learning to lend for off-grid solar power: policy lessons from world bank loans to india, indonesia,...

*Corresponding author. Tel.: #44-1223-339591; fax: #44-1223-339701.

E-mail addresses: [email protected] (D. Miller),[email protected] (C. Hope)

1Many studies emerged in the early 1990s in response to the risingspectre of climate change, and the United Nations Conference onEnvironment and Development (UNCED). (see for instance Anderson,1992; O$ce of Technology Assessment, 1992; Johansson et al., 1993;Jhirad and Mintzer, 1993; Messenger, 1994).

Energy Policy 28 (2000) 87}105

Learning to lend for o!-grid solar power: policy lessons fromWorld Bank loans to India, Indonesia, and Sri Lanka

Damian Miller, Chris Hope*Judge Institute of Management Studies, University of Cambridge, Cambridge CB2 1AG, UK

Received 12 November 1999

Abstract

The World Bank has sought to advance the di!usion of solar photovoltaic (PV) technology for o!-grid applications in thedeveloping world. As these systems are fundamentally di!erent to centralised power stations and conventional rural electri"cation,the World Bank has been learning how best to lend for such technology. This study seeks to highlight the lessons learnt from theWorld Bank's "rst loans for o!-grid PV to India, Indonesia, and Sri Lanka. It uses lifetime cost analysis to justify continuedintervention in this sector, and it draws on theories of innovation di!usion to guide analysis and ultimately policy recommendations.Because of the special role of entrepreneurial start up companies in the rural PV sector, the paper also uses a company cash #owmodel to demonstrate the e$cacy of various supply-side policies. Finally, the study concludes with a checklist of policy lessons anda consideration of the role of the International Finance Corporation in this sector. ( 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Renewable energy policy; Rural electri"cation; Solar photovoltaic technology; Multilateral lending institutions

1. Introduction

Renewable energy technology is seen by many as apotential solution to the environmental challengespresented by the developing world's burgeoning electric-ity requirements.1 Such technology has particular ap-plicability in remote, rural areas, where it can make use oflocally available resources, such as sunlight, biomass,wind, and hydro power. Often, complete electri"cationthrough grid extension has been restricted by budgetaryconstraints, leaving large numbers of rural householdswithout access to electricity.

As the largest multilateral lender in the power sector,the World Bank has come under pressure to expand itslending for renewable energy. In the search for re-newables that were near-commercial, the Bank has found

that, in the absence of a connection to the grid, ruralhouseholds were willing to pay commercial prices forsolar photovoltaic home-lighting systems, and that suchsystems could be a cost-e!ective substitute for kerosenelighting and battery charging for radio and TV. Thus,with the help of the Global Environment Facility (GEF),the Bank set out to advance the di!usion of the solarhome system (SHS), "rst in India in 1994, followed byIndonesia in 1997, and more recently, Sri Lanka in 1998.

The experience of the World Bank in the power sectorhas been predominantly one of lending for large central-ised, fossil or hydro-based power stations, while loans forrural electri"cation have gone to conventional grid ex-tension, and the development of diesel powered, micro-grid stations. As such, lending for the SHS, where powergeneration, transmission, and distribution is located atthe level of the rural household, has necessitated policyinnovations. The demand for policy innovation has, inturn, meant that the World Bank has been on a learningcurve in lending for o!-grid solar power.

After "rst considering the background to the WorldBank entering the o!-grid SHS market, this study goeson to use a lifetime cost analysis to provide a rationale forthe World Bank and GEF's continued intervention inthis sector. To further guide our analysis of theWorld Bank's PV loans and, ultimately, our policy

0301-4215/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved.PII: S 0 3 0 1 - 4 2 1 5 ( 9 9 ) 0 0 0 7 1 - 3

2 Indeed an internal report within the World Bank concluded thatthere were considerable limitations to such systems (Solar IndustryJournal, 1991).

3 Interview with former renewable energy specialist in World BankWashington, DC, 11 December 1996.

4 Interview with member of ASTAE, World Bank, 19 July 1994.5 Interview with member of Energy and Industry Division, the World

Bank, 20 July 1994.6 In 1994, at the time of the India loan for PV technology, the

minimal internal economic rate of return required to use funds from theInternational Development Agency in the electricity sector was 12%.Early World Bank assessments projected that a 10 million USD grantfrom the GEF would lift the economic return of the PV component ofthe India loan from 1.5 } 3.3% to 14.0 } 14.6% (World Bank, 1992a,b).

prescriptions, we then turn brie#y to theories on innova-tion di!usion. Because of the historically signi"cant roleof the entrepreneurial start up company in this sector,and the limitations of cash-#ow, our analysis also em-ploys a company cash #ow model to assess the e$cacy ofvarious supply-side policy options. Finally, we concludewith a checklist of policy recommendations and a consid-eration of the future of World Bank lending in this sector,with particular attention given to the emerging role of theInternational Finance Corporation (IFC).

2. Background to World Bank lending for the solarhome system

2.1. An opening for renewable energy technology

The World Bank recognised early on that investmentsin rural electri"cation through conventional grid exten-sion would have to be selective (World Bank, 1975).Thus, by implication, many rural households in the de-veloping world would not receive a grid connection. Theconventional alternative to grid extension is an isolatedmicro-grid supplied by a diesel generator. However, theremote location of these systems has made it di$cult toacquire spare parts and costly to transport diesel oil,resulting in service that is generally both costly andunreliable (Lovejoy, 1992).2 Thus, in short, the WorldBank was open to alternative forms of rural electricitysupply.

At the same time, as one former sta! member of theBank re#ects, the environmental question was starting to`creep into the lexicona.3 In the run up to UNCED in1992 the World Bank's environmental record wascoming under ever-increasing scrutiny. This was parti-cularly the case in the power sector where large-scalepower plants were accused of exacerbating global warm-ing (Bosshard, 1994), while large-scale dams were seen toboth damage local environments and cause social dislo-cation (Morse, 1992).

The World Bank and other multilateral institutionswere called upon to increase their lending for renewableenergy technologies (Johansson et al., 1993; O$ce ofTechnology Assessment, 1992). The World Bank re-sponded with a$rmation of the increasing viability ofsuch technologies in developing countries:

2the clean, reliable, and increasingly cost-com-petitive characteristics of renewable energy tech-nologies makes them ideal candidates for displacing

other conventional energy options in a variety ofdeveloping countries in the coming years. (TheWorld Bank 1993; p. 92)

Furthermore, organisational changes were made withinthe Bank to give renewable energy technology the sup-port it would need to attract loan o$cers' attention. In1993, the Asia Alternative Energy Unit was establishedwithin the World Bank to generate con"dence amongBank sta! (particularly loan o$cers) in a brand of tech-nology that had formerly been seen as either anR & D exercise, or the sole domain of aid agency demon-stration projects.4 As one sta!member said at the time, itwas really a question of instigating a paradigm shiftwithin the Bank.5

2.2. The solar home system gains in credibility

In the mid- to late- 1980's, entrepreneurs and otherprivate sector companies were demonstrating thatthere was a market for PV technology among rural,unelectri"ed households for small-scale lighting and en-tertainment services (Hankins, 1993). The fact that ruralhouseholds were willing to pay commercial prices for theSHS was seen as the strongest signal of this technology'spotential. Kenya received particular attention withinthe Bank (Van der Plas, 1994), given thatthe private sector had been largely responsible for thedi!usion of 20}40,000 solar home systems by 1993; anexample reinforced by private sector sales in countriessuch as the Dominican Republic, Sri Lanka, andZimbabwe.

The credibility of the SHS was further enhanced byreports by the Industry and Energy Department of theWorld Bank (Ahmed, 1994; Anderson and Ahmed, 1995),which then served as an advisory body to loan o$cers inthe energy sector. In short, these reports found that undercertain circumstances, PV technology was more coste!ective than grid electri"cation and diesel mini-grids forrural electricity supply (Anderson, 1992).

Furthermore, with the establishment of the GlobalEnvironment Facility (GEF) in 1991, World Bank sta!now had access to grants to increase the projected eco-nomic returns on loans for renewable energy technolo-gies to the levels required by in-house guidelines.6 Thus,with the SHS attracting both greater credibility and

88 D. Miller, C. Hope / Energy Policy 28 (2000) 87}105

Table 1

Monthly amortized costs of SHS and diesel genset in India

Product option Cost to customer37 Wp SHS, Growing to 74 Wp SHS 10.1 USD500 W diesel genset 30.4 USD

concessional money, the stage was set for the WorldBank to enter the o! grid solar power market.

3. The rationale for lending for o4-grid solar power

Some have challenged the idea that the World Bankand the GEF should be giving assistance to the SHS(Erickson and Chapman, 1995; Drennen et al., 1996).These authors question whether a 48 W peak SHS is themost cost-e!ective solution for rural unelectri"ed house-holds when compared to a 650 W portable petrol genset.

3.1. Comparison with genset technology

Certainly on a per kilowatt hour basis, the isolatedpetrol/diesel genset is a more cost e!ective option thanthe SHS. However, many households in the developingworld have quite basic electricity requirements whichonly extend to lighting, radio, and television (WordBank, 1996). For such households genset technologywould provide vastly excess capacity, and thus, excesscosts.

