aspirations and poverty in rural ethiopia
TRANSCRIPT
1
Aspirations and Poverty in Rural Ethiopia
Tanguy Bernard1, Stefan Dercon2, Kate Orkin3, Fanaye Tadesse1,
Alemayehu Seyoum Taffesse1, and Ibrahim Worku1
1International Food Policy Research Institute2 University of Oxford, 3 University of Cambridge
iiG Conference: Improving Institutions for Growth
20-21 March 2015
St Catherine’s College, University of Oxford
27/03/2015
Motivation and research agenda;
Aspirations
Aspiration and poverty – suggestive evidence
Field experiment – design and findings
Report on the direct effects on aspirations;
Summarize results related to beliefs, preferences, and future-oriented
behaviour
Outline
27/03/2015 2
Motivation - Initial
27/03/2015 3
Fatalism
Examples:
“We live only for today”;
“It is a life of no thought for tomorrow”;
“We have neither a dream nor an imagination”
Rahmato and Kidanu (1999)
General - lack of proactive and systematic effort to better one’s own
life;
Economic perspective – not making the ‘investments to better one's
life’ or exploit and/or create opportunities.
Evidence: underinvestment by the poor is common and can be a source
of persistence in poverty and inequality
Step 1: Correlations
Step 2: Measurement
Step 3: Treatment
Step 4: Experiment
Step 5: Replications
Research program
With parallel implementation as
demanded by circumstances
Aspirations – what, why?
27/03/2015 5
Aspirations:
are goals or boundary-states sought after with respect to a relevant
domain of choice (future-oriented);
Aspirations and expectations – preferences vs. beliefs;
Aspirations are important for analysing and/or addressing
poverty:
Condition individual behaviour and well-being (motivators );
Are distributed unevenly within communities – beliefs, experience,
personality;
Are context-dependent and changing;
Correlates and Selected Outcomes
27/03/2015 6
Improved
seed use
Fertilizer
use
Radio
ownership
Girls
school
ratio
School
ratio – all
school-age
children
LoC - Internal0.0064** 0.0060* 0.0095*** 0.0071*** 0.0050**
(0.003) (0.003) (0.002) (0.002) (0.002)
LoC - Chance-0.0015 -0.0144*** -0.0034* 0.0031 0.0016
-0.003 -0.004 (0.002) (0.003) (0.003)
LoC-Others-0.0035 -0.0044 -0.0040** -0.0097*** -0.0086***
-0.003 -0.003 (0.002) (0.003) (0.003)
Self-reported
wealth
0.0361*** 0.0647*** 0.0496*** 0.0375*** 0.0307***
Controls – sex, age, education; village clustered standard errors.
Suggestive Evidence - Wealth Aspiration I
Ethiopian Rural Households Survey (ERHS)
Spatial coverage: 15 Kebeles (villages);
Temporal coverage: 1993/94-2009 (7 rounds the last three roughly one every 5
years)
Wealth Aspirations
Round 7 Question: We would now like you to think of your own wealth.
Thinking of a scale from 1 (the lowest or worst level) to 10 (the highest or
best level):
Q39a. At what level do you believe you are currently?
Q39b. At what level would you like to be?
Estimation
Use ordered responses to Q39b as the dependent variable;
o ordered probit model (basic, generalized, semi-nonparametric);
Robust/Clustered standard errors
27/03/2015 7
Marginal Effects – All Rounds
dy/dx per one SD change (%)
Shock
Median income
growth of
neighbors
(round7-round6)
Median
income
growth (all
rounds)
Number of
rounds
respondent
was poor
Pr(Wealth Aspiration = 1) 1.92 -3.44 0.00 2.30
Pr(Wealth Aspiration = 2) 0.79 -3.44 0.00 0.88
Pr(Wealth Aspiration = 3) 1.36 0.00 -2.60 0.00
Pr(Wealth Aspiration = 4) -3.84 -6.88 5.19 -2.65
Pr(Wealth Aspiration = 5) -0.23 13.77 0.00 -0.71
Mean (SD) 2.4 (1.1) -18.3 (34.4) 3.3 (26.0) 2.8 (1.8)
Note: Figures in red are statistically significant at least at 10% level of significance. Controls
include: sex, age, marital status, education, participation in non-farm activities, Iddir
membership.
Comparable results when assets or self-reported wealth rank is used instead of
income, as well as when the analysis is restricted to rounds 6 and 7 only;
27/03/2015 8
Suggestive Evidence - Wealth Aspirations II
dy/dx per one SD change (%)
Self-reported
wealth
Median self-reported
wealth in the village
Log (Asset Aspiration)
Mean = 11.02, SD = 1.37
29.0 9.1
Mean (SD) 4.4 (1.1) 4.13 (0.35)
Note: Figures in red are statistically significant at least at 1% level of significance.
Controls include: sex, age, marital status, education, participation in non-farm activities,
Iddir membership.
27/03/2015 9
Insurance Data
Surveys of the Index Insurance Study 2011-2014;
Baseline and four follow-up household surveys of 1760 randomly
selected households.
