Politics &Society

EDITORIAL BOARD

FRED BLOCK MIEKE MEURS University of California, Davis American University JOHN BOWMAN MOLLY NOLAN Queens College, CUNY New York University ARCHON FUNG DAVID PLOTKE Harvard University New School for Social Research IRA KATZNELSON GAY SEIDMAN Columbia University University of Wisconsin–Madison ELIZABETH KIER NIKHIL PAL SINGH University of Washington University of Washington CLAIRE JEAN KIM SUSAN STOKES University of California, Irvine University of Chicago MAGALI SARFATTI LARSON MARGARET WEIR Temple University University of California, Berkeley University of Urbino, Italy ERIK OLIN WRIGHT MARGARET LEVI University of Wisconsin–Madison University of Washington

MARY-ANN TWIST Managing Editor

Politics & Society publishes original analyses of politics, including its social roots and its consequences. In its broadest sense, politics encompasses conflicts over the shape of social life, whether on the shop floor, within the family, or in the realms of the state and the world economy. The quest for a good society is also enduringly a part of political life. Politics & Society is committed to developing Marxist, post-Marxist, and other radical perspectives and to examining what Robert Lynd once called “some outrageous hypotheses.”

For Sage Periodicals Press: Stephanie Lawrence, Ani Ayvazian, Tammy Ditmore, and Corina Villeda Politics & Society Volume 29, Number 4, December 2001

Contents

Contributors 483

Collective Action, Property Rights, and Decentralization in Resource Use in India and Nepal ARUN AGRAWAL ELINOR OSTROM 485

National Political Community and the Politics of Income Taxation in Brazil and South Africa in the Twentieth Century EVA N S. LIEBERMAN 515

Thinking Globally, Acting Locally: Local Governments, Industrial Sectors, and Development in China ADAM SEGAL ERIC THUN 557

Fee-Splitting Revisited: Concealing Surplus Value in the Temporary Employment Relationship GEORGE GONOS 589

Index 613

Sage Publications Thousand Oaks • London • New Delhi Politics & Society publishes original analyses of politics, including its social roots and its consequences. In its broadest sense, politics encompasses conflicts over the shape of social life, whether on the shop floor, within the family, or in the realms of the state and the world economy. The quest for a good society is also enduringly a part of political life. Contributions are welcome from people of many disciplines, and they may take the form of theoretical essays, historical investigations, philosophical reflections, and empirical research. The journal emphasizes the use of lucid English in its articles. Politics & Society is committed to developing Marxist, post-Marxist, and other radical perspectives and to examining what Robert Lynd once called “some outrageous hypotheses.” Manuscript Submission: Contributions and other editorial comments should be sent to Mary-Ann Twist, Managing Editor, Politics & Society, 123 Van Deusen Street, Madison, WI 53715. Manuscripts may be transmitted electronically to [email protected] in most formats. Please include your name, address, phone, fax, and e-mail address on the cover letter. Politics & Society will not consider manuscripts currently under review with any other journal. Manuscripts must be typed, double-spaced, and submitted in duplicate. Notes must be double-spaced and placed at the end of the text (not at the bottom of pages). Politics & Society follows Chicago style (A Manual of Style, 14th ed., Chicago: University of Chicago, 1993). Notes must be complete: References to journal articles must include volume and issue numbers, as well as month and year of publication; references to books must list the author, title, city and year of publication, and publisher; references to contributed books must list the contributor, title of the contributor’s essay, and full book information, including editor(s). All publication decisions are made by the editorial board as a whole at its quarterly meetings. There, one of four possible decisions is made: (1) acceptance; (2) conditional acceptance, pending revisions; (3) revision and resubmission; or (4) rejection. In the case of a decision to accept pending revisions, the coordinating editor usually determines if the author’s changes meet the board’s requests. In the case of a decision to revise and re- submit, the article is treated as a new submission and is reread and rediscussed by the entire board. Contribu- tors should receive a decision within five months of submission. Politics & Society (ISSN 0032-3292) is published four times annually—in March, June, September, and December by Sage Publications, 2455 Teller Road, Thousand Oaks, CA 91320. Telephone: (800) 818- SAGE (7243) and (805) 499-9774; fax/order line: (805) 375-1700; e-mail: [email protected]; web site: www.sagepub.com. Copyright © 2001 by Sage Publications. All rights reserved. No portion of this work may be reproduced in any form without written permission from the publisher. Subscriptions/Inquiries: Annual subscription rates for institutions and individuals are based on the current frequency. Prices quoted are in U.S. dollars and are subject to change without notice. Canadian subscribers add 7% GST (and HST as approptiate). Outside U.S. subscription rates include shipping via air-speeded de- livery. Institutions: $333 (within the U.S.) / $346 (outside the U.S.) / single issue: $95 (worldwide). Individ- uals: $72 (within the U.S.) / $88 (outside the U.S.) / single issue: $29 (worldwide). Noninstitutional orders may be paid by personal check, VISA, or MasterCard. Address all correspondence regarding subscriptions to Sage Publications, 2455 Teller Road, Thousand Oaks, California 91320. All subscription inquiries, orders, and renewals with ship-to addresses in North America, South America, and Canada must be addressed to Sage Publications, 2455 Teller Road, Thousand Oaks, CA 91320, U.S.A., telephone (800) 818-SAGE (7243) and (805) 499-9774, fax (805) 375-1700. All subscription inquiries, orders, and renewals with ship-to addresses in the U.K., Europe, the Middle East, and Af- rica must be addressed to Sage Publications, Ltd, 6 Bonhill Street, London EC2A 4PU, England, +44 (0)20 7374 0645, fax +44 (0)20 7374 8741. All subscription inquiries, orders, and renewals with ship-to addresses in India and South Asia must be addressed to Sage Publications Private Ltd, P.O. Box 4215, New Delhi 110 048 India, telephone (91-11) 641-9884, fax (91-11) 647-2426. Address all permissions requests to the Thou- sand Oaks office. Periodicals postage paid at Thousand Oaks, California, and at additional mailing offices. The journal is abstracted in Academic Abstracts, Academic Search, the Alternative Press Index, Amer- ica: History and Life, Applied Social Sciences Index & Abstracts (ASSIA), Biological Sciences Abstracts, Canadian MAS, Corporate ResourceNET, Current Citations Express, CSA Political Sci- ence & Government: A Guide to Periodical Literature, Current Contents: Social & Behavioral Sciences, Expanded Academic Index, Health & Safety Sciences Abstracts, Historical Abstracts, International Bibliography of the Social Sciences, International Political Science Abstracts, ISI Basic Social Sciences Index, MAS FullTEXT, MasterFILE FullTEXT, Military Library FullTEXT, Periodical Abstracts, Political Science Abstracts, Public Library FullTEXT, Risk Abstracts, Safety Science & Risk Abstracts, Sage Public Administration Abstracts, Sage Race Relations Abstracts, Social Science Source, Social Sciences Index Full Text, Social Services Abstracts, Sociological Abstracts, Standard Periodical Directory (SPD), TOPICsearch, Wilson OmniFile V, and Wilson Social Sciences Index/Abstracts and is available on microfilm from University Microfilms, Ann Arbor, Michigan. Back Issues: Information about availability and prices of back issues may be obtained from the publisher’s order department (address below). Single-issue orders for 5 or more copies will receive a special adoption discount. Contact the order department for details. Write to London office for sterling prices. Advertising: Current rates and specifications may be obtained by writing to the Advertising Manager at the Thousand Oaks office (address above). Claims: Claims for undelivered copies must be made no later than six months following month of publication. The publisher will supply missing copies when losses have been sustained in transit and when the re- serve stock will permit. Change of Address: Six weeks’advance notice must be given when notifying of change of address. Please send old address label along with the new address to ensure proper identification. Please specify name of journal. POSTMASTER: Send address changes to: Politics & Society, c/o 2455 Teller Road, Thousand Oaks, CA 91320. Printed on recycled, acid-free paper POLITICSCONTRIBUTORS & SOCIETY

Contributors

Arun Agrawal ([email protected]) is an associate professor in the Department of Political Science at Yale University. His research focuses on the institutional political economy of development and environmental conservation. His first book, Greener Pastures, on pastoralists in western India was published by Duke and Oxford University Presses. He has also coedited Agrarian Environ- ments (Duke University Press), Regional Modernities (Stanford University Press), and Communities and the Environment (Rutgers University Press, forthcoming). His research has been published in various journals, among them, Comparative Political Studies, Development and Change, Journal of Asian Studies, Journal of Developing Areas, Journal of Theoretical Politics, and World Development. George Gonos ([email protected]) is an associate professor of economics and employment relations at the State University of New York at Potsdam. He is writing a book on the rise of the temporary help industry and the new wave of for-profit labor market intermediaries in the United States. Evan S. Lieberman ([email protected]) is a Robert Wood Johnson Health Policy Scholar at Yale University and will begin an appointment as an assistant professor of politics at Princeton University in September 2002. He is currently completing a book on the politics of taxation and state development (Cambridge University Press, forthcoming) and is the author of articles forthcoming in Comparative Political Studies, Studies in Comparative International Development, and the Journal of Development Studies. His current research examines the political determinants of national strategies for preventing the trans- mission of HIV/AIDS. Elinor Ostrom ([email protected]) is the Arthur F. Bentley Professor in the Department of Political Science at Indiana University. She is the codirector with Vincent Ostrom of the Workshop in Political Theory and Policy Analysis and with Emilio Moran of the Center for the Study of Institutions, Population, and Environ- mental Change. She is a member of the American Academy of Arts and Sciences and a recipient of the Frank E. Seidman Prize in Political Economy and the Johan

Skytte Prize in Political Science. Her books include Crafting Institutions for Self-Governing Irrigation Systems (1992), Governing the Commons (1990), Local Commons and Global Interdependence (1995, coauthored with Robert Keohane), Rules, Games, and Common-Pool Resources (1994, coauthored with Roy Gardner and James Walker), and Institutional Incentives and Sustainable Development (1993, coauthored with Larry Schroeder and Susan Wynne). Adam Segal ([email protected]) is a Next Generation Fellow with the Council on Foreign Relations. He is also a visiting scholar at the Center for International Studies, MIT. He received his Ph.D. from the government department at Cornell University in January 2000 and is currently finishing a book, Digital Dragon: High Technology Enterprises in China, to be published by Cornell University Press. Eric Thun ([email protected]) is an assistant professor in the politics department and Woodrow Wilson School at Princeton University. He received his Ph.D. from Harvard University in 1999. His current research focuses on the globalization of manufacturing networks and the ability of host societies to capture the gains from foreign investment. He is also completing a book on the auto sector development efforts of local governments in China. POLITICSAGRAWAL & and SOCIETY OSTROM

Collective Action, Property Rights, and Decentralization in Resource Use in India and Nepal

ARUN AGRAWAL ELINOR OSTROM

National governments in almost all developing countries have begun to decentralize policies and decision making related to development, public services, and the environment. Existing research on the subject has enhanced our understanding of the effects of decentralization and thereby has been an effective instrument in the advocacy of decentralization. But most analyses, especially where environmental resources are concerned, have been less attentive to the political coalitions that prompt decentralization and the role of property rights in facilitating the implementation of decentralized decision making. By comparing decentralization in four cases in SouthAsia—Forest Councils in Kumaon in India, Joint Forest Management in India, the Parks and People Program in Nepal’s Terai, and Community Forestry legislation in Nepal—this article provides answers to two questions: When do governments decentralize environmental decision making? and Which types of property rights must be devolved if decentralized decision making is to be effective?

Decentralization has emerged as a major strategy for many nation-states to achieve development goals, provide public services, and undertake environmental conservation. Political leaders and bureaucrats, international donors, and in many places local residents and their leaders have begun to see decentralization as a means to gain political-economic and policy objectives. According to a recent report from the World Bank, “Of seventy-five developing and transitional countries with populations greater than five million, all but twelve claim to be

The findings reported in this article are based on ongoing comparative research projects on decentralization of control over forest resources in India and Nepal. The research has involved extensive POLITICS & SOCIETY, Vol. 29 No. 4, December 2001 485-514 © 2001 Sage Publications 485 486 POLITICS & SOCIETY embarked on some form of transfer of political power to local units of governments.”1 Indeed, decentralization has periodically been an important objective of state policy.2 This article focuses on decentralization of central state control over natural resources, especially forests. The subject is important for at least three reasons. The lives of millions of households are affected by how governments manage forests and admit local claims.3 Furthermore, the factors that lead to durable decentralization of forests may be relevant in other arenas where decentralization is occurring. And finally, decentralized decision making can be seen as an inher- ently important concern of democratic life. Current studies of decentralization, however, suffer from two significant lacu- nae, especially where natural resources are concerned. They are relatively inatten- tive to the political question of why states attempt decentralization, and they do not pay sufficient attention to the nature of property rights that states devolve to local actors. Much scholarship on decentralization of natural resources asserts the superiority of decentralized solutions versus centralized solutions on grounds of efficiency, equity, or sustainability.4 But such advocacy can only show why decentralization should be pursued, not why political actors choose to pursue it. It remains relatively unclear why central government actors, known for their pursuit of power, should initiate actions to take power away from themselves. Inattention to the nature of property rights that central states devolve means that we do not well understand the relationship of property rights to political authority, or even which forms of authority are necessary for successful decentralization. As a result, we often cannot be sure if a government’s claims that it is pursuing thoroughgoing decentralization can be taken at their face value. In the ensuing discussion, we first focus on the politics of decentralization to identify the actors most likely to initiate decentralization activities. To address our second concern, we draw from the literature on property rights and propose that the extent of decentralization can be assessed by examining the rights and capacities that are transferred to actors at lower levels of political organization. Powers and rights that are devolved to lower levels have important effects on resource management-related outcomes. Our approach leads to a more precise framework to understand the nature of decentralization and its impact. interviews with more than 200 households in Nepal and India and a careful examination of village and district level records maintained by local forest management councils. We are appreciative of the time and hospitality extended to us repeatedly in the field. This article is a vastly revised version of an earlier draft prepared for the System-Wide Program on Property Rights and Collective Action (CAPRi) unit of the International Food Policy Research Institute. We thank Ruth Meinzen-Dick for inviting us to think more carefully about our ideas on decentralization, property rights, and collective action. We have benefited greatly from the comments of Archon Fung, Clark Gibson, Amy Poteete, Jesse Ribot, and other colleagues at the Workshop in Political Theory and Policy Analysis. We gratefully acknowledge the always-sterling editorial help from Patricia Zielinski and Shaun McMahon. Financial support from a John D. and Catherine T. MacArthur Foundation grant (00-63798-GSS) and a National Science Foundation grant (SBR 9905443) is also thankfully acknowledged. AGRAWAL and OSTROM 487

Comparison of the processes and outcomes of decentralization initiatives regarding forests in South Asia helps us answer the two questions this article addresses: What roles do local populations play in ensuring that property rights and decision making are actually devolved? And what types of property rights must be devolved to local populations for effective decentralization? The four comparative case studies we use are (1) the Forest Councils of Kumaon in India, (2) Joint Forest Management in India, (3) Community Forestry in Nepal, and (4) the Parks and People Program in Nepal. The four cases exemplify different origins of pressures for decentralization of forest governance, different property rights that central governments devolved, and different outcomes.

THE POLITICS OF DECENTRALIZATION

At its most basic, decentralization aims to achieve one of the central aspira- tions of just political governance: democratization, or the desire that humans should have a say in their own affairs. But decentralization is also synonymous with redistribution of power, resources, and administrative capacities through different territorial units of a government and across local groups.5 Thus, decentralization can be seen as a strategy of governance, prompted by external or domestic pressures to facilitate transfers of power closer to those who are most affected by the exercise of power.6 This understanding of decentralization implies that it is critical to identify the interests of the major actors involved in pursuing it. Among the most important actors who affect decentralization are international donors, local organizations, and NGOs, and factions within central governments. These three categories of actors are large aggregations, and their decentralization- related interests are likely to remain unpredictable without close attention to the context. This is especially true for central government actors. Consequently, the relationship between central and local governments can be ambiguous because local governments can be viewed by their superiors either as trusted agents or as contestants for power.7 To understand why a central government might initiate decentralization reforms and give up control over collective and operational choices in some instances and not in others, it is necessary to examine the political relationship between different local and central actors, the role of international donors and NGOs, and the ways in which decentralization reforms may serve the interests of these actors. An additional question that requires attention is, When is it likely that decentralization reforms will continue after initiation? To explain why a central government may pursue decentralization and voluntarily give up some of its powers to less powerful local government actors, we suggest that central governments are best seen as congeries of actors who have different and perhaps conflicting objectives as they pursue a diversity of goals including power. Once the center is recognized as being potentially divided, decentralization can be hypothesized to take place when actors at the central level compete for power among themselves and find in decentralization a mechanism to enhance 488 POLITICS & SOCIETY their access to resources and power in comparison to other political actors at the central level. Such a conceptualization of decentralization is at one level merely a restatement of the tautology that actors undertake those actions that gain them the highest returns. But even this simple statement allows us to suggest that decentralization of power and decision making is likely when some central political actor(s) or a coalition of such actors find(s) that decentralization makes it possible to reduce costs (and/or improve revenues), deflect blame, or extend state reach further into social processes. Without such calculations on the part of some central actors and their support, meaningful decentralization is unlikely to be initiated. In this sense, policy choices about decentralization are no different from policy choices in other domains. Even if there are effective coalitions in favor of decentralization, and programs are initiated to pursue decentralization, setbacks and retreat may be possible in the process of implementation. Once again, the support of some central state political actor may be essential, but for long-term success, it is as important to examine how local-level politics articulates with decentralization reforms. For effective implementation of decentralization programs that are initiated by central actors, our case studies suggest that local mobilization is critical. Because local political actors are less powerful than international donors or central government actors, their long-term interests are often ignored in decentralization initiatives unless they can form effective lobbying groups through networking or organization. The trick for advocates of decentralization, therefore, is to align the interests of powerful decision makers who make policy choices with organization and mobilization of local actors who can create additional pressures in favor of reforms. With such alignment, more thoroughgoing decentralization of powers and rights would become possible.8

PROPERTY RIGHTS AND DECENTRALIZATION

The multidimensionality of decentralization is reflected in the many terms that refer to it: for example, deconcentration, delegation, devolution, deregulation, privatization, and even denationalization.9 In the context of the management of natural resources, decentralization can be defined as any act by which a central government cedes rights of decision making over resources to actors and institutions at lower levels in a politico-administrative and territorial hierarchy.10 Usually, the means through which new actors and institutions come to gain new powers of decision making are different combinations of property rights.11 Prop- erty rights, themselves, can be defined as relationships among actors with respect to things such as natural resources.12 Property rights are always contestable. Contestation attenuates the capacity of right holders to exercise effectively the rights they hold. Typically, successful decentralization of resource management results in the creation of new commons as central governments delegate rights and powers to AGRAWAL and OSTROM 489 new actors who can make decisions about the disposition of these resources. The creation of new commons is an important development in natural resource policies worldwide.13 As governments formulate new rules, they allow district and other lower-level actors greater leeway in deciding the fate of locally situated resources. When decentralization is relatively far-reaching, as in the case of the Forest Councils of Kumaon, or to an extent, in the case of Nepal’s Community Forestry, even local residents in village communities and their representatives can gain new powers of decision making and new property rights over local forest resources. Because environmental policy decentralization in developing countries is ostensibly designed to manage resources better by granting rights to communities, the large body of literature on common property theory is especially useful to analyze the extent to which decentralization programs are successful. Different types of property rights over resources—whether held juridically, exercised authoritatively, or both—can create varying consequences for use and management. Drawing from existing scholarship on the commons, we focus here on four types of property rights that are most relevant for the use of common-pool resources: withdrawal, management, exclusion, and alienation.14

Withdrawal is the right to enter a defined physical area and obtain resource units or products of a resource system (e.g., cutting firewood or timber, harvesting mushrooms, diverting water). Management is the right to regulate internal use patterns and transform the resource by making improvements (e.g., planting seedlings and thinning trees). Exclusion is the right to determine who will have right of withdrawal and how that right may be transferred. Alienation is the right to sell or lease withdrawal, management, and exclusion rights.

The above rights can be exercised at several distinct levels of analysis. The exercise of withdrawal rights corresponds to an operational level of analysis.15 Management, exclusion, and alienation rights, however, require that the right holders operate at a collective-choice level of analysis that affects future operational decisions. Furthermore, for communities to possess collective choice- making capabilities, some rules at a constitutional level (set locally or by a national government) must give them this authority. The analytical distinction between operational level rules, and collective and constitutional choice arenas should not create the impression that these correspond to three different formal levels of authority in a political or legislative system.16 A single political body can use operational rules, create them by deliberating at the collective choice level, and may have powers in the constitutional-choice arena as well. Conversely, a number of levels of authority may exist, corresponding perhaps to the village, district, province, and nation, where specific rules are created and powers are exercised. It is crucial to understand that for any resource, some rules affect day-to-day 490 POLITICS & SOCIETY use and consumption, others structure the creation of operational level rules, and still others at a higher constitutional level affect collective choices. In a highly centralized regime, almost all authority for making constitutional, collective-choice, and operational-level rules is concentrated in a national government. Local officials and citizens are viewed as rule followers, not as rule makers. In regimes that have undergone forms of deconcentration, the authority to make rules at the three levels is still lodged in national government officials, even if government officials working in field offices are permitted some latitude in implementing rules. Forms of decentralization that go beyond deconcentration usually involve local populations and representatives in creating rules at multiple levels. Our four case studies illustrate what happens when central governments do not relinquish some powers of collective and constitutional level decision making to lower level actors in managing resources. In such cases, activities and outcomes related to forest management do not change.

Classes of Property-Rights Holders and Outcomes

In examining the set of property rights listed above—withdrawal, management, exclusion, and alienation—it is clear that forms of ownership do not hinge on a single type of property right. Private property is frequently defined as a well-defined right of alienation. Then the right to alienate is assumed to be necessary to promote efficient use of resources.17 It is also often assumed that property rights systems that include the right to alienation permit voluntary transfers of assets through market exchanges from lower valued to highest valued uses.18 In our analysis, we do not find alienation to be an essential right. We find it more useful to define four classes of property rights holders, as shown in Table 1. Those who have the right to enter and to harvest some forms of products are “authorized users.” Operational rules (created in collective choice processes) determine whether there are constraints on the timing, technology and purpose of use, and quantity of resource units harvested.19 “Claimants” possess the operational rights of access and withdrawal plus a collective-choice right of management. The right to manage includes decisions concerning the construction and maintenance of facilities, the authority to devise limits on harvesting rights, and the ability to enforce limits.20 When local users are able to make their own rules concerning how to limit the timing, location, and technology of use, they can begin to devise rules that fit local circumstances.21 A major problem in trying to devise management rules for an entire country from a central governmental office is that the characteristics of diverse ecological zones in a country can vary dramatically. The effectiveness of management depends on an enormous number and range of variables that centralized decision making simply cannot take into account. The importance of local knowledge has been ignored in much of the forest policy of developing countries. The scientific AGRAWAL and OSTROM 491

Table 1 Bundles of Rights Associated with Positions Owner Proprietor Authorized Claimant Authorized User

Withdrawal ×× × × Management ×× × Exclusion ×× Alienation ×

Source: Adapted from Ostrom and Schlager (1992, 252).

forestry principles on which forestry policies are based are usually ineffective without local knowledge that can help identify the specific scientific findings that are relevant to a location or problem. Furthermore, when national bureaucracies are underpaid and understaffed, developing effective management plans for a large number of local forests becomes even more difficult. In many cases, therefore, the devotion to centrally designed, scientifically informed forest policy has meant open access to forests and degradation over time because local users do not have any powers or interest to devise rules that limit use or require monitoring. “Proprietors” hold the same rights as claimants with the addition of the right to determine who may harvest from a resource. Most of the property systems that are called common property regimes involve participants who are proprietors and have three of the above rights but do not possess the right to sell their management and exclusion rights even though they most frequently have the right to bequeath it to members of their family.22 Empirical studies have repeatedly found that—con- trary to the presumption that only those with full ownership rights can efficiently manage resources—the rights proprietors possess can be sufficient to make decisions that promote long-term investment and harvesting from a resource.23 Existing scholarship thus leads us to expect that decentralization programs empowering local users to be proprietors—even without the right to sell these rights to others—are likely to generate sufficient incentives to improve outcomes over time. Those who jointly hold proprietorship rights can manage a resource and also exclude those who are unwilling to contribute to the costs of management. A crucial problem to be solved, however, is how local users can gain sufficient confidence that the rights they begin to exercise after decentralization will not be taken away.24 “Owners” possess the right of alienation—the right to transfer a good in any way the owner wishes, but without harming the physical attributes or uses of other owners, and also all the rights held by a proprietor. An individual, a private corporation, a government, or a communal group may possess full ownership rights to any kind of good including a common-pool resource.25 The rights of owners, however, are never absolute. Even private owners have responsibilities not to generate 492 POLITICS & SOCIETY particular kinds of harms for others.26 Some policy recommendations for complete decentralization advocate that local users be given full ownership rights. This would be the strongest form of decentralization because local users could then do anything they wanted with their forests, including selling all the timber or the land itself. The foregoing discussion is by no means an advocacy of the position that only full ownership rights permit effective exploitation of resources. Nor are we sug- gesting that the likelihood of solving collective action problems related to resource management increases monotonically as users enjoy greater rights. Indeed, empirical studies have found that even proprietors have sufficient rights to make decisions that promote long-term investment and harvesting from a resource.27 Especially where common-pool resources are concerned, studies of inshore fisheries, self-organized irrigation systems, forest user groups, and groundwater institutions show that proprietors can develop strict boundary rules to exclude noncontributors, establish authority rules to allocate withdrawal rights, devise methods for monitoring conformance, and use graduated sanctions against those who do not conform to rules.28 The outcomes associated with the decentralization of different levels of property rights depend in significant measure on the certainty of users about their rights and the nature of governance arrangements to protect users’ rights. Many decentralization proposals assign no more than the operational-level right of being authorized users to those whom the program is supposed to benefit. All other significant operational and collective choice rights continue to be held by government officials, with all the attendant uncertainties that their opaque decision-making processes can generate for users. Even when government officials work in a local office instead of the nation’s capital, they may not have a long-term interest in sustaining the resource. Obtaining at least some rights to continued use encourages local users to perceive long-term interests in a local resource. Yet, such limited property rights do not establish strong incentives for sustainable management. Without the collective choice right to manage a resource, local users cannot consider and decide on various ways of growing and planting seedlings, thinning noncommercial trees for use as firewood, restricting the grazing of cattle in a forest, and so on. Without the collective choice right to exclude others, local users will fear that any effort they make to limit harvesting will benefit others who can then assert a future right to harvest. The absence of constitutional choice rights means that existing rights of local users can be taken away by distant power-holders without consultation. Therefore, decentralization can be said to have occurred only when governments devolve property rights over resources that conform to the collective and constitutional choice levels. Simply granting rights to undertake operational level actions is insufficient to justify claims of decentralization.29 AGRAWAL and OSTROM 493

DECENTRALIZATION OF FOREST CONTROL IN INDIA AND NEPAL

Decentralization of forest resources around the world is occurring for the most part under the general rubric of community-based conservation, where communities and their representatives gain varying degrees of collective control over forests.30 At one end of the continuum, we can distinguish those circumstances whereby national governments relax control sufficiently to allow local users rights corresponding to those of the proprietor or the owner. At another end are initiatives that permit users greater rights of access and use (authorized user) but few claimant or proprietorial rights. In the middle would be a host of other situations in which local residents may be allowed some managerial or decision-making rights but in a highly attenuated fashion. The following case studies illustrate different coalitions of actors involved in pressing for decentralization policies, variations in forms of decentralization, and differences in outcomes related to forest management. In Kumaon, India, villagers mobilized against the efforts of the British colonial state to appropriate forests and won significant property rights to become proprietors of large areas of forests through community based Forest Councils. The second study shows how, in the 1980s, the central Indian government and many provincial governments initiated a decentralization initiative known widely as the Joint Forest Management Pro- gram. Despite much wider reach, the institutional results of Joint Forest Manage- ment efforts have lagged far behind what a colonial government achieved (or was forced to undertake) in the 1920s and the 1930s in Kumaon. The third study examines how in Nepal’s Middle Hills, widespread devolution of forested areas in the 1990s under the Community Forestry program created thousands of active forest user groups. But in the Terai region of Nepal, where the government has sought to decentralize control over national parks and protected areas by involving residents in the Parks and People Program, the results have fallen far short of what might be called effective decentralization. The four case studies together illustrate the elements we have highlighted as necessary to analyze decentralization. Pres- sures for decentralization have varying levels of support from central government actors, local users, and international donors. The rights devolved to users range from those of an authorized user to a proprietor. The outcomes range from no change in use patterns and decision making to significant success at the local level. The cases provide an initial indication that decentralization is likely to be implemented more successfully when local users mobilize to support it and when they gain at least proprietorship rights.

Decentralization and Forests in Kumaon, India

The forest councils of Kumaon were created in 1931 through an act of decentralization by the British colonial state. They constitute perhaps the oldest surviv- 494 POLITICS & SOCIETY ing examples of formal collaboration between communities and the state to manage resources. Local collective action by villagers and bureaucratic politics within the state were critical in the formation of the forest councils.

Origins, Actors, and Their Interests Between 1911 and 1917, the British colonial state transferred more than 3,000 sq. km of forests in Kumaon to the Imperial Forest Department (Kumaon Forest Grievances Committee [KFGC] 1921). The government had made a number of inroads between 1865 and 1910 to curtail progressively the area of forests that local communities used. Its incursions at the turn of the century, however, raised the special ire of villagers because the elaborate new rules specified strict limits on grazing rights, restricted use of non-timber forest products, prohibited the extension of cultivation, increased the labor extracted from the villagers, and aug- mented the number of forest guards. The new laws goaded villagers into widespread protest. The incessant, often violent, protests forced the government to appoint the KFGC to look into the local “disaffection.” Comprising government officials and local political leaders, the KFGC examined more than 5,000 wit- nesses from all parts of Kumaon. It used the resulting evidence to make two major recommendations: (1) dereserve the larger part of the newly created Reserved Forests and (2) lay the foundations for establishing community forests that would be managed under a broad set of rules framed by the government but for which villagers themselves would craft the specific rules for everyday use. The government took both recommendations seriously. At first, it reclassified Reserved Forests that had been taken over by the Forest Department between 1911 and 1917 into Class I and Class II forests. Class I Reserved Forests, about half of the total, were all transferred to the Revenue Department and, in time, could come to be controlled by villagers. Class II Reserved Forests remained with the Forest Department. The government also passed the Forest Council Rules of 1931, permitting village residents to create forest councils and to bring under their own control forest lands that had been transferred to the Revenue Department as Class I Reserved Forests and Civil Forests. The division of forests into two categories—Class I/Civil Forests under the control of the Revenue Department and Class II Forests under the control of the Forest Department—resulted in part from the interdepartmental rivalry that was sparked into being by the creation of the Kumaon Forest Department in 1868, and by the passing of a huge swath of territory under its control in the name of the protection of forests. The increasing revenues from timber auctions soon rivaled and outgrew the tax revenues from land. The transfer of all the Class I forests to the Revenue Department was the outcome of a bureaucratic struggle. Protests by villagers for greater access to and use of their forests fed into this struggle. Villager protests helped the case for the transfer of a significant proportion of territory AGRAWAL and OSTROM 495 back to the Revenue Department, and from the Revenue Department to villagers in the form of forests managed by village level councils.

Degree of Decentralization of Property Rights Nearly 3,000 forest councils today formally manage and control about a quarter of the forests in the three districts of Kumaon: Nainital, Almora, and Pithoragarh. The Forest Council Rules of 1931, as amended in 1976, specify state-defined limits to local autonomy. Villagers cannot clear fell the forest or impose fines beyond a specified amount. They can raise revenues only through certain limited sources, and they must take recourse to established legal procedures to resolve conflicts. Conflicts over interpretation and application of rules are resolved by district and provincial level revenue or judicial authorities. Collec- tively, the Forest Council Rules constitute more a framework for the management of forests than a defining straitjacket. Rural residents, through their elected forest councils, possess substantial powers to create concrete restrictions to prevent certain types of forest use and facilitate others. Villagers vote to elect between five and nine council members and the council leader. Councils in many of the villages meet frequently. Their members discuss, craft, and modify specific rules that govern withdrawal of forest products and create monitoring and sanctioning mechanisms in an effort to enforce the Forest Council Rules as well as locally specific rules crafted by councils. Councils select guards, fine rule-breakers, manage finances, and maintain records of their meetings, accounts, and local rule infrac- tions. In many cases, households contribute in cash or kind to pay the council- appointed guards. The Forest Council Rules also provide support to the councils from the Reve- nue and the Forest Departments to facilitate rule enforcement and the maintenance of vegetation in the forests. Elections to the forest council are held under the supervision of the forest council inspector, an employee of the Revenue Depart- ment. The council is expected to keep records of meetings and maintain accounts. Whereas Revenue Department officials underwrite the enforcement of rules, the Forest Department coordinates the commercial harvest of forest products from community forests and provides technical assistance to develop them. Before the council can sell any of its timber or resin, it must seek approval from the relevant authorities in the Forest Department. Like the interactions with the Revenue Department officials, these can take a long time because of other duties that receive greater priority. The above description of the rights to forest management in Kumaon shows that the rural residents of Kumaon not only have the rights to access and use local forests but also can exercise claimant and proprietor rights. The forest councils’ ability to harvest firewood for commercial purposes and their access to markets for timber are mediated by the Forest Department. But otherwise, the councils control the use of their forests to a large degree. 496 POLITICS & SOCIETY

Outcomes Existing studies of forest councils’effectiveness in protecting forests show the importance of managerial responsibilities related to exclusion, monitoring, and enforcement.31 The forest councils can be seen as locally situated partners in the management of forests, subordinate to the employees of the Forest and the Reve- nue Departments but with substantial control over local management. Their asymmetric relations with government officials cast the officials into the role of arbiters in case of disputes between villagers and forest council officeholders. Forest users can also question the authority of the councils implicitly by not limiting their harvests of forest resources. They also do so more explicitly by contest- ing the fines imposed by the councils. In each of these situations, the councils need to invoke the cooperation of government officials, demonstrating their links to the state, their weaker position in this managerial relationship, and their relative autonomy in everyday management. The condition of vegetation on many council-managed forests compares favorably with the best Reserved Forests in the hills that are under the control of the Forest Department.32 A survey of more than 300 councils in the early 1990s showed that nearly half of the councils had forests that they rated as being in good to very good condition.33 Such conservation of forests, under conditions of acute scarcity, is especially impressive because council-managed forests supply villagers with anywhere between 25 to 50 percent of their requirements of fodder, firewood, and construction timber, a level that is far higher than what villagers get from Reserved Forests.34 Villagers’lives would be far closer to subsistence levels were they to be denied access to their forests, as was so graphically demonstrated at the beginning of the previous century. The council-managed forests are in far better condition than those managed by the Revenue Department. Overall, the experiment to decentralize forest management in Kumaon seems to have been a significant success.

Joint Forest Management in India

Since the founding of the Forest Department in several states of India in the 1860s and the passage of the Indian Forest Acts in 1878, the Indian state has attempted to define and allocate property rights over forests ever more precisely. The goal of more precise definition has been the generation of higher revenues by asserting and implementing higher levels of control over valuable timber resources. Where attempts to control forests have led to social unrest or protests, however, the Indian state has also shown itself capable of retreating and sharing benefits with new claimants. The forest councils in Kumaon, described above, are a well-researched example, but other instances are available from Bengal, Madhya Pradesh, and Gujarat.35 AGRAWAL and OSTROM 497

Origins, Actors, and Their Interests Joint Forest Management, with its objective of better management of forests through cooperation between the Forest Department and village communities, marks an important shift in the strategy of environmental conservation in India.36 The impetus for a nationwide program that could lead to an altered relationship between the state and village residents has come from several sources. Among these are a sustained critique of national forest policy by environmentalists, examples of successful community-based forest management in the states of West Ben- gal and Orissa, the increasing difficulty that the Forest Department faced in enforcing coercive forest conservation, efforts of many NGOs and social movement actors to influence government policy toward greater participation, and pressures and monetary inducements from multilateral and bilateral donors like the Ford Foundation, Swedish International Development Cooperation Agency (SIDA), and the World Bank.37 One of the main reasons why the Indian central government has been willing to accept communities into the management process has been the limited success of the Forest Department in protecting forests under its control. For example, in the early 1980s, satellite pictures of vegetation cover revealed that although the forest department controlled about 25 percent of India’s land, only about 11 percent had significant cover. The different state governments have had other reasons as well, among them, substantial aid from donors for community-based conservation. The National Forest Policy of 1988, government orders by many states, and the Government of India policy instructions of 1990 are the formal instruments that have launched Joint Forest Management.38 By 2000, 25 of 28 Indian states had passed resolutions to encourage the setting of village-level forest protection committees.

Degree of Decentralization of Property Rights The primary mechanisms of decentralization are village-level Forest Protec- tion Committees. They can either be constituted afresh, or Forest Department officials can register existing committees that predate Joint Forest Management. Offi- cials are also empowered to dissolve the committees.39 Because of the substantial powers forest officials have in forming, registering, and dissolving local committees, many villagers perceive the process to be arbitrary and biased. Some communities that organized informally to protect local forests have resisted joining the Joint Forest Management Program because they fear that the standardized approach of Joint Forest Management will not allow them to continue using rules that they found to be successful in their efforts to protect local forests.40 Members of forest committees gain nontimber forest products from their village forests and a share of the proceeds from the sale of timber once forests have matured. Villagers do not have the right to harvest timber products by themselves, 498 POLITICS & SOCIETY unless the wood is dry or has fallen to the ground. In some cases, their share of the proceeds from timber harvests is limited and has been reduced after successful protection. The degree to which villagers can frame their own rules of use and management is significantly shaped by Forest Department officials, and interventions in day-to-day management of village forests are common.41 Villagers have the responsibility to protect trees and vegetation and the ability to allocate nontimber benefits within the community. But where more significant issues are at stake—such as transfer of power to villagers, formation of forest protection committees, share in timber sales, or negotiation of rule changes—villagers are highly constrained in their ability to act.42

Outcomes There are no comprehensive assessments of the results of Joint Forest Manage- ment across Indian forests.43 According to some recent estimates, nearly 53,000 forest protection committees in India cover approximately 18 percent of the total forest area.44 Overall, the condition of Indian forests has improved in the past decade, despite significant population growth, but it is difficult to say how much, if any, of the improvement is a result of the Joint Forest Management initiative. The effects of Joint Forest Management on villager livelihoods is clearer, despite remarkable variations in the available figures. Joint Forest Management has provided millions of villagers more legitimate access to important livelihood resources and has given them a stake in environmental conservation. Significant and well-founded criticisms of Joint Forest Management point to the limited share of rural households in proceeds from sales of forest timber, but even critical evaluations of Joint Forest Management concede that rural residents have gained economically from participating in Joint Forest Management programs.45 Thus, at least at the local level, rural poor have benefited from a relaxation of state control over forest resources. Increased access to nontimber resources has occurred in tandem with continuing efforts by the Forest Department to control silvicultural aspects of forestry and commercial revenues from sale of timber. Within communities, participation and allocation of benefits vary greatly by caste, gender, class, and occupation. For example, because of membership rules that specify a single member from a house- hold to the forest protection committee, women often have little say in the limited rights of management that villagers exercise. Finally, an accurate assessment of the program is difficult because of conflicting objectives of foresters who most often see it as a means to protect forests and of villagers who view it as a solution to problems of shortage of firewood and fodder or as a way to improve income.

Community Forestry in Nepal

Perhaps the first systematic effort to centralize forest management in Nepal took place in 1957 when the government nationalized all forests. This assertion of AGRAWAL and OSTROM 499 control was cemented through a series of measures between 1961 and 1970 when the state tried to curtail even the use rights of rural residents in forests. In the absence of effective monitoring and enforcement systems, however, the new laws had perverse effects. They undermined existing local systems of management and led to widespread deforestation as people came to view forests as state property.46 The overwhelming evidence on deforestation showed that the existing policy needed rethinking.

Origins, Actors, and Their Interests Today Nepal is often seen as a leader among developing countries in setting conservation goals and priorities and in creating progressive programs and legislation related to resource management and conservation.47 New steps toward decentralization of forest control began in the late 1970s. The precursors of current community forestry legislation were the Panchayat Forest Rules of 1978 and the Community Forestry program of 1980. The limited conservation objectives of these initiatives were revised when the government realized that deforestation was approaching epidemic proportions. The pace of reforms accelerated with the widespread movement for democratization and the restoration of democracy after 1990. The current framework for community forestry legislation is represented by the Master Plan for the Forestry Sector in 1989, the Forest Act in 1993, and new Forest Regulations of 1995. Under the impact of these new pieces of legislation, the area of forests managed by local user groups and the number of these groups have increased exponentially.

Degree of Decentralization of Property Rights The major objectives of the new legislation are to provide forests to willing community groups, especially in the hill areas, and establish and promote community plantations in open and degraded areas. Community user group members are identified by the District Forest Officers. These user groups then prepare their own constitutions that govern day-to-day functioning and management. Follow- ing the demarcation of a forested area that can be handed over to a community, a five-year operational plan is prepared for the forest. User groups frequently play a direct role in preparing and implementing the plan.48 The District Forest Officer can hand over any part of a national forest to a user group in the form of a community forest, entitling it to develop, conserve, use, and manage the forest, and to sell and distribute forest products by independently fixing the price in the market. User groups can thus legally use their forests for subsistence, cultivating nontimber forest products, growing trees, and harvesting forest products for commercial processing and sale. Users are not permitted to clear the forest for agricultural purposes. But control over commercial profits from sale of timber products is already a major departure from forestry policies around the world. 500 POLITICS & SOCIETY

Executive committees of ten to fifteen members are elected by the general membership of the forest user groups. They undertake most of the everyday tasks associated with the management of the community forest. These tasks include protection of the forest (either directly or by a guard the committees appoint), allocation of benefits from the forest, steps to improve the condition of vegetation cover, and sanctioning rule breakers. Rural residents in many areas have begun to generate substantial benefits from their community forests, including cash revenues. Revenues are not taxed, but user groups are required to spend 25 percent of all cash income on collective development activities.49

Outcomes By 1999, the new legislation had led to the formation of 8,500 community forest user groups comprising nearly a million households. These user groups were managing more than 6,200 sq. km of forests. This is about 10 percent of the total forest area of Nepal. Unofficial estimates of these numbers are even higher. New user groups are being formed at the rate of nearly 2,000 a year, and they are now active in seventy-four of seventy-five districts in Nepal. In some areas of Nepal hills, a slow reversal of earlier deforestation can also be witnessed.50 The nature of property rights that community forest user groups have received places them in the category of proprietors. They are able to use all the products from their forests, buy and sell in markets, manage how the forest is to be used, and finally, create collective choice level rules that affect the nature of management. At least in the Middle Himalaya in Nepal, where the Community Forestry initiative is the most widespread, rural households have registered significant increases in their reliance on local forests. But a potential problem is the question of succession. At present, most groups, mainly because they have been formed relatively recently, have the same leaders that were selected at the time of their creation. But as the groups grow older, issues of who will lead the group and how transitions will occur will become increasingly important. An important development in Nepal’s community forestry is the emergence of a nationwide federation of community user groups (Federation of Community Forestry Users of Nepal, or FECOFUN), which seeks to lobby the government on behalf of its members and to disseminate information about community forestry more widely.51 It has already led active protests against government signals that users’ rights to commercial profits from forests may not be available in the Terai region of Nepal (because Terai forests are commercially more valuable). Indeed, efforts by the government to limit commercial use of community forest products to only the hill regions of Nepal signify the limits of the willingness of forest departments to devolve control. They also demonstrate that in the absence of influence over constitutional level rights, the property rights of local users can be limited quite easily. AGRAWAL and OSTROM 501

The Parks and People Program in Nepal’s Terai

Decentralization of forest rights in Nepal’s Terai, especially in the buffer zone of the national parks, is a somewhat different story in comparison to Community Forestry. Beginning in the 1960s with the protection of wildlife in royal hunting reserves, Nepal has created an extensive network of fifteen national parks, wildlife areas, hunting reserves, and conservation areas that cover nearly 15 percent of the country’s total area. Of this network of protected areas, five are in the Terai. These five protected areas face the greatest problems related to conservation because each of them is in close proximity to significant human populations.

Origins, Actors, and Their Interests Nepal’s Department of National Parks and Wildlife Conservation came into being in 1980, with authority to administer the protected areas system in Nepal. It is part of the Ministry of Forestry, along with the Department of Forestry. The Parks and People Program was established after negotiations among officials in the Ministry of Forestry and the United Nations Development Program (UNDP). The area covered by the Program is the buffer zone of the five National Parks and Wildlife Reserves in Nepal’s Terai, and comprises nearly 2,000 sq. km with a population of more than 600,000 people. An important factor in the launching of the Program is support from the UNDP. Local residents were unhappy with the coercive forms of protection that were used for conservation of wildlife and forest resources in the protected areas. But there have been no systematic efforts at collective protests nor any attempts to pressure the government into changing its protected area-related policies.

Degree of Decentralization of Property Rights To address the human-nature conflicts in relation to protected areas, the Parks and People Program aims at three objectives: attempts to develop alternatives to the use of park resources for neighboring households, seeks to devise compensation mechanisms for local communities in exchange for their exclusion from resources upon which they relied prior to the formation of the protected area in question, and tries to create incentives for local populations to change their actions in the protected areas. Development of buffer zones that “separate a protected area from direct human or other pressures and provide valued benefits to neighboring rural communities”52 is a key element of the conservation strategy in the Parks and People Program. Parks and People authorities have attempted to create user groups in all the buffer zones of the protected areas. The Warden of a protected area and other officials in the Parks and People Program can constitute user groups in the buffer zone to coordinate the management of fallen trees, firewood, fodder, and other grasses. Members of the user groups can use some forest resources in the buffer zone for 502 POLITICS & SOCIETY subsistence and commercial benefits. For specified times during the year, buffer zone residents are permitted to enter the protected area itself and to harvest products such as thatch grass and firewood and, in some cases, to graze their animals. Typically, the period for which they can harvest thatch grass, used for roofing, varies between ten and fifteen days in a year. Rules about harvesting firewood and grazing animals are strict. Of the total income earned through sales of forest products, 30 to 50 percent can be used for community development in consultation with local agencies and community members. Parks and People Program officials had created approximately 400 community user groups in the buffer zones of the protected areas by the end of 1999. Although these groups have limited rights of access and use in the forests, they have little control over how forest resources in the protected area are allocated. Most rules continue to be crafted by protected area officials without the involvement of local residents. Thus, villagers in the buffer zone do not possess any rights of collective or constitutional choice to create rules at the operational level, nor are they involved in the enforcement of rules.

Outcomes The most visible outcome of the Parks and People Program has been an increase in the types and density of authorities that try to affect how users living in the buffer zone will live their lives. More forest department officials, development offices, projects, and formal user groups are now present and interact with each other more often. The politics and procedures that have produced these changes have primarily been born within the offices of the Nepali government, and foreign aid agency programs, rather than being prompted by local demands. Government and park officials and aid agency personnel have counted local interactions with residents living near protected areas as relevant mainly because they believe that consumption pressures generated by the poor have the potential to adversely affect protection of natural resources. As a result, the focus in this decentralization effort has been on strategies that would shift consumption pressures away from the protected areas. Officials have paid less attention to involving residents in the management of the protected areas, let alone transferring property rights corresponding to those of proprietors or owners. Decentralization through the Parks and People Program has mainly affected users at the operational level. They can now enter the parks and use some park resources such as fodder legally for a few weeks during the year. But they have little influence on management or conservation outcomes. Conflicts between residents and Park officials occur regularly and are unlikely to diminish without substantial change in existing institutional arrangements or reallocation of the property rights that residents currently exercise. Although buffer zone residents have now become authorized entrants and users from being illegal users, the actual effect of the Parks and People Program on rural livelihoods has been lim- AGRAWAL and OSTROM 503 ited. The condition of forests and wildlife inside the protected areas has remained the same. Indeed, in some areas, consumption pressures on protected areas may have increased because the economic projects of the Parks and People Program have facilitated agricultural intensification and may have encouraged rural residents to increase grazing and fodder harvests.53 According to the most recent evaluation of the Parks and People Program, those residing close to the boundaries of the protected areas continue to use local resources despite the efforts of officials to shift harvesting pressures elsewhere.54

COMPARING DECENTRALIZATION OF FOREST USE IN INDIA AND NEPAL

The nature of decentralization varies greatly in the four cases in terms of origins, actors involved, the extent to which subnational actors have come to exercise new property rights, and outcomes related to forests. Overall, the greatest levels of decentralization seem to have occurred in Kumaon. The Community Forestry program in Nepal is relatively recent in origin, but the formal aspects of the program—and many of the effective examples of its operation—involve substantial levels of decentralization. The Parks and People Program in Nepal, despite claiming participatory decentralization, has devolved such limited property rights that it can scarcely be classified as a case of decentralization. Joint Forest Manage- ment in India is spread over the largest area of the four cases and has highly variable outcomes, but overall, the Forest Department continues to exercise significant managerial control over local actors. Unlike the case of Community Forestry in Nepal, participants in Joint Forest Management have little control over how commercial forest products such as timber will be sold or how the proceeds will be allocated. We can infer that the decentralization of property rights in this case falls somewhere between the case of Community Forestry in Nepal and of the Parks and People Program in Nepal’s Terai. Table 2 summarizes the discussion of the four cases and highlights their differences. The extent of decentralization in these four cases is closely tied to their origins. High levels of local collective mobilization meshed with interdepartmental rivalries within the British colonial state to facilitate the birth of the forest councils in Kumaon. Protests by hill residents helped the Revenue Department gain control over a significant proportion of forestland. By creating new officials to supervise the forest councils, the Revenue Department also gained a measure of control over local activities. In the case of community forestry in Nepal and Joint Forest Man- agement in India, there were no significant local pressures for decentralization, but effective national lobbying groups existed. Furthermore, the existing government policy generated clear evidence that the Forest Department had to change its orientation drastically if environmental conservation was to be successful. In Nepal’s Terai, government officials were able to launch the Parks and People Pro- 504 Table 2 Comparing Decentralization of Forest Control in India and Nepal Axes of Comparison Forest Councils, Kumaon Joint Forest Management Community Forestry, Nepal Parks and People Program, Nepal

Nature of macrolevel politics No donor pressures; high High donor pressures; some High donor pressures; some High donor pressure to create the around decentralization interest among state actors interest in decentralization interest in decentralization program; no interdepartmental in decentralization among state actors among state actors rivalries Nature of local level High levels of local mobilization; Donor pressures match some Donor pressures and limited No collective action at local level; collective action and resonates with Land Revenue interests of state government; collective action locally little connection with higher its relationship with Department actors in low levels of local collective matches some central levels of politics macrolevel politics government action match the interest of government interests; some central government emergence of new factions actors against decentralization Property rights Rights to withdrawal Granted to local groups, some Granted to local groups, some Granted to local groups, but Granted to local grou ps, very limits limits future uncertain limited Rights to manage Granted to local groups, some Granted to local groups, Granted to local groups, some Not available to local groups limits; limited supervision by significant limits in practice; limits Forest Department active Forest Department intervention Rights to exclude Granted to local groups Granted to local groups, some Granted to local groups, some Not available to local groups limits limits Rights to transfer Not available to local groups Limited rights granted to Limited rights granted to Not available to local groups local groups local groups Outcomes Participation and High levels of decentralization Middling levels of decentralization High levels of decentralization Almost no decentralization; decentralization and participation and participation and participation limited participation Resource use patterns Decentralization limits resource Decentralization limits resource Decentralization limits resource Institutionalization is limited, and use and improves resource use and improves resource use and improves resource local use patterns remain same condition condition in some cases condition in some cases AGRAWAL and OSTROM 505 gram in the name of decentralization. The chief motivation behind this program, however, seems to be that funds were available from an international donor. Local protests or evidence of drastic failure had little role in the initiation of the program. The counterintuitive comparative lesson from the four case studies about the origins of decentralization seems to be that local demands are not critical in the launching of a decentralization initiative. Such demands were present only in Kumaon, but decentralization programs were initiated in all the cases. Continued local pressures and lobbying from actors outside the state machinery, however, seem to be very important if decentralization of property rights is to continue and be effective over time. The one case in which local demands for decentralization are virtually absent—the Parks and People Program in Nepal’s Terai—is also the case in which local users have found their status as resource users to have changed the least. The absence of initial and continuing demands for more control over resources and the lack of government officials who would gain from greater decentralization of control has meant that powerful officials have found it easy not to give up managerial control over resources in protected areas. It is also not surprising that Kumaon villagers and members of community forestry user groups in Nepal have gained significantly wider latitude in the property rights they exercise over forests in comparison to the rights exercised by villagers in the buffer zones of Nepal’s Terai. In Kumaon, villagers have clear rights to access and use forest products and almost complete control over management of their forests. The state has placed some limits on managerial discretion, such as prohibiting clear felling or requiring the involvement of the Forest Department in commercial sales, but in daily management, villagers have a free hand. They are entitled to exclude others from their forests and can even buy, sell, and bequeath to others the use and access rights they hold. In addition, the presence of government officials such as the forest council inspector, and the numerous other existing officials to whom they can appeal, also helps in the effective implementation of their rights over forests.55 The Community Forestry initiative in Nepal formally grants to villagers even more far-reaching property rights. In addition to all the rights that forest councils of Kumaon enjoy, Nepalese villagers can also buy and sell timber and retain the revenues for themselves. They are not required to involve the Forest Department or any other government officials in market transactions. But the initiative suffers from two limitations. Existing rules are unclear about selection of local leadership and succession. Over time, existing leaders can entrench their positions. Two, forest policies can change in Nepal without much notice because the government has not devolved constitutional level rights. Thus, a recent announcement that taxes will be imposed on revenue from forest products has raised genuine concerns about the nature of decentralization that Community Forestry embodies. 506 POLITICS & SOCIETY

Joint Forest Management in India grants to villagers the property rights of access and use, and management, but in actual functioning of the forest protection committees, Forest Department officials intervene pervasively. In addition, no security of tenure exists for the committees since they must be registered by Forest Department officials who also possess the power to dissolve a committee. Because of these limitations, the degree of decentralization of property rights in Joint Forest Management conforms only to that of claimants. User groups in the buffer zones of the protected areas in Nepal can only claim to have somewhat attenuated use and access rights. They have almost no managerial discretion and few powers of exclusion or alienation. Collective and constitutional choice authority has been retained by the Department of National Parks and Wildlife and the Parks and People Program officials. The differing levels of decentralization in the four cases have had clear results in terms of the ability of local residents to use and manage local resources. Mem- bers of the Kumaon forest councils in Kumaon and Community Forestry user groups in Nepal have witnessed the greatest change in their ability to use and manage forest resources. Like them, villagers who participate in Joint Forest Manage- ment can also manage their forests, exclude other users, and determine how their forests will be used. But the Joint Forest Management initiative gives rural residents fewer rights over the commercial exploitation of forests in comparison to Forest Council Rules or the Community Forestry Program in Nepal. The evidence on outcomes related to changes in forest conditions is prelimi- nary. But the existing evidence suggests that the cases in which there has been more effective decentralization of property rights are also the ones in which forest conditions have improved more. Recall that in Kumaon, the Forest Department had kept the best land to itself as Class II forests. It transferred only the somewhat more fragmented and often more degraded patches of forests to the Revenue Department for eventual management by forest councils. Yet,today, the condition of forests on land held by the forest councils compares favorably with that of For- est Department–managed lands.56 In the Nepal hills, where only two decades ago many scholars were raising an alarm about the degradation of forests,57 improvement in the condition of vegetation is already apparent. The Parks and People Pro- gram has been in existence only for five years, and therefore, it is harder to judge whether it has made a significant difference in forest cover. In the case of Joint Forest Management, although there have been significant improvements in some areas, the sheer scale of the initiative makes it harder to make general statements about its vegetation-related outcomes in the absence of systematic large scale studies. The exercise of property rights that resource users have come to hold depends, in significant measure, on the larger context of governance arrangements within which their rights are located. Thus, governments can easily launch decentralization initiatives that promise to give users wide latitude in using and managing AGRAWAL and OSTROM 507 resources. On the other hand, the same governments can pull the teeth of these initiatives by not providing administrative and enforcement support that users need to exercise rights effectively. For example, users may have the de jure right to protect forests. These rights would have no meaning, however, if the users are poor, have no resources to hire guards, have no authority to raise revenues by selling forest products, and existing mechanisms to resolve disputes are expensive and cumbersome. These conditions are all present in the Parks and People Program. By contrast, the Kumaon forest councils also face the problem of a cumbersome legal system but can raise revenues locally and can appeal to forest council inspectors and officers that the colonial state appointed specifically to supervise the working of the councils. In the case of the Community Forestry Program in Nepal, and the Joint Forest Management initiative, Forest Department officials are supposed to help local forestry organizations when problems arise. Such help is especially crucial when conflicts arise between groups. But in neither case do the existing policies make provisions for how intergroup disputes at various levels are to be addressed. The effectiveness with which local organizations exercise their property rights depends then on how much support they receive from forestry officials empowered to interpret and enforce their rights.

CONCLUSION

This article began by posing two questions: What role do local populations play in ensuring devolution of different categories of property rights? and What types of property rights must be devolved to local populations for effective decentralization? The evidence we have on the first question is somewhat mixed. In one of our cases—forest councils of Kumaon—the effective mobilization of local actors was instrumental in the initiation of decentralization. In the case of the Parks and Peo- ple Program, there were no local demands, and there was no effective decentralization. But in the other two cases—Community Forestry in Nepal and Joint For- est Management in India—we witness significant decentralization without local actors mounting a widespread social movement or nationwide agitation. We must conclude that active participation of local actors is not a prerequisite for decentralization programs to be launched. However, local groups do actively have to pursue opportunities opened up by decentralization reforms, or else they are likely to find that decentralization has been retracted (as in the case of Community Forestry in Nepal) or discover that their rights are greatly limited by active interventions of more powerful officials (as in the case of Joint Forest Management). It is notewor- thy that in the case of the Parks and People Program, pressures for decentralization were generated from outside the government, through the intervention of an international donor agency. In the case of both Joint Forest Management in India and Community Forestry in Nepal, there were sufficient numbers of actors within 508 POLITICS & SOCIETY the government who were active in pursuing decentralization, even if no clear pressures from local populations were visible. In both cases, international NGOs have also provided extensive financial assistance for decentralization of forest management. However, it is a different story when we come to maintenance of decentralized property regimes. For continued existence of decentralized control over resources, active participation of local actors seems to be quite necessary. Nepal’s Community Forestry Program has seen the emergence of a national-level networking organization, FECOFUN, that is prepared to voice protests against efforts of the Ministry of Forestry to limit the territorial expansion of existing decentralization reforms. Neither the forest councils of Kumaon nor the protection committees formed via the efforts of Joint Forest Management in India have found a similar organization or representation. The emergence of such macrolevel networks and organizations is likely crucial to ensure that decentralization processes, once initiated, are not retracted. The evidence on the second question is more conclusive. The case studies con- firm that it is necessary for local users and their representative institutions to possess property rights that transform them into claimants and proprietors to achieve effective decentralization. The delegation of only the operational-level rights of access and use of forests and forest products does not produce much change in either the condition of vegetation or the relationships between state and community actors. The decentralization of control over forests in Kumaon has meant that local residents in the region have come to interact in very different ways with local and government officials as compared to the past. There is significant enforcement of locally crafted rules. Variationsin the conditions of forests are primarily a result of microlevel differences in institutional and bio-physical factors. A similar story can also be told for Nepal’s Community Forestry Program. As rural residents have begun to mange their forests actively, the nature of their relationships with government officials has changed. In the case of Joint Forest Management, the results are sufficiently variable that a blanket assertion of improvement would be hasty at best. In Nepal’s Terai, villager participation in the Parks and People Program, and their ability to make decisions about the use and management of forests, are overshadowed by state officials. The findings of our empirical study, thus, match the theoretical expectation that when local users do not exercise significant control over collective and constitutional-level choices related to rule design, management, and enforcement, the impact of decentralization is limited. To create an impact, governments need to allow local users and their representatives at least the rights to manage resources and make decisions about resource use and exclusion. At the same time, effective implementation of decentralization reforms and enforcement of local property rights depend, at least in part, on the governance arrangements within which specific decentralization reforms are located. The effective enforcement of decentral- AGRAWAL and OSTROM 509 ization in Kumaon is attributable in no small measure to the ability of forest council members to appeal to Revenue and Forest Department officials when conflicts arise. Such appeals for help and avenues of enforcement are not available to users living in the buffer zone of protected areas in Nepal’s Terai. It is worth pointing out in passing that the conclusions of our comparative analysis do not change if we restrain analytical focus to a single country. The findings of the study may, therefore, be more widely applicable in other South Asian countries as well. To conclude, decentralization reforms imply changes in property rights over resources that gain local users’rights and capacities to make operational rules. As important, decentralization initiatives should allow users to make collective and constitutional choices. Successful decentralization initiatives related to forests are difficult to accomplish without the existence of state actors who see some of their interests being fulfilled by decentralization processes. Furthermore, the likelihood of success is enhanced by promoting the conditions that generate self-organization among local groups. As the examples of the Kumaon forest councils and Community Forestry in Nepal make clear, only with some level of organization and pressure from local groups are decentralization processes likely to move beyond the control of elite actors.

NOTES

1. William Dillinger, Decentralization and Its Implications for Urban Service Deliv- ery. Urban Management Program Discussion Paper No. 16 (Washington, DC: The World Bank, 1994), 8. 2. The global pursuit of decentralization, even if it appears novel, has had several precursors. Beginning in 1917, four waves of decentralization can be discerned in Francophone West Africa according to Jesse Ribot, “Decentralization, Participation, and Accountability in Sahelian Forestry: Legal Instruments of Political-Administrative Con- trol,” Africa, 69 (August 1999): 23-65. Arun Agrawal and Jesse Ribot, “Accountability in Decentralization: A Framework with South Asian and West African Cases,” Journal of Developing Areas, 33 (summer 1999): 495 identify at least three waves of decentralization in south Asia since the mid-1800s. 3. There is a close relationship between the willingness of governments to decentralize powers to lower level decision makers in a polity and the ability of citizens to influence decision making by participating more intensively in governance, even if the relationship is hard to specify precisely. The ability of governments to foster participation has been termed “state-fostered civic organizing” by Rebecca Abers, “From Clientalism to Cooper- ation: Local Government, Participatory Policy, and Civic Organizing in Porto Alegre, Brazil,” Politics and Society, 26, no. 4 (December 1998): 511. 4. See the essays in Tariq Banuri and Frederique Apffel Marglin, eds., Who Will Save the Forests? Knowledge, Power, and Environmental Destruction (London: Zed Books, 1993); and Michael Kaufman and Haroldo Dilla Alfonso, eds., Community Power and Grassroots Democracy: The Transformation of Social Life (London: Zed Books, 1997). A number of scholars have also made similar arguments about development in book-length studies. See Julie Fisher, Non Governments: NGOs and the Political Development of the Third World (West Hartford, CT: Kumarian Press, 1998); and Richard C. Crook and James 510 POLITICS & SOCIETY

Manor, Democracy and Decentralisation in SouthAsia and West Africa: Participation, Accountability and Performance (Cambridge, UK: Cambridge University Press, 1998). William Ascher, Why Governments Waste Natural Resources: Policy Failures in Developing Countries (Baltimore, MD: The Johns Hopkins University Press, 1999), 269-71 reviews the natural resource management policies of a host of governments, and recommends decentralization to private and communal actors. 5. For a careful political study of decentralization that highlights the territorial aspects of the exercise of power, see Brian C. Smith, Decentralization: The Territorial Dimension of the State (London: George Allen and Unwin, 1985), especially pp. 1-6. 6. Peter Evans, “A Virtuous Cycle Model of Rural-Urban Development: Evidence from a Kenyan Small Town and its Hinterland,” Journal of Development Studies, 28, no. 4 (April 1992): 135-53; Philip Mawhood, Local Government in the Third World:The Experi- ence of Tropical Africa (Chichester, UK: Wiley, 1983); and James S. Wunsch and Dele Olowu, The Failure of the Centralized State: Institutions and Self-Governance in Africa (Boulder, CO: Westview, 1990). 7. M. Goodwin, S. Duncan, and S. Halford, “Regulation Theory, the Local State, and the Transition of Urban Politics,” Environment and Planning D: Society and Space, 11, no. 1 (February 1993): 67-88; and R. Rhodes, Beyond Westminster and Whitehall: The Sub-Central Governments of Britain (London: Unwin Hyman, 1988). 8. This article brackets the question of when local groups will self-organize. For a careful analysis of this question, see Elinor Ostrom, Governing the Commons: The Evolu- tion of Institutions for Collective Action (New York: Cambridge University Press, 1990), chap. 6; and Elinor Ostrom, “Reformulating the Commons,” In J. Burger, E. Ostrom, R. B. Norgaard, D. Policansky, and B. D. Goldstein, eds., Protecting the Commons: A Frame- work for Resource Management in the Americas (Washington, DC: Island Press, 2001), 17-41. 9. For definitions of these terms and their relationship to decentralization, see Dennis Rondinelli, James McCullough, and Ronald Johnson, “Analyzing Decentralization Pol- icies in Developing Countries: A Political-Economic Approach,” Development and Change, 20, no. 1 (January 1989): 76; Elinor Ostrom, Larry Schroeder, and Susan Wynne, Institutional Incentives and Sustainable Development: Infrastructure Policies in Perspec- tive (Boulder, CO: Westview, 1993), 164-68; Michael Maniates, “Organizational Designs for Achieving Sustainability: The Opportunities, Limitations, and Dangers of State-local Collaboration for Common Property Managemen,” (presented at the First Annual Meeting of the International Association for the Study of Common Property, Durham, NC, 27-30 September 1990), 1; Devraj Dahal, The Challenge of Good Governance: Decentralization and Development in Nepal (Kathmandu, Nepal: Center for Governance and Development Studies, 1996), 15; and Paul Smoke, “Local Government Fiscal Reform in Developing Countries: Lessons from Kenya,” World Development, 21, no. 6 (June 1993): 901-3. It is clear that a very large literature elaborates the meanings of decentralization. 10. Mawhood, Local Government; Smith, Decentralization, 1-3. 11. As John R. Commons, Legal Foundations of Capitalism (Madison: University of Wisconsin Press, 1968) has observed, a property right is an enforceable authority to undertake particular actions in a specific domain. 12. Daniel Bromley et al., eds., Making the Commons Work: Theory, Practice and Pol- icy (San Francisco: Institute for Contemporary Studies Press, 1992). 13. According to a survey of Forest Departments by the Food and Agriculture Organi- zation, more than 50 countries claim to be moving toward involving user communities in some form in the management of forest resources. See Food and Agriculture Organization, AGRAWAL and OSTROM 511

“Status and Progress in the Implementation of National Forest Programmes: Outcomes of an FAO Worldwide Survey,” mimeo (Rome, Italy: FAO, 1999). 14. For a detailed discussion of these different types of rights, see Edella Schlager and Elinor Ostrom, “Property Rights Regimes and Natural Resources: A Conceptual Analy- sis,” Land Economics, 68, no. 3 (August 1992): 249-62. In response to referees’requests to keep this analysis as simple as possible, we do not discuss a fifth right identified in the above article—the right of access. A decentralization policy gives local users effective access rights when they receive some rights of harvest. But access as a separable right is important if one is analyzing parks and other institutional arrangements where access alone is granted to some actors. 15. Elinor Ostrom, Roy Gardner, and James Walker, Rules, Games, and Common-Pool Resources (Ann Arbor: University of Michigan Press, 1994), chap. 2. 16. Analysis of decisions as occurring at nested levels is developed at greater length in James Buchanan and Gordon Tullock, The Calculus of Consent: Logical Foundations of Constitutional Democracy (Ann Arbor: University of Michigan Press, 1962); George Brennan and James Buchanan, The Reason of Rules (Cambridge, UK: Cambridge Univer- sity Press, 1985); and Larry Kiser and Elinor Ostrom, “The Three Worlds of Action: A Metatheoretical Synthesis of Institutional Approaches,” in Elinor Ostrom, ed., Strategies of Political Inquiry (Beverly Hills, CA: Sage, 1982), 179-222. A concise treatment is available in Ostrom, Governing the Commons, 50-55. 17. See Harold Demsetz, “Toward a Theory of Property Rights,” American Economic Review, 57 (May 1967): 347-59. 18. Not all scholars agree. Bruce Larson and Daniel Bromley, “Property Rights, Exter- nalities, and Resource Degradation: Locating the Tragedy,” Journal of Development Eco- nomics, 32, no. 2 (October 1990): 235-62, show that much more information about the specific values of a large number of ownership parameters is necessary before judgments can be made concerning the efficiency of a property right. 19. Operational rights may be finely divided into quite specific and, sometimes, overlapping “tenure niches” that vary by season, by use, by technology, and by space as discussed by John Bruce, Louise Fortmann, and Calvin Nhira, “Tenures in Transition, Tenures in Conflict: Examples from the Zimbabwe Social Forest,” Rural Sociology, 58, no. 4 (1993): 626-42. Operational rules can permit authorized users to transfer access and withdrawal rights either temporarily through a rental agreement, or permanently when these rights are assigned or sold to others. 20. For studies of how collective management-related rights are exercised in relation to fisheries, see Bikash Raychaudhuri, The Moon and the Net: Study of a Transient Commu- nity of Fishermen at Jambudwip (Calcutta: Anthropological Survey of India, Government of India Press, 1980); for irrigation, see Shui Yan Tang, Institutions and Collective Action: Self-Governance in Irrigation (San Francisco: Institute for Contemporary Studies Press, 1992); and for forestry, see George Varughese and Elinor Ostrom, “The Contested Role of Heterogeneity in Collective Action: Some Evidence from Community Forestry in Nepal,” World Development, 29, no. 5 (May 2001); George Varughese, “Villagers, Bureaucrats, and Forests in Nepal: Designing Governance for a Complex Resource,” Unpublished Ph.D. Dissertation, Department of Political Science, Indiana University, Bloomington, 1999. 21. See Ostrom, Governing the Commons, 204-5; Ostrom, Schroeder, and Wynne, Institutional Incentives, 51-54. 22. The essays in several collected volumes show the importance of institutional regulation of common property at the local level. See Daniel Bromley, David Feeny, Margaret McKean, Pauline Peters, Jere Gilles, Ronald Oakerson, Carlisle Ford Runge, and James 512 POLITICS & SOCIETY

Thompson, eds., Making the Commons Work:Theory, Practice, and Policy (San Francisco: Institute for Contemporary Studies Press, 1992); and Bonnie J. McCay and James Acheson, eds., The Question of the Commons: The Culture and Ecology of Communal Resources (Tucson: University of Arizona Press, 1987). 23. In a series of studies of inshore fisheries, self-organized irrigation systems, forest user groups, and groundwater institutions, proprietors tended to develop strict boundary rules to exclude noncontributors, established authority rules to allocate withdrawal rights, devised methods for monitoring conformance, and used graduated sanctions against those who do not conform to these rules. See the empirical essays in Ostrom, Gardner, and Walker, Rules, Games, and Common-Pool Resources, 219-300. 24. J. Arnold and Gabriel Campbell, “Collective Management of Hill Forests in Nepal: The Community Forestry Development Project,” in Proceedings of the Conference on Common Property Resource Management (Washington, DC: National Research Council, National Academy Press, 1986): 425-54, point to the effect uncertainties of tenure produce among agricultural producers. 25. Robert A. Dahl and Charles E. Lindblom, Politics, Economics, and Welfare: Planning and Politico-Economic Systems Resolved into Basic Social Processes (New York: Harper, 1963). 26. See Demsetz, “Toward a Theory,” 350-52. 27. Frank Place and Peter Hazell, “Productivity Effects of Indigenous Land Tenure Sys- tems in Sub-Saharan Africa,” American Journal of Agricultural Economics, 75 (1993): 10-19 conducted surveys in Ghana, Kenya, and Rwanda to ascertain if indigenous land-right systems were a constraint on agricultural productivity. They found that having the rights of a proprietor as contrasted to an owner in these settings did not affect investment decisions and productivity. Some other studies from Africa have also found little difference in productivity, investment levels, or access to credit purely as a result of more rights. See Shem E. Migot-Adholla, Peter Hazell, Benoit Blarel, and Frank Place, “Indige- nous Land Rights Systems in Sub-Saharan Africa: A Constraint on Productivity?” World Bank Economic Review, 5, no. 1 (1991): 155-75; John W. Bruce and Shem E. Migot- Adholla, eds., Searching for Land Tenure Security in Africa (Dubuque, IA: Kendall/Hunt, 1994). In densely settled regions, however, proprietorship over agricultural land may not be sufficient for efficient exploitation as argued by Gershon Feder and David Feeny, “Land Tenure and Property Rights: Theory and Implications for Development Policy,” World Bank Economic Review, 5, no. 1 (1991): 135-53; and Gershon Feder, T. Onchan, Y. Chalamwong, and C. Hangladoran, Land Policies and Farm Productivity in Thailand (Bal- timore, MD: Johns Hopkins University Press, 1988). 28. See Arun Agrawal, “Rules, Rule Making, and Rule Breaking: Examining the Fit between Rule Systems and Resource Use,” in Elinor Ostrom, Roy Gardner, and James M. Walker, eds., Rules, Games, and Common-Pool Resources (Ann Arbor: University of Michigan Press, 1994), 267-82; William Blomquist, Dividing the Waters: Governing Groundwater in Southern California (San Francisco: ICS Press, 1992); Edella Schlager, “Fishers’ Institutional Responses to Common-Pool Resource Dilemmas,” in Elinor Ostrom, Roy Gardner, and James M. Walker, eds., Rules, Games, and Common-Pool Resources (Ann Arbor: University of Michigan Press, 1994), 247-65; Shui Yan Tang, “Building Community Organizations: Credible Commitment and the New Institutional Economics,” Human Systems Management, 13 (1994): 221-32; and Wai Fung Lam, Gov- erning Irrigation Systems in Nepal: Institutions, Infrastructure, and Collective Action (Oakland, CA: ICS Press, 1994). 29. We can point out in passing that decentralization of all the rights—use (or withdrawal), management, exclusion, and alienation—only facilitates self-organization. Not AGRAWAL and OSTROM 513 all local groups will self-organize to manage local forests sustainably just because central governments are willing to devolve authority. Self-organization by users requires over- coming local barriers to collective action as well. 30. For a comprehensive review, see Arun Agrawal and Clark Gibson, “Community and Conservation: Beyond Enchantment and Disenchantment,” World Development, 27, no. 4 (1999): 629-49. 31. See Arun Agrawal and Gautam Yadama, “How do Local Institutions Mediate Mar- ket and Population Pressures on Resources? Forest Panchayats in Kumaon, India,” Devel- opment and Change, 28, no. 3 (July 1997): 435-66. 32. E. Somanathan, “Deforestation, Property Rights, and Incentives in Central Himalaya,” Economic and Political Weekly, 26 (1991): 37-46. 33. Agrawal and Yadama, “How do Local Institutions,” 447. 34. See Arun Agrawal, “State Formation in Community Spaces? Decentralization of Control over Forests in the Kumaon Himalaya, India,” Journal of Asian Studies, 60, no. 1 (February 2001): 1-32. 35. For a discussion of changing colonial forestry policies in Bengal, see K. Sivaramakrishnan, Modern Forests: Statemaking and Environmental Change in Colonial Eastern India (Stanford, CA: Stanford University Press, 1999). Careful studies of forest history in Gujarat and Madhya Pradesh are available respectively in Ajay Skaria, Hybrid Histories: Forests, Frontiers, and Wilderness in Western India (Delhi: Oxford University Press, 1999); and Nandini Sundar, Subalterns and Sovereigns: An Anthropological History of Bastar, 1854-1996 (Delhi: Oxford University Press, 1998). 36. Shashi Kolavalli, “Joint Forest Management: Superior Property Rights?” Economic and Political Weekly, 30 (29 July 1995): 1933-38; and Ajit Krishnaswamy, “Sustainable Development and Community Forest Management in Bihar, India,” Society and Natural Resources, 8 (1995): 339-50. 37. Nandini Sundar, “Unpacking the ‘Joint’in Joint Forest Management,” Development and Change, 31, no. 1 (January 2000): 255-60. 38. Dolly Arora, “From State Regulation to People’s Participation: Case of Forest Man- agement in India,” Economic and Political Weekly, 29 (19 March 1994): 691-98. 39. Mark Poffenberger and Chhatrapati Singh, “Communities and the State: Reestab- lishing the Balance in Indian Forest Policy,” in Mark Poffenberger and Betsy McGean, eds., Village Voices, Forest Choices (Delhi: Oxford University Press, 1996), 58-59. 40. Rucha Ghate, “The Role of Autonomy in Self-Organizing Processes: A Case Study of Local Forest Management in India,” mimeo (Bloomington: Indiana University, Work- shop in Political Theory and Policy Analysis, 2000), 10. 41. Richard Mahapatra, “Community Forest Management: The Nepalese Experience,” Down to Earth, 8, no. 9 (September 2000): 1-10; See also the essays in Roger Jeffery and Nandini Sundar, eds., A New Moral Economy for India’s Forests? Discourses of Commu- nity and Participation (Sage: New Delhi, 1999). 42. Neeraj Kumar, “All is Not Green with Joint Forest Management in India,” Forests, Trees, and People Newsletter, 42 (2000): 46-50; Sundar, “Unpacking the ‘Joint,’” 275-77. 43. Several case studies available in Mark Poffenberger and Betsy McGean, eds., Vil- lage Voices, Forest Choices (Delhi: Oxford University Press, 1996), provide local assessments. See also the recent study by Arvind Khare, Madhu Sarin, N. C. Saxena, Subhabrata Palit, Seema Bathla, Farhad Vania, and M. Satyanarayana, Joint Forest Management: Pol- icy, Practice and Prospects (London: International Institute for Environment and Develop- ment, 2000) for a more recent, general account. 44. “State Forestry Action Programme,” Public speech by Inspector General of Forests, Indian Institute of Forest Management, Bhopal, Madhya Pradesh, 21 January 2000. 514 POLITICS & SOCIETY

45. Khare et al., Joint Forest Management, 2-8; Kumar, “All is not Green,” 48-50. 46. Arnold and Campbell, “Collective Management,” 440-44. 47. Joel Heinen and B. Kattel, “A Review of Conservation Legislation in Nepal: Past Progress and Future Needs,” Environmental Management, 16 (1992): 723-33. 48. As with any policy of this kind, the manner of implementation of the law is sometimes more arbitrary and less participatory than it is at other times. A 1995 study of 419 “chairpersons” of forest committees uncovered that “most of them did not know if they were members of a forest committee, or what they were expected to do.” Charla Britt, “For- estry and Forest Policies,” Working paper (Bloomington: Indiana University, Workshop in Political Theory and Policy Analysis, 2000), 22. 49. There is an initiative currently under consideration to tax the revenue that user groups obtain under this program; see Mahapatra, “Community Forest Management,” 7-8. 50. Mahapatra, “Community Forest Management,” 8; see also George Varughese, “Population and Forest Dynamics in the Hills of Nepal: Institutional Remedies by Rural Communities,” in Clark Gibson, Margaret A. McKean, and Elinor Ostrom, eds., People and Forests: Communities, Institutions, and Governance (Cambridge, MA: MIT Press, 2000), 193-227. 51. Britt, “Forestry and Forest Policies,” 27. 52. Sanjay Nepal and K. Weber, “A Buffer Zone for Biodiversity Conservation: Viabil- ity of the Concept in Nepal’s Royal Chitwan National Park,” Environmental Conservation, 21 (1994): 333. See also N. Iswharan and W. Erdelen, “Conserving Sunharaja: An Experi- ment in Sustainable Development in Sri Lanka,” Ambio, 12 (1990): 237-44. 53. See Arun Agrawal, Shree Govind Shah, Mukunda Karmacharya, and Birendra Karna, “Conservation with Communities: The Parks and People Program in Nepal,” Eval- uation Report of the Parks and People Program, UNDP (Kathmandu, Nepal: International Forest Resources and Institutions Program, Nepal, 1999). 54. Ibid., 42-47. See also “Terai’s Forest-Challenge for Management,” Spotlight: The National News Magazine, 20, no. 35 (March 16-22, 2001), for a critical review of the level of illegal harvesting occurring in the Terai of Nepal, where the People and Parks Program is located. 55. That the forest councils of Kumaon have existed the longest and were created by a colonial government also sets them apart from the other cases in our set. However, these two distinguishing factors seem to us less important in explaining their success. The British colonial state had also initiated programs of environmental decentralization in Madras and Burma. In both these places, the effort at decentralized management of forests was launched before it was in Kumaon. However, the programs in Burma and Madras were scrapped in favor of centralized management within four decades. The determining factors in the success of the councils seem to be the match between local mobilization and support within the Revenue Department and the institutional measures that this congruence of interests provoked. 56. Somanathan, “Deforestation, Property Rights,” pp. 42-45. 57. For a review of alarmist arguments in the 1970s and early 1980s, see Jack Ives and Bruno Messerli, The Himalayan Dilemma: Reconciling Development and Conservation (London: Routledge, 1989). POLITICSEVAN S. LIEBERMAN & SOCIETY

National Political Community and the Politics of Income Taxation in Brazil and South Africa in the Twentieth Century

EVAN S. LIEBERMAN

Why was the South African state so much more successful than the Brazilian state in its attempts to collect income taxes during the twentieth century? Nationally distinctive tax policies and patterns of administration can be explained by examining the impact of contrasting definitions of National Political Community, specified in critical constitutions written around the turn of the century. The ways in which racial and spatial cleavages were addressed in the 1891 Brazilian constitution and the 1909 SouthAfrica Act influenced thedevelopment of interclass and intraclass relations, which in turn affected the willingness of upper-income groups to accept state demands for income tax payments. Varied patterns of state development were largely predicated on contested notions of “us” and “them.”

