The Impact of Rural Electrification:
Challenges and Ways Forward
Maximo Torero
[email protected]
IFPRI
11th Conference FD-PROPARCO / EUDN
3 December 2014, Maison de la RATP - 189 rue de Bercy, Paris
12ème
According to the International Energy
Agency’s (IEA) World Energy Outlook
(2013), more than 1.2 million people
worldwide did not have access to
electricity in 2011. Almost all of them live
in developing countries (1,257,000 out of
1,258,000)
The IEA reports that only 65.1 percent of
rural areas in developing countries had
access to electricity in 2011, while rural
electrification rates of transition
economies and OECD countries was
99.7 percent.
Page 2
Outline –
Moving Forward on Rural Electrification
What we know on rural electrification
What we know on expected outcomes and
impacts
Challenges in evaluating the impacts
Importance of complementarities
Key messages to move forward
Page 3
Outline –
Moving Forward on Rural Electrification
What we know on rural electrification
What we know on expected outcomes and
impacts
Challenges in evaluating the impacts
Importance of complementarities
Key messages to move forward
Page 4
What we know
All relied on sub national data comparisons to draw some
conclusions from differences in access to electricity;
• Fan et al. (2002)
• Balisacan and Pernia (2002)
• Balisacan et al. (2002) did the same for Indonesia in
1990
• Taylor (2005) and Escobal and Torero (2005)
• Kanagawa and Nakata, 2008
• Khandker, Barnes, and Samad (2009)
What we know
New papers look at the impact of electricity interventions
in terms of:
• Large development gains consistent with productivity
improvements [Lipscomb et al (2013)]
• The impact of privatization of electricity on health [GonzalezEiras and Rossi (2007)],
• Rural electrification rollout on female labor participation and
cooking technologies [Dinkelman (2011)], and also on
schooling of girls [van de Walle et.al, 2013]
• Rural electrification roll out on different dimensions of welfare
[Bernard and Torero 2011, 2014] in Ethiopia; [Davis, 1998,
Dinkelman 2011 and Spalding-Fecher and Matibe 2003] in
South Africa; and in Bangladesh and Vietnam [Khandker,
Barnes and Samad, 2009 and Khander et.al 2009]
• Rural electrification impact on employment [Davis 1988]
What we know
A quasi-meta evaluation of the evidence on the impact of
rural electrification in 12 countries by IEG showing:
• generally successful in establishing electricity infrastructure but
weak in strengthening supplier institutions;
• the largest share of benefits from rural electrification is captured
by the non-poor;
• high connection fees and community selection criteria that
emphasize economic returns continue to be barriers to reaching
the very poor;
• consumer education and promotion of productive uses would
enhance the benefits of electrification; and
• properly calculating willingness to pay can demonstrate good
rates of return on rural electrification projects.
Outline –
Moving Forward on Rural Electrification
What we know on rural electrification
What we know on expected outcomes and
impacts
Challenges in evaluating the impacts
Importance of complementarities
Key messages to move forward
Page 8
Impact Pathways of Rural Electrification
El Salvador – Measuring Reduction of Indoor
Pollution
Cuelgue el cable de un
clavo o protuberancia en la
pared, columna, o techo
(marco de foto)
detector
1 metro
más de 1.50 metros de la
ventana o puertas operables
1.50 metros de
altura
Source: Dalberg Global Development Advisors,
editor. Solar lighting the base of the pyramid:
Overview of an emerging market. IFC and the
World Bank; 2010. Available
athttp://www.lightingafrica.org/files/Solar_Lighting_f
or_the_Base_of_the_Pyramid_Overview_of_an_E
merging_Market.pdf.
How does infrastructure affect
welfare?
Δ
≈
Δ
Change in the
proportion of time
for activity “j”
+Δ
+
Change in the total
number of hours
worked by the
household
Δ
+Δ
Change in
returns to labor
Δ
Interaction
effect
Page 11
Expected Results of Rural Electrification
Term
Theme
Immediate
Coverage and
Access
Coping costs
Health
Short term
Education,
Leisure, and
Information
Productivity
Long term
Economic Growth
Indicator
Expected Impact
Gender heterogeneity
•
Percentage of households connected to the grid
Positive
No differentiated effect
•
Cost of electricity
Negative
No differentiated effect
•
Reliability of electric services
Positive
No differentiated effect
•
Number of sources used
Negative
No differentiated effect
•
Consumption of electricity
Positive
No differentiated effect
•
Energy input collection time use
Negative
Larger effect for females
•
Coping expenses in other energy sources
Negative
No differentiated effect
•
Indoor pollution
Negative
No differentiated effect
•
Incidence of acute respiratory disease among
vulnerable groups
Negative
No differentiated effect
•
Hours in education or studying in the home
Positive
No differentiated effect
•
Hours spent in childcare
No change
No differentiated effect
•
Hours spent in entertainment and other leisure
activities
Positive
Larger effect for females
•
Total hours of work
Positive
Larger effect for females
•
Percentage of hours of agricultural
Negative
Larger effect for females
•
Percentage of hours of non-agricultural work
Positive
Larger effect for females
•
In home business productivity/revenue
Positive
Larger effect for females
•
Change in total income and expenditure
•
Percentage of poor households
Positive
Negative
Larger effect for females
Larger effect for females
Results of Rural Electrification
Term
Theme
Immediate
Coverage and
Access
Coping costs
Health
Short term
Education,
Leisure, and
Information
Productivity
Long term
Economic
Growth
Ethiopia
El Salvador
15% points more likely to connect 11% to 19% more likely to connect
Spillover effects: 2% from baseline
25% of the effect of the voucher
of 41% connection rate
Changes in use of kerosene for
Changes in use of kerosene for
lighting
lighting
No changes in cooking practices No changes in cooking practices
65% reduction in overnight air
N.A
pollutants
Reduction of 37-44% on acute
N.A
respiratory infections incidence
among children < 6
No effect
Increase hours of studying in 7%
No effect
More appliance ownership
Leisure reduced in average by 0.7
No effect
hours per day
Non agricultural independent
No changes
activities increased by 13%
Annual per capita income increased
N.A
in $ 186 (34% of baseline income)
N.A
Positive distributional effects
Outline –
Moving Forward on Rural Electrification
What we know on rural electrification
What we know on expected outcomes and
impacts
Challenges in evaluating the impacts
Importance of complementarities
Key messages to move forward
Page 14
Key problems in existing work
Problem 1: Causality between the intervention
and the impact
Problem 2: Endogenous infrastructure
placement
Problem 3: Implementer solve a cost
minimization problem when deciding where to
extend an existing grid
Problem 4: Impact pathways and proper
measure of impact variables
Solution to Problem 1:
Randomization on the last mile connection
Objective: Randomize the probability of connecting to the
grid through incentive mechanisms
Instrument to used: Discount voucher
Procedure:
• Between 10-50% of eligible survey respondents are
randomly assigned vouchers for a discount charge to
cover the connection of the HH or business
Methodologically:
• Vouchers serve as an IV for intensity of electricity
access across communities and for the probability of a
household connecting to the grid.
• Difference in difference estimators.
Solution to Problem 1:
Randomization on the last mile connection within
identified treated and control towns
Potential problems:
How to define the size of the voucher
Local providers must comply with protocol established for
the distribution of the vouchers.
How to avoid contagion effects/anticipation effects
Competitive underground market of vouchers
Solution to Problem 1:
Randomization on the last mile connection within
identified treated and control towns
Potential problems:
Problems of non compliance
• Alternatively we can use random assignment as an IV
for actual connection (Duflo and Kremer 2003; Angrist
and Krueger, 2001).
• Local treatment effects paper of Imbens and Angrist
(1994)
• Political feasibility of limited number of vouchers –
implement a sequencing in the distribution.
Example: Randomized- Barriers to connection in
Ethiopia (Bernard and Torero 2010)
Connection fees range between USD 50
and USD 150 (drop down line and meter).
Need to find ways to facilitate connection
for the poorer.
What is best: 2 years loan or 5 years loan
for connection fee?
Pilot study on 20 towns to assess optimal
subsidies.
Experimental approach (randomize
encouragement through distribution of
vouchers of 10% and 20%)
Vouchers requirements: official, clear and
understandable, non-transferable,
allocated publicly, and allocated randomly
Impossible d’afficher l’image.
Random selection…
Public distribution
Evaluation Sample
Only 10 Electrified Towns
Village ID
1
2
3
4
5
6
7
8
9
10
Total
# HH surveyed
81
83
87
81
85
88
82
62
78
75
804
61.7
53.0
55.2
69.1
79.3
81.8
68.3
62.9
59.0
57.3
65.0
% 10%
56.0
34.8
38.6
46.4
47.7
47.7
51.9
61.5
57.8
70.8
50.7
% 20%
44.0
65.2
61.4
53.6
52.3
52.3
48.1
38.5
42.2
29.2
49.2
% HH connected
43.2
42.2
28.7
23.5
46.0
34.1
35.4
59.7
44.9
26.7
37.9
52.0
50.0
41.7
25.0
50.7
36.1
32.1
79.5
41.3
37.2
43.4
29.0
33.3
12.8
20.0
27.8
25.0
42.3
26.1
50.0
12.5
27.8
0.04
0.12
0.00
0.62
0.08
0.39
0.37
0.00
0.45
0.01
0.00
% vouchers
Of which
Of which:
1. % with voucher
2. % without
voucher
Test of diff (1-2): pval
Relatively large sample (804)
Power of voucher - recipients 60% more likely to connect
15 percentage points more likely on average
Solution to Problem 2:
Combine Incentives and Implementation Timeline
Example from Salvador
Identify areas of intervention
Determine evaluation areas based on sequence
of interventions
Strategy 1: Identify control group and treatment
group based on intervention sequencing
Strategy 2: Using the household certification
cost as a mechanism to identify impacts through
incentives (vouchers)
Intervention Areas- San Miguel
Electrification (1st) Electrification (2nd) Electrification (3rd)
# villages
Beneficiary
San Miguel
San Miguel
Canton
30 minute influence area to Longitudinal
# villages
# villages
Intervention Areas- Chalatenango
Electrification (1st) Electrification (2nd) Electrification (3rd)
# villages
1
1
2
2
2
3
3
5
Chalatenango
Chalatenango
Canton
30 minute influence area to Longitudinal
# villages
1
4
Beneficiary
# villages
Solution to Problem 3:
Modeling Isoprofits
Page 26
Solving Problem 3: Modeling Isoprofits
Roads versus Electric Grid:
Northern Zone of El Salvador
B
A
Page 27
Solving Problem 3: Modeling Isoprofits
Graphic representation of a stochastic production
frontier in the single-output, single-input case
B
A
Page 28
Solving Problem 3: Modeling Isoprofits
Agricultural typology areas versus electric grid:
Northern Zone of El Salvador
B
A
Page 29
Outline –
Moving Forward on Rural Electrification
What we know on rural electrification
What we know on expected outcomes and
impacts
Challenges in evaluating the impacts
Importance of complementarities
Key messages to move forward
Page 30
Complementarities of infrastructure
Peru, 2002
Bangladesh,
2000-2004
% change of PC HH Income
% change of PC HH Exp
60%
60%
50%
50%
40%
40%
Infrastructure does
seem to have an
impact on
household’s
welfare
30%
30%
20%
20%
10%
10%
0%
0%
Pipeline
water
Electricity
Water
+
Water
Elec
+ phone
Elec+ +elect
road+ Water
Elec++elect
road++
electricity
phone
phonephone
+ road
There exists
complementarities
in the provision of
different types of
infrastructure
Source:Source:
EscobalChowdhury
and Torero,and
2004.
Torero, 2006
Page 31
How Does Infrastructure Affect Welfare?
PERU, 2002
Propensity Score Matching; comparison group: HH with no assets
A) Households work more hours
B) Households increase non-agricultural hours
of work
2 infrastr
10
3.5
% change in time allocation
additional weekly hours of work
3+ infrastr
15
4
3
2.5
2
1.5
1
0.5
0
-5
-10
-15
-20
0
1 infrastruct
2 infrastruct
3+ infrastruct
1 infrastr
5
Ag salaried
Ag self-empl
Non-ag salaried
Non-ag self empl
-25
Source: Escobal and Torero, 2005.
Page 32
Infrastructure Impacts and
Gender Heterogeneity
Bangladesh, 2004: ATT effects of infrastructure among men and women
(PSM among men and women)
16
density
12
0.02
10
8
Male
Female
6
density
14
16
Treatment: 1 infrastructure
Control: No infrastructure
Diff =
14
Diff =
12
0.04
10
8
4
2
2
0
0
0
0.1
ATT
0.2
6
5
0.3
-0.2
0.4
-0.1
0
0.1
20
Treatment: 3 infrastructures
Control: No infrastructure
Diff =
Male
Female
6
4
-0.1
Treatment: 2 infrastructures
Control: No infrastructure
0.2
0.3
0.4
0.5
Treatment: More than one infrastructure
Control: No infrastructure
Diff =
15
-0.08
ATT
0.02
3
Male
Female
2
1
0
-1 -0.4
density
density
4
10
Male
Female
5
0
-0.2
0
0.2
ATT
0.4
0.6
0.8
-0.2
-0.1
0
ATT
0.1
0.2
0.3
0.4
Page 33
Outline –
Moving Forward on Rural Electrification
What we know on rural electrification
What we know on expected outcomes and
impacts
Challenges in evaluating the impacts
Importance of complementarities
Key messages to move forward
Page 34
Lessons Learned
Rigorous impact evaluation that includes
appropriately selected control groups must be a part
of rural electrification program designs
From the beginning of the program specify the
expected outcomes and the plausible sizes of
impacts given the local context
The impact methodology to be used must be
homogeneous
It is crucial to understand the technology and/or to
bring into the evaluation team the technology
expertise
Lessons Learned
We have focus mostly on internal validity – but we
also need to focus on external validity
Continuous treatment methodologies need to be
further developed to address:
• Contamination
• Quality of the service
• Price issues linked to continuous treatment
Focusing solely on cost minimization can result in
missed opportunities. When deciding where to deploy
the electric grid in rural areas it is imperative to take into
account the potential profits