winner, loser, or innocent victim? has renewable energy performed as expected?

Solar Energy Vol. 68, No. 3, pp. 237–255, 20002000 Elsevier Science Ltd

Pergamon PII: S0038 – 092X( 99 )00073 – 0 All rights reserved. Printed in Great Britain0038-092X/00/$ - see front matter

www.elsevier.com/ locate / solener

WINNER, LOSER, OR INNOCENT VICTIM? HAS RENEWABLE ENERGYPERFORMED AS EXPECTED?

,†JAMES McVEIGH*, DALLAS BURTRAW** , JOEL DARMSTADTER** andKAREN PALMER**

*School of Public and Environmental Affairs, Indiana University, Bloomington, IN, USA**Resources for the Future, 1616 P Street, NW, Washington, DC 20036, USA

Received 13 August 1999; revised version accepted 18 October 1999

Communicated by ARI RABL

Abstract—This study provides an evaluation of the performance of five renewable energy technologies usedto generate electricity: biomass, geothermal, solar photovoltaics, solar thermal, and wind. We compared theactual performance of these technologies against stated projections that helped shape public policy goals overthe last three decades. Our findings document a significant difference between the success of renewabletechnologies in penetrating the US electricity generation market and in meeting cost-related goals, whencompared with historic projections. In general, renewable technologies have failed to meet expectations withrespect to market penetration. They have succeeded, however, in meeting or exceeding expectations withrespect to their cost. To a significant degree, the difference in performance in meeting projections ofpenetration and cost stem from the declining price of conventional generation, which constitutes a movingbaseline against which renewable technologies have had to compete. 2000 Elsevier Science Ltd. All rightsreserved.

1. INTRODUCTION AND OVERVIEW This study does not address the merits of thisclaim nor its implication for public policy; instead

At least since the oil spill off the coast of Santait focuses on the premise of the argument — that

Barbara, California in 1969, the energy pricerenewable technologies have not met the goals

upheavals of the 1970s, and the Three Mile Islandand claims of their proponents. We evaluate the

meltdown of 1979, the choice of technologies toperformance of renewable technologies for elec-

meet US energy demand has appeared prominent-tricity generation measured against stated timet-

ly on the agenda of US public opinion and policy.ables that helped shape public policy goals, and

Concerns about the environment, the economy,we also evaluate these technologies against pro-

equity, monopoly power, and the role of thejections and trends in conventional electric power

public sector have been manifest in the publicgeneration. The renewable energy technologies

debate. One outcome has been public policy andwe investigated are biomass, geothermal, solar

public-sector support, albeit sometimes falteringphotovoltaics, solar thermal, and wind.

support, for the development of renewable energyOur findings refute the premise that renewable

technologies.technologies have failed to meet public policy

Nearly 30 years into this public discussion, thegoals. To summarize briefly, we confirm that

reality is that renewable technologies have failedpenetration into the market has fallen far short of

to emerge as a prominent component of the USprojections. However, the costs of these tech-

energy infrastructure. This has created the percep-nologies generally have fallen in accordance with

tion that renewables have not met the goals andthe projections of their proponents, sometimes

claims of proponents. The implication is that afterexceeding the projected decline. This is remark-

several decades of support without success, it isable, given that renewable technologies have not1time to pull the plug on renewables.significantly penetrated the market, nor have theyattracted large-scale investment and productionthat can contribute to technological development

†Author to whom correspondence should be addressed. Tel.: or economies of scale in production, as many1 1-202-328-5087; fax: 1 1-202-939-3460; e-mail:

analysts anticipated when forming their [email protected] projections.For example, see Management Information Services (1998)

and Bradley (1997). Our analysis indicates that the small market

237

238 J. McVeigh et al.

share of renewable technologies appears to have about the possible future role of renewable energymore to do with changes outside their develop- technologies, and about whether public policy canment than with their own technological perform- contribute effectively to the direction and pace ofance. Over roughly three decades, regulatory technological change.reform has swept the energy industry. Prominent Although we reject the negative, we cannotpolicy changes such as the deregulation of natural make an unambiguously positive assertion. Thatgas and oil coupled with declining technical costs is, we do not attempt to attribute the successfulof production have made these conventional achievement of projected technological develop-energy fuels less expensive. An increasingly ment and cost declines to a specific governmentcompetitive world petroleum market has led to a policy or any other factor. We do not make adecline and stabilization in the price of oil, such direct case for continued government support ofthat currently the real price of oil is at its lowest these technologies. Nonetheless, the successful