By contrast a SHS is modular, allowing a household toadd another solar module as and when its electricitydemand increases. A one or two module system cannotrun power tools, or provide refrigeration, but it cansupply many of the basic home electricity needs fora rural household: compact #uorescent lights, table fan,radio/stereo, and television. Comparing the monthlyamortised costs of the SHS and the diesel genset over 10years, where the household adds an additional panel inyear "ve and increases the size of their battery, the SHS isvastly more competitive (Table 1).

Furthermore, "eld work in India (Miller, 1998) re-vealed that in addition to their concern about the run-ning costs of genset technology, existing SHS customersshared concerns about the relative inconvenience of ac-quiring diesel fuel and switching the system on and o!,the hazards involved, the noise and pollution, and rela-tively high demands of maintenance. These "ndings aresupported by "eld work in Kenya where compared todiesel gensets, SHS customers perceived their system wasless costly, required less maintenance, was more conve-nient to turn on and o!, was less vulnerable to fuel price#uctuations, saved journeys, and was far less hazardous

(Acker and Kammen, 1996). Even those who question theeconomics of the SHS recognise that unelectri"ed house-holds may prefer the SHS due to the regular expense ofpurchasing fuel, the burden of transporting the fuel, andthe risk of non-availability of fuel (Erickson andChapman, 1995).

3.2. Comparison with kerosene lighting and batterycharging

The more immediate competition to the SHS in ruralareas comes from battery charging for black and whiteTV and radio, and kerosene lanterns for lighting. In thisregard, studies published by the World Bank have foundthat the SHS is more cost e!ective for the rural house-holds than kerosene lanterns and battery charging; albeitthat such claims should be made country speci"c (Foley,1995; Cabraal et al., 1996).

We have found that in all three countries under consid-eration, the SHS is cost competitive where a household'sdemand is for a black and white television and higherquality lighting (the equivalent of using battery chargingfacilities and petromax lanterns) (Table 2).

Table 2 does not include the cost of transportingbatteries for recharging and collection. Moreover, usersof the SHS have stated other non-monetary bene"tscompared to kerosene and battery charging which in-clude: higher quality lighting than kerosene, allowingchildren to study and members of the family to moreeasily engage in income producing activities at night;light at the #ip of a switch compared to regular purchasesof kerosene and preparation of the lantern; reducedhealth hazards from inhalation of fumes and risk ofhousehold "re; greater convenience than transportingbatteries and waiting 1}2 days for collection; increasedself-su$ciency; and perceptions of enhanced status(Cabraal et al., 1996; Acker and Kammen, 1996; Miller,1998).

This does not mean that rural households would notrather have a connection to the electricity grid. What itmeans is that in the absence of the grid, the SHS iscompetitive with the technologies to which rural house-holds would otherwise turn.

4. Theories on innovation di4usion pertaining to the SHS

Having a rationale for lending for the SHS does notmake it any easier to do so. In so far as theories oninnovation di!usion consider why technological innova-tions spread throughout society, they can provide a use-ful framework for analysing and assessing World Bankpolicy in the o!-grid PV sector, and making policy rec-ommendations.

Reviews of innovation di!usion literature have tendedto conclude that explanations of a technology di!usion

D. Miller, C. Hope / Energy Policy 28 (2000) 87}105 89

Table 2

Monthly amortized cost of SHS and kerosene/battery charging

Product 10 year amortized cost(USD)

15 year amortized cost(USD)

20 year amortized cost(USD)

India 2 basic lanterns and batterycharging

6.1 6.2 6.2

India 2 petromax lanterns andbattery charging

12.5 12.7 12.7

India 2 lights 37 Wp SHS 8.6 7.7 7.6

Indonesia 2 basic lanterns and batterycharging

8.2 8.4 8.4

Indonesia 2 petromax lanterns andbattery charging

15.0 15.1 15.2

Indonesia 2 lights 37 Wp SHS 6.7 5.9 5.7

Sri Lanka 2 basic lanterns and batterycharging

5.9 6.0 6.0

Sri Lanka 2 petromax lanterns and batterycharging

11 12.3 12.3

Sri Lanka 2 lights 37 Wp SHS 7.4 6.5 6.2

Table 3!

Country Year Size (Wp) Price ($)

Kenya 1993 53 1378China 1994 10 93

20 160}280Indonesia 1994 12 215

40 40053 425}700

Philippines 1993 48 64053 900

Sri Lanka 1995 20 34030 46040 56050 674

Brazil 1994 50 700100 1100

Dominican Republic 1993 25 45035 57548 700

Mexico 1994 50 700

!Source: Foley (1995, p. 10.).

need to do a better job of integrating both supply anddemand-side factors (Brown, 1981; Heertje, 1983; Jones,1982; Jain et al., 1991; Miller, 1998). The logical implica-tions of such "ndings are that policies to accelerate di!u-sion must equally seek to integrate the supply anddemand side of the di!usion process.

4.1. The demand side

On the demand side, the dominant literature toexplain the di!usion of technological innovations indeveloping countries derives from the sociological tradi-tion (Agarwal, 1983), as represented mainly by Rogers andShoemaker, 1971, Rogers and Kinkaid, 1981 and Rogers,1995. This body of literature notes that an adopter's feltuncertainty when confronted with an innovation canlead to perceptions of risk, and thus non-adoption(Rogers and Shoemaker, 1971). Such perceptions of riskare even more applicable where an innovation requiresa large "nancial investment up-front (Rogers, 1995). Cer-tainly, the above factor is relevant to the SHS whichinvolves new, relatively complex technology with con-siderable up-front investment (see Table 3).

Yet, even if a household is convinced that they wouldlike to invest in an innovation, other analysts have foundthat access to "nancial resources can be a fundamentalconstraint on adoption (Brown, 1981; Roling, 1982;Agarwal, 1983; Roy, 1994; Harrison, 1994). Speci"cally, ithas been found that access to credit is critical to theup-take of capital intensive technologies in rural areas ofdeveloping countries (Roy, 1994). For example, greenrevolution technology, with its requirements for irriga-tion and arti"cial fertilisers, was found to have occurred

among farmers who began with more capital, while non-adoption was found to be primarily due to `perceived orreal institutional blocks to credit availabilitya such asland-holdings for collateral (Havens and Flinn, 1975;p. 479). As the SHS requires that the household invest in15}20 years of power generation up-front, it is not sur-prising that studies published both within and outside ofthe Bank view access to credit as the single most impor-tant factor in#uencing the di!usion of the SHS (GTZ,1995; Cabraal et al., 1996).


7Certainly, the state level utilities in India and the national-levelutilities in Sri Lanka and Indonesia have historically shown littleinterest in entering the SHS market. A more recent exception is the jointventure between the pro"table South African utility* ESKOM* andShell Renewables. The venture was launched in 1998, and seeks toinstall 50,000 solar home systems within three years on a fee for servicebasis.

8Background information provided by former renewable energyspecialist in World Bank, who participated as a consultant in theproject design phase, Washington, DC, 11 December 1996.

4.2. The supply side

Although it seems to have received comparatively lessattention, analysts have also noted that an innovation'sdi!usion will be slowed without adequate supply(Griliches, 1957; Brown, 1981; Simon and Sebastian,1987; Jain et al., 1991). As one analyst remarks on thedi!usion of green revolution technologies in rural areasof LDCs:

in the underdeveloped parts of the world, fertiliser,pesticides, better seed and simple tools cannot befound in local villages .... consequently even whena producer becomes convinced that it would pay forhim to use more fertiliser in combination with otherimproved methods, he cannot do so because theessential materials and services are not available tohim. (quoted in Kearl, 1976, p. 169)

Introduction of the SHS will rely on a distribution andmarketing infrastructure, while its proper functioningwill require an installation and servicing infrastructure(for simplicity both infrastructures will be referred to asthe market infrastructure). That the SHS currently lacksa market infrastructure is seen as a factor strongly limit-ing its di!usion (Hurst, 1990; Huacuz, 1991; Hankins,1995; Barozzi and Guidi, 1993; Hill et al., 1994; Cabraalet al., 1996). The infrastructure needs are not extensive;only a small shop, several technicians, motorbikes anda tool kit. Nonetheless, the di!usion of the SHS is depen-dent on it.

As to who will develop this market infrastructure, giventheir knowledge of where unelectri"ed customers arelocated and their existing network of trained technicians,national electric utilities are ideal propagators of SHStechnology. However, utilities have a knowledge basethat pertains to grid extension, around which they haveestablished managerial and technical practices, and thus,have generally been reluctant to enter the SHS market,especially where experiencing "nancial di$culties(Huacuz, 1991; Foley, 1995).7

In the absence of the utility, entrepreneurial start-upcompanies have played a pivotal critical role in the earlyintroduction of the SHS throughout the developingworld; particularly in India, Indonesia, and Sri Lanka(Miller, 1998). Similarly, entrepreneurs played a centralrole in introducing grid connected electric lighting tourban and rural customers in the United States (Hughes,1979), as well as subsequent innovations in the lighting

industry (Utterback, 1994). As such, policies to promotethe growth of the market infrastructure for SHS techno-logy should logically seek to promote the growth of suchentrepreneurial start-ups, and ultimately, the entry ofmore established companies.

