cash transfers mechanisms - abdul latif jameel …...documented positive effects of cash transfers...
TRANSCRIPT
The Impact of Cash Transfers on Labor Supply
Gabriel Kreindler
(joint with Abhijit Banarjee, Rema Hanna, and Ben Olken)
J-PAL SEA Policy Conference on Social ProtectionJanuary 12, 2016
2
• 52 countries with conditional cash transfers in 2013
• Up from 27 countries in 2008
• 119 countries with unconditional cash transfers in 2013
• Indonesia:• Program Keluarga Harapan (PKH) with 3.2 mil beneficiary families (2014)
• Bantuan Langsung Tunai (BLT) with 15.5 mil beneficiary HHs (2013)
• Bantuan Siswa Miskin (BSM) with 11.13m beneficiaries (2014)
• Bantuan Langsung Sementara Masyarakat (BLSM)
Cash Transfers Important Worldwide
Documented Positive Effects of Cash Transfers
• Income support and poverty reduction (Fiszbein and Schady 2009)
• Increase school attendance (Schultz 2004, Glewwe and Olinto 2004, Maluccio and Flores 2005)
• Boost early childhood cognitive development (Macours, Schady, and Vakis 2012)
• Decrease child labor (Edmonds and Schady 2012, Skoufi as and Parker 2001, Maluccio and Flores 2005)
• Increase use of health services(Gertler 2000, Gertler 2004, Attanasio et al. 2005)
3
The Effects of Cash Transfers in Indonesia
World Bank RCT evaluation of PKH (2011):
• Increased food consumption
• No change in temptation goods
• Increased usage of primary healthcare services (natal check-ups, immunizations, etc.)
• Little impact on education behavior
• No effect on child labor.
4
Potential Negative Effects
1. Do cash transfer recipients work less?
2. Do recipients spend the money on temptation goods? (tobacco, alcohol, sugary drinks)
• World Bank meta-analisys of 19 programs: no effect(Evans and Popova 2014)
5
Do Transfers Discourage or Encourage Work?
• Reasons to discourage work:
• Recipients can afford to work less
• Working may lower chance to get benefits
• Reasons to encourage work:
• Nutritional poverty traps
• Can be used for investments
• Ability to hire child care
• Ability to fund seasonal migration
• Draw on rigorous RCT evidence from Indonesia and worldwide.
6
Results from 7 Randomized StudiesDo cash transfer programs discourage work?
7
Evidence from Indonesia and 5 other countries
• PKH in Indonesia
• Each study is a randomized control trial (RCT).
8
Mexico Honduras
Nicaragua
Morocco
Indonesia(PKH program)
Philippines
Cash transfers are large and last several years
• Across the seven programs:
• Transfer size 4-30% of household consumption.
• Households receive transfers for 2-9 years.
• PKH program:
• ~11% of household consumption
• Disbursed for 6 years
9
PKH recipients work as much as non-recipients
Data from SUSENAS 2013. Restricted to 162 kabupatens with >1% PKH recipients. Households with 5% highest expenditure excluded. (N=88,138) 10
0%
50%
100%
Household head Household head spouse
Fraction Working
Non-recipients PKH recipients
Data from SUSENAS 2013. Restricted to kabupatens with >1% PKH recipients. Households with 5% highest expenditure excluded. (N=88,138) 11
0
10
20
30
40
Household head Household head spouse
Hours of work per week
Non-recipients PKH recipients
PKH recipients work as much as non-recipients
Randomized evaluations ensure that detected
outcomes are caused by the policy or program
• When comparing program impacts between people that
do or do not receive a program, outcomes may be due to
differences between the people, not the program itself.
• Randomized evaluations overcome this challenge:
• Randomly assign program to different areas by lottery
• Some areas randomly assigned as “Program Group,” others
to “Comparison Group”
• Areas in the program and comparison groups are
statistically equivalent before the experiment
12
Program
Group
Groups are statistically identical before program
Comparison
Group
=
Any differences at follow-up can be attributed to program
Randomized evaluations ensure that detected
outcomes are caused by the policy or program
For almost every program, the effect of transfers on work is not distinguishable from zero
14
0%
20%
40%
60%
80%
100%
Honduras(4%)
Morocco(5%)
Philippines(11%)
Mexico PAL(11.5%)
IndonesiaPKH
(11%)
Nicaragua(20%)
MexicoProgresa
(20%)
Program Country (transfer size as % of household consumption)
Fraction who Worked Last Week
Comparison Group Program Group
Pooling all the studies: No impact on whether the recipient works
• A cash transfer program decreases the fraction working by 0.6 percentage points (not significant).
