Chapter Five: Autonomy and Secessionism

Chapter Six: Conclusion

Theory and Evidence

Chapter One set forth an ambitious program for this dissertation:

[to study] the demand for secession from diverse perspectives, including economic, political-institutional, and cultural-demographic accounts. The result is an empirically robust model of secessionist demands that can be used to predict when and where secessionist movements will arise and achieve electoral success – and what the implications of secessionist electoral success are for the balance of power between central governments (centralization and decentralization).

The purpose of this chapter is to evaluate the success of this dissertation in meeting these objectives.

The theoretical development and empirical analysis of the dissertation were divided into three stages. In the first stage, variables explaining the difference in average secessionist support across democratic regions were developed and tested (Chapters Two and Four). In the second stage, variables explaining year-to-year variation in secessionist support in those regions that have stable secessionist parties were developed and tested (Chapters Two and Five). In the third stage, the effects of secessionist electoral success on subsequent decentralization were examined (Chapter Five).

Important variables in the first stage include: regional language speakers, history of independence, proximity to linguistic homeland in another country, relative regional income, population, multi-party political system, and lack of road connections with the rest of the country. A region that scores favorably on these variables is likely to develop a secessionist party – and one that is particularly strong relative to other regions. Once a region possesses a secessionist party, different factors explain its year-to-year electoral success (the second stage): changes in relative regional unemployment, changes in a regional misery index, changes in relative regional income, the progress of globalization in terms of world exports, and policy changes or announcements or electoral results affecting the region’s autonomy status. Regions with stronger secessionist parties receive more autonomy than regions without such parties (the third stage).

The results on the cross-sectional analysis (the first stage) were largely as expected, though there were also some surprises. The theory and evidence could be summarized as follows.

Regional language speakers. The greater the number of speakers of a region-specific language, the higher is average secessionist vote. Speakers of regional languages see themselves as culturally and perhaps ethnically different from people in the rest of their country. The classical theory of nationalism holds that each “nation” or geographically concentrated ethnicity should have its own state. Classical nationalism therefore seems to play an important role in contemporary secessionist movements.

History of independence. Regions that have declared themselves or been recognized as independent at any point since 1648 have higher average secessionist vote. A history of independence can serve as an identifier of a separate nationality in lieu of or in addition to a regional language.

Proximity to linguistic homeland. Regions that are adjacent or proximate[1] to other countries in which the regional language is an official or majority language have lower average secessionist vote. The theory argued here is that this variable tracks irredentist potential, and that irredentist potential is negatively correlated with secessionist potential. If a region is a linguistic enclave abutting a country where its language is a majority language, there exists a possibility for people in both the region and the nearby country to claim the region for this country. The country in which the region currently resides is of course unlikely to recognize such a claim. Therefore, the two countries will likely undertake negotiations to prevent international conflict from developing over the issue. These negotiations would be directed at eliminating any grievances the linguistic enclave may have against its current country. Thus, Austria and Italy negotiated autonomous status for South Tyrol. In some cases, for historical reasons, such negotiations are unlikely. Belgium, for example, is divided between two major linguistic groups that abut other countries with the same language.[2] However, because Belgium fought both France and the Netherlands during the 19^th century, these countries have avoided involving themselves in Belgian affairs, and Walloons and Flemings both have been leery of incorporating themselves into either neighboring state. Furthermore, since there is no “Belgian language” and thus no central government representing the interests of an overwhelmingly dominant “Belgian nationality,” there is no possible domestic partner for international negotiations: both Flemings and Walloons see themselves as aggrieved minorities. For all these reasons, secessionism tends to be a live issue in Belgium, while irredentism is not – apart from a small group of Walloons.

Relative regional income. Regions with a higher ratio of gross regional product per capita to gross domestic product per capita (of the country of which it is a part) see higher average secessionist vote. In democracies, regions with a higher per capita income than the rest of the country tend to support the rest of the country with their tax money. Inter-regional redistribution flows from high-income regions to low-income regions. In Europe, the European Union’s Structural Funds reinforce this trend. Secession allows high-income regions the prospect of “cutting off the spigot,” of retaining their tax resources for their own use. Secession is unattractive to low-income regions because they wish to see the transfers from the rest of the country continue to flow.

Population. Regions with higher populations tend to have higher average secessionist vote, all else equal. This finding reflects the perception that larger regions are more viable as independent countries than smaller regions. While a large population does not “promote” secessionism, it does remove a psychological or economic barrier to seeking independence, if factors are otherwise favorable to this course. Put another way, a small population may dampen secessionism in a region where it might otherwise be strong.