5. India: the 1994 renewable resources development loan

The need to integrate demand and supply-side policieswas implicitly recognised in the World Bank's policydocuments for the PV loan to India in 1994 (World Bank,1992a,b). However, in practice, both demand and supplyof the SHS would be restricted under this loan.

5.1. Background to the PV market development loan

Di!erent stakeholders within the Bank sought to pro-mote a renewable energy loan to India.8 The India coun-try desk had already identi"ed a high return loan forwaste-fuelled paper manufacturing plant and waslooking for other renewable energy technologies tocomplement this loan. Simultaneously, as the "nancialmechanism to the Climate Change Convention, the GEFwas looking to fund a large renewable energy project,and was particularly interested in supporting PV techno-logy. The GEF ultimately selected India (and not China)on the basis of the Government's long-running supportfor renewables (in 1992 the Department of Non-Conven-tional Energy Sources was up-graded to a ministry), andits early contribution to GEF funds. Although WorldBank sta! recognised that a PV loan would be di$cult toimplement, it was included to guarantee GEF participa-tion. In the end the GEF provided a grant of USD 26million to accelerate the di!usion of both PV and windenergy technologies, of which USD 10 million was de-voted to PV.

The GEF initially wanted to develop a large utility-scale-PV project. However, pre-investment studies re-vealed that such projects were still uneconomic and thatIndia's electricity sector did not yet make provisions forprivate power generation. Instead, the studies found thatthe most economically and "nancially viable applicationsof PV lay in the home lighting sector; for two and fourlight solar home systems, as well as centralised PV sta-tions for 100 rural households (World Bank, 1993, Annex5.3, p. 3).

The loan o$cer in charge did not want to de"ne themarket ahead of time. As such the loan was deliberatelycalled `the PV Market Development Loana. Nonethe-less, this loan o$cer is equally clear that


9 Interview with India loan o$cer, World Bank, Washington, DC,2 October, 1995.

10 Interview with India loan o$cer, World Bank, Washington, DC,11 December 1996.

Table 4!

India's rural credit infrastructure: number of branches

Commercial banks Regionalrural banks

Number of banks 28 196Number of semi-urbanbranches

10900 1079

Number of rural branches 21952 13290

!Source: Swiss Development Co-operation (1996).

11 Interview with managing director of IREDA, New Delhi, 22 June1996.

12 Interview with ASTAE adviser, World Bank, Washington, DC, 16December 1996.

the conception was that the rural market was wherewe would "nd the demand for PV technology and itwas in this market that PV technology made themost sense. Solar home systems and solar lanterns,assuming a willingness to pay, were the viable ap-plications.9

Given the up front costs of PV technology, the Bank'spre-investment study identi"ed the need to provide low-interest credit: `because PV systems are characterized byhigh capital, yet low operating costs, "nancing is a keyingredient in making PV a!ordablea (World Bank,1993; p. x). In addition, the same study saw a need tosupport the supply side:

because the industry is still new, marketing net-works and support systems for PV are lacking.There is a need to stimulate demand by promotingtheir availability2 (World Bank, 1993, p. 1}2).

Similarly, the GEF appraisal document (World Bank,1992b, p. 3) stated that `the solar photovoltaic compon-ent has a high potential for replicability2but it requiresthe building up of basic market infrastructurea.

However, World Bank policies did not ensure thatcredit #owed to rural customers. Nor did they su$cientlysupport those companies coming forward to develop themarket infrastructure.

5.2. The demand side: an aversion to rural credit

Responsibility for managing the PV line of credit (andsmall hydro and wind) lay with the Indian RenewableEnergy Development Agency (IREDA), which was estab-lished by the Indian Government in 1987 to lend forrenewable energy projects. Yet two years after the WorldBank loan to India, while both the small hydro and windloans were fully taken-up, IREDA had not sanctionedany loans under the PV line of credit (Deviah, 1995).When the money for PV technology did start to move, itwent to the corporate rather than the rural sector. Al-though this did not contradict the intentions of the Bank,the loan o$cer nonetheless expressed the view that `weare still worried about the loan * we are not seeing itmove into the rural sectora.10

IREDA was not predisposed to lending for rural ap-plications of renewable energy technologies. Prior to theloan, a study commissioned by the Dutch aid agencyfound that `2IREDA has done little to help identifyand develop sound renewable energy projects and energy

e$ciency projects that are relevant to rural developmentand the disadvantaged.a (Hassing and Mendes, 1992,p. 30) As the managing director of IREDA later ex-plained in relation to "nancing sales of PV in rural areas:

you can do it, you can reach remote users. But toreach and then recover your money is another prob-lem - the rural credit risk is very high2.Agricul-tural co-operatives, commercial banks, and ruraldevelopment banks cannot recover their loans forfertilisers, animals, tractors, etc2.[rural people]take all these loans but do not have su$cient in-come to repay them. We might call this the over-debt burden of the rural person2.This is a majorrealistic problem211

Yet, this &problem' could have been addressed by theWorld Bank before the loan by hand-picking the ruralcredit organisations * banks, private "nance organisa-tions, and NGOs * that were su$ciently robust tomanage the rural credit risk. India has a tremendousnetwork of rural banks, which could have been broughtinto the project preparation process (Table 4)

Yet, those who prepared the loan are clear that theresimply was not su$cient time to undertake such prepara-tory work:

people say to us that we did not take enough time indeveloping the PV line of credit. But if we had takenthe required amount of time, there would have beenno PV component. There would have been no pro-ject to attach it to, as the others would have gonethrough. This is the reality of the situation. We arenot working in the ideal world.12

Equally pressured to meet disbursement deadlines,IREDA did not set about trying to identify the viablerural credit organisations. As IREDA sta! made clear in


13 Interview with assistant to MD of IREDA, New Delhi, 26 June1996.

14The interest of such companies was to use the 2.5% interest fromIREDA, repayable over ten years, and the 100% accelerated deprecia-tion. Together this made PV technology very attractive to companiesthat could reinvest the IREDA loan and use the depreciation incentiveto reduce their tax payments.

15 Interview with managing director of IREDA, New Delhi, 22 June1996.

16 Interview with Karnataka dealer to PV manufacturer, Bangalore,5 June 1996.

17 Interview with Marketing Manager and General Manager of PVmanufacturer, Hyderabad, 21 February 1996. This company wasknown to have gained the major share of Government orders.

18 Interview with India loan o$cer, World Bank, Washington, DC,2 October, 1995.

19 Interview with India loan o$cer, World Bank, Washington, DC,11 December 1996.

1996, `we cannot adopt this approach as we have to meetthe set project terminal dates * the money from theWorld Bank lapses in 19982.a13 Instead IREDA con-centrated on lending directly to corporations who couldavail themselves of Government tax incentives (100%accelerated depreciation in the "rst year for owners of PVtechnology).14 Such loans entailed fairly large transac-tions to corporations who could meet IREDA's securityrequirements, and who were felt by IREDA to improvethe credibility of PV technology. One day, IREDA wouldconcentrate on the rural sector, but not now:

the rural market is a high potential market, but it isdi$cult to reach2our strategy is to encouragerural adoption through a step by step process. Bythis I mean go from urban, to semi-urban, to ruraland then to remote, such that the PV would reachthe remote user over a period of time, at which pointthe costs will have come down and awareness willhave gone up. You cannot go straight to the tail end- to the remote areas. This is an accepted deliverypattern for all services and products in this country.All items such as TVs, communications, etc. beganin the urban end and went to the remote end.Reaching out now is very challenging.15

5.3. The supply side: lacking a market infrastructure

In the early 1980s the Government of India set out tonurture the growth of a domestic PV industry behinda tari! wall. Government demand for PV technology ledto the rapid growth of this industry: with installed pro-duction capacity of 18 MW per annum in 1995, a total of"ve cell manufacturers, eight module manufactures, and50 system integrators (ECN, 1995). Yet, despite this sig-ni"cant supply network, none of the private and publicsector PV companies saw an interest in developing therural market infrastructure.

Suppliers of modules and complete systems were ableto generate more secure revenue under the Government'sPV programmes, as well as industrial applications of PV(such as remote oil platforms and gas pipelines). As wascon"rmed by the dealer to a PV supplier:

government orders are easy as there is a chunk to beallotted, with few decision-makers, which can be

in#uenced. But in the rural markets, there are mil-lions of decision-makers to try and in#uence, and todo this requires more e!ort, meaning that yourcosts go up and at least in the short term, yourpro"ts go down.16

Similarly the marketing director of a major Indian PVmanufacturer noted that:

the consolidated market is easier to serve2as therural markets are dispersed. Here, the problem is`how do I reach them?a The reach is where the costscome in for us. You need many of these little, little,little centres, and then one needs to manage thesecentres, and they will grow and how to orchestratethem?17

Thus, as a World Bank review in late 1996 revealed,although `there were projects that would lead to PVbeing placed in rural areas, such as lighting for teaplantationsa, there were `none that lead to building ruralsector delivery mechanismsa (World Bank, 1996). Theonly exception, was an entrepreneurial company inSouth India, who, seemingly alone at this time, had thevision of establishing a rural network of PV servicecentres.

