Forty-two individual-level questionnaires and five focus groups were administered in each of six case study districts, yielding a total sample size of 252 for the survey and thirty focus groups. The six case study districts were Lusaka Urban, Luanshya, Livingstone, Mongu, Kasama, and Chipata (see Figure B.1). These districts were chosen because they offered variation in urban/rural location and because they constitute urban/rural pairs with the same dominant language groups, thereby making it possible to assess the effects of urban/rural location while controlling for language. Lusaka Urban, Luanshya, and Livingstone are all urban districts located on Zambia's industrial rail line. Mongu, Kasama, and Chipata are overwhelmingly rural districts, though they all contain the administrative headquarters of their respective provinces. Kasama and Luanshya are both Bemba-speaking. Chipata and Lusaka Urban are both principally Nyanja-speaking. Mongu and Livingstone are both Lozi-speaking.

Measuring Ethnic Identities and Attitudes

Collecting reliable data on respondents' ethnic attachments and attitudes about ethnic politics is fraught with methodological obstacles. The simple-minded solution of simply asking respondents “who they are” or “what they think” about the role that ethnicity plays in political life is undermined by the fact that individuals possess multiple ethnic identities – they “are” many things – and the salience of each identity varies with the context in which the respondents are asked to reflect on who they are and what it means to them.

Many of the analyses reported in this book involve comparing the tribal backgrounds of parliamentary candidates with the tribal demographics of the constituencies in which they competed. Undertaking these analyses required collecting information about candidates' tribal affiliations. Due to the very large number of candidates whose tribal backgrounds I was interested in ascertaining, it was impossible to contact each candidate or, when they were deceased, their living family members directly. Instead, I took advantage of the fact that a person's tribal affiliation can be discerned from their name with a high degree of accuracy in Zambia, particularly by those who come from the same part of the country.

I began by drawing up a list of every candidate who had run in the elections of 1968, 1973, 1978, 1983, 1988, 1991, and 1996, as well as in all Third Republic by-elections held through the end of 1999. I then sorted this master list by district, generating 57 separate worksheets containing inventories of every candidate that had ever run for Parliament in that district. These worksheets contained a total of 2,231 different candidates.

The data-collection exercise took place during July and August 1999. My strategy was to track down, in Lusaka, people from each district who could serve as informants on the tribal backgrounds of the candidates from their district. I especially sought out older men or women who had spent a major portion of their lives in the district in question.

As a consequence of the administrative policies of the Northern Rhodesian colonial state and its missionary and mining company partners, Zambians in the post-independence era came to view both their own ethnic identities and the country's larger social landscape in terms of two competing schemes of ethnic classification: one defined by affiliation with one of the country's six dozen tribes, and the other defined by membership in one of the country's four principal language groups. Each scheme of ethnic classification implies a very different landscape of political competition and conflict. It also provides people with a very different set of criteria for determining, in the course of that competition and conflict, who is and is not a member of their ethnic group.

The question is: given this ethnic multi-dimensionality, can we generalize about the conditions under which Zambians understand politics as a struggle among the country's tribes rather than as a conflict among its language groups? Can we construct an explanation to account for when one of these ethnic cleavages will become politically salient rather than the other? Part I furnished the background necessary to ask these questions. Part II provides the empirical and theoretical tools with which to answer them.

Chapter 4 provides the foundation for our explanation by showing why ethnicity matters in Zambia and how it is used politically. The chapter shows that ethnicity matters because of the widespread expectation that politicians will channel patronage resources to members of their own ethnic groups.

Much transpired in Poland's politics and economy after 1997. Economic growth slowed and unemployment increased dramatically. The rightist coalition of church and trade-union leaders that dominated the 1997 election disintegrated to the point it could not gain enough votes to be seated in the parliament after the 2001 elections. A casual inspection might suggest that these events contradict our previous analysis and conclusions, and this might be true. The detailed economic data that form the basis for the preceding chapters are not available for the years after 1997, preventing a careful analysis of the four years following 1997 and a rigorous testing of the propositions developed earlier. We use what data are available to explore several explanations for what transpired between 1997 and 2001. The goal is to offer some insight into whether the conclusions we reached about the period from 1990 to 1997 are only applicable for that stage of the transition or whether the model we develop is consistent with this later period as well. After a brief summary of the main economic and political events, we examine the economic changes, compare the voting patterns in the 2001 with the votes cast in 1997, and then evaluate the model and its usefulness in understanding the later stages of the Polish transformation.

THE ECONOMY

Table 8.1 shows the level of real GDP relative to 1989, the unemployment rate, and the change in GDP for the period 1989 to 2001.

In this chapter, I explain why, when contemporary Zambians think about who they are, and about who is and is not a member of their ethnic group, one of the ways they do so is as members of one of the country's six dozen tribes. I begin by showing how the institutions of Northern Rhodesian colonial rule created incentives for rural Africans to invest in their identities as tribe members. Then I account for the perpetuation (and reinforcement) of these identities among urban migrants. I conclude by explaining how the landscape of tribal divisions came to have its present contours.

