Inefficient RedistributionDARON ACEMOGLU Massachusetts Institute of TechnologyJAMES A. ROBINSON University of California, BerkeleyThere are many well-developed theories that explain why governments redistribute income, but very fewcan explain why this often is done in a socially inefficient form. In the theory we develop, comparedto efficient methods, inefficient redistribution makes it more attractive to stay in or enter a group thatreceives subsidies. When political institutions cannot credibly commit to future policy, and when the politicalinfluence of a group depends on its size, inefficient redistribution is a tool to sustain political power. Ourmodel may account for the choice of inefficient redistributive policies in agriculture, trade, and the labormarket. It also implies that when factors of production are less specific to a sector, inefficient redistributionmay be more prevalent.There are many normative and positive theoriesthat explain why governments redistribute in-come. For example, most positive theories ofpolitics typically involve a group that redistributesresources and income away from other groups to itself.Redistribution also may be undertaken for normativereasons; for example, the distribution of income andwelfare generated by market outcomes may be judgedunfair or undesirable by some ethical criterion. Welack a satisfactory understanding, however, of whyredistribution often takes an inefficient form.1A common example of income redistribution thattakes an inefficient form is farmers’ receiving pricesupports or input subsidies. Such policies distort rela-tive prices and discourage the reallocation of produc-tive resources away from agriculture and into othersectors, such as manufacturing, where they could bebetter used. Similarly, despite economists’ convictionthat free trade is typically efficient, domestic industriesare often protected by tariffs and quotas. A particularlyinteresting and relatively neglected example in whichthe form of redistribution appears to be inefficient islabor market regulation. Although firing costs and suchrestrictive labor practices as closed shops are wide-spread in most countries, they are thought to be highlyinefficient because they disrupt labor reallocation andcause unemployment.In all these cases, it is difficult to argue that theparticular form of the policy is correcting a marketfailure. Rather, it seems aimed simply at redistributingincome. For instance, no scholars appear to argue thatprice supports for farmers, which have the effect ofincreasing farm output, promote efficiency becausewithout them there would be too few resources inagriculture. This might be the case if farm outputgenerated positive externalities, but that seems implau-sible. Instead, it is widely agreed that price supports aresimply a way to raise farmers’ incomes. If this is correct,then they are Pareto inefficient in the sense that farmincomes could be maintained, and everyone else madebetter off, by a form of redistribution that did notinvolve resource misallocation. A simple transfer toraise the income of the farmers by as much as theinefficient policy yields would constitute an actualPareto improvement.We present a theory of inefficient redistribution thatbuilds on two basic assumptions. First, the politicalsystem cannot commit today to future policies, sincethey will be determined by whomever has politicalpower in the future. Second, at least over some range,political power increases with group size. Under theseconditions, inefficient redistribution may arise as a wayto expand or maintain the size of a group in order toguarantee its future political power.Consider the example of price support for agricul-ture. Imagine that farmers have sufficient politicalinfluence to induce the government to redistributeincome to them, and this can take the form of a simplemoney transfer to current farmers or a price subsidy.The latter is relatively inefficient as it potentially avoidsthe reallocation of resources to sectors in which theycan be used more productively.2Our key observation isthat the political equilibrium may nonetheless entailprice subsidies because that form of redistributionaffects the decision to remain in farming and encour-ages new agents to enter, in a way that lump-sumtransfers would not. Everything else equal, farmerswould not want to encourage newcomers, who increasecompetition both for transfers and in the marketplace,but if future political power and ability to extractDaron Acemoglu is Professor of Economics, MassachusettsInstitute of Technology, E52-371, Cambridge MA 02319([email protected]). James A. Robinson is Associate Professor ofPolitical Science, University of California, Berkeley, CA 94720([email protected]).We are grateful to three anonymous referees and the editor forvaluable recommendations. We also thank Jeff Frieden for hisencouragement and suggestions, and Pranab Bardhan, MiriamGolden, Dani Rodrik, Andrei Shleifer, and participants in a Depart-ment of Political Science seminar at Northwestern University, par-ticularly Michael Wallerstein.1Rodrik (1996, 204) provides a detailed discussion of trade policyand notes that the prevalence of inefficient redistribution is a majorpuzzle in need of an explanation: “Saying that trade policy existsbecause it serves to transfer income to favored groups is a bit likesaying Sir Edmund Hillary had to climb Mt. Everest because hewanted to get some fresh air. There was surely an easier way ofaccomplishing that objective!”2Notice, however, that in a dynamic world the expectation of futurelump-sum transfers also makes farming a more attractive professionand may inefficiently keep resources there. Nevertheless, other typesof redistribution keep more resources in farming and therefore aremore inefficient.American Political Science Review Vol. 95, No. 3 September 2001649further redistribution depends on the number of farm-ers, price subsidies may be preferred.In some sense, our analysis extends Becker’s (1985,338) insight that “a satisfactory analysis of the choice ofmethod must consider whether the influence functionitself depends on the methods used.” In our example,to ensure future transfers it is necessary for farmers toretain their political power, and they achieve this bychoosing a relatively inefficient method of redistribu-tion; it discourages farmers from changing sectors andencourages new agents to enter agriculture. The sameargument may apply to other