
Paul Collier on Democracy and Violence 07/06/2009
What this covers
This is a conversation between Paul Collier and Russ Roberts about Collier's book Wars, Guns, and Votes. It starts from the puzzle of why democracy has delivered so little in the poorest countries, and Collier's answer is that Western observers treated elections as the whole of democracy. Roberts pushes back throughout, mostly on whether outsiders can be trusted to intervene well. He raises the Bootleggers and Baptists problem and notes that the "international community" is no single decider with responsibility or feedback. He also floats the view that aid has worked, but for the wrong people.
The middle of the discussion works through what an incumbent can do when he holds an election. Collier covers bribery, which he says works, partly because of patron-client relationship traditions that make a taken payment feel like an obligation. He covers intimidation, which works because a ruler can observe turnout even when he cannot observe votes, and he cites a randomized controlled experiment in Nigeria. He also covers excluding rivals, with Côte d'Ivoire as the example, where the ban backfired. Ethnic voting gets its own treatment: Kenya's 2007 election, and the contrast between Kenya and Tanzania drawn by Ted Miguel, in which Julius Nyerere built a national identity. Collier also argues that a good government is an unattractive option for an entrenched incumbent.
The remainder turns to evidence and policy. Collier defends his statistical work on conflict, noting that later work by Tim Besley and others supports the income and natural-resource findings. He concedes the evidence is not definitive and says policy cannot wait for it. He then argues, contentiously, for threatening rapid military intervention to deter coups against elected governments, and says that seeking accountability abroad is not colonialism. Postwar Germany and Mexico under the PRI serve as contrasting reference points for when democracy takes hold.
Collier argues that democracy's spread failed because elections are easy to hold while the institutional checks that make them meaningful are costly public goods that incumbents have every incentive to resist, so outside help aimed at accountability is both ethically defensible and feasible in some contexts.
- Elections are easy to conduct almost anywhere, but checks and balances are public goods nobody has an incentive to provide
- Incumbents have a menu of electoral manipulation options (bribery, intimidation, exclusion, scapegoating, miscounting), and each has costs that push them toward manipulation anyway
- Statistical evidence on development and conflict is suggestive but not definitive, so policy should combine it with case-study judgment and local knowledge
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Checks and balances are processes rather than events and are public goods, so no individual has an incentive to provide them, while governments have strong incentives to resist them; by contrast, the private incentives for parties to take part in elections are very strong.
“checks and balances of public goods nobody's incentive to provide them and indeed it's there's quite strong incentives for governments to resist them”
The international community is not a conscious agent or decider with responsibility or feedback incentives; the UN is not under anyone's control, so interventions done in its name face public choice problems that individual decision-makers do not.
“the international community is not an entity it is not a conscious being it is not a decider it does not bear responsibility it faces no incentives that individuals face”
Foreign intervention in poor countries creates a bootleggers-and-Baptists problem: high-minded advocates' justifications can enable others with less attractive agendas, and because the 'international community' is not a conscious decider bearing responsibility, it faces weaker feedback incentives than individuals.
“the argument here would be one that our listeners will recognize as a bootleggers and Baptists argument”
Vote-buying works in poor democracies partly because many voters treat accepting a payment as creating a moral obligation, which fits the patron-client tradition: having taken money, a voter feels ethically bound to vote for the patron, so these societies are coming to democracy from a history of patron-client relationships rather than clean democratic traditions.
“a lot of people in these societies kind of regarded deal as as giving some sort of moral responsibility”
Intimidation is effective because a ruler can observe whether people vote even when he cannot observe how they vote, and in ethnically stratified societies it is clear which groups will vote against him, so small forces of thugs can significantly depress turnout.
“the one big advantage of intimidation is that you can observe whether people vote even if you can't observe how they vote”
The statistical propositions linking conflict to income, growth, and resource dependence cannot yet be nailed down definitively, and statistical findings alone should not be used to settle policy because policy cannot wait for the evidence to mature.
“policy can't wait for a hundred years whilst economic source itself out policy is being made on the hoof day by day”
A randomized controlled experiment during the Nigerian presidential elections, run with a local NGO's anti-violence campaign, found that reducing violence and fear of violence in treated constituencies shifted votes: votes for peaceful politicians rose and votes for violent politicians fell, showing that intimidation works.
“the politicians that were using violence it worked so that where we were able to reduce that affect the vote for the peaceful politicians went up in the vote for the violent politicians went down”
National identity is not something that grows naturally out of the soil; it is a political construct built by leadership, which scholars do not understand terribly well.
