recommendations. Top management has also increasingly used the bully pulpit to persuade both the institution and the borrowers that environmentally unsustainable development is uneconomical, even immoral.
Learning addressed two requirements: rebuilding favorable political coalitions and improving performance. The Bank approached both tasks first through error correction, which took different forms depending on the political dynamics at work. The EA directives enabled REDs to stop projects in their tracks. In some cases, the Bank took dramatic action as when it suspended disbursements for the first time in 1985 for environmental reasons or cancelled a loan on similar grounds for the first time in 1989. This procedure remains rare, however. Only after years of controversy, pressures from donors and NGOs, and the release of a critical report ( Morse and Berger, 1992), did the Bank withdraw its financial support of the Sardar Sarovar project in India. (This report demonstrated a remarkable and unprecedented willingness to look publicly and extensively into its record.) In other cases, error correction meant banning some activities outright. In June 1986 it decided to stop lending for projects in fragile ecosystems and in July 1991 declared an end to support for commercial logging in tropical forests. Although blanket interdictions are probably the easiest way to avoid controversies and promote bureaucratic compliance, they are not necessarily the best means of promoting environmentally sustainable development. Countries need foreign exchange. Yet Operations finds it more appealing since such blanket policy rids it of complicated headaches.
Environmental assessments seek to change the norms that govern behavior. The Bank promotes a new model of the world around a classical economics approach to environmental issues. Much has been learned about the sociological, cultural, and political contexts of sustainable development. EAs, especially sectoral ones, are potentially powerful instruments of change and can serve as primary vehicles for reorientation.
More important though, learning is essentially a political process. Consensual knowledge and stable coalitions are important when they foster common expectations about the organization's performance. The Bank has acted chiefly when it could make an economic, social, or political link between environmental degradation and human welfare. EAs aim to facilitate the development of political consensus inside and outside the Bank in order to minimize future political controversies; their content then becomes secondary. Organizations that learn are able to create environments they can control.
The Bank hopes to improve performance by limiting its involvement in controversial projects, by creating the conditions for effective national implementation of agreed-upon procedures, by internalizing environmental