Multi-party political system. Regions with higher “effective numbers of electoral parties” tend to have higher average secessionist vote. The reason for this result is presumably that most secessionist parties are small, and they suffer from the “wasted-vote syndrome” in political systems with few parties: voters do not want to cast votes for parties that do not have a chance of winning seats. In the empirical analysis, the substantive impact of this variable was small but statistically significant.

Lack of road connections. Regions that lack road connections with the rest of the country have higher average secessionist vote. This finding was an unexpected one, but it was both strong in substantive impact and robust to different specifications. This result may reflect a cultural difference that correlates with distance, a military or strategic advantage in being unconnected to the country from which the region may secede, or the fact that island regions develop their own largely self-sufficient economies and that the transition costs associated with independence would therefore be lower for these economies.

Some variables expected to have an impact did not. Higher levels of regional representation in the national legislature do not stimulate secessionism, greater levels of ideological difference between region and country do not correlate strongly with secessionist vote (perhaps because of the globalization considerations mentioned in Chapter 2), and the evidence that secessionist parties do better in regional than national elections is not quite strong enough to make a definitive conclusion. It was thought that regional elections afford secessionist parties a greater opportunity to form a government, and that voters trust secessionist parties to govern better on regional issues than countrywide ones. The coefficient on the regional elections variable is always positive, but it sometimes falls just short of statistical significance.

The findings on over-time determinants of secessionist vote are generally far less conclusive. This shortcoming is not surprising, since political scientists do not generally believe that they can predict how all parties of a specific ideological orientation will fare over time in all democracies. The limit of electoral prediction has generally been to forecast how incumbents of any party will fare on the basis of economic growth, length of incumbency, and other factors. Nevertheless, Chapter Five attempted to predict the vagaries of vote share for all major secessionist parties in Western democracies – without accounting for pre-election polling data, party strategy, and other “micro-political” factors. The findings can be summarized as follows.

Relative regional unemployment. In years when regional unemployment grows relative to countrywide unemployment, secessionist vote may increase. The result on this variable is significant only under the Generalized Least Squares regression format, which emphasizes year-to-year changes. If the positive correlation is meaningful, it probably captures the fact that when a region is doing notably poorly relative to the rest of the country, the region’s secessionist party can claim that the central government is discriminating against the region or, at least, failing to manage the country’s economy with appropriate sensitiveness to the trends of the regional economy.

Regional misery index. In years when regional unemployment rate plus national inflation rate is high, secessionist parties may do poorly. This result also appears only in the GLS regressions. In general, secessionist parties seem to do poorly when the economy is doing poorly in an absolute sense (rather than, relative to the rest of the country). When the economy is bad, voters may look to traditional left-right parties for immediate solutions.

Relative regional income. For the same reasons highlighted above, in years when relative regional income is high, secessionist parties do better. This result is more robust than the previous two. In addition, the fact that it is strongly confirmed in the cross-sectional regressions should give us confidence that the same dynamic is at work over time, just in a weaker or more diffuse form.

World exports. In years when world exports divided by world production are high, secessionist parties do better. In practice, this finding means that secessionist party vote has increased dramatically in the last decade. Globalization does not cause secessionism to arise where it previously did not exist; rather, globalization merely lowers a barrier to independence and thus aids secessionist parties where they already exist.

Policy changes and electoral results. Policy changes and electoral results viewed as unfavorable to the region tend to promote temporary but often quite large increases in secessionist support, according to case-study evidence. It is important that these policy changes originate externally to the region. “Internal” policy changes or failures, such as the failure of referendums on autonomy or independence, tend to trigger declines in secessionist support, because they highlight divisions in the regional community and suggest that the relevant issues are settled for some time. Despite all this, there is no quantitative evidence that increases in autonomy cause declines in secessionist support, at least at the next election.

Governments apparently believe that autonomy offers can forestall secessionism, however. Chapter Five finds strong evidence that regions with strong secessionist parties receive more autonomy than regions without such parties.

Thus, we do know much more now about where and when secessionist parties will succeed than we previously did. The findings of this dissertation have various implications both for the way future research is conducted and for our understanding of secessionist appeals and central-government responses.

Implications for Secessionist Politics

Chapters Four and Five used the empirical findings of the dissertation to venture some forecasts of future secessionist performance. The Sardinian nationalist movement has been weaker than expected, and an opportunity for secessionist growth appears to exist there. Likewise, the Frisian regionalist movement in the Netherlands seems primed to develop a small secessionist offshoot. Secessionism should do much better in the Balearic Isles and Valencia if the CiU were to begin contesting elections there, though the vote share for the independentist ERC and smaller radical parties would likely decline. The results of Chapter Five seem to indicate that in regions already possessing secessionist movements, secessionist vote will continue to increase as global economic integration proceeds.

