This article presents a study of the cost-effectiveness of government inputs in raising achievement scores in Bangladeshi primary schools. The analysis itself involves three steps: regression analysis, cost-analysis and simulation analysis--each built on the previous step to identify the most cost-effective government inputs. The paper describes each step and the results of each step. The results are, nonetheless ambiguous. The paper posits that these ambiguities derive from the policy context in which the study was designed and executed. This study suggests that there are hidden costs when research designs and methodologies are chosen in order to minimize political/policy risk.