During the twentieth century, government leaders around the world attempted to collect income taxes as part of a broader effort to finance rapidly growing expenditures. Income taxation proved to be an incredibly lucrative source of revenue for many governments, while having the desirable property of being “equita- ble,” as it could be collected with progressive rate structures. In their attempts to collect this tax, however, governments found that they faced high monitoring costs and required relatively high levels of cooperation and “quasi-voluntary”

I would like to acknowledge financial support from the Fulbright Commission, the Social Science Research Council, the National Science Foundation (NSF No. 9724055), and the African Studies and Latin American Studies Centers at the University of California, Berkeley. For their valuable feedback, I would like to thank Ruth Berins Collier, David Collier, Marc Morjé Howard, David Leonard, Anthony Marx, Robert Price, John Quigley, Ian Shapiro, and Sven Steinmo. Thanks also to the editorial board of Politics & Society, especially to Margaret Levi, for excellent comments and suggestions. POLITICS & SOCIETY, Vol. 29 No. 4, December 2001 515-555 © 2001 Sage Publications 515 516 POLITICS & SOCIETY compliance, particularly from the economically dominant groups within society.1 Unlike other forms of revenue, such as consumption taxes and social contributions, which tend to be paid more indirectly and/or with some promise of specific benefit, income tax payments tend to be highly visible and unrequited. Because the very demand for taxation provides incentives for citizens to free ride on the tax payments on others, initially, citizens have tended to resist the enactment of such policies. And while states have generally threatened coercion against those refusing to pay, coercion can be both expensive and ineffective in the wake of extensive avoidance and evasion schemes. Thus, state leaders attempting to collect income taxes have been highly dependent on the willingness of upper-income groups—who control the lion’s share of taxable resources—to pay such taxes. Over the course of the twentieth century, levels of cooperation with state demands for income tax payment have varied widely. In some countries, citizens have essentially ignored the state’s income tax policies, and virtually no tax has been collected at all. In others, some income tax revenue has been collected, but with significant challenge from those liable during the policy-making and/or compliance stages. And in still other countries, the tax has been collected effectively and efficiently, with few significant challenges from within society. In turn, such variation has influenced the size of the state treasury, the functioning of markets, and the after-tax distribution of income within society. Such patterns of income tax policy and administration also reflect more generally on the relationship between the state and upper class groups within society and the degree to which such relations can be described as either adversarial or cooperative.2 Brazil and South Africa are cases of highly contrasting outcomes in terms of levels of state success in collecting income taxes from upper groups during the 20th century. On one hand, despite repeated attempts, the Brazilian state has never managed to collect huge sums of income tax, and by the 1990s, income tax collections amounted to only about 5 percent of GDP. Instead, Brazilians have paid a wide range of complicated, hidden, and largely regressive taxes, levies, and other charges. By contrast, the South African state has managed to become one of the world’s most successful income tax collectors, collecting approximately 15 percent of its GDP in the form of progressive, direct income taxes, largely from high-income earners. Such contrasting outcomes are remarkable in light of enormous social, political, and economic similarities between the two countries (see Table 1). In terms of most development indicators, the two countries are virtually identical, with similar levels of per capita income and similar levels of industrial development. In both countries, the size of the state is relatively large when compared with other upper middle-income countries, and total central state expenditure as share of GDP is about the same. These two societies are also characterized by the most unequal distributions of income in the world, as measured by the GINI coefficient, and in both, such economic inequalities are highly correlated with racial differ- EVAN S. LIEBERMAN 517

Table 1 Development Indicators in Brazil and South Africa (1994) Indicator Brazil South Africa

GDP/Capita ($PPP) 5,400 5,130 Manufacturing/GDP (%) 25 23 GINI Index 63.4 58.4 Race × Income correlation High High Government expenditure/GDP (%) 32 33

Source: World Bank. World Development Indicators on CD-ROM. Ronald Inglehart et al. World ValuesSurveys and European ValuesSurvey, 1981-1984, 1990-1993, and 1995-1997 [Computer file]. ICPSR version. Ann Arbor, MI: Institute for Social Research [producer], 2000. Ann Arbor, MI: Interuniversity Consortium for Political and Social Research [distributor], 2000.

ences. Moreover, these two countries share legacies of colonialism, European immigration, and slavery. In the wake of such similarities, it is surprising that the mix of tax policies and their application should be so different in the two countries. In this article, I argue that historically constructed definitions of National Polit- ical Community (NPC), varying in terms of how racial and regional cleavages were addressed in critical constitutions, influenced the development of class relations, political strategies, and ultimately the willingness of upper-income groups to cooperate with state demands for tax payment. The National Political Commu- nity is the official, state-sponsored definition of the nation, which is specified in constitutions or other key policy documents during critical moments of political change. Although for most of the twentieth century, repressive regimes governed both countries to the benefit of the wealthy, I demonstrate that contrasting modes of inclusion and exclusion from the political community shaped very different political dynamics, and were ultimately responsible for very different paths to inequality. Through a series of intervening causal steps, the explicit form of exclusion that was embodied in South Africa’s institutionalized white supremacy ultimately legitimated the state in the eyes of white-owned firms and high-income individuals. Although lower-income whites constituted only a minority of all low-income people in Southern Africa during the first half of the twentieth century, the definition of a white NPC eventually facilitated strong cross-class linkages between the white working class and the virtually all-white, upper-class groups. Ironically, this produced the legacy of a set of progressive tax policies and administrative practices that are today far more beneficial to blacks in that country than what their black counterparts inherited in modern Brazil. By contrast, in Brazil, where race chauvinism has been expressed in more subtle ways, but not through official state policy, class relations unfolded in almost the exactly opposite manner. The federal constitution helped make regional identities politically salient, and the virtually all-white upper groups came to see their interests as more 518 POLITICS & SOCIETY competing than shared. Such dynamics further exacerbated the distance between economic classes by providing idioms for mobilizing a sense of difference. In such a context, upper groups have not identified strategically or normatively compelling reasons to cooperate with the state executive and have largely resisted enforcement of the income tax. Without such “quasi-voluntary” compliance, state efforts to collect have been far less successful. Over time, these political patterns, and the tax policies and administrative practices they engendered, became increasingly institutionalized within these two countries. The remainder of this essay examines this argument through comparative analyses of political, economic, and social factors, and their influence on income tax collections. It begins by considering the limitations of several possible explanations to account for the variation between the two countries. It proceeds to investigate the alternative explanation by analyzing the impact of varied definitions of National Political Community on the development of tax politics in the respective countries. The comparison reveals that in response to similar sets of political problems, two very different political solutions were codified in founding constitutions, generating nationally distinctive patterns of class relations. The article shows how such class relations affected upper group calculations and responses to new demands for taxation during three key moments in the twentieth century, ultimately leading to the development of very different sets of income tax policies and patterns of collection—what I will refer to as the “tax state.”3

NATIONAL PATTERNS OF INCOME TAX COLLECTION AND RIVAL EXPLANATIONS

To account for contemporary differences in levels of income tax collection, it is necessary to investigate the historical trends and turning points in Brazilian and South African collections in the twentieth century. Prevailing theories of the political economy of taxation can account for common, over-time patterns of income tax collection, but critical differences in the trajectory of those collections remain unexplained, providing room to consider alternative hypotheses. Figure 1 plots the central state’s collection of income taxes in the two countries from 1912 to 1997 and it reveals important patterns of similarity and difference in the timing and quality of tax state development. In both countries, income tax collections as share of GDP rose from virtually nil at the beginning of the century to substantial shares of much larger economies by century’s end. In particular, these collections rose dramatically during the period of approximately 1939 to 1945, to be followed by slight declines until the mid-1960s. After this period, collections rose fairly steadily for the rest of the century. Beyond these similarities, however, there are critical differences. The most prominent contrast in the two-country, over-time comparison is that the rate of increase in total collections of income and property taxes was more than three times greater in South Africa than in Brazil. By 1917, South African collections EVAN S. LIEBERMAN 519

10 South Africa Brazil 8

Percent GDP 6

2 1912 1917 1922 1927 1932 1937 1942 1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997

Figure 1. Income tax collections in Brazil and South Africa (1912-1997). Source: Escola de Administração Fazendaria, Sistema Tributário: Características Gerais, Tendências Internacionais e Administração (Brasília: Secretaria da Receita Federal, 1994). Fundação Instituto Brasileiro de Geografia e Estatísticas—IBGE, Estatísticas Históricas do Brasil: Series Econômicas, Demográficas e Sociais de 1550 a 1988 (Rio de Janeiro, Brasil: IBGE, 1990). South Africa. Bureau of Census and Statistics, Union Statistics for Fifty Years, 1910-1960 (Pretoria, South Africa: Union of South Africa, 1960). Republic of South Africa, Budget Review (various years). Note: Does not include taxes on mines.

grew quickly to almost 2 percent of GDP, while in Brazil they remained at a frac- tion of 1 percent until the 1940s. During both subsequent periods of increased collections, the gains achieved in the South African case were much greater than those in Brazil. These divergent trends were fairly well established by the first half of the century, and nationally distinctive patterns have persisted largely until the present day. In the remainder of this section, I consider several possible determinants of the varied trajectories—including economic, international, and cultural factors, as well as the problem of corruption—and I demonstrate that no single one, or even a combination of these explanations, is fully sufficient.

Economic Influences

Analysts from a range of theoretical perspectives have argued that the processes of industrialization and economic development have strongly influenced the development of taxation systems around the world.4 Empirically, it is easy to verify that wealthier and more developed economies tend to generate not just more tax revenue in an absolute sense, but even relative to the size of their econo- 520 POLITICS & SOCIETY mies. The central state’s collection of direct taxes on income and property also tends to increase with levels of economic development.5 As shown in Figure 2, the correlation between per capita GDP and income and property tax collections as share of GDP during the period 1990-1994 was strong.6 On the other hand, from the perspective of the best-fit line in Figure 2, it is clear that South Africa collected a larger-than-expected sum of such taxes for a country at its level of development, while Brazil’s take was smaller-than-expected, highlighting the central puzzle of this article and the limits of a purely structural economic argument to predict particular country outcomes. While the “noise” around the best-fit line could be interpreted as randomly generated error, it is necessary to probe further to determine whether other factors may be responsible for such differences.7 From a more dynamic, historical perspective, the development of the income tax (and other domestic taxes) in both countries was associated with the process of economic development, but again, most evidence suggests that patterns of economic development across the two countries were very similar. Low levels of tax collections in Brazil and in Southern Africa at the end of the 19th century can easily be explained by the structure of their respective economies: Up until that period, both had been largely rural, nonindustrialized economies, with few easily taxable bases, and without great needs for centrally provided public goods. Most scholars date the onset of industrialization at around 1870 for both countries along with financial windfalls from primary commodities sold on international markets—coffee in the case of Brazil and gold from South Africa. Secondary “infant” industries were established with the protection of high tariffs, particularly beginning in the 1920s and 1930s, and both countries are notable for having achieved exceptional patterns of economic growth, particularly in the 1960s and early 1970s.8 Between 1963 and 1972, South Africa’s rate of return on invested capital was the highest in the world, and of the capitalist economies, only Japan expanded more rapidly than South Africa.9 Brazil’s “economic miracle” included rates of economic growth among the highest in the world, with sustained periods of double-digit growth. In both countries, businesses and economic elites reaped significant material rewards from the labor repressive regimes of the mid-1960s and 1970s, and patterns of socioeconomic privilege were reproduced and reinforced. These patterns of development help to explain rapidly increasing needs for public goods. They also gave rise to potentially large and increasingly concentrated income tax bases, lowering the transaction costs of collection in both countries.10 In short, they explain similarities in taxation outcomes, but not differences.

International Influences

Beyond economic factors, analysts have also emphasized the role of international influences on the development of tax systems.11 Indeed, it is possible to identify the advent of particular sets of taxation policies and administrative practices around the world with particular wars, periods of economic expansion or EVAN S. LIEBERMAN 521

20 South Africa

Brazil 0 400 800 600 1000 2000 4000 6000 8000 10000 20000

R=.56 ** GDP/capita ($PPP) N=71 Figure 2. Income and property tax collections plotted as share of GDP/Capita (1990-1994). Source: World Bank. World Development Indicators on CD-ROM (1998). **Significant at the .01 level.

contraction, and new economic ideas and models. The process of taxation capacity building is ongoing in all states, but punctuated moments of significant political, social, and/or economic change have been responsible for establishing trajectories of over-time development. Such factors have also been important in the adoption and implementation of the income tax in Brazil and South Africa. Both countries were participants in both World Wars, were deeply affected by the Cold War, and were deeply entrenched in the twentieth century state system, as signa- tory members of the League of Nations and the United Nations. Moreover, as shown in Figure 3, both countries faced a common twentieth century pressure of declining trade revenues. Yet again, the similarity of these influences can help explain only the similarities in over-time trends, but these factors do not provide any insights into why levels of collections varied so widely across the two countries.

Cultural Influences

Both in casual and more rigorous analyses, a conventional explanation for cross-national differences in tax policy and tax administration is a theoretical 522 POLITICS & SOCIETY

South Africa 5 Brazil

4 Percent GDP

1 1900 1906 1909 1915 1921 1924 1936 1948 1951 1954 1903 1912 1918 1927 1930 1933 1939 1942 1945 1957 1960

Figure 3. Trade revenues in Brazil and South Africa (1900-1960). Source: IBGE, Estatísticas Históricas do Brasil. South Africa, Bureau of Census and Statistics, Union Statistics for Fifty Years. black box often labeled “culture.” Cultural arguments tend to be resilient because they are so often difficult to prove wrong. Indeed, as a last resort, many analysts of taxation identify cultural differences as the basis for unexplained variations that tend to be sustained across countries over time. Of course, this is a broad, catch-all category with quite varying notions of how to define the concepts, let alone the specification of the causal mechanisms involved. To evaluate culture as an explanatory variable, it is critical to separate cause and effect. If patterns of taxation are described as tax “cultures,” then culture cannot be said to explain patterns of variation. This would be true by definition. Ana- lysts have observed that across time and space, different norms and practices of taxation develop in particular places. In other words, they argue that the very set of policies and practices which comprise the tax state constitute their own nationally distinctive cultures. Webber and Wildavsky’s discussion of tax systems around the world is an important contribution in its identification of differences in budgetary systems across places and history, but this presentation lacks a true theory of why such patterns emerge in the ways they do.12 As a result, no testable hypothesis can be derived when culture is defined as the outcome. Nonetheless, two potentially plausible cultural explanations could be mar- shaled to explain the central differences in the Brazilian and South African tax structures: value orientations and colonial legacies. In both cases, however, apparent covariation between these cultural factors and taxation outputs can be dis- EVAN S. LIEBERMAN 523 missed as spurious when considered in over-time comparisons and comparisons with additional country cases. In the first case, we can assess the degree to which overall cultural systems and dominant sets of values may be associated with the development of the state and economy.13 But this poses the thorny question of how to measure culture. Since religious values generally play a central role in the construction of culture, one can usefully contrast these two cases in terms of religious memberships: Brazil is a predominantly Catholic country, and South Africa is not—it is largely Protestant, with a mix of other religions. Following Weber, some have argued that Catholic societies tend to have less effective state capacities than others because the norms of citizenship are far weaker. Yet, various measures of country-level Catholic membership revealed that there is no substantively or statistically significant relationship between that factor and levels of income tax collections, even when con- trolling for other factors.14 Such analysis suggests that it is highly unlikely that relatively low income tax collections in Brazil can be explained by its religious culture. More thorough measures of value orientations, such as those conducted by Inglehart and his collaborators in the World Values Study (1995) reveal that in fact Brazil and South Africa are remarkably similar in terms of the “key cultural dimensions” of traditional versus secular-rational authority and survival versus material well-being.15 In the wake of such similarities, we cannot conclude that such conceptions of culture can provide an independent explanation of national patterns of taxation. As for “colonial legacies,” the South African colonies were governed under British colonial rule, and Brazil was governed under Portuguese rule prior to independence, and again, it is plausible that such differences could account for different tax structures, or the early formation of tax policies and administrative practices. If these legacies were truly influential on contemporary patterns of taxation, we would expect to see some identifiable relationship between colonial and postcolonial patterns of taxation. However, there is an unambiguous disjuncture in these patterns on either side of the turn of the twentieth century, when founda- tional constitutions were written in the two countries. Neither Brazil nor any of the four political entities that would later become the Union of South Africa were able to collect significant sums of direct taxation in the late nineteenth century, and ordinary income taxes never amounted to more than one percent of national income in either country. Instead, the Brazilian emperor and the various colonies and nascent states of Southern Africa relied heavily on trade duties, various indirect taxes and user fees to fund government expenditures. In both countries, several attempts to levy broad-based income and other consumption taxes were rebuked in the face of societal resistance.16 If anything, central state authority was far more centralized and consolidated in Brazil, where the monarchy had built a centralized tax administration. A central state did not even exist in Southern Africa before the twentieth century, and prior to the formation of Union, 524 POLITICS & SOCIETY intercolonial competition over revenue sources was at times rather fierce. Again, an explanation emphasizing colonial legacies can hardly account for South Africa’s much more successful collections of the income tax in recent years.

State Corruption

A fourth alternative account of variations in revenue collections and structures, and of citizen compliance more generally, focuses on the trustworthiness and credibility of the state executive and the bureaucracy.17 Specifically, one could hypothesize that the problems of corruption and of the failure to gain the trust of upper-income citizens have been more acute in Brazil than in South Africa, and that this variation explains important differences in the amount of direct income tax revenues collected. Although perceptions of corruption, credibility, and trustworthiness influence taxation outcomes, we gain more analytic leverage if we consider these factors as part of the puzzle that need to be explained, rather than as exogenous determinants of tax compliance and levels of collection. First, corruption on the part of tax collectors—in particular, the deliberate undervaluation of the tax liabilities of a firm or individual in return for a bribe—is simply one aspect of the outcome under investigation. I assume that most tax collectors are vulnerable to graft, but it is only when citizens actively attempt to evade their tax liabilities that tax collectors engage in such behavior as a regular practice. Tax fraud, whether carried out in partnership with the state’s agents or without it, is a phenomenon that is conceptually so proximate to the outcome of measured levels of collection that it is hardly useful to say that the former “explains” the latter. Rather, we can restate the more general puzzle, which is, why is there more tax corruption and less income tax collection in Brazil than in South Africa? At a deeper level, to assess the arguments posed by Levi and by others who emphasize the influence of “credible commitments” made by government leaders, we need to explore the hypothesis that citizens are less likely to meet their tax obligations quasi-voluntarily if they believe that the national treasury is being pil- fered for private or narrow gains. In both Brazil and in South Africa, and certainly in many other societies, the citizens who challenge the state’s demands for tax payment in the policy arena or through aggressive avoidance and evasion strategies typically justify their resistance to taxation with the claim that they are not getting a “fair deal,” and that the money is being wasted, either on bad policies, poor implementation, or through outright stealing by individuals in government. At the extreme, when state executives and their agents demonstrate a wanton dis- regard for public interests, and when virtually no government services are provided, it would be reasonable to predict that citizens will resist tax payments. When comparing Brazil and South Africa, we can establish that perceptions of corruption have covaried in the predicted manner with income tax collections. The best-known attempt to provide a cross-national measure of perceptions of EVAN S. LIEBERMAN 525 corruption for more recent years is the Transparency International (TI) survey, for which the earliest available measure is from the period 1980-1985, in which fifty-four countries were compared. On this survey, South Africa scored a 7.35 (rank 20) and Brazil a 4.67 (rank 32) on a 10-point scale, with ten representing the most trustworthy and transparent business environments.18 In short, this finding implies that the claim that perceptions of corruption among business elites are associated with the state’s ability to collect income taxes is a plausible one. The problem with an explanation that links historically higher levels of income tax collection in South Africa to perceptions of a more trustworthy and less corrupt state is that it too easily suggests that “perceptions” of corruption are tantamount to measuring “actual” corruption. By its very nature, “actual” corruption is virtually impossible to measure at a macro-level scale, and there is no reason to believe that citizens or economic elites have accurate information across countries. Ironically, transparent democratic institutions, such as a free press, may lead the average citizen to perceive higher levels of corruption simply because such information is made available to an extent that is not true in less transparent but potentially more corrupt polities. Certainly, in the cases of Brazil and South Africa, for most of the twentieth century, the respective governments have demonstrated through substantial expenditure projects that the treasury has not been used exclusively for the personal gains of government leaders. Massive “modernizing” development projects were carried out in both societies,19 major infrastructure projects were completed, and in both cases, these largely served the (upper-income) citizens who were liable for the payment of income taxes. In both countries, there exists documented accounts of illegal and corrupt behavior on the part of state leaders as well as of low level bureaucrats, but such anecdotes provide little clue as to the extent to which actual corruption has varied across the two countries.20 Neither country has had a profligate dictator like the former Zaire’s Sese Seko Mobutu, and particularly during more authoritarian periods, the respective states have been governed by regimes committed, at least to a degree, to the development of professional, “legal-rational” bureaucracies. In short, there is little empirical evidence that can lead us to conclude that a more corrupt or untrustworthy state is to blame for Brazil’s relatively weaker record of income tax collections, despite over a century of attempts to increase collections. If not actual state corruption, what can account for cross-national variation in perceptions of corruption? As Levi points out, “political and cultural organizations can also affect perceptions of the trustworthiness of government and of the extent of ethical reciprocity.”21 This suggests the need to look more closely at such factors in comparative perspective. Perceptions of state action on the part of individual22 and collective actors may be largely determined by political context, independent of the goods and services the state actually provides. If we can develop a plausible account of political-institutional variation that leads societal actors to interpret state action in predictable ways, then we will have made corrup- 526 POLITICS & SOCIETY tion and trustworthiness endogenous variables in the explanation. Such an analytic move is made in the general explanation developed below.

THE INFLUENCE OF NATIONAL POLITICAL COMMUNITY ON INCOME TAX COLLECTION

This section advances an argument that relates definitions of National Political Community (NPC) to the development of income tax policies and administrative practices. It provides an explanation for the cross-national variation in levels of collections depicted in Figure 1. Comparative-historical analysis reveals how distinctive definitions of NPC affected the organization of class relations and the political interactions that would ultimately produce such different relationships between the state and upper groups in Brazil and South Africa.

The Argument

The logic of the argument is a critical junctures model of state development (see Figure 4), which highlights the influence of the period during which definitions of NPC are defined. A critical juncture is a “period of significant change, which occurs in distinct ways in different countries...andwhich is hypothesized to produce distinct legacies,” and which generally follows a period of social, political, and/or economic crisis.23 The NPC is the group of people who are officially entitled to the rights and responsibilities of citizenship of a particular state.24 As Reinhard Bendix points out, “A core element of nation-building is the codification of the rights and duties of all adults who are classified as citizens. The question is how exclusively or inclusively citizenship is defined.”25 At its core, nationhood constitutes a bounded, political identity tantamount to the notion of a “people.” Given that there is no “natural” or genetic basis for nationhood—as Benedict Anderson points out, they are “imagined communities”26—state leaders must specify rules and definitions of membership. While the process of defining the community may be constantly evolving, at critical moments, truly novel definitions are advanced and accepted in the form of political settlements that become accepted as basic “rules of the game” or institutions within society. In most societies, the written constitution is the most important specification of the NPC, though supplementary policies and legislation may be used to amend the framework laid out in that document. I argue that the different ways in which groups are either included in or excluded from the NPC explains a great deal of the variation in the development of the tax state. Such inclusions and exclusions influence the cohesion of economic classes and the likelihood that cross-class alliances will get formed. This is because important political institutions such as party systems, employer organizations and unions tend to reflect the structure of the NPC and to allocate power according to the political salience of different groups within society. Depending EVAN S. LIEBERMAN 527

Pre-Existing Brazil: Race, Region Cleavages So Africa: Race, Region

Crisis Brazil: Fall of the Emperor (1889) So Africa: Boer War (1899-1902)

Brazil: Constitutional convention (1889-1891) Critical Juncture So Africa: Constitutional Convention (1902-1909)

Definition of Upper-Group Class National Political Calculations, Tax State Relations Community Strategies

Brazil 1891 Constitution Fragmentation and State serves “them”; Adversarial: Officially Non-racial Polarization Challenge the state - Complex policies Federation - Difficult collections Region is salient

South 1909 Constitution Cohesion and (white) State serves “us”; Cooperative: Africa Racially-exclusionary Cross-class linkages Cooperate with state - Transparent policies Union - Easy collections Race is salient

Figure 4. The influence of national political community on the tax state in Brazil and South Africa.

on how groups are included or excluded, the definition of NPC also affects the types of political idioms that are used for making claims about moral obligations and notions of “fairness” and “equity.” In turn, these political institutions,27 and class relations more generally, frame the calculations and strategies of upper class actors who are largely responsible for paying the income tax in highly unequal societies. Within the public economy, individuals are more willing to pay when they are confident that “our” group will enjoy the benefits of state action. If there is a perception that the state will transfer benefits to “them,” or some other group, or if the barriers to in-group membership are permeable, citizens are less likely to perceive tax payment as a beneficial or rational move. In such cases, they are more likely to either perceive the state as behaving in an unfair or “corrupt” manner, or they are more likely to engage in tax fraud as part of a larger strategy to minimize tax payments. Citizens are more willing to pay if the state can make credible promises that the money will go to “us,” either in the near-term, or to the in-group’s offspring. When the National Political Community is defined in exclusionary terms—that is, when it is explicitly defined in terms of “ins” versus “outs” or in terms of a common group identity versus an excluded group, this leads to higher levels of upper class cohesion. In this case, upper-income actors are much more likely to perceive the state as “our” state, that serves “us,” and they are more likely to cooperate with state demands for taxation at both the policy-making and collections stages. Additionally, exclusion can increase the coherence of lower groups 528 POLITICS & SOCIETY within the NPC, and such strength provides opportunities to make strong claims for progressive taxation, particularly when lower groups can invoke a shared identity as the basis for a cross-class project. Alternatively, an inclusive definition, which recognizes multiple groups within the political community, tends to create class fragmentation and polarization. In this case, upper-income actors are less likely to perceive their interests as shared with one another. They are more likely to calculate that benefits of state action will go to one of “them,” not to “us,” and as a result, they are more likely to challenge the state’s demand for tax payment with narrowly defined notions of equity and obligation. Inclusive definitions also provide a weaker basis for organization and demand-making from below, and for the development of broad cross-class projects. Together, these factors lead to more tenuous state-society relations, in which the tax state is comprised of more complex policies and less effective administration. The political legacies of contrasting definitions of NPC run deeply, and at several levels. Over time, organizations and political discourse become institutionalized within society, reproducing the political patterns set in motion by the critical juncture, even in the wake of changing economic and international conditions.28 Once the definition of the NPC has been established by the state, it may become so embedded in the organizations and institutions of the society that it is taken for granted by the individual and collective actors within the polity. Moreover, the early development of the tax state—its policy framework and the bureaucracy’s relationship with society—may become institutionalized to the extent that other factors influence change only at the margins.29 As a result, much of the contemporary variation in cross-national taxation can be explained with a turn toward history, through an identification of the critical juncture, and a comparative analysis of the hypothesized impact on taxation outcomes.

Critical Juncture: Defining National Political Community in Brazil and South Africa

To evaluate this argument in the cases of Brazil and South Africa, it is necessary to identify the periods when modern definitions of NPC were forged in these countries. In the decades surrounding the turn of the twentieth century, Brazilian and South African political elites were presented with historically unique opportunities to codify definitions of NPC. The critical junctures (Brazil: 1889-1991; South Africa: 1902-1909) followed important moments of political crisis. In Brazil, modernizing urban and military groups came to view the imperial government as an anachronism which stifled progress, and following the end of the Para- guayan War (1865-1870), they overthrew the emperor on 15 November 1889. A constituent congress was held, in which each of 20 states and the federal district were represented, and on 24 February 1891, the first Republican constitution was promulgated in Brazil. In Southern Africa, the bloody Anglo-Boer war (1899-1902) EVAN S. LIEBERMAN 529 finally came to an end with a surprisingly difficult British victory, and eventually gave way to the 1902 peace Treaty of Vereeniging. After seven years of constitutional conventions and conferences, in which the four former colonies were represented, the South Africa Act was given Royal Assent on 20 September 1909. During the constitutional conventions in Brazil (1889-1891) and in South Africa (1902-1909), political elites in both countries faced a similar set of questions regarding how to define the NPC. For constitutional planners, legacies of immigration from Europe, Asia, and Africa; slavery; miscegenation; and centuries of internal conflict rendered the project of establishing clear and cohesive criteria for nationhood extremely difficult.30 Specifically, racial and regional cleavages were highly divisive and stood in the face of the very notion of nationhood. First, racial differentiation had become important in both South Africa and Brazil from the beginning of European contact with the indigenous populations in those countries. As is well known, in the case of South Africa, white supremacy was mobilized as the basis for social and political organization almost from the day the Dutch East India Company representative Jan Van Riebeeck landed at the Cape of Good Hope in 1652.31 The conquering of various African populations as part of ongoing European territorial expansion in Southern Africa was legitimized with racial prejudice. Soon after European arrival, the institution of slavery was established at the Cape with slaves taken from Angola, Madagascar, and Mozam- bique. Although slavery was abolished in 1834, the institution of white supremacy permeated virtually all aspects of social relations in Southern Africa. Simi- larly, racial conflict in Brazil began when Portuguese immigrants arrived on the South American continent in 1500 and differentiated themselves in racial terms with respect to the sizeable indigenous population. Moreover, approximately 3.5 million slaves32 were brought to Brazil from Africa, an institution whose foundation was located in racial prejudice. Similar to the case of South Africa, in Brazil, “the relationship between race and politics in that country has been a close and integral one. Portuguese state policy made black slavery the very foundation of Brazil’s social and economic order during three centuries of colonial rule.”33 “Sci- entific” ideas about race, imported from Europe, reinforced prejudice and provided a workable political strategy for maintaining social hierarchies in which white minorities rested at the top.34 A parallel set of critical questions for defining NPC in Brazil and South Africa stemmed from the legacies of regional political fault-lines also associated with the colonial past. Regionally based political claims may be advanced by political leaders, when they mobilize a sense of territorially based collective interest and identity, proposing special claims for group rights, often including a degree of political sovereignty. National unity and feelings of commonality are clearly jeopardized by such claims. In Brazil and South Africa, prior regional administra- tions overlapped with real economic inequalities and cultural differences, providing a basis for serious political competition in the second half of the nineteenth 530 POLITICS & SOCIETY century. Nothing analogous to Southern Africa’s bloody three-year Anglo-Boer war (1899-1902) took place in Brazil, but strong regional identities and the model of state fragmentation within the rest of the South American continent posed stark regional challenges to national unity across vast expanses of Brazil’s national territory. Despite the similarity of these constitutional questions, and the availability of a similar set of options for addressing such questions, elites ultimately constructed very different definitions of NPC (see Table 2). The South Africa Act of 1909 defined it as a racially exclusionary union. Although the formal system of apartheid, which codified a host of draconian legislation for separating the race groups, was not put in place until four decades after the 1909 constitution was signed, that strategy can be seen as the ultimate extension of original decisions made in the South Africa Act. Blacks had hoped for a more tolerant racial order following the British victory, but “the price of trying to reconcile the whites was paid by the blacks and browns.”35 The 1902 signing of the Treaty of Vereeniging, ending three years of bloody warfare marked the beginning of a critical period of smoothing Brit- ish-Afrikaner antagonism through the idiom of white national unity in what would become the Union of South Africa. The 1909 South Africa Act was highly explicit in its racial codification of citizenship, reserving the privileges and responsibilities of voting and leadership only to those of “European” descent,36 and mandating a five-yearly census which would be the basis of allocating provincial representation on the basis of the number of European (male) adults. In the years following, several acts deepened the original ideas established in the constitution. For example, the 1927 Immorality Act made it illegal to have sexual inter- course across the color line. Meanwhile, the Southern African territory was defined in racial terms, demobilizing “region” as a viable political idiom. The 1913 Natives Land Act restricted the purchase and lease of land by Africans to the reserves, which constituted a mere 7 percent of the land—while this group represented more than 60 percent of the population living in the country at the time. In response to the question of federalism in South Africa, it was agreed that the “Native question...needed to be confronted as a whole by a centralized state capable of pooling all available ides and implementing them rapidly.”37 Accord- ing to the constitution, the Senate would provide equal representation for provinces of unequal size, but it was to be indirectly elected and weaker in many critical ways.38 Perhaps most importantly, “money bills” (those appropriating or raising revenue) were designated to originate in the House of Assembly and were immune to amendment (but not rejection) from the Senate.39 Provinces would have councils and governors to manage certain internal affairs, but they were to be clearly subordinate to national government. Their status was completely unprotected in the constitution, and in all ways, parliamentary sovereignty was established. EVAN S. LIEBERMAN 531

Table 2 Political Cleavages and Definitions of National Political Community in Brazil and South Africa Definitions of National Political Community Cleavage Brazil (1891 constitution) South Africa (1909 constitution)

Race Official nonracialism Racial exclusion Region Federalism Unitary system

The Brazilians resolved the question of NPC in virtually the opposite manner, by creating an officially nonracial federation. Most prominently, there was no mention of race or racial categories in the constitution,40 and as such, no subsequent legal authority was created for allocating resources or political representation based on the size of racial subgroups. In the 1900 and 1920 censuses, no question about race or color was asked.41 At least in rhetoric, the state committed itself to the integration of Brazilians of various skin colors into a single nation. Indeed, it is critical to point out that race has remained a socially salient category in Brazil up until the present, and persistent race chauvinism in various facets of social and economic life have been at work to maintain a racial hierarchy in terms of economic opportunity and achievement in Brazil.42 Yet, as will be described below, the tension between prevailing sentiments of race chauvinism with official state policies prohibiting explicit race-based discrimination has been central to understanding politics in Brazil. Unlike in South Africa, where provincial powers were carefully enumerated and restricted, in the 1891 Brazilian constitution, the provincial states (estados) gained the power, “To exercise in general any and every power or right not denied them by express provision of the constitution or contained by implication in such express provision.” Under this federal constitution, states could form their own military forces, directly elect their own governors, and were responsible for carry- ing out the fundamental functions of government. Indeed, a few important limitations were placed on state powers, particularly where federal powers were concerned, but this estado sphere of government was identified as the one deserving of all “residual” powers, making it appear the most fundamental power of all the government spheres. Unlike the parliamentary sovereignty accorded in South Africa, in Brazil, the federal government was highly restricted in its ability to intervene in state affairs, and all central state powers were enumerated in the constitution. In addition to this autonomy, states gained significant voice within the structure of national decision making. Like in South Africa, the parliament was bicameral, but the Senate had much greater powers—as all bills could originate in either house and could be amended or rejected by either house. 532 POLITICS & SOCIETY

Mechanism: Class Relations and Upper-Group Calculations and Strategies

The critical junctures argument does not imply that once the critical juncture is completed that we should expect a particular type of outcome to materialize immediately. Rather, legacies materialize after a series of intervening steps, and are transmitted through ongoing political processes.43 In particular, the different ways in which the NPC was defined in founding constitutions in Brazil and South Africa around the turn of the twentieth century helped make a particular ascriptive identity highly salient. In turn, those identities structured the development of interclass and intraclass relations through organizational development, political idioms, and the larger political environment in which calculations and strategies would be made. South Africa’s explicit, discursive form of racial exclusion was instrumental in generating high levels of class cohesion. The racial coordination of whites in the political arena, mobilized alongside the notion of a threatening black “other,” created a sense of collective interest, transcending regional, sectoral, and linguistic lines, making it easier to solve the free rider problem generally associated with taxation. South Africa’s definition of NPC provided an institutional basis for societal organization and strategy. For most of the twentieth century in South Africa, upper groups found themselves linked to one another through national business organizations, such as the Association of Chambers of Commerce (Assocom) and the Chamber of Mines; through strong, national political parties;44 and through the idioms of “whiteness,” and “Europeanness.”45 This level of class cohesion facilitated the state executive’s ability to strike effective bargains with upper groups by closing off opportunities for narrowly defined groups to make claims for “special interests.” Moreover, the salience of race as a political idiom helped link South Africa’s upper groups to lower-income whites, though this relationship was initially characterized by confrontation as much as by cooperation. White workers used race as the basis for high levels of mobilization, and under the banner, “Workers of the world unite and fight for a White South Africa,” organized a highly disruptive political challenge in the early decades of the twentieth century culminating with the exceptionally violent Rand Revolt of 1922—which amounted to a small-scale civil war, in which over 1.3 million man days were lost to strike activity.46 In the face of such challenges, upper groups were forced to recognize claims to address the “poor White” problem, and the demands of “European Labour.” As Arendt observes, “The poor whites in South Africa...demanded and were granted charity as the right of a white skin; having lost all consciousness that normally men do not earn a living by the color of their skin.”47 Employers and other representatives of upper-groups found that it was strategically useful to co-opt this group with inducements in various policy areas.48 EVAN S. LIEBERMAN 533

The combination of intraclass and interclass cohesion provided enormous leverage for the state executive, whose hand was strengthened by “legitimate” claims to represent white interests, and to enact policies that would equalize opportunity and income within the white polity. Upper groups calculated that the state was serving “us,” and elected to cooperate with the state in the construction of policies and in the collection of taxes. By contrast, Brazil’s officially nonracial federation, which gave significant political salience to regional claims, created the institutional basis for class fragmentation and polarization. Political parties developed as mainly local organizations, and upper-group interests were largely expressed within the framework of the provincial states.49 No business organization developed that could articulate a truly national set of business or class interests.50 Strong regional idioms, overlapping with race chauvinism, reinforced a sense of intraclass competition. A sense of ethno-regional heterogeneity—the problem that in many ways was “solved” by the mobilization of race in South Africa—was a source of division among upper groups in Brazil.51 In other words, there was no political “glue” to hold together upper groups in that country, and without explicit forms of political exclusion it appeared more likely that “others” might someday profit from tax payments made to the central state, making the state executive much less able to convince citizens of the collective benefits of tax payment. At the extreme, the period of the First Republic (1889-1930) was marked by a series of attempts at secession by states in the South and Southeast regions, which had continued to develop with strong regional identities and significant political and financial autonomy. Unlike in South Africa, where race suggested some solidarity between upper- and lower-income whites, in Brazil, the existence of “poor whites” has had virtually no emotive or strategic implications for the white elite. In retrospect, the final acceptance of abolition on the part of large rural landowners depended on a calculation—which turned out to be correct—“that abolition need not endanger their social and economic dominance.”52 Because official discrimination was made illegal in Brazil, poor whites have been treated simply as “poor,” rather than as “white.” Nevertheless, racial stereotyping and prejudice were perpetuated in Brazil through social institutions and through the state’s deliberate attempts to “whiten” the population through racial mixing, immigration and other policies,53 and inequalities both within and between race groups were tolerated as part of a more “acceptable” socioeconomic hierarchy. Brazil’s definition of NPC established an institutional environment that weakened the ability of lower groups to make demands for progressive taxation when compared with South Africa. Although European workers displaced Afro-Brazilians from jobs, and came to Brazil with racist views, in the political context of postabolition Brazil, ideas of white-or European-supremacy could not be used explicitly as a basis for mobili- zations from below.54 Though workers did stage important strikes in São Paulo and Rio de Janeiro between 1917 and 1920, these did not amount to anything 534 POLITICS & SOCIETY approximating the challenge advanced in South Africa.55 Without an analogous basis for solidarity, Brazilian organized labor was incorporated through far more repressive tactics, including police raids and the jailing of union leaders.56 The state laid down the terms of how unions could organize and the extent to which unions were provided opportunities to participate in decision making. Regional- ism was a key principle of social control: under the Presidency of Getúlio Vargas (1930-1945), corporatist arrangements divided unions along sectoral lines, creating federations at the state level and confederations at the national level. Even within sectors, the confederations have not been unifying organizations, the manifestation of a blueprint designed to divide the working class. As a result, even with a larger share of unionized workers than in South Africa, Brazilian lower groups have been politically weaker and less successful at making class-based demands for more progressive social policies, including more progressive allocations of the tax burden during the twentieth century.

