Is Empowerment Efficient?: A Data Envelopment Analysis of 260 Local Associations in Bangladesh

“This report presents one of the first formal analyses, outside the microfinance area, of the efficiency (as different from the effectiveness) of a development NGO program. The author [Aldo Benini], who invites comments and suggestions, offers this summary:

“Empowerment, a concept with a successful twenty-century cultural career, has been recognized for its relevance and, increasingly, effectiveness in liberating the poor, both at the individual and local community level. Efforts to create valid measurement tools have advanced, with a focus on causality, thus on effectiveness of empowerment programs. The efficiency of such programs, in other words considerations of optimal resource use, has not been investigated widely, with the exception of microfinance projects. Such programs are sheltered from efficiency pressures by the subsidies of aid chains and by the need to work out, in precarious social environments, organizational arrangements that produce credible empowerment effects in the first place.

I take advantage of a unique dataset on 260 local associations of poor people in northwestern Bangladesh, founded and fostered by the NGO “RDRS Bangladesh”, with a homogeneous basic structure and contiguous working areas. A capacity building grant from the EU in 2002 helped further revitalize these so-called “Union Federations” or, synonymously, “RDRS Federations”. Using Data Envelopment Analysis, a technique suited to relate diverse inputs and outputs without pre-existing weights or prices, I examine how efficiently the federations transformed membership and financial inputs in 2003 into female activism and business profits in 2005. I compare these efficiency estimates with effectiveness estimates from federation self-assessments and from a NGO-defined index. I also relate them to institutional and environmental factors.

The 260 federations are fairly distinctly separated into a large group of inefficient operators, and a smaller group of efficient ones, with a noticeable mid-field gap. For this I lack a cogent explanation. Congestion in 2003 with inactive members and fluctuations in tree sales may account for some of it. More importantly, efficiency and effectiveness are unrelated or, depending on the effectiveness measure, weakly negatively related. Their predictors have opposite effects. For example, loans from RDRS had effects that are positively related to self-assessed effectiveness. The effect on efficiency is negative. The aid chain shelters effectiveness, but not efficiency, from the impacts of higher ambient poverty. Better NGO support does improve federation efficiency whereas the relationship with effectiveness is less clear.

I seek a coherent interpretation of these various findings with the help of Dani Rodrik’s allegory of “second-best institutions”. The organizational arrangements that prove effective in poor societies are not always those recommended by context-free best-practice thinking. Moreover, aid chains tolerate vast inefficiencies, in order to keep flows predictable and hedge against various contingencies. While empowerment programs struggle with effectiveness, efficiency takes a back seat. This raises a moral dilemma about how to treat partners who use their modest resources efficiently, but, due to accidents of the aid chain or local environment, are still far from being effective.

This article chiefly addresses NGO managers and monitoring staff, for a greater conceptual awareness of effectiveness and, newly, efficiency. Most technicalities have been relegated to the appendix, but a sidebar offers a first introduction to Data Envelopment Analysis. This technique deserves further exploration as a tool to help mitigate reporting and evaluation challenges typical of multi-objective NGOs, an organizational form common to many empowerment programs.”


Comments?