Pre-Analysis Plan Report for Twenty Year Economic Impacts ...emiguel.econ.berkeley.edu/.../98/...PAP_Report.pdf · Pre-Analysis Plan Report for Twenty Year Economic Impacts of Deworming

Pre-Analysis Plan Report forTwenty Year Economic Impacts of Deworming

PAP Report Authors Michelle Layvant Edward Miguel Michael Walker

Paper authors Joan Hamory r Edward Miguel r Michael Walker rMichael Kremer r Sarah Baird

July 17 2020

Executive Summary

This document outlines all of the results pre-specified as part of the ldquoTwenty YearEconomic Impacts of Dewormingrdquo pre-analysis plan Many of these results are pre-sented in the corresponding paper due to space constraints not all outcomes androbustness checks are presented in the paper This document is meant to provide thesein an easy-to-verify format for transparency We briefly note the specifications we useand present tables of results organized by outcome family

1 Introduction

Estimating the returns to childhood investments is of central importance in development and

labor economics and there is growing recognition of the long-run importance of childhood

health and education interventions This study builds on the work of Miguel and Kremer

(2004) and Baird et al (2016) to provide experimental evidence on the long-run returns

to improved child health in Africa In particular this project analyzes the 10 to 20-year

economic impacts of a school-based childhood deworming program in Kenya using outcome

data from the most recent rounds of the Kenya Life Panel Survey namely KLPS-3 collected

during 2011-2014 and KLPS-4 collected from 2017-2019

In Baird et al (2017) we pre-specified analyses to be conducted with data from KLPS

rounds 2 3 and 4 This plan was filed in advance of analyzing KLPS-4 data This report

presents results of all of these pre-specified analyses It is meant to supplement the twenty-

year paper (Hicks et al 2020) and pre-analysis plan (Baird et al 2017) interested readers

should see these documents for more details of the study design data and interpretation of

results1

We outlined five groups of outcomes (1) consumption (2) wealth (3) earnings (4) labor

supply and (5) occupational choice In addition we denoted two primary outcomes total

per-capita consumption in the last 12 months and total respondent earnings in the last 12

months We briefly outline our main econometric approach and then present results for each

of these families in turn

2 Econometric Specifications

We refer the reader to the twenty-year paper (Hicks et al 2020) and pre-analysis plan (Baird

et al 2017) for details of the data and experimental design As described in those documents

our main sample drops vocational training winners from the analysis of the KLPS-3 and

KLPS-4 data and cash grant winners are dropped from the analysis of the KLPS-4 data as

these programs occurred prior to the start of data collection

Our main regression equation takes the following form The dependent variable Yijt is

an outcome for individual i in original PSDP school j as measured in survey round t

Yijt ldquo α ` λ1Tj ` λ2Cj ` λ3Pj `X1ij0β ` εijt (1)

The outcome is a function of Tj P t0 1u the assigned deworming program treatment sta-

1 In the interest of brevity the paper presents 21 out of 54 outcomes that we show here

tus of the individualrsquos school The pre-specified main coefficient of interest is λ1 which

captures gains accruing to individuals in the 50 treatment schools relative to the 25 control

schools We consider two secondary sources of exogenous variation in exposure to deworming

namely the 2001 cost-sharing school indicator Cj P t0 1u and the proportion of students

in neighboring schools within 6 km that received deworming Pj P r0 1s which we call local

deworming saturation The vector Xij0 of individual and school covariates includes base-

line school characteristics (average test score population number of students within 6 km

and administrative zone indicators) baseline individual characteristics (gender and grade)

indicators for the KLPS survey calendar month wave and round and an indicator for the

vocational training and cash grant control group Estimates are weighted to maintain rep-

resentativeness with the baseline PSDP population taking into account the sampling for

KLPS the two-stage tracking methodology and inclusion in the vocational training and

cash grant program Finally εijt is the error term clustered at the school level allowing for

correlation in outcomes both across individuals in those schools and across survey rounds

We pre-specified two main approaches for estimating the data the cross-sectional ap-

proach (Approach 1 in the PAP) restricts attention to KLPS-4 while the pooled approach

(Approach 2 in the PAP) makes use of longitudinal data from all rounds where available We

also noted that where longitudinal data is available we may look at effects round-by-round

in order to better understand the evolution of outcomes over time these regressions also

take the same form as above

We look at two main sources of heterogeneity ndash heterogeneity by gender and age at

baseline Heterogeneity by the respondentrsquos gender is estimated from a single regression by

including an indicator for female and treatment-female cost-sharing-female and saturation-

female interaction terms Heterogeneity by baseline age is also estimated from a single

regression including an indicator for those older than 12 at baseline and treatment-older

cost-sharing-older and saturation-older interaction terms

In addition to these main results we present several robustness checks First for mone-

tary outcomes the pre-analysis plan notes that our primary approach trims the top 1 of

observations to reduce the influence of outliers here we also present untrimmed results for

the pooled specification Second we include respondents that participated in the vocational

training (VocEd) and cash grant (SCY) interventions including indicators for treatment

groups for both interventions Lastly for some earnings-related outcomes we collect data

about both the last 12 months and the last month our main results focus on the last 12

months and here we also present results for the last month

The PAP notes that we will present both ldquoper-comparisonrdquo p-values as well as FDR

q-values within the families of outcomes that we have defined following Anderson (2008)

and Casey Glennerster and Miguel (2012) In addition FDR q-values are calculated over

the two key primary outcomes (per-capita consumption and respondent earnings) outlined

3 Price Conversion

The PAP notes that consumption earnings and wealth meausures will be converted to 2017

USD at Purchasing Power Parity (PPP)

All monetary amounts that are not already reported in Kenyan Shillings (KES) are first

converted to KES using the average exchange rate by survey wave in KLPS-4 This means

that if the survey was conducted during Wave 1 all amounts reported in Ugandan Shillings

(UGX) are converted to KES using the average exchange rate of UGX to KES for the time

period January 5 2017ndashMay 21 2018 (352684 UGX 1 KES) Similarly if the survey

was conducted during Wave 2 all amounts reported in UGX are converted to KES using

the average exchange rate for the time period June 2 2018ndashJune 18 2019 (370597 UGX

1 KES)2 All amounts in KLPS-2 and KLPS-3 that are not already reported in KES are

converted to KES using the average exchange rate between 2006ndash2016 for each respective

currency

Values that are reported or have been converted to KES are then converted to USD at

PPP for that survey year3 In other words surveys that took place in 2018 are then converted

to 2018 USD by dividing the total amount in KES by the 2018 KES to USD exchange rate

Surveys that took place during 2019 are converted to USD using the 2018 exchange rate

as the 2019 exchange rate had not been released when PPP data was downloaded from the

World Bank

Once values have been converted to USD values are inflated (or deflated) to 2017 USD

using Consumer Price Index (CPI) inflation rates4 All values are multiplied by the 2017

CPI and divided by the CPI in that survey year Surveys conducted in 2019 are deflated to

2017 USD using the 2018 CPI inflation rate

We also adjust consumption expenditures using an urban-rural price deflator for respon-

dents living in Nairobi and Mombasa See Section 5 for additional details on the construction

of the urban-rural price deflator

2 Average exchange rates by wave were obtained at wwwfxtopcom3 Exchange rates ndash PPP conversion factor GDP (LCU per international $) ndash were obtained from the

World Bank at httpsdataworldbankorgindicatorPANUSPPPlocations=KE during June 20194 The CPI inflation rates were obtained from the Bureau of Labor Statistics (All Urban Consumers -

US city average All items ndash CUUR0000SA0) at httpswwwblsgovdataprices during June 2019

4 Primary Outcomes

We designated two outcomes as primary annual per-capita consumption and annual indi-

vidual earnings Details on the construction of these outcomes can be found in Sections 5

and 71 respectively Here we present main table that reports FDR q-values across each of

these for additional robustness checks see the corresponding tables in the aforementioned

sections

List of tables

1 Table 1 Pooled top 1 trimmed

5 Family 1 Household Consumption

Consumption data is only available in KLPS-4 and for a representative subset of KLPS-

3 respondents We thus present pooled estimates and then results by round Our primary

specification trims the top 1 of observations For robustness we present untrimmed pooled

results as well as including individuals that participated in the vocational training (VocEd)

and cash grant voucher (SCY) programs

Consumption expenditure measures are adjusted for urban-rural price differences using

information from price surveys that were collected contemporaneously at various local mar-

kets (in western Kenya as well as in the major urban areas of Nairobi and Mombasa) Total

consumption expenditures for individuals living in Nairobi and Mombasa are divided by an

urban-rural price deflator or price index This price deflator is calculated by fixing consump-

tion quantities as in a Laspeyres-style price index The urban-rural price deflator equals 111

in KLPS-3 and 119 in KLPS-4

Consumption expenditure measures are converted to 2017 USD PPP (see Section 3 for

a detailed explanation of the PPP conversion)

Notes on outcome construction

1 Per-Capita Consumption The sum of the monetary value of goods consumed by the

household through purchase gift barter or home production in the last 12 months

divided by the number of household members Households which indicated that they

purchased an item but did not report a total price for these items nor a total quan-

tity purchased are assumed to have consumed the mean amount consumed by other

households adjusted for household size

2 Log Per-Capita Consumption As in (1) and logged

3 Per-Capita Food Consumption The sum of the monetary value of food items consumed

by the household through purchase gift barter or home production in the last 12

months divided by the number of household members Food purchases reported in

quantities rather than a monetary value (eg kilograms liters etc) are converted to

KES by multiplying the quantity purchased for that household by the average price

of this item for that month and market region Food that is reported in non-standard

units relative to the price data (eg household reporting 1 kg of potatoes as opposed

to the number of potatoes purchased) are converted back to the rdquostandard unitrdquo and

then converted to KES using the price data for that month and market region

4 Log Per-Capita Food Consumption As in (3) and logged

5 Per-Capita Non-Food Consumption The sum of the monetary value of non-food items

consumed by the household through purchase gift barter or home production in the

last 12 months divided by the number of household members Non-food consump-

tion includes frequent and non-frequent non-food purchases excluding contributions to

informal credit schemes losses due to theft and contributions to savings accounts

6 Log Per-Capita Non-Food Consumption As in (5) and logged

7 Average Meals Eaten This is equal to the average number of meals eaten by the

respondent in the last 3 days for KLPS-4 observations and meals eaten yesterday for

KLPS-3 observations

8 Household Tax Spending The sum of local council taxes taxes and fees community

group fees and bribes paid by the household to the government in the last 12 months

KLPS-3 uses E-Module data

9 Log Household Tax Spending As in (8) and logged

List of tables

2 Table 3 Pooled untrimmed

3 Table 4 KLPS-4 top 1 trimmed

5 Table 6 Pooled top 1 trimmed SCYVocEd included

6 Family 2 Household Wealth

Wealth data is only collected in KLPS-4 Our primary specification trims the top 1 of

observations For robustness we present untrimmed results as well as including individuals

in the VocEd and SCY programs

Wealth measures are converted to 2017 USD PPP (see Section 3 for a detailed explanation

of the PPP conversion)

1 Per-Capita Household Wealth The sum of total household durable asset ownership

and livestock ownership divided by the number of household members The total value

of durable assets and livestock is found by multiplying the number of durable assets

and livestock owned by each household by the average price of that item purchased by

the household If the household did not purchase this item within the last 12 months

we employ the median price of that item across all households (separately for urban

versus rural households) Urban household are considered those living in Nairobi and

Mombasa

2 Log Per-Capita Household Wealth As in (1) and logged

3 Per-Capita Household Assets The sum of total household durable asset ownership

divided by the number of household members The total value of durable assets is

found by multiplying the number of durable assets owned by each household by the

average price of that item purchased by the household If the household did not

purchase this item within the last 12 months we employ the median price of that item

across all households (separately for urban versus rural households)

4 Log Per-Capita Household Assets As in (3) and logged

5 Per-Capita Livestock Ownership The sum of total livestock ownership divided by the

number of household members The total value of livestock is found by multiplying

the number of livestock owned by each household by the average price of that item

purchased by the household If the household did not purchase this item within the last

12 months we employ the median price of that item across all households (separately

for urban versus rural households)

6 Log Per-Capita Livestock Ownership As in (5) and logged

List of tables

2 Table 8 KLPS-4 untrimmed

3 Table 9 KLPS-4 top 1 trimmed SCYVocEd included

7 Family 3A Individual Earnings Outcomes

Here we list out earnings outcomes broken down into non-hourly and hourly earnings out-

comes All multiple-testing adjustments (FDR adjustments) are done across all earnings

outcomes

Individual wage earnings and self-employment profits are collected in KLPS-2 KLPS-3

and KLPS-4 Farming profits are only collected in KLPS-3 and KLPS-4 Earnings mea-

sures are converted to 2017 USD PPP (see Section 3 for a detailed explanation of the PPP

conversion)

We present pooled estimates and results by round (for KLPS-3 and KLPS-4) Our pri-

mary specification trims the top 1 of observations For robustness we present untrimmed

pooled results as well as including individuals that participated in VocEd and SCY pro-

grams We also present results where we calculate earnings focusing on the most recent

month only (and hence not using any recall data)

71 Part 1 Non-Hourly Earnings Measures

1 Individual Earnings The sum of wage employment across all jobs non-agricultural

self-employment profit across all businesses and farming profits within the last 12

months

2 Log Individual Earnings As in (1) and logged

3 Individual Wage Earnings The sum of wage earnings across all jobs within the last 12

months

4 Log Individual Wage Earnings As in (3) and logged

5 Self-Employment Profit The sum of self-employment profits across all jobs (excluding

agricultural businesses) within the last 12 months

6 Log Self-Employment Profit As in (5) and logged

7 Individual Farming Profits Individual farming profit for KLPS-3 is measured as the

sum of all crop-specific production (valued in cash) minus input costs within the last 12

months for farming activities for which the respondent provided all reported household

labor hours and was the main decision-maker The input cost is calculated as the sum of

the salaries for workers from outside the household and the amount spent on tools and

machinery Individual farming profit for KLPS-4 includes the net profit generated from

non-crop and crop farming activities within the last 12 months for which the respondent

provided all reported household labor hours and was the main decision-maker This

is the value of everything produced minus expenses for the activity including hired

workers land rental storage and purchase of inputs such as raw materials fuel and

electricity Farming profits are not available for KLPS-2

8 Log Individual Farming Profits As in (7) and logged

9 Taxes on Individual Wages and Profits The sum of licenses and taxes and bribes to

the government for current self-employment jobs plus the total amount deducted in

taxes from the respondentrsquos salary for current wage-earning jobs

10 Log Taxes on Individual Wages and Profits As in (9) and logged

List of tables

6 Table 15 Pooled top 1 trimmed last month

72 Part 2 Hourly Earnings Measures

1 Individual Hourly Earnings The sum of wage employment across all jobs non-agricultural

months divided by 52 divided by the total hours worked across all activities during

the last week among those with at least 10 work hours across all activities Total

hours worked in each job within job categories (ie wage-earning self-employment

and farming) and across all jobs are top-coded at 100 hours per week

2 Log Individual Hourly Earnings As in (1) and logged

3 Hourly Individual Wage Earnings The sum of wage earnings across all jobs within

the last 12 months divided by 52 divided by the total hours worked across all wage

employment during the last week among those with at least 10 work hours Total

hours worked in each wage job and across all wage-earning jobs are top-coded at 100

hours per week

4 Log Hourly Individual Wage Earnings As in (3) and logged

5 Hourly Self-Employment Profit The sum of self-employment profits across all jobs

(excluding agricultural businesses) within the last 12 months divided by 52 divided

by the total hours worked across all non-agricultural self-employment during the last

week among those with at least 10 work hours Total hours worked in each self-

employment job and across all self-employment jobs are top-coded at 100 hours per

6 Log Hourly Self-Employment Profit As in (5) and logged

7 Hourly Individual Farming Profits The net profit generated from non-crop and crop

farming activities for which the respondent provided all reported household labor hours

and was the main decision-maker within the last 12 months divided by 52 divided by

hours worked in farming during the last week among those with at least 10 work hours

Farming profits are not available for KLPS-2 Total hours worked in each farming job

and across all farming jobs are top-coded at 100 hours per week

8 Log Hourly Individual Hourly Farming Profits As in (7) and logged

List of tables

8 Family 3B Household Earnings Outcomes

Household earnings are only collected as part of KLPS-4 Household tax spending is collected

in KLPS-3 and KLPS-4 Household earnings measures are converted to 2017 USD PPP (see

Section 3 for a detailed explanation fo the PPP conversion)

We present pooled estimates and results by round (for KLPS-4 only) Our primary

results as well as including individuals that participated in VocEd and SCY programs We

also present results where we calculate household earnings focusing on the most recent month

only (and hence not using any recall data)

1 Per-Capita Household Earnings The sum of wage employment earnings self-employment

profits and agricultural profits across household members in the last 12 months di-

vided by the number of household members Household earnings are only available in

KLPS-4

2 Log Per-Capita Household Earnings As in (1) and logged

3 Household Tax Spending The sum of licenses and taxes on self-employment profits

bribes paid to government officials police or other authorities local council taxes

and fees and community group fees in the last 12 months divided by the number of

household members Bribes local council taxes and fees and community group fees

are collected at the household level whereas licenses and taxes on self-employment

profits are only collected for the KLPS respondent

List of tables

9 Family 4 Labor Supply

Labor supply data is collected in KLPS-2 KLPS-3 and KLPS-4 We present pooled esti-

mates and results by round (for KLPS-4 and KLPS-3) For robustness we present results

that include individuals that participated in the VocEd and SCY programs

1 Total Hours Worked The sum of hours worked in agriculture wage-earning activities

and self-employment in the last 7 days Total hours worked in each job within job

categories (ie wage-earning self-employment and farming) and across all jobs are

top-coded at 100 hours per week

2 Log Total Hours Worked As in (1) and logged

3 Non-Zero Hours An indicator variable for non-zero hours worked in the last 7 days

4 Farm Hours Worked The sum of hours worked in all non-crop and crop farming

activities in the last 7 days Total hours worked in each farming job and across all

farming jobs are top-coded at 100 hours per week

5 Log Farm Hours Worked As in (4) and logged

6 Wage Hours Worked The sum of wage hours worked across all wage jobs in the last 7

days Total hours worked in each wage job and across all wage jobs are top-coded at

