Economic Impact of HIV/AIDSin 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 ActionHarvard School of Public Health

Sept 13-15 2006

Keith Jefferis and Anthony Kinghorn

Background

Previous macroeconomic impact study 2000Roll-out of ARTNational Strategic Framework costing – considered “unaffordable”Subsequent work on macroeconomic impact in Botswana (IMF) and elsewhere in Southern Africa

Ongoing Study - 2006

Funded by UNDP, on behalf of GoB/NACAParallel demographic impact studyReview of earlier studies

Accuracy of projectionsMethodology

ComponentsUpdating of macroeconomic modelsFirm/industry reviewCosting/fiscal impactHousehold/poverty impact

Macroeconomic Modelling

Macroeconomic Modelling

Aim to capture variety of macro impact channels:Labour force

slower growth (demographics)changed age & experience structurelabour productivity (illness/absence)

Broader macro impacts overall productivity growthexpenditure diversionsavings & investment

Macroeconomic Modelling

Dual approach: Aggregate production function (Solow growth model) incorporating formal and informal sectors, skilled & unskilled labourComputable 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)

Macroeconomic Modelling

Scenario modelling:No AIDSwith AIDSAIDS with treatment (ART)

Solve annually and roll forward to 2021Outputs include GDP, growth, per capita incomes, employment, wages

Model Structure (Agr. PF)

FormalSector

InformalSector

Capital

SkilledLabour

UnskilledLabour

Population & AIDS

OUTPUTProductivity

(TFP)

Illustrative GDP Growth Impact

0%

1%

2%

3%

4%

5%

6%

2002

2004

2006

2008

2010

2012

2014

2016

2018

2020

No AIDS AIDS no ART AIDS with ART

Contributions to GDP GrowthNo-AIDS vs AIDS with ART

TFP, 31%

Skilled, 14%

Unskilled, 6%

Capital, 49%

Illustrative Impact - Real GDP per capita

10,00011,00012,00013,00014,00015,00016,00017,00018,00019,000

2001 2004 2007 2010 2013 2016 2019

P m

illio

n (2

001

pric

es)

No AIDS AIDS No ART AIDS with ART

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 growthOnly partially alleviated by ART

Household-level Impact

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 trendsSimulate income and expenditure effects and calculate impact on poverty headcount rates

CGE Results - Poverty

19

20

21

22

23

24

2003 05 07 09 11 13 15 17 19 21

Nat

iona

l pov

erty

hea

dcou

nt (%

)

With AIDS

Without AIDS

Treatment

Costing & Fiscal Impact of HIV/AIDS

Methodology

Demographic projectionsART, No-ART, No-AIDS

UtilisationVarious protocols, policies, site dataCalibration to empirical data - plausible Limitations

CostsUnit costs of ART, Orphan Grant, program expenditure history, step down for in- and outpatient

Projected Total Number of adults and children on ART(Provisional - illustrative)

0

20

40

60

80

100

120

140

160

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

2021

thou

sand

s

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

Projected Number of Total deaths per year(Provisional - illustrative)

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

No AIDS No ART ART Best estimate

• Needs for terminal care should not increase substantially beyond recent levels

Costs

Preliminary projected Costs – % contribution of selected interventions No ART(Best estimate)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

P m

illio

n

Hospital In-patient Ambulatory excl ART ARTHBC Prevention Prog. mgt.OVC

Preliminary projected Costs – % contributed by selected interventions combined with ART (Best estimate)

0%10%20%30%40%50%60%70%80%90%

100%

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

P m

illio

n

Hosp. in-patient Ambulatory excl ARTART HBC Prevention Prog. mgtOVC OA pensions (cost vs. no ART)

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 remainThe double orphan epidemic should reach a plateau soon under high ART coverage scenariosPrevention expenditure is uncertain but costing shows importance of effective prevention for sustainabilityCapacity requirements of sustainable, effective ART models are still unclear

Current models and implications for e.g. HBC and hospital loads are not clear

Preliminary Conclusions: Impact on Government Budget

Overall fiscal impact of HIV/AIDS expected to be substantial, but (just) manageableBulk 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 generatedOverall costs of HIV/AIDS cannot be financed from budget deficitsNeed to reprioritise expenditures within health budget, HIV and AIDS program and elsewhereTougher trade-offs required if ART is providedDonor resources needed to keep fiscal burden manageable

Summary of Preliminary Conclusions

Preliminary Conclusions –Methodological issues

Policy making advantages of combined macroeconomic, sectoral and poverty analysis

Shows linkages between sectoral decisions and effectsClearer tradeoffs for prioritisation

Fiscal analysisMacro planning – establishing “common language” with health and programme plannersDeveloping implicit policy scenarios and interpreting them for different audiences and purposes

Preliminary Conclusions –Methodological issues

Macroeconomic analysisMacro modelling approaches valid and usefulCGE + micro-simulation particularly useful in providing integrated approachSome 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 processesTrade-off between cuts in recurrent and investment spending in fiscal decisions

Preliminary Conclusions – Policy making implications

Risks of inadequate NSF costingPrioritisationObjectives of costingCost vs cost benefit focus

Cost control essential (ART, welfare)Consider cost & clinical effectiveness of ART distribution channels; innovative solutions necessaryExploring implications of Abuja Declaration targets – Health as 15% of public expenditureAdvocacy to donor community

Implications – other countries

Botswana somewhat exceptional (in sub-Saharan Africa):

Very high HIV prevalence rateHigh income, GDP growthSavings surplus (over investment)Capital intensiveFiscal, BoP surplusesDomestically-financed ART provision feasible but tough even in favourable environment

Methodological approaches useful and transferable depending on quality of dataResults elsewhere could well be different elsewhere