This thesis analyses fiscal effects of aid, first of health aid on health spending for a sample of developing countries and then broadly for Ethiopia and Tanzania. Particular attention is paid to data quality and the severe difficulties in achieving a reliable disaggregation of aid into its on-budget and off-budget components. The first essay assesses the sensitivity of estimated health aid fungibility to how the missing data (often considerable) are treated and explores a novel (at least in economics) method of multiple imputation. The second essay provides a conceptual framework for the disaggregation of (sector) aid into its on-budget and off-budget components. Given that complete binary distinction is not feasible, the aid-spending relationship is explored from a broader fiscal effects angle. This yields new insights on assessing the effect of health aid on health spending. Contributing to the growing body of evidence based on time-series methods, two essays adopt a case study approach to analyse distinct fiscal dynamics in Tanzania and Ethiopia, invoking detailed understanding of qualitative economic and political context to complement the quantitative data. Both essays employ current Cointegrated Vector Auto-Regressive (CVAR) techniques to distinguish long run equilibrium relationships from short term adjustment mechanisms, test for variable exogeneity and identify which variables adjust to disequilibrium. The fifth and final essay addresses the differences between donor and recipient data records for the two countries, demonstrating that the direction of the discrepancies is not necessarily predicted from the outset and affects the estimated fiscal effects of (and on) aid. These essays contribute to the growing literature using country case studies to assess the fiscal effects of aid.