100 hours per week

7 Log Wage Hours Worked As in (6) and logged

8 Self-Employment Hours Worked The sum of wage hours worked across all self-employment

jobs in the last 7 days Total hours worked in each self-employment job and across all

self-employment jobs are top-coded at 100 hours per week

9 Log Self-Employment Hours Worked As in (8) and logged

10 Non-Zero Wage or Self-Employment Hours An indicator variable for non-zero hours

worked in wage-earning activities and self-employment in the last 7 days

List of tables

1 Table 27 Pooled

2 Table 28 KLPS-4

3 Table 29 KLPS-3

4 Table 30 Pooled SCYVocEd included

10 Family 5 Occupational Choice

Occupational choice data is collected in KLPS-2 KLPS-3 and KLPS-4 and includes indica-

tors for working a wage-earning job in various sectors

We present pooled estimates and results by round (for KLPS-4 and KLPS-3) For robust-

ness we present results that include individuals that participated in the VocEd and SCY

programs

1 Employed - Agriculture An indicator for employed in the agricultural sector as the

primary job (excluding subsistence agriculture and self-employment)

2 Employed - Fishing An indicator for employed in the fishing sector as the primary job

(excluding subsistence fishing and self-employment)

3 Employed - Manufacturing An indicator for working for wages in the manufacturing

sector as the primary job

4 Employed - ConstructionCasual Labor An indicator for employed in the construc-

tioncasual labor sector as the primary job

5 Employed - Services An indicator for employed in the services sector as the primary

6 Employed - Retail and Wholesale Trade An indicator for employed in the retail and

wholesale trade sector as the primary job

7 Employed - Trade Contractor An indicator for employed in the trade contractor sector

as the primary job

List of tables

1 Table 31 Pooled

2 Table 32 KLPS-4

3 Table 33 KLPS-3

References

Anderson Michael L 2008 ldquoMultiple Inference and Gender Differences in the Effects

of Early Intervention A Reevaluation of the Abecedarian Perry Preschool and Early

Training Projectsrdquo Journal of the American Statistical Association 103 (484) 1481ndash

Baird Sarah Joan Hamory Hicks Michael Kremer and Edward Miguel 2016

ldquoWorms at Work Long-run Impacts of a Child Health Investmentrdquo The Quarterly

Journal of Economics 131 no 4 (July) 1637ndash1680

2017 ldquoPre-Analysis Plan for ldquoThe 20-year Impacts of Child Deworming in Kenyardquordquo

RCT ID AEARCTR-0001191

Casey Katherine Rachel Glennerster and Edward Miguel 2012 ldquoReshaping Insti-

tutions Evidence on Aid Impacts Using a Preanalysis Planrdquo The Quarterly Journal

of Economics 127 (4) 1755ndash1812

Hicks J Hamory E Miguel M Walker M Kremer and S Baird 2020 ldquoTwenty

Year Economic Impacts of Dewormingrdquo Working paper June

Miguel Edward and Michael Kremer 2004 ldquoWorms identifying impacts on education

and health in the presence of treatment externalitiesrdquo Econometrica 72 (1) 159ndash217

Table 1 Primary Outcomes Annual Per-Capita Consumption and Annual Individual Earn-ings

(1) (2) (3) (4) (5)All Female Male Older Younger

Panel A Annual Per-Capita Consumption (KLPS-3 and 4)Treatment (λ1) 305˚ 89 513˚ 886˚˚˚ -179

(159) (134) (304) (223) (185)Control Mean 2156 1715 2594 1908 2381Treatment Effect () 1415 521 1976 4644 -752FDR q-value 132 630 623 001 290Number Observations 4794 2473 2321 2402 2341

Panel B Annual Individual Earnings (KLPS-2 3 and 4)Treatment (λ1) 80 41 118 258˚˚ -75

Notes Panel A reports annual per-capita total consumption calculated as the sum of the monetary value of goodsconsumed by the household through purchase gift barter or home production in the last 12 months dividedby the number of household members The consumptionexpenditure module was administered to a subset ofthe sample during round 3 and the full sample during round 4 Consumption is adjusted for urban-rural pricedifferences for respondents living in Nairobi and Mombasa Panel B reports annual individual earnings calculatedas the sum of wage employment across all jobs non-agricultural self-employment profit across all business andindividual farming profit defined as net profit generated from non-crop and crop farming activities for whichthe respondent provided all reported household labor hours and was the main decision-maker within the last 12months Wage earnings and self-employment profits were collected in KLPS rounds 2 3 and 4 agricultural profitswere collected in KLPS 3 and 4 All outcomes are converted to constant 2017 USD at PPP rates and the top1 of observations are trimmed Treatment is an indicator variable equal to 1 for PSDP Worm Groups 1 and 2which received an additional 24 years of deworming on average compared to Group 3 Columns (2) through (5)report estimates separately by gender and age at baseline (older than 12 12 or younger) Columns (2) and (3)report estimates for Female and Male are constructed from a single regression including treatment-female cost-sharing-female and saturation-female interaction terms Columns (4) and (5) also report results from a singleregression using an indicator for those older than 12 at baseline and analagous interaction terms to Columns (2)and (3) The pre-analysis plan (PAP) specified annual per-capita consumption and annual individual earningsas primary outcomes Following the PAP the FDR adjustment in column (1) is carried out across the two λ1

coefficient estimates from column (1) The FDR adjustment in columns (2) and (3) are carried out across thefour λ1 coefficient estimates from columns (2) and (3) Similarly the FDR adjustment in columns (4) and (5) arecarried out across the four λ1 coefficient estimates from columns (4) and (5) Covariates follow Baird et al (2016)and include controls for baseline 1998 primary school population geographic zone of the school survey waveand month of interview a female indicator variable baseline 1998 school grade fixed effects the average schooltest score on the 1996 Busia District mock exams total primary school pupils within 6 km and a cost-sharingschool indicator Those treated in a separate vocational training intervention (VocEd) which occurred prior toKLPS-3 are dropped from the KLPS-3 and KLPS-4 sample Those treated in a separate small grant intervention(SCY) which occurred after KLPS-3 are dropped from the KLPS-4 sample Observations are weighted to berepresentative of the original PSDP population and include KLPS population weights SCY and VocEd controlgroup weights and KLPS intensive tracking weights Standard errors are clustered at the 1998 school level denotes statistical significance at 10 pct at 5 pct and at 1 pct level

Table 2 Consumption - Pooled (trimmed)

Per-CapitaConsumption

Log Per-CapitaConsumption

(3)Per-Capita

FoodConsumption

(4)Log Per-Capita

FoodConsumption

(5)Per-CapitaNon-Food

Consumption

(6)Log Per-Capita

Non-FoodConsumption

AverageMeals Eaten

HouseholdTax Spending

Log HouseholdTax Spending

Panel A Full SampleTreatment (λ1) 305˚ 10˚ 22 03 116 10˚ -00 -0 17

(159) (05) (47) (04) (71) (05) (03) (1) (25)Cost Sharing (λ2) -136 -04 10 -00 -101 -06 -04 -1 -06

(144) (05) (51) (04) (67) (05) (03) (1) (24)Saturation (λ3) 957 02 -555 -21 -63 -03 10 4 125

(1408) (37) (397) (30) (390) (30) (14) (5) (126)Control Mean 2156 732 882 65 1138 655 251 3 37Treatment Effect () 142 91 25 33 102 98 -1 -3 158Joint F-Test (p-value) 259 220 263 542 310 277 259 582 735Treatment FDR q-value 237 237 855 709 237 237 1000 1000 709Number Observations 4794 4794 4794 4791 4794 4794 4835 4811 169

Panel B FemalesTreatment (λ1) 89 05 49 05 12 05 04 -0 21

(134) (06) (49) (05) (88) (08) (04) (1) (41)Cost Sharing (λ2) -179 -03 -15 00 -48 -03 -08˚˚ 0 52

(152) (06) (53) (05) (78) (07) (04) (1) (37)Saturation (λ3) -896 -34 -373 -07 -687 -47 23 -4 -144

(1056) (37) (395) (35) (552) (46) (27) (5) (325)Control Mean 1715 714 735 634 866 634 255 1 344Treatment Effect () 52 46 67 48 13 48 17 -377 188Joint F-Test (p-value) 0471 0473 0238 0555 0336 0519 0151 0878 0174Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2473 2473 2464 2461 2471 2471 2475 2471 56

Panel C MalesTreatment (λ1) 513˚ 14 -5 02 219 15 -05 0 15

(304) (09) (82) (07) (153) (10) (05) (2) (28)Cost Sharing (λ2) -113 -06 35 -00 -158 -10 00 -2 -29

(263) (08) (70) (06) (127) (08) (04) (2) (25)Saturation (λ3) 2422 30 -693 -32 437 31 -00 10 168

(2340) (63) (594) (54) (689) (49) (18) (8) (120)Control Mean 2594 749 1021 665 1402 676 247 5 377Treatment Effect () 198 133 -5 19 156 144 -18 79 143Joint F-Test (p-value) 0412 0427 0557 0869 0509 0440 0685 0537 0405Treatment FDR q-value 537 537 1000 1000 537 537 539 1000 967Number Observations 2321 2321 2330 2330 2323 2323 2360 2340 113

Panel D Older than 12Treatment (λ1) 886˚˚˚ 30˚˚˚ 153˚˚˚ 20˚˚˚ 383˚˚˚ 28˚˚˚ 05 2 40

(223) (07) (57) (05) (98) (08) (03) (2) (38)Cost Sharing (λ2) -347 -16˚˚ -100˚ -11˚˚ -263˚˚˚ -19˚˚ -06 -1 21

(228) (07) (59) (05) (94) (08) (04) (2) (32)Saturation (λ3) 3095 56 -129 26 230 19 -13 6 139

(1978) (49) (408) (33) (524) (42) (14) (7) (163)Control Mean 1908 72 821 64 981 643 244 2 338Treatment Effect () 464 262 187 182 390 248 21 873 334Joint F-Test (p-value) 0002 0001 0034 0004 0002 0004 0100 0665 0142Treatment FDR q-value 001 001 006 001 001 001 067 097 110Number Observations 2402 2402 2392 2391 2403 2403 2412 2399 96

Panel E 12 or YoungerTreatment (λ1) -179 -07 -80 -10˚ -89 -04 -05 -2 -02

(185) (05) (61) (05) (87) (07) (04) (1) (31)Cost Sharing (λ2) -33 02 79 07 11 02 -01 -0 -24

(155) (05) (67) (05) (83) (06) (03) (1) (30)Saturation (λ3) -1424 -56˚ -985˚˚ -68˚˚ -370 -26 28 0 -04

(911) (29) (421) (30) (407) (31) (19) (6) (159)Control Mean 2381 742 938 659 1274 667 257 3 389Treatment Effect () -75 -70 -86 -104 -70 -40 -19 -512 -24Joint F-Test (p-value) 0355 0258 0144 0072 0618 0844 0047 0247 0749Treatment FDR q-value 765 765 765 765 765 765 765 765 765Number Observations 2341 2341 2350 2348 2340 2340 2370 2359 71

Notes Analysis pools the last 12 months of retrospective data from KLPS-3 and KLPS-4 KLPS-3 consumption data is only measured for a subsample of KLPS-3 individuals Columns (1)-(6) (8) and (9) aretrimmed at the top 1 of observations See Section 5 for notes on outcome construction Consumption is adjusted for urban-rural price differences for respondents living in Nairobi and Mombasa Treatment is anindicator variable equal to 1 for PSDP Worm Groups 1 and 2 which received an additional 24 years of deworming on average compared to Group 3 Reported estimates for Female and Male are constructed froma single regression including treatment-female cost-sharing-female and saturation-female interaction terms Reported estimates for Older than 12 and 12 or Younger also report results using a single regressionincluding an indicator for those older than 12 at baseline and analogous interaction terms to Panels B and C Covariates follow Baird et al (2016) and include controls for baseline 1998 primary school populationgeographic zone of the school survey wave and month of interview a female indicator variable baseline 1998 school grade fixed effects the average school test score on the 1996 Busia District mock exams totalprimary school pupils within 6 km and a cost-sharing school indicator Those treated in a separate vocational training intervention (VocEd) which occurred prior to KLPS-3 are dropped from the KLPS-3 andKLPS-4 sample Those treated in a separate small grant intervention (SCY) which occurred after KLPS-3 are dropped from the KLPS-4 sample Observations are weighted to be representative of the original PSDPpopulation and include KLPS population weights SCY and VocEd control group weights and KLPS intensive tracking weights Standard errors are clustered at the 1998 school level The Joint F-Test (p-value)gives the p-value associated with an F-test on the joint significance of the treatment cost-sharing and saturation coefficients against the null hypothesis that all three coefficients are jointly equal to zero The FDRadjustment is carried out across the nine outcomes within this family separately by panel (full sample females males older than 12 and 12 or younger) denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 3 Consumption - Pooled (untrimmed)

(3)Per-Capita

FoodConsumption

(4)Log Per-Capita

FoodConsumption

Consumption

(6)Log Per-Capita

Non-FoodConsumption

AverageMeals Eaten

Panel A Full SampleTreatment (λ1) 380 06 309 03 118 09 -00 -944054 -45

(498) (06) (405) (05) (165) (06) (03) (782112) (50)Cost Sharing (λ2) -374 -03 -385 00 53 -05 -04 -431906 24

(484) (05) (413) (04) (160) (06) (03) (422339) (31)Saturation (λ3) -1292 -11 -2292 -32 1017 08 10 2455641 96

(2355) (36) (1917) (32) (1213) (38) (14) (2384526) (184)Control Mean 2740 739 1045 652 1415 661 251 954974 481Treatment Effect () 139 57 296 29 83 91 -1 -989 -603Joint F-Test (p-value) 859 642 636 431 759 526 259 666 712Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 4842 4842 4842 4839 4842 4842 4835 4865 223

Panel B FemalesTreatment (λ1) 91 06 -97 04 135 07 04 50370 -00

(234) (06) (240) (05) (122) (08) (04) (343130) (82)Cost Sharing (λ2) -11 -03 155 01 -147 -04 -08˚˚ -367864 09

(250) (06) (242) (05) (138) (08) (04) (368140) (68)Saturation (λ3) -4160 -50 -3074 -21 -582 -51 23 3941688 -351

(3298) (41) (2856) (36) (1029) (51) (27) (3633497) (367)Control Mean 1715 714 774 635 866 634 255 3 386Treatment Effect () 53 56 -125 39 156 67 17 19727518 -0Joint F-Test (p-value) 0401 0326 0620 0525 0433 0436 0151 0370 0787Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2477 2477 2477 2474 2477 2477 2475 2485 70

Panel C MalesTreatment (λ1) 653 06 705 02 93 12 -05 -1925633 -59

(982) (10) (900) (08) (302) (11) (05) (1759686) (65)Cost Sharing (λ2) -735 -03 -903 00 230 -05 00 -480252 30

(990) (08) (956) (07) (275) (09) (04) (508451) (36)Saturation (λ3) 852 19 -1749 -41 2271 53 -00 1118567 203

Panel D Older than 12Treatment (λ1) 1877˚ 29˚˚˚ 1211 21˚˚˚ 504˚˚ 30˚˚˚ 05 -131378 36

(1104) (08) (984) (06) (235) (09) (03) (233332) (47)Cost Sharing (λ2) -1448 -17˚˚ -1146 -11˚ -174 -19˚˚ -06 -300530 33

(1108) (07) (1002) (06) (226) (08) (04) (335426) (41)Saturation (λ3) 1221 43 -1288 12 2856 65 -13 2458809 161

Panel E 12 or YoungerTreatment (λ1) -860˚˚ -12˚˚ -451 -12˚˚ -189 -07 -05 -1748901 -138˚

(406) (06) (278) (05) (199) (08) (04) (1492971) (73)Cost Sharing (λ2) 438 06 235 08˚ 164 04 -01 -578405 38

(362) (06) (197) (05) (242) (07) (03) (547311) (47)Saturation (λ3) -3899 -67˚˚ -3306˚ -78˚˚˚ -789 -46 28 1623100 -07

(3083) (32) (1947) (29) (1409) (40) (19) (2245248) (257)Control Mean 3233 752 1170 661 1618 673 257 1781384 534Treatment Effect () -266 -131 -386 -123 -117 -67 -19 -982 Joint F-Test (p-value) 0184 0076 0360 0020 0731 0650 0047 0685 0289Treatment FDR q-value 128 128 178 128 293 293 273 273 178Number Observations 2370 2370 2370 2368 2370 2370 2370 2382 97

Notes Analysis pools the last 12 months of retrospective data from KLPS-3 and KLPS-4 KLPS-3 consumption data is only measured for a subsample of KLPS-3 individuals Results are untrimmed See Section5 for notes on outcome construction See Table 2 for notes on the regression specification denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 4 Consumption - KLPS-4 (trimmed)

(3)Per-Capita

FoodConsumption

(4)Log Per-Capita

FoodConsumption

Consumption

(6)Log Per-Capita

Non-FoodConsumption

AverageMeals Eaten

Panel A Full SampleTreatment (λ1) 199 06 -12 01 70 07 00 -1 -03

(130) (05) (44) (04) (75) (06) (03) (1) (30)Cost Sharing (λ2) -103 -02 20 02 -38 -02 -05˚ 1 07

(128) (05) (38) (04) (70) (06) (03) (1) (26)Saturation (λ3) -716 -29 -278 -27 -377 -26 14 0 112

(693) (26) (241) (24) (390) (31) (14) (3) (146)Control Mean 2044 731 878 65 1125 656 255 2 368Treatment Effect () 97 62 -13 13 62 64 1 -490 -34Joint F-Test (p-value) 091 200 652 473 436 449 195 275 860Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 4076 4076 4076 4073 4076 4076 4110 4090 139

Panel B FemalesTreatment (λ1) 50 03 20 03 19 05 05 -0 -18

(141) (07) (42) (05) (92) (08) (04) (0) (62)Cost Sharing (λ2) -64 00 3 04 -35 -01 -08˚ 1 86˚˚

(131) (06) (47) (05) (79) (08) (04) (1) (42)Saturation (λ3) -1429˚ -55 -307 -27 -849 -61 28 -3 -215

(766) (36) (366) (38) (533) (48) (24) (4) (389)Control Mean 1655 713 725 633 844 633 258 1 322Treatment Effect () 30 31 27 31 22 47 18 -667 -194Joint F-Test (p-value) 0154 0207 0559 0363 0293 0337 0206 0488 0123Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2102 2102 2096 2093 2100 2100 2104 2101 42

Panel C MalesTreatment (λ1) 345 09 -43 -01 119 08 -04 -2 01

(242) (08) (86) (08) (145) (10) (05) (1) (30)Cost Sharing (λ2) -148 -04 37 00 -46 -04 -01 1 -21

(227) (08) (55) (06) (120) (08) (05) (1) (28)Saturation (λ3) -138 -07 -254 -27 5 02 04 3 176

(1228) (46) (462) (47) (674) (49) (22) (6) (130)Control Mean 2440 749 1028 666 1403 679 253 4 376Treatment Effect () 142 90 -42 -6 85 79 -17 -469 11Joint F-Test (p-value) 0392 0660 0789 0931 0869 0863 0665 0406 0477Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 1974 1974 1980 1980 1976 1976 2006 1989 97

