Energy Policy 108 (2017) 657–672

Contents lists available at ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Scaling up finance for off-grid renewable energy: The role of aggregation MARK
and spatial diversification in derisking investments in mini-grids for rural
electrification in India
⁎
Abhishek Malhotraa, , Tobias S. Schmidta, Leonore Haelga, Oliver Waissbeinb
a
ETH Zurich, Department of Humanities, Social and Political Sciences, Energy Politics Group, Haldeneggsteig 4, CH-8092 Zurich, Switzerland
b
United Nations Development Programme, 304 E 45th Street, FF-924, New York, NY 10017, USA

A R T I C L E I N F O A BS T RAC T

Keywords: Today, about 1.1 billion people lack access to electricity worldwide. It is estimated that annual investments of 48
Energy access billion USD are required to meet the target of the Sustainable Development Goals of providing universal
Financial risk electricity access by 2030. The need for private investments to meet this target is evident, but small-scale
Poverty alleviation electrification projects are often unattractive for private investors due to unfavourable risk-return profiles and
Investment attractiveness
small investment volumes. Both issues can potentially be addressed by aggregating projects into diversified
Portfolio diversification
portfolios – an approach commonly used by investors in several contexts, but little investigated in the context of
rural electrification. This paper addresses the question of how spatial diversity in a portfolio can be used to
reduce investment risks and increase investment volumes through a mixed-method approach involving three
steps: (i) identification and classification of investment risks for renewable energy mini-grid projects, (ii)
qualitative analysis of the correlations between investment risks for different projects through interviews, and
(iii) quantitative estimation of the cost of capital and derisking effects of spatial diversification strategies for an
experimental portfolio in India. We discuss the implications for policymakers in promoting and facilitating the
ability of private sector investors to aggregate small-scale electrification investments.

1. Introduction
About 1.1 billion people lack access to electricity worldwide. An
estimated 48 billion USD of annual investments are required to provide
electricity access to them by 2030 (IEA and World Bank, 2015).
Compared to the investment of 13 billion USD in 2013 (IEA, 2015),
this represents a need to scale up investments significantly. Private
investment is likely to play a major role in meeting this target
(UNCTAD, 2014). However, how to attract these investments is an
aspect that has not received sufficient attention in the literature on
rural electrification. Off-grid electrification projects are often unattrac-
tive for private investors due to unfavourable risk-return profiles and
small investment volumes (Schmidt, 2015). Both issues can potentially
be addressed by aggregating projects into larger, diversified portfolios.
The potential of aggregation to redirect financial flows into small-scale
renewable energy is beginning to be recognised internationally. For
instance, solar home system developers in East Africa such as BBOXX
and M-Kopa have closed aggregation transactions in 2016.
Internationally, the Climate Aggregation Platform (CAP) initiative
was announced at the COP21 climate talks in Paris (UNDP; GEF;
CBI, 2015). However, there are very few studies investigating financial
aggregation of renewable energy assets in developing country or off-
grid contexts.
Most existing studies analysing investment risks for renewable
mini-grid projects focus on measures to reduce risks, and to increase
stable revenue streams for mini-grid developers (Schmidt et al., 2013;
Aggarwal et al., 2014; Schnitzer et al., 2014; Williams et al., 2015;
Comello et al., 2017). These studies focus on single projects, and do not
consider the role of investment risks for portfolios of such projects.
Lowder and Mendelsohn (2013) and Alafita and Pearce (2014) evaluate
the potential for aggregation of distributed renewable power genera-
tion, but only for grid-connected PV in the United States. The potential
for aggregation of off-grid power in developing countries, a theme that
is missing in these studies, is taken up by Davidsen et al. (2015),
Gershenson et al. (2015) and (Orlandi et al., 2016). They discuss the
benefits of structuring pooling facilities which can aggregate projects to
reduce risks, increase investment volume, and lower transaction costs.
All the studies evaluating the potential for aggregation discuss the
benefits of spatial diversification, i.e., investing in a portfolio of projects
located across different geographic distances and jurisdictions.

⁎
Corresponding author.
E-mail address: abmalhot@ethz.ch (A. Malhotra).

