# -*- coding: utf-8 -*-

"""

Created on Thu Jun 09 23:35:58 2016

@author: dan

"""

import json

import scipy.stats as stats

import numpy as np

import matplotlib.pyplot as plt

plt.style.use('ggplot')

import pandas as pd

def get_rvs(p, n):

return p['val']

def calc_npv(r, ubi, inputs, n):

return z

def calc_deworming(s, deworming, inputs):

return z

if __name__ == '__main__':

n = 100000

m = 1000.0

key = 'DALYs per $' + str(m)

#key = 'Cost per equivalent life saved'

with open('params.json') as fp:

params = json.load(fp)

inputs = {}

for k in params.keys():

d = {}

for p in params[k].keys():

d[p] = get_rvs(params[k][p], n)

inputs[k] = d

cash = {}

cash['Total size of transfer (in nominal USD)'] = 1000.0

cash['Average household size'] = 4.7

cash['Size of transfer per person'] = cash['Total size of transfer (in nominal USD)']/ \

cash['Average household size']

cash['Amount invested'] = cash['Size of transfer per person']* \

inputs['GD']['Percentage of transfers invested - Standard program']

cash['Initial increase in consumption'] = (1.0 - inputs['GD']['Percentage of transfers invested - Standard program'])* \

cash['Size of transfer per person']

cash['Annual increase in consumption made possible by investment returns'] = cash['Amount invested']* \

inputs['GD']['Return on investment - Standard program']

cash['Baseline annual consumption per capita (in nominal USD)'] = 285.922288106034

cash['Initial increase in ln(consumption)'] = np.log(cash['Initial increase in consumption'] + \

cash['Baseline annual consumption per capita (in nominal USD)']) - \

np.log(cash['Baseline annual consumption per capita (in nominal USD)'])

cash['Future annual increase in ln(consumption)'] = np.log(cash['Annual increase in consumption made possible by investment returns'] + \

cash['Baseline annual consumption per capita (in nominal USD)']) - \

np.log(cash['Baseline annual consumption per capita (in nominal USD)'])

cash['Present value of future increase in ln(consumption) (excluding the last year)'] = cash['Future annual increase in ln(consumption)']/ \

(1.0 + inputs['Shared']['Discount rate'])* \

(1.0 - 1.0/np.power((1.0 + inputs['Shared']['Discount rate']), (inputs['GD']['Duration of investment benefits (in years) - Standard program'] - 1.0)))/ \

(1.0 - 1.0/(1.0 + inputs['Shared']['Discount rate']))

cash['Present value of increased ln(consumption) in the last year'] = \

(np.log(cash['Baseline annual consumption per capita (in nominal USD)'] + \

cash['Amount invested']* \

(inputs['GD']['Return on investment - Standard program'] + inputs['GD']['Percent of investment returned when benefits end - Standard program'])) - \

np.log(cash['Baseline annual consumption per capita (in nominal USD)'])) / \

np.power((1.0 + inputs['Shared']['Discount rate']), inputs['GD']['Duration of investment benefits (in years) - Standard program'])

cash['Present value of total future increase in ln(consumption)'] = cash['Present value of future increase in ln(consumption) (excluding the last year)'] + \

cash['Present value of increased ln(consumption) in the last year']

cash['Total present value of cash transfer'] = cash['Present value of total future increase in ln(consumption)'] + \

cash['Initial increase in ln(consumption)']

cash['Proportional increase in consumption per dollar'] = cash['Total present value of cash transfer']* \

inputs['GD']['Transfers as a percentage of total cost - Standard program']/ \

cash['Size of transfer per person']

ubi = {}

ubi['Annual transfer size per person over 18 (nominal USD)'] = 0.75*365.25

ubi['Average number of people over 18 in each household'] = 2.395555556

ubi['Total amount transfered to each household'] = ubi['Annual transfer size per person over 18 (nominal USD)']* \

ubi['Average number of people over 18 in each household']

ubi['Household size'] = 4.7

ubi['Transfer size per person'] = ubi['Total amount transfered to each household']/ubi['Household size']

ubi['Annual quantity of transfer money used for immediate consumtion (pre-discounting)'] = \

