Available now: USAvailable now: UK

Democracy, What Is It Good For?

In an earlier post, we reported on our research joint with Suresh Naidu and Pascual Restrepo, “Democracy, Redistribution and Inequality”, which showed very limited effects of democracy on inequality.

So one would be excused for paraphrasing Edwin Starr’s famous song and Ian Morris’s forthcoming book, War! What Is It Good for?, and ask “democracy, what is it good for?”

Certainly not economic growth, most would reason.

This conclusion is based on a consensus engulfing both academia and the popular press that democracy is at its best irrelevant for growth, and perhaps even a hindrance.

For example, Tom Friedman wrote in the pages of The New York Times:

One-party nondemocracy certainly has its drawbacks. But when it is led by a reasonably enlightened group of people, as China is today, it can also have great advantages. That one party can just impose the politically difficult but critically important policies needed to move a society forward in the 21st century,”

Friedman wasn’t making this up. Robert Barro, who has written several papers on the topic, argued in his book Getting it Right: Markets and Choices in a Free Society:

More political rights do not have an effect on growth… The first lesson is that democracy is not the key to economic growth.

A recent survey of the recent literature similarly concludes:

The net effect of democracy on growth performance cross-nationally over the last five decades is negative or null.

Equally dominant is the view that democracy isn’t right for low-income countries (which are often the ones trying to turn their societies into democracies). The pages of The New York Times again summarize what most of the popular press seems to have accepted as axiomatic, this time in the words of David Brooks defending the Egyptian military coup,

It’s not that Egypt doesn’t have a recipe for a democratic transition. It seems to lack even the basic mental ingredients.

Judge Posner also agrees with this conclusion (even though, to the best of our knowledge, he does not go so far as supporting Egypt’s murderous generals), and writes

Dictatorship will often be optimal for very poor countries. Such countries tend not only to have simple economies but also to lack the cultural and institutional preconditions to democracy.

Our paper Democracy, Redistribution and Inequality” was in fact part of a broader project investigating the implications of democracy for both economic growth and inequality.

Our main paper (again joint with Suresh Naidu and Pascual Restrepo) is out, and as the title suggests “Democracy Does Cause Growth, it sharply disagrees with this consensus.

The curious thing is that our paper is not actually the first one to find a positive effect of democracy and economic growth, but it is true that the literature contains many papers that find no effects or sometimes even negative effects.

We think there is a simple reason for this, which can be seen in the next figure. This figure shows the evolution of GDP per capita following a democratization event compared to nondemocracies (all democratizations are lined up to date 0 so as to visually trace out average growth following democratization relative to the control countries in which there is no democratization).


The first thing that jumps out from the figure is that a typical democratization takes place when a country is undergoing an economic crisis (a point first emphasized in an earlier paper we wrote with Simon Johnson and Pierre Yared, “Income and Democracy”, and later investigated in greater detail in work by Markus Brückner and Antonio Ciccone.

What this implies is that unless one takes care of modeling the dynamics of GDP per capita carefully, one can reach any conclusion one likes from this pattern. Imagine, for example, aggregating the data into five-year intervals (as much of the previous literature does) and consider shifting where these five intervals start from in this figure. You will quickly convince yourself that one could easily find that democracy could be bad for growth (because GDP per capita is declining in the five-year interval in which democratizations are taking place).

Another major problem plaguing the previous literature is that many of the measures of democracy are ridden with measurement error. To deal with this problem, we follow work by Elias Papaioannou and Gregorios Siourounis, and constructed a 0-1 index of democracy using multiple sources, which minimizes the extent of measurement error (in particular by removing spurious movements in the democracy score of many low-income countries).

Using such an index, annual observations, country fixed effects (so as to control for various institutional and other country-level determinants of both democracy and economic growth) and econometric models that control for the dynamics of GDP per capita, we find quite well-estimated positive effects of democracy on growth.

In fact, once these 0-1 measures are used and the dynamics of GDP per capita are control for (even in a very rudimentary fashion), the positive effects of democracy on growth are very robust.

We also report instrumental-variables estimates, exploiting the fact that democratizations often occur in the form of regional waves (as noted by Samuel Huntington in The Third Wave). These estimates also show robust positive effects of democracy on GDP per capita of similar magnitude to the other models we report.

Our baseline estimates suggest that a country that democratizes increases its GDP per capita by about 20% in the next 20-30 years. Not a trivial effect at all.

Is there any evidence that democracy is only good for already developed economies? The answer is no. Though we do find that democratizations are associated with larger increases in GDP per capita in countries with higher levels of secondary schooling, there is no evidence that democracy is bad for economic growth in low income economies or even in economies with low levels of schooling.

In all, the evidence seems to be fairly clear that democracy is good for economic growth.

Why? This is a harder question to answer. Our evidence shows that democracies are better at implementing economic reforms, and also increase education. They also probably increase the provision of public goods (though the evidence here is a little less robust).

