«in Designing Democratic Institutions (I. Shapiro & S. Macedo eds.), New York : NYU Press (Nomos XLII), 2000, pp. 296-320. It is a pleasure to comment ...»
POWER-SHARING VERSUS BORDER-CROSSING
IN ETHNICALLY DIVIDED SOCIETIES
Philippe Van Parijs
Université catholique de Louvain
Chaire Hoover d'éthique économique et sociale
Program in Ethics, Politics and Economics
in Designing Democratic Institutions (I. Shapiro & S. Macedo eds.),
New York : NYU Press (Nomos XLII), 2000, pp. 296-320.
It is a pleasure to comment on such an instructive and gloomy paper.1 Its instructiveness was particularly pleasurable, because it helped me see in a completely new light whatever I knew about the subject, not, as it happens, by virtue of any expertise I might possess in political theory, but rather by virtue of having lived for most of my life in what can plausibly be characterized (see below) as a severely divided society: Belgium. Less predictably, the paper's gloominess too was a source of pleasure as I prepared this comment, not at all because I enjoy learning that things go wrong, let alone understanding that they are bound to go wrong, but – quite the contrary – because the little I knew and understood about the subject implied, I thought, that I had some good news for the author. For his paper's central message I understood as follows: while we can get a pretty definite image of the coherent constitutional package needed by a severely divided multiethnic society, there are deep-seated reasons that such societies will adopt instead incoherent hybrids, which will do them no good. The good news will take the form of an argument to the effect that this grim message needs to be drastically qualified. Unsurprisingly (coming from a philosopher), it will rest on two small exercises in conceptual clarification, the crucial relevance of which will be illustrated by my reading of Belgium's constitutional development and debate.
1. WHAT IS A SEVERELY DIVIDED SOCIETY ?Red spots and red spheres.
First of all, what is it, in Donald Horowitz's view, that makes a poly-ethnic society qualify as severely divided? By definition, its being prone to (acute, violent) conflict between ethnic groups. Let us take for granted that the notion of an "ethnic group" is clear enough and concentrate on the concept of "conflict-proneness". Conflict-proneness is clearly a dispositional property of the society concerned. But for our purposes, the term "society" is crucially ambiguous. Do we mean "society" in a comprehensive sense that encompasses a country's current constitutional arrangements? Or do we mean it in a lean sense, which counterfactually strips a country of these arrangements? In either case, a society's characterization includes the specifics of its territory and its economy, its ethnic features, including their geographical and social distribution, the overall level and distribution of income and wealth, etc. But unlike the first interpretation, the second one excludes "constitutional design", understood roughly and pretty narrowly as those rules that directly organize the distribution of political power? The comprehensive interpretation makes the notion of a severely divided society fairly simple, while the counterfactual definition makes it unavoidably tricky. Nonetheless, choosing the comprehensive interpretation would be most unwise, for present purposes. It would soon prove a recipe for depression, as it would turn into an oxymoron, not constitutional design as such, but any successful constitutional design for a severely divided society: the very success of the design disqualifies the society as a severely divided one.2 Therefore, unless one takes some perverse pleasure in pursuing the logically impossible, there is no sensible way out of some variant of the counterfactual definition.
But what does it mean to
counterfactually from a country's constitutional design? Does a poly-ethnic society count as severely divided if and only if it would be torn by acute ethnic conflict if it had no constitutional design at all, or perhaps if and only if there exists at least one (sufficiently absurd) constitutional design under which the society would be prone to acute ethnic conflict? Under such characterizations, any poly-ethnic society – indeed, presumably, any society under a sufficiently broad definition of an ethnic group – would count as severely divided. On the other hand, if a society were severely divided only if it was prone to acute conflict whatever its political institutions, we would be back to making successful constitutional design an oxymoron. The appropriate definition must obviously lie somewhere in between. Here is one way of making it precise.
-2Consider a particular society at a particular time, as characterized by the current values of its non-constitutional parameters (in a sense that matches the definition of constitutional design adopted above), and think of the set of all logically possible constitutional arrangements for this society as a multidimensional hyperspace, each point in which represents such an arrangement. To make this more concrete, think of this space as a sphere, with each constitutional arrangement represented by a small spot within this sphere, and specified by the values taken by three continuous variables – for example, what percentage of the total vote is required for representation in the Parliament, how much veto power there is for ethnic minorities, and how strong the government is with respect to the Parliament. Next, colour in red any spot that represents an arrangement under which acute conflict is likely, while leaving in white any point that represents an arrangement under which acute conflict is most unlikely, and colour the rest in shades of pink. Under the comprehensive interpretation of what "society" means in that expression, a severely divided society would be one that happens to be in a red area of the sphere. Under the absurdly broad version of the counterfactual interpretation, it would be one whose sphere has at least one red spot: under a sufficiently broad conception of imaginable arrangements, any society, however safely lodged in the middle of a large white area, is severely divided in this sense. Under the self-defeatingly narrow version of the counterfactual interpretation, on the other hand, a severely divided society would be one whose sphere is completely red: no conceivable institutional arrangement could alleviate its conflict-proneness and only the delicate, often painful surgery of secession may enable the red to recede, as the one sphere is turned into two or more.
Finally, under the intermediate counterfactual interpretation I propose to adopt, a severely divided society is defined as one whose sphere has a large red area: it is conflict-prone under a large proportion of the constitutional arrangements. The redder the sphere, the more severely divided the society: like fragility or vulnerability, severe division is a dispositional property that admits of degrees. For some countries, the red spots may be so few that the sphere looks white. The desperate cases are those in which red is all over and deep down. Constitutional engineering for deeply divided societies is concerned with the intermediate case, in which there is a serious risk of being in the red area, but also a serious chance of sticking to the white one.