ubi['Transfer size per person']*(1.0 - inputs['UBI']['Percentage of transfers invested - UBI'])

ubi['Size of transfer invested each year'] = ubi['Transfer size per person']*inputs['UBI']['Percentage of transfers invested - UBI']

ubi['Annual return for each year of transfer investments (pre-discounting)'] = \

ubi['Size of transfer invested each year']*inputs['UBI']['Return on investment - UBI']

ubi['Value eventually returned from one years investment (pre-discounting)'] = ubi['Size of transfer invested each year'] * \

inputs['UBI']['Percent of investment returned when benefits end - UBI']

ubi['Duration of program - Long Term Arm'] = 12.0

ubi['Duration of program - Short Term Arm'] = 2.0

ubi['Baseline consumption per capita'] = 285.922288106034

ubi['Inflation for lack of targeting'] = (0.86-0.5)/0.5*0.3

ubi['Expected baseline per capita consumption (nominal USD)'] = ubi['Baseline consumption per capita']* \

(1.0 + ubi['Inflation for lack of targeting'])

ubi['Adjusted per capita consumption (nominal USD)'] = ubi['Expected baseline per capita consumption (nominal USD)']* \

inputs['UBI']['Work participation adjustment']

ubi['Initial benefit (in terms of ln[consumption])'] = np.log(ubi['Annual quantity of transfer money used for immediate consumtion (pre-discounting)'] + \

ubi['Adjusted per capita consumption (nominal USD)']) - \

np.log(ubi['Expected baseline per capita consumption (nominal USD)'])

ubi['Years Post Transfer'] = 100

ubi['Net present value of benefits beyond initial year of program (in terms of ln[consumption]) - Long Term Arm'] = \

calc_npv(ubi['Duration of program - Long Term Arm'], ubi, inputs, n)

ubi['Net present value of entire program (in terms of ln[consumption]) - Long Term Arm'] = \

ubi['Net present value of benefits beyond initial year of program (in terms of ln[consumption]) - Long Term Arm'] + \

ubi['Initial benefit (in terms of ln[consumption])']

ubi['Efficiency of UBI program relative to standard program'] = 0.93

ubi['Transfers as a percentage of total cost - UBI'] = inputs['GD']['Transfers as a percentage of total cost - Standard program']* \

ubi['Efficiency of UBI program relative to standard program']

ubi['Per capita transfer cost over entire program - Long Term Arm'] = ubi['Transfer size per person']* \

ubi['Duration of program - Long Term Arm']/ubi['Transfers as a percentage of total cost - UBI']

ubi['Proportional increase in consumption per dollar - Long Term Arm'] = \

ubi['Net present value of entire program (in terms of ln[consumption]) - Long Term Arm']/ \

ubi['Per capita transfer cost over entire program - Long Term Arm']

ubi['Net present value of benefits beyond initial year of program (in terms of ln[consumption]) - Short Term Arm'] = \

calc_npv(ubi['Duration of program - Short Term Arm'], ubi, inputs, n)

ubi['Net present value of entire program (in terms of ln[consumption]) - Short Term Arm'] = \

ubi['Net present value of benefits beyond initial year of program (in terms of ln[consumption]) - Short Term Arm'] + \

ubi['Initial benefit (in terms of ln[consumption])']

ubi['Per capita transfer cost over entire program - Short Term Arm'] = ubi['Transfer size per person']* \

ubi['Duration of program - Short Term Arm']/ubi['Transfers as a percentage of total cost - UBI']

ubi['Proportional increase in consumption per dollar - Short Term Arm'] = \

ubi['Net present value of entire program (in terms of ln[consumption]) - Short Term Arm']/ \

ubi['Per capita transfer cost over entire program - Short Term Arm']

ubi['Percent of transfers going to short term arm'] = 7.9/30.0

ubi['Percent of transfers going to long term arm'] = 1.0 - ubi['Percent of transfers going to short term arm']

ubi['Weighted proportion increase in consumption per dollar'] = ubi['Percent of transfers going to short term arm']* \

ubi['Proportional increase in consumption per dollar - Short Term Arm'] + \

ubi['Percent of transfers going to long term arm']*ubi['Proportional increase in consumption per dollar - Long Term Arm']

