gov-
ernments cannot signi¬cantly trim the social protection system without
triggering strong political opposition from entrenched groups, and without
endangering the complementarities between the social protection system
and the successful production system in the export-oriented sector of the
economy. In Paul Pierson™s ¬tting metaphor, the European welfare model
is an immovable object meeting an irresistible force (Pierson 1998). In the
clash of these forces, a new strategy, which I tentatively call selective and
shielded deregulation, is being attempted.
The essence of the strategy is very simple: (i) keep the full set of pro-
tections in place for workers in the internationally oriented manufactur-
ing sector, (ii) allow ¬‚exibilization of product and labor markets in the
domestically oriented service sector, while (iii) compensating low-wage
workers through tax-bene¬t subsidies and opportunities to acquire skills.
Deregulating only part of the labor market is not as impossible as it sounds
because the incidence of part-time and temporary employment is much
higher in services than in manufacturing. Deregulating the labor mar-
ket for temporary and part-time employment therefore does not neces-
sarily interfere with the protection system that underpins most employ-
ment in the international sector. Thelen and Kume (1999) have made a
related argument from the perspective of employers. In addition, changes
in the tax and social transfer system can cushion the inequalizing effects of
The political logic for selective and shielded deregulation is also quite
compelling. Private employers in low-paid services are in the best posi-
tion to take advantage of more ¬‚exible labor markets and have long fa-
vored deregulation (Schwartz 2001). Workers with a peripheral relation to
the labor market, especially among women, often prefer the ¬‚exibility of
Forces of Change

temporary or part-time jobs and see selective deregulation as a welcome
opportunity to enter the labor market. Shielding, however, may encounter
opposition among better-paid workers who must accept a shift in the tax
burden to make it possible to reduce the pretax wages of low-paid part-time
and temporary workers without altering their take-home pay. On the other
hand, there may be signi¬cant savings on bene¬ts for the long-term unem-
ployed, and presumably it will allow prices on services to fall as implied by
the analysis in Section 6.1.1.
For full-time workers, especially in manufacturing, a principal concern is
that insurance against loss of skills remains high, in particular that employ-
ment and unemployment protection is not being undermined for skilled
workers. The former entails that an existing full-time job cannot be re-
placed by temporary or part-time contracts, while the latter requires that
workers who lose their jobs are not forced to accept temporary or part-time
employment. With such guarantees, full-time workers bene¬t from more
¬‚exible markets in services including longer opening hours, a more diver-
si¬ed choice of services, and possibly lower prices. Employers and unions
in the exposed sectors have long been critical of cross-sector coordination
of wages and working conditions, especially with the public sector, pre-
cisely because such coordination raises the relative price of services and
may undermine competitiveness (Swenson 1991b; Iversen 1999). Selective
deregulation, by creating a second-tier labor market in low-end services
could address some of those problems.
The trend toward selective deregulation is picked up in OECD data
on employment protection for full-time regular employment, and for tem-
porary and part-time employment. Figure 6.5 compares the stringency of
employment protection for full-time workers in the late 1990s to the mid-
1980s. Assuming that the OECD methodology, which relies heavily on
legal regulations, accurately re¬‚ects practice, there is practically no change.
With the exception of Finland, all points lie on the 45—¦ status quo line.
Compare this pattern to changes in the stringency of regulations of tempo-
rary and part-time employment during the same period (Figure 6.6). In the
majority of high-protection countries (France is a notable exception), the
possibilities for temporary employment, which comes with no employment
protection, and part-time employment, which sometimes comes with little
protection, have been expanded. Indeed, it is reasonable here to talk about
convergence to a more deregulated model. The mean for the protection
index has dropped from 2.2 to 1.5, and the standard deviation has dropped
from 1.7 to 1.1.
New Tradeoffs, New Policies


Ger Swe
Jap Ita
Status quo line

1.5 Ire

1 Aus

0 0.5 1 1.5 2 2.5 3 3.5

Figure 6.5 Stringency of employment regulation for regular full-time employ-
ment, 1980s versus 1990s.
Source: OECD (1999b).

Figure 6.6 Stringency of employment regulation for temporary and part-time
employment, 1980s versus 1990s.
Source: OECD (1999b).