http://dx.doi.org/10.1016/j.enpol.2017.06.037
Received 21 November 2016; Received in revised form 3 May 2017; Accepted 16 June 2017
Available online 23 June 2017
0301-4215/ © 2017 Elsevier Ltd. All rights reserved.
A. Malhotra et al.
However, there is a lack of studies conducting a systematic assessment
of all possible risks, their respective contributions to financing cost as
well as correlations between the risks (Schmidt, 2014). A better
understanding of risks, correlations, and aggregation strategies can
help private investors and policymakers formulate strategies to aggre-
gate assets in order to attract private investments.
This study, focusing on the role of mini-grids in achieving decen-
tralised rural electrification, addresses the question: How can different
project aggregation strategies be used to enable private sector
investment in renewable energy mini-grids in developing countries?
It presents a first analysis of the derisking effect of portfolio diversifica-
tion on decentralised renewable energy mini-grid projects in the two
Indian states of Uttar Pradesh and Bihar. Specifically, it analyses how
spatial diversity in a portfolio can be used to reduce investment risks
and increase investment volumes. This is done through a mixed-
method study involving (i) identification and classification of the major
investment risks for decentralised renewable mini-grid projects based
on literature and expert interviews, (ii) qualitative analysis of the
correlations between perceived investment risks for mini-grid projects
through structured interviews with investors, project developers and
industry experts, and (iii) estimation of the derisking effects of different
diversification strategies across four levels (from village to national
level) for an experimental portfolio by applying modern portfolio
theory in a bottom-up techno-economic model.
The remainder of this paper is structured as follows: Section 2
reviews the existing literature on aggregation and diversification of
risks in investments for rural electrification. Section 3 describes the
research case for this study. Section 4 describes the methods used, and
Section 5 presents the main findings of the study. The implications for
policymakers, limitations for the current analysis, and avenues for
further research are discussed in Section 6.
2. Literature review
This section provides a review of investment needs and barriers for
mini-grid based rural electrification projects (Section 2.1). It provides
an overview of concepts related to risk diversification, and of studies
applying them to investments in the power sector in general and rural
electrification in particular (Section 2.2).
2.1. Investment needs and barriers for mini-grid based rural
electrification
With the launch of the Sustainable Energy for All (SE4All) initiative
in 2011, the UN General Assembly's declaration of the decade 2014–
2024 as the Decade of Sustainable Energy for All, and the announce-
ment of the Sustainable Development Goals (SDGs), achieving uni-
versal access to modern energy services by 2030 has become a major
goal for sustainable development. Given the unprecedented scale of
this undertaking, several studies have estimated the level of invest-
ments required to achieve this goal. Bazilian et al. (2010) estimate
annual costs till 2030 ranging from 12 to 134 billion USD, with the
wide range largely owing to variation in underlying assumptions such
as the level of electrification, technologies employed, and fuel costs.
The IEA (2011), on the other hand, provides a point estimate of 48
billion USD annual investment. In a more recent report, they note that
actual investments in 2013 only amounted to 13 billion USD, which is
projected to increase over time, averaging 30 billion USD annually
between 2015 and 2030 (IEA, 2015).1 Despite the large variation in
estimates, it is commonly acknowledged that the private sector
investments will play an indispensable role in meeting global targets
1
Regional estimates further highlight the disparity in estimates of investment needs.
See, for example, 2.4–3 billion USD for India estimated by Banerjee et al. (2015); 922
million to 12.7 billion USD for India estimated by CKinetics (2013); 7.8–78 billion
estimated for two extreme scenarios in sub-Saharan Africa by Bazilian et al. (2012).
658
Energy Policy 108 (2017) 657–672
(IEA, 2011; Bhattacharyya, 2013a; UNCTAD, 2014; Pueyo et al.,
2015).
However, rural electrification projects in developing countries,
especially off-grid projects based on renewable power, are often
unattractive for private investors. These projects typically have un-
favourable risk-return profiles and small investment volumes (Schmidt
et al., 2013; Schmidt, 2015; Williams et al., 2015). Further, renewable
based projects, owing to their high up-front cost, are influenced to a
much larger extent by higher financing costs as compared to fossil fuel-
based generation (Schmidt, 2014). Consequently, there has been
considerable attention towards identifying, classifying and understand-
ing risks and measures to address them.
Several studies have discussed investment risks for mini-grid
projects, applying different classification criteria. For example,
Williams et al. (2015) review and classify barriers to private sector
investment into three interdependent categories – financial, institu-
tional and policy, and technical barriers. Similarly, Hazelton et al.
(2014) identify technical, organizational, social, sustainability, finan-
cial, and safety risks for solar PV hybrid mini-grids. Franz et al. (2014)
organise risks at the macro level into political, social, economic, and
financing risks. Ahlborg and Hammar (2014) identify investment
barriers in Tanzania and Mozambique, which they classify into the
categories ‘weak institutions and barriers’, ‘economy and finance’,
‘social dimensions’, ‘technical system and local management’, ‘technol-
ogy diffusion and adaptation’, and ‘rural infrastructure’. Schmidt et al.
(2013) distinguish between risks arising from barriers at the local,
national and international level. Waissbein et al. (2013) classify
investment risks by associating them with a specific stakeholder whose
actions have the potential to result in negative impact on the project
finances, and quantify their impact on cost of financing for renewables.
They also propose policy and financial derisking instruments, i.e.
public derisking instruments aimed at mitigating investment risks
and transferring them to a public institution, respectively. Thus the
existing literature provides a good qualitative understanding of differ-
ent investment risks and barriers for mini-grid projects. However,
these studies generally focus either on single projects, or on barriers in
the investment environment, without analysing how they might affect a
portfolio of projects. Further, none of them indicate the relative
importance of different risks in terms of impact on financing costs
for a diversified portfolio.
2.2. Diversification of risk
Modern portfolio theory (MPT), first introduced by Markowitz
(1952), provides a framework to quantify the risk reduction associated
with diversification of assets in an investment portfolio. According to
MPT, the expected return of a portfolio is the weighted average of the
expected returns of the individual assets. The portfolio risk, defined as
the variance of returns of the portfolio, depends on the variance of the
returns of each of the assets, and on the correlations between the
assets’ returns. Mathematically it can be represented as:
n n n
V (R ) = ∑ xi 2σi 2 + ∑ ∑ xi xj σi σj ρij ,
i =1 i j≠i (1)
where V (R ) is the variance of the expected returns of the portfolio, xi is
the fraction of the total portfolio investment amount invested in asset i,
σi is the standard deviation of the returns of asset i, and ρij is the
correlation coefficient between the returns of assets i and j. When the
returns of individual assets in a portfolio do not perfectly correlate
(i.e. ρij <1), the variance of the returns of the portfolio is lower than that
of the individual assets.
Investment risks can be broadly classified as unique risks and
market risks (see Fig. 1). According to MPT, within one market, only
unique (or project-specific) risks, are not perfectly correlated and may
be addressed by diversification (Brealey et al., 2012). In order to reduce
A. Malhotra et al.
Fig. 1. Unique and market risks for investment portfolios.
market risk, assets from different markets, e.g. other geographies or
industries, need to be included in the portfolio. While aggregation of
electrification projects in developing countries therefore can potentially
decrease unique risks and, depending on the extent of spatial diversi-
fication, market risks, we are not aware of academic studies investigat-
ing portfolio diversification in this context. However, here we discuss
existing studies that apply concepts of portfolio theory to study the
effects of diversification of investments in power generation assets.
Wüstenhagen and Menichetti (2012) provide an overview of exist-
ing applications of MPT to investments in power generation assets.
They include the evaluation of the diversification effect of including
renewable power generation assets in conventional power mixes
(Awerbuch, 1993, 2000; Bhattacharya and Kojima, 2012), identifica-
tion of optimal generation portfolios in the power sector (Roques et al.,
2008), and evaluation of strategies for diversifying plant-specific risks
for renewable energy technologies (Laurikka and Springer, 2003).
While there have been concerns about the use of historical variances
and correlations to make predictions about portfolio performance
(Jobson and Korkie, 1981), especially in project portfolios (Hubbard,
2014), most analyses rely on historical data of costs and electricity
prices not to predict, but to demonstrate the effects of portfolio
diversification.
There is a notable lack of analyses for aggregation of distributed
generation, which could be attributed to several related reasons – the
immaturity of markets leading to lack of substantial historical data
(Alafita and Pearce, 2014); the rate of technological change and policy
uncertainty leading to concerns about the applicability of such data;
and the fact that distributed generation has largely been financed by
the end-users in the case of grid-connected generation, or by grants in
the case of off-grid generation for rural electrification (Bhattacharyya,
2013a; Pueyo et al., 2015). Despite this, with the emergence of new
aggregative financing models for distributed solar in the US (Hess,
2013), there has been an increased interest in investigating the benefits
of project aggregation. Lowder and Mendelsohn (2013) estimate the
potential for aggregation of distributed solar PV generation in the US,
and discuss the potential reduction in financing costs for the asset
originator due to diversification of credit risk and geographic risks.
Mendelsohn and Feldman (2013) calculate the levelised cost of
electricity (LCOE), assuming lower financing costs for aggregative
financing structures due to increased access to low cost capital,
increased liquidity allowing access to a wider class of investors, and
increased diversity of the underlying assets. Alafita and Pearce (2014)
identify and model the influence of some key parameters on the cost of
financing for aggregated residential solar PV assets. They note that
“natural or manmade disasters, decline of regional industry, or change
in state policy” could be arguments for geographic diversification of
659
Energy Policy 108 (2017) 657–672
investments for further risk reduction. Overall, we observe that spatial
diversification across geographies and jurisdictional boundaries is a
common theme touched upon across these studies. However, none of
them analyse in detail the mechanisms by which it can lead to reduced
risk correlations.
Three studies discuss aggregation and diversification in the specific
case of rural electrification (Davidsen et al., 2015; Gershenson et al.,
2015; Orlandi et al., 2016). They identify several advantages of this
approach: First, portfolio diversification results in reduction in invest-
ment risk. Second, investment volumes are increased. Third, the
transaction costs are reduced on a per project basis thanks to the
centralisation of expenses. Gershenson et al. (2015) and Orlandi et al.
(2016) qualitatively assess the risks of off-grid electrification projects
as well as the transaction costs, and propose possible ways to structure
pooling facilities. Davidsen et al. (2015) propose a strategy in which a
pooling facility creates portfolios of mini-grid assets with telecom
towers as base loads, and sells asset-backed securities to capital market
investors. All three studies indicate that risks that are specific to
projects, geographic locations, and jurisdictions can be addressed to
some extent by diversification across multiple projects in different
locations. Yet these studies do not provide a systematic and quantita-
tive assessment of investment risks and how diversification individually
acts on each risk and the overall project financing cost. Furthermore, in
order to make use of imperfect risk correlations for project pooling and
diversification of risks, it is important to understand the causes and
mechanisms for correlation between risks for electrification projects
located at different distances from each other and across different
jurisdictions. Thus we find that there is a lack of studies conducting a
systematic assessment of all possible risks, their respective contribu-
tions to financing cost (Schmidt, 2014), as well as correlations between
risks for projects located in different geographies and jurisdictions – a
gap that we address in this study.
3. Research case and background
3.1. Research case
As the research case for this study, we chose solar PV based mini-
grids in the Indian states of Uttar Pradesh and Bihar. Solar PV based
mini-grids are chosen because of three main reasons. First, according
to the IEA, 40% of the installed capacity required to provide universal
electricity access would need to come from off-grid sources, including
mini-grids, in order to achieve universal electrification by 2030 (IEA,
2011). With their fast declining costs, modularity, no emissions, and
high resource potential, solar PV based mini-grids in particular are fast
becoming a viable alternative to other electrification technologies (see
Table A1; Kempener et al., 2015; Nykvist and Nilsson, 2015). Second,
mini-grids can support higher levels of electrification as compared to
other off-grid technologies, and can create opportunities for productive
use of electricity for income generation activities (Blum, 2013; Pueyo,
2013; Chattopadhyay et al., 2015). However, their high upfront
investment cost and low variable cost (as compared to fossil fuel based
power generation) result in the cost of finance forming a significant
portion of the LCOE in developing countries (Blum et al., 2013;
Ondraczek et al., 2015). Third, we focus only on a single technology
because we are interested in analysing the effects of spatial diversifica-
tion in a portfolio. Theoretically, other strategies such as diversifying
across different technologies, or across projects by different project
developers can also be used to reduce risks. However, the increased
transaction costs due to lack of streamlining and standardisation of due
diligence processes across different technologies and corresponding
business models can be prohibitive (Gershenson et al., 2015).
The study analyses the states of Uttar Pradesh and Bihar in India
because of three reasons. First, India's multi-level governance structure
allows us to study the effects of spatial diversification across different
jurisdictions within the same country. The organisation of institutions
A. Malhotra et al. Energy Policy 108 (2017) 657–672

related to rural electrification in India is complex and overlapping, with 4. Methods

a mix of ministries, agencies, utilities and administrators operating at
the national, state and district level (see Section 3.2). Each of these This study uses a mixed methods approach in order to identify,
stakeholders is capable of affecting different investment barriers, classify, and analyse investment risks for mini-grid portfolios. We use
leading to the possibility of investment risks occurring at different qualitative interview based data and a literature review to analyse
governance levels.2 Second, we analyse Uttar Pradesh and Bihar as they investment risks and their correlations, and quantitative methods to
have a relatively high number of firms active in the solar PV-based explore the potential effects of different diversification strategies on an
mini-grid electrification space (see Table A1 and Davidsen et al., 2015). experimental portfolio.
Third, India, with 263 million people lacking electricity access, has the
highest absolute number of unelectrified households globally (IEA and
4.1. Qualitative analysis of risks and their correlations
World Bank, 2015), with Uttar Pradesh and Bihar accounting for about
180 million people of this total (Banerjee et al., 2015).
In the first step, qualitative methods are used to investigate the
research question. This is because qualitative analysis allows us to
understand the mechanisms by which investment risks for mini-grid
3.2. Background
projects in different locations correlate in greater depth as compared to
quantitative methods (Eisenhardt, 1989; Yin, 2003). First, following
India is a federal state, with power as a “concurrent” subject,
Yin (2003), we used a literature review to develop first hypotheses on
meaning that it falls within the purview of both central and state
the classification of investment risks. For this analysis, it is important
legislatures and governments. This, along with a long history of rural
that investment risks are categorized such that they are mutually
electrification efforts, has resulted in a complex institutional landscape
independent to the greatest extent possible, so that each of them can be
for rural electrification with several actors and institutions at the
analysed independently. Waissbein et al. (2013) propose such a risk
national, state and local levels. At the national level, India's power
classification for large-scale grid-connected renewable energy projects,
sector (including rural electrification) is governed and overseen by the
in which each risk category is associated with a specific stakeholder
Ministry of Power (MOP). The Central Electricity Regulatory
who can potentially affect project finances negatively. In this study, we
Commission (CERC) is responsible for regulation of tariff of generating
apply this classification scheme to investment risks for mini-grid
companies owned or controlled by the central government and
electrification projects.
companies active in more than one state. It also advises the central
Second, we conducted six exploratory semi-structured interviews
government in formulation of National Electricity Policy and Tariff
with Indian and international investors, project developers and in-
Policy. The Ministry of New and Renewable Energy (MNRE) is
dustry experts, each lasting 60–90 min. They were used to refine and
responsible for developing and promoting renewable energy sources
adjust the classification of risks,3 and to understand whether the
in general. Additionally, the Indian Renewable Energy Development
magnitude of correlations of risk categories depends more on the
Agency (IREDA), a non-banking financial institution under the MNRE,
extent of geographic distance between projects, or on diversification
provides term loans for off-grid, renewable energy and energy effi-
across jurisdictional boundaries. A list of potential interview partners
ciency projects. Currently, activities in the power sector are guided by
was compiled on the basis of a literature research and personal
the Electricity Act 2003. Even though the Act was framed with
contacts. In preparation for the interviews, we scanned newspaper
centralised, non-intermittent power generation sources as the primary
articles, case study reports and online company statements for
focus, mini-grids are still recognised as an alternative to grid extension
information related to the company and the interviewee. The informa-
for electricity supply in rural areas. It provides for exemptions for tariff
tion was used to tailor the interview guidelines to the interviewee's
approval and license requirements, and prescribes technical standards
organisation and experience.
and safety measures for mini-grids.
Third, we conducted 12 structured interviews with project devel-
At the state level, the power department of the state government is
opers and investors active in Uttar Pradesh and Bihar, each lasting
responsible for formulating state-level power sector policies. Individual
between 60 and 120 min, in order to deepen our understanding of each
State Electricity Regulatory Commissions (SERCs) regulate the power
one of the risk categories. At least two interviewers conducted the
sector at the state level. State-level institutions for electrification and
interviews, with one interviewer taking notes and recording the inter-
renewable energy (usually known as State Renewable Energy
view proceedings (for a full list of interviews, refer to Table 1).
Development Agencies) serve as nodal agencies for developing, im-
plementing and promoting renewable energy and off-grid electrifica-
tion projects. Implementation of rural electrification projects is usually 4.2. Quantitative modelling of effects of portfolio diversification
the responsibility of the public utilities in the states, which are executed
at the sub-district (or sectional) level (Banerjee et al., 2015). Finally, The structured interviews were also used to get quantitative ratings
private developers need approval from the village-level governing body of likelihood and financial impact of each risk category, and the extent
(the village panchayat) to deploy mini-grid projects. of correlation between projects either distributed across different
Overall, national regulations and policies related to mini-grids are jurisdictions, or at different geographic distances. The method pro-
guided by institutions and legislations at the national level, which are posed by Waissbein et al. (2013) was applied to calculate the attribu-
taken up, adapted, and implemented by institutions at the state level. tion of each risk category to the cost of capital. First, the interviewees
The flexibility of the policy framework in accommodating private mini- rated the likelihood and financial impact of each risk category on a
grids has resulted in the rise a number of for-profit mini-grid Likert scale from 1 to 5. The product of the two ratings gave a measure
developers, largely funded by grants, impact investments and venture of the scoring of each risk category for the model calculations. Second,
capital (see Table A2). However, commercial debt has so far been for data on the correlation ratings, the interviewees stated for each risk
relatively inaccessible for mini-grid enterprises (Interviews; Davidsen category whether the correlation between any two projects depends on
et al., 2015). jurisdictional boundaries or on geographical distance. Then, following
the elicitation methods of Garthwaite et al. (2005) we asked the
2
We follow Fukuyama (2013) in defining governance as “a government's ability to
3
make and enforce rules, and to deliver services”. This can vary across jurisdictions, We limit the scope of our analysis to project developers, investors and industry
depending on the stakeholders involved. Thus, we take both the structures, and the experts since they have operational experience of barriers and risks for investments into
mechanisms of governance into account (Levi-Faur, 2012). off-grid rural electrification projects in this specific context.