cash['Weight of UBI program in overall cost-effectiveness of GiveDirectly'] = 1.0 - \

inputs['GD']['Weight of standard program in overall cost-effectiveness of GiveDirectly']

cash['Weighted proportional increase in consumption per dollar'] = inputs['GD']['Weight of standard program in overall cost-effectiveness of GiveDirectly']* \

cash['Proportional increase in consumption per dollar'] + ubi['Weighted proportion increase in consumption per dollar']* \

cash['Weight of UBI program in overall cost-effectiveness of GiveDirectly']

cash['Cost per equivalent life saved'] = inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']* \

inputs['Shared']['1 DALY averted is equivalent to increasing ln(consumption) by one unit for one individual for how many years?']/ \

cash['Weighted proportional increase in consumption per dollar']

cash['DALYs per $' + str(m)] = m*inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

cash['Cost per equivalent life saved']

# long term effects

deworming = {}

deworming['Benefit on one year\'s income (discounted back because of delay between deworming and working for income)'] = \

inputs['Deworming']['Treatment effect of deworming on ln(consumption)']/ \

np.power((1.0 + inputs['Shared']['Discount rate']), inputs['Deworming']['Average number of years between deworming and the beginning of long term benefits'])

deworming['Present value of the sum of the lifetime benefits per worker (in terms of Ln(income))'] = \

deworming['Benefit on one year\'s income (discounted back because of delay between deworming and working for income)']* \

(1.0 - 1.0/np.power((1.0 + inputs['Shared']['Discount rate']), inputs['Deworming']['Duration of long term benefits of deworming (in years)']))/ \

(1.0 - 1.0/(1.0 + inputs['Shared']['Discount rate']))

deworming['Adjusted long term benefits per year of treatment (in terms of ln $), assuming income supports household consumption (before adjusting for alternate funders)'] = \

deworming['Present value of the sum of the lifetime benefits per worker (in terms of Ln(income))']* \

inputs['Deworming']['Number of household members that benefit - Deworming']* \

inputs['Deworming']['Adjustment for El Nino']* \

inputs['Deworming']['Proportion of child-years that are as helpful (in terms of developmental effects) as the years in Baird et al.']* \

inputs['Deworming']['Proportion of dewormed children that benefited from long term gains in Baird et al.']* \

inputs['Deworming']['Replicability adjustment for deworming']/ \

inputs['Deworming']['Additional years of treatment assigned to Baird\'s treatment group']

deworming['Adjusted long term benefits per year of treatment (in terms of ln $), assuming income supports household consumption'] = \

deworming['Adjusted long term benefits per year of treatment (in terms of ln $), assuming income supports household consumption (before adjusting for alternate funders)']* \

inputs['Deworming']['Deworming alternate funders adjustment']

# short term effects

deworming['Short term health benefits of deworing in terms of ln(income)'] = \

inputs['Shared']['1 DALY averted is equivalent to increasing ln(consumption) by one unit for one individual for how many years?']* \

inputs['Deworming']['Short term health benefits of deworming (DALYs averted per person treated)']* \

inputs['Deworming']['Deworming alternate funders adjustment']

# deworming specific charities

dtw = calc_deworming('DtW', deworming, inputs)

dtw['DALYs per $' + str(m)] = m*inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

dtw['Cost per equivalent life saved']

sci = calc_deworming('SCI', deworming, inputs)

sci['DALYs per $' + str(m)] = m*inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

sci['Cost per equivalent life saved']

ss = calc_deworming('SS', deworming, inputs)

ss['DALYs per $' + str(m)] = m*inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

ss['Cost per equivalent life saved']

