Wednesday, February 16, 2011

Links between Serbian, Tunisian, and Egyptian activists

From a very interesting article in the NY Times:
After a strike that March in the city of Mahalla, Egypt, Mr. Maher and his friends called for a nationwide general strike for April 6. To promote it, they set up a Facebook group that became the nexus of their movement, which they were determined to keep independent from any of the established political groups. Bad weather turned the strike into a nonevent in most places, but in Mahalla a demonstration by the workers’ families led to a violent police crackdown — the first major labor confrontation in years.

Just a few months later, after a strike in Tunisia, a group of young online organizers followed the same model, setting up what became the Progressive Youth of Tunisia. The organizers in both countries began exchanging their experiences over Facebook. The Tunisians faced a more pervasive police state than the Egyptians, with less latitude for blogging or press freedom, but their trade unions were stronger and more independent. “We shared our experience with strikes and blogging,” Mr. Maher recalled.

For their part, Mr. Maher and his colleagues began reading about nonviolent struggles. They were especially drawn to a Serbian youth movement called Otpor, which had helped topple the dictator Slobodan Milosevic by drawing on the ideas of an American political thinker, Gene Sharp. The hallmark of Mr. Sharp’s work is well-tailored to Mr. Mubark’s Egypt: He argues that nonviolence is a singularly effective way to undermine police states that might cite violent resistance to justify repression in the name of stability.

The April 6 Youth Movement modeled its logo — a vaguely Soviet looking red and white clenched fist—after Otpor’s, and some of its members traveled to Serbia to meet with Otpor activists.

Another influence, several said, was a group of Egyptian expatriates in their 30s who set up an organization in Qatar called the Academy of Change, which promotes ideas drawn in part on Mr. Sharp’s work. One of the group’s organizers, Hisham Morsy, was arrested during the Cairo protests and remained in detention.

“The Academy of Change is sort of like Karl Marx, and we are like Lenin,” said Basem Fathy, another organizer who sometimes works with the April 6 Youth Movement and is also the project director at the Egyptian Democratic Academy, which receives grants from the United States and focuses on human rights and election-monitoring. During the protesters’ occupation of Tahrir Square, he said, he used his connections to raise about $5,100 from Egyptian businessmen to buy blankets and tents.
This fits with Mark Beissinger's thesis about the "color revolutions": activists learn from one another and spread protest across borders, as I mentioned below. (I was interested to learn that Egyptian activists even modeled their logo on the Otpor logo, just as Ukrainians and Georgians did). The article's description of how these activists used facebook  also suggests that the more important contribution of "social media" to the protests in Tunisia, Egypt, and elsewhere is less the ability to organize people than the spread of tactical information across borders (basically, how to outsmart the police and energize a small cadre of dedicated individuals).

The whole piece is well worth reading.

Inequality, injustice, and democratization

(Warning: an epically long post that meanders through the literature on inequality and democratization and comes to conclusions that probably sound unsurprising. Written partly as an attempt to construct a workable set of lecture notes for my course this term).

Was economic inequality important in triggering the anti-regime protests in Tunisia and Egypt? A number of news articles I’ve read mention, often in passing, that rising inequality was one of the causes of the unrest in both countries. This inequality was manifested in the large fortunes accumulated by both the Ben Ali and the Mubarak clans and other influential insiders (including senior government figures in both countries) and in the lack of opportunity for relatively well educated people, who struggled to get jobs even with university educations. And many people believe inequality had been rising there (for reasons that are common to a lot of other countries – some “liberalizing” reforms that basically produced forms of crony capitalism that enriched well-connected insiders at the expense of most people).

This seems plausible enough, even if the evidence is scattered and anecdotal; reliable and recent statistical estimates of inequality in Egypt and Tunisia do not appear to exist. What does exist provides inconsistent information. For example, the CIA World Factbook reports a Gini of 0.34 for Egypt in 2001 (which is below average), but without giving sources, and a gini of 0.4 for Tunisia in 2005 (which middling), yet a more complete dataset of inequality measures developed by the University of Texas Inequality Project  (which is pretty complete as these things go, i.e., not that complete) suggests inequality was higher in both Egypt in 2001 (gini of .47 in 1999, and an average of .42 for the period 1963-1999; .47 is above average) and Tunisia in 2000 (gini of .48 in 1998, and an average of .47 for the period 1963-1999). (If you ask Wolfram|Alpha, you seem to get yet another set of numbers, without sources, though they are probably based on some complicated computation involving the CIA factbook.)