2since 1973. Deregulation of the railroads led to achievement of cost-related goals provides somedramatic cost reductions for coal use in electricity reason for optimism with respect to the role ofgeneration. The Public Utility Regulatory Policies renewable technologies and of public policy inAct (PURPA) of 1978 opened the door for non- meeting future challenges such as reducing green-utility generation of electricity for resale, and the house gas emissions.Energy Policy Act of 1992 paved the way forcompetitive wholesale generation. These changes

2. METHODSand other technological developments have pro-duced a dramatic decline in the price of fossil- To evaluate the performance of renewablefueled electricity generation. In addition, public technologies, we documented projections made bypolicy and technological changes have led to a a variety of organizations, and compared themdramatic improvement in the environmental per- with what has actually transpired over the lastformance of these technologies (especially for three decades. These forecasts varied with respectnewly constructed facilities). to their intervals and ultimate time horizon. We

The ultimate impacts of these changes in the report projections in five-year increments begin-regulation, technology and market structure of ning in the 1970s, with forecasts as far as 2020.fossil fuels have been mostly favorable for elec- Our principal focus is performance to the mid-tricity consumers; they have also been frustrating- 1990s. We follow up that discussion with aly disappointing for the fate of renewable tech- retrospective look at how projections for conven-nologies. Renewable technologies have had to tional power systems compare with actual out-

3compete in this changing environment. Hence, comes over the same time horizon.renewable technologies may be seen as a relative

2.1. Technologiesloser — perhaps the innocent victim — amidst thewidespread success of a wide array of public Each category of renewable energy tech-policies aimed at energy markets. nologies for electricity generation reviewed en-

We conclude that many significant expectations compasses a variety of technologies, but for easeand public policy goals regarding development of of discussion they have been aggregated intorenewable technologies for electricity generation biomass, geothermal, solar photovoltaics, solarhave been achieved. Any argument that public thermal, and wind. Hydropower has been ex-policy support for renewable technologies should cluded for two reasons: first, it has been de-be ended because ‘past efforts have been unsuc- veloped so extensively that typically it is consid-cessful’ is based on a faulty premise. These ered a conventional source of electricity, and,findings should be of interest in the policy debate second, significant expansion of this resource

would face severe opposition based on environ-mental concerns. Similar considerations apply to

2 our exclusion of nuclear power, whose underlyingAlthough the role of oil in electricity generation has di-resource base could, like geothermal energy, beminished over time, it remained important to generation

during peak periods of demand until recently. More viewed as virtually unlimited.important, in the 1970s and early 1980s the choice amongconventional fuels was between oil and coal, as theavailability of natural gas was in question. Hence, thedecline in oil prices had an influence on the pace of

3technological change and cost reductions in the coal Conventional power systems include all forms of generation,industry. of which renewables are just a small portion.

Winner, loser, or innocent victim? Has renewable energy performed as expected? 239

2.2. Subsidies and incentives tent, or was dominated by support for convention-al technologies, this could have weakened theThe studies we reviewed differed in their levelperformance of renewable technologies in com-of sophistication of potentially important assump-parison with the projections.tions, including the treatment of subsidies and

incentives for the development of renewable and 2.3. Characteristics of the technologiesnon-renewable technologies. On the federal level,

Another missing and potentially importantthese incentives included investment tax credits,piece of the analysis is an accounting of the

production credits, and accelerated depreciation ofspecial characteristics of renewable technologies

capital. Many states offered various incentives forthat make their marginal cost in the delivery of

renewables, such as California’s Interim Standardenergy services (and their environmental charac-

Offer contracts, which were offered to Qualifying teristics) differ in qualitative ways from otherFacilities and Cogenerators under PURPA, and technologies. One aspect that detracts from theiradditional investment tax credits of 10–15%. value to an electricity grid is the intermittentDirect expenditure by government on research generation potential of solar and wind resources.and various other subsidies contributed to the Absent a technology such as hydroelectricdevelopment of not only renewable technologies pumped storage or batteries to store electricity4but non-renewable technologies as well. and/or potential energy, and the energy from the