660
A. Malhotra et al. Energy Policy 108 (2017) 657–672

Table 1
List of interviews.

Interview type Interview Stakeholder Interviewee Interview codea

number

Exploratory 1 Project developer Executive director of mini-grid company active in Tanzania DEV
interviews 2 Investor Co-founder and managing director of a non-profit impact investor firm with INV
clean energy assets in developing countries
3 Industry expert Senior researcher in the field of electrification and small-scale renewable energy EXP
4 Industry expert Consultant for small-scale renewable energy projects and rural electrification EXP
5 Industry expert Executive director of industry association and policy advocacy for the EXP
decentralised clean energy sector
6 Public donor Advisor of public development aid organisation PUB
Structured 7 Project developer and Researcher on renewable mini-grids, and founder and executive director of DEV
interviews industry expert mini-grid company
8 Project developer Director of a mini-grid company DEV
9 Project developer Executive director of a mini-grid company DEV
10 Project developer Managing director and investment lead of mini-grid company (2 interviewees) DEV
11 Project developer Executive director of a mini-grid company DEV
12 Project developer Former head of business development of a mini-grid company DEV
13 Project developer Director of a mini-grid company DEV
14 Project developer Director of a mini-grid company DEV
15 Investment advisor Director of an investment consulting company INV
16 Public investor Deputy general manager (Renewable energy), assistant general manager INV
(Corporate planning) and assistant manager (Corporate planning) of a public
debt investment company (3 interviewees)
17 Impact investor Regional investment manager of nonprofit debt investor INV
18 Industry expert Chairman and principal analyst of charitable foundation and policy advisory EXP
supporting and researching the mini-grid sector (2 interviewees)

a
DEV = Project Developer; INV = Investor; EXP = Industry Expert; PUB = Public donor.

interviewees to rate the correlations of the individual risk categories
between two projects located at varying degrees of separation in terms
of jurisdictions and geographic distances on a Likert scale from 1 to 5
(for survey questions refer to Table B1 in Appendix B).4 The survey
data thus obtained is a proxy for expectations of future asset perfor-
mance. While the precise values of actual risk correlation coefficients
cannot be obtained by such a method, the mean of the elicited values
taken relative to each other can certainly be used to illustrate the
efficacy of different spatial diversification strategies in order to address
specific risks. Finally, data on cost of financing for mini-grid projects
was obtained from the interviewees, which was triangulated with other
data sources obtained through desk research.
In the second step, to calculate the effect of spatial diversification on
risks, we calculated the variance of the returns of the individual project,
and correlations between different projects distributed across different
jurisdictions, or at different geographic distances. The correlation of
returns ρij of two assets i and j was calculated as:
n t is the year and WACC is the weighted average cost of capital.8
ρij = ∑ xr ρij,r
r =1 (2)
where xr is the score of risk r (the product of the likelihood and
financial impact ratings obtained from the interviewees) and ρij, r is the
correlation of risk r between projects i and j . These values were used to
calculate the standard deviation of the returns of the overall portfolio,
and the resulting cost of financing, assuming that the financing cost
depends linearly on the standard deviation within this range.5 A
conservative estimate of the new risk premium was calculated as:
1 ⎛ 1⎞
rp, a ≥ ra + ⎜1 − ⎟ ra ρav, a
n ⎝ n⎠ (3)
where rp, a and ra are the risk premium of risk category a for a portfolio
and for one project, respectively. n is the number of projects in the
portfolio and ρav, a is the average correlation of risk category a between
4
It was assumed that the risk categories do not have negative correlations.
5
This is an approximation which we use for high-risk assets with small changes in
standard deviation of returns resulting from aggregation. In general, the cost of capital is
seen as a concave upward investor utility function of the risk (Lintner, 1965).
all the projects in the portfolio.6 In order to estimate the impact of
uncertainty in the elicited correlation coefficients, we modelled them as
stochastic inputs in a Monte Carlo simulation. We used a PERT
distribution to model the correlation coefficients,7 using the minimum,
mean, and maximum values elicited in the survey as inputs for the
parameters (minimum, most likely, and maximum value) for the PERT
distribution. Each Monte Carlo simulation was conducted with 10,000
iterations.
We calculated the pre-tax LCOE for a single mini-grid project and
for the mini-grid portfolio. We further calculated the pre-tax LCOE for
a mini-grid powered by a diesel generator to serve as a baseline for
comparison. The LCOE was calculated as:
T Expenditures
CAPEX + ∑t =1 (1 + WACC )tt
LCOE = Electricitygeneratedt
T
∑t =1 (1 + WACC )t (4)
where CAPEX is the initial capital expenditure, T is the project lifetime,
Assumptions related to the technological parameters, cost assump-
tions, loads and capacities for various mini-grid components are
outlined in Appendix D.
5. Results
In this section we first present the qualitative results of the analysis of
the identified risk categories. We describe in detail the mechanisms by
which the risks are correlated. Second, we present the results of the
quantitative analysis, which indicate the contributions of different risks to
the cost of financing, and the effects of diversification across jurisdictional
boundaries and geographic distances in terms of risk reduction.
6
Eq. (3) applies to the specific case of equal standard deviations and equal shares of
investment volume for each project in the portfolio. The derivation is presented in
Appendix C.
7
The PERT (Programme Evaluation and Review Technique) distribution is particu-
larly suited for modelling of expert elicitations (Brown et al., 2001; Vose et al., 2016;
Battke et al., 2013).
8
In this study, the WACC is equal to the cost of equity, since each project is assumed to
be 100% equity financed.

661
A. Malhotra et al. Energy Policy 108 (2017) 657–672

Table 2
Risk categories for renewable energy mini-grid projects identified through literature review and exploratory interviews.

Risk category Description Key stakeholder group

Power market risk Risk arising from limitations and uncertainty in the energy market (off- and on-grid) Power market-related policymakers (civil servants);
regarding market outlook, price, access and competition legislators; regulators
Permits risk Risk arising from the public sector's inability to efficiently and transparently administer Permit administrators
mini-grid-related licensing and permits
Technology sourcing Risk arising from limitations in the quality and availability of mini-grid hardware, as well Technology supply chain; technical regulator; customs
risk as the customs treatment of hardware (excise)
Financing risk Risks arising from scarcity of domestic investor capital (debt and equity) for renewable Domestic investors (equity and debt), financial sector
energy, and domestic investors' lack of familiarity with renewable energy and regulator
appropriate financing structures
Currency risk Risks arising from currency mismatch between hard currency debt/equity and domestic No specific stakeholder
currency revenues
Grid extension risk Risks arising from uncertainty in grid extension plans and technical regulations for Utility, distribution company; utility/grid regulator
integration into the main grid
Social acceptance risk Risks arising from lack of awareness and resistance to renewable energy and mini-grids General public; NGOs
in communities
Payment and credit Risk arising from customers' willingness, ability, and methods of payment for electricity End users; consumer finance actors (consumer banks,
risk credit data actors, end-consumer finance regulator)
Sovereign risk Risk arising from a mix of cross-cutting political, economic, institutional and social No specific stakeholder
characteristics in the particular country which are not specific to mini-grids
Labour inputs risk Risks arising from the lack of skilled and qualified potential employees Labour force, educational institutions
Developer risk Risks arising from limitations in the developer's capability to efficiently and effectively Mini-grid developer (BOO)
design, install, operate, maintain and monitor the project

Fig. 2. Schematic diagram of the jurisdictional boundaries and geographic distances across which a mini-grid portfolio needs to be diversified to address specific risks. For projects
located across the jurisdictions and geographic distances below the arrows, the correlation of the specific risk category is maximum and constant.

