"Estimating the cost and financing gap for meeting CAADP growth and MDG1 targets", presentation by Sam Benin at the USAID, IFPRI Financial Gap Analysis Workshop held at the World Bank, January 7, 2010.

1. IFPRI
Estimating the cost and financing gap for
meeting CAADP growth and MDG1
targets
Sam Benin
International Food Policy Research Institute
USAID/World Bank Workshop on
“Agricultural investment priorities and financing gaps for achieving growth and
poverty reduction targets: Review of evidence and methodology”
January 7, 2010
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE

2. Introduction
 Existence of several cost estimates for attaining
the MDGs has raised the need the most
appropriate methodology to obtain consistent
and reliable projections
 However, the issue is not merely technical. There
is need to also consider the political motivations
 As relatively “large” or “small” estimates will
generate different reactions in donor and
developing countries, developing accurate
methodologies appears critical for both parties
 Four main approaches have been used to cost
MDGs
IFPRI

3. Approaches and limits to MDGs costing
Source: NallariIFPRI Heuty, 2004
and

4. Estimated required resources to meet
MDG1
Methodology Studies Estimates Remarks
Intervention-based Anti-poverty $24 billion
program
Aggregate unit costs Rosegrant et $238 billion from
al. (2005) 1997-2025
UN Reports $ per capita in
(2005) 2006: Ghana=80;
Tanzania=96;
Uganda=92
ICOR Devarajan et $54-62 billion
al. (2002) per year
Input-outcome Zedillo $20 billion per
elasticity report year
IFPRI

5. Sources of discrepancies in MDGs costs
estimates
 Interpretation of targets and baselines
 Countries covered
 Underlying assumptions (economic
growth, population growth, resource
mobilization and allocation, institutional
reform, etc.)
 Data sources
 Unit costs and elasticity parameters
 Alternative scenarios
…
IFPRI

6. Estimating agricultural spending required to achieve
CAADP growth and the MDG1
 Rationale
» Previous studies focused on costing the MDGs
(whether at the global, regional, or country level)
have ignored agricultural financial resources
 Elasticity approach
» From the policy perspective of using public spending
for stimulating growth and reducing poverty,
methods based on expenditure-growth, expenditure-
poverty, and growth-poverty elasticities are
conceptually sound
» Elasticity measures of the relative change in the
outcome with respect to change in expenditures (or
inputs), taking into account any conditioning and
confounding factors, including lag between
expenditures and realization of the outcome
IFPRI

7. Issues to consider
 Relative effect of public investment in the agricultural
and non-agricultural sectors
 Public investment is not be growth-neutral: different
types of public investment (across and within sectors)
affect growth and poverty differently via different
pathways and at different levels
 Relative productivity or efficiency of public versus
private investment in overall economic growth
 Plausible crowding-out effect of public investment on
private investment
 Interaction effects among the different types of
investment
 Initial conditions of development and pattern of growth
IFPRI

8. Estimation of required growth and spending
 Poverty-growth elasticity
» decompose “elasticity of poverty with respect to growth”
into effects of agricultural and non-agricultural growth
and an interaction term that captures a linkage or
multiplier effect
 Growth-spending elasticity
» decompose “elasticity of agricultural (and non-
agricultural) growth with respect to public spending” into
the effects of growth in different types agricultural and
non-agricultural spending and interaction terms that
captures complementarity (substitution) effects among
different types of spending
 Initial conditions of development and pattern of growth
» Resource endowments, climate, institutions, etc.
IFPRI

