Flying Blind? Constructing Evidence-based Poverty Reduction Policies in PRSP Adopting Countries

The World Bank, alongside other development institutions and leading donors, has increasingly emphasised the importance of evidenced-based policymaking in addressing entrenched poverty. Additionally, widespread adoption of Poverty Reduction Strategy Papers (PRSPs) has necessitated establishing poverty baselines and updating them on a regular basis, to enable effective impact evaluation. To facilitate this, considerable resources have been devoted to providing sound longitudinal poverty data via the implementation of the World Bank’s flagship Living Standards Measurement Survey (LSMS) programme. While a lively debate has emerged over the conceptual basis of poverty measurement within LSMSs, rather less has been said about the wider impact of basic changes in survey design and measurement approaches. We argue that the latter have been as important in shaping controversies over the comparability of national poverty aggregates and the value of LSMS data for policy analysis. Supported by a review of World Bank Poverty Assessments carried out since 2004, and a case study of the Mongolian experience, we find that variations in survey methods and subsequent major revisions are far from rare, and have generally been accompanied by inadequate disclosure. This paper shows that where revisions have been made, the lack of transparency undermines the application of standard analytical techniques and prevents the replication of key results. We conclude that insufficient efforts are being made to establish consistent and verifiable poverty data; and, moreover, that the lack of openness and rigour has permitted the fitting of results to predicted policy objectives and weakened the level of external scrutiny. An implication of this is that effective policymaking is made more difficult and national ownership is compromised. This supports the case for alternative, more eclectic approaches to the monitoring and evaluation of poverty reduction policies.