At any rate, most of the recent theoretical work on democratization supports a role for inequality in regime change, though not without qualifications, and it certainly would  not support a "high inequality leads to democracy" thesis. Boix and Acemoglu and Robinson, among others, have argued that the level of inequality is a pretty important factor in whether or not a country moves from dictatorship to democracy. Their arguments are slightly different, but the logic is similar. The basic idea is that democracy is more responsive to the wishes of the “median individual” than dictatorship. As inequality increases, the median voter in a democracy will be poorer and will benefit more from redistribution. Under some reasonable assumptions (e.g., voters vote their interests), democracy will thus tend to be, all other things equal, more redistributive than a dictatorship controlled by the economic elite. (The key word is “tend”. Democracies vary greatly in the degree to which they are actually redistributive, as we will see below, for all sorts of reasons you can probably figure out for yourself: ethnic and religious diversity, different constraints on taxation, etc .).

The poor should thus normally prefer democracy to dictatorship; demand for democracy should come from “below” not from “above,” and should be higher the higher the level of inequality. In this framework, moreover, the poor prefer democracy not just because of the redistribution it offers “today” but also because it acts as a commitment device: sure, the rich may offer some economic concessions in the face of unrest, but without an institutional voice the poor cannot be certain that the rich will continue to offer such concessions in the future. Conversely, the rich should normally prefer a dictatorship in which they control the state to democracy, and this preference will be stronger the higher the level of inequality (and thus the expected redistributive effects of democracy). This is obviously a highly simplified story about why people prefer one regime over another – it hardly accounts, for example, for the humiliations of arbitrary power that appear to be the immediate and evident causes of the anger against dictators that we see in actual episodes of mass protest, or for the complexity of class alliances in actual transitions to democracy. But it is not a terrible starting point for thinking about the incentives of key actors in processes of potential democratization.

For one thing, dictatorships do seem to be, on average, more unequal than democracies. Though measures of inequality are not always of very great quality (and should always be taken with a grain of salt), correlations between democracy and equality seem to persist across a range of measures. Using the DD measure of democracy and dictatorship by Cheibub, Gandhi, and Vreeland and the UTIP data we can see that democracies  in the period 1963-1999 had a mean Gini coefficient of  0.4, whereas dictatorships had a mean Gini coefficient of 0.45. Not an enormous difference, perhaps, but at least expected from the theory. Measures of “capital shares” – which are more appropriate in this context, since they basically measure what part of the national income goes to the “capitalist class” – tell the same story. Using a somewhat truncated version of the dataset on capital shares compiled by Rodriguez and Ortega, we find that in the 1963-2001 period the share of the national income going to capital was on average 60% in dictatorships, and 56% in democracies. (I should actually give you confidence intervals for that sort of thing, but I am not quite sure how to produce them).

The actual distribution of inequality across regime types is fairly wide, however: some democracies are more unequal than most dictatorships, and some dictatorships are more equal than most democracies. Consider the following density plot (essentially a smoothed histogram), using the UTIP data and the DD measure of democracy and dictatorship:

(Democracies are represented by the purple line, dictatorships by the blue line). The graph basically tells you that though democracies are clustered towards the low inequality end of the spectrum (the median Gini for democracies is 0.39) and dictatorships towards the high inequality end of the spectrum (median Gini  0.46), there is still a fairly wide spread, with many democracies with high inequality and some dictatorships with low inequality. We can dig deeper, however. If we consider the dictatorships that have lower inequality than most of the democracies (lower than the median level of inequality for democracies), most of them are communist or former communist countries: Albania 1988-1990, Azerbaijan 1991, Bulgaria 1963-1989, Bosnia and Herzegovina 1991, China 1977-1986, Cuba 1977-1989, Hungary 1963-1989, Poland 1970-1988, Romania 1963-1969, Russia in 1993. A few other countries round out the list: Afghanistan (which could be counted as a formerly communist country, and anyway figures for Afghanistan are likely to be nonsense), Singapore 1978-1999, the Seychelles 1976-1986, Syria 1983-1992, Algeria 1976-1994 (which also had a “leftist” revolution), Egypt 1974-1978 (the Sadat years, I believe, though the foundation for this would have been laid during Nasser’s rule), Iran 1981-1989 (another country that experienced a revolution), Malaysia 1993-1999, Portugal 1975, Senegal 1974-1986, South Korea 1979-1987, and Uruguay  1976-1979. The rest of the dictatorships for which there is inequality data in this sample have higher inequality than most democracies. (If you run the same exercise for the democracies that have higher inequality than most dictatorships, you find basically younger democracies – an average age of 13 years, compared to 37 in the entire sample- or democracies that have been repeatedly undermined by coups, though India and Venezuela are clear outliers, Venezuela probably for reasons discussed in this post.).

None of this proves anything (correlation is not causation and all that, besides the fact that the data on inequality is poor and I have not controlled for anything, though other people who know more about this than I do have done so), but it’s interesting (for some values of interesting, I suppose). There is also, among other things, the fact that most coups in Latin America in the 20th century were supported by economic elites fearful of redistribution and confiscation, and that historically objections to the extension of the suffrage focused on the threat to property more than anything else. This is what we would expect from the highly simplified models of Boix or Acemoglu and Robinson: when inequality is high, the wealthy elite have large incentives to invest substantial resources in controlling the state to prevent redistribution, up to staging a coup (in a democracy) or otherwise supporting high levels of repression (in dictatorship), at least so long as they do not have an “exit” option (perhaps because their assets are mobile or otherwise easily hidden from taxation). But when inequality is high, the poor non-elite also have the most to gain from redistribution, so high levels of inequality should be associated with high levels of class conflict and ultimately with dictatorship (either of the “right” or of the “left,” depending on the balance of forces: so we see revolutionary dictatorships with relatively low levels of measured inequality). Very high inequality seems to lead to irreconcilable class conflict, which a small wealthy elite is more likely to “win” (since its collective action problems are smaller; hence the rarity of great social revolutions). Hence it is no surprise that high levels of inequality appear to be statistically associated with the breakdown of democracy (see, for example, the recent work of Christian Houle [gated]).  One may think that this logic probably works against democracy in Egypt; if the military is as highly embedded in the economy as recent stories note, then it would seem to be less likely that they will agree to relinquish enough power to a genuinely democratic government, especially when much of that wealth funds the lifestyles of senior officers. (The same does not appear to be true in Tunisia, where the military did not appear to have been part and parcel of the “winning coalition” in recent years).

But if the theory is very good at telling us when democracy breaks down or when it is unlikely to emerge (namely, when inequalities are large and visible and wealthy elites cannot easily take their assets elsewhere, as in agrarian economies, so that they have very strong incentives to prevent the poor from gaining control or substantial influence over the state) it seems to be less good at telling us when it is likely to emerge. There is no clear statistical relationship between the level of inequality and transitions to democracy: democracy has emerged in countries with low, high, and medium levels of inequality [Houle again], even though the high inequality democracies have tended to have higher rates of breakdown afterwards. (I do not know if the data is good enough to bear a look at the within-country pattern: do countries that experience an increase in inequality have higher or lower probabilities of transitioning to democracy? Anybody wants to help me look at that? How would you do that?). Part of the problem is that it isn’t clear how lower levels of inequality affect both the “demand” for democracy (Acemoglu and Robinson suggest demand would be lower, but then Boix suggests that the wealthy would have less incentive to oppose it, which seems to produce an indeterminate result) and the ability of non-elite groups to coordinate action (my guess is that low levels of status inequality would make it easier to coordinate collective action and to generate collective protest identities, but I don’t really know). There may also be income effects: high or low inequality may produce different political outcomes at different levels of development (just-so story: it may be that as income increases throughout society, and its marginal contributions to happiness and physical security decrease, direct redistribution becomes less and less important, and other things like physical repression may be more important).  

