ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
CEA of steady-state grant, EvAc TA costs only; Technical assistance (TA) marginal impact, excluding any additional costs incurred by govt
2
3
Increase in coverage per year without TA, in percentage points3%Assumption
4
Increase in coverage per year with TA, in percentage points6%Assumption
5
Implied % of coverage increase each year attributable to TA50%Calculation
6
7
201920202021202220232024
8
Coverage with TA34%40%46%52%58%64%
9
Coverage without TA31%34%37%40%43%46%
10
11
Coverage increase due to TA3%6%9%12%15%18%
12
13
Cost for TA$2,401,730$2,401,730$2,401,730$2,401,730$2,401,730$2,401,730
14
Population targeted28,739,54028,739,54028,739,54028,739,54028,739,54028,739,540
15
16
Increase in person-years of coverage due to TA862,1861,724,3722,586,5593,448,7454,310,9315,173,117
17
18
Cost per person-year of coverage due to TA$0.80
19
20
Implied: Beta's TA is X times more cost-effective than GiveDirectly13.9
21
22
Cost of TA per person targeted per year$0.08
23
24
Rough estimate to account for leverage/funging
25
Equivalent lives saved by coverage increase18,221
26
27
Counterfactual use of money, cost per life saved equivalent for government financial costs
$15,000
28
Additional money spent by government to achieve this coverage increase that's attributable to TA
$5,345,442
29
Counterfactual lives saved with govt spending356
30
Offsetting lives as % of lives saved by program2.0%
31
32
Factors left out:
33
- Number of years included is very important and I've made an arbitrary assumption. For slightly more information, see https://www.givewell.org/node/3109#Cost-effectiveness
34
- Govt staff time is largely excluded (except for what is implicitly factored into rows 27-30 above). For more information, see https://www.givewell.org/node/3106#Is_the_program_cost-effective
35
- If TA stops, coverage increases may continue without additional investment
36
- Did not discount costs and benefits in future years. Expected impacts to be small and offsetting.
37
- Did not include any benefits from national-level TA
38
- I don't account for declining counterfactual anemia rates over time via interventions aside from IFA properly. I've roughly factored it into the coverage rate above just by adding a percentage point for it, but this isn't the perfect mathematical way to incorporate it into the model. For slightly more information, see https://www.givewell.org/node/3109#Cost-effectiveness
39
- Did not factor in any impacts on birthweight, birth defects
40
41
42
CEA of Phase 2 grant, direct impacts only
43
Source
44
Cost of TA for this grant, Rajasthan$1,510,243
See https://www.givewell.org/node/3109#Budget
45
Cost of TA for this grant, MP$1,553,193
See https://www.givewell.org/node/3109#Budget
46
Cost of TA for this grant, Total$4,008,259
See https://www.givewell.org/node/3109#Budget
47
48
Target population, Rajasthan1,570,628
From an unpublished email from Evidence Action ("Questions about Beta IFA", sent Sept 12, 2018): Rajasthan Statewide target population: 13,350,342; Average target population at district level: 392,657; Average number of districts in a region: 4; Average regional-level target population: 1,570,628
49
Target population, MP1,508,740
From an unpublished email from Evidence Action ("Questions about Beta IFA", sent Sept 12, 2018): MP Statewide target population: 15,389,198; Average target population at district level: 301,748; Average number of districts in a region: 5; Average regional-level target population: 1,508,740
50
Number of kids targeted in relevant region3,079,368Calculation
51
52
Baseline coverage, Rajasthan17.5%
This is a rough best guess. The unpublished "Government of India, presentation on coverage of IFA program" reports 16-19% coverage in Rajasthan. Note: we are unsure how baseline coverage in chosen regions will differ from state-level coverage estimates.
53
Baseline coverage, MP45%
This is a rough best guess. The unpublished "Government of India, presentation on coverage of IFA program" reports 70% coverage in Madhya Pradesh, but according to Evidence Action's survey in MP (see survey discussed in "Evidence Action, WIFS National Concept Note" Pg. 5), 44% of schools are unaware of the program and only 1/3rd of schools reported implementing it regularly. Note: we are unsure how baseline coverage in chosen regions will differ from state-level coverage estimates.
54
Baseline coverage, overall31.0%Calculation
55
56
Increase in coverage per year without TA, in percentage points3%Assumption
57
Increase in coverage per year with TA, in percentage points7%Assumption
58
Implied % of coverage increase each year attributable to TA57%Calculation
59
60
2019202020212022
61
Coverage with TA38%45%52%59%
62
Coverage without TA34%37%40%43%
63
64
Coverage increase due to TA4%8%12%16%
65
66
Cost for TA$2,401,730$2,401,730$0$0
67
Population targeted3,079,3683,079,3683,079,3683,079,368
68
69
Increase in person-years of coverage due to TA123,175246,349369,524492,699
70
71
Cost per person-year of coverage due to TA$3.90Calculation
72
73
Implied: Beta's TA is X times more cost-effective than GD2.8See formula
74
75
Cost of TA per person targeted per year$0.65Calculation
76
77
78
CEA of Phase 2 grant, indirect effects on learning
79
80
Cost of the grant$4,328,259
See formula. Added $320K for the planning grant. We'll likely need to spend more than just this to figure this out.
81
Probability of this project ending up as 10x+ cash (at a level of analysis similar to current top charities)
35%
(Note: this is a conservative assumption. Implies that it's either 10x or not valuable at all.)
82
83
Funding that this project will be able to absorb at 10x+ cash (i.e., amount of funding that would be reallocated to this project if this grant succeeds)
$25,000,000
Very rough assumption of ~$1m per state per year, and that we'll fund 5 states for 5 years each at that level
84
85
Cost per under-5 life equivalent saved, GiveDirectly$11,663.00
Median from https://docs.google.com/spreadsheets/d/13b_qt-G_TQtoYNznNak3_5dzvzgCSUPJnk3l5dMisJo/edit#gid=1034883018
86
Cost per life saved, 10x cash$1,166.30Calculation
87
How much more cost-effective than cash will marginal GW recommendations be in the absence of this grant?
5
Guess based on our current estimate of cost-effectiveness of marginal dollars to GW top charities.
88
Cost per life saved, counterfactual use of marginal GW dollars$2,332.60Calculation
89
90
Lives saved if funding amount inputted above is spent on this project21,435
Assumes this project will end up 10x cash
91
Lives saved if funding amount inputted above is spent on counterfactual10,718Calculation
92
93
Net lives saved10,718Calculation
94
Net lives saved, adjusted for probability of the project succeeding3751Calculation
95
96
Cost per life saved from this grant$1,154Calculation
97
This grant is X times more cost-effective than cash10.1Calculation
98
This grant is X times more cost-effective than marginal spending on GW top charities now
2.0Calculation
99
100