“national identity is not something that grows out of the ground out of the soil it's something that's a political construct”
Being a good government is a poor option for an entrenched incumbent because good economic performance would require suppressing the patronage of elite groups, weakening the incumbent's own power base, and the incumbent's skill set and support organizations are not designed to run the economy well.
“the incumbent president might find himself weakening his own power base even as the performance of the economy improved”
In Kenya's 2007 election, ethnic voting was extremely strong: among the Luo, 98% voted for Raila Odinga, even though they gave President Kibaki and the vice president good approval ratings, and voters in denial attributed their choices to economic performance rather than ethnicity.
“amongst the amongst the Luo 98% of the Luo voted for a dinga the luo candidate bang”
Bribery is effective: the researchers in Collier's group found that bribery works on the whole, and new technologies such as mobile phones can favor bribery by letting voters photograph the ballot to prove how they voted.
“on the whole we've they found that the bribery is effective”
Ethnic and subnational identities are very strong relative to national identity, and this makes cooperation harder, which reduces the supply of public goods, and also contaminates politics because votes freeze into blocs locked in by allegiance rather than performance, reducing the role of competition.
“it's harder for people to cooperate and that that process reduces the supply of public goods”
Ethnic identity is not an automatic constraint, and national identity can be built: Ted Miguel's comparison of Kenya and Tanzania shows that Tanzania's first president Julius Nyerere built national identity while Kenya's leaders spent 40 years playing on ethnic differences, leading to ethnically polarized violence in 2007.
“Ted Ted McCoy has this very nice work comparing Kenya with with Tanzania”
Democracy's disappointing results occurred because Western observers emphasized elections as the fundamental element of democracy, missing that elections are only one institutional technology, and a properly functioning democracy also needs a range of checks and balances.
“we missed the the true nature of democracy which is we emphasized elections”
Democratic processes in developed countries are not reliably disciplining, and US foreign aid has not been particularly attractive in impact; the alternative view, that aid has been done all too well but to serve the wrong people, may be correct, which Roberts raises in the spirit of Bruce Bueno de Mesquita's argument.
“I'm less sanguine about the democratic process generally even in the developed countries”
International donors should protect democratically elected governments from coups by threatening timely military intervention: in the last year in Africa there were four successful coups, two against elected governments, and the African Union's condemnations and refusals to recognize the new governments did not reverse them, while a credible threat of rapid intervention could deter coups.
“the very threat of timely military intervention would have discouraged these things in the first place”
Excluding the strongest rival candidates is a tactic with a long record of success, and in Côte d'Ivoire incumbent Laurent Gbagbo banned the two candidates most likely to win (Henri Konan Bédié and Alassane Ouattara) and then ran against a weak opponent, Bédié's partner Gbagbo being the replacement, but he miscalculated how voters would perceive the bans and lost.
“he banned both of them and then he looked around for some somebody who he could run against to ensure a win against”
The first statistical analysis of UN peacekeeping by Collier, Hegre and Vogt-type collaborators finds that peacekeeping is effective, but this is not definitive, since assembling basic peacekeeping data took a long time and the field is open for much more research.
“we find that it's sizing its effective is that a definitive study very much not”
External intervention to strengthen government accountability to citizens is ethically legitimate, because the struggle for accountability has historically involved international assistance; the 'colonialism' charge conflates foreign empires that were undemocratic and extractive with temporary UN-sanctioned interventions aimed at increasing accountability.
“if you can't tell the difference between that and a temporary intervention by the international community under the auspices of the United Nations with the express purpose of increasing the accountability”
Elections are remarkably easy to conduct almost anywhere: even in post-conflict Iraq, Afghanistan, and the Democratic Republic of the Congo, it was possible to hold elections with high levels of turnout because political parties have strong incentives to take part.
“elections are remarkably easy to conduct almost anywhere”
Elections are necessary but not sufficient for democracy: the checks and balances that really matter are institutional and cultural, subtle and complex, and not dramatic events like elections.
“the election themselves are somewhat they're necessary but they're not sufficient”
Statistical evidence on causes of conflict has been strengthened: Collier and Anke Hoeffler's three propositions (higher income reduces conflict risk, income growth matters, and natural resource dependence raises risk) have been tested with better instruments by later work, including Tim Besley and Torsten Persson, which still supports the level-of-income and natural-resource propositions.
“the evidence still points still supports those propositions the on the level of income and the risks from natural resource dependence”
A functioning democracy can look like Mexico under one-party PRI rule, which was a democracy 364 days a year: the real constraints on incumbents are subtle institutional and cultural checks that matter more than the dramatic election day itself.