5.1. Qualitative results: Investment risks for mini-grid projects and
their correlation mechanisms
As described in Section 3, we used an adapted version of the
classification scheme proposed by Waissbein et al. (2013) to classify
investment risks into categories based on the primary stakeholder with
which they are associated (Table 2 summarises the risks). We then
describe the correlations of these risks for different projects, and their
dependence on geographic distances and jurisdictions. These findings
are schematically represented in Fig. 2.9
9
Developer risk was excluded from the qualitative analysis and from Fig. 2, since its
correlation is independent of spatial dimensions.
5.1.1. Risks with correlations depending purely on jurisdictions
Power Market Risk was generally perceived to have a high
impact on financing costs. As described in Section 3.2, electrification
targets, legislations, and policies related to planning, regulation and
provision of incentives for electrification projects are decided at the
national level in India. The Ministry of Power currently targets to
provide electricity access to every household in India by 2022, and
every household in Uttar Pradesh by 2019. However, the lack of clarity
regarding the role of mini-grids in the target creates uncertainty in the
market outlook for project developers and investors (Gram Oorja,
2013). Under Section 8.5 and 8.6 of the Rural Electrification policy of
2006, mini-grids are exempt from tariff approvals by regulators and
can set the retail tariffs based on mutual agreements with customers.
Under Section 8.5 of the Policy, mini-grids are subject to technical

662
A. Malhotra et al.
standards and safety measures implemented by the Central Electricity
Authority. Given that the current mini-grid enterprises are operating
under this regime, ongoing discussions regarding a national mini-grid
policy (Bhattacharyya and Palit, 2016) mean that risks arising from
changes in policies and regulations affect all projects located within the
country, contributing to risk correlation at the national level. In
addition, state power departments and nodal agencies have jurisdiction
over state-specific deployment policies, and state regulatory commis-
sions each set their own regulations and electricity tariffs. Uncertainties
regarding the mini-grid policy for the state of Uttar Pradesh under
discussion by the Uttar Pradesh New and Renewable Energy
Development Agency (UPNEDA) at the time of the interviews were
perceived to affect all projects within the state equally (INV, DEV).10
Therefore, power market risk was perceived by the interviewees to
correlate highly between projects within one state, since any policy or
regulatory changes at either the state or national level are equally likely
to affect all projects within a state. For the same reason, the correlation
is lower for projects located in different states.11
Permits Risk was generally perceived to have a low impact on
financing costs. Administration of permits was found not to be an issue
in India since Section 14 of the Electricity Act of 2003 specifies that
operators for generation and distribution in rural areas are exempt
from any license requirement, except in special cases where they need
to be deployed in protected areas such as forests (DEV; Palit et al.,
2014). Even though the risk itself was perceived to be low, the
interviewees indicated that in case permits would be needed from
different authorities in the future, correlation would be high within a
district, since this would be handled at the district level. As observed by
one interviewee, “If you are looking at a district, then it is the same
district administration. So for different districts you may even say that
the district administration is easier to deal with than another, but I
think across the district it's likely to be the same” (EXP). Three
interviewees highlighted the human aspect of the permitting process,
pointing out that since different officers would administer the permits,
even within a district, correlation would not be perfect (DEV, EXP).
However, it was commonly agreed that correlation would reduce by
diversifying across districts and states.
Technology Sourcing Risk was generally perceived to have a
moderate impact on financing costs. Several interviewees highlighted
the homogeneous technology market within India, implying equal but
imperfect correlation of technology sourcing risk across different
jurisdictions. According to one interviewee, “When it comes to procur-
ing technology, [the market] is the same for every state” (DEV). One
interviewee mentioned a potential gap between rural and urban areas
implying that, if a developer in a rural area faces problems procuring
the necessary technology, a developer in an urban area may not face the
same problem (EXP). This rather geographical aspect was however
dismissed by other interviewees referring to the low transportation cost
within India.
Financing Risk was generally perceived to have a high impact on
financing costs. Most interviewees rated financing risk to be similar in
the whole country and to highly correlate across all jurisdictions within
India (INV, DEV). However, there are some perceived differences
across states due to differences in maturity of financial sectors and
proximity to major cities with concentrations of human and financial
resources. This reduces the correlation of risk arising from capital
10
This is exemplified by the observation by an interviewee: “Power Market is pretty
clearly a state and national issue. […] The more conventional risks around subsidies and
tariff regulation are likely to be set at the national level and some are set at the state level”
(DEV).
11
Unlike the opinion of the majority of the interviewees, one interviewee rated the
correlation of Power Market Risk to depend on geographical distance rather than
governance levels and to be almost zero at distances of 20–150km. He stated, “We are
active in such tiny villages that are not even considered villages. State legislation hence
does not have an influence on our activity. I therefore see correlation only for very close-
by projects.” (DEV).
663
Energy Policy 108 (2017) 657–672
scarcity across different states to a small extent.
Currency Risk was generally perceived to have a moderate impact
on financing costs. This risk category applies in cases where equipment
is imported and the purchase is made in foreign currency, and in cases
where investments are made in a foreign currency, while revenues are
generated in the local currency. Since India has one currency, the
Indian Rupee, currency risk correlates perfectly for projects located
across all jurisdictions within the country. The risk can potentially be
addressed by diversifying the portfolio with projects from different
countries and therefore different currencies (INV, EXP). However, the
correlation can only be partially reduced even when diversifying across
national borders since different currencies may move similarly depend-
ing on the global economy.
5.1.2. Risks with correlations depending on both jurisdictions and
geographic distance
Grid Extension Risk was generally perceived to have a high
impact on financing costs. In India, the planning and execution of grid
extension is the outcome of a complex process involving several
stakeholders at the national and state levels, creating a high degree
of uncertainty regarding which projects might be affected (Chandran-
Wadia et al., 2015). The interviews revealed that the correlation of grid
extension risk between projects depends on jurisdictions as well as
geographical distances, since it is influenced by both political factors
(electoral cycles and the political will of the state government) and
geographic factors (proximity to pre-existing grid). Formulation of
electrification strategy and allocation of financial resources for grid
extension is done at the national and state level. State utilities plan and
execute grid extension projects, dividing the states into sections, which
are at the sub-district level (EXP). While theoretically the utilities are
independent entities, free to pursue their own grid extension plans, the
practical implementation is the outcome of a political process in which
local politicians may, as one interviewee put it, “decide to invest in grid
extension to win elections” (DEV; Baskaran et al., 2015). This is
exemplified in the observation that “The actual effect on your portfolio
of micro-grids is going to be felt at the district or village level where the
physical build-out of the grid is occurring […], but the decision was
made at the state or national level” (DEV). Considering the geographic
dimension, correlations were found to be high at small geographic
distances. Since grid extension occurs in batches of several villages,
extension of the distribution grid to one village makes it more likely
that a neighbouring village is likely to be electrified as well (DEV). This
is supported by the findings of Kemmler (2007) and Oda and Tsujita
(2010), who show that proximity to the grid and subsequent access to
grid electricity are highly correlated.
Social Acceptance Risk was generally perceived to have a
moderate impact on financing costs. Though most interviewees per-
ceived barriers due to resistance from local communities as a minor
issue, they agreed that social acceptance risk correlates at a local level
due to factors related to both jurisdictions and geographical distances.
Hierarchical structures associated with informal and formal institu-
tions are important, and village heads and district politicians can be
influential in determining social perceptions of renewable technologies
and mini-grids (EXP), leading to high correlation within a village or
even district. The geographical dimension arises because in case there
are negative perceptions towards mini-grids in one location, “word of
mouth can travel to spread negative perceptions” (DEV). At short
distances, “the market is very homogeneous” (DEV) and perception of
renewables amongst villagers very similar. Correlation of social accep-
tance risk was therefore considered high at small geographic distances
and within lower governance levels (villages and districts). With
increasing diversification across jurisdictions and geographical dis-
tances, the correlation between projects reduces significantly.
Payment and Credit Risk was generally perceived to have a
moderate impact on financing costs. It was also found to correlate due
to factors related to jurisdictions as well as geographical distance.
A. Malhotra et al.
Potential mini-grid customers in rural areas are mainly dependent on
agriculture as their source of income. In case a customer is unable to
pay electricity bills due to a bad harvest, it is very likely that other
customers in nearby areas engaged in agriculture will be affected as
well, making payment risk highly correlated in the entire region.
Additionally, local communication channels play an important role in
increased correlation of non-payment at small geographic distances.
This is highly influenced by “…word of mouth. If the neighbouring
village doesn't pay, they also don’t want to pay” (DEV). The correlation,
however, reduces with increasing distance between projects. To a lesser
extent, the correlation of payment and credit risk across jurisdictions
arises from the collective behaviour (DEV) of villagers to whom local
politicians or village heads set an example. If one of them has a low
willingness to pay for mini-grid based electricity, it is likely to influence
the willingness to pay for customers within the village (EXP).
Sovereign Risk was generally perceived to have a moderate
impact on financing costs. Sovereign risk includes barriers arising
from general political, economic, and geographic conditions, which are
not specific to mini-grids and may lead to negative financial impact for
any investment in a country. This includes risk arising due to conflict,
political instability, legal governance, economic performance and
natural disasters, which are found to correlate across jurisdictions as
well as geographical distances. With the risk spanning from local up to
national events, the correlation is high at the local level (district level as
well as neighbouring projects located less than 20 km away from each
other) and decreases when diversifying across jurisdictional and
geographical dimensions.
5.1.3. Risks with correlations independent of spatial diversity
Labour Inputs Risk was generally perceived to have a moderate
impact on financing costs. The interviewees generally did not perceive
labour inputs risk to be addressable by diversification across either
spatial dimension. While labour for basic electrical and civil work is
abundantly available in most places, more skilled labour may be harder
to acquire in some areas. Given the early stage of market development
for mini-grids, all project developers without exception have to under-
take in-house training for new employees for installation, operation
and maintenance of mini-grids. One interviewee rated the correlation
to be high in the entire country because “[rural] areas have similar
demographics. If one area needs education, another will need educa-
tion as well” (PUB). Another interviewee stressed on the “urban-rural
dichotomy” (EXP) with high correlations for areas equally far from
urban centres and decreasing when moving closer to cities.
Fig. 2 schematically summarizes the results of the qualitative
analysis.
5.2. Quantitative results: Indicative calculation for an experimental
portfolio
This section presents the results of a model which demonstrates the
potential effects of spatial diversification for different portfolios,12
showing the effects of increasing portfolio size and spatial diversifica-
tion on portfolio risk and cost of equity. We present the contributions
of the individual risks to the cost of equity, and the LCOE, before and
after diversification.
Fig. 3 indicates the reduction in cost of equity resulting from
increasing the portfolio size (Fig. 3a), with all projects in a given
portfolio distributed across different jurisdictions (Fig. 3b), geographic
distances (Fig. 3c), and both jurisdictions and geographic distances
(Fig. 3d). Fig. 3a shows the exponential decay of the risk with
increasing portfolio size at a single location, resulting in a reduction
in the cost of equity from 21.7% for one project to 19.6% for a portfolio
12
Interview responses used as inputs for the model are summarized in Table E1 and
Table E2.
664
Energy Policy 108 (2017) 657–672
of 10 projects. Assuming market risk to be approximated by the risk of
a portfolio with 1000 projects, unique (or diversifiable) risks were
found to account for 2.5% points (pp) of the cost of equity.
Fig. 3b shows the effect of diversification across jurisdictions on the
risk. As expected, the unique risk reduces as the number of projects
increases. The contribution of market risk to cost of equity reduces by
0.3 pp and 0.9 pp when the projects are located in different districts of
the same state and in different states, respectively. A noticeable aspect
of diversification across jurisdictions is the fact that the incremental
derisking effect of diversification across different districts is smaller
than the incremental derisking effect of diversification across different
states. This reflects the Indian governance structure in which states are
responsible for the implementation of most policies concerning the
electricity sector whereas districts only have minor executive and no
legislative power.
The derisking effect of geographic diversification is displayed in
Fig. 3c. The reduction in market risk is higher in this case, with a
reduction of 1.3 pp and 2.2 pp for portfolios with projects spread across
distances of 20–150 km and greater than 150 km, respectively. Thus
diversifying across moderate distances (20–150 km) has a higher
incremental derisking effect than high distances ( > 150 km). Even
though more risk categories correlate due to jurisdictions than
geographical distances, the correlations of geographic risks are com-
parably lower and contribute more to overall risk reduction.
Finally, Fig. 3d shows the combined derisking effects of the
different aggregation strategies including increasing portfolio size,
diversification across jurisdictional boundaries and geographic dis-
tances. Spatial diversification can hence result in a maximum theore-
tical reduction in cost of equity from 21.7% to 16.4%. This is not
realisable in practice since each project is in a different state and at a
distance of greater than 150 km from every other project.
In addition, we calculate the cost of equity for a ‘moderate scenario
portfolio’ with 10 projects. Five of the projects are located in one
district of one state within a distance of 20 km from each other. The
other five are located in one district of a different state, also within a
distance of 20 km from each other, but greater than 150 km away from
the first five projects. The combined effect of increased portfolio size,
and diversification across jurisdictional boundaries and geographic
distances in this moderate scenario results in a cost of equity of 17.8%.
Fig. 4 illustrates the derisking effect and hence the reduction in cost
of equity, and the extent of the individual risk reductions for the
‘moderate scenario portfolio’. Six risk categories (power market risk,
permit risk, technology sourcing risk, financing risk, currency risk and
sovereign risk) correlate at the national or state level, which means the
reduction in correlation by diversification across two states, and hence
the effectiveness of diversification across jurisdictional boundaries is
relatively low. In contrast, three risk categories (grid extension risk,
social acceptance risk, and payment and credit risk) correlate mainly
across distance with their source often being very local. The rate at
which the correlation reduces with increasing distance between two
projects of these three risks is hence relatively high and geographic
diversification across small distances is an efficient tool for risk
reduction. Thus even though the number of risks addressed by
geographic diversification is lower, in this case it is overall more
effective than diversification across jurisdictional boundaries.
The results of the LCOE calculations described in Section 4.2 are
illustrated in Fig. 5.13 The LCOE of solar PV-battery mini-grids in Uttar
Pradesh was found to be 0.73 USD/kWh for one project and 0.63 USD/
kWh for the portfolio described above. This corresponds to a reduction
of 13.8% in the LCOE, making solar PV based mini-grids more
competitive with conventional diesel generation based mini-grids.
13
We present the results of LCOE calculations for two additional scenarios in
Appendix F as sensitivities. One scenario incorporates state investment subsidies in
the capital structure, and the second scenario models the impact of incorporation of debt
in the capital structure of the diversified portfolio.
A. Malhotra et al. Energy Policy 108 (2017) 657–672