Panel D Older than 12Treatment (λ1) 575˚˚˚ 22˚˚˚ 69 14˚˚ 293˚˚˚ 22˚˚˚ 06 -1 -18

(199) (07) (52) (06) (102) (08) (04) (1) (43)Cost Sharing (λ2) -398˚ -15˚ -52 -08 -186˚ -16˚ -07 2˚˚ 58

(204) (08) (48) (05) (100) (08) (04) (1) (39)Saturation (λ3) 263 09 54 14 22 03 -01 -1 55

Panel E 12 or YoungerTreatment (λ1) -96 -06 -73 -09˚ -100 -05 -05 -2 05

(132) (06) (62) (05) (93) (07) (05) (1) (34)Cost Sharing (λ2) 96 06 55 08 60 07 -02 -0 -25

(134) (05) (55) (05) (88) (07) (04) (1) (29)Saturation (λ3) -1791˚˚˚ -70˚˚˚ -643˚˚ -70˚˚ -820˚ -57˚ 24 -0 06

(676) (26) (312) (27) (443) (32) (19) (6) (172)Control Mean 2204 74 937 659 1262 667 262 3 384Treatment Effect () -44 -65 -78 -95 -79 -56 -19 -529 46Joint F-Test (p-value) 0074 0065 0222 0038 0302 0304 0080 0385 0834Treatment FDR q-value 743 743 743 743 743 743 743 743 743Number Observations 1974 1974 1982 1980 1973 1973 1999 1989 60

Notes Analysis pools the last 12 months of retrospective data from KLPS-4 Columns (1)-(6) (8) and (9) are trimmed at the top 1 of observations See Section 5 for notes on outcome construction See Table 2for notes on the regression specification denotes statistical significance at 10 pct at 5 pct and at 1 pct

(3)Per-Capita

FoodConsumption

(4)Log Per-Capita

FoodConsumption

Consumption

(6)Log Per-Capita

Non-FoodConsumption

AverageMeals Eaten

Panel A Full SampleTreatment (λ1) 1163˚ 33˚˚ 278 17 411˚˚ 35˚˚ 01 7 184

(664) (14) (183) (12) (193) (13) (06) (5) ()Cost Sharing (λ2) -405 -17 5 -07 -490˚˚ -31˚˚ 00 -9˚˚ -55

(562) (14) (230) (14) (196) (12) (06) (4) ()Saturation (λ3) 9526 175 -1819 28 1467 111 -09 11 5556

(6328) (142) (2342) (129) (1089) (73) (50) (22) ()Control Mean 2878 739 904 653 1222 648 223 5 392Treatment Effect () 404 286 307 157 336 297 4 1249 1045Joint F-Test (p-value) 313 141 022 437 095 040 987 139 Treatment FDR q-value 145 106 163 163 106 106 287 163 1000Number Observations 718 718 718 718 718 718 725 717 30

Panel B FemalesTreatment (λ1) 707 19 303 17 12 08 04 1 298

(585) (15) (220) (15) (259) (18) (09) (5) ()Cost Sharing (λ2) -549 -20 -73 -16 -164 -17 -06 -3 -317

(576) (16) (265) (16) (225) (17) (10) (5) ()Saturation (λ3) 3007 143 232 173 936 88 22 -2 00

Panel C MalesTreatment (λ1) 1581 46˚˚ 248 16 807˚ 61˚˚ -03 12 -13

(1175) (23) (228) (17) (420) (23) (11) (10) ()Cost Sharing (λ2) -246 -16 57 -01 -808˚˚ -44˚˚ 06 -16˚ 292

(895) (20) (266) (17) (321) (19) (10) (9) ()Saturation (λ3) 14018 210 -3016 -53 1959 146 -27 20 00

(8993) (189) (2369) (146) (1833) (106) (54) (37) ()Control Mean 3488 753 980 661 1393 657 215 6 384Treatment Effect () 453 381 253 152 580 474 -13 2009 -136Joint F-Test (p-value) 0377 0265 0064 0538 0100 0036 0874 0334 Treatment FDR q-value 354 186 390 390 186 103 658 354 1000Number Observations 347 347 350 350 347 347 354 349 16

(781) (19) (251) (14) (350) (22) (13) (10) ()Cost Sharing (λ2) -713 -34˚ -408 -35˚˚ -943˚˚˚ -54˚˚˚ -00 -23˚˚ -605

(751) (17) (305) (14) (346) (20) (12) (10) ()Saturation (λ3) 19744˚˚ 381˚˚ -859 141 1802 148 -64 25 00

Panel E 12 or YoungerTreatment (λ1) -486 -09 -62 -15 -87 02 01 -1 -2262

(769) (14) (179) (12) (202) (14) (08) (4) ()Cost Sharing (λ2) -522 -11 275 10 -162 -14 03 -1 67

(540) (14) (232) (15) (214) (15) (08) (3) ()Saturation (λ3) -2759 -57 -2589 -78 1201 74 51 4 00

(4087) (114) (2025) (135) (1215) (89) (55) (16) ()Control Mean 3375 753 947 659 1348 662 23 3 485Treatment Effect () -144 -93 -66 -157 -65 16 4 -324 Joint F-Test (p-value) 0662 0658 0321 0698 0415 0434 0819 0925 Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 367 367 368 368 367 367 371 367 11

Table 6 Consumption - Pooled (trimmed including SCY and VocEd)

(3)Per-Capita

FoodConsumption

(4)Log Per-Capita

FoodConsumption

Consumption

(6)Log Per-Capita

Non-FoodConsumption

AverageMeals Eaten

Panel A Full SampleTreatment (λ1) 172 06 32 03 116˚ 08 -03 -0 13

(132) (04) (44) (04) (62) (05) (03) (1) (22)Cost Sharing (λ2) -79 -03 -3 -02 -43 -03 00 -0 -12

(130) (05) (50) (04) (64) (05) (03) (0) (22)Saturation (λ3) 177 -06 -455 -14 75 06 -08 -2 66

Panel B FemalesTreatment (λ1) 47 05 47 06 8 05 -00 -0 -05

(134) (06) (50) (04) (87) (08) (04) (1) (41)Cost Sharing (λ2) -86 -03 -41 -04 -2 -00 -01 0 72˚

(153) (06) (56) (05) (85) (08) (04) (1) (41)Saturation (λ3) -1413 -33 -352 -01 -685 -28 -12 -6˚ -175

Panel C MalesTreatment (λ1) 299 07 18 01 227˚ 10 -05 -0 21

(248) (07) (67) (06) (120) (07) (04) (1) (26)Cost Sharing (λ2) -89 -04 33 -00 -88 -05 02 -1 -44˚

(222) (07) (65) (06) (119) (07) (04) (1) (23)Saturation (λ3) 1467 16 -533 -24 706 35 -05 1 100

Panel D Older than 12Treatment (λ1) 664˚˚˚ 22˚˚˚ 162˚˚˚ 18˚˚˚ 311˚˚˚ 20˚˚˚ -01 1 30

(188) (06) (53) (05) (81) (06) (03) (1) (45)Cost Sharing (λ2) -312 -13˚ -104˚ -12˚˚ -166˚ -12˚ -00 -1 12

(207) (07) (61) (05) (89) (07) (03) (1) (35)Saturation (λ3) 2247 56 -16 38 916 54 -31˚ 0 77

(1713) (48) (402) (29) (605) (44) (16) (5) (167)Control Mean 1926 722 804 64 980 644 246 2 35Treatment Effect () 345 203 201 167 317 186 -4 349 266Joint F-Test (p-value) 0006 0004 0013 0005 0004 0013 0203 0600 0607Treatment FDR q-value 002 002 003 002 002 002 346 229 229Number Observations 2857 2857 2843 2842 2855 2855 2913 2903 110

(177) (05) (60) (04) (81) (06) (04) (1) (30)Cost Sharing (λ2) 33 01 50 03 24 02 02 0 -20

(134) (05) (63) (05) (79) (06) (04) (1) (30)Saturation (λ3) -1983˚˚ -68˚˚ -899˚˚ -64˚˚ -725˚ -39 10 -5 -03

Notes Analysis pools the last 12 months of retrospective data from KLPS-3 and KLPS-4 Columns (1)-(6) (8) and (9) are trimmed at the top 1 of observations Analysis includes KLPS respondents whoparticipated in SCY or VocEd with indicators for receiving a SCY grant or a vocational training voucher See Section 5 for notes on outcome construction See Table 2 for additional notes on the regressionspecification Observations are weighted to be representative of the original PSDP population and include KLPS population weights and KLPS intensive tracking weights denotes statistical significance at 10pct at 5 pct and at 1 pct

Table 7 Household Wealth - KLPS-4 (trimmed)

(1)Per-Capita

Household Wealth

(2)Log Per-Capita

Household Wealth

(3)Per-Capita

Household Assets

(4)Log Per-Capita

Household Assets

(5)Per-Capita

Livestock Ownership

(6)Log Per-Capita

Livestock OnwershipPanel A Full SampleTreatment (λ1) 69 10 61 10 -5 12

(50) (07) (45) (06) (6) (12)Cost Sharing (λ2) -60 -09˚ -47 -10˚˚ 5 10

(39) (05) (33) (05) (7) (13)Saturation (λ3) -394˚ -40 -383˚˚ -26 -113˚˚˚ -138˚˚

(215) (32) (180) (29) (42) (65)Control Mean 522 583 460 571 53 352Treatment Effect () 133 92 133 96 -88 112Joint F-Test (p-value) 041 180 029 153 055 021Treatment FDR q-value 362 362 362 362 362 362Number Observations 4085 4084 4085 4084 4093 2208

Panel B FemalesTreatment (λ1) 36 08 29 09 -8 15

(51) (07) (47) (07) (5) (14)Cost Sharing (λ2) -45 -12˚ -31 -13˚ 2 -09

(46) (07) (41) (06) (6) (17)Saturation (λ3) -489˚ -37 -485˚˚ -28 -83˚˚ -168˚˚

Panel C MalesTreatment (λ1) 102 11 93 11 -1 09

(97) (11) (89) (11) (10) (16)Cost Sharing (λ2) -76 -05 -64 -07 8 26

(81) (07) (71) (07) (14) (17)Saturation (λ3) -317 -43 -299 -24 -136˚˚ -109

Panel D Older than 12Treatment (λ1) 253˚˚˚ 23˚˚ 222˚˚˚ 23˚˚ -1 23

(89) (10) (82) (09) (9) (16)Cost Sharing (λ2) -186˚˚ -18˚˚ -153˚ -19˚˚ -0 05

(86) (09) (78) (08) (8) (18)Saturation (λ3) 22 03 6 23 -142˚˚˚ -190˚˚

Panel E 12 or YoungerTreatment (λ1) -93 -02 -80 -01 -8 00

(59) (09) (54) (08) (7) (15)Cost Sharing (λ2) 44 -02 37 -03 11 11

(49) (06) (43) (06) (9) (16)Saturation (λ3) -976˚˚˚ -105˚˚˚ -906˚˚˚ -91˚˚ -117˚ -111

(293) (37) (267) (39) (62) (81)Control Mean 579 59 519 579 51 363Treatment Effect () -161 -24 -155 -10 -155 4Joint F-Test (p-value) 0012 0033 0011 0112 0253 0495Treatment FDR q-value 740 1000 740 1000 740 1000Number Observations 1985 1984 1986 1985 1989 1027

Notes Analysis uses KLPS-4 data Results are trimmed at the top 1 of observations See Section 6 for notes on outcome construction Treatment is an indicator variable equal to 1for PSDP Worm Groups 1 and 2 which received an additional 24 years of deworming on average compared to Group 3 Reported estimates for Female and Male are constructed froma single regression including treatment-female cost-sharing-female and saturation-female interaction terms Reported estimates for Older than 12 and 12 or Younger also report resultsusing a single regression including an indicator for those older than 12 at baseline and analogous interaction terms to Panels B and C Covariates follow Baird et al (2016) and includecontrols for baseline 1998 primary school population geographic zone of the school survey wave and month of interview a female indicator variable baseline 1998 school grade fixedeffects the average school test score on the 1996 Busia District mock exams total primary school pupils within 6 km and a cost-sharing school indicator Those treated in a separatevocational training intervention (VocEd) which occurred prior to KLPS-3 are dropped from the KLPS-3 and KLPS-4 sample Those treated in a separate small grant intervention (SCY)which occurred after KLPS-3 are dropped from the KLPS-4 sample Observations are weighted to be representative of the original PSDP population and include KLPS populationweights SCY and VocEd control group weights and KLPS intensive tracking weights Standard errors are clustered at the 1998 school level The Joint F-Test (p-value) gives the p-valueassociated with an F-test on the joint significance of the treatment cost-sharing and saturation coefficients against the null hypothesis that all three coefficients are jointly equal to zeroThe FDR adjustment is carried out across the six outcomes within this family separately by panel (full sample females males older than 12 and 12 or younger) denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Table 8 Household Wealth - KLPS-4 (untrimmed)

(1)Per-Capita

Household Wealth

(2)Log Per-Capita

Household Wealth

(3)Per-Capita

Household Assets

(4)Log Per-Capita

Household Assets

(5)Per-Capita

Livestock Ownership

(6)Log Per-Capita

Livestock OnwershipPanel A Full SampleTreatment (λ1) -36 06 -40 07 4 16

(154) (08) (152) (08) (13) (13)Cost Sharing (λ2) -185˚ -12˚˚ -173˚ -13˚˚ -12 03

(101) (06) (99) (06) (14) (14)Saturation (λ3) -953˚ -67˚ -754 -48 -199˚˚˚ -151˚˚

(553) (39) (541) (36) (66) (63)Control Mean 881 594 811 582 70 359Treatment Effect () -41 59 -50 64 63 148Joint F-Test (p-value) 062 072 089 071 019 012Treatment FDR q-value 1000 1000 1000 1000 1000 1000Number Observations 4135 4134 4135 4134 4135 2250

(121) (08) (118) (08) (9) (14)Cost Sharing (λ2) -47 -12 -56 -13˚ 9 -06

(76) (07) (74) (07) (9) (18)Saturation (λ3) -1114 -62 -974 -48 -141˚˚ -176˚˚

Panel C MalesTreatment (λ1) -86 06 -106 06 20 18

(322) (14) (318) (14) (23) (17)Cost Sharing (λ2) -314 -12 -281 -13 -32 11

(192) (09) (185) (09) (27) (17)Saturation (λ3) -852 -71 -605 -48 -247˚˚˚ -130

Panel D Older than 12Treatment (λ1) 151 19˚ 144 20˚˚ 8 29

(138) (10) (125) (10) (21) (19)Cost Sharing (λ2) -290˚˚ -20˚˚ -261˚˚ -23˚˚ -28 -04

(119) (09) (108) (09) (21) (19)Saturation (λ3) -625 -19 -416 07 -209˚˚ -196˚˚

(671) (54) (640) (52) (105) (91)Control Mean 685 583 608 571 77 349Treatment Effect () 221 178 236 186 97 253Joint F-Test (p-value) 0104 0128 0117 0066 0054 0023Treatment FDR q-value 287 214 287 214 502 214Number Observations 2071 2071 2071 2071 2071 1168

Panel E 12 or YoungerTreatment (λ1) -160 -03 -162 -03 2 03

(233) (10) (235) (10) (16) (16)Cost Sharing (λ2) -116 -07 -118 -07 2 07

(160) (08) (161) (08) (17) (16)Saturation (λ3) -1321˚ -129˚˚˚ -1101 -110˚˚˚ -220˚˚ -133˚

(711) (41) (713) (41) (86) (79)Control Mean 1040 602 977 592 63 368Treatment Effect () -153 -34 -165 -26 33 28Joint F-Test (p-value) 0114 0013 0166 0044 0099 0372Treatment FDR q-value 1000 1000 1000 1000 1000 1000Number Observations 2011 2010 2011 2010 2011 1049

Notes Analysis uses KLPS-4 data Results are untrimmed See Section 6 for notes on outcome construction See Table 7 for notes on the regression specification denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Table 9 Household Wealth - KLPS-4 (trimmed including SCY and VocEd)

(1)Per-Capita

Household Wealth

(2)Log Per-Capita

Household Wealth

(3)Per-Capita

Household Assets

(4)Log Per-Capita

Household Assets

(5)Per-Capita

Livestock Ownership

(6)Log Per-Capita

Livestock OnwershipPanel A Full SampleTreatment (λ1) 21 02 19 03 -10˚ -06

(39) (06) (34) (05) (6) (09)Cost Sharing (λ2) 3 -00 12 -01 7 13

(33) (04) (26) (04) (7) (10)Saturation (λ3) -377 -53˚ -317 -32 -148˚˚˚ -207˚˚˚

(227) (30) (198) (27) (41) (57)Control Mean 531 586 465 574 58 362Treatment Effect () 39 21 40 29 -179 -67Joint F-Test (p-value) 175 296 127 543 003 005Treatment FDR q-value 1000 1000 1000 1000 900 1000Number Observations 4949 4948 4949 4948 4950 2747

(42) (06) (39) (05) (6) (11)Cost Sharing (λ2) -28 -06 -12 -06 -0 -08

(35) (05) (31) (04) (7) (14)Saturation (λ3) -546˚˚ -72˚˚ -492˚˚ -51 -131˚˚˚ -268˚˚˚

Panel C MalesTreatment (λ1) 31 -00 34 01 -11 -15

(62) (08) (56) (08) (9) (13)Cost Sharing (λ2) 29 05 33 04 14 32˚˚

(55) (06) (45) (05) (13) (16)Saturation (λ3) -235 -37 -170 -16 -162˚˚ -154˚

(295) (37) (258) (35) (64) (78)Control Mean 617 6 538 588 63 378Treatment Effect () 50 -1 64 11 -178 -162Joint F-Test (p-value) 0449 0533 0320 0732 0075 0096Treatment FDR q-value 1000 1000 1000 1000 1000 1000Number Observations 2433 2432 2435 2434 2432 1350

Panel D Older than 12Treatment (λ1) 161˚˚ 12 140˚˚ 13˚ -10 -05

(65) (08) (56) (07) (8) (13)Cost Sharing (λ2) -92 -07 -58 -07 -3 04

(64) (07) (55) (07) (7) (14)Saturation (λ3) 22 -19 73 07 -195˚˚˚ -265˚˚˚

Panel E 12 or YoungerTreatment (λ1) -103˚˚ -07 -90˚ -05 -11 -11

(50) (07) (46) (07) (7) (12)Cost Sharing (λ2) 74˚ 04 62˚ 03 18˚ 20˚

(42) (05) (35) (05) (10) (12)Saturation (λ3) -896˚˚˚ -107˚˚˚ -807˚˚˚ -84˚˚˚ -127˚˚ -171˚˚