9. Review of the evidence
Elasticities and growth rates
IFPRI

10. Elasticity of poverty with respect to agricultural
and non-agricultural growth
 Import table from Fan et al
IFPRI

11. Elasticity of agricultural productivity with respect to
public agricultural spending
Indicator of public agricultural investment Dependent variable Elasticity Source/Country
Government investment:
Agriculture Ag Output 0.085 Fan et al., 2008a (44 Developing countries, including 17
Research Ag Output 0.038 from Africa)
Non-research Ag Output –0.070
Research (R&D) Ag GDP per hectare Thirtle et al. 2003 (48 developing countries, including 22
All countries 0.442 from Africa)
SSA 0.363
Asia 0.344
Latin America 0.197
Research (R&D) Ag GDP per capita
All countries 0.304
SSA 0.264
Asia 0.231
Latin America 0.093
Research and extension Ag output per capita 0.189 Fan et al., 2004 (Uganda)
Agriculture Ag output per capita 0.153 Benin et al., 2008b (Ghana)
Research Ag GDP per capita 0.085 Fan et al., 2002 (China)
Irrigation Ag GDP per capita 0.101
Research Ag output per worker 0.464 Fan et al., 2008c (Thailand)
Research TFP 0.049–0.066 Rosegrant and Evenson, 1995 (India)
Research TFP 0.255 Fan, Hazell and Thorat, 2000 (India)
Irrigation TFP 0.036
Soil and water conservation TFP 0.002n
Irrigation TFP 0.003 Teurel and Kuroda, 2005 (Philippines)
Non-government investment:
Official development assistance (ODA) Ag GDP 0.03 Schuh and Norton, 1991 (98 developing countries)
Other indicators:
Agricultural extension (staff per 1000 TFP 0.041–0.063 Rosegrant and Evenson, 1995 (India)
farms)
Domestic research (scientists per ha of TFP 2.69 Johnson and Evenson, 2000 (90 Least developed
arable land) countries)
Foreign research (spending per ha of TFP 10.27
arable land)
IFPRI

12. Elasticity of agricultural productivity with respect to
public non-agricultural spending
Indicator of public non-agricultural investment Dependent variable Value of coefficient Source/Country
Education
Literacy rate Ag Output 0.362n Fan et al., 2008a (44 Developing countries, including 17 from
Africa)
Rural literacy rate Ag output per capita 0.332 Fan et al., 2004 (Uganda)
Share of people completed at least primary Ag output per capita –0.11 Benin et al., 2008b (Ghana)
education
Spending on education Ag GDP per capita 0.197 Fan et al., 2002 (China)
Expenditure on rural education TFP 0.047 Fan, Hazell and Thorat, 2000 (India)
Spending on education Ag output per worker 0.578 Fan et al., 2008c (Thailand)
Health
Share of people sick last month Ag output per capita –0.465 Fan et al., 2004 (Uganda)
Share of people living more than 15 minutes of Ag output per capita –0.81 Benin et al., 2008b (Ghana)
a health center
Spending on public health and welfare TFP 0.012n Fan, Hazell and Thorat, 2000 (India)
Roads
Density (km/1000km2) Ag Output –0.092n Fan et al., 2008a (44 Developing countries, including 17 from
Africa)
Distance to feeder road Ag output per capita –0.139 Fan et al., 2004 (Uganda)
Feeder road density Ag output per capita 0.13 Benin et al., 2008b (Ghana)
Spending on rural roads Ag GDP per capita 0.037 Fan et al., 2002 (China)
Road density TFP 0.042 Zhang and Fan, 2004 (India)
Investment on rural roads TFP 0.057 Fan, Hazell and Thorat, 2000 (India)
Spending on rural roads Ag output per worker 0.119 Fan et al., 2008c (Thailand)
Investment on roads TFP 0.015 Teurel and Kuroda, 2005 (Philippines)
Other public investments
Spending on rural power TFP 0.004n Fan, Hazell and Thorat, 2000 (India)
Spending on rural power Ag GDP per capita 0.009n Fan et al., 2002 (China)
Spending on rural power Ag output per worker 0.198 Fan et al., 2008c (Thailand)
Investment on electrification TFP 0.002 Teurel and Kuroda, 2005 (Philippines)
Spending on rural development TFP 0.022n Fan, Hazell and Thorat, 2000 (India)
Crop area under public irrigation TFP 0.036 Fan, Hazell and Thorat, 2000 (India)
Spending on rural telecommunications Ag GDP per capita 0.021 Fan et al., 2002 (China)
IFPRI

13. Effect of public spending on factors of
agricultural production and input use
Dependent variable Value of Source/Country
coefficient
Indicator of public agricultural
investment
Investment on irrigation Agricultural labor –0.233 Teurel and Kuroda, 2005 (Philippines)
Investment on irrigation Intermediate inputs –0.501
Investment on irrigation Agricultural capital 0.650
Government expenditures on Household total agricultural 0.148 Benin et al., 2008b (Ghana)
agriculture expenditures per capita
Indicator of public non-agricultural
investment
Share of people completed at Household total agricultural 0.459 Benin et al., 2008b (Ghana)
least primary education expenditures per capita
Share of people living more than Household total agricultural –0.359
15 minutes of a health center expenditures per capita
Feeder road density Household total agricultural –0.045n
expenditures per capita
Investment on roads Agricultural labor –1.189 Teurel and Kuroda, 2005 (Philippines)
Investment on roads Intermediate inputs –1.052n
Investment on roads Agricultural capital 1.806
Investment on electrification Agricultural labor –0.099 Teurel and Kuroda, 2005 (Philippines)
Investment on electrification Intermediate inputs –0.216
Investment on electrification Agricultural capital 0.499
IFPRI

14. Crowding-in and crowding-out effects of
public on private investments
Indicator of public investment Dependent variable (Indicator of Value of coefficient Source/Country
private investment or market)
Public investment Private investment 0.027–0.067n Ashipala and Haimbodi, 2003 (South Africa)
Public investment Private investment 0.312–1.108n Ashipala and Haimbodi, 2003 (Namibia)
Public investment Private investment –0.021 to 0.022n Ashipala and Haimbodi, 2003 (Botswana)
Expenditures on public applied Expenditures on private applied 0.25–0.28 Malla and Gray, 2005 (USA)
research research
Expenditures on public basic Expenditures on private applied 0.20–0.22
research research
Subsidy on research Expenditures on private research 0.10 Görg and Strobl, 2006 (Ireland)
Stocks of public R&D Stocks of private R&D 0.035–1.918 Sadraoui and Ben Zina, 2006 (23 countries
including 3 from Africa)
Share of public investment in GDP Share of private investment in GDP –0.082 Ramirez and Nazmi, 2003 (9 Latin American
countries)
Ratio of public to private investment Overall TFP –0.23 del Mar Salinas-Jimemez, 2004 (Spain)
Ratio of public to private investment Ag TFP –0.001n
Expenditures on public irrigation Crop area under private irrigation 0.08 Fan, Hazell and Thorat, 2000 (India)
(%)
Crop area under public irrigation (%) Crop area under private irrigation 0.92
(%)
Spending on research Rural wages 0.033 Fan, Hazell and Thorat, 2000 (India)
Public wages Private wages 0.212–0.357 Afonso and Gomes, 2008 (16 OECD countries)
IFPRI

15. Interaction effects among different types of
public spending
Explanatory variable Dependent variable Value of coefficient Source/Country
Interactions
Fertilizer and stone terrace Household agricultural output –0.804; –0.076n Pender and Gebremedhin, 2006
per acre (Ethiopia). Estimates are for two
different methods.
Fertilizer and soil bund Household agricultural output 0.369n; –0.455
per acre
Fertilizer and irrigation Household agricultural output 0.663n; 0.131n
per acre
Neighborhood effects
Tax rate of neighbors Tax rate 0.158–0.314 Hauptmeier et al., 2009 (Germany)
Public spending of neighbors Public spending 0.178–0.507
Public social spending Public education spending 0.265–0.410 Busemeyer, 2007 (21 OECD
Decentralization Public education spending 0.134–0.271 countries)
Decentralization Public health spending 0.015n
Decentralization Public social spending –0.042 to –0.099
Decentralization Public total spending 0.046
Public total spending Ratio of spending on other –0.82 to –1.51 Ramajo et al., 2007 (Spain)
services to spending on
economic services
IFPRI

16. Public agricultural spending growth rates
Total Agriculture
Country Expenditure Expenditure
Benin 7.66 12.98
Botswana 2.41 -2.48
Burkina Faso 21.42 11.05
Burundi 16.84 19.80
Cameroon 3.83 8.21
Central African Republic 15.69 -4.46
Chad -0.18 3.70
Congo, Dem. Rep. 26.95 30.21
Congo, Rep. -21.78 -1.09
Cote d'Ivoire 3.09 4.26
Djibouti 7.17 51.90
Egypt, Arab Rep. -0.19 3.84
Ethiopia 10.97 38.62
Ghana 21.47 35.32
Guinea-Bissau 18.03 5.57
Kenya 16.60 13.91
Lesotho 10.16 -2.37
Madagascar 19.10 21.86
Malawi 12.13 36.44
Mali 11.09 6.76
Mauritania 0.20 -4.42
Morocco 8.52 -7.66
Mozambique 9.26 -20.12
Namibia 8.94 -1.64
Niger -1.36 -13.96
Nigeria -0.10 13.55
Sao Tome and Principe 28.09 56.47
Senegal 11.07 23.33
Seychelles -2.36 5.80
Sierra Leone 0.52 -1.41
Swaziland 12.25 20.99
Tanzania 15.20 17.72
Source: Nin Pratt Togo Yu, 2009
and 5.48 14.48
Tunisia 5.30 3.85
IFPRI

17. Growth rates of factors and productivity, and
information on other parameters
Fertilizer Tractors Animal stock Worker Output Output TFP
per hectare per per hectare per per per
hectare hectare hectare worker
Benin 6.37 -3.58 -2.73 -3.60 0.07 3.81 1.67
Burkina Faso -1.26 -1.07 0.52 -1.46 1.25 2.76 1.32
Cameroon 1.29 0.00 2.19 0.69 2.62 1.91 1.84
Chad 10.71 -0.31 2.52 0.89 2.71 1.80 2.48
Congo 4.14 -0.54 1.59 0.23 1.68 1.45 1.39
Cote d’Ivoire 4.75 -0.25 1.01 -0.40 2.09 2.50 1.60
Ethiopia 1.70 -1.79 1.63 0.68 2.49 1.79 2.55
Gabon -2.76 -0.73 0.00 -1.53 0.75 2.31 2.31
Ghana 5.27 -3.96 -1.48 -1.31 1.57 2.92 1.79
Guinea -3.05 -0.27 2.41 -0.72 0.87 1.60 0.42
Guinea-Bissau 10.36 -2.73 -1.17 -0.80 -0.13 0.67 0.45
Kenya 0.29 0.77 -0.82 1.18 1.30 0.11 1.05
Malawi 7.34 -2.63 -1.20 -1.48 3.23 4.78 3.35
Mali 3.97 -0.44 2.04 0.52 2.25 1.72 2.85
Mauritania -2.55 1.01 3.87 1.60 2.02 0.41 1.44
Mauritius -2.12 0.00 2.75 -1.75 0.97 2.76 0.93
Mozambique 4.63 -1.85 -1.01 0.04 2.79 2.75 3.32
Nigeria -5.45 0.85 0.82 -0.06 2.02 2.08 2.12
Sudan 0.19 -0.29 1.39 -0.14 1.64 1.78 3.19
Tanzania -14.94 0.38 1.69 0.66 0.74 0.09 2.79
Togo 2.97 -2.27 1.33 0.55 0.96 0.41 0.59
Zambia 1.25 -0.03 0.46 0.77 1.23 0.46 0.03
Source: Nin Pratt and Yu, 2009
IFPRI

18. Application
 Successful application depends on the extent to which
information on the different parameters is available
 It is unlikely, actually unrealistic, to obtain information on all
the parameters for every country in Africa
 Parameter estimates from similar countries or the regional
level would have to be used in the cost calculations for
countries where such information is lacking
 How the value of the parameters change over time (or do not
change) would have to be decided upon
 Obtaining a range of estimates would be more prudent than
point estimates
» the lower end of the range would correspond to an optimistic
spending scenario characterized by (e.g. high spending
efficiency, greater crowding-in effect on private investments,
and positive interaction effect with other types of spending)
» vice versa for the upper end of the range
IFPRI

19. Africa-wide estimates
 Michael
IFPRI

20. Country-level estimates
 Use evidence from different countries to assess the aggregate public
agricultural expenditures (PAE) required to reach the CAADP and MDG1
growth targets in the next 10 years (2005-15) for selected countries
 Elasticity of agricultural productivity with respect to public agricultural
spending: 0.15 and as low- and high-end values or a less and more optimistic
public spending efficiency scenario, respectively.
 Scenarios:
» Baseline: public agricultural and non-agricultural spending in 2004 constant prices
continue to grow according to their respective recent (1999-2005) trends. Other factors
(e.g. interactions between different types of spending, crowding effects of public spending
on private investments, non-spending factors affecting agricultural growth) remain
unchanged.
» Accelerated public agricultural and non-agricultural expenditure growth speeds up too to
match with the higher growth rate required in the agricultural and non-agricultural
GDP. For the latter, we use low-end and high-end elasticity values of 0.15 and 0.25,
respectively.
 Other assumptions
» Interaction effects remain unchanged as in the baseline scenario and are already reflected
in the estimated elasticities with respect agricultural and non-agricultural spending
» Non-spending factors that affect agricultural growth (e.g. weather, policies, prices) are
difficult to model and so are assumed to remain unchanged as in the baseline scenario.
IFPRI

21. Annual average growth (%) in aggregate public agricultural expenditures
required to achieve CAADP growth and MDG1 (2005-15)
CAADP MDG1
baseline low high low high
Malawi 13.8 34.8 24.1 37.2 24.1
Rwanda -6.5 30.3 15.2 45.6 22.6
Uganda 14.8 35.1 23.1 35.1 23.1
Zambia 8.4 31.9 20.1 44.6 26.4
IFPRI