Workplace based Insurance against Malaria
Experimental Evidence from Nigeria
Pieter Serneels, University of East AngliaOladele Akogun, Modibbo Adama University of Technology
Andrew Dillon, Michigan State UniversityJed Friedman, The World Bank
1
6th Development Economics Workshop Tilburg & Wageningen Universities 12-13 March 2015
Motivation� Ill health can have substantial negative effects on agricultural worker income, labour supply and productivity� HIV/AIDS reduces labour supply in Kenya, Botswana (Fox et
al. 2004; Thirumurti et al., 2006;Habyarimana et al 2010)
� Schistomosiasis reduces labor supply in Mali (Audibert and Etard, 1998)
� Pollution reduces worker output in US (Zivin 2010)
� Malaria treatment increases earnings with 11-14% in Nigeria, due to increased labor supply and productivity (Dillon et al., 2015)
Motivation
Motivation� Ill health can have substantial negative effects on agricultural worker income, labour supply and productivity
� But demand for health care is often failing
� Supply side failures: poor health care provision
� Demand side failures: still poorly understood - credit constraints, perceived returns to health care (information, trust, understanding)
� We offer a very specific insurance giving workers access to malaria testing and treatment at the workplace
1. Analyse willingness to pay and its correlates and clinic use and its correlates
2. Impact: Do access and use to this type of insurance impact worker income, labour supply and productivity?
Motivation
� Aim of the paper: Increase understanding of take up and use of health insurance in rural settings
� Literature suggests low take up of insurance
� poor quality of the services offered (at health facilities)
� lack of information about the insurance and its modalities
� ill-designed contracts (De Bock & Ugarte Ontiveros 2014)
� trust (Dercon, Gunning, Zeitlin 2011)
� understanding (Platteau, Ugarte Ontiveros 2013)
=> Offer a specific service that addresses these
� Little knowledge of the economic impact of health insurance
=> Study impact on worker income, ls and productivity
5
Motivation� Focus on malaria
� One of top three diseases worldwide, particularly in sub-Sahara Africa
� Estimated 210 million infections and ~1 million deaths per year
� 51% of households in Nigeria reported at least one episode of malaria (NLSS 2003/4) – in endemic settings, “malaria” often becomes a general term for illness/fever
� Of special relevance for agricultural growth in low income countries as investments in favourable agro-ecological areas or irrigation may increase mosquito breeding
� Both the biological and economic impacts can be severe
� Malaria is characterized by recurring fever, headache, muscle pain, and weakness/fatigue; severe cases can result in encephalopathy and death (especially for young children)
� Malaria typically reoccurs unless treated
� In past 4 years, dramatic global gains in malaria-related morbidity and mortality: ACTs, almost 100% effective
6
The malaria parasite lifecycle
Motivation� Investigate promise of a workplace based approach
� Part of search for sustainable provision of malaria care, where costs are shared by government, workers, and employers
8
Framework for analysis
� Workers derive utility from income and disutility from effort; they maximize:
� = � �� , ��; ��
� A worker will take insurance if it increases his utility:
�� = � ��
, ��; �� > ��
� = � ��� , ��
�; ��
� willingness to pay for insurance will be a function of changes in income and effort, cost of use, preferences
��� = � �� − ��
� , �� − ��
� , �; ��
� We can derive a labor response function
�� = � � ℎ� , �, �� , ��
where R is the piece rate, A is ability
Study setting
� One large sugarcane plantation in rural Nigeria –5,700 hectares