(254) (30) (230) (31) (59) (72)Control Mean 586 594 524 583 54 371Treatment Effect () -175 -71 -171 -54 -201 -115Joint F-Test (p-value) 0008 0008 0008 0067 0107 0053Treatment FDR q-value 194 278 194 278 194 278Number Observations 2393 2392 2394 2393 2397 1273

Notes Analysis uses KLPS-4 data Results are trimmed at the top 1 of observations Analysis includes KLPS respondents who participated in SCY or VocEd with indicators forreceiving a SCY grant or a vocational training voucher See Section 6 for notes on outcome construction See Table 7 for notes on the regression specification Observations are weightedto be representative of the original KLPS population and include KLPS population weights and KLPS intensive tracking weights denotes statistical significance at 10 pct at 5pct and at 1 pct

Table 10 Earnings Part 1 - Pooled (trimmed)

IndividualEarnings

Log IndividualEarnings

(3)Individual

WageEarnings

(4)Log Individual

WageEarnings

(5)Self-

EmploymentProfit

(6)Log Self-

EmploymentProfit

(7)IndividualFarmingProfits

(8)Log Individual

FarmingProfits

(9)Taxes on

Individual Wagesand Profits

(10)Log Taxes on

Panel A Full SampleTreatment (λ1) 80 09 81 07 41˚ 12 -0 06 9 21

(76) (06) (68) (07) (24) (09) (2) (11) (10) (16)Cost Sharing (λ2) -32 -04 -63 -06 -7 -05 2 12 -12 06

(76) (06) (67) (08) (25) (10) (2) (13) (9) (15)Saturation (λ3) -366 -14 -280 27 255 09 -23˚ -192˚˚ -49 28

(463) (28) (506) (43) (195) (64) (12) (83) (40) (94)Control Mean 1218 673 887 697 212 613 9 446 51 545Treatment Effect () 65 88 92 68 193 109 -38 62 172 187Joint F-Test (p-value) 427 297 316 816 314 608 308 035 117 307Treatment FDR q-value 609 609 609 609 609 609 803 763 609 609Number Observations 13624 7698 13628 5103 13638 2652 13707 780 13655 1381

Panel B FemalesTreatment (λ1) 41 13 22 11 32 08 -1 12 -7 04

(62) (10) (68) (12) (26) (14) (3) (18) (9) (28)Cost Sharing (λ2) -0 -03 -49 -11 13 -07 -0 09 9 -01

(67) (10) (63) (11) (31) (18) (3) (15) (10) (26)Saturation (λ3) -270 -33 2 88 151 -35 -31˚˚ -205˚ -57 174

(403) (46) (361) (69) (206) (99) (15) (106) (59) (155)Control Mean 674 62 507 664 123 561 9 417 27 502Treatment Effect () 60 124 44 107 258 82 -137 110 -246 40Joint F-Test (p-value) 0550 0206 0884 0567 0360 0888 0174 0027 0753 0712Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 6826 3175 6826 1807 6825 1275 6816 444 6817 511

Panel C MalesTreatment (λ1) 118 06 138 04 51 13 1 -01 24 30

(133) (07) (110) (09) (48) (13) (3) (21) (22) (18)Cost Sharing (λ2) -63 -04 -77 -02 -27 -03 4 15 -33 12

(119) (08) (112) (10) (43) (13) (3) (20) (20) (20)Saturation (λ3) -439 -05 -502 -02 340 40 -16 -183˚ -41 -27

(643) (34) (780) (52) (264) (75) (16) (106) (89) (139)Control Mean 1728 711 1243 716 296 656 9 488 74 568Treatment Effect () 68 62 111 40 171 125 59 -14 326 260Joint F-Test (p-value) 0562 0786 0266 0965 0570 0669 0578 0311 0110 0101Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 6798 4523 6802 3296 6813 1377 6891 336 6838 870

Panel D Older than 12Treatment (λ1) 258˚˚ 19˚˚ 162˚ 20˚ 70˚ 23˚˚ -3 -14 37˚ 21

(108) (08) (89) (11) (39) (10) (3) (19) (22) (25)Cost Sharing (λ2) -76 -01 -105 -12 6 06 3 08 -42˚˚ -08

(99) (07) (74) (12) (38) (12) (2) (18) (20) (25)Saturation (λ3) 201 38 -134 41 333 86 -31˚˚ -295˚˚ -26 03

(610) (49) (518) (60) (283) (82) (13) (117) (83) (125)Control Mean 1177 665 846 688 244 611 9 444 44 551Treatment Effect () 219 173 192 181 289 210 -307 -154 846 193Joint F-Test (p-value) 0045 0041 0154 0322 0173 0061 0065 0104 0040 0823Treatment FDR q-value 071 071 084 084 084 071 153 201 089 191Number Observations 6791 4195 6793 2613 6797 1592 6836 436 6799 768

Panel E 12 or YoungerTreatment (λ1) -75 -01 28 -02 0 -09 2 29 -13 35˚

(100) (08) (90) (09) (32) (12) (3) (18) (11) (18)Cost Sharing (λ2) 4 -07 -26 00 -16 -19 1 17 13 18

(98) (08) (92) (09) (31) (16) (3) (20) (9) (17)Saturation (λ3) -820 -82 -278 36 139 -97 -15 -66 -47 116

(705) (53) (729) (48) (204) (94) (18) (101) (53) (121)Control Mean 1242 682 906 703 187 62 9 445 57 538Treatment Effect () -61 -15 31 -21 2 -94 229 257 -232 299Joint F-Test (p-value) 0675 0433 0934 0843 0773 0209 0494 0075 0535 0031Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 6780 3461 6782 2467 6788 1043 6818 337 6803 602

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12 months prior) Results aretrimmed at the top 1 of observations See Section 71 for notes on outcome construction Treatment is an indicator variable equal to 1 for PSDP Worm Groups 1 and 2 which received an additional 24 years of deworming onaverage compared to Group 3 Reported estimates for Female and Male are constructed from a single regression including treatment-female cost-sharing-female and saturation-female interaction terms Reported estimates forOlder than 12 and 12 or Younger also report results using a single regression including an indicator for those older than 12 at baseline and analogous interaction terms to Panels B and C Covariates follow Baird et al (2016)and include controls for baseline 1998 primary school population geographic zone of the school survey wave and month of interview a female indicator variable baseline 1998 school grade fixed effects the average schooltest score on the 1996 Busia District mock exams total primary school pupils within 6 km and a cost-sharing school indicator Those treated in a separate vocational training intervention (VocEd) which occurred prior toKLPS-3 are dropped from the KLPS-3 and KLPS-4 sample Those treated in a separate small grant intervention (SCY) which occurred after KLPS-3 are dropped from the KLPS-4 sample Observations are weighted to berepresentative of the original PSDP population and include KLPS population weights SCY and VocEd control group weights and KLPS intensive tracking weights Standard errors are clustered at the 1998 school level TheJoint F-Test (p-value) gives the p-value associated with an F-test on the joint significance of the treatment cost-sharing and saturation coefficients against the null hypothesis that all three coefficients are jointly equal to zeroThe FDR adjustment is carried out across the 18 outcomes within this family separately by panel (full sample females males older than 12 and 12 or younger) denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 11 Earnings Part 1 - Pooled (untrimmed)

IndividualEarnings

(3)Individual

WageEarnings

(4)Log Individual

WageEarnings

(5)Self-

EmploymentProfit

(6)Log Self-

EmploymentProfit

(8)Log Individual

FarmingProfits

(9)Taxes on

(10)Log Taxes on

Panel A Full SampleTreatment (λ1) -254 11˚ -29 09 -110 11 -115 11 228 21

(1064) (06) (1056) (08) (165) (10) (148) (15) (6348) (16)Cost Sharing (λ2) 2360 -08 2147 -08 -178˚ -10 391 -08 20373˚ 06

(2177) (06) (2147) (09) (102) (10) (415) (15) (11617) (15)Saturation (λ3) 2667 03 2642 46 1188 -34 -1164 -183˚ 5386 28

(7933) (30) (7610) (48) (1538) (64) (1523) (103) (47222) (94)Control Mean 2738 681 1983 705 656 63 98 478 137 545Treatment Effect () -93 104 -15 87 -167 108 -1177 103 1668 187Joint F-Test (p-value) 754 337 774 661 314 505 772 294 348 307Treatment FDR q-value 1000 933 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 13794 7868 13794 5269 13794 2808 13796 869 13809 1381

(999) (10) (1009) (12) (176) (16) (171) (23) (6242) (28)Cost Sharing (λ2) 4900 -02 4238 -06 -132 -12 793 12 -945 -01

(3962) (10) (3879) (12) (125) (19) (815) (18) (11581) (26)Saturation (λ3) -3290 -16 -3641 83 1729 -50 -1378 -72 -4678 174

Panel C MalesTreatment (λ1) -565 12 -99 09 -381 12 -86 25 472 30

(1949) (08) (1911) (10) (401) (13) (155) (26) (7047) (18)Cost Sharing (λ2) 86 -12 255 -08 -200˚ -08 31 -25 39190 12

(2276) (09) (2285) (11) (117) (13) (112) (26) (25446) (20)Saturation (λ3) 7304 13 7603 30 709 -23 -1007 -277˚˚ 14730 -27

Panel D Older than 12Treatment (λ1) 522 24˚˚˚ 482 25˚˚ 204 33˚˚˚ -165 11 3227 21

(2074) (08) (2066) (12) (209) (12) (199) (20) (8478) (25)Cost Sharing (λ2) 470 -09 -40 -20 -252 -03 761 -12 39239 -08

(2751) (08) (2647) (13) (153) (12) (773) (17) (27218) (25)Saturation (λ3) 4098 58 4278 82 1658 53 -1839 -314˚˚ 14721 03

Panel E 12 or YoungerTreatment (λ1) -879 -03 -367 -02 -406 -21 -106 15 -3711 35˚

(1176) (08) (1128) (09) (327) (13) (142) (23) (6007) (18)Cost Sharing (λ2) 4077 -07 4107 03 -115 -19 86 -02 3777 18

(3748) (09) (3740) (10) (108) (16) (152) (21) (9544) (17)Saturation (λ3) 2723 -67 2722 35 737 -160 -735 -23 -21106 116

(7835) (54) (7628) (53) (1175) (102) (1357) (110) (41807) (121)Control Mean 2626 689 1710 709 819 644 97 485 84 538Treatment Effect () -335 -29 -214 -25 -496 -232 -1095 136 -44273 299Joint F-Test (p-value) 0557 0572 0640 0845 0438 0024 0902 0916 0908 0031Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 932Number Observations 6850 3531 6850 2535 6850 1105 6850 369 6858 602

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12 months prior) Resultsare untrimmed See Section 71 for notes on outcome construction See Table 10 for notes on the regression specification denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 12 Earnings Part 1 - KLPS-4 (trimmed)

IndividualEarnings

(3)Individual

WageEarnings

(4)Log Individual

WageEarnings

(5)Self-

EmploymentProfit

(6)Log Self-

EmploymentProfit

(8)Log Individual

FarmingProfits

(9)Taxes on

(10)Log Taxes on

Panel A Full SampleTreatment (λ1) 85 11 106 -00 113˚ 19 2 08 25 31˚

(171) (09) (138) (11) (58) (14) (5) (13) (29) (16)Cost Sharing (λ2) -34 -05 -98 -04 -57 -12 5 19 -38 -09

(175) (08) (132) (12) (65) (15) (6) (13) (26) (14)Saturation (λ3) -879 -78 -1015 -48 496 -09 -71˚˚ -250˚˚˚ -210 01

(1183) (48) (980) (75) (357) (80) (36) (84) (132) (106)Control Mean 2133 687 1488 731 394 623 21 432 139 564Treatment Effect () 40 104 71 -3 287 170 109 75 177 272Joint F-Test (p-value) 745 148 284 723 248 544 180 005 036 138Treatment FDR q-value 568 461 563 770 365 389 568 563 563 365Number Observations 4072 3330 4074 1944 4077 1433 4078 640 4086 924

Panel B FemalesTreatment (λ1) -0 13 22 -04 50 23 -1 19 -30 02

(141) (15) (154) (16) (49) (18) (7) (21) (26) (26)Cost Sharing (λ2) 182 03 -23 -08 40 -19 -2 11 27 07

(164) (14) (153) (15) (79) (26) (7) (16) (30) (24)Saturation (λ3) -1110 -108 -607 52 443 -51 -87˚˚ -226˚ -344˚˚ -82

(1022) (79) (865) (110) (460) (131) (42) (120) (152) (146)Control Mean 1136 629 858 707 210 568 19 397 67 516Treatment Effect () -0 127 26 -39 239 208 -65 171 -440 20Joint F-Test (p-value) 0432 0146 0803 0820 0445 0432 0092 0015 0144 0926Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2099 1500 2099 770 2103 673 2096 343 2106 322

Panel C MalesTreatment (λ1) 174 09 194 01 176 14 6 -05 78 44˚˚

(306) (10) (235) (13) (120) (19) (10) (25) (64) (21)Cost Sharing (λ2) -244 -11 -169 -01 -150 -06 11 27 -101˚ -16

(280) (10) (226) (14) (113) (16) (11) (20) (57) (20)Saturation (λ3) -715 -61 -1344 -97 523 21 -57 -270˚˚ -109 18

Panel D Older than 12Treatment (λ1) 479˚˚ 32˚˚ 296˚ 27 201˚ 40˚˚ -1 -03 103˚ 54˚˚

(223) (14) (172) (16) (102) (17) (8) (23) (61) (24)Cost Sharing (λ2) -257 -10 -330˚˚ -26 -57 -04 6 05 -125˚˚ -42

(223) (11) (136) (17) (109) (18) (6) (18) (56) (25)Saturation (λ3) 81 12 -499 -22 421 119 -88˚˚ -315˚˚˚ -116 53

Panel E 12 or YoungerTreatment (λ1) -252 -09 0 -16 -4 -12 6 19 -35 28

(278) (13) (225) (16) (76) (16) (7) (22) (34) (23)Cost Sharing (λ2) 157 00 101 15 -56 -24 5 33 37 12

(255) (12) (216) (13) (78) (21) (11) (22) (31) (16)Saturation (λ3) -1737 -190˚˚ -1257 -39 481 -160˚ -58 -164 -270 41

(2035) (79) (1669) (92) (455) (94) (59) (124) (199) (152)Control Mean 2433 707 1675 745 409 639 22 435 155 565Treatment Effect () -103 -99 0 -169 -10 -130 286 173 -226 245Joint F-Test (p-value) 0748 0128 0829 0645 0454 0211 0394 0021 0536 0306Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 1979 1592 1978 1017 1985 617 1990 282 1988 445

Notes Analysis uses KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12 months prior) Results are trimmed at the top1 of observations See Section 71 for notes on outcome construction See Table 10 for notes on the regression specification denotes statistical significance at 10 pct at 5 pct and at 1 pct

IndividualEarnings

(3)Individual

WageEarnings

(4)Log Individual

WageEarnings

(5)Self-

EmploymentProfit

(6)Log Self-

EmploymentProfit

(8)Log Individual

FarmingProfits

(9)Taxes on

(10)Log Taxes on

Panel A Full SampleTreatment (λ1) 83 08 77 11 -4 15 -2 08 0 -20

(83) (08) (75) (10) (36) (12) (2) (15) (3) (31)Cost Sharing (λ2) -88 -04 -128 -09 35 01 -0 -41˚˚ 3 79˚

(91) (08) (80) (10) (30) (13) (1) (18) (2) (41)Saturation (λ3) -781˚ -28 -378 31 287 -01 8 49 53˚˚ 272

(446) (49) (479) (53) (249) (104) (11) (104) (26) (184)Control Mean 1165 667 884 682 207 601 6 499 9 524Treatment Effect () 71 74 87 101 -18 141 -303 78 17 -229Joint F-Test (p-value) 051 503 217 746 260 419 249 091 117 101Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 4525 2897 4526 2030 4534 898 4552 140 4541 285

Panel B FemalesTreatment (λ1) 47 11 18 17 25 07 -1 06 6 -37

(99) (12) (87) (17) (43) (21) (3) (21) (5) (65)Cost Sharing (λ2) -199˚ -10 -163˚ -10 -11 -09 -1 -21 2 04

(100) (13) (83) (18) (34) (21) (2) (22) (5) (47)Saturation (λ3) -761 -27 -330 70 -53 10 1 -30 125 833˚˚˚

(553) (65) (448) (93) (239) (161) (15) (134) (75) (277)Control Mean 682 613 505 637 121 542 8 491 7 486Treatment Effect () 70 100 36 158 207 71 -159 54 808 -462Joint F-Test (p-value) 0027 0645 0119 0768 0834 0969 0690 0793 0273 0015Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2247 1215 2249 722 2243 462 2232 101 2249 121

Panel C MalesTreatment (λ1) 112 05 133 06 -32 20 -3 19 -6 -10

(159) (12) (140) (13) (51) (15) (3) (26) (4) (45)Cost Sharing (λ2) 8 01 -102 -08 76˚ 10 -0 -73˚˚ 4 109˚

(144) (11) (124) (11) (44) (16) (2) (34) (4) (61)Saturation (λ3) -787 -29 -402 10 558˚ 03 13 229 -6 -68

(743) (65) (733) (62) (286) (118) (12) (241) (21) (230)Control Mean 1601 704 1227 708 284 647 5 511 11 548Treatment Effect () 70 52 108 59 -114 185 -519 174 -498 -101Joint F-Test (p-value) 0393 0849 0520 0927 0020 0208 0322 0174 0487 0293Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2278 1682 2277 1308 2291 436 2320 39 2292 164

Panel D Older than 12Treatment (λ1) 136 07 89 12 -19 21 -6˚˚˚ -05 1 -99˚

(123) (11) (117) (15) (52) (17) (2) (19) (3) (55)Cost Sharing (λ2) 27 05 -28 -04 81˚˚ 02 3˚ -31 6 119˚˚

(124) (10) (108) (14) (39) (19) (1) (20) (5) (47)Saturation (λ3) -351 04 -571 06 482 86 0 -111 11 -49

(604) (44) (553) (60) (296) (131) (11) (128) (17) (234)Control Mean 1257 668 903 684 288 612 9 5 8 57Treatment Effect () 108 71 99 116 -67 187 -712 -52 87 -4669Joint F-Test (p-value) 0128 0437 0313 0816 0055 0513 0024 0157 0346 0073Treatment FDR q-value 376 519 519 519 651 318 094 651 651 236Number Observations 2255 1553 2254 1027 2259 541 2265 85 2257 166

Panel E 12 or YoungerTreatment (λ1) 29 06 60 06 6 06 2 19 -1 48

(103) (10) (93) (09) (45) (21) (2) (30) (4) (42)Cost Sharing (λ2) -189˚ -11 -209˚˚ -13 -7 -02 -3 -43 1 48

