Horizontal Inequality

In the last post, we heard Kendra Dupuy talk about different theories of why conflict, and especially civil war, begins. Common approaches involve looking at greed, grievances, weak states, organizational structures and other factors, but these mostly focus on how individuals are prompted to pick up arms against the state.

Today, Solveig Hillesund, a professor at the University of Oslo, is here to talk about group dynamics. And, in particular, about a phenomenon known as horizontal inequality and how it can improve our understanding of how and why wars start.

This isn't going to be a seminar about inequality generally, defined here as the unequal distribution of a resource, but about how perceptions of inequality can mobilize people to violent acts. We're also going to be expanding the definition of 'resource' here to not only include income, wealth or land, but also education, health care access, social services or political influence.

One way of talking about inequality is talking about how individuals in a society measure up against each other. If you take all the people in a country, town or demographic group (Society A) and rank them for some factor (wealth, mobility, life expectancy), the corresponding graph might look like this:

If we're measuring for wealth, then this would describe a society with a relatively large middle class, with a few people who are significantly richer and a few people who are significantly poorer. This is known as vertical inequality, as everyone is being compared to everyone else on a single, vertical axis.

Let's say that you compare the vertical inequality in Society A with Society B. Society B has recently seen the recurrence of armed conflict, and researchers might want to investigate whether or not there's a statistical difference between them when it comes to the measurable factor (in this case, total wealth). The comparison would look like this:

There are some differences between the two societies, with a slightly smaller middle class in Society B and greater degrees of wealth inequality. But, for the most part, the pattern doesn't look exceedingly different than in Society A. Looking only at vertical inequality, researchers might then conclude that wealth discrepancies aren't a relevant factor and then go on to look at something else.

This is something that happened in the academic literature – people started looking at how much income goes to the top ten percent (or one percent) in a given society and tried to measure for the likelihood of a war starting. The first studies were made from the 1960's to the 1980's and gave mixed results. For some, inequality was positively related to conflict, for others it turned up negative (or even in a U-shaped graph).

Researchers were criticised for having small or even biased samples and, generally speaking, grievance-related theories (and inequality is a grievance) were going out of style in favour of studying factors like weak states, geographical terrain and others.

But there were other indications that suggested that it was still too early to put the inequality-debate to rest. For one, while the number of total conflicts was going down, for example, the percentage of them that were ethnically-based was rising.

There's also how some researchers felt that the typical, economics-based solutions to the collective action problem didn't hold up.

As mentioned in the last seminar, the collective action dilemma refers to how people need a reason to put aside their personal interests for the sake of the group and do something risky. This is especially the case in the context of revolutions or civil wars, because you always have freeloaders who don't take risks and will still benefit if the revolution wins. And if you can get benefits without risking, why risk anything at all? Theories that explain why conflicts start have to give a compelling reason why people take these risks.

Economic-based explanations (like greed) say that people are getting paid in shelter, drugs, money or what have you, but that doesn't take into account more psychological factors like how community, anger, collective identity, solidarity and other non-material rewards can prompt people to fight the government. People can become criminals for economic reasons (as individuals), but to form groups capable of armed fighting takes a reason that holds them together.

For certain researchers, there was still a missing piece that wasn't showing up in the data. So they tried assembling the data a little differently.

What if you take Societies A and B and measure wealth inequality not in terms of everyone, but by separating them into ethnic, class or religious groups? When you do that, Society A might look like this:

Three separate groups within one society that are relatively equal with some variation. But what if we take Society B and the data shows something entirely different?

Society A, on the left, shows relative equality between the respective groups, but the same groups in Society B, once separated and displayed alone, are subject to far wider economic disparities. This is what's called horizontal inequality, or inequality between groups as compared to between individuals.

While vertical inequality didn't seem to correlate with armed conflict, horizontal inequality proved a much more relevant factor and is now a standard way of analyzing societies at potential risk of armed conflict.

You don't just have to study wealth, too: certain groups may have less access to education (due to language or geography), less representation in government (due to class or religious issues), or something else. The bigger the gap, it turned out, the bigger the possibility that someone's going to reach for a gun.

