The Problem of Social Choice
If societies and their economic systems evolve over time, this is a very impersonal process based on the four conditions described above. Such a process may be rational in hindsight, in that strategies, institutions, technologies, behaviors or systems chosen by competition were by definition more successful in the given environment. If instead we say that societies adapt to changing circumstance, this implies somehow that society as a group is able to choose a rational path with foresight. While individuals may be able to make rational choices in their best long-term interest, given their particular preferences, certain constraints, and limited information about the future, economists are very skeptical about the ability of groups to make rational choices. In fact, much of what is interesting about economics is the conflict between the individual and the group (e.g., having more babies may be in the interest of the parents but the resulting population growth may not be in the interest of society).
It is easy to see how oligarchies, autocracies, or dictatorships may make choices that benefit a narrow segment of society to the detriment of society overall. Consider, however, a democratic society in which individuals are able to jointly make social choices. Though we would expect that such a society would act in its overall best interest, social choice theory tells us that this may not only not be so, but may in fact be virtually impossible to determine. How would such a society choose the policies its government would carry out? Individuals might directly vote on specific policies (as Ross Perot has proposed), but it is costly for individuals to become well-informed on the arcana of the many political issues that get decided everyday, or they might also elect representatives to make the day-to-day decisions on the basis of agreed-upon general principles (sometimes, representatives may decide for society as if they know best). How can we take different individuals with different preferences and somehow determine what society prefers?
The Marquis de Condorcet recognized that majority voting, where the people either vote on specific policies or on the representatives to decide for them, is an inconsistent method of social choice. In a simple example of the voting paradox, suppose there are three persons (named 1, 2, and 3) and three possible choices (called A, B, and C). Suppose person 1 prefers A to B and B to C; his preferences are consistent, since he will also prefer A to C. Now suppose that person 2 prefers B to C and C to A, while person 3 prefers C to A and A to B. If we were to use majority voting to decide what this three-person society prefers, then we would find that A is preferred to B, B is preferred to C, and C is preferred to A.
Maybe democracy is the problem, perhaps? There are many different methods of social choice, and many different voting rules. We could rely on dictatorship (where one person has the only vote that counts), we could rely on consensus (where any one individual has veto power), we could have run-off elections or rank-order voting, or we could discover an optimal social choice rule that has never before been used. Might one of these work best?
Arrow defined several conditions for a optimal social choice rule. The rule must work for any large set of consistent individual preferences over a wide range of possible options. The rule must be consistent, so that the ordering is transitive (if society prefers A to B and B to C, then it also prefers A to C) and the choice is independent of irrelevant alternatives (when comparing A to B, we don't need to know how C compares to D). The rule must be decisive, so that if at least one person prefers option A over B and nobody prefers B over A, then society can be said to prefer A over B (this is the so-called Pareto rule). Finally, the rule must be fair, so the decision is nondictatorial and individual preferences cannot be imposed on others. In what became known as Arrow's General Impossibility Theorem, he proved that no social choice rule could possibly satisfy these conditions. No method of making social choices could in general be fair, consistent, and decisive; for example, dictatorship isn't fair, majority voting isn't consistent, and consensus isn't decisive.
Many people find Arrow's Theorem to once again highlight economics as the dismal science, and interpret it to mean that democracy doesn't work. The political scientist W. H. Riker takes another view, however. Riker takes issue with the idea of populism, in which a representative claims to speak for the people, since Arrow's Theorem proves that it is impossible to know the will of the people with any confidence. Instead, he argues that we should rely instead on what he calls Madisonian Liberalism, named after one of the key figures in the writing of the U.S. Constitution. Madison clearly understood that voting was subject to all sorts of problems, and that it would not be able to reveal what policies society most preferred; instead, he argued that voting was necessary as a check on the use of government power, since it would prevent the continuation of policies that society clearly least preferred. And thus democratic societies are less likely to make the worst choices, but no more likely to make the best choices.
Another interpretation of Arrow's Theorem comes from his method of proof, since he was able to show that any decision by a coalition depended on agreement on a narrow range of issues, that any coalition could be split up by the right issue, and that voting outcomes can be manipulated by changing the agenda. Indeed, how political innovators gain power is by asking questions that split up ruling coalitions. With unidimensional policies, policies tend to be somewhat centrist (a political coalition wins by capturing the median voter, if it does not lose its base by appearing too centrist), but with multidimensional policies, policy becomes much less stable. The swings and cycles due to political innovation may temporarily lead governments towards extreme policies, though in such cases there are still opportunities to those who innovate middle-ground policies favored by the median voter, as long as voting remains a check on power.
Government policy is largely determined then by the agreement of coalitions that are themselves made up of smaller coalitions, and so on until we reach the level of individuals who don't agree on everything. Coalition behavior is thus important. Mancur Olsen explained the incentives of coalitions as they attempt to affect government policy. Specifically, he divided coalitions into two types -- distributional and encompassing -- and he models coalition behavior by comparing marginal costs and benefits. The marginal cost of running an organization rises as it increases in size. The distributional coalition is trying to increase the benefits received by its members through political action, and while its chances of success increase with size, the average benefit falls as it is shared with larger numbers. The encompassing coalition seeks benefits that cannot be excluded to only its members, and so it would like to be larger, but the poor benefits of membership discourage participation.
Using this theory, Olsen develops a model of economic growth based on the stability of the society and the ability of any individual coalition to affect social choice. He predicts that stable societies tend to accumulate distributional coalitions, or special interest groups, and that over time these will get a stranglehold over the economy. Societies may be able to establish constitutional rules beforehand that prevent this stranglehold. Larger societies may be able to prevent this stranglehold if no single coalition is able to dominate, but is instead forced to compete with many other coalitions, but there is no guarantee that coalitions won't organize into meta-coalitions to achieve their political and economic dominance. Smaller societies may in fact find that the optimal distributional coalition is big enough that it becomes an encompassing coalition, especially when faced with external competition.
In an extension of this model, Olsen argues that social choices that
lead to economic development in the interest of society as a whole may
come from a variety of political settings. In an extreme case, Olsen
suggests that a society dominated by thieves is likely to be the worst
if the thieves compete with each other in plunder. On the other hand,
if a robber chieftain were able to eliminate his competition and have a
long-term hold over society (perhaps by crowning himself an hereditary
king), then policies encouraging overall economic development might be
in his own self-interest if it maximized the long-run potential of his
plunder. A smaller share of a growing pie may in time add up to more
than a large share of a shrinking pie, and strong stable nation-states
may be more successful than unstable states made up of squabbling powers.
Even democratic societies where government officials have the ability to
make choices in their self-interest may be susceptible to tenure problems,
since short-termers in unstable governments would be tempted to plunder
quickly, while long-term governments in stable situations would find their
officials more interested in retaining power longer, and promoting long-term
economic growth.