This latter statement deserves qualification, however. Most secessionist parties are left-leaning and base part of their appeal on the idea that independence would afford their regions the opportunity to develop distinctive policies for economic development. However, as Chapter Two noted, standard political-economic theory suggests that as national economies are opened ever wider to international competition, governments will be less able to pursue significantly different economic policy approaches, particularly those approaches that reduce the return to capital in exchange for other social objectives. If fiscal competition does proceed this far, globalization will present a serious challenge to left-leaning secessionist movements. They have a choice of either moving to a basically liberal ideological orientation (and possibly losing much of their appeal) or holding fast to their socioeconomic ideology and moderating or eliminating their secessionist message (and losing the support of committed nationalists).

So far, the impact of globalization has been to reduce the efficiency barriers to independence and thus promote secessionist vote. If the “race to the bottom” claims made about globalization ever do come to fruition, however, the result may be a decline in secessionist vote in many Western regions, as these parties must come to grips with a cluster of policy proposals that no longer make sense together. This factor may have something to do with the recent decline of the Northern League into obscurity. The Northern League was a basically liberal secessionist movement in the mid-1990s but adopted a far-right and progressively non-secessionist ideology beginning in late 1998.[3] The move away from liberal economic ideology seems to have required a later move away from secessionism as well. Arguably, the League’s electoral support would have been much higher in 2000 and 2001 had the party not alienated its core support of “Padanian” nationalists and peripheral supporters in small business due to its anti-secessionist and anti-liberal trends, respectively. Bull and Gilbert argue, however, that Bossi has been rational, and that the entry of Italy into the European Monetary Union hampered the League’s attempts to use Italian deficit spending as an argument for Padanian secession from Italy and entry into EMU in their own right.[4] Thus, while the League’s decentralist and liberal ideology had dovetailed nicely with global trends and northern Italian sympathies in the early and mid 1990s, the League has become a victim of its own success. By forcing the other Italian parties to embrace federalism and fiscal responsibility, the League reduced the very grievances that sustained its existence.

So far globalization seems on balance to have benefitted secessionists. It has also possibly created incentives for moderate secessionists to become more radical, that is, to support independence outright. According to the BBC, the new leader of the moderate CiU, Artur Mas, is calling for “shared sovereignty” in Catalonia between the Spanish government and the Generalitat.[5] This position is certainly more radical than Jordi Pujol’s inward-looking nationalism, aimed at restoring and fortifying Catalan identity. The “shared sovereignty” formula follows the demands of the government of the Basque Country for a referendum on shared sovereignty, demands that the Spanish Prime Minister, Jose Maria Aznar, has opposed.[6] As of this writing, it is unclear whether the referendum will go ahead, and if so, whether the Spanish government will recognize the outcome.

In Flanders, Volksunie has moved from a non-secessionist position in the 1970s to a moderate secessionist position in the late 1980s (advocating a confederal solution for Belgium), and in 2001 split into moderate and radical parties. New-Flemish Alliance (N-VA), headed by Geert Bourgeois, “took over the actual party apparatus,” according to the Vlaams Volksbeweging press-clippings website.[7] N-VA favors independence for Flanders within a confederal Europe and holds to a centrist ideology. SPIRIT, the breakaway faction, retains the old confederal-Belgium policy but emphasizes a left-liberal ideology. It has hemorrhaged members to both the Flemish Liberals and Flemish Socialists since its founding, and reports from the region suggest that it is likely to disappear soon. According to the Economist Intelligence Unit, the Christian Democratic and Flemish Party (CD&V, formerly CVP: Christian People’s Party), traditionally the largest political party in Flanders, is now advocating the transformation of Belgium into a confederation.[8] Since Chapter Five showed that higher secessionist vote correlates with increases in regional autonomy, we should expect that increases in secessionist vote due to globalization will put additional pressure on central governments to decentralize.