a. 23.0 b. 23.0

22.0 22.0

21.0 21.0
Cost of equity [%]

Cost of equity [%]

20.0 20.0
Diversify across 19.0
19.0 jurisdiconal
boundaries 18.0
18.0
17.0
2.0
Same district
1.0 1.0 Different district, same state
Same district, < 20 km Different state
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 Number of projects 0 1 2 3 4 5 6 7 8 9 10 Number of projects
Diversify across

Diversify across
geographic

geographic
distance

distance
c. d.
23.0 23.0

22.0 22.0

21.0 21.0

20.0
Cost of equity [%]

Cost of equity [%]

20.0

19.0 Diversify across 19.0

jurisdiconal
18.0 boundaries 18.0

17.0 17.0

16.0 < 20 km 16.0 Same district, < 20 km

20-150 km Different district, same state, 20-150 km
1.0 1.0
> 150 km Different state, > 150 km
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 Number of projects 0 1 2 3 4 5 6 7 8 9 10 Number of projects

Fig. 3. Derisking effect of increasing portfolio size and spatial diversification on cost of equity.

Fig. 4. Derisking effect of spatial diversification on individual risk categories. The labels above the individual risk categories indicate the reduction in risk premium in the moderate
portfolio scenario. The residual risk premium is the component of the risk premium remaining after spatial diversification.

6. Conclusions and policy implications
Off-grid energy sources will play a significant role in achieving
universal electricity access in the near future. However, lack of
financing – due to unfavourable risk-return profiles, small investment
volumes, and absence of appropriate financing sources and structures
– continues to be one of the major bottlenecks in scaling up their
deployment. Previous studies such as those by Gershenson et al. (2015)
and Davidsen et al. (2015) have discussed the potential benefits of
aggregation of mini-grid investments, primarily focusing on the
resulting reduction in transaction costs for private investors, increased
access to institutional capital, and diversification of risks. Our key

665
A. Malhotra et al.
-13.8%
0.8
0.73
0.7
0.63
0.61
0.6
Pre-tax LCOE [USD/kWh]
0.5
0.4
0.3
0.2
0.1
0.0
LCOE Diesel LCOE LCOE
mini-grid Single Porolio
Project
Fig. 5. Reduction in LCOE resulting from spatial diversification.
findings indicate how aggregation of mini-grid assets can play a
significant role in addressing investment risks (Fig. 3a). Further,
aggregating mini-grid assets strategically into diversified portfolios
can be even more effective at addressing risks (Figs. 3b, c, d). In this
section, we discuss three aspects related to the analysis carried out in
this study that need to be taken into consideration. First, risk-return
profiles for mini-grids can potentially be made more attractive through
policies aimed at facilitating or instituting structures for aggregation of
projects into diversified portfolios. Second, policymakers can help
mitigate risks arising from experimentation and learning required to
develop and adapt these policy measures by facilitating diversification
of investments across jurisdictions. Third, while designing deployment
policies to attract private investment into mini-grids, policymakers
need to assess how diversification interacts with other derisking
measures – namely risk mitigation and risk transfer – and how they
can be effectively combined.
6.1. Institutions for aggregating investments into mini-grids
Our analysis indicates that the high cost of financing solar PV based
mini-grids (especially when compared to the financing cost for large
scale grid-connected renewables – see Nelson et al., 2012), can be
addressed to some extent by aggregation and diversification of invest-
ments across different geographic distances and jurisdictional bound-
aries. Policymakers at the national level can utilise this by taking
measures to enable diversification of investments within the country.
For instance, granting of concessions or provision of franchisee
contracts to mini-grid developers under national rural electrification
policies should be done across geographic distances and jurisdictional
boundaries to incentivise the creation of a spatially diverse market for
rural electrification, providing opportunities for intermediaries or
investors to structure spatially diverse portfolios.
International organizations and development finance institutions
(DFIs) can potentially support such efforts through measures aimed at
supporting the creation of structures for aggregation of mini-grids into
spatially diverse portfolios, or through the direct implementation of
such initiatives. For instance, the United Nations Development
Programme (UNDP), Global Environment Facility (GEF), Climate
Bonds Initiative (CBI) and Inter-American Development Bank (IDB)
are setting up the Climate Aggregation Platform (CAP) to facilitate
financial aggregation of distributed grid-connected renewable energy
generation sources by increasing information flow, developing stan-
dardised toolkits (template contracts, performance metrics, transaction
structures) and creating awareness via demonstration programs.
Energy Policy 108 (2017) 657–672
Similar measures such as supporting standardisation of contracts and
95% error bar legal documentation, and sharing of best practices for mini-grid
policies can support intermediaries in aggregation of investments,
and even in structuring of asset portfolios across multiple countries
(which, as indicated in Fig. 2, can further reduce risk correlations).
Thus public efforts devoted to enable structuring of diversified asset
portfolios by mini-grid developers or third parties such as DFIs can
benefit both investors and end-users: because of reduced risks and
hence increased interest from a wider range of more risk-averse
investors, and because of reduced cost of electricity provision resulting
from lower financing cost.
6.2. Supporting policy experimentation and learning
Our results show that there is a high level of uncertainty associated
with barriers arising from public sector stakeholders (leading to power
market risk, permit risk, grid expansion risk), which can be addressed to
some extent by diversifying investment portfolios across jurisdictional
boundaries. It has been previously discussed in literature that policy
intervention is necessary for regulation and scaling up of mini-grids: First,
mini-grids can be seen as a natural monopoly which need regulation to
protect customers and ensure quality of service (Bhattacharyya, 2013b).
Second, they are complex systems and have the potential to benefit from
deployment policies due to learning-by-using (Blum et al., 2015).
However, policy frameworks for private sector mini-grids are still at a
nascent stage, with relatively few examples to be found. In such situations,
it has been proposed that policy experimentation by state and local
governments might be critical to promote policy learning and adaptation
in the specific context (Shipan and Volden, 2008), including policies
aiming to promote technological innovation (Chaminade et al., 2011).
However, when such policies take the form of deployment policies, the
inherent uncertainty associated with the experimentation process also
creates negative effects in the form of investment risks. Policymakers can
address this trade-off by enabling portfolio diversification across jurisdic-
tions, and by transferring risks to the public sector through financial
derisking. This could be done by international agencies with mini-grid
portfolios spanning several countries, each with country-specific mini-grid
deployment policies in place. This could also be taken up by policymakers
at the national level in nations with regional or state-specific policy
frameworks for mini-grids such as those in India.
6.3. Risk diversification as a complement to other public derisking
instruments
In our analysis, we focus on portfolio diversification as a means of
addressing investment risks. But as seen in Fig. 4, there are still
residual risks, which can potentially be addressed by policy and
financial derisking instruments discussed in Section 2.1. These differ-
ent approaches can complement each other in two ways. First, while
policy derisking measures can be very effective at addressing risks by
systematically removing the underlying barriers (Waissbein et al.,
2013; Schinko and Komendantova, 2016), most policy derisking
instruments require several years to take effect (Schmidt, 2014).
Furthermore, their viability depends to a large extent on the presence
of strong policy commitment and institutional capacity (Carafa et al.,
2015). Thus in the short term, diversification across jurisdictional
boundaries (in combination with financial derisking instruments) can
be effective in catalysing investments by addressing risks arising from
policy uncertainty, before policies conducive to private sector invest-
ment can stabilise and take effect. Second, aggregation of projects can
act as an enabler for financial derisking instruments. Such instruments
do not address underlying barriers, but transfer them to public actors,
catalysing private sector investments in the short term. However,
instruments such as loan guarantees and political risk insurance can
have high transaction costs, which can be justified by achieving larger
investment scales through project aggregation.
666
A. Malhotra et al. Energy Policy 108 (2017) 657–672

