UUnionn iMoenm bMersheipmbership
Union membership statistics
in 24 countries
An analysis of "adjusted" union membership data
in 24 countries yields past and present union density rates;
the data provide explanatory factors
for the differences and trends in unionization
Jelle Visser In 1991, Monthly Labor Review published an ating the strength of trade unions. If defined andoverview of union membership statistics in 12 measured in a comparable way, it describes howcountries, presenting broad trends in union- the position of unions changes over time and dif-
ization from 1955 to 1990 and raising various fers across countries, industries or social groups.
critical issues concerning the comparability of the If large variations or swings in union density
data.1 In this article, the analysis is extended to a rates are observed, then there have been major
wider set of 24 developed countries and to recent changes in the legal-political, social, or eco-
years. Unlike the 1991 article, only “adjusted” nomic environment of labor unions. In this sense,
membership data are presented, satisfying mini- the union density statistic provides a useful com-
mum comparability criteria and used as a basis parative indicator in industrial relations research,
for calculating union density rates, defined as as was claimed by George Bain and Bob Price in
union membership as a proportion of wage and their seminal work on union growth.2
salary earners in employment. Like the previous It does not tell, of course, the whole story.
article, this one starts with a discussion of com- Other relevant indicators of “union presence”
parability issues—related to the use of sources, include the following:  bargaining coverage—
definitions, data coverage, reporting errors, spe- that is, the share of workers covered by labor
cial groups outside employment, and the selec- contracts negotiated by one or more labor
tion of the employment base for calculating den- union(s); election results of union candidates in
sity rates.  Next, the main findings for 1970, 1980, employee works councils; union representation
and 1990–2003 regarding union membership and in advisory, consulting, and legislative councils;
density are presented and evaluated. The final and the standing of labor unions and union lead-
part discusses some explanatory factors for the ers in public opinion.3 Although the union den-
differences and trends in unionization, and con- sity rate captures a major aspect of union bar-
fronts union membership statistics with data on gaining power—it is probably more difficult to
bargaining coverage, measuring the proportion of replace striking workers in the short run when
employed wage and salary earners directly cov- most of the firm’s or industry’s workers are
Jelle Visser is the ered or affected by union-negotiated collective unionized—as a full measure of “what unions
scientific director agreements. do” it is inadequate. For instance, the organiza-
at the Amsterdam tion and coordination of collective bargaining
Institute for Advanced
Labour Studies ( ), Use and comparability over employment conditions, probably the mainAIAS
University of activity of labor unions everywhere, varies a
Amsterdam, Union membership, relative to the potential of great deal even in developed economies. Esti-
Netherlands.
E-mail: those eligible to join a labor union, is the most mating the effects that labor unions have on eco-
Jelle.Visser@uva.nl commonly used “summary measure” for evalu- nomic performance and distribution of income
38     Monthly Labor Review January   2006
requires a great deal of knowledge about union structure and insurance, assistance with job search, or help with adminis-
government, bargaining practice and collective action among trative issues such as tax forms or sickness benefit claims.
employers, the aims of unions, legal rules, and public policy.4 Other unions, on the other hand, offer no tangible individual
Whereas union density is closer to measuring potential union benefits except a moral or ideological sense of belonging.
bargaining pressure, the other measures, especially bargain- Comparative research in Europe has shown that density rates
ing coverage, are closer to measuring the effectiveness of are 20 to 30 percentage points higher if unions, rather than
unions in providing and defending minimum standards of the state, assess unemployment insurance claims even where
income and employment protection in labor markets. Be- the insurance itself is fully subsidized and nonmembers have
tween the two measures there are considerable differences, legally the same entitlements as members.8  It has been noted,
as will be shown in the final section. and is shown below, that in Europe many members, after re-
In this article, great care is taken to assure minimum com- tiring from the labor market, retain their membership in the
parability of the membership data. However, even when high union, usually on the basis of very low or no financial contri-
comparability standards of counting union members are met, butions. In addition to a continued sense of belonging and the
“membership” of a labor union may not mean the same thing possibility to meet old friends and colleagues, unions may
in different countries. Obviously, membership can involve offer assistance with various administrative chores or help
variable degrees of personal commitment, sacrifice, social manage occupational and disability pension claims. The num-
pressure, and coercion, and it may come with various collec- ber of these members who are no longer “active” in the labor
tive and individual benefits. The often-cited example is market for paid work has increased in all European trade
France, where union membership is sometimes taken to mean unions, in part as a consequence of the practice of early re-
active engagement in the union as “lay representative” and tirement before the compulsory pension age of 65 or 67 years,
“militant.”  Elsewhere, but also according to the rulebooks of as well as the ageing of union membership. Self-evidently, in
French unions, membership implies no other obligation than cross-national comparisons of union density rates, members
the monthly payment of dues, usually with little effort, through without an active status in the labor market must be taken
automatic withdrawals, possibly in direct transfer (“check out.9
off”) from the wage check by the employer. Other activities,
including the willingness to support the union in industrial Comparability issues
action, are voluntary.
In the new democracies—which previously belonged to In this section, specific comparability issues are discussed—
the Communist bloc (here represented by the Czech and Slo- related to the use of sources, definitions, data coverage, re-
vak Republics, Hungary, and Poland)—membership was porting errors, special membership groups outside the labor
hardly a free choice, and it does not surprise that the high force, and the selection of the employment base for calculat-
membership numbers before 1989 proved unsustainable after ing density rates.
the transition to democracy.5 Compulsory membership upon
taking the job has been common in some occupations (artists, Sources.  As was explained in the 1991 article, union mem-
printers, dockworkers) and among manufacturing workers in bership data can be derived from two types of sources: house-
some countries, like Britain, Australia, and New Zealand. But hold surveys and administrative data obtained from the
these practices have been made illegal or unenforceable in unions.  Currently, survey data based on household surveys
the 1980s and 1990s, and in all countries in this comparison are available on an annual basis in the United States, Canada,
the freedom of association includes the right “not to join.” Australia, the United Kingdom, Sweden, Finland, and the
Similarly, labor unions vary in the services rendered to their Netherlands—and on a nonannual basis in Norway and the
members. In most countries, union-negotiated contracts are Republic of Ireland.
