Read PDF Multi-Ethnic Coalitions in Africa (Cambridge Studies in Comparative Politics)

Free download. Book file PDF easily for everyone and every device. You can download and read online Multi-Ethnic Coalitions in Africa (Cambridge Studies in Comparative Politics) file PDF Book only if you are registered here. And also you can download or read online all Book PDF file that related with Multi-Ethnic Coalitions in Africa (Cambridge Studies in Comparative Politics) book. Happy reading Multi-Ethnic Coalitions in Africa (Cambridge Studies in Comparative Politics) Bookeveryone. Download file Free Book PDF Multi-Ethnic Coalitions in Africa (Cambridge Studies in Comparative Politics) at Complete PDF Library. This Book have some digital formats such us :paperbook, ebook, kindle, epub, fb2 and another formats. Here is The CompletePDF Book Library. It's free to register here to get Book file PDF Multi-Ethnic Coalitions in Africa (Cambridge Studies in Comparative Politics) Pocket Guide.

Visit Seller's Storefront. All books are shipped in New condition promptly, we are happy to accept returns up to 30 days from purchase. Orders usually ship within business days. Please contact the seller directly if you wish to return an order.


  • Suonatina - Score.
  • Recently Viewed?
  • Discrete Random Signal Processing and Filtering Primer with MATLAB (Electrical Engineering & Applied Signal Processing Series).

We are happy to accept returns upto 30 days from purchase. List this Seller's Books. Payment Methods accepted by seller. Home Leonardo R. Stock Image.

[PDF] Elite Parties, Poor Voters: How Social Services Win Votes in India - Semantic Scholar

New Condition: New Soft cover. Save for Later. About this Item Language: English. Brand new Book.

Why are politicians able to form electoral coalitions that bridge ethnic divisions in some countries and not others? This book answers this question by presenting a theory of pecuniary coalition building in multi-ethnic countries governed through patronage.

Focusing on sub-Saharan Africa, the book explains how the relative autonomy of business from state-controlled capital affects political bargaining among opposition politicians in particular. While incumbents form coalitions by using state resources to secure cross-ethnic endorsements, opposition politicians must rely on the private resources of business to do the same.

In This Article

This book combines cross-national analyses of African countries with in-depth case studies of Cameroon and Kenya to show that incumbents actively manipulate financial controls to prevent business from supporting their opposition. It demonstrates that opposition politicians are more likely to coalesce across ethnic cleavages once incumbents have lost their ability to blackmail the business sector through financial reprisals. About this title Synopsis: Incumbent leaders in African countries typically keep themselves in power, winning election after election, by using state resources to enlist the support of politicians from other ethnic groups.

Store Description Book Depository is an international bookseller. Thus to conclude on the basis of our data that ethnicity is more salient in Country A than Country B because a larger share of survey respondents in Country A ranked ethnicity first is not quite right. It is conceivable, though we think unlikely, that ethnicity might be more salient in absolute terms to people in Country B, even though a larger share of them rank some other category of identity as even more important than ethnicity. Finally, legitimate questions can be raised about the generalizability of our findings.

Although broadly representative of Africa as a whole, the 10 countries included in our study are not a substitute for a continent-wide sample. Our sample includes just one Francophone country Mali , no countries that have failed to introduce at least some democratic or market reforms over the last decade a precondition for an Afrobarometer survey ,.

Although strong social pressure against publicly expressing a preference for one's ethnic identity over one's national identity may account for these results, the close proximity of the survey to the election should have made this a particularly likely moment for respondents to have ranked their ethnic identities first. Notes: Political rights from Freedom House.

Months to election is the number of months to the nearest national election, with negative numbers signaling that the nearest election is in the past. Electoral margin is defined as the gap between the vote share of the winner and the runner-up in the most recent presidential election; if no presidential elections were held within five years e. Average for SSA is not meaningful as not all countries hold regular elections. Rates of urbanization are roughly comparable to the regional average.