Aspiration module added in the last round;
Observations Results
Poorer individuals have on average lower aspirations;
Results persist across specifications;
Panel data used, but happy to consider them as correlations;
Issues
Measurement – revealed vs. declared – develop an instrument
Identification – correlations vs. causal links (poverty–low
aspirations) – field experiment
27/03/2015 10
Specific Question
27/03/2015 11
Is it possible to alter poor individuals’ aspirations and link
changes thereof to behavior using an experimental design in a
real-world setting?
Measure aspirations;
Introduce an exogenous shock to aspiration;
Estimate impact on aspirations, correlates/determinants, and
behaviour;
Four dimensions
income, assets, education, social status
weights
Five questions;
What is the maximum (minimum) level of (dimension k) that one can have in
have in your current neighborhood?
What is the minimum level of (dimension k) that one can have in your current
current neighborhood?”
What is the level of (dimension k) that you have at present?”
What is the level of (dimension k) that you would like to achieve in your life?”
Measurement of Aspiration
27/03/2015 12
Experimental design: individual treatment
27/03/2015 13
64 villages. Random selection of 6 treatment HH, 6 placebo HH, 6
control HH. Head and spouse treated.
3 arms:
Treatment: ticket to view mini-documentaries about similar people who were
successful in agriculture or small business.
o No other intervention.
o 4 x 15 minute documentaries (2 men, 2 women) = 1 hour in Oromiffa
o Examples on CSAE Oxford YouTube channel
Placebo: local Ethiopian TV show in 15 minute segments.
Control: surveyed at their home.
3 rounds of data collection:
baseline (Sept-Dec 2010),
aspirations immediately after treatment,
follow-up six months later (Mar-May 2011).
Measures of aspirations
27/03/2015 15
Four dimensions:
Annual income in cash
Assets: house, furniture, consumer goods, vehicles
Social status: do villagers ask advice
Level of education of oldest child
Aspirations vs. Expectations:
What is the level of ___ that you would like to achieve?
What is the level of ___ that you think you will reach within ten years?
Overall aspiration index:
𝐴𝑖 = 𝑘w𝑖𝑘 𝑎𝑖𝑘 − 𝜇𝑘𝜎𝑘
𝑎𝑖𝑘 = individual 𝑖’s aspiration response to dimension 𝑘.
𝑤𝑖𝑘 = weight individual 𝑖 assigned to dimension 𝑘.
𝜇𝑘 , 𝜎𝑘 = village sample mean and standard deviation for dimension 𝑘.
Specification
27/03/2015 16
𝐴𝑖 = 𝑘w𝑖𝑘 𝑎𝑖𝑘 − 𝜇𝑘𝜎𝑘
𝑎𝑖𝑘 = individual 𝑖’s aspiration response to dimension 𝑘.
𝑤𝑖𝑘 = weight individual 𝑖 assigned to dimension 𝑘.
𝜇𝑘 , 𝜎𝑘 = village sample mean and standard deviation for dimension 𝑘.
Results
27/03/2015 17
After screening (t=1)
Aspirations Expectations
Treated individual 0.13* 0.13* 0.12* 0.12** 0.12** 0.11**
0.07 0.07 0.06 0.06 0.05 0.05
Placebo individual 0 0 0 0.02 0.03 0.03
0.03 0.03 0.03 0.04 0.04 0.03
Village F.E. Yes Yes Yes Yes Yes Yes
Lagged outcome No Yes Yes No Yes Yes
Controls No No Yes No No Yes
Respondents 1959 1957 1957 1959 1954 1954
Small treatment effects on aspiration immediately (about 20% of SD).
No placebo effect;
Results
27/03/2015 18
After six months (t=2)
Aspirations Expectations
Treated individual 0.04* 0.04* 0.03* 0.06*** 0.06*** 0.05**
0.02 0.02 0.02 0.02 0.02 0.02
Placebo individual 0.03 0.02 0.03 0.02 0.02 0.03
0.02 0.02 0.02 0.02 0.02 0.02
Village F.E. Y es Yes Yes Yes Yes Yes
Lagged outcome No Yes Yes No Yes Yes
Controls No No Yes No No Yes
Respondents 2063 2058 2058 2062 2054 2054
Small effects on aspiration after 6 months (about 3-5% of SD);
No placebo effect;
27/03/2015 19
Observations
Watching documentaries about role models improves
aspirations compared to a control group and, in some cases,
compared to a placebo group.
Driven by those with above-median aspirations at baseline.
No changes in risk aversion and time preferences.
Improvements in individuals’ sense that they control their lives (LoC,
causes of poverty).
Small effects on ‘forward-looking behaviour’ - children’s
school enrolment, spending on schooling, hypothetical desire
for credit - that are robust to multiple testing.
Effects on savings, credit are not robust to multiple testing.
27/03/2015 20
Observations
Note:
The treatment is “weak“ – an aspirational intervention alone;
Effects on behaviour are small and diminishing – but can have lasting
effects (e.g. one additional year of schooling)
Repeat survey to be conducted soon