5.4. Inadequate Support for the Entrepreneur

The entrepreneur launched his operations in late 1994with less than 30,000 USD. Thus it is not surprising thatafter establishing 3 service centres in two di!erent statesof South India, his start up company, like many otherstart up companies, faced a cash #ow crunch (Fig. 1).

The World Bank's loan o$cer had felt that the entre-preneur should play a critical role in developing themarket infrastructure: `it only takes an entrepreneur tosettle in, to establish themselves, and to work with thelocal people.a18 But under the India loan there were to beno working capital loans to public or private sectordevelopers. Instead the World Bank wanted &to see' thetransaction between supplier and customer, and thusmandated that the funds be used for consumer (and notsupplier) "nance.19

This meant that the South India entrepreneur "rst hadto identify rural customers to whom his company wouldon-lend IREDA "nance, and then demonstrate how his


Fig. 1. India base case * a cash #ow crunch.

20 Interview with PV adviser to IREDA, New Delhi, 19 May 1996.21The Bank had in part selected IREDA due to its `satisfactory

collection ratea (World Bank, 1992a, p. 27). Yet, perversely, the securityrequirements to ensure this collection rate e!ective precluded the entre-preneurial start-up company.

22Former renewable energy specialist within the World Bank andindependent consultant to India Loan, Washington, DC, 11 December1996

23 Interview with India loan o$cer, World Bank, Washington, DC,11 December 1996

Fig. 2. Long-term loan.

24 Interview with ASTAE adviser, World Bank, Washington, DC, 16December, 1996.


company would recover the loan and repay IREDA. Notsurprisingly, IREDA's assessment of this loan was one of`high institutional riska, since the company did not yethave the managerial skills to administer a rural creditprogram; nor was it su$ciently capitalised to sustain thedefaults that IREDA perceived were likely in the ruralareas.20

Furthermore, the entrepreneur struggled withIREDA's security21, and paper-work requirements. Asa consultant to the loan commented, `the borrowers haveto be so credit-worthy that they should hardly need to goto IREDA in the "rst place * it makes no sense.a22Indeed, the Dutch report had earlier recognised that `thecontinued strict loan requirements tends to bias ap-provals toward the large-scale and better establishedindustrial applicantsa (Hassing and Mendis, 1992, p. 22).In addition this report found that IREDA's procedureswere `long and bureaucratic and that smaller organisa-tions "nd them complicated and costlya. Thus, as theBank's loan o$cer lamented:

this is di$cult, because while a big company hastheir own legal sta! and can handle the documentsthat IREDA requires, [the entrepreneur] is not pre-pared for these requirements. Consequently theloan [to the entrepreneur] has taken time.23

The entrepreneur applied for the loan in November1995, but due to IREDA's reservations, the loan was not

sanctioned until mid 1996, and funds were not disburseduntil early 1997, by which point sales had stagnated(Miller, 1998). Thus, as the ASTAE adviser concludes,`the India loan was not well suited to businesses that areselling a product, and are concerned about their workingcapital requirements and cash #owsa.24

5.5. Counterfactual scenarios

5.5.1. Availability of long-term loansAssuming that the entrepreneur had been able to ac-

quire a long term, low-interest loan in November 1995,when he applied to IREDA, and that it was used todevelop a series of 10 service centres and establish aninventory of systems, historic levels of sales suggest thatthe company would have been able to sell 830 systems bythe end of 1996 (instead of the actual "gure of 250systems). Fig. 2 shows that this alone would not trans-form the company's fortunes.

5.5.2. Business advisory supportThe start-up company was still in the process of devel-

oping its managerial competence, and thus could haveused what the IFC now refers to as `business advisorysupport servicesa (IFC, 1998). As the ASTAE adviser tothe India loan recognised in retrospect:

the need is to work at the dealer/distributor level toprepare them2The point is that you need to bringin the business rigour and management expertise tothese enterprises.25

The Bank put the onus on IREDA to `o!er at its discre-tion, technical and marketing experts to their borrowers


Fig. 3. Expansion with lower cost battery.

26 Interview with assistant to MD of IREDA, New Delhi, 1 March1996.

Table 5!

Import duties on PV technology in India

Product 1992 1994 1996PV modules 110% 45% 30%PV systems 110% 60% 30%

!Source: NREL (1994).

Fig. 4. Expansion without import duty on PV Module.

27 Interview with ASTAE adviser, World Bank, Washington, DC, 16December, 1996.

to help in product quality improvement and to establishe!ective marketing and after-sales service networksa(World Bank, 1993, p. xiii). IREDA o!ered three entre-preneurial development programmes a year, butIREDA's sta! reported that nobody developed fromthese and in the end concluded that `it was just an excusefor a free luncha.26 Instead the main thrust of IREDA'sassistance programmes were technical as opposed tobusiness services, devoted to teaching how to design andinstall PV systems.

The entrepreneur's start-up company could have bene-"ted tremendously from assistance with unit gross mar-gins. The entrepreneur felt that he had hit a price ceiling,and thus found his margins squeezed by the rising cost ofcomponents such as batteries. He was paying USD 129(1996 rupee rate) for deep cycle batteries (suited to PVtechnology), while international prices were roughly halfthis amount. Assuming that IREDA had helped theentrepreneur identify an alternative supplier to reducebattery costs by 25%, and that it had provided a lowinterest loan in November 1995, the entrepreneur couldhave expanded to 830 systems sold with a more favour-able cash #ow situation (Fig. 3).

5.5.3. Elimination of taxes on the PV moduleIn addition, the World Bank could have used its lever-

age to eliminate the import duties on PV modules, whichhad declined, but were still signi"cant (Table 5).

Whereas international prices were USD 4 } 4.5/W in1996, the entrepreneur was paying USD 6.9 (at 1996rupee rate). With the low-interest loan, and no importduties on the PV module, the cost of the module from theIndian supplier to the company would have been at least30% lower, which would have again allowed the com-pany to expand more comfortably (Fig. 4).

6. Indonesia: the 1997 solar home system loan

The World Bank learned valuable lessons from theIndia loan, and through the ASTAE organisation, ap-plied them to the subsequent SHS loan to Indonesia. Asthe ASTAE adviser to both loans re#ects:

there is a learning curve. We are all trying to learnhow best to implement these projects. I have e!ec-tively served as a bridge between the India and theIndonesia loan.27

6.1. Background to the Indonesia Loan

The $44.3 million solar home system loan to Indonesiasought to support the sale of 200,000 systems to ruralunelectri"ed homes in West Java, Lampung and SouthSulawesi. Unlike the India project, the push for PV didnot come from the GEF to address climate change, butfrom the Indonesia country desk for rural electri"cation.The loan o$cer responsible for the World Bank's secondrural electri"cation loan to Indonesia recognised that


28 Interview with Indonesia loan o$cer, World Bank, Washington,DC, 17 December 1996.





even after this loan, there would be 30,000 villages inIndonesia to be electri"ed, and that a lot of these wouldwait 20}30 years to receive electricity, while some villagesnever would. To "ll the gaps * what the loan o$cerdescribed as `the swiss cheese e!ecta * a mission wassent out in 1995. At this time, an entrepreneur in WestJava was identi"ed that had sold 4000 solar home sys-tems on credit in the "rst year of operation. According tothe loan o$cer `after seeing this, I went on the gut feelingthat this project could be donea28

6.2. The demand side: preparing the way for rural credit

The Bank was compelled to spend no less than twoyears ensuring that rural SHS customers could accesscredit. Internal pressures were placed upon the loano$cer to increase the speed of preparation, but he de#ec-ted such pressure, recognising,

that is why the solar component of the [India] loannever took o!2They are only now awakening tothe problems in meeting the rural areas2It isa question of pipeline "lling and doing your home-work. In India it was a question of the Bank rushingto get in, and then leaving. So it is vital that theBank does its homework, only then can the loan bee!ective. That is precisely why it took us two yearsto prepare this loan.29

The rural banks in Indonesia were deemed to be inappro-priate intermediaries, as they only o!ered short-termloans at high rates of interest for consumer durables. Bycontrast, the entrepreneur in West Java had been able tosell the SHS on company credit of 4 years, at commercialrates of interest, with high rates of recovery. Thus, Banksta! decided to make private sector dealers the inter-mediaries for SHS "nance.

This meant investing considerable time in #ushing outother dealers that would sell the SHS on credit. Whilesome were new to the market, they were selected on thebasis that they already operated in related markets* such as consumer electronics. To ensure that thesenew entrants could access "nance to on-lend to ruralSHS customers, the World Bank invested heavily in theirdevelopment. The loan o$cer reports that many of thenew entrants could not even use a spread-sheet, and thatthe `lack of such competencea was `critical to a bank'sdecisiona. And he goes on to say,

this was heavy duty work for us! We literallybrought the dealers to the stage and then rehearsed

them for their performance in front of thebanks.30

As recognised by the ASTAE adviser, this contrastedstrongly with the India loan:

In Indonesia it has taken a huge e!ort, and a hugeamount of preparatory money and work to get thedealers in place to apply for the loan. In India, wehad neither the time nor the resources to developthis pipeline of borrowers.31

In addition, the World Bank sta! carefully selected fourbanks that indicated a willingness to re"nance the parti-cipating dealers' sales of the SHS on credit.