6.4. Limitations and future research
The results of our analysis indicate that aggregation and spatial
diversification of renewable energy mini-grid projects can be effective
in addressing investment risks and hence lowering the cost of finan-
cing. We demonstrate how the correlation of investment risks between
individual projects in a portfolio depends on geographic distances and
jurisdictions. It is, however, a first attempt at estimating these effects,
and there are some promising avenues for future research on aggrega-
tion and diversification of investments into rural electrification pro-
jects. While this study has focused on spatial diversification, future
studies could investigate diversification across different technologies,
business models, and project developers. Further, future research could
guide policymakers about the specific policy instruments and designs
required to enable aggregation of other rural electrification technolo-
gies such as solar home systems, within and across different country
contexts. Finally, country case studies analysing the trade-off between
risk diversification and increased transaction costs for different aggre-
gation strategies could help better inform policymakers about how to
incentivise aggregation by lowering transaction costs efficiently.
Acknowledgements
We gratefully acknowledge the support of our interview partners.
We would like to thank Hande Bayraktar and Christoph Henrich for
their valuable inputs to this study. We would also like to thank
participants of the INOGOV Workshop on “Climate Policy Innovation
and the Access to Clean Energy Technology in Developing Countries”
held on 27th May 2016 in Geneva, Switzerland, and two anonymous
reviewers for their valuable feedback.

Appendix A

Table A1 provides an overview of private sector mini-grid developers active in India. The data has been collected from interviews, company
websites, and news articles. The relatively high activity of mini-grid developers in Uttar Pradesh and Bihar, and the prevalence of solar PV based
mini-grids has guided the research case selection presented in Section 3.1.
Table A2 provides an overview of some private sector investors with exposure in the mini-grid sector in India. The data has been collected
through interviews, company websites, news articles and industry reports.

Table A1
Sample of private mini-grid developers in India.

Mini-grid developer States active Technology System size (Nominal generation capacity in
kW)

Mera Gao Power Uttar Pradesh Solar PV 0.12–0.25

OMC Power Uttar Pradesh Solar PV, biogas, diesel 25–100
Naturetech Infra Uttar Pradesh Solar PV 1–3
Gram Oorja Karnataka, Maharashtra, Uttar Pradesh, Jharkhand, Solar PV 8–10
Haryana
Husk Power Uttar Pradesh, Bihar, Uttarakhand Biomass gasification, solar PV 1–35
Kuvam Microgrid Uttar Pradesh, Bihar Solar PV 1–3
Mesh Power Bihar Solar PV 1
DESI Power Bihar Biomass gasification, solar PV 0.5–30
Gram Power Rajasthan Solar PV 1–5
Mlinda West Bengal Solar PV 0.2–8

Table A2
Sample of investors in mini-grid enterprises in India.

Investor Investor type Investee

Campanille Impact Fund Venture Capital Gram Power

Vestergaard Fransen Angel Investor Gram Power
Tenshi Peak Angel Investor Gram Oorja
Khattar Holdings Angel Investor OMC Power
Energy Investment Tech Energy Investment Fund OMC Power
Pte. Ltd.
Acumen Equity Fund Husk Power
LGT Venture Philanthropy Impact Investor Husk Power
ICCO Investments Impact Investor Mera Gao Power
Insitor Impact Fund Impact Investor Mera Gao Power
DOEN Foundation Impact Investor SELCO
IFC International DFI Husk Power
Intellegrow Non-Banking Financial Husk Power
Company (NBFC)
IFMR Capital NBFC DESI Power
ICICI Bank Commercial Bank DESI Power

667
A. Malhotra et al. Energy Policy 108 (2017) 657–672

Appendix B

See Appendix Table B1

Table B1
Structured interview and survey questions used in the study.

Type Question

Scoring of risk categories On a scale of 1–5 (with 1 meaning unlikely and 5 meaning very likely), how would you rate the probability that the events
underlying the particular risk occur? Why?
On a scale of 1–5 (with 1 meaning low impact and 5 meaning high impact), how would you rate the financial impact of the
events underlying the particular risk, should the events occur? Why?

Rating of risk correlations across jurisdictions
Imagine that the events underlying the particular risk occur in project A, resulting in a financial loss. On a scale of 1–5 (with 1
meaning no change and 5 meaning high increase), how would the probability of the events underlying the particular risk change
• for a project in the same district?
• for a project in a different district, but in the same state?
• for a project in another state?
Why?

Rating of risk correlations across geographic Imagine that the events underlying the particular risk occur in project A, resulting in financial loss. On a scale of 1–5 (with 1
distances meaning no change and 5 meaning high increase), how would the probability of the events underlying the particular risk change
• for a neighbouring project, less than 20 km away?
• for a project about 150 km away?
• for a project much further than 150 km?
Why?

Appendix C. Derivation of Eq. (3)

The risk premium for a portfolio of assets can be expressed in terms of the risk premia of the individual assets, the average risk correlation
between all the assets in the portfolio, and the number of assets in the portfolio. Starting with the equation for the variance of returns of a portfolio,
n n n
V (R ) = ∑ xi 2σi 2 + ∑ ∑ xi xj σi σj ρij ,
i =1 i j≠i (C1)
we make the following simplifications and assumptions:

• Each asset in the portfolio comprises an equal share of the total amount invested, i.e., xi =
1
for i ∈ {1,…, n}.
•
n
The variance of the returns of each of the assets in the portfolio is equal, i.e., σ f = σ 2 for i ∈ {1,…, n}.
2

• The risk premium is linearly proportional to the variance of returns, i.e., r = k (σ 2 − σ f2 ), where r is the risk premium, σ f2 is the variance of a mini-
grid investment in a low-risk environment.

Using these assumptions, we can rewrite Eq. (C1) as:

n
⎛ 1 ⎞2 2 n n
⎛ 1 ⎞2 2
V (R ) = ∑ ⎜⎝ ⎟σ + ∑ ∑ ⎜⎝ ⎟ σ ρij ,or
i =1
n⎠ i j≠i
n⎠

⎛1 ⎞2
rp + kσ f2 = ⎜ ⎟ (n + n (n − 1) ρav )(r + kσ f2 )
⎝n⎠

1 ⎛ 1⎞ ⎛ 1⎞
rp = r + ⎜1 − ⎟rρav + kσ f2⎜1 − ⎟(ρav −1)
n ⎝ n⎠ ⎝ n⎠

Since the second term on the right hand side is always negative, we can write:
1 ⎛ 1⎞
rp ≥ r + ⎜1 − ⎟ rρav
n ⎝ n⎠

Appendix D

Table D1 shows the assumed loads for the village mini-grid. They have been adapted from the load profile used for a basic electrification scenario
by Blum et al. (2013) using inputs from interviews with mini-grid developers to suit the context of our case study. The resulting load profile for the
mini-grid is shown in Fig. D1.
Table D2 and D3 show the cost assumptions used for the LCOE calculations for the diesel and the solar PV based mini-grids. The nominal
capacity of the diesel generator is determined by calculating the peak load for the mini-grid and adding a safety margin of 20% to it. The capacity of
the solar PV module and battery is determined by optimizing their values such as the LCOE is minimized under the constraint that the system is
capable of meeting the demand in 90% of the cases.

668
A. Malhotra et al. Energy Policy 108 (2017) 657–672

Table D1
Assumptions for diesel mini-grid.