(105) (10) (90) (11) (43) (22) (2) (33) (3) (74)Saturation (λ3) -1203 -68 -164 56 33 -123 14 183 90˚ 640˚˚

(724) (87) (678) (60) (259) (155) (16) (122) (51) (255)Control Mean 1082 666 867 68 134 587 4 497 10 487Treatment Effect () 27 62 69 62 48 60 370 178 -58 391Joint F-Test (p-value) 0060 0377 0142 0576 0999 0800 0553 0235 0043 0005Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2270 1344 2272 1003 2275 357 2287 55 2284 119

Table 14 Earnings Part 1 - Pooled (trimmed including SCY and VocEd)

IndividualEarnings

(3)Individual

WageEarnings

(4)Log Individual

WageEarnings

(5)Self-

EmploymentProfit

(6)Log Self-

EmploymentProfit

(8)Log Individual

FarmingProfits

(9)Taxes on

(10)Log Taxes on

Panel A Full SampleTreatment (λ1) 136˚ 10˚ 116˚ 07 42˚ 15˚ -1 -03 1 18

(77) (06) (67) (07) (24) (09) (2) (10) (8) (12)Cost Sharing (λ2) -71 -06 -71 -06 -17 -08 2 15 -8 14

(76) (06) (64) (07) (26) (09) (2) (11) (6) (11)Saturation (λ3) 136 19 146 38 252˚ 65 -19 -156˚˚ -70˚ 79

(74) (10) (78) (11) (24) (15) (3) (16) (8) (24)Cost Sharing (λ2) -55 -07 -81 -09 -0 -15 -0 13 4 -03

(80) (11) (76) (11) (24) (16) (3) (14) (9) (23)Saturation (λ3) -113 -01 46 85 227 49 -26˚ -162˚˚ -35 130

(130) (06) (106) (07) (45) (11) (3) (19) (13) (15)Cost Sharing (λ2) -90 -06 -64 -04 -32 -02 3 18 -17 25

(107) (06) (97) (08) (44) (11) (4) (17) (12) (16)Saturation (λ3) 351 31 234 15 274 78 -13 -152 -99 64

Panel D Older than 12Treatment (λ1) 354˚˚˚ 22˚˚˚ 256˚˚˚ 22˚˚ 58 20˚ -3 -21 18 21

(108) (08) (89) (09) (38) (11) (3) (17) (11) (21)Cost Sharing (λ2) -102 -05 -97 -13 -14 -01 1 06 -26˚˚˚ 11

(95) (07) (72) (11) (38) (10) (2) (17) (10) (22)Saturation (λ3) 1206˚ 83˚ 857 79 288 107˚ -25˚ -246˚˚ -43 81

Panel E 12 or YoungerTreatment (λ1) -52 -02 12 -04 14 03 1 15 -11 29

(107) (08) (96) (08) (31) (13) (3) (17) (11) (19)Cost Sharing (λ2) -54 -07 -58 -00 -16 -18 2 23 8 15

(105) (08) (99) (08) (31) (15) (3) (17) (10) (18)Saturation (λ3) -839 -60 -450 16 198 -03 -12 -68 -74 128

(656) (43) (651) (37) (187) (89) (20) (99) (52) (105)Control Mean 1253 684 921 704 183 615 10 448 54 547Treatment Effect () -42 -16 13 -42 77 29 109 143 -202 252Joint F-Test (p-value) 0561 0446 0784 0806 0740 0623 0674 0088 0503 0049Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 7512 3992 7513 2789 7517 1251 7543 428 7535 696

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12 months prior) Resultsare trimmed at the top 1 of observations Analysis includes KLPS respondents who participated in SCY or VocEd with indicators for receiving a SCY grant or a vocational training voucher See Section 71 for notes onoutcome construction See Table 10 for notes on the regression specification Observations are weighted to be representative of the original KLPS population and include KLPS population weights and KLPS intensive trackingweights denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 15 Earnings Part 1 - Pooled (trimmed last month)

IndividualEarnings

(3)Individual

WageEarnings

(4)Log Individual

WageEarnings

(5)Self-

EmploymentProfit

(6)Log Self-

EmploymentProfit

(8)Log Individual

FarmingProfits

(9)Taxes on

(10)Log Taxes on

Panel A Full SampleTreatment (λ1) 18˚˚ 09 7 07 3 08 0 25˚ 1 16

(9) (07) (5) (07) (2) (08) (0) (14) (1) (14)Cost Sharing (λ2) -14 -09 -6 -06 -0 -06 -0 -38˚˚ -1 06

(9) (07) (5) (08) (2) (09) (0) (15) (1) (14)Saturation (λ3) 2 01 -12 62˚ 15 -26 0 -07 -3 57

(39) (31) (35) (33) (19) (45) (0) (85) (3) (93)Control Mean 99 457 70 491 25 407 0 233 5 323Treatment Effect () 184 83 96 70 105 82 211 224 207 152Joint F-Test (p-value) 119 487 347 312 679 573 393 074 251 432Treatment FDR q-value 000 000 000 000 000 000 000 000 000 000Number Observations 12881 6326 13625 4036 12893 2624 13711 268 13659 1390

Panel B FemalesTreatment (λ1) 6 07 1 06 1 03 0 23 -1 -05

(8) (11) (5) (10) (3) (12) (0) (17) (1) (25)Cost Sharing (λ2) -1 -06 -4 -08 3 -05 -0 -20 1 03

(8) (12) (5) (10) (3) (16) (0) (16) (1) (23)Saturation (λ3) -29 -23 8 133˚˚ 4 -65 0 -58 -5 205˚˚

(45) (47) (27) (52) (21) (72) (1) (123) (5) (101)Control Mean 53 412 39 476 15 363 0 22 2 287Treatment Effect () 110 72 30 55 86 30 292 209 -235 -51Joint F-Test (p-value) 0305 0679 0831 0063 0644 0792 0911 0424 0551 0185Treatment FDR q-value 000 000 000 000 000 000 000 000 000 000Number Observations 6378 2472 6830 1274 6375 1267 6813 192 6820 514

Panel C MalesTreatment (λ1) 30 10 12 08 4 13 0 36 2 28˚

(19) (09) (9) (09) (5) (10) (0) (22) (2) (15)Cost Sharing (λ2) -26 -10 -7 -05 -3 -07 -0 -79˚˚ -3˚ 09

(17) (08) (9) (10) (4) (10) (0) (32) (2) (18)Saturation (λ3) 28 13 -27 34 23 00 1 123 -1 -00

Panel D Older than 12Treatment (λ1) 41˚˚ 22˚˚˚ 13˚ 22˚˚ 6 17˚ -0˚ -15 3˚ 12

(16) (07) (7) (11) (4) (09) (0) (20) (2) (22)Cost Sharing (λ2) -30˚˚ -08 -9 -09 1 06 0 -02 -4˚˚ -07

(15) (07) (6) (11) (4) (10) (0) (21) (2) (22)Saturation (λ3) 107 51 1 63 37 50 0 -236˚ -0 21

Panel E 12 or YoungerTreatment (λ1) -0 -07 3 -05 -2 -09 0˚ 66˚˚˚ -1 35˚˚

(10) (09) (7) (08) (4) (12) (0) (24) (1) (16)Cost Sharing (λ2) -1 -10 -3 -03 -1 -18 -0˚˚ -65˚˚ 1˚ 18

(10) (09) (7) (09) (3) (13) (0) (26) (1) (14)Saturation (λ3) -91 -71 -11 90˚˚ -12 -157˚ 1 193˚ -3 154

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 30 days from the interview date Results are trimmed at the top 1 of observations Results See Section 71 fornotes on outcome construction See Table 10 for notes on the regression specification denotes statistical significance at 10 pct at 5 pct and at 1 pct

(1)Individual

HourlyEarnings

(2)Log Individual

HourlyEarnings

(3)Hourly

IndividualWage Earnings

(4)Log Individual

WageEarnings

(5)Hourly

Self-EmploymentProfit

(6)Log Hourly

(7)Hourly

IndividualFarming Profits

(8)Log HourlyIndividual

Farming ProfitsPanel A Full SampleTreatment (λ1) 14˚ 09˚ 07 07 14˚ 12 05 23

(08) (05) (11) (07) (08) (10) (05) (18)Cost Sharing (λ2) -22˚˚˚ -08 -19˚ -10 -14˚ 02 -13˚˚˚ -42˚˚

(07) (06) (11) (08) (08) (10) (05) (20)Saturation (λ3) 06 -08 46 36 -28 -65 31 153

(36) (29) (49) (39) (50) (56) (29) (101)Control Mean 107 -74 126 -45 73 -126 25 -201Treatment Effect () 127 87 52 71 193 112 195 204Joint F-Test (p-value) 021 352 310 588 110 202 039 058Treatment FDR q-value 609 609 763 609 609 609 609 609Number Observations 6096 5887 3745 3584 2242 2168 305 282

Panel B FemalesTreatment (λ1) 01 11 02 19 07 17 02 13

(11) (12) (15) (13) (11) (16) (07) (20)Cost Sharing (λ2) -09 -07 -16 -19˚ -07 -06 -04 -23

(09) (11) (12) (11) (09) (17) (06) (24)Saturation (λ3) -32 -28 -14 83 -36 -85 21 56

Panel C MalesTreatment (λ1) 22 08 09 01 20 06 12 58

(15) (08) (18) (10) (15) (13) (10) (44)Cost Sharing (λ2) -30˚˚ -08 -21 -06 -21 09 -30˚˚˚ -81˚˚

(13) (09) (16) (10) (15) (13) (09) (37)Saturation (λ3) 30 02 74 13 -22 -51 41 387˚˚

Panel D Older than 12Treatment (λ1) 32˚ 24˚˚˚ 28 30˚˚ 25˚˚ 29˚˚ 05 02

(16) (07) (25) (12) (11) (11) (08) (31)Cost Sharing (λ2) -36˚˚ -14˚ -39˚ -24˚ -16 02 -11 -34

(14) (07) (21) (13) (10) (12) (08) (33)Saturation (λ3) 84 48 145 59 01 47 33 -01

Panel E 12 or YoungerTreatment (λ1) -05 -08 -12 -11 00 -14 04 55

(10) (08) (14) (09) (17) (14) (08) (33)Cost Sharing (λ2) -06 -01 -01 02 -14 -00 -15˚ -32

(09) (09) (12) (11) (16) (16) (09) (34)Saturation (λ3) -82˚ -79 -29 47 -69 -199˚ 16 333˚˚

(48) (51) (62) (46) (111) (111) (42) (164)Control Mean 113 -65 13 -35 84 -118 24 -206Treatment Effect () -47 -86 -93 -115 6 -156 166 437Joint F-Test (p-value) 0270 0409 0745 0211 0483 0308 0217 0043Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2646 2545 1760 1675 849 822 116 104

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus12 months prior) Results are trimmed at the top 1 of observations See Section 72 for notes on outcome construction Treatment is an indicator variable equal to 1 for PSDP Worm Groups 1 and2 which received an additional 24 years of deworming on average compared to Group 3 Reported estimates for Female and Male are constructed from a single regression including treatment-femalecost-sharing-female and saturation-female interaction terms Reported estimates for Older than 12 and 12 or Younger also report results using a single regression including an indicator for those olderthan 12 at baseline and analogous interaction terms to Panels B and C Covariates follow Baird et al (2016) and include controls for baseline 1998 primary school population geographic zone of theschool survey wave and month of interview a female indicator variable baseline 1998 school grade fixed effects the average school test score on the 1996 Busia District mock exams total primary schoolpupils within 6 km and a cost-sharing school indicator Those treated in a separate vocational training intervention (VocEd) which occurred prior to KLPS-3 are dropped from the KLPS-3 and KLPS-4sample Those treated in a separate small grant intervention (SCY) which occurred after KLPS-3 are dropped from the KLPS-4 sample Observations are weighted to be representative of the originalPSDP population and include KLPS population weights SCY and VocEd control group weights and KLPS intensive tracking weights Standard errors are clustered at the 1998 school level The JointF-Test (p-value) gives the p-value associated with an F-test on the joint significance of the treatment cost-sharing and saturation coefficients against the null hypothesis that all three coefficients arejointly equal to zero The FDR adjustment is carried out across the 18 outcomes within this family separately by panel (full sample females males older than 12 and 12 or younger) denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

(1)Individual

HourlyEarnings

(2)Log Individual

HourlyEarnings

(3)Hourly

(4)Log Individual

WageEarnings

(5)Hourly

(6)Log Hourly

(7)Hourly

Farming ProfitsPanel A Full SampleTreatment (λ1) -107 04 -176 15˚ -52 -03 22˚˚ 30

(115) (07) (182) (09) (75) (13) (11) (33)Cost Sharing (λ2) -29 -04 -04 -12 -53 11 -25˚˚˚ -08

(135) (08) (217) (09) (38) (17) (07) (26)Saturation (λ3) 37 -27 -210 70 467 -242˚˚ 24 253˚

(636) (51) (1065) (57) (636) (103) (35) (136)Control Mean 34 -81 421 -68 189 -123 25 -17Treatment Effect () -315 35 -420 142 -277 -28 906 261Joint F-Test (p-value) 769 833 777 307 474 136 007 245Treatment FDR q-value 1000 1000 1000 933 1000 1000 933 1000Number Observations 6170 3267 3790 2152 2272 1036 308 127

Panel B FemalesTreatment (λ1) 74 14 24 52˚˚˚ 44 -14 13 04

(77) (13) (105) (17) (47) (22) (09) (32)Cost Sharing (λ2) -19 -13 -20 -37˚˚ -41 01 -08 38

(59) (13) (88) (15) (45) (26) (07) (34)Saturation (λ3) 153 -09 -280 123 731 -188 58 219

(637) (62) (930) (75) (759) (113) (45) (190)Control Mean 79 -132 95 -124 61 -153 23 -179Treatment Effect () 943 131 247 418 715 -154 560 35Joint F-Test (p-value) 0818 0680 0969 0015 0780 0378 0284 0278Treatment FDR q-value 1000 1000 1000 071 1000 1000 1000 1000Number Observations 2370 1158 1193 596 1048 487 220 94

Panel C MalesTreatment (λ1) -225 -03 -278 -03 -127 07 37˚˚ 52

(214) (09) (293) (10) (154) (20) (18) (63)Cost Sharing (λ2) -27 02 06 -01 -58 17 -61˚˚˚ -114˚

(208) (09) (317) (09) (47) (20) (17) (64)Saturation (λ3) -60 -42 -216 29 294 -276 -92 113

(910) (69) (1416) (61) (570) (173) (88) (241)Control Mean 506 -52 59 -42 288 -99 29 -128Treatment Effect () -445 -35 -471 -31 -441 65 1279 419Joint F-Test (p-value) 0653 0942 0773 0895 0371 0270 0002 0209Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 3800 2109 2597 1556 1224 549 88 33

Panel D Older than 12Treatment (λ1) 11 12 -77 29˚˚ 65 10 26 91˚

(189) (09) (319) (12) (44) (17) (18) (47)Cost Sharing (λ2) -17 -07 13 -20 -71 05 -28˚˚ -55

(239) (11) (410) (12) (48) (23) (14) (44)Saturation (λ3) -780 05 -2004 75 535 -90 51 148

(1106) (57) (1915) (81) (659) (88) (49) (192)Control Mean 232 -78 315 -67 9 -113 25 -189Treatment Effect () 48 116 -244 257 727 97 1055 646Joint F-Test (p-value) 0793 0532 0529 0114 0493 0566 0186 0241Treatment FDR q-value 1000 435 924 118 368 831 368 214Number Observations 3454 1946 1992 1218 1394 682 189 82

Panel E 12 or YoungerTreatment (λ1) -239 -06 -288 00 -188 -20 19˚˚ -48

(201) (11) (300) (11) (173) (25) (09) (46)Cost Sharing (λ2) -35 -00 19 -05 -26 20 -23˚˚ 30

(71) (12) (119) (13) (42) (24) (11) (47)Saturation (λ3) 971 -62 1805 81 236 -430˚˚ -07 347˚˚

(857) (79) (1388) (62) (495) (176) (67) (151)Control Mean 464 -85 537 -68 308 -135 24 -141Treatment Effect () -515 -63 -538 0 -611 -225 786 -651Joint F-Test (p-value) 0597 0857 0530 0512 0419 0109 0051 0032Treatment FDR q-value 1000 1000 1000 1000 1000 1000 932 1000Number Observations 2682 1321 1780 934 864 354 117 45

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus twelvemonths prior) Results are untrimmed See Section 72 for notes on outcome construction See Table 16 for notes on the regression specification denotes statistical significance at 10 pct at 5 pctand at 1 pct

(1)Individual

HourlyEarnings

(2)Log Individual

HourlyEarnings

(3)Hourly

(4)Log Individual

WageEarnings

(5)Hourly

(6)Log Hourly

(7)Hourly

Farming ProfitsPanel A Full SampleTreatment (λ1) 26˚ 14˚ 14 -04 26˚˚ 24 08 42

(14) (08) (22) (12) (12) (16) (06) (28)Cost Sharing (λ2) -35˚˚˚ -12˚ -31 -11 -20˚ -04 -15˚˚˚ -98˚˚˚

(12) (07) (20) (11) (11) (16) (04) (21)Saturation (λ3) 30 -02 75 -01 51 46 09 -43

(61) (37) (94) (69) (58) (76) (35) (162)Control Mean 128 -6 166 -09 76 -124 2 -234Treatment Effect () 207 133 85 -36 345 216 425 349Joint F-Test (p-value) 026 270 402 477 134 462 011 000Treatment FDR q-value 365 365 563 677 365 379 389 379Number Observations 2718 2662 1501 1459 1168 1146 173 156

Panel B FemalesTreatment (λ1) -01 09 -04 -13 15 46˚˚ 04 24

(17) (15) (26) (19) (13) (21) (09) (31)Cost Sharing (λ2) -01 -07 -11 -09 01 -17 -12˚˚ -79˚˚˚

(15) (14) (21) (16) (12) (21) (06) (28)Saturation (λ3) -73 -71 -32 29 -04 01 06 -122

(87) (75) (150) (118) (99) (110) (41) (179)Control Mean 78 -112 116 -35 43 -184 17 -261Treatment Effect () -18 83 -34 -142 346 376 217 212Joint F-Test (p-value) 0613 0375 0781 0598 0570 0170 0219 0053Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 1139 1102 536 508 538 527 124 113

Panel C MalesTreatment (λ1) 45˚ 18 24 02 35 07 21 108˚

(26) (13) (35) (15) (22) (18) (13) (56)Cost Sharing (λ2) -58˚˚ -16 -43 -11 -38˚ 06 -23˚˚ -147˚˚˚