Legacy: The Development of the Income Tax in Brazil and South Africa

The development of the income tax in Brazil and South Africa during the twentieth century involved a punctuated bargaining process between the state executive and upper-income groups within those societies. The nature of that interaction was strongly influenced by the class relations and associated upper-group calculations and strategies produced by nationally distinctive definitions of NPC, as described above. Although each state executive made broadly similar demands for tax payment in response to similar exogenous pressures, upper-group responses to those demands were quite different across the two countries, both in terms of support for policies during the policy-making process and in terms of levels of “quasi-voluntary” compliance during the collections process. Such responses profoundly influenced the quality of tax policy, tax administration, and ultimately, tax collections. Fragmented and polarized class relations in Brazil led upper groups to repeatedly challenge the state’s demands during the policy-making process and during the process of collection. As a result, an “adversarial” tax state developed, in which the bureaucracy has not been able to uniformly collect taxes from individuals and firms, who engaged in deliberate avoidance and evasion schemes, and noncooperation in the formulation of policy. Coherent class relations among whites in South Africa, by contrast, led analogous upper groups to largely accept the state’s demand for progressive taxes. A “cooperative” tax state emerged, in which the tasks of bureaucrats and policy-makers were largely facilitated by the free exchange of information, and the minimization of challenges to the state’s claim to authority. Three historical episodes reveal that even in the wake of similar structural conditions, different political logics and patterns of engagement between state and society resulted in very different taxation outcomes (see Table 3). At each junc- EVAN S. LIEBERMAN 535

Table 3 The Development of the Income Tax in Brazil and South Africa in the twentieth Century Country Outcomes Brazil South Africa (Nonracial (Racially Exogenous “Shock” State Initiative Federation) Exclusionary Union)

International acceptance Introduce the general Slow adoption Rapid adoption of income tax; WWI income tax (33 + 4 years) (1 year) WWII Increase income tax Minor gains in Significant gains revenues for war needs revenue (+1% GDP) in revenue (1939-45) (+6% GDP) Escalation of the Modernize the tax system Partial modernization Thoroughgoing Cold War and increase revenues modernization (1960-75)

ture, similar exogenous international and economic pressures generated revenue “needs” in both countries, leading the executive to propose new taxes or major tax reforms. During the first decades of the twentieth century, particularly around the First World War, executives acted to adopt a general income tax; during the Sec- ond World War, they were prompted to expand and to consolidate the tax to generate additional revenues; and during the 1960s, rapid industrialization and the politics of the Cold War inspired modernizing reforms of the respective tax systems. During each episode, the legacy of the original critical juncture was manifest in the nationally distinctive responses on the part of upper groups, and ultimately, in the content of tax reforms and patterns of collection.57

Introduction of the General Income Tax Internationally, the first three decades of the twentieth century witnessed the adoption of the general income tax around the world. In both Brazil and South Africa, initiatives to develop more modern states, the First World War, and associated changing economic conditions, provided strong incentives to establish this important, progressive tax. The South African Minister of Finance stressed new expenditure demands and the longer-term financial health of the state as the ratio- nale for making demands for income taxation in his 1914 budget speech. Simi- larly in Brazil, plummeting export prices and trade revenues in 1913,58 alongside rapid industrial capital formation, led to renewed initiatives by the Finance Minis- ter for the enactment of a general income tax.59 Differing class relations and associated political calculations structured very different responses to these otherwise largely similar demands, however, and the income tax was passed and adopted much more rapidly and efficiently in South Africa than in Brazil. Brazil’s definition of NPC created regional fault lines along which upper-income actors challenged proposals for the new tax. Upper groups resisted its enactment, largely in the name of “unfairness,” and regional competition 536 POLITICS & SOCIETY blocked effective passage for more than three decades after it was first proposed during the constitutional convention. Without the ability to reach consensus on a general, uniformly applied policy, the state selectively taxed income through a series of piecemeal policies, none of which generated significant amounts of revenue.60 Even some of these taxes were challenged. For example, the Industrial Cen- ter of Brazil in 1919 vehemently protested the imposition of a 3 percent tax on profits and dividends on the manufacturing industry.61 Despite the persistence of several legislators in the Chamber of Deputies Budget Committee, who believed that the income tax was an appropriately “modern” form of state revenue for the Republican government, sufficient consensus around a general income tax could not be built immediately following the outbreak of war, as several legislators argued that collection would be difficult.62 A general income tax on personal and company income was eventually passed through the Chamber of Deputies in 1922, but because of various political objections over detail, it was not even effectively implemented until 1926.63 The compromise tax legislation was complex: Part of the tax was proportional (a flat rate), which differed according to the source of the income, and the other part was progressive on a calculated base of net income with rates ranging from 0.5 percent to 8 percent. The initial law was ridden with exemptions, particularly on agricultural profits and any interest on loans associated with agriculture or extractive industries. Moreover, several proposals to extend the calculation of taxable income were quickly revoked.64 Eight different methods of computation were adopted for arriving at taxable income or profits—a complex and potentially leaky basis for calculation even in the early law. Once enacted, resentment toward the tax was expressed by business leaders and through the expression of regional group interests. The São Paulo newspaper Folha argued that the income tax was “absurd,” and predicted, “The income tax is destined to fail. And all the country will profit from this.”65 Oliveira Passos of the Centro Industrial do Brasil—the organization that would become the Centro Industrial do Rio de Janeiro—argued in a 1928 meeting of the organization that the income tax was a tax that should not have been introduced in Brazil, “a new country, an importer of foreign capital and labor....Thetaxrepels capital and can only be properly introduced in countries with excess capital.”66 (Similar claims could have been made in South Africa but were not.) Strong resistance to the tax from representatives of such business organizations as well as from (estado) state governors, forced the national executive to grant a panoply of regional and sectoral privileges and exemptions.67 Not surprisingly, collecting the income tax in this environment proved difficult: in a dispatch to the American State Department, a representative from the American Embassy in Rio de Janeiro wrote, “The collection of income tax is very imperfect, and it is unlikely that a great deal of profit will be shown, were the government to consider the amount spent in administration and the number of individuals and firms who evade payment.”68 EVAN S. LIEBERMAN 537

In the decade that followed the enactment of the income tax, additional political fights ensued over the national government’s right to collect the tax, and various legislators proposed that the estados—rather than the federal government— should collect at least part of the income tax. Ultimately, under the dictatorial regime of Getúlio Vargas, it was resolved that the income tax—as was standard international practice—would be collected exclusively by the federal government. Nonetheless, amidst the plethora of challenges to payment, early income tax collections were relatively paltry. By contrast, in South Africa, where upper groups were both ideologically and organizationally unified, there was much less political space to challenge the general income tax during the first years of the Union. Political authority was firmly within the grasp of the Finance Minister, whose South African Party (SAP)—a conciliatory party of Afrikaners and Englishmen—had won 67 of the 121 seats in the National Assembly. Without strongly defined regional party factions, as was the case in Brazil, the parameters of the new tax were not challenged in the name of regional balancing, or the need for greater regional autonomy. It is critical to highlight how surprising such an outcome was from the vantage point of only a few years earlier, as intercolonial relations prior to the formation of the Union had been characterized by sharp economic differences and political conflicts over pubic finance. Instead, employers and business leaders expressed support for the war and a willingness to sacrifice in financial terms.69 The formation of a union, as opposed to the federation that most had expected, had worked quickly to smooth over regional and intraethnic divides within the white polity.70 Also important for understanding the enactment of the income tax was the “push” from white lower groups in South Africa. Following the 1913 strike on the Rand, in which white workers challenged the possibility of threats from black competition on the labor market, upper groups calculated the strategic advantages of progressive taxation, and the state recognized the need to create policies that would be seen as benefiting this group. In announcing the income tax, the Minis- ter of Finance argued against the alternatives, explaining, “indirect taxation...bears more heavily on the poor man than on the rich.”71 If shear numbers alone determined political strategies, lower groups would have been equally influential in Brazil, but it was race-based organization, and solidarity in South Africa that made this collective actor far more influential over the shape of policy. Once established, South African income tax rates tracked up quickly through the use of income tax surcharges, and by 1922, the top individual income tax rate was 35 percent. Moreover, income tax collections grew rapidly over the course of this period because of the excess profits tax initiated in 1917, which was also accepted quite easily within the business community. This tax was applied on a base of profits that was considered above the “normal” prewar profitability of a firm. The state and business associations generally agreed that no one should 538 POLITICS & SOCIETY profit disproportionately from the war economy—in which supply was limited and the opportunity for price gouging was great—and that taxation would be the fairest way to redistribute the economic effects of the war within the white population. Although the excess profits tax was slated to be removed six months following the end of the war, in 1919, it was increased to 5 percent of excess profits for “meeting war expenditure.” The “ratcheting” effect that fiscal historians have noted for other contexts certainly characterized this wartime adjustment. In the 1921-1922 Budget, the excess profits tax and special war levies were removed—but they were replaced with higher normal rates of tax on companies (7.5 percent instead of 5 percent) amounting to a slight overall increase in the tax rate, and those rates were raised yet again in 1925 to 12.5 percent. These early proposals and the expressed willingness of upper groups to pay facilitated the development of the tax bureaucracy. By virtually all accounts, the South African state’s efforts to collect the income tax were as successful as treasury officials could have hoped. In a report on the working of the income tax act in its first year, the Commissioner of Taxes reported that projections for numbers of taxpayers and revenues collected were met or surpassed. Even in describing the details of building the bureaucracy, the Commissioner made clear the importance of taxpayer views of the tax as legitimate: “the public as a rule have recognized the need for fresh taxation, and have accepted their liability to contribute towards the country’s requirement.”72

Increases of Income Tax Revenues for War Needs During the Second World War, state executives in Brazil and South Africa responded to new pressures and opportunities to expand the scope of the state by demanding special “war taxes,” largely in the form of extensions of the general income tax base. Although both states sought similar ends—to increase revenues and to control inflationary pressures—varied class relations and strategic calculations continued to structure very different responses, ultimately producing different outcomes. By the end of the war, South Africa was collecting more than three times what the Brazilian state was collecting in income taxes (7.5 percent of GDP in South Africa compared with 2.2 percent of GDP in Brazil, not including mining revenues). And, South Africa more successfully stemmed the tide of inflation, as the price index grew only 35 percent during the six-year period, compared with 120 percent in Brazil. The Brazilian state’s relatively much weaker ability to expand income taxation during this period is surprising from the perspective of standard accounts of Bra- zilian state development. Students of Brazilian politics have long noted that the presidency of Getúlio Vargas (1930-1945) and particularly the seven-year period known as the Estado Novo (1937-1945) during which Vargas assumed dictatorial powers, were periods of remarkable expansion of the size and scope of the central EVAN S. LIEBERMAN 539 state. From the time of his 1930 installation in the presidency, Vargas sought to achieve a more modern state, less dependent on local influences, and a key policy goal was to increase tax collections to the central state. In fact, the very tenentistas (young, modernizing military officers) that helped bring Vargas to power in a 1930 coup d’etat argued that budgetary imbalances were “evil,”73 and quite clearly, increased taxation on income was a potential weapon against such deficits. Efforts to collect such taxes were advanced throughout the 15-year presidency. Moreover, the economic conditions were favorable for increased income taxation. Although the pace of economic growth slowed somewhat from the prior period, the economy was still growing and industrializing, and improved terms of trade actually meant a tangible increase in living conditions between 1939-1945.74 Industrial production grew at rate of 5.4 percent during 39-45; and several key industries, in particular textiles, expanded during these years because of shortages in world production.75 These favorable circumstances, including extremely high coffee prices, led to consistent exchange surpluses, and industrial firms made large profits and accumulated significant reserves.76 Undoubtedly, a potentially important “handle” for taxing domestically generated income became available during these years, and as direct taxation was clearly becoming a popular tool around the world, Brazilian finance minister, Octâvio Gouvêa de Bulhões, sought to follow suit during the Second World War. Initiatives to consolidate the authority of the Brazilian central state during this period were consistently thwarted by the reproduction of fragmented class relations. The salience of regional identity, reinforced through employer and other organizations, and through political discourse, could not be easily erased from the political landscape, particularly without a “black threat” or any nationally unifying idiom for upper groups. After seven years in office, the proliferation of competing interests became unmanageable for Vargas, and despite earlier efforts to consolidate and to centralize state power during his first term, he initiated a period of even more extreme political change, arguing that the country must forego the “threatening” aspects of national political competition. With the support of the military, he preempted the scheduled elections of 1938 and initiated a coup d’etat in 1937 to retain power, in the defense of national unity and security.77 He implemented a dictatorial regime, introducing a new constitution that was centralizing in spirit, and made investments in education and transportation infrastructure to develop a much stronger national identity. Nevertheless, unlike in South Africa, employer organizations remained divided along both state and sectoral lines. Despite the peak-level integration of employer and labor organizations into “confederations,” these national organizations remained extremely weak when compared with the state level federations, such as FIRJAN and FIESP, the state-level industrial associations of Rio de Janeiro and São Paulo. Similarly, the urban middle class appeared heterogeneous—“divided in terms of income, social mobility, racial origin and degree of dependence on the regional ruling class.”78 They also 540 POLITICS & SOCIETY varied in terms of their alignment with the working classes—more so in Rio de Janeiro and less so in São Paulo.79 Moreover, Vargas did not eliminate the power of state political leaders; he merely accommodated them into the framework of the federal bureaucracy.80 The prevailing logic of regionalism suggested to South- eastern elites that compliance with central state demands would benefit their Northern counterparts—not benefiting “us,” but “them.” Within this political context, upper groups again challenged initiatives to raise new taxes. While the state executive deployed its dictatorial powers to declare a series of tax hikes, and “excess profits” taxes to increase revenue and to guard against wartime inflation,81 upper groups vehemently challenged the “fairness” of these taxes by making special claims to the particular needs and burdens of their states, regions, and sectors. Employer representatives repeatedly traveled to Rio de Janeiro to negotiate “compromises” that effectively diluted the overall burden on firms and high-income individuals. In a telegram to the ministry of finance, the Acting President of the Federation of Commerce of the State of São Paulo complained that the “interested classes” were “overloaded” with taxes. João Daudt D’Oliveira, president of the Federation of Brazilian Business Associations affirmed that, “The news of the recent creation of the tax on extraordinary profits is the source of profound apprehension on the part of the productive classes of São Paulo.”82 The negotiations in Rio ultimately produced a complex piece of legislation, with highly arbitrary rules and extremely high thresholds for assessing excess profits, and ultimately, relatively little additional revenue was collected. Moreover, industrialists secured the option to avoid war taxes by purchasing equipment certificates to import capital equipment when international conditions would allow it. Even under a dictatorship, the state executive was still forced to bargain with the leaders of organized business, and in practice, possessed insufficient bargaining power to force significant additional taxation on the diverse business interests of the various Brazilian regions. Through aggressive and often illicit reporting of accounts, high-income individuals and firms further minimized or eliminated their tax burden. At every turn, upper groups resisted the tax burden. Writing in 1946, a frustrated, high-ranking Brazilian tax official described the process of getting taxpayers to pay as a “struggle.”83 During the same period, class relations in South Africa remained highly distinctive from those of Brazil, even though the socioeconomic bases for organization and conflict were rather similar. Like in Brazil, the South African economy was also characterized by profound regional inequalities, and huge disparities of income and wealth. Yet, the political logic of South Africa, which gave low salience to regional identities and sectoral interests, pointed toward relatively much greater cohesion within the white polity, and particularly among the business elite. In 1938, the United Party won an overwhelming victory of the white electorate, claiming 111 of 153 seats in the National Assembly, with a support base that was largely similar to that held by the SAP during the first Parliamentary EVAN S. LIEBERMAN 541 election. Although it lost some Afrikaner support in the 1943 election, the United Party maintained a strong majority with eighty-nine seats.84 As a result, South African policy-making resembled that of a cohesive, single-party dominant, European parliamentary political system. “Local” concerns were increasingly absent from the national political arena, and political power was increasingly consolidated in the hands of the state executive, limiting the opportunities for narrowly defined groups to make claims with any clout. In the South African context, the state’s demand for taxation from upper groups went largely unchallenged. Ironically, just as the government promoted the “noble” and “humanitarian” causes of the war, only whites were allowed to join combat units.85 Largely English-speaking upper-groups viewed their fate as shared with that of the state and the larger war effort, and the executive was empowered to make demands in broad, collective terms and did not have to appeal to more narrowly defined interests.86 In a rousing speech to the Chamber of Com- merce, State President Smuts appealed to this sentiment, arguing, “Let us keep alive the spirit of sacrifice,” and emphasized the need for cross-class empathy within the white polity when proposing progressive tax policies.87 He pointed to the need to continue using income taxation, even if on a very narrow base of individuals, because this would be more “just” with respect to the economically less well off—at the time a clear euphemism for poor white Afrikaners. The massive 1922 strike, (the “Rand Revolt”) and the Labor Party’s 1924 electoral victory loomed in the minds of employers and high-income individuals as examples of the potential threats of class-based resistance if the white working-class were squeezed too hard in economic terms. In this light, sacrifice benefiting other whites was simultaneously normatively appealing and a strategically rational move that would facilitate the labor peace viewed as necessary for further capitalist development within the country. Although English-speaking upper-income groups expressed more liberal views with respect to the race question, strategic conciliation with white labor prevented significant backtracking on racially exclusionary practices. With each successive “war budget” presented by the Minister of Finance, tax rates were increased and revenues skyrocketed. By 1941, the top marginal income tax rate for an individual earning over £100,000 was increased to 72.4 percent. In fact, when compared with the USA, Canada, Australia, and Britain, South Africa’s income tax system was considerably more progressive in its rate structure.88 From all available evidence, it appears that such high rates did not induce massive corruption or evasion—a relationship many contemporary tax analysts argue is almost axiomatic. Rather, the response to state demands for taxation was impressive. The President of the Chamber of Mines explained,

The issues at stake, being what they are, require that all that can be done should and must be done. It is from that standpoint that we regard and accept taxation and other measures which, during this time of war, apply to and affect the Mining Industry.89 542 POLITICS & SOCIETY

Unlike in Brazil, there was no political space within which more narrow capitalist interests could publicly claim that their burden was truly unfair, and taxpayers simply accepted the proposals in a spirit of “grin and bear it.”90 Between 1939 and 1945, income tax collections increased by a factor of three. Although consumption taxes increased slightly, as a share of GDP these taxes still only accounted for about 13 percent of total domestic taxation.

Modernization of Tax Systems During the “War on Communism” A mix of Cold War tensions and rapid economic development implied a host of new administrative and revenue needs that would lead the respective state leaders to initiate a new round of tax reform in the 1960s. Once again, upper groups in the respective countries responded in very different ways to otherwise similar state initiatives to increase their tax burden, shaping very different outcomes. The political idioms of race and region, and their organizational and discursive manifestations, structured the political exchange between strong executives and these economic elites during the process of policy-making and collection. The South African definition of NPC continued to reproduce high levels of class cohesion, facilitating the additional extraction of revenue from upper groups. Between 1960 and 1975, collections of income tax as share of GDP more than doubled, from 4.9 percent of GDP to 11.5 percent of GDP. Efficiency gains in South Africa from modernizing reforms implemented in the early 1960s led to sharp declines in the costs of administration relative to revenue collected to as low as 0.3 percent of collections (compared with costs of 1.0 to 1.4 percent in Brazil.) In Brazil, although a strong military government was able to push through important policy and administrative reforms within a short period of time, the continued salience of regionalism resparked the flame of intraelite competition over the tax burden. Tax incentives proliferated, and compliance rates returned to abysmally low levels, ultimately impeding the state’s ability to collect from those groups. The rate of increase in collections in Brazil was much slower, and from a much lower base during the same period, increasing from just 2.0 to 3.0 percent of GDP. Initiatives for tax reform in both countries during this period were advanced within the context of massive state-sponsored repression and deliberate attempts to establish more “modern” political orders, with increased bureaucratization of the state.91 High levels of inequality, when combined with rapid urbanization and industrialization had sowed the seeds of challenge from below during the postwar years.92 In both cases, such challenges were struck down with a strong fist: In 1960, the South African state responded violently to a peaceful protest against the pass laws in the township of Sharpeville, and sixty-nine Africans were killed.93 In Brazil, responding to increased political polarization within society, the military seized the reigns of government in a 1964 coup, installing a military leadership that would remain for more than two decades. Thus, in both countries, the early EVAN S. LIEBERMAN 543

1960s marked turning points in which the state went to new lengths to impose its authority in the name of anti-Communist rhetoric, taking new powers such as the censoring of information, and the capacity to detain suspects without trial as tools for maintaining “internal security.” In both countries, tax reform was heralded as an important initiative necessary to modernize the state and the respective capitalist economies. State executives attempted to achieve greater revenue yields, more efficiently, with fewer exemptions and loopholes, and with more modern technologies. During the period since the end of the Second World War, the respective states had offered various tax expenditures to encourage new investment. These included increasingly generous allowances for deducting capital expenditures, and selective exemptions for various types of organizations deemed to be in the public interest.94 The accumulation of reform needs combined with concrete political challenges provided state executives specific opportunities to close these loopholes and to make new demands for sacrifice. Initially, the support of the business sector in both countries was achieved through the idiom of Cold War standoffs that essentially pitted an “Anti-Commu- nist” state against “Communist” segments of society. Rather quickly, however, “Cold War” political conflicts were reinterpreted with preexisting definitions of NPC and associated identities and political idioms, producing different patterns of cooperation and compliance. In South Africa, Cold War political conflict was interpreted in largely racial terms. Apartheid state leaders articulated a vision in which White South Africa was an anti-Communist bastion within Black Africa, and this frame helped to cement together white interests. Although in South Africa, many business leaders (particularly within the manufacturing and commercial sectors) espoused a mix of pragmatic and truly empathic disdain for the degree to which racial legislation had been codified in South Africa, they did this while accommodating this order, often arguing that they were simply too politically weak to challenge the apartheid system.95 Although there was disagreement within the business community concerning its commitment to apartheid, such schisms were not manifest in more narrow sectoral or regional interests that might make the functioning of the tax system a highly politicized affair among whites. As Greenberg points out, soon after Sharpeville, commercial interests explicitly engaged the National Party government with words like, “accommodation,” “trusted partners,” and “cooperation.”96 White South Africans found themselves more unified than ever before, particularly as the “poor White” problem was being stamped out, and the state found it could count on business leaders to share the responsibility of defending the country against Communism. The glue of racial identity helped weld together potentially conflicting ethnic, regional, and sectoral interests.97 Moreover, apartheid legislation and strategies reinforced the degree to which physical space or territory became politically meaningful in racial terms. The regulation of the move- 544 POLITICS & SOCIETY ment of black people into and out of “white” areas, and the creation of a homeland system further widened the gap between white and black, further minimizing perceived regional differences within white areas, and closing out the possibility for cross-racial alliances within regions. The war on Communism was also advanced in Brazil as a way of legitimizing the state and its repressive policies, but internal political cleavages that had divided the economic elite in previous years were rearticulated only a few years after the military coup of 1964. A parochial logic soon crept in to the politics of taxation, as regional patterns of business organization, and discourse surrounding distributive politics led to intraelite conflicts. Political power bases remained at this level, and even the centralizing military government still relied heavily on governors and other state (estado) elites to carry out its policies.98 Moreover, racial inequalities and race chauvinism played a very different role than in South Africa as these factors served to further undermine collective solidarity in Brazil. Offi- cially, Brazilians sought to build on the idea of racial tolerance. Gilberto Freyre, a leading Sociologist and social commentator who became closely tied to the Bra- zilian nation-building project argued that anyone who challenged the notion of a racial democracy was Communist and anti-Brazilian—again, exactly the opposite configuration of race and nation when compared with South Africa’s state policies. In this vein, the military government went on to declare as subversive the research of several scholars exposing racial inequality and discrimination in the country.99 Nevertheless, survey research and anthropological research have identified the existence of pervasive race chauvinism and racial inequalities pervading Brazil during this period.100 Together, the state’s definition of the NPC and pervasive race chauvinism emerged as contradictory forces, undermining collective solidarity. In a society that prided itself in being a “racial democracy,” the only acceptable channel for expressing prejudices and feelings of difference continued to be the regionalist discourse that pitted the wealthy South and Southeast against the poorer North and Northeastern states. The varied salience of political identities was instrumental in reproducing nationally distinctive sets of class relations, which in turn affected the calculations and strategies of upper groups responding to state demands for tax reform and additional revenue. White, upper groups in South Africa reasoned that the state was advancing common class interests, particularly with its war on Communism. For example, the chief economist of one of the country’s leading industrial holding companies recalled in an interview that his firm was insistent on cooperation and compliance when it came to paying taxes, because the objectives of the firm and the government were basically the same, and that they needed one another.101 Another leading businessman from a different firm in a different region recalled with respect to the state demands, “We were fighting a war in those days and we had to pay for this war to keep the Communists out, and the good citizens didn’t mind paying for EVAN S. LIEBERMAN 545 that. It was a sort of conscience money to a certain extent.”102 South African business leaders either actively engaged the state as a partner, or at the very least, did not challenge its legitimacy as the dominant political authority in the country. The state managed to increase tax rates, particularly on top earners, and to remove dated tax incentives when necessary without significant political repercussions. The state also enjoyed high levels of cooperation during the collections process. Beginning in April 1962, the Data Processing Section of the Department of Inland Revenue installed a computer system to calculate and issue assessments, to record payments, and maintain controls. Government reports of the 1960s and the 1980s both reflected on the enormous efficiencies gained from the installation of such equipment.103 A withholding pay-as-you-earn system (PAYE) of tax collection was implemented in 1963, making employers responsible for withholding taxes on a regular basis from employee paychecks. South African companies accepted the extra administrative burdens associated with such reforms, reflecting a long-standing pattern of recognizing the state’s initiatives as ultimately serving the collective interest. For example, the Commissioner of Inland Revenue wrote in his March 1964 report,

The PAYE scheme, which has been under way now for some months, has had its inevitable crop of teething troubles but the indications are that it has been well received on all sides and has the full support of the overwhelming majority of taxpayers.

In Brazil, certain reforms were initially successfully implemented following the 1964 coup. First, the government centralized tax collection. Delfim Netto, the Minister of Finance who orchestrated the centralizing project of the military period, had argued for a much greater level of centralization and more uniform application of the tax system throughout the country.104 The Brazilian government initiated multiple reforms to modernize its tax administration. Most importantly, this involved cracking down on tax fraud. During the Castello Branco regime, the tax authorities automated record keeping and prosecuted delinquents with much tougher inflation-adjusted fines. Such measures helped to increase the real yield of taxes by 34 percent between 1963-1966.105 Also, in 1968, the old tax revenue administration was reconstituted as the Secretaria da Receita Federal (SRF) and began to employ more modern organizational structures and more sophisticated information technologies. As part of these reforms, between 1965 and 1969, the number of federal income tax payers rose from 400,000 to 1.5 million.106 However, the benefits of the initial reforms were short-lived. Particularly after 1974, with the more moderate political approach of President Ernesto Geisel, it became evident that without a clearly defined Communist “other,” Brazilian business did not have anything analogous to the racial “glue” that bound together upper groups in South Africa, and support for the authoritarian regime was more pragmatic than ideological.107 Despite the need for reform, and the presence of 546 POLITICS & SOCIETY highly skilled technocrats and advisors—arguably far more skilled and better trained than in the South African case—state-level elites reclaimed their powerful position relative to the soft-line military government of the 1970s, which was loosening its control on electoral outcomes.108 Netto argues that early successes in achieving state reform were quickly derailed with regional political challenges that undercut the military’s claim to embody the goals of the entire nation.109 In particular, Southern and Southeastern elites complained of a loss of autonomy, reflecting their desire to maintain higher levels of economic resources in their own regions.110 In response to such pressures, the executive granted various regional and sectoral tax incentives during this period, reducing the efficiency and progressivity of the income tax as well as other tax bases, and in turn, these factors led to higher deficits, inflation, and slowed growth.111 Given the breaks received by corporate taxpayers and the numerous exemptions that could be claimed by high-income earners, let alone the evasion of self-employed professionals, the income tax fell mainly on the middle class and sectors of better-off workers.112 Finally, in the wake of multiple conflicts and challenges over how the burden should be allocated, upper groups became increasingly less compliant during the process of collection, and new levels of avoidance and evasion were practiced by the mid-1970s.113

CONCLUSION

This comparative historical analysis of the rise of the income tax in Brazil and South Africa provides substantial insights into the varied trajectories of twentieth century state-building: Economic, international, and cultural factors, as well as a focus on corruption within the state, provided limited explanatory power for understanding the roots of cross-national variation. Rather, the central finding is that definitions of National Political Community were ultimately responsible for substantial differences in the development of national income tax policies and administrative practices. The relative salience of racial and regional identities in South Africa and Brazil framed political competition within and across economic classes, creating incentives and disincentives for upper groups to comply with demands for tax payment. Across three distinctive time periods, during which different external factors motivated state initiatives for tax reform, nationally distinctive political institutions shaped political responses and outcomes in increasingly predictable ways. While both countries underwent democratic transitions in the 1980s and 1990s, the legacies of tax state development inherited from the first three-quarters of the century were very different and continued to influence collections during the democratic era (see Figure 1). In South Africa, the most important source of tax revenue continued to come from direct taxes on income, levied at highly progressive rates, and ultimately paid by a very small minority of the population— EVAN S. LIEBERMAN 547 mostly by “middle class” and wealthy whites. By contrast, in Brazil, although the relative size of the state in the economy was approximately similar, the total tax burden was spread out over a much wider array of taxes, and only a very small portion came from direct income taxes, and levied at much less progressive rates.114 In other words, political contests fought out in much earlier periods largely determined the types of tax systems that were inherited in these highly unequal, recently democratized societies. Given important political changes in South Africa, including the promulgation of a nonracial constitution, there is good reason to believe that upper-income whites will resist payment in the future—and there is good reason to believe that many are trying—but they must resist a well-institutionalized tax system that has managed to collect significant amounts of information and has developed effective rules and procedures for collection during almost a century of state-building. The comparative analysis has revealed that even when race and control of material resources are highly correlated, upper groups do not always perceive their interests as shared—an observation that is sometimes taken for granted when analyzing politics in societies where race is highly salient, such as in the United States and South Africa. Rather, as was demonstrated above, a largely white economic elite in Brazil did not solve the collective action problem of taxation because the more salient regional identity was ultimately a source of division. The South African case reveals the willingness of high-income whites across diverse regions to pay quite a bit in income tax to maintain a coalition with one another and with lower income whites. Through such a coalition, the “poor white” problem was largely solved in South Africa. Yet, it is important to recognize that the willingness of upper-income whites to pay this bill must be regarded as an historically contingent outcome in light of the Brazilian evidence.115 It was only in the context of a socially constructed black “other” that whites in South Africa were willing to pool their resources in the public economy, foregoing the possibility of shifting the tax burden downwards or horizontally through interregional political competition. The legacy of such different patterns of state-building is highly ironic: Today, low-income blacks in South Africa benefit from the progressive income tax that was developed in the wake of this history of deliberate racial exclusion. By contrast, their poor, black Brazilian counterparts continue to face a far more regressive tax system. These findings are suggestive for a broader understanding of the relationship between institutions, identities, and their impact on state development. Spe- cifically, the analysis highlights an important example of “path dependency” in the process of state-building, as early divergences in the specification of national constitutions set states on quite different trajectories, making subsequent convergence on policies and policy outcomes enormously difficult, even when other circumstances and pressures were highly similar. Looking beyond Brazil and South Africa, the argument could be extended to make sense of other country cases, but 548 POLITICS & SOCIETY within limits. The impact of varied definitions of NPC as described above are relevant only for other cases in which the problems of racial and regional heterogeneity have been historically relevant—namely, the “fragment societies” identified by Hartz.116 The impact of exclusionary and inclusionary definitions of NPC could be applied more generally, but to do so, it would be necessary to reconceptualize possible categories of variation with reference to specific potential “insiders” and “outsiders.” Because the notion of a critical juncture is useful only in a structured, comparative analysis, it would also be necessary to specify a group of country cases that share a set of historical circumstances in many of the ways that characterize the comparison between Brazil and South Africa. The argument about federalism applies more widely and helps to explain why countries such as Switzerland and India collect relatively less income tax revenue than other countries at similar levels of economic development. At a more general level, the article demonstrates that negotiated definitions of political community shaped the logic of interclass and intraclass relations in turn affecting the ability of the state executive to gain the cooperation of economically powerful groups. In other words, the authority and efficacy of the modern state is strongly influenced by how group identities get constructed and institutionalized within the political arena.