“those checks and balances are what often really matter they're institutional and cultural and all kinds of subtle and complex processes and the election themselves are somewhat they're necessary but they're not sufficient”
Germany's democratic success after World War II was easier than the earlier Weimar experiment because it was a high-income, prosperous society, was embedded in a neighborhood of democracies, had a prior democratic history, and was tied into European institutions where democracy was a condition of membership, along with sustained American troop presence and support.
“it was easier in a high-income society on a prosperous society it was certainly easier in a neighborhood where other countries were democratic”
Miscounting votes is the final and most tempting option, because the incumbent controls local institutions and local authorities may compete to deliver as many votes as possible to curry favor, so the manipulation can emerge without any top-down order.
“there might even be competition amongst local leaders to to deliver as many votes as possible and even if the incumbent president doesn't pressure for it”
Economic performance is very hard to measure from the voter's perspective: in the United States, it is impossible to distinguish whether rising unemployment reflects an insufficient stimulus or a president who inherited a worse situation than anyone thought, so voters and the dictator's 'be a good government' option is a weak basis for tenure.
“distinguishing between those two is I'd say impossible”
Policymakers should combine statistical evidence with practitioner and case-study judgment: the Copenhagen Consensus panel of Nobel-winning economists and experts, including Andrew Mack, chose peacekeeping as good value for money, a judgment mixing statistical and policy evidence.
“the panel of Nobel prize-winning economists who formed the assessment team for the Copenhagen consensus chose peacekeeping as one of the ways in which spending international public money looked to be good value”
The feasibility of international intervention is the real issue: intervention is often not feasible, but the fact that it cannot help everywhere does not justify helping nowhere, just as failing to save all children does not justify saving none.
“because we can't save the lives of all children we shouldn't save the lives of any children”
Within the UN Security Council, the three democracies with veto power (America, Britain, and France) provide a check, because their domestic democratic processes can discipline interventions, and these processes would have to break down in all three simultaneously for abuse to happen.
“within even within the Security Council there are three big democracies that have the power of veto America Britain France”
Bottom billion countries have the potential for rapid transformation to prosperity, and two meta public goods, accountability and internal security, are necessary preconditions for that movement to prosperity.
“there are two meta public goods that are one of which is accountability and the other of which is internal security without which that movement to prosperity is difficult”
Scapegoating carries a major cost because incumbents often rely on prominent ethnic minority businesspeople for financial support, since minorities cannot build independent political power bases and are therefore safer patronage recipients; attacking them can threaten the incumbent's own patronage system.
“the big con is that very often the incumbent rulers have kind of depended upon some prominent minorities especially ethnic business minorities for their own financial support”
Scapegoating a minority is a disturbingly successful electoral strategy because majorities have a long history of turning on minorities in the mistaken belief that the minority is the cause of their problems, but it creates a con because autocrats often depend on prominent ethnic minority businesspeople for financing.
“this is disturbingly successful there's a long history of majorities turning on minorities”
The main con of intimidation is that it is a two-player game: the incumbent has an advantage only because he can use the police and army, and in less extreme conditions the opposition may have an advantage in violence because it likely has more supporters than the government.
“the two can play at that game and and it's not clear that the incumbent has an advantage ultimately”
A dictator facing democratic pressure has a cost-benefit menu of options: being a good government, lying, scapegoating a minority, bribery, intimidation, excluding strong candidates, and miscounting votes, each with pros and cons.
“you list the various options in a cost-benefit analysis can you walk us through those”
Statistical findings give a pattern of scientific reliability that they may not deserve, and sophisticated statistics often fail to convince opponents, so Collier has become increasingly skeptical of the value of sophisticated statistical analysis.
“the danger is that a time statistical findings give a pattern of scientific reliability that it may not deserve”
In the first years of democracy, elections had big payoffs because some old autocrats stood for election believing their people loved them and lost, but this did not persist because incumbents learned there were easier ways to win than being a good government.
“a lot of incumbent leaders woke up to the fact that there were easier ways of winning an election than than the obvious one of being a good government”
Aid should follow the money: providing budget support without independent verification of budget systems empowers crooks, so donors should only give budget support where governments report to their own citizens on how the money is used, and independent international verification should be used where budget systems are weak.
“by providing aid without sufficient budget through transparency we're inadvertently empowering the crooks”
Building a more informed citizenry is possible and would improve policy outcomes: ordinary citizens' understanding of development is poor, and the bottom billion itself shows an appetite for better information, since over a hundred thousand copies of the book have circulated.
