Working papers & reports

Abstract: Prevalent counterinsurgency theories posit that small development aid projects can help stabilize regions in conflict. A widely assumed mechanism runs through citizen attitudes, often known as “winning hearts and minds.” In this formulation, aid leads to economic benefits and sways public perceptions about the government, leading to more cooperation and eventually to less violence. Following a pre-registered research design, we test this claim using a difference-in-differences approach. We leverage original survey data and new geo-coded information about small German-sponsored infrastructure projects in northern Afghanistan. We find that aid improves perceived economic conditions, but erodes attitudes towards government and improves perceptions of insurgents. This finding challenges the “heart and minds” theory, but comple- ments the wider literature on legitimacy in developing states, suggesting that providing public goods improves legitimacy only when done transparently and accountably. We test alternative mechanisms, finding evidence that aid gatekeeping and credit claiming by insurgents are occurring.

Abstract: While we have a good understanding of how states become fragile, we know next to nothing about how states exit from fragility, or even which states exited. In this paper we provide a novel measure for reliably identifying states which exited fragility. Our measure is derived from a principal component analysis of the World Governance Indicators, which we apply against various threshold bands. When using the threshold band with the highest concept validity, we find that, since 1996, only eight countries made it out of fragility, and only one of these is a developing poor country of the global South. The other seven are either post-socialist, or middle-income, countries, and they are structurally vastly different from the majority of fragile states. Our results suggest that existing fragility appears to be associated with being a middle-income country and being able to reconnect to historically solid state institutions.

There are 14  sub-reports to this summary paper:

On South Sudan

South Sudan Part I Summary Paper"

South Sudan Part II Country Level Evaluations"

South Sudan Part III Program and Project Evaluations"

 "Method report - Effects of Development Aid to South Sudan and Mali"

On Mali

Mali Part I Summary Paper"

Mali Part II Country Level Evaluations"

Mali Part III Program and Project Evaluations"

"Method report - Effects of Development Aid to South Sudan and Mali"

On Afghanistan

"Summary paper meta-review of evaluations Afghanistan"

"Part 1 Systematic Review of Impact Evaluations"

"Part 2 Summary Report of Eleven Bilateral Country-Level Evaluations"

"Part 3 Summary Report of Selected SIGAR Reports, Afghanistan"

"Part 4 Summary Report of Evaluation Reports by the Asian Development Bank"

"Part 5 Summary Report of Selected Evaluation Reports by Multilateral Organizations and NGO"

Since the adoption of UNSC Resolution 1325 in October 2000, the Women, Peace and Security (WPS) agenda has been a top priority for the United Nations and for many of its member states. UNSC Resolution 1325 is credited with two achievements. First, it established a gendered perspective on violence and war, by stating that women and girls suffer disproportionately from the impacts of war and violence, and called for more prevention and better protection. Second, resolution 1325 also recognized that women can be resourceful and effective actors in the field of peacebuilding. This latter perspective is instrumentalist. In this view, women's inclusion and participation in peacebuilding activities will lead to better outcomes; hence women's inclusion and participation are instrumental for more effective peacebuilding activities. Three claims dominate the instrumentalist discourse: (1) the participation and inclusion of women in UN-led peacekeeping can make peacekeeping more effective (“operational effectiveness claim”). (2) the participation and inclusion of women in formal peace negotiations can make peace more durable (“better peace agreements claim”). (3) the agency of women in local peacebuilding activities can make peacebuilding more effective (“better local peacebuilding claim”). This systematic review aims to collect and summarize the existing evidence on these politically influential instrumentalist claims. This is the first systematic review on this topic. Given the importance of the WPS agenda and the prominence of the three causal claims, we believe it is important to collect and present the available evidence on these instrumentalist claims and to highlight existing gaps. The review systematically collects and synthesizes qualitative and quantitative evidence from studies meeting specific inclusion criteria. The study protocol was registered with OSF (Open Science Framework) on February 19, 2021 as OSF-Standard Pre-Data Collection Registration.

Purpose 

To provide a systematic assessment of existing literature about the impact of development aid on violence in fragile and conflict-affected states. 