NOTES

1. Margaret Levi describes this form of compliance as “voluntary because taxpayers choose to pay. It is quasi-voluntary because the noncompliant are subject to coercion—if they are caught.” Margaret Levi, Of Rule and Revenue (Berkeley: University of California Press, 1988), 52. 2. As Joel Migdal points out, a central focus of analysis of the state is concerned with the “yawning gap between state rhetoric and performance.” Joel Migdal, “Studying the State,” in Marc Lichbach and Alan Zuckerman, eds., Comparative Politics (Cambridge, UK: Cambridge University Press, 1997), 211. The notion that state finances are reflective of larger patterns of state development is a central tenet of fiscal sociology. See, for example, Rudolf Goldscheid, “A Sociological Approach to Problems of Public Finance,” in Richard Musgrave and Alan T. Peacock, eds., Classics in the Theory of Public Finance, (London: Macmillan, 1964). 3. Joseph A. Schumpeter, “The Crisis of the Tax State,” in International Economic Papers, Number 4, ed. Alan T. Peacock and et. al. (London: Macmillan, 1954). 4. See, for example, Charles Tilly, ed., The Formation of National States in Western Europe (Princeton, New Jersey: Princeton University Press, 1975). Vito Tanzi, “Quantita- tive Characteristics of the Tax Systems of Developing Countries,” in David Newberry and Nicholas Stern, eds., The Theory of Taxation for Developing Countries, (Washington, DC: Oxford University Press, 1987), 205-241. Harley H. Hinrichs and Harvard University International Program in Taxation, A General Theory of Tax Structure Change During Eco- nomic Development (Cambridge, MA: Law School of Harvard University, 1966). 5. Income and property taxes are reported together in the World Bank dataset because they both tend to be direct, progressive taxes. In most countries, including Brazil and South Africa, the property tax is an insignificant source of revenue at the national level. EVAN S. LIEBERMAN 549

6. Excludes oil-exporting, and former Communist countries, and countries with populations of less than 1 million because income tax collections likely reflect different types of political and administrative solutions in these countries. Nonetheless, the correlation remains moderately strong (above .50) even when these countries are included and for other time periods. 7. Other “economic” variables, such as levels of trade, or agricultural share of value-added, provide no significant additional explanatory power in multivariate regres- sion analyses. 8. In this sense, the South African economy developed with a strategy that was quite similar to that used in Brazil and elsewhere in Latin America. See, for example, Victor Bulmer-Thomas, The Economic History of Latin America Since Independence (New York: Cambridge University Press, 1994). For the South African case, see Stuart Jones and Andre Muller, SouthAfrican Economy, 1910-90 (New York: St. Martin’s Press, 1992), and N. Nattrass and E. Ardington, eds., The Political Economy of South Africa (Cape Town: Oxford University Press, 1990). 9. Dan O’Meara, Volkskaapitalisme: Class, Capital and Ideology in the Develop- ment of Afrikaner Nationalism, 1934-1948 (Johannesburg: Ravan Press, 1983), 247. 10. Levi, Of Rule and Revenue. 11. For a discussion of the influence of “emergencies” on the development of the American tax system, see W. Elliot Brownlee, Federal Taxation in America: A Short His- tory (New York:Cambridge University Press, 1996). See also, Sven Steinmo, Taxationand Democracy (New Haven, Conn.: Yale University Press, 1993), and Levi, Of Rule and Revenue. 12. Carolyn Webber and Aaron B. Wildavsky, A History of Taxation and Expenditure in the WesternWorld (New York:Simon and Schuster, 1986). See also various references to “culture” in Newbery and Stern, The Theory of Taxation for Developing Countries. 13. In extensive field research, I have found this hypothesis to be the most dominant “folk theorem” of cross-national variation in tax structures. Inglehart advances a variant of this argument in his various reports of the World Values Survey. See, for example, Ronald Inglehart and Marita Carballo, “Does Latin America Exist? (And is There a Confucian Culture?): A Global Analysis of Cross-Cultural Differences,” PS: Political Science and Politics, no. March 1997 (1997): 34-47; and Ronald Inglehart, Modernization and Postmodernization: Cultural, Economic, and Political Change in 43 Societies (Princeton, NJ: Princeton University Press, 1997). 14. The Pearson’s correlation between “Catholics as Percent of Population,” and “Income and Property taxes as share of GDP” was R = –.128, p = .42. 15. Inglehart, Modernization and Postmodernization, 45. 16. Gilberto de Ulhôa Canto, “Study of the Brazilian Tax System in View of the Fed- eral Regime,” Bulletin for International Fiscal Documentation 3, no. 49 (1949), 72-104. Maria Valéria Junho Penna, “The Formation of the Tax State in Brazil: The Genesis of Income Tax,” Textos Para Discussão no. 281 (Rio de Janeiro: Instituto de Economia Indus- trial, Universidade Federal do Rio de Janeiro, 1992). Minstério da Fazenda, Imposto de Renda: 60 Anos No Desenvolvimento (Rio de Janeiro: Ministério da Fazenda, 1982). Thomas Pakenham, The Boer War (London: Futura, 1988). Marius Cloete Van Blerck, Mining Tax in SouthAfrica (Rivonia, South Africa: Taxfax CC, 1992), C-2. Leonard Thompson, The Unification of South Africa 1902-1910 (London: Oxford University Press, 1960), 13. Thomas E. Skidmore, Brazil: Five Centuries of Change, Latin American His- tories (New York:Oxford University Press, 1999), 47. Dauril Alden, Royal Government in Colonial Brazil (Berkeley and Los Angeles: University of California Press, 1968). 550 POLITICS & SOCIETY

17. In particular, see Levi, Of Rule and Revenue, and Margaret Levi, Consent, Dissent, and Patriotism (New York: Cambridge University Press, 1997). 18. This measure is itself a dubious indicator of perceptions of corruption, as it represents the results of evaluations by staff of the American-based political risk service of the, “likeliness to demand special and illegal payments in high and low levels of government.” Nonetheless, these findings resonate with my own extensive interviews with Brazilian and South African business elites which suggest that historically, Brazilians have perceived state corruption as a much greater problem than the South Africans have. 19. Though in both cases, with tragic consequences for human development. See the discussions of these projects as described by James C. Scott, Seeing Like a State: How Cer- tain Schemes to Improve the Human Condition Have Failed (New Haven, Conn.: Yale Uni- versity Press, 1998). 20. To date, there has been no scandal in South Africa which has been widely recognized as evidence of “corruption” that can equal the 1992 resignation of Brazilian President Fernando Collor de Melo in the wake of impeachment proceedings prompted by corrupt behavior. However, not only did this event follow the period under primary investigation in this article, but it was largely hailed as a triumph of democracy because of the demonstrated ability of a system of due process to weed out such behavior. In South Africa, the Truth and Reconciliation Commission has uncovered examples of massive abuse of power during the apartheid era, in which various state actions constituted crimes against humanity. Other recent research suggests that between 1948 and 1990, government contracts were regularly awarded on a nepotistic basis, rather than on the basis of merit. See, Chris Heymans and Barbara Lipiez, “Corruption and Development: Some Perspectives,” Institute for Security Studies Monographno. 40 (http://www.iss.co.za/Pubs/MONOGRAPHS/40/ Corruptiondevelopment.html, 1999). 21. Levi, Consent, Dissent, and Patriotism, 27. 22. As I have shown elsewhere, individual perceptions of getting a “fair deal” and of their inclinations to comply with tax obligations are strongly influenced by their level of agreement with the definition NPC. Evan Lieberman, “Through Rainbow-Tinted Glasses: How South African Citizens Evaluate Their Economic Obligations to the State,” Journal of Development Studies (forthcoming). 23. Ruth Berins Collier and David Collier, Shaping the Political Arena (Princeton: Princeton University Press, 1991), 29. 24. Thus, in my lexicon, a nation may exist independently of a state, but the National Political Community is the state-sponsored definition, which reflects its attempt to structure the membership of the nation. 25. Reinhard Bendix, Nation Building and Citizenship: Studies of Our Changing Social Order (New York: Wiley, 1964), 90. 26. Benedict Anderson, Imagined Communities: Reflections on the Origin and Spread of Nationalism (London: Verso, 1996). 27. This argument reflects an attempt to extend the core logic of historical institutional analysis outside the institutional environment of the advanced, industrialized countries, generally characterized by stable democratic institutions. The paradigmatic example of this work for the area of taxation is Steinmo, Taxation and Democracy. 28. For more general discussion of path dependency, see Kathleen Thelen, “Historical Institutionalism in Comparative Politics,” Annual Review of Political Science 2 (1999): 369-404; Paul Pierson and Theda Skocpol, “Why History Matters,” APSA-CP Newsletter 10, no. 1 (1999): 29-31; Paul Pierson, “Increasing Returns, Path Dependence, and the Study of Politics,” American Political Science Review 94, no. 2 (2000): 251-267. The para- EVAN S. LIEBERMAN 551 digmatic example of this work is Barrington Moore, Social Origins of Dictatorship and Democracy (Boston: Beacon, 1966). 29. A similar argument about the staying power of tax regimes is made in Edward Ames and Richard T. Rapp, “The Birth and Death of Taxes,” Journal of Economic History 37, no. 1 (1977): 161-178. 30. Louis Hartz describes both Brazil and South Africa as “European fragment” societies. Louis Hartz, The Founding of New Societies: Studies in the History of the United States, Latin America, SouthAfrica, Canada, and Australia (New York: Harcourt Brace & World, 1964). 31. See, for example, Richard Elphick and Hermann Giliomee, eds., The Shaping of SouthAfrican Society , 1652-1820 (Middletown, CT: Wesleyan University Press, 1989), and George M. Fredrickson, White Supremacy: A Comparative Study in American and SouthAfrican History (New York: Oxford University Press, 1982), for two of the best historical accounts of the early manifestations of white supremacy in Southern Africa. 32. Anthony Marx, Making Race and Nation (Cambridge: Cambridge University Press, 1998), 49. 33. George Reid Andrews, “Black Political Protest in São Paulo, 1888-1988,” Journal of Latin American Studies 24 (1992): 147-171. 34. The 1911 South African census reported that 21 percent of the population was white, while the 1872 Brazilian census cited 38 percent white. Indeed, a much higher degree of “mixing” had taken place in Brazil, producing a relatively much larger mulatto population (only one-third of the nonwhite population in Brazil was counted as “pure” black, while more than 80 percent of the nonwhite population in South Africa was considered “Black” or “African”). South Africa Bureau of Census and Statistics, Union statistics for Fifty Years, 1910-1960 (Pretoria, Union of South Africa: 1960). Fundação Instituto Brasileiro de Geografia e Estatísticas—IBGE, Estatísticas Históricas do Brasil: Series Econômicas, Demográficas e Sociais de 1550 a 1988 (Rio de Janeiro, Brazil: IBGE, 1990). Since all of these categories are socially constructed without any clear scientific divide, the categories and criteria for racial group membership are arbitrary and vary over time. See Andrews, “Black Political Protest,” and Cleusa Turra and Gustavo Venturi, eds., Racismo Cordial (São Paulo: Editora Ática, 1995), for a discussion of these labels. 35. Pakenham, The Boer War, 576. 36. Only in the Cape Province, where blacks previously had been allowed to vote, was the franchise initially afforded on a nonracial basis. In 1934, African voters from the Cape were removed from the voters’ role, and in 1948, the remaining Coloured voters from the Cape were removed from the voter role. 37. John Cell, The Highest Stage of White Supremacy: The Origins of Segregation in SouthAfrica and theAmerican South (Cambridge Cambridgeshire; New York:Cambridge University Press, 1982), 64. 38. In administrative terms, the Union of South Africa was a federal government, characterized by three “tiers” of government (and a parallel native administration). In political terms, however, it was still a unitary state because subnational tiers were afforded very little say in decision making and virtually no autonomy. 39. Thompson, The Unification of South Africa, 243. 40. In the first two decades of the Republic, a proposal to establish a formal color-bar was debated in the parliament and ultimately rejected. Marx, Making Race and Nation, 166. 41. Melissa Nobles, Shades of Citizenship: Race and the Census in Modern Politics (Stanford, CA: Stanford University Press, 2000), 104. 552 POLITICS & SOCIETY

42. See, for example, Thomas E. Skidmore, Black into White: Race and Nationality in Brazilian Thought (Durham: Duke University Press, 1995). Gay Seidman, Manufacturing Militance: Workers’Movements in Brazil and SouthAfrica 1970-85 (Berkeley, California: University of California Press, 1994), 22. 43. Collier and Collier, Shaping the Political Arena, 31. 44. For example, between 1910 and 1961, the top two parties secured more than 63 percent of the seats in the parliament, and on average, the top two parties secured a full 88 percent of the seats. Author analysis of election returns from TRH Davenport, South Africa: A Modern History (Toronto: University of Toronto Press, 1991), 564-65. 45. White supremacy surely reaped concrete material gains for capital interests across sectors. “The alliance of gold and maize”—the cooperative arrangements between mining magnates and commercial farmers—helped to maintain institutions that provided a steady stream of cheap labor: the large black population living in the region, yet deprived of the rights of citizenship. See, Cell, The Highest Stage, 63. 46. South Africa. Bureau of Census and Statistics, Union Statistics. 47. Hannah Arendt, The Origins of Totalitarianism (USA: Harcourt, Brace, Jovanovich, 1979), 194. 48. Robert Davies et al., “Class Struggle and the Periodisation of the State in South Africa,” Review of African Political Economy 7, (Sept-Dec 1976), 11. The best discussion of the incorporation of white labor in South Africa is, David Yudelman, The Emergence of Modern South Africa: State, Capital, and the Incorporation of Organized Labor on the South African Gold Fields, 1902-1939 (Westport, Conn.: Greenwood Press, 1983). 49. Boris Fausto, “Brazil: The Social and Political Structure of the First Republic,” in Leslie Bethell, ed., The Cambridge History of Latin America (Cambridge: Cambridge Uni- versity Press), 788. 50. See Fernando Luiz Abrucio, Os Barões da Federação: Os Governadores e a Redemocratização Brasileira (São Paulo: Editora Hucitec, 1998), 35-41. Virtually all studies have concluded that the corporatist organization, the National Confederation of Industries (CNI) has been neither powerful, nor able to broadly represent national business interests in Brazil. Philippe Schmitter, Interest Conflict and Political Change in Brazil (Stanford, CA: Stanford University Press, 1971). Nathaniel Leff, Economic Policy Making and Development in Brazil 1947-64 (New York: John Wiley, 1968). Leigh A. Payne, Bra- zilian Industrialists and Democratic Change (Baltimore: Johns Hopkins University Press, 1994). 51. Leff, Economic Policy Making, 116. 52. Skidmore, Black into White, 39. 53. Brazil’s bizarre strategy for addressing the racial problem involved the promotion of miscegenation to achieve a “whiter” population. Skidmore, Black Into White, 173. 54. In fact, as Andrews argues, the strategy was quite the opposite: “Acutely aware of the tactical opportunities which an ethnically and racially divided working class offered to employers and the state, and inspired by the egalitarian doctrines of socialism, anarchism, and anarchosyndicalism, labor organizers repeatedly invoked the goal of eliminating such divisions.” George Reid Andrews, Blacks & Whites in São Paulo, Brazil, 1888-1988 (Mad- ison, Wis.: University of Wisconsin Press, 1991), 61. 55. Fausto, “Brazil: The Social and Political Structure,” 810-11. 56. Seidman, Manufacturing Militance. Collier and Collier, Shaping the Political Arena, 186. 57. To be clear analytically: During all three periods, the key explanatory variable, the definition of NPC remains unchanged; each period is marked by a distinct change in a “background” explanatory variable; and the outcomes, measured in terms of tax policy, EVAN S. LIEBERMAN 553 administration, and collections are compared in terms of similarities and differences in over-time change. For a discussion of strategies for causal inference in Historical Institutionalism see, Evan Lieberman, “Causal Inference in Historical Institutional Analy- sis: A Specification of Periodization Strategies,” Comparative Political Studies 34, no. 9 (November 2001), 1011-1035. 58. Annibal V. Villela and Wilson Suzigan, Government Policy and the Economic Growthof Brazil, 1889-1945, Brazilian Economic Studies no. 3 (Rio de Janeiro: Instituto de Planejamento Econômico e Social/ Instituto de Pesquisas (INPES), 1977), 95. 59. In fact, as early as 1889, during the Brazilian constitutional convention, the idea of a general income tax had been raised and proposed within legislative meetings and in the broader public arena. Penna, “The Formation of the Tax State in Brazil,” 31. 60. Benedito Ferreira, Legislação Tributária: A Historia da Tributação no Brasil (Brasília: Senado Federal, Centro Grafico, 1986), 74-5. 61. Penna, “The Formation of the Tax State in Brazil,” 40. 62. Ibid., 36. 63. Alcides Jorge Costa, “História do Direito Tributário,” in Revista Especial do Tri- bunal Regional Federal (São Paulo: Imprensa Oficial do Estado, 1995), 45. 64. Escola de Administração Fazendaria (ESAF), Sistema Tributário: Características Gerais, Tendências Internacionais e Administração (Brasília: Secretaria da Receita Fed- eral, 1994), 44-45. 65. Folha de São Paulo (19 October 1926). 66. Edgard Carone, O Centro Industrial do Rio de Janeiro e Sua Importante Participação na Economia Nacional, 1827-1977, (Rio de Janeiro: Centro Industrial do Rio de Janeiro, 1978), 97. 67. Ibid. Penna, “The Formation of the Tax State in Brazil,” 41. Villela and Suzigan, Government Policy and the Economic Growth of Brazil, 285-6. 68. “Course of Legislation: Income Tax Law,” Declassified letter from the American Embassy, Rio de Janeiro, dates 29 May 1926. Published on Microfilm Reel 31 of United States. Dept. of State., Records of the Department of State Relating to the Internal Affairs of Brazil. 69. In particular, very supportive statements for the new taxes were articulated in the meetings of the Associated Chambers of Commerce and Chamber of Mines. 70. Thompson, The Unification of South Africa. Marx, Making Race and Nation. 71. Budget Speech, Union of South Africa. House of Assembly Debates (24 April 1914), 1936. 72. JB Moffat, “Report on the Working of the Income Tax Act, 1914, for the Year ended 30th June, 1915,” (Pretoria: Union of South Africa, 1915). 73. Fausto, “Brazil: The Social and Political Structure,” 815. 74. Villela and Suzigan, Government Policy and the Economic Growth of Brazil, 174. 75. Werner Baer, The Brazilian Economy: Growth and Development (Westport: Praeger Publishers, 1995), 38-40. Villela and Suzigan, Government Policy and the Eco- nomic Growth of Brazil, 167. 76. Ibid., 181. 77. Thomas Skidmore, Politics in Brazil, 1930-1964 (New York: Oxford University Press, 1967), 30. 78. Fausto, “Brazil: The Social and Political Structure,” 807. 79. Ibid., 808. 80. See Abrucio, Os Barões da Federação, 46-8, for a discussion of the corporatist- federative arrangements of the Estado Novo. 81. Canto, “Study of the Brazilian Tax System,” 70. ESAF, Sistema Tributário, 49. 82. “Associaçao Comercial de São Paulo,” O Estado de São Paulo (6 January 1944), 2. 554 POLITICS & SOCIETY

83. Valentim F. Bouças, “Os Tributos e sua Arrecadação em Face da Constituição,” Boletim do Conselho Técnico de Economia e Finanças, no. 68-9 (1946): 4-6. 84. Davenport, South Africa: A Modern History, 564. 85. Many Africans and Coloureds participated in the war, as unarmed members of South African auxiliary units. Davenport, SouthAfrica: A Modern History , 299. Of the approximately 5,500 South African battle deaths, approximately half were African. Leon- ard Thompson, A History of SouthAfrica (New Haven, Conn.: Yale University Press, 1990), 177. 86. A large segment of the economically weaker, Dutch-speaking population challenged the state’s support for the Allied cause. 87. J. C. Smuts, 4 November 1941 speech reported in the 1941 Assocom minutes. 88. Social and Economic Planning Council, Report No. 7 of the Social and Economic Planning Council into the Tax System (Pretoria: Government Printers, 1946). 89. PM Anderson, President of the Chamber of Mines. Annual Report, 1941. 90. (Editorial) Rand Daily Mail (26 February 1942). 91. See the various contributions in David Collier, The New Authoritarianism in Latin America (Princeton: Princeton University Press, 1979), for a discussion of the “Bureau- cratic-Authoritarian” model in Brazil. The descriptions for this ideal-type are in many respects also appropriate for the South African case. See Seidman, Manufacturing Militance, 94. 92. In South Africa, in 1952, the ANC launched a large passive resistance campaign, and in 1955, approximately 3,000 people gathered to adopt the Freedom Charter, calling for an end to institutionalized racism. In Brazil, political polarization in the early 1960s emerged between the more conservative congress on the one hand, and workers and peasants with the support of President Goulart on the other. In March 1964, Goulart led a Rio de Janeiro rally of working class people, calling for “basic reforms,” a stance interpreted as highly threatening to the Brazilian business community. 93. The pass laws were a key pillar of apartheid legislation that regulated the flow of people into and out of various areas within Southern Africa based on identity documents and racial classification. 94. By the late 1950s and early 1960s, policy makers in both countries were expressing increased desires to reform the respective tax systems, and to modernize them with new technologies, and in fact, study commissions from as early as the mid-1940s presented arguments for long-term tax reform. For example, the study reports issued by Special Com- mission of the Tributary Code (1950s and 1960s) and the joint Fundação Getúlio Vargas/ Inter-American Development Bank commission (1960s) in Brazil; and the Social and Eco- nomic Planning Council (1940s), and the Holloway (1946), Steyn (1954), and Schumann Commissions (1966) in South Africa. 95. Stanley Greenberg, Race and State in Capitalist Development: SouthAfrica in Comparative Perspective (Johannesburg: Ravan Press, 1980). Thompson, A History of South Africa, 206-7. 96. The Manufacturer (January 1966, 14), as cited by Greenberg, Race and Nation, 205. 97. Marx, Making Race and Nation; Thompson, A History of SouthAfrica . In particular, see the discussion of Verwoerd’s conciliatory stance toward the English-speaking population following the declaration of the South African Republic in Davenport, South Africa: A Modern History, 360-2. 98. Abrucio, Os Barões da Federação. Frances Hagopian, Traditional Politics and Regime Change in Brazil, Cambridge studies in comparative politics (Cambridge; New York: Cambridge University Press, 1996). 99. Andrews, Blacks & Whites in São Paul, 7. EVAN S. LIEBERMAN 555

100. See, for example, Carlos A. Hasenbalg, “Desigualdades Raciais no Brasil,” Dados, no. 14 (1977): 7-33. Darcy Ribeiro, O Povo Brasileiro (São Paulo: Companhia das Letras, 1995). 101. Personal Interview, Sandton, South Africa (31 July 1998). 102. Personal Interview, Johannesburg South Africa (1 June 1998). 103. See the Report of the Secretary for Inland Revenue for the year 1965-1966 (Preto- ria, March 1968), and the Margo Report (1986). 104. Revista de Finanças Públicas, (September 1968), 2 105. Howard S. Ellis, The Economy of Brazil. (Berkeley: University of California Press, 1969), 182. 106. Hagopian, Traditional Politics and Regime Change, 70. 107. Leigh A. Payne, Brazilian Industrialists and Democratic Change (Baltimore: Johns Hopkins University Press, 1994), 51. 108. Abrucio, Os Barões da Federação, 82-90. 109. Personal Interview, Delfim Netto, Brasília, Brazil (15 April 1999). 110. Hagopian, Traditional Politics and Regime Change, 120. 111. Ricardo Varsano, “The Tax System of 1967: Is It Still Adequate for Brazil in the 1980s,” in Brazilian Economic Studies (Rio de Janeiro: Instituto de Planejamento Econômico e Social, 1984), 338, 345. Dennis J. Mahar, “Fiscal Incentives and the Eco- nomic Development of the Western Amazonia,” Brazilian Economic Studies 2 (1975), 149. 112. Kurt Weyland, Democracy Without Equity: Failures of Reform in Brazil (Pitts- burgh: University of Pittsburgh Press, 1996), 81-2. 113. Extensive interviews with SRF officials, tax consultants from various “Big-Six” firms, and business leaders. See also, Ferreira, Legislação Tributária, 139. 114. In 1998, the top personal income tax rate in South Africa was 45 percent, and in Brazil, it was 25 percent. 115. Moreover, although absolute and per capita expenditure on whites was always significantly greater than on blacks, within this highly unequal society, blacks still received more on the expenditure side than they contributed in the form of taxes on the revenue side. M. D. McGrath, “The Racial Distribution of Taxes and State Expenditures,” (Durban: Department of Economics, University of Natal, 1979). 116. Hartz, The Founding of New Societies. Beyond Latin America, the United States, Canada, Australia, and South Africa, other possible cases that would meet the definition of a “fragment” society are New Zealand, Rhodesia/Zimbabwe, Southwest Africa/Namibia, and Israel. POLITICSSEGAL and & THUN SOCIETY

Thinking Globally, Acting Locally: Local Governments, Industrial Sectors, and Development in China

ADAM SEGAL ERIC THUN

This article argues that studies of late-development should be altered in two respects: the unit of analysis should increasingly be the subnational economy, and an understanding of economic outcomes should be sector specific. Variation in the developmental outcomes of particular sectors can be best understood by analyzing the fit between local institutions and firms within a particular sector. The specific development needs of firms vary across sectors, and the institutional structures required to meet firm-level needs are local more often than national. The importance of regional and sectoral variation is demonstrated by comparing Beijing and Shanghai in two sectors.

The story of economic development is often the story of the nation-state. Varia- tion in developmental outcomes across nation-states, according to the standard literature, is the result of national level differences in policy, institutional structures, history, or position in the world economy. Successful national “models” of development are studied, and national policy prescriptions are drawn-up for less successful nations. While the existence of differences within individual nation-states is acknowledged in these standard accounts, it is generally a matter of specific regions lagging behind the rest of the nation. In Italy, for example, the southern states have lagged behind the northern; in Japan, prefectures along the

The authors would like to thank the following for their helpful comments and suggestions: Rawi Abdelal, Suzanne Berger, Mary Gallagher, Peter Katzenstein, Elizabeth Remick, Vivienne Shue, and the editorial board of Politics & Society. All errors, either of fact or interpretation, remain the authors’. POLITICS & SOCIETY, Vol. 29 No. 4, December 2001 557-588 © 2001 Sage Publications 557 558 POLITICS & SOCIETY

Sea of Japan regions have developed more slowly than those along the Pacific; in the United States, the south has traditionally been a step behind the north. Regional variation is a result of national policies not penetrating deeply enough, not providing equal opportunities to all regions. This article takes a different view in two respects. First, although national institutional characteristics provide the overall framework for growth and regulate the overall process, variation in developmental outcomes is the result of the specific characteristics and abilities of local institutions. We argue that local governments do not simply try to reproduce and catch up with development efforts initiated by the central government, but are often the actual architects of growth, designing and implementing development policies that are conducive to local institutional frameworks and specific development needs. In this respect, our work follows in the tradition of a growing number of studies that have reoriented the study of political economy to the local level.1 Second, we push the local development argument further, arguing that even at the local level there is no “one size fits all” development policy: different industrial sectors have different developmental needs, and policies that work for one sector will not necessarily work for another. It is not enough to say that we must pay attention to local governments. We must also understand what type of local governments can play what types of roles. Given the differing needs of sectors, and that the institutional structures that are required to meet these firm-level needs are local more often than national, it is only by focusing on the “fit” between local institutions and firms within a particular sector that we can understand variations in developmental outcomes. Analysis must be both local and sector specific. The importance of regional and sectoral variation as a means of explaining developmental outcomes is demonstrated empirically by comparing the development efforts of the Chinese municipalities of Beijing and Shanghai in two sectors. Policy-makers in Beijing and Shanghai produced regionally distinct strategies to promote both the automotive and information technology sectors as “pillars” (zhizhu chanye) of the local economy. Local officials failed to differentiate between the organizational needs of firms within the two sectors and failed to understand how the institutional endowments of the locality meshed with these needs. Each city adapted the same developmental approach to the two sectors— despite the differences between them—and succeeded in the one sector for which this approach, serendipitously, was appropriate. This is not to say that geography is destiny, that general lessons cannot be drawn from the specific case studies. But comparisons across cases must be sensi- tive to basic differences across sectors and the institutional endowments of a particular region. In the first section of the paper we make the case for an approach to comparative political economy that focuses on the subnational and the sectoral level. Understanding development outcomes, particularly in less developed economies, requires a focus on the institutional endowments and policy traditions of SEGAL and THUN 559 different regions, and the degree to which these features meet the development needs of firms within different industrial sectors. This argument is then supported in the second section with the empirical evidence from the case studies of Beijing and Shanghai. Finally, we should note that while this paper analyzes how different regional developmental approaches lead to different outcomes, it does not address the question of why each of these regions initially chose to pursue very different policies. This equally interesting question is the subject of another paper.2

REGIONS, SECTORS, AND ECONOMIC DEVELOPMENT

Understanding the relationship between institutions and economic performance has always been central to the field of comparative political economy. Inspired by the work of Andrew Shonfield, the dominant approaches to political economy take the nation-state as the basic unit of analysis and explain economic outcomes by analyzing the relationship between domestic state institutions, patterns of industrial policy, and social actors.3 Successive generations of this approach have analyzed how alternative national institutional structures respond to different challenges in the developed world. Early variants used distinctions between “strong” and “weak” state structures to understand industrial intervention;4 later approaches used the concept of neo-corporatism to explain the ability of governments to forge cooperative relationships with producer groups in an open economy.5 Most recently, the focus has been on the varieties of capitalism in the developed world, and whether the increasing integration of the global economy will ultimately lead to the convergence of different national models.6 Studies of late-development have generally followed from this traditional approach in that successful development is seen as a function of getting national political and economic institutions right. Lessons are drawn from analyzing the institutional arrangements of successful cases of development, and prescriptions are drawn for less successful cases. Advocates of the developmental state approach, for instance, argue that rapid growth in countries such as Japan, Korea, and Taiwan was the result of an activist government and strong institutions. “Where Korea differs from other late industrializing countries,” argues Alice Amsden, “is the discipline its state exercises over private firms.”7 The state was able to direct capital into targeted sectors, and large multidivisional enterprises structured the economy and served as the agent of expansion and development. In Taiwan, according to Robert Wade, corporatist and authoritarian political arrangements provided the basis for market guidance, “augmenting the supply of investible resources, spreading or ‘socializing’ the risks attached to long-term investment, and steering the allocation of investment by methods that combine government and entrepreneurial preferences.”8 Others, of course, make the opposite argument. Success was the result not of an activist government, but a government that allowed markets to operate freely; protection was minimized, exports 560 POLITICS & SOCIETY were promoted, and a stable macroeconomic was maintained. As the World Bank concluded, high growth was achieved by getting the basics right.9 Although these approaches clearly differ (and often bitterly) over what the “right” policies are, they all agree that the study of economic development should be focused on the national level.

Regional Diversity

A number of political economists working on Europe have already questioned the traditional account of state building as the almost inevitable absorption of previously independent administrative units by a centralizing power. As Richard Locke argues, the actual historical experiences of most modern nation-states were far more complicated than the gradual assimilation of regional units into a central economy.10 Moreover, regional economies continue to exist within the contemporary nation-state. National economies are in fact made up of disparate regional economies.11 Recent research on subnational units of advanced capitalist economies analyzes the manner in which “the micro-agents of capitalist systems— companies, customers, employees, owners of capital—organize and structure their relationships, within a framework of incentives and constraints or ‘rules of the game’ set by a range of market-related institutions within which the micro- agents are embedded.”12 Studies that focus broadly on national macroeconomic planning miss the heterogeneity of regional development efforts. As Gary Herrigel writes with respect to Germany, “different groups of industrial actors...will invariably conceptual- ize, organize, and enact industrial activity in ways that reflect their own pasts.”13 In Baden Württemberg, for example, small and large firms have developed a regionally based comparative advantage lacking in other parts of Germany. Firms rely on extensive links that facilitate information flows, allow for specialization, mediate conflict, and define the terms of cooperation and competition. The regional economy has developed faster than other parts of the national economy by “socializing risk across a broad array of public and private organizations.”14 Similarly, in high-technology sectors, firms in Silicon Valley created a regional economy better able to respond to market changes in the late 1980s than corporations around Route 128 in Boston. Silicon Valley’s dense social networks and open labor markets promoted collective learning and encouraged experimentation and entrepreneurship; firms competed intensely with each other while simultaneously learning from one another about changing technology through informal communication. In contrast, independent firms internalized a wide range of productive activities in the Route 128 region, reinforcing a high degree of self-reliance and market stability, but reducing the ability of firms to react to rapid changes.15 In the developing world, an exclusive focus on the national level risks missing many of the mechanisms that are actually driving development. Because there is SEGAL and THUN 561 less uniformity in countries that are undergoing rapid economic and social change than in countries that have enjoyed long periods of stability, a micro-political approach is even more necessary than in the developed world. India’s federal structure ensures that the process of economic liberalization has not affected the country uniformly, but rather promoted regional variation.16 Even in countries with more centralized government institutions, pockets of development, or “industrial clusters,” have emerged within very low rates of national growth. These clusters are characterized by vertical and horizontal relations through which product arrangements are made, information exchanged, skills transferred, and capital accumulated.17 Economic reform in China produced a similar mosaic effect. The central government expanded both the decision-making authority of local governments and enterprises and their ability to retain the revenue earned within their respective jurisdictions.18 The result was a system that encouraged localities to promote local development, sometimes by implementing a local industrial policy that ignored national objectives. Even on issues officially controlled by the center, a local government’s interpretation and/or the degree of compliance with the dictates of the central government was often the more important determinant of actual policy. Much of the literature on reform and development in China, of course, focuses on local level development. Jean Oi, for instance, argues that increased fiscal incentives gave rise to a new form of state-led growth in rural China, what she calls local state corporatism: a system in which local governments “treat enterprises within their administrative purview as one component of a larger corporate whole.”19 Victor Nee focuses on how increased market pressure leads local governments and private firms to form alliances as protections against an uncertain environment.20 Andrew Walder points to the organizational differences between state-owned firms operating in rural and urban areas as a primary reason that growth in Chinese townships and villages was so much faster than in large municipalities.21 While these studies do not ignore regional differences, they do tend to be grounded in one particular region (or type of region, i.e., urban or rural) focusing on the relationship between a certain set of institutional arrangements and economic development. Few, if any, of the previous studies of China (or of more advanced industrial economies) compare how local governments respond to the challenges of different economic sectors, and so fail to capture the institutional variation between regions. Susan Whiting, for example, looks explicitly at the variation in local institutional structures in her study of rural industry, but rather than relate these difference to development outcomes, her primary purpose is to explain why property rights arrangements evolve in different ways and how this variation affects the extractive capacity of the state.22 We argue that local state corporatism may be an effective developmental strategy, but it depends on the institutional structure of a region and the developmental objectives. The goal of 562 POLITICS & SOCIETY this paper is to begin linking local institutions with the failure or success of specific sectors. The development histories of Beijing and Shanghai are too complex to cover adequately here. The important differences between the two cities, however, can be condensed into two primary variables: the structure of the local bureaucracy (unified or fragmented) and the dominant form of industrial organization (small firms or large state-owned enterprises). Shanghai, with a unified municipal bureaucracy and a local economy dominated by large conglomerates, was essentially a local developmental state. Faced with the challenge of developing a new industrial sector, local officials’ first impulse was to try and build large business groups (jituan gongsi). Large amounts of capital were funneled to state-owned enterprises, and the city government actively participated in the establishment of producer and supplier networks. Beijing, on the other hand, had a more fragmented local bureaucracy, and smaller firms dominated the economy. Pulled between the needs of the city and the national economy, local bureaucrats were unable to work closely with the managers of large state-owned enterprises or to build new business groups. Instead, they promoted a more diffuse form of development that linked smaller firms through personal networks rather than governmental structures. In place of the large, municipally controlled business groups, they praised “innovation networks” (chuangxin wangluo).

Sectoral Differences

Our central point is not simply that the bureaucratic and economic structures differ at the local level in most economies, but that these differences create opportunities that vary by sector. In addition to moving the unit of analysis to the subnational level, it is critical to understand sectoral differences and the developmental needs of individual firms within them. Focusing on economic sectors is increasingly common in the study of advanced capitalist economies and is premised on the belief that the organization, or “governance,” of economic activity varies not only between countries, but also between sectors. Sectors are those subsets of the economy that involve a combination of highly complementary activities (tasks, services, goods, etc.) that must operate in coordination to produce a certain end product.23 Firms within the same sector address similar questions of development and control both inside and outside the enterprise; these firms are more likely to adopt the same production processes, to allocate resources and information in a similar way, and to create equivalent property rights structures. Across sectors, firms face different problems of coordination and regulation. The example of the automotive and information industries illustrates this. There are at least three key characteristics that distinguish automotive firms from those involved in information industries. The first is the capital intensity of SEGAL and THUN 563 the development process. All firms need investment capital during the early stages of development, but in the automotive industry these needs are particularly intense due to the importance of economies of scale and the need for complementary investment. The massive investment in an auto assembly plant, usually in the hundreds of millions of dollars, must be rapid (so that it will be able to quickly ramp up production), and it must be matched with a coordinated investment program in supply firms. A typical automotive supply base consists of hundreds of firms, all of which must be developed simultaneously. Firms involved in the production of computer hardware, software, components, and peripherals — what is broadly known as information technologies (IT)—although certainly in need of capital, have neither the same linkages, nor the compressed time frame. The IT sector is populated more often than not by stand-alone firms that are competing with each other for investment, rather than dependent on each other. The IT industry has space for more firms, including assemblers, Internet providers, chip and software designers, as well as a range of smaller “niche” producers. Furthermore, investment can be in distinct stages, rather than all up-front. High-tech firms need money, but not as much, not all together, and not as quickly. A second key distinguishing factor between auto and IT firms is the relationship between property rights, firm incentives, and the key challenges of the sector. In the auto industry, the key problem during the first stage of development is asset specificity: supply firms are often reluctant to invest in the new technologies and equipment necessary to form linkages with a new joint venture assembly plant not only because the assembly plant initially has very low volumes (and a very uncertain future), but also because they will be totally dependent on only one customer after they have undergone the process of technical upgrading. The mistrust and opportunism that may potentially result from these circumstances can be eliminated by vertical integration24—incentives are aligned when an assembler and suppliers share a common owner—but there is also a cost: a vertically integrated firm can quickly fall prey to the organizational rigidities common to any large organization. It can be difficult both to drive down costs and to promote innovation. In the auto sector, the negative impact of vertical integration during the early stage of development is minor. Because the development process involves learning well-established manufacturing techniques, the hierarchical controls within a vertically integrated firm can be beneficial. In high-tech firms, however, the key challenge in this early stage of development is quite the opposite: the rate of innovation in the industry is extremely rapid, and firms must never lose touch with the market. Rather than making huge investments in stable manufacturing processes that serve a small number of customers, the firms are constantly struggling to keep up with the market in an intensely competitive industry. In an environment of this sort, the ideal organizational structure is one that creates incentives to inno- 564 POLITICS & SOCIETY vate—the less bureaucratic insulation separating the forces of innovation within a firm from the market, the better. Rather than unifying the incentives across the parts of a sector, as in the auto industry, the objective is to maximize the incentives of each individual actor within the sector. A final characteristic that distinguishes auto and high-tech development involves the form of relationships within the sector. This refers to the relationships that a firm manager might depend upon for assistance when a manufacturing problem is encountered, or relationships that might be relied upon to gain information or new ideas about markets, technology, or business. When analyzing the relationships in auto and high-tech sectors, the contrast is between vertical and horizontal patterns. In the auto sector, for instance, firm managers generally rely on the head office of their corporate group. This again is partly a function of the learning process inherent in auto sector development; at this stage, managers are not being taught to be innovative, but are being instructed in established procedures much the way a student is taught mathematical formulas by a teacher. Although it can be extremely useful for managers to compare notes with each other—a form of horizontal interaction—the head office of a corporate group or a local government ministry usually coordinates contact between managers of individual firms. In the absence of the vertical ties, in other words, horizontal ties are rare. In the IT sector, on the other hand, firms benefit from horizontal and often informally organized interpersonal linkages to other firms. Innovation is no longer viewed as a linear process but as the result of complex interactions between various types of actors. These networks replace vertically integrated production as the source of comparative advantage and provide access to highly specialized but complementary knowledge about production, markets, and innovation.25 Even within firms there tends to be an absence of strict hierarchy: authority is based on skill rather than seniority, there are fluid and personal links between departments, and there are flexible definitions of authority and absences of sharp distinctions between levels of authority. The contrast with an auto firm could not be sharper. These sectoral differences have real implications for policy makers. The absence of initial institutional endowments and capabilities limits their ability to produce real sectoral transformation. They must have the ability to respond to firm-specific needs that vary from high capital investment to ensuring the steady flow of highly trained scientists. In the case of IT, governments must intervene in new markets in ways that do not stifle innovation. In developing countries, these may include defining property rights, creating efficient labor markets, and providing venture capital for the most competitive firms. For the development of the automotive sector, government services may be more interventionist in the initial stage of growth and include the provision of investment capital, coordination of SEGAL and THUN 565 the localization process, and the monitoring of technology transfer. The emphasis is on the promotion of learning rather than innovation. In the case studies that follow, our intent is to explain the divergence of developmental outcomes in Shanghai: why could Shanghai successfully develop autos but not high-tech, while Beijing did the reverse? In the auto sector, we define the successful development of a supply sector by analyzing the sourcing patterns of the primary auto assembly projects in each city. The sourcing patterns of the assembly plants provides an appropriate measure of successful development for two reasons: first, the assembly plants are intent on finding high quality parts for the lowest possible price; and second, the local partners in both cases—the municipal governments—have strong financial incentives to give business to local firms. Given the overwhelming preference for using local supply firms, a heavy reliance on outside suppliers is strong evidence of an extremely weak local supply base. After 10 years of development, Shanghai Volkswagen was sourcing 88 to 90 percent of domestically sourced parts within the municipality of Shanghai, while Beijing Jeep could only source 20 to 30 percent of its components from within Beijing.26 Shanghai clearly had the stronger auto sector. In the technology sector, we focus on the growth of a network of competitive, nonstate (or nongovernmental) enterprises in the computer and software industries. Beijing has led Shanghai in both the numbers of these firms as well as their share of national production. By 1993 the nongovernmental enterprises in Beijing as a whole had a total output of 6.75 million RMB—84 percent of the national total for nongovernmental enterprises.27 Total income of all nongovernmental firms in Shanghai equaled less than 1/10 of those in Beijing; exports were a little more than half; and only 16 percent of products were designated “high tech,” as opposed to 25 percent in Beijing.28 From 1988 to 1994, Beijing nongovernmental firms received 32.5 percent of all awards given to National Technology Entrepre- neurs, more than in any other part of the country.29 China’s most successful computer and software companies—Legend and Founders—are both in Beijing, and Shanghai has yet to produce a firm that impacts national or international markets. These patterns appear to have been reproduced in the newest Internet start-ups as well. In July 2000, Shanghai had only 8,457 of the mainland’s domain-name reg- istrations, 8.6 percent of the whole country, and less than a quarter of the number registered in Beijing. It perhaps should not be surprising that Shanghai was more successful in auto sector development—it had a long history in heavy industry—but why did this not translate into successful information technology sector development as well? If Shanghai is a successful local developmental state, shouldn’t this success be enjoyed across sectors? Not only would existing theoretical approaches suggest this outcome, but common sense would as well. First, Shanghai was both the country’s most important supplier of electronics products and a key R&D base at 566 POLITICS & SOCIETY the start of the reform period. Second, Shanghai already played a leading role in the consumer electronics sector. In 1981, for example, Shanghai produced 22 percent of all televisions and 37 percent of all tape recorders in China.30 Finally, even popular stereotypes (for what they are worth) favored Shanghai: Shanghainese are usually viewed as entrepreneurial business people, while their Beijing counterparts are viewed as candid peasants more attuned to politics than business.31 Understanding the mixture of success and failure in Beijing and Shanghai, however, forces us to move beyond generalizations about developmental roles and focus more closely on how the capacity of state institutions and economic organizations at the local level determine a region’s developmental capacity in different sectors.

