Did local democracy help or hinder post-2001 Afghanistan? An MIT study comes up with some surprising insights.
- By Fotini ChristiaFotini Christia is an associate professor of political science at MIT and author of Alliance Formation in Civil Wars. Her research focuses on conflict and cooperation in the Muslim world. , Ruben EnikolopovRuben Enikolopov is a junior researcher at Barcelona Institute for Political Economy and Governance and an assistant professor of economics at Pompeu Fabra University and New Economic School, Moscow.
At the end of 2014, most NATO combat forces will have been withdrawn from Afghanistan, marking a disappointing end to an extraordinary exercise in nation-building. Despite over $100 million in development aid to the country, achievements in strengthening Afghanistan’s economy and institutions have fallen short of expectations. Paeans to an "Afghan miracle" have given way to hand-wringing post-mortems. Attracting particular criticism have been efforts — at both the international and local level — that promoted Jeffersonian democratic ideals over the hierarchical decision-making structures that have governed Afghanistan for hundreds of years.
But did the imposition of democratic structures in Afghanistan really make things worse? In a recent experimental study, "Do Elected Councils Improve Governance Quality?" we examine whether externally-imposed local democratic institutions improved or worsened governance in rural Afghanistan. Our results indicate that democracy per se was not the problem; when democratic bodies were charged with a specific activity, they produced more equitable outcomes. However, such institutions were often created in parallel to existing customary structures, and this had a different effect. With multiple local bodies, there was a lack of clarity as to who was in charge, which weakened leader accountability and provided opportunities for fraudulent behavior. As an Afghan proverb remarks, "the calf with two mothers gets no milk."
As an Afghan proverb remarks, "the calf with two mothers gets no milk."
Our study was a combination of two experiments: one long and one short.
In the long experiment, we created of democratically-elected, gender-balanced development councils in 2007. As per the randomized controlled trial methodology, we randomly selected 250 villages from a pool of 500 villages to receive democratic councils. These councils supplemented informal yet sophisticated customary structures (which usually include the village headman and an ad hoc council of tribal leaders), but assumed responsibility for implementing development projects in the villages. The remaining 250 villages formed the control group and did not receive elected councils, with governance structures remaining centered on customary local authorities.
We conducted the short experiment four years later, in which we distributed wheat to examine the effects of the councils on leader behavior and local governance quality. Across the 500 villages, we varied the distribution procedures to find out whether directly involving different groups also affected distribution outcomes. In half of the 250 villages with elected councils, the council was directly asked to manage the distribution, while, in the other half, we delegated responsibility to the "village leaders," a deliberately vague term which, depending on who was more powerful in the village, could mean either the elected council or the customary leadership. To examine the effects of involving women, we induced similar variation in the control villages. In half of these 250 villages, we specifically asked women to oversee the distribution, while, in the other half, we did not.
Two weeks after the distribution, we surveyed 10,000 villagers across the 500 villages to obtain data on how leaders behaved in their oversight role and who in the village received wheat. We collected data specifically to enable us to assess whether the wheat truly went to the most vulnerable recipients; how much wheat ended up in the hands of the leaders and their relatives; and the degree of participation by villagers in the decision-making process.
Although somewhat unusual, the wheat distribution outcomes did provide good measures of the quality of local governance. First, unlike other local governance activities, wheat distributions generate objective, quantifiable outcomes that can easily be compared across villages. Second, malign local leaders stand to benefit significantly from distributions at the expense of vulnerable villagers, with anecdotal evidence indicating that as much as a third of wheat is diverted for sale in district markets. Third, as wheat distributions occur relatively regularly in rural Afghanistan, the behavior of leaders in the distribution is representative of their general behavior. Thus, by measuring how much wheat reaches vulnerable villagers, we get a picture of how the creation of elected councils affects the tendency of local leaders to act in an equitable or malign manner.
The results of the study were unexpected, but all the more interesting.
In villages where we specifically asked the democratically-elected council to oversee the distribution, vulnerable households were more likely to receive wheat (relative to vulnerable households in villages in which elected councils did not exist and customary leaders alone managed the distribution). The elected councils had no effect, however, on the level of embezzlement or the inclusiveness of decision-making. We can reservedly say, then, that creating elected councils and putting them in charge of the distribution improves outcomes for vulnerable villagers.
The interesting part, though, is what happened when the prescribed procedures made it less clear who was in charge — that is, in villages with elected councils where we delegated responsibility to "village leaders," and in villages where we asked women to participate alongside customary leaders. In the first case, levels of embezzlement were higher and decision-making was less inclusive than in villages in which elected councils did not exist and customary leaders alone managed the distribution. In villages where we required that women oversee the distribution alongside customary leaders, levels of embezzlement were similarly higher than in villages in which customary leaders alone managed the distribution.
Embezzlement increased when the responsibility for the oversight of the distribution was not clearly defined.
The key takeaway from the results is that embezzlement increased when the responsibility for the oversight of the distribution was not clearly defined. In villages with elected councils, embezzlement increased when it was not clear whether elected councils or customary leaders were in charge. Likewise, for villages without elected councils, embezzlement increased when customary leaders were asked to share responsibility with women leaders. On the other hand, outcomes were best when responsibility was the most precisely assigned — that is, when a specific institution, the elected councils, was placed in charge of the distribution.
These results turn out to be quite powerful in demonstrating how development interventions might have inadvertently weakened accountability and governance in rural Afghanistan.
Development programs in Afghanistan have commonly followed a practice of creating new local bodies (commonly termed councils or shura) to manage local projects or interactions. Water management shura were established to manage irrigation projects; the National Solidarity Program created the development councils
we studied; the International Security Assistance Force created "reintegration shuras" to transfer detainees back to communities; and a multitude of NGOs created their own shura to manage local interventions. All of this occurred on top of customary structures comprising a village headman, a customary shura, and local clergy. One scholar coined the phrase "shura fatigue" in lamenting the rampant practice that made spaghetti soup out of local governance.
This avalanche of shura splintered governance authority across an array of village institutions, old and new, which made it difficult for villagers to figure out who was in charge. Our results suggest that this in turn triggered a "common pool" problem, where leaders reacted to the diffusion of accountability by engaging in rent-seeking behavior that left ordinary villagers worse off.
Our results relay a solid piece of advice: if development actors want to empower communities to sustain change, they should think twice about creating new institutions when options exist to strengthen the accountability of existing institutions. While democratic institutions like elected councils can improve outcomes, reformers must first clarify these institutions’ relationship with existing bodies. And whatever the institutional form — be it democratic or customary — governance outcomes are best when the population is able to tell who is in charge and who will hold them accountable.