applied erga omnes and non-members gain the same wage In the United States, data for 1973–81 come from the May
increases, reduction in working hours, holiday entitlements, Current Population Survey, and 1983–2003 data come from
and benefits as members do. This obviously creates a consid- the CPS Outgoing Rotation Group Earnings Files of the Bu-
erable temptation to take a “free ride” as the benefits of col- reau of Labor Statistics.10 For 1984–87, 1989–90 and 1992–
lective action can be obtained without sharing in the costs.6 93 there are data on union membership based on the Cana-
For example, in the Netherlands as many as 70 percent of all dian Labor Market Activity Survey, and from 1997 Statistics
employees and more than half of all nonmembers approve of Canada included a question on union membership in the La-
unions and judge the activities of unions as “necessary” and bor Force Survey (LFS).  The first series is not strictly compa-
“beneficial.”7 Some labor unions have been effective in offer- rable, because it includes membership in all jobs whereas it is
ing “selective benefits,” for instance through unemployment common in LFS (household) surveys to consider only one
Monthly Labor Review January  2006 39
Union Membership
membership per person. As proposed by Chang and veys of trade unions in December 1991. These surveys con-
Sorrentino in their 1991 article, the series has been adjusted tinue to the current date and have been used here.18  The new
to the first-job ratio, using OECD data derived from the Cana- Employment Relations Act 2000 reinstated the obligation for
dian LFS.  In Australia, information about union membership labor unions to submit an annual return of members to the
and various characteristics of members and nonmembers Registrar of Unions, and the return to official collection of
comes from the August LFS since 1986. Similar surveys were data on union membership in New Zealand began in 2001.19
previously conducted in November 1976 and during the For all other countries, the data on union membership are
March-May 1982 period. In the United Kingdom, an annual obtained from union confederations, in some cases published
question on trade union membership was introduced into the in national statistical yearbooks (Germany, Switzerland), the
August LFS in 1989, and an annual series is available from annual report of the Chamber of Labor (Austria), or the an-
1995 (without Northern Ireland, from 1992).11 In Sweden and nual report of union research centers (Italy). In the case of
the Netherlands, the LFS includes questions about union mem- Belgium, France, and Spain, and in the four Central and East-
bership since 1988 and 1992 respectively, presented as an- ern countries, information is “pieced together” from various
nual averages.  In Finland, data on union membership can be sources, including annual reports or statements of union con-
derived from the annual Income Distribution Survey (IDS) federations, independent unions, Web sites, financial reports,
conducted by Statistics Finland since 1991.  In addition, in tax office declarations (in the case of France), and occasional
the case of Norway, special surveys on union membership surveys.20
conducted as part of the LFS are available for the second quar- While each of the aforementioned sources has its particu-
ter of 1995 and 1998.12  Based on a special module on union lar problems and errors (see below), household surveys have
membership contained in the Quarterly National Household the clear advantage of allowing individual-level analysis of
Survey of 2004 and the Labor Force Surveys of 1994–97, union membership characteristics and the calculation of de-
data released by the Central Statistical Office of the Republic tailed union density rates—for instance, by sex, race, employ-
of Ireland allows an authoritative estimate of recent trends.13 ment status, industrial branch, enterprise size, educational at-
Moreover, representative employee surveys on union mem- tainment, level of earnings, or other characteristics. Data ob-
bership and various characteristics of members and nonmem- tained from recorded administrative sources are at a more
bers outside the structure of the LFS are available in France aggregate level, and probably more vulnerable to distortion.
for 1996–2003,14 for the Republic of Ireland in 2003,15 and However, when studying membership developments in rela-
for the Netherlands in 1992–93.16 tion to union type, size, inter-union competition, the position
Membership data based on administrative sources or files of peak federations, union politics, or union ideology, one
reported by the unions come in different forms. In some coun- must rely on administrative data.
tries, the National Statistical Bureaus have conducted an an-
nual survey of union organizations and their membership be- Definitions.  What is a labor union and who counts as a union
ginning as early as the 19th century.  Such series exist or ex- member?  For comparative statistics, reasonably consistent
isted in the United States (discontinued after 1980), Canada, definitions are needed.  Following the definition of the Aus-
Australia (discontinued after 1996), Japan, Korea, Denmark, tralian Bureau of Statistics, a labor union may be defined as
Finland, the Netherlands, Norway, and Sweden. In the United “an organization, consisting predominantly of employees, the
Kingdom, Ireland, and New Zealand, the official register is or principal activities of which include the negotiation of pay
has been the basis of these statistics. The British data are avail- and conditions of employment for its members” or, slightly
able from the annual report of the Certification Officer and different, as “an organization which consists wholly or mainly
published by the Department of Trade and Industry in con- of workers ... and whose principle purposes includes the regu-
junction with the Office for National Statistics. The Irish data lation of relations between workers and employers or
are not published and come from two sources, the Registrar employer’s associations.”21  A union member is a person who
of Friendly Societies for Irish-based unions operating in the self-defines that he or she belongs to a labor union, employee
Republic and Northern Ireland, and the U.K. register for U.K.- or staff organization (in the case of household surveys), or a
based unions operating in the Republic.17  When the Employ- person who pays his or her dues and is recognized as a mem-
ment Contracts Act of 1991 ended the practice of union regis- ber by a union organization (in the case of administrative
tration in New Zealand, it not only removed the distinct legal data).
status of trade unions but it also brought to an end the official These definitions include management staff unions and
collection of data on trade union membership. In the absence professional associations, but exclude associations that do not
of official data, the Industrial Relations Center at Victoria (seek to) regulate employment relations with employers.
University of Wellington began to undertake voluntary sur- However, collective bargaining, albeit the principle method
40     Monthly Labor Review January   2006
of regulation of employment relations, is not a defining char- cluded in the aggregate statistics. In historical statistics such
acteristic.  Unions may further members’ interests through as- artifacts may misrepresent union membership growth, but in
sistance in individual bargaining, representation of members the 1970-to-present period the problem is fairly small. How-
in legal courts or consultation with employers, and through ever, across countries, coverage of the usually independent or
social and political action. unaffiliated staff and professional associations differs between
Unions are worker or employee organizations, even though a very comprehensive coverage in Scandinavia, Finland, the
some of them include members who work on their own ac- United Kingdom, Ireland, and the Netherlands, to less than
count. This is common among professional associations that complete coverage in Germany, Switzerland, Austria, Bel-
combine salaried staff and self-employed members (for ex- gium, France, Spain, and Italy. These differences in coverage,
ample, medical doctors, engineers, architects, artists, journal- however, may reflect more widespread unionization in North-
ists, and so forth).  In recent times, following the trend toward ern Europe, where managers and professional groups like art-
“market mediated employment relations,” contracting out and ists, doctors, architects, lawyers, church ministers, or soccer
freelance work—for instance, in trades like building and con- players have formed their own unions and employee associa-
struction, hairdressing, nursing, business and household ser- tions.22  Possibly, some of these groups are in miscellaneous
vices—the boundary between dependent employment and or general organizations in Austria, Germany, Belgium,
self-employment is blurring.  In many European countries, France, or Italy.