Presidential elections appear to be similarly un competitive in our 10 sample countries as in Africa as a whole the average margin of victory in presidential contests is 32 and 34 percentage points, respectively , but citizens in our sample enjoy slightly more extensive political rights than.


  • INSANE TALES.
  • Leonardo Arriola | Egap.
  • Join Kobo & start eReading today;
  • African Politics Conference Group (APCG);

Our findings therefore must be interpreted with the caveat that they may not be entirely representative of Africa as a whole. This said, the fact that Bossuroy reports similar results to ours in a parallel study using comprehensive survey data from a quite different set of African countries lends confidence to the generalizability of our findings. Notes: Average values for each column weight each survey round equally, so respondents from countries with larger sample sizes are weighted less.

Table 2 reports the frequency distribution of responses to the "which specific group do you feel you belong to first and foremost? In addition, responses vary tremendously across countries and, perhaps even more strikingly, within countries over time—a finding consistent with theories of ethnic identification that stress contextual variability. The variation we. The raw unweighted share of respondents identifying in ethnic terms is The variation within countries over time is, of course, central to our identification strategy: our main interest is in ascertaining whether or what share of that variation can be explained by the proximity and competitiveness of the nearest presidential election.

Since the surveys are repeated cross-sections rather than panels of individuals, we cannot reject completely the possibility that sampling variation is behind some of the changes that we observe within countries across survey rounds. However, since the Afrobarometer employs the same sampling methodology in all survey rounds, and given the large, nationally representative sample of individuals included in each survey, we can be fairly certain that sampling variation is not primarily behind these shifts.

The robustness of our findings to dropping countries one at a time also allays fears that sampling variation in a single country might be driving our results. A crucial, and slightly different, question relates to the timing of the Afrobarometer surveys, which provides the source of variation in our key proximity variable. One concern is that surveys might have been deliberately scheduled close to exciting, hotly contested elections perhaps because they represent moments when political attitudes are particularly interesting and worth surveying but not close to less hotly contested elections.

Fortunately, there is little evidence that the timing of surveys was in any way related to electoral cycles, in part because the enormous logistical task of selecting enumeration sites and setting up field teams requires that preparations be made many months or even years in advance. Moreover, this timing would not account for the strong interaction effects between election proximity and competitiveness that we document below.

To the extent that survey timing was in any way endogenous to election timing, it was through what appears to have been a conscious decision by the Afrobarometer organizers after round 1 not to schedule surveys right near elections. While this would have been a uniform and thus unproblematic policy change, the worry is that such a change in the timing of surveys away from elections might have combined with a downward secular trend in the salience of ethnic identities.

This could produce a spurious correlation between electoral proximity and ethnic identity salience. We deal with this possible confounding story, as well as the possibility that changes in survey implementation14 across survey rounds might have generated changes in reported levels of ethnic identification, by including fixed effects in our regressions for each survey round 1, 1. Using this framework, we can examine empirically the extent to which a respondent's identity function Sict is systematically related to his or her observable characteristics and his or her country's political environment.

Combining observable and unobservable heterogeneity, we express the salience of ethnic identity for individual i in country c during survey round t as. Thus the hypothesized change in the strength of ethnic identity as elections draw nearer is allowed to depend on the competitiveness of those elections. Table 3 presents the results of four regressions of ethnic identification on our main independent variables: proximityct, competitivenessct, and proximityct x.

What, then, accounts for the variation we observe in the tendency of survey respondents to identify in ethnic terms? To answer this question, we model each individual respondent i living in country c taking part in survey round t as attaching a salience Sict to his or her ethnic identity recall that "salience" is operationalized as the likelihood that a respondent answers the "with which group do you identify first and foremost?

In the case of Botswana, this is because the country does not hold presidential contests; in the case of Zimbabwe, it is because presidential and parliamentary elections are not held concurrently and the most proximate national election to both of the Afrobarometer surveys we use was the parliamentary contest of June In analyses not shown , we tested for asymmetrical effects i.