Table 6.9. Employment and Unemployment in Denmark, the Netherlands, and Europe, 1990“1999

Denmark Netherlands OECD“Europe
Employment Government Unemployment Employment Government Unemployment Employment Government
Unemployment Ratea Ratiob Consumptionc Ratea Ratiob Consumptionc Ratea Ratiob Consumptionc
1990 7.7 75.4 23.1 6.2 61.1 12.1 8.0 61.2 16.1
1995 7.3 73.9 23.4 6.9 64.2 12.3 10.5 59.6 16.5
1996 6.8 74.0 23.5 6.3 65.4 11.9 10.5 59.7 16.6
1997 5.6 75.4 23.0 5.2 67.5 11.8 10.3 59.6 16.2
1998 5.2 75.3 23.0 4.0 69.4 11.8 9.7 60.3 16.0
1999 5.2 76.5 n.a. 3.3 70.9 n.a. 9.2 61.5 n.a.
Change 1.1 9.8 1.2 0.3
’2.5 ’0.1 ’2.9 ’0.3 ’0.1
Standardized unemployment rates.
Total employment as a percentage of the working-age population.
Government consumption as a percentage of GDP.
Source: OECD (2000).
New Tradeoffs, New Policies

If we look closer at the development in particular countries, two stand
out: Denmark and The Netherlands. Under governments with social demo-
cratic participation, both have experienced a signi¬cant improvement in
employment during the 1990s (see Table 6.9) without resorting to increased
government spending and without signi¬cantly increasing the inequality
of take-home pay. Unemployment rates have also fallen signi¬cantly be-
low the European average, and The Netherlands has seen employment
as a share of the adult population rise from 61 to 71 percent during the
1990s, an impressive increase of 16 percent in the total number of employed
The Dutch efforts to deregulate temporary and part-time work during
the 1980s and early 1990s have gone further in recent years, with notable
effects on performance. Replacement rates for workers under 50 on dis-
ability pensions have been cut, while eligibility rules have been signi¬cantly
tightened, and more of the burden on ¬nancing has now been placed on
employers (Becker 2001; Green-Pedersen, van Kersbergen, and Hemerijck
2001). Single parents with children over 5 years of age lose bene¬ts if they
do not accept a part-time job, and the de¬nition of suitable jobs for unem-
ployed has been notably broadened, now including employment with lower
pay than in a worker™s previous job. As noted, the minimum wage has also
been signi¬cantly cut as a result of a more restrictive indexation (Salverda
Because most long-term unemployed and disability recipients are low
skilled, one would expect these reforms to cause considerable downward
pressure on wages at the lower end of the distribution. There is some evi-
dence to this effect. Hemerijck, Unger, and Visser report that hourly wages
for part-time employees is a modest 7 percent lower in the private sector
than for similar workers in full-time jobs (2000, p. 227), but documentation
is not very good on this issue.28 That said, industry wage dispersion data
appear to tell a rather clear story. As noted previously, there seems to be a
trend break in wage dispersion in 1983, and the coef¬cient of variation in
wages has increased by about 25 percent since then (the last data point is
1995). Because these data are based on all wages divided by the number of
“man years” (or full-time equivalents), they pick up the effects of declin-
ing hourly wages for part-time jobs as well as any other changes in pretax

28 They offer no data on temporary employees. Temporary and part-time employed are ex-
cluded from OECD™s dispersion statistics, so this source cannot be used.

Forces of Change

A closer scrutiny of the data by industry shows that relative wages decline
in all services except ¬nance (measured as the average wage in an industry
divided by the average wage for the economy), whereas wages have been
rising in all manufacturing industries (see Figure 6.7). At the same time,
nearly all gains in employment “ shown at the end of each bar as the change
in employment as a percentage of the working-age population “ have come
in services with declining relative wages. From the perspective of these

Figure 6.7 Relative wages in The Netherlands, 1980“1995.
Key: FNS: ¬nancial institutions and insurance (Isic-81/2); FOD: food, beverages, and tobacco
(Isic-31); CST: construction (Isic-5); CHE: chemicals and chemical petroleum, coal, rubber,
and plastic products (Isic-35); TEX: textile, apparel and leather industries; MEQ: fabricated
metal products, machinery, and equipment (Isic-38); RWH: wholesale trade and retail trade
(Isic-61/2); HOT: restaurants and hotels (Isic-63); SOC: community, social, and personal
services (Isic-9); RES: real estate and business services (Isic-83).
Notes: The bolded number at the base of each bar is the average wage in that industry divided
by the average wage in the economy. The italicized number at the end of each bar is the
change in employment as a percentage of the working-age population between 1980 and 1995
(in sectors where relative wages have been falling the number is below the bar).
Source: Calculated from OECD (1999a).

