Estimating Greenhouse Gas Emissions Reducitons from Replacing Fuel-Based Lighting with LED

From carbon to light: a new framework for estimatinggreenhouse gas emissions reductions from replacingfuel-based lighting with LED systems

Evan Mills & Arne Jacobson

Received: 13 September 2010 /Accepted: 16 March 2011 /Published online: 21 April 2011# Springer Science+Business Media B.V. 2011

Abstract There is considerable well-intended, yetwishful anticipation about reducing greenhouse gasemissions by replacing fuel-based lighting in thedeveloping world with grid-independent light-emittingdiode (LED) lighting systems. Most estimates glossover important practical realities that stand to erode agenuinely significant potential. The Clean DevelopmentMechanism (CDM) is the leading system for quantify-ing the benefits of such projects in developing countriesand embodying them in a market-based platform fortrading carbon credits. However, compliance withmethodologies for highly decentralized, small-scaleenergy saving projects currently employed in theCDM is viewed by developers of as onerous, time-consuming, and costly. In recognition of the problem,the CDM has recently placed priority on improvedmethodologies for estimating carbon dioxide reductions

from displacement of fuel-based lighting with energy-efficient alternatives. The over-arching aim is tomaintain environmental integrity without stifling sus-tainable emission-reduction projects and programs inthe field. This article informs this process by laying out anew framework that shifts the analytical focus fromhighly costly yet narrow and uncertain baseline estima-tions to simplified methods based primarily on deemedvalues that focus on replacement lighting system qualityand performance characteristics. The result—manyelements of which have been adopted in a newmethodology approved by the CDM—is more struc-tured and rigorous than methodologies used for LEDprojects in the past and yet simpler to implement, i.e.,entailing fewer transaction costs. Applying this newframework, we find that some off-grid lighting technol-ogies can be expected to yield little or no emissionsreductions, while well-designed ones, using productsindependently certified to have high quality anddurability, can generate significant reductions. Enfold-ing quality assurance within the proposed frameworkwill help stem “market spoiling” currently underway inthe developing world—caused by the introduction ofsubstandard off-grid lighting products—thereby ensur-ing carbon reduction additionality (emissions reductionsthat would have not occurred in the absence of theCDM program).

Keywords Clean development mechanism .

Energy efficiency . LED lighting . Carbon emissions .

Developing countries

Energy Efficiency (2011) 4:523–546DOI 10.1007/s12053-011-9121-y

E. Mills (*)Lawrence Berkeley National Laboratory,1 Cyclotron Road,Berkeley, CA 94720, USAe-mail: [email protected]

A. JacobsonOffice of Policy and International Affairs,US Department of Energy,1000 Independence Avenue SW,Washington, DC 20585, USA

A. JacobsonHumboldt State University,Schatz Energy Research Center,Arcata, CA, USA

Introduction

One in four people on Earth lack electricity, derivingillumination for their homes—and often their busi-nesses—from the flame.1 According to the Interna-tional Energy Agency, in the absence of seriousintervention, this number will decline only slightlyover the following two decades and will in fact risesignificantly in sub-Saharan Africa (World EnergyOutlook 2009).

While off-grid lighting users spend nearly$40 billion per year (almost 20% of all global lightingexpenditures) on ad hoc and polluting methods forobtaining illumination, they receive only 0.1% of thetotal lighting services consumed by the electrifiedworld (Fig. 1). The carbon dioxide emissions emittedin producing this inferior illumination are equivalentto that of about 30 million cars.2

As such, the primary by-product of illuminatingnon-electrified homes and businesses in the develop-ing world with fuels is greenhouse gas emissions andonly secondly light. This state of affairs contributes topoverty as well as to global warming. Other adverseimpacts include suppressed productivity when fuel-based lighting is used in market or productioncontexts, adverse effects on health through burn risksand indoor air pollution (Apple et al. 2010),3 poorreading conditions, excessive costs for unelectrifiedbusinesses, and reduced nighttime security. With acombination of new technologies and appropriatemarket-delivery solutions, this situation can be reversedto a profound degree.

The single-most promising of these new technolo-gies are solid-state light sources that emit white light,or, as they are more commonly known, light-emittingdiodes (LEDs; Mills 2005; Lighting Africa 2010).LEDs offer many attributes that are superior to orotherwise provide a more appropriate fit todeveloping-country lighting needs than fluorescentlighting technology—which, prior to the advent ofLED lighting, has been rightfully promoted as the

best way to improve efficiency in comparison totraditional incandescent or fuel-based light sources.

Properly applied, the additional energy savingsfrom LEDs compared to fluorescents can be on a parwith those historically gained through the conversionfrom incandescent to fluorescent lamps. LEDs alsooffer a number of other attributes that are highlydesirable in a developing-country context, including:long service life, ruggedness, absence of mercury,low-voltage operation, compact/portable size, and aform factor well-suited toward directing light on therequired task with very high optical efficiencies.

Although diffuse compared to lighting markets inthe industrialized world, the existing fuel-based off-grid lighting market is present in nearly 200 countriesand has an established distributed energy deliverysystem (typically kerosene).

Low-income consumers in developing countrieshave demonstrated the ability to adopt new lightingtechnologies rapidly. For example, 90% of flashlightsin parts of Kenya are now based on LEDs (Johnstoneet al. 2009). Properly designed and manufactured, awide diversity of LED lighting systems could displace

1 This article builds on Mills (2010).2 At US average conditions of approximately 20,000 vehiclekilometers traveled per year at 0.81 l/100 km.

Fig. 1 Lighting cost and services for electrified and un-electrified populations (Mills 2005)

3 Note that many proponents of off-grid lighting conflate thedocumented health impacts and mortalities associated withfuelwood with those from lighting. However, while cook stovesno doubt pose a far greater threat to health and life than dolighting fuels, those from lighting are not trivial.

524 Energy Efficiency (2011) 4:523–546

large quantities of kerosene lighting. This shift couldbe driven by dramatically lower operating costs and ahost of other end-user benefits. The efficiency of fuel-based lighting strategies can be as low as 0.04 lm/Wor less than 1/1000th that of a modern LED lightsource. Put another way, a typical kerosene lanternconsumes kerosene at a rate of approximately 200 W,while a quality LED-based lamp using 1 W ofelectricity produces five times the light output.

Eliminating and monetizing all greenhouse gasemissions associated with global fuel-based lightingwould correspond to as much as a $4 billion annualmarket.4 At the level of the individual consumer, theper-lantern value of the carbon offset could be asubstantial fraction of the improved lantern’s cost,providing a compelling impetus for large-scale markettransformation.

The Clean Development Mechanism (CDM) hasbeen instrumental in creating a massive and fast-growing market for carbon emissions reductions.5

Governments of industrialized countries can useCertified Emissions Reductions (CERs) from CDMprojects to fulfill their commitments under the KyotoProtocol. Likewise, companies can use them inemissions trading systems like that introduced in theEU. The tradable emissions can be supplied bycertified projects that improve energy systems indeveloping countries. The financing of projects thatdeliver CERs into these markets in theory overcomemarket barriers and failures that would otherwisethwart investment in low-emissions energy systems inthe developing world.

Currently, however, CDM projects addressingsmall-scale emissions such as those in householdlighting (on or off the electrical grid) are playing avanishingly small role in carbon-trading markets(Michaelowa et al. 2009). This is due in part to thehigh transaction costs of attaining and documentingthese savings in comparison to larger centralizedprojects such as those in the power or industrialsectors. Two proposed off-grid lighting projects (both

in India) have recently been approved for CDMcredits.6 This article explores means for fosteringincreased activity via an improved and less onerouscarbon-accounting methodology than those utilizedby the CDM thus far.

Greenhouse gas emissions from fuel-based lighting

People without access to electricity grids (or distrib-uted electricity generation) obtain light in a remark-able variety of ways. The predominant fuel iskerosene, but other ubiquitous sources include diesel,candles, various forms of biofuels, and battery-powered flashlights. Users commonly employ morethan one type of fuel and consume them in varioustypes of lamps (Fig. 2a–b). Patterns differ by countryand at far smaller scales within countries. Each lamp–fuel combination results in a different carbon intensity(emissions per hour of utilization) and lighting servicelevel. Figure 3 provides an example limited to afamily of kerosene-burning lamps demonstratingdifferences in fuel-use rates and associated carbonemissions.

The single published global estimate of greenhousegas emissions from fuel-based lighting places thevalue at 190 million metric tons of CO2 per year(Mills 2005). This could well be an under estimate asit did not explicitly include biomass, other greenhousegasses, or the global warming potential of associatedblack carbon (“soot”), which is not treated as agreenhouse gas under the Kyoto Protocol. Non-household uses were only roughly estimated, andresults were not broken out by geography ordemographic factors. The estimate is also built upfrom the nominally unelectrified population, whereaselectrified households and businesses revert to kero-sene during power outages, which are frequent inmany areas. In a recent market test in Kenya, just over25% of those who purchased LED lamps intended toreplace kerosene lanterns occupied homes that wereon the grid (Tracy et al. 2010a).

The intensity of use varies widely within countriesand even specific demographics (Fig. 4a–b). Acompilation of 28 surveys from around the world

4 This amount derives from estimated carbon emissions of190 MT CO2/year (Mills 2005) at the current selling price ofapproximately US$20 per ton CO2.5 While this article focuses on the CDM, the principlesdeveloped herein apply equally well to the various voluntarymarket emissions-reduction systems, and could in fact add rigorto such programs and thus increase the valuation of carbonoffsets they attain.

6 S e e h t t p : / / c dm . u n f c c c . i n t / P r o j e c t s /DB /TUEV-SUED1245158196.62/view and http://cdm.unfccc.int/Projects/DB/DNV-CUK1226479189.57/view.

Energy Efficiency (2011) 4:523–546 525

http://cdm.unfccc.int/Projects/DB/TUEV-SUED1245158196.62/view

http://cdm.unfccc.int/Projects/DB/TUEV-SUED1245158196.62/view

http://cdm.unfccc.int/Projects/DB/DNV-CUK1226479189.57/view

http://cdm.unfccc.int/Projects/DB/DNV-CUK1226479189.57/view

showed a variation of 3–30 l/month of lighting fueluse (Mills 2005). The drivers of these wide differ-ences are not primarily attributable to geography.7 Forexample, in Ghana (and no doubt elsewhere), somenight vendors use lamps with very large wicks thatconsume fuel at the rate of 0.06 l/h. This use rate,combined with very long hours of use, results inannual fuel consumption of about 180 l/lantern ascompared to approximately 20 l/lantern for ordinary

households using conventional lamps for shorterperiods of time each day.

