Você está na página 1de 16

Society for Latin American Studies (SLAS)

The Analysis of Labour Movements in Latin America: Typologies and Theories Author(s): Ian Roxborough Reviewed work(s): Source: Bulletin of Latin American Research, Vol. 1, No. 1 (Oct., 1981), pp. 81-95 Published by: Blackwell Publishing on behalf of Society for Latin American Studies (SLAS) Stable URL: http://www.jstor.org/stable/3338621 . Accessed: 05/03/2012 20:10
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Blackwell Publishing and Society for Latin American Studies (SLAS) are collaborating with JSTOR to digitize, preserve and extend access to Bulletin of Latin American Research.


The Analysis of LabourMovementsin Latin America:Typologies and theories

Departmentof Sociology, LSE Institute of LatinAmericanStudies Universityof London This article addressesthe question of whetheran adequatetheory for explaining the historical development of Latin Americanlabour movements is currently available.The importance of the question derivesfrom the fact that empirical studies of labour movements (however restrictedin time and space) must necessarily refer (even if only implicitly) to some wider context in which the monothe primaryfocus of attention is graphicstudy is situated. For most researchers, the and the delimited context case; comparisonis usually(and incor(correctly) There a need to turn our attention, taken as is, therefore, unproblematic. rectly) everynow and again,explicitly to the largerpicture. In recent yearsthere seems to have been somethingof a minorboom in studies of the working class. For the first time, an impressivequantityof monographic materialis becoming available.!One ratherironic result of this recent flurry of activity has been to highlight the discrepancybetween general theories about labour and our concrete knowledge. It is to this issue that the presentarticleis addressed. As an illustration of the present situation, it may be useful to begin with some comments on a recent English-language work on Latin Americanlabour Laborin LatinAmerica.2 Original history, HobartSpalding's Spaldingis an historianwith considerableknowledge of his area, who works within a dependency framework.He has written the first detailed treatment of Latin Americanlabour history from this perspective.This attempt to move beset by monographic yond the narrowconfines and arbitrary comparisons analysis is laudible, and there is a greatdeal that is of value in Spalding'saccount. As our present concern is with Spalding'stheoretical framework,we will pass over the detailed historiographicissues raised by his book, and immediately proceed to examine the interpretative schemawhich he uses to organizethe data. Spalding claims to detect three 'stages' in the development of the labour movement in Latin America. These he identifies as: (1) formative(2) expansive and explosive (3) co-optive-repressive.3 The first question concernsthe analytical of these what do categories: power they tell us? The answer, unfortunately,is little. Let us examinethem in more detail. remarkably The first phase simply states that things have a beginning.It is difficult to imagine not being able to talk about a formativeperiod for any phenomenon. This is not a useful conceptual (theoretical) category. It is just a statementthat

RESEARCH OF LATINAMERICAN BULLETIN 82 an historianis going to start at the beginning.The second phase is a little less will be explotautological,in that it assertsthat the periodof growth(expansion) sive ratherthan smooth and tranquil.The third period is such a catch-all that it is hard to see what explanatoryor descriptive power it has. Is there any labour movement anytherethat cannot be describedin some sense as either being reare there? In sum, it seems pressed or being co-opted? Whatother alternatives that Spaldinghas told us that labourmovements havea beginning,a middle,and a period of transitionfrom the beginningto the middle. This is true, but what does it tell us? Sincethe first task of anyanalysis of the historyof LatinAmerican movementsmust be to describethem in a meaningfulway this is not a trivial point. Spalding's categoriesaredevoidof substance,for a very good reason. That quarrelsomehistorian, A.J. Hexter, once claimed that all historians could be dividedinto 'lumpers' and 'splitters',those who sawa commonthread in apparentlydiversephenomena,and those who saw majordifferencesamong This is, of course, a dilemmawhich is intrinisic seemingly similarphenomena.4 to organizedknowledge.Put in other terms,the issue is, what amountof fuzziness around the edges of a paradigm is sufficient to warrantits abandonment? Now, in terms of Latin Americanhistory, Spaldingis a 'lumper',seeinga single commonpatternthroughoutthe continent.Onthisissue,Iam a 'splitter'. Against on a singletheme, I see rathera varietyof distincthistorical Spalding'svariations this is by no meansan asserexperiences.5 (However,as will shortlybe apparent, tion that each LatinAmerican countryis unique.) The historicalexperienceof labourin a country like Peru is quite different from Argentina; and both are profoundlydifferentfromChile,etc.; consequently the elements of commonality can only be conceptualizedat the most general level. If all LatinAmericancountriesareto be squeezedinto the Procrustean bed of a single, unitaryhistory, then the analytic categoriesmust be so broad as to be virtuallymeaningless. This is, I think, the case with Spalding's threestagesof development.The frameworkcould be fitted to virtuallyany labourmovement anywherein the world. If it 'explains'everything,then it explainsnothing.Perto cracka nut. PerhapsSpalding is a haps this seems like using a sledgehammer strawman. Perhaps. But it mustbe bornein mindthat only Spalding haspresented a fully developedtheory of the historical of labourin LatinAmerica. development The fact that Spaldingis alone in this field is an indicator of the poverty of theorizingin this area. If Spalding's threestagesof development arerejectedasan explanatory schema, what alternativesremain?Most of the alternativetheories of Latin American labourare either nondevelopmentalor country-specific. By non-developmental, I mean those theories which simply state an opposition between LatinAmerica and developed countries in terms of a static contrast.The explanationthen, is the differencebetween labourin LatinAmericaand labourin developedcountries. The natureof this differenceis conceptualizedin a varietyof ways, most It is usually in terms of the centralityof the role of the state in LatinAmerica.6 frequently assertedthat, in contrastto the liberalmodel of industrialrelations which is held to prevailin the countriesof advancedcapitalism, in LatinAmerica the state activelyintervenes in, and profoundlyshapes,labourrelations.That the liberal model is largely a myth in terms of its applicabilityto WesternEurope and the USA seemsto escapethese writers.7