6.3. The supply side: developing the market infrastructure

Unlike India, the loans in Indonesia were explicitly forworking capital. The dealers could use the loan to both"nance their sales and develop their rural marketingnetwork; for example through investments in motor-cycles, tool kits, training of technicians, etc. As the In-donesia loan o$cer says:

In Indonesia, the provision of working capital bythe Bank is made explicit. The Bank has designateditself in Indonesia as explicitly re"nancing the work-ing capital requirements of Indonesia's dealers.32

The four participating banks in Indonesia required secur-ity on such loans. However, unlike IREDA, where therequirements were for 100% of the value of the loan, inIndonesia, the banks were seeking 2/3 collateral, plus 1/3equity investment by the dealer. Because this was stillstringent, the Bank also identi"ed `private investorsa totake an equity stake in the participating companies, whilethe remaining 2/3 was to be provided in the form of land,panels, and for some banks that agreed, receivables oncredit sales.

In addition, the World Bank introduced a signi"cantpolicy innovation. To attract private sector dealers to theSHS market, and to ensure that unit gross margins weresu$cient for the rapid expansion of the market infras-tructure, the GEF grant was designated as a supply-sidesubsidy of USD 100 per system installed. The result wasthat 14 dealers came forward to participate under theloan, and by the end of 1996, the loan o$cer reported


Fig. 5. Indonesia base case * managing sales on credit. Fig. 6. Further expansion with long-term loan.

Fig. 7. Further expansion with grant per system sold.

that eight of these had received the approval of workingcapital loans from the participating banks.

6.4. Counter-factual scenarios

In the end, Indonesia's "nancial crisis came just daysafter the loan was signed by the Government. Withinterest rates soaring to prop up the Rupiah, selecteddealers could no longer a!ord a working capital loan.Furthermore, devaluation dramatically increased theprice of PV modules, which were imported, and whichmade up roughly 50% of the cost of a system. Today,only one or two dealers are selling the SHS on a cashbasis without accessing the working capital facility. Thus,we cannot look to Indonesia to assess the e$cacy of thepolicy innovations adopted by the Bank. However, wecan refer to counter-factual scenarios to see the potentialinherent in this loan.

6.4.1. Long-term working capital loansAs an indication of the importance of working capital

loans, we can turn to the entrepreneur in West Java whoinspired the Indonesia loan. His rural PV energy servicecompany was able to sell 8000 systems in 2.5 years. Yet,because his company sold 90% of their systems on fouryear credit terms, cash #ow became the single mostsigni"cant constraint on sales (see Fig. 5).

The entrepreneur did not have su$cient cash reservesto "nance a large inventory, and thus there were oftensupply lags. He had been seeking a USD 2 million loanfor several years, but bankers and private investors hadbeen scared away by the rural credit risk. Had the entre-preneur been able to acquire such a loan when he laun-ched the business in January 1994, and had it been usedto double the amount of service centres established, thecompany's historic sales "gures suggests that it wouldhave sold more than 20,000 units in three years. Albeitthat, as in India, this rate of expansion would have stillpresented the company with cash #ow di$culties (seeFig. 6).

6.4.2. Enhanced gross margins with supply-side grantTraining of the participating dealers in not only tech-

nical, but business skills, was the "rst important steptaken by the World Bank in Indonesia to ensure that theparticipating companies were "nancially strong. Yet,even more critical, was the supply-side grant. Due to thecompetition inspired by the loan, the Bank was con"dentthat part of the supply-side grant would be passed on tothe consumer. Assuming that the entrepreneur passed on50% of the grant, his company would have been able toexpand to 20,000 systems sold in three years, and wouldhave been in a more comfortable cash #ow position foryet further expansion (see Fig. 7).

6.4.3. Lower tax on PV modulesFurthermore, unlike India, the World Bank found

a policy environment that enhanced, rather than limited,unit gross margins. Government o$cials were clear thatthey desired in country manufacturing of PV very much,but were not prepared to tax the PV module with animport duty:

we have been proposing a reduction in the tari!since 1991. So far we have managed to have it


Fig. 8. Limited expansion with import duty.

33 Interview with Director of Energy Technology, BPPT, Jakarta, 26July 1996.

34 Interview with consultant to Indonesia loan, World Bank,Washington, DC, 6 October, 1998.

reduced from 5 to 0% on the cell and from 15 to 5%on the module. We are still pushing for 0% on thepanel.33

Had the entrepreneur been faced with 30% importduties as in India, Fig. 8 shows that his company couldnot have even achieved its sales of 8000 systems in 2.5years.

7. Sri Lanka: the 1998 Energy Services Delivery Project

Although the Indonesia loan could not realise its po-tential, the consultant to this loan found that the In-donesia loan o$cer's e!ort assisted in securing approvalfor the 1998 loan to Sri Lanka, and essentially provided`a templatea for the PV component of the Sri Lankaloan.34

7.1. Background to the Energy Services Delivery Project

Roughly half of Sri Lanka's population of 18 millionare unelectri"ed, equating to 2 million households. Ofthese, 300,000 households are estimated to use batteryrecharging facilities for television and radio services, inaddition to kerosene for lighting. In the 1980s, entrepre-neurs entered this market, and demonstrated that suchhouseholds were willing to switch to the SHS. The WorldBank estimates that between 1982 and 1997, approxim-ately 5000 systems (varying from 300 to 700 USD persystem) were sold by various small PV businesses in SriLanka (World Bank, 1998).

In an e!ort to further the di!usion of the SHS, theBank sta! included a PV component in the 1998 Energy

Services Delivery Project (which includes 24.2 millionUSD from the International Development Agency and5.9 million USD from the GEF). The target adoptedunder this loan was to promote sales of 30,000 solarhome systems by the end of 2002. As of September 1999however, only 500 systems have been installed under thisloan.

7.2. The demand side: restricted credit and customerconxdence

On the demand side, the Bank has sought to encourageprivate sector dealers, NGOs and commercial banks to"nance SHS sales. While one entrepreneurial start-upcompany has established a credit scheme for its cus-tomers, others shy away due to the perceived risks ofcustomer default. A similar view is shared by the com-mercial banks, who are also concerned about transactioncosts and lending for what they consider consumptiverather than productive bene"ts of PV. The only excep-tion is a prominent micro-credit NGO, with branchesthroughout the country, which after 3 months of lendingin this sector, is processing up to 90 SHS loans permonth. With only this "nance organisation o!eringcredit for the SHS, companies "nd that demand is limitedto the rate at which this organisation can process andapprove loan applications.

The chairman of a publicly held commercial bank hasstated that it is interested in accessing the World Bankloan facility. Moreover, this bank has the largest numberof rural outlets of any bank in Sri Lanka, making it idealfor lending for SHS sales. However, under the WorldBank scheme, the World Bank has e!ectively precludedthe participation of publicly held banks, stating in theproject document that: `in order to become eligible toparticipate in the ESD Credit Program, and to maintaintheir eligibility, credit institutions must be privatelyowned and controlled2a (World Bank, 1998, Annex 5,p. 1)

A further demand-side restriction has emerged underthe Sri Lanka loan: limited customer con"dence in pur-chasing the SHS due to the possibility of a grid connec-tion. Households are concerned that they will make aninvestment in the SHS only to "nd that their local politi-cian has "nally ful"lled the promise of a connection tothe grid. Although rural people are sceptical of thesepromises, participating companies report that manyhouseholds are unwilling to take the risk of investing 400USD in an o!-grid system based on the lingering hope ofbeing connected to the grid.

7.3. The supply-side: successful policies

The World Bank has been relatively more successful indeveloping the market infrastructure. Participating


35 Interview with managing director of Indian NBFC (non-banking"nancial company) which plans to enter the o! grid solar power marketin West Bengal and Anhra Pradesh, in India, Amsterdam, 31 August,1999.

companies are able to access long-term loans to "nancethe development of their market infrastructures. Further-more, the World Bank has again instituted a grant, thistime graded according to the wattage of the system (fora 45 Wp system and above, companies receive USD 120,while for the smallest 20 Wp system, they receive USD70). Finally, gross margins are further enhanced by im-ports of PV modules at zero percent tari! (although theycould be further improved by the removal of a generalsales tax of 12.5%). Critically, the perceived potential ofthe market, combined with the supply-side grant and lowimports has encouraged two larger players to enter themarket: Shell Renewables, which can access Shell GasLanka's network of over 600 dealers, and a large-scaledistributor of batteries, with a total of over 1400 dealerscountry-wide.

8. Synthesising the policy lessons

The World Bank is progressing with a forthcomingrenewable energy loan to China which includes a PVcomponent. Furthermore, with the reorganisation of theWorld Bank, ASTAE's functions now go beyond Asia, toassisting loan o$cers in other regional divisions. Thus, itseems timely to consider the policy lessons to date.

8.1. Recommended demand-side policies for the SHS

8.1.1. Enhancing the yow of rural creditThe key lesson for the Bank is to take time to prepare

the pipe-line for #ows of rural credit. The World Bankneeds to "rst identify which credit organisations arewilling and able to lend to rural SHS customers. Oncethese organisations have been identi"ed, it is a questionof making sure that the Bank's counterpart * such asIREDA in India* is willing to accept the risk of lendingto these organisations. Preferably, the World Bankshould contractually bind the counter-part organisationand "nancial intermediary prior to disbursement of theloan.

