Arrunada Seminar: Paul Dower – Centralized vs. Decentralized Allocation
10 January 2013 at 9:49 am Lasse 3 comments
| Paul Dower |
Centralized vs. Decentralized Allocation
In Benito Arruñada’s insightful new book, Institutional Foundations of Impersonal Exchange: Theory and Policy of Contractual Registries, the widespread failure of titling programs in developing countries is used as motivation for a greater appreciation of the role of contractual registries. In many developing countries, immovable assets, especially land, are initially and subsequently allocated using a centralized mechanism as opposed to a decentralized market mechanism implicitly assumed in the book.
The conflict between those holding property and those acquiring property is different under a centralized allocation mechanism. Sara Berry in Chiefs Know Their Boundaries, an interesting work on an agricultural region of Ghana, describes the political process involved in determining the complicated overlapping and competing property claims in a system where land is allocated by a centralized mechanism. Here, the relevant asset is not exactly land but community membership. This asset consists of various rights, one of which entails a kind of social insurance that functions through land allocated based on perceived need. The chief simultaneously serves as the contractual registry, performing public reallocation of rights when necessary, as well as the steward of the community members’ rights in rem, enforceable against all parties. Since need is imperfectly observable, this allocation mechanism suffers from a moral hazard problem, in which the acquiring party has private information putting the holding party at a disadvantage. In this setting, the registry is and can not be independent but it can aim to be impartial.
This example highlights the institutional specialization required for impersonal exchange, a point made well in the book, but it also points to several difficulties not apparent in the analysis. First, the judgment proof problem is more complicated. Power and social status can create a judgment proof problem that is independent or even negatively correlated with the standard one of not having enough wealth to compensate the victim of a violation of rights. The judgment proof problem can create problems for the voluntary registration of property claims. Second, the asset that is transferred or involved in transactions in a centralized system may not easily map into assets exchangeable in a decentralized system. Here, there is a parallel to the informational externality discussed in the book concerning transactions of rights in rem. The lack of institutional specialization leads to significant information costs if rights in rem are transferred. Third, since local legal orders are usually less specialized and serve multiple purposes under a centralized allocation mechanism, they may appear weaker than they actually are. On one hand, the apparent favoritism of a local may merely reflect the fact that an outsider does not have a legitimate claim to rights in rem because the local that transacted with the outsider did not possess rights in rem (even though, as shown above, rights in rem exist and can be transferred). On the other hand, due to the social insurance role of land allocation, the local property holder commonly has a superior claim to land in the abstract than what can be acquired at any moment in time by another local or an outsider. Thus, local legal orders can be in better positions to track the competing or overlapping claims than a public registry based on a state-backed legal order, even though the political process required to adjudicate competing claims under the local legal order restricts trade opportunities.
Paul Dower
Kinross Assistant Professor of Development Economics, New Economic School (NES). Research Economist, Center for Financial and Economic Research (CEFIR).
Entry filed under: - Lien -, Institutions, Law and Economics, New Institutional Economics, Theory of the Firm.
Organizations and Markets 
1.
Peter Klein | 14 January 2013 at 10:51 am
Paul, is this ultimately a statement about scalability? When the relevant region is small, these problems are handled via repeated interaction among a close-knit group, as in the cases described by Bob Ellickson. There the “centralized” solution (decisions by the chief) and the decentralized one (peer-to-peer negotiation) end up being the same. The question, then, is what are the limits to customary law. Can this be framed in terms of group characteristics — size, heterogeneity, etc. — a la Ostrom?
2.
arrunada | 15 January 2013 at 4:12 pm
Thanks a lot, Paul. Starting with Peter’s question, yes, I think there is indeed an issue of scalability in the sense that local institutions such as those described in Paul’s post, are suitable for local markets (or, better, for local economies, not necessarily grounded on market exchange). These local economies, included their limited local markets, are mostly personal: participants and even traders know or have the possibility of knowing each other at low cost. Impersonal exchange works in a different scale.
Paul’s post poses many other fascinating issues. In particular, I am somewhat puzzled by the suggestion to apply the framework in the book to a centralized land allocation mechanism. In particular, doubts emerge on the in rem nature of rights and the meaning of independence in the contexts of centralized land reallocation. My difficulties come from the fact that the book focuses on the institutions needed to make property compatible with impersonal markets, Paul’s “decentralized market mechanism”. Most of the book therefore assumes the existence of private individual property, which implies an allocation of property rights to individuals. Reallocation on the basis of need or power are implicitly assumed away.