# bednets

bednets = {}

bednets['Deaths averted per protected child under 5 according to Lengeler 2004\'s Summary Effect'] = 5.53/1000.0

bednets['Under 5 all mortality rate 1995 in the countries where RCT studies were done - Burkina Faso, Gambia, Ghana, and Kenya (per 1000)'] = 123.565

bednets['Pre-distribution wastage'] = 0.05

bednets['Under 5 all mortality rate 2015 (per 1000)'] = 63.2938

bednets['Ratio of mortality rate in 2015 compared to 1995'] = bednets['Under 5 all mortality rate 2015 (per 1000)'] / \

bednets['Under 5 all mortality rate 1995 in the countries where RCT studies were done - Burkina Faso, Gambia, Ghana, and Kenya (per 1000)']

bednets['Under 5 all mortality rate 2004 (per 1000)'] = 113

bednets['Ratio of mortality rate in 2004 compared to 1995'] = bednets['Under 5 all mortality rate 2004 (per 1000)']/ \

bednets['Under 5 all mortality rate 1995 in the countries where RCT studies were done - Burkina Faso, Gambia, Ghana, and Kenya (per 1000)']

bednets['Percent of mortality decline between 2004 and 2015 attributable to ITNs'] = 0.25

bednets['Adjustment for ratio of current child mortality to historical child mortality when RCTs were conducted'] = bednets['Ratio of mortality rate in 2015 compared to 1995'] + \

(bednets['Ratio of mortality rate in 2004 compared to 1995'] - bednets['Ratio of mortality rate in 2015 compared to 1995'])* \

bednets['Percent of mortality decline between 2004 and 2015 attributable to ITNs']

bednets['Efficacy reduction due to insecticide resistance'] = 0.26

bednets['Percent of country population with IRS coverage (Current inputs based on 2015 Malaria Atlas Project estimates)'] = 0.0826

bednets['Reduction in ITN efficacy when ITNs are used in conjunction with IRS'] = 0.5

bednets['Efficacy reduction attributable to IRS'] = bednets['Percent of country population with IRS coverage (Current inputs based on 2015 Malaria Atlas Project estimates)']* \

bednets['Reduction in ITN efficacy when ITNs are used in conjunction with IRS']

bednets['Adjustment for differences in malaria burdens between geographic areas'] = 0.86

bednets['Deaths averted per protected child under 5 - adjusted for today\'s lower rates of child mortality, insecticide resistance, and IRS'] = \

bednets['Deaths averted per protected child under 5 according to Lengeler 2004\'s Summary Effect']* \

bednets['Adjustment for ratio of current child mortality to historical child mortality when RCTs were conducted']* \

bednets['Adjustment for differences in malaria burdens between geographic areas']* \

(1.0 - bednets['Efficacy reduction due to insecticide resistance'])* \

(1.0 - bednets['Efficacy reduction attributable to IRS'])

bednets['Adjustment for net use in AMF distributions relative to net use in Lengeler'] = 0.8

bednets['Effective deaths averted per under 5 person-year in community-wide distributions'] = \

bednets['Deaths averted per protected child under 5 - adjusted for today\'s lower rates of child mortality, insecticide resistance, and IRS']* \

inputs['AMF']['Replicability adjustment - ITNs']* \

bednets['Adjustment for net use in AMF distributions relative to net use in Lengeler']* \

inputs['AMF']['External validity adjustment for declines in malaria mortality (to make ITN model consistent with SMC model) - Autocalculated']

bednets['Percent of population under 5 (used for mortality effects and development effects)'] = 0.1740241

bednets['Percent of population ages 5-14'] = 0.30426622

bednets['Cost per LLIN'] = 4.85

bednets['Number of LLINs distributed per person in the community'] = 1.0/1.8

bednets['Cost per person covered in a universal distribution'] = bednets['Cost per LLIN']* \

bednets['Number of LLINs distributed per person in the community']

bednets['Percent of all individuals owning nets (in absence of a distribution)'] = 0.3

bednets['Proportion of "extra" nets distributed that are eventually used'] = 0.5

bednets['Adjustment for unused nets'] = (1.0 - bednets['Proportion of "extra" nets distributed that are eventually used']* \

bednets['Percent of all individuals owning nets (in absence of a distribution)'])

bednets['Adjustment for unused nets and alternate funders'] = inputs['AMF']['ITN alternative funders adjustment']* \

bednets['Adjustment for unused nets']

bednets['Arbitrary donation size'] = 1000000

bednets['Donation size after costs from pre-distribution wastage'] = bednets['Arbitrary donation size']* \