SHANGHAI: THE DEVELOPMENTAL STATE AT THE LOCAL LEVEL

Shanghai was in many respects a developmental state at the municipal level. The local bureaucracy was organized so as to be able to coordinate effectively firm development, it collected the capital necessary to finance development, and it carefully monitored the overall process. Large state-owned corporate groups (jituan gongsi) dominated the local economy, a legacy of Shanghai’s historical role as the “industrial workhorse” of the planned economy, and this structure com- plemented the interventionist approach of the government. Managers in a head office controlled firms within the hierarchically structured groups, these managers maintained tight linkages to the municipal government, and the government protected the industry from outside competition. In the auto sector, these characteristics facilitated firm development: the government bureaucracy and local auto conglomerate coordinated the financing of local firms, assisted in the process of technology transfer, and monitored the development process as a whole. The city took the same approach to the information industries, but with less satisfactory results. Municipal officials funneled local and foreign direct investment to large state-owned firms and generally ignored the smaller, more flexible nongovernmental technology firms (minban keji qiye) founded by scientists. The municipality of Shanghai has a long history as the industrial center of China. Prior to 1937, approximately half of the large-scale modern industrial production in China was located in Shanghai.32 The city was a booming metropolis, and dominated the national production of textiles, ships, chemical, and metal- working. Both the skills necessary to run these large enterprises and the free- wheeling reputation of the metropolis came in large part from foreigners; the city’s international settlement served as a conduit for foreign technology and managerial and entrepreneurial skills. After the communist victory in 1949, the new leaders of China sought to harness Shanghai’s industrial prowess for the benefit of the nation, and the city was turned into the industrial core of the planned economy. The new regime systematically subjected Shanghai business to state control. SEGAL and THUN 567

The reach of state planning within China varied considerably, and as Lynn White has argued, in Shanghai the planning system was at its most extensive. Fac- tory managers were hemmed in not only by more rules and regulations than existed in other regions, but also by more attentive supervising agencies (zhuguan). In the case of most firms, according to White, a municipal agency (ju) or a state holding company (gongsi) would supervise—they were commonly referred to as “mother-in-laws” (popo)—a group of factories or stores that were spread out across the city. To a far greater extent than in other municipalities, these supervising agencies took the primary responsibility for arranging both upstream and downstream product flows for the factories.33 The local state controlled enterprise operations, but revenue was remitted to the center: between 1949 and 1980, approximately one-sixth of overall central government revenue came from Shanghai, and the city was allowed to retain less than 13 percent of this amount.34 Shanghai continued to be China’s cash cow during the reform era, and as such, the city’s industrial structure was deemed too important to tinker with. It was left to other cities to experiment with new tax rates, stock systems, bankruptcy laws, and short-term money markets during the early reform period. The state leaned on Shanghai as a dependable source of revenue, and Shanghai firms increasingly leaned on the state. Shanghai managers did not search for new forms of promoting growth, and avoided risk whenever possible. It was easier to depend on government subsidies and bailouts than to improve the efficiency of firms, easier to pass problems on to higher level officials. Shanghai gained the reputation of being a “house that follows the law.”35 Although the era of state planning was officially over, its influence endured.

Shanghai Bureaucratic Structure

The past patterns of economic and political associations in Shanghai, when combined with the fiscal incentives the central government designed to promote local development during the reform era, led to the creation of a local developmental state. Power flowed from the top of the municipal bureaucracy down to firms through a tightly organized hierarchical structure; top municipal officials controlled the head offices of the corporate groups in the city, and these head offices, in turn, controlled the individual firms within the group. The unified bureaucratic structure allowed the local government to compensate for underde- veloped financial networks by providing investment capital to firms, and large-scale hierarchical corporations were able to coordinate internal activity, properly structure internal incentives, and monitor the performance of firms. It was a pattern of development that evoked the development approach of East Asian countries such as Japan and Korea. The bureaucratic structure of the Shanghai municipal government was well suited for auto sector development. When the municipality created Shanghai Volkswagen (SVW) in 1984, a joint venture (JV) between the Shanghai Automo- 568 POLITICS & SOCIETY bile Industry Corporation (SAIC) and Volkswagen AG, the objective was not simply to make SVW the dominant assembly plant in China, but also to make the automobile industry a “pillar” of the local economy. Local officials hoped to capi- talize on the extensive backward linkages of the auto industry to increase the technical capability and manufacturing skills of hundreds of supply firms. Given the capital intensity of the industry, however, this was a daunting task. State-owned firms in Shanghai had little capital of their own, and even when they did have access to sources of capital, there was a great reluctance to assume any risk. In essence, the classic problem of late development—the need for broad, simultaneous investment—gave rise to a collective action problem within the auto sector. Acting individually each individual supply firm had no guarantee that coordinated investment would occur, and the risk of investing alone was very high. If production volumes did not increase, no supply firm would be able to earn a return on an investment. But if supply firms upgraded their technical capacities in a coordinated manner, the increase in domestic content would allow SVW to rapidly raise production volumes, capture market share, and both SVW and the suppliers would prosper.36 The answer to this coordination problem came through a slight reorganization of the municipal bureaucracy and the creation of a “localization” tax. In China, governmental structures and decision making are usually fragmented, and thus coordination and control can be achieved across separate ministries only if a coordinating office has sufficient power. Shanghai created such an office—the Auto- mobile Industry Leading Small Group—in 1987. It was situated directly under- neath the mayor’s office and composed of the directors of every government office that might potentially affect development of the city’s automobile industry.37 This Leading Small Group supervised the activities of what became known as the Local- ization Office, the office that controlled the day-to-day operations of the Shanghai auto industry. In 1988, with a unified institutional structure in place, the municipal government began levying a “localization” tax of 28,000 RMB (at the time an amount equal to 16 percent of the total price) on every VW Santana that was sold, the proceeds going into a fund controlled by the Localization Office. From 1988 to 1994 the municipality collected more than 5 billion RMB, all of which was ear- marked for an import substitution program for local supply firms.38 Although the institutional structure of the Shanghai municipal government effectively directed investment capital to state-owned enterprises in the auto sector, applying the same strategy in the IT sector led to the marginalization of the smaller, more flexible nongovernmental firms. It was these firms that were most likely to be innovative and able to effectively respond to a rapidly changing market. Officially, these firms were not part of the state, collective, or private sectors; instead, “nongovernmental” came to cover those firms which did not receive state funding and maintained an independent ability to develop new technologies and SEGAL and THUN 569 handle internal management. In many cases these firms were commercial ven- tures established by the Chinese Academy of Science which, facing reductions in its state-sponsored budget, used them to market new technologies. Nongovernmental firms often had very unclear property relations with their original units, and tensions often arose over profit retention, development strategies, and management styles. From the beginning of the reform process in Shanghai, it was clear that nongovernmental technology firms were not at the center of the local development plan. As one prominent official explained: “the main part of scientific progress is the enterprise, especially those large state-owned enterprises that occupy a dominant position in the national economy.”39 In Shanghai, reforms of the central state science and technology system did not result in the devolution of economic power to independent enterprise level units, as occurred in Beijing, but in a strengthening of the administrative controls of the municipal government. For example, in response to the March 1985 Central Com- mittee Decision on Reforming S&T Management, the Ministry of Electronics (MEI) institutionalized a system of branch management to overcome the vertical barriers that separated the various units involved in the electronics industry.40 All enterprises under the direct control of the MEI were to be placed under local supervision in Shanghai. After being removed from MEI control, these enterprises were linked to local government institutions to form branch corporations that would integrate research, production, sales, and service institutions.41 A policy decision that helped shore up the growing nongovernmental industry in Beijing was used to increase local government efficacy in Shanghai. This focus on economic planning and government guidance in Shanghai is reflected in the balance of power among local government institutions. Within the local government, ministries traditionally active in the state plan and sympathetic to the plight of SOEs, like the planning and economic commissions (jiwei and jingwei), dominate bureaucratic politics. Members of the Science and Technol- ogy Commission rarely work their way up the ladder of city politics, and they have never been appointed mayor or vice mayor. In fact, one interviewee reported that members of the STC were the least respected in city government, seen to have little real economic knowledge, and often derided as “bookworms” (shudaizi).42 To be fair, the Shanghai municipal government had a limited number of financial tools available to support the growth of technology enterprises. Bank lending had to be directed to foundering SOEs both for political reasons—primarily the fear of mass worker discontent and unemployment—and because lending to nonstate firms was believed to be more risky. Between 1985 and 1994, Shanghai financial organs and banks issued loans for only 300 million RMB to high-tech enterprises.43 Most of the small technology firms were forced to rely on the capital raised by their founders and their friends, or on the limited investment made by the supervisory agency (zhuguan danwei). But even when the Shanghai local govern- 570 POLITICS & SOCIETY ment pursued other means of raising capital, it almost always directed the funds at large state-owned enterprises. Foreign investment, for example, focused on large projects and on large firms such as Shanghai Belling, a joint venture between a large SOE and the Belgian Bell Company, involved in more manufacturing intensive technologies like integrated circuits. By 1993, multinational corporations (MNCs) had invested a total of 190 million US in twenty-seven firms, or an average of 7 million US per venture. In Beijing, by contrast, smaller investments were distributed among a wider variety of players: MNCs averaged an investment of 710,000 US in a total seventy firms for a total of 49 million US.44 Funding for research and development work in Shanghai also reinforced the dominant role of SOEs and the vertical relationships that linked scientific institutes and production units. R&D activities in Shanghai were concentrated in large and medium sized state-owned enterprises. Shanghai invested six times more than Beijing in the R&D activities of large and medium sized SOEs, but only half the amount of Beijing in scientific research and development organizations.45 Focusing R&D funding within the labs of SOEs hampered the growth of cooperative links between research institutes and productive units, insulated the R&D work from market forces, and made the diversion of funds to worker bonuses or other unproductive diversions more likely.46 A unified bureaucratic structure concentrated investment in the IT sector, just as it did in the auto sector, but because of the different developmental requirements of an IT sector it had the opposite effect: channeling investment to large SOEs allowed auto firms to import technology and rapidly increase volumes—critical ingredients of success for this sector—but this same approach in the IT sector interfered with the development of small, innovate firms that are so central to high-tech success.

Shanghai Economic Structure

The Shanghai economy was dominated by large, state-owned corporate groups, and these groups were the primary beneficiaries of government support. In the auto sector this did not pose a problem because these corporate structures eased the problem of asset specificity and unified incentives within the sector in favor of supplier development. The Shanghai Automobile Industry Corporation, a large hierarchically organized corporate group fully owned by the municipal government, was not only the local partner of Shanghai Volkswagen, but also the owner of the vast majority of supply firms in the municipality. It held all the relevant strings in the local auto sector: it controlled both purchasing at SVW and the finances and personnel of supply firms within the group.47 Consequently, when a supply firm was deciding to take the risk of investing in the technology necessary to supply the new JV assembly plant (that was operating at very low volumes due to the exorbitant cost of importing components) it could be assured not only that it would have a guaranteed customer in SVW, but also that other firms in the network would simultaneously be making the same investments. SEGAL and THUN 571

The head office of SAIC did everything possible to assist firms within the group that were struggling. A firm that was not performing well might be given a new manager, provided with a new product to manufacture, given a slightly higher price for the component it sold SVW (since SAIC also controlled SVW purchasing), or perhaps provided with the capital necessary for expansion and technical upgrading. Because SAIC made sure that there was no duplication within the supply network and also controlled the purchasing office at SVW, the supply firms knew that they would have a guaranteed market when they finished the process of technical upgrading. Furthermore, each individual supply firm realized that it was a simultaneous process. Using the capital collected by means of the localization tax, the head office of SAIC, working with the Localization Office (the director of the Localization Office served simultaneously as the president of SAIC) directed a straightforward import substitution policy, with import defined as any component purchased outside of the Shanghai municipality; a component sourced from a Beijing firm was viewed in the same manner as a component sourced from an American firm. The two offices examined the list of components that SVW was “importing,” and determined which could most successfully be produced by Shanghai factories.48 Potential factories were considered, meetings were held with factory managers to discuss a particular factory’s needs and capabilities, and an evaluation made. Based upon the evaluation, a decision was made as to whether a particular factory had the capability to manufacture a given component, advice was given concerning the equipment and technology that had to be imported, and the necessary investment capital was provided.49 Because all firms in the group were going through the same process, the assembly plant would be able to rapidly increase local content, raise volumes and lower costs, and each investment would pay off. The hierarchical structure of the Shanghai auto sector also proved to be very conducive to learning new manufacturing techniques. Both the Localization Office and the SAIC corporate group managed each step of the development process for individual firms. Both provided technical advice and cultivated horizontal relationships between firms within the network. When improvements were made in one firm, the rotation of managers, overlapping memberships of each firm’s board of directors, or the activity of the head office meant that others would soon adapt the same techniques. Similarly, when a supply firm confronted a particular vexing problem, it could turn to the conglomerate head office (or the Local- ization Office).50 Although managers in the head office might not have the answer to the question, they often could point to someone within the conglomerate who could. A Shanghai tire factory was not trying to reinvent the wheel; it sought simply to manufacture a reliable wheel. The oversight and assistance of a tightly structured corporate group helped it to achieve this end. The challenges of development in the Shanghai IT sector were completely different. The primary problem was not coordinating investment across firms and 572 POLITICS & SOCIETY disseminating well-established manufacturing techniques, it was to create highly innovative firms that were in touch with a rapidly changing market. A hierarchically organized corporate group, while very good at the former, is notoriously poor at the latter. In the Shanghai IT sector, the large corporate groups were supported at the expense of smaller, nongovernmental firms. Since the founding of the PRC, Shanghai has had the country’s second largest science and technology base after Beijing. In contrast to Beijing, where the local government provided a range of informal and market enhancing services like introducing entrepreneurs to potential investors, Shanghai nongovernmental firms like Forward and Fuxing operated in an environment of neglect. The municipality’s continued focus on large state-owned enterprises and its inability (or unwillingness) to support smaller scale, more independent firms fragmented technology networks and iso- lated entrepreneurs from the local government and from each other. As one manager at a high technology enterprise explained: “State-owned enterprises are so big in Shanghai there is room for nothing else—they are like trees blocking out the sun.”51 Like in Beijing, small firms had few sources of capital and unclear property rights. Unlike Beijing, the local government did little to provide supervision and services that would help nongovernmental firms overcome these market barriers. Shanghai officials focused on strengthening the position of SOEs in the local economy and discriminated against smaller and more flexible nongovernmental enterprises through funding, regulatory, and supervisory policies. The most important manifestation of this discrimination against smaller, nongovernmental firms was the reluctance with which municipal officials accepted new forms of property rights arrangements that the new enterprises in the information industries were experimenting with. Spun off from local research institutes, universities, or even public enterprises, these firms operated in a “no-man’s land” between the officially recognized categories of “state-owned” and “collective.” According to one newspaper report, technological development lagged because the city lacked policies and regulations for nongovernmental development and local officials failed to provide “systematic guidance” to the sector; policies in support of S&T development made up less than 10 percent of all new regulations issued during the reform period until 1988. To maneuver through the maze of regulations—to receive business licenses as well as establish resi- dence permits for employees—“nongovernmental” had to be recognized by local bureaucrats. Most importantly, failure to define “nongovernmental” as a separate ownership category meant that firms were taxed at the higher rate levied on individual (getihu) enterprises. A technology firm set up by an individual scientist with support of the local Science and Technology Commission had to pay annual taxes of 63,074 RMB; if the firm had registered as an SOE its taxes would have only SEGAL and THUN 573 equaled 16,000 RMB.52 As a result, entrepreneurs were setting up companies elsewhere; all of the firms established by scientists from the Shanghai branch of the Chinese Academy of Science were located outside of the city.53 The local government also made little effort to create a social, political, and economic environment that supported entrepreneurship, or at least did not stifle it.54 While the local government did sponsor a number of citywide meetings of S&T personnel, there do not seem to have been in the 1980s any equivalent of the specialized meetings held for nongovernmental entrepreneurs in Beijing in 1984 and beyond, or for that matter the efforts to organize Shanghai SOEs in the auto sector. It was not until 1993 that the city established a consultative committee for S&T development that included Vice Mayor Xu Kuangdi and 26 members of the Acad- emy of Science, and it was not until 1995 that the municipal government held a citywide meeting about nongovernmental enterprises.55 While the process of working with large SOEs was institutionalized in Shanghai, and city officials did so very effectively, when it was necessary to work outside of this structure, officials had little interest or ability. Even with the meetings they did hold, there was a sense among local nongovernmental entrepreneurs that local officials did not understand that the rules of the game for high technologies were different. As one entrepreneur stated: “local officials repeat what they are supposed to say about high-tech development, but they do not truly understand.”56 Shanghai entrepreneurs were aware of the dense social and personal connections that seem to characterize high-tech development; they used scientists in both Silicon Valley and Beijing as reference groups. Yet it was seen to be particularly difficult to develop social relations in Shanghai, either with local officials or with other entrepreneurs. One interviewee complained that those in local government, especially those in the planning commission, knew nothing about technological development. This was understand- able, given their position. What was worse was “that the people in the science commission did nothing for us either: no meetings, no dinners, and no exchanging ideas.”57 Although in the early 1980s local officials did speak out in defense of S&T personnel’s right to look for new jobs outside of the state sector, most entrepreneurs in Shanghai spoke of a continued sense of “social discrimination” (shehui qishi). Residents did not view nongovernmental enterprises as important to the local economy or as a respected career path; in a survey of 4,000 middle-aged S&T workers, only 12 percent stated they would like to work in nongovernmental enterprises.58 Local officials did little to change that view. One author from the Commercial and Industrial Bureau wrote that nongovernmental technology firms often arbitrarily changed their business addresses, embezzled money, and with- held employee salaries.59 In Shanghai, quite clearly, large state-owned enterprises were preferred at all levels. 574 POLITICS & SOCIETY

BEIJING: LOCAL GOVERNMENT AND NATIONAL CAPITAL

The municipal government in Beijing, in contrast to Shanghai, took a more “hands off” approach to development, preferring to rely on market signals and smaller enterprises. This was in part a result of its role as the national capital— local officials were reluctant to blatantly promote the local interests over the national—but also a result of the structure of the local economy. Firms tended to be smaller and were not as likely to be wholly owned branches of a larger corporate structure. The more independent atmosphere of the municipal economy proved to be quite conducive to high-tech development, and it was not necessary to choose between local and national interests because the two overlapped. Local officials promoted a wide range of ownership structures in the IT sector, directed capital to smaller nongovernmental firms, maintained internal management authority, and helped scientists build a network that facilitated information and personnel flows. Local auto firms, however, suffered. They did not have the capital or the skills necessary to develop basic manufacturing capabilities, and because neither the local government nor the head office of a corporate group was monitoring the development process, firms floundered helplessly when they encountered problems. Local political and economic structures must match the organizational needs of a sector, and as in Shanghai, Beijing could not hope to use the same developmental approach successfully with two sectors that had diamet- rically different needs.

Beijing Economic Structure

At the start of the reform era, the economic structure of Beijing began to evolve in a manner starkly different from Shanghai’s. Prior to 1978, Beijing’s role in the national economy, like other large cities in China, was determined by the central plan. In 1956, the first “Urban Development Plan” intended to emphasize Beijing’s role as the political and cultural center of China, but when the rapid industrialization efforts of the Great Leap Forward began two years later, these plans were revised in favor of turning Beijing into a “comprehensive industrial base.” Investment began to be directed toward heavy industry at the expense of urban services and infrastructure. Between 1957 and 1978 investment in heavy industries increased 474 percent with a particular focus on developing iron and steel, petrochemicals, transportation, and machine-building; the number of SOEs grew from 140 in 1950 to 1,124 in 1983.60 The local auto industry benefited greatly from this emphasis, and until the late 1980s the Beijing automobile sector was considered to be one of the strongest in the country. As late as 1992, the production output value of the local auto industry (8 billion RMB) was the largest of any sector in the city.61 With the start of the reform process, however, central planners began to shift back to the pre–Great Leap development plan for the city. In April of 1980, the SEGAL and THUN 575

State Council pointed out that “development of the Beijing urban and rural economy should be subordinate to and serve the requirements of Beijing as the political and cultural center of the country.”62 Factories in the city center occupied valuable real estate, generated choking pollution, and contributed to increasing congestion. New guidelines began to emphasize the promotion of economic prosperity based on the service sector and new technologies, knowledge-based sectors that capitalized on the city’s strong educational infrastructure.63 Local officials began closing down some of the largest heavy industry enterprises, and between 1980 and 1984 twenty enterprises ceased operation.64 In effect, local planners in Beijing began to see small technology enterprises as a source of great opportunity. While all development plans in Shanghai focused on the large-scale corporate groups, small firms became central to Beijing’s plans. As in Shanghai, nongovernmental firms in Beijing had difficulty securing bank loans. In a 1993 survey of 213 enterprises within the city’s high technology development park, the Beijing Experimental Zone (BEZ), only 17 received funding from a bank.65 But unlike in Shanghai, branches of the municipal government, often in coordination with the enterprise’s supervisory unit, stepped in to arrange funding for small nongovernment technology firms from third sources. In Haidian, the neighborhood where most technology firms are located, science commission officials arranged for communes or SOEs in the district to invest in nongovernmental firms.66 Government officials also acted as matchmakers between entrepreneurs and more informal capital sources. In Beijing’s Chaoyang district, for instance, the head of the science office sought out his own friends to raise a 200,000 RMB investment for a firm unable to raise the capital itself.67 Infrastructure projects and investment in the local science budget were also used to support small, nonstate firms. These investments were not firm specific and were not based on the political or administrative connections: benefits were either public goods or distributed by competitive means. In 1991, for example, the State Science and Technology Commission and the Beijing municipal government began construction of the Shangdi Information Industry Base, China’s first zone dedicated to developing information technologies.68 And unlike in Shang- hai, the majority of research funds in Beijing went to independent scientific organizations, not to labs within SOEs. In 1992 for example, only 14 percent of the city’s S&T appropriations went to medium and large-sized SOEs, while 58 percent went to research institutes.69 Such funding ensured that research funds were not diverted to welfare concerns within the firm, but instead served to promote horizontal links between production and research units. Through property rights reform, local officials gradually focused on defining, refining, and institutionalizing the concept of “nongovernmental” firm. In contrast to Shanghai, they did not use property rights reform solely as a method to regenerate moribund state-owned factories, but instead sought to create a framework that would support the growth of small firms. Unhindered 576 POLITICS & SOCIETY by the need to support large SOEs, the municipal government began praising the nongovernmental enterprises as an important engine of growth as early as 1984.70 In part, officials were reacting to an explosion in the number of small firms in the city—by the end of 1984, there were forty nongovernmental technology firms in Beijing, with an operating volume of 18 million RMB71—but they also worked to encourage the growth of these firms, both by supporting changes in ownership structures and quickly legitimating experiments made by the high-tech entrepreneurs themselves. The Beijing city reform commission approved use of an internal stock system (gufen hua) in 1988 for Stone, a company that produced word-processors and was known as China’s IBM, almost five years before similar systems were promoted at the national level. By 1992, research reports produced by local officials and policy analysts argued that stock systems would clarify ownership structures, modernize management, and increase investment in nongovernmental enterprises. The city’s goal was gradually to convert firms that had originally registered as “collective” to “stock” companies. Eventually “stock companies” created internal incentive systems that tied the most talented scientists to the enterprise and increased employee loyalty and productivity. By emphasizing informal means of supporting small, nongovernmental firms, the Beijing government increasingly de-emphasized the importance of exactly the sort of corporate groups that Shanghai was supporting, and that seemed to be so conducive to auto sector development. Although on the surface the Beijing Auto- mobile Industry Corporation (BAIC) appeared to be very similar to SAIC, it func- tioned in a very different manner. The problem was essentially one of property rights. In a process that began in the mid-1970s and intensified during the reform era, auto sector firms were taken from disparate ministries—primarily the Minis- try of Machinery and Electronics, the Agriculture Ministry, and the Beijing Auto- mobile Company—and placed under the control of the Beijing Automobile Industry Corporation.72 The idea was to create a unified corporate group like Shanghai’s, but because the firms that were transferred to BAIC from other ministries were “managed” (daiguan qiye) by it rather than wholly owned (quanzi qiye), the head office of BAIC did not have the same degree of power as its coun- terpart in Shanghai.73 Revenue from the managed firms was not controlled by the head office of the group, but either by the supply firm itself or the ministry or bureau to which the firm originally belonged. Because complete ownership and control was not transferred, the internal structure of the group—unlike Shang- hai—was quite loose, and the head office did not perceive there to be the same incentive to invest in supplier development; it invested in the development of supply firms that were fully owned by the group because it was from these firms that it collected profits. The manufacturing operations within the group operated as independent firms, receiving little investment from either the head office or the local government.74 SEGAL and THUN 577

The property rights arrangements that prevailed in Beijing neither created the incentives necessary to promote coordinated development within the auto sector nor eased the problem of asset specificity. In Shanghai, in addition to the investment coming from the municipal government’s auto office, the corporate group reinvested its profit in the supply sector. It did this both directly and through the pricing policy of SVW—the prices the assembly plant paid to local supply firms for components were artificially inflated so as to give the supply firms a financial cushion. The auto group in Beijing did just the opposite. Rather than provide supply firms with a financial cushion during the lean years of early development and ease the risk of making large up-fronts investments with only a limited number of customers—a classic problem of asset specificity—the assembly plant actually used multiple suppliers for each component as a means of using competitive pressure to force down the prices. Contracts were short, subject to frequent change, and awarded to the lowest bidder.75 It was the market that governed relations between firms. The plight of one joint venture supply project in Beijing typified the strategy.76 The Beijing Jeep assembly plant awarded the supply firm a one year “open” contract to supply 50 percent of the volume of a particular component; the remaining 50 percent was sourced from a Shanghai supplier. If either supplier increased its prices or began to have quality problems, it received less business. The competing Shanghai supply firm could handle this pressure because it was simultaneously providing 100 percent of SVW’s volumes. It had high volumes and was doing very well. But the Beijing supplier, like most supply firms in Beijing, had a deadly combination of high costs (due to low volumes) and low sales prices. Although this competitive pricing strategy caused problems for the suppliers, from a short-term financial perspective, it made sense for BAIC because it did not want to invest heavily in firms that it did not wholly own, particularly when capable suppliers could be found outside of Beijing—usually Shanghai.

Beijing Bureaucratic Structure

In Shanghai, the bureaucracy of the municipal government worked in a unified manner to direct investment capital into the state-owned sector and nurture the development of these firms—it played the role of a local developmental state. In Beijing, the municipal government remained largely on the sidelines. The Beijing municipal government took a relatively passive role for two reasons. First, because the city was not forced by the central government to remit as large a percentage of local revenue to the center as Shanghai, it did not have to worry quite as much about supporting large SOEs. In Shanghai, given its importance as a revenue source for the nation as a whole, widespread experimentation could not be risked; in Beijing, on the other hand, officials had the leeway to encourage the development of nongovernmental enterprises and experiment with 578 POLITICS & SOCIETY different types of ownership structures. This willingness to experiment, when combined with the general belief high-tech development did not conflict with the city’s role as the cultural and political capital of China, worked greatly to the benefit of technology enterprises: government officials would not stand in their way, and they might even begin to remove barriers to growth. Second, the more interventionist instruments of industrial policy were either unavailable or unattractive to officials in Beijing. Local government actors did not have administrative control over many of the most important players in either technological or automobile development. In the auto sector, as was mentioned in the previous section, the long histories of individual firms meant they were spread across ministerial boundaries, and unlike in Shanghai, the municipal government did not have coordinated control over firm operations. The municipal auto office in Beijing, rather than being granted separate status within the municipal government as in Shanghai, was simply the head office of the Beijing Automobile Indus- try Corporation (BAIC). This office reported to the Beijing Economic Commis- sion, and because it was nested within a ministerial level organization it did not have the ability to coordinate relations between ministries as the Shanghai office did.77 In the high-tech sector, most of the earliest technology enterprises drew their personnel, technology, equipment, and investment capital from the Chinese Academy of Science, a branch of the central government that local officials had little control over. Moreover, these enterprises entered new markets where no government bureaucracies, either central or local, existed. Rather than bemoan their inability to intervene in the economy in the manner of Shanghai, officials in Beijing began to talk of a Silicon Valley model of development. Even though reality fell well short of the model—Beijing was very far from developing the multiple levels of intellectual property rights protection and well-financed capital markets necessary to support long-term sustainable innovation—it was certainly an objective that worked well for technology firms. Local officials in Beijing came closer to understanding the needs of the information technologies than their counterparts in Shanghai. They sought to implement development policies that did not require extensive government supervision and that would turn the strength of nongovernmental firms (and the weakness of large corporate groups) into an advantage. An important example was the effort to link scientific research more closely with industrial production through the creation of nongovernmental enterprises. Even though state-owned enterprises make up almost twenty percent of nongovernmental firms in Beijing, as early as April 1984 municipal officials acknowledged individual scientists and entrepreneurs, not state-owned units, as the driving force behind technology development.78 Later the municipality trumpeted interorganizational cooperation and entrepreneurial activity as the basis of an “innovation network” (chuangxin wangluo) linking firms, research institutes, and universities to each other and to the local govern- SEGAL and THUN 579 ment.79 In Shanghai R&D units were all nestled within the confines of a single corporate group. The Beijing municipality used other types of regulation to reinforce and duplicate nongovernmental forms of organization. The Beijing Experimental Zone for New and High Technologies was “formulated for the purpose of fostering the growth of new technology industries.”80 The management office of the BEZ criti- cized relationships between firms and supervisory agencies that clung to traditional methods of management and thinking; the zone noted that “management and personnel decisions should not be interfered with by the administrative unit.”81 Moreover, the management office insisted that firms registering in the zone use modern accounting techniques, technology management, and enterprise rules. Those who did not meet these requirements could attend seminars given by the management office in all of these skills. The local government also adopted a range of more informal roles to encourage entrepreneurship, facilitate information flows, and to foster economic and research links between firms. Market supporting activities included sponsoring meetings where entrepreneurs could exchange ideas, publicizing new technological or organizational innovations, organizing jobfairs for graduates of businessor computer science programs, and awarding prizes to the most innovative and successful enterprises. In 1987, to give one example, the city government established the Beijing Leading Small Group on Science and Technology (Beijing Shi Keji Lingdao Xiaozhu). This small group, under the direction of Chen Xitong, who would later become Beijing mayor, fostered meetings with over 100 S&T personnel and district officials in an attempt to create a network of firms in the city.82 Not insignificantly—particularly in the Chinese context—these government sponsored activities also served to give nongovernmental firms ideological legitimiza- tion, something they did not receive in Shanghai. In the IT sector, firms benefited from a government that was forced to seek cre- ative policy solutions because it did not have the tools to be direct growth. In the capital-intensive auto sector, however, there was not the same potential for cre- ative solutions: if Beijing firms were to keep up with Shanghai, they needed investment capital and they had to raise production volumes. But auto firms received very little support from the municipal bureaucracy, and a chronic shortage of capital was the result. “We are a big enterprise, yet we get absolutely no help from the Beijing Municipal government—they have invested nothing,” complained the manager of one Beijing supply firm. Simply to keep the enterprise run- ning, he continued, it was necessary to take annual bank loans of 10 million renminbi at interest rates of 12 percent. No preferential policies were given.83 Other auto firms in Beijing uniformly reported that they had to rely on bank loans for capital, although in some instances the government would help them obtain loans at preferential rates. Rather than fight the central government to retain funds 580 POLITICS & SOCIETY for local investment as Shanghai did, Beijing officials believed they had to take the perspective of the national government. “Beijing is China’s capital,” explained Yi Xiqun, an assistant to the Beijing mayor, “it is the political and cultural center for the entire nation, and it is this which is its most important function.”84 The municipal government took a laissez-faire approach to local auto sector development and at times even enacted traffic policies that discriminated against locally manufactured vehicles. The competitive environment within the sector affected the form of relationships that surrounded firms. Rather than having multiple sources to turn to when problems arose, whether they be technical or managerial, Beijing firms were forced to rely on themselves. The assembly plant, on the one hand, had little incentive to assist struggling supply firms: when price was the primary criteria for choosing a supplier, and there were no equity ties, an assembler would switch suppliers rather than work with one that had manufacturing or technical problems. The municipal government and the corporate group, on the other hand, made little effort to nurture firms that they did not own. The manager of one Beijing firm lik- ened the structure of the local auto sector to a family—the municipal government is the grandmother, the head office of the auto group is the mother, the individual firms are children. The parents, he argue, have a responsibility toward the children that goes beyond the financial: “it is important for a child to eat, but a parent must cultivate and teach as well.”85 When asked what sort of assistance would be helpful, he emphasized the need for technical support and coordination. Beijing firms would prefer to cultivate relations with other local firms simply because it is cheaper and more convenient, especially given transportation difficulties within China, but quality was simply inferior. Supply firms did not have the internal resources to overcome problems, and because neither the municipal government nor the head office of the group made an effort to pool knowledge within the sector or facilitate horizontal ties between firms, the suppliers had nowhere to turn.

CONCLUSION

A subnational approach to development is necessary not simply to reveal a level of detail and nuance that would be impossible when the nation-state is the primary unit of analysis, but more importantly because institutions at the local level will often determine the success or failure of development efforts. Beijing and Shanghai, despite being municipalities within the same national framework, had very different political and economic histories, and these histories influenced how the localities organized, promoted, and supported individual firms. In Shang- hai, local leaders had both the desire and ability to act like a developmental state at the local level. During the prereform era, firms in the city had been more closely controlled by municipal officials than in Beijing; these officials were accustomed to being involved in every facet of firm operations and they had the bureaucratic SEGAL and THUN 581 tools to do so. Policy makers preferred to have large state-owned enterprises dominate new markets and actively intervened to coordinate the vertical relations between the highly integrated units. By contrast, local leaders in Beijing usually did not have the bureaucratic authority or the capital necessary to support large conglomerates, and they had to limit themselves to promoting horizontal links between production and research units. Although local governments in China may possess what seem to be similar institutional frameworks, they often operate quite differently. These differences play an important role in determining whether new markets consist of large or small enterprises, vertical or horizontal relations, and interventionist or more flexible government regulations. When faced with new industrial challenges, municipal officials often rely on well worn, widely accepted policy traditions. To create both automotive and IT industries, local officials in Beijing and Shanghai had to address the distinct development needs of firms within two different sectors. In the auto sector, a massive investment of capital was required up-front in both an assembly plant and supply firms, and the structure of the sector had to be conducive to a development process characterized by the learning of well-established manufacturing techniques. Hierarchical corporate structures were useful not only for easing asset specificity and facilitating a coordinated investment approach, but also for promoting the exchange of information between firms. In the IT sector, the creation of innovative firms required a very different development approach. Firms needed technology, skilled personnel, and capital (although not as quickly as in the auto sector and not in a coordinated fashion); they also require a regulatory framework that encouraged innovation and the creation of formal and informal networks between firms. The horizontal ties and the external economies that resulted were a critical element of firm competitiveness. Given the dramatically different needs of an IT sector and an auto sector, it should perhaps not be surprising that neither Beijing nor Shanghai was able to adopt a development strategy conducive to the successful development of both simultaneously. As Peter Evans notes, “sectoral characteristics define what roles are likely to work; the nature of the state determines whether a role can be carried out.”86 Brazil, Korea, and India all embraced a highly interventionist approach toward the steel industry, but only Korea succeeded. The Korean state could intervene effectively because its bureaucratic structures fit the needs of the sector. This paper demonstrates that an argument about roles and sectors holds within individual countries as well as across them; in fact, the lack of institutional uniformity at the regional level in many developing countries makes local level analysis a necessity. Neither Beijing or Shanghai had an institutional endowment for all aspects of economic development, each did certain things well. Beijing had an institutional endowment that lent itself to the development of networks of small, innovative firms; Shanghai had an institutional endowment that lent itself to the 582 POLITICS & SOCIETY development of large, hierarchically organized corporate groups. In a sense, localities as well as firms, have “core competencies” that are important to recognize and exploit. In many respects, this conclusion is quite the opposite of those offered in the literature on late-development. Usually, the emphasis is placed on imitating the most effective model, on choosing an effective “blueprint” for development, whether it is a market-based or state-led approach to growth.87 Applying development strategies from abroad is extremely difficult for local policy makers to do, however. Institutions change slowly—history played a key role in determining the development approach of both Beijing and Shanghai—and there rarely is one model that meets all of a country’s development needs. In some cases, the developmental objectives of a state might require mutually contradictory approaches, highly interventionist in one sector, “hands-off” laissez-faire in another. Each sector provides different opportunities and obstacles to state involvement; there is no one best approach. The difficulty of designing institutions to match the development needs of targeted sectors is reinforced by the changes that sectors undergo over time. The rate of change within a sector, or what Charles Fine calls the “clockspeed” of an industry, varies by sector, but no industry is static. In fact, as Fine argues, the dynamics that lead to organizational changes within a sector are the same across sectors.88 An industry with a horizontal structure will move toward vertical integration when technical advances in one part of a manufacturing chain give market power to a firm making a particular subsystem because market power in the one subsystem encourages bundling with other subsystems and movement toward an integral engineering solution (so as to add value and increase control over the final product). The result is increased vertical integration. In the opposite direction, competition from niche competitors, the challenge of advancing on many technological fronts and markets simultaneously, and the organizational rigidities of large companies lead vertically integrated structures to disintegrate. The process may be more rapid in high-tech sectors than in automotive, but the same dynamics are at work. Although this article examines a relatively brief period of time, it is already possible to see these dynamics at work in China after a decade of development. In fact, the context of late-development might increase the speed with which the needs of a particular sector change in China—because development is more rapid than in earlier cases—and an institutional arrangement that is conducive to the development of a particular sector during one stage of development may actually inhibit growth in the next stage of development. During the first decade of auto sector development in Shanghai, for instance, the tightly controlled hierarchical groups of the city provided a nurturing environment for small SOEs that were trying to upgrade their technical capabilities. Once the supply base had developed a basic manufacturing capability, however, the emphasis shifted to decreasing SEGAL and THUN 583 costs, and the same institutional characteristics that proved so useful in the first stage of development became a severe hindrance. There was no competition within the supply network, inadequate mechanisms to identify inefficient firms (since prices were determined by administrative fiat rather than markets), and little incentive for individual supply firms to lower costs. The implication is that it makes little sense for a local government to make a tremendous effort to reinvent itself for the sake of developing one particular sector. Even if the attempt is successful—and this is by no means guaranteed—the advantage that is gained is likely to be all too transitory. It makes far more sense to understand the inherent advantages of the locality, the needs of different sectors, and to make development choices accordingly. What will begin to emerge is a nation composed of regions with a multiplicity of talents. Different cities attract different firms in different sectors, and, as the development process proceeds and the needs of firms change, a process of migration begins: firms seeking to satisfy changing needs relocate to the regions that best satisfy these needs. Already, high-tech firms that were formed in Beijing have started to open factories and branch offices in Guangzhou and Shenzhen. The local governments in these cities have aggressively courted firms of this sort and have offered more advanced venture capital approaches and the ability to list on technology stock markets as incentives. Finally, it should be noted that this argument, although perhaps most powerful in large states with numerous and diverse localities, is also increasingly important in a world characterized by regional integration. When the nation-state is the basic unit of the world economy, there are incentives for states to duplicate each other’s economic capabilities. Although in theory, free trade should lead states to concen- trate on their comparative advantage, in reality barriers to trade, politics, and the desire to control what are perceived to be highly profitable industries leads to a duplication of capabilities. Free-trade zones, however, transform the playing field from one consisting of numerous small players to one consisting of a few big players. The European Community (and perhaps someday North America), for instance, might be looked at as a unit very similar to China: internally it is a free trade zone with a single currency and a central policy-making organization, but it is composed of localities that have different customs, languages, institutional strengths, and interests. It makes just as much sense to speak of Baden Wurttemberg or Milan in such an environment, if that is the internal unit with the most pronounced institutional variation, as it does to speak of Germany or Italy. Paradoxically, as the units of the world economy increase in size, the importance of localities may also increase.