But none of this is conclusive evidence.

Our results from the two papers combined thus suggest an intriguing pattern: contrary to what many have presumed, democracy doesn’t have a huge effect on inequality. But also contrary to what seems to have been almost a consensus, democracy does have a robust and fairly sizable positive effect on economic growth.



Some brief comments on Ukraine.


The Benefits of Social Capital? Bowling for Hitler.

In our discussion of David Laitin’s explanation for why Basque nationalism became violent but Catalan did not, we pointed out that this is partially explained by the Basque country having the type of “horizontal” or “bonding” social capital that Robert Putnam and his collaborators argued in their famous work, Making Democracy Work, to have been crucial for promoting democracy and good governance.

In our research on the politics of the paramount chieftaincy in Sierra Leone with Tristan Reed, which we discussed last year, we found a great deal of evidence to be skeptical about this. In Sierra Leone, the evidence suggests that no matter how you measure social capital, it is negatively correlated with both less accountable local political institutions and economic development.

Another powerful example of the drawbacks of social capital is provided in the recent paper “Bowling for Fascism: Social Capital and the Rise of the Nazi Party in Weimar Germany, 1919-33” by Shanker Satyanath, Nico Voigtländer and Hans-Joakim Voth. The authors collected data on the extent of social capital in Germany in the 1920s as measured by the “density of associational life” in effect the presence of different social groups such as sports clubs, choirs, animal breeding associations, or gymnastics club. Their measure of social capital in a city is the total number of such associations per 1,000 people. They show that where social capital was higher, the Nazi party rose faster in terms of membership and it also recorded higher vote totals.

Historical evidence suggests that like ETA in the Basque country, the Nazi party were very adept at exploiting the possibilities provided by social capital to recruit new members.

One conclusion to draw from all of this work is not that Putnam was wrong. Indeed, his arguments are plausible in the Italian case and have received empirical support there (for instance, as shown in the paper “Long-Term Persistence” by Luigi Guiso, Paola Sapienza and Luigi Zingales).

Rather, the correct conclusion would be that the impact of social capital is highly heterogeneous and crucially depends on how it interacts with other aspects of a society’s institutions and politics.

Satyanath, Voightländer and Voth capture a bit of this idea since they show that in Prussia, which had stronger institutions, the relationship between social capital and the rise of the Nazi party is much weaker. They conclude

Our results therefore suggest that strong, inclusive institutions can keep the “dark side” of social capital in check.

Music to our ears, though obviously it’s not just inclusive institutions that keep the dark side of social capital in check, since the Prussian state of the 1920s that did control this type of social capital was far from inclusive. Clearly, there is more to the mystery of social capital.


Why is the Basque Country more Violent than Catalunya?

In our last post, we examined why Catalans in Spain had developed a distinct Catalan identity, to the point of pushing for an independent state, while those in France had not. Looking at the wider Spanish scene one sees the failure to create a Spanish identity more broadly. This is true not simply in Catalunya but in Galicia and perhaps most obviously in the Basque country.

But this observation raises other interesting questions because there is obviously a huge difference between the nationalism of Catalunya and that of the Basque Country: the former is not violent while the latter is. What could explain this?

Exactly this issue was addressed by political scientist David Laitin in his paper “National Revivals and Violence”. Laitin argued that in the Basque country, the nationalist movement ETA was very successful at tapping into local social capital and recruiting young men from small towns based on mountain-climbing clubs, called mendigoitzale, or youth gangs, or cuadrillas. In contrast Catalunya lacked these types of social groups and social capital. (This is related to the idea that social capital can be used for good or ill purposes, so its political implications are much more subtle than what one might at first conjecture, as we discussed in this post a while ago).

Laitin argues that, in the Basque country, small town life was dominated by patron-client ties to political parties and that economic groups, like trade unions were much more important. This provided a social base for political parties that were much more inclined to negotiate than fight.

Yet Laitin also argues that these different social structures only made recruitment easier in the Basque country, and they also leaned more towards violence in the sense that ETA, based on local social capital, did not have simple and effective channels of communication with Madrid, which could have facilitated negotiation (unlike the situation in Catalunya).

Paradoxically, what further precipitated violence was that it was “more costly” for people in the Basque country to become Basque nationalists, particularly from a linguistic point of view. Basque is a non-Indo European language and completely different from Spanish, while Catalan is very close. Hence in the Basque case it was more difficult to get local people to become nationalistic without using violence to ‘encourage’ them. In Catalunya it was much easier to speak Catalan, and moreover people probably also expected others to switch to speaking Catalan, something that was much less likely with Basque.

Laitin uses this language example as a bit of a metaphor for the fact that being Basque was harder than being Catalan, which meant that those trying to promote Basque nationalism were more prone to resort to violence.