Results on variables other than globalization also have implications for secessionist and central-government strategies. The results on GDPRATIO, for example, suggest that secessionist parties should emphasize fiscal drains from the region, while central governments should emphasize subsidies to the region. As Chapter Four noted, one perverse implication of this finding is that central governments should offer subsidies to potentially secessionist regions that do not promote long-run growth, as long-run growth could turn the region into a net taxpayer in the future. Nevertheless, the central government should not allow itself to be seen as following such a cynical strategy. Subsidies present secessionist parties with a catch-22. On the one hand, they would like to take credit for forcing the central government to take notice of regional needs, but on the other hand, they wish to minimize the extent to which the region is subsidized in order to make the case for independence or at least full fiscal sovereignty. This catch-22 displays itself particularly well in the case of Plaid Cymru, which stresses that subsidies to Wales are the result of its electoral successes, but that subsidies would not be necessary if Wales had control of its political and economic destiny. The net-subsidized status of Wales may be one factor in Plaid Cymru’s choice of a moderate, gradualist secessionist strategy that rejects “independence” in favor of “self-government.”

The result on political system (effective number of electoral parties) also has important implications for secessionist strategy. In particular, secessionist parties in first-past-the-post systems with few competitive political parties should try to seek the regional median voter on socio-political issues, while secessionist parties in more proportional, splintered systems may try to carve out a niche in the electorate by adopting an extreme ideology. The SNP, PC, and PQ are all examples of basically centrist (by the standards of their regions) secessionist parties, and all are located in first-past-the-post systems. The most ideologically extreme secessionist parties are the Vlaams Blok in Flanders (far right) and Herri Batasuna in Euskadi (Marxist-Leninist). Both of these parties have benefited by attracting voters and activists who are “soft” on the independence issue but in agreement with these parties on issues like immigration and proletarian revolution that other parties do not touch. Both parties participate in proportional systems. The Lega Nord is drifting toward the extreme right in an Italian system that is now only very modestly proportional. What allows them to do this is the system of electoral pacts in Italy, in which certain constituencies are reserved to the Lega as part of the terms of their pact with Forza Italia and Alleanza Nazionale. In the Faroe Islands, where regional elections are proportional (the Faroes also send two representatives to the Danish parliament, the top two vote-getters in the election winning seats), a parallel system has developed, with secessionist and anti-secessionist parties on both the right and left. Catalonia is the exception to the rule, where two parties dominate in a moderately proportional system. The moderate (both ideologically and on the sovereignty issue) CiU consistently wins regional elections, while the anti-secessionist Socialists win most elections to the Spanish parliament in Catalonia. However, parties that are more radical have been able to stake out small sections on the political spectrum, from the left-leaning, pro-independence ERC to the Communists, who are themselves not altogether unsympathetic to Catalan sovereignty. The appropriate conclusion for secessionist political strategy thus seems to be that proportionality allows for ideologically extreme strategies where they appeal to substantial proportions of the electorate but does not cause them to become viable. First-past-the-post systems force secessionist parties to seek the center on socio-political issues, but notably, not necessarily on the independence issue.

Cultural variables like regional language, history of independence, and irredentist potential do not have as significant implications for secessionist political strategy, as they simply create a general feeling of regional distinctiveness and make a secessionist movement possible. In a given election campaign, emphasizing linguistic and cultural issues generally does not work unless there have been recent, notorious cases of linguistic discrimination or an alarming downward trend in language use. The only cases in which a general strategy of stressing regional identity might promote secessionism over the long run are those in which a regional identity exists but seems not to have been politically activated yet. The archetypal case is Sardinia. The empirical findings in this dissertation suggest that Sardinian secessionists might benefit their cause by increasing awareness about the Sard language, through music festivals, classes, literary clubs, and so on. This strategy has worked to promote Breton nationalism in Brittany, after decades of (sometimes self-imposed) suppression.

Generally, it would seem that a platform of independence is a more viable option than a more moderate platform of autonomy or confederation when linguistic distinctiveness is very high, and irredentist potential does not exist. The archetypal cases here are the Faroe Islands, Quebec, and Puerto Rico, where regional languages are overwhelmingly dominant.[9] In the Faroes all secessionist parties are now in favor of full independence, in Quebec the independence (“sovereignty-association”) option was the only alternative to federalism until the recent rise of the ADQ, and in Puerto Rico the three main parties support statehood, status-quo autonomy, and independence. A large population and distance from the mainland should also promote more radical variants of secessionism, where linguistic distinctiveness is already present, since both variables increase secessionist vote. Secessionist parties will see less need to compromise or moderate their program where the social and economic conditions are highly favorable for independence.