With regard to pre-selection of credit organisations,the Bank should avoid blanket statements that publiclyheld or controlled banks cannot participate under a loan.Often, these institutions have the largest network of ruralbranches and may be in a position to secure a govern-ment guarantee that would satisfy World Bank lendingcriteria. Finding a credit organisation that is willing andeager to lend for the SHS is di$cult * every organisa-tion that can meet the stated "nancial criteria should beenabled to participate.

Finally, to ensure that su$cient credit organisationscome forward to participate, the Bank should considerusing GEF grants to secure a portion of the rural defaultrisk. In India private companies operating in the ruralsector have internal estimates that only 5% of all loans

will result in defaults.35 In such cases, a GEF grant for5% of each SHS loan could #ow to a guarantee fund,which simply reverts to the credit organisation at the endof the loan period.

8.1.2. Transparency in grid extensionYet, as has been shown in Sri Lanka, it will also be

necessary for the Bank to address Government policieson grid extension. PV businesses in Sri Lanka have foundwould-be customers are dissuaded by unrealistic Gov-ernment promises of grid extension. Where the Bank hassome leverage it should require that the national Govern-ment publicise the areas which will not be served by theGrid for "ve years or more, and make a public commit-ment not to extend the grid to these areas for thisduration.

Of course, the reality is that grid extension is a highlypoliticised process, and thus di$cult to regulate. Forinstance, the entrepreneur in Indonesia reported thatpoles were routinely placed in the ground during elec-tions, while the wires would not arrive for another four to"ve years. Where greater transparency is not forthcom-ing from the Government, the World Bank could con-sider a limited buy}back scheme where customers areo!ered the depreciated value of the module alone (as theother components depreciate much more quickly) for 2}3years after their purchase. Given the unreliable nature ofthe grid in all three countries considered, customers arelikely to retain the SHS even after a grid connection, andthus the "nancial burden of a buy}back scheme is likelyto be minimal. In India for instance, the SHS currently"nds its strongest demand in the rural electrixed sector,where chronic, prolonged power shortages mean thathouseholds desire a reliable back-up system, often notprovided by suppliers of 220 V battery inverter systems(Miller, 1998). Whatever remained from a buy}back fundat the end of the loan period, would be handed over tothe Government to continue the scheme.

8.2. Policies to stimulate the supply of SHS

8.2.1. Long-term loansAs shown through company cash #ow simulations in

India and Indonesia, it is critical for the World Bank too!er long-term loans to entrepreneurial start-up com-panies. Where a grant per system sold is o!ered, as inIndonesia, such loans can be o!ered at commercial ratesof interest, though preferably with a year moratorium onrepayment. Start-up companies typically lack su$cientcapital for rapid expansion, in which case a loan becomescritical to rapid development of a market infrastructure



37 Interview with India loan o$cer, World Bank, Washington, DC,2 October 1995. Indeed other sta! have more openly questionedwhether the recent restructuring of the World Bank will allow loano$cers to pursue such a time intensive strategy.

for the SHS. Furthermore, a company with su$cientcapital to develop and sustain its infrastructure, will givecon"dence to banks to begin "nancing SHS customers(Miller, 1998). Of course, there is a risk that low-interestworking capital can be misused. But this can be safe-guarded against with adequate monitoring, and stag-gered disbursement.

As in India, access to such loans can be hindered byexcessive paper-work and bureaucratic procedures. Inaddition, if the security requirements are too stringent,the smaller scale businesses will be precluded from suchloan facilities. Therefore, the World Bank also needs totake time to in#uence the assessment procedures prior todisbursing the loan. To encourage the entry of manylocal, smaller scale players, the Bank should considera ceiling (e.g. 50,000 USD), under which loans would be`fast-trackeda and require limited security.

8.2.2. Business advisory servicesThe Bank needs to adopt a holistic approach to advis-

ory and support services which sees the establishment ofin-country incubators for PV businesses. It is by nomeans su$cient to concentrate on technical support.Given the importance of unit gross margins, companiesneed to be advised on lowest cost suppliers of essentialcomponents, such as batteries and charge controllers. Inaddition, sta! from companies that propose to managetheir own credit scheme should undergo a mandatorytraining exercise in rural loan collection.

8.2.3. Eliminate taxation on the systemThe provision of a World Bank loan and GEF grants

should be made conditional on eliminating, or at min-imum, phasing out import duties and all local taxes onPV modules and complete systems. Not only do suchtaxes harm the consumer, but critically, for companiesoperating in price-sensitive markets, they limit unit grossmargins. It is clear from "nancial modelling in India andIndonesia, that changes to import duties on PV modulesstrongly a!ected the rate at which the start-up enterprisecould expand their sales.

8.2.4. Supply-side grantA supply-side grant per system sold is extremely bene-"cial to the rapid development of a market infrastructurefor the SHS. First it attracts new players to the market,and second it allows participating companies to expandrapidly. Moreover, where there is competition, such asIndonesia and Sri Lanka, there is pressure to pass on partof the grant to the customer. Furthermore, where alltaxes on PV systems and modules are eliminated,

the World Bank could a!ord to reduce the supply-sidegrant.

9. The future of World Bank lending for o4-grid solarpower

9.1. The demands of lending for ow-grid solar power

Of course, it remains to be seen whether further loansto the rural PV sector has broad-based appeal amongWorld Bank loan o$cers. As the ASTAE adviser to theIndia and Indonesia loans now warns:

[loan o$cers] must plan for and be prepared tojustify increased project preparation time as theseactivities take considerable e!ort and cost.36

The costs of preparing the Indonesia loan, for instance,were managed by bilateral aid agencies, which should notpose a problem for future lending. However, the time ande!ort involved may prove a more intractable issue. As theIndia loan o$cer re#ects, the PV component took themost `micro-managinga since `the rural market in par-ticular needs to be scaled upa, and that fundamentally,this is at odds with Bank convention:

the Bank does not normally do this much micro-managing for a project. The Bank would ideally justlike to provide the money and leave it to the mar-ket237

9.2. Enter the International Finance Corporation

The IFC has now entered the o!-grid solar powermarket through a host of initiatives, which are brie#ydescribed below:

1. The renewable energy and energy ezciency fund. Toprovide a satisfactory return, the majority of this fundwill be invested in large-scale grid connected wind-farm,and small hydro projects. However up to 20}30% of themoney has also been set aside for smaller, o!-grid sectorloans. Given that such loans are projected to achievereturns of 10% or less, the GEF has approved the alloca-tion of USD 30 million for `smaller more complex,


38 Interview with lead consultant to the REEF, IFC, Washington,DC, 12 December 1996.

39For background information on PVMTI see IFC (1995).

leading edge projects, with proportionally higher trans-action costsa.38

2. The IFC/GEF photovoltaic market transformation in-itiative (PVMTI)39. The IFC and GEF have selected threecountries* India, Kenya, and Morocco* based on thestrength of private sector sales of SHS technology anda government commitment to the PV market. Under thisscheme USD 30 million is available to re"nance con-sumer credit schemes, provide long-term low-interestloans to participating companies, and provide a grant ofup to 10% of the total loan for one o! market develop-ment costs.

3. The IFC/GEF small-to medium-scale enterprise pro-gramme. The GEF has provided a grant of USD 20.8million for debt and equity loans to small- and mediumscale enterprises who, through their activities, are reduc-ing emissions of greenhouse gases. To date this pro-gramme has invested a total of USD1.6 million in threerural PV electri"cation projects: to a leasing company inthe Dominican Republic; to a Bangladeshi NGO; and toa Vietnamese co-operative involved in "nancing PV sales.

4. The Solar Development Corporation (SDC). The SDCis an entirely independent, separate holding company,with its own management structure. The IFC will investin the fund with additional funding provided by theWorld Bank, the GEF, and private investors. The initialcapitalisation goal is USD 50 million, of which USD18 million will be used for business advisory services,while the balance will be used for low interest, mediumterm "nancing of organisations that sell and "nance PVsales (IFC, 1998).

Certainly, such schemes o!er great hope to the rapiddi!usion of SHS technology, given that the IFC canmove more quickly than the traditional lending arms ofthe World Bank, can lend directly to private sector or-ganisations selling and "nancing the SHS, and can lendfor relatively small amounts, using GEF grants to coverthe transaction costs. However, the IFC cannot lend topublicly owned rural banks, which, as we discussed, areoften well suited to "nance the SHS. Furthermore, it hasnot been willing or able to o!er a supply-side grant toclients per SHS installed * a policy step that it shouldcertainly consider. Finally, it is not clear that the IFC hasthe same leverage in reducing import duties and othertaxes on the SHS. In the absence of the IFC being able toimplement the full range of policy measures recommen-ded in this study, traditional lending arms of the WorldBank, such as the IBRD and IDA should continue toparticipate in the o!-grid solar power market.

9.3. Conclusions

The di$culties experienced by the World Bank inlending for the SHS to date should not dissuade it fromstaying the course. As we have seen, the SHS is a morecost-e!ective solution for many rural households com-pared to equivalent services, it does much to enhancetheir quality of life, and it provides a new source of skilledemployment for many rural technicians. It was inevi-table, given the decentralised nature of o!-grid solarpower, that the Bank would need time to learn how tosuccessfully integrate and implement demand and sup-ply-side policies in support of the SHS.

