Needling the entitlement balloon: Assumptions, cost projections and flaws in the Centre for Global Development’s AIDSCost model
Working Paper Number:
The recently developed AIDSCost model was designed to forecast the future financial burden of global antiretroviral treatment (ART) up until 2050, and has been used to claim that ART treatment is too expensive for developing countries. This paper interrogates the structure and outputs of the AIDSCost model. Firstly, we investigate the model's assumptions and find a number of flaws in its makeup, the most significant being the unrealistic way in which ART coverage expands. These make it inappropriate for use in many real-world policy settings. Secondly, we compare the model's outputs for South Africa to those of ASSA2003, South Africa's most highly developed AIDS modelling tool, for the period 2007-2016 as a test of the former's accuracy. We find that, even when applying a number of different ART coverage and costing scenarios, AIDSCost overestimates the future burden of ART by as much as 100%. Though the model's costing function is disputable, the most serious errors underlie the calculation of vital outputs on which costing depends, most notably AIDS death rates, the number of those on ART and HIV prevalence. Accordingly, we argue that the model should be subjected to thorough refinement before it is used by anyone. Further, we argue that AIDSCost co-author Mead Over, in employing the model to show that "ballooning" ART burdens will overwhelm US aid budgets, generates unreliable figures which severely overestimate the future financial burden of global ART.