1

Economic Impact of HIV/AIDS

Stakeholder Workshop March 2, 2007

Structure of Presentation

Review of Terms of ReferenceAssessment of BIDPA (2000) ModelMacroeconomic ModelsFirm/industry reviewFiscal impactHousehold/poverty impactConclusions

2

Terms of Reference

Terms of Reference

Review and update the BIDPA (2000)macroeconomic impact study, in particular:

the methodology, assumptions and choice of models;evaluate the findings of the study vis a vis subsequent trends, and ascertain the predictive capabilities and suitability of the models chosen

3

Terms of Reference

Analyse the likely impact of HIV/AIDS on the Botswana economy to 2021 using quantitative models; contrast the findings with those of BIDPA (2000)Estimate the trend paths of key economic variables under alternative HIV/AIDS scenarios, including the without-AIDS scenario, specifically:

economic growth, savings, investment, human resource capacity, labour supply, productivity, competitiveness and poverty

Terms of Reference

Estimate the disaggregated current and future costs, direct and indirect, to the Government and the economy, of HIV/AIDS, with implications for the Government budget. Reconcile model predictions of the micro and macro level impacts of HIV/AIDS. This will involve estimating the household and sectoral impacts of HIV/AIDS.

4

Terms of Reference

Determine what policy levers the Government has at its disposal to mitigate the economic impact of HIV/AIDS, the extent to which such levers have been used and to what effect.Investigate the strategies that firms have employed to protect their businesses from HIV/AIDS and the extent to which they have been successful in this regard.

Review of BIDPA Study (2000)

5

Review of BIDPA Study

Macro Model StructureBIDPA model – aggregated growth function, formal & informal sectors, skilled & unskilled labourAccommodates “with” & “without” AIDS scenariosRelevant parameters (infection rates, productivity, labour force growth) can varyCalibrated to 1995/96 actual data, simulations to 2021 based on demographic projectionsProjections of real GDP growth; per capita incomes; wages; employmentHousehold (poverty) impact using HIES data

Diagram of Model Structure

FormalSector

InformalSector

Capital

SkilledLabour

UnskilledLabour

Population & AIDS

OUTPUT

6

Survey of HIV & AIDS Economic Impact Studies – Model Types

No.Type of Model

3Computable general equilibrium

3Macro-econometric model

5Aggregate growth model

2Econometric estimation

Review of BIDPA Study

Methodology was sound – aggregate growth model is most widely used; notably IMF studies on Botswana, 2001 & 2004, also Malawi & TanzaniaModel is transparent, data requirements modest, maths & programming tractableAssumptions used were based on best available data at the time, although subsequent developments not always as assumedBIDPA study has been widely referenced and quotedHousehold impact analysis (simulation based on HIES data):

BIDPA study was first of its kindUsed in other studies subsequently

7

Review of BIDPA Study

Other methodologies also useful where data is availableDisaggregated approach can be useful – more detailed simulation of economic changesCGE models used for SA, Tanzania, ZambiaMacro-econometric models in SA

Actual Outturn vs BIDPA (2000) Projections & Assumptions(period averages)

1995/6 - 2000/01

2000/01 - 2005/06

Economic GrowthBIDPA 3.1% 2.9%Actual (non-mining) 5.9% 4.7%Actual (non-mining private sector) 5.4% 3.7%GDP per capita (growth)BIDPA 1.3% 1.1%Actual 2.8% 3.8%Population GrowthBIDPA 2.5% 0.8%Actual (CSO) 2.4% 0.9%Actual (CARe) 2.5% 1.7%Labour Force Participation RateBIDPA 48.5% 48.3%Actual 49.8% 56.5%*Investment (% GDP)BIDPA 25% 25%Actual 30% 21%Productivity (TFP) GrowthBIDPA 0.25% 0.25%Actual 1.3%*HIV prevalence (15-64 yrs, %)BIDPA 31% 30%Actual 24%** different time period

8

Summary of Model vs Outcomes

Average GDP growth over 2001-2005 higher than predictedHowever, recent growth of non-mining private sector close to predicted rates

0123456789

'97/98

'99/00

'01/02

'03/04

'05/06

%

Summary of Model vs Outcomes

Population growth higher than predicted (+)Higher labour force participation (+)Investment close to predicted valueHIV prevalence lower than forecast (+)Productivity (TFP) higher (+)ART available

9

Choice of Macroeconomic

Modelling Approaches

Channels of Potential Economic Impact

MorbidityProductivity (sickness, time off)Expenditure (health care, training)Savings (diversion of incomes)Investment (uncertainty, profits, savings)