confederations have set up new sections or unions, and ad- In Germany, Belgium, and Austria the size of independent
justed their rulebooks to widen their recruitment basis to “eco- unionism, outside the main confederations, is believed to be
nomically dependent workers,” (that is, workers who are for- small or negligible.23  The size of “autonomous” unions out-
mally self-employed but usually depend on a single employer side the two (Spain), three (Italy), or five (France) main union
for their income). This phenomenon is still relatively small- confederations is significant, but reliable data are hard to come
scale but growing (see table 1). by. In Spain, the phenomenon is associated with regional in-
dependence or separatism—in France and Italy, with political
Statistical coverage.  Both sampling and nonsampling errors rivalries and loyalties. In the case of Spain and France, in
may affect statistical coverage of union membership in house- addition to data published by these organizations, the size of
hold surveys. Questions may be differently phrased and sur- these independent unions may be estimated from their share
veys may suffer from nonresponse in general or, specifically, in the vote in works council elections.24  On this basis, we
with regard to the “union question.” Due to the wording of the estimate an 18-percent membership share for independent
question, surveys, unlike administrative data, could count a unions in Spain and a 24-percent share in France. If this
member of a staff association that was not recognized, identi- method is applied to Italy, the three main confederations rep-
fied, or defined as a labor union. resent between 90 and 95 percent of all members in the coun-
The main problem of administrative data is varying statis- try.  Unfortunately, membership claims of independent unions
tical coverage:  the identification of small and unregistered in Italy are absurdly large and wholly uncontrollable. In this
unions, administrative arrears, and the misrepresentation of case, only the membership data of the three main confedera-
paying membership. The problem of varying coverage is es- tions are presented, even though this may understate the true
pecially worrying in the case of data that is only obtained from size of union membership in Italy, especially in the public
main confederations and labor unions. But even in the case of sector, by as much as 10 percentage points.25
an official registrar, some unions may have chosen not to regis-
ter or declare their membership, although this problem is prob- Reporting errors.  Union membership data are inevitably
ably negligible in the democratic countries represented here. based on self-reporting:  by individual workers or employers
In the case of U.K. and Irish registration data, the main prob- in the case of household surveys, and by union administrators
lem is that union members working outside the country are in the case of recorded data. The results may be inaccurate
also counted. If uncorrected, this leads to distorted density because of sampling and nonsampling errors; nonresponse and
statistics. Another general problem, common to all adminis- memory failure in the case of surveys; and because of out-
trative data, is that persons who are members of two unions dated record, financial interests or deliberate misrepresenta-
will be recorded twice, whereas they would be identified only tion in the case of administrative data. With computerized
once in household surveys. This problem is probably small, files, now used by most unions, the difficulty of keeping files
however, as few people will hold two (costly) memberships. up to date may have become smaller, but the problem of mis-
In the course of time, the coverage of unions and member- representation for reasons of prestige, recognition claims, or
ship by national statistical offices has broadened, and over political gain is still present, especially in countries with rival
time more professional and staff associations have been in- unionisms and without some external checking or recording
Monthly Labor Review January  2006 41
Union Membership
of files.  Thus, in France, Poland, and Hungary, estimates culating union density rates.  The average proportion of mem-
based on as many independent sources as possible have to be bers who have retired from the labor market is 17.2 percent of
provided.  In the case of the Netherlands, Italy, Switzerland, total membership, varying from 4.5 percent in Spain to as
and Spain, administrative membership data may be inaccu- much as 48.0 percent in Italy. Inflated membership figures
rate or incorporate membership in arrears with payment, but and counting nonpaying adherents as full members is a factor
they are not deliberately misrepresented. In the case of Bel- mainly in France and Belgium. Fairly large numbers of unem-
gium, each of the three union confederations tends to inflate ployed members are observed where one would expect—that
its membership statistics with the same amount, currently es- is, where unions are directly involved in the management of
timated at 13 percent.26  Another source of error consists in unemployment funds (Belgium) or provide such funds them-
double counting, the reporting of nonpaying members or “sup- selves (Denmark, Finland, and Sweden).28  Elsewhere the share
porters” outside the labor force (see table 1). of unemployed workers in union membership is very small or
negligible. The proportion of self-employed workers is also
Special groups and membership adjusted to employed wage fairly small, though rising in Finland (associated with the
and salary earners.  Historically, union movements in Eu- membership of full-time students)29 and the United Kingdom
rope, often in alliance with Social-Democratic or Christian (where self-employment in services and construction has risen
Parties, have tried to achieve “comprehensive” or “inclusive” more than elsewhere in Europe).  In Italy, where the main
representation, extending beyond wage earners in employ- confederations used to organize tenant farmers, the share of
ment. Many European unions allow or often actively seek the the self-employed in total membership has decreased. In Nor-
retention of those members who retire from the labor market way, the professional associations include a significant num-
(pensioners, early retirement, fully disabled workers), the self- ber of self-employed members.  (In Denmark, Sweden, and
employed, full-time students and apprentices, workers becom- the Netherlands, they are not included in the statistics reported
ing unemployed or first-time job-seekers, persons in volun- by the national statistical bureaus.)
tary (unpaid) work, and spouses or women’s groups.27 In the 14 countries shown in table 1, the total adjustment on
As shown in table 1, a sizeable share of the reported mem- account of these “special groups” amounts, on average, to 24.2
bership of European unions is outside the employed depen- percent—with a large variation across countries.  Taking these
dent labor force, the denominator usually applied when cal- members out from the total count, “adjusted” membership sta-
Table 1. Union membership in 14 countries, total and adjusted membership
Adjustment, of which on account of:
Country Year
In percent Nonfinancial Retired from Unemployed Self-employedof reported membership labor market and students
membership
Austria ........................................... 2002 18.2 0.0 118.2 – –
Belgium ......................................... 2002 41.7 12.9 18.2 10.6 0.2
Denmark ....................................... 2003 20.4 .0 14.2 5.9 .3
Finland .......................................... 2003 29.7 .0 11.5 8.2 1, 210.0
France ........................................... 2003 33.0 13.0 20.0 – –
Germany ....................................... 2003 19.8 .0 119.8 – .0
Ireland ........................................... 2003 8.0 – 18.0 – –
Italy ............................................... 2004 53.1 3.1 48.0 .7 1.3
Netherlands .................................. 2003 20.1 .0 119.8 – .3
Norway .......................................... 2002 26.0 .0 124.0 – 2.0
Spain ............................................. 2003 6.0 . . 4.5 1.5 –
Sweden ......................................... 2003 20.7 .0 14.7 5.6 .4
Switzerland ................................... 2001 13.0 .0 13.0 .0 .0
United Kingdom ............................ 2003 12.8 .0 110.0 – 2.8
......................................................
Average ......................................... 24.2 17.2
SOURCE: Own estimates, based on administrative data obtained from maandstatistiek March 2003, pp. 17–23, The Hague: Centraal Bureau voor
unions, following the estimation methods in Ebbinghaus and Visser 2000. de Statistiek (Statistics Netherlands).
For Finland: study of Finnish Ministry of Labour covering 89 percent of all 1 Includes unemployed and disabled workers.
unions, published in February 2003 (www.eiro.eurofound.eu.it/2003/02/
feature/fi0302204f.html). For the Netherlands: Jo van Cruchten and Rob  2 Of which, 6.1 percent are are students.
Kuipers, “Organisatiegraad van werknemers, 2001”, Sociaal-economische
42     Monthly Labor Review January   2006
tistics are obtained and can be compared with the data ob- tential membership and the base for calculating union den-
tained from surveys, usually reporting the estimated mem- sity rates, having excluding all other groups from the mem-
bership of wage and salary earners in their main job.  Differ- bership statistic.32  Table 2 presents the adjusted membership
ences between the two series may still occur on account of statistics (employed wage and salary earners only), and table
varying reporting dates throughout the year, varying defini- 3 the union density rates calculated from these adjusted sta-
tions of the dependent labor force, and the exclusion of cer- tistics. The data on civilian employment of wage and salary
tain occupations from the survey. earners is from the OECD Labour Force Statistics, published
annually by the Organisation of Economic Cooperation and
Base for union density rate statistics.  Union density ex- Development and available online.33
presses the rate of “actual” to “potential” membership, usu-
ally as a percentage.  For any one union, potential member- Comparable statistics
ship is given by eligibility criteria, usually defined in the
union rulebook or constitution. Practices vary massively This overview presents adjusted data on union membership and
across unions, occupations, industries, and countries, and union density for 1970, 1980, and 1990–2003 in 24 developed
they have changed in the course of time, usually widening economies belonging to the Organization for Economic Coop-
the definition of those eligible for membership.  In some but eration and Development (OECD).34  In addition, a series has
not all countries, the law excludes particular categories (for been calculated for the European Union defined by its size as of
instance, the military and security staff).30  Following the “eli- May 2004, before the recent enlargement with eight member
gibility” criterion would render the comparison of numbers states from the former Communist bloc and two small island
impractical, as was recognized by Chang and Sorrentino in states in the Mediterranean.35  The data and statistics presented
their 1991 article in this journal. It is for this reason, in line in table 2 (pages 43–44) are, to the largest extent possible, the
with their article and the OECD database,31 to use the size of net of total members who are unemployed, self-employed, full-
civilian wage and salary employment as the domain of po- students, pensioned or disabled, or not part of the labor market.
