Social Limits to Redistribution

In many countries, equality of political rights, including the right to vote over taxes and transfers, goes along with substantial inequality of economic conditions. This observation lies at the root of one fundamental issue in political economy, one about the factors that limit the extent of redistribution in democracies. The conventional economic wisdom on this issue can roughly be summarized by two arguments. Firstly, redistributive taxation may be limited by various kinds of incentive costs of taxation as, e.g., in James A. Mirrlees (1971), Kevin W.S. Roberts (1977), Dennis N. Epple and Thomas Romer (1991), and Thomas Piketty (1995). Secondly, redistribution may be dampened by the lobbying activity of high-income groups and by various imperfections inherent to the political process that determines the tax system as, e.g., in Gary S. Becker (1983) and John E. Roemer (1995). While both arguments offer illuminating insights into the phenomenon of redistribution, they are also based on a very crude description of human behavior, one which ignores social motivations of action, both in the economic and the political sphere. The present paper develops an explanation that takes such motivations into account. Specifically, it shows that limits to redistribution may arise when economic inequality has an informational value for social decisions. Our approach is based upon the following observation: many goods and decisions that heavily affect an individual’s quality of life are not allocated or made through markets but through social interactions (Thomas Scitovsky, 1976). For instance, although people have strong preferences over how they are treated by others and over whom they mate with, these things are not the object of market transactions. A number of sources of satisfaction, like conversation, dinners and parties, playing with others, and being observed and admired, are nonmarket goods, for which a keen social competition between individuals often develops. As shown by Harold Cole et al. (1992), the existence of social—rather than market—competition for some goods can endogenously generate a concern for relative position in the income distribution. When information about relative income is private, individuals become interested in the observable consumption differentials between them and their social competitors. Since consumption differentials are influenced by redistributive taxation, the political attitudes of people toward redistribution will be shaped by its expected impact on social competition. In some identifiable environments, this concern for social success may be the crucial factor that limits the extent of redistribution desired by a majority of voters. Specifically, while the middle class may obtain economic benefits from a large amount of redistributive taxation, it may oppose an equalization of living standards since this would harm its social success. The fear of losing social status in favor of the poor induces the middle class to enter a political alliance with the rich which supports conservative taxation programs. In the sociological literature, scholars of voting behavior have often suggested that the need for social recognition plays a crucial role in shaping political attitudes. Seymour M. Lipset (1967) pointed out that white-collar workers tend to be socially valued similarly as those higher in the system and, although their income may be only slightly larger than that of manual workers, white collars are much more likely to * Corneo: Department of Economics, University of Osnabruck, Rolandstrasse 8, D-49069 Osnabruck, Germany, CESifo, Munich, and CEPR, London; Gruner: Department of Economics, University of Mannheim, D-68131 Mannheim, Germany, IZA, Bonn, and CEPR, London. We would like to thank Emmanuelle Auriol, Dieter Bos, Peter Funk, Mike Hout, Olivier Jeanne, Peter Jonas, Georg Noldeke, Thomas Piketty, Regis Renault, John Roemer, Jens Weidmann, and three anonymous referees for insightful comments and suggestions. We have also benefited from the comments of participants at workshops and conferences in Barcelona, Bonn, Davis, Jena, Maastricht, Magdeburg, Mannheim, Montreal, Munich, Rotterdam, Silvaplana, Tel Aviv, Tilburg, Toulouse, and Warwick, at which we presented earlier versions of this work. Financial support from the Deutsche Forschungsgemeinschaft, SFB 303 at the University of Bonn, is gratefully acknowledged.