THE ORIGINS OF TRIBAL IDENTITIES IN RURAL ZAMBIA

The British South Africa Company (BSA Co.) took over the administration of the territory that later became Zambia in 1894. The company sought to administer the territory with two goals in mind. First, it sought to extract labor from the local population to sustain its mining efforts and satisfy the demands of its white settler population for African workers. Second, it sought to minimize its costs. Taxation was embraced from the beginning as the key vehicle for achieving both of these goals. It not only generated revenue but, because taxes were payable in cash only, it also induced large numbers of African men to take up wage employment for Europeans. First imposed in the northeastern part of the territory in 1900 and in the northwestern part in 1904, taxation quickly became “the heart and soul of Company government in Northern Rhodesia, and the constant preoccupation of district officials” (Coombe 1968: 12).

The creation of a pro-reform constituency is a necessary but not a sufficient part of choosing a pro-reform government. The method for allocating seats in parliament once votes have been cast is a critical concern. In any system, but particularly in a parliamentary system with proportional representation, such as Poland's, the electoral rules creating voting and parliamentary districts and determining how votes are converted to seats for these districts have a critical bearing on what influence pro-reform interests can exert on national policy. At this point the findings about the clustering of new enterprises and the agglomeration effects identified in Chapter 3 become very important. The location effects lead to new businesses being concentrated in certain areas, even beyond what traditional factor prices and resources would predict. A consequence of this clustering is that the pro-reform voters will also be concentrated rather than being widely distributed across the country. Thus, how districts are drawn and how seats are apportioned will affect the representation of liberal interests.

SEAT ALLOCATION RULES

A critical set of rules was adopted prior to the 1993 election at the urging of the coalition led by the UD. The political instability experienced after the 1991 elections was attributed, with some justification, to the presence of so many small parties in the parliament. Three critical rules were designed to reduce the representation of smaller parties and to push consolidation of existing parties: parties had to attain 5 percent and coalitions 8 percent of the national vote in order to hold seats in the new Sejm; there was a national list of sixty-nine seats allocated to parties with more than 7 percent of the national vote; and the d'Hondt formula was used to allocate seats within each voivodship among the eligible parties.

Do the Zambian findings presented in this book travel to other African nations? Is there any evidence that transitions between one-party and multi-party rule in other African countries have brought changes in the kinds of ethnic cleavages that structure political competition and coalition-building? More than two dozen African nations have held parliamentary elections under both single-party and multi-party political systems, so in principle it should be possible to test whether the Zambian results hold elsewhere in the region. For them to be eligible as comparison cases, the countries would have to have multi-dimensional ethnic cleavage structures (so variation in the cleavage outcome is possible) and to be places where ethnicity is understood by voters to convey information about how politicians distribute patronage. Nearly all African countries meet both of these criteria. Almost all have local cleavages defined by tribal affiliation or clan membership and national-scale divisions based on religion, language, or region. A handful of African countries – Sudan, Zimbabwe, Mauritius, Tanzania, South Africa – have significant racial cleavages as well.

As for the role that ethnicity plays, Africa is a region whose poverty and weak government institutions lead citizens to view the state as a resource to be consumed by the ethnic kin of those who control its offices. Young and Turner summarize the partrimonial basis of African politics well:

The motivation for the model presented in the last chapter was the observation that changes in regime type in Zambia seem to co-vary with changes in the relative political salience of linguistic and tribal identities in national elections. During multi-party contests, ethnic politics revolves around language group divisions, whereas during one-party elections it revolves around tribal differences. Given this pattern of co-variation, it is natural to assume that it is something about the multi-party or one-party nature of the electoral regime that is driving the salient cleavage outcome. However, it is at least possible that other factors that happen to co-vary with regime type could be responsible for the changes we see in the salience of tribal and linguistic identities. If so, these factors would offer competing explanations for the argument presented in Chapter 5. The first part of the present chapter explores this possibility.

The second part of the chapter takes up another potential problem: endogeneity. Even if we are able to rule out the possibility that something other than regime change has caused Zambian politicians and voters to shift their focus from one ethnic cleavage dimension to the other, we might still have the causal arrows backward. It is possible that changes in the salience of tribal and linguistic identities, driven by factors other than changes in regime type, are what caused the transitions from multi-party to one-party rule and back. This possibility needs to be ruled out for the argument advanced in the book to hold.

In Identity in Formation, David Laitin writes that “ethnic entrepreneurs cannot create ethnic solidarities from nothing. They must, if they are to succeed, be attuned to the micro incentives that real people face” (1998: 248). Having identified the micro-incentives that Zambians face in Chapters 4 and 5, this chapter tests whether or not politicians are attuned to them. It investigates whether political elites behave in the way that the model would predict in each institutional setting.

The chapter is divided into three sections. In the first, I present evidence that politicians make different sorts of ethnic appeals in one-party and multi-party political campaigns: tribal in the former and linguistic in the latter. In the second, I show that the shift from multi-party to one-party rule alters parliamentary candidates' choices about the constituencies in which it will be most advantageous for them to run. Whereas running in a constituency where one is a member of the dominant tribe is of paramount importance in one-party elections, it is much less important in multi-party contests, since electoral success in those races is much more a function of a candidate's party affiliation than of his or her ethnic background. Drawing on information about the ethnic demographics of electoral constituencies and the tribal backgrounds of each of the more than 2,200 candidates that ran for Parliament between 1968 and 1999, I present a series of quantitative analyses that show that the different rules for what it will take to win leads candidates to choose to run in different kinds of constituencies in one-party and multi-party elections.