The varied treatment of incentives in these sun and the wind are only available a portion ofstudies is relevant in two ways. If possible, it each day. Often the availability coincides withwould be best to control for the assumption about periods of peak energy demand, but not always.the level of public-sector incentives over the The possible unavailability of these resources,horizon contained in each study with respect to especially at peak periods, detracts from theirthe cost projections and reported cost of each potential contribution to a system grid.study, as well as with respect to projected market On the other hand proponents point to certainpenetration. Unfortunately, most of the studies characteristics of these technologies — theirfailed to make their assumptions explicit. Those relatively small scale and independence from fuelthat did so did not offer enough complementary supply – as virtues that also are missing from ourdetail to allow us to disentangle the effect of these analysis. These attributes make siting easier andassumptions from those of others in the study. especially practical in remote areas not served by

6Therefore, on this and many other issues we the electricity grid. This feature enhances theaccept the projections at face value. That is, we ‘niche market’ appeal of renewable technologies,do not adjust the projections for potential differ- particularly in remote areas of the Third World.

5ences in their underlying assumptions. Clearly, Hence renewable energy will often compete notprojections of cost and market penetration some- on the cost of energy, but on the basis of valuetimes were built with anticipation of sustained provided to the customer. In addition, the distrib-high levels of government support. To the extent uted nature of these generation resources can bethat this support did not materialize, was intermit- used to ease congestion and loop-flow problems

on an electricity grid, thereby adding to theirvalue within an electric system.

4Expenditure through the Department of Energy on research,development, and demonstration projects for renewable 2.4. Selecting studies for reviewand fossil technologies were about equal in 1980 and both

In designing the study, we tried to cover a widewere in excess of $1.3 billion. Both fell by nearly three-range of the projections that were cast into thequarters by 1985, but by 1990 expenditure on renewables

continued to fall to $129 million while expenditure on public debate, though of course we could not dofossil rose to over $1.1 billion. In 1995 they were again so exhaustively. About 60 studies were located.similar, with renewables receiving $342 million and fossil Not all of those found could be given equalreceiving $504 (OTA, 1995, p. 33, 1995 dollars).

5 weighting, because the rigor of the analysesComments on preliminary versions of this paper have been onvaried tremendously. To account for this variationopposite sides of this issue. Renewable advocates have

argued that we should look to contracting issues, subsidies,and market imperfections as obstacles to renewable tech-

6nologies. A critic of renewables has argued the opposite, There may be other impediments to siting of renewablethat we should look to these same issues to find preferen- generating facilities. For example, objections to noisetial treatment of renewables. Indeed, these are important levels and potential damages to migrating birds canissues that have significant bearing on an evaluation of complicate the siting of wind turbines (Gipe, 1995;renewables, but they are not the focus of this study. Bradley, 1997).


we developed a qualitative scheme to evaluate the Our primary focus is a chronological organizationstudies. by the decade when the studies were written

First, on subjective but fairly transparent (1970s, 1980s and 1990s). A secondary focus isgrounds, we reduced the number of studies we the affiliation of the authors: government agen-

7reviewed in detail to 25. Second, we constructed cies, research institutions (including national labsexplicit criteria and evaluated the studies in light and academic groups), the Electric Power Re-of these in order to develop weights that were search Institute (EPRI), and non-governmentalapplied to each study in the aggregate analysis. organizations (NGOs). In our sample, projections

8The criteria are listed in Tables 1 and 2. from within the renewable energy technologyFor each result, the median value among the industry itself are not represented due to our

10studies (after accounting for the weighting) is the inability to locate original sources.9point used as an estimate. It is noteworthy that

2.5. Evaluation criteriathe results are not highly sensitive to weightsgiven to the studies. The overall results displayed 2.5.1. Market penetration. The first evaluationin the next section are largely unaffected by our criterion we considered was penetration into therating scheme compared with one that would have market or, the equivalent, the contribution ofapplied equal weights to each of the 25 studies. technologies to electricity supply. This is mea-This is partly due to the use of a median value sured by electricity generation and installedrather than a mean value of the weighted studies. capacity. We placed primary emphasis on elec-This also indicates that the rigorous and not-so- tricity generation — the measure of how muchrigorous studies were distributed about equally in energy is actually produced by a specific technol-the sample. ogy over the course of a year, reported in million

Also reported in Tables 1 and 2 is an affiliation kiloWatt hours (kW h), or gigaWatt hours (GW h).of the author or authors by categories described We occasionally refer to capacity, which is thebelow. The column Broad Technical Specification measure of how much electricity is available atindicates whether the study addressed all, many, any one point in time reported in megaWattsor just one of the renewable technologies we (MW), and is the sum of all nameplate ratings of

