Urban Violence or Urban Peace: Why Are Some Cities Safer than Others?

By: Claire Mc Evoy and Gergely Hideg

Cities continue to increase in importance, acting as magnets for migration, innovation, and economic concentration. In 2015, almost four billion people lived in cities, a number that’s expected to increase to five billion by 2030. The UN has noted that challenges for growing cities include growing slums, increased air pollution, and heightened risks of disasters for the population. But what about violence? City dwellers are often perceived as being at more risk of violence than those in rural areas, but this is not always the case.

In 2016, at least 560,000 people were killed violently, which corresponds to about 7.50 violent deaths per 100,000 population. Cities often present significant risk factors that encourage violence. These include mass unemployment, gang violence, weakened security institutions, organized crime, firearms proliferation, limited government capacity, and rising inequality. Seventy-five per cent of the world’s cities have higher levels of income inequalities than two decades ago, and research shows that urban violence generally concentrates in areas of strong disadvantage, social exclusion, and poverty. In addition, the size of cities concentrates perpetrators and victims of violence in confined spaces.

Having said that, cities also offer ‘protective’ factors that can help to prevent violent criminality and reduce its impact — such as more policing, street lighting, services, infrastructure, medical facilities, and prevention programmes.

Setting the scene: what do we know about what we know?

To support the New Urban Agenda and the 2030 Agenda for Sustainable Development (in particular, Goal 11 on inclusive, safe, sustainable cities), the Small Arms Survey’s Database on Urban Homicides — containing data from 928 cities (see Endnote 1) worldwide in 76 countries with an aggregated population of more than one billion people (see Endnote 2) — aims to provide timely data on emerging urban issues.

Map 1: Coverage of the Small Arms Survey’s Database on Urban Homicides

Source: Small Arms Survey Database on Urban Homicides

As can be seen on Map 1, coverage of city-level data on homicides is patchy or unavailable in many countries, particularly in Africa, Eastern and Southern Asia, and Oceania — making global and regional analysis particularly challenging.

In countries where city-level data is available, several factors affect the data:

· Inflated/deflated numbers. City homicide numbers may be inflated due to better data-gathering systems in urban — as opposed to rural — areas. Numbers may also be underreported in cities where large geographical areas — such as informal settlements — are outside the authorities’ control.

· Different definitions of city. Countries use different definitions of ‘city’ to gather crime data. These may be restricted to a district in the city. i other cases, the reporting unit is the actual city; while fairly often data is also provided for the urban agglomeration surrounding the city. Even within countries, reporting units can vary. Most data from Buenos Aires in Argentina is from the city itself, for example, while in Mendoza, homicide data covers a much wider urban area.

· Unrealistic population figures. Reported city homicide rates based on incidents per 100,000 population may be too high because they are not adjusted to count ‘commuter-adjusted populations’ or visitors who increase a city’s (day-time) population.

· Different categorization of crimes. Depending on where and in what context a person is killed, a violent act may be considered ‘justified’ or ‘lawful’ for cultural, ethical, political, or legal reasons, and may therefore not appear in homicide statistics. The Small Arms Survey estimates that approximately 3 per cent of violent deaths at the global level are due to ‘legal interventions’, which are the ‘killing of civilians by law enforcement officials, or killings of law enforcement officials on duty’. Urban settings may affect this category. For instance, urban law enforcement operations in some countries may result in large numbers of people being killed. The 920 killings by police in Rio de Janeiro in 2016 resulted in a police-inflicted death rate of over 14.5 per 100,000 population. This rate is higher than the total violent death rate in 185 of the 222 countries worldwide. Because these deaths are recorded separately it may be difficult to assess the full scope of these ‘interventions’.

· Capacity gap. The capacity gap between more and less developed countries in terms of accurately recording urban homicides — or recording them at all — negatively affects the cross-national comparison of urban crime statistics.

Therefore, the interpretation of data on city homicides must proceed carefully. While recognizing these limitations, analysis of the Small Arms Survey’s city data has yielded a number of interesting results. As a starting point, Table 1 provides an overview of the number of cities covered in the database, per category of city size.

Table 1: City size categories and number of cities covered by the Small Arms Survey’s Database on Urban Homicides

Source: Small Arms Survey Database on Urban Homicides

And what do we know?