Load type Load Power Quantitya Usage

consumption duration
per day

Household Lamp (inside 6 2 18:00–00:00

house)
Lamp (outside 6 1 18:00–00:00
house)
Phone 5 1 18:00–23:00
Charging
Fan 10 1 18:00–23:00
Television 60 0.2 18:00–23:00
Productive use Refrigerator 36 1 00:00–24:00
Agricultural 1500 1 11:00–16:00
Mill
Water pump 250 1 11:00–16:00
Sewing 120 1 09:00–13:00
machine
Social School 6 6 08:00–15:00
infrastruc- lighting
ture School fan 60 1 08:00–15:00
Street lamps 6 10 18:00–07:00

a
Quantities are expressed as number of loads per household for household loads and number of loads per village for productive use and social infrastructure.

Fig. D1. Assumed load profile for village.

Table D2
Assumptions for diesel mini-grid.

Parameter Value Source

Diesel generator cost (USD/kW) 550 (Blum et al., 2013; Palit et al., 2014)
Diesel generator size (kW) 5.8 Authors’ assumption
Diesel generator efficiency (%) 30 Manufacturers’ technical
specifications
Diesel price (USD/L) 0.66 Indian Oil Corporation website
(November 2016)
Annual change in diesel price 1.8 (NACS, 2014)
(%)
Diesel generator lifetime 50,000 (Kost et al., 2013)
(runtime in hours)
Low voltage distribution line 2500 (Palit and Chaurey, 2011)
cost (USD/km)
Low voltage distribution line 3 Interviews
length (km)
Distribution losses (%) 5 (Blum et al., 2013)
Connection and labour cost 20 Interviews
(USD/household)
Operation and maintenance cost 0.02 (Kost et al., 2013)
(USD/kWh)

669
A. Malhotra et al. Energy Policy 108 (2017) 657–672

Table D3
Assumptions for solar PV mini-grid.

Parameter Value Source

Solar PV Modules cost (USD/kW) 1200 Interviews

Solar PV module size (kW) 10 Authors’ assumption
Solar PV module lifetime (years) 20 (NREL, 2012)
Solar PV degradation rate (%) 0.5 (Jordan and Kurtz, 2013)
Battery cost (USD/kWh) 220 (Fürstenwerth and
Waldmann, 2014)
Battery size (kWh) 35 Authors’ assumption
Battery lifetime (years) 5 (Fürstenwerth and
Waldmann, 2014)
Battery roundtrip efficiency (%) 89.5% (Akhil et al., 2013)
Inverter cost (USD/kW) 200 (Akhil et al., 2013)
Inverter lifetime (years) 10 (NREL, 2012)
BOS (mounting structure, battery 20 Interviews
room, etc.) (% of system cost)
Low voltage distribution line cost 2500 (Palit and Chaurey, 2011)
(USD/km)
Low voltage distribution line length 3 Interviews
(km)
Distribution losses (%) 5 (Blum et al., 2013)
Connection and labour cost (USD/ 40 Interviews
household)
Operation and maintenance cost (% of 1.50 (Blum et al., 2013)
total investment)

Appendix E

Table E1 shows the mean ratings for the probability and impact of each of the risk categories discussed in Section 5.1, as well as the risk
premium above the cost of equity in a low-risk environment. Based on the interviewee ratings, we find that the risk categories contributing the most
to the high cost of capital for mini-grid investments are developer risk, financing risk, payment and credit risk, grid extension risk, and power
market risk. The high influence of developer risk and financing risk on the cost of financing is due to the early stage of the mini-grid market of the
country, and cannot be addressed effectively through spatial diversification. On the other hand, as discussed in Section 5.1, the effect of payment
and credit risk, grid extension risk, and power market risk can be reduced through spatial diversification (Table E2).

Table E1
Mean survey ratings for risk categories and calculated risk premia for a single mini-grid investment.

Risk Category Mean incremental probability rating (on a scale of 1–5) Mean financial impact rating (on a scale of 1–5) Risk premium [%]

Power market risk 2.08 3.58 1.2%

Grid extension risk 2.29 3.63 1.3%
Permits risk 0.95 1.90 0.3%
Social acceptance risk 1.68 3.32 0.9%
Technology sourcing risk 1.46 3.14 0.7%
Labour inputs risk 2.08 3.08 1.0%
Developer risk 3.00 4.50 2.1%
Payment and credit risk 2.58 3.75 1.5%
Financing risk 2.91 4.64 2.1%
Currency risk 2.06 3.22 1.0%
Sovereign risk 1.28 3.67 0.7%

Table E2
Average correlation ratings for risk categories by interviewees.

Risk category Governance level

District State National

Currency risk 0.96 0.96 0.96

Financing risk 0.93 0.93 0.79
Technology sourcing risk 0.67 0.67 0.63
Power market risk 0.92 0.75 0.48
Permits risk 0.78 0.66 0.50
Sovereign risk 0.65 0.45 0.25

Geographic distance
< 20 km 20–150 km > 150 km
Grid expansion risk 0.68 0.31 0.03
Social acceptance risk 0.61 0.32 0.14
Payment and credit risk 0.77 0.36 0.18
Labour inputs risk 0.68 0.68 0.68

670
A. Malhotra et al. Energy Policy 108 (2017) 657–672

Appendix F

Fig. F1 illustrates the results of alternate scenarios to that presented in Section 5.2, Fig. 5 (labelled here as “unsubsidized scenario”). The
“subsidized scenario” replicates the calculations for the “unsubsidized scenario”, and also assumes that the mini-grid developer avails of the 30%
subsidy offered under the Uttar Pradesh state mini-grid policy. The “debt scenario” models the impact of incorporation of debt in the capital
structure in addition to and as a result of the derisking effect of the investment subsidy and diversification. The remaining capital outlay after
availing of the subsidy is assumed to have a capital structure of 50% equity and 50% debt, with the cost of debt assumed to be 14%.

Fig. F1. Reduction in LCOE resulting from spatial diversification.

References
Aggarwal, V., Fahey, A., Freymiller, H.S., Li, S., Huang, C.C., Moilanen, S., 2014. Rural
Energy Alternatives in India: Opportunities in Financing and Community
Engagement for Renewable Energy Microgrid Projects. Woodrow Wilson School for
Public and International Affairs.
Ahlborg, H., Hammar, L., 2014. Drivers and barriers to rural electrification in Tanzania
and Mozambique – grid-extension, off-grid, and renewable energy technologies.
Renew. Energy 61, 117–124.
Akhil, A.A., Huff, G., Currier, A.B., Kaun, B.C., Rastler, D.M., Chen, S.B., Cotter, A.L.,
Bradshaw, D.T., Gauntlett, W.D., 2013. DOE/EPRI 2013 Electricity Storage
Handbook in Collaboration with NRECA. Sandia National Laboratories,
Albuquerque, New Mexico.
Alafita, T., Pearce, J.M., 2014. Securitization of residential solar photovoltaic assets:
costs, risks and uncertainty. Energy Policy 67, 488–498.
Awerbuch, S., 1993. The surprising role of risk in utility integrated resource planning.
Electr. J. 6 (3), 20–33.
Awerbuch, S., 2000. Investing in photovoltaics: risk, accounting and the value of new
technology. Energy Policy 28 (14), 1023–1035.
Banerjee, S.G., Barnes, D., Singh, B., Mayer, K., Samad, H., 2015. Power for All:
Electricity Access Challenge in India. The World Bank, Washington DC.
Baskaran, T., Min, B., Uppal, Y., 2015. Election cycles and electricity provision: evidence
from a quasi-experiment with indian special elections. J. Public Econ. 126, 64–73.
Battke, B., Schmidt, T.S., Grosspietsch, D., Hoffmann, V.H., 2013. A review and
probabilistic model of lifecycle costs of stationary batteries in multiple applications.
Renew. Sustain. Energy Rev. 25, 240–250.
Bazilian, M., Nussbaumer, P., Haites, E., Levi, M.A., Howells, M., Yumkella, K.K., 2010.
Understanding the scale of investment for universal energy access. Geopolit. Energy
32 (10 and 11), 21–42.
Bazilian, M., Nussbaumer, P., Rogner, H.-H., Brew-Hammond, A., Foster, V., Pachauri,
S., Williams, E., Howells, M., Niyongabo, P., Musaba, L., et al., 2012. Energy access
scenarios to 2030 for the power sector in Sub-Saharan Africa. Util. Policy 20 (1),
1–16.
Bhattacharya, A., Kojima, S., 2012. Power sector investment risk and renewable energy: a
Japanese case study using portfolio risk optimization method. Energy Policy 40,
69–80.
Bhattacharyya, S.C., 2013a. Financing energy access and off-grid electrification: a review
of status, options and challenges. Renew. Sustain. Energy Rev. 20, 462–472.
Bhattacharyya, S.C., 2013b. To regulate or not to regulate off-grid electricity access in
developing countries. Energy Policy 63, 494–503.
Bhattacharyya, S.C., Palit, D., 2016. Mini-grid based off-grid electrification to enhance
electricity access in developing countries: what policies may be required? Energy
Policy 94, 166–178.
Blum, N., 2013. Fostering Rural Electrification: The Case of Renewable Energy-Based
Village Grids in South East Asia (DISS. ETH NO. 21265), ETH Zurich.
Blum, N.U., Sryantoro Wakeling, R., Schmidt, T.S., 2013. Rural electrification through
village grids – assessing the cost competitiveness of isolated renewable energy
technologies in Indonesia. Renew. Sustain. Energy Rev. 22, 482–496.
Blum, N.U., Bening, C.R., Schmidt, T.S., 2015. An analysis of remote electric mini-grids
in laos using the technological innovation systems approach. Technol. Forecast. Soc.
Change 95, 218–233.
Brealey, R., Myers, S., Allen, F., Mohanty, P., 2012. Principles of Corporate Finance. Tata
McGraw-Hill Education.
Brown, L., Brown, S., Jarvis, S., Syed, B., 2001. An inventory of nitrous oxide emissions
from agriculture in the UK using the IPCC methodology: emission estimate,
uncertainty and sensitivity analysis. Atmospheric.
Carafa, L., Frisari, G., Vidican, G., 2015. Electricity transition in the middle East and
North Africa: a de-Risking governance approach. J. Clean. Prod. 128, 34–47.
Chaminade, C., Lundvall, B., Vang, J., Joseph, K.J., 2011. Designing innovation policies
for development: towards a systemic experimentation-based approachHandbook of
Innovation Systems and Developing Countries: Building Domestic Capabilities in a
Global Setting. Edward Elgar Publishing, 360–379.
Chandran-Wadia, L., Deorah, S.M., Nair, S., Lath, A., 2015. Prospects for electricity
access in rural india using solar photovoltaic based mini-grid systems. In: Groh, S.,
van der Straeten, J., Edlefsen Lasch, B., Gershenson, D., Leal Filho, W., Kammen,
D.M. (Eds.), Decentralized Solutions for Developing Economies. Springer
Proceedings in Energy; Springer International Publishing: Cham, 53–64.
Chattopadhyay, D., Bazilian, M., Lilienthal, P., 2015. More power, less cost: transitioning
up the solar energy ladder from home systems to mini-grids. Electr. J. 28 (3), 41–50.
CKinetics, 2013. Financing Decentralized Renewable Energy. cKinetics, New Delhi, .
Comello, S.D., Reichelstein, S.J., Sahoo, A., Schmidt, T.S., 2017. Enabling mini-grid
development in rural india. World Dev. 93, 94–107.
Davidsen, A., Pallassana, K., Singh, J., Shiv, J., Walker, P., Parrish, S., Sitsabeshan, S.,
2015. The business case for off-grid energy in India. The Climate Group, London,
UK.
Eisenhardt, K.M., 1989. Building theories from case study research. Acad. Manag. Rev.
14 (4), 532–550.
Franz, M., Nico Peterschmidt, Rohrer, M., Kondev, B., 2014. The Mini-Grid Policy
Toolkit. GIZ, Eschborn.
Fukuyama, F., 2013. What is governance? Governance 26 (3), 347–368.
Fürstenwerth, D., Waldmann, L., 2014. Stromspeicher in der Energiewende -
Untersuchung zum Bedarf an neuen Stromspeichern in Deutschland für den
Erzeugungsausgleich, Systemdienstleistungen und im Verteilnetz. Agora
Energiewende, Berlin.
Garthwaite, P.H., Kadane, J.B., O’Hagan, A., 2005. Statistical methods for eliciting
probability distributions. J. Am. Stat. Assoc. 100 (470), 680–701.
Gershenson, D., Tilleard, M., Cusack, J., Cooper, D., Monk, A., Kammen, D., 2015.
Increasing private capital investment into energy access: the case for mini-grid
pooling facilities. UNEP.
Gram Oorja, 2013. Rural electrification in India – issues and learnings. Gram Oorja
Solutions Private Limited, Pune.
Hazelton, J., Bruce, A., MacGill, I., 2014. A Review of the Potential Benefits and Risks of
Photovoltaic Hybrid Mini-Grid Systems. Renew. Energy 67, 222–229.