11considered. Further, the tables indicate the range the respective generation sources.of years covered in the study. These two columnswere not used in weighting the studies. 2.5.2. Cost. The second evaluation criterion

We organized the studies in two different ways. was cost, measured by the levelized cost ofelectricity generation and by capital costs. Thecost of electricity at point of production was our

7 primary measure, and it incorporates capital, fuel,For example, several studies were excluded because we wereunable to locate the entire report. Others were largely and operation /maintenance (O/M) costs, as welljournalistic in nature, lacking the technical detail to make as expected lifetime and capacity factors. Thetheir inclusion useful. total costs of production over the lifetime of the8Criteria concerning projections of electricity production in-

facility were amortized in a straight-line fashionvolved whether the following assumptions were explicit:(just as payments for a standard home mortgagepolicy initiatives, total electricity /energy demand, cost of

conventional generation / fuel, and the continued existence would be). This annual cost was divided by theof tax credits. Also we considered whether the assumptions average annual amount of electricity producedor results were specified by region, and whether the study over that lifetime to calculate the levelized cost ofwas original work. Criteria having to do with projections

electricity generation (COE). We assume no infla-of cost involved whether the following were explicit: thetion. Levelized cost is reported in mills. perdiscount rate, year in which dollars are denominated,

operating and maintenance cost, and capacity utilizationrates. Again, we considered whether the study was original

10work. A considerable effort was made to contact industry associa-9A ‘score’ of 0.5, 1.0 or 1.5 was given to each study with tions and individual firms, some of which provided us with

respect to each of the criteria, and the scores were tallied to technical background or served as reviewers of this study.develop an overall weight for each study with respect to However, off-the-shelf estimates by firms in the renewableproduction projections, and separately with respect to cost technology industries were not readily available to us.

11projections. Weights for projections of cost of capital and We restrict our focus to the US market, to the exclusion ofcost of electricity differ because the studies differed with the expanding opportunities for US technology in foreignrespect to whether this information was explicit. Similarly, markets. However, we do not believe this was the primarythe weights for projections of capacity and generation context for public policy debates over the period wediffer because of information that may not have been considered, nor was it a primary element of the studies weexplicit in the study. reviewed.

Winner,

loser,or

innocentvictim

?H

asrenew

ableenergy

performed

asexpected?

241

aTable 1. Weighting of studies in cost projections

Group Study Broad Projected Criteria considered Final weightingaffiliation technology years Discount Dollar year Operation Capacity Original Cost of Cost ofspecification given rate specified /maintenance factor work capital electricity

indicated

GOVT FEA (1974a,b) d 1980, 1985, 1990 d d d s d d dGOVT ERDA (1975) s 1985, 2000 s s s s d s sGOVT FEA (1976) d 1980, 1985, 1990 d d ( s d ( dEPRI EPRI (1977) s 1975–2022 d d s s ( ( (NGO Lovins (1977) ( 1975–1990 ( d s s ( d s

GOVT OTA (1978) s 1980, 2000 s d ( ( ( ( (GOVT NAS-CONAES (1979a,c) d 1980, 1985, 1990, 2000 d d ( d d d d

Research Stanford (1979) d 1990, 2025 d d d ( s d dGOVT NAS-CONAES (1979b) d 1980, 1985, 1990, 2000 d d ( d d d dGOVT EIA (1980) d 1975–2020 s d s s d ( s

Research SERI (1981) ( 1980–2000 d d d d d d sResearch Stoughbaugh and Yergin (1979) ( 1980–2000 s s s d s d s

EPRI EPRI (1982) ( 1982, 1986, 1990 d d s s ( d sGOVT USDOE (1982) d 1985–2000 s d s s d s sGOVT USDOE (1983) d 1985–2010 s d s s d s sNGO Deudney and Flavin (1983) d 1980–2000 s s s ( ( d (

GOVT OTA (1985) d 1990–2000 d d d d d d (GOVT EIA (1990) d 1990–2030 ( d d d ( d sNGO Flavin and Lessen (1990) d 1980–2030 s d s s ( s d

GOVT OTA (1995) d 2000–2030 d d d d d d dNGO Gipe (1995) s 1990–2000 d d ( d ( d d

GOVT EIA (1997) d 1995–2020 ( d s d d s sGOVT PCAST (1997) d 1995–2010 ( d ( ( ( d dEPRI EPRI–USDOE (1997) d 1997–2020 d d d d d d s

GOVT EIA (1997) d None s d O ( d ( sa Weights: solid, 1.5; partially solid, 1.0; empty, 0.5.

