1. Economic Impact of HIV/AIDS
in Botswana: Linkages between
Macroeconomic, Sector and
Household levels
HIV/AIDS intervention in developing
countries: use of Cost Effectiveness and Cost
Benefit analysis to guide Policy and Action
Harvard School of Public Health
Sept 13-15 2006
Keith Jefferis and Anthony Kinghorn
2. Background
Previous macroeconomic impact
study 2000
Roll-out of ART
National Strategic Framework
costing – considered “unaffordable”
Subsequent work on
macroeconomic impact in Botswana
(IMF) and elsewhere in Southern
Africa
3. Ongoing Study - 2006
Funded by UNDP, on behalf of GoB/NACA
Parallel demographic impact study
Review of earlier studies
Accuracy of projections
Methodology
Components
Updating of macroeconomic models
Firm/industry review
Costing/fiscal impact
Household/poverty impact
5. Macroeconomic Modelling
Aim to capture variety of macro impact
channels:
Labour force
slower growth (demographics)
changed age & experience structure
labour productivity (illness/absence)
Broader macro impacts
overall productivity growth
expenditure diversion
savings & investment
6. Macroeconomic Modelling
Dual approach:
Aggregate production function (Solow growth
model) incorporating formal and informal
sectors, skilled & unskilled labour
Computable General Equilibrium (CGE) model
incorporating range of economic sectors and
labour and household categories
Both solve for macroeconomic equilibrium
on the basis of calibrated model & input
assumptions (e.g. demographics)
7. Macroeconomic Modelling
Scenario modelling:
No AIDS
with AIDS
AIDS with treatment (ART)
Solve annually and roll forward to
2021
Outputs include GDP, growth, per
capita incomes, employment, wages
8. Model Structure (Agr. PF)
Formal Skilled
Sector Labour
Population
Capital
& AIDS
Informal Unskilled
Sector Labour
OUTPUT
Productivity
(TFP)
9. Illustrative GDP Growth Impact
6%
5%
4%
3%
2%
1%
0%
10
02
04
06
08
12
14
16
18
20
20
20
20
20
20
20
20
20
20
20
No AIDS AIDS no ART AIDS with ART
10. Contributions to GDP Growth
No-AIDS vs AIDS with ART
TFP, 31%
Capital, 49%
Skilled, 14%
Unskilled,
6%
11. Illustrative Impact - Real GDP per
capita
19,000
P million (2001 prices)
18,000
17,000
16,000
15,000
14,000
13,000
12,000
11,000
10,000
2001 2004 2007 2010 2013 2016 2019
No AIDS AIDS No ART AIDS with ART
12. Key Modelling Results & Conclusions
Labour market effects through:
demand (investment, wage levels,
productivity)
supply (size & composition of LF)
Result: less favourable employment
trends (reduced demand outweighs
reduced supply)
Higher un/under-employment and slower
wage growth
Only partially alleviated by ART
14. Household Impact
Poverty impact simulated through use of
household survey data (income &
expenditure, 2002/03 & AIDS impact,
2004)
Superimpose HIV/AIDS on population in
accordance with demographic prevalence
trends
Simulate income and expenditure effects
and calculate impact on poverty
headcount rates
15. CGE Results - Poverty
24
National poverty headcount (% )
23
With AIDS
22
Treatment
21
Without AIDS
20
19
2003 05 07 09 11 13 15 17 19 21
17. Methodology
Demographic projections
ART, No-ART, No-AIDS
Utilisation
Various protocols, policies, site data
Calibration to empirical data - plausible
Limitations
Costs
Unit costs of ART, Orphan Grant, program
expenditure history, step down for in- and
outpatient
18. Projected Total Number of adults and children on ART
(Provisional - illustrative)
160
140
120
100
thousands
80
60
40
20
0
01
03
05
07
09
11
13
15
17
19
21
20
20
20
20
20
20
20
20
20
20
20
ART Best estimate ART 10% lower ART 10% higher
• There will continue to be large, rapidly rising
numbers on ART
• Some uncertainty about length of survival on ART,
uptake rates that may affect scenarios
19. Projected Number of Total deaths per year
(Provisional - illustrative)
40,000
35,000
30,000
25,000
20,000
15,000
10,000
5,000
0
91
93
95
97
99
01
03
05
07
09
11
13
15
17
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
No AIDS No ART ART Best estimate
• Needs for terminal care should not increase
substantially beyond recent levels
22. Preliminary projected Costs – % contributed by
selected interventions combined with ART (Best estimate)
100%
90%
80%
70%
P million
60%
50%
40%
30%
20%
10%
0%
97
99
01
03
05
07
09
11
13
15
17
19
19
19
20
20
20
20
20
20
20
20
20
20
Hosp. in-patient Ambulatory excl ART
ART HBC
Prevention Prog. mgt
OVC OA pensions (cost vs. no ART)
23. Key preliminary findings
Terminal care and hospital bed needs are unlikely
rise substantially above 2001/2 levels until after
2015, but substantial backlogs and referral
system inefficiencies remain
The double orphan epidemic should reach a
plateau soon under high ART coverage scenarios
Prevention expenditure is uncertain but costing
shows importance of effective prevention for
sustainability
Capacity requirements of sustainable, effective
ART models are still unclear
Current models and implications for e.g. HBC and
hospital loads are not clear
24. Preliminary Conclusions: Impact on
Government Budget
Overall fiscal impact of HIV/AIDS expected to be
substantial, but (just) manageable
Bulk of HIV/AIDS-related costs required whether
or not ART is provided (ART adds 50% to costs)
Incremental costs of ART can probably be
partially – but not completely - funded from taxes
on extra GDP generated
Overall costs of HIV/AIDS cannot be financed
from budget deficits
Need to reprioritise expenditures within health
budget, HIV and AIDS program and elsewhere
Tougher trade-offs required if ART is provided
Donor resources needed to keep fiscal burden
manageable
26. Preliminary Conclusions –
Methodological issues
Policy making advantages of combined
macroeconomic, sectoral and poverty
analysis
Shows linkages between sectoral decisions and
effects
Clearer tradeoffs for prioritisation
Fiscal analysis
Macro planning – establishing “common
language” with health and programme
planners
Developing implicit policy scenarios and
interpreting them for different audiences and
purposes
27. Preliminary Conclusions –
Methodological issues
Macroeconomic analysis
Macro modelling approaches valid and useful
CGE + micro-simulation particularly useful in
providing integrated approach
Some key input parameters – investment and
productivity impacts – have uncertain empirical
basis – key areas for further, micro-level
research
HIV impact on impact on firms’ decision
making processes
Trade-off between cuts in recurrent and
investment spending in fiscal decisions
28. Preliminary Conclusions – Policy
making implications
Risks of inadequate NSF costing
Prioritisation
Objectives of costing
Cost vs cost benefit focus
Cost control essential (ART, welfare)
Consider cost & clinical effectiveness of
ART distribution channels; innovative
solutions necessary
Exploring implications of Abuja
Declaration targets – Health as 15% of
public expenditure
Advocacy to donor community
29. Implications – other countries
Botswana somewhat exceptional (in sub-Saharan
Africa):
Very high HIV prevalence rate
High income, GDP growth
Savings surplus (over investment)
Capital intensive
Fiscal, BoP surpluses
Domestically-financed ART provision feasible but
tough even in favourable environment
Methodological approaches useful and
transferable depending on quality of data
Results elsewhere could well be different
elsewhere