MortalitySmaller population and labour forceChanged age structure (experience)Loss of skills

10

Macro Modelling ApproachesUpdating of BIDPA model

Calibrate to 2001 (from 1996) with new economic dataIncorporate 2006 demographic projectionsIncorporate “with ART” & “no ART” scenarios along with “No AIDS” counterfactualPay more attention to costs of HIV/AIDS treatment, impact on savings, investment & growth Impact of ART on labour force, productivityImprove modelling of productivity growthUse 2002/03 HIES data, but no new labour force data (since 1996)

Macro Modelling Approaches

Other macro modelling approachesMacro-econometric model

needs pre-existing model – not available in Botswanamodel building a long and complex process

CGE modelfeasible to build CGE for this projectwell-suited to analysis of HIV/AIDS impact

11

Basis of CGE Model

As with aggregate growth model, also works by simulating behaviour of economyMore detailed economic structure –disaggregated by sector, labour category, household income groupCan model many interaction channels simultaneouslyBased on Social Accounting Matrix (SAM)Very demanding data requirementsCan be linked with HIES for simulations

Key Findings

Macroeconomic Impact

12

Simulated GDP Growth Rates, 2002-2021 (Fig. 5-8)

0%

1%

2%

3%

4%

5%

6%

2002

2004

2006

2008

2010

2012

2014

2016

2018

2020

No AIDS AIDS-ART AIDS - No ART

Simulated Real GDP per capita 2002-2021 (Fig. 5-7)

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

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

2021

No AIDS AIDS-ART AIDS - No ART

13

Simulated Underemployment, 2002-2021 (Fig. 5-9)

20%22%24%26%28%30%32%34%36%

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

2021

No AIDS AIDS-ART AIDS - No ART

Contributions to GDP GrowthNo-AIDS vs AIDS with ART

TFP, 22%

Skilled, 21%

Unskilled, 9%

Capital, 48%

14

Sectoral Impact

012345678

AgrMine

MfgW&E

ConstrTrad

eH&R

Transp

Fin serv

Bus serv

Oth serv

Pub Adm

E & H

Avg

. gro

wth

200

3-21

(%)

NO AIDS AIDS ARTLabour intensive sectors dependent on less-skilled workers most affected

Key Findings

Household Impact

15

Household (Poverty) Impact

Wide range of possible impacts on HHIncome and ExpenditureDirect and Indirect ChannelsTemporary and Permanent Effects

Possible channelsCosts of medical provisionFuneral costsChanged household composition (fewer or more members; income-earners vs dependents)Loss of income as breadwinners fall sick or dieChanged employment opportunitiesImpact on general wage levelsGovernment orphan support

Household (Poverty) Impact

Modelled through:Simulating impacts on HIES source data over 10 years (as per BIDPA study)CGE modelling to 2021 (new)

16

Simulated Poverty Impact (HIES)

0%10%20%30%40%50%60%

Gab FT Othurb

R SE R NE R NW R SW NatPove

rty

Hea

dcou

nt (H

H P

DL)

Without AIDS AIDS with ART

Simulated Poverty Impact to 2021 (CGE)

18

19

20

21

22

23

24

25

2003 '05 '07 '09 '11 '13 '15 '17 '19 '21

Pove

rty

Hea

dcou

nt ($

a d

ay, %

)

With AIDSNOAIDS

17

Household (Poverty) Impact

HIV & AIDS has clear negative impact on povertyPoverty headcount up to 3% higher due to HIV/AIDSART provision offsets this by 1/3 to 1/2.Orphan welfare provision also has significant poverty benefits

Key Findings

Firm-levelSurvey

18

Firm-level Survey - Introduction

25 firms were interviewed in different sectorsSurvey was not intended to be nationally representative but was sufficient to bring out the salient issues about HIV/AIDS

General Results

Generally a bigger loss of unskilled workers due to illness and death than skilled workers75% of firms reported negative impact of HIV & AIDS on output and productivityMost firms (56%) responded that HIV/AIDS has no significant impact on investment:

other factors affecting profitability more importantsome firms reported delays in expansion and diversion of spending

Difference in impact across sectors – level of skills a major factorSectoral impact similar to SAFirms reported a reduced effect of the disease due to the availability of ARV since 2001/2002, esp. for skilled workers.