ANALYSISOF LABOURMOVEMENTS IN LATINAMERICA 83 There seem to be two important objections to such a global comparison. Firstly, even if it were possibleto specify some modal patternapplicable,grosso modo, to Latin America, and another modal pattern applicable to developed countries, certain assumptionsabout the ranges of variationwould have to be made before useful comparisonscould be drawn.Clearly,the notion of a modal pattern implies a certain rangeof variationof empiricalcasesabout the mode. If the two modes are relativelyclose to each other, and the rangesof variationare large, then there will be a substantialoverlappingof cases. Conceivably,the majorityof cases could fit either pattern. Whentheorists develop modal patternsand ideal types they are, 'of course', aware of such possible objections. But we are all awareof how easy it is to reify such modal patternsand ignore the rangeof variation.This is particularlyeasy to do with that half of the comparisonwhich is not our own particular concern. In this case, it is only too easy to use a model of industrialrelationsin advanced capitalistcountrieswhich is a travestyof historicalreality. In the interestof highlightingthe role played by the state in industrialrelationsin LatinAmerica,the very important role played by the state in industrialrelationsin Britain,France, Germany,the USA, Italy, etc., etc., is practicallyignored.It may perhapsbe the case that the state intervenes in different ways in industrialrelationsin Latin America,but the contrastcannot be drawnso boldly.8 exercisesare almost invariably Secondly, these comparative non-dynamic,and do not deal with the question of change over time. In so far as developmentis treated in these comparativetypologies, it is almost invariablytreated as a unilinear progression from politicized forms of trade union bargainingtoward a liberal model. The working class becomes more 'responsible'and'incorporated' as developmentoccurs. The most debatablepoint in Spalding's argumentis his assertionthat all Latin American labour movements go through these same three phases, and for the same basic reasons. This is a familiartheme in LatinAmericanstudieswhich has been accentuatedby the generalizedimpact of 'dependencytheories'. Spalding argues that Latin American labour movements share common featureslargely because of the homogenizingimpact of international variables.However,he also arguesthat two sets of 'internal'variables (the natureof the dominantclassesand the structureof the working class) also affect the historicalpatternsof development of Latin Americanlabour movements.9This point would seem confusing since it is not clearif these two sets of factors arebroughtinto account as differences in Latin Americanlabour movements,or indeed quite what their explanatory status actually is. It seems that Spaldingis caught in a contradiction:if he wants to say that Latin Americanlabourmovementsare basicallythe same then this is best done by emphasizingdependency theory and givingexternalfactors the central explanatory role. The two sets of internalfactorswould then be relegated to a purely residual role of explaining what Spalding sees as essentially minor differencesbetween various countries. But Spaldingis not at all clearon this, and it is possibleto readwhat he is sayingas an assertionthat these internal factors (which tend to differentiate LatinAmericancountries,one fromanother) are importantexplanatoryvariables.If this is the case, then it seemsthatit would be difficult to arguethat LatinAmericanlabourmovementsexperiencea similar patternof development.



The following comments are intended to show the lack of any sufficiently for describing labourmovements.Fourvery sophisticatedconceptualframework broad dichotomiesare availableto us: (1) reformversusrevolution(a reformist labourmovementversusa revolutionary labourmovement); (2) oligarchy/bureaucracy versus democracy/spontaneity (labour movementscontrolled by conservativeleadersversuslabourmovementswhich areresponsive to the militantrank and file); (3) politicalversuseconomic ( labourmovementorientedtowardsthe state versusa labourmovementorientedtowardswage bargaining with employers); (4) co-opted versus independent(labourmovementswhich are supportive of the regimeversuslabourmovementswhich adopt a criticalstancevis-a-vis the regime). The first dichotomy (reform versus revolution) seems pretty limited for it puposes of investigation.In so far as a labourmovementis institutionalized, and critically,the status quo.10To that exmust accept, howeverprovisionally tent, it may be describedas reformist.This behaviourcan coexist with a verbal commitmentto revolutionor with a verbalstatementof belief in the legitimacy of the status quo. In neither case can we infer much about action from such statements. between The second dichotomy, which is concernedwith the relationships and the rankand file indicateswhat I see as a centralissue, but union leaderships as usuallyformulatedthe dichotomyis too crudeto be useful.Weneeda typology which is more complex than the black and white categorieswhich tend to crop on labourmovementsin LatinAmerica.1The imporup in a lot of the literature tance of this issue relatesto the predictabilityof certainkinds of institutional To the extent that a union leadershipmust be responsiveto the arrangements. wishes of the rank and file (or, alternatively,must take into account possible must reflect movementsin challengesfrom rival leaderships),wage bargaining as they affect that industry.12 the economicvariables The third dichotomy deservesa moreextended discussion.Since it was formhas enjoyed conalized by Payne in 1965, the notion of political bargaining In brief, it is arguedthat because unemploymentis so siderablepopularity.13 high (as comparedto the situationin Europeancountrieswhen they began to workerscannoteasilystrike,because they could be quicklyreplaced. industrialize) the employers,the employersare However,while the workersare weak vis-a-vis is only a smallsectorof the economy), the state(sincemanufacturing weak vis-a-vis and the state is weak vis-a-visorganizedurban opposition. This enables the in the workersto threatenthe politicalstablilityof the regimeby demonstrating on streets. The state will then attemptto resolvethe conflict by puttingpressure This the employers to settle on terms relatively favourableto the workers.14 notion has been widely accepted. But despite its intuitive appeal, the Payne model is open to challengeon a numberof grounds.The most obviouspoint is which thereby that the state may not be responsiveto threatsof urbandisorder reducesthe applicabilityof the model. The extent to which employersareweak the state will vary from situationto situation.Finally,whetherworkers vis-a-vis are weak vis-a-visemployers depends (as the model states, of course) on the labour market in that industry. When entry is restricted(either by skill or by the bargaining institutionalbarriers), powerof labouris a factorto be contended with.15These commentsare not meant to deny that the model has some utility;