(23) (12) (30) (15) (21) (16) (10) (38)Saturation (λ3) 84 36 117 -15 90 75 20 239

(107) (62) (140) (80) (81) (79) (74) (317)Control Mean 163 -24 197 07 102 -76 26 -183Treatment Effect () 273 164 121 16 345 64 799 731Joint F-Test (p-value) 0040 0503 0509 0773 0227 0758 0138 0000Treatment FDR q-value 723 723 816 1000 723 1000 723 723Number Observations 1579 1560 965 951 630 619 49 43

Panel D Older than 12Treatment (λ1) 51 38˚˚˚ 52 29 32˚ 45˚˚ 07 05

(31) (13) (53) (22) (18) (22) (11) (43)Cost Sharing (λ2) -57˚˚ -22˚˚ -72 -32 -16 01 -19˚˚ -80˚

(25) (09) (44) (23) (17) (19) (09) (41)Saturation (λ3) 116 65 232 08 17 139 -23 -362˚

Panel E 12 or YoungerTreatment (λ1) 04 -10 -13 -24 22 -03 10 102˚˚

(19) (12) (26) (15) (22) (16) (09) (42)Cost Sharing (λ2) -16 -04 -01 09 -31 -19 -09 -90˚˚

(16) (10) (23) (14) (20) (22) (06) (41)Saturation (λ3) -74 -93˚ -48 36 107 -38 29 318

(90) (51) (116) (83) (108) (100) (51) (221)Control Mean 139 -4 173 08 9 -103 1 -272Treatment Effect () 32 -105 -74 -275 244 -31 975 702Joint F-Test (p-value) 0542 0231 0932 0256 0335 0783 0408 0011Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 517Number Observations 1270 1242 775 751 483 474 67 59

Notes Analysis uses KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus twelve months prior) Resultsare trimmed at the top 1 of observations See Section 72 for notes on outcome construction See Table 16 for notes on the regression specification denotes statistical significance at 10 pct at 5pct and at 1 pct

(1)Individual

HourlyEarnings

(2)Log Individual

HourlyEarnings

(3)Hourly

(4)Log Individual

WageEarnings

(5)Hourly

(6)Log Hourly

(7)Hourly

Farming ProfitsPanel A Full SampleTreatment (λ1) -01 03 -03 14 02 03 04 00

(07) (09) (08) (11) (15) (15) (09) (27)Cost Sharing (λ2) -09 -02 -13˚ -15 -01 10 -05 07

(06) (07) (07) (10) (16) (17) (08) (26)Saturation (λ3) -35 -72 47 55 -149 -253˚ 27 212˚

(42) (57) (48) (63) (139) (135) (47) (123)Control Mean 1 -77 112 -57 77 -129 29 -17Treatment Effect () -6 30 -23 132 31 31 123 3Joint F-Test (p-value) 139 451 085 495 578 208 867 388Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2316 2211 1484 1406 768 724 132 126

Panel B FemalesTreatment (λ1) -05 09 01 46˚˚ -04 -14 -02 -13

(12) (14) (15) (21) (22) (23) (11) (30)Cost Sharing (λ2) -17˚˚ -09 -27˚˚ -39˚˚ -14 02 13 41

(08) (12) (12) (18) (16) (22) (13) (33)Saturation (λ3) -21 -36 13 99 -31 -117 -08 101

(54) (68) (69) (88) (116) (154) (71) (163)Control Mean 75 -127 78 -114 75 -153 28 -179Treatment Effect () -65 84 7 376 -51 -149 -58 -138Joint F-Test (p-value) 0085 0688 0052 0101 0681 0818 0656 0449Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 899 850 446 407 387 367 96 94

Panel C MalesTreatment (λ1) 02 -02 -04 -03 09 23 01 04

(09) (11) (10) (12) (24) (23) (20) (53)Cost Sharing (λ2) -05 03 -08 -03 06 08 -37˚˚ -62

(08) (09) (09) (11) (27) (25) (15) (49)Saturation (λ3) -41 -95 53 16 -208 -331 11 267

(69) (83) (57) (67) (205) (219) (79) (210)Control Mean 117 -46 128 -3 78 -108 34 -128Treatment Effect () 18 -16 -30 -29 118 205 29 35Joint F-Test (p-value) 0627 0663 0439 0940 0509 0125 0095 0204Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 1417 1361 1038 999 381 357 36 32

Panel D Older than 12Treatment (λ1) 22˚˚˚ 16˚ 12 32˚˚ 31˚ 25 21 75˚

(08) (09) (12) (15) (18) (18) (15) (39)Cost Sharing (λ2) -26˚˚˚ -09 -26˚˚ -24˚ -18 -05 -07 -46

(09) (10) (12) (14) (17) (22) (17) (42)Saturation (λ3) 49 -22 84 76 07 -88 77 200

Panel E 12 or YoungerTreatment (λ1) -21˚˚ -09 -16˚ -02 -30 -20 -23˚ -95˚˚

(11) (13) (09) (13) (32) (29) (11) (44)Cost Sharing (λ2) 08 06 -02 -08 24 29 -18 38

(09) (13) (09) (15) (28) (25) (21) (46)Saturation (λ3) -111 -121 23 43 -293 -399˚ -39 238

(76) (108) (58) (88) (227) (202) (56) (150)Control Mean 102 -78 111 -54 85 -14 39 -141Treatment Effect () -208 -96 -142 -23 -346 -227 -583 -3074Joint F-Test (p-value) 0143 0728 0253 0852 0460 0267 0093 0012Treatment FDR q-value 474 1000 575 1000 1000 1000 474 474Number Observations 1039 984 714 671 301 284 49 45

(1)Individual

HourlyEarnings

(2)Log Individual

HourlyEarnings

(3)Hourly

(4)Log Individual

WageEarnings

(5)Hourly

(6)Log Hourly

(7)Hourly

Farming ProfitsPanel A Full SampleTreatment (λ1) 47˚˚ 06 04 02 17 13 19 15

(19) (05) (08) (05) (11) (09) (14) (16)Cost Sharing (λ2) -72˚˚˚ -09˚ -10 -09˚ -18˚ -03 -40˚˚˚ -40˚˚

(20) (05) (08) (05) (10) (09) (12) (19)Saturation (λ3) 23 -19 18 13 -17 -41 -14 -11

(89) (24) (39) (28) (48) (48) (70) (87)Control Mean 296 1 133 -23 89 -11 74 -103Treatment Effect () 160 54 26 19 186 123 259 138Joint F-Test (p-value) 004 089 648 287 287 246 017 191Treatment FDR q-value 361 424 543 554 361 361 361 424Number Observations 8082 7759 5276 5008 2827 2727 449 415

Panel B FemalesTreatment (λ1) 11 03 -08 00 -01 13 16 13

(26) (10) (11) (10) (13) (12) (20) (20)Cost Sharing (λ2) -36 -09 -09 -08 -05 -10 -29˚ -17

(25) (10) (10) (10) (10) (14) (16) (24)Saturation (λ3) -32 -58 14 26 -86 -84 -14 -90

(137) (50) (72) (52) (54) (61) (90) (116)Control Mean 201 -28 103 -51 64 -151 57 -132Treatment Effect () 57 31 -81 4 -15 124 277 123Joint F-Test (p-value) 0350 0410 0431 0822 0364 0330 0325 0693Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 3219 3054 1802 1662 1314 1270 312 294

Panel C MalesTreatment (λ1) 72˚˚ 08 10 03 31 12 25 20

(31) (06) (12) (07) (19) (12) (30) (33)Cost Sharing (λ2) -96˚˚˚ -09˚ -11 -09 -28 03 -59˚˚ -74˚˚

(28) (06) (10) (06) (18) (11) (26) (31)Saturation (λ3) 63 04 23 08 32 -08 -33 138

Panel D Older than 12Treatment (λ1) 77˚˚ 16˚˚ 15 14˚ 22 24˚ 17 -00

(30) (07) (13) (08) (16) (13) (23) (28)Cost Sharing (λ2) -95˚˚˚ -15˚˚ -25˚˚ -17˚˚ -30˚˚ -12 -31 -35

(26) (06) (11) (08) (14) (10) (21) (29)Saturation (λ3) 218 39 86 44 62 64 -58 -152

(149) (32) (60) (40) (67) (62) (102) (116)Control Mean 264 -01 133 -29 9 -111 69 -108Treatment Effect () 294 150 113 132 247 219 248 -1Joint F-Test (p-value) 0006 0079 0122 0205 0162 0231 0256 0092Treatment FDR q-value 052 076 202 126 202 126 260 442Number Observations 4422 4268 2712 2597 1719 1658 269 249

Panel E 12 or YoungerTreatment (λ1) 16 -06 -04 -07 10 -06 20 34

(36) (08) (13) (08) (19) (12) (28) (30)Cost Sharing (λ2) -45 -03 04 -00 -04 05 -49˚˚ -35

(33) (07) (13) (08) (16) (14) (23) (28)Saturation (λ3) -204 -88˚˚ -21 06 -113 -174˚˚ 03 131

(155) (39) (57) (38) (93) (86) (117) (119)Control Mean 324 21 132 -19 9 -103 79 -96Treatment Effect () 48 -60 -33 -68 114 -60 253 289Joint F-Test (p-value) 0339 0096 0972 0719 0458 0234 0142 0231Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 3626 3457 2546 2393 1094 1055 178 164

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12months prior) Results are trimmed at the top 1 of observations Analysis includes KLPS respondents who participated in SCY or VocEd with indicators for receiving a SCY grant or a vocationaltraining voucher See Section 72 for notes on outcome construction See Table 16 for notes on the regression specification Observations are weighted to be representative of the original KLPS populationand include KLPS population weights and KLPS intensive tracking weights denotes statistical significance at 10 pct at 5 pct and at 1 pct

(1)Individual

HourlyEarnings

(2)Log Individual

HourlyEarnings

(3)Hourly

(4)Log Individual

WageEarnings

(5)Hourly

(6)Log Hourly

(7)Hourly

Farming ProfitsPanel A Full SampleTreatment (λ1) 13 06 08 06 12 06 05 01

(08) (05) (10) (07) (10) (08) (07) (28)Cost Sharing (λ2) -21˚˚˚ -07 -18˚ -09 -19˚˚ -02 -06 -10

(07) (05) (10) (07) (08) (09) (06) (24)Saturation (λ3) 04 -15 44 29 -47 -82˚ 50 325˚˚

Panel B FemalesTreatment (λ1) 04 05 00 09 12 03 02 -01

(11) (10) (14) (11) (11) (12) (09) (28)Cost Sharing (λ2) -06 -03 -14 -14 -05 01 -01 02

(09) (10) (11) (09) (09) (14) (08) (29)Saturation (λ3) -40 -42 -10 55 -14 -100 42 285˚

(15) (08) (16) (09) (19) (12) (12) (55)Cost Sharing (λ2) -31˚˚ -10 -20 -07 -30˚ -04 -16 -41

(13) (08) (14) (09) (15) (11) (12) (49)Saturation (λ3) 31 -00 69 17 -68 -69 65 401˚

Panel D Older than 12Treatment (λ1) 29˚ 18˚˚˚ 28 25˚˚ 17 16˚ 09 29

(16) (07) (22) (11) (13) (09) (11) (34)Cost Sharing (λ2) -36˚˚˚ -13˚˚ -35˚ -18˚ -17 -01 -03 -26

(13) (06) (19) (10) (12) (10) (10) (32)Saturation (λ3) 93 28 127 37 37 26 109˚ 388˚

Panel E 12 or YoungerTreatment (λ1) -02 -08 -09 -11 06 -13 -03 -34

(12) (09) (12) (08) (20) (13) (10) (38)Cost Sharing (λ2) -06 -01 -02 00 -21 -03 -12 08

(09) (08) (11) (09) (16) (14) (08) (36)Saturation (λ3) -88˚ -73 -12 51 -163 -234˚˚ -18 295˚

(50) (58) (62) (41) (99) (101) (48) (163)Control Mean 116 -43 134 -11 99 -82 2 -149Treatment Effect () -18 -83 -64 -112 58 -134 -128 -414Joint F-Test (p-value) 0300 0624 0828 0135 0116 0138 0441 0129Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2453 2325 1758 1663 850 822 116 67

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Each observation is for an individual in the last 30 days from the interview date Results are trimmed at the top 1 of observations ResultsSee Section 72 for notes on outcome construction See Table 16 for notes on the regression specification denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 22 Household Earnings - Pooled (trimmed)

(1)Per-Capita

Household Earnings

(2)Log Per-Capita

Household Earnings

(3)Household

Tax Spending

(4)Log HouseholdTax Spending

Panel A Full SampleTreatment (λ1) 239˚ 12 15 20

(129) (11) (26) (15)Cost Sharing (λ2) -157 -04 -28 -16

(120) (08) (22) (13)Saturation (λ3) -1011˚ -92˚˚ -205˚ -65

(604) (44) (114) (87)Control Mean 1296 62 79 521Treatment Effect () 184 110 184 180Joint F-Test (p-value) 018 075 052 199Treatment FDR q-value 380 380 569 380Number Observations 4074 3946 4803 1064

Panel B FemalesTreatment (λ1) 36 -01 -26 -06

(107) (12) (22) (25)Cost Sharing (λ2) 63 07 30 00

(105) (12) (25) (21)Saturation (λ3) -1691˚˚˚ -174˚˚ -331˚˚ -142

(621) (82) (143) (140)Control Mean 973 599 40 485Treatment Effect () 37 -8 -656 -57Joint F-Test (p-value) 0013 0067 0146 0761Treatment FDR q-value 1000 1000 1000 1000Number Observations 2099 2003 2478 390

Panel C MalesTreatment (λ1) 439˚ 24 55 30˚

(252) (16) (55) (18)Cost Sharing (λ2) -375 -16 -84 -23

(229) (13) (51) (18)Saturation (λ3) -484 -28 -115 -50

Panel D Older than 12Treatment (λ1) 565˚˚ 31˚˚ 92˚ 59˚˚

(232) (15) (53) (23)Cost Sharing (λ2) -424˚ -26˚˚ -109˚˚ -43˚

(219) (12) (50) (23)Saturation (λ3) 865 -18 -108 -05

Panel E 12 or YoungerTreatment (λ1) -22 -04 -43 -02

(171) (13) (29) (22)Cost Sharing (λ2) 15 14 41 10

(188) (12) (25) (19)Saturation (λ3) -2967˚˚ -174˚˚ -255 -52

(1138) (69) (167) (160)Control Mean 1501 637 88 524Treatment Effect () -15 -37 -496 -24Joint F-Test (p-value) 0049 0085 0305 0943Treatment FDR q-value 1000 1000 1000 1000Number Observations 1982 1909 2356 506

Notes Columns (1) and (2) use KLPS-4 data and columns (3) and (4) use KLPS-3 and KLPS-4 data Each observation isfor an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12 monthsprior) Results are trimmed at the top 1 of observations See Section 8 for notes on outcome construction Treatment isan indicator variable equal to 1 for PSDP Worm Groups 1 and 2 which received an additional 24 years of deworming onaverage compared to Group 3 Reported estimates for Female and Male are constructed from a single regression includingtreatment-female cost-sharing-female and saturation-female interaction terms Reported estimates for Older than 12 and 12or Younger also report results using a single regression including an indicator for those older than 12 at baseline and analogousinteraction terms to Panels B and C Covariates follow Baird et al (2016) and include controls for baseline 1998 primary schoolpopulation geographic zone of the school survey wave and month of interview a female indicator variable baseline 1998 schoolgrade fixed effects the average school test score on the 1996 Busia District mock exams total primary school pupils within 6km and a cost-sharing school indicator Those treated in a separate vocational training intervention (VocEd) which occurredprior to KLPS-3 are dropped from the KLPS-3 and KLPS-4 sample Those treated in a separate small grant intervention(SCY) which occurred after KLPS-3 are dropped from the KLPS-4 sample Observations are weighted to be representativeof the original PSDP population and include KLPS population weights SCY and VocEd control group weights and KLPSintensive tracking weights Standard errors are clustered at the 1998 school level The Joint F-Test (p-value) gives the p-valueassociated with an F-test on the joint significance of the treatment cost-sharing and saturation coefficients against the nullhypothesis that all three coefficients are jointly equal to zero The FDR adjustment is carried out across the four outcomeswithin this family separately by panel (full sample females males older than 12 and 12 or younger) denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Table 23 Household Earnings - Pooled (untrimmed)

(1)Per-Capita

Household Earnings

(2)Log Per-Capita

Household Earnings

(3)Household

Tax Spending

(129) (11) (30) (16)Cost Sharing (λ2) -157 -04 -27 -14

(120) (08) (25) (14)Saturation (λ3) -1011˚ -92˚˚ -241˚ -92

(107) (12) (26) (26)Cost Sharing (λ2) 63 07 37 12

(105) (12) (30) (23)Saturation (λ3) -1691˚˚˚ -174˚˚ -385˚˚ -202

(252) (16) (64) (19)Cost Sharing (λ2) -375 -16 -89 -27

(229) (13) (57) (19)Saturation (λ3) -484 -28 -131 -74

Panel D Older than 12Treatment (λ1) 565˚˚ 31˚˚ 97 56˚˚

(232) (15) (61) (25)Cost Sharing (λ2) -424˚ -26˚˚ -120˚˚ -42˚

(219) (12) (56) (25)Saturation (λ3) 865 -18 -129 -26

Panel E 12 or YoungerTreatment (λ1) -22 -04 -45 00

(171) (13) (33) (23)Cost Sharing (λ2) 15 14 54˚ 10

(188) (12) (30) (19)Saturation (λ3) -2967˚˚ -174˚˚ -320 -74

(1138) (69) (205) (170)Control Mean 1501 637 161 541Treatment Effect () -15 -37 -279 1Joint F-Test (p-value) 0049 0085 0293 0904Treatment FDR q-value 1000 1000 1000 1000Number Observations 1982 1909 1989 474

Notes Columns (1) and (2) use KLPS-4 data and columns (3) and (4) use KLPS-3 and KLPS-4 data Each observation isfor an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12 monthsprior) Results are untrimmed See Section 8 for notes on outcome construction See Table 22 for notes on the regressionspecification denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 24 Household Earnings - KLPS-4 (trimmed)