Standard errors clustered at the country level in parentheses. All logit specifications include country fixed effects and trend and survey round controls; OLS country-level regression includes country fixed effects only. Clustering error terms at the country level should deal appropriately with the dependence of the key independent variables for individuals in the same country and the same survey round. The advantages of HLM lie in the explicit structure placed on the error and hence the ability to estimate directly the correlations in the error terms across "related" observations.

in-holod.ru/includes/72.php

Leonardo Arriola

However, HLM is not suitable for our purposes because it makes the strong assumption that model random effects are uncorrelated with the independent variables of interest, and the likelihood of this assumption being violated is precisely the motivation for our empirical approach. In addition, we are studying a discrete choice of primary identity from a menu of options, which makes the nonlinear, maximum-likelihood logit a natural modeling framework. Linear models like HLM are less suitable for discrete choice settings. The fact that all three versions of our main specification columns generate almost identical results speaks to the robustness of the relationship between ethnic identification and the political factors we are investigating.

The results reported in column 1 suggest that, on average, neither the proximity of the survey to a presidential election in months, absolute value nor the competitiveness of that election the margin of victory, in percentage points has any independent impact on the likelihood that a survey respondent will identify him- or herself in ethnic terms. Some caution must be taken, however, in interpreting the "electoral competitiveness" term given the relatively small degree of within-country variation we observe in our data on this variable see the "vote margin" column in Table 1.

Indeed, in four of our 10 countries, the same election serves as the most proximate contest to the two country surveys we use, so there is no variation on this term. Since all of the explanatory leverage in our specification comes from within-country comparisons, the coefficient estimates on the "competitiveness" variable are being produced by only a subset of our already small set of country cases. Given these considerations, we do. Rather, we focus on the interaction term between proximity and competitiveness, and it is the substantial cross-country variation in electoral competitiveness that allows us to estimate this effect.

When we add such an interaction term to our initial specification column 2 , we find that the coefficients on proximity and the interaction term are statistically significant. Taken together, the interpretation of the point estimates in column 2 is that the likelihood that a person will identify him- or herself in ethnic terms increases by 1. Thus a survey respondent asked within a month of a closely fought presidential election how she identifies herself would be nearly 22 percentage points s.

However, if the election was won in a landslide, her answer would be unaffected by the proximity of the election. These results are confirmed in column 3, which adds a host of individual-level controls for respondents' age, gender, occupation, education, media exposure, and urban or rural residence coefficients not shown , and column 4, which replicates the analysis at the country level.

The findings are nearly identical across all three specifications; more-. Bootstrapped standard errors for the electoral proximity and interaction terms were somewhat larger than those reported but still within conventionally accepted limits of statistical significance. For those interested, the bivariate correlation between competitiveness and ethnic salience conditional on country fixed effects is not statistically significant results not shown ; we thank a referee for this suggestion.

The main results are presented graphically in Figure 1, where the proximity to the closest country election is presented on the x-axis demeaned by country, which is equivalent to our country fixed-effects regression specification and the extent of ethnic identification is on the y-axis also demeaned by country. Two plots are presented: one pattern for relatively competitive elections cases where the electoral margin is less than the sample median of The relationships come through clearly: the plot is strongly negative for competitive elections meaning that ethnic identification falls sharply when surveys are conducted farther away in time from competitive elections but is nearly flat for landslide elections.

All of this is consistent with a story whereby the salience of ethnic identities is correlated with the electoral cycle, but only in settings where elections constitute meaningful contests for political power. Our main dependent variable based on the "with which group do you feel you belong to first and foremost? This makes it a. To model this process, we modify slightly the framework we introduced earlier. Thus respondents who attach high salience to their ethnic identity have large values of SEthn;c, ict; those who attach low salience to their gender identity have small values of SGenderi;ct; and so on.

When asked to report the group that they feel they belong to first and foremost, respondents choose the identity dimension j with the highest salience:. Using this framework, we can examine the extent to which the identity functions Sjict are systematically related to Xict and Zct. We can thus express the strength of social identity category j for individual i in country c during survey round t as. Note that the coefficients Pj and yj now have j subscripts as the independent variables could have different impacts across different social identity dimensions.