• Effect not distinguishable from zero. 95% confidence interval: -1.6% to 0.4%.
15
Indonesia: PKH did not significantly affect work
• Difference not distinguishable from zero. (N=80,851)
16
55%
60%
65%
Before the program After the program
Fraction who Worked Last Week
Comparison group
Program group
Indonesia: Larger transfers did not affect work
17
50%
55%
60%
65%
70%
1st quintile(<6%)
2nd quintile(6-8%)
3rd quintile(8-11%)
4th quintile(11-15%)
5th quintile(>15%)
Transfer Size Quintile (% of household consumption)
Fraction who Worked Last Week
ComparisonGroup
Programgroup
Cash transfers do not lead to fewer hours per day
18
0
5
10
15
20
25
30
Honduras(4%)
Morocco(5%)
Philippines(11%)
Mexico PAL(11.5%)
Nicaragua(20%)
MexicoProgresa
(20%)
Hours of Work per Week
Comparison Group Program Group
Pooling all the studies:No impact on how long recipient works
• A cash transfer program leads recipients to work 15 minutes less per week (not significant).
• This effect is not distinguishable from zero. The 95% confidence interval is between one hour less and 30 minutes more per week.
19
• How does higher household consumption decrease the expected amount of PKH transfer received?
• Relevant numbers:
• Average yearly benefit 2.1mil IDR (2013)
• Approx. 11% of beneficiary yearly consumption
• 6 years of benefits
• 4.3% of population receive PKH
20
Indonesia: The implicit PKH tax on consumption
The implicit PKH tax on consumption is small
• A household with 1mil IDR higher yearly consumption will receive, on average, 40k IDR less in benefits over 6 years.
21
Conclusion
• Despite stereotypes, the poor recipients of social benefits work as much as the rest of the population.
• Transfer do not induce recipients to work less (nor more). True in Indonesia and worldwide.
• Cash transfers have positive impacts:
• Income support
• Education
• Health service utilization
22
Terima kasih!
23
Sources
• World Bank State of the Safety Net
• World Bank (2011), Program Keluarga Harapan: Main Findings from the Impact Evaluation of Indonesia’s Pilot
Household Conditional Cash Transfer Program.
• World Bank (2009), Conditional Cash Transfers – Reducting Present and Future Poverty, A World Bank Policy
Research Report.
• CCTs on income support and poverty reduction (Fiszbein and Schady 2009)
• CCTs on health services: (Gertler 2000, Gertler 2004, Attanasio et al. 2005)
• CCTs on school enrollment and attendance (Schultz 2004, Glewwe and Olinto 2004, Maluccio and Flores 2005)
• CCTs on early childhood cognitive development (Macours, Schady, and Vakis 2012)
• CCTs on child labor: Edmonds and Schady 2012, Skoufi as and Parker 2001, Maluccio and Flores 2005.
24
Impact of transfers on work (men)
25
0%
20%
40%
60%
80%
100%
Honduras(4%)
Morocco(5%)
Philippines(11%)
Mexico PAL(11.5%)
Indonesia(11%)
Nicaragua(20%)
MexicoProgresa
(20%)
Worked Last Week (Men)
Comparison Group Program Group
Impact of transfers on work (women)
26
0%
20%
40%
60%
80%
100%
Honduras(4%)
Morocco(5%)
Philippines(11%)
Mexico PAL(11.5%)
Indonesia(11%)
Nicaragua(20%)
MexicoProgresa
(20%)
Worked Last Week (Women)
Comparison Group Program Group
Impact of transfers on hours per week (men)
27
0
5
10
15
20
25
30
35
40
45
Honduras(4%)
Morocco(5%)
Philippines(11%)
Mexico PAL(11.5%)
Indonesia(11%)
Nicaragua(20%)
MexicoProgresa
(20%)
Hours Worked per Week (Men)
Comparison Group Program Group
Impact of transfers on hours per week (women)
28
0
2
4
6
8
10
12
14
16
18
Honduras(4%)
Morocco(5%)
Philippines(11%)
Mexico PAL(11.5%)
Indonesia(11%)
Nicaragua(20%)
MexicoProgresa
(20%)
Hours Worked per Week (Women)
Comparison Group Program Group
Follow-up data collected 1.5 years after program started
• In Indonesia: data collected after 2 years and 4 months.
29
0
1
2
3
4
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Freq
ue
ncy
Years after Implementation
Follow-up Timing