Moreover, many historical episodes of democratization seem to have been driven by emerging economic elites who wished to avoid predation by other elites controlling the state, and who therefore enlisted the lower classes in their struggles against these predatory state elites, something that does not fit neatly with the “redistributive” models of Boix or Acemoglu and Robinson (as discussed, for example, by Ansell and Samuels here [gated]). Consider the fact that in Egypt the main protests seem to have been “led” (to the extent that they were led at all) by young professionals who would not necessarily be well connected to the state but instead probably suffered quite a bit from its predation. The point is more general: in actual episodes of democratization, the demands for an end to arbitrary treatment by the state – for “dignity”, an end to corruption or police harassment, for free expression, etc. – seem pretty prevalent, even though we can also find demands for redistribution driven by more economic interests.

My guess is that the effects of inequality on democratization are mediated by beliefs about justice or fairness. Here’s a sketch of a theory. There is some cultural equilibrium, produced by long-lasting regimes, between beliefs about what constitutes a “fair” distribution and the actual level of redistribution in society, determining the long-run level of inequality in a society. This equilibrium can be partly genuine (there are differences in what different societies accept as fair processes of distribution, leading to different levels of inequality; see this article by Alesina and Angeletos for more [gated]), but in non-democratic societies the equilibrium is also enforced by coercion and opacity (the people do not know how rich the wealthy really are). But sharp departures from this equilibrium lower the mobilization threshold of people dissatisfied with the status quo – they make more people angry, and hence more likely to mobilize (anger is a much better spur to take risks than plain vanilla self-interest). These departures might be produced by increases in the “visibility” of elite wealth (“ostentation”) or by increases in the incidence of wealth due to unfair processes of distribution (“corruption”). Hence the importance of incidents like the Shah’s grandiose party on the anniversary of the Persian monarchy for triggering mobilization, or even things like the publicity surrounding the comparatively luxurious living conditions of the leaders of the GDR early during the 1989 fall of the communist regime there; in Eastern Europe, where people had been socialized for decades into an ethos of “equality,” even the relatively small privileges of the nomenklatura compared to the rest of the population were seen as galling. (Honecker and Krenz lived luxuriously by egalitarian GDR standards, but not that luxuriously). In general then, we should expect that it is not the level of inequality that matters for mobilization, but sharp changes in inequality, relative to the “fairness” baseline. Yet the level of inequality may still be related to mobilization, since it may be that at high levels of inequality, even small departures from the fairness baseline will be easily perceived as forms of injustice. So that high inequality societies should be, on this view, more prone both to mobilization leading to democratization (as we see in the data assembled by Houle, referenced above, where high levels of inequality seem to be associated with more transitions to democracy: see table 2) and to the breakdown of democracy (because of its threat to the interests of the wealthy).

To finish this epically long post, if you are still reading, consider this interesting statistic from Egypt: according to the World Values Survey of 2005 (if I’m reading the answers to question V120 correctly – link may not take you to the question directly), an astonishing 52% of Egyptians thought that “In the long run, hard work usually brings a better life” – much higher numbers than in the USA or New Zealand. And yet this belief was bound to be tested by the way the Egyptian economy worked (where only the well connected ultimately prospered, and where large numbers of college graduates apparently failed to find jobs). I do not know why the Egyptians seemed to be so optimistic in 2005, but I can imagine that in the face of the realities of the Egyptian economy, a lot of (especially young and previously unmobilized by existing parties) people became very angry, especially young people. Egypt was a very equal society by some measures, but given its “fairness” baseline, it was probably failing spectacularly in the last few years. 

Wednesday, February 02, 2011

Why does the spread of protest follow regional and cultural lines?