New Tradeoffs, New Policies

data, growth in private service employment can scarcely be characterized
as a “miracle” (cf. Visser and Hemerijck 1997). Rather, it looks suspiciously
similar to a neoliberal choice in the trilemma, and Becker (2001) suggests
that deregulation is in fact the main source of new employment in The
There is, however, an important departure from a traditional neolib-
eral strategy that recalls Scharpf™s tax wedge argument. Thus, there has
been a deliberate effort to change the tax burden to maintain the income
of low-wage workers and reduce the costs to employers of hiring low-skill
workers. Social security contributions have been lowered for all jobs paying
115 percent of the minimum wage or less. New tax reforms will also lower
value-added taxes for employers in labor-intensive services (Hemerijck et al.
2000). There is little doubt that such reforms shield some of the inequal-
izing effects of lower wages for part-time employees, but the strategy is
perhaps not as clearly distinguished from the workfare reforms in Britain
and the United States as some of its advocates would like to have it. There
is, as several scholars have noted, a trend toward mixing policies from dif-
ferent welfare models (Green-Pedersen et al. 2001; Manow 2001). Yet, even
though there is some convergence in policy reforms, most would agree that
fundamental differences remain. It is this pattern of lasting divergence and
convergence at the margin that the regime argument in a deindustrializing
world captures.
Politically, selective deregulation works up to the point where the savings
on transfers to the nonworking population are exceeded by the costs to
the pivotal voter of shifting the tax burden away from the low-skilled, low-
productivity occupations toward the rest of the economy. Beyond that point,
additional subsidization requires either a shift in political power or in the
preferences for redistribution. There is no way to know with certainty
where this point lies, but enough time has elapsed from the onset of the
employment crisis and the initiation of new policies to get some sense
of the scope for improvement in employment performance through these
“shielded” policies.
For the sake of argument, assume that reforms have not increased posttax
and posttransfer inequality. To address the question of impact, we then need
to look at whether the volume of work (as opposed to the number of people
working) has changed, and by how much. If most of the employment gains
have come through reducing hours, adopting work sharing, and splitting
full-time jobs into part-time ones, it may signify an important change in

Table 6.10. The Volume of Work in Twelve OECD Countries, 1990“1999

Social Democratic Liberal Christian Democratic

Year Hours 1990 Hours 1990 Hours 1990
1990 Finland 1,306 100 Australia 1,269 100 Germany 1,021 100
1995 1,081 83 1,266 100 992 97
1999 1,165 89 1,271 100 996 98
1990 Norway 1,047 100 New Zealand 1,225 100 Italy 902 100
1995 1,039 99 1,290 105 826 92
1999 1,088 104 1,289 105 854b 95
1990 Sweden 1,285 100 United Kingdom 1,279 100 France 993 100
1995 1,165 91 1,206 94 952 96
1999 1,176 92 1,233 96 953b 96
Netherlands 876 100
1990 Denmark 1,192 100 United States 1,403 100
889 101
1995 1,182 99 1,415 101
921c 105
1999 1,222 103 1,460 104
Average Average Average
1990 1,207 100 1,294 100 948 100
1995 1,117 93 1,294 100 915 97
1999 1,163 97 1,314 101 931 98
a Average annual hours worked per person in the working-age population.
b 1998.
c 1997.

Source: OECD (2000).
New Tradeoffs, New Policies

the distribution of work, but that would merely be a movement within the
trilemma. We thus need numbers for the volume of work, not the number
of workers.
In Table 6.10, I have used OECD™s measures of the number of hours
actually worked by employed workers, and then multiplied this number by
the employment ratio to get a measure of the volume of work available. The
number is similar to the employment ratio, except that the unit of analysis is
hours worked instead of number of people working. Table 6.10 shows the
results for The Netherlands (bold faced) as well as for eleven other OECD
countries for which more or less complete data are available (organized
into social democratic, liberal, and Christian democratic clusters). Note
that The Netherlands does indeed experience an increase in the volume
of work between 1990 and 1997, but the gain is considerably smaller than
the increase in the number of employees. During this period, the number
of employed persons increased by 16 percent, while the volume of work
increased by only 5.2 percent. About two thirds of the increase in employ-
ment is, therefore, accounted for by a greater spread in the distribution of
available work, not by an increase in the volume of work.
Whether a 5 percent increase in the volume of work during the 1990s
is a lot depends on your perspective. If one compares The Netherlands
to the other Christian democratic countries, it appears that The Nether-
lands has simply been catching up to the average level in that cluster. In
1990, the hours worked were signi¬cantly below the other countries in
the cluster, whereas in 1997 they were similar. The main reason that The
Netherlands outperforms the other countries in terms of the employment
ratio, as we saw earlier, is that the Dutch have accepted a much larger
proportion of part-time employees than other countries, either inside or
outside the Christian democratic cluster (see Table 6.11). In 1998, part-
time employment accounted for a remarkable 30 percent of total employ-
ment, which is roughly twice as high as the other countries in the cluster
(which range from 17 percent in Germany to 12 percent in Italy). It is
also signi¬cantly higher than in any other country in Table 6.11, includ-
ing the most deregulated market economies. Note also that during the
1990s the part-time workforce became increasingly feminized, with the
share of women in part-time employment rising from 70 to 76 percent.29
In other words, selective deregulation in The Netherlands has created a
dual labor market of part- and full-time employment, divided along gender