Two important aspects of this econometric specification bear mention. First, the multinomial logit model cannot estimate the level of the coefficients yj directly because, as noted above, the choices we observe only contain information about relative preferences. We therefore cannot distinguish absolute effects on the level of the identity strengths Sjict, only the degree to which explanatory variables make a respondent more or less likely to say that identity j is the one that they feel they belong to. Note, however, that this assumption is potentially problematic for country-level political variables zjc if there are omitted variables for example, unobserved country characteristics that are correlated with both political characteristics and individuals' identity choices.

This is particularly worrisome if the unobserved characteristics are correlated with either or both of our key independent variables, electoral competitiveness, andprox-imity to elections.

Crude Democracy Natural Resource Wealth and Political Regimes Cambridge Studies in Comparative Polit

In such a situation, the estimated y j coefficients will be biased. However, our use of country fixed effects allows us to, at least partly, address this problem for time-invariant country characteristics. The logit model identifies coefficients of the form yj — yk, or effects on identity j relative to a reference identity k. These coefficients give rise to an equivalent set of marginal effects, or impacts of the independent variables on the probabilities of choosing each of the five primary identity categories.

Second, the probabilities that particular social identities are chosen are not independent of one another. As the probability rises that a particular social identity is chosen, the probability of others being chosen necessarily falls since only one identity can be indicated in the survey. In particular, the marginal effects must mechanically sum to zero, because probabilities must always add to one. As we have stressed, a major advantage of our multinomial approach is that, if the salience of one dimension of social identification increases in response to a particular explanatory variable, we can simultaneously estimate which identity dimensions are becoming less salient.

That is, our method estimates substitution patterns among social identities in response to changes in the characteristics of individuals and in their political environment. Reading across the first row of Table 4 allows us to discover which identity dimensions lose salience as elections come closer. Effects for gender identity are also.

Additional analyses not shown also include controls for the characteristics of the interview and enumerator, with nearly identical results. Notes: Multinomial logit. All specifications include country fixed effects, individual-level covariates, and trend and survey round controls. For reasons described earlier, while we do not read too much into the lower-order coefficient estimates on the competitiveness variable, the interactive effect of competitiveness and electoral proximity is informative.

Thus far, we have focused on general trends among all survey respondents; our coefficient estimates can be interpreted as applying to the "average" person. However, both the theoretical literature on ethnicity and the cov-. We test four of these expectations in the bottom panel of Table 4. A first hypothesis derives from the classic literature on modernization. These second-generation theorists would therefore expect us to find the strongest ethnic attachments among those in the modern sector, since it is there that competition for scarce resources is most intense, and thus there that the incentives for people to use their ethnic identities as tools to acquire those resources should be most strong.

Farmers and fishermen are, on average, about 4 percentage points less likely than people in the modern sector to identify themselves first and foremost in terms of their language or tribe column 1. We also investigate the impact of group size on ethnic identification. Arguments developed by Posner and others lead to the expectation that respondents will be more likely to identify themselves in ethnic terms to the extent that they belong to ethnic groups that are large enough to constitute viable political coalitions.

An alternative perspective, however, would lead us to expect members of smaller ethnic groups to be more likely to identify in ethnic terms because fears of being dominated by members of other groups will heighten their ethnic identities Horowitz ; Rabushka and Shepsle We speculate that this may be because rural location is only a rough proxy for participation in nontra-ditional economic sectors.

Teachers, factory workers, government officials, and people with a range of educational attainment are well represented in both rural and urban areas in our data. We next investigate the impact of partisanship. Several scholars have identified a pattern in contemporary Africa whereby multiethnic ruling parties occupy the center of the political space, where they are surrounded by ethnically defined opposition parties seeking their overthrow e. To the extent that this is an accurate portrayal of African politics, we might expect to find respondents who report supporting the ruling party to have lower levels of ethnic identification than those who report supporting opposition parties.