The dataset confirms that in a significant number of countries (49 out of the 76 included in the database, or two thirds), the urban homicide rate is indeed higher than the national one (see Figure 1, which shows the difference between urban and national homicide rates). In some otherwise low-risk countries, the urban crime rate is considered high or extremely high. Examples include Greece (with a homicide rate of 12.9 per 100,000 population in Athens, compared to an average of 0.8 homicides per 100,000 nationally) and Belgium (with a rate of 20.7 per 100,000 in Brussels, compared to 1.9 nationally).

However, in one-third of the countries represented in the database the urban crime rate is either close to the national figure or below it. Notable examples of this trend are Ireland (with a homicide rate of 0.3 per 100,000 population in the Dublin metropolitan area, compared to 1.5 nationwide), and large cities in Turkey: Istanbul, Ankara, Izmir, and Bursa have an average rate of 1.2 homicides per 100,000 population, compared to 4.3 in Turkey in general.

City size and the risk of homicide

Large cities are generally safer. People living in them are — relatively speaking — least likely to be exposed to extremely high levels of violence (a homicide rate of 20 or more per 100,000 population). Indeed, only 10 per cent of large-city dwellers inhabit cities experiencing extremely high violence. Eighty-six per cent of large-city dwellers live in cities that experience fewer than 10 homicides annually per 100,000 residents.

In contrast, many small towns (those with fewer than 250,000 residents) and especially small cities (250,000–500,000 residents) are much less safe: 35 per cent of residents of small towns and 44 per cent of residents of small cities face a high or extremely high risk of lethal violence (10 homicides or more per 100,000).

Generally, the data indicates that the larger the settlement, the lower the homicide rate. Small towns experience an average homicide rate of 14.6 per 100,000 population, while in large cities the rate is 7.5 homicides per 100,000.

But as we know, correlation does not necessarily imply causation. Many factors affect patterns of violence in cities, including changes in socio-economic and demographic factors, migration flows, prevention programmes, security reforms, security responses, and access to local resources.

And the relationship between city size and homicide rates is not consistent across regions. For example, in Central America the database shows the homicide rate is lower in cities with a population of over 1.2 million than in small cities. Yet in South America, the opposite is true (see Figure 2).

Figure 2: Homicide rate by city size, globally and in selected regions

Source: Small Arms Survey Database on Urban Homicides

Previously published research has indicated that some cities that are growing rapidly — with some exceptions — are more prone to violence than those that are not. However, globally, the Small Arms Survey has not found evidence that world regions with a higher urban growth rate consistently display a higher homicide growth rate than those with slower urban growth. The relationship appears to hold true in some world regions, but not in others.

Figure 3: Population living in settings of different violence levels (defined by homicide rate categories), by city size

Source: Small Arms Survey Database on Urban Homicides

Figure 3 shows that large cities — those with more than 1.2 million inhabitants — are home to more people facing extreme violence. This is because they concentrate such large populations, not because they are less safe. Overall, extremely high homicide rates are found only in 21 of the 215 large cities in the database.

And what about the women?

In 2016 about 18 per cent of globally recorded homicide victims were female. Sex-disaggregated data on homicide victims is available for only 177 cities (19 per cent of the total).

In these cities, about 20 per cent of homicide victims are female, which indicates that, generally, the proportion of female victims tends to be slightly higher in more urbanized settings. One notable exception is the proportion of female victims in large cities, at 13 per cent, which is well below the average (Figure 4).

The proportion of female victims varies according to the overall level of violence and the size of the city. In countries or territories with relatively low rates of violent deaths (3 per 100,000 population or fewer), the female proportion of homicides is generally much higher in urban settings, whatever the size of the city, reaching 40 per cent in big cities (500,000–1.2 million inhabitants).

In contrast, where populations experience extremely high levels of lethal violence, the proportion of female victims is the lowest — below 10 per cent — with a minimum of 7 per cent in large cities.

While the underlying causes of female homicide often lie in unequal power relations, particular risk factors accompany urbanization. These include gendered urban poverty, a lack of sanitary facilities, cramped living conditions, chronic unemployment and poverty that fuels domestic violence, including intimate-partner violence, alcohol and drug abuse, fragmented social relations, and a lack of prevention programmes and policing in informal settlements.