29 Data are from OECD Employment Outlook (1999b, p. 240).

Table 6.11. Part-Time Employment in Twelve OECD Countries, 1990“1999

Social Democratic Liberal Christian Democratic

Part-Time Women™s Part-Time Women™s Part-Time Women™s
Year Employment Shareb Employment Shareb Employment Shareb
1990 Finland 7.6 66.8 Australia 22.6 70.8 Germany 13.4 89.7
1998 9.7 63.1 25.9 68.8 16.6 84.1
1990 Norway 21.3 82.7 New Zealand 19.6 77.1 Italy 8.8 70.8
1998 21.0 79.1 22.8 74.3 11.8 70.4
1990 Sweden 14.5 81.1 United Kingdom 20.1 85.1 France 12.2 79.8
1998 13.5 97.3 23.0 80.4 14.8 79.3
Netherlands 28.2 70.4
1990 Denmark 19.2 71.5 United States 13.8 68.2
30.0 75.8
1998 17.0 68.5 13.4 68.0
1990 Average 15.7 75.5 Average 19.0 75.3 Average 15.7 77.7
1998 15.3 77.0 21.3 72.9 18.3 77.4
a Part-time employment as a share of total employment.
b Women™s share in part-time employment.
Source: OECD (1999b, p. 240).
New Tradeoffs, New Policies

lines.30 This may not be an unambiguously bad thing because many women
prefer to work part-time, but the low volume of work continues to be a
distinctly Christian democratic pattern.
Denmark is the country that stands out in the social democratic cluster
because it exhibits signi¬cant gains in the volume of work during the 1990s,
despite starting from a relatively high level in 1990. As in the case of The
Netherlands, these gains, which occurred under social democratic govern-
ments, have partly been achieved through a combination of more ¬‚exi-
ble work-time and opening hours, and signi¬cant tightening of eligibility
requirements for unemployment bene¬ts among long-term unemployed
and unemployed workers under 25 without a qualifying education.31 As
in Britain, a series of tax reforms has reduced income taxation, especially
for low-income people, while tax deductions, especially for home own-
ers, have been scaled back. To alleviate some of the pressures on the low-
skilled labor market, wage subsidies and upskilling among young and low-
paid workers have been promoted (“Handlingsplan for Beskaeftigelse”).
The wage subsidies to employers hiring unemployed workers amount to
about 50 percent of wages (Martin 2003), and expanded training opportu-
nities have allowed ¬rms to hire long-term unemployed while their core
workers receive training (Tor¬ng 1999). The fall in vacancy rates dur-
ing the 1990s can at least in part be seen as a result of better matching
of skills to jobs, and the reforms have met with support among employ-
ers (Martin 2003). Finally, an element of the reforms that clearly distin-
guishes the reforms under the social democrats from those in the Christian
democratic model is the signi¬cant scale-back of the early retirement
scheme, which had induced a large number of workers to leave the labor
In the Danish case, however, one also needs to control for changes in the
macroeconomic conditions. Open economies can sustain several different
equilibrium rates of unemployment at the cost of a deterioration in the
trade balance. Soskice (2000) suggests that, as a rule of thumb, a 1 percent
change in the external current account is equivalent to about a 1 percent
change in the rate of unemployment. Because Denmark went from a 0.9

30 This association between part-time employment and gender segmentation is a general
phenomenon across the OECD, as can clearly be seen by comparing across regime clusters
in Table 6.11.
31 But the improvements are clearly not coming from an expansion of part-time employment,
which is an important contrast to The Netherlands.

Forces of Change

surplus to a 1.2 percent de¬cit on the current account between 1990 and
199832 “ with most of this change resulting from the tax reforms “ some
of the 2.5 percent reduction in unemployment is accounted for by ¬scal
If we look beyond the Danish and Dutch cases, the data on the volume of
work in Table 6.10 show that there has been no overall gain in the volume
of work in Europe during the 1990s, and the employment gaps between
clusters of countries persist and have in fact widened. While bearing in
mind that the numbers are not necessarily fully comparable across coun-
tries, the variance across clusters is much greater than within clusters, and
the differences across clusters are entirely consistent with the differences
in employment ratios that were apparent by the early 1980s. Speci¬cally,
the volume of work in the liberal cluster in 1999 was 41 percent higher
than in the Christian democratic cluster and 16 percent greater than in the
social democratic cluster. These numbers were up from 37 and 7 percent,
respectively, in 1990, and 25 and 5 percent in 1979. Concerning the gap
between social democratic and Christian democratic countries (25 percent
in favor of the Nordic countries in 1999), this can largely be accounted for
by higher levels of public employment in the former countries, which is a
familiar story by now.
So while the Danish and Dutch cases point to the potential for real em-
ployment gains through selective deregulation, the big picture presented in
Table 6.10 hardly suggests that the tradeoffs captured by the trilemma are
a thing of the past. Indeed, it is hard to escape the conclusion that selective
deregulation of labor markets, as it has been practiced up to now at least,
has had little effect on the employment gap with liberal countries. In the
case of The Netherlands, at least part of the employment gain, which is
less impressive when measured in hours than in workers, has been accom-
plished at the expense of a somewhat more dispersed wage structure. On
the other hand, shielding low-wage workers through tax incentives, wage
subsidies, and training opportunities “ especially in the Danish case “ has
prevented the inequality to rise nearly as sharply as in the Anglo-Saxon