but to caution againsta global contrastbetween 'political' LatinAmerican labour movementsand 'economic' Europeanlabour movements. I think the contrastis overdrawn. The validity of the Payne model is restrictedto earlyindustrialization with regimeswhich are vulnerableto urban protest. Although this may cover quite a broad spectrum of Latin Americanhistory, but it is by no means the entire picture.'6 Crucially,the Payne model underplaysthe role of corporatist labourrelationsinstitutions. The final dichotomy (independentversus co-opted) presentsa problemsince it lumps together all forms of unionism with some supportiverole with respect to the state. It could reasonablybe arguedthat the apparentlysimilarcorporatist regimes in Mexico and Brazil concealed entirely different roles for the union movements in those countries.'7 In the independent/co-opteddichotomiesthe termsof the dichotomy also conceal importantvariations. If none of these commonly availableconceptualcategoriesare reallyadequate for the task in hand,what would be a preferable approach?Althoughthis article does no more than search for an answer, one fact is clear. It is importantto get tendency automaticallyto create ideal-types,that away from the neo-Weberian is so easy to slide into. The debatesover populismare a good example of the ease with which people adopt ideal-types.'8Thereificationof the supposeddichotomy of the economy into a marginal pole and dynamic manufacturingsector is another,19 while the modal patternmodel is the best exampleof all.2 Insteadof rapidlybuildingup ideal types, or theoreticalmodels (if the notion of ideal-types is offensive), it might be more useful to proceed more cautiously via attempts to define variablesin isolation. It would then be an open question as to how the variablesfitted togetherin realityto form concretemodels. I am arguingthat we havemoved too directlyfrom empiricalreality and labelswith commoncurrency to theoreticalconstructs.All too often we have taken terms such as "tradeunion charro,pelego, Vandorista,'businessunionism', and 'reformism', bureaucracy", and more or less uncriticallyincorporatedthem into our theoreticalframework. Beforethey can be useful these sorts of concepts do requiremajorreworking. The widely used Mexican term charrismomay serve as an example (though similarcomments could be made about such terms as pelego, Vandorista, etc.). Whilethere is no universally accepted definitionof the term, Alonso's comments providea useful startingpoint: Charrismo is a particularform of tradeunion control which is characterized by: a) the use of the repressiveforces of the state to supporta tradeunion leadership; b) the systematic use of violence; c) the permanentviolation of workers'union rights; d) misuse and theft of trade union funds; e) dishonest dealingwith the workers'interests;f) connivancebetween union leadersand the governmentand capitalists; g) corruptionin all its forms.2' There are many elements in this 'definition'. Perhapsthe most importantis the penultimate criterion: anti-workingclass policies of the union leadership which providesthe evaluativeconnotation.Charrismo merelymeansunion leadership disapprovedof by the speaker. No serious attempt is made to specify in what ways behaviour is anti-workingclass, or to clarify what would constitute pro-workingclass politics (which would be, at the same time, possible). The that theory of workingclass behaviourwhich is implicitin this definitionsuggests



if the charro leadershipwere removed,the workerswould pursue 'authentic' and government working-class politics, and it is only throughmanipulation supcontinuesin office. Thisis a manifestabsurdity. It ignores port that the leadership the extent to which all union leaderships of must operatewithinthe parameters the existingsystem. The questionof corruption is not unimportant. It indicatesboth a motive for and a office source of It also retaining power. provides,perhaps,a motive for settlementfor the 'selling'a contractto an employer- acceptinga less favourable workersin returnfor a bribe. The extent to which such practicesoccurin Latin Americais difficultto determine,but they cannotbe dismissed out of hand. The use of state interventionto impose a particular in a trade union (as leadership occurredin the aftermathof the railwayworkers'strikesof 1958-9) is perhapsa feature of Latin Americanunionismwhich does not occur elsewhere.This,perif we took the impoHowever, haps,might be the definingelement of charrismo. sition of a specificleadership of charrismo we would by the state as the hallmark find that this practicewas relativelyinfrequent,both in Mexicoand elsewherein LatinAmerica.In this caseperhapsthe use of violenceto repressinternalopposition within the union mightbe a better definingcharacteristic of charrismo. It is haveused certainlythe case throughoutthe world that many union leaderships physicalviolenceagainstsectionsof their rankandfile,most frequentlyoccurring in unionswhere corruption has been important.Nevertheless, it seemsunwiseto take the occurrenceof violenceas a definingcharacteristic becausemany unions which we might wish to describeas oligarchical are not characterized by overt violence. Leaderships which are not authentic exist for a varietyof reasonsand seek to perpetuatethemselvesin office througha varietyof mechanisms which do not necessarily rest on the use of violence. One of the reasonswhy trade union leaderscontinue in office is that their see them as delivering the goods. There can be little doubt that, in membership many instances,unions are able to operate in the labourmarketto alter wages and conditionsof work, at least in the short run. Thisis, however,largelyunexplored territory,with a dearthof concrete studies of the impact of unions on to supposethat wages in Latin America.Nevertheless,it would be unreasonable to thisprocess. are completelyirrelevant unions, and thereforeunion leaderships, Whatemergesfrom this discussionis the need for a multidimensional approach to the subject. Analystshave tended to work with ideal-typicalconstructs,presenting a list of union types (often only two types), each of which is definedby a clusterof variables. Thisarticleargues thattheseideal-types shouldbe unpacked, and the constituentvariables treated separately.Firstly, it seems usefulto treat variablesrelatingto the internalgovernmentof unionsseparatelyfromvariables which describethe relationships betweenunionsandotheractorsandinstitutions. with the analysisof unionleadership, (Therearesome problemshere, particularly since this is the principal point of contact between internal and external variables.)Turningto union government,the simple dichotomy of oligarchyvs an imdemocracyneeds to be redefined.In the structureof uniongovernment, portant question concernsthe intermediatestrata of union officers. To what of the top leadership? Are extent, and in what ways, do they act independently the shop-floorofficials primarilythe executiveagentsof top leaders,or arethey primarilyresponsibleto the rank and file? Do they havetheir own independent



