NOTE: Another contribution by Fred Thompson:
A recent paper by Martin Gilens and Ben Page entitled “Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens” and forthcoming Fall 2014 in Perspectives on Politics has garnered a lot of commentary in the blogosphere. But some have hailed the study for doing things Gilens and Page do not do: prove scientifically that America is basically an oligarchy/plutocracy. Moreover, while I find Gillen and Page’s empirical analysis very impressive – a real tour de force –, it is less than fully conclusive.
On the first point, America may well be an oligarchy/plutocracy. But Givens and Page do not claim that this is what their analysis shows; at most they claim that their findings are consistent with the notion that wealthy elites and businesses run things, which they acknowledge is their view. Nevertheless, they are very careful to distinguish between their opinions and their empirical findings.
On the second point, let’s start with what they find and how they got there. Gilens and Page find that the preferences of economic elites and organized groups, especially those representing business interests, influence (to a degree) the shape and content of public policies, while those of average citizens do not. They also find that business-oriented groups play a bigger role in shaping policy outcomes than do other kinds of groups, primarily because more of them are engaged on each issue (roughly twice as many, on average), and also because they are almost always in agreement on proposed policy changes. Both sets of findings are based upon a multivariate analysis of 1,779 cases, in which the a-priori policy preferences of average citizens, economic elites, mass-based interest groups, and business-oriented interest groups were regressed upon legislative outcomes.
These findings are interesting and plausible, but they aren’t very strong or robust. Their strongest model has an R-squared of just over .07, which is to say, all of the predictor variables in the analysis taken together explain only 7 percent of the variance in legislative outcomes; taken individually none of them explain more than 5 percent.
There is a second reason for my questioning the conclusiveness of their findings: I don’t think their analysis shows what they think it does. Their analysis focuses on the shape and content of public policies, not their effects overall. It is hard for me to see how ordinary citizens could or would want to shape the public policies of a huge and complex government like that of the United States. Nor does democratic theory require them to.
As Gilens and Page themselves explain, rational choice theories of electoral democracy presume that competitive vote-maximizing parties or candidates in a two-party system converge at the mid-point of the distribution of voters’ most-preferred positions. Consequently, where preferences are jointly single peaked so that they can be arrayed along a single dimension, the public policy outcome consistent with the preferences of the median voter is the empirically predicted equilibrium result of two-party electoral competition.
In economics, this outcome is usually referred to as Bowen equilibrium (after Howard Bowen, one of my teachers in graduate school). Note, however, that the naïve median-voter outcome is rarely an efficient outcome (the losers could compensate the winners to move to a different place on the continuum and still be better off) and under most straightforward sets of circumstances the inefficiency usually takes the form undersupply of publicly provided goods and services. In contrast, if one accounts for side payments (think tax prices), an outcome can be imagined that is both efficient and consistent with the median-voter result, but compared to the voters’ starting positions, would tend to shift the policy outcome in the direction of the preferences of those who have the most to gain/lose from the policy outcome in question (just what Gilens and Page tend to observe).
The efficient outcome that satisfies the median-voter outcome is, in fact, the true “Condorcet winner:” the only outcome that would be preferred to any alternative policy in head-to-head majority-rule voting by all voters. Except where this knife-edged position is realized there is always an alternative position and set of side-payments that can defeat the last head-to-head winner. Hence, it is arguably the “most democratic” policy.
Under most rules of order, the status quo remains in effect until the true Condorcet winner is located or until someone brings the process to a premature conclusion. Because the division of authority and jurisdiction in most legislative arenas permits leaders to manage/direct the process, real-world policy choices depend not only upon the preferences of the participants but also upon how the rules and organization of the legislative process. This means that real-world institutional arrangements can produce outcomes, which are neither democratic nor efficient. For example, where concentrated, salient, short-term payoffs (pork, jobs, etc.) trump diffuse, low-visibility, long-term consequences.
In theory, policy outcomes also depend upon how preferences are configured. As Gilens and Page explain: If voters’ preference orderings are not unidimensional and are sufficiently diverse, there may be no Condorcet winner (p. 5). Economists typically finesse this issue by assuming that any preference that matters will have measurable consequences and that anything that can be measured can be expressed in terms of willingness and ability to pay (money), which means that preference orderings on any issue can be treated as unidimensional – but this may merely reflect our own one-dimensional notions of what matters.
Gilens and Page acknowledge that there is a version of democratic control in which deviations from naïve or a priori policy preferences might be entirely consistent with the outcomes they observe (also those suggested by the informal economic theorizing above). In this version of democratic control, the marginal voter has weak policy preferences but instead cares about the consequences of policies – they ask only how well their interests have been served and vote accordingly; politicians enact policies in anticipation of those judgments. Consequently, rather than responding to citizens’ stated policy preferences, legislators are oriented to the electorate’s interests (which is unfortunately not necessarily the same as the public interest).
Gilens and Page simply dismiss this possibility, both because they don’t know how to test it and because they cannot imagine it to be true.
I am willing to grant that elected officials often fail to pursue the public interest, otherwise how can one possibly explain Congressional dithering over TARP in the late summer of 2008 or the absence of further fiscal stimulus in 2010. However, I am as inclined to attribute this to an excess of responsiveness to public opinion with respect to policy content as to too little.