32 The trade balance went from a 5.9 percent surplus to a 2.9 percent surplus in the same
period. The ¬gures are calculated by dividing the trade balance in dollars with the GDP
in dollars. Both current account and trade balances are from OECD Employment Outlook

New Tradeoffs, New Policies

6.4.3. The Coalitional Dynamics of Labor Market Reforms
The national experiences discussed in the previous two sections suggest
systematic differences in the politics of reform in speci¬c as opposed to
general skills countries. As I have suggested, these differences re¬‚ect struc-
tural, institutional, and partisan in¬‚uences on political processes. But they
also re¬‚ect differences in the logic of political coalition building, and in this
section I brie¬‚y outline what these logics might look like, building on the
theoretical framework developed in Part II.
A useful starting point is to imagine a two-dimensional policy space: One
dimension is protection through redistributive transfers; the other is pro-
tection through labor-market regulation, especially legislated job security.
Now distinguish three economic classes: (i) low-skilled and low-income
workers, (ii) high speci¬c skill and middle-income workers, and (iii) gen-
eral skills, and high-income, professionals. These are similar to the classes
identi¬ed in Chapter 4, except that they here combine both skill speci¬city
and income (similar to the model in Chapter 3.)
In terms of preferences, low-skilled workers unambiguously bene¬t from
redistributive transfers. They may or may not bene¬t from labor-market
regulation because although such regulation can improve their security in
the labor market, they may have signi¬cant costs in terms of job opportu-
nities. The latter depends on the extent to which low-skilled workers are
complements to high-skilled workers in production. If skilled workers ben-
e¬t from regulation, as I will assume, and skilled and unskilled workers are
close complements, unskilled workers may also bene¬t from, and therefore
prefer, regulation.
Because it serves as a protection of their skill investments, speci¬c skill
workers will prefer high job security, but they may or may not bene¬t from
redistributive spending. The latter depends on the progressivity of such
spending as explained in Chapter 4. Finally, professionals are unambigu-
ously hurt by both redistributive transfers (because they are net contrib-
utors), as well as by high job security (because it restricts mobility and
Assume further that speci¬c skills countries have PR electoral systems,
whereas general skills countries have majoritarian systems. As discussed
in Chapter 4, this is a good approximation to the empirical reality. If the
former are dominated by center-left governments, it corresponds to an
alliance between low skilled and skilled workers. As long as these workers

Forces of Change

are strong complements in production, and such production is facilitated
by job protection (owing to the effects on skill investments), both will have
an interest in a highly regulated labor market. As explained in Chapter 4,
center-left coalitions can also use a combination of ¬‚at-rate and targeted
transfers to produce relatively high levels of redistribution. By contrast,
in general skills countries, the two-dimensional space essentially collapses
into a single dimension where the median income voter will tend to vote
for the center-right party.
Now introduce an economic shock that undermines the complementar-
ity of skilled and unskilled workers in production. Empirically, this corre-
sponds to the rise of low-productivity and low-skilled services, replacing
“good” manufacturing jobs. The tradeoff between employment regulation
and jobs now becomes steeper, and low-skilled workers will prefer less regu-
lation. Because professionals and skilled workers are consumers of services,
and also pay the bill for long-term unemployed workers, they both pre-
fer lower protection for unskilled workers. In general skills countries with
majoritarian electoral systems, this unambiguously leads to labor market
reforms designed to maximize the ¬‚exibility of low-skilled labor markets.
Insofar as such reforms cause signi¬cant increases in inequality, they may
be accompanied by some redistribution (e.g., negative income taxes), but
such redistribution ultimately depends on middle-class altruism.
The dynamic in special skill countries with PR is more complicated,
but distinct. On the one hand, center-left governments have an incentive
to endorse selective deregulation of low-skilled labor markets (especially
part-time and temporary employment) because the tradeoff between regu-
lation and employment is now steeper. On the other hand, unskilled work-
ers will be worse off than before the shock, and the question is therefore
whether deregulation will be accompanied by agreements to increase trans-
fers (“shielding” as I have called it).
The answer turns on whether skilled workers (M in terms of the model
in Chapter 4), will be willing to accede to more redistribution. And this,
it seems to me, depends on whether low-skilled workers (L in the model)
can credibly threaten to ally with professionals (H in the model). Because
their preferences on employment protection are similar to the preferences
of professionals, an alliance between L and M could undermine the entire
employment protection system, which would serve as a strong incentive for
skilled workers to agree to “compensating” unskilled workers when part-
time and temporary job markets are deregulated.The idea is not far-fetched.
As documented by Kitschelt (1994), many unskilled workers in Europe have
New Tradeoffs, New Policies