Sample 

32 high-quality studies published since 2001 covering 36 aid interventions were analysed. These studies investigate the impacts of eight different aid types: Commander emergency response program; community driven development; conditional cash transfer; employment and training; large-scale infrastructure; humanitarian aid; food aid; and the impact of aggregated aid /all aid types. All reviewed aid programs were DAC eligible. Regionally, the sample is dominated by cases from Afghanistan and Iraq. 21 out of 36 cases refer to either Afghanistan or Iraq. Other cases refer to India, Philippines, Colombia, Liberia, Democratic Republic of Congo, Ethiopia, and Sudan. Furthermore, there is regional evidence from samples consisting of all civil war countries between 1969 and 2008, 125 non-OECD countries, 22 sub-Saharan African states and African countries with more than 1 million inhabitants, 1995 – 2012. 13 out of the 36 cases list stabilization as one of the main objectives of the programs. The other cases refer to normal development programs without a specific focus on stabilization. 

Main Results a) Of the analyzed cases1 , • Fourteen find a violence-increasing effect • Eight find no effect at all • Seven find that aid has a violence-reducing effect • Five find a heterogeneous treatment effect (meaning that a specific type of aid can, depending on the context, either increase or reduce violence b) These results suggest that the probability that aid has a violence increasing effect in regions in conflict is higher than the probability that aid has a violence-reducing effect. c) There is a widespread assumption in the literature that some types of aid / sectors are more likely to reduce violence than others. We find no support for this. None of the aid types covered in the sample is systematically associated with less violence. * This study was originally prepared for Global Affairs Canada (GAC), International Assistance Evaluation division. The author gratefully acknowledges financial support from GAC. 1 The studies cover 36 cases. However, two studies are essentially update versions of previous studies. In order to not overrepresent their findings we collapse the two versions of the same study into one observation. This leaves us with 34 cases. ii d) Vice-versa, we find that every type of aid, depending on the context, can have a violence-increasing effect. e) The studies suggest that aid can, depending on the context, reduce violence by (1) winning hearts and minds; (2) incentivizing local communities to share information about the insurgency with the government; (3) reducing grievances and (4) providing employment thereby increasing opportunity costs for insurgencies. f) However, all of these mechanisms can be undermined (“sabotaged”) by insurgents: When insurgents have enough capacity, they may violently sabotage projects. Thus, aid can trigger a strategic reaction by insurgents, which leads to more violence. g) Aid can also lead to more violence when aid flows are misappropriated by insurgents (“predation”). Predation can be violent per se, or it can provide insurgents with resources for organizing violence. h) Predation can be indirect and may be hard to detect by donors. i) No aid type is immune to predation: All types of aid can be “looted” or “taxed. j) Whether a specific aid intervention has a positive, a negative, or no impact, depends on the context (“scope conditions”). This meta-review has identified a number of important scope conditions. It appears that in order to maximize the probability that aid has a stabilizing effect, the following is required: • Aid is given in a fair, transparent and equitable way by a respected authority • Aid is locally meaningful, perceived to be beneficial, and ideally contributes to better livelihoods and more employment • The beneficiary group is internally coherent and not fractionalized • The aid project is relatively small, low-tech and implemented with the participation of the community In addition, the insurgency o is locally rooted o is mainly grievance driven o has little or no capacity to sabotage k) It is evident that these scope conditions are rarely met, which explains why aid in conflict zones is more likely to increase violence than to reduce violence. Nevertheless, these scope conditions can serve as important guidelines for aid programming in fragile states. l) Most studies equal “success “with “more security”, which they define as a reduction in physical violence. Studies employ measurements for both “objective” and “subjective” security. Objective security is measured by a count of security incidents, often disaggregated by the target of violent attacks, the perpetrator of violent attacks, the means of the violent attacks, and the number of victims of violent attacks. These measurements are taken from existing databases. “Subjective” measurements for security are self-reported, based on surveys. These surveys inquire about the perceptions of respondents with regard to their own security, or the security of their households and communities. m) Other measured concepts for “success” refer to perceptions of legitimacy of the national and subnational government, the economic situation and the provision of public goods. iii n) In rare cases, the studies also employ measures of the economic situation of individuals or households and how respondents assess the provision of basic services by government and development actors. o) The measured concepts refer to different spatial and temporal units. Spatially, the preferred unit is the district, followed by the village or the municipality. Occasionally researcher also use grid cells.

Executive Summary

This report provides results from a new monitoring & evaluation instrument, developed by BMZ. The instrument uses data from the German Development Tracker, the Risk Management Office, a booster survey in 18 districts in North Afghanistan and publicly available data sets.

The objectives of the instrument are (1) to report trends in the perceptions of Afghans (male and female) in regions where Germany development cooperation is active, and (2) to identify impacts of development cooperation on a number of outcomes.

Trends

Survey results show than on average over the whole sample (n=3273 in 18 districts)  there were little changes between 2016 and 2018, but differences between districts and changes within districts are often swift and large.  This is a strong reminder that generalizations are difficult, and that sound analysis has to take place at district level.