Table 2. Union membership in 24 countries and the European Union, adjusted data, 1970–2003, in thousands
Year United Canada Australia New Japan Republic European Germany France Italy United Ireland
States Zealand of Korea Union Kingdom
1970 ............. 118,088.6 2,211.0 92,512.7 14529.0 11,605.0 473.3 33,939.5 6,965.6 3,458.0 4,736.2 10,068.3 381.7
.....................  
1980 ............. 217,717.4 63,543.3 102,567.6 714.0 12,369.0 948.1 43,663.6 8,153.6 3,282.0 7,189.0 11,652.3 490.7
.....................  
1990 ............. 16,739.8 3,897.6 2,659.6 603.2 12,265.0 1,932.4 39,261.6 8,013.8 1,968.0 5,872.4 8,952.3 441.5
1991 ............. 16,568.4 — – 514.3 12,397.0 1,886.9 43,093.0 11,969.4 1,935.0 5,913.3 8,626.5 441.1
1992 ............. 16,390.3 3,802.8 2,508.8 428.2 12,541.0 1,803.4 41,707.8 11,083.1 1,940.0 5,906.1 8,142.9 437.9
1993 ............. 16,598.1 3,768.0 2,376.9 409.1 12,663.0 1,734.6 40,084.7 10,264.9 1,870.0 5,661.0 7,831.3 428.6
1994 ............. 16,740.3 — 2,283.4 375.9 12,699.0 1,667.4 38,742.2 9,709.5 1,800.0 5,489.5 7,450.2 432.9
1995 ............. 16,359.6 — 2,251.8 362.2 12,614.0 1,659.0 37,558.4 9,334.8 1,780.0 5,341.2 6,791.0 453.4
1996 ............. 16,269.4 — 2,194.3 339.0 12,451.0 1,614.8 36,677.7 8,826.5 1,650.0 5,266.4 6,631.0 475.0
1997 ............. 16,109.9 3,517.0 2,110.3 327.8 12,285.0 1,598.6 36,286.9 8,538.0 1,650.0 5,142.3 6,643.0 472.6
1998 ............. 16,211.4 3,553.0 2,037.5 306.7 12,093.0 1,484.2 36,335.8 8,326.9 1,650.0 5,123.4 6,640.0 491.6
1999 ............. 16,476.7 3,595.0 1,878.2 302.4 11,825.0 1,401.9 36,620.4 8,218.3 1,720.0 5,276.8 6,622.0 —
2000 ............. 16,258.2 3,740.0 1,901.8 318.5 11,539.0 1,480.7 36,640.5 8,067.0 1,780.0 5,212.2 6,636.0 —
2001 ............. 16,288.8 3,831.3 1,902.7 329.9 11,212.0 1,527.0 36,361.9 7,601.8 1,800.0 5,332.6 6,558.0 512.3
2002 ............. 15,978.7 3,923.6 1,833.7 334.8 10,801.0 1,568.7 36,261.2 7,433.9 1,840.0 5,308.5 6,577.0 519.7
2003 ............. 15,776.0 4,036.5 1,866.7 — 10,531.0 1,606.0 — 7,120.0 1,830.0 5,327.7 6,524.0 515.7
.....................
1970–1980 ... 31,034.8 1,276.2 954.9 15185.0 764.0 474.9 9,724.1 1,188.1 –176.0 2,452.8 1,584.0 109.0
1980–1990 ... 4–977.6 7354.3 1092.0 –110.8 –104.0 984.3 –4,402.1 –139.8 –1,314.0 –1,316.6 –2,700.0 –49.2
1990–2003 ... –963.8 138,9 –792,9 16–268.4 –1,734.0 –326.4 16–3,003.3 –893.8 –138.0 –544.7 –2,428.3 74.4
1970–2003 ... 5–1,940.4 8493.2 11–646.0 17–194.2 –1,074.0 1,132.7 172,321.7 154.4 –1,628.0 591.5 –3,544.3 134.2
Percent
 change ........
1970–1980 ... 35.4 57.7 112.2 1535.0 6.6 100.3 28.7 17.1 –5.1 51.8 15.7 28.6
1980–1990 ... 4–5.5 710.0 123.6 –15.5 –.8 103.8 –10.1 –1.7 –40.0 –18.3 –23.2 –10.0
1990–2003 ... –5.8 3.6 –29.8 16–44.5 –14.1 –16.9 16–7.6 –11.2 –7.0 –9.3 –27.1 16.9
1970–2003 ... 5–11.3 822.3 13–25.7 17–36.7 –9.3 239.3 176.8 2.2 –47.1 12.5 –35.2 35.2
NOTES:  11973; 21983; 31973–1981; 41983–1990; 51983–2003; 61984; 71984–1990; 81984–2003; 91976; 101982; 111976–1982; 121982–1990; 131976–2003;
141971; 151971–1980; 161990–2002; 171970–2002.
Monthly Labor Review January  2006 43
Union Membership
Where available on an annual basis, the use of survey data according to survey data, or by a little more than half a million
has been preferred;36 elsewhere, administrative data has been from 1970 to 1980 according to administrative data. Cana-
adjusted to approach as much as possible the same definitions dian unions, instead, grew spectacular in this period, by more
and coverage. Changes in sources, constituting possible breaks than 50 percent.
in the series, are underlined in table 2, but generally very small. In the 1980s, the unions gaining members were in the fol-
A major break of a substantive nature did occur in Germany in lowing countries:  Spain (where unions return to democracy
1990, following unification with former East Germany, when after the Franco era); Korea (where union organizing activi-
large numbers of “card-holding” members were added to those ties are eased); Australia; Canada; and the four Northern Eu-
of former West Germany (but soon dwindled as a consequence ropean countries—with unions in Japan, Germany, Belgium,
of transition to a free market economy). Austria, and Switzerland being relatively stable. In contrast,
U.S., French, Italian, British, Irish, and Dutch unions suffer
Union membership.  Looking at the membership statistics in large membership losses; in the European Union, half of the
table 2, membership decreased in 18 countries (and on aggre- membership gained in the preceding decade is lost. In the
gate in the European Union) since 1990, and increased in 6: 1990s, in addition to the large membership losses in the four
Canada, Ireland, Norway, the Netherlands, Belgium, and transition economies (but largely reflecting the change from
Spain. Considering each of the three decades, from the point compulsory to voluntary membership), there are very large
of view of unions and especially in Western Europe, the 1970s membership reductions in Australia, New Zealand, the United
were associated with large gains (France being the only ex- States, Japan, Germany (both West and East), Italy, Sweden,
ception, with losses setting in around 1976). Depending on Austria, and Switzerland, whereas decline seems to have “bot-
the sources used, unions in the United States increased their tomed out” in France—and unions in Ireland, Spain, the Neth-
membership by just more than 1 million from 1973 and 1981 erlands, and Belgium made significant membership gains.
 Table 2. Continued—Union membership in 24 countries and the European Union, adjusted data, 1970–2003, in thousands
    Year Finland Sweden Norway Denmark Nether- Switzer- Czech Slovaklands Belgium Spain  land Austria Hungary Republic Republic Poland
1970 ........... 828.4 2,325.2 683.2 1,107.7 1,429.9 1,230.6 — 759.8 1,355.4 — — — —
....................   
1980 ........... 1,332.2 3,038.7 937.5 1,604.5 1,517.2 1,650.5 1,030.0 852.6 1,443.5 — — — —
....................   