turned to the New Right, thereby abandoning social democracy. The in-
centive to limit such exit is an important source of support for redistribution
among skilled workers.
What this analysis provides is some theoretical pieces to the puzzle of
explaining deregulation. But it certainly does not offer a complete picture.
The reason is that the coalition game is occurring in at least four dimensions
simultaneously: Redistribution can be divided into ¬‚at-rate and targeted
transfers, and regulation can be divided into regulation for L and regula-
tion for M. Because there are also multiple players (and I have not even
considered the role of employers), this is an inherently very complicated,
and potentially intractable, game. What I have done here is to combine
some theoretical building blocks from Part II with the empirical discussion
in this chapter to suggest how the game may be played out. I suspect that
ideology plays a critical role as focal point for the outcome, but I leave this
as an open-ended question.

6.4.4. Service Trade Liberalization as a Long-Term Solution?
Adam Smith™s famous dictum that the division of labor is limited by the
size of the market is as applicable to services as it is to manufactures.33
What has become painfully clear to many European governments is that
you cannot have high social protection with a compressed wage structure
while simultaneously cultivating a vibrant private service sector unless you
are able to specialize in high value-added services through international
trade. Specialization in product markets that require an intensive use of
workers with good vocational skills raises the relative price on these skills
and makes them more attractive to invest in. But specialization requires
trade, and services have traditionally been shielded from trade.
This is one of the most important reasons for the concerted effort in
recent years by many European governments to liberalize service trade.
There are contributing factors, of course. Technological change, especially
in telecommunications and the internet, has opened up possibilities for
trade where previously they did not exist (Vogel 1996). Lower costs of
setting up foreign branches have also increased competition in domestic

33 It is interesting to note that Smith had little regard for the value of services. He claimed that
service workers were “unproductive of any value” and included in this category “churchmen,
lawyers, physicians, men of letters of all kinds, players, buffoons, musicians, and opera
singers” (quoted in Knudsen 2000).

Forces of Change

markets and provided ¬rms with stronger incentives to sell their products
in international markets. In addition, liberalization of telecommunication
and air transport in the United States has intensi¬ed competitive pressures
on European ¬rms and demonstrated the possibility of cheap, privately
provided services (Vogel 1996). The notion that many services, especially
telecommunications, were natural monopolies was simply unsustainable in
the face of technological change and the emergence of ¬erce competition
in the American market.
But as Knudsen persuasively shows, liberalization would have been very
unlikely had it not been for a reinforcing change in the policy outlook
of governments (Knudsen 2000). In the cases of the United States under
Reagan and Britain under Thatcher, change was not simply driven by ideo-
logical zeal but also by a pragmatic drive to exploit and deepen comparative
advantages in a world that was only just waking up to the realities of two
decades of postindustrialization. Opening up to international competition
came early in Britain and applied to services across the board. British ¬rms
had strong competitive positions in insurance and many ¬nancial services
owing to decades of more liberal policies, and the government aimed at cre-
ating new ones in air transport, telecommunications, and postal services.
Britain had been falling behind continental Europe in telecommunications,
and the reforms seem to have greatly increased the competitiveness of for-
mer monopolies such as British Telecom, which now command a signi¬-
cant share of the European and North American markets. Likewise, Britain
developed an early competitive advantage in the airline industry (Lehrer
It is important to point out that the change in government policy
was not limited to right-leaning governments of the neoliberal variety. In
Britain, the new Labour government under Blair came to power with a
¬rm commitment to competition and free trade in services (Knudsen 2000,
Chapter 6). Likewise, Christian democratic governments in Germany have
made the Dienstleistungswueste “ the “service desert” “ a major target of
policy reforms by inducing competition and by opening up world markets
for German telecommunication and insurance companies. The German
Social Democratic Party eventually came to the same view, and this was a
critical change since some deregulation required changes in German Basic
Law (Grundgesetz) that could only be passed with social democratic support
(Knudsen 2000, Chapter 3).
Liberalization in Germany has led to a rapid expansion in exports of
telecommunications and insurance services, with leading German ¬rms
New Tradeoffs, New Policies