Figure 4: Proportion of females among homicide victims, by size of city and violent death levels in the country

Source: Small Arms Survey Database on Urban Homicides

At the same time, large cities also provide opportunities for work and access to education, services and facilities, which may help to protect women and girls from violence.

Understanding urban violence better

As cities become increasingly attractive, their expansion needs to be carefully managed to ensure that safety forms part of that attraction; and there are a number of challenges related to rapid and uncontrolled urbanization growth in many parts of the world.

Cities are not necessarily more violent than rural areas but they can easily set the scene for situations that encourage violence. Better monitoring of urbanization would help understanding its various dimensions, including violent ones. Unfortunately, authorities in many countries — even where urban violence is endemic — have a poor grasp of it, and if the problem is not diagnosed, it will be very difficult to solve it.

Looking ahead, it is important to monitor, analyze and understand such violence, which is unevenly concentrated both geographically and socio-economically. Only then can it be tackled.

Building on existing successes and the sharing of best practices on violence prevention and reduction, especially through South-South learning, will be critically important in order to manage urban violence. The global spotlight on improving safety in urban areas is yielding valuable lessons learned that can be adapted to local conditions. The High Level Political Forum in July 2018, when progress will be officially reviewed on SDG 11, is one such opportunity to both share and learn.

The experiences of cities and towns in developing countries are particularly important, as this is where the challenge is greatest. While all regions are expected to continue to urbanize, developing countries in Africa and Asia are urbanizing faster than others, many without parallel industrialization. In the process, tracking statistics — including crime rates, is hardly perceived as a priority. A radical shift is needed so that data on urban violence, including homicides, is routinely gathered and analyzed in support of the New Urban Agenda and Agenda 2030.

The research behind this blog post and the unpublished background paper on urban violence was carried out by the Small Arms Survey with support from the Swiss Agency for Development and Cooperation.

Disclaimer: Blog posts are intended as a way for various Small Arms Survey collaborators and researchers to discuss issues of small arms and armed violence, and do not necessarily reflect the views of the Small Arms Survey, nor the Government of Switzerland.


1. Most of the data — 78 per cent — is from 2015 or 2016, although some dates back to earlier years.

2. Global estimates of urban populations differ greatly, including due to the use of different definitions of ‘city’. The United Nations Statistical Division reported that there were 3,732 cities with a population of at least 100,000 globally in 2015, with an aggregated population of about 1.6 billion people. The World Bank estimated the global urban population to be just over four billion people in 2016.

Useful resources:

Centre on Conflict, Development and Peacebuilding. 2012. Graduate Institute. ‘CCDP (Centre on Conflict, Development and Peacebuilding). 2012. Urban Resilience in Situations of Chronic Violence. Geneva: Graduate Institute of International and Development Studies.

Frost, Emilia and Matthias Nowak. 2014. Inclusive Security, Inclusive Cities. Policy Paper 1. Geneva: Geneva Declaration Secretariat. March.

Geneva Declaration Secretariat. 2015. Global Burden of Armed Violence 2015: Every Body Counts. Cambridge: Cambridge University Press.

ICRC (International Committee of the Red Cross). 2016. War in Cities: International Review of the Red Cross. Vol. 98. №901. April.

Mc Evoy, Claire and Gergely Hideg. 2017. Global Violent Deaths 2017: Time to Decide. Small Arms Survey report. Geneva. December.

Milliken, Jennifer. 2016. ‘Urban safety and security: Lessons from the last two decades and emergent issues.’ Synthesis report of the conference Reviewing the State of Safety in World Cities: Safer Cities +20. Geneva, 6–8 July. Geneva Peacebuilding Platform. Paper №22.

Nowak, Matthias. 2012. Urban Armed Violence. Research Note №23. Geneva: Small Arms Survey. November.

Small Arms Survey. 2007. Small Arms Survey 2007: Guns and the City. Cambridge: Cambridge University Press.

Quero, Yann-Cédric. 2017. Homicide and Cities. Unpublished background paper. Geneva: Small Arms Survey.