Hello everybody! My name is Joshua Choi and I’m a 4th year Economics student at UC Berkeley. This semester I’m working on a cost-effectiveness project with the Berkeley EA group. My project is in collaboration with Solar4Africa, a charity that distributes subsidized solar products in rural Malawi and the goal of my particular project is to estimate the cost-effectiveness (CE) of solar pump irrigation systems and determine if it is competitive with other top-level Global Health and Welfare interventions or general EA charity donations. In this post I will cover the following: (1) background on solar pumps, (2) the potential value of solar pumps, and (3) how I will be valuing solar pump cost-effectiveness.
Background
To begin, I will outline the nature of the Solar4Africa solar pump intervention in Malawi. Rural areas in Malawi experience a dry season during which lower-income farmers or women gardening groups are typically unable to provide their crops sufficient water for a large harvest. Many if not most small farmers and women’s gardening groups in rural Malawi carry irrigation water by hand. Solar4Africa distributes and subsidizes small solar pump irrigation systems that allow these target groups to utilize the characteristic harsh sunlight of the dry season to power irrigation systems for their crops, providing an opportunity for them to grow a larger area of crops for consumption or sale. Following the intervention, Solar4Africa collected preliminary data on the impact of having solar pumps during the dry season from a non-random sample. With this early data, we can begin to build a picture of the potential benefits of offering these solar pump systems.
Preliminary estimates of cost-effectiveness
Preliminary data taken from interviews of a non-random sample of 8 beneficiaries indicates that introduction of solar pump irrigation for use during the dry season increases the annual cash income of Malawi farmers and gardening groups by an average of 48%. That being said, this number does have a high variance for different pump owners.
Currently the system costs approximately $200 and is subsidized approximately $100/pump by philanthropic donors.
An alternative engineering-based model that estimates income impacts shows that the pumps (which receive a $100 subsidy) may generate anywhere between $300 to $9,500 (90% confidence interval) of new farm income with an expected median value of $2,100. The model assumes 3 to 10 years of solar pump use. Another way of characterizing the cost-effectiveness is to estimate the benefit multiplier which is the total net benefits divided by the pump subsidy. The area irrigated is estimated by measuring the volume of water that is pumped with solar energy against the water requirements for the crop field, and the income is estimated from the expected yield per unit area and the expected gross profit per unit of crop harvested. A Monte Carlo simulation is used to calculate a range of outcomes based on the expected range of input parameters. The model creates a probability distribution that estimates a median benefit multiplier of $21 net benefit per $1 of subsidy for the solar pump.
One might ask why don’t farmers buy the solar pumps on the market if they are so beneficial? Solar pumps are available on the Malawi market but tend to be at least 5X more expensive than the ones distributed with this project because they are geared to much richer farmers. The smaller, subsidized version is affordable to a much larger number of low-income farmers. This also allows for bulk distribution of the pumping systems which lowers costs.
Though the cost-effectiveness numbers initially appear promising, it is important to refine the estimates and solidify their validity through more detailed cost-effectiveness analyses. I describe the more detailed cost-effectiveness methodology in the next section.
(Inverse marginal cost per unit of impact = Cost to donors / Net benefits of the interventions)
Each of MCI’s terms can be broken down further.
Cd, or more fully, the cost to donors per unit of intervention, can be estimated through the following equation:
Cd = Cprod • Psub • Md
(Cost to donors = Cost of product • Percent subsidy to beneficiaries • Markup factor)
NBs, or, the net benefit of the subsidized intervention that can be attributable to the impact of donor financing, can be estimated through the following equation:
NBs = NBann • Fattrib • UF • NPVsum
(Net benefits = Annualized benefit minus alternative benefit without intervention • fraction of
In the project data will be collected from Solar4Africa and compared to the academic literature to try to set reasonable ranges for each of the parameters in the CE equation.
Finally, to calculate the distribution of possible CE results, I will use a “poor man’s” Monte Carlo calculation in which each of the above parameters will be broken down into low, medium, and high values based on the average of the 0-33%, 33-66%, and 66-100% respectively. These numbers will come from a rough assessment and evaluation of reasonable ranges for the various parameters that we get from an overview of the literature and what has been done in other CE calculations such as those published by GiveWell.
Conclusion
Given the apparently high CE of this project, I am excited to see how the CE can be maximized. Any input and suggestions that people may have on factors and approaches that I can take in my analysis would be greatly appreciated. I plan to take any suggestions and finish the CE analysis in the next month.
Solar4Africa Project 3: Solar pumps for income generation–Project overview and methodology
Introduction
Hello everybody! My name is Joshua Choi and I’m a 4th year Economics student at UC Berkeley. This semester I’m working on a cost-effectiveness project with the Berkeley EA group. My project is in collaboration with Solar4Africa, a charity that distributes subsidized solar products in rural Malawi and the goal of my particular project is to estimate the cost-effectiveness (CE) of solar pump irrigation systems and determine if it is competitive with other top-level Global Health and Welfare interventions or general EA charity donations. In this post I will cover the following: (1) background on solar pumps, (2) the potential value of solar pumps, and (3) how I will be valuing solar pump cost-effectiveness.
Background
To begin, I will outline the nature of the Solar4Africa solar pump intervention in Malawi. Rural areas in Malawi experience a dry season during which lower-income farmers or women gardening groups are typically unable to provide their crops sufficient water for a large harvest. Many if not most small farmers and women’s gardening groups in rural Malawi carry irrigation water by hand. Solar4Africa distributes and subsidizes small solar pump irrigation systems that allow these target groups to utilize the characteristic harsh sunlight of the dry season to power irrigation systems for their crops, providing an opportunity for them to grow a larger area of crops for consumption or sale. Following the intervention, Solar4Africa collected preliminary data on the impact of having solar pumps during the dry season from a non-random sample. With this early data, we can begin to build a picture of the potential benefits of offering these solar pump systems.
Preliminary estimates of cost-effectiveness
Preliminary data taken from interviews of a non-random sample of 8 beneficiaries indicates that introduction of solar pump irrigation for use during the dry season increases the annual cash income of Malawi farmers and gardening groups by an average of 48%. That being said, this number does have a high variance for different pump owners.
Currently the system costs approximately $200 and is subsidized approximately $100/pump by philanthropic donors.
An alternative engineering-based model that estimates income impacts shows that the pumps (which receive a $100 subsidy) may generate anywhere between $300 to $9,500 (90% confidence interval) of new farm income with an expected median value of $2,100. The model assumes 3 to 10 years of solar pump use. Another way of characterizing the cost-effectiveness is to estimate the benefit multiplier which is the total net benefits divided by the pump subsidy. The area irrigated is estimated by measuring the volume of water that is pumped with solar energy against the water requirements for the crop field, and the income is estimated from the expected yield per unit area and the expected gross profit per unit of crop harvested. A Monte Carlo simulation is used to calculate a range of outcomes based on the expected range of input parameters. The model creates a probability distribution that estimates a median benefit multiplier of $21 net benefit per $1 of subsidy for the solar pump.
One might ask why don’t farmers buy the solar pumps on the market if they are so beneficial? Solar pumps are available on the Malawi market but tend to be at least 5X more expensive than the ones distributed with this project because they are geared to much richer farmers. The smaller, subsidized version is affordable to a much larger number of low-income farmers. This also allows for bulk distribution of the pumping systems which lowers costs.
Though the cost-effectiveness numbers initially appear promising, it is important to refine the estimates and solidify their validity through more detailed cost-effectiveness analyses. I describe the more detailed cost-effectiveness methodology in the next section.
Proposed cost-effectiveness (CE) Methodology
The lead organizer of the Solar4Africa charity has posted a proposed CE methodology on the EA forum: “A simplified cost-effectiveness estimation methodology for use in Solar4Africa’s Fall-23 student projects,” the solar pump CE analysis will utilize the following equation:
MCI = Cd / NBs
(Inverse marginal cost per unit of impact = Cost to donors / Net benefits of the interventions)
Each of MCI’s terms can be broken down further.
Cd, or more fully, the cost to donors per unit of intervention, can be estimated through the following equation:
Cd = Cprod • Psub • Md
(Cost to donors = Cost of product • Percent subsidy to beneficiaries • Markup factor)
NBs, or, the net benefit of the subsidized intervention that can be attributable to the impact of donor financing, can be estimated through the following equation:
NBs = NBann • Fattrib • UF • NPVsum
(Net benefits = Annualized benefit minus alternative benefit without intervention • fraction of
intervention adoption attributable to donations • usage fraction • discounted present value)
In the project data will be collected from Solar4Africa and compared to the academic literature to try to set reasonable ranges for each of the parameters in the CE equation.
Finally, to calculate the distribution of possible CE results, I will use a “poor man’s” Monte Carlo calculation in which each of the above parameters will be broken down into low, medium, and high values based on the average of the 0-33%, 33-66%, and 66-100% respectively. These numbers will come from a rough assessment and evaluation of reasonable ranges for the various parameters that we get from an overview of the literature and what has been done in other CE calculations such as those published by GiveWell.
Conclusion
Given the apparently high CE of this project, I am excited to see how the CE can be maximized. Any input and suggestions that people may have on factors and approaches that I can take in my analysis would be greatly appreciated. I plan to take any suggestions and finish the CE analysis in the next month.