Having finally turned the corner on a brutal, 11-day (and counting) cold, I feel up to getting back to my blogging routine. First up: a followup to last month’s post, “Why So Little Decentralization?”
To review, that post posed a puzzle (a problem for political scientists to ponder, you might say). The puzzle is this: developing countries are far more centralized than developed countries. That is so despite the fact that some developing countries are much larger and more diverse than developed countries, and many of them have now been democratic for quite some time. Furthermore, if decentralization were simply a relict of post-medieval state-building (some might venture that sort of claim about Switzerland, for instance), then the fact that developing countries have lower state capacity and a more recent independence than almost all developed countries deepens the puzzle.
I went through two explanations that do not actually explain the puzzle very well: shallow local talent pools and illiberalism. In particular, they cannot explain why developing countries are often very decentralized along some dimensions (allowing discrimination against goods and workers from other regions, linguistic and cultural rights, etc.), but not others (chiefly tax policy).
I think there are two explanations that actually work: secession prevention (in ethnic federations) and excessively personalist electoral systems (in nonethnic federations). In this post I’ll talk about secession prevention.
Some developing democracies are ethnoregionally diverse, that is, they contain minority ethnic homelands that could form the basis of independent states. Examples include India, Indonesia, Nepal, Pakistan, Thailand, Ghana, Moldova, the Philippines, and South Africa. Others are not very ethnoregionally diverse, like Mexico, Brazil, Argentina, Chile, Colombia, Costa Rica, South Korea, and Botswana. (Some of these have very small indigenous groups and/or ill-defined racial minorities, but historically neither type of group has been likely to make autonomy demands.) In highly ethnoregionally diverse societies, central governments will worry a great deal about preventing secession.
In general, smaller countries, diverse or not, are highly centralized. Even Thailand, at 66 million, is highly centralized. Apart from some hill tribes in the far north, its main ethnic minority is the Malay-Muslims in the south, from whom a long-running secessionist/irredentist “Sultanate of Pattani” insurgency has spawned. Moldova is tiny and is generally centralized but conceded tremendous autonomy to the truly minuscule Gagauz minority (about 120,000 people) after a rebellion that the failing Moldovan state could not vanquish. Smallish Ghana (24 m.), Costa Rica, and Botswana are highly centralized. Smallish Nepal (26 m.) is looking at a federal system, but is both highly diverse and highly mountainous, inhibiting communication among regions.
Other than Thailand, the states with more than 40 million population above all have some kind of autonomous regional government tier: India, Indonesia, Pakistan (though it briefly experimented with centralization prior to 1973), the Philippines, South Africa, South Korea, Mexico, Argentina, Colombia, and Brazil. But these countries sort into those with substantial ethnoregional diversity (the first five) and those without it (the last five).
Large, democratic countries with ethnoregional minorities have incentives to offer autonomy over cultural policies to those minorities. Thus, India, Indonesia, Pakistan, the Philippines, and South Africa all grant at least some of their regional governments a role in education; the Philippines differs from the rest in offering it to only one region, the Autonomous Region of Muslim Mindanao. Education is critical to cultural reproduction; it allows ethnic minorities to bring up their children in the local language. Furthermore, breaking the common market by allowing regional governments to discriminate against non-indigenes in employment, as Indonesia allows, helps prevent ethnic minorities from being “swamped” by ethnic strangers in their homeland, a common cause of violence. It may be economically inefficient, but there is a certain conflict reduction logic to the policy.
By contrast, none of these countries except India gives its regional governments substantial tax-raising powers. Why not? If regional governments were able to raise their own taxes, they could fund their own armed units and cultivate a nationalist clientele inaccessible to the central government. The central government would prefer to hold the key to government benefits and be able to withdraw funding for recalcitrant regional governments. For instance, as Yugoslavia democratized, republican leaders in Slovenia and Croatia used their powers to smuggle weapons and establish new agencies needed for an independent state (like customs agencies). Again, keeping taxation powers out of the hands of local governments is probably economically inefficient, because it prevents local voters from responsibly matching expenditure demands to revenue-raising responsibilities. Instead, regions will tend to overspend and go to the central government for help. Moreover, tax centralization prevents beneficial fiscal competition: regions’ keeping corruption and taxes low to attract investment.
So what’s up with India? India allows states to collect major taxes, and about 35% of all taxes paid in India are raised autonomously by state governments (local governments have no fiscal autonomy). But the Indian federal government has an ace in the hole: President’s Rule. Any time a state government gets unruly, the federal government imposes President’s Rule, allowing it to rule a state directly. It’s been used dozens of times since independence, often for nakedly political reasons. It has quickly been applied whenever a state government started to make noises about secession, as in Punjab between 1983 and 1992. Thus, India’s exception actually proves the rule: India doesn’t need to fiscally handicap its states, because it can use a “nuclear option” of taking over a state government, and has in fact been very willing to use that instrument.
There is also direct archival evidence that India’s framers were concerned about secession and designed federalism accordingly. Jawaharlal Nehru, India’s first prime minister (1947-1964) and more than any other individual, the architect of India’s constitution, presided over the Union Powers Committee of the Constituent Assembly drawing up a constitution for India after Pakistan’s secession. Its report stated, as a reason for backing away from a decentralized federalism:
Momentous changes have since occurred. Some parts of the country are seceding to form a separate state, and the plan put forward in the statement of the 16th May. . . is, in many essentials, no longer operative. In particular, we are not bound by the limitations in the scope of Union powers. . . Now that partition is a settled fact we are unanimously of the view that it would be injurious to the interests of the country to provide for a weak central authority which could be incapable of ensuring peace, of coordinating vital matters of common concern, and of speaking effectively for the whole country in the international sphere.
(Quoted in Rao and Singh, p. 46)
Even before partition, Nehru’s Congress Party favored a dominant fiscal role for the central government if the provinces were to enjoy the residual power (and President’s Rule had not yet been devised). In short, fears of empowering ethnic-minority regions with secessionist potential led the Indian framers to create a centralized, quasi-federal constitution.
By comparison, two of the ethnoregionally homogeneous countries I listed above — Argentina and Brazil — are highly fiscally decentralized. They can afford to be fiscally decentralized because there isn’t a secession threat. However, decentralization is “economically perverse” in other ways in those two countries, and that will be the topic of the third and final post in this series.
One thought on “Why So Little Decentralization? Part Two: Secession Prevention”
Reblogged this on Stormfields and commented: