VIP Call Girls Service Bikaner Aishwarya 8250192130 Independent Escort Servic...
Pathways Less Explored – Locus of Control and Technology Adoption
1. 24/03/2017 1
Pathways Less Explored – Locus of Control and
Technology Adoption in Ethiopian Agriculture
Alemayehu Seyoum Taffesse (IFPRI),
Fanaye Tadesse (IFPRI)
ESSP-EDRI Seminar
March 24, 2017
EDRI
2. Motivation
State of Ethiopian Agriculture (no statistics);
Pathways less explored – psychological and social influence
that can complement, accelerate;
Concepts and Measures
Locus of control
Technology adoption
A broad schema
Applications to Ethiopia
LOC and adoption of modern farm inputs;
Observations – so what?
Outline
24/03/2017 2
3. Motivation – State of Agriculture
24/03/2017 3
Summary – Outcome state
low but growing productivity,
Summary – Conditioning states
Rising but limited capital stock (physical, human, infrastructural,
natural),
Improving but weak institutions (imperfect markets, agencies of public
service delivery, social protection, …, beliefs and norms);
Frequent exposure to shocks (natural, market, policy);
Policy Instruments
public investment (research and extension services, education, health,
infrastructure);
‘reforms’ – effective policy making process, land reform, public sector
reform, incentives (taxes and subsidies, interest rates, regulation);
Question: Are there complementary pathways not yet used? How about
beliefs and norms
4. Motivation – why do poor people do not adopt?
24/03/2017 4
Underinvestment (non-adoption) by the poor – a source of
persistence in low productivity, poverty, and inequality
Focus 1 - ‘external circumstances’ and ‘opportunities’.
Low returns to investments; Unexploited opportunities due to lack of
information or knowledge; Social constraints;
Feder, Just, and Zilberman (1982); and Besley and Coate (1993); WDR (2008);
Suri (2011); Sheahan and Barrett (2014)
Opportunities – existent (exploit), new (create)
Focus 2 - constraints associated with the manifested attributes of
decision makers
Identity issues: sense of self;
Psychological issues: impatience, commitment, and psychological barriers;
WDR (2008), WDR (2015);
Aspirations failure, ‘external’ locus of control (Tanguy et al.)
5. Conceptual Issues – Locus of control
24/03/2017 5
Specific focus: Locus of control
Locus of Control (Rotter, Levenson, Bandura, Hill):
a person’s belief regarding the primary causation of events in his or
her life in general (or in a specific area?);
‘internal’ vs. ‘external’ – continuum;
deemed a powerful influence on personality and behaviour;
used to predict behaviour in a lot of spheres (health, education,
employment …)
See: Borghans, Duckworth, Heckman, ter Weel (2008). “The Economics
and Psychology of Personality Traits,” Journal of Human Resources,
Volume 43, Number 4, pp.972-1059.
7. 24/03/2017 7
Premise – Poorer households use less modern inputs
Low productivity and poverty
persist;
.3.4.5.6.7
1 2 3 4 5
5 quantiles of wealth_index
95% CI predicted fert_use
.1
.15
.2
.25
.3
.35
1 2 3 4 5
5 quantiles of wealth_index
95% CI predicted improved_seed
36pp - Fertilizers
17pp - Improved Seeds
Note: Chemical fertilizers
and improved seeds = 20%
of crop output growth during
2005-2013 in Ethiopia
(Bachewe et al. (2015))
8. 24/03/2017 8
Premise – LOC a possible pathway
Measuring LOC: Binary
Survey
LOC-Destiny
(%)
Number of
Observations
Ethiopia - PSNP 2008 25.8 4,360
Ethiopia - ERHS 2009 30.9 2,068
Ethiopia - PSNP 2010 32.3 4,619
Ethiopia - Aspirations Survey 2010-11 37.7 2,068
Ethiopia – AGP Baseline Survey 2011 35.3 7,896
Ethiopia – AGP Midline Survey 2013 33.1 7,495
IFPRI Pakistan Household Survey 2011 58.1 1,546
Malawi Rural Household Survey-2011 27.0 671
Ethiopia - FTF Baseline Survey 2013 30.3 6,903
Ethiopia – Transport Survey 2014 31.4 775
Ethiopia - FTF Midline Survey 2015 26.5 6,685
“Each person is primarily responsible for his/her success or failure in life.”
“A person’s success or failure in life is a matter of his/her destiny.”
9. 24/03/2017 9
Premise – LOC a possible pathway
Locus of control – pared down version of Levenson (1981)
C To a great extent my life is controlled by accidental/chance happenings.
O I feel like what happens in my life is mostly determined by powerful people.
I When I make plans, I am almost certain/guaranteed/sure to make them work.
C Often there is no chance of protecting my personal interests from bad luck happenings.
C When I get what I want, it’s usually/mostly because I’m lucky.
C My experience in my life has been that what is going to happen will happen.
O My life is chiefly controlled by other powerful people.
O
People like myself have very little chance of protecting our personal interests when they
conflict with those of more powerful people.
C
It’s not always wise for me to plan too far ahead because many things turn out to be a matter
of good or bad fortune.
O
Getting what I want requires making those people above me (people with higher status)
happy with me.
I I can mostly determine what will happen in my life.
I
I am usually able to protect my personal interests (I can usually look after what is important to
me)
I When I get what I want, it’s usually because I worked hard for it.
O
In order to have my plans work, I make sure that they fit in with the desires of people who
have power over me.
I My life is determined by my own actions.
Measuring LOC: Four-level (Likert-type) semantic scale (Strongly disagree,
Disagree, Agree, Strongly agree)
10. 24/03/2017 10
Results – LOC and Wealth
7
7.5
8
8.5
9
1 2 3 4 5
5 quantiles of wealth_index
95% CI predicted LOC_others
11.5
12
12.5
13
13.5
14
1 2 3 4 5
5 quantiles of wealth_index
95% CI predicted LOC_internal
8.28.48.68.8
9
9.2
1 2 3 4 5
5 quantiles of wealth_index
95% CI predicted LOC_chance
Poorer individuals have
lower (higher) internal
(external) locus of control (7-
17%) – FTF (2013) survey;
Holds for the AGP, PSNP4,
and Transport Surveys;
7
7.5
8
8.5
9
1 2 3 4 5
5 quantiles of wealth_index
95% CI predicted LOC_others
11.5
12
12.5
13
13.5
14
1 2 3 4 5
5 quantiles of wealth_index
95% CI predicted LOC_internal
11. 24/03/2017 11
Results – LOC and Wealth
The LOC-Wealth relations hold after controlling for age, gender, and
schooling of the respondent (OLS regression)
LOC-Internal LOC-Chance LOC-Others
FTF 2013 PSNP4
2016
FTF 2013 PSNP4
2016
FTF 2013 PSNP4
2016
Gender
(Male=1)
0.272** 0.367*** -0.301** -0.211*** -0.370*** -0.31***
(0.125) (0.093) (0.124) (0.071) (0.131) (0.077)
Age (Years)
-0.005 -0.016*** 0.005 0.008*** -0.002 0.007***
(0.004) (0.002) (0.004) (0.002) (0.004) (0.002)
Schooling
(Years)
0.039 0.024*** -0.004 0.005 -0.021 0.001
(0.026) (0.005) (0.024) (0.006) (0.023) (0.006)
Wealth Index
0.466*** 0.199*** -0.167*** -0.137*** -0.265*** -0.159***
(0.081) (0.03) (0.061) (0.027) (0.064) (0.026)
Constant
12.69*** 12.9*** 8.896*** 9.25*** 8.465*** 9.32***
(0.257) (0.176) (0.239) (0.165) (0.223) (0.145)
Number of
Observations
5907 6800 5907 6800 5907 6800
Notes: EA –clustered standard errors in parentheses.
12. 24/03/2017 12
Hypothesis – LOC and Poverty
Poorer individuals have lower internal (higher external) locus of
control;
A feedback loop linking LOC and poverty?
Poverty lowers internal LOC, but is not a complete determinant;
Weak internal (or strong external) LOC discourages ‘investment’
(including modern technology adoption) by the poor;
Poverty persists …
Explore the hypothesis in relation to adoption of modern inputs by
farmers in rural Ethiopia
Six surveys – AGP (2011, 2013), FTF (2013, 2015), Transport (2014),
PSNP4 (2016)
13. 24/03/2017 13
Specification – Propensity to adopt Modern Inputs
Specification – Ordered Probit (semi-nonparametric) regression of
adoption of modern inputs (also generalized orderd probit)
Dependent variable:
‘No Fertilizer or Improved Seeds used’ = 0; ‘Either Fertilizer or Improved
Seeds used’ = 1; ‘Both Fertilizer and Improved Seeds used’ = 2
Why order?
specific way of testing the proposition that LOC attributes are likely to
affect the overall propensity to adopt modern inputs;
a simple way of examining why most farmers appear to fail to benefit
from joint fertilizer-improved seed use;
exploit well-known ordered choice models/techniques;
Adoption
Status
FTF1 FTF2 AGP1 AGP2 PSNP4 Transport
0 43.59 32.30 39.90 44.06 51.13 14.71
1 39.92 37.96 37.86 39.25 35.50 50.97
2 16.49 29.74 22.24 16.68 13.37 34.32
Total 5969 6117 7316 7505 7287 775
14. 24/03/2017 14
Specification – Modern Input Use
Controls:
Gender (Male=1),
Fraction of Landholding Deemed
Flat,
Age (Years),
Fraction of Landholding Cultivated
with Cereals,
Schooling (Years), Rainfall (mm in logs)
Number of Working-age Male Members in
the Household,
Access to Extension Services
(Yes=1),
Total Household Landholding (Hectares in
logs),
Access to Credit (Yes=1),
Fraction of Plots with Land Certificate, Distance to a Permanent Market (km),
Average Distance of Parcels from the
Homestead (minutes),
Wealth Quintile,
Fraction of Landholding Deemed Fertile,
Off-farm and/or Non-farm Income
(Yes=1)
15. 24/03/2017 15
Results – LOC and Modern Input Use - AGP (2011)
dy/dx per one SD change (%)
0 1 2
Sex of the Household Head (Male=1) -6.8*** 2.7*** 4.1***
Age of the Household Head (Years) -13.6*** 4.5*** 9.1***
Schooling of the Household Head (Years) -2.6*** 1.1*** 1.6***
Proportion of Male Adults in the
Household
-1.1 0.4 0.7
Area Cultivated (Hectares in logs) -2.9*** 1.2*** 1.8***
Proportion of land deemed fertile (%) 2.2** -0.8* -1.4**
Access to Extension Services (Yes=1) -20.0*** 6.3*** 13.7***
Wealth index -1.7*** 0.8** 1.1***
LOC – Chance 1.7* -0.7* -1.0*
LOC – Others 0.4 0.0 0.0
LOC – Internal -2.9*** 1.1*** 1.8***
Number of observations 7212
Note: Robust Standard errors used; *** <0.1%, ** <1%, *<5%
Note: Statistical and ‘economic’ significance;
16. 24/03/2017 16
Results – LOC and Modern Input Use - FTF (2013)
dy/dx per one SD change (%)
0 1 2
Sex of the Household Head (Male=1) 2 -1.1 -0.9
Age of the Household Head (Years) 2.9 -1.5 -1.5
Schooling of the Household Head (Years) 1.96* -0.9* -0.9*
Proportion of Male Adults in the
Household
-2.4** 1.3** 1.1**
Area Cultivated (Hectares in logs) -9.1*** 5.1*** 4.0***
Proportion of land deemed fertile (%) 8.9*** -5.0*** -3.9***
Access to Extension Services (Yes=1) -17.9*** 9.6*** 8.2***
Wealth index -2.7*** 1.4*** 1.1***
LOC – Chance 1.9* -1.1* -0.8*
LOC – Others 3.1*** -1.5*** -1.2***
LOC – Internal -2.9*** 1.6*** 1.2***
Number of observations 5641
17. 24/03/2017 17
Results – LOC and Modern Input Use – PSNP4 (2016)
dy/dx per one SD change (%)
0 1 2
Sex of the Household Head (Male=1) 1.7 -1.1 -0.6
Age of the Household Head (Years) -17.2*** 10.9*** 6.2***
Schooling of the Household Head (Years) -1.9*** 1.3*** 0.6***
Proportion of Male Adults in the
Household
-3.4*** 2.3*** 1.2***
Area Cultivated (Hectares in logs) -6.3*** 4.2*** 2.2***
Proportion of land deemed fertile (%) -0.5 0.4 0.2
Access to Extension Services (Yes=1) -15.3*** 8.9*** 6.4***
Wealth index -6.6*** 4.5*** 2.3***
LOC – Chance 0.0 0.0 0.0
LOC – Others 1.6* -1.1* -0.5*
LOC – Internal -1.7** 1.1** 0.6**
Number of observations 6796
18. 24/03/2017 18
Results – LOC and Modern Input Use - Transport (2014)
dy/dx per one SD change (%)
0 1 2
Sex of the Household Head (Male=1) -19.8** 3.5 16.3***
Age of the Household Head (Years) -7.1 -4.3 11.4
Schooling of the Household Head (Years) 1.3 0.6 -1.9
Proportion of Male Adults in the
Household
-0.4 -0.2 0.6
Area Cultivated (Hectares in logs) -17.2** -8.7 25.9**
Proportion of land deemed fertile (%) -1.9* -1.0 2.9*
Access to Extension Services (Yes=1) -13.1*** -2.4 15.5***
Wealth index -3.3 -1.7 5.1***
LOC – Chance 0.3 0 -0.3
LOC – Others 0.4 0.4 -0.7
LOC – Internal -2.3** -1.3 3.7**
Number of observations 762
19. 24/03/2017 19
Summary
PSNP4
(2016)
AGP
(2011)
FTF
(2011)
AGP
(2013)
Transpo
t (2014)
AGP
(2013)
(Lagged)
10.32 14.75 19.25 38.88 8.4 21.43
p-value 0.016 0.002 0.000 0.000 0.038 0.000
All LOC coefficients: 32 out of 54 are significant below the 5% level with
the hypothesized sign;
Internal LOC coefficients: 14 out of 18 are significant below the 5% level
with the hypothesized sign;
The null hypothesis that the three LOC indicators are jointly insignificant is
rejected in all cases.
2
(3)χ
20. 24/03/2017 20
Summary
‘Standard’ associations detected – extension, wealth, farm size, credit,
labour availability, age, gender are correlated with adoption;
LOC and adoption:
Sign: Internal’ locus of control – positive; ‘chance’ and ‘powerful
others’ locus of control – negative.
Size 1: size of the correlation is comparable to that with schooling.
Size 2: association averages across datasets at -12 percent (LOC-
Chance), -15 percent (LOC-Others), and 18 percent (LOC-Internal) of
the corresponding association for ‘access to extension’.
Gender
Women heads appear less likely to adopt; and
Women heads have lower LOC-Internal and higher LOC-Chance and
LOC-Others (not huge but statistically significant differences);
21. 24/03/2017 21
Observations
So what? Any policy implications?
Ascertain the nature and extent of “psychological and social influences” that
affect behaviour – “desirable, possible, ‘thinkable’” (WDR (2015));
Relevance to policy design (complementary to incentives and widening
opportunities):
Focus both on ‘what’ and ‘how” – timing, labelling, simplifying,
reminding;
Understand target communities – norms, identity (internal constraints);
Examples from the suggestive evidence above:
Poor vs. non-poor – same delivery modality may not work;
Male vs. female – additional reason to enhance women empowerment in
agriculture;
Motivational devices, Role models
A lot to be learnt – more research and experimentation
22. 24/03/2017 22
Caveats
Association, not causation – correlations uncovered in
multiple datasets;
Costs and benefits of adoption;
Distance to markets – access, prices;
Education, farm size, availability of labour;
Access to credit, wealth;
Cross-sectional data – learning, other dynamics (risk);
Risk and time preferences;
wealth, education, agro-ecological factors (rain, soil
fertility, temperature), correlated errors;