In the case of non-renewable biomass, the amountsof net greenhouse gas emissions have not beenquantified. Biofuel light sources include raw plantand wood fuels (from grass to resins), vegetable oil,biogas, yak butter, and animal oils. Highly resinousplants (e.g., the African Olive) are used exclusively forlighting. Wood cooking fires are used to an unknowndegree for illumination globally, and at least in somecontexts for this reason are burned longer than requiredfor cooking. The degree to which these fuels aresustainably produced versus net carbon producers hasnot been quantified. Of the five countries surveyed by

7 However, geography can be taken into consideration forthings like baseline lighting fuel mix, prevailing fuel prices, andwillingness to pay for alternative technologies.

Fig. 2 a–b Wide variancein the types of lightingsources used by consumersand traders (night marketvendors) in off-grid areas.These country specific datawere collected by LightingAfrica (2009, 2011) throughsurveys of 2,831 consumersand 1,261 traders. In mostcases, users employ morethan one type of light source(totals>100%). Consumervalues are for light used theprevious night


the Lighting Africa Project, half the households reportusing fuelwood daily for illumination, and two thirdsreport doing so two or three times a week (Fig. 5). Asseen in Fig. 2a, nearly 20% of homes in Ethiopia reportusing these fuels for lighting (Lighting Africa 2009,2011). The Tanzania household survey reports that 7%of rural households use firewood as a primary sourceof lighting, and the value runs as high as 24% in onedistrict (National Bureau of Statistics Tanzania 2002).

The potential for LED replacement technologies

For multiple reasons, properly designed and manu-factured LED lamps are vastly superior to theincumbent fuel-based lighting technology, as well ashighly appealing alternatives to the nearest competingtechnology alternative (fluorescent or compact fluo-rescent lamps, CFLs):

& Unlike most other lighting technologies, which havematured and reached efficiency plateaus, LEDs forwhite light are relatively new and are undergoingrapid efficiency improvements and cost reductions.

& They are much more rugged and longer-lived thanfluorescent lamps.

& They provide better quality illumination forcertain tasks.

& At over 100 lm/W, LED peak efficiencies havealready surpassed those of CFLs, and the USDepartment of Energy has set a target of 165 lm/Wby the year 2025 (USDOE 2009).

& Low-power requirements mean that chargingsystems and batteries can be much smaller thanthose in conventional household solar electriclighting systems (e.g., AA size rechargeablebatteries instead of batteries of similar size tothose used in motorcycles and cars).

& A low-voltage platform is especially suitable for asolar power supply.

& Products are typically portable “integrated sys-tems” (including lights, charging, and storage),requiring no assembly and, if well-made, relative-ly little maintenance in the field.8

& Grid-independent lighting systems are not subjectto the risks of voltage fluctuation that have created

Fig. 3 Rates of fuel use vary widely among lamps. Thoseshown in this figure vary from 0.018 to 0.089 l of kerosene perhour. Annual estimates are based on 4-h/day use, a 5-year timehorizon, an emissions factor of 2.4 kg CO2/l, and an illustrativeemissions price of $20/ton CO2. Note that the vast preponder-

ance of users fall into the small or medium wick or hurricanelamp categories, implying a factor of three variance in fuel-userates. Source: Field measurements—timed using a digitalbalance—by Lawrence Berkeley and Humboldt State Univer-sity (Lumina Project)

8 The analysis in this report focuses on integrated systems.Custom-made LED lighting systems (e.g., with technician-installed batteries, lights, and charging devices) are notcommon in this market and the associated risks would need tobe treated in the CDM assessment framework in much the samefashion as traditional Solar Home Systems.


uncertainty as to the service life of grid-connectedcompact fluorescent lamps in prior CDM efforts(Michaelowa et al. 2009).

& The cost of ownership (including initial cost andoperating costs over time) of LED lightingsystems can be lower than the cost of fuel-basedlighting.

Off-grid LED lighting systems can also offerhighly compelling non-energy benefits, includingsuperior light quality, improved fire safety, elimina-tion of adverse indoor air pollutants, and promotion ofgood conditions for studying and learning.9

The time is ripe for accelerating the market forimproved off-grid lighting technologies in develop-ing countries. Arguably, these markets will bereceptive to LEDs well before those in industrializedcountries. The baseline technology (fuel) cost ishigher and their performance requirements (level andextent of lighting) are lower. Several major public–

private initiatives have been established to pursuethis goal. They include the World Bank Group’sLighting Africa10 program and the US Departmentof Energy’s Solar and LED Access Program, whichare collaborating, as well as the Lighting a BillionLives initiative and the Asian Development Bank’sEnergy for All initiative.11

While the deployment of quality assured off-gridlighting systems and significantly larger solar homesystems (SHS) both represent decentralized pathwaysfor achieving modest greenhouse gas emissionsreductions, there are good reasons to believe thatstand-alone LED systems have greater near-termpotential to deliver reliable and affordable results forillumination. (Of course, where other energy servicesare required, SHS can be the preferred approach.)First, the initial cost of small, LED-based off-gridlighting systems, which are now available at retail

Fig. 4 a–b Distribution of kerosene lamp fuel consumptionrates in Karnataka (left) (CDM 2009). Distribution of monthlykerosene lantern fuel consumption (right) and daily hours of

use (inset), inferred from liters-per-month data, assumingaverage consumption rate of 0.030 l/h. Figure 4b data furnishedvia personal communication by Stewart Craine, Barefoot Power

9 One study claims that average study time of students rosefrom 1.47 to 2.71 h/day, with a positive effect on schoolperformance (Agoramoorthy and Hsu 2009).

11 See http://energy.gov/news2009/8391.htm, http://labl.teriin.org/, and http://www.adb.org/Clean-Energy/energyforall-initiative.asp.

10 See http://www.lightingafrica.org/.


http://energy.gov/news2009/8391.htm

http://labl.teriin.org/

http://labl.teriin.org/

http://www.adb.org/Clean-Energy/energyforall-initiative.asp

http://www.adb.org/Clean-Energy/energyforall-initiative.asp

http://www.lightingafrica.org/

prices on the order of $10–70,12 are frequently five to20 times lower than conventional solar home systems,which often sell for $300–1,000. This was born out ina recent side-by-side demonstration of LED lightingversus traditional solar fluorescent lighting in chickenproduction (Tracy and Mills 2010a). The lower costof LED lighting systems makes them more broadlyaffordable to the lowest-income families and busi-nesses that do not have direct access to grid power.This suggests much greater potential for widespreaddeployment that, in turn, could lead to greaterdisplacement of fuel-based lighting and associatedcarbon emissions. Second, while LED lighting sys-tems are designed primarily to provide off-gridlighting (and, in some cases, mobile phone charging),the energy generated from SHS can be used in manyways (e.g., for powering televisions) that often takepriority over lighting (e.g., see Jacobson 2007). Third,although quality assured versions of both system typescan perform well, small LED lighting systems tend tobe simpler and, by extension, easier to install, maintain,and use. This simplicity can be an important factor forenabling widespread adoption and maximizing servicelife (Jacobson et al. 2000; Nieuwenhout et al. 2000).

While LED lighting solutions have emerged as aviable “disruptive technology” alternative, seriousissues of product quality and durability are impedingthe development of potentially immense markets foralternatives. There is considerable wishful and well-intended anticipation of the environmental and socialbenefits that may be captured from the widespreaduse of LED lighting systems. However, most claimsgloss over important practical realities that stand to

erode this assumed potential. Many claims do notexpressly address the means for maximizing savingsand minimizing the risks of under-attainment. Thereare a variety of specific performance and qualityissues related to the LED light sources, optics, drivercircuits, batteries, and charging systems, as well as theruggedness of the switches and housings (Mills andJacobson 2008; Tracy et al. 2009).

Most current commodity LED systems are low-price/low-quality products (Mills and Jacobson 2008;Johnstone et al. 2009; Mink et al. 2010). Marketsurveys have shown that end-users are satisfied withsome current products, although the fit is not to betaken for granted (Mills and Jacobson 2007; LightingAfrica 2009, 2011; Tracy et al. 2009, 2010b). Surveysof early adopters in Kenya showed that 87% of LEDflashlight buyers had problems within 6 months(Tracy et al. 2009). A market trial conducted in2008 found that many of the lamps had failed by thetime of a return visit 2 years later (Tracy et al. 2010a).Fortunately, private companies are beginning to offersuperior choices.

Under the most disadvantageous conditions, few ifany carbon savings can be expected to result from lowperformance LED products, while in well-designedapplications the value of the carbon reductions up toapproximately US $15 in our analysis represents asubstantial fraction of cost of the product itself.Although baseline fuel-based lighting assumptions(e.g., hours of use) are important, far larger uncer-tainties exist in the attributes and viability of thereplacement LED technologies, their patterns of use,and particularly their useful service life. For example,products with low-quality construction can corrode orprematurely fail in any number of other ways. Battery12 See http://light.lbl.gov/products.html.

Fig. 5 Use of firewood forlighting purposes (LightingAfrica 2009)


http://light.lbl.gov/products.html

technology choice also influences life in cases wherebatteries are not easily replaceable.13 In some cases,products are intentionally designed for a short life,such as counterfeited “hand-cranked” lights with non-rechargeable batteries. Inability to replace batteries,emissions associated with grid charging, multifunc-tion uses that do not offset carbon, and other factorscan also de-rate the nominally assumed greenhousegas emissions savings.

This discussion highlights the important relation-ship between product quality and emissions reduc-tions potential for LED-based off-grid lightingsystems. In fact, we find that systems for quantifyingand valuing greenhouse gas savings from alternativesto fuel-based lighting would do well to focusprimarily on the attributes of the replacement tech-nologies (rather than the fuel-based baseline technol-ogy). Moreover, by incorporating product quality intothe determination of emissions valuation, the dualobjectives of persistent savings and fostering technol-ogy innovation are productively reinforced.14

Evaluation of replacement technologies shouldthus be an integral component of a new carbonaccounting framework. In the next section, we

discuss existing carbon accounting frameworks andpropose an alternative that is based on deemedvalues that consider the attributes of the replacementtechnologies.

Adequacy of existing carbon accountingframeworks

The apparent simplicity of flame-based light sourcesand the compelling nature of the alternatives make thederivation of carbon savings seem straightforward.The temptation is to stipulate the savings as thedifference between a baseline that is simply theproduct of an assumed fixed rate of fuel and a fixedlevel of use. This value is then compared to anassumption of zero emissions for a replacementelectric light that has an assumed (frequently optimis-tic) product service life. Finally, the calculationnaively assumes full substitution wherein each hourof electric light corresponds to 1 h of displaced fuel-based lighting.

This highly simplified approach might be summa-rized as shown in the following example:

Baseline ¼ 0:025 l=h � 4h=day � 365 days=y � 10 years lifeð Þ¼ 365 l of kerosene

Energy saved ¼ 365 l of kerosene 100% offsetð ÞEmissions reduction ¼ 0:876 metric tonnes CO2 over the replacement product0s

lifetime assuming a 100% kerosene� fuel baseline; and 2:4 kg CO2=liter of keroseneð Þ

As we show below, this method for estimatingsavings defines an unreasonably optimistic upperlimit on emissions reduction rather than an expectedvalue.

Within the Clean Development Mechanism, untilrecently the approved methodology known as “AMS-

I.A. Electricity generation by the user” (UNFCCC2010) has been applied to several projects.15 Somebelieve this was not an appropriate methodology. Inany event, it has been used and approved repeatedly.

Following are some aspects of the methodologythat could be improved:

& The language is highly technical in places(including complicated mathematical formulas),which could create a deterrent to its use.

& Because the methodology attempts to cover to avery wide range of technologies and end-use

15 Methodology AMS II.C. "Demand-side energy efficiencyactivities for specific technologies" may be applicable for grid-recharged products with battery storage.

13 Cautious estimates might be 6–9 months for sealed lead–acidbatteries, 2 years for nickel–metal hydride, and 3 years forlithium ion technology.14 Incorporation of quality assurance into carbon emissionreduction schemes could be achieved through collaborationwith emerging quality assurance efforts such as that underdevelopment by the World Bank Group’s Lighting Africainitiative (http://www.lightingafrica.org/node/78).


http://www.lightingafrica.org/node/78

contexts, many passages are not applicable to off-grid lighting and thus impede the method’s useand precision.

& The methodology treats the baseline technologyas having a highly predictable set of uniformattributes, when in fact there may be many typesof sources with varying characteristics that affectthe amount of greenhouse gas emission offsets.

& The methodology focuses on the household sector.Non-household users (such as night market vendors,cottage industry, schools, clinics, and fishermen) aresignificant and should be accommodated. Often, asingle light is moved between both locations.

& The potential for increases in future emissions inthe baseline is not addressed.

& The methodology calls for measurement andverification that could be too cumbersome forproject developers and in cases not possible(Michaelowa et al. 2009).

& Section 7(c) Option 3 recommends a default dailyusage value of 3.5 h, which is appears to be overlycautious, at least in sub-Saharan Africa.16 This isparticularly true for certain cottage industry uses,such as poultry production, which are closer to12 h/day (Tracy and Mills 2010b).

& LED systems introduced under a given programmay be characterized uniformly, even if there is amaterial difference in the mix of specific productsdeployed under the program.

& Quality assurance of the replacement lamps isrelegated to the post-deployment period, but itcould be more effective if done before deployment.

& The methodology is silent on product service lifeand has accepted dubiously long default valuesproposed by project developers.

& The method does not accommodate the prevalentbaseline case of electrified consumers that rely onfuel-based lighting during power outages.

& The case of centralized grid-charged replacementlights is not addressed. This is a significant

limitation as grid-based charging is sometimesthe preferred strategy in locations where there issufficient infrastructure (typically in the form ofdistributed mobile phone charging microenter-prises). Thus, this provision inadvertently discrim-inates against one of the more popular technologyoptions among end-users. Moreover, this gridcharging bears a carbon footprint that should beaccounted for (Fig. 6).

& The method implicitly assumes perfect (100%)substitution of the electric light source for the fuelused in the baseline.

When the aforementioned factors are regardedin a highly favorable manner—or disregardedaltogether—projects will, not surprisingly be as-sumed to attain larger levels of carbon reductionsthan may be defensible. On the other hand, somelegitimate technology options or use cases are notaccommodated in the current method. Silence on keyfactors also invites widely varying estimates ofimpacts. In evidence of this, the two approvedCDM projects for off-grid lighting differ by a factorof four in their stipulated per-lamp savings (from0.31 to 1.17 metric tons CO2 per lamp over theproject life), while there is no obvious difference in thetarget markets or deployment strategy that wouldexplain such a large variance (UNFCCC 2006, 2009).

The Small Scale Working Group of the CDMExecutive Board (SSC WG) has been mandated toimprove the methodologies for small-scale, end-userenergy-efficiency projects. At its twenty-first meeting,the SSC WG placed priority on improved methodol-ogies for estimating displacement of fuel-basedlighting with efficient lighting technologies. Anexpress goal was to reduce “the transaction costrelated to monitoring aspects and to establish baselineemissions at the same time as maintaining theenvironmental integrity of the methodology.” Thus,the twofold goals are on the one hand increasedsimplicity with a cautious approach to estimatingemissions reductions.

Toward a viable methodology for assessing CO2

reductions from integral off-grid lighting alternatives

Identifying a widely applicable methodology isimportant given the large but diffuse target popula-

16 This is the current CDM default value, which is a low valuebased on recent survey results from five sub-Saharan Africancountries from the Lighting Africa (2009) market research. Thissurvey encompassed 5,000 end-users across five countries.Evening use alone averaged 3.5 h/day in Ethiopia, Zambia,Kenya, and Tanzania, and 4 h/day in Ghana. Additional use inthe early mornings was not quantified, but is frequentlyreported at 0.5–1.5 h/day, which we have observed usingembedded loggers as shown in Fig. 7.


tions, the diversity of replacement technologies, andthe low potential revenues per participant compared tomany other carbon-reduction technologies.

Important design principles could include:

1. The methodology is easy to understand and apply.2. The methodology recognizes technology, user,

and market drivers of outcomes.3. Assumptions and variables can be independently

verified.4. Carbon valuation is linked to project quality and

technologies employed.5. The cost of implementation is not a barrier to its

application.

We propose an improved methodology based on acautious standardized set of basic deemed baselineand upgrade defaults that could be selected in lieu of

costly field assessments. Recommended default val-ues for the fuel-based and LED technologies areoutlined and defined (Table 1), along with variousfactors for adjusting performance based on LEDtechnology choices.

Alternate values should be permitted if adequateresearch/monitoring/documentation is provided. Inter-ested third parties, non-governmental organizations,and governmental bodies could choose to improve theaccuracy and functioning of this market by conduct-ing strategic surveys and research to improve thebasis for alternate assumptions, thereby reducing oreliminating the transaction cost of doing so faced byprivate businesses attempting to operate in the market.The rapid rate at which LED technologies areevolving, combined with extensive new marketresearch yielding new information, should be consid-ered in regular updates to the methodology.

Standardized baseline assumptions

One of the statistical benefits of small-scale projectswith large numbers of participants is that a givenproject may be highly randomized as compared to,for example, a single large power plant. In evaluat-ing the acceptability of variance in baseline values(Fig. 4a–b), taking the central value can accuratelyrepresent a population of lighting users or an array oflighting technologies. Recently, a new CDM meth-odology (AMS II J) for CFL projects pioneered theconcept of including cautious default operatingparameters as an alternative to costly continuousmonitoring (Michaelowa et al. 2009).

While there is an overall fivefold variance in thestandardized hourly rates of emissions from fuel-based lighting products as seen in Fig. 3, the vastmajority of products are of the small-to-medium wickand hurricane lantern type, which vary by closer to afactor of 2–3. In practice, additional lamp-to-lampvariation is added by end-user wick managementpractices, wind conditions, and daily lamp usepatterns. Self-reported values for these types ofvariables are not necessarily reliable.17 Efforts to

Fig. 6 Greenhouse gas emissions associated with grid-chargingLED lighting systems depend on the power consumption of thesystem, conversion efficiencies, and emissions factors. Powersupply efficiencies vary from ~3% to ~95% (Alstone et al. 2011;Ecos 2002). Minimum efficiency standards in California are 50%(California Energy Commission 2009). SLA battery efficienciesvary from 50% to 90%, depending on the charging strategy(Stevens and Corey 1996). This assessment assumes a grid-electricity emissions factor of 1,000 g/kWh (gCO2/kWh) and20% transmission and distribution losses. Values in developingcountries range from to 600 to 1,800 (gCO2/kWh), includingtransmission and distribution losses (EIA 2007). For comparison,a typical kerosene lantern results in emissions of approximately40 g/h. In the example given, losses range from 5% to 100% ofbaseline lantern emissions, but losses rise steeply at the low-efficiency end of the scale. These values do not include standbypower consumption that occurs when grid-connected lampsremain plugged in after the battery has been fully charged

17 In a recent study (Tracy et al. 2010a), night watchmenreported an estimated time of 3.5 h of flashlight use per night;however, preliminary results from digital data logging indicatesthat nightly time of use is closer to 1.5 h on average. Radecskyet al. (2008) also reported higher than actual measured rates ofuse.


Table 1 Deemed baseline assumptions: illustrative recommendations

Baseline fuel-based lighting technology Default value

Fuel use rate There is a wide range of fuel-based lighting sources, and each requires its own deemedfuel-consumption baseline. For kerosene lanterns, fuel use rates range from 0.01 to0.10 l/h with most products operating in the 0.02–0.04 l/h range (i.e., the small/medium wick lamps and larger lanterns). A value of 0.025 is a reasonableconservative approximation in lieu of superior local data

0.025 l/ha,b

Daily hours of use Recent surveys of 5,000 households across five sub-Saharan countries foundaverage lantern-use values of 3.5–5 h for evenings only (excluding earlymorning lighting) (Lighting Africa 2009). Irrespective of the value assumed,fixing this value without option for petitioned alternative levels wouldinadvertently create a disincentive for program developers to identify and targetparticularly high-use groups

4 h/daya

Days of use For fully unelectrified users, daily use can be assumed. For unreliablyelectrified contexts, lower values must be used based on acceptable publishedinformation (presumably available directly from the power productionauthorities). For this purpose, it would be reasonable to take the average over amulti-year outage history for an appropriate region (city, sub-grid) rather than atthe household level. A major challenge still would remain, however, inascertaining whether a given buyer was using a light in a grid-connected context

365 days/yeara,c

Fuel emissions factor In practice, a variety of light sources may be replaced by the new technology. Acontext-specific blended fuel-mix could be proposed by users of the methodol-ogy. Where data are available in the open literature, UNFCCC may choose todevelop official fuel mixes (emissions factors) for use as default values inspecific regions or contexts. Developing appropriate emission factors for biofuelsis particularly difficult. The rate given here is for kerosene

2.4 kgCO2/la,b

Suppressed-demand multiplier The CDM does not formally permit adjustments for suppressed demand 1.00a

Annual dynamic baselinemultiplier

A number of factors can be expected to alter baseline consumption oflighting fuels upward or downward during the service life of carbon-reducingproducts. These include fuel-price increases/decreases and the effects ofsubsidies and taxes, numbers of people per household, income, andelectrification rates. If there is a basis for estimating these factors amongthe user population, the value can be specified as a net annual rate. At aminimum, in cases where there is increasing income, the consumption of lightingfuels will likely increase, and thus the baseline would grow during the time thealternative lamp was in service. A study of Ethiopian households estimatedthat kerosene use grows considerably faster than income (Mulugeta 2004)

10% per yeara

LED replacement technology

Leakage factor In practice, some proportion of users will move their original fuel-based light to adifferent location or to use it in conjunction with the new light source. A cautiousdefault substitution efficacy of 50% may be applied (Tracy 2010a). It could beargued that this relocated fuel-based light source is just reducing suppresseddemand, and that no carbon penalty should be assessed, however no literal carbonreductions will occur in this event

50% fuel usereductiona

Number of fuel-based lampsreplaced per LED

Well-designed LEDs may be able to replace multiple fuel-based lamps in someinstances, thereby increasing the carbon offset considerably. A cautious averagedefault assumption of 1:1 should be assumed in lieu of acceptable alternate datafrom the applicant

1a

LED service life All electric lighting products experience a reduction in light output over time, aprocess known as lumen depreciation. The rate of decrease varies widely by typeof lamp and quality. The Alliance for Solid State Illumination Systems andTechnologies (ASSIST) recommends defining the useful lifetime for LEDs as thetime at which initial light output has declined by 30%, which would beapproximately 2,500 h for a fully-powered high-quality 5 mm component. At4 h/day of operation (fully driven), this is about 2 years. The service life of larger“High-power” LEDs is on the order of 10–20 times this number (USDOE 2006).Given many other factors that can shorten product life, a more cautiousassumption of seven years for products with high-power LEDs would beappropriate, given that other components of the product are likely to fail withinthat time, effectively terminating the product's service life

2 yearsa


accurately measure these variables at the end-userlevel in a way that is cost-effective for an isolatedCDM project, especially with repeated measurementsover time, are likely to be futile.

Deemed-savings via standardized replacementtechnology assumptions

As indicated above, we recommend choosing acautious set of default assumptions for LED replace-ment technologies, and then applying performance-adjustment factors to reflect varying attributes thatcan determine the amounts of fuel-based lighting thatare ultimately offset. We note that shifting the analysisto the replacement technology addresses a number ofuncertainties overlooked in the AMS-1.A methodol-ogy, and does so by applying readily available datathat do not involve costly and fallible house-by-housemeasurement and verification processes. In this case,verification can be performed at the point of sale oreven further upstream in the product manufacture/delivery process. Importantly, this approach alsoincorporates incentives for improving the technologyand program delivery (which are absent from thecurrent methodology).

Performance adjustments

The deemed-savings approach must be performed inthe context of various uncertainties that are difficult(or costly) to measure or otherwise manage. Asidefrom these factors are a set of technology factorsassociated with the replacement systems that collec-

tively introduce far greater uncertainty, yet, fortunate-ly, are easier to quantify and incorporate into anassessment of real-world energy savings and carbonoffsets.

These include factors influencing the product’sservice life, performance, and level of offsets, andproduct quality and reliability variables that determineuser acceptance and the level of utilization.

The concepts of increases in future baselineemissions, additionality, and leakage require deeperconsideration and—in some cases—adjustments tothe deemed default values. These issues will beconsidered in the sections that follow.

Increases in future baseline emissionsduring the project period

As an energy end-use, lighting has unique character-istics and complexities compared to many otherenergy services. Lighting users’ needs vary widely,from small incidental applications to higher intensityand continuous ones. The quality and quantity ofacceptable illumination also vary. For some, taskscolor rendition or glare are not important; while forothers, they are critical.

The level of lighting service provided by fuel-based lighting is very small; light levels from fuel-based lamps are commonly 1–5% of those called forin illumination standards in industrialized countries(Mills and Borg 1999; Mills and Jacobson 2007). Forexample, the Western standards for light levels(“illuminance”) can vary from 100 to 10,000 lx(lumens/square meter) depending on the activity

Table 1 (continued)

Baseline fuel-based lighting technology Default value

Net-to-Gross value (NTG) NTG is a value ≤1, which represents the ratio of products obtained through theprogram to the total obtained in or out of the program. Because LED systemsorganically entering the target markets are of very limited use (virtually allflashlights) and of such exceptionally low quality that they garner negligible, ifany, carbon reductions, a NTG of 1.00 can safely be assumed in the near term

1.00a,d

Deemed lifetime emissions(tons CO2 over lamp life)

Product of all preceding factors in this table, with the exception of grid-charging,applied instead in Table 2

0.106 tons CO2

(lifetime)

a Alternative value can be used with qualifying data, or stipulated by program evaluatorsb Separate fuel and carbon-accounting methods must be employed for other baseline fuels, including biomass, candles, and dieselc A lower value should be used for grid-connected customers using fuel-based lighting during power outagesdWith time, or in specific contexts, the use of default Net-to-Gross (NTG) values <1.0 will become appropriate. While LED systemsare currently entering the market, few if any are of the quality that would be promoted in CDM programs using this methodology


being illuminated. In contrast, a kerosene lantern maydeliver only 1–10 lx. There is no consensus as to theminimum acceptable illuminance levels or lightoutput from qualified LED systems.

As can be seen from Fig. 1, there is massive pent-up demand for lighting services in the developingworld: a quarter of the world’s population consumesfar less than 1% of the utilized illumination services.Moreover, users of fuel-based lighting obtain less thana thousandth of the illumination energy services perunit of money spent on illumination as do those inindustrialized countries. The poorest of the poor thuspay far more than the rich for each unit ofillumination.

Consumers surveyed in sub-Saharan Africa reporttwo to three rooms kept dark in the evenings, whileboth consumers and off-grid businesses report inad-equate illumination and rank improved lighting high-est among a set of improvements desired for theirpremises (Lighting Africa 2011). Conversely, aproject in Malawi found that lighting use increasedfrom 2.7 to 4.4 h/day (63%) after the introduction ofLED systems (Adkins et al. 2010). Householdsurveys conducted under a CDM project based inKarnataka, India, found that existing households hadone to three lamps and would acquire an additionalfour lamps if they could afford the initial cost andoperating fuel (CDM 2009). They estimated thatactual kerosene consumption was about one seventhof what it would be if they could afford to operatemore lamps for more hours each day.

While it may not be appropriate to explicitlyaccount for these factors in the determination ofgreenhouse gas emissions reductions, they must beconsidered in the selection of replacement technolo-gies and the design of deployment programs in orderto maximize the chances for customer acceptance,retention, and persistence of the change.

On a lamp-for-lamp basis, a high-quality LEDlighting system designed for developing countryapplications can produce illumination levels for tasklighting that are 10–100 times higher than thoseproduced by the baseline flame-based lanterns. Ifusers then aspired to extend that higher lighting levelthroughout their homes or businesses, the impliedincreases in future baseline emissions during thecourse of the project grows again many-fold. Theamount of lighting fuel required to replicate thisexpanded level of service would amount to many

thousands of times of current usage. However,equating all of this phantom fuel offset by LEDlighting systems would result in hundreds of dollarsof notional carbon value for each lantern—tens oftimes the total price of that lantern—severely distort-ing the market while crediting imaginary carbonsavings.

A potentially more defensible treatment forincreases in future baseline emissions (because itcould plausibly change/manifest during the projectperiod) would be to consider and quantify two factors:

1. Rising emissions due to technical as opposed toeconomic factors. These would include temporar-ily curtailed use of the baseline lantern due torestricted kerosene availability (rather than insuf-ficient income to purchase).

2. Growth in the fuel-based lighting baseline in theabsence of the LED alternative over the proposedproject period, with growth indexed to increasedincomes as well as kerosene price effects thatcould boost (or shrink) demand for kerosene.Geographically based indices for kerosene pricescould be based on price elasticities from theliterature or on field research conducted express-ly to determine the relationship. Linking correc-tions to these historical socioeconomic factorswould also be a more quantitatively rigorousapproach insofar as the time horizon for growthin illumination consumption is not practicallymeasurable.

In thinking about the dynamics of lighting choicesover time, it should be noted that households andbusinesses using fuel-based lighting tend to purchasemore expensive lamps that generate more light at ahigher rate of efficiency (but with higher absoluterates of fuel use) if and when their purchasing powerincreases (either through increased income or de-creased prices). For example, a user could upgradefrom a wick to kerosene to pressurized lantern, whileincreasing the number of lanterns and hours of use.The upper limit is the point at which the user is wellenough off to switch to grid-based electricity, and thusrepresents an absolute cap on the potential increasesin future baseline emissions during the course of theproject.

The briefer the project period, the lower theexpected effect either of these factors. Both factorswould also be difficult to quantify in practice, and for


this reason it may not be practical to producedefensible estimates.

Additionality

Lighting systems based on white LEDs are only justbeginning to penetrate markets in the developingworld. The question appropriately arises as to whethersavings from programs under the CDM would yieldnet benefits compared to business as usual andthereby meet the requirement of “additionality”. Theterm “net-to-gross” savings is also used to describethis effect.

The cost of LED systems represents an establishedmarket barrier to natural adoption, particularly for thelowest-income target audiences. By effectively in-creasing the payback time by many-fold, fuel pricesubsidies (particularly high in India18 and Indonesia)create a significant barrier to the uptake of newlighting technologies. Import duties magnify thisdeparture from true–cost relationships among con-sumer choices. Taken together, these factors amplifyintrinsic economic barriers by depressing the cost ofthe polluting baseline and magnifying the relative costof the alternatives.

The prices of LED technologies being offered todeveloping countries vary widely. Commodity, low-quality products (generally in the form of flashlights)are typically priced at under $5. Higher-quality,higher-performance products fall in the $10–50 range(some much higher, which is not realistic for thelowest-income markets). Recent market research hasestimated end-user willingness to pay for suchproducts (Fig. 7). While there may be some excep-tions, it appears that current retail prices often exceedthe willingness to pay, suggesting a role for incentivessuch as those that may be generated through CDMprojects. However, this relationship could changedramatically in the future as LEDs become lessexpensive. In any event, the availability of carbon-credit incentives should not be allowed to biasmanufacturers against seeking lower cost productionmethods.

There are several other modes by which LEDprojects under the CDM can be expected to achieveadditionality. The first is by accelerating marketpenetration. While LEDs have become common inbattery-powered flashlights (Johnstone et al. 2009),they are virtually non-existent in other lightingcontexts in which fuels are the baseline energy source.The extreme income sensitivity of the target audien-ces and the relatively high first cost compared toincumbent baseline technologies (which could behigher by 100-fold, e.g., $0.20 versus $20.00)suggests that baseline “unaided” market penetrationmay be quite slow. However, as LED prices fall theymay cross a point at which demand is large withoutthe benefit of carbon valuation. A recent study foundthat the uptake rate of LED alternatives was verysensitive to income (Adkins et al. 2010).

The second and more important argument foradditionality has to do with product quality. Thecurrent trend is toward very low-quality LED prod-ucts, which are spoiling the market and thus slowingdemand (Mills and Jacobson 2007; Tracy et al. 2009,2011). In lieu of interventions aimed at restoringquality and end-user trust in this technology category,penetration will be very low and cumulative savingswill be diminished due to the minimal service livesachieved before products fail. Conversely, productsand programs that embed high quality standards willsecure durable emissions that are strongly additionalto those in the business-as-usual scenario.

A third factor that argues for significant addition-ality in CDM-based off-grid lighting programs is theabsence of adequate market conditions to enableimproved off-grid lighting technologies to gain afoothold. Programs enabled through CDM couldpotentially create or improve local institutions, sup-port financing mechanisms to overcome first-costsensitivities, mount information and education cam-paigns to better equip sellers and buyers to engage,and create or improve supply chains for criticalcomponents (e.g., replacement batteries). Doing socan clearly accelerate market uptake (Adkins et al.2010). Companies developing off-grid lighting prod-ucts can have difficulty accessing investment capital.Securing carbon payments is one way to reduceperceived risk by investors.

Another criterion for additionality—that improvedproducts not be required by law—is clearly met inmost cases. There will be exceptions and the CDM

18 Households targeted by a CDM project in Karnataka weresaid to pay as little as 12 Rupees/l ($0.25/l) for kerosene (CDM2009), which is substantially lower than prices of $1–2/l observed in sub-Saharan Africa.


should monitor this for use in evaluating prospectiveprojects. For example, in late 2009, Peru outlawed theuse of kerosene for lighting and cooking.19 They arepromoting integrated off-grid electric lighting in itsplace, probably through product give-away programs(Centeno et al. 2009).

One contingency to be considered is a programrecipient’s home or business eventually becomingelectrified. A grid-based incandescent lamp is respon-sible for about 10–100 g of CO2 emissions per hour,

20

which is the same order of magnitude as theemissions rates kerosene lanterns depicted in Fig. 3(but the electric lamp of course generates vastly morelight). If off-grid LED systems cease being used whenan entire home becomes electrified, then there may nolonger be carbon savings. On the other hand,decommissioned LED systems would, in practice, beplaced into use by others (even in the same family)who remained off the grid. Moreover, the presence ofthe LED systems, especially where other modestfunctions were served (e.g., television and cell phonecharging) could defer for a period of time a consum-er’s move to the grid.

If current trends in technology development(lower cost, higher performance components) and

policy efforts are successful, the role for CDM inoff-grid lighting will be limited to the near term.Important differentiators of LED product lifetimenotwithstanding, LED lighting systems are generallymuch shorter-lived products, which entail lessspeculation and need for long-term monitoring toensure that deemed lifetime performance is main-tained in practice. Moreover, high-quality LEDsystems are generally designed to be nearly mainte-nance free (the primary maintenance is an occasionalbattery change), thereby reducing uncertainties aboutdurability. In contrast, many other types of CDMprojects claim emissions that are based on very longasset lives (e.g., 20–30 years), which of courseincrease uncertainties about the full lifetime emis-sions being attained.

Taken together, the aforementioned factors suggestthat, given their current miniscule market penetrationdue to a combination of economic and institutionalfactors, integrated LED lighting systems are far lesssusceptible to additionality concerns in the near termthan many more well-established technologies cur-rently deployed within carbon markets. Given theshort product innovation lifecycle, baseline condi-tions, and methodology assumptions can be revisedregularly with little risk of inappropriately grand-fathering legacy projects. That said, the technologies,their costs, and other market factors are changingrapidly. It would be prudent to revisit the issue ofadditionality regularly and to make adjustments tothis aspect of the methodology as necessary. CDM’sapplicability in a given country or region could evenbe benchmarked to a specific rate of market penetra-tion for quality products, e.g., 20%.

Fig. 7 The willingnessof households and traders topay for rechargeable off-grid LED lighting systemsvaries by country, user type,and type of lighting serviceprovided. Countries:Ethiopia (9.95 ETB/US$),Ghana (1 GHC/US$),Kenya (66 Ksh/US$),Tanzania (1,181 Tsh/US$),and Zambia (3,333.3 ZK/US$). Source: LightingAfrica (2009)

19 In legal rule D. S. No. 045-2009-EM on 29 April 2009, thePeruvian government banned the sale of kerosene nationwide.20 This range is defined depending on lamp type and gridcarbon emissions factors. A 100-W incandescent lamp and anemissions factor of 1,000 g CO2/kWh of electricity wouldcorrespond to about 100 g CO2/h, a 15-W compact fluorescentlamp and an emissions factor of 500 g CO2/kWh of electricitywould correspond to emissions of 8 g CO2/h.


Leakage

Leakage is defined as the net change of anthropogenicemissions by sources of greenhouse gasses whichoccur outside the project boundary, and which aremeasurable and attributable to the CDM projectactivity (3/CMP.1, Annex, paragraph 51). This prob-lem would arise in the case of fuel-based lighting if,for example, the fuel-based lantern displaced by theLED light was transferred out of the project boundary.

Alternatively, some fuel-based lanterns are likelyto remain in use within the project area to illuminate apreviously dark area. A recent study (short-term, self-reported data) found more than an 80% kerosenereduction—and an even higher rate for candles—across 54 homes in Malawi (Adkins et al. 2010).Observations in Kenya found approximately 50%reduction in expected savings for night traders(Alstone et al., in press) and 14% for households(Tracy et al. 2010a).

There is a strong likelihood that baseline lampswill be kept in use to some degree. We suggest aprovisional 50% “leakage factor” assumption default.A lower factor could be allowed based on additionalresearch or if the project developer implements apersuasive means of reducing this risk. Smallerfactors could be applied for user certain groups thatare unlikely to continue using their kerosene lanterns(e.g., poultry producers using lights to extend theeating period for their birds, per Tracy and Mills(2010)). Given their very low cost (e.g., $0.20 for astandard wick lamp), destroying the baseline lamp isunlikely to be beneficial, given that these lamps areeasily remanufactured at an extremely low cost.

Monitoring

AMS-1.A (Section 14) calls for monitoring in the formof “[a]n annual check of all systems or a samplethereof to ensure that they are still operating (otherevidence of continuing operation, such as on-goingrental/lease payments could be a substitute).” Suchmonitoring is highly onerous and cannot, in mostcases, be expected to be cost-effective for distributedenergy projects such as off-grid lighting deployment.This condition serves to discourage the development ofprojects, which limits the CDM to larger scale efforts.

Our proposed framework provides a more prag-matic alternative in the form of a very short deemed

service-life proxy (2 years) and other deemed perfor-mance defaults. Projects that opt to institute monitor-ing can benefit by being assigned a longer service life,provided that their product demonstrably lasts beyondthe default service life. Note that many other factorsalso affect service life and are taken into account inthe framework.

Longer service lives could be assumed in the caseof ongoing rental/lease payments, as provided in theexisting methodology. Warranty or insurance-backedproducts could also be allowable mechanisms fordeeming longer service lives.

Market factors

A variety of “soft” factors also influence projectsuccess. Perhaps the most fundamental of these isconsumer acceptance. Many lighting users are highlydiscriminating in their preferences and willingness topay (Mills and Jacobson 2007). This is not surprising,given the high proportion of income spent on lighting,very specific expectations on product performance,the front-loaded cost of the replacement technologies,and the widespread existence of low-quality electron-ics in most developing country marketplaces.

Related factors include the quality of the marketinfrastructure in which the products are nested, suchas an adequate variety and availability of products,financing, and the efficacy of product sellers inhelping users match the right light to their needs.After-sales service and a viable supply chain forreplacement parts and warranty are also important.For grid-charged devices, the availability, reliability,and affordability of charging services are crucial.

For well-manufactured products, the componentwith the shortest service life will commonly be thebattery. Thus, availability of matching replacementbatteries (and users’ ability to open the product inorder to easily replace them) is important to ensurethat products remain in service for their expectedperiod of time. Both currently approved CDMprojects include at least one product with a batterythat is soldered into the light, drastically reducing therespective lights’ service lives to on the order of 2–3 years. Those projects, however, assume servicelives of 5 and 10 years.

The relative product prices and availability ofalternatives will create an elasticity effect on demand,with higher prices and/or limited choice leading to


reduced uptake of the new technology. Conversely,delivery mechanisms (such as charity models) thatcircumvent traditional market processes may result inmistreatment of products and attenuated service life(and thus reduced cumulative emissions reductions).

Risk management and financing

Risk management can occur at various points in thevalue chain. The effort should begin at the point ofmanufacture. An in-house quality control, qualityassurance system is critical, and should be augmentedby independent rating and labeling. Factories manu-facturing qualifying projects could be subject to CDMfield inspections.

In carbon markets, insurance products are emerg-ing to manage non-delivery risks. Examples includeMunich Re’s (2007) Kyoto Multi-risk product. Insur-ance and warranty projects for the underlying tech-nologies can also be appropriate, especially whenproactively based on an engineering-based assessmentof product quality.21 A number of insurers offerrenewable-energy performance or energy savingsinsurance instruments (Mills 2003, 2009). Products

have not been fashioned expressly for small-scaleCDM projects, but may be in the future.

Micro-insurance is already used by nearly 80 mil-lion people globally (Mills 2009), and micro-financeby an even larger population. Application of thesefinancial services to small-scale carbon abatementtechnologies is a natural extension. In this context,carbon-performance insurance for off-grid LED proj-ects would represent an interesting market mechanismfor managing risks of the attainment of emissions.Insurers would be compelled to conduct their owndue diligence of products and projects, which wouldintroduce an additional layer (albeit unconventional)of quality assurance that could achieve some of thesame objectives as conventional project monitoring.Claims “paid” with equivalent CERs could be ofadditional interest.

Hypothetical application of the proposed framework

In the proposed framework, default values would bestipulated, and only over-ridden if acceptable datawere provided. Research and surveys by neutralparties could be periodically reviewed so as toimprove the deemed default values or make them21 See http://www.insurance4renewables.com/

Fig. 8 Based on hypotheti-cal inputs for the proposedsystem, the value of emis-sions varies widely depend-ing on product attributes.Assumes carbon price is$20/ton of CO2. Valuesshown are summed over thelife of the lamp. All prod-ucts offset the same baselinescenario. See Table 2 fordefinitions of productcharacteristics


http://www.insurance4renewables.com/

Tab

le2

Adjusted-performance

carbon

valuation

DefaultValue

Produ

ctA

Produ

ctB

Produ

ctC

Produ

ctD

Produ

ctE

Produ

ctF

Techn

olog

ymod

ifiers

LED

Techn

olog

yservicelife

Userof

5mm

techno

logy

canpetitionforextend

edlifeifacceptable

documentatio

nof

under-drivingthelig

hts(toextend

life)

isprov

ided.High-po

wer

LEDsassumed

7-year

life

2.0

2.0

2.0

2.0

7.0

7.0

7.0

Rechargeable

batteries

Rechargeablebatteries

have

alim

itedlife,

which

varies

bythe

techno

logy.Goo

d-qu

ality

nickel-m

etal-hyd

ride

batteries

canbe

expected

tolastperhaps2–

3yearsin

practice,

andless

than

1year

forlead-acidbatteries.Ifthebattery

compartmentcann

otbe

opened,

then

thebattery

end-of-lifedeterm

ines

theentireprod

uct’sendof

life.

Con

sumer

mustbe

able

tochange

battery

with

outtools;

otherw

iselifecapp

edat

2years

No

No

Yes

No

No

Yes

Yes

Warranty

Absence

ofawarrantyor

otherrisk-guarantyprod

uct(e.g.,carbon

-offset

insurance,

orprod

uct/com

ponent

performance

insurance)

may

reflectthe

manufactureror

interm

ediary’sdegree

ofconfidence

inthe

prod

uctand

theuser’sability

tohave

itrepaired

orreplaced

ifit

malfunctio

ns.A

bsence

ofawarrantyor

insurancecanform

thebasis

ofde-ratingthedefaultprod

uctlifetim

e,e.g.,b

y25

%in

this

illustration,

orexclud

ingitaltogether

No

0.75

1.00

0.75

1.00

1.00

1.00

Adjustedprod

uct

servicelife

Calculated

1.5

2.0

1.5

2.0

7.0

7.0

Baselinelig

hting

energy

source

Defaultvalues

may

assumeafossilfuelas

baselin

e,bu

tin

somecases

otherenergy

sourcesareused.Biofuelscouldhave

high

eror

lower

netem

ission

sthan

fossilfuels.Replacementlig

htingsystem

sthat

prim

arily

replaceabattery-pow

ered

lightingbaselin

e(e.g.,tradi-

tionalflashlightsor

“torches”)

couldbe

expected

tosave

very

little

fuel.De-ratin

gthebaselin

eassumptionby

90%,may

beapprop

riate

toaccoun

tforthis.How

ever,aflashlight

form

-factorisno

tnecessarily

prob

lematic

ifitprov

ides

effectivehy

brid

mod

esof

operationinclud

ingam

bientlig

htor

non-hand

held

task

lightingthat

end-usersdeem

adequate

forreplacingfuel-based

lights

Kerosene

0.10

1.00

1.00

1.00

1.00

1.00

Multifun

ction

capability

Som

einno

vativ

elig

htingtechno

logies

beingbrou

ghtto

market

supp

ortno

n-lig

htingfunctio

nssuch

ascell-ph

onecharging

orradios.

Inthiscase,thereiscompetitionbetweenuses

that

displace

fuel

and

thosethat

dono

t.Lessthan

a1:1offset

ofthebaselin

elig

htingmay

resultfrom

redu

cedlig

htingho

ursfrom

thedevice.Thiseffect

will

bestrong

estin

larger

solarho

mesystem

s(SHS)where

sign

ificant

non-lig

htingloadsarebeingmet.Savings

couldbe

de-rated

by25

%to

accoun

tforthiseffect

No

1.00

0.75

1.00

1.00

0.75

1.00

Pow

erconv

ersion

losses

(for

grid

charging

)

Solar-pow

ered

charging

saving

swou

ldbe

regarded

as"off-grid".In

manyareas,ho

wever,end-usersprefer

prod

uctsthat

canbe

grid-

charged,

e.g.,viacell-ph

onecharging

shop

sor

otherbattery-

charging

metho

ds.Ifthelocalgrid

uses

fossilfuelsandthecharging

efficiency

islow,then

ano

n-trivialam

ount

ofgreenh

ouse

gas

1.00

0.75

0.75

0.90

1.00

1.00

1.00


Tab

le2

(con

tinued)

DefaultValue

Produ

ctA

Produ

ctB

Produ

ctC

Produ

ctD

Produ

ctE

Produ

ctF

emission

swill

beem

itted

(Figure6).In

thiscontext,efficiency

isbasedon

thedifferentialbetweenpo

wer

deliv

ered

totheACadapter

andthat

ultim

atelyreleased

bythebattery

tothelig

ht.A

cautious

defaultem

ission

srate

foraprod

uctthat

isalwaysgrid-charged

is25

%of

that

from

astandard

kerosene

lantern(10%

iftheprod

uctis

chargeable

onor

offthegrid).High-efficiency

charging

yields

negligiblelosses.C

onversely,ifoff-grid

lightsareused

byelectrified

consum

ersdu

ring

power

outages,carbon

saving

smay

resultifthe

alternativebaselin

etechno

logy

choice

isback-upfuel-based

lanterns.

Ineither

case,grid-based

emission

scanbe

calculated

usingthe

CDM

metho

dology

know

nas

“AMSI.D”

Qualitymod

ifiers

Truth

inadvertising

Researchhasshow

nthat

manyoff-grid

lightingprod

uctsdo

not

perform

asadvertised

(MillsandJacobson

2007

).Und

erperforming

(orcoun

terfeit)prod

uctswill

disapp

oint

theuser

andareno

tlikelyto

remainin

useas

long

asaccurately

advertised

ones.Factors

tobe

considered

includ

ebattery

capacity,lig

htou

tput,andprod

uctlife.

Defaultem

ission

svalues

couldbe

de-rated

by25

%(orprod

ucts

disqualified)

ifthereisevidence

offailu

reto

prov

idetruthin

advertisingwith

regard

toprod

uctcharacteristicsor

performance

0.85

0.75

0.75

1.00

1.00

0.75

1.00

Certification

Anindepend

entprod

uctqu

ality

ratin

g(e.g.,that

beingdevelopedby

theInternationalFinance

Corpo

ratio

nandtheWorld

BankGroup

’sLightingAfricaProject)wou

ldmakeitpo

ssible

todifferentiate

amon

gprod

uctson

thebasisof

performance

anddu

rability,

which

arestrong

determ

inantsof

howlong

theprod

uctislik

elyto

remainin

service.

One

ofmanyexam

ples

offactorsthat

wou

ldno

totherw

ise

bevisibleto

abu

yerwou

ldbe

thedu

rabilityandlifetim

eof

the

embedd

edLEDs,which,asdiscussedabov

e,couldvary

from

weeks

toyears.Assum

ingamulti-levelrating,

de-ratingfactorson

theorder

of10

%couldbe

appliedfortheabsenceof

anygivenlevel

Level-1

Derated

10%

ifno

certificationat

thislevel

0.90

0.90

Level-2

Derated

10%

ifno

certificationat

thislevel

0.90

0.90

0.90

Level-3

Derated

10%

ifno

certificationat

thislevel

0.90

0.90

0.90

0.90

0.90

Effectiv

eprod

uct

servicelife

Produ

ctof

individu

almultip

liers

(tobe

appliedto

baselin

edeem

edsaving

sfrom

Table

1)Calculated

0.06

0.68

1.22

2.00

3.54

7.00

Cum

ulativedy

namic

baselin

emultip

lier

Calculated

1.15

1.21

1.15

1.21

1.95

1.95

Carbo

nem

ission

sredu

ctionov

erprod

ucteffective

servicelife

Ton

s0.00

40.04

40.07

40.12

80.36

60.72

3

Marketvalue

ofcarbon

(atUS$2

0/ton)

Metrictonn

es0.1

0.9

1.5

2.6

7.3

14.5


more case-specific. Large research and deploymentprograms currently underway (Lighting Africa andthe Solar and LED Access Program) are producingextensive information of this sort, as exemplified byLighting Africa’s surveys of thousands of householdsin Ethiopia, Ghana, Kenya, Tanzania, and Zambia(Lighting Africa 2009, 2011). For example, themajority of households in Zambia report that theirflashlights and solar-powered lanterns last only 1 yearor less (Lighting Africa 2009). These reports alsoprovide information on lighting fuel mix by country(Fig. 2a–b).

One of the benefits of the proposed framework isthat all of the deemed input values can be determinedbefore the technology is deployed and without marketsurveys. Petitions for alternate values can be sup-ported by market observations that do not requirecostly and intrusive visitations to individual users.

In evaluating the persistence of specific LEDproducts, disclosure of product returns and repairsunder warranties may provide justification for adjust-ing service-life assumptions.

Figure 8 provides an illustrative implementation ofthe proposed framework. The figure shows results forsix hypothetical products, spanning a wide range ofoperating conditions and product quality. The exam-ples are developed in greater detail in Table 2.

& The worst-case product receives essentially novaluation for carbon emissions reductions. Thishypothetical product uses shorter-lived “5 mm”LED lights, is grid-charged, has no performancewarranty, has a non-replaceable battery, substi-tutes for a battery-powered baseline technology (aconventional flashlight, which defers little kero-sene), is a multi-function device with a built-incell phone charger (which diverts some batterypower), and bears no independent quality rating.The product’s advertised claims could not bereplicated with lab tests. The product receivesnearly no carbon credits.

& The best-case product employs long-lived “PowerLEDs”, has a replaceable battery, provides awarranty, is strictly solar charged, complies withtruth-in-advertising criteria, and has been certifiedat the highest quality assurance level by anindependent testing body. This product producessubstantial carbon offsets valued at about US$15over the life of the lamp.

It should be noted that a given CDM project islikely to incorporate multiple brands and/or models ofLED lighting systems, each of which may scoredifferently in the proposed framework.

The CDM is not a panacea. There exist a variety ofissues that, while not intrinsic barriers to the deploy-ment of improved off-grid lighting technologies,confound efforts to perfect a methodology within thebounds of CDM. Examples of these are accountingfor use among weakly electrified consumers who uselamps during outages, treating biomass fuels used forlighting, accounting for grid-charging of otherwiseoff-grid products, quantifying increases in futurebaseline emissions during the course of a project,defining dynamic baselines, and adjusting for theadditionality concerns associated with the degree towhich these technologies would be adopted in theabsence of the ability to monetize the carbonreductions. Moreover, the rate of evolution of CDMprotocols is not keeping pace with the rapid develop-ment of the technologies they are intended to support.Indeed, LED products have evolved even during agiven project evaluation and approval process, suchthat technologies described in project proposals differfrom those available at the time of ultimate deploy-ment. The extremely long lead-time for approval ofCDM projects (~2 years) is thus in itself a formidablebarrier to the use of this mechanism.

Enabling tools and analyses

The framework offered in this article is conceptual innature and the examples given are indicative—ratherthan prescriptive—for the purposes of discussion andrefinement. Various sorts of research and analysis cansupport implementation of the recommended systemand determination of actual values to be used in theframework.

Given the popularity and likely large market shareof grid-charged systems, it is important to develop abetter sense of the associated emissions. A key factoris to quantify the losses between grid power andenergy delivered to the light. These include ACadaptors and battery chargers, batteries, and circuitrythat might be in the pathway.

Independent product assessments and ratingswould provide one of the most valuable streams ofinformation for use in de-rating or enhancing the


default offset assumptions. The Lighting AfricaProject and the Renewable Energy DeploymentInitiative are currently developing such systems.Encouraging more of this sort of “public goods”research would result in data independent of specificfor-profit validator interests (Schapiro 2010), whilealleviating considerable financial disincentives fromindividual companies for whom collecting such datain support of a single project would be cost-prohibitive or a conflict of interest.

Product labeling systems could prove to be a keytool for simplifying the application of the qualityassessment elements of the framework proposed here.Such systems would provide an objective set ofmetrics in turn used as a proxy for product qualityand service life. Moreover, this would enable projectapplicants to ensure and verify performance inadvance of applying for CDM project qualification,which in turn, would result in higher quality projects,deeper and more durable carbon reductions, andenhanced additionality compared to a business-as-usual pathway that is currently introducing manysuboptimal products into these markets.

Although not included in the existing CDMprotocol, the embodied energy and associated green-house gasses of off-grid products could potentially beincorporated in the analytic framework. One investi-

gation determined that the embodied energy of oneoff-grid LED product is recovered in 1–2 months ofoperation, a modest value that can be expected todecline over time (Alstone et al. 2011).22

Improved estimates of baseline global carbonemissions from off-grid lighting would help incharacterizing the overall potential market. Theone value in the literature (Mills 2005) should beupdated and refined to include intervening demo-graphic changes, new data and understandingsabout the technologies, and specific examinationof biofuels in the provision of illumination. Onefactor that has not been previously evaluated is theperhaps significant role of “black carbon” (soot) inthe overall climate-forcing impact of fuel-basedlighting. The various fuel-based lighting sourcesall emit some amount of black carbon, but blackcarbon’s global warming potential is not currentlyrecognized in CDM projects.

22 It should be noted that new LED systems with rechargeablebatteries that replace conventional flashlights offset significantsolid waste production in the form of non-rechargeable batteries(implying additional embodied carbon reductions). Moreover,there are carbon emissions associated with producing anddistributing liquid fuels, which may alone outweigh thoseembodied in the manufacture of LED lighting systems.

Fig. 9 Micro-logger formonitoring on-time foroff-grid LED lighting prod-ucts. The chart shows anexcerpt of data collected forone user in Kenya over a3-day period. For thisparticular trial, the averageLED lamp utilization (bluecurve) was 2.6 h/day. Notedual morning and eveninguse patterns. As batteryloses charge on the thirdnight, the light output canbe seen to decrease. Thelogger was designed byKyle Palmer and others atHumboldt State University(HSU); photo/data gatheredby Peter Alstone at HSU


Improved data on the utilization of baselinetechnologies can help refine the default values andperhaps provide different authorized datasets fordifferent geographies or demographics. Cost-savingmethods for collecting field data could provide avaluable basis for adjusting default assumptions.Low-cost light loggers have been developed andfield-tested in off-grid lighting products, but not yetcommercialized (Radecsky et al. 2008; Fig. 9). If non-intrusive data recovery (e.g., through short-rangewireless networks) could be applied, then utilizationassumptions could be validated at the project level ata lower cost (and with less self-reporting error) than ifin-person interviews were required. However, safe-guards would be necessary to manage risks of gamingor fraud. Independently orchestrated surveys (e.g.,conducted by governmental or non-governmentalorganizations for public interest applications) wouldbe less susceptible to these concerns.

Conclusions

There is massive need for improved lighting servicesin the developing world. Current efforts to meet thoseservices involve the use of highly inefficient liquidand solid fuel combustion, which results in substantialgreenhouse gas emissions as well as other adverseimpacts such as compromised indoor air quality.

Grid-independent lighting systems based on LEDlight sources offer a very promising alternative forsimultaneously improving lighting services and reducinggreenhouse gas emissions.

Some developers of projects to promote thistechnology seek to monetize the carbon emissionsreductions that are achieved. However, only two off-grid lighting projects under the CDM have beenpreviously approved (CDM 2008, 2009). The lengthyproject documentation and methodology proposed byproject developers varies considerably; there is verylittle standardization. It would create greater transpar-ency for policymakers and remove barriers for projectdevelopers if a more uniform and cost-effectivemethodology was implemented.

A more accurate and effective CDM methodologycould eliminate the need for costly field investigationsby relying instead on certain deemed baselineparameters combined with consistent adjustmentsbased on more readily available market data and

quality assessment of the incoming LED technolo-gies. These adjustments manifest largely with respectto effective product lifetime and thus its cumulativeemissions reductions, and can lead to much moreinternally consistent estimates of carbon savings thanis the case at present with divergent methods designedby project developers.

Aside from its traditional role of directingcapital from wealthy countries toward highlycost-effective carbon-reduction projects in thedeveloping world, in the case of off-grid lightingsystems the CDM, if properly applied using aframework along the lines of what is proposed inthis article, can play a highly meaningful role inpromoting improvements in the quality of productsoffered to the marketplace. The logical outcomewould be significantly higher uptake and end-usersatisfaction with improved lighting systems thancould occur through sole reliance on existingimperfect market forces.

Epilog: A new CDM methodology for off-gridLED lighting

Based on the analysis presented in this article, theCDM has released a new approved methodologyAMS-III-AR for quantifying the carbon reductions ofLED lighting systems in off-grid contexts (UNFCCC2010). The methodology incorporates much of theframework set out in this article.

The guiding principles are to provide a methodmore well-suited to LED projects and reducing thetime and cost of qualifying a project and documenting thecarbon savings, while requiring performance disclosureand embedding new criterion for minimum productquality while rewarding those products that exceed theseminimums. In most cases, independent testing is requiredin order to demonstrate performance.

Based on the minimum performance criteriaspecified in the new approved methodology, thedeemed savings would be appropriately modest:0.16 tons of CO2 per lamp (over a 2-year deemedservice life). Moreover, low-quality products (or thoselacking a warranty) are ineligible for any level ofCERs. Conversely, alternate values for many “de-fault” factors can be used if adequately justified bythe project developer, which could bring the avoidedemissions significantly higher.


Acknowledgments A longer version of this report was preparedat the request of The United Nations Framework Convention onClimate Change (UNFCCC), Small Scale Working Group of theClean Development Mechanism (CDM) Executive Board. Thiswork was also supported by The Rosenfeld Fund of theBlum Center for Developing Economies at UC Berkeley,through the US Department of Energy under Contract No.DE-AC02-05CH11231. Art Rosenfeld has been a keysupporter of this work. This project benefitted from valuablecollaborations with Gaj Hegde of the UNFCCC Secretariat; PeterAlstone, Kristen Radecsky, Jennifer Tracy, and Dustin Poppendieckat Humboldt State University; Jessica Granderson, Jim Galvin, andFrancis Rubinstein at Lawrence Berkeley National Laboratory; andMaina Mumbi and Francis Ngugi in Kenya. Steven Schiller ofthe CDM Small Scale Working Group provided constructivereview comments and consultation.

References

Adkins, E., Eapen, S., Kaluwile, F., Nair, G., & Modi, V.(2010). Off-grid energy services for the poor: introducingLED lighting in the millennium villages project in Malawi.Energy Policy, 38, 1087–1097.

Agoramoorthy, G., & Hsu, M. (2009). Lighting the lives of theimpoverished in India’s rural and tribal drylands. HumanEcology, 37(4), 513–517.

Alstone, P., Mills, E., Jacobson, A. (2011). Embodied energyand off-grid lighting. Lumina Project Technical Report #9.Lawrence Berkeley National Laboratory and HumboldtState University. Available from: http://light.lbl.gov/pubs/tr/lumina-tr9-summary.html.

Apple, J., Vicente, R., Yarberry, A., Lohse, N., Mills, E.,Jacobson, A., et al. (2010). Characterization of particlematter size distributions and indoor concentrations fromkerosene and diesel lamps. Indoor Air, 20(5), 399–411.

California Energy Commission (2009) 2009 Appliance effi-ciency regulations (cell phone chargers discussed on page166). CEC-400-2009-013. Available from: www.energy.ca.gov/appliances/index.html.

CDM (2008) Clean Development Mechanism Project DesignDocument–d. Light Rural Lighting Project, Version 2, 13/10/2008.

CDM (2009) Clean Development Mechanism Project DesignDocument–Rural Education for Development Society(REDS) CDM Photovoltaic Lighting Project, Version 4,4-Aug-09, and associated baseline survey spreadsheet.

Centeno, Z.C., Cesar, R.A., and Franchini, C. (2009) ProjectPico Solar Lamps for Peruvian Amazonia.

Ecos Consulting (2002) Power supplies: a hidden opportunityfor energy savings. Prepared for that National ResourcesDefense Council.

EIA (2007) Voluntary reporting of greenhouse gases: appendixA: Electricity emissions factors. US Energy InformationAdministration, form EIA-1605.

Jacobson, A., D.M. Kammen, R. Duke, and M. Hankins (2000)Field performance measurements of amorphous siliconphotovoltaic modules in Kenya. Proceedings of theAmerican Solar Energy Society, Madison, WI, USA, June16–21.

Jacobson, Arne. (2007). Connective power: solar electrificationand social change in Kenya. World Development, 35(1),144–162.

Johnstone, P., J. Tracy, and A. Jacobson (2009) Pilot baselinestudy—Report: market presence of off-grid lightingproducts in the towns of Kericho, Brooke, and Talek.Prepared for Lighting Africa.

Lighting Africa (2009) Lighting Africa market assessmentresults: quantitative assessment. Series includes Ethiopia,Ghana, Kenya, Tanzania, and Zambia. InternationalFinance Corporation and the World Bank Group. Avail-able from: www.lightingafrica.org/node/191/.

Lighting Africa (2010) Briefing notes: light emitting diode (led)lighting basics. Available from: www.lightingafrica.org/files/Lighting_Africa_Briefing_Notes_21.Dec_.2009.pdf.

Lighting Africa (2011) The off-grid lighting market in Sub-Saharan Africa: market research synthesis report. pp. 92

Michaelowa, A., Hayashi, D., & Marr, M. (2009). Challengesfor energy efficiency improvements under the CDM: thecase of energy-efficient lighting. Energy Efficiency, 2,353–3678. doi:10.1007/s12053-009-9052-z.

Mills E. (2010) From carbon to light. Prepared for the UnitedNations Framework Convention on Climate Change,Clean Development Mechanism Executive Committee,Small-Scale Working.

Mills, E. (2003). Risk transfer via energy savings insurance.Energy Policy, 31, 273–281. Available from: http://evanmills.lbl.gov/pubs/pdf/energy_savings_insurance.pdf.

Mills, E., & Borg, N. (1999). Trends in recommended lightinglevels: an international comparison. Journal of theIlluminating Engineering Society of North America, 28(1), 155–163.

Mills, E. (2005). The specter of fuel-based lighting. Science, 308,1263–1264. Available from: http://eetd.lbl.gov/newsletter/nl21/2fuel_lite.htm.

Mills, E. (2009). From risk to opportunity: insurer responses toclimate change. Boston: Ceres.

Mills, E. and A. Jacobson (2007) The off-grid lighting marketin Western Kenya: LED alternatives and consumerpreferences in a millennium development village. LuminaProject Technical Report #2. Lawrence Berkeley NationalLaboratory and Humboldt State University. Availablefrom: http://light.lbl.gov/pubs/tr/lumina-tr2.pdf.

Mills, E., & Jacobson, A. (2008). The need for independentquality and performance testing for emerging off-gridwhite-LED illumination systems for developing countries.Light & Engineering, 16(2), 5–24.

Mink, T., P. Alstone, J. Tracy, and A. Jacobson (2010) LEDFlashlights in the Kenyan Market: quality problemsconfirmed by laboratory testing. Prepared for LightingAfrica, IFC and World Bank.

Mulugeta, E. (2004). The demand for kerosene and per capitaincome in Ethiopia. Ethiopian Journal of Economics, 13(2), 35–60. Available from: http://ajol.info/index.php/eje/article/view/39807.

Munich RE (2007) Topics: cycle management; climateneutrality: Kyoto Multi Risk Policy. Munich Reinsur-ance Company. Available from: www.munichre.com/publications/302-05473_en.pdf.

National Bureau of Statistics Tanzania (2002) HouseholdBudget Survey: 2000/01.


http://light.lbl.gov/pubs/tr/lumina-tr9-summary.html


http://www.energy.ca.gov/appliances/index.html

http://www.energy.ca.gov/appliances/index.html

http://www.lightingafrica.org/node/191/

http://www.lightingafrica.org/files/Lighting_Africa_Briefing_Notes_21.Dec_.2009.pdf

http://www.lightingafrica.org/files/Lighting_Africa_Briefing_Notes_21.Dec_.2009.pdf

http://dx.doi.org/10.1007/s12053-009-9052-z

http://evanmills.lbl.gov/pubs/pdf/energy_savings_insurance.pdf

http://evanmills.lbl.gov/pubs/pdf/energy_savings_insurance.pdf

http://eetd.lbl.gov/newsletter/nl21/2fuel_lite.htm

http://eetd.lbl.gov/newsletter/nl21/2fuel_lite.htm

http://light.lbl.gov/pubs/tr/lumina-tr2.pdf

http://ajol.info/index.php/eje/article/view/39807

http://ajol.info/index.php/eje/article/view/39807

http://www.munichre.com/publications/302-05473_en.pdf

http://www.munichre.com/publications/302-05473_en.pdf

Nieuwenhout, F. D. J., A. van Dijk, V. A. P. van Dijk, D.Hirsch, P. E. Lasschuit, G. van Roekel, H. Arriaza, M.Hankins, B. D. Sharma, and H. Wade. (2000) Monitoringand evaluation of solar home systems: experiences withapplications of solar PV for households in developingcountries. ECN-C--00-089.

Radecsky, K., P. Johnstone, A. Jacobson, and E. Mills (2008)Solid-state lighting on a shoestring budget: the economicsof off-grid lighting for small businesses in Kenya. LuminaProject Technical Report #3. Lawrence Berkeley NationalLaboratory and Humboldt State University. Availablefrom: http://light.lbl.gov/pubs/tr/lumina-tr3.pdf.

Schapiro, M (2010) Conning the climate. Harpers. pp. 31–39.Stevens, J.W. and G.P. Corey (1996) A study of lead-acid battery

efficiency near top-of-charge and the impact on PV systemdesign. Photovoltaic Specialists Conference. Record of theTwenty Fifth IEEE.Washington, DC, USA. p 1485–1488

Tracy, J., P. Alstone, A. Jacobson, E. Mills, and P. Avato (2011)Low-cost LED flashlights and market spoiling in Kenya’soff-grid lighting market. Energy Policy (in press).

Tracy, J. and E. Mills (2010) Illuminating the pecking order inoff-grid lighting: a demonstration of LED lighting forsaving energy in the poultry sector. Lumina ProjectTechnical Report #8. Lawrence Berkeley National Labo-ratory and Humboldt State University. Available from:http://light.lbl.gov/pubs/tr/lumina-tr8-summary.html.

Tracy, J., P. Alstone, A. Jacobson, and E.Mills (2010a)Market trial:off-grid LED lighting product market potential. LuminaProject Technical Report No. 6. Lawrence Berkeley NationalLaboratory and Humboldt State University. Available from:http://light.lbl.gov/pubs/tr/lumina-tr6-summary.html.

Tracy, J., A. Jacobson, and E. Mills (2010b) Use patterns ofLED Flashlights in Kenya and a one-year cost analysis offlashlight ownership. Lumina Project Research Note #5.Lawrence Berkeley National Laboratory and Humboldt

State University. Available from: http://light.lbl.gov/pubs/rn/lumina-rn5-torch-costs.pdf.

Tracy, J. and E. Mills (2010) Illuminating the pecking order inoff-grid lighting: a demonstration of LED lighting forsaving energy in the poultry sector. Lumina ProjectTechnical Report #8. Available from: http://light.lbl.gov/pubs/tr/lumina-tr8-summary.html.

Tracy, J., A. Jacobson, and E. Mills (2009) Quality andperformance of LED flashlights in Kenya: common enduser preferences and complaints. Lumina Project ResearchNote #4. Lawrence Berkeley National Laboratory andHumboldt State University. Available from: http://light.lbl.gov/pubs/rn/lumina-rn4-torches.pdf.

US Department of Energy (2006) Lifetime of white LEDs.PNNL-SA-50957.

US Department of Energy (2009) Multi-year programplan FY’09–FY’15: Solid-State Lighting Research andDevelopment.

UNFCCC (2010) Approved methodologies for small-scaleCDM project activities. AMS-III.AR: substituting fossilfuel based lighting with LED lighting systems–version1.0. Available from: http://cdm.unfccc.int/methodologies/DB/UMZGFR9COL8J0SRQXBVYR3DEM9F4TM.

UNFCCC (2006) D.light Rural Lighting Project: project designdocument form (CDM-SSC-PDD)–version 03." 22 Dec.pp. 68. Available from: http://cdm.unfccc.int/Projects/DB/DNV-CUK1226479189.57. Accessed on 3 Sep 2010.

UNFCCC (2009) Rural Education for Development Society(REDS) CDM Photovoltaic Lighting Project: ProjectDesign Document Form (CDM-SSC-PDD)–version 03.August 4, pp. 29.

World Energy Outlook (2009) Number of people withoutaccess to electricity in the reference scenario. Availablefrom: www.iea.org/country/graphs/weo_2009/fig2-10.jpg.


http://light.lbl.gov/pubs/tr/lumina-tr3.pdf



http://light.lbl.gov/pubs/rn/lumina-rn5-torch-costs.pdf

http://light.lbl.gov/pubs/rn/lumina-rn5-torch-costs.pdf



http://light.lbl.gov/pubs/rn/lumina-rn4-torches.pdf

http://light.lbl.gov/pubs/rn/lumina-rn4-torches.pdf

http://cdm.unfccc.int/methodologies/DB/UMZGFR9COL8J0SRQXBVYR3DEM9F4TM

http://cdm.unfccc.int/methodologies/DB/UMZGFR9COL8J0SRQXBVYR3DEM9F4TM

http://cdm.unfccc.int/Projects/DB/DNV-CUK1226479189.57

http://cdm.unfccc.int/Projects/DB/DNV-CUK1226479189.57

http://www.iea.org/country/graphs/weo_2009/fig2-10.jpg

http://www.iea.org/country/graphs/weo_2009/fig2-10.jpg

Estimating Greenhouse Gas Emissions Reducitons from Replacing Fuel-Based Lighting with LED

Documents

Transcript of Estimating Greenhouse Gas Emissions Reducitons from Replacing Fuel-Based Lighting with LED