(1.0 - bednets['Pre-distribution wastage'])

bednets['Number of people covered'] = bednets['Donation size after costs from pre-distribution wastage']/ \

bednets['Cost per person covered in a universal distribution']

bednets['Under 5\'s covered'] = bednets['Number of people covered']*bednets['Percent of population under 5 (used for mortality effects and development effects)']

bednets['Children 5-14 covered'] = bednets['Number of people covered']*bednets['Percent of population ages 5-14']

bednets['Lifespan of an LLIN'] = 2.22

bednets['Person-years of coverage for under 5\'s'] = bednets['Under 5\'s covered']*bednets['Lifespan of an LLIN']

bednets['Person-years of coverage for ages 5-14'] = bednets['Children 5-14 covered']*bednets['Lifespan of an LLIN']

bednets['Person years of protection for under 14\'s'] = bednets['Person-years of coverage for under 5\'s'] + bednets['Person-years of coverage for ages 5-14']

bednets['Cost per person year of protection for under 14\'s'] = bednets['Arbitrary donation size']/ \

bednets['Person years of protection for under 14\'s']

bednets['Cost per person-year of protection, adjusted for insecticide resistance: under-14\'s'] = bednets['Cost per person year of protection for under 14\'s']/ \

(1.0 - bednets['Efficacy reduction due to insecticide resistance'])

bednets['Total under 5 lives saved by AMF nets and pre-existing nets'] = bednets['Person-years of coverage for under 5\'s']* \

bednets['Effective deaths averted per under 5 person-year in community-wide distributions']

bednets['Cost per death averted (before adjusting for alternate funders and pre-existing nets)'] = bednets['Arbitrary donation size']/ \

bednets['Total under 5 lives saved by AMF nets and pre-existing nets']

bednets['Under 5 lives saved on the margin'] = bednets['Total under 5 lives saved by AMF nets and pre-existing nets']* \

bednets['Adjustment for unused nets and alternate funders']

bednets['Marginal cost per under 5 death averted'] = bednets['Arbitrary donation size']/ \

bednets['Under 5 lives saved on the margin']

bednets['Proportional increase in consumption per dollar from under 5 deaths averted'] = \

inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']* \

inputs['Shared']['1 DALY averted is equivalent to increasing ln(consumption) by one unit for one individual for how many years?']/ \

bednets['Marginal cost per under 5 death averted']

bednets['Ratio of 5 and over malaria deaths to under 5 malaria deaths'] = 0.49

bednets['Value of over 5 mortality reduction relative to under 5 mortality reduction'] = \

inputs['AMF']['Value of an adult malaria death prevented relative to that of a young child']* \

inputs['AMF']['Relative efficacy of ITNs for reducing adult mortality']* \

bednets['Ratio of 5 and over malaria deaths to under 5 malaria deaths']

bednets['Proportional increase in consumption per dollar from 5 and over deaths averted'] = \

bednets['Proportional increase in consumption per dollar from under 5 deaths averted']* \

bednets['Value of over 5 mortality reduction relative to under 5 mortality reduction']

bednets['Proportional increase in consumption per dollar from developmental benefits'] = \

bednets['Adjustment for unused nets and alternate funders']* \

deworming['Adjusted long term benefits per year of treatment (in terms of ln $), assuming income supports household consumption (before adjusting for alternate funders)']/ \

inputs['AMF']['Relative value of a year of deworming in Baird to development benefits from a year of ITN coverage']/ \

bednets['Cost per person-year of protection, adjusted for insecticide resistance: under-14\'s']

bednets['Proportional increase in consumption per dollar equivalent overall'] = \

bednets['Proportional increase in consumption per dollar from developmental benefits'] + \

bednets['Proportional increase in consumption per dollar from 5 and over deaths averted'] + \

bednets['Proportional increase in consumption per dollar from under 5 deaths averted']

bednets['Cost per equivalent life saved'] = \

inputs['Shared']['1 DALY averted is equivalent to increasing ln(consumption) by one unit for one individual for how many years?']* \

inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

bednets['Proportional increase in consumption per dollar equivalent overall']

bednets['DALYs per $' + str(m)] = m*inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

bednets['Cost per equivalent life saved']

bednets['Percent of total benefit coming from development effects'] = \

bednets['Proportional increase in consumption per dollar from developmental benefits']/ \

bednets['Proportional increase in consumption per dollar equivalent overall']

bednets['Percent of total benefit coming from adult mortality reduction'] = \

bednets['Proportional increase in consumption per dollar from 5 and over deaths averted']/ \

bednets['Proportional increase in consumption per dollar equivalent overall']

bednets['Percent of total benefit coming from child mortality reduction'] = \

bednets['Proportional increase in consumption per dollar from under 5 deaths averted']/ \

bednets['Proportional increase in consumption per dollar equivalent overall']