(1)Per-Capita

Household Earnings

(2)Log Per-Capita

Household Earnings

(3)Household

Tax Spending

(129) (11) (30) (16)Cost Sharing (λ2) -157 -04 -27 -14

(120) (08) (25) (14)Saturation (λ3) -1011˚ -92˚˚ -241˚ -92

(107) (12) (26) (26)Cost Sharing (λ2) 63 07 37 12

(105) (12) (30) (23)Saturation (λ3) -1691˚˚˚ -174˚˚ -385˚˚ -202

(252) (16) (64) (19)Cost Sharing (λ2) -375 -16 -89 -27

(229) (13) (57) (19)Saturation (λ3) -484 -28 -131 -74

(232) (15) (61) (25)Cost Sharing (λ2) -424˚ -26˚˚ -120˚˚ -42˚

(219) (12) (56) (25)Saturation (λ3) 865 -18 -129 -26

(171) (13) (33) (23)Cost Sharing (λ2) 15 14 54˚ 10

(188) (12) (30) (19)Saturation (λ3) -2967˚˚ -174˚˚ -320 -74

Notes Analysis uses KLPS-4 data Each observation is for an individual in the last 12 months from the interview date(ie the month in which they were interviewed plus 12 months prior) Results are trimmed at the top 1 of observationsSee Section 8 for notes on outcome construction See Table 22 for notes on the regression specification denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Table 25 Household Earnings - Pooled (trimmed including SCY and VocEd)

(1)Per-Capita

Household Earnings

(2)Log Per-Capita

Household Earnings

(3)Household

Tax Spending

Panel A Full SampleTreatment (λ1) 257˚˚ 08 -5 05

(115) (09) (20) (12)Cost Sharing (λ2) -48 -02 -16 -12

(112) (07) (15) (10)Saturation (λ3) 350 -49 -243˚˚ -49

(662) (34) (118) (76)Control Mean 1295 622 79 532Treatment Effect () 199 75 -57 51Joint F-Test (p-value) 124 246 064 438Treatment FDR q-value 132 1000 1000 1000Number Observations 4936 4778 5755 1242

Panel B FemalesTreatment (λ1) 25 -02 -17 01

(102) (10) (20) (24)Cost Sharing (λ2) 74 09 17 -04

(102) (10) (21) (17)Saturation (λ3) -1268˚˚ -128 -257˚˚ -151

(635) (79) (114) (115)Control Mean 969 598 39 485Treatment Effect () 26 -24 -429 5Joint F-Test (p-value) 0066 0062 0134 0505Treatment FDR q-value 1000 1000 1000 1000Number Observations 2511 2395 2927 456

Panel C MalesTreatment (λ1) 463˚˚ 16 7 07

(207) (13) (34) (15)Cost Sharing (λ2) -165 -12 -46 -15

(188) (10) (30) (15)Saturation (λ3) 1737 21 -237 -18

Panel D Older than 12Treatment (λ1) 608˚˚˚ 25˚˚ 43 30

(198) (12) (29) (22)Cost Sharing (λ2) -230 -19˚ -72˚˚˚ -32

(192) (10) (25) (20)Saturation (λ3) 3184˚˚ 57 -137 -23

(182) (12) (29) (22)Cost Sharing (λ2) 13 11 29 05

(186) (10) (27) (20)Saturation (λ3) -2613˚˚ -160˚˚˚ -303˚ 27

Notes Columns (1) and (2) use KLPS-4 data and columns (3) and (4) use KLPS-3 and KLPS-4 data Each observationis for an individual in the last 12 months from the interview date (ie the month in which they were interviewed plus 12months prior) Results are trimmed at the top 1 of observations Analysis includes KLPS respondents who participated inSCY or VocEd with indicators for receiving a SCY grant or a vocational training voucher See Section 8 for notes on outcomeconstruction See Table 22 for notes on the regression specification Observations are weighted to be representative of theoriginal KLPS population and include KLPS population weights and KLPS intensive tracking weights denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Table 26 Household Earnings - Pooled (trimmed last month)

(1)Per-Capita

Household Earnings

(2)Log Per-Capita

Household Earnings

(3)Household

Tax Spending

Panel A Full SampleTreatment (λ1) 21 15 1 15

(13) (11) (1) (14)Cost Sharing (λ2) -18 -12 -2 -11

(12) (09) (1) (13)Saturation (λ3) -64 -00 -6 -65

Panel B FemalesTreatment (λ1) 1 12 -1 -10

(9) (12) (1) (24)Cost Sharing (λ2) -1 -01 1 09

(9) (11) (1) (19)Saturation (λ3) -131˚˚ -03 -11 -93

(59) (62) (7) (124)Control Mean 87 369 4 261Treatment Effect () 12 114 -217 -108Joint F-Test (p-value) 0139 0622 0466 0887Treatment FDR q-value 1000 1000 1000 1000Number Observations 2103 1899 4360 396

Panel C MalesTreatment (λ1) 40 19 3 25

(27) (16) (3) (15)Cost Sharing (λ2) -34 -22 -4˚ -21

(24) (13) (2) (17)Saturation (λ3) -11 01 -2 -66

(25) (17) (3) (21)Cost Sharing (λ2) -39˚ -22 -6˚˚ -38˚

(23) (15) (2) (20)Saturation (λ3) 59 85 -3 -35

(13) (12) (1) (21)Cost Sharing (λ2) -3 -07 2˚ 11

(15) (12) (1) (17)Saturation (λ3) -202˚˚ -100 -7 -34

Notes Columns (1) and (2) use KLPS-4 data and columns (3) and (4) use KLPS-3 and KLPS-4 data Each observationis for an individual in the last 30 days from the interview date Results are trimmed at the top 1 of observations ResultsSee Section 8 for notes on outcome construction See Table 22 for notes on the regression specification denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Table 27 Labor Supply - Pooled

TotalHours Worked

Log TotalHours Worked

Non-ZeroHours

Farm HoursWorked

Log FarmHours Worked

Wage HoursWorked

Log WageHours Worked

Self-EmploymentHours Worked

(9)Log

(10)Non-Zero Wage

or Self-EmploymentHours

Panel A Full SampleTreatment (λ1) 1 08˚˚˚ 00 -1˚˚ -05 1˚˚ 03 -0 -10 04˚˚

(1) (03) (01) (0) (08) (1) (04) (1) (07) (01)Cost Sharing (λ2) -0 -01 -00 0 01 -1 01 1 09 -01

(1) (03) (01) (0) (07) (1) (04) (1) (07) (01)Saturation (λ3) 2 10 -03 -3 -49 3 -02 3 23 07

(4) (14) (07) (2) (38) (5) (28) (4) (50) (07)Control Mean 24 328 64 4 2 14 327 7 329 52Treatment Effect () 43 73 6 -218 -20 103 26 -41 -105 70Joint F-Test (p-value) 471 031 910 243 646 186 825 545 386 070Treatment FDR q-value 165 060 434 100 434 100 434 434 222 060Number Observations 13807 8477 13807 13807 2768 13807 4872 13807 2598 13807

Panel B FemalesTreatment (λ1) -0 06 -02 -1˚˚ -06 0 -00 1 07 02

(1) (05) (02) (0) (10) (1) (09) (1) (08) (02)Cost Sharing (λ2) 1 06 -00 1 04 0 07 0 -02 -01

(1) (05) (02) (0) (10) (1) (08) (1) (10) (02)Saturation (λ3) 1 27 -05 -4 -48 1 33 2 86˚ 06

(6) (24) (10) (3) (64) (5) (35) (5) (49) (10)Control Mean 20 314 58 4 2 10 32 6 31 42Treatment Effect () -7 57 -43 -277 -22 19 -0 158 65 59Joint F-Test (p-value) 0913 0117 0409 0117 0890 0975 0509 0399 0382 0545Treatment FDR q-value 995 682 682 248 820 995 995 682 820 682Number Observations 6860 3672 6860 6860 1558 6860 1671 6860 1213 6860

Panel C MalesTreatment (λ1) 2˚˚ 09˚˚ 03˚ -1 -04 3˚˚ 04 -1 -24˚˚ 05˚˚

(1) (04) (02) (1) (11) (1) (06) (1) (11) (02)Cost Sharing (λ2) -1 -05 -01 -0 -03 -2˚ -02 1 19˚ -02

(1) (05) (02) (0) (10) (1) (07) (1) (11) (02)Saturation (λ3) 2 -01 -00 -1 -52 5 -17 3 -21 08

Panel D Older than 12Treatment (λ1) 2˚ 07 01 -0 -10 2˚ 04 1 -10 04˚

(1) (04) (01) (1) (09) (1) (05) (1) (09) (02)Cost Sharing (λ2) -1 01 -03˚ 0 11 -2˚˚ -00 1 24˚˚˚ -03

(1) (04) (02) (0) (09) (1) (05) (1) (08) (02)Saturation (λ3) 0 -14 -00 -2 -67 5 -29 -3 -60 04

(5) (19) (08) (3) (51) (5) (25) (4) (47) (08)Control Mean 26 332 69 4 3 15 33 8 323 57Treatment Effect () 68 64 17 -112 -40 108 41 89 -106 65Joint F-Test (p-value) 0216 0124 0226 0838 0475 0077 0370 0119 0027 0261Treatment FDR q-value 425 425 434 434 434 425 434 434 434 425Number Observations 6898 4578 6898 6898 1476 6898 2507 6898 1574 6898

Panel E 12 or YoungerTreatment (λ1) 1 09˚ 00 -1˚ -00 2 02 -1 -09 04˚

(1) (05) (02) (1) (11) (1) (06) (1) (12) (02)Cost Sharing (λ2) -0 -02 01 0 -09 -1 01 0 -08 -01

(1) (04) (02) (0) (11) (1) (07) (1) (12) (02)Saturation (λ3) 4 34 -02 -4 -35 3 26 8 126 13

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data See Section 9 for notes on outcome construction Treatment is an indicator variable equal to 1 for PSDP Worm Groups 1 and 2 which received an additional 24 years ofdeworming on average compared to Group 3 Reported estimates for Female and Male are constructed from a single regression including treatment-female cost-sharing-female and saturation-female interaction terms Reported estimatesfor Older than 12 and 12 or Younger also report results using a single regression including an indicator for those older than 12 at baseline and analogous interaction terms to Panels B and C Covariates follow Baird et al (2016) andinclude controls for baseline 1998 primary school population geographic zone of the school survey wave and month of interview a female indicator variable baseline 1998 school grade fixed effects the average school test score on the 1996Busia District mock exams total primary school pupils within 6 km and a cost-sharing school indicator Those treated in a separate vocational training intervention (VocEd) which occurred prior to KLPS-3 are dropped from the KLPS-3and KLPS-4 sample Those treated in a separate small grant intervention (SCY) which occurred after KLPS-3 are dropped from the KLPS-4 sample Observations are weighted to be representative of the original PSDP population andinclude KLPS population weights SCY and VocEd control group weights and KLPS intensive tracking weights Standard errors are clustered at the 1998 school level The Joint F-Test (p-value) gives the p-value associated with an F-teston the joint significance of the treatment cost-sharing and saturation coefficients against the null hypothesis that all three coefficients are jointly equal to zero The FDR adjustment is carried out across the ten outcomes within thisfamily separately by panel (full sample females males older than 12 and 12 or younger) denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 28 Labor Supply - KLPS-4

TotalHours Worked

Non-ZeroHours

Farm HoursWorked

Wage HoursWorked

(9)Log

(10)Non-Zero Wage

Panel A Full SampleTreatment (λ1) -0 06 -02 -2˚˚ -08 2 -05 -2 -15˚ 03

(1) (04) (02) (1) (10) (1) (06) (1) (08) (02)Cost Sharing (λ2) 1 -01 03 1 04 -1 01 2˚ 19˚˚ 01

(1) (05) (02) (1) (09) (1) (07) (1) (08) (02)Saturation (λ3) -13 -60˚˚ -03 -5 -80 -6 -80˚˚ 1 -11 00

(9) (26) (09) (4) (51) (10) (34) (8) (57) (13)Control Mean 38 352 86 8 2 20 376 14 349 67Treatment Effect () -6 59 -28 -263 -31 101 -56 -156 -164 46Joint F-Test (p-value) 380 016 382 146 448 271 134 145 117 296Treatment FDR q-value 349 285 285 285 317 285 317 285 285 304Number Observations 4135 3518 4135 4135 1589 4135 1618 4135 1343 4135

Panel B FemalesTreatment (λ1) -2 02 -06˚˚ -2 -02 -0 -14 -0 07 00

(2) (07) (03) (1) (13) (2) (10) (2) (11) (03)Cost Sharing (λ2) 5˚˚ 11 06˚ 2 05 2 08 2 04 05˚

(2) (07) (03) (1) (14) (2) (10) (2) (13) (03)Saturation (λ3) -10 -27 -08 -6 -61 -11 -30 1 59 -11

(13) (40) (17) (7) (84) (12) (68) (11) (73) (18)Control Mean 33 34 8 8 2 14 384 11 339 54Treatment Effect () -70 24 -77 -239 -7 -2 -147 -22 69 3Joint F-Test (p-value) 0042 0202 0127 0375 0837 0633 0566 0665 0644 0164Treatment FDR q-value 889 1000 469 889 1000 1000 889 1000 1000 1000Number Observations 2112 1653 2112 2112 917 2112 580 2112 618 2112

Panel C MalesTreatment (λ1) 2 09 01 -2 -15 4˚ -01 -4˚˚ -31˚˚ 06˚

(2) (05) (02) (2) (17) (2) (08) (2) (12) (03)Cost Sharing (λ2) -2 -10˚ 00 -0 03 -4˚˚ -03 2 30˚˚ -03

(2) (05) (02) (1) (16) (2) (07) (2) (11) (03)Saturation (λ3) -17 -85˚˚˚ 01 -4 -105 -2 -103˚˚˚ 1 -52 09

(11) (30) (10) (5) (75) (14) (38) (10) (67) (14)Control Mean 44 362 92 8 3 25 372 17 356 8Treatment Effect () 41 85 14 -285 -61 158 -14 -239 -369 73Joint F-Test (p-value) 0107 0006 0902 0363 0485 0059 0040 0205 0055 0347Treatment FDR q-value 273 228 331 228 317 201 522 201 190 220Number Observations 2023 1865 2023 2023 672 2023 1038 2023 725 2023

Panel D Older than 12Treatment (λ1) 2 06 -00 -1 -10 4˚˚ 06 -1 -22˚ 03

(2) (05) (03) (1) (13) (2) (08) (2) (13) (03)Cost Sharing (λ2) -1 -03 -00 1 21˚ -5˚˚ -14˚ 4˚˚ 38˚˚˚ -02

(2) (05) (03) (1) (11) (2) (08) (2) (12) (03)Saturation (λ3) -12 -86˚˚˚ 10 -2 -86 5 -106˚˚ -14 -65 -02

(13) (32) (15) (6) (68) (13) (44) (10) (66) (19)Control Mean 40 355 89 8 3 19 372 14 343 7Treatment Effect () 39 61 -6 -95 -39 219 60 -103 -251 42Joint F-Test (p-value) 0386 0004 0875 0779 0281 0069 0021 0143 0019 0688Treatment FDR q-value 800 800 1000 859 800 673 800 800 673 800Number Observations 2071 1812 2071 2071 860 2071 762 2071 756 2071

Panel E 12 or YoungerTreatment (λ1) -1 07 -03 -3˚˚ -08 1 -12 -3 -08 04

(2) (08) (03) (1) (16) (2) (08) (2) (13) (04)Cost Sharing (λ2) 2 00 04˚ 1 -12 1 10 1 02 02

(2) (06) (03) (1) (15) (2) (08) (1) (13) (04)Saturation (λ3) -13 -36 -10 -9˚ -93 -14 -43 15˚ 49 10

Notes Analysis uses KLPS-4 data See Section 9 for notes on outcome construction See Table 27 for notes on the regression specification denotes statistical significance at 10 pct at 5 pct and at 1 pct

TotalHours Worked

Non-ZeroHours

Farm HoursWorked

Wage HoursWorked

(9)Log

(10)Non-Zero Wage

Panel A Full SampleTreatment (λ1) 2˚ 07 -00 -0 05 2˚ 04 0 01 04

(1) (04) (02) (1) (08) (1) (07) (1) (10) (03)Cost Sharing (λ2) -2˚˚ -02 -03 -0 -07 -2˚˚ -04 0 03 -04˚

(1) (05) (02) (0) (07) (1) (06) (1) (11) (02)Saturation (λ3) 4 33 -23˚ -2 06 3 -22 3 21 -02

(4) (29) (13) (3) (44) (6) (45) (5) (93) (13)Control Mean 25 312 75 4 2 15 305 6 297 6Treatment Effect () 63 70 -5 -84 21 121 39 11 13 60Joint F-Test (p-value) 087 362 054 764 803 036 769 814 979 274Treatment FDR q-value 540 540 1000 915 915 540 915 1000 1000 540Number Observations 4595 3472 4595 4595 1179 4595 2089 4595 902 4595

Panel B FemalesTreatment (λ1) 0 08 -05 -1˚˚ -04 1 02 1 22 03

(2) (08) (03) (1) (09) (2) (13) (1) (14) (04)Cost Sharing (λ2) -2 -02 -03 0 -03 -2 -04 -0 -19 -04

(2) (08) (03) (0) (08) (2) (15) (1) (16) (04)Saturation (λ3) -2 18 -25 -5 -11 -1 -10 4 115 04

(8) (45) (19) (4) (55) (7) (79) (6) (114) (19)Control Mean 19 29 71 5 2 10 283 4 262 5Treatment Effect () 25 73 -65 -314 -18 52 16 324 199 60Joint F-Test (p-value) 0709 0795 0026 0077 0893 0835 0994 0478 0394 0748Treatment FDR q-value 1000 673 441 125 1000 1000 1000 441 441 702Number Observations 2260 1543 2260 2260 641 2260 750 2260 455 2260

Panel C MalesTreatment (λ1) 3 07 04 1 16 3˚ 05 -1 -21 04

(2) (09) (03) (1) (12) (2) (12) (1) (15) (04)Cost Sharing (λ2) -2 -02 -03 -1 -11 -3˚ -04 1 26˚ -04

(2) (07) (02) (1) (09) (2) (10) (1) (15) (04)Saturation (λ3) 8 43 -21 0 27 7 -27 1 -37 -08

(6) (32) (19) (3) (52) (8) (49) (6) (90) (19)Control Mean 30 329 79 3 2 19 318 7 326 69Treatment Effect () 88 68 47 260 69 159 52 -184 -242 60Joint F-Test (p-value) 0258 0562 0175 0567 0549 0196 0802 0494 0381 0498Treatment FDR q-value 563 563 563 563 563 563 563 563 563 563Number Observations 2335 1929 2335 2335 538 2335 1339 2335 447 2335

Panel D Older than 12Treatment (λ1) -0 05 -04 -0 -02 -0 01 0 03 -01

(1) (07) (03) (1) (09) (1) (11) (1) (13) (03)Cost Sharing (λ2) -1 02 -03 -1 -05 -1 05 1 12 -02

(1) (07) (03) (1) (10) (1) (09) (1) (13) (03)Saturation (λ3) -8 -11 -34˚ -0 -44 -5 -67 -3 -113 -17

