Secessionism and the Quality of Government: Evidence from a Sample of OECD Countries

Abstract

In this article we test the hypothesis that secessionism reduces government quality because secessionist threats elicit a response from central governments concerned with the territorial integrity of the state and this, in turn, channels attention and resources away from necessary governance reforms. We consider the link between secessionism and government quality based on an original data set that reflects the electoral success of secessionist parties in national elections. Our empirical results, drawn from a sample of 22 OECD countries over the period from 1980 to 2007, support the expectation that secessionism will tend to reduce the quality of government even after controlling for the influence of potentially confounding variables and the possibility that government quality may itself affect the electoral fate of secessionist parties.

Extracto

En este artículo comprobamos la hipótesis de que el secesionismo reduce la calidad gubernamental porque la amenaza de los separatistas motiva una respuesta de los Gobiernos centrales que se preocupan de la integridad territorial del Estado, lo que desvía la atención y los recursos que normalmente se dedicarían a las reformas necesarias del Gobierno. Consideramos el vínculo entre el secesionismo y la calidad gubernamental basándonos en un grupo original de datos que reflejan el éxito electoral de los partidos separatistas en las elecciones nacionales. Nuestros resultados empíricos, a partir de una muestra de 22 países de la OCDE durante el periodo de 1980 a 2007, confirman las expectativas de que el secesionismo contribuye a reducir la calidad del Gobierno, incluso después de tener en cuenta la influencia de posibles variables relevantes y la posibilidad de que también la calidad del Gobierno pueda afectar al éxito electoral de los partidos separatistas.

摘要

我们在本文中测试以下假说：来自分离主义者的威胁，会引发考量国家领土完整性的中央政府作出回应，并随之将关注与资源从必要的治理改革中挪开，因此分离主义会降低治理素质。我们根据反映分离主义政党在全国性选举中胜选的原始数据集，考量分离主义和治理素质之间的连结。我们根据採样自1980年至2007年的二十二个OECD国家所获得的经验结果，支持分离主义将会降低治理素质的预期，即便在控制了可能的混淆变项之影响，以及治理素质本身便影响了分离主义政党选举结果的可能性之后，该研究结果仍为真。

Résumé

Ce présent article cherche à tester l'hypothèse selon laquelle le sécessionnisme réduit l'efficacité gouvernementale parce que les menaces sécessionnistes provoquent une réponse de la part des administrations centrales chargées de l'intégrité territoriale de l’État et cela, à son tour, détourne l'attention et les ressources des réformes nécessaires en matière de gouvernance. On considère le lien entre le sécessionnisme et l'efficacité gouvernementale à partir des données originales qui reflètent la réussite électorale des partis sécessionnistes au moment des élections législatives. Les résultats empiriques, tirés d'un échantillon de 22 pays de l'OCDE pour la période allant de 1980 à 2007, confirment la prévision que le sécessionnisme aura tendance à réduire l'efficacité gouvernementale même après avoir tenu compte de l'influence des variables qui risquent de prêter à la confusion et la possibilité que l'efficacité gouvernementale puisse influer elle-même sur le sort électoral des partis sécessionnistes.

Keywords:

• Secessionism
• governance
• crowding-out
• panel data
• OECD

JEL Classification:

• D72
• D74
• H77

Acknowledgements

Andreas Kyriacou would like to acknowledge financial support from the Institut d'Estudis Autonòmics (Generalitat de Catalunya) and both authors would like to thank two anonymous referees for their helpful comments. The authors dedicate this article to the memory of Santi Pons Hita (1969–2014).

SUPPLEMENTARY DATA

The Supplementary data of this article can be view on doi.10.1080/21622671.2014.978883.

Notes

1. Our chosen measure may underestimate the strength of secessionist movements based in regions which are smaller compared to the country as a whole. It means that the strength of secessionism in Catalonia or the Basque Country may appear weaker than, say, that in Flanders. Thus, the cross-section differences in the electoral success of secessionist movements at the national level are imperfect measures of the cross-section variation in the strength of secessionist movements. We thank an anonymous reviewer for pointing this out to us.

2. See the supplementary material for the sample of countries and election years, the list of secessionist parties, and for the summary statistics and sources of all the variables employed in the article.

3. In order to maximize our sample, we maintain the secessionist party vote share of the previous period in cases where national elections did not occur during a four-year interval. Our substantive empirical findings remain unchanged if, alternatively, we treat the absence of an election as a missing value.

4. Applying fixed effects would help control for unobserved time-constant variables thereby potentially reducing omitted variable bias. However, fixed-effects estimates rely exclusively on the within variation of the key variables: in the case of our main government quality indicator, the within and between standard deviations are, respectively, 0.2794 and 0.6239.

5. Note that we transform our vote variable to Log (3 + Secessionist Vote) because the logarithmic transformation helps smooth out the skewness of the original variable and adding three ensures that the lowest value of the variable is at least one (see also Sorens, 2004, 2005).

6. Thus, fiscal and political decentralization may provide a platform for the mobilization of secessionist parties at the same time as they can diffuse secessionist demands by allowing distinct communities to decide on issues over which their preferences diverge from the majority. Recall also that the presence of secessionist parties may have an impact on the degree of decentralization (Heller, 2002).

7. Specifically, we employ two-stage feasible generalized least squares (TS-FGLS) and two-stage least squares with panel corrected standard errors (TS with PCSE). The suitability of employing lagged values of secessionist vote as instruments, is supported by the significant variability of our measure of secessionism over time as attested by a within standard deviation of 2.1419 and thus a coefficient of variation of 1.1956.

8. When we drop fiscal decentralization from the regressions the negative point estimate of secessionism increases markedly. This is consistent with the argument that secessionism can contribute towards fiscal decentralization and vice versa.

9. Murray (2006) warns that more distant lags may also be weaker instruments. The first-stage regressions, however, confirm that all the lags employed are strong instruments since the F-statistic obtained is always well above the critical value of 10 (Staiger and Stock, 1997).

10. Nothing changes if instead we control for economic globalization by way of a standard control variable which captures the degree of openness of a country's economy to international trade.