One of the things I find interesting about the recent events in Tunisia and Egypt is the fact that protest tends to spread along regional and cultural lines. Though there is some evidence that the fall of Ben Ali in Tunisia and the huge demonstrations in Egypt are making authoritarian leaders nervous in China (which has apparently started to censor searches on Egypt), Cuba, Azerbaijan, Kazakhstan, and many other countries, their main effect has been on other Middle Eastern countries – Jordan in particular, which has also seen large demonstrations and where the king just fired his entire government. The same phenomenon was seen in 1989, where protest spread primarily in (mostly European) communist countries, and in the early 2000s, with the original “color revolutions,” where protest spread among post-Soviet competitive authoritarian regimes and resulted in regime change in four of them (Serbia, Georgia, Ukraine, and Kyrgyzstan). And we could go further back in time: there is an age of guerrilla insurrection in Latin America in the 50 and early 60s catalyzed in part by the Cuban revolution, a wave of European revolutions in 1848, and so on (see also the maps of political change below). But why should protest spread primarily along cultural and geographical lines? Why shouldn’t the fall of an autocrat randomly impact all other autocrats, as people update their beliefs in the efficacy of protest?

In a 2007 piece in Perspectives on Politics, Mark Beissinger pointed out that a similar cultural or political context seems to be important for people to update their beliefs in the efficacy of protest. Suppose that citizens in country A successfully manage to overthrow their autocratic government. How should citizens of country B, also under an autocratic government, process this information? If they think the situation in A is very similar to their own, they might think that overthrowing their own autocrat is easier than they thought, and hence be more willing to run the risks of protest, even if, objectively speaking, nothing has changed in their situation. But if people in B believe that their situation – their political institutions, culture, etc. – is very dissimilar from the situation in A, the information provided by the overthrow of the government in A will not be as relevant to their decision regarding whether or not to risk protesting against their own government. As a consequence, protest will tend to spread in regimes that appear to be in culturally and politically similar situations: Soviet satellite regimes, post-Soviet competitive authoritarian regimes, and now apparently non-oil-rich Arab regimes experiencing economic stagnation and high levels of corruption.  This does not mean the protests in B will be successful – many color “revolutions” failed (in Belarus, Armenia, Kazakhstan, and elsewhere) in part because the situation was not actually as similar as people in B believed. But the beliefs of people in B regarding the similarity of the situation in A could, so the argument goes, affect their chances of success in B. Until recently, few people really thought Mubarak was going to be shaken from power by popular revolt (see, for example, this long and well informed piece by Adam Shatz in the LRB); but the success of the revolution in Tunisia actually increased the chances of success in Egypt by lowering the thresholds for protest of many people (leading to a bandwagon or snowball effect of the sort described by Timur Kuran); and it lowered the thresholds of many people in part because people in Egypt saw enough similarities between their situation and that of the Tunisians to revolt.

There is more to Beisinger’s piece than this. He also notes that explicit person to person connections among activists are also very important in ensuring the spread not only of protests but of specific tactics of contention (e.g., what to do in case of a fraudulent election): activists in Serbia, for example, were in contact with activists in Georgia and Ukraine, and many of them were also in contact with people in other post-Soviet regimes like Kyrgyzstan, Kazakhstan, and so on. Cultural context mattered here as well: many of these activits could talk to each other in Russian, and they could find many similarities among their different regimes. I assume that similar connections will be found among Arab activists. And he has a very interesting discussion of how autocratic learning interacts with the spread of protest, a topic which I discussed briefly below. But let’s just talk about the implications of the first part of the argument for current events.

First, the argument implies that protest will spread in the Middle East but not necessarily elsewhere, and will become progressively less effective, not just because autocrats will learn to counter such protest, but because people in other regimes will find less and less in their situation that is analogous to the situation in the Middle East. The Chinese and Cuban regimes thus would seem to be safe for the moment, for example.

Second, the argument implies that the more citizens of a country think their situation is unique and without parallel, the less likely they are to follow the lead of protests in other countries. Bad analogical reasoning might actually increase the chances of revolutionary success elsewhere. Even if the situation in Libya, for example, is quite different from the situation in Egypt, if enough people in Libya believe otherwise, they might actually increase their chances of overthrowing their own government. (What are the chances of Qaddafi becoming the Arab Ceausescu?). By contrast, if people in North Korea, for example, (wrongly) believe their own situation and culture is especially unique, they might be less likely to take advantage of a wave of protest. 

What do people think?