� Employs ~1000 sugarcane cutters who work throughout the season, and are organized into 10 work groups, managed by a supervisor and headmen
� Workers are paid (piece rate) 2.04 naira per “measured rod” of sugarcane they cut
Study setting
� One large sugarcane plantation in rural Nigeria –5,700 hectares
� Employs ~1000 sugarcane cutters who work throughout the season, and are organized into 10 work groups, managed by a supervisor and headmen
� Workers are paid (piece rate) 2.04 naira per “measured rod” of sugarcane they cut
Study setting
� One large sugarcane plantation in rural Nigeria –5,700 hectares
� Employs ~1000 sugarcane cutters who work throughout the season, and are organized into 10 work groups, managed by a supervisor and headmen
� Workers are paid (piece rate) 2.04 naira per “measured rod” of sugarcane they cut
Study setting
� One large sugarcane plantation in rural Nigeria –5,700 hectares
� Employs ~1000 sugarcane cutters who work throughout the season, and are organized into 8 work groups, managed by a supervisor and headmen
� Workers are paid (piece rate) 2.04 naira per “measured rod” of sugarcane they cut
� The plantation records for each worker the daily amount cut, days worked, and earnings – average daily wage: 1428 Naira (~ US$9.5)
Study setting
� One large sugarcane plantation in rural Nigeria –5,700 hectares
� Employs ~800 sugarcane cutters who work throughout the season, and are organized into 8 work groups, managed by a supervisor and headmen
� Workers are paid (piece rate) 2.04 naira per “measured rod” of sugarcane they cut
� The plantation records for each worker the daily amount cut, days worked, and earnings – average daily wage: 1428 Naira (~ US$9.5)
Study setting
� One large sugarcane plantation in rural Nigeria –5,700 hectares
� Employs ~1000 sugarcane cutters who work throughout the season, and are organized into 8 work groups, managed by a supervisor and headmen
� Workers are paid (piece rate) 2.04 naira per “measured rod” of sugarcane they cut
� The plantation records for each worker the daily amount cut, days worked, and earnings – average daily wage: 1428 Naira (~ US$9.5)
Field Experimental Method
� We offer each worker one chance to get access to malaria testing and treatment two times during six weeks in the harvest period.
� Only covers worker himself, not family or friends� Only covers malaria: treated after test� Allows two tests during six weeks period� No refund or rebate
� We elicit each worker’s willingness to pay using Becker-De Groot-Maarschak (BDM) method� Incentive compatible: “worker puts his money where his mouth is”
� No bargaining� Price paid is determined exogenously through a draw� Careful explanation and double checking of understanding (extensive training, use 6 page script)
� Performs well (Berry, Fisher, Gutierrez, 2014)
Field Experimental Method
� Extensive pilot� Develop script for our service and context� Pre-pilot (January 2013)� Translate and back translate in Hausa by registered translators
� First pilot: 3 rounds with 24 workers to revise script and price levels (Sep 2013)
� Second pilot: 5 rounds with 40 workers to fine tune price levels and logistics of game (Jan 2014)
� Final price levels� Capture the (hypothetical) distribution of wtp observed during the pilot
� Is deliberately nontransparent to avoid information spillovers
� 12 price levels with a nontrivial step of 190 (50, 240, 430, 620, 810, 1000, 1190, 1380, 1570, 1760, 1950, 2140)
Method� Enumerators trained to follow detailed script
1. Explanation & demonstration - small group session:
1. Explain service, contract, game
2. Practice round with mosquito coil
3. Test understanding of game and re-explain
4. Emphasize worker has to buy service if he draws a price lower or equal than his bid
2. Decision – individual session:
1. Explain service and contract again
2. Elicit wtp
3. Several checks and opportunities to revise wtp
4. Draw a price from a bag
5. Buy service if price drawn ≤ bid
6. Ask whether worker has enough money in his account
7. Worker signs a paper where he agrees that funds can be taken from earnings from work at the plantation
8. Ask worker to keep price and outcome confidential for 6 weeks
Method� Trained enumerators followed a detailed script
1. Small group session:
1. Explain service, wtp, game,
2. Practice round with mosquito coil
3. Test understanding of game and re-explain
4. Emphasize worker has to buy service if he draws a price lower or equal than his bid
2. Individual session:
1. Explain service
2. ask wtp
3. several checks and opportunities to revise wtp
4. Draw a price from a bag
5. Buy service if price drawn ≤ bid
6. Ask whether worker has enough money in his account
7. Worker signs a paper where he agrees that funds can be taken from earnings from work at the plantation
8. Workers is asked to keep price and outcome confidential for 6 weeks
Method� Enumerators trained to follow detailed script
1. Explanation & demonstration - small worker group session:
1. Explain service, contract, game,
2. Practice round with mosquito coil
3. Test understanding of game and re-explain
4. Emphasize worker has to buy service if he draws a price lower or equal than his bid
2. Decision – worker individual session:
1. Explain service and contract again
2. Elicit wtp
3. Several checks and opportunities to revise wtp
4. Draw a price from a bag
5. Buy service if price drawn ≤ bid
6. Ask whether worker has enough money in his account
7. Worker signs a paper where he agrees that the funds can be taken from his earnings at the plantation
8. Worker is asked to keep price and outcome confidential for 6 weeks
Field Experimental Method
� If worker gains access:� He receives a card entitling him to 2 visits to our mobile health facility
� Worker chooses time of visit� During a visit the worker is tested and, if positive, treated with ACT
� Implementation is overseen by a committee consisting of 2 worker representatives, a harvest manager, the head nurse, and a senior researcher
� No workers refused to participate� Small worker survey on work, health, malaria� Successful implementation in at times demanding conditions
� Random draw of 12 price levels was successful
Field Experimental Method
� If worker gains access:� He receives a card entitling him to 2 visits to our mobile health facility
� Worker chooses time of visit� During a visit the worker is tested and, if positive, treated with ACT
� Implementation is overseen by a committee consisting of 2 worker representatives, a harvest manager, the head nurse, and a senior researcher
� No workers refused to participate� Small worker survey on work, health, malaria� Successful implementation in at times demanding conditions
� Random draw of 12 price levels was successful
0.0
2.0
4.0
6.0
8.1
Fra
ction
50 240 430 620 810 1000 1190 1380 1570 1760 1950 2140Drawn Price Y
Econometric Analysis
1. Demand and correlates
�� ���� = � + � !� + �"#� + �$�� + �%�� + �&'(� + �)*� + �+�� + ��
2. Use of the service
�� = � + �,���� + � !� + �"#� + �$�� + �%�� + �&'(� + �)*� + �+�� + ��
*� = � + ��� + ��
I=InformationE=experience and exposureT=TrustR=Risk preferences
SH=Self assessed healthD=DemographicsW=Wealth
3. Impact of having access to the insurance on income, labor supply and productivity?
4. Impact of clinic use (when gained access) on income, labor supply and productivity?
25
�� = � + � �-� + �".� + ���� = � + � �� + �".� + ��
�� = � + � �/� + �".� + ���� = � + � ���'�� + �"�#'�#*0#12�#� + ��
DemandWillingness to pay
0.1
.2.3
.4F
raction
100350 7001000 1500 2000 2500 3000 3500 4000 4500 5000Max WTP for Malaria Health Service?