from the top leadership? sourcesof power, or is all power derivative In additionto the question concerningthe extent to which the unionrankand file can influence the selection of leaders and the policies pursued,there is the issue of their perceptions and attitudes to the union government.Do the rank and file accept the leadershipas legitimate?Do they think the leadershipis doing a good job? Is the leadershippopularor not? These dimensionsof support for the leadershipneed to be investigatedin their own right.It would be premature to assume that a democraticallyelected leadershipwas popular, or that an oliwas not. Indeed,one of the sourcesof oligarchicruleis a widegarchicleadership does deliverthe goods spreadbelief among the rank and file that the leadership and has a legitimaterightto representthe membership. in LatinAmerica Until very recently, few studies of internaluniongovernment existed. The over-politicized conception of the natureof the labourmovement, togetherwith a Michelianpessimismconcerningthe possibilitiesof union democThe racy, suggestedthat empirical investigationwould be largely redundant.22 received wisdom was that trade unions in Latin America were oligarchical machines,run by self-serving leaders, be they populist, verbally'revolutionary', or conservativebureaucrats.However, those few empirical studies which have been publishedrecently suggestthis pictureto be overly simple.23 Although many, possibly most, unions in LatinAmericaare run by oligarical cliques, there are, and alwayshave been, exceptions to this rule. In Brazil,in the 1950s and again in the 1970s, a number of unions, particularlyin the metalworking industries, appear to have produced leadershipswhich were directly responsibleto their constituents.24In Argentina,importantunions such as the meatpackersfought bitterly (but ineffectively) to forestall the imposition of and within Peronism,militant currentshave often provided Peronist keaders,25 the vehiclesfor oppositionalcurrentswithin unions. A similarsituationoccurs in Mexico, where despite all the talk of charrismo,substantialelementsof democracy exist in certainunions in the automobileindustry,in electricity, and in the union.26Oppositionalcurrentsexist, or have existed in the mining-metallurgical past, in other importantunions, with internalunion democracya salientfeature of the labour movementsin Chile,Peruand Bolivia.The extent of union democracy, and its perdurability,remain to be examined. That it exists (alongside oligarchicalpractices) cannot be denied, while, as suggested above, studies of union government need to move beyond the formal dichotomy of oligarchydemocracy,towardsa more refinedtypology of formsof union government. on the workingclass Up to now, we have concentratedour attention primarily and the labour movement, and the state and dominant classes have been mentioned only in passing.Yet many analystswould claim that, in the case of Latin of labour America,one should beginwith the state. The corporatistorganization in Latin Americatestifies to, and derivesfrom, the preponderant role played by the state in these societies. Generallyspeaking,the strengthand omnipresenceof the state vis-a-viscivil society is the starting point for any analysis of labour movementsin Latin America. However,the balanceneeds redressing againstthis since the state not is all. it is Moreover, over-politicizedimage, easy to underestimate the direct impact of both the workingclassand industrialists on policy formation and implementation.Even in the stronglycorporatistetatiste regimes, Just as the corporatist the direct influence of these classes is often discernible.27



so alsoarethe corporatist societiesareoften overstated, aspectsof LatinAmerican aspects of Europeanand North Americansocieties understated.To say this is does not exist, but merely that it is neitherunique not to say that corporatism of those societies. nor the key to the understanding to LatinAmerican It is not difficult to show how LatinAmerican states,with varyingdegreesof success,have attemptedto control labour relations.As importantis to attempt to determinethe role playedby labourand capitalin this process.It is very easy to see labouras a passiveor purelyreactiveforce. However,a somewhatdifferent readingof the historical record is possible. As Skidmorehas noted in a recent essay,28it is paradoxicalto claim that labouris essentiallypassivewhen it can plausiblybe arguedthat manymajorpoliticalcrisesare due largelyto the action a few illustraof the labour movement.Perhapsit might be useful to enumerate tive examples.Bolivia:Since the revolutionof 1952, the COB,and in particular the tin miners,have been a majorpoliticalforce. The COBwas the leadingforce in the governmentsof Paz Estenssoro,an importantforce in the Siles Suazo while the Barrientos administration, coup of 1964 was, in many aspects, a resregime, ponse to the powerof the union movement.Later,underthe briefTorres in a situationwhichhas been describedas dual the COBonce againparticipated power. Brazil:Despite widespreadagreementamong academicsas to the weakness of the Brazilianlabour movement,it was, as a result of the 'strikeof the 300,000' in Sao Paulo in 1953, largely responsiblefor Vargas'downfall the following year.29Strike activity continued, with another massivestoppage in to 1957, while duringthe Goulartpresidencystrikeswere a factor contributing labour movement the military interventionof 1964. Since 1977, the Brazilian has once again experiencedan upsurgeof militancy,viewed with considerable circles.Chile:Mentionof the Popular concernin governmental Unitygovernment is sufficientto note the importanceof organizedlabourin Chilean history.Perhaps it should also be amphasizedthat the victoryof the PopularUnity in 1970 of work.Mexico: of decades was not a bolt out of the blue;it wasthe culmination as a countrywherelabouris totally subservient to the governregarded Generally at the of look a brief relations ment, history government-union suggeststhat these have been more problematicthan is sometimesassumed.The generalprogovernmentposture of the official labourmovementwas only securedas a result of massivepurgesof the Left in 1948.30Evenso, oppositionto government polin icies continuedin a numberof importantunions, and flaredup dramatically the railwayworkers'strike on 1958-9. In the late 1970s, the government with some difficulty persuadedthe labour movementto acceptan incomespolicy, as policy. Argentina:The history of Argentinasince part of an anti-inflationary and hence, of the tradeunions.Whenthe 1943 has been the history of Peronism, Peronistshave been out of power,the labourmovementhas (at least until 1976) of 1969).31In power, been able to bring down government (as in the Cordobazo Peronismhas alwayshad to try to controla militantrankand file. As in Mexico, on the government. dissidentunions have been ableto exert substantial pressure These, of course, are the strongestcases.Similararguments mightbe difficult to make for countriessuchas Ecuador, Venezuelaor Colombia.But it is not the purpose of this article to arguethat the labour movementis a majorpolitical force in all LatinAmericancountries,only that theoriesthat emphasizethe element of control neglect an importantpart of the picture- continuingmilitancy