1990 ........... 1,526.8 3,259.9 1,033.7 1,755.5 1,347.8 1,645.6 1,193.4 820.2 1,374.6 3,000.0 3,820.0 1,920.0 66,300.0
1991 ........... 1,510.2 3,198.0 1,022.5 1,762.7 1,381.1 1,657.8 1,424.1 821.0 1,364.5 — — — —
1992 ........... 1,451.0 3,146.3 1,022.6 1,762.5 1,459.0 1,651.4 1,545.4 823.1 1,359.8 — — — —
1993 ........... 1,396.1 2,965.4 1,023.5 1,757.4 1,502.0 1,649.1 1,613.9 807.2 1,343.2 — 2,680.0 — —
1994 ........... 1,376.1 2,923.2 1,042.1 1,749.3 1,491.0 1,636.1 1,586.7 802.8 1,325.1 – — — —
1995 ........... 1,419.7 2,943.1 1,061.2 1,784.6 1,536.0 1,680.7 1,517.5 789.5 1,310.5 1,860.0 2,000.0 1,150.0 3,420.0
1996 ........... 1,442.7 2,920.1 1,080.7 1,809.7 1,533.0 1,695.7 1,544.3 787.9 1,269.6 — — — —
1997 ........... 1,461.6 2,875.7 1,103.7 1,814.0 1,578.0 1,715.6 1,582.9 769.7 1,237.6 — — — —
1998 ........... 1,478.8 2,892.1 1,128.2 1,822.6 1,606.0 1,728.9 1,741.0 753.2 1,221.5 1,000.0 — — 2,700.0
1999 ........... 1,499.5 2,931.6 1,121.3 1,799.3 1,661.0 1,745.2 1,852.0 731.1 1,209.3 — — — —
2000 ........... 1,504.4 2,950.5 1,114.3 1,803.5 1,578.0 1,805.7 1,963.6 687.3 1,187.3 — — — —
2001 ........... 1,529.0 2,976.9 1,103.6 1,780.9 1,571.0 — 2,040.6 642.6 1,165.2 650.0 1,075.2 700.0 1,500.0
2002 ........... 1,513.4 2,985.1 1,114.4 — 1,578.8 1,849.8 2,117.5 — 1,151.0 — — — —
2003 ........... 1,495.0 2,984.2 1,108.7 1,710.5 1,575.2 — 2,196.8 — — — — — —
Absolute
  change
1970–1980 . 503.8 713.5 254.3 496.7 87.3 419.9 — 92.7 88.1 — — — —
1980–1990 . 194.6 221.2 96.2 151.0 –169.4 –4.9 163.4 –32.4 –68.9 — — — —
1990–2003 . –31.8 –275.7  75.0 –45.0 227.4 1204.2 1,003.4 4–177.6 1–223.6 7–1,210.0 7–924.8 7–450.0 7–1,920.0
1970–2003 . 666.6 659.0 425.5 602.8 145.3 2619.2 31,166.8 5–117.2 2– 204.4 — — — —
Percent
change
1970–1980 . 60.8 30.7 37.2 44.8 6.1 34.1 — 12.2 6.5 — — — —
1980–1990 . 14.6 7.3 10.3 9.4 –11.2 –.3 15.9 –3.8 –4.8 — — — —
1990–2003 . –2.1 –8.5  7.3 –2.6 16.9 112.4 84.1 4–21.7 1–16.3 7–65.1 7–46.2 7–39.1 7–56.1
1970–2003 . 80.5 28.3 62.3 54.4 10.2 250.3 3113.3 5–15.4 2–15.1 — — — —
NOTES: 11990–2002; 21970–2002; 31980–2003; 41990–2001; 51970–2001; 61989; 71995–2001.
44     Monthly Labor Review January   2006
Union density.  These statistics provide a much better compari- density rates in 2002 or 2003 are lower than in 1970 in all but
son when measured against the size of the employed wage and four small European economies (Finland, Sweden, Denmark,
salary earners.  Table 3 presents the union density rates. Now the and Belgium). These four happen to be the only ones in which
picture becomes more sobering for labor unions. In fact, union unions are involved in the administration and execution of un-
Table 3. Union density in 24 countries and the European Union, adjusted data, 1970–2003, in percent
Year United Canada Australia New Japan Republic European Germany France Italy United Ireland
States Zealand of Korea Union Kingdom
...........
1970 ............. 123.5 31.6 950.2 1455.2 35.1 12.6 37.8 32.0 21.7 37.0 44.8 53.2
.....................
1980 ............. 219.5 634.7 1049.5 69.1 31.1 14.7 39.7 34.9 18.3 49.6 50.7 57.1
.....................
1990 ............. 15.5 32.9 40.5 51.0 25.4 17.6 33.1 31.2 10.1 38.8 39.3 51.1
1991 ............. 15.5 — — 44.4 24.8 16.1 34.1 36.0 9.9 38.7 38.5 50.2
1992 ............. 15.1 33.1 39.6 37.1 24.5 15.1 33.4 33.9 9.9 38.9 37.2 49.8
1993 ............. 15.1 32.8 37.6 34.5 24.3 14.5 32.7 31.8 9.6 39.2 36.1 47.7
1994 ............. 14.9 — 35.0 30.2 24.3 13.4 31.7 30.4 9.2 38.7 34.2 46.2
1995 ............. 14.3 — 32.7 27.6 24.0 12.9 30.4 29.2 9.0 38.1 32.6 45.8
1996 ............. 14.0 — 31.1 24.9 23.4 12.2 29.5 27.8 8.3 37.4 31.7 45.5
1997 ............. 13.6 28.8 30.3 23.6 22.8 11.9 28.8 27.0 8.2 36.2 30.6 43.5
1998 ............. 13.4 28.5 28.1 22.3 22.5 12.1 28.2 25.9 8.0 35.7 30.1 41.5
1999 ............. 13.4 27.9 25.7 21.9 22.2 11.1 27.8 25.6 8.1 36.1 29.8 —
2000 ............. 12.8 28.1 24.7 22.7 21.5 11.1 27.3 25.0 8.2 34.9 29.7 —
2001 ............. 12.8 28.2 24.5 22.6 20.9 11.2 26.6 23.5 8.1 34.8 29.3 36.6
2002 ............. 12.6 28.2 23.1 22.1 20.3 11.1 26.3 23.2 8.3 34.0 29.2 36.3
2003 ............. 12.4 28.4 22.9 — 19.7 11.2 — 22.6 8.3 33.7 29.3 35.3
Absolute
change .......
1970–1980 ... 3–2.5 3.3 11–.7 1513.9 –4.0 2.0 1.9 2.9 –3.4 12.6 5.9 3.9
1980–1990 ... 4–4.0 7–1.8 12–9.0 –18.1 –5.8 3.0 –6.7 –3.7 –8.1 –10.8 –11.4 –6.1
1990–2003 ... –3.1 –4.7 –17.6 –28.9 –5.6 –6.5 –6.7 –8.6 –1.9 –5.1 –10.0 –15.8
1970–2003 ... 5–11.1 8–6.5 13–27.3 16–33.1 –15.4 –1.5 17–11.5 –9.5 –13.4 –3.3 –15.5 –17.9
NOTES:  11973; 21983; 31973–1981; 41983–1990; 51983–2003; 6 1984; 71984–1990; 81984–2003; 91976; 101982; 111976–1982; 121982–1990; 131976–2003;
141971; 151971–1980; 161990–2002; 171970–2002.
Table 3. Continued—Union density in 24 countries and the European Union, adjusted data, 1970–2003, in percent
Finland Sweden Norway Denmark Nether- Belgium Spain Switzer- Austria Hungary Czech Slovak Poland
lands land Republic Republic
1970 .................. 51.3 67.7 56.8 60.3 36.5 42.1 — 28.9 62.8 — — — —
..........................
1980 .................. 69.4 78.0 58.3 78.6 34.8 54.1 12.9 31.1 56.7 — — — —
..........................