As the results presented in Table 4 indicate, however, we find no support for this conjecture. Nor does support for the ruling party appear to be associated with a greater tendency toward any other type of social identification. Finally, media portrayals of African ethnic conflicts as being carried out by armies of unemployed young men might lead us to expect young men to have higher than average rates of ethnic identification.

However, the literature also generates expectations about the kinds of respondents that are likely to be more or less susceptible to ethnic mobilization—that is, more or less likely to identify themselves in ethnic terms in proximity to competitive elections. Indeed, given the focus of this article, the issue of whether young men, opposition party supporters, or members of large ethnic groups are more likely to identify themselves in ethnic terms generally is less relevant than whether members of such categories are more easily mobilized along ethnic lines in the.

The nested nature of language groups in Africa makes coding group size less than straightforward for some groups.

Supporting Information

Our main specification codes a respondent's ethnic group memberships based on the size ofthe group to which he or she reports belonging, based on data from Ethnologue n. An alternative specification codes group sizes based on the aggregated, politically relevant eth-nolinguistic groups within which the respondents' reported groups are nested as in Posner The results are robust to using either coding rule not shown.

Many respondents who are effectively unemployed indicate that they are traders, hawkers, or vendors. Although our estimation framework is ideally suited to addressing these questions, data limitations make it impossible for us to do so in a meaningful way. We are already constrained in our main analyses by the fact that we are exploiting variation over time within just 10 countries.


  1. A Snowman for Christmas?
  2. Goran Hyden.
  3. Article Metrics.
  4. The Apothecary Murders?
  5. Gregory Luebbert Book Award;
  6. Using triple interactions to test the joint impact of proximity and competitiveness on particular subpopulations—young men, ruling party supporters, etc. In analyses not shown we did run the triple interactions between proximity, competitiveness, and the indicator variables for each of these respondent categories, and none of these effects is statistically significant at traditional confidence levels.

    However, limited statistical precision means we are unable to rule out even quite substantial effects. The robust relationship we find between ethnic salience and exposure to political competition provides strong support for instrumental understandings of ethnicity. The fact that ethnic identities become systematically more important to people at the time that competitive elections are being held suggests that ethnicity plays a role in the struggle for political power.

    But exactly what role does ethnicity play? And for whom? One prominent answer in the African politics literature emphasizes the role of political elites. By this account, politicians find it advantageous to "play the ethnic card" as a means of mobilizing supporters to acquire or retain political power e. These efforts are also likely to be particularly vigorous when the elections are close and the advantage to be gained by mobilizing supporters will be greatest. Thus, to the extent that politicians' ethnic appeals make ethnicity more salient for voters, and to the extent that, once made salient, ethnic identities take some time to return to baseline levels, we would expect to find exactly the pattern that we do: stronger ethnic attachments during the periods preceding and following competitive national elections than at other times.

    For examples from other regions, see Chandra , Gagnon , Horowitz , Mendelberg , and. An alternative explanation for the link between political competition and ethnic identification focuses not on elites but on regular citizens—specifically, on their beliefs that jobs, favors, and public goods will be channeled disproportionately to coethnics of the person who is in a position to allocate them Barkan ; Posner ; Throup and Hornsby ; van de Walle ; Wantchekon Since elections are the moment when the people who will control the allocation ofresources are chosen, they are also the occasion when people should be most mindful of their ethnic identities and of the match between their own identity and that of the candidates vying for power.

    The association we find between ethnic identification and the electoral cycle is, again, consistent with this story. Unfortunately, our data do not permit us to adjudicate between these two explanations. To do so would require systematic information collected at different points in each country's electoral cycle about the kinds of ethnic appeals politicians make—data that the Afrobarometer surveys do not collect and that is difficult to gather systematically in a single country, let alone in When politicians in the run-up to Sierra Leone's presidential election promised that "if you help your kinsmen you will survive; we will give you jobs, opportunities and education" Manson , were they manipulating voters or simply playing to their expectations?

