Table 1.

The modelled constraints on allocative efficiency.

Name	Representation	Assumptions
‘Earmarking’	Multiple criteria apart from efficiency (such as equity and cultural), and priorities of differing stakeholders must be taken into account by decision makers, often in an ad-hoc manner [14]. For example, due to specific criteria donors may attach specific conditions on how funds are spent, thereby limiting the health interventions, for specific diseases or key populations, decision makers can implement [15]	We stipulated that if a province received funding the first intervention funded would be PrEP for heterosexual women (excluding FSWs)
‘Meeting targets’	Externally imposed targets and the political context may influence the allocation process [11]	We stipulated that if a province received any prevention funding then, as a proxy for UNAIDS 90–90–90 target implementation [16], 90% of PLHIV must receive UTT
‘Minimizing change’	Decision makers may have limited capacity to modify existing HIV prevention programmes or redirect funding to different provinces due to the cost of making changes, such as the investments required for new clinics or personnel training [11]	We defined a simple weighted capitation resource allocation between provinces, approximated by the resource allocation that would arise if funds were distributed to maximize HIV prevention but with a single nationwide strategy. Under the ‘minimizing change’ constraint the total budget allocated to a province was the same as that in the previously defined weighted capitation allocation
All constraints on allocative efficiency		The earmarking, meeting targets and minimizing change constraints were applied simultaneously

FSW, female sex worker; PLHIV, people living with HIV; PrEP, preexposure prophylaxis; UTT, universal test-and-treat; UNAIDS, The Joint United Nations Programme on HIV/AIDS.