1990 .................. 72.5 80.8 58.5 75.3 24.3 53.9 12.5 24.3 46.9 — 78.8 78.7 653.1
1991 .................. 75.4 80.6 58.1 75.8 24.1 54.3 14.7 22.7 45.5 — — — —
1992 .................. 78.4 83.3 58.1 75.8 25.2 54.3 16.5 23.0 44.3 — — — —
1993 .................. 80.7 83.9 58.0 77.3 25.9 55.0 18.0 22.9 43.2 — — — —
1994 .................. 80.3 83.8 57.8 77.5 25.6 54.7 17.6 23.3 41.4 — — — —
1995 .................. 80.4 83.1 57.3 77.0 25.7 55.7 16.3 22.8 41.1 63.4 46.3 57.3 32.9
1996 .................. 80.4 82.7 56.3 77.1 25.1 55.9 16.1 22.9 40.1 — — — —
1997 .................. 79.5 82.2 55.5 75.3 25.1 56.0 15.7 22.6 38.9 — — — —
1998 .................. 78.0 81.3 55.5 75.6 24.5 55.4 16.4 21.7 38.4 32.8 — — 24.2
1999 .................. 76.3 80.6 54.5 74.1 24.6 55.1 16.2 21.0 37.4 — — — —
2000 .................. 75.0 79.1 53.7 73.3 23.1 55.6 16.1 19.4 36.5 — — —
2001 .................. 74.5 78,0 52.8 72.5 22.5 — 16.1 17.8 35.7 19.9 27.0 36.1 14.7
2002 .................. 74.8 78.0 53.0 — 22.4 55.4 16.2 — 35.4 — — — —
2003 .................. 74.1 78.0 53.3 70.4 22.3 — 16.3 — — — — — —
—
Absolute change
1970–1980 ........ 18.1 10.3 1.5 18.3 –1.7 12.0 — 2.2 –6.0 — — — —
1980–1990 ........ 2.9 2.8 .2 –3.3 –10.4 –.2 –.3 –6.8 –9.8 — — — —
1990–2003 ........ 1.6 –2.8 –5.2 –4.9 –2.0 11.4 3.7 4–6.5 1–11.5 7–43.6 7–19.3 7–21.2 7–18.2
1970–2003 ........ 22.8 10.3 –3.5 10.1 –14.2 213.3 33.4 5–11.2 2–27.3 — — — —
NOTES: 11990–2002; 21970–2002;31980–2003; 41990–2001; 51970-2001; 61989; 71995–2001.
Monthly Labor Review January  2006 45
Union Membership
employment insurance. Also, each decade became progressively such prediction, a reasonably accurate idea about what caused
worse from the perspective of union organizing (except in Spain the current decline and variation in union organizing is needed.
where the unions, after a difficult start after the fall of the Franco
dictatorship, managed to acquire organizing rights and suc- Some explanations and further data.  Explaining the varia-
ceeded to build a reasonably loyal membership base among tions and differences in union membership and density is be-
permanent workers in large firms).  Thus, even in countries in yond the scope of this article, which has its focus on evaluat-
which unions made strong membership gains in the 1990s, as ing the state of comparative statistics on the subject. How-
was the case in Ireland or the Netherlands, the rapid employ- ever, some explanations, aided by some analytical data, may
ment growth caused the union share in wage and salary employ- be mentioned here. The combination of a general downward
ment to fall. Elsewhere in Europe—for instance, in Germany, trend or a general trend reversal occurring in recent decades,
France, or Austria—union density fell in spite of extremely slow and the observation of cross-national divergence, shown by
employment growth. the data in table 3, suggests that structural, cyclical, and insti-
The density statistics in table 3 show a very large degree of tutional factors are at work.37  A common trend reversal sug-
variation—from very low rates in the United States, Korea, gests similar structural forces and economic and/or political
France, Poland, and Spain to very high rates in Finland, Swe- cycles with roughly similar timing and impacts. Persistent and
den, and Denmark, closely followed by Belgium and Nor- increasing cross-national differences are prima facia evidence
way. Union density is twice as high in the European Union as that unions and union membership must be seen in the con-
in the United States, but trends are similarly downward and text of institutions specific to national labor markets.
may be expected to converge somewhat when current mem- Table 4 presents data on union density for specific groups or
bership trends in the largest European economy (Germany) categories of employees. In the case of the United States, Canada,
and the largest of the new Member States in Central and East- Australia, the United Kingdom, Ireland, the Netherlands, Swe-
ern Europe (Poland) continue. Also, current levels of union- den, and Norway, these disaggregated statistics are derived from
ization in Australia, New Zealand, Germany, the Netherlands, surveys; for Finland, France, Spain, Austria, Germany, and Ja-
and Switzerland—with just more than one-fifth of the em- pan, they are based on adjusted administrative records.
ployed wage-earning population joining a union—tend toward One striking finding is that in a number of countries the fe-
the lower end of the spectrum. It may be that union decline male unionization rate is equal with (Canada, the United King-
has “bottomed out” in France or Britain, or that there will be dom, Ireland) or even higher (Sweden, Norway, Finland) than
reversals in union fortunes in the near future, but to make any the male unionization rate. The rapid advance of female union
Table 4.  Union density rates and bargaining coverage in 14 countries – analytical table
 Survey data                                            Administrative data
Category United
States Canada Australia
United
Kingdom Ireland
Nether- Sweden Norway Finland France Spain Austria Germany Japan
lands
  2004 2004 2004 2004 2003 2001 1997 1998 2001 2003 1997 1998 1997 2003
Total .................. 12.5 30.3 22.7 28.8 37.7 25.0 82.2 55.5 71.2 8.2 15.7 38.4 27.0 19.6
..........................
Men ................... 13.8 30.6 25.9 28.5 38.0 29.0 83.2 55.0 66.8 9.0 — 44.0 29.8 22.0
Women .............. 11.1 30.3 21.7 29.1 37.4 19.0 89.5 60.0 75.6 7.5 — 26.8 17.0 17.0
..........................
16–24 ................ 4.7 — – 9.7 27.8 11.0 45.0 25.0 453.5 — — — — —
..........................
Full-time ............ 13.9 32.0 25.0 31.5 39.6 27.0 90.0 362.0 — — — — — —
Part-time ........... 6.4 23.6 17.0 21.1 29.2 19.0 83.0 357.0 49.1 — — — — —
..........................
Standard ........... — — 136.0 29.5 40.8 26.0 — 361.0 — — — — — —
Casual ............... — — 113.8 17.2 22.1 10.0 — 335.0 — — — — — —
..........................
Private ............... 7.9 17.8 17.4 17.2 30.4 22.4 77.0 43.0 555.3 5.2 14.5 29.8 21.9 17.9
Public ................ 36.4 72.3 46.4 58.8 68.0 38.8 93.0 83.0 86.3 15.3 32.0 68.5 56.3     58.1
..........................
Manufacturing ... 12.9 30.5 235.0 24.6 40.0 28.0 95.0 54.0 683.8 67.5 24.0 57.0 45.0    27.0
..........................
Coverage .......... 13.8 32.4 50.0 35.0 — 82.0 92.0 77.0 95.0 95.0 81.0 99.0 63.0    23.5
NOTES: 11997; 22002; 3 1994; 416-29 years; 5private services only; 6including mining and construction.