establishing strong positions in integrated and customized communication
services, some branches of business consulting, and specialized business in-
surance. On the other hand, German service producers have been losing
out to British and American ¬rms in some markets such as personal insur-
ance, advertising, corporate ¬nance, and entertainment (Hall and Soskice
2001). A trade pattern that is not unlike that in manufacturing seems to be
developing here, although the data are woefully inadequate.34
Not surprisingly, the initial business preferences for and against trade
liberalization depended on actors™ perceptions of their ability to thrive in
a global marketplace. Small ¬rms in highly protected market segments
typically opposed liberalization, and so did many unions fearing rational-
izations, more job insecurity, and downward pressures on wages. Large
internationally oriented ¬rms with strong core competencies, on the other
hand, were eager to expand their markets and to assist their clients in for-
eign markets. Likewise, large corporate consumers bene¬ted from better
products and lower prices, and although mass consumer groups often wor-
ried about lack of transparency, inequality of service provision (especially
across regions), and nonuniform standards, they were often swayed by the
reality of lower prices, greater choice, and improved quality and ef¬ciency.
However, it was only after governments turned their backs on old pro-
tectionist policies that sustained coalitions of domestic interests emerged
to push for further liberalization. As noted by Vogel (1996), interest groups
that eventually bene¬ted from reforms often play no role in advocating the
policy. This shift in policy was signi¬cant because services, as noted ear-
lier, have traditionally been viewed by left-leaning governments as a buffer
against swings in the business cycle, as public goods or natural monopo-
lies, and even as a social policy instrument where gross inef¬ciencies could
be justi¬ed by the need to provide jobs or equalize access to services. But
with deindustrialization gaining steam, and more women entering the labor
market, this strategy had unacceptable ¬scal consequences. Precisely when
European monetary integration demanded ¬scal austerity, the protected
sector approach to services required more spending in order to deal with
the growing employment problem. The tension between these opposing

34 Knudsen quotes Mike Moore, Director-General of the World Trade Organization, as say-
ing that “we are still surprisingly ignorant about services trade ¬‚ows. Compare the material
published on services trade with the abundance of information available on merchandise
trade! It is certainly easier to ¬nd data on pork ¬lets in EC intervention stocks, by year
and origin (and, possibly, by name of the pig), than on the Community™s external trade in
computer or accountancy services” (Knudsen 2000, footnote 8, pp. 23“4).

Forces of Change

forces gave way to a shift in thinking toward the view that ef¬ciency and
trade are prerequisites for sustainable growth in employment. In Hall™s ter-
minology (Hall 1993), this is a potentially paradigmatic change in ideas.
Was this shift in thinking justi¬ed? That is to say, is international trade
truly a long-term avenue out of the trilemma? The answer has to be very
tentative because we have yet to see the full impact of trade liberalization,
and research in the area is very limited. Theoretically, however, the trade
liberalization strategy strikes at the two key problems that gave rise to
the trilemma: an incomplete division of labor and low productivity. In that
sense, I am inclined to view trade liberalization in services as a profound and
important step in the right direction. However, one must be cognizant of
the limitations of the strategy. Trade liberalization has occurred primarily in
insurance, telecommunications, ¬nancial services, and air transport. These
services all fall into either the Isic-4 (transportation) or the Isic-8 (business
services) industrial classes, and these constitute only about one quarter of
the growth in service employment between 1970 and 1995. Moreover, they
explain little of the variance in employment performance among countries
(especially between the United States and continental Europe). Also, ser-
vices continue to make up only a small portion of total trade. Of course,
being a very recent phenomenon, this fact could change signi¬cantly over
the next decade as the rate of growth in services trade is much faster than
in manufacturing (Paemen and Bensch 1995). Changes in the retail trade
sector by such companies as Amazon promise to add to this change, as do
business-to-business transactions and other services that can be done over
the internet.
That said, the employment heavy sectors of social and personal services
are inherently less susceptible to trade. In these areas, workfare, selective
deregulation, and continued government provision will likely play a much
more important role. Nor should it be forgotten that trade liberalization
is accompanied by signi¬cant short- and medium-term pain as greater ef-
¬ciency is accomplished primarily by laying off workers. The long-term
dynamic gains may be very signi¬cant, but in the short term there could be
signi¬cant losses. Ironically, it is precisely these losses that may be the best
bet for social democracy to remain in power.