“any democracy you you can't get better policies than ordinary citizens understand and at the moment the understanding ordinary citizens about development is pretty poor”
Early optimism that democratization would rapidly produce economic reform and prosperity in poor countries was misplaced, because the bottom billion's core problem was political: power concentrated in self-serving elites who ran economies for their own benefit.
“like many people I'd diagnosed the sort of a core problem as being a political problem of the concentration of power in self-serving elites but it didn't turn out so well”
Lying is the second option for a dictator: it is even easier in autocracies than in democracies because the media is restricted and controlled by the authorities, but it is an unreliable tool because after long use people discount what the ruler says.
“you've been doing it such a long time that people don't believe you they people will discount what you say”
Vote-buying happens in developed democracies too, such as in Chicago-style practices where people give out money or gifts before voting and voters feel warm toward the giver and follow through even under a secret ballot; the difference is that these practices are more pronounced in poor countries because institutions that would curb them are weaker.
“they're more pronounced there because the institutions which could tail that behavior a weaker”
Lyndon Johnson's career as portrayed by Robert Caro shows an election-stealer who in almost every election he was in engaged in vote manipulation, illustrating that electoral fraud is a recurring feature of American political culture, not only of poor countries.
“his portrait of LBJ as an election stealer in almost every election he was in pretty amazing”
Standard regression-based statistical analyses of democracy, violence, and income face two problems: one never observes all the relevant control variables, and causation between correlated variables is often difficult or impossible to determine.
“one problem with the statistical approach is that one never has the data on all the other characteristics that are likely to matter”
Japan's democratization after World War II also succeeded because it was imposed with substantial external support, especially American support, and it had prosperity and tools for growth, suggesting that rapid transformation to both democracy and prosperity is possible under the right conditions.
“it's true of Japan as all as well a country that had very little democratic history and suddenly is transformed by Fiat essentially into a democracy”
A former Nigerian president set up five political parties, each of which nominated him as its presidential candidate, a 'constitution of innovation in multi-party dictatorship' that collapsed only because he died of a heart attack before carrying it out.
“he set up five political parties each of which happened to choose him as its presidential candidate”
Good economic performance does increase an incumbent's expected duration in office: where elections are properly conducted, people are more likely to vote for an incumbent when the economy has recently been growing than when it has been stagnant.
“people are more likely to vote for an incumbent if if the economy is recently been growing than if it's recently been stagnant”
Even with good economic performance, an incumbent's chances of winning are far from certain, so there may be more effective ways than good governance to secure a long tenure in office.
“even with good economic performance your chances of winning of very far from certain so there might be more effective ways of insuring the long duration in office”
The spread of democracy in the 1990s was a direct result of the fall of the Soviet Union: democracy first spread in Eastern Europe, and that gave courage to democracy movements across the world, moving many societies of the bottom billion to democracy quite rapidly.
“the spread occurred as a result of the fall of the Soviet Union”
The struggle is not a them-versus-us story: the story is about building institutions that restrain the strong incentives incumbent politicians have to use these tactics, which apply everywhere but are more pronounced where institutions are weak.
“it's a story of the struggle to build institutions which restrain the very strong incentives for incumbent politicians to use these tactics”
Politicians exploit shallow citizen understanding of development by making simple political gestures like kissing the baby, which lets them appear responsive while pursuing other interests.
“it's not moved beyond the the simple political gestures of the of the politician kissing the baby”
Democratic citizens in the rich countries have no appetite to reestablish empire but do want to ameliorate the condition of disempowered citizens in the world's poorest places, so the democratic process can be relied on to discipline interventions, though Collier concedes the process is imperfect.
“do the citizens of America of Britain or France have an appetite to reestablish Empire I think not”
Statistics without knowledge of context can be very dangerous, so the right approach combines statistics with 40 years of accumulated knowledge about Africa, since his judgments are not just based on statistics.
“I believe that statistics without knowledge of context can be very dangerous”
Informing citizens about development is far less complicated than teaching them financial regulation or inflation and unemployment economics, so the task is not as difficult as it seems.
“this is far less complicated than getting people up to speed on say financial regulation or inflation unemployment issue far less complicated”
Collier receives large volumes of email from people in these societies, which indicate that they are not gullible and understand what their governments are doing.
“I get a huge amount of email traffic from these societies on the hold of people who are sending me emails and not don't founded me to be gullible”
Democracy in the bottom billion might be advanced by external actors in the 2009 context, illustrated by the Iranian election whose results Collier said showed citizens' preferences being ignored, meaning internal accountability is often not working in these societies.
“there's a cause grave reason to think that citizens preferences are being ignored”