(6) (35) (18) (3) (52) (6) (43) (5) (91) (22)Control Mean 27 318 8 4 2 16 312 7 295 64Treatment Effect () -1 47 -46 -60 -8 -4 8 43 28 -11Joint F-Test (p-value) 0429 0725 0054 0281 0661 0570 0376 0736 0368 0848Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 2291 1814 2291 2291 616 2291 1042 2291 542 2291

Panel E 12 or YoungerTreatment (λ1) 3˚˚ 09 03 -0 13 3˚˚ 05 -0 -03 07˚˚

(1) (05) (03) (1) (10) (1) (08) (1) (17) (03)Cost Sharing (λ2) -3˚ -05 -02 0 -07 -3˚˚ -10 0 -09 -05˚

(1) (06) (02) (0) (08) (1) (09) (1) (16) (03)Saturation (λ3) 16˚˚ 76˚˚ -12 -3 62 12 22 7 140 13

(6) (34) (13) (3) (51) (10) (77) (10) (119) (16)Control Mean 22 305 71 3 2 14 299 5 3 56Treatment Effect () 128 85 36 -98 55 239 49 -39 -30 127Joint F-Test (p-value) 0031 0128 0410 0702 0550 0014 0661 0765 0605 0094Treatment FDR q-value 110 230 636 722 364 106 722 784 784 106Number Observations 2304 1658 2304 2304 563 2304 1047 2304 360 2304

Table 30 Labor Supply - Pooled (including SCY and VocEd)

TotalHours Worked

Non-ZeroHours

Farm HoursWorked

Wage HoursWorked

(9)Log

(10)Non-Zero Wage

Panel A Full SampleTreatment (λ1) 1 06˚˚˚ 00 -1 -01 2˚ 02 1 01 03˚˚

(1) (02) (01) (1) (06) (1) (03) (1) (06) (01)Cost Sharing (λ2) -1 -01 -00 0 -02 -1 01 0 -02 -01

(1) (03) (01) (1) (05) (1) (03) (1) (06) (01)Saturation (λ3) 3 07 -02 -5 -56 3 -10 9 78˚ 06

(6) (11) (06) (4) (39) (7) (19) (6) (44) (06)Control Mean 40 387 64 8 3 24 395 10 37 52Treatment Effect () 35 58 5 -136 -4 68 24 98 14 54Joint F-Test (p-value) 493 048 969 463 493 322 507 349 289 079Treatment FDR q-value 316 070 638 316 638 238 517 375 638 144Number Observations 15334 9741 15334 15334 3374 15334 5457 15334 3049 15334

Panel B FemalesTreatment (λ1) 0 06 -02 -2˚˚ -03 1 01 2˚ 14 02

(1) (04) (02) (1) (09) (1) (06) (1) (08) (02)Cost Sharing (λ2) -0 01 00 1 02 -0 05 -1 -14 -00

(1) (05) (01) (1) (08) (1) (06) (1) (10) (02)Saturation (λ3) 5 35˚ -08 -8 -44 3 17 12 145˚˚˚ 02

Panel C MalesTreatment (λ1) 3˚˚ 06˚ 02˚ 0 00 3˚ 03 -0 -09 04˚˚

(1) (03) (01) (1) (09) (1) (04) (1) (08) (02)Cost Sharing (λ2) -1 -02 -00 -0 -06 -1 -00 1 07 -01

(1) (03) (01) (1) (07) (2) (05) (1) (08) (02)Saturation (λ3) 1 -11 03 -3 -72˚ 3 -21 6 29 09

Panel D Older than 12Treatment (λ1) 3˚˚ 05˚ 02 -1 -05 2˚ 05 2˚˚ -01 04˚

(1) (03) (01) (1) (07) (1) (05) (1) (07) (02)Cost Sharing (λ2) -2˚˚ -01 -02 -1 06 -2 -01 -1 10˚ -02

(1) (03) (01) (1) (07) (1) (04) (1) (06) (02)Saturation (λ3) 6 -01 02 -3 -49 9 -19 6 33 07

Panel E 12 or YoungerTreatment (λ1) 0 07˚˚ -01 -1 02 1 01 -0 03 03

(1) (04) (02) (1) (09) (2) (05) (1) (10) (02)Cost Sharing (λ2) 0 -02 01 1 -09 -1 02 0 -15 00

(1) (04) (01) (1) (09) (2) (05) (1) (12) (02)Saturation (λ3) 1 15 -03 -7˚ -68 -1 01 12 135˚ 09

(7) (17) (08) (4) (53) (13) (41) (9) (79) (09)Control Mean 36 381 59 7 3 22 391 9 369 48Treatment Effect () 13 72 -9 -146 8 60 9 -28 32 53Joint F-Test (p-value) 0954 0225 0828 0333 0350 0775 0911 0428 0240 0315Treatment FDR q-value 1000 648 1000 1000 1000 1000 1000 1000 1000 1000Number Observations 7591 4446 7591 7591 1537 7591 2638 7591 1203 7591

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Analysis includes KLPS respondents who participated in SCY or VocEd with indicators for receiving a SCY grant or a vocational training voucher See Section 9 for notes onoutcome construction See Table 27 for notes on the regression specification Observations are weighted to be representative of the original KLPS population and include KLPS population weights and KLPS intensive tracking weights denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 31 Occupational Choice - Pooled

Employed -Agriculture

Employed -Fishing

Employed -Manufacturing

(4)Employed -

ConstructionCasual Labor

Employed -Services

(6)Employed -Retail and

Wholesale Trade

(7)Employed -

TradeContractor

Panel A Full SampleTreatment (λ1) -01˚˚ 00 -00 -00 01 -00 00

(00) (01) (00) (00) (02) (01) (01)Cost Sharing (λ2) 00 00 00 00 -02 -00 00

(00) (01) (00) (00) (02) (01) (01)Saturation (λ3) -03 -07 -00 00 -05 03 -01

(02) (06) (03) (04) (09) (04) (05)Control Mean 02 03 03 02 19 04 03Treatment Effect () -406 117 -56 -29 46 -83 129Joint F-Test (p-value) 109 371 693 876 416 665 730Treatment FDR q-value 252 1000 1000 1000 1000 1000 1000Number Observations 13768 13768 13760 13768 13760 13762 13762

Panel B FemalesTreatment (λ1) -00 -00 -01 -00 01 -01 -00

(00) (01) (00) (00) (02) (01) (00)Cost Sharing (λ2) 00 -00 02˚˚˚ 00 -02 00 00

(01) (01) (01) (00) (02) (01) (00)Saturation (λ3) -06˚˚ -01 -02 01 -04 02 -00

(03) (06) (04) (03) (09) (05) (03)Control Mean 01 0 01 0 16 03 0Treatment Effect () -229 -1288 -407 -1248 43 -327 -1086Joint F-Test (p-value) 0261 0847 0029 0489 0492 0392 0635Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000Number Observations 6847 6847 6844 6847 6844 6844 6844

Panel C MalesTreatment (λ1) -01˚˚ 01 00 00 01 00 01

(01) (01) (01) (01) (02) (01) (01)Cost Sharing (λ2) 01 01 -01 01 -01 -00 00

(00) (02) (01) (01) (02) (01) (01)Saturation (λ3) -01 -11 01 00 -06 04 -02

(03) (08) (04) (06) (14) (05) (09)Control Mean 02 05 04 03 21 06 05Treatment Effect () -501 195 62 99 47 44 169Joint F-Test (p-value) 0158 0181 0775 0793 0791 0912 0789Treatment FDR q-value 509 1000 1000 1000 1000 1000 1000Number Observations 6921 6921 6916 6921 6916 6918 6918

Panel D Older than 12Treatment (λ1) -01 01 00 -01 01 -01 -01

(01) (01) (01) (01) (02) (01) (01)Cost Sharing (λ2) -00 -01 01 01 -02 00 01

(00) (01) (01) (01) (02) (01) (01)Saturation (λ3) 00 -07 02 -02 -04 00 -02

(03) (07) (04) (05) (11) (05) (07)Control Mean 03 03 03 02 18 05 04Treatment Effect () -276 366 95 -281 65 -236 -234Joint F-Test (p-value) 0408 0358 0675 0614 0454 0611 0598Treatment FDR q-value 978 978 978 978 978 978 978Number Observations 6879 6879 6875 6879 6875 6877 6877

Panel E 12 or YoungerTreatment (λ1) -01 -00 -01 00 01 00 01

(00) (01) (01) (00) (02) (01) (01)Cost Sharing (λ2) 01˚˚ 01 00 00 -02 -00 -00

(00) (01) (01) (01) (02) (01) (01)Saturation (λ3) -07˚˚˚ -06 -03 03 -06 06 -01

(02) (07) (03) (03) (12) (07) (05)Control Mean 01 03 03 02 18 04 02Treatment Effect () -710 -80 -193 186 46 69 738Joint F-Test (p-value) 0038 0776 0763 0708 0772 0831 0245Treatment FDR q-value 939 1000 1000 1000 1000 1000 939Number Observations 6836 6836 6832 6836 6832 6832 6832

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data See Section 10 for notes on outcome construction Treatment is an indicator variable equal to 1 forPSDP Worm Groups 1 and 2 which received an additional 24 years of deworming on average compared to Group 3 Reported estimates for Female and Male areconstructed from a single regression including treatment-female cost-sharing-female and saturation-female interaction terms Reported estimates for Older than 12and 12 or Younger also report results using a single regression including an indicator for those older than 12 at baseline and analogous interaction terms to PanelsB and C Covariates follow Baird et al (2016) and include controls for baseline 1998 primary school population geographic zone of the school survey wave andmonth of interview a female indicator variable baseline 1998 school grade fixed effects the average school test score on the 1996 Busia District mock exams totalprimary school pupils within 6 km and a cost-sharing school indicator Those treated in a separate vocational training intervention (VocEd) which occurred prior toKLPS-3 are dropped from the KLPS-3 and KLPS-4 sample Those treated in a separate small grant intervention (SCY) which occurred after KLPS-3 are droppedfrom the KLPS-4 sample Observations are weighted to be representative of the original PSDP population and include KLPS population weights SCY and VocEdcontrol group weights and KLPS intensive tracking weights Standard errors are clustered at the 1998 school level The Joint F-Test (p-value) gives the p-valueassociated with an F-test on the joint significance of the treatment cost-sharing and saturation coefficients against the null hypothesis that all three coefficients arejointly equal to zero The FDR adjustment is carried out across the ten outcomes within this family separately by panel (full sample females males older than 12and 12 or younger) denotes statistical significance at 10 pct at 5 pct and at 1 pct

Table 32 Occupational Choice - KLPS-4

Employed -Fishing

(4)Employed -

Employed -Services

Wholesale Trade

(7)Employed -

TradeContractor

Panel A Full SampleTreatment (λ1) -01 00 -01 -00 03 -01 01

(01) (01) (01) (01) (03) (01) (01)Cost Sharing (λ2) 01 -00 01 00 -03 -00 -00

(01) (02) (01) (01) (03) (01) (01)Saturation (λ3) -03 -13˚ 00 05 -21 -01 04

Panel B FemalesTreatment (λ1) -01 -00 00 00 02 -01 -00

(01) (01) (01) (00) (04) (01) (01)Cost Sharing (λ2) 00 -01 02˚ 00 -03 00 01

(01) (01) (01) (00) (04) (01) (01)Saturation (λ3) -07 -03 -03 00 -24 04 05

(06) (07) (06) (04) (19) (09) (06)Control Mean 01 0 01 0 24 03 0Treatment Effect () -620 34 86 -383 -316Joint F-Test (p-value) 0620 0810 0205 0956 0370 0582 0389Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000Number Observations 2102 2102 2102 2102 2102 2102 2102

Panel C MalesTreatment (λ1) -01 01 -01 -01 03 -01 02

(01) (02) (01) (01) (05) (02) (03)Cost Sharing (λ2) 01 01 -01 00 -03 -00 -02

(01) (03) (01) (01) (05) (02) (02)Saturation (λ3) 00 -21˚ 03 08 -19 -06 03

Panel D Older than 12Treatment (λ1) -01 00 -00 -01 03 -02 -01

(01) (01) (01) (01) (03) (02) (01)Cost Sharing (λ2) 00 -00 03˚ 01 -05 -01 01

(01) (02) (02) (01) (04) (01) (01)Saturation (λ3) -00 -09 00 00 -21 02 01

(09) (09) (07) (05) (20) (10) (09)Control Mean 03 02 02 02 28 06 05Treatment Effect () -422 216 -16 -311 101 -283 -199Joint F-Test (p-value) 0711 0641 0287 0771 0198 0630 0879Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000Number Observations 2059 2059 2059 2059 2059 2059 2059

Panel E 12 or YoungerTreatment (λ1) -01 01 -01 -00 03 -01 03

(01) (01) (02) (01) (04) (02) (02)Cost Sharing (λ2) 01 -00 -01 00 -02 01 -01

(01) (02) (02) (01) (04) (02) (02)Saturation (λ3) -06 -13 -01 09 -19 -06 07

Notes Analysis uses KLPS-4 data See Section 10 for notes on outcome construction See Table 31 for notes on the regression specification denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Employed -Fishing

(4)Employed -

Employed -Services

Wholesale Trade

(7)Employed -

TradeContractor

Panel A Full SampleTreatment (λ1) -00 01 -01 00 01 -00 00

(01) (01) (01) (01) (02) (01) (01)Cost Sharing (λ2) -00 00 00 00 -03˚ 01 01

(01) (01) (01) (01) (02) (01) (01)Saturation (λ3) -06 03 -07 -03 02 07 -07

(04) (07) (06) (06) (10) (06) (06)Control Mean 02 03 04 02 19 04 03Treatment Effect () -7 224 -205 168 34 -86 84Joint F-Test (p-value) 392 873 381 824 277 402 475Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000Number Observations 4593 4593 4593 4593 4593 4593 4593

Panel B FemalesTreatment (λ1) 00 -00 -02 -01˚ 00 -03˚˚ -01

(01) (01) (01) (01) (03) (01) (01)Cost Sharing (λ2) -01 00 01 00 -03 02˚ 00

(01) (01) (01) (01) (03) (01) (01)Saturation (λ3) -12 03 -11 03 08 -03 -05

(08) (06) (07) (06) (15) (07) (04)Control Mean 01 0 03 01 17 04 0Treatment Effect () 369 -1034 -648 -1241 13 -674 -3413Joint F-Test (p-value) 0247 0910 0250 0309 0648 0150 0553Treatment FDR q-value 685 685 354 297 685 297 409Number Observations 2259 2259 2259 2259 2259 2259 2259

Panel C MalesTreatment (λ1) -00 02 00 02 01 02 01

(01) (02) (01) (02) (03) (02) (02)Cost Sharing (λ2) 00 01 -01 00 -03 00 01

(01) (02) (01) (02) (02) (02) (02)Saturation (λ3) -01 03 -04 -07 -03 15˚˚ -09

Panel D Older than 12Treatment (λ1) 00 02 -01 -01 01 -02 -02

(01) (02) (01) (01) (02) (01) (01)Cost Sharing (λ2) -01 -02 -01 02 -02 02 02

(01) (02) (01) (01) (03) (02) (02)Saturation (λ3) -01 00 -05 -09 08 -08 -11

(06) (09) (06) (09) (13) (07) (10)Control Mean 03 03 05 02 17 05 05Treatment Effect () 71 527 -185 -501 44 -393 -411Joint F-Test (p-value) 0459 0814 0305 0535 0891 0411 0395Treatment FDR q-value 1000 1000 1000 1000 1000 1000 1000Number Observations 2290 2290 2290 2290 2290 2290 2290

(01) (02) (01) (01) (03) (01) (01)Cost Sharing (λ2) 01 02 01 -01 -04 00 -00

(01) (01) (01) (01) (03) (01) (02)Saturation (λ3) -10˚˚˚ 06 -08 03 -02 21˚ -03

Table 34 Occupational Choice - Pooled (including SCY and VocEd)

Employed -Fishing

(4)Employed -

Employed -Services

Wholesale Trade

(7)Employed -

TradeContractor

Panel A Full SampleTreatment (λ1) -01˚˚ 00 -00 00 01 -00 00

(00) (01) (00) (00) (01) (01) (01)Cost Sharing (λ2) 01˚˚ -00 00 00 -02 00 00

(00) (01) (00) (00) (01) (01) (01)Saturation (λ3) -04˚ -08 00 03 -05 -01 01

Panel B FemalesTreatment (λ1) -00 -00 -00 -00 01 -01˚ -00

(00) (01) (00) (00) (02) (01) (00)Cost Sharing (λ2) 01 -00 01˚˚˚ 00 -03 01 00

(01) (01) (00) (00) (02) (01) (00)Saturation (λ3) -07˚˚ -01 -02 01 -04 -02 -00

(00) (01) (01) (01) (02) (01) (01)Cost Sharing (λ2) 01˚ 00 -01 00 -01 -00 00

(00) (02) (01) (01) (02) (01) (01)Saturation (λ3) -01 -13 02 04 -06 -01 01

(00) (01) (01) (01) (02) (01) (01)Cost Sharing (λ2) 00 -01 00 00 -02 00 00

(00) (01) (01) (01) (02) (01) (01)Saturation (λ3) -01 -06 03 02 00 -06 -01

Panel E 12 or YoungerTreatment (λ1) -01˚ -00 -00 00 00 -00 01

(00) (01) (01) (00) (02) (01) (01)Cost Sharing (λ2) 01˚˚˚ 01 00 00 -02 -00 00

(00) (01) (01) (00) (02) (01) (01)Saturation (λ3) -07˚˚˚ -08 -03 03 -10 03 02

(02) (07) (03) (03) (11) (06) (04)Control Mean 01 03 03 01 18 04 02Treatment Effect () -824 -164 -166 294 23 -28 340Joint F-Test (p-value) 0021 0682 0766 0100 0461 0927 0388Treatment FDR q-value 951 1000 1000 1000 1000 1000 1000Number Observations 7570 7570 7566 7570 7566 7566 7566

Notes Analysis uses KLPS-2 KLPS-3 and KLPS-4 data Analysis includes KLPS respondents who participated in SCY or VocEd with indicators for receiving aSCY grant or a vocational training voucher See Section 10 for notes on outcome construction See Table 31 for notes on the regression specification Observationsare weighted to be representative of the original KLPS population and include KLPS population weights and KLPS intensive tracking weights denotes statisticalsignificance at 10 pct at 5 pct and at 1 pct

Introduction

Econometric Specifications

Price Conversion

Primary Outcomes

Family 1 Household Consumption

Family 2 Household Wealth

Family 3A Individual Earnings Outcomes

Part 1 Non-Hourly Earnings Measures
Part 2 Hourly Earnings Measures
Family 3B Household Earnings Outcomes
Family 4 Labor Supply
Family 5 Occupational Choice