in the more importantcountries.In a period when militarydictatorshipsdo their best to confine union activity to narrowlimits, it is perhapssalutaryto stressthe potential for conflict and destabilizationwhich is characteristicof organized labour.On the other hand,it would be quite wrongto suggestuniversal militancy and political combativity.Labourmovementsvaryconsiderablyin their political behaviour,both over time, and from country to country, and thereforemust be discussedin termsof the system of political partiesspecific to each country. It may seem strangethat the notion of populismhas hardlybeen mentioned in this article up to this point. The omission is deliberate.The term is used so loosely, and in so many ways, that it generallybringswith it confusion rather than enlightenment.Ratherthan enter a necessarilylengthy discussionabout the possible meaningof the term, I will merely state how I intend to use it in this of the term populismis that it refersto an ideology, or article. My understanding element in an ideology, which assertsthat the principalconflict in society is between the people and the oligarchyor imperialism.32 As such it is usuallycounterposed to a vision of society as made up of classes.Movementsor governments which espouse such an ideology may be referredto as populist, providingthat this is taken as descriptiveof theirideologyalone,and carries no otherimplication such as a loose organizational leader,or a mass (rather structure,or a charismatic than class) base, or a multi-classcoalition, etc. Given this definition, a great many political movementsmake some sort of appealto the people, andthus have populist elementsin their ideology.33In generalterms, I suspect that the analysis of labour movements in terms of differences in their professedideologies and beliefs is probablyof limited use in understanding their actualbehaviour.If this is so, then the differentiation,in ideological terms, between populist and classconsciouslabourmovementsis probablyof little utility. Whatis more useful, perhaps,is a distinctionin terms of organizational structure and the classoriginsof the membership, which is, of course,what is implied in some definitionsof populism.Unfortunately,the correlationbetweenideology and organizational structureis usually asserted,ratherthan demonstrated,with considerableslippagebetween the two. It is probablymore useful to examine organizationalstructureseparately.This must be examinedin terms of real relastructure.It cannot tionshipsbetween the component partsof the organizational be inferredfrom an organizationaldiagramor from a set of statutes. This said, the significanceof organizationalstructureappearsto residein two sets of questions: (1) internal union government,which is discussedelsewherein this article, and (2) size and compositionof bargaining units. The organizationof collective bargaining varieswidely in LatinAmerica.Not are some countries much more in this respectthan others, there centralized only is also considerablevariationwithin certain countries. Disentanglingthe effects of bargaining structurefrom other variables, althoughcomplicated,canbe neglected only at the analyst'speril. Another closely relatedfactor which differentiates labour movements is the state of the labour market.Grossomodo, it seems reasonable to account for the greater bargainingstrength of unions in Chile and Argentina,comparedwith other Latin Americancountries,in terms of the early formation of more or less homogeneouslabourmarketswith relativelylow levels of unemployment.34 At a sectoral level, variousinsitutional controls over entry into the labour force act to tighten labour markets in situations of apparent



labour surplus. In some industries(mining, for example) collective contracts sometimes contain clauses stipulatingthat sons of workersmust be given preference in hiring. Over time, this segmentationof labour marketsis likely to produce cumulative effects, diminishing social mobility within the working has of privileged strata.This phenomenon class and leadingto the crystallization in termsof the notion of a dichotomous been much commentedon, particularly and a marginalized But the existence mass.35 split between a labour aristocracy of sucha divisionin the labourforce cannotbe taken as given.Muchwill depend in the automoon patternsof job tenure.For example,in Mexico,whereturnover bile industry is very low and wages quite high, many firms are contractually obliged to give preferencein hiringto sons of employees. In this situation,one stratumof workers.However,in both might expect the formationof a privileged Braziland Argentina,where labourturnoverin the automobileindustryis high, and wages are not exceptionallyhigh, one would not, therefore,expect the forstratum.36 mationof a privileged One important set of factors in relationto labourmarketsis the processof urbanization. Thereare significant between LatinAmerican differences countries in terms of the size of the urbansectorvis-a-vis the ruralsector, the rateof ruralthe availability of employmentoutsidethe metropolis(regional urbanmigration, industrialor mining centres), etc. These factors affect not only the supply of of the labourforce. labour,but also the previousexperiences Another differentiatingfactor is industrial development, not merely the absolute size of the manufacturing sector, important thoughthis is, but also the and mining.Hereit is importantto structureand compositionof manufacturing of enterprises, as well as theirgeographical considersize distribution distribution, and the processof developmentof the leadingsectors. It has been arguedthat sectorsof patternsof changein industrialrelationsbegin in the export-oriented into the international societieswhich areintegrated Thisis a plausible economy.37 and interestinghypothesis,though possibly a ratherrestrictive one, and I would of notion at a broader which leadingsectors, might times, but need not, suggest coincide with the export sector. Industries will differ in terms of their political importance,with the state more concernedin some than others about growth and about labour relations.The state is likely to intervenein the settlementof labourdisputesin those industrieswhere it has interests,althoughthis does not necessarilymean it will intervenedirectly on behalf of the employers.It is not difficult to envisagesituationsin which the state views the employersas an obindustrial stacleto regularized relationsand economicgrowth. Let us considerthe leadingsector industries(those which aredefinedby the state as leading the growth process). These tend to be industriesexperiencing for an approfairly rapidexpansion,supported by the state in termsof measures are likely,if successful, Suchleadingsectorindustries priategrowthenvironment. for periodsof twentyto fortyyears.Thereafter to retaineconomic predominance to are enter a they likely long period of slow decline. The textile industryin many Latin Americancountriesprovidesa good exampleof this phenomenonin the first half of the twentieth century though miningwould obviouslybe the leadingsector in some countries.Wagesand workingconditionsin these leading sectors will not necessarily be higheror lower than in other industries. It may be hypothesizedthat in the initial stages of these leadingsector industries,when