# smc

smc = {}

smc['Total cost for all of ACCESS-SMC in 2015'] = 19441346.4311218

smc['Total ACCESS-SMC target population in 2015'] = 3423337.0

smc['Coverage rate: % of targeted children who received at least 1 cycle of treatment (in 2015)'] = 0.8695

smc['Coverage rate: % of targeted children who received at least 3 cycles of treatment (in 2015)'] = 0.6588

smc['Coverage rate: % of targeted children who received at least 2 cycles of treatment (in 2015)'] = \

(smc['Coverage rate: % of targeted children who received at least 1 cycle of treatment (in 2015)'] + \

smc['Coverage rate: % of targeted children who received at least 3 cycles of treatment (in 2015)'])/2.0

smc['Coverage rate: % of targeted children who received all 4 cycles of treatment (in 2015)'] = 0.4628

smc['% of kids who got exactly 1 round'] = smc['Coverage rate: % of targeted children who received at least 1 cycle of treatment (in 2015)'] - \

smc['Coverage rate: % of targeted children who received at least 2 cycles of treatment (in 2015)']

smc['% of kids who got exactly 2 rounds'] = smc['Coverage rate: % of targeted children who received at least 2 cycles of treatment (in 2015)'] - \

smc['Coverage rate: % of targeted children who received at least 3 cycles of treatment (in 2015)']

smc['% of kids who got exactly 3 rounds'] = smc['Coverage rate: % of targeted children who received at least 3 cycles of treatment (in 2015)'] - \

smc['Coverage rate: % of targeted children who received all 4 cycles of treatment (in 2015)']

smc['% of kids who got all 4 rounds'] = smc['Coverage rate: % of targeted children who received all 4 cycles of treatment (in 2015)']

smc['Adjustment to get from coverage survey figures to actual coverage'] = 1.0

smc['Overall adjustment for lack of adherence to treatment regimen'] = 0.88

smc['Total person-months of coverage'] = smc['Total ACCESS-SMC target population in 2015']* \

(1.0*smc['% of kids who got exactly 1 round'] + 2.0*smc['% of kids who got exactly 2 rounds'] + \

3.0*smc['% of kids who got exactly 3 rounds'] + 4.0*smc['% of kids who got all 4 rounds'])* \

smc['Adjustment to get from coverage survey figures to actual coverage']* \

smc['Overall adjustment for lack of adherence to treatment regimen']

smc['Equivalent 4-cycles of person-months of coverage'] = \

smc['Total person-months of coverage']/4.0

smc['Sanity check: Effective coverage rate for 4-cycles of person-months of coverage'] = \

smc['Equivalent 4-cycles of person-months of coverage']/smc['Total ACCESS-SMC target population in 2015']

smc['Cost per equivalent child treated with 4 cycles'] = smc['Total cost for all of ACCESS-SMC in 2015']/ \

smc['Equivalent 4-cycles of person-months of coverage']

smc['Sanity check: Cost per child targeted (i.e., cost for 4 treatments per child if coverage and adherence were 100%)'] = \

smc['Total cost for all of ACCESS-SMC in 2015']/smc['Total ACCESS-SMC target population in 2015']

smc['Sanity check: overall adjustment on cost per child targeted'] = smc['Cost per equivalent child treated with 4 cycles']/ \

smc['Sanity check: Cost per child targeted (i.e., cost for 4 treatments per child if coverage and adherence were 100%)']

smc['Hypothetical cohort'] = 1000.0

smc['Cost to cover hypothetical cohort with 4 cycles each'] = smc['Cost per equivalent child treated with 4 cycles']*smc['Hypothetical cohort']

smc['2015 3- to 59-month old all-cause deaths expected in cohort, in absence of SMC'] = 15.36