242J.

McV

eighet

al.

aTable 2. Weighting of studies in production projections

Group Study Broad Projected Criteria considered Final weightingaffiliation technology years Policy Total Cost of Tax Regional Original Capacity Production

specification givenassumptions energy/ conventional credits /production specifications work

electricity generation / fuel incentivesneeds

GOVT FEA (1974a,b) d 1980, 1985, 1990 d d d s s d d d

GOVT ERDA (1975) s 1985, 2000 d d ( s s d ( (

GOVT FEA (1976) d 1980, 1985, 1990 d d d ( s d d s

EPRI EPRI (1977) s 1975–2022 ( d d ( d ( s d

NGO Lovins (1977) ( 1975–1990 ( d d ( s ( s s

GOVT OTA (1978) s 1980, 2000 ( ( s d ( ( ( s

GOVT NAS-CONAES (1979a,c) d 1980, 1985, 1990, 2000 d d ( ( ( d d d

Research Stanford (1979) d 1990, 2025 d d d ( ( s s d

GOVT NAS-CONAES (1979a,b,c) d 1980, 1985, 1990, 2000 d d ( ( ( d d d

GOVT EIA (1980) d 1975–2020 d d d s s d s d

Research SERI (1981) ( 1980–2000 d d d ( d d d (

Research Stoughbaugh and Yergin (1981) ( 1980–2000 ( ( ( ( s s s s

EPRI EPRI (1982) ( 1982, 1986, 1990 ( ( s s d ( s s

GOVT USDOE (1982) d 1985–2000 d d d s s d s d

GOVT USDOE (1983) d 1985–2010 d d d s s d s d

NGO Deudney and Flavin (1983) d 1980–2000 ( ( ( s ( ( d s

GOVT OTA (1985) d 1990–2000 d d d d d d ( s

GOVT EIA (1993) d 1990–2030 d ( d s s ( s d

NGO Flavin and Lessen (1990) d 1980–2030 ( ( s ( s ( s s

GOVT OTA (1995) d 2000–2030 d ( ( d ( d s s

NGO Gipe (1995) s 1990–2000 ( ( ( d ( ( s d

GOVT EIA (1997) d 1995–2020 d d d d d d d d

GOVT PCAST (1997) d 1995–2010 d ( ( d d ( ( s

EPRI EPRI–USDOE (1997) d 1997–2020 d ( ( ( d d s s

GOVT EIA (1997) d None d ( ( s ( d d d

a Weights: solid, 1.5; partially solid, 1.0; empty, 0.5.


kiloWatt hour (mills /kW h), where one mill is source of ‘actual’ values in the graphs. The dataequal to one-tenth of one cent. We occasionally tables are sensitive to our assumptions and inter-refer to capital costs, measured by the dollar pretations of previous studies; the trends illus-

12expenditure for the rated capacity in dollars per trated in the graphs are much less sensitive.kiloWatt ($ /kW). Cost data are reported in con-

3.1. Windstant 1995 dollars. To normalize costs we used theconsumer price index, and assumed values in each 3.1.1. Production. In the 1970s, projections forstudy were denominated in dollars for the year the wind generating capacity were high due to as-study was published if no other information was sumptions influenced by the energy market dis-

13given. ruptions of that decade. Projections were as highNone of the studies reviewed offered a com- as 45 000 MW for 1995 and 140 000 MW by

plete set of projections. Transformations were 2000 (NAS-CONAES, 1979c).made from capacity to generation and vice versa, Studies of generation and capacity during theand from capital cost to the levelized cost of following decades offered projections that wereenergy for each of the technologies using standard lower by an order of magnitude, due in large partcapacity and utilization factors. When data were to declining fossil fuel prices. In Fig. 1 this isplotted on a logarithmic scale, geometric means illustrated by a large shift downward in projec-were used to interpolate estimates for missing tions of generation after the 1970s. Projections oftime periods between projected years in each generation and capacity from the 1990s are con-study; otherwise, arithmetic means were used. sistent with those from the 1980s. The industry

For each technology we also constructed a experienced a brief decline in capacity in the earlymeasure of actual generation and cost. For the 1990s, due in part to retiring standard-offerearly years, assessments by different organizations contracts under PURPA and a decline in otherof the actual generation and costs of renewable public-sector incentives. Actual wind energy gen-technologies often disagreed. In such cases, we eration has been on the rise since the mid-1970s,relied on whatever assessments were available but specific estimates prior to 1990 are quitefrom the Energy Information Administration uncertain and are not included in the graphs. In(EIA). 1995, wind production was approximately 3196

GW h.

3. FINDINGS3.1.2. Costs. Fig. 2 illustrates that projections

In this section we summarize projections and of a decline in the capital cost and COE of windactual performance for each renewable technolo- have been realized or exceeded over time. Windgy, and end with a review of projections for energy (along with geothermal) is the least-costelectricity generation from conventional sources. renewable technology currently, but its competi-Note that many of the graphs use a log scale to tive standing depends on the availability of sitesdisplay results. Note also that sharp changes in the with strong wind resources and with access toslopes of some of these graphs are in part an transmission lines. Some early projections ex-artifact of our chosen methodology, which selects pected that the exhaustion of good wind resourcea median value among projections for each five- sites early in the development of the technologyyear increment. There is considerable noise in the would prevent costs from falling. For instance, thedata (i.e. a random aspect to the estimates due to Committee on Nuclear and Alternative Energyour interpretation of the studies and translation of Systems of the National Academy of Sciencestheir results into common forms of measurement) envisioned rising capital costs and COE afterdue to reasons we have outlined. Hence, we chose 1995 for wind energy due to the need to use sitesto report our findings as graphs, which communi- ‘with lower average wind speeds because the bestcate the qualitative nature of the findings better sites have already been used’ (NAS-CONAES,than data tables would. In some cases, inconsis- 1979c). The exhaustion of valued wind sites hastencies are apparent between ‘actual’ values and not occurred, however, because of the unmet‘projected’ values, or among ‘historic’ valuesprojected at various points in time. This stems

12Data tables are available from the authors.from the difficulty faced by early studies in 13This analysis does not include cost projections for small-assessing actual market performance, due in part scale wind turbines intended for distributed applications —to heterogeneity in technology and suppliers. We i.e. applications located close to the user that tend to berely primarily on government studies as the less than 50 kW in size.


Fig. 1. Wind generation by date of forecast.

Fig. 2. Wind cost of electricity by date of forecast.

penetration of wind into the market and because the average cost of generation from conventionalthe inventory of sites identified to have strong sources.

14wind resources has expanded. In addition, tech-nological advances such as lower start-up speeds 3.2. Solar photovoltaicshave improved profitability at lower wind speeds.

Wind has a current cost of about 52 mills /kW h 3.2.1. Production. Like wind, photovoltaic costat existing facilities (Gipe, 1995; OTA, 1995), projections have been revised downward from theand a recent bid for 30 mills /kW h for a 100-MW 1970s. Projections from the 1970s for solarwind farm was submitted to Northern States photovoltaic range tremendously, as seen in Fig.Power in Minnesota (Regulatory Assistance Pro- 3. Some authors forecast nearly 35 000 MW of

15ject, 1998). Current cost estimates are close to capacity (FEA, 1974b) and 150 000 GW h ofgeneration by 2000 (NAS-CONAES, 1979c);

14The siting process includes a trade-off between the highest while others forecast nearly zero capacity andquality wind resource locations and proximity to transmis- generation by 2020 (EPRI, 1977). Viewing thesion lines. median value of projections of generation15Presumably this bid takes advantage of the 15 mills /kW h

chronologically reveals a ‘fan diagram’ that re-Renewable Energy Production Credit currently available tosults from downward revisions of expected pene-wind and closed-loop biomass, which would be reflected in

the bid price. tration. This pattern appears often in the figures


Fig. 3. Solar photovoltaic production by date of forecast.

on other technologies as well. Capacity and costs at around $3000/kW and of COE at aroundgeneration have grown more than 10-fold since 150 mills /kW h by 1990 (NAS-CONAES,the early 1980s, but market penetration has been 1979c). Fig. 4 indicates these early projections ofconfined to niche markets and remote applica- costs were not achieved.tions. In 1995, about 89.2 MW of capacity were Since the 1980s, however, projections of COEinstalled in the United States. declines have been met or exceeded. Despite

limited market penetration, capital costs and COEhave dropped significantly. The current cost of

3.2.2. Costs. Early studies were relatively op- capacity is about $6000/kW, and the cost oftimistic regarding trends in costs for photovoltaic generation is nearly 300 mills /kW h. Recenttechnology compared with those developed later. studies project continuing declines in cost inStoughbaugh and Yergin (1979) projected that by coming decades due to efficiency improvements1990 the cost of electricity from photovoltaic in both manufacturing the cells and capturingtechnology would be about 100 mills /kW h solar radiation and transforming it to electricity(Stoughbaugh and Yergin, 1979). Another early (OTA, 1995; PCAST, 1997; EPRI–USDOE,study projected an eventual leveling off of capital 1997).

Fig. 4. Solar photovoltaic cost of electricity by date of forecast.


Fig. 5. Solar thermal generation by date of forecast.

However, reductions in public-sector financial3.3. Solar thermalincentives and government R&D spending on

3.3.1. Production. Projections of solar thermal solar thermal hit this technology particularly hard.electricity production also form a fan diagram. Luz International entered bankruptcy, contributingSolar thermal electricity production began in the to a decline in production in the beginning of thelate 1970s with the 10-MW Solar One, a central 1990s. As with photovoltaics, the outcome is astation receiver, in the desert of southern Califor- fan diagram of declining forecasts of generationnia. Projections made during the late 1970s and apparent in Fig. 5. Recent projections anticipateearly 1980s put solar thermal capacity at any- capacity and generation to increase to twicewhere from 240 MW to 3400 MW, and expected current levels by 2020 (EIA, 1997).generation to range from 480 GW h to 5400GW h by the end of the 1980s (EIA, 1980; NAS- 3.3.2. Costs. Few projections exist for theCONAES, 1979c). Luz International’s facilities capital costs of solar thermal technology, and(SEGS I–IX) provided research on commercial those we found varied greatly with regard to theadaptability of the technology, and demonstration type of technology modeled. There is also sub-of the first Stirling dish engine in 1984 signaled stantial variation in the measure of COE. Projec-that these goals might be attainable by the end of tions from the 1970s for 1990 ranged from 36 tothe decade. Projections for capacity during the 198 mills /kW h (FEA, 1976; NAS-CONAES,1980s reflected this expectation. 1979c). Fig. 6 illustrates that the median projec-

Fig. 6. Solar thermal cost of electricity by date of forecast.


Fig. 7. Geothermal generation by date of forecast.

tions have been tracked closely by the actual generating electricity from geothermal in the16COE. future (EPRI, 1977; NAS-CONAES, 1979c).

The earliest development of geothermal resource3.4. Geothermaland technology used dry steam resources, histori-cally the least expensive form of the resource. It3.4.1. Production. Projections of electricitywas anticipated that as dry resources wereproduction from geothermal produce a muchexhausted, subsequent development would have toweaker version of the familiar fan diagram.turn to more expensive resources. However, tech-Generating electricity from geothermal energynological advances have expanded the types ofinvolves capturing naturally heated steam to drivegeologic settings that can be tapped. Fig. 8turbines. ‘Dry steam’ is the easiest to use andindicates that recent predictions have projected acheapest, and is the resource type found at Thedeclining path from 5.5 to 4 cents /kW h over theGeysers, California, the largest geothermal elec-next 20 years, a smaller decline than projectedtricity generation facility in the US. Increasingly,today for other renewable technologies (EIA,advances in geophysical theory and mapping, as1990; Flavin and Lessen, 1990; EPRI–USDOE,well as more economical drilling technology, have1997).increased the reservoir of geothermal energy that

can be tapped cost-effectively. Also, valuableminerals (e.g. zinc) in the briny by-product can be 3.5. Biomasscaptured and sold to improve the economics of a

3.5.1. Production. Biomass resources can begeothermal project. These advances have notconverted into energy and can be divided intocome quickly compared to expectations in thethree categories: wood and agricultural wastes,1970s (NAS-CONAES, 1979a,c). However, in amunicipal solid waste, and closed-loop biomasstrend similar to petroleum extraction, they havegrown specifically for energy content. All threeenabled new installations in the outer areas ofsources can be used for generating electricityexisting fields that compensate for the moderate(‘biopower’) as well as heat, while the last sourcedecline in production at long-established facilitiesalso is frequently converted into fuel to powersuch as The Geysers.transportation (as in the blending of ethanol withRecent forecasts for new geothermal energygasoline). Currently wood and agricultural wastecapacity and generation have been more moderateaccount for about 70% of biomass capacity.than previous ones (OTA, 1985; EIA, 1997). This

produces in Fig. 7 a weak version of the familiar16fan diagram. The graph of COE displays an unusual pattern for the 1980s.

This is because only one study is used (OTA, 1985), and it3.4.2. Costs. Many early reports from the projected a significant drop in COE between 1990 and

1970s forecast higher costs than those realized for 2000 (150 to 77.8 mills /kW h).


Unfortunately, few early studies dealt specifi- to the advances in co-firing biomass with fossilcally with biopower. Instead, most early studies fuels (e.g. replacing a small percentage of coalaggregated all the varied uses of the fuel and burned in a power plant with biomass), use ofreported future market penetration as the energy waste-to-energy plants to deal with municipal

solid waste, and technological developments forcontent of the fuels it was to replace (includingclosed-loop biomass generation systems, whichinputs to electricity generation). It is thereforehas been developed but is not yet in use.difficult to separate the electricity contribution

The actual production numbers used here arefrom the heat and fuel contribution of thesebased on utility generation alone until 1990. Non-projections, so few studies from before 1990 areutility data (such as the use of wood waste toincluded in our survey.generate electricity in pulp and paper production)Fig. 9 indicates that generation projectionswere not collected by the EIA until 1989, and nodecreased from the 1970s to the 1980s, but thenreliable source could be found for estimation ofexceeded all prior levels in the 1990s. This is due

Fig. 8. Geothermal cost of electricity by date of forecast.

Fig. 9. Biomass generation by date of forecast.


Fig. 10. Biomass cost of electricity by date of forecast.

the generation by these sources before that date. Fig. 10 shows that projections of biomass COEBoth utility and non-utility generation are in- have fallen over time, and expectations have beencluded after that, accounting for the jump in met or exceeded. Currently, biomass costs areactual capacity and generation between 1985 and slightly higher than wind and geothermal, at about1990. 63 mills /kW h. However, biomass is the largest

provider of renewable energy due mainly to its3.5.2. Costs. As in the case of production availability 24 h a day and its ability to co-fire

statistics, few studies segregated the cost data with traditional fossil fuels.among the various uses for biomass. Costs typi-cally are reported in dollars per million BTU 3.6. Conventional generation($ /MBTU) to represent the cost of fuel input in

To provide an estimate of projections forbiomass applications, including production of heatconventional technologies, we relied on a recentand electrical generation. For the evaluation ofEIA document that evaluates forecasts over thebiopower, we converted these estimates to the 19last two decades (EIA, 1998a). The EIA evalua-standard mills /kW h to estimate the COE when ittion begins with forecasts made in the 1982was appropriate to consider it as equivalent to

17 Annual Energy Outlook for all electricity genera-electricity generation.tion. This included non-hydroelectric renewableBiomass costs as a whole parallel more closelytechnologies, which made up a very small part ofthose of conventional fossil fuels because this istotal generation (at most 0.3% in 1982). It alsothe only renewable technology with fuel costs. Aincluded hydroelectric (14%), nuclear (13%), andlarge portion of delivered biomass costs stems 20fossil-fired power production (73%).from the transportation of biomass resources for

18combustion or fuel production. The capital costs 3.6.1. Production. Electricity sales increasedof biomass are similar to those of fossil fuel by about 2.7% a year or about 80% overall fromgenerators. The other renewable technologies 1975 to 1997. Fig. 11 indicates that from 1982reviewed have high initial capital costs and no until now, government forecasts in the Annualfuel costs. Energy Outlook of 2–3% a year annual growth

have been largely accurate. The succession of

17The conversion rate was 10 353 BTU/kW h 5 10 353BTU 3 (0.2931 W h/BTU) fuel per kW h electricity 5

193.034 W /W . Cohen et al. (1995) provide a review of energy price andfuel electricity18This applies to dedicated feedstock fuel, and to wood and production projections for a number of sectors including

agricultural wastes if these are being purchased. Often for electricity beginning in the 1970s. Their finding that pricemunicipal solid wastes there is a tipping fee received by forecasts have been less accurate than production forecaststhe plant for taking in the waste that offsets transportation is reinforced in our review in this section.

20costs. Based on data in EIA (1998b).


Fig. 11. Conventional generation (retail sales) by date of forecast.

forecasts displayed in the figure appear mutually the forecast. During 1983–1995, fuel costs were21consistent and predict well the actual sales. projected to rise 21%; they actually fell nearly

65%. Indeed, the drop in fuel prices exaggerates3.6.2. Costs. In contrast to the agency