Mean 350 Naira
Standard Deviation 405 Naira
Frequencies
0-99 15%
100-199 20%
200-299 23%
300-499 12%
500-999 20%
1000-5000 10%
Demand
-.2
0.2
.4.6
.8ta
ke
up
0 500 1000 1500 2000price drawn
Linear prediction lpoly smooth: takeup
0.2
.4.6
.81
Pro
babili
ty >
max_w
tp_health
200 500 1000 1500 2000 2500 3000 3500 4000 4500 5000willingness to pay
Demand
� Demand is nonlinear: workers are more responsive to a change in price at low levels of price. Especially price reductions below 700 Naira will lead to increased demand
-.2
0.2
.4.6
.8ta
ke
up
0 500 1000 1500 2000price drawn
Linear prediction lpoly smooth: takeup
� Demand is price sensitive: ε=-0.25
Correlates of willingness to pay
29
wtp
Education 0
Wealth 0
Experience with and exposure to malaria care +
Information: knowledge malaria treatment +
Information: expected cost of treatment +
Trust +
Risk preferences 0
Time preferences 0
Self assessed health +
Demographics 0
30
(2) (4) (5) (9) (10) (11) (13)
VARIABLES lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1
age2 -0.003 -0.003 -0.002 -0.002 -0.002 -0.004
(0.005) (0.005) (0.005) (0.005) (0.005) (0.006)
firmexpy 0.024** 0.023** 0.018** 0.020** 0.020** 0.017*
(0.008) (0.008) (0.007) (0.007) (0.007) (0.008)
yos 0.000 -0.003 -0.000 -0.002 -0.002 0.001
(0.018) (0.018) (0.018) (0.017) (0.017) (0.018)
hhasset_m1D000 0.012 0.010 0.008 0.008 0.008 0.007
(0.007) (0.008) (0.008) (0.008) (0.008) (0.008)
selfrepmallm 0.279** 0.258** 0.256** 0.267**
(0.109) (0.110) (0.113) (0.112)
evertested 0.273 0.261
(0.153) (0.157)
lastyear -0.005 0.022 -0.002
(0.088) (0.098) (0.088)
everprescribedaftest 0.363** 0.359*
(0.158) (0.161)
hh_members_diagnosed 0.031
(0.028)
hhsize2 0.012
(0.013)
Constant 5.344*** 5.188*** 5.234*** 5.129*** 5.166*** 5.161*** 5.148***
(0.350) (0.171) (0.377) (0.378) (0.385) (0.381) (0.383)
Observations 825 971 825 825 825 825 825
R-squared 0.014 0.005 0.018 0.040 0.041 0.041 0.043
31
(7) (6) (8) (9) (10) (11) (13) (14) (17)
VARIABLES lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1 lwtpp1
age2 -0.002 -0.002 -0.004 -0.006 -0.004 -0.003 -0.003 -0.002 -0.002
(0.005) (0.005) (0.006) (0.005) (0.006) (0.005) (0.005) (0.005) (0.005)
firmexpy 0.022** 0.022** 0.021** 0.019** 0.021** 0.023** 0.023** 0.022** 0.021**
(0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.007)
yos -0.004 -0.005 -0.001 -0.006 -0.001 -0.003 -0.004 -0.006 -0.005
(0.019) (0.019) (0.018) (0.018) (0.018) (0.018) (0.017) (0.019) (0.019)
hhasset_m1D000 0.008 0.008 0.007 0.008 0.007 0.010 0.009 0.008 0.008
(0.008) (0.009) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008)
knowmaltotpos 0.029
(0.017)
knowmalsymp -0.053 -0.060* -0.054
(0.031) (0.031) (0.031)
knowmaltreatpos 0.108* 0.102* 0.106*
(0.053) (0.054) (0.054)
knowmalprev 0.033 0.026 0.032
(0.063) (0.066) (0.064)
lnexpected_cost 0.118**
(0.047)
lnexpected_cost_treat 0.120**
(0.046)
lnexpected_cost_test 0.046
(0.047)
time_min_facility 0.000
(0.001)
selfassesmalinfo 0.019* 0.015
(0.010) (0.012)
knowhi 0.050
(0.185)
Constant 5.075*** 5.164*** 4.499*** 4.729*** 5.011*** 5.224*** 5.169*** 5.148*** 5.174***
(0.338) (0.347) (0.586) (0.409) (0.614) (0.377) (0.377) (0.343) (0.355)
Observations 825 825 798 784 797 822 825 825 825
R-squared 0.020 0.026 0.022 0.022 0.016 0.018 0.020 0.027 0.026
32
(1) (7) (11)
VARIABLES lwtpp1 lwtpp1 lwtpp1
age2 -0.002 -0.001 -0.002
(0.005) (0.005) (0.005)
firmexpy 0.017** 0.016** 0.018**
(0.006) (0.007) (0.007)
yos 0.004 0.005 -0.002
(0.018) (0.019) (0.019)
hhasset_m1D000 0.006 0.006 0.007
(0.008) (0.008) (0.008)
lastyear -0.019 -0.016 0.013
(0.093) (0.091) (0.091)
selfrepmallm 0.272** 0.274** 0.302**
(0.106) (0.106) (0.107)
evertested 0.257 0.251 0.282
(0.156) (0.159) (0.161)
lntrust 0.187*** 0.184***
(0.052) (0.050)
crra -0.016
(0.032)
twoweeki 0.120
(0.370)
perceivedhealth2d2 -0.221*
(0.110)
perceivedhealth2d3 -0.006
(0.087)
Constant 4.111*** 4.125*** 5.226***
(0.452) (0.468) (0.381)
Observations 811 811 823
R-squared 0.057 0.057 0.049
Use of care when gained accessHow often is the service used, once gained access?