    When voters in recent elections in Kenya Gibson and Long , Malawi Posner , or South Africa Ferree overwhelmingly supported presidential candidates from their own ethnic or racial groups, were they responding to the candidates' ethnic appeals or simply channeling their votes to the politicians who they thought would best look out for their interests? The answer is almost certainly "both. Similarly, although most citizens do not need to be reminded that their ethnic connection with the election's winner is likely to affect the level of resources they will receive in the election's aftermath, politicians' ethnic appeals almost certainly reinforce such expectations.

    The result is an equilibrium in which expectations ofethnic favoritism by voters generate ethnic appeals by politicians which, in. The link between political competition and ethnic identification is characterized by a second sort of equilibrium as well. Rational politicians should target their ethnic appeals to the voters they believe will be most receptive to them. Thus if we can identify the kinds of voters that politicians should be seeking to mobilize, we should expect to find higher levels of ethnic identification among these voters than others.

    This was the intuition behind our test for higher levels of ethnic identification among members of large ethnic groups since these are the groups that constitute sufficiently large votingblocs to be able to affect the outcome of the election and among opposition party supporters since one of the surest ways to mobilize opposition to a ruling party in Africa is to claim that the party discriminates against members of ethnic groups that are underrepresented in its leadership.

    The fact that we find no differences between members ofthese groups and others with respect to their likelihood of identifying in ethnic terms suggests either that such targeting is not taking place which we believe is unlikely or that, consistent with a pattern long noted by scholars of ethnic politics e.

    Our central result is that exposure to political competition powerfully affects whether or not survey respondents identify themselves in ethnic terms. The finding—based on precisely the kind of cross-national data that has hitherto been lacking—provides strong confirmation for sit-uational understandings of ethnicity and for theories that link the salience of particular social identities to instru-.

    However, this is not a group that African politicians have traditionally sought to mobilize, ethnically or otherwise Bates Beyond their relevance for this academic literature, the article's results also have important implications for policy makers and researchers interested in elections and ethnicity. It might be tempting to interpret our findings as suggesting that, by heightening the salience of ethnic identities, the reintroduction of multiparty elections in Africa in the s—widely celebrated as a positive development— may have a conflict-inducing downside.

    Kenya's presidential contest, which triggered weeks of violence that left more than 1, people dead and , displaced International Crisis Group , would seem to provide strong support for this thesis. Yet it would be wrong to construe our results as endorsing this position. While we do find strong evidence that ethnic identities are heightened by exposure to political mobilization, our findings do not support the proposition that political competition accounts for the baseline levels of ethnic salience that make mobilizing ethnicity so politically useful in many African countries—indeed, our fixed-effect estimation strategy makes it impossible for us to test such a claim.

    Nor do our results suggest that the increasing competitiveness of African elections Diamond will necessarily instigate ethnic violence. Our findings suggest that countries with periodic competitive elections should experience fluctuations in ethnic salience that are correlated with their electoral cycle, not that they will exhibit higher levels of ethnic identification, on average, than countries without competitive elections.

    The relationships we uncover would be consistent with such a pattern, but establishing such a relationship would require a different research design than the one we adopt here. Yet the fact that elections make ethnicity even momentarily more salient does suggest the need for African governments to develop policies and institutional mechanisms that are capable of dealing with ethnic divisions.

    Policies and institutions such as those in place in Tanzania—a country known for its efforts at nation building through the promotion of Swahili as a national language, civic education, and institutional reforms like the abolition of chiefs, as described by Miguel — might serve as a model for how Kenya and other African countries might dampen destructive ethnic divisions. Acemoglu, Daron, James A. Robinson, and Simon Johnson. Barkan, Joel. Joel Smith and Lloyd Musolf. Bates, Robert H. Donald Rothchild and Victor A.

    Boulder, CO: Westview, Bossuroy, Thomas. Bratton, Michael, Robert Mattes, and E. New York: Cambridge University Press. Chandra, Kanchan. Why Do Ethnic Parties Succeed? Patronage and Ethnic Head Counts in India. Cheeseman, Nicholas, and Robert Ford. Collier, Paul. Diamond, Larry. Ferree, Karen.