In a sense, centralized reallocation of land is a characteristic institution of what could be labeled as a “redistributive” economic system. Even the limited exchange it allows (with chiefs potentially acting as registries) is likely to be mostly local, for the reasons Paul explains. I am inclined to think that exploiting the specialization advantages available in today’s economies often requires reaching further—i.e., contracting with non-locals, with strangers, impersonally. The challenge is to make such impersonal contracting possible without expropriating customary rights. The book explores some specific palliatives, such as the effective interface provided by community responsibility, which has been effective at least in some credit contracts. However, these are palliatives. The main solution is the development of legally reliable and long-term sustainable registries, two requirements that are often absent in titling projects. Consequently, such projects do not achieve their goal of making impersonal transactions possible; they pose, however, substantial expropriation hazards.
A last point. Using land reallocation to provide social insurance seems hardly compatible with land and mortgage markets. Let me suggest that in developed market economies we have observed a centuries-long development of increasingly impersonal markets—moving away from redistribution systems. Interestingly, this development has occurred in parallel with the creation of new and wider institutions for social insurance. This hints that our goal in less-developed economies might lie in developing institutional specialization that separates in different institutions the provision of resource allocation and social insurance.
3.
Paul Dower | 15 January 2013 at 7:08 pm
Hi Peter, Thank you for your comment. The scalability issue is certainly related to the issues I was discussing in my post. However, I disagree that my post is ultimately about scalability. I can hardly say it better than what is found in Benito’s response to my post when he speaks about institutional specialization. But let me respond in more detail to fully express my opinion.
First, I would argue that most resource allocation systems make use of both centralized and decentralized mechanisms. But let’s assume that we understand whether a particular resource allocation system is better characterized by one or the other.
Second, centralized mechanisms, whether traditional institutions or modern ones, can accommodate impersonal exchange and rights in rem. The Ghana example is not the best one to illustrate these points but even in this case one can observe both rights in rem and impersonal (or much less personal) exchange. Rights in rem exist for the stool. The rights are enforceable against all and are alienable (although, very much in contrast to the decentralized allocation mechanism envisioned in Benito’s book, these are centrally concentrated with a single entity, for which the chief is the steward). Impersonal exchange of certain rights does occur. An active land rental market supports the entrance of “outsiders” to the local economy. Even mortgage markets exist.
Third, in the case of developing countries, where multiple markets are undeveloped, a centralized mechanism of land allocation can be less specialized than a decentralized land market. In the Ghana example, land is allocated not merely based on productive efficiency but also according to need, a kind of social insurance. So the question of whether a centralized land allocation is the same as a decentralized one, and how well customary law can scale up, depends on the presence of other markets and not just peer-to-peer transactions of land rights.
Fourth, and now I begin to more closely address your comment, in most cases, the centralized solution is not the same as the decentralized one for at least two reasons. First, given the third point above, the two mechanisms solve different problems. Second, even if they solve the same problem, power affects allocation differently under the two mechanisms. Under the decentralized system, an exchange would occur if there were surplus to be had from the transaction. Power then would only affect the distribution of this surplus. Under the centralized mechanism, power affects which transactions occur as well as the distribution of the surplus of transactions that do happen. To see this in the Ghana example, some legitimate claims for the redistribution of rights will not occur simply because they might fall across social status lines.
Finally, I would add that, for centralized mechanisms, often the main issue is not one of scale but one of information and incentives due to moral hazard. (By problem of scale, I restrict attention to the ability to enforce as the number of group members increase.) Information certainly has a scale component as Benito has already discussed in his response to my post but the essential problem is not one of enforcement. A hierarchy of chiefs could be organized to discipline/reward local chiefs who make the mechanism work well at the local level, similar to the China case. The information of whether or not a particular household is in need (relative to other households) is not easily extracted or observed (even at the local level). Another example that better illustrates the information vs. scale problems under a centralized mechanism is the collectivization of agricultural production in the early Soviet era. Collectivization is easy to scale up. Nearly all of the Soviet Union was collectivized in a year’s time. (Of course, those who resisted were deported or executed). But information/incentives was the actual constraint on the system. The imposition of too harsh output requirements by the central authorities generated widespread famine.