46     Monthly Labor Review January   2006
membership and density, combined with the fall in male union- especially where there is pattern bargaining or if wage setting is
ization, is probably the “biggest and most profound transforma- coordinated nationally across industries (as is still the case in
tion in union membership”38 and not only in Canada.  It prob- many if not most European economies, with the exception of
ably reflects the greater attachment of women to the market for the United Kingdom and all but few of the new member states
paid labor, as shown in rising participation rates and longer ten- of the European Union).39
ure; the higher female share in public services (in Europe); and These differences are reflected in the coverage rates—that
the adoption of equal opportunity policies. In the German-speak- is, the share of employed wage and salary earners whose terms
ing countries and the Netherlands, female membership is still of employment are affected by collective agreements negoti-
relatively low, though it is rising there as well. One factor is the ated between unions and employers.  Bargaining coverage is
rise of part-time jobs, mostly held by women. Interestingly, the only slightly above union membership in the United States,
gap in unionization between part-time and full-time employees Canada, or—with a wider margin—the United Kingdom. This
is narrowing in some countries in Northern Europe—most reflects the fact that bargaining is mostly organized on a de-
strongly in those wherein a part-time job is both widely diffused centralized basis, as company bargaining. The union-negoti-
and “normalized” in the sense of being covered by the same ated contract applies only to union members and some
rights, benefits, and employment conditions that apply to full- nonunionized employees in the same bargaining unit (possi-
time workers. This is increasingly the case in, for instance, Nor- bly with the right to opt out of membership).  Multi-employer
way, Sweden, and the Netherlands; whereas in the United King- bargaining and public policies extending the negotiated con-
dom or the United States, or Japan, part-time jobs are more of- tract to nonorganized firms guarantees very high coverage
ten flexible and less covered by union contracts. rates in most European countries, far in excess of union den-
A rather universal research finding is the decline of union sity rates. It is likely that such contracts are less detailed—
density among the young. This is observed even in the Scan- and that in countries such as Spain or France, with low union-
dinavian countries. Whether this represents a lower demand ization rates outside large firms and the public sector, em-
for unionization among the young, is a cohort or age effect, or ployers have much leeway to disregard the letter if not the
reflects the increased use of part-time and flexible employ- spirit of the contract. On the other hand, research in a country
ment contracts and lower pay rates for those that enter the like the Netherlands has shown that general application and
labor market is hard to say and requires further study.  The extension of contracts still have the support of a large major-
lower unionization rate among those that hold casual or tem- ity of employers. These factors tend to lower the opposition of
porary jobs is also a general finding across countries and may employers against unions, as all share the same costs inflicted
reflect the greater difficulty of union organizing (“union sup- by unions (as well as benefits from union cooperation).40
ply”) and/or a lower attachment to the labor market, and pos- In conclusion, it can be argued that sharper international com-
sibly a lower “demand” for union representation. petition (“globalization”), the rise of service employment, slower
The decline in unionization is concentrated very strongly in growth—or even decline of government employment
the market or private sector of the economy, with rates of union- (“privatization”), much higher (long-term) unemployment rates
ization in the public or government sector remaining very high (especially in Europe), the increased use of flexible employ-
in most countries. Depending on the size of the public sector— ment contracts, also the lower inflation rates and the control of
which is usually much larger in Europe (including the new tran- inflation by means of tighter monetary policies—have limited
sitional economies) than in, for instance, the United States— union power and union recruitment. However, these influences
this has been an important resource for labor unions and federa- are mediated by labor market institutions, legal rules, and poli-
tions. Union rates in manufacturing, although often above aver- tics. Most cross-national comparative and longitudinal studies
age (and always above rates calculated for private services, with- on the subject find that such institutional factors as union-ad-
out the public sector), have decreased in many countries, in par- ministered unemployment funds, the accepted presence of
ticular the United States, the United Kingdom, Australia, Ire- unions in the workplace, coordinated nationwide bargaining, and
land, the Netherlands, France, and in recent years, Germany. consultation correlate positively with union density—because it
But unionization in manufacturing, together with public-sector provides direct incentives for membership, underpins the “so-
unions, still constitute the vertebrate of today’s labor unions and cial custom” of membership in the workplace, and lowers em-
federations in terms of bargaining power and wage setting— ployer opposition.41
Notes
1 Clara Chang and Constance Sorrentino, “Union Membership Sta- 2 George Sayers Bain and Robert Price, Profiles of Union Growth:
tistics in 12 Countries,” Monthly Labor Review, December 1991, pp. A Statistical Portrait of Eight Countries. (Oxford, Basil Blackwell,
46–53. 1980).
Monthly Labor Review January  2006 47
Union Membership
3 For Western Europe, data on these indicators have been presented 17 This data is unpublished and sorting out trade union membership
in Lars Calmfors, Alison Booth, Michael Burda, Daniele Checchi, in Irish- and British-based unions in the Republic has been a laborious
Robin Naylor, and Jelle Visser, “The Role of Collective Bargaining in task.  See Ebbinghaus and Visser, The Societies of Europe ...., chapters
Europe,” pp. 1–156 in T. Boeri, A. Brugiavini and L. Calmfors, eds.,The 9 and 17. Fortunately, since 1990 the Irish Congress of Trade Unions
Role of the Unions in the Twenty-First Century (Oxford, Oxford Uni- has published separate membership statistics for its British- and Irish-
versity Press, 2001). based affiliates operating in the Republic and Northern Ireland, cover-
4 See Robert Flanagan, “Macroeconomic Performance and Collec- ing about 97 percent of total membership in the Republic.
tive Bargaining: an international perspective,” Journal of Economic 18 Robyn May, Pat Walsh, Raymond Harbridge, and Glen Thickett,
Literature, Vol. 37, 1999, pp. 1150–75. Unions and Union Membership in New Zealand:  Annual Review for
5 See, for instance, the collection of studies in Steve Crowley and 2002, Working Paper (Wellington, New Zealand, Victoria University’s
David Ost, eds., Workers After Workers’ States (Lanham, , Rowman Industrial Relations Centre).  Albeit provided on a voluntary basis, thisMD
and Littlefield, 2001). survey appears to have a very high compliance rate.
19
6 The classical source is Mancur Olson, The Logic of Collective  See the New Zealand Department of Labour Web site, on the
Action. Public Goods and the Theory of Groups (Cambridge, , Internet at www.ers.dol.gov.nz-union-registration.MA
Harvard University Press, 1965). 20 The 2001–02 figures for the Czech and Slovak Republics, Hun-
7 According to a national survey, reported and analyzed in Bert gary, and Poland are based on the “Representativity Survey of Unions
Klandermans and Jelle Visser, De vakbeweging na de welvaartsstaat and Employers Associations” conducted by the Institut des Sciences
(Assen, van Gorcum, 1995). du Travail of the Catholique University of Louvain (Belgium) on be-half of the European Commission. Older figures are from the global
8 Bernhard Ebbinghaus and Jelle Visser, “When Institutions Matter: unionization survey by Jelle Visser on behalf of the ILO and published
Union Growth and Decline in Western Europe, 1950–1995,” European in the World Labour Report 1997–98:  Industrial Relations, Democ-
Sociological Review, Vol. 15, February 1999, pp. 1–24. See also Bertil racy and Social Stability (Geneva, International Labour Organization,
Holmlund and Per Lundborg, “Wage Bargaining, Union Membership, 1998).  For the other countries, sources are listed and discussed in
and the Organization of Unemployment Insurance,” Labour Econom- Bernhard Ebbinghaus and Jelle Visser, The Societies of Europe:  Trade
ics, Vol. 6, March 1999, pp. 397–415. Unemployment insurance funds Unions in Western Europe Since 1945 (London, Macmillan, 2000), with
administered by the unions are still common, though not exclusive, in CD-ROM database.