industrialconflict emergeson a large scale, the state is likely to intervenein disputes between labour and management.Interventionwill not be an ad hoc manlabourrelationsin the long term, it will institutionalizea ner, but, to restructure specific pattern of labour relations. The next step in the argumentis to assert that this model will be diffused to the rest of the economy in a relativelyshort time. This diffusionmay occur in two ways. Eitherthe workersin otherindustries will take the leadingsector (correctlyor incorrectly)as a model of whatis possible and effective, or the state will impose this pattern throughoutthe rest of the economy by legal enactment. As the leadingsector shifts over time from one industryto another,there will be a break in the institutional pattern of classrelations.As a new leadingsector emerges, the state will once again intervenein this sector to regulatethe pattern of class conflict. The older patternwill almost certainlybe substantially modified in the process, and labour organizationswill be restructured. This model asserts that discontinuity, rather than continuity, will characterizeLatin Amercianlabour movements,which means that any attempt to find an originalmoment in history when the pattern of labour relations was set, once and for all, will be fruitless. This is worth emphasizing because some analystsseem to think that, at least in severalkey countries, the patternof labourrelationswas definitivelyset at some key juncturein the first half of this century,and that this somehow sets and defines the essence of that country's labour movement.38 Obviously,I disagree fundamentallywith such a perspective.The point that is often made, that the institutionalpatternsof the labourmovementwere profoundlyalteredin the first decades of this century in severalcountries,is quite valid. But this did not set industrial relations in an immutable mould. In particular,it could well be arguedthat there were major shifts in the 1940s and 1950s in some of the more industriallyadvancedcountries, such as Brazil,Mexico and Argentina.And most analysts would accept that the late 1960s and 1970s saw widespreadattempts at of restructuringlabour relations as part of the process of internationalization capital. So far, the model has been presentedin an entirely formal way: discussion of the patterns of institutionalized class conflict has been omitted. Which form of labour relationsis adopted in the leading sector will depend partly on the models availableto the state, andpartlyon the particular form of organization of the working class in that industry. Whatis meant by the models availableis that Latin Americanstates, by and large, drawtheir inspirationfrom the stock of ideas and practicescurrentlyavailable.The obvious exampleis the influenceof Italianfascismon the adoption of corporatistpracticesin severalLatinAmerican countries in the 1930s, the form of working class organizationin the industry relatesto the so-calledobjective factorsmentionedabove. It will be apparent that this model has little space for certaineconomicvariables: wages;economiccycles; inflation; rate of profit; rate of capital accumulation;etc. These factors enter into the model only in two ways: (a) as backgroundfactors contributingto the development of a leading sector and affecting the dimensionsand timing of industrialconflict in that sector;(b) as factors which influence the volume and timing of conflict once an institutional pattern has been established,but which do not directly determinewhich institutional solution is adopted.



Other factors,however,do play an importantrole in this model. Thesehave to do with the degreeof homogeneityor heterogeneityof the work force in the leadingsector and its relationswith the work force in other sectorsof the economy.3 It has been arguedthat work forcesvaryfrom industryto industry,with the most well-known analysisalong these lines perhapsthe Kerr-Siegal hypothesis.40In attemptingto accountfor what they believedto be the high levelsof in certainindustries(mining,lumber,etc.), Kerrand Siegaldestrike-proneness veloped what they call the isolated mass hypothesis.Drawingdirectly on masssociety theory (best exemplifiedby Komhauser),41 they postulatedthat isolated and homogenouswork forceswere likely to develophigh levelsof solidarity,and that this would be a factor disposingthem to high levels of collective conflict with management.42 Othertheoristshave talked about dual markets,or about a mass and a labour supposedsplit in the subordinateclassesbetween a marginal aristocracy.Whatall these distinctionshavein commonis some notion of heterTorre,forexample,makeshomogeneity ogeneity of the workingclass.JuanCarlos one of the lynchpinsof his work, stressing, for example,the homogeneityof the with the Bazilian BuenosAiresworkingclassas compared workingclass.Hetero/homogeneityis clearly a useful notion, but it is often ratherimprecise.If we are to talk of a workingclass, some kind of homogeneityis clearlyimpliedeitherin terms of positionin the labourmarket,or in termsof life chances,or in termsof sharedperceptions. In discussionsabout homogeneity in terms of life chancesand experiences, many writershave stressedthe existenceof a radicalcut withinthe workingclass between a stable labour force in manufacturing and an unskilled,migrant,marin of labour the the force rest haveeven impliedthat resiSome ginal economy. dentialpatternsmirrorthis disjuncture, with the economicallymarginal livingin labourforce presumably favelasand the core industrial livingin some other form of housing.As far as the housing questiongoes, however,a largebody of literature criticizes such a straightforward dichotomy as far as housing is concerned with the recognitionthat residentialneighbourhoods contain varyingmixes of occupational types.43 the occupationaldisjuncture continuesto be acceptedwidely. I Nevertheless, think we need to reconsiderthis matter carefully.This supposed disjuncture within the Latin Americanworking class is based on assumptionsabout the tenure of occupationalroles. Specifically,it is assumedthat once a workerhas a industry,he or she will keep it permanently,i.e. job in modernmanufacturing that the rate of turnoverof the labourforce will be verylow. Thisis an empirical question, which for instance may be true in Mexico. However,as mentioned evidence as there is for Argentinaand Brazilsuggests above, such fragmentary that labourturnoverin the modernmanufacturing sectoris quite high,44though it should be stressed aboutthe operajust how little we actuallyknowempirically tion of labourmarketsin LatinAmerica.If it is assumedthat turnoverratesare high, what does this imply for workingclasshomogeneity?Surelya highrate of turnovermust increasethe homogeneityof experiencewithin the workingclass, as people move betweenoccupationalroles. The homogeneityof the workingclass has two importanteffects: within any given industry or labour force, it increasesthe workers'capacity to organize effectively againstthe employer.(I am not sayingthat homogeneityis the only,



or indeed, the most important factor in determining workerresistance,nor am I saying that there will be no worker resistancewhen the labour force is heterogenous.) Secondly,within the workingclassas a whole, the degreeof homogeneity will be one of the factors makingfor a rapidtransmission of the lessonsof industrial conflict in the leading sector to the rest of the economy. (Again,I am not sayingthat homogeneityis the only factor involvedhere.) the preceding To summarize argument: any adequateanalysisof LatinAmerican labour movementsmust beginwith a multivariate approach.In this articleI have attempted to discuss some of the factors which account for variationsamong labourmovementsin LatinAmerica,factorssuchas type of internaluniongovernment, the degreeof integrationof the labourmarket,the degreeof homogeneity of the workingclass,rates of labourturnover,differingformsof corporatism, etc. is a long one,and many combinationscould be devised Clearly,the list of variables to createtypologies. I have refrained from suggesting suchtypologiesherebecause I wish to stress the complexity and variability,both in time and space, of Latin Americanlabour movements.Whilethe elaborationof ideal-typesis a necessary part of intellectual enquiry, it should not lead to prematurecodification and oversimplification.At this stage, we are a long way from even being able adequately to describeLatinAmericanlabourmovements,let alone explain them.

the case with Braziland Mexico,and to some extent Argentina and 1. This is particularly a 17-volume UNAMand SigloXXI are currently Peru.To give an illustration: publishing collection on the history of the Mexicanworkingclass. This articleis not intendedas a ratherthe aim is to illustrateselectivelywhat I believe systematicsurveyof the literature; in the areaof labourhistory. to be generaltheoreticalproblems Labor in LatinAmerica,New York University 2. Hobart A. Spalding(1977), Organized Press,(New York). 3. Ibid. p. 282 andp. ix. Collins(London),p. 242. 4. A. J. Hexter(1979), OnHistorians, 5. At a broaderlevel, the differencebetween 'lumpers'and 'splitters'is exemplifiedby James Malloy (1977), 'LatinAmerica,the modal pattern'.In: J. Malloy(ed.) Authoriin LatinAmerica,University tarianism and Corporatism of Pittsburgh Press(Pittsburgh), versus the multiplepath analysisof F. H. Cardosoand E. Faletto (1979), Dependency and Development in LatinAmerica,University of California Press(Berkeley). 6. I would include as examples of this perspectiveL. MartinsRodrigues(1974), Trabalhadores Sindicatose Industrializacao, Brasilience(Sao Paulo), K. P. Erickson(1977), The BrazilianCorporative Stateand Working ClassPolitics,University of California Press (Berkeley);H. Wiarda(1978), 'Corporative Originsof the Iberianand Latin American Labor Relations Systems', Studies in Comparative International Development,vol. 13, No. 1. 7. The analysisof the role of the state, and of corporatistinstitution, in the USA and WesternEurope is hardly a novelty. Workswhich deal with this include, inter alia, A. Shonfield (1965), Modern Capitalism,Oxford UniversityPress(London), N. Harris and the Corporate (1972), Competition Society, Methuen(London),C. Crouch(1979), ThePoliticsof Industrial Relations,Fontana(London). 8. Thatis, the contrastis not between stateintervention andits absence,but betweentypes of state intervention. 9. Spalding (1977), p. 282. 10. Cf. R. Hyman(1975), Industrial Relations,Macmillan (London).