On the background of this conceptual clarification, I can now try to express my first bit of good news. In the dispositional interpretation for which I have argued above, there are far more severely divided societies than is revealed by overt conflict. The United States is hardly less deeply divided than South Africa, or Holland than Ulster. It just so happens that some countries have chosen or stumbled upon institutions that have kept them safely in the white
-3area. Compare the Netherlands and Northern Ireland, for example. Both were carved out of a larger territory (the Spanish Lower Countries, British Ireland) in which Catholics were an overwhelming majority, to form a territory in which the Protestants came to form roughly two-thirds, and the Catholics roughly one third, of the remaining total. By the beginning of the 20th century, both had a history of pretty ruthless domination by the Protestant majority and of anti-Catholic discrimination. But in 1917, the Netherlands adopted a Pacification settlement that introduced proportional representation, protected both Protestant and Catholic school systems and ended discrimination against Catholics in access to public sector positions3. In Ulster, instead, no such pacification deal was struck. Discrimination and domination continued, at least partly as a direct effect of the political institutions.
Proportional representation (in the form of Single Transferable Vote) was introduced by Lloyd George in 1920 and kept in place in the Republic of Ireland, where the Protestant minority soon dissolved, politically speaking, into a number of Catholic-majority parties. But it was repealed in Ulster in 1929 by the Protestant prime minister James Craig, precisely in order to hinder transconfessional parties4. The good news, for Donald Horowitz and his profession, which is illustrated by this contrast is of course not, as such, that there are more severely divided societies than they think, but that constitutional design (whether deliberate or not) can be so successful in some societies that one loses sight of the fact that they are just as severely divided as others in which conflict rages. Once severe division is interpreted, as it must, as a reddish sphere (of potentialities) rather than a reddish spot (in which one happens to find oneself), constitutional engineering holds great promise.
The goodness of this news should not be overstated, however, and the illustration I just gave is not meant to suggest that there are quick fixes. The job is promising but it is not easy.
The constitutional engineer's first task obviously consists in locating with some precision the red and white areas, whose sizes and shapes will vary a great deal from one society to another. When naively advocating the import of a ready-made package of rules that has proven its value through years, the clumsy Western do-gooders stigmatized by Horowitz in his contribution are simply oblivious to this fact: a point safely located in an immaculate area of the sphere associated with one country — for example, a presidential system with an Assembly elected by first-past-the-post — may be deep inside a dark red portion of the sphere associated with another, in which the ethnic set up is crucially different. The fact that the red patterns vary from one sphere to another does not mean that countries cannot learn
-4from one another. Quite the contrary: there is a lot to be learned from other countries' successes and failures, providing one does not make conflict-proneness an attribute of isolated constitutional devices, nor even of whole constitutional frameworks, but of a combination of a system of devices and the background non-constitutional conditions. Even though no two countries are anything like identical along these dimensions, insight into the mechanisms that underlie conflict-proneness and conflict-inhibition in one country can help guide choices in another. This is exactly what is at work when Horowitz ventures to say, for example, to South Africa: "Don't go there, it's red. Go there, you'll be safe."
The job does not stop at identifying the contours of the red and white areas, however, for the wisest recommendation is not always that one should move to the nearest white spot.
Often reaching the red area will require moving along two or more dimensions at once. If one moves along one of these dimensions and gets stuck, one may end up in a darker red area than the one one was trying to steer away from. When making recommendations, one should therefore anticipate the possibility that one may be able to go only part of the way. One must also try to guess what the winds and slopes will be, driving the reform further or pushing it back to where it started.
All this seems hard enough. But there are more sophisticated tasks still.5 For the contours of the red area are not fixed. Demographic or economic changes, for example, may upset the ability of current institutions to keep conflict-proneness under check, and constitutional engineering should anticipate such shifts of the red area and design institutions accordingly.6 As a very simple illustration, take a country, such as Belgium, in which the constitution can be changed only with a two-thirds majority. If the majority ethnic group represents 60 percent of the population (which is currently the case in Belgium), this rule protects the minority group against a constitutional change unilaterally imposed by the majority. But if demographic trends lead to the majority ethnic group forming more than twothirds of the population, then the current arrangements, without undergoing any change themselves, may suddenly find themselves in the turbulent red area.7 Or take the following, slightly more complex illustration, also taken from the history of Belgium. Throughout the 19th century, Belgium was marked by a sharp contrast between its mainly rural North (Flanders) and its far more industrialized South (Wallonia), with the result that, from 1884, Flanders sends 100 percent of Catholics to Parliament, and Wallonia a majority of liberals. In 1893, the country moves from highly restricted male suffrage (only taxpayers vote) to universal male suffrage with plural voting (one additional vote for married taxpayers, one or two additional votes for the educated), using a plurality type of electoral
-5system with small multi-member constituencies and a double ballot. As a result, the newly created socialist party obtains representation in Parliament, where Liberals and Socialists together win 40 of the 62 seats, while the Catholic party wins all 90 seats in Flanders and Brussels at the 1994 national election.8 Obviously, the Catholics can retain power with a comfortable, overwhelmingly Flemish parliamentary majority. The government, which had no Wallone member at all in the 1880s, will have no more than one in the 1890s9. As the population of Flanders kept growing faster than that of Wallonia, while Wallonia remained far more industrial, there was no prospect of a change in the underlying situation.