Those in the business of giving policy advice know the surest way to guarantee a policymaker ignores their counsel is to say the problem is “complicated” or “there are no easy solutions” and that the best way to see the advice is accepted is to cast it in the form of a simple, straightforward solution that fits easily onto a single power point slide. World Bank economists have learned this lesson well as their recent report on how developing countries can cure corruption and related governance ills demonstrates.  Making Politics Work for Development: Harnessing Transparency and Citizen Engagement manages to state the solution to the corruption problem in one sentence: Give citizens more information about politicians so they will know which ones to vote out and which ones to keep at the next election.

The authors are able boil the complex problems of corruption and bad governance down into such a neat solution thanks to application of principal-agent theory.  But in avoiding the “it’s complicated”/“no easy solution” Scylla have they veered into the Charybdis of oversimplification?

From off-line correspondence I know I will lose one important GAB reader (the editor-in-chief) if it is not immediately clear that this post is not an attack on the value of principal-agent analysis per se.  It is not.  As Matthew explained in an earlier post, p-a is an invaluable tool for dissecting questions of political economy and designing political institutions and offers “many insights . . . into the study and practice of anticorruption.”  It bears emphasizing, however, as Professor Sean Gailmard explains in a chapter for the Oxford Handbook of Public Accountability, that what makes principal-agent theory so powerful is that it is not “a single encompassing theory with a specific set of assumptions and conclusions.”  Rather, its power derives from the fact that it is actually “a family of models with a related perspective” that allows different family members to be deployed depending upon the circumstances.

At its simplest the theory involves one actor (the agent) retained to act for the benefit of a second actor (the principal) where the agent has interests of its own.  When those interests diverge from its principal, the agent may further its interests at the expense of the principal’s, and where that is the case the solution the model dictates is to provide the principal with more information so it can monitor the agent’s performance. One principal with a single interest, one agent with a single interest, and the possibility that for selfish reasons the agent will not do as the principal instructs.

This is the version of p-a Making Politics Works applies to democratic states.  The electorate is the principal; the leader it elects is the agent, and the divergence of interest comes when the leader’s desire for self-enrichment is not held in check.  The solution to the corruption problem neatly follows.  Make the voters better principals.  Furnish them more information about office holders and those seeking office through right to information, income and asset disclosure and other transparency mechanisms and then be sure they use the information to vote the crooked out and keep the honest in (“healthy engagement” in the report’s somewhat awkward terminology).

The Bank is not the first to apply p-a to electoral democracy, and Professor Gailmard summarizes others that have done so as well.  The difference is that these authors apply variations to the basic model that more closely track the reality of electoral politics: voters are never perfectly informed about politicians’ performance, they cannot they accurately gauge whether a candidate is honest, and so forth.  What these more sophisticated models lose in simplicity they make up for with more nuanced conclusions about the relationship between citizen-principals and politician-agents.  One, at odds with the conclusion in the Bank’s report, is “that public accountability. . . is inherently limited, regardless of any imperfections voters may exhibit in their decision making process” (Gailmard p. 9).

In other words, no matter how much information voters have about politicians and how diligently they apply it in the voting booth, elections cannot ensure those elected will faithfully implement the will of the people.  One immediate consequence: combating corruption requires more than making citizens more informed, and more diligent, voters.

The report contains a valuable review and synthesis of the massive amount of research on combating corruption and related governance ills, discussing everything from the impact of media campaigns on election outcomes to when social accountability mechanisms produce better, less corrupt social services.  Policymakers, their advisors, and academics looking either for a summary of what interventions can make a difference or a way into the field of governance studies will find this part of the report of great value.

One body of research the report ignores, however, is that by political scientists which tests the fundamental assumption underpinning accountability in electoral democracy: that citizens have both the capacity and the inclination to vote for candidates that advance the general welfare. Professors Christopher Achen and Larry Bartels sum up the findings in their 2016 volume Democracy for Realists: Why Elections Do Not Produce Responsive Government:  “(E)lection outcomes are mostly just erratic reflections of the current balance of partisan loyalties in a given political system. In a two-party system with competitive elections, that means the choice between the candidates is essentially a coin toss.”

Whether, if Achen and Bartel’s conclusion stands, a more nuanced p-a model of electoral accountability can accommodate it I leave to others to argue.  Whatever the conclusion, though, Achen and Bartels throw a rather large curve ball at the simple form of p-a analysis Making Politics Works advances.

To be sure, that simplified framework is not without benefit.  It drives home the need for greater transparency and accountability, areas where so many developing country governments fall so short. And it shows why the obstacles to achieving these objectives are political not technical.  Getting policymakers in developing countries listen to this counsel is no mean feat and so perhaps the Bank can be excused for the way it decided to package it.

On the other hand, economists long ago found they could abandon the simplest model of a market, one where participants were perfectly informed and transaction costs were zero, and still get the ear of policymakers.  Isn’t it time that simple models of electoral accountability give way to those that are closer to real world conditions?  Won’t they produce more realistic policy advice?  And aren’t there ways to offer that advice without saying “it‘s complicated” or “there is no easy solution”?