smc['2015 3- to 59-month old malaria mortality rate (i.e., expected deaths in the specified cohort, in absence of SMC)'] = \

smc['2015 3- to 59-month old all-cause deaths expected in cohort, in absence of SMC']* \

inputs['SMC']['If malaria were eliminated, the fraction of all-cause mortality that would be averted in 3- to 59-month olds in ACCESS-SMC countries']

smc['Fraction of annual malaria burden occurring in SMC period'] = 0.7

smc['2015 3- to 59-month old malaria deaths during malaria season (in the cohort)'] = \

smc['2015 3- to 59-month old malaria mortality rate (i.e., expected deaths in the specified cohort, in absence of SMC)']* \

smc['Fraction of annual malaria burden occurring in SMC period']

smc['Deaths averted in hypothetical cohort (including replicability, external validity, and alternate funders adjustments)'] = \

smc['2015 3- to 59-month old malaria deaths during malaria season (in the cohort)']* \

(1.0 - inputs['SMC']['Relative risk for malaria outcome, intention to treat effect'])* \

inputs['SMC']['Ratio of the reduction in malaria mortality to the reduction in malaria incidence']* \

inputs['SMC']['Replicability adustment - SMC']* \

inputs['SMC']['External validity adjustment - SMC']* \

inputs['SMC']['Alternate funders adjustment - SMC']

smc['Cost per 3- to 59-month old death averted'] = smc['Cost to cover hypothetical cohort with 4 cycles each']/ \

smc['Deaths averted in hypothetical cohort (including replicability, external validity, and alternate funders adjustments)']

smc['Sanity check: Value of SMC death averted relative to AMF death averted'] = \

inputs['SMC']['DALYs averted per death of a 3- to 59-month old child averted - SMC']/ \

inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']

smc['Proportional increase in consumption per dollar from under 5 deaths averted'] = \

inputs['SMC']['DALYs averted per death of a 3- to 59-month old child averted - SMC']* \

inputs['Shared']['1 DALY averted is equivalent to increasing ln(consumption) by one unit for one individual for how many years?']/ \

smc['Cost per 3- to 59-month old death averted']

smc['Proportional increase in consumption per dollar equivalent from developmental benefits'] = \

inputs['SMC']['Alternate funders adjustment - SMC']* \

deworming['Adjusted long term benefits per year of treatment (in terms of ln $), assuming income supports household consumption (before adjusting for alternate funders)']/ \

inputs['SMC']['Relative value of a year of deworming in Baird to development benefits from 4-months of SMC coverage']/ \

smc['Cost per equivalent child treated with 4 cycles']

smc['Proportional increase in consumption per dollar equivalent overall'] = smc['Proportional increase in consumption per dollar equivalent from developmental benefits'] + \

smc['Proportional increase in consumption per dollar from under 5 deaths averted']

smc['Cost per equivalent life saved'] = inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']* \

inputs['Shared']['1 DALY averted is equivalent to increasing ln(consumption) by one unit for one individual for how many years?']/ \

smc['Proportional increase in consumption per dollar equivalent overall']

smc['DALYs per $' + str(m)] = m*inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

smc['Cost per equivalent life saved']

smc['Percent of SMC\'s cost-effectiveness coming from development effects'] = smc['Proportional increase in consumption per dollar equivalent from developmental benefits']/ \

smc['Proportional increase in consumption per dollar equivalent overall']

iodine = {}

iodine['Benefit on one year\'s income (discounted back 10 years because of delay between deworming and working for income)'] = \

(inputs['Iodine']['% of benefit of iodine that lasts for the long term']* \

inputs['Iodine']['Equivalent increase in wages from having iodine throughout childhood'])/ \

np.power((1.0 + inputs['Shared']['Discount rate']), inputs['Deworming']['Average number of years between deworming and the beginning of long term benefits'])

iodine['Present value of the sum of the lifetime benefits per worker (in terms of Ln(income))'] = \

iodine['Benefit on one year\'s income (discounted back 10 years because of delay between deworming and working for income)']* \

(1.0 - 1.0/np.power((1.0 + inputs['Shared']['Discount rate']), inputs['Deworming']['Duration of long term benefits of deworming (in years)']))/ \