NOTES

1. For work on advanced industrialized countries in this tradition, see Richard M. Locke, Remaking the Italian Economy (Ithaca, NY: Cornell University Press, 1995), 22; 584 POLITICS & SOCIETY

Gary Herrigel, Industrial Constructions: The Source of German Industrial Power (Cam- bridge: Cambridge University Press, 1996). For recent work on China, see Jean C. Oi, Rural China Takes Off: Institutional Foundations of Economic Reform (Berkeley: Univer- sity of California Press, 1999); Andrew Walder, “Local Governments as Industrial Firms: An Organizational Analysis of China’s Transitional Economy,” American Journal of Soci- ology, vol. 101 (September 1995): 263-301; and Susan Whiting, Power and Wealthin Rural China: The Political Economy of Institutional Change (Cambridge: Cambridge Uni- versity Press, 2001). 2. Adam Segal, Digital Dragon: High Technology Enterprises in China (Ithaca, NY: Cornell University Press, forthcoming) and Eric Thun, Changing Lanes in China: Indus- trial Development in a Transitional Economy (Ph.D. Dissertation, Harvard University 1999). 3. Andrew Shonfield, Modern Capitalism: The Changing Balance of Public and Pri- vate Power (New York: Oxford University Press, 1965). 4. See Peter J. Katzenstein, ed., Between Power and Plenty: Foreign Economic Pol- icies of Advanced Industrial States (Madison: University of Wisconsin Press, 1978); John Zysman, Governments, Markets and Growth: Financial Systems and the Politics of Indus- trial Change (Ithaca, NY: Cornell University Press, 1984). 5. See Philippe Schmitter and Gerhard Lehbruch, eds., Trends Towards Corporatist Intermediation (Beverly Hills, CA: Sage, 1979); Suzanne Berger, ed., Organizing Interests in Western Europe (New York: Cambridge University Press, 1982); Peter J. Katzenstein, Small States in World Markets (Ithaca, NY: Cornell University Press, 1985). 6. See Suzanne Berger and Ronald Dore, eds., National Diversity and Global Capital- ism (Ithaca, NY: Cornell University Press, 1996); Peter A. Hall and David Soskice, eds., Varieties of Capitalism, forthcoming. 7. Alice Amsden, Asia’s Next Giant: SouthKorea and Late Industrialization (New York: Oxford University Press, 1989), 14. 8. Robert Wade, Governing the Market: Economic Theory and the Role of Govern- ment in East Asian Industrialization (Princeton: Princeton University Press, 1990), 27. 9. World Bank, The East Asian Miracle: Economic Growth and Public Policy (New York: Oxford University Press, 1993). 10. Richard M. Locke, Remaking the Italian Economy (Ithaca, NY: Cornell University Press, 1995), 22. 11. William Skinner made this point with respect to China in the 1960s. See G. W. Skin- ner, “Marketing and Social Structure in China,” Journal of Asian Studies, November 1964, February 1965, May 1965. 12. David Soskice, “Divergent Production Regimes: Coordinated and Uncoordinated Market Economies in the 1980s and 1990s,” in Herbert Kitschelt et al., eds., Continuity and Change in Contemporary Capitalism (Cambridge: Cambridge University Press. 1999). 13. Gary Herrigel, Industrial Constructions: The Source of German Industrial Power (Cambridge: Cambridge University Press, 1996), 26. 14. Gary Herrigel, “Large Firms, Small Firms, and the Governance of Flexible Special- ization: The Case of Baden Wurttemberg and Socialized Risk,” in Bruce Kogut, ed. Coun- try Competitiveness (New York: Oxford University Press, 1993), 17. 15. Annalee Saxenian, Regional Advantage Culture and Competition in Silicon Valley and Route 128 (Cambridge: Harvard University Press, 1994), 2-3. 16. Aseema Sinha, “From State to Market via the State Governments: The Political- Federal Logic of Liberalization in India.” Paper presented at the 1999 Annual Meeting of the Association of Asian Studies. SEGAL and THUN 585

17. Khalid Nadvi and Hubert Schmitz, Industrial Clusters in Less Developed Coun- tries: Review of Experience and ResearchAgenda , Institute of Development Studies, Dis- cussion Paper 339 (January 1994), 36. 18. Susan L. Shirk, The Political Logic of Economic Reform in China (Berkeley: Uni- versity of California Press, 1993). 19. Jean Oi, “Fiscal Reforms and the Economic Foundations of Local State Corporatism in China,” World Politics 45 (October 1992): 99-126. 20. Victor Nee, “Organizational Dynamics of Market Transitions: Hybrid Forms, Prop- erty Rights, and Mixed Economy in China,” Administrative Science Quarterly 37 (1992): 1-27. 21. Andrew G. Walder, “Local Governments as Industrial Firms: An Organizational Analysis of China’s Transitional Economy,” American Journal of Sociology, vol. 101, no. 2 (September 1995): 263-301. In particular, Walder makes the persuasive argument that the degree of internal diversification within a government jurisdiction affects its ability to monitor firms’ performance and enforce financial discipline. 22. Susan Whiting, Power and Wealthin Rural China: The Political Economy of Institu- tional Change (Cambridge: Cambridge University Press, 2001). 23. Herbert Kitschelt, “Industrial Governance Structures, Innovation Strategies, and the Case of Japan: Sectoral or Cross-National Comparative Analysis?” International Orga- nization 44, no. 4 (Autumn 1991); and J. Rogers Hollingsworth, Philippe C. Schmitter, and Wolfgang Streeck, Governing Capitalist Economies: Performance and Control of Eco- nomic Sectors (New York: Oxford University Press, 1994). The definition of a sector is from Hollingsworth, Schmitter, and Streeck, 8. 24. See Oliver E. Williamson, “The Modern Corporation,” Journal of Economic Litera- ture 19 (December 1981). 25. OECD Proceedings, Boosting Innovation: The Cluster Approach (Paris, France: OECD, 1998), 11. 26. These percentages are from interviews with the supply division of each JV.A locally sourced part is defined as being manufactured within the municipality in which the JV assembly project is located. It is possible that Shanghai was simply more protectionist than Beijing—and this is partially true—but the fact that auto projects throughout China began to source from Shanghai (rather than their home locality) lends further credence to the assertion that Shanghai had the strongest auto supply sector. Eric Thun, Changing Lanes in China: Industrial Development in a Transitional Economy (Ph.D. Dissertation, Harvard University 1999). 27. 1997 Niandu Beijing Keji Qiye Gongzuo Yaoloan [1997 Overview of Beijing Tech- nology Enterprises] (Beijing: Zhongguo Jingji Chubanshe, 1998), 70. 28. Zhu Jianjiang, “Shanghai Minban Keji Qiye de YunxingJizhi ji Guanli” [The Oper- ation and Management of Shanghai Nongovernmental Science Enterprises] Zhonggu Keji Luntan [China S&T Forum] 5(1993): 42. 29. “Beijing Minying Keji de Chenggong Moshi,” [Patterns of Beijing nongovernmental firm’s Success] Zhongguo Keji Chanye Yuekan [Chinese Technology Industry Monthly], 9 (1995): 27-28. 30. Detlef Rehn, “Organizational Reforms and Technology Change in Electronics Industry: The Case of Shanghai,” in Denis Fred Simon and Merle Goldstein, eds., Science and Technology in Post–Mao China (Cambridge: Harvard University Press, 1989), 144. 31. Yang Dongping, Chengshi Jifeng: Beijing he Shanghai de Wenhua Jingshen [City Monsoon: The Cultural Spirit of Beijing and Shanghai] (Beijing: Dongfang Chubanshe, 1994). 586 POLITICS & SOCIETY

32. Roads Murphey, Shanghai: Key to Modern China (Cambridge: Harvard University Press, 1953), 166. 33. Lynn T. White, Shanghai Shanghaied? Uneven Taxes in Reform China (Hong Kong: University of Hong Kong, 1989), 24 and 64-66; Lynn T. White, Unstately Power Volume I: Local Causes of China’s Economic Reforms (Armonk: M. E. Sharpe, 1998), 170-174. 34. Zhimin Lin, “Shanghai’s Big Turnaround since 1985,” in Peter T. Y.Cheung, Jae Ho Chung, and Zhimin Lin, eds., Provincial Strategies of Economic Reform in Post–Mao China: Leadership, Politics, and Implementation (Armonk, NY:M. E. Sharpe, 1998), 51. 35. White, Shanghai Shanghaied? 29. 36. The assembly plant could not use imported components to raise production volumes because the government restricted the foreign exchange needed to purchase these components. 37. Shanghai qiche gongye bianweihui, Shanghai Qiche Gongye Shi [History of the Shanghai Automotive Industry] (Shanghai: Shanghai Renmin Chubanshe, 1992), 28. 38. Interview no. S12, 26 May 1997. The tax decreased to 2,300 RMB in 1990 and was abolished completely in 1994 when the VAT tax was created. 39. Huang Ju, “Jiasu Keji Jinbu, Zujin Jingji Fazhan he Shehui Quanmian Jinbu,” [Speed S&T Progress, Improve Economic Development and Social Progress] in Shangahi Kexue Jishu Gongzuo Nianbao: 1995 [Yearbook of Shanghai Science and Technology Work: 1995] (np, Shanghai: 1996), 4. 40. “Guanyu Tuijin Dianzi Gongye Jingji Guanli Tizhi Gaige de Baogao,” [Report on the Implementation of the Reforms of the Management System of the Electronics Industry] Zhongguo Dianzi Bao [Electronics Daily] (21 March 1986). 41. Rehn, “Organizational Reforms and Technology Change,” 141. 42. Interview no. S29, 28 October 1996. 43. Lin Huangu, “Shanghai Gaojiao Keji Chanye Xiankuang he Fazhan Duice de Yanjiu,” [Shanghai High Technology Industry Development Policy] Zhongguo Minying Keji Yu Jingji [China Nongovernmental S&T and Economy] no. 2 (1995): 15. 44. “Beijing Shiyan Qu Yao Jiduan Pandeng Xin Gao Shan,” [Beijing Experimental Zone Must Continue to Scale New Heights] Keji Luntuan [S&T Forum] no 3. (1994): 43-47. 45. Shanghi Kexue Jishu Gongzuo Nianbao: 1995, 39-41. 46. Yuan Jianmin, “Shanghai Chuangban Gao Xin Jishu Fengxian Touzi Gongsi de Yanjiu yu Fenxi,” [Analysis of Shanghai Establishing a High Technology Venture Capital Company] Shanghai Tongji [Shanghai Statistics], 3(1993): 10-13. 47. Although many companies in China refer to themselves as corporate groups (jituan gongsi, literally “group companies”), in many of these, individual firms control their own profits. By most common definitions, a multidivisional enterprise is an umbrella company with more than one factory under the control of a head office that controls profits and internal pricing. On this point, see Doug Guthrie, Dragon in a Three-Piece Suit: The Emergence of Capitalism in China (Princeton: Princeton University Press, 1999), 34. 48. The SVW assembly plant also played a key role in this process. See Wang Rongjun, “Chenggong zhi lu” [The Road to Success], in Nian Yi Sun, Zhongguo Qiche Jixing [Notes on the Chinese Auto Industry] (Beijing: Zhongguo Tongji Chubanshe, 1995), 145. 49. The Localization Office consisted of approximately 10 officials, each responsible for 10-20 factories in a particular subsector of the auto industry, and in most cases the official had extensive background in his or her particular jurisdiction (light industry, textiles, machine tools, electronics, etc.). There was also a program to bring retired engineers from SEGAL and THUN 587 the Senior Expert Service in Germany to serve as short-term consultants to firms that were in the process of upgrading their technical capabilities. 50. See YePing, Zai Santana gongtongti di jiu ci chuanti huiyi shang de jianghua [Min- utes of the Ninth Meeting of the Santana Localization Community], 6 November 1996. 51. Interview no. S35, 17 June 1998. 52. Wang Pusheng, “Minban Keji Fazhan Zhong de Wenti Yu Jianyi,” [Some Problems in Developing Nongovernmental S&T] Minban Keji [Nongovernmental S&T] 1991 (6): 8. 53. “Xia yi bu Wang Nar Mai? Guanyu Shanghaishi Shenhua Keji Tizhi Gaige Jige Wenti de Sikao,” [Where to Next? Some thoughts on deepening the reforms of the Shang- hai S&T system] Keji Ribao [Science and Technology Daily] (31 December 1988). 54. “Shanghai Shi Minban Keji Jigou Guanli Banfa,” [Managing Shanghai’s Nongovernmental S&T Structure] in Zhongguo Minban Keji Shiyejia Xiehui Huixun [Pro- ceedings of the Conference of Chinese Nongovernmental High Technology Entrepre- neurs], 3 June 1988. 55. “Shanghai Sets Up S&T Consultative Body,” Wen Hui Bao, 14 July 1993 in Foreign Broadcast Information Service China 145-16, 1993, and “Shanghai Minying Keji Jinru Kuai Che Dao,” Minban Keji, 8(1995): 33. 56. Interview no. S18, 22 June 1996. 57. Interview no. S11, 16 April 1996. 58. “Shanghai Minban Keji Jigou Shige Ri Meng Zeng Bai Yu Jia,” [Shanghai Nongovernmental Institutions Rapidly Increase to over a Hundred] Wen Hui Bao (10 Octo- ber 1988). 59. Wang Enshou, “Minban Keji zai Shanghai,” [Nongovernmental S&T in Shanghai] Minban Keji 6(1992): 33. 60. Han Guang, “Beijing Municipal Government’s Management of State-Owned Enterprises,” in George Totten and Zhao Shulian, eds., China’s Economic Reform: Admin- istering the Introduction of Market Mechanisms (Boulder, CO: Westview, 1992), 77; Brian Hook, Beijing and Tianjin: Towards a Millennial Megalopolis (Hong Kong: Oxford Uni- versity Press, 1998), 137. 61. Lao Chang, “Life in the Fast Lane,” China Daily (17 January 1993). 62. Han, “Beijing Municipal Government’s Management of State-Owned Enterprises,” 78. 63. Zixuan Zhu and Reginald Kwok, “Beijing: The Expression of National Political Ideology,” in Won Bae Kim, ed., Culture and the City in East Asia (New York:Oxford Uni- versity Press, 1997): 136. 64. Han, “Beijing Municipal Government’s Management of State-Owned Enter- prises,” 80. 65. “Jinrong Tizhi Gaige Dui Gaoxinjishu Qiye de Fazhan Yingxiang ji Duice,” [Sug- gestions on Reforming the High-Technology Finance System] Beijingshi Xinjishu Chanye Kaifa Shiyan Qu Yanjiu Baogao: 1995 [Research Report on the Beijing Experimental Zone:1995] (Beijing: n.p., 1996), 16. 66. Interview no. B2, 3 May 1996. 67. Zhang Yi, “Minying qiye de ‘hao popo,’” [Nongovernmental Firm’s “Good Mother-in-Law”] Zhongguo Minying Kexue yu Jishu, no. 6 (1997): 39. 68. Office of the Beijing Experimental Zone for the Development of New Technologies, “Beijing City Experimental Zone for the Development of New Technologies: Investment Guide,” no date. 69. Beijing Keji Tongji Nianjian: 1992 [Beijing S&T Yearbook: 1992], 107. 70. See the “Inquiry into reforming the city’s S&T system,” Beijing Minban Keji Shiye Dashiji: 1980-1990 [A Chronicle of Beijing’s NonGovernmental Technology Industry: 1980-1990] (Beijing: Zhonghua Gongshang Lianhe Chubanshe, 1994). 588 POLITICS & SOCIETY

71. 1997 Niandu Beijing Keji Qiye Gongzuo Yaolan [1997 Overview of Beijing Tech- nology Enterprises] (Beijing, Keji Chubanshe, 1997), p. 67. 72. By the early 1990s, BAIC consisted of 31 enterprises and research institutes. 73. Interview, no. B37, 14 August 1998. The primary fully owned BAIC enterprise was BAW, the Chinese partner in Beijing Jeep, and even BAW was required during contract negotiations to consult with over 10 separate ministries on a regular basis. Jim Mann, Beijing Jeep: How Western Business Stalled in China (New York: Simon and Schuster, 1989), p. 68. 74. Interview, no. B9, 11 August 1997. 75. Interview, no. B40, 17 August 1998. 76. This account is based on Interviews B25, 22 April 1997, and B34, 6 August 1998. 77. In Shanghai, by contrast, the Auto Industry Leading Small Group was above ministerial status and hence had the ability to coordinate relations between ministries. 78. “Kaichuang Zhongguoshi de Guigu de Tansuo,” [Initiating a Chinese-style Silicon Valley] Beijing Ribao [Beijing Daily] (11 September 1984). 79. “Peiyu Zhongguancun Quyu Chuangxin Wangluo, Zengqiang Chuangxin Nengli, Cujin Gaoxinjishu Chanye Fazhan,” [Foster Zhongguancun’s Innovation Network, Strengthen Innovative Ability, Promote High-tech Industry Development] Beijingshi Xinjishu Chanye Kaifa Shiyan Qu Xiezuo Xiaozu [Beijing Experimental Zone Small Work Group] (n.p., Beijing, 1997); 7-25. 80. Beijing Municipal People’s Government, “Interim Regulations on the Beijing Municipality’s Experimental Zone for the Development of New Technology Industries,” 20 May 1988. 81. “Shoubi Xinjishu Qiye Zhengshi Dansheng” [First Batch of High-Tech Firms Emerge] Beijing Ribao, 9 September 1988, p. 1. 82. “Dongyuan Wange Keji Renyuan Zhiyuan Jingjiao Jianyi” [Mobilizing the Sugges- tions of S&T Personnel for Beijing’s suburbs] Keji Ribao 20 May 1987. 83. Interview, no. B2, 22 April 1997. 84. Yi Xiqun, “Beijing fazhan kaifangxing jingji de zhanlue sikao,.” [Strategic Thinking on Developing Beijing’s Open Economy] Zhanlue yu Guanli [Strategy and Man- agement] 3 (1996): 117. 85. Interview, no. B27, 5 May 1997. 86. Peter Evans, Embedded Autonomy: States and Industrial Transformation (Prince- ton: Princeton University Press, 1995), 94. 87. In Asia’s Next Giant, Amsden points to Korea as an example from which others can learn. 88. This paragraph draws from Charles H. Fine, Clockspeed: Winning Industry Control in the Age of Temporary Advantage (Reading, MA: Perseus Books, 1998), 48-50. POLITICSGEORGE GONOS & SOCIETY

Fee-Splitting Revisited: Concealing Surplus Value in the Temporary Employment Relationship

GEORGE GONOS

Based on a historical study of fee regulation in the employment agency business, the claim that temporary help agencies charge “no fees” to workers is challenged. It is found that the claim rests on technical changes in the statutory definitions of fee and employment agency won through industry lobbying efforts, changes that simply mask traditional fee-charging methods. The article traces the evolution of fee-charging practices in the post–World War II period and points to a new version of “fee splitting” as a prevalent wage setting practice in the staffing industry. The implications of these developments for wage depression and income redistribution are explored.

You always have to expect emotional reactions on the part of trade unions against temporary work, unless you can provide us with security that profits obtained at the expense of two employers are not greater than the profits made by a single employer. —German trade unionist (1979)1

Despite the view of many analysts that the “temporary help” arrangement is an exploitative one2 and increasing statistical evidence of the substantial losses of earnings and benefits incurred by temps,3 the subject of agency fees has scarcely been mentioned in the growing literature on temporary work and remains a much overlooked aspect of the “flexible” labor market in the United States. One reason for this is the general acceptance of the staffing industry’s position that “tempo-

Earlier versions of this article were presented at the Annual Meetings of the Society for the Study of Social Problems, 1996, and the Industrial Relations Research Association, 1997. POLITICS & SOCIETY, Vol. 29 No. 4, December 2001 589-611 © 2001 Sage Publications 589 590 POLITICS & SOCIETY rary employees are never charged a fee of any kind,”4 that fees are borne entirely by employers, the business clients of staffing firms. This article excavates the political and historical foundations of this virtually taken-for-granted claim and suggests that with the rapid growth in recent years of the temporary help and staffing industry (hereafter, “the staffing industry”), fees for jobs, shouldered largely by workers, have surged. It explores current fee-charging practices and their implications for income redistribution and wage depression. Through the early 1990s, a jobseeker going through the classified ads placed by temp and staffing firms would have been assuaged by the following kinds of inducements: “No fees,” “Not an agency—Never a fee,” “Never a fee to the job seeker,” “ALL FEES COMPANY PAID,” “A fee-paid service.” From its birth in the late 1940s, the staffing industry used these claims to build what became by the 1970s the fastest growing industry in the country. Nowadays, industry representatives say, the idea that staffing agencies charge no fees to workers has become so widely accepted and understood that these printed claims have become unnecessary. Still, the “no fees” claim is continually reinforced in TV and radio ads and reiterated in industry literature and in briefings with job applicants in agency offices. Moreover, state laws across the country now uphold the industry’s position. If the existence of fees is acknowledged at all, they are said to be “client-paid,” that is, covered in their entirety by the industry’s business clients, the users of temporary help. Furthermore, changes in state laws brought about by industry lobbying have allowed staffing agencies to conceal their “mark-ups,” thereby hiding the amount they garner from a worker’s labor. The staffing firm’s mark-up is the biggest secret in the industry, “the great untouchable topic.”5 In an openly self-serving and transparent practice of dubious legality, agencies forbid temps from ever inquiring about the wages of other workers or about the firm’s billing rate. The agreement that temps sign typically contains a provision such as the following: “I understand that all matters relating to wages and rates are necessarily confidential and will never discuss same with clients or others.”6 Such warnings are to a surprising extent effective in keeping temps (and organizers) in the dark about what regular employees or other temps make on the same job and about what their agency is receiving for their work, and hence inhibit bargaining and the operation of a free labor market. Due to effective concealment, it is safe to say that workers paying fees out of their own pockets early in the 20th century knew more about the actual cost to them of working through an agency than their counterparts do today. Recently, however, anger over fees among temps and contract workers has grown, and workers’groups in some states have backed “right-to-know” legislation that would require labor market intermediaries to reveal to workers the hidden fees in their employment contracts. A movement for “full disclosure” agencies has been spawned.7 GEORGE GONOS 591

Based on a sociolegal study of fee regulation over the past century, this article shows that the claim currently made by the staffing industry that no fees are charged to workers rests solely on technical changes in the statutory definitions of fee and employment agency quietly won by the industry through its aggressive lobbying efforts over the past four decades, changes that simply mask traditional fee-charging methods that have been prevalent since the emergence of private employment agencies in the 1890s. After providing some historical background, the article traces the evolution of fee-charging practices in the post–World War II Untied States through three important mutations: (1) the elimination of early definitions of fee that included an agent’s “mark-up”; (2) the rapid spread and acceptance of the doctrine of “client paid” fees; and (3) the use of this doctrine to legal- ize referrals from public employment offices to commercial agencies and to promote staffing agencies as legitimate providers of employment for former welfare recipients. At the center of these developments was the staffing industry’s effort to reshape the legal environment to legitimate its social “takings.”8 As will be shown, these alterations of law meant a change in wage setting practices for the growing “temporary” workforce—in effect, a revival of the formerly illegal practice of “fee splitting”—helping to effect the redistribution of bargaining power and income between workers and employers in recent decades.

STATE REGULATION OF EMPLOYMENT AGENCY FEES

Early in the twentieth century, concern over excessive employment agency fees was intense. The problem commonly appeared on the lists of industry abuses compiled by state labor bureaus and federal commissions, and a broad movement grew up in opposition to “the vampire system.”9 In the words of Justice Louis Brandeis in 1917:

There gradually developed a conviction that the evils of private agencies were inherent and ineradicable, so long as they were permitted to charge fees to the workers seeking employment. And many believed that such charges were the root of the evil.10

As early as 1890, states began to pass regulatory legislation addressing all aspects of employment agency operations, including their fee-charging practices.11 Prior to the passage of national labor legislation in the 1930s, which sanc- tioned collective bargaining as the primary means of protecting workers’ rights, this state level regulation of labor market intermediaries comprised a significant part of labor law and provided the only protection for the majority of workers engaged in “individual bargaining.”12 State legislative efforts to guard workers from excessive fees were further intensified after the U.S. Supreme Court declared in 1917 that a state’s attempt to abolish commercial fee-charging agencies was unconstitutional.13 By 1928, 29 states had placed statutory ceilings on 592 POLITICS & SOCIETY employment agency fees. When, in that year, the Supreme Court outlawed fee ceilings as an illegal form of price fixing,14 many states forced disclosure by requiring agencies to post their fee schedules in a “conspicuous place” in their offices and to file them with the state. The free public employment office, or labor exchange, many of which had been established by states and municipalities in the pre-World War I period, was seen as a natural alternative to the commercial fee-charging firm. In 1933, the charter for the United States Employment Service forbade the referral of workers to agencies that charged either workers or employers a fee. But the growing number of private labor scouts kept the issue on the national agenda into the New Deal period, when Congressional hearings in 1940 focused attention again on the “excessive fees, paid by workers for employment, out of all proportion to the wages earned.”15 And in 1941 the Supreme Court reversed its earlier ruling, once again allowing states to set ceilings on employment agency fees, a practice that continued into the post–World War II period.16 Through the mid-1960s, state departments of labor strongly pursued the regulation of private employment agencies, and fee regulation in specific, over vehement industry opposition.17 “One of the most important reasons for regulating private employment agencies is to protect applicants against excessive fees,” the U.S. Department of Labor stated in 1960,18 and recommended that states set maximum fees “for both temporary and permanent jobs.”19 Along with the strengthening of the public employment service during and after World War II, and the broad support for continued government manpower policies, these renewed efforts at fee regulation were compelling reasons why the future of the employment agency business looked problematic to its practitioners in the immediate postwar period. But now, collective bargaining had become the main agenda for labor, as opposed to the protection of individual rights vis-a-vis private labor agents. State employment agency laws were relegated to minor importance, receiving little attention from the either the public, the media, or organized labor, thus making it possible for a quiet industry-run campaign to reverse the trend toward strict state regulation. Beginning in the 1960s, when the temporary help industry turned to aggressive lobbying in the state legislatures,20 the regulation of employment agencies “deteriorated.”21 The trend now turned to the oversight of employment agencies by generally friendly industry boards and, in many states, toward the deregulation of placement fees.22

FEE SPLITTING AND JOBS OF SHORTENED DURATION

To appreciate the temp industry’s solution to the post-WWII predicament of the employment agency business, we must examine another aspect of that business early in the 20th century, namely, the practice of fee-splitting, widely reviled by workers and reformers, and cited prominently on all the lists of employment GEORGE GONOS 593 agency abuses generated by government investigations. Fee-splitting referred to a form of collusion in which an employer (or foreman) agreed to hire workers from a certain agent in return for a share of the fees collected from these workers. Such collusive arrangements resulted in increased fee rates, since employment agents were caused to charge enough to cover both their own costs and the gratuity paid the employer.23 In addition to inflated charges, fee splitting almost inevitably resulted in another problem for workers, that of accelerated turnover (or jobs of shortened duration), as workers were discharged solely to make way for others, thereby multiplying the fees to be collected and divided.24 Artificially shortening work assignments, that is, making jobs more temporary, increased the immediate profits of both employers and employment agents.25 Thus the practice of fee-splitting was widely seen as exacerbating the irregularity of employment. Workers often testified before government commissions that they had been sent to positions which employment agents had represented as permanent or long-term but which turned out to be of short duration.26 As early as 1913, legislation in eight states and the District of Columbia prohibited fee-splitting. By 1940 the number had grown to twenty-one states.27 The arrangements between labor agents and employers engaged in fee splitting could take several forms.28 Most simply, discharged workers could be sent back to the employment agent for reassignment and, after paying another fee, were frequently sent back to the same work site.29 In a somewhat more sophisticated version of the practice, workers would be forced to pay the agent a certain percent of their wages each week or month, that is, a regular fee, for the promise of holding their jobs against other workers.30 Taking it one step further, some agents took on the role of paymaster, collecting payments from the employer and deducting their fees before compensating the workers (as the staffing industry does today), often making additional charges for transportation, supervision of workers, housing, commissary, or other “services.” With these improvisations, many employment agents had devised means of collecting a continuous stream of fees— whether workers stayed on a single jobor worked at a series of different work sites, and whether for the same or different employers. In effect, a segment of the employment agency business had moved beyond the idea that agency fees were one-time payments made for job placement, that is, matching a worker with an employer, the traditional role of what became known as the “permanent” employment agency. This segment of the business was based on forming an ongoing triangular relationship with the worker on one side and the employer on the other. This triangular arrangement not only served to increase the financial gain for the agent, but was also beneficial to the user of labor in that it served to weaken his ties or obligations to these “agency workers” relative to what would be expected toward his direct employees, and therefore also decreased long-term pressures toward higher wages (a point which became even more salient after New Deal legislation endorsed collectivization 594 POLITICS & SOCIETY and increased the obligations of employers). This possibility of distancing or shielding the actual users of labor from obligation to workers was another reason (along with screening for union preferences) that the “services” of private employment agents appealed to employers, that is, why they might even pay premium hourly rates for this arrangement, and conversely why free public agencies, which could not serve the same functions, were not as useful.31 The triangular arrangement was optimal, then, for both employers seeking labor “flexibility” and employment agents seeking steady income. But, as the post–World War II period began, this system, along with the employment agency business as a whole, had for half a century faced broad public enmity and been vexed by the extensive efforts of state regulators and reformers. The amount of fees that could be charged workers was limited by ceilings set in most states, and fee-splitting, as a mechanism for multiplying fees, was outlawed in 32 states.32 At the same time, the strengthening of the public employment service posed the question of whether that competitor would become the primary way that the nation marketed its labor. These conditions constituted a problem to be solved not just for the employment agency business itself, but for many corporate employers concerned with the rigidity of labor markets in a period of apparent union strength. This challenge was taken up by the emerging temporary help industry, which followed the past practice of that segment of the employment agency business that used an ongoing triangular relationship, collecting payment from employers and serving as paymaster to workers. To avoid state regulation, however, the industry put itself forward as a “new type of service.”33 not connected with the old employment agency business, basing this claim on the premise that temp firms were themselves employers, not employment agencies.34 In taking this position, the “new” business faced obstacles in the prevailing legal definitions of employment agency and fee. The industry thus turned its attention to reshaping this legal environment.

THE EVOLUTION OF FEE-CHARGING PRACTICES

Eliminating the Definition of Fee as Mark-Up

For the type of triangular practice represented by the temporary help industry, the fee was commonly defined in state law as the difference between what an employment agency charged its business client for supplying workers and what it paid those workers. For instance, New Jersey’s 1918 employment agency law specified that the term fee, in addition to direct payment for employment, “shall also mean and include the difference between the amount of money received by any person who furnishes employees . . . and the amount paid by him to said employees.”35 This definition of fee (what I call the “fee as mark-up”) would remain part of employment agency law for much of the nation into the post–World GEORGE GONOS 595

War II period. In the early years of temp industry growth this prevailing definition supported the intuitive understanding of state regulators that temp firms were in fact fee-charging employment agencies. Thus, during the 1960s temp firms found they were “viewed as simply another version of employment agencies, that is, it is contended that the spread between the wage rate and the billing rate is a ‘fee.’”36 And, like the early law that defined fee as mark-up, temp workers were from the beginning prone to see the billing rate as their “gross wage” from which the agency took a “cut,”37 a view that found support in some state courts.38 The young “temporary help industry” fought hard against this understanding of fee.39 Defining the mark-up as a fee, it was said, “could mean the demise” of the industry,40 since it would force temp agencies to reveal their most important secret, the spread between billing rates and wages. Avoiding this, the industry organized nationally for a fight to exempt temp firms from state regulation. In a state-by-state blitz between 1963-1971, the industry pushed through quick, often crude, amendments to state employment agency laws across the country, amendments that defined temp firms as legal employers, categorically exempted them from state regulation, and specifically eliminated the definition of fee as mark-up.41 This could be accomplished in many states with the addition or sub- traction of just a few lines in the state statute, and without any public debate. Insuring its ability to hide the mark-up provided the industry with considerable relief. In 1960, prior to deregulation, no less than twenty states required that maximum “fee schedules” be posted in agency offices for workers to see.42 For temp agencies the required fee schedule would display the hourly amount to be billed the business client (the billing rate) on one side and the hourly wage paid the worker on the other, thus clearly revealing the mark-up. In 1957, the N.J. Superior Court found the statutory requirement to post fee schedules entirely appropriate. “It is important,” the Court stated, for the worker to know what Manpower [the agency] is charging the customer for his work, or, to put it another way, it is important for the worker to know how the money paid by the customer will be divided as between [the agency] and himself.43

The industry despised the requirement to post fee schedules, perhaps more than any other provision of law. Community organizers found they simply flaunted the law, refusing to post them. Winning exemption from this requirement, along with fee ceilings, produced the situation faced by temporary workers today, in which there is no regulation of the amount of the mark-up, and no requirement for agencies to make it known. Thus, in hearings on a bill that would reregulate agencies under the U.S. Department of Labor (DOL), Senator Walter Mondale spoke in 1971 of the “unconscionable fees” charged to temp workers, and bemoaned the fact that “there are no controls or limits on what private temporary help supply firms may...charge for what are in substance placement fees.”44 (In negotiations with staffing firms over an industry “code of conduct” in recent 596 POLITICS & SOCIETY years, grass-roots groups have found that disclosure of mark-ups is practically the only provision to which agencies adamantly refuse to agree. In contrast, labor law prohibits unions from charging excessive dues or initiation fees.45) In practice, the passage of legislation defining temp agencies as employers rendered the notion of fee-as-markup meaningless. If temp workers are their direct employees, it follows that the paychecks issued by temp firms — at an hourly rate apparently agreed to by workers — represent their pay in full (less only the standard deductions required by law). Hence the notion that temp agencies charge no fees to workers was neatly inscribed in law and became part of the accepted logic of employment relations in the United States. In effect, “no fees” came to mean that there is no need for workers to take money out of their pockets to make a payment to the agency, as they normally did in the earlier history of the industry. Soon, this logic would become institutionalized for other types of staffing and “payrolling” firms as well, as the industry moved to ratify the notion of “client paid” fees for a wide range of personnel firms.

The Spreading Acceptance of “Client-Paid” Fees

The next stage in the evolution of fee charging practices, occurring from the mid-1970s through the early 1990s, consisted of the rapidly spreading acceptance of the veracity of “client-paid” fees as a basis for exemption from state regulation. This included other “new” types of personnel placement firms, for example, staffing, executive search, personnel consulting, and payrolling firms, and PEOs (“professional employer organizations”), even many engaged in traditional permanent placement activities. Like “temporary help” firms, many of these firms were now beginning to collect fees from business clients rather than workers, whether these workers were treated as W-2 employees of the personnel firm or not. Their argument (to state legislators and regulatory agencies) was that since they charged no fees directly to workers, then the original reason for their regulation, that is, the protection of workers from abuses such as exorbitant fees, had been eliminated. To put the emergence of “client-paid” fees in bold relief, note that as recently as 1962 the U.S. DOL had reported that fees were paid by employers only in “rare circumstances.”46 But thereafter the legal environment changed rapidly. In the mid-1970s, the Book of States noted the legislative trend toward “shifting fee charges from jobapplicants to employers for all or certain types of placements.” 47 Throughout the later 1970s and 1980s legislative measures in many states, presented by their backers as pro-worker initiatives, excluded from regulation as employment agencies firms whose fees were “employer paid.”48 Based on the assumption that the amount of any “client-paid” fees did not matter to, or affect, workers, the fee as mark-up as well as the traditional one-time fee for placement were deregulated as long as they were viewed as paid for by employers.49 The GEORGE GONOS 597

“about face” in economics from institutionalism to a free market perspective in the 1970s supported these developments.50 Such was the extent of fee deregulation under this logic that in 1990 the industry trade association could say that, “[e]mployment agencies are still licensed in many states although the number requiring a license has dropped and is often limited to the relatively few agencies that charge applicants a fee.”51 Thus the notion of client paid fees became pervasive in the culture of the personnel industry, and trade groups made the most of it in advertising and public relations.

Who really pays? Many pro-worker observers, however, had long realized that “employer paid” fees (fees collected “on the back end” rather than up front) could not be viewed as a solution to the problem of agency fees. As early as 1914 it had been pointed out that agencies advertising “positions furnished free” ultimately found ways to charge workers,52 that is, that fees ostensibly “charged to employers” could easily be passed on to workers by reducing wages or through other means.53 This lesson was internalized into new employment agency laws passed by states in the wake of the 1918 Supreme Court decision nullifying Washington state’s initiative to outlaw the practice of charging agency fees to workers.54 These new laws specified that both worker-paid and employer-paid fees would be recognized as fees. For instance, beginning in 1918 New Jersey defined an employment agency as the business of procuring . . . help or employment . . . where a fee or privilege is exacted, charged or received directly or indirectly ...whethersuchfeeiscollected from the applicant for employment or the applicant for help.55

In this way lawmakers early in the twentieth century explicitly recognized that whether charges were collected “directly or indirectly” — that is, from the worker or the user of labor — they were nonetheless fees in need of regulation, since such charges had the potential effect of diminishing real wages by diverting a certain amount of the total income flow away from workers. Similarly, delegates to the International Labour Organization, deliberating in 1932 over that organization’s position on employment agencies, reached the conclusion that charges collected indirectly, rather than out of the worker’s pocket, nonetheless constituted a fee requiring state regulation. Thus ILO convention No. 34 defined a “fee charging employment agency” as inclusive of any business that supplies workers to employers “with a view to deriving either directly or indirectly any pecuniary or other material advantage from either employer or worker.”56 As interpreted by the ILO Director-General in 1966 (when the issue of temporary help firms was raised) this language clearly signified that a triangular arrangement in which an intermediary paid workers out of revenues collected from the user of labor represented a segment of the fee-charging employment agency business.57 598 POLITICS & SOCIETY

Since early in the twentieth century, then, employment agency law in the United States and internationally reflected the realization of pro-worker advocates that so-called “employer-paid” fees were nothing but an illusion. Experi- ence taught that it was a relatively simple matter for labor market intermediaries to pass along to workers the fees ostensibly charged to their business clients by adjusting pay rates in accordance with desired margins, a process that was facilitated by the natural alliance long established between employment agents and the users of labor.58 But that history was “forgotten” and the myth of employer-paid fees reborn when, through political lobbying in the 1970s and 1980s, the idea was legitimized again and put to work by the staffing industry on a widespread basis. Through public relations work and changes in law, the idea that “no fees” are extracted from workers through staffing and other personnel firms was raised to truth value and official policy across the country. As a result of this change in policy, temporary help and other personnel placement firms that use the device of client paid fees no longer have those fees subject to state regulation, and hence may legally conceal them. The official account now holds that if fees are unseen, that is, if workers do not take money out of their own pockets to pay them, then they do not exist.