The perceived economic well-being of households has not changed on average. Households are still very vulnerable. 13% say it’s hard to buy food, and 37% say they can buy food, but nothing more. We see a slight trend that there are more very poor households and more (relatively) well-to-do household, whereas the number of households in the middle categories is decreasing. There are considerable differences between districts and changes within districts.  This points to the fact that households are very vulnerable to economic shocks and can very quickly slide into poverty.

Around 28% of households consume water from piped water schemes (unchanged since 2016). However, water consumption from unsafe sources such as ponds and rivers slightly decreased.

The number of households which received electricity (from any source) rose from 80.43% in 2016 to 90.83% in 2018. The number of households which were connected to landlines rose from 40.43% to 45.77%. The number for individual household generators, arguably the most expensive and most polluting energy source, decreased by half from 10.6% to 5%.

Looking at a wide range of government services, we find that the number of people how went and requested services has not change for most services, the exception being requests for IDs / Tazkerasa, which increased from 33% to 50%.

General satisfaction with received services has not changed, with the exception of requests for  IDs/ tazkeras, and for government agricultural offices (more satisfied) and  office of provincial Governor's office and Department of Women's Affairs (less satisfied).

Satisfaction with quality of primary education is decreasing. This trend begins in 2012. It may point to the fact that insecurity makes it difficult to provide good quality teaching, or that the initial enthusiasm about rebuilding of schools has been replaced my more critical views of quality of schooling.

Huquq courts increased their acceptance and were perceived to be more effective than in 2016.

Perceptions of security for the village, and for the household, slightly increased in 2018.

Compared to 2016 fewer people thought that the ANA had a positive impact on security

The number of respondents who said that their community received food aid, capacity building or job creation programs decreased.

Overall satisfaction with developed efforts by development organization and by the Afghan government is unchanged. Development actors are perceived to be most effective in increasing quality of drinking water, schooling, health, and least effective in job creation and agriculture.

Still very few people think that subnational government is responsive to the needs of the population, but  there is a positive but very slow trend.

In 2018, 44% of responders said they would leave Afghanistan, given the opportunity. This is more than in 2016 (37%). The main reason for wanting to leave is insecurity. 28% would prefer to go to Iran, 25% to Turkey, and 10% to Germany (down from 12%).

Overall skepticism about Western value (measured as the ratio of those who agreed that “foreign development aid is threatening our local way of life and Islamic values in our community, although it may bring material benefits”) has steadily grown since 2007 and was in 2018 at 67%. However, skepticism among women is lower, and actually deceasing.

Impacts

German development projects lead to perceptions of improved livelihoods. These effects are slightly larger in zones fully controlled by government. This suggests that German development has a positive impact on the economic well-being of beneficiaries.

Development projects do not increase positive attitudes towards the government. To the contrary, we find evidence that development projects increase negative attitude towards government (all levels) and can increase sympathy for armed groups. This is especially the case for “government buildings” and “road” construction”. It is less the case for education and health projects.

Also, these effects are stronger in zones controlled by government than in zones contested zones. This is an unexpected finding, and puts the widely shared “hearts-and-minds” theory on its head. We can only speculate about the reasons for this. Previous research has shown that the construction sector in Afghanistan is widely perceived as being corrupt. Local contractors have often a very bad reputation. Also, projects which are perceived to benefit only a few – the local elite - have been shown to increase inter-communal tensions It is possible that construction projects, even if they produce economic benefits, also produce apprehension towards authorities and elites.

Using an endorsement experiment, we find that the Taliban, although less popular than the Kabul government, are much more popular than people are willing to state explicitly. This is particularly the case in government-held areas. In contrast, in Taliban-held areas the insurgents are implicitly less popular than respondents said explicitly.

We find some marginal evidence that projects reduce violence, but the effects are not consistent or strong.

We don’t find evidence that police projects increased attitudes toward ANP.

Link to the WP site here https://www.washingtonpost.com/politics/2021/03/30/afghans-want-peace-new-data-show-theyre-open-surprising-number-options/

There are 5  sub-reports to this summary paper:

"Part 1 Systematic Review of Impact Evaluations"

"Part 2 Summary Report of Eleven Bilateral Country-Level Evaluations"

"Part 3 Summary Report of Selected SIGAR Reports, Afghanistan"

"Part 4 Summary Report of Evaluation Reports by the Asian Development Bank"

"Part 5 Summary Report of Selected Evaluation Reports by Multilateral Organizations and NGO