Belgium, Denmark, Finland, and Sweden. This explains why in these 21
countries the unemployed retain membership and union membership  Annual Report of the Certification Officer 2002–2003 for the
tends to go up in recessions, contrary to the “pro-cyclical” movement United Kingdom, page 8.
of union membership and density found in other countries. See Bruce 22 From the special survey, reported by Lipset and Katchanovski, it
Western, Between Class and Market – Post-war Unionization in the appears that in the United States, against a background of union de-
Capitalist Democracy (Princeton, NJ, Princeton University Press, cline, union density among professionals has doubled from 9 to 19 per-
1997); also, Daniele Checchi and Jelle Visser, “Pattern Persistence in cent in the four decades since the late 1950s with large advances among
European Trade Union Density – A Longitudinal Analysis 1950–1996,” teachers, nurses, physicians, psychologists, social workers, librarians,
European Sociological Review, Vol. 21, January 2005), pp. 1–22. and speech therapists. Seymour Martin Lipset and Ivan Katchanovski,
9 As proposed by Chang and Sorrentino in their 1991 article, and by “White-Collar and Professionals – their attitude and behavior towards
Jelle Visser, “Trends in Trade Union Membership,” Employment Out- unions II,” (Research Paper, George Mason University, Washington,
look 1991 (Paris, Organisation for Economic Co-operation and Devel- DC, 1999).
opment), pp. 97–134. 23 In the case of Germany, in addition to the membership statistics
10 There were no union questions in the 1982 . on the German Confederation of Trade Unions and the Civil Servant’sCPS Federation published in the Statistical Yearbook of the Federal Statisti-
11 In 2004, the series has been revised and no longer allocate people cal Office, data on the smaller organizations (a Christian union confed-
who did not report their union status on a pro-rata basis.  Thus, figures eration, a federation of manager unions, of medical, court, and military
for previous years are lower than was previously reported.  Heidi staff, and various occupational unions) have been obtained from the
Grainger and Heather Holt, “Trade Union Membership 2004,” (Lon- Institut der deutsche Wirtschaft (IW) in Düsseldorf. In the case of Swit-
don, Dept. of Trade and Industry, April 2005). zerland, in addition to data included in the Statistical Yearbook, we
12 Kristine Nergaard, Organisasjonsgraden målt gjennom AKU 2. rely on the extensive survey of organizations by Robert Fluder of the
Kvartal 1995 (Oslo, FAFO Institute for Applied Social Research, 1996); University of Zürich, reported in Bernhard Ebbinghaus and Jelle Visser,
and Kristine Nergaard, Organisasjonsgrad og tariffavtaledekning målt The Societies of Europe ..., chapter 16.  In the case of Belgium, a small
ved AKU 2. Kvartal 1998 (Oslo, FAFO Institute for Applied Social Re- federation of manager unions (with an estimated 2 percent of total mem-
search, 1999). bership) has been left out. In Austria, there seem to be no independent
13 Central Statistical Office (CSO) of the Republic of Ireland, data unions or at least no recognized ones.
from the Quarterly National Household Survey, Dublin, September 24 In fact, it is often argued that rather than the (not always verifi-
2005. able) membership claims, voting in these elections (usually with a high
14 Derived from the “Enquêtes permanentes sur les conditions de turnout) establishes the credibility and representation legitimacy of
vie de ménages,” a representative household survey conducted by the Spanish and French unions. This argument must of course be seen
French official statistical office .  See Thomas Amossé, “Mythes against the very low membership and density figures in both countries.INSEE
et réalités de la syndicalisation en France,” publication of French Min- 25 For an overview of independent unions in Italy, see Bernhard
istry of Labour, in DARES:  Premières synthèses et informations, no. Ebbinghaus and Jelle Visser, The Societies of Europe ..., chapter 10.
44.2, October 2004. 26 Patrick Pasture and Jo Mampuys, In de ban van het getal:
15 Philip O’Connell, Helen Russell, James Williams, and Sylvia Ledenanalyse van het ACV 1900–1990 (Louvain, Acco); and chapter 4
Blackwell, The Changing Workplace:  A Survey of Employees’ Views in Bernhard Ebbinghaus and Jelle Visser, The Societies of Europe ...
and Experiences (Dublin, 2004). Report published by the National 27 The latter applies to the Netherlands, but the Central Bureau of Statistics
Centre for Partnership and Performance in cooperation with the Eco- published aggregate data on union membership without such “secondary” af-
nomic and Social Research Institute (ESRI). filiations of, for instance, spouses and women outside the labor force.
16 Bert Klandermans and Jelle Visser, De vakbeweging na de 28 In each of these countries, an estimated 80 percent of the unem-
welvaartsstaat (Assen, van Gorcum, 1995). ployed are unionized, although this percentage has declined somewhat in
48     Monthly Labor Review January   2006
recent years in Denmark, Sweden, and Finland, following the availability comparison. Estimates and data on these countries can be found in the
and increased take-up of nonunion-related unemployment insurance. OECD data set on the Internet at www.oecd.org.
29 In the case of Finland, use has been made of a special survey, carried 35 The EU figures combine those of Austria, Belgium, Denmark, Finland,
out by the Ministry of Labor in 1989, 1994, and 2002, on the membership France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Por-
of students, pensioners, the self-employed, and the unemployed. tugal, Spain, Sweden, and the United Kingdom (estimates for Greece and
30 This is the case in, for example, Italy, Spain, Poland, and the Portugal, and data for Luxembourg included).
United Kingdom. 36 Unfortunately, the IDS data for Finland are not currently available in any
detailed form, and the statistics in table 2 are based on adjusted administrative data.
31 Jelle Visser, “Trends in trade union membership”; and Jelle Visser,
37
Sebastien Martin, and Peter Tergeist, “Trade Union Members and Union  Bernhard Ebbibnghaus and Jelle Visser, “When Institutions Mat-
Density” (Paris, OECD, 2004), on the Internet at www.oecd.org. ter …”; Bruce Western, Between Class and Market …
38
32 In some countries—for instance, the Netherlands, Sweden, or Nor-  Statistics Canada, “Fact Sheet on Unions,” Perspectives on
way—this means that military staff, often with extremely high union- Labour and Income, August 2004.
ization rates, are to be taken off the membership count. 39 European Commission, Industrial Relations in Europe 2004.  Di-
33 Using these data rather that the national figures may cause a small rectorate-General for Employment and Social Affairs (Luxembourg,
difference in the published figures by national sources, on account of differ- Office for Official Publications of the European Communities).
ent reference dates.  For instance, the union density figures for the United 40 Cited in European Commission, Industrial Relations in Europe
States published by BLS tend to be .1 or .2 of a percentage point higher than 2004, chapter 1.
those presented in table 3.  In the case of the United Kingdom, using the 41
OECD averages causes a drop of more than 1 percentage point in the union  Daniele Checchi and Jelle Visser. “Pattern Persistence …“;  Bruce
density figures; I have therefore decided to use the August LFS figures used Western, Between Class and Market …; see also Sven Oskarsson, “Class
by the Department of Trade and Industry. Struggle in the Wake of Globalisation – Union Organization in an Era ofEconomic Integration,” in L. Magnussenand J. Ottosson, eds., Europe: One
34 Of four OECD members (Greece, Mexico, Portugal, Turkey) we Labour Market? (Brussels, Peter Lang, 2003); and Linda Scruggs and Peter
have only rough estimates of union membership, and we have decided Lange, “Where have all members gone? Globalization, Institutions and Union
not to include the two smaller ones (Iceland and Luxembourg) in this Density,” Journal of Occupational Psychology, Vol. 64, pp. 125–53.
Monthly Labor Review January  2006 49