Indeed, from the case of o!-grid solar power, it istempting to draw yet broader conclusions. Namely, thatwhere technological innovation necessitates policy in-novation, the organisational learning process insidemultilateral and bilateral lending organisations is inevi-tably a factor a!ecting the speed with which developingworld economies make the transition to renewable en-ergy technologies.

Acknowledgements

The authors would like to acknowledge the invaluablesupport of World Bank sta! who since 1994 have co-operated with repeated requests for interviews. The viewsexpressed in this document are not necessarily those ofShell Renewables, Ltd. The authors would also like tothank the MacArthur Foundation and the former GlobalSecurity Programme for the funding that made this re-search possible.

Appendix A. Assumptions for life-time cost analysis

(a) Assumptions common to all countries for solar homesystem

f All costs in 1999 prices.f 37 Wp module assembled in India.f 6 A charge controller assembled in Indonesia.f 2 light "xtures from Indonesia: 10.5 USD.f Module, bracket, wire, switches: 19 USD.f 12% real discount rate.f Company's unit gross margin: 20%.f Dealer/sales commission: 10%.f 15 USD installation fee.f 3 USD annual service fee.f Lifetime of battery: 3 years.f Lifetime of charge controller: 5 years.f Lifetime of light "xtures: 3 years.


(b) Assumptions distinctive to particular countries for solar home system

India Sri Lanka Indonesia

1 USD"43 Indian rupees 1 USD"70 Sri Lankan rupees 1 USD"8000 Indonesian rupees40% import duty on import of PVcells and modules

0% import duty on PV modules;12.5% sales tax

5% import duty on PV modules

25% import duty on systemcomponents for PV plus clearing

0% import duty on balance ofsystem components; 12.5% sales tax

System components sourced in country

37 Wp PV module"141 USD(assembled in India)"

37 Wp PV module"139 USD(imported from India)

37 Wp PV module"139 USD(imported from India)

6 A charge controller"45.17 USD 6 A charge controller"36.8 USD 6 A charge controller"35.12 USD100 A h battery"46.8 USD(imported)

90 A h battery"43 USD (sourcedlocally)

100 A h battery"34.7 USD (sourcedlocally)

Total up-front cost of System"399USD

Total up front cost of system"392USD

Total up front cost of system"359USD!

!Note: 1999 prices used for SHS in Indonesia, as prices have generally increased in line with devaluation. By comp-arison, kerosene and battery charging prices are now distorted (see below), for which 1996 prices are used.

"The price of PV module is more expensive to use in India as the PV cells are charged an import duty. When modulesare assembled in India and exported, PV cells do not attract an import duty, and thus the same module cheaper in SriLanka and Indonesia despite shipping and clearing costs.

(c) Assumptions for diesel genset in India

f To compare with the SHS, diesel genset calculationsinclude two incandescent lights plus installation/wiring.

f The lowest wattage diesel genset that is readily avail-able to rural customers in India is 500 kva

f Locally available 500 kva diesel gensets are low cost;priced at 186 USD.

f Reported life time of these gensets can be as low as twoyears; our calculations assume 5 years

f Annual cost of 2 incandescent bulbs: 3.5 USDf Installation and wiring fee: 15 USD.f Price of litre of diesel fuel: 0.35 USD.f Diesel genset is assumed to run at rated capacity for

4 hours per day in line with hours of lighting andentertainment required by a typical rural household inIndia

f Locally available 500 kva gensets are highlyine$cient, reported to consume 0.75 litres per hour; weconservatively assume consumption of 0.5 litres perhour

f Diesel genset assumed to run every day of the year;annual consumption of 730 litres

f Maintenance costs per annum, including oil changesand repairs: 9.3 USD (5% of genset price).

f Rural customers have a further choice of 500 kva kero-sene genset, which are also widely available; up frontcosts are greater } 409 USD } but consumption isroughly the same and fuel costs are lower !0.28USD. Making the same assumptions as above, themonthly amortised cost of a 500 kva kerosene genset is33.8 USD.

(d) Assumptions common to all countries for kerosene/bat-tery charging

f 12% real discount rate.f 4 h usage of petromax and chimney lantern per

day.f Lifetime of chimney lantern: 5 years.f Lifetime of petromax lantern: 5 years.f Consumption of petromax: 0.13 l per hour.f Consumption of chimney lantern: 0.03 l per hour.f 70 Ah car battery used for recharging.f Lifetime of car battery: 2 years.f One recharge per week: 52 per year.f Transportation costs of battery to charging station

not included.


(e) Assumptions distinctive to particular countries for kerosene/battery charging

India (USD prices) Sri Lanka (USD prices) Indonesia (USD prices)

1999 prices, 1 USD"43 Rs 1999 prices, 1 USD"70 Rs 1996 prices, 1 USD"2330Price of kerosene per litre"0.12 Price of kerosene per litre"0.105 Price of kerosene per litre"0.20Basic Chimney lantern"5.1 Basic chimney lantern"4.6 Basic chimney lantern"4Maintenance costs of basicchimney lantern"0.7 per annum

Maintenance costs of basic chimneylantern"0.7 per annum

Maintenance costs of basic chimneylantern"0.7 per annum

Petromax lantern"27.9 Petromax lantern"18.57 Petromax lantern"20Maintenance costs of petromaxlantern"2.9

Maintenance costs of petromaxlantern"2.9

Maintenance costs of petromaxlantern"2.9

70 Ah car battery"51.2 70 Ah car battery"43 70 Ah car battery"40Per charge of 70 Ah carbattery"0.5

Price per charge of 70 Ah carbattery"0.5

Price per charge of 70 Ah carbattery"1

Note: 1996 prices used for kerosene and battery charging, as current prices are wildly distorted: despite devaluation,kerosene is lower nominal price in 1999 than 1996 due to government intervention, while battery charging is currentlyreported at three times 1996 price.

Appendix B. Assumptions for cash 6ow analysis of PVstart-up companies

(a) Assumptions for Financial Modelling of India PVCompany

For Fig. 1f 1 USD " Rs 34 (1996 rate of exchange).f Company launched October 1994.f Monthly cash #ow analysis until end of 1996.f USD 26,000 start-up capital.f 250 four light 35 Wp SHSs sold by end of 1996: 60

systems sold through rural bank, who paid India PVcompany within one month. Rest of the systems soldon cash basis.

f 50% drop in sales during the monsoon season (fromJuly through September).

f Unit gross margin per system including installa-tion"USD 85.29.

f 20 systems o!ered on credit in "rst month"3 yearscredit, 10% down-payment, 7% rate of interest perannum; 2 of 20 customers defaulted.

f One service centre established in 1994, one in 1995,and one in 1996.

f At head-quarters, costs rise by 10% per year to ac-count for increase travel and telecommunications bythe entrepreneur.

f At service centres, operational costs rise by 15% peryear to account for greater fuel use with increasedmarketing and installations, and a general rise in theprice of the fuel.

f Annual increase of 15% in service centre operatingcosts per annum

f Due to cash #ow constraints company maintains lim-ited inventory of only 5 systems

For Fig. 2f USD 50,000 long-term working capital loan.f Disbursed up-front by IREDA with one year mora-

torium on repayment. Interest of 3.5% per annum(2.5% rate plus 1% administration fee) over 10 years.

f Company invests in jeeps which allows for more doorto door sales. Average sales per service centre permonth in base case.

f 5.6 sales per month, per service centre, based on actualhistoric sales of company between October 1994 andJune 1996.

f 1 service centre added in 1994, 3 service centres addedin 1995, and six service centres added in 1996.

f Cumulative sales of 830 four light solar home systems(35 Wp).

For Fig. 3f Same as Fig. 2 except 25% reduction in cost of battery.f Unit gross margin increases to 118 USD per system sold.

For Fig. 4f Same as Fig. 2 except 30% reduction in cost of mod-

ule.f Unit gross margin increases to USD 158 per system

sold.

(b) Assumptions for xnancial modelling of Indonesia PVCompany

For Fig. 5f 1 USD "2330 Rupiah (1996 rate of exchange).f Two sources of seed capital: a personal investment of

$500,000 and 1000 systems from management buy-out.f Company launches beginning of 1994 and begins to

sell systems directly to rural customers on credit inMay of 1994 (month "ve).


f Customer credit is o!ered at 30% per annum, withminimum 25% down-payment, repayment over4 years, with monthly instalments.

f The most popular credit scheme was the 4-year term,with 25% down-payment. This is used as standardcredit sale for purpose of the model.

f Average sales per service centre"200 per annum.f Sales at end of 1996"10,090.f Service centres opened by end of 1996"50.f Credit is provided for the system * not installation

which is paid in cash. Fee for installation"13 USD.f 90% of the systems sold were the 50Wp SHS*model

takes as standard system sold.f Company has only 5% month's sales as inventory.f Receives 6 months supplier credits after starting May

1994.f 20% annual increase in headquarter operating costs

and 20% annual increase in service centre operatingcosts * higher than India to account for much morerapid rate of expansion per annum.

f The entrepreneur raised the price from 890,000 Rp to990,000 Rp after the end of 1995. Unit gross marginper system increases from 103 USD to 129 USD persystem.