betweena formtwo-orthree-party 11. WhatI havein mindhereis the difference democracy and the differencebetween bureaucratic and plebicitarian and democracy, oligarchies mafia-likeunion bosses.Cf. the discussionin J. Edelsteinand M. Warner (1975), ComUnionDemocracy, AlienandUnwin(London). parative 12.It is sometimes often on the basisof somekindof Fei-Ranis, with asserted, 'development unlimitedsuppliesof labour'model, that the economicbehaviourof trade unions in has no discernible LatinAmerica effectson wages.Thisis, I think,an open question.The sameappliesto the possibleeffect of the stateof the economyon wages. andPoliticsin Peru,YaleUniversity 13.JamesPayne(1965), Labour Press(New York). CorPayne'smodel has been pickedup by interalia, K. Erickson(1977), TheBrazilian ClassPolitics, Universityof CaliforniaPress(Berkeley); porative State and Working YaleUniversity M. Urrutia (1969), TheDevelopment of the Colombian LaborMovement, Press (New Haven);S. Sigaland J. C. Torre'Unareflexi6nen torno a los movimientos and J. L. Reyna(eds.) (1979), Fuerzade en AmericaLarina'. In: R. Katzman laborales en America Laborales rabajo Latina,Colegiode Mexico(Mexico). y Movimientos 14. See also L. Martins (1974). Rodrigues 15. By institutionalrestrictionson labour marketentry,I havein mind, for example,the in a unionized by provisionin Mexicanlabourlaw that job applicants plantbe proposed the tradeunion. 16. The Payne model may also have some utility for earlyperiodsof industrialization in someEuropean countries. and exclusionaryforms of corporatism is now 17. The distinctionbetween intergrative widely accepted.Cf. the importantarticleby R. B. Collierand D. Collier(1979), 'Invs. Constraints: ducements APSR,vol. 73, No. 4. Disaggregating Corporatism', of the term.Oneof the clearest debateon the meaning i8. Thereis a considerable statements of the standard is NicosMouzelis notion of populism andClass Politics: (1978), 'Ideology a critiqueof Eresto Laclau', New Left Review,No. 112. ThatI disagree fundamentally with this conception of populismwill be apparentfrom the discussionin my book, I. Roxborough(1979), Theoriesof Underdevelopment, Macmillan (London).I have searchedin vain OctavioIanni(1972), La Formacion del EstadoPopulistaen America Latina,ERA (Mexico),for a definitionof the term.It is not clearwhetherIanniis referthat happenedin Latin ring to an ideology,a movement,a state, or merelyeverything America after1930. 19. This will be discussedin more detailbelow. A typical expositionof this dichotomyis A. Quijano(1974), 'The Marginal Pole of the Economy and the Marginalized Labour Force',Economyand Society,vol. 3, No. 4. 20. Malloy(1977). en Mexico,ERA(Mexico),p. 98. Ferrocarrilero 21. A. Alonso(1972), El Movimiento 22. R. Michels(1962), PoliticalParties,Collier(New York). Michelsarguedthat a number of factors would predisposetrade unions towardsoligarchy. This thesis was basically accepted by S. M. Lipset et al (1956), Union Democracy,Doubleday(New York). andWarner, Edelstein differentpicture. op cit., presenta rather 23. H. Handelman within the (1979), 'Unionization, Ideology,and PoliticalParticipation Class'.In: M. Seligsonand J. Booth (eds.) PoliticalParticipation MexicanWorking in LatinAmerica,vol. 2, Holmesand Meier(New York);H. Handelman (1977), 'Oligarchy and Democracyin Two MexicanLabourUnions',ILRR, vol. 30, No. 2; J. C. Torre en la Argentina', Sindical Desarrollo (1974), 'LaDemocracia Economico,vol. 14, No. 55; S. Gomez Tagle (1980), Insurgencia en los SindicatosElectrecistas, El y Democracia Colegiode Mexico(Mexico). 24.T. Harding(1973), 'The PoliticalHistoryof Organized Laborin Brazil',Ph.D. Thesis, Stanford University. and Proletarianiz7tion 25.C. Berquist(1979), 'Bourgeoisfication in the Semi-Periphery: ClassPoliticsin Argentina andChileCompared', MS. Working unpublished andI. Roxborough 26. M. Thompson andUnionDemocracy in (forthcoming), 'Corporatism Mexico'. In: K. Coleman(ed.), The Politics of Laborin Latin America,Holmesand Meier(New York). 27. See, for example,in the Brazilian EstadoE Capitalcase, Eli Diniz (1978), Empresario, ismo No Brasil,Paz e Terra(Rio de Janeiro); e L. Werneck Vianna(1977), Liberalismo no Brasil,Paze Terra(Rio de Janeiro). Sindicato



andElite Responses and Soldiers:Urban LaborMovements 28. T. Skidmore(1979), 'Workers andModernin TwentiethCenturyLatin America'.In: V. Bernhard (ed.) Elites,Masses of TexasPress(Austin). izationin LatinAmerica1850-1930, University de Massae Crise 29. Ibid;J. A. Moises(1978), Greve Politica,Polis (Sao Paulo). vol. 20 of Historiade la 30. L. Medina(1979), Civilismo y Moderizacion del Autoritarismo, RevolucionMexicana,El Colegiode Mexico(Mexico). 31. F. Delich (1970), Crisesy ProtestaSocial, B. A. Signos;B. Balveet al (1973), Luchade Calles,Luchade Clases,B.A. RosaBlindada. 32. This is, in many respects,similarto the positionadvanced by E. Laclau(1977), Politics and Ideologyin MarxistTheory,NLR (London). of populist strains,it is hardly 33. This is preciselyLaclau'spoint. Giventhe omnipresence worthwhiletalkingabout movementswhich areexclusivelypopulistin supposedopposition to movementswhichareuntaintedby populistideologicalthemes. en laborales 34.S. Sigaland J. C. Torre (1979), 'Una reflexion en tono a los movimientos and J. L. Reyna(eds.),Fuerzade Trabajo AmericaLatina'.In: R. Katzman y Movimientos en AmericaLatina,El Colegiode Mexico(Mexico). Laborales y Fuerzade Trabajoen la (1978), Transnacionales 35.Quijano, op. cit.; R. Trajtenberg ILET(Mexico). Periferia, de Sociologia,vol. RevistaMexicana automotrizArgentina', 36.J. Nun (1979), 'Laindustria da industriaautomobilisticano Brasil',Estudos XL, no. 1, J. Humphrey,'Operarios no. 23. Cebrap. 37.J. Cronin(1979), Industrial Conflictin Moder Britain,CroomHelm(London). Maritime Strikeof 38.For example, Skidmore(1979); P. de Shazo (1979), 'The Valparaiso 1903', JLAS,vol. II, no. 1. 39.For analyseswhichuse a homogeneity/heterogeneity dichotomy,seeinteralia,J. C. Torre mimeo;E. Jelin (1977), 'Orientaciones (1979), 'El movimientosindicalen la Argentina', y ideologiasobrerasen AmericaLatina'.In: R. Katzmanand J. L. Reyna(eds.), Fuerza en AmericaLatina,El Colegiode Mexico(Mexico); de Trabajo Laborales y Movimientos e ConflitoSocial,DIFEL(Sao Paulo). Urbano B. Fausto(1977), Trabalho to Strike'.In:A. Kornhauser 40.C. Kerrand A. Siegal(1954), The Inter-industry Propensity et al (eds.), Industrial (New York). Conflict,McGraw-Hill 41.W. Kornhauser (1959), ThePoliticsof MassSociety, FreePress(New York). featuresof this theory it has been stronglycriticizedon empirical 42.Despite some attractive of the Kerr-Siegal as well as theoreticalgrounds.See P. Edwards (1977),'A Critique Hypothesis',Sociological Review,vol. 25, no. 3. Arnold(London). 43.BryanRoberts(1978), Citiesof Peasants,Edward 44.Cf. n. 36.