671
A. Malhotra et al.
Hess, D.J., 2013. Industrial fields and countervailing power: the transformation of
distributed solar energy in the United States. Glob. Environ. Change 23 (5),
847–855.
Hubbard, D., 2014. How to measure anything: finding the value of intangibles in
business. John Wiley and Sons.
IEA, 2011. Energy for all: financing access for the poor, special early excerpt of world
energy outlook 2011. IEA, Paris.
IEA, 2015. World Energy Outlook. International Energy Agency (IEA), Paris.
IEA; World Bank, 2015. Sustainable Energy for All 2015 – Progress Toward Sustainable
Energy. World Bank, Washington, DC.
Jobson, J.D., Korkie, R.M., 1981. Putting Markowitz theory to work. J. Portf. Manag. 7
(4), 70–74.
Jordan, D.C., Kurtz, S.R., 2013. Photovoltaic degradation rates-an analytical review.
Prog. Photovoltaics Res. Appl. 21, 12–29. http://dx.doi.org/10.1002/pip.1182.
Kemmler, A., 2007. Factors influencing household access to electricity in India. Energy
Sustain. Dev. 11 (4), 13–20.
Kempener, R., D’Ortigue, O.L., Saygin, D., Skeer, J., Vinci, S., Gielen, D., 2015. Off-grid
renewable energy systems: status and Methodological Issues. IRENA, Bonn.
Kost, C., Mayer, J.N., Thomsen, J., Hartmann, N., Senkpiel, C., Philipps, S., Nold, S.,
Lude, S., Saad, N., Schlegl, T., 2013. Levelized Cost of Electricity: Renewable Energy
Technologies. Freiburg, Germany.
Laurikka, H., Springer, U., 2003. Risk and return of project-based climate change
mitigation: a portfolio approach. Glob. Environ. Chang. 13 (3), 207–217.
Levi-Faur, D., 2012. From “Big Government” to “Big Governance”? In: The Oxford
Handbook of Governance; Oxford, UK, pp 3–18.
Lintner, J., 1965. The valuation of risk assets and the selection of risky investments in
stock portfolios and capital budgets. Rev. Econ. Stat., 13–37.
Lowder, T., Mendelsohn, M., 2013. The Potential of Securitization in Solar PV Finance.
National Renewable Energy Laboratory, Golden, CO.
Markowitz, H., 1952. Portfolio selection. J. Finance 7 (1), 77–91.
NACS, 2014. Future of Fuels: an Analysis of Government Projections Through 2040.
Alexandria, VA, USA.
Nelson, D., Shrimali, G., Goel, S., Konda, C., 2012. Meeting India's Renewable Energy
Targets: The Financing Challenge. Climate Policy Initiative, San Francisco.
NREL, 2012. Cost and Performance Data for Power Generation Technologies. National
Renewable Energy Laboratory, Golden, CO.
Nykvist, B., Nilsson, M., 2015. Rapidly falling costs of battery packs for electric vehicles.
Nat. Clim. Change 5 (4), 329–332.
Oda, H., Tsujita, Y., 2010. The Determinants of Rural Electrification in Bihar, India. IDE
Discussion Paper No. 254.2010.9, Chiba.
Ondraczek, J., Komendantova, N., Patt, A., 2015. WACC the dog: the effect of financing
costs on the levelized cost of solar PV power. Renew. Energy 75, 888–898.
Orlandi, I., McCrone, A., Battley, D., Tyabji, N., Falzon, J., Lerner, A., 2016. How can
pay-as-you-go solar be financed?. Bloomberg New Energy Finance, London, UK.
Palit, D., Chaurey, A., 2011. Off-grid rural electrification experiences from South Asia:
status and best practices. Energy Sustain. Dev. 15, 266–276. http://dx.doi.org/
10.1016/j.esd.2011.07.004.
672
Energy Policy 108 (2017) 657–672
Palit, D., Sarangi, G., Krithika, P., 2014. Energising rural India using distributed
generation: the case of solar mini-grids in Chhattisgarh state, India. In:
Bhattacharyya, S.C., Palit, D. (Eds.), Mini-Grids for Rural Electrification of
Developing Countries. Springer International Publishing, 313–342.
Pueyo, A., 2013. The Evidence of Benefits for Poor People of Increased Renewable
Electricity Capacity: Literature Review. Institute of Development Studies (IDS): IDS
Evidence Report 31, Brighton.
Pueyo, A., Orraca, P., Godfrey-Wood, R., 2015. What Explains the Allocation of Aid and
Private Investment for Electrification? IDS March 1.
Roques, F.A., Newbery, D.M., Nuttall, W.J., 2008. Fuel mix diversification incentives in
liberalized electricity markets: a mean–variance portfolio theory approach. Energy
Econ. 30 (4), 1831–1849.
Schinko, T., Komendantova, N., 2016. De-risking investment into concentrated solar
power in North Africa: impacts on the costs of electricity generation. Renew. Energy
92, 262–272.
Schmidt, T., 2015. Will private-sector finance support off-grid energy? In: Heap, R. (Ed.),
Smart Villages: New Thinking for Off-grid Communities Worldwide. Banson,
Cambridge, 81–87.
Schmidt, T.S., 2014. Low-carbon investment risks and de-risking. Nat. Clim. Chang. 4
(4), 237–239.
Schmidt, T.S., Blum, N.U., Sryantoro Wakeling, R., 2013. Attracting private investments
into rural electrification – a case study on renewable energy based village grids in
Indonesia. Energy Sustain. Dev. 17 (6), 581–595.
Schnitzer, D., Lounsbury, D.S., Carvallo, J.P., Deshmukh, R., Apt, J., Kammen, D.M.,
2014. Microgrids for Rural Electrification: A Critical Review of Best Practices Based
on Seven Case Studies. United Nations Foundation.
Shipan, C.R., Volden, C., 2008. The mechanisms of policy diffusion. Am. J. Pol. Sci. 52
(4), 840–857.
UNCTAD, 2014. World Investment Report 2014 – Investing in the SDGs: An Action
Plan. United Nations Conference on Trade and Development, Geneva.
UNDP; GEF; CBI. 2015, The climate finance aggregation initiative for developing
countries – concept note. United Nations Development Programme, Global
Environment Facility, Climate Bonds Initiative, New York, USA.
Vose, D., Koupeev, T., Van Hauwermeiren, M., Smet, W., van den Bossche, S. Modeling
Expert Opinion. 〈http://www.vosesoftware.com/ModelRiskHelp/index.
htm#Modeling_expert_opinion/Modeling_expert_opi〉. (Accessed Oct 6, 2016).
Waissbein, O., Glemarec, Y., Bayraktar, H., Schmidt, T.S., 2013. Derisking Renewable
Energy Investments 2013. United Nations Development Programme (UNDP), New
York.
Williams, N.J., Jaramillo, P., Taneja, J., Ustun, T.S., 2015. Enabling private sector
investment in microgrid-based rural electrification in developing countries: a review.
Renew. Sustain. Energy Rev. 52, 1268–1281.
Wüstenhagen, R., Menichetti, E., 2012. Strategic choices for renewable energy
investment: conceptual framework and opportunities for further research. Energy
Policy 40, 1–10.
Yin, R.K., 2003. Case Study Research – Design and Methods 3rd ed.. Sage Publications,
Thousand Oaks, USA; London, UK; New Delhi, India.