But how do you study something like this, and what factors do you look for when seeking horizontal inequalities in a society?

When we talk about inequality among groups, there are three questions that researchers are interested in:

Leadership, and the way that cultural leaders frame issues, is another important factor in the literature. This is particularly true when leaders interpret or reinterpret a group's history, circumstances or present situation in a way that one particular identifier becomes emphasized and a grievance becomes salient. Think about a society that perceived itself as divided by economic status but then being encouraged by certain groups to think about themselves in terms of race or religion.

This grievance or group identifier might not have been relevant even a few years before a leader mobilizes the population, but one its in play it can create narratives among people that support collective action. This is known in the literature as collective action framing.

Through this act of reframing, new salient groups can be formed and destabilize a country's order.

We don't know all the ins and outs of this process, but it has been observed that there are at least four steps though which horizontal inequalities can lead to salient grievances and group dynamics:

These steps can generate a grievance that can, provided the right circumstances (resources, weak states, access to weapons, etc), lead to a civil war if certain claims are made and then not met, or are met by repressions from the government. After this, groups may be mobilized based on networks with a shared emotional framework.

When preparing to measure for horizontal inequality, in addition to asking the three main questions above, you need to think of how you collect and compare the data. What group identity markers you select to measure for is key, as a factor that might appear at first to be salient might not be. You might also try measuring for salient factors across different countries.

One place to start is with survey and census data. There's an initiative known as the Demographic Health Survey (DHS) program that conducts comprehensive surveys in many countries, and in some they measure for ethnicity or other variables. You can take data like this, plug in different factors and compare groups to see if inequalities emerge.

There are also local datasets in different countries that you can use. But there's always the issue that you're depending on the people who designed the survey, and if that organization failed to measure for something salient then their data likely won't reflect it either.

You can also try to figure out if there's a proxy for salience, like political relevance, representation in parliament, officially recognized language status and the like. These might still leave out relevant group divisions, but it could be a good place to start.

There are also NGOs or international research programs, like Minorities at Risk, that also look at social as well as political relevance.

When looking for political factors, you also have to be aware of nuances like tokenism. Is a group actually represented in parliament, for example, or are there merely a few powerless MPs that give the impression of representation? You can also look at representation as compared to group size or general demographics.

Some proxies for socioeconomic development and inequalities could be education rates, health or infant mortality. Economic proxies could look for whether people have water in their homes, or TVs and other kinds of assets as compared to other members of that society. Then, of course, there's GDP per capita within specific groups.

Measuring for political inequality has produced some solid data and correlations, as well as for economic factors relating to deprivation (but not as much for privilege, which has generated more mixed evidence). Social indicators are still something new and there haven't been enough (or diverse enough) studies to create solid recommendations just yet.

Group identifiers with a growing amount of evidence include geographical region, migrant status, age and gender. While these are promising developments, the precise nature of these links is still being analyzed, as are their links with structural conditions.

Inequality within groups is also a new factor that's currently being studied, but so far there are only mixed results.

And while links are being shown in certain contexts, it's another thing to prove causality. In other words: is conflict causing the inequality, or is the inequality causing the conflict? Sometimes it goes both ways and it's hard trying to pin down the dynamics exactly.

When it comes to making policy implications from this kind of data, there's the temptation to say "reduce inequality and we get less conflict!" But it's more complicated than that. You might get less violence from underprivileged and deprived groups, but maybe privileged groups might take up arms to protect their resources.

That, and when you address inequality and try to even things out in one way, you might trigger another problem somewhere else. So it's complicated. And, even if you do find some great evidence and make solid links, all you'll be able to bring to the table is a probability model: if you make a concrete change in society, then the probability of a conflict goes down. Nothing's for certain.

And people like certainties, especially politicians who make decisions. But that is not, Solveig reminds us, how social sciences work. We make do with what we have, test our models, look for the missing pieces and try, in the end, for a less-incomplete picture of the world.