� Workers do not all use their full entitlement
01
02
03
04
0P
erc
en
t
0 .5 1 1.5 2useservice
Parasite
count
First visit Second visit
# of
workers
Percentage of
total
# of
workers
Percentage
of total
Malaria negative rate =
1st
visit: 76.7%
2nd
visit: 66.3%
0 14 7.1 7 7.1
1 62 31.5 27 27.6
2 75 38.1 31 31.6
Malaria positive rate =
1st
visit: 23.3%
2nd
visit: 33.7%
3 25 12.7 19 19.4
4 11 5.6 6 6.1
5+ 10 5.0 8 8.2
197 workers assessed 98 workers assessed
Correlates of use of care
34
Use
Education +
Wealth 0
Experience with and exposure to malaria care +
Information: knowledge malaria treatment 0
Information: expected cost of treatment -
Trust +
Risk preferences -?
Time preferences 0
Self assessed health +
Demographics 0
Overview correlates
35
wtp Use
Education 0 +
Wealth 0 0
Experience with and exposure to malaria care + +
Information: knowledge malaria treatment + 0
Information: expected cost of treatment + -
Trust + +
Risk preferences 0 -?
Time preferences 0 0
Self assessed health + +
Demographics 0 0
Causal effects on labour outcomes
ITT: impact of access to insurance on labour outcomes
(1) (2) (3) (4) (5) (6)
VARIABLES lny lny ls ls lnw lnw
OLS IV OLS IV OLS IV
havecare 0.034 0.067** 0.176 -0.366 0.032 0.075***
(0.026) (0.031) (0.282) (0.548) (0.023) (0.024)
Observations 969 969 969 969 969 969
� Having gained access by level of wtp
38
0.2
.4.6
.81
lpoly
sm
ooth
: havecare
0 1000 2000 3000 4000 5000lpoly smoothing grid
� Having gained access by level of wtp
39
0.2
.4.6
.81
lpoly
sm
ooth
: havecare
0 1000 2000 3000 4000 5000lpoly smoothing grid
Table: Willingness to pay
Low 0-199 35%
Low-medium 200-499 35%
Medium- High 500-999 20%
High 1000-5000 10%
� ITT by level of wtp
40
Low wtp Low-med wtp Med-high wtp High wtp
lny lny lny lny
OLS -0.002 0.015 0.063 0.003
(0.066) (0.065) (0.067) (0.053)
IV 0.231** -0.034 0.078 0.023
(0.096) (0.056) (0.061) (0.146)
ls ls ls ls
OLS -1.817 -0.522 0.876 2.161
(1.543) (1.156) (1.351) (1.778)
IV 0.296 -1.092 0.851 5.089*
(1.738) (0.788) (1.195) (2.877)
w w w w
OLS 0.054 0.021 0.034 -0.008
(0.049) (0.059) (0.043) (0.042)
IV 0.236** -0.007 0.047 -0.079
(0.093) (0.049) (0.041) (0.093)
Effect of clinic use on labour outcomes
Effect of clinic use on labour outcomes
(1) (2) (3) (4) (5) (6) (7)
VARIABLES lntotalamount lntotalamount work_days work_days lnseasonwage lnseasonwage useservice
OLS IV OLS IV OLS IV 1st
Stage
Use 0.105*** 0.270* 2.331*** 8.149*** 0.047 0.083
(0.032) (0.160) (0.683) (3.138) (0.028) (0.108)