6.5. Conclusion
While deindustrialization spurred much of the rise in the level of social
protection during the 1970s and 1980s, it also gave rise to new tradeoffs
New Tradeoffs, New Policies

and political con¬‚ict. In particular, the virtuous circle between employ-
ment and high levels of protection and wage compression became harder
to sustain. The initial political response was to either (i) limit labor market
entry (e.g., through tax disincentives for women to work) and maximize
exit (e.g., through long maternity leave and early retirement), (ii) vastly ex-
pand employment in the public sector, or (iii) deregulate the labor market
and undermine the power of the unions. These responses in many respects
deepened the division between high- and low-protection countries, but they
also sharpened the distinction between countries dominated by Christian
democratic and social democratic ideas and governments.
As the costs of each strategy became increasingly apparent, governments
embarked on reforms that sought to limit the severity of the tradeoffs of
potentially overcoming trilemma altogether. Liberal countries have sought
to restrict the opportunities to opt out of an unattractive labor market
for the low-skilled by tying bene¬ts to employment (workfare instead of
welfare), while governments have used tax subsidies to somewhat improve
the plight of the working poor. The result has been labor markets with stark
income inequalities but ample employment opportunities and a thriving
service economy. Workfare has done little to improve the skills of low-
paid workers, but the ¬‚uidity of the labor market, a good post-secondary
educational system, and the availability of low-paid social and personal
services have limited gender-based inequalities, which tend to rise with
high job protection and speci¬city of skills.
In the high-protection, speci¬c skills countries, recent efforts have con-
centrated on expanding employment in the private sector while maintain-
ing ¬scal discipline. Governments have sought to accomplish this through
some measures of deregulation and privatization of services (longer open-
ing hours, private daycare, hospitals, etc.) combined with ¬‚exibilization of
especially part-time and temporary employment. As in the liberal coun-
tries, bene¬ts have become more closely tied to work, but governments
have generally gone farther in ensuring that greater dispersion in pretax
income is offset by tax subsidies and transfers bene¬ting low-paid workers.
The generous social protection system for a core, skilled full-time labor
force has largely been retained, creating a pattern of selective and shielded
deregulation. This has preserved many of the key differences between the
high-protection European welfare states and the low-protection Anglo-
Saxon ones described in Chapter 2. In addition, although countries like
Denmark and The Netherlands share many new policies, differences in the
in¬‚uence of Christian democracy and social democracy continue to shine
Forces of Change

through, especially in terms of the position of women in the labor market
and the role of the public sector.
It is not dif¬cult to see that the character of reforms is closely tied to the
nature of politics in different countries. In the liberal cases, reforms cater
to an increasingly well-educated and professionalized middle and upper-
middle class who face little risk of poverty but bene¬t from inexpensive
services and relatively low taxes. Poverty in this system is limited by the
desire of pivotal voters to ensure that the low-skilled choose regular jobs
instead of “careers” on the streets or in crime. In high-protection coun-
tries, reforms require the consent of organized labor and business, and
coalitions that exclude the left from political in¬‚uence are more dif¬cult
to sustain in multiparty PR electoral systems. Even though skilled workers
and employers in the exposed sector bene¬t from deregulation of services
and ¬‚exibilization of low-skilled labor markets, they want to retain most
protections for full-time core workers but need to compromise with unions
representing service sector and the low-skilled workers. Again, selective
and shielded deregulation is a descriptive term that captures this political
reality, as well as a plausible representation of the underlying coalitional
The differences between market-based general skills systems and pro-
tective speci¬c skills systems are likely to endure because of (not in spite
of ) the international division of labor. Social protection is a complement
to different types of production, and specialization on international mar-
kets makes convergence unlikely. The pressure for convergence has instead
come in sectors of the economy and the labor market where trade and di-
vision of labor have traditionally been limited. This places governments
under pressure to adopt similar policies, and the rise in direct or indirect
subsidization of low-skilled labor in many countries is an expression of this
pressure. It can only be relieved in a permanent way by increasing trade
and specialization in services, but precisely because the scope for such trade
is limited, chances are that we will continue to observe tendencies toward
dualism in labor and product markets.
The resilience of existing differences between social democratic and
Christian democratic varieties of this dualism seems less certain. In par-
ticular, countries dominated in the past by Christian democracy are bound
to come under greater pressure to provide job opportunities for women,
both because many young women increasingly prefer an active life outside
the family and because paid work is a source of independence and therefore
an insurance against the adverse effects of divorce. It is not inconceivable, in
New Tradeoffs, New Policies

my view, that governments will respond by expanding public employment
as happened in the Scandinavian countries. This may even be facilitated by
the greater ¬scal autonomy afforded by EMU and the single currency. If
this does not happen, the dualistic tendencies in the private labor market
will be exacerbated, producing a large second tier of part-time and tem-
porary jobs primarily occupied by women. To some extent, as discussed
earlier, the Dutch case already exempli¬es this possibility.



. 9
( 11)