State Subsidized Fees

The latest stage in the evolution of fee-charging practices involves the entrance of the state into the employment triangle, as exemplified in the expanding partnership between the commercial staffing industry and the public employment service. The newly elevated doctrine that staffing firms do not charge workers a fee was instrumental in this development, as it provided the justification for allowing free public agencies to refer job applicants to staffing agencies and other private personnel offices. This constituted a change in long-standing policy. Based on the record of employment agency abuse and on the principle that job placement services should be free, federal regulations since 1933 had provided that the state employment service (ES) could “make no referral as a result of which a charge would be made to either the worker or the employer for filling the job.”59 For many years, the private employment agency business had “objected strenuously” to this prohibition,60 but in the 1980s, the widespread acceptance of the notion of client-paid fees provided a new opening. In Congressional hearings on the 1982 Job Partnership Training Act, the private personnel business, and the staffing industry in particular, pressed for inclusion of a specific provision that would encourage the public ES to make referrals to staffing firms and other private agencies.61 JPTA thus included language — in effect an amendment to the 1933 Wagner-Peyser Act—that allowed public employment offices to make such referrals “as long as the applicant is not charged a fee.” The provision allowed public offices to fill job orders from any placement firm that claimed its fees were client paid, and with GEORGE GONOS 599 passage of JPTA, the private personnel industry was eager for a surge of business from this new market. Yet it was only under continuing political pressure, bolstered by the 1986 Gen- eral Accounting Office study entitled “More Jobseekers Should be Referred to Private Employment Agencies,” that the Department of Labor, initially reticent, finally agreed to issue new guidelines to regional offices directing implementation.62 State implementation took different forms. In New Jersey, private agencies seeking referrals from public offices are required to sign an agreement stating that no jobseeker referred will be charged a fee. (Of course, the state accepts the premise that any fees charged by private staffing agencies are “client-paid.”) According to regional ES staff in several states, the program of referrals to temp firms and other private agencies has grown substantially in recent years.63 However, since these referrals are not separately categorized by the ES, no data on their number are currently available.64 It is also on the basis of the “client paid” fees doctrine that public social service agencies other than the Employment Service have been transformed into labor brokers.65 The prime example comes from the states’burgeoning welfare-to-work programs, in which public funds are used to pay subsidies—a portion of the fees—to private agencies that either “place” welfare recipients or, posing as “employers,” hire them directly. In this connection, a staffing industry representative states quite candidly,

The training and jobplacement activities in the staffing industry have a lot of appeal for those who are looking at welfare reform strategies. . . . The ability to create a successful alliance between these governmental organizations and the staffing services industry can go a long way in helping to alleviate the recruiting shortages in our industry.66

Since passage of federal welfare reform in 1996, staffing firms have offered proposals to many states eager to implement expanded “workfare” programs, and now profit from the large number of former recipients entering the workforce.67 In the first year of Wisconsin’s program, staffing firms handled 42 percent of the state’s placement efforts, and 30 percent of the total jobs held by former recipients were through staffing firms.68 Though extensive, the precise extent of the staffing industry’s national role in welfare reform is currently unknown, again because pertinent statistics are not made public. Returning to our discussion of the evolution of fee charging practices, we see that another mutation of policy has occurred. With the developments discussed above, the notion of client paid fees has been stretched again, to include not only employer-paid fees but publicly paid fees as well. But government subsidies paid to staffing firms represent just the first installment of the fee program. As long as these workers remain “employed” by staffing agencies, a significant part of what they earn each hour goes for the agency mark-up, undermining workers’ability to 600 POLITICS & SOCIETY gain self-sufficiency. The typical agency mark-ups of 30 to 50 percent are literally the difference between poverty and a living wage.69

FEE-SPLITTING REVISITED

As noted earlier, fee splitting was among the most reviled abuses of the early employment agency era. State laws prohibited this collusion whereby workers were “given a few days’work and then discharged to make way for new workmen, the agent and employer dividing the fee.”70 During the staffing industry’s recent boom period, however, the issue has disappeared from public discussion. The spread of “client paid” fees has seemed to make regulations or concerns pertain- ing to fee-splitting obsolete.71 But a close look at how billing rates and pay rates are set in the staffing industry casts serious doubt on this optimistic view. Current industry practices, though more sophisticated, are in many respects indistinguishable from those which state law defined as illegal fee splitting early in the century. A strong case can be made that the practice exists today on a much greater scale than it did in its earlier manifestation and has only been hidden in the dynamics of the staffing business. In private negotiations, staffing firms and their clients share information on current labor costs and other operating expenses and decide where to set billing rates and wages so as to allow for “cost savings” to the client and a reasonable operating margin for the agency. These “savings” and the agency’s take, including its fee, are paid for by the benefits that “temp” employees will forego and the difference between the old (“premium”) and the new (“market”) wage, as positions are con- verted from “regular” to “temporary.” The two parties literally decide, in the words of one agency manager, to “split the difference” between them (though not necessarily in equal proportions). This picture is supported by the industry’s own advertising claims, in which “cost savings” are guaranteed for business clients. (“Comforce has helped some clients cut labor costs by as much as 50%.”) For instance, the ad targeted to potential business clients by Comforce Corporation (the “fastest growing” staffing firm from 1996-98, according to Staffing Industry Report) graphically represents the current cost of the workers’ benefits package (including health insurance, holidays, sick days, et al.) as “your [the client’s] savings” (see Figure 1). As shown, other employment-related costs (unemployment insurance, worker’s compensation, payroll administration) become part of the revenue stream of the staffing firm. What was formerly part of the employees’ share of total income has been redistributed or split between client firm and agency.72 Along with the client’s savings on labor costs, the agency’s fee has come out of workers’ pockets. From a business standpoint, fee-splitting in its current form is clearly superior to the cumbersome early twentieth century practice in which employment agents had to collect fees directly, over and over again, from workers, and then rebate a portion of these fees to allied employers. First, although it allows the turnover of GEORGE GONOS 601

Your Costs without COMFORCE

Payroll Administration & Recruitment: 6.0% Moving Expense: 2.9% Federal, S.S., Unemployment: 8.5% Workers Compensation: 2.0% Pension Contribution: 6.4% State Unemployment: 2.7%

Base Hourly Pay

Sick Days (Average, 8 Days): 3.2% Life Insurance & Accident: 2.5% Vacations (2 Weeks): 4.0% Holidays (10 Days): 4.0% Medical/Health Insurance, w/ Dependents: 9.8% Savings & Security: 2.7% Other Accident & Compensation: 0.5%

Your Costs with COMFORCE

Our Cost to Labor

Base Hourly Pay

Your Savings

Figure 1. COMFORCE ad. Source: COMFORCE Corporation (1999). Retrieved from http://comforce.com/comforce/ci.htm workers as frequently as may be desired by the user-firm, rapid turnover is unnecessary to achieve the desired results, since long-term assignments generate continuous fees from every hour worked. Second, since fees are paid indirectly, they are invisible, making collection more efficient and the potential for exploitation 602 POLITICS & SOCIETY even greater. There is no need to collect fees from workers, and no need to make a “kickback” to the client firm. In effect, the “fee-split,” or redistribution of income, in proportions agreed upon ahead of time through negotiation between the agency and client-firm, is consummated when the worker accepts a paycheck and the client firm pays its bill to the agency. As “employer,” the agency issues a paycheck from which all the necessary deductions have already been made—both those required by law, which the worker finds marked, and those unmarked deductions covering the agency’s fee and the “savings” of the corporate client. Depending on the duration of a “temporary” assignment, the worker often ends up paying much more for leasing a job than what would have been the maximum allowable placement fee under state law before deregulation. Tied to the agency, what the long-term temp actually pays in fees (or, if one prefers, in reduced compensation) is not only for placement, but for the privilege of continuing her employment, a situation in no way different from the extortion practiced early in the twentieth century when agents charged periodic fees to “keep” or hold a jobfor one worker against another. By winning the legal endorsement of the temp agency as “employer,” and the acceptance of the notion of client-paid fees, the staffing industry had not only solved the problem posed by government fee ceilings. It had also avoided the old prohibition against fee-splitting by submerging the practice within the private relationship between the employment agency and its client.

The public-private split. One reason that employers have always preferred to utilize private agencies rather than the public service has usually remained unspo- ken. As Grace Abbott was told in 1908 by the superintendent of Chicago’s free public agencies, it is that public agencies have “no fees to divide with contractors.”73 But, now, with the policy changes (reviewed above) that allow public agencies to hand off jobapplicants and former welfare recipients to commercial labor market intermediaries, these “problems” with the public service may have been rectified. Thus, in addressing the issue of how public referrals to private agencies would work, the GAO considered the following suggestion:

Private agencies should split employer-paid fees with the Employment Service if it sup- plied jobseekers for private agency job openings, some state Employment Service officials believe. (This is what private agencies currently do among themselves when they pool resources to match jobseekers with openings.) Such fee-splitting would be an acceptable concept, private agency industry representatives told us.74

Since the term fee-splitting had never before been used in connection with the employment agency business except to refer to an illegal and abusive practice, the suggestion to lend it legitimacy and involve the state as partner would seem particularly striking. For the state, the payoff or “kickback” associated with using commercial labor brokers is the presumed reduction in expenditures for its employment or “welfare” programs. (The expanding practice of leasing prison labor is a particularly salient example of this, as a portion of inmates’ wages goes to offset GEORGE GONOS 603 the costs of prison administration, a public expense.) The alliance of social service agencies with commercial labor market intermediaries favors employers’ interests and has “mainly guided the poor into a low-road in the labor market.”75 Since, thus far, neither private or government agencies have effectively monitored job placements to determine either their quality or duration, the program of referrals to private agencies contributes to the problem of accelerated turnover (“temporary jobs”) that, as we have seen, has always been associated with the practice of fee-splitting. Interestingly, rather than wage a fight to displace the public service to extend its market (as earlier generations of the employment agency business had done), the staffing industry has “partnered” with it, helping to improve state agencies’ “placement” statistics while continuing to serve employers’ interests. From this view, it is no coincidence that the boom in “temporary work” has occurred con- comitantly with the restructuring of the welfare state since the 1970s, since the expansion of the contingent labor market (otherwise called “job creation”) was a “solution” for both the private personnel industry and the state itself. Though the industry does not provide figures on the proportion of its business that is represented by the public sector, government (at the federal, state and local levels) is a significant, and may actually be its largest, client. Thus the state’s heavy reliance on the staffing industry to implement the latest “welfare reform” initiative was a natural choice. In a profound way, the state’s answer to welfare reform literally is temporary work.

Fee-splitting and wage depression. Some observers have recently been critical of such broad concepts as globalization, new technology, and skill differences as causes for the growth of inequality and have pointed to wage structure as the largest contributing factor.76 The mechanics of the wage setting process in the staffing industry (outlined above) is one overlooked locus within which wage depression and income redistribution has taken place. The private negotiation of billing rates and pay rates in the “external” labor market, mediated by the commercial staffing industry, has provided an institutional mechanism by which labor markets have been made to function more like commodity markets in recent years. “Floating” pay rates, the daily adjustment of wages to the supply and demand conditions of the external market, serve the cause of wage flexibility, one of the central functions of the staffing industry for its business clients. The introduction and spread of this wage setting practice in industries and occupations throughout the economy may be seen as one important means by which the older norms of equity and stability characteristic of the post–World War II institutional model of wage setting have been undermined.77 “Splitting the difference” in every transaction, the staffing industry has in the last quarter century played an important role in the “relentless shaving of labor costs,”78 and func- tioned on a macro level to redivide a portion of the total national income paid out to labor and capital. Simply put, the new wave of “fee splitting,” like its historical 604 POLITICS & SOCIETY predecessor, acts not only to exacerbate the irregularity of work but to depress wage scales toward what classical economists call the “correct” market rate.79 Wage setting by the staffing industry not only reduces income for temps, but erodes traditional wage scales for all workers in the occupations and economic sectors where the industry functions to create a secondary labor market. Thus, the involvement of the staffing industry is another reason (in addition to flooding the low-wage labor market) why workfare policies exert downward pressure in the low-wage sector.80 It now appears clear that the rapid expansion of the staffing industry helps account for the relative decline in the growth of compensation in the United States beginning in the early 1970s and the continuation of this trend even in the tight labor markets of recent years. Thus, researchers find consistently lower wage growth in states with a greater share of temporary help employment.81

CONCLUSION

With union decline and the growing externalization of labor markets since the 1970s, “individual bargaining” has been reinstituted as the norm for American workers. Part and parcel of this change has been the resurgence of unregulated labor agents. Historically, periods of high labor mobility or “churning” in Amer- ica have provided the opportunity for private labor agents to place what Veblen called an “overhead charge” on the price of labor. This cost has always come out of the pockets of workers.82 With the widespread return of agency fees and the spread of a new brand of fee-splitting, today’s labor market is eerily reminiscent of the “secret rates and rebates” of a past era of employment relations.83 Filling the “structural hole” left by the loosening of the employer-employee relationship,84 the staffing industry has served as a catalyst for the secondary labor markets burgeoning within companies, occupations, and industries85 and as a medium for the upward redistribution of revenues from workers to firms. What appears to many as a small percent of workers in “temporary” employment has given employers tremendous leverage in the bargaining relationship. In his study of a union machine shop, Burawoy pointed to the crucial role of concealment in the process by which surplus value is appropriated.86 Temporary work and other emergent forms of employment relations depend on new means of obscuring power and exploitation.87 An aura of secrecy has long surrounded the employment agency business.88 The extensive, often blatant, measures taken by staffing agencies to keep their billing rates and the wages of others a secret from workers testifies to the importance of concealment in the temporary employment relationship. By fashioning changes in law and public perception, the staffing industry gained legitimacy for its claim on a share of the revenues flowing from economic enterprise, that is, for its social “takings.”89 The staffing industry’s overall “take” has been growing steadily. If an old fashioned fee schedule for the entire industry were presented to workers for 1998, it would indicate fees of $15.2 billion, a figure that is no doubt conservative, since the trade association reporting GEORGE GONOS 605 the data from which the figure is derived does not represent or count the personnel business as a whole. What would have been “big money crime” in an earlier era of employment agency regulation is now merely the legalized looting of workers throughout the economy.90 Recalling the early history of state regulation lends support to the current fight of temporary workers for disclosure of agencies’ mark-up rates. Due to staunch industry opposition, efforts to pass state “right-to-know” legislation that would require the disclosure of agency mark-up rates have not succeeded. In the meantime, grass roots groups (recently joined together in the National Alliance for Fair Employment) seek voluntary compliance through an industry code of conduct. The old adage to “follow the dollar” holds weighted importance for workers today. As one organization’s Website proclaims,

We believe we have a fundamental right to know how much is being charged for our work....Wearetheones—not the staffing agencies—who add value to the products on which we work. Knowing bill rates empowers contractors to make informed decisions about contracts, benefit options, job classifications and career paths.91

Some “discount brokers” at the high end of the staffing industry now say they provide access to billing information and allow workers to negotiate their own rates. At the low end of the labor market, hidden agency fees undermine the ability of workers to escape poverty. Labor markets mediated by for-profit agencies is not an effective antipoverty strategy. Compensation from work must be increased if employment-based welfare reform is to succeed.92 The significant slice of compensation costs siphoned off by staffing firms belies the assumption that employers cannot afford to pay a living wage. Thus, the analysis presented above suggests certain policy options, all of which have historical precedents in the United States: (1) minimally, requiring the disclosure of agency billing rates would allow workers to bid down agency fees, and would be relatively easy to implement; (2) rein- stating ceilings on the amount of fees that can be extracted from workers is a more difficult measure to implement, but would help address the problem of “permatemping,” that is, keeping workers on staffing agency payrolls for extended periods; (3) once again banning the use of fee-charging agencies by public employment programs and public sector employers might spur the growth of nonprofit agencies and union hiring halls providing alternatives. These measures may be implemented without immediately raising labor costs to employers that now use private agencies. They would, however, result almost immediately in better wages, and ultimately help workers gain greater leverage in bargaining over the terms and conditions of their employment.

NOTES

1. Quoted in Randy Albeda and G.M.J. Veldcamp, Temporary Work and Modern Soci- ety: Part III—Proceedings of the International Conference (Boston: Kluwer, 1979), 123. 606 POLITICS & SOCIETY

2. See, for example, Jackie Krasas Rogers, Temps: The Many Faces of the Changing Workplace (Ithaca, NY:ILR Press, 2000); Jamie Peck and Nik Theodore, “Contingent Chi- cago: Restructuring the Spaces of Temporary Labor,” International Journal of Urban and Regional Research (forthcoming); Heidi Gottfried, “In the Margins: Flexibility as a Mode of Regulation in the Temporary Help Service Industry,” Work, Employment and Society,6, no. 3 (1992). 3. Susan N. Houseman, “Temporary, Part-time, and Contract Employment in the United States: A Report on the W. E. Upjohn Institute’s Employer Survey on Flexible Staffing Policies” (Kalamazoo, MI: W. E. Upjohn Institute for Employment Research); Arne L. Kalleberg, Edith Rasell, Ken Hudson, David Webster, Barbara F. Reskin, Naomi Cassirer, and Eileen Appelbaum, Nonstandard Work, Substandard Jobs: Flexible Work Arrangements in the U.S. (Washington, DC: Economic Policy Institute 1997). 4. Edward A. Lenz, “Apples and Oranges: Some Key Operational and Legal Differ- ences Between Temporary Help and Other Services,” Contemporary Times (Fall 1990): 15. 5. Joseph B. Darby, “The Untouchable Topic: Agencies Explain Why They Charge Fees and What is Reasonable,” Contract Professional (July-August 1998): 37. 6. From the “employees’ agreement” signed by workers sent out by Career Blazers, quoted in William Lewis and Nancy Schuman, The Temp Worker’s Handbook: How to Make Temporary Employment Work for You (New York: American Management Associa- tion, 1988), 62. 7. Washington state House Bill 2756 (2000); Mike Blain, “We Need Full Disclosure of Agency Fee Rates,” Contract Professional (November-December 1998): 72; Brian O’Connell, “The Untouchable Topic Part II: Agency Fees and Lack of Disclosure Feed Contractors’ Anger,” Contract Professional (November-December 1998): 33-7. 8. Elizabeth Mertz, “The Uses of History: Language, Ideology, and Law in the United States and South Africa,” Law & Society Review, 22, no. 4 (1988): 682. 9. A Massachusetts critic referred to fee-charging agencies as “leeches engaged in sucking the life blood from the poor.” Quoted by Alexander Keyssar, Out of Work:The First Century of Unemployment in Massachusetts (Cambridge: Cambridge University Press, 1986), 259. “The necessity of paying for the privilege of going to work,” the U.S. Commis- sion on Industrial Relations stated in 1912, “seems foreign to the spirit of American freedom and opportunity.” Quoted in Adams v. Tanner, 244 U.S. 590, 604 (1917). 10. Adams v. Tanner 244 U.S. 590, 606 (1917). 11. E. L. Bogart, “Public Employment Offices in the United States and Germany,” Quarterly Journal of Economics, 14 (1900): 341-377. 12. John R. Commons and John B. Andrews, Principles of Labor Legislation (New York: Augustus M. Kelley, 1967 [1916, 1936]). 13. Adams v. Tanner, 244 U.S. 590; John B. Andrews, “Fee-charging Employment Agencies Must Be Effectively Regulated,” American Labor Legislation Review, 19 (1929): 367-70. 14. Ribnick v. McBride, 277 U.S. 350 (1928). 15. Chairman Tolan of the Select Committee of the House of Representatives, quoted by W. S. Woytinsky and Associates, Employment and Wages in the United States (New York: Twentieth Century Fund, 1953), 180. Seeking to close the gap in the Wagner Act which left workers using private employment agents unprotected, the Committee introduced legislation (backed by the Roosevelt administration) to impose federal regulation on employment agents under the supervision of the U.S. Department of Labor. The legislation failed to materialize. 16. Olsen v. Nebraska, 313 U.S. 236 (1941). In once again setting ceilings on employment agency fees, the states were guided by a uniform bill prepared by the U.S. DOL. See GEORGE GONOS 607

U. S. Department of Labor, Private Employment Agencies: Laws Relating to Their Regula- tion as of Sept. 1, 1943, Bulletin No. 57 (Washington, DC: U.S. Department of Labor, 1943). 17. Council of State Governments [hereafter CSG], The Book of the States, Volumes 8-28 (Lexington, KY: Council of State Governments, 1950-1970). 18. U.S. Department of Labor. State Laws Regulating Private Employment Agencies, Bulletin 209 (Washington, DC: Bureau of Labor Standards, 1960), 14. 19. U.S. Department of Labor, State Laws Regulating Private Employment Agencies, Bulletin 252 (Washington, DC: Bureau of Labor Standards, 1962), 18; Arthur Goldberg, Growthof Labor Law in theUnited States (Washington, DC: U.S. Department of Labor, 1962). 20. See George Gonos, “The Contest over ‘Employer’ Status in the Postwar United States: The Case of Temporary Help Firms,” Law & Society Review, 31, no. 1 (1997): 81-110. 21. CSG 1972-73, 508-509. 22. CSG 1970-71, 502; 1974-75, 501; 1976-77, 535; Tomas Martinez, The Human Marketplace: An Examination of Private Employment Agencies (New Brunswick, NJ: Transaction, 1976). 23. The standard split in 1915 was reported as 40 percent of the fees to the foreman or employer. Frances A. Kellor, Out of Work: A Study of Unemployment (New York: Arno Press, 1971 [1904, 1915]), 187. 24. As explained by the U.S. Commission on Industrial Relations in 1915: “The foreman agrees to hire men of a certain employment agent on condition that one-fourth or one-half of every fee collected from men whom he hires be given to him. This leads the foreman to discharge men constantly in order to have more men hired through the agent and more fees collected.” Quoted by Goldberg, Growth of Labor Law, 140. 25. Grace Abbott, “The Chicago Employment Agency and the Immigrant Worker,” American Journal of Sociology, 14, no. 3 (1908); Bogart, “Public Employment Offices.” 26. George H. Trafton, “Employment Agencies Officially Exposed: Sworn Testimony Shows Urgent Need of State Action,” American Labor Legislation Review, 20 (1930): 27-36. 27. Goldberg, Growth of Labor Law, 1962. 28. Woytinsky, Employment and Wages in the United States. 29. This arrangement came to be known as the “three gang system,” as it meant having one group of workers on the job, another on its way from the agency, and a third group returning there to be rehired. Thus, workers complained of employment agents having “three men for one job . . . and receiving compensation (fees) from all.” Commons and Andrews, Principles of Labor Legislation, 7; see also Adams v. Tanner, 244 U.S. 590, 602-3 (1917). 30. Kellor, Out of Work, 187. 31. In many cases, employers refused to hire directly, insisting that workers secure their jobthrough a preferred agency, and possiblyeven one the employer himself owned or con - trolled. By arrangement with employers, some employment agents could receive fees for workers they never saw, that is, those who were actually hired directly, a practice the contemporary staffing industry calls “payrolling.” See Kellor, Out of Work, 140-1. 32. U.S. DOL, State Laws, 1962. 33. Florida Industrial Commission v. Manpower of Miami, 91 So.2d 197, 199 (1956). 34. Gonos, “The Contest Over ‘Employer’ Status.” 35. N.J.L. 1918 Chap. 227, Sec. 1(b). 36. Mack A. Moore, “The Legal Status of Temporary Help Services,” Labor Law Jour- nal, 16 (1965): 622. 608 POLITICS & SOCIETY

37. Mack A. Moore, “Proposed Federal Legislation for Temporary Labor Services,” Labor Law Journal, 26 (1975): 773. 38. Nebraska v. Manpower of Omaha, 73 N.W.2d 692 (1955); Manpower, Inc. of New Jersey v. Richman (Attorney General of New Jersey), Superior Court of New Jersey, Law Division, Essex County, Docket No. L-22576-56 (June 24, 1957, unpublished). 39. “The mark-up included in the billing rate is as legitimate as profit is in any other kind of business venture, and should not be treated as a fee under any conditions.” Moore, “The Legal Status of Temporary Help Services,” 634. 40. Ibid., 633. 41. See Gonos, “The Contest Over ‘Employer’ Status.” 42. U.S. DOL, State Laws, 1960. 43. Manpower v. Richman, 12. 44. U. S. Congress, Hearings Before the Special Subcommittee on Labor of the Commit- tee on Education and Labor on H.R. 10349, A Bill to Establish and Protect the Rights of Day Laborers (92nd Congress). (Washington, DC: Government Printing Office 1971), 177. 45. Labor Management Relations Act, Section 8(b)(5). 46. Goldberg, Growth of Labor Law, 135. 47. CSG, 1976-77, 535. New York, for example, eliminated statutory ceilings on “employer-paid” fees provided the worker was not also charged a fee. Ibid., 536. 48. CSG, 1974-1991. Throughout this period a continuing narrow battle was waged in the state legislatures between “competing bills” or “sharply divergent proposals” brought by state regulatory agencies supporting continued regulation and industry forces seeking a policy of deregulation or “self-regulation.” CSG, 1974-75, 501; 1976-77, 535. Most new legislation “incorporated industry objectives almost entirely.” CSG, 1972-73, 508. 49. The same reasoning was applied also to what the staffing industry called “liquidated damages,” the charges made by staffing firms when temps are hired “away from them” by client firms. Many states provided that such charges would not be considered as “fees” (and would therefore be exempt from regulation) as long as client firms and not workers were held liable. For example, CSG, 1972-73, 509; 1976-77, 536. Since “liquidated damage” charges appeared to many observers to be identical to placement fees, their use threatened to undermine staffing firms’ claim to be something other than traditional employment agencies. The industry’s position, which became widely accepted in practice, was that staffing firms made these charges “not as payment for arranging the hire of their temporary employees, but for breach of the customer’s promise not to hire them.” Edward A. Lenz, “Liquidated Damages Revisited: A Discussion of the Legalities and Pitfalls of Liquidated Damages Provisions in Temporary Help Agreements,” Contemporary Times (Spring 1986): 10, emphasis in the original. 50. Bruce Kaufman, “The Postwar View of Labor Markets and Wage Determination,” in B. Kaufman, ed., How Labor Markets Work (Lexington, MA: Lexington Books, 1988), 180. 51. Lenz, “Apples and Oranges,” 15. Interestingly, at the same time states were deregu- lating firms that relied on client-paid fees, they were simultaneously establishing more restrictive regulations for the “relatively few” employment agencies that were not exempted from coverage. To maintain their legal capacity in this area, it was important for the states to keep the regulation of employment agencies “on the books,” even though the law now applied to a very small segment of the industry. This strategy allowed states to keep a regulatory unit open, and to continue to utilize the rhetoric of consumer protection, even in a deregulatory phase, and served the industry’s needs as well. Indicative of this was the practice devised in New Jersey and other states whereby staffing firms that were no longer required to seek licensure continued to be “registered” by the state, a practice that kept the “regulatory” office busy, and allowed the state to continue to charge annual fees to GEORGE GONOS 609 deregulated firms. (Agency operators complain about these fees, referring to them as “G and C”—graft and corruption.) Quite keenly, staffing firms had found a way out of regulation, not by defeating the idea of regulation itself, but by dissociating specific types of practices and firms from identification with the regulated “employment agency” business. Surely this way around fee ceilings was smarter than a direct assault on fee regulation (or worker protection) as such. Regulation was fine, as long as it was aimed at someone else. 52. Kellor, Out of Work, 164-5. 53. The same point was made early on by labor’s opponents of the plan for compulsory “employer-paid” unemployment insurance. Especially in the nonunion sector, such employers could unilaterally reduce wages to pay for premiums. Kenneth Casebeer, “Unemployment Insurance: American Social Wage, Labor Organization and Legal Ideol- ogy,” Boston College Law Review, 35, no. 2 (1994). 54. Andrews, “Fee Charging Employment Agencies.” 55. N.J.L. 1918, chap. 227, sec. 1(d), author’s emphasis. 56. Convention No. 34 Concerning Fee-Charging Employment Agencies, 1933, Article 1, author’s emphasis. 57. International Labour Office. “Interpretation of Decisions of the International Labour Conference: Fee-charging Employment Agencies Convention (Revised),” Official Bulletin, 49, no. 3 (1966): 393. 58. Mark Granovetter, Getting a Job (Chicago: University of Chicago, 1995); Martinez, The Human Marketplace. See also the discussion of “the pass-along economy” in Mark Green and John F. Berry, The Challenge of Hidden Profits: Reducing Corporate Bureau- cracy and Waste (New York: William Morrow and Co., 1985), 296, 315. As one study puts it, low business failure rates in the temporary help industry reflect the “efficiency with which agencies are able to pass on the costs of order-book fluctuations to the workforce.” Peck and Theodore, “Contingent Chicago,” 30. 59. Quoted in U. S. General Accounting Office, Employment Service: More Jobseekers Should Be Referred to Private Employment Agencies, Washington, DC: Report to the Chairman, Committee on Governmental Affairs, U.S. Senate, 1986, 13-4, author’s emphasis. When the question of whether public employment offices should fill job orders from temp agencies was raised, state representatives at the Interstate Conference of Employment Security Agencies in 1959 objected on the grounds that such practice would run against long-standing policy and tradition. William Haber and Daniel H. Kruger, The Role of the United States Employment Service in a Changing Economy (Kalamazoo, MI: Upjohn Institute for Employment Research, 1964), 67. 60. GAO, Employment Service, 30. 61. U.S. Senate, Employment and Training Policy: Joint Hearings before the Subcom- mittee on Employment and Productivity of the Senate Committee on Labor and Human Resources and the Subcommittee on Employment Opportunities of the House Committee on Education and Labor, Parts I and II (97thCongress, 2nd session) (Washington, DC: Government Printing Office, 1982). 62. At least one DOL official raised doubts about the reality of “employer-paid” fees. What, he asked, would “preclude the private agency from referring applicants from the State agency to employers who pay the private agency but recoup all or part of the fee from the employee?” GAO, Employment Service, 33. 63. ES staffers said in interviews that staffing firms make more extensive use of ES services than any other group of employers, some using it almost daily as an extension of their own offices. One staffer said he suspected that private agencies use “America’s Talent Bank” (the ES’s on-line jobsearch service, which permits employers to search a pool of resumes) to circumvent the rule that they must have a signed agreement with the state to get 610 POLITICS & SOCIETY

ES referrals. Michigan contracts with Kelly Services to run the computerized “Talent Bank” for that state. 64. Some public ES offices have added their own “fee-based enhanced services” to their free “core” services. 65. See Michael Indergaard, “Retrainers as Labor Market Brokers: Constructing Net- works and Narratives in the Detroit Area,” Social Problems, 46, no. 1 (1999): 67-87. 66. Bruce Steinberg, “Temporary Help Services: An Annual Update for 1994,” Con- temporary Times (Spring 1995): 11-16, my emphasis. 67. Ellen Reese, “From Welfare Rights to Workers’ Rights: Political Struggles over Welfare Reform in Los Angeles” (paper presented at the Annual Meeting of the American Sociological Association, Chicago, 1999). 68. John Pawasarat, “The Employer Perspective: Jobs Held by the Milwaukee County AFDC Single Parent Population, January 1996-March1997,” University of Wiscon- sin-Milwaukee, Employment and Training Institute, 1999. 69. The author’s calculation of agency mark-ups is based on industry data from 1990-1998. 70. U.S. Bureau of Labor Bulletin No. 109, quoted in Adams v. Tanner, 244 U.S. 590, 601. 71. Martinez’study of the employment agency business concluded that, “The extent of fee-splitting today is probably less than it was in the early 1900s, largely because of the dis- appearance of large labor contractors and the increased risks of detection by a government agency or moral crusader.” Martinez, The Human Marketplace, 48. Martinez’ study considered only the traditional or permanent segment of the employment agency business and did not include the temporary help and staffing industry. 72. “Secondary sourcing” arrangements, in which a “primary vendor” subcontracts with other staffing agencies, a frequent practice in the high-tech field, amounts to splitting in multiple ways. Income may be split with other businesses as well, as when check cashing firms locate within agency offices in the industrial sector, or when transportation to work sites is contracted to an outside firm. 73. Abbott, “The Chicago Employment Agency,” 294. 74. GAO, Employment Service, 22. 75. Indergaard, “Retrainers as Labor Market Brokers,” 77. 76. James K. Galbraith, Created Unequal: The Crisis in American Pay (New York:Free Press, 1998); Michael Hout, “Inequality at the Margins: The Effects of Welfare, the Mini- mum Wage, and Tax Credits on Low-Wage Labor,” Politics & Society, 25, no. 4 (1997): 513-524; Lucy Williams, “Rethinking Low-Wage Markets and Dependency,” Politics & Society, 25, no. 4 (1997): 541-550. 77. Paul Osterman, Securing Prosperity: The American Labor Market: How It Has Changed and What to Do About It (Princeton: Princeton University Press, 1999), 20. 78. Robert Kuttner, Everything For Sale: The Virtues and Limits of Markets (Chicago: University of Chicago Press, 1999), 80. 79. Given the staffing industry’s wage-setting practices, it becomes clear that the industry’s use of the term “mark-up” is misleading. The term mark-up implies a certain ordering of events, namely, that the cost of the “merchandise” to be sold—here, the temp worker’s wage—has been determined first by the agency, and then “marked-up” a certain percentage before being sold to the corporate user. For all but a small group of temps at the high end of the market, however, the flow of events is closer to being the other way around. In common practice, the prevailing pay rate at the time the client firm hires temps has been used as a guide in setting billing rates, that is, as a starting point for negotiations. Client firms, however, are normally unwilling to add the agency mark-up to the original quote. Hence, after GEORGE GONOS 611 the final purchase price (billing rate) is set in negotiation with the corporate customer, it is marked down by the staffing agency to arrive at the temp worker’s wage, insuring the agency a certain margin. This process accounts in part for why temps usually make a lower wage than regular employees doing the same work, and for the downward pressure on wages caused by this wage setting process. Further, since the hourly wage of the clients’ regular employees has been used as a guide in setting billing rates, the cost of temporary labor to client firms has typically been the same or very close to the hourly wage of their regular (or former) employees in the same jobpositions. The greatest part of the corporate customer’s immediate “savings” have typically come not from lower hourly rates, but from savings on benefits. See Arne Kalleberg and Kathryn Schmidt, “Contingent Employment in Organizations: Part-time, Temporary, and Subcontracting Relations,” in A. L. Kalleberg, D. Knoke, P.V.Marsden, and J. L. Spaeth, eds., Organizations in America: Analyzing Their Structures and Human Resource Practices (Thousand Oaks, CA: Sage, 1996), 253-275; Mack Moore, “The Temporary Help Service Industry: Historical Development, Operation and Scope,” Industrial and Labor Relations Review, 18 (1965): 569; Houseman, “Tempo- rary, Part-time, and Contract Employment.” 80. Elaine McCrate, “Welfare and Women’s Earnings,” Politics & Society 25, no. 4 (1997): 417-442. 81. Lawrence F. Katz and Alan B. Krueger, “The High-Pressure U.S. Labor Market of the 1990s,” Working Paper No. 416, Princeton University Industrial Relations Section, 1999, 43. 82. E. Epstein and J. Monat, “Labor Contracting and its Regulation: I,” International Labour Review, 107 (1973): 462. The labor market intermediary “makes a profit out of the earnings of the workers for whom he finds employment.” 83. Louis Adamik, Dynamite: The Story of Class Violence in America (New York: Vin- tage, 1958 [1931]), 23. 84. See Robert Tillman and Michael Indergaard, “Field of Schemes: Health Insurance Fraud in the Small Business Sector,” Social Problems, 46, no. 4 (1999): 572-590. 85. Randy Hodson, “Companies, Industries, and the Measurement of Economic Seg- mentation,” American Sociological Review, 49 (1984): 335-348. 86. Michael Burawoy, Manufacturing Consent: Changes in the Labor Process Under Monopoly Capitalism (Chicago: University of Chicago Press, 1979). 87. Vicki Smith, “The Fractured World of the Temporary Worker: Power, Participation, and Fragmentation in the Contemporary Workplace,” Social Problems, 45, no. 4 (1998): 411-30. 88. Martinez, The Human Marketplace, 43. 89. Elizabeth Mertz, “The Uses of History,” 682. The “overall division between worker incentives and capitalist incentives is embedded in the structure of the legal system.” Arthur L. Stinchcombe, Economic Sociology (New York: Academic Press, 1983), 183, 140-43. On “legitimized distribution,” see also Pierre Bourdieu, “The Force of Law: Toward a Sociology of the Juridical Field,” Hastings Law Journal, 38 (1987): 837. 90. See Kitty Calavita, Henry N. Pontell, and Robert H. Tillman, Big Money Crime: Fraud and Politics in the Savings and Loan Crisis (Berkeley, CA: University of California, 1997) for analysis of an illegal example of “looting” or “collective embezzlement.” 91. Washington Alliance of Technology Workers (www.washtech.org). 92. Fred Block and Jeff Manza, “Could We End Poverty in a Postindustrial Society? The Case for a Progressive Negative Income Tax,” Politics & Society, 25, no. 4 (1997): 473-511; Kevin Duncan, “Incentives and the Work Decisions of Welfare Recipients: Evi- dence from the Panel Survey of Income Dynamics, 1981-1988,” American Journal of Eco- nomics and Sociology, 59, no. 3 (2000): 433-449. POLITICSINDEX & SOCIETY INDEX

POLITICS & SOCIETY

VOLUME 29

Number 1 (March 2001), 1-168 Number 2 (June 2001), 169-332 Number 3 (September 2001), 333-480 Number 4 (December 2001), 481-616

Authors:

AGRAWAL, ARUN, and ELINOR OSTROM, “Collective Action, Property Rights, and Decentralization in Resource Use in India and Nepal,” 485. BACCARO, LUCIO, “‘Aggregative’ and ‘Deliberative’ Decision–Making Procedures: A Comparison of Two Southern Italian Factories,” 243. BAIOCCHI, GIANPAOLO, “Participation, Activism, and Politics: The Porto Alegre Experiment and Deliberative Democracy Theory,” 43. BECKWITH, KAREN,“Gender Frames and Collective Action: Configurations of Mascu- linity in the Pittston Coal Strike,” 297. BIERNACKI, RICHARD, “Labor As an Imagined Commodity,” 173. CHANDRA, KANCHAN, “Ethnic Bargains, Group Instability, and Social Choice The- ory,” 337. DIAMANT, NEIL J., “Making Love Legible in China: Politics and Society during the Enforcement of Civil Marriage Registration, 1950-66,” 447. FUNG, ARCHON, “Accountable Autonomy: Toward Empowered Deliberation in Chi- cago Schools and Policing,” 73. FUNG, ARCHON, and ERIK OLIN WRIGHT, “Deepening Democracy: Innovations in Empowered Participatory Governance,” 5. GONOS, GEORGE, “Fee-Splitting Revisited: Concealing Surplus Valuein the Temporary Employment Relationship,” 589. HELLER, PATRICK, “Moving the State: The Politics of Democratic Decentralization in Kerala, South Africa, and Porto Alegre,” 131. KRIPPNER, GRETA R., “The Organization of Disorganization in Agricultural Labor Markets,” 363. LIEBERMAN, EVANS., “National Political Community and the Politics of Income Taxa- tion in Brazil and South Africa in the 20th Century,” 515. OSTROM, ELINOR, see Agrawal, A. ROTHSTEIN, BO, “Social Capital in the Social Democratic Welfare State,” 207.

RUIZ, JAVIERASTUDILLO, “Without Unions, but Socialist: The Spanish Socialist Party and Its Divorce from Its Union Confederation (1982-96),” 273. SEGAL, ADAM, and ERIC THUN, “Thinking Globally, Acting Locally: Local Govern- ments, Industrial Sectors, and Development in China,” 557. STONER-WEISS, KATHRYN, “The Limited Reach of Russia’s Party System: Underinstitutionalization in Dual Transitions,” 385. THOMAS, CRAIG W., “Habitat Conservation Planning: Certainly Empowered, Some- what Deliberative, Questionably Democratic,” 105. THUN, ERIC, see Segal, A. WANG, NING, “The Coevolution of Institutions, Organizations, and Ideology: The Longlake Experience of Property Rights Transformation,” 415. WRIGHT, ERIK OLIN, see Fung, A.

Articles:

“Accountable Autonomy: Toward Empowered Deliberation in Chicago Schools and Policing,” Fung, 73. “‘Aggregative’ and ‘Deliberative’ Decision–Making Prodecures: A Comparison of Two Southern Italian Factories,” Baccaro, 243. “The Coevolution of Institutions, Organizations, and Ideology: The Longlake Experience of Property Rights Transformation,” Wang, 415. “Collective Action, Property Rights, and Decentralization in Resource Use in India and Nepal,” Agrawal and Ostrom, 485. “Deepening Democracy: Innovations in Empowered Participatory Governance,” Fung and Wright, 5. “Ethnic Bargains, Group Instability, and Social Choice Theory,” Chandra, 337. “Fee-Splitting Revisited: Concealing Surplus Value in the Temporary Employment Rela- tionship,” Gonos, 589. “Gender Frames and Collective Action: Configurations of Masculinity in the Pittson Coal Strike,” Beckwith, 297. “Habitat Conservation Planning: Certainly Empowered, Somewhat Deliberative, Ques- tionably Democratic,” Thomas, 105. “Labor as an Imagined Commodity,” Biernacki, 173. “The Limited Reach of Russia’s Party System: Underinstitutionalization in Dual Transi- tions,” Stoner-Weiss, 385. “Making Love Legible in China: Politics and Society during the Enforcement of Civil Mar- riage Registration, 1950-66,” Diamant, 447. “Moving the State: The Politics of Democratic Decentralization in Kerala, South Africa, and Porto Alegre,” Heller, 131. “National Political Community and the Politics of Income Taxation in Brazil and South Africa in the 20th Century,” Lieberman, 515. “The Organization of Disorganization in Agricultural Labor Markets,” Krippner, 363. “Participation, Activism, and Politics: The Porto Alegre Experiment and Deliberative Democracy Theory,” Baiocchi, 43. “Social Capital in the Social Democratic Welfare State,” Rothstein, 207. “Thinking Globally, Acting Locally: Local Governments, Industrial Sectors, and Develop- ment in China,” Segal and Thun, 557. “Without Unions, but Socialist: The Spanish Socialist Party and Its Divorce from Its Union Confederation (1982-96),” Ruiz, 273.