Age -0.006* -0.005* -0.099 -0.093* -0.004 -0.004 -0.003
(0.003) (0.003) (0.059) (0.050) (0.002) (0.002) (0.007)
experience 0.002 -0.001 -0.146 -0.251*** 0.006 0.006 0.014*
(0.003) (0.006) (0.088) (0.083) (0.004) (0.005) (0.008)
yos -0.018 -0.024 -0.349 -0.574 -0.013 -0.014 0.041**
(0.016) (0.021) (0.382) (0.497) (0.008) (0.011) (0.019)
lntrust 0.198***
(0.073)
Ever tested 0.288**
(0.114)
Constant 11.287*** 11.150*** 50.104*** 45.452*** 7.422*** 7.388*** -0.265
(0.250) (0.181) (6.215) (3.166) (0.122) (0.137) (0.478)
Observations 189 186 189 186 189 186 194
R-squared 0.048 0.055 0.042 0.031 0.114
Conclusion
� Average wtp for a very specific insurance that provides access to malaria testing and treatment is modest but has considerable variation.
� Demand is price sensitive, especially at lower price levels
� Correlates of demand: experience and exposure, information, trust
� Correlates of use: experience and exposure, information, trust
� Access to insurance has strong causal effects on income and productivity, but not ls, in particular for workers with low willingness to pay
� Use of clinic has strong causal effects on income, through labour supply
� workplace based health insurance may provide useful
� As complement for malaria control and eradication
� As a way to provide health insurance
Coda
� Thanks for listening!
Appendix
Knowledge about malaria
back
47
.1.2
.3.4
.5.6
Fra
ctio
n
0 1 2 3 4knowing the symptoms of malaria
.1.2
.3.4
.5.6
Fra
ctio
n
0 1 2 3 4knowing the prevention of malaria
.1.2
.3.4
.5.6
Fra
ctio
n
0 1 2 3 4knowing the treatment of malaria
(a) Prevention (b) symptoms (c) treatment
Knowledge about malaria
48
0.2
.4.6
.81
Pro
bab
ility
<=
kn
ow
mals
ym
p
0 1 2 3 4knowmalsymp
0.2
.4.6
.81
Pro
bab
ility
<=
kn
ow
maltre
atp
os
0 1 2 3 4knowmaltreatpos
0.2
.4.6
.81
Pro
bab
ility
<=
kn
ow
malp
rev
0 1 2 3 4knowmalprev
back
Trust
49
0.0
02
.004
.006
.008
.01
De
nsity
0 200 400 600 800 1000trust
back
Risk preferences
50
1,800
Naira
Lottery A
1,800
Naira
6000
RWF
Lottery C
5,400
Naira-200
Naira
Lottery F
3,400
Naira1000
Naira
5000
Naira
200
Naira
Lottery E
4,200
Naira
600
Naira
Lottery D
2,600
Naira
Lottery B
1,400
Naira
back
Risk preferences
51
low
payoff
high
payoff
expected
payoff
Implied range crra
fraction of
subjects
Lottery A 1800 1800 1800 2.24 < r 39%
Lottery B 1400 2600 2000 0.75 - 2.24 17%
Lottery C 1000 3400 2200 0.44 - 0.75 10%
Lottery D 600 4200 2400 0.28 - 0.44 3%
Lottery E 200 5000 2600 0.00 - 0.28 10%
Lottery F -200 5400 2600 r < 0.00 22%
back