(1.0 - 1.0/(1.0 + inputs['Shared']['Discount rate']))

iodine['Adjusted long term benefits per year of treatment (in terms of ln $), assuming income supports household consumption'] = \

iodine['Present value of the sum of the lifetime benefits per worker (in terms of Ln(income))']* \

inputs['Iodine']['Replicability']*inputs['Iodine']['External validity']* \

inputs['Iodine']['% of children that benefit']* \

(inputs['Iodine']['Percent of population under 15']/inputs['Iodine']['Years of Childhood (for iodine)'])* \

inputs['Deworming']['Number of household members that benefit - Deworming']

iodine['Short term health benefits (deworming only) in terms of Ln(income)'] = 0.0

iodine['Adjusted total benefits, per person dewormed (Ln(income))'] = \

iodine['Adjusted long term benefits per year of treatment (in terms of ln $), assuming income supports household consumption']

iodine['Proportional increase in consumption per dollar'] = \

inputs['Iodine']['Probability that GAIN/ICCIDD has an impact']* \

iodine['Adjusted total benefits, per person dewormed (Ln(income))']/ \

(inputs['Iodine']['Cost per person per year']/inputs['Iodine']['Leverage (dollars of impact per dollars spent)'])

iodine['X as cost effective as Cash'] = \

iodine['Proportional increase in consumption per dollar']/cash['Proportional increase in consumption per dollar']

iodine['Cost per equivalent life saved'] = \

inputs['Shared']['1 DALY averted is equivalent to increasing ln(consumption) by one unit for one individual for how many years?']* \

inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

iodine['Proportional increase in consumption per dollar']

iodine['DALYs per $' + str(m)] = m*inputs['AMF']['DALYs averted per death of an under-5 averted - AMF']/ \

iodine['Cost per equivalent life saved']

'''

# generate pdfs

x = np.linspace(0.0, 25.0, 100)

cash_y, cash_x = np.histogram(cash[key], bins=x, density=True)

bednets_y, bednets_x = np.histogram(bednets[key], bins=x, density=True)

dtw_y, dtw_x = np.histogram(dtw[key], bins=x, density=True)

sci_y, sci_x = np.histogram(sci[key], bins=x, density=True)

ss_y, ss_x = np.histogram(ss[key], bins=x, density=True)

iodine_y, iodine_x = np.histogram(iodine[key], bins=x, density=True)

smc_y, smc_x = np.histogram(smc[key], bins=x, density=True)

# generate plots

colors = plt.cm.jet(np.linspace(0.0, 1.0, 7))

plt.figure(0, figsize=(8, 6))

plt.plot((cash_x[:-1] + cash_x[1:])/2.0, cash_y, label='cash', linewidth=2.0, color='k')

plt.plot((bednets_x[:-1] + bednets_x[1:])/2.0, bednets_y, label='bednets', linewidth=2.0, color='b')

plt.plot((dtw_x[:-1] + dtw_x[1:])/2.0, dtw_y, label='dtw', linewidth=2.0, color='g')

plt.plot((sci_x[:-1] + sci_x[1:])/2.0, sci_y, label='sci', linewidth=2.0, color='r')

plt.plot((ss_x[:-1] + ss_x[1:])/2.0, ss_y, label='ss', linewidth=2.0, color='y')

plt.plot((iodine_x[:-1] + iodine_x[1:])/2.0, iodine_y, label='iodine', linewidth=2.0, color=colors[5])

plt.plot((smc_x[:-1] + smc_x[1:])/2.0, smc_y, label='smc', linewidth=2.0, color='m')

plt.xlim([np.min(x), np.max(x)])

plt.ylim([0, 0.5])

plt.xlabel(key)

plt.ylabel('Probability')

plt.title('PDF of cost effectiveness')

plt.legend(loc='upper right')

plt.show()

'''

# export data

data = np.array([dtw[key], sci[key], ss[key], cash[key], bednets[key], smc[key], iodine[key]]).transpose()

df = pd.DataFrame(data, columns=['dtw', 'sci', 'ss', 'cash', 'bednets', 'smc', 'iodine'])

df.to_pickle('gw_data.pickle')