The last piece of Laitin’s argument is more idiosyncratic. He also argues that early successes were important in sustaining the violent strategy of ETA. ETA commandos assassinated Luis Carrero Blanco, the Spanish Prime Minister and heir apparent to Franco and ETA membership quickly doubled in its wake. The execution of two ETA prisoners in 1975 yielded a general strike and turned the victims into martyrs, further boosting support for ETA. 

In the end Laitin’s analysis blends structural factors, the different nature of social capital, the different cost of being Basque, with idiosyncratic shocks which allowed particular violent strategies to consolidate themselves and appear successful.


What’s the Problem with (Spanish) Catalunya?

Most of the research by economists on the state has focused on a quite narrow subset of things the state does. For example, Daron’s paper “Politics and Economics in Weak and Strong States” on the ability to raise taxes as the key dimension of whether a state was weak or strong. This was subsequently a key theme in the 2011 book Pillars of Prosperity by Tim Besley and Torsten Persson who focused on the development of fiscal systems to raise taxes and legal institutions to efficiently enforce contracts.

In our work with Rafael Santos on Colombia, “Monopoly of Violence: Evidence from Columbia, we focused on the creation of a monopoly of violence. (We will also soon report on some other new work of ours on the economic consequences of state capacity and state building).

But all of this research ignores one of the most fundamental things that modern states do — to create a national identity eliminating alternative identities.

One of the reasons, for example, that modern Tunisia is so different from neighboring Libya is that after independence the first president, Habib Bourguiba, invested heavily in creating a national identity. Instead, the rulers of Libya, such as Muammar Gaddafi, exploited and exacerbated different identities as a way to stay in power.

Modern states obviously differ a lot in how successful they have been at creating national identities, with huge consequences, as the previous example shows (it is Libya, not Tunisia, that is tottering on the verge of civil war at the moment). This is even true within Europe, a place that we tend to think of as having very successful consolidated nation states. Take the case of Spain and France. Though some people in the southwest of France still speak Occitan and grumble about being controlled from Paris, the reality is that the French state did a very effective job at forging a national identity (the locus classicus on this process is the 1976 book Peasants into Frenchmen by Eugen Weber which we mentioned in an earlier post, where we discussed why the English had been so bad at obliterating the identity of the Scots – who may ungratefully declare independence this year).

In recent research Laia Balcells, a political science professor at Duke University, has investigated the differences between Catalans south of the Spanish border and those north. In her paper “Mass Schooling and Catalan Nationalism”, she points out that Catalunya was split between France and Spain by a treaty in 1659. The Catalans had a distinct history with their own language yet the extent to which people identify as Catalan today differs greatly north and south of the border.

In Spanish Catalonia, Balcells shows that Catalan is the main language of communication between members of the family for 37% of the population; also, 7% say that Catalan is not the only language, but that it is more usual than Spanish.

In French Catalonia, in contrast, only 0.5% of the population speaks Catalan within the family: French is the main language in family communications for 87.6% of the population. This use of language is one way of seeing the different facts about identity.

The question then is this: why is there this divergence with Catalans in Spain currently demanding a referendum on becoming an independent country while nothing of the sort is taking place in France?

At some loose level this is obviously connected with what Eugene Weber was writing about. The French created a very effective top down state and socialized everyone into being French, particularly through the educational system where French was the only language which could be used.

Balcells does not dispute this claim but makes a more subtle argument. She puts it in the context of “nationalistic revivals” which take place in three phrases:

  1. Phase A which she calls “scholarly interest” led by intellectuals who discover and celebrate some lost or repressed identity
  2. Phase B which she describes as “patriotic agitation” where people become much more aware of the issues and more general nationalists sentiments surface;
  3. Phase C, finally, is the “rise of a mass movement” where collective action for national recognition and even independence takes place.

Balcells’ argument is that what is critical is the interaction between these social dynamics and what she calls, following the terminology of political scientist Keith Darden, a “scholastic revolution”. This corresponds to the first generation of people to receive mass education and when a community first shifts from an oral to a literate mass culture. The important point about France was that the strong state was around at the time of the scholastic revolution that meant that Catalan nationalistic sentiments got no air time in school.

In Spain, where the central state had not effectively exerted itself in the periphery, mass schooling arrived at just the moment when Phase B was in full flow. In consequence the state could not control the teaching of Catalan nationalism in school. Hence the first generation of people who received schooling was socialized into being highly sympathetic to the cause of Catalan nationalism. In Darden’s view the “scholastic revolution” is a “critical juncture” in identity formation that then massively persists over time even in the face of efforts to repress it. Balcells shows this is precisely what happened in Catalonia during the dictatorship of Franco.

The critical factor causing the divergence between the north and south of the French-Spain border in terms of Catalan nationalism was that the Spanish state was weak at exactly the wrong time — when the scholastic revolution interacted with a wave of patriotic agitation.