This chapter lays the foundation for the argument presented in Chapter 5 by accomplishing three tasks. In its first part, the chapter establishes the relevance of ethnicity in post-independence Zambia by showing that ethnic group memberships underlie people's perceptions of how patronage resources are distributed by those who enjoy access to them. I show that, in a context where all politicians promise to distribute jobs and development resources to the people whose votes they are seeking, voters use ethnicity as a cue to help them distinguish promises that are credible from promises that are not. I argue that it is the information that ethnicity is assumed to convey about likely patterns of patronage distribution – not atavism or tradition – that explains why it plays such an important role in Zambian political life.

The second part of the chapter shows how these expectations shape the strategies that politicians, parties, and individual voters employ to construct and secure membership in winning political coalitions. Specifically, I show that, because Zambians assume that having a member of their own ethnic group in a position of power will increase their access to patronage resources, they are inclined to join coalitions led by members of their own ethnic groups, to be sympathetic to electoral appeals couched in ethnic terms, and to be skeptical of promises made by leaders of groups other than their own. I show that, knowing this, politicians and parties appeal to voters' ethnic affiliations in predictable ways.

Introduction

Swiss federalism shares attributes with both United States and German federalism. As in the United States, an essential goal of the federalist project is to allow for differences in living conditions among the constituent territorial units. When the Swiss cantons formed a federal state in 1848, they did so on the basis of a constitutional structure that was designed to allow for diversity of social, economic and political organization at the cantonal level. On the other hand, Swiss federalism is hardly competitive. As in Germany, cantons co-operate with each other, and above all the federal government co-operates with the cantons because it relies on their administration for the implementation of most policies. Finally, as in both the US and Germany, the emergence of national social security systems has shifted power and resources from the local and the state to the federal level.

Unsurprisingly, this peculiar institutional context has contributed to the shaping of social policy over the years. Overall, we can identify three different forces underlying the territorial dimension of the Swiss welfare state and working in different directions: first, a unifying and centralizing force related to the rise of the national welfare state in response to the imperatives of industrialization and societal modernization in the nineteenth and twentieth centuries; second, a unifying – but not centralizing – force arising from the co-operation of cantonal and local administrations with a fiscally and politically weak central government; and, third, a force of diversity and decentralization stemming from the combination of cantonal competencies with different resources, polities, politics and policies.

Introduction

Conventional wisdom strongly suggests that federalism is inimical to high levels of social spending. Two arguments are prominent in this context: a veto point thesis and a ‘competition of jurisdictions’ thesis. The veto point thesis is quite straightforward: federal systems have more veto points than unitary systems ceteris paribus. This increases the probability that groups opposed to welfare state expansion can exert some influence in the legislative process. Veto points would then give these groups the opportunity to block or substantially water down redistributive legislation. ‘Competition of jurisdiction’ arguments hold that welfare redistribution is limited in federal systems because those who would pay more than they would gain in a given jurisdiction (high income earners, ‘capital’) can credibly threaten to exit highly redistributive jurisdictions and join those that are less égaliste. At the same time, those who gain more than they would pay (e.g. low income earners) are attracted to regions with higher levels of redistribution and these would therefore develop into ‘welfare magnets’. Thus, a redistributional policy stance is self-defeating in a federal context.

Indeed, many econometric studies of the determinants of welfare state spending have found that federalism exerts a statistically significant, stable and negative influence on social spending. Prominent country cases are Switzerland and the United States, both strongly federalist countries and historically, prominent welfare ‘laggards’ (although since 1980 Switzerland has moved rapidly from laggard to leadership status).

Introduction

Canadians developed their version of the welfare state in the context of a vibrant federal state, with strong governments at both the federal and provincial level. Their experience highlights in fascinating ways the reciprocal interplay between federalism and social policy. In comparative context, the Canadian case underscores the need for more nuanced analysis than is found in much of the comparative literature of the welfare state, which is summarized in the introduction to this book. Attention normally focusses on simple dichotomies: federal versus non-federal, centralized versus decentralized, concentrated power versus multiple veto points. It is widely argued that federal, decentralized and/or fragmented decision-making inhibited the expansion of the welfare state in the twentieth century, but has slowed the processes of restructuring in the contemporary period. Such propositions do find echoes in Canada. For example, decentralization helped to slow the pace of development in the first half of the twentieth century.

The primary lessons to be drawn from the Canadian experience, however, emerge from the modern social programmes put in place in the second half of the twentieth century. Canada did not develop a single, integrated public philosophy of federalism in this period, and federal–provincial relations in social policy incorporated three distinct models, each with its own decision rules. At any point in time, governments were shaping or reshaping different programmes according to different rules and processes. Canada therefore constitutes a natural laboratory in which to analyze the implications of different models of federalism.