19

Firms’ Responses

Firms have been innovative in their responses, especially those that have been impacted most by the disease:

training more workers than needed;keeping additional workers on standby.over-employ for critical positionsmulti-skillingmechanisationmore overtimetemporary staff

Although output could be maintained, training costs increased significantly

Response in training by skill level

0

10

20

30

40

50

60

70

80

No training response Train more multi-skilling Total

% o

f ski

lled

wor

kers

20

Response in hiring by skill level

0

10

20

30

40

50

60

70

80

No response Hire more Hiring temporaryworkers

Total

% o

f ski

lled

wor

kers

Severity of Impact by Sector

0 20 40 60 80 100

Retail

Financial

Service

Mining

Manufacturing

Construction

(Least)

- Sev

erity

- ( Most)

% of skilled workers

21

Fiscal Impact of HIV/AIDS

Cost Implications

ARTHospital in-patientAmbulatoryOrphan careHome-based carePreventionProgramme managementOld age pension

22

Projected Total Number of adults and children on ART

0

20

40

60

80

100

120

140

160

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

2021

thou

sand

ART Best estimate ART 10% lower ART 10% higher

Hospital bed needs for HIV and AIDS per year

0

500

1,000

1,500

2,000

2,500

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014

2016

2018

2020

No ART ART Best estimate ART 10% lower ART 10% higher

23

Projected Number of Total deaths per year

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 ART Best estimate less 10%

Projected Costs – with ART

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

2021

P m

illio

n (r

eal,

2004

/05

pric

es)

OAPOVCOtherPreventionHBC ARTAmb exc ARTIn-patient

24

Projected ART costs

0

100

200

300

400

500

600

700

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

2021

P m

illio

n (r

eal,

2004

/05

pric

es)

ART Best estimate ART 10% lowerART 10% higher Actual from Dev expenditure

Projected Costs – selected interventions No ART

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

2019

2021

P m

illio

n (r

eal,

2004

/05

pric

es) OVC

Prog. mgt.

Prev

HBC

ART

Amb exc.ARTIn-patient

25

Total costs by scenario (P million)

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

P m

illio

n (2

004/

05 p

rice

s)

No ART ART best estimate

Total costs by scenario (% of GDP and Gov Exp)

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

%

No ART (% GDP) ART (% GDP) No ART (% Gov Exp) ART (% Gov Exp)

26

Conclusions, Policy Implications and

Recommendations

Macroeconomic Implications

Real GDP growth reduced by 1.5% -2% a year without ARTEconomy will be up to one-third smaller by 2021 due to HIV & AIDSResult of reduced labour force growth, younger LF, reduced productivity & investmentGDP/capita growth 0.5%-1% lower

27

Macroeconomic Implications

ART provision adds 0.4% - 0.8% to average GDP growth (cf. no-ART)Eliminates apprx one-third of negative growth impactEconomy still 20%-25% smaller by 2021Avg. incomes growth higher with ARTIn both scenarios investment channel is most important

Labour Force & Employment

Reduced labour supply and labour demand – so overall effect uncertainModels suggest that demand effects dominateLeading to lower emloyment and lower wages with HIV & AIDS

28

Macroeconomic Recommendations

Efforts to improve economic efficiency and reduce costs crucial to offset negative HIV & AIDS impactsImplement measures supportive of private sector investment & economic diversificationSkills development, shared training costsMake it easier for firms to recruit citizens and non-citizensPoverty reduction and social welfare policies crucial to minimise poverty impact

Fiscal Implications

HIV & AIDS is having a major impact on govt budget – approx 6% of govt spendingCost will rise by 60% in real terms by 2021, peaking at 8% of spending/3% of GDPART drugs largest single component (40% of total)No-ART scenario costs are lower, but ART savings offset by higher other costs (health, OVC, HBC etc.)Economic growth and govt revenues would be lower in No-ART scenarioHence incremental ART (as % of GDP and govt spending) costs are small

29

Fiscal Recommendations

Costs are manageable but large & imply fiscal adjustments if budget is to be sustainableFully funding HIV & AIDS costs from budget deficits not feasible – needs trade-offs & cuts in spending elsewherePrioritising of expenditures crucial to make cuts in lower priority areasFocus on cutting costs of HIV & AIDS programmes e.g. generics, lower cost servicesWork with donors to secure resources to maintain programme

Fiscal Recommendations

Consistent data a problem – spread across many spending departmentsNeed for NASA/NAANSF costings rough and readyNeed for more accurate and better documented NSF costings to:

Allow more accurate assessment of resource needsEnable updating using consistent methodsFacilitate consensus approachEngage meaningfully with donors

30

Thank You