The Chapter Five section on policy proposals and changes has perhaps the most relevance of any empirical work in this dissertation for central government strategies, as this area is one where central governments have apparently made serious mistakes in the past. The major lesson from this section is that central governments should follow through on autonomy plans once proposed (though there is little evidence that increases in autonomy actually reduce secessionist party vote at the next election), and should not allow other regions to have veto power over the status of the region to whom autonomy is offered. If the central government fails to follow through on its proposals, or if other regions veto them, the situation is ripe for a secessionist resurgence in the aggrieved region. In addition, little harm should ensue from a central government’s proposing a referendum on independence that it knows it can win. Secessionist vote declines after referendum failures, due both to a feeling that the issue is settled for the near future and to the divisiveness conjured up within a region during the referendum campaign. Therefore, if a central government wants to nip a secessionist movement in the bud, a good strategy seems to be to put on the ballot a referendum on complete independence that will certainly fail. The reason central governments have not tried this strategy in the past is perhaps that they do not wish to concede any obligation on their part to abide by the referendum’s results, thus implying a “right to secede.” This concern is probably now anachronistic in most countries, however, as a precedent of orderly, internationally recognized secession has now been set. Threatening to crush a peaceful, democratically authorized secession with military invasion is simply no longer an option in most democratic states.

Implications for Future Research

This dissertation should have implications for how future research on nationalism and political parties is conducted.

First, it is necessary to separate secessionism from other variants of “ethno-nationalism” (linguistic defense, regionalism, irredentism, and of course state nationalism). Many of the findings of this dissertation apply only to movements for substantial autonomy or independence. For example, relative regional affluence, size of population, and irredentist potential should not affect linguistic-defense and regionalist parties, and the former two should not affect irredentist movements. They can pursue their goals regardless of how their regions are situated on these measures, for these measures have to do with the costs and benefits of an independent state. Voters care about the costs and benefits of independence, and this concern manifests itself in secessionist vote shares. Almost all previous studies have lumped together regionalist, linguistic, irredentist, and secessionist parties; this error in sample selection has undoubtedly retarded the discovery of cross-regional patterns in secessionist vote.

Second, using aggregate variables in quantitative analyses of vote shares can yield real insights. In the past, most electoral studies have focused either on predicting incumbent re-election, regardless of party, or on uncovering the motivations of the individual voter. These studies have yielded some interesting findings. With regard to secessionist parties, a consistent finding is that voters with more risk-acceptant attitudes are more likely to vote for secession than voters with risk-averse attitudes. Consequently, groups that tend to risk-aversion are, on average, weak demographics for secessionist parties: particularly, women, the elderly, and top managers in very large corporations.[10] Still, predicting vote share across time and countries requires the use of aggregate variables.

Secessionist parties are perhaps unique in that a few economic factors seem to affect the year-to-year variation in their vote share in the same way in all countries. Even with secessionist parties, the findings from Chapter 5 are mostly tentative. Additional data available in the future may confirm the hypotheses of this dissertation more strongly. However, the cross-regional results were quite strong, and some of the results contradicted intuition or conventional wisdom. Using aggregate variables to predict vote shares for different kinds of parties across countries or regions should prove to be a useful strategy in the future.

Third, using regional data is a useful strategy in quantitative analysis. Using regions as data units rather than countries increases the number of observations and improves the chances for finding the true, underlying relationships among factors. Spurious correlations can usually be weeded out with appropriate controls. Sometimes, a country is not the sum of its parts, and using national rather than regional data is misleading. That is clearly the case for the subject of this dissertation: using country-level variables to predict countrywide secessionist vote would have missed all the regional factors going into secessionist electoral success. The more sophisticated analyses of electoral systems also are now using district-level data rather than country-level data. If a national legislature has a large number of political parties represented in its parliament, does this fact reflect a reasonably proportional electoral system with multiple social cleavages, or a multiplicity of regional party systems, each with a small number of parties? There is no way to answer this question using country-level data; as a result, electoral studies must focus on election districts to determine how the electoral system impacts political party formation and success.

Much research remains to be done on the phenomenon of secessionism itself. More comparative work needs to be done on secessionism in autocracies, which this dissertation omitted. Some of the causal mechanisms should operate differently in autocracies: for example, low-income regions may well be more likely than high-income regions to foster secessionism under autocratic rule – a reversal of the correlation found here. The relative military capacities of the central government and of the region should be important in autocracies.

More case studies can be done on how specific secessionist movements mobilize their supporters and reach out to the whole electorate, and how central governments typically respond to these movements under various conditions. Some of the speculation found in this dissertation (on growing ideological tensions within secessionist movements, on differential support for autonomy among central government parties – why Conservatives should be more uncompromising on Scottish and Welsh autonomy than Labour, for example) remains just speculation until key players can be interviewed.

More time needs to elapse, too, before we can reach definitive conclusions about the determinants of secessionist electoral performance from election to election. Chapter Five’s analyses yielded some results, but they were sensitive to regression specification. As more data come in, stronger conclusions will be possible.