Many think that development is mainly hampered by what is known as a “poverty trap”. Farmers do not buy fertiliser even though they know it will produce a better harvest. If there is no road, they reason, their bumper crop will just rot in the field. The way out of such a trap is to build a road. And if poor countries cannot build it themselves, rich donors should step in.

Yet the World Development Report suggests that the main constraint on development these days may not be a poverty trap but a violence trap. Peaceful countries are managing to escape poverty—which is becoming concentrated in countries riven by civil war, ethnic conflict and organised crime. Violence and bad government prevent them from escaping the trap.

To see the impact, compare two small African states. Until 1990 Burundi and Burkina Faso had similar rates of growth and levels of income. But in late 1993 civil war erupted in Burundi after the assassination of the president; 300,000 people died in the next dozen years, most of them civilians. Placid Burkina Faso is now two-and-a-half times richer.

That may sound like a special case. Civil wars are obviously damaging, and not many countries suffer them. True, but a lot of others are trapped in persistent, pervasive lawlessness. The report reckons that 1.5 billion people live in countries affected by political violence, organised crime, exceptionally high murder rates or low-intensity conflicts. All this falls short of civil war, but the effects can be as bad.

Many of these people are caught in cycles of violence. Almost all the 39 countries which have suffered civil wars since 2000 also had one in the previous three decades—something that was true of far fewer in the 1960s. Moreover, “lesser” forms of violence are worsening to a point where they can be more deadly than civil war itself. In Guatemala, more people are now being murdered each year (mostly by gangs) than were killed in the country’s civil war in the 1980s (see article).

This is more remarkable because “traditional” state violence is subsiding. Although the world’s population has expanded since 1990, numbers of interstate wars, civil wars and coups have fallen—as have the numbers of deaths in them. Fewer countries suffer large-scale violence, but the ones that do suffer repeatedly.

As a result, people in these countries are more than twice as likely to be malnourished, three times as likely to miss primary school and almost twice as likely to die in infancy as people in other developing countries. They are also more vulnerable to shocks. Protests during the 2007-08 food-price crisis were more frequent and more likely to turn violent in countries with the most fragile governments.

The gap between the two sets of countries is widening. Nearly all have cut infant mortality since 1990. But conflict-ridden states have reduced it by only 19% compared with 31% elsewhere. No poor, violent country has achieved a single one of the millennium development goals (MDGs), targets set by the UN in 2000. As a rule of thumb, concludes the report, countries that suffer large-scale violence lose almost 1% in poverty reduction each year.

But perhaps these countries are violent because they are poor, rather than poor because they are violent? To some extent this is true. As a rebel leader in South Sudan once said, life is so cheap “It pays to rebel”. Growth would presumably reduce the incentive to fight. To test the importance of wealth, the authors of the report asked why young people joined gangs and rebel groups in half a dozen countries. The biggest share, about two-fifths, cited unemployment as the main reason; only a tenth said belief in the cause (in contrast, belief was cited as the main motivation half the time for members of militant Islamist groups in Mali and the West Bank).

But even if poverty is a cause of violence, it is not the only one. Legitimacy of government also matters. The report tries to disentangle its effects from income by looking at countries with similar incomes and identifying those that stand out for governance. It finds that countries with good governance are much less likely than their peers to have suffered from civil conflict or high murder rates in 2000-05. Perhaps unsurprisingly, Libya scores worse on indicators measuring government accountability and tribal loyalties than Egypt or Tunisia—which may help explain the course the Arab spring has taken there. The more personal a government, the harder it finds to change.

The implications of this analysis are wide-ranging. First, it suggests that preventing violence should be given much higher priority than it is now. At the moment, the MDGs that guide as well as measure development do not even mention things like justice and people’s security.

Next, countries should learn from the mass of evidence about what works to reduce violence. Quickly restoring people’s confidence in government is key. This can be done by getting a reasonable number of conflicting parties to sign up to a peace deal or otherwise break with the recent past, as Ghana did in 2003. Governments also need to signal good intentions straight away by making credible appointments (as Nigeria did in picking its new electoral commissioner). And they need some quick results, like new jobs. But, as the report says, “development assistance is easier to obtain for macroeconomic policy, health or education…than for job creation.”

Third, outsiders should stop treating new conflicts as if they were traditional interstate or civil wars, with clearly defined roles for diplomats, soldiers and human-rights or aid workers. People need to act together, the report says—though most aid workers would not be seen dead working with a police or army officer.

Lastly, people need more patience—a lot more. It took the best-reforming countries since 1985 27 years to reduce corruption to acceptable levels. Few involved in post-conflict settlements are willing to wait that long. Haiti tried to create an effective government in 18 months. Naturally, it failed. The moral is that the countries of the Arab spring are likely to face multiple transitions, not just one. Alas, it is perhaps easier to urge patience if you are a World Bank economist than if you are the leader of an impatient or violent country under threat from your people or your rivals.