For Fig. 6f $2 million working capital loan in month one.f Terms of repayment of long-term loan: as per terms

under 1998 Indonesia loan, the loan repaid quarterlyat an annual rate of 20% interest, over three years.

f Company reduces down-payment from 25 to 15% toincrease sales. 100% increase in sales over levels in Fig. 5.

f Inventory of 15% on one months sales to allow formore rapid expansion.

f The enterprise sells 20,180 over 3 years, with a total of101 service centres.

f Increase in operating costs at head-quarters and ser-vice centres from 20% to 30% per annum, respectively.

For Fig. 7f Same as Fig. 6 except 100 grant per system sold now

available.f Company increases inventory to 50% of one month's

sales.f Company passes on 50 USD to customer to reduce up

front price and keeps 50 USD to improve unit grossmargins.

For Fig. 8f Same as Fig. 6, except imposition of 30% import duty

on exiting price of PV module.f Unit gross margin until end of 1995 reduced to USD

60, and for year 1996 reduced to USD 103.

References

Acker, R., Kammen, D., 1996. The quiet energy revolution; analysingthe dissemination of photovoltaic power systems in Kenya. EnergyPolicy 24 (1), 81}111.

Agarwal, B., 1983. Di!usion of rural innovations: some analytical issuesand the case of wood-burning stoves. World Development 11 (4),359}376.

Ahmed, K., 1994. Renewable Energy Technologies: A Review of theStatus and Costs of Selected Technologies. The World Bank, Wash-ington, DC.

Anderson, D., 1992. The Energy Industry and Global Warming; NewRoles for International Aid. Overseas Development Institute.

Anderson, D., Ahmed, K., 1995. The Case for Solar Energy Invest-ments. World Bank, Washington, DC.

Barozzi, L., Guidi, D., 1993. Prospects for the Expansion of Solar PVTechnology in the Developing World; Financial Mechanisms andTechnology Transfer Optimisation. WWF, Italy.

Bosshard, P., 1994. Energy from Dante's Inferno; A Memorandum ofthe Berne Declaration and Greenpeace Switzerland on the Contra-dictory Energy Policies of the World Bank and the GEF in India.Erklarung von Bern, Zurich, 1993.

Brown, L., 1981. Innovation Di!usion. Methuen, New York.Cabraal, A., et al., 1996. Best Practices for Photovoltaic Household

Electri"cation Programs. World Bank, Washington, DC.Deviah, M.A., 1995. India may lose $55 m WB loan. Indian Express,

Bangalore, 3 June.Drennen, T.E. et al., 1996. Solar power and climate change policy in

developing countries. Energy Policy 24 (1), 9}16.ECN, 1995. Market Development for Solar Home Systems. Nether-

lands Energy Research Foundation (ECN), The Netherlands.Erickson, J., Chapman, D., 1995. Photovoltaic technology: markets,

economics, and rural development. World Development 23 (7),1129}1141.

Foley, G., 1995. Photovoltaic Applications in Rural Areas of the Devel-oping World. World Bank, Washington, DC.

Griliches, Z., 1957. Hybrid corn: an exploration in the economics oftechnological change. Econometrica 25 (4), 501}522.

GTZ, 1995. Basic electri"cation for rural households, GTZ, Eschborn,Germany.

Hankins, M., 1993. Solar Electri"cation in the Developing World, FourCountry Case Studies: Dominican Republic, Kenya, Sri Lanka, andZimbabwe. SELF, Washington, DC.

Hankins, M., 1995. Solar Electric Systems for Africa. Nairobi,Kenya.

Harrison, S., 1994. Technology adoption in rural development: the caseof non-adopters with special reference to credit. In: Roy, K., Clark,C. (Eds.), Technological Change and Rural Development in PoorCountries. Oxford University Press, Calcutta.

Hassing, P., Mendis, M., 1992. Indian renewable Energy DevelopmentAgency. Evaluation Report, Ministry of Development Corporation,September.

Havens, A., Flinn, W., 1975. Green revolution technology and com-munity development; the limits of action programs. Economic De-velopment and Cultural Change (23), 469}481.

Heertje, A., 1983. Can we explain technical change? In: MacDonald etal., (Eds), The Trouble with Technology; Explorations in the Pro-cess of Technical Change. Frances Pinter, London.

Hill, R. et al., 1994. The successful implementation of photovoltaics indeveloping countries. Presented at The 12th European PhotovoltaicSolar Energy Conference, Amsterdam, April. 1994.

Hurst, C., 1990. Establishing new markets for mature energy equipmentin developing countries: experience with windmills, hydro-poweredmills, and solar water heaters. World Development 18 (4), 605}615.

Huacuz, J., 1991. Rural electri"cation with renewable energies inMexico: "nancial, technical, social and institutional challenges. Pre-sented at SADCC Annual Technical Seminar, Swaziland, 26-28November.

Hughes, T., 1979. The electri"cation of America: the system builders.Technology and Culture 20, 124}161.

IFC, 1995. Cultivating the Green Carrot: A Market Stimulus forPhotovoltaic Technology, IFC/GEF, Washington, DC.


IFC, 1998. The Evolution of the Solar Development CorporationConcept in Relation to World Bank Group/GEF Financing ofSolar Photovoltaic Technology and Related Experiences. IFC Envi-ronment Division, Washington, DC.

Jain, et al., 1991. Innovation di!usion in the presence of supply restric-tions. Marketing Science 10, 83}90.

Jhirad, D., Mintzer, I., 1993. Electricity: technological opportunitiesand management challenges to achieving a low emissions future. In:Mintzer, I. (Ed.), Confronting Climate Change: Risks, Implications,and Responses. Cambridge University Press, Cambridge.

Johansson et al., (Eds.), 1993. Renewable fuels and electricity for a grow-ing World Economy. In: Renewable Energy: Sources for Fuels andElectricity. Earthscan Publications, London.

Jones, G., 1982. Reviews: Brown, Lawrence, A. Innovation di!usion:a new perspective. Journal of Agricultural Economics 33 (2),252}253.

Kearl, B., 1976. Communication for Agricultural Development. In:Schramm, W., Lerner, D. (Eds.), Communication and Change: TheLast Ten Years and the Next. The University of Hawaii Press,Honolulu.

Lovejoy, D., 1992. Electri"cation of rural areas by solar PV. NaturalResources Forum, May, 101}110.

Messenger, H., 1994. Models for success: the Asia experience. Presentedat Conference on Renewable Energy in the Americas '94, PuertoRico, 26 July-1 July.

Miller, D., 1998. Agents of sustainable technological change; the case ofsolar electri"cation in the developing world. Ph.D. Thesis. Cam-bridge University.

Morse, B., 1992. Sardar Sarovar: The Report of the Independent Re-view. Resources Future International.

NREL, 1994. NREL India Initiative; Phase One Visit Report. NationalRenewable Energy Laboratory (USA), Golden, CO.

O$ce of Technology Assessment, 1992. Fuelling Development: EnergyTechnologies for Developing Countries, United States Congress,Washington, DC.

Rogers, E., 1995. Di!usion of Innovations, 4th Edition. The Free Press,New York.

Rogers, E., Kinkaid, L., 1981. Communication Networks: Towardsa New Paradigm for Research. Free Press, New York.

Rogers, E., Shoemaker, F., 1971. Communication of Innovation;A Cross Cultural Approach. Free Press, New York.

Roling, N., 1982. Alternative approaches in extension. In: Jones, G.,Rolls, M. (Eds.), Progress in Rural Extension and CommunityDevelopment, Vol. 1. Wiley, New York.

Roy, K., 1994. Neglected issues in technological change and ruraldevelopment: an overview. In: Roy, K., Clark, C. (Eds.), Technolo-gical Change and Rural Development in Poor Countries. OxfordUniversity Press, Calcutta.

Simon, H., Sebastian, K., 1987. Di!usion and advertising: the Germantelephone campaign. Management Science 33, 451}466.

Solar Industry Journal, 1991. The limitations of diesel power. SolarIndustry Journal, 26}43.

Van der Plas, 1994. Solar Energy Answer to Rural Power inAfrica (FPD Note No. 6). The World Bank, Washington, DC, April.

World Bank, 1975. Rural Electri"cation; Policy Paper, Washington,DC.

World Bank, 1992a. Sta! Appraisal Report; India Renewable Re-sources Development Project (Report No 11240-IN). World Bank,Washington, DC, November.

World Bank, 1992b. Global Environment Facility Recommendation tothe Regional Vice President on a Proposed Grant from the GlobalEnvironment Trust Fund for an Alternate Energy Project. (ReportNo 11315-IN). Washington, DC, November.

World Bank, 1993. India: Pre-Investment Study of the PV MarketDevelopment Project. (Vols. I and II). Washington, DC,January.

World Bank, 1996. World Bank Mission to India Nov 4-15; IREDAReview. World Bank, Washington, DC.

World Bank, 1998. Republic of Sri Lanka; Energy Services DeliveryProject. World Bank Project Document, Washington, DC.


learning to lend for off-grid solar power: policy lessons from world bank loans to india, indonesia,...

Documents