Jobs, unemployment and violence

(This 10-minute presentation accompanying the reading pack was recorded at the GSDRC Seminar on Conflict and Development, London, 23 March 2015.)

The notion that unemployment is a strong probable cause or motivating factor behind violence and violent conflict is remarkably pervasive in international development. It is believed that unemployment triggers participation in insurgencies, prompts people to join violent gangs, drives people to extremism, and that it is the primary reason behind domestic violence. More remarkable is that this idea is based more on intuition and assumption than on evidence.

A good starting point for a discussion of this assumed relationship is the assumption that rational calculations of opportunity cost matter most in driving people to join armed groups. The argument that unemployed people (especially young men) are those most likely to be recruited voluntarily runs through much of the influential ‘economic perspective’ on violent conflict. It was one of the main pillars of Paul Collier’s early versions of his models of the causes of civil wars, and it was a key feature of Collier’s argument about the key causes of sustained post-war peace.

The argument is that unemployed young men have a low opportunity cost for engaging in violence and joining armed groups. Therefore, where there is high unemployment, especially among young males, this is one of the main factors ‘predicting’ civil war incidence. But since it is so difficult to measure young male unemployment in low-income countries, this was often measured indirectly, via a proxy: average years of education.

In Collier’s later papers on post-conflict reconstruction, he argues that the economic growth rate is the single most (statistically) influential variable in determining whether a country returns to war within a few years of a peace settlement. Employment, he says, is a likely route through which growth might affect the risk of violence. Again, then, an assumption is made that growth translates into employment, and that increased employment reduce s the risk of civil war recurring.

These arguments – and the reasoning they are built on, together with the evidence they present – are important to understand and to discuss in thinking through the assumed relationship between unemployment and conflict. For this reason, Paul Collier’s pieces are important first readings to inform a discussion.

Reading 1: Collier, P. (2000). Doing well out of war: an economic perspective. In M. Berdal and D. Malone (Eds.), Greed and grievance: Economic agendas in civil wars (pp. 91-112). Boulder, CO: Lynne Rienner. Available as a conference paper (1999).

Reading 2: Collier, P., & Hoeffler, A. (1998). On economic causes of civil war. Oxford Economic Papers, 50(4), 563-573.

Reading 3: Collier, P. (2007). Post-conflict recovery: How should policies be distinctive? Oxford: Centre for the Study of African Economies, University of Oxford.

However, there is rather little direct evidence to confirm that unemployment has played as important a role as this kind of economic logic would predict. This is partly because data on unemployment across many low-income countries, and even more so data on youth unemployment, are often highly unreliable and a poor basis, especially, for cross-country comparisons or regression analysis. Labour market data in developing countries have many problems, including infrequent labour force surveys, and survey questionnaires whose design is based on inappropriate OECD economic structures. An example of an analysis showing the problems specifically in rural labour market statistics in Africa is Carlos Oya’s article:

Reading 4: Oya, C. (2013). Rural wage employment in Africa: Methodological issues and emerging evidence. Review of African Political Economy, 40(136), 251-273.

The implication is that we have to be very wary of leaping to conclusions about the role that labour markets, employment and unemployment play in violent conflict (and this extends to the study of other forms of violence such as terrorism, membership of and participation in violent gangs, and indeed intimate partner violence). It also puts the onus on researchers, development agencies, and policy makers to engage with a wider array of carefully produced types of evidence. As an increasingly sophisticated literature on violent conflict in developing countries has suggested, there is in fact evidence that it is not simply opportunity cost, or, directly, unemployment, that motivates individuals to participate in collective (or private) violence.

A particularly interesting example is the work of Francisco Gutierrez Sanin on the FARC in Colombia. The FARC present what may look like a rational choice puzzle: for many years recruitment was voluntary and at high levels, but there were no direct material rewards to joining the FARC and the risks of death were higher than not joining, or than joining alternative militia groups. Gutierrez Sanin has published a number of articles on this issue, but one that nicely draws on multiple sources of data (recovered CD-ROMs with socio-economic data on members of one part of the FARC, judicial records, etc.) addresses the employment/unemployment issue directly. It finds that many recruits to Colombian armed groups were not only employed at the time of joining, but were paid above average wages.

Similarly, Berman, Callen, Felter and Shapiro (2009) argue – on the basis of panel data for Iraq and the Philippines – that unemployment does not predict participation in political violence. They find that if there is an opportunity cost effect it is not dominant.

There is also little decent evidence on the extent to which the employment-related schemes that often play some part in post-war Disarmament, Demobilisation and Reintegration programmes actually make a substantial difference. One interesting exercise that has produced some limited evidence worth considering is Chris Blattman and Jeannie Annan’s recent working paper reporting on an evaluation of the (self-)employment, incomes, and socio-economic behaviour outcomes of an agricultural training programme for Liberian ex-fighters. And in a useful review of donor approaches to addressing armed violence through youth employment programmes, Oliver Walton (2010, p.1) finds that ‘both the theoretical and empirical cases for using youth employment programmes as a stand-alone tool for reducing violent conflict are extremely weak.’

There is a parallel literature on the ‘economic perspective’ on terrorism, and also a critical literature on this approach. One example of an argument, based on evidence (albeit subject to some criticism), that poverty and unemployment are not the main motivators of participation in ‘terrorist’ attacks is Alan Krueger’s work.

Reading 7: Bourgois, P. (1989). Crack in Spanish Harlem: Culture and economy in the inner city. Anthropology Today, 5(4), 6-11.

One point that emerges clearly from Bourgois (and that is a feature of a wide range of literature on gangs, insurgent groups and other organisations of collective violence) is that it is impossible to understand the role of employment, unemployment, labour markets and labour institutions in collective violent conflict without seeing labour within the broader relationship between institutions, policies, economic trends, and social relations. An example of a study – one among a number coming to varied conclusions, it must be said – of xenophobic violence in South Africa in 2008 that argues that the causes were more political than straightforwardly economic (or that could not be just ‘read off’ from economic indicators) is Wa Kabwe-Segatti and Fauvelle-Aymar.

Reading 8: Wa Kabwe-Segatti, A., & Fauvelle-Aymar, C. (2011).People, space and politics: An exploration of factors explaining the 2008 anti-foreigner violence in South Africa. In L. B. Landau (Ed.) (2011), Exorcising the demon within: Xenophobia, violence, and statecraft in contemporary South Africa (chapter 3). Johannesburg: Wits University Press.

Labour markets, employment status, and labour relations often do matter for violent conflict. But they matter in significantly varying and complex ways. And there is still far too little good large-scale statistical data available to assess this at the large-N comparative level. There is also still scope for much more research using different methodologies. For one attempt to assess the state of the literature and the available evidence, see Cramer (2010).

Reading 9: Cramer, C. (2010). Unemployment and participation in violence (World Development Report 2011 background paper). Washington: World Bank.

Questions to guide readings

  • What are the theoretical foundations of the idea that unemployment is likely to be associated with a higher statistical incidence of violent conflict or participation in violent gangs or terrorism?
  • How good is the evidence on employment and unemployment levels in the countries where many armed conflicts take place? Are these data problems so serious that they undermine confidence in otherwise plausible claims about the relationship between labour markets and violence?
  • Are there grounds to think the link between unemployment and participation in violence may be far less straightforward? What is the reasoning behind different possible causal connections? What do different kinds of evidence suggest?
  • To what extent – and why – should donor agencies and governments prioritise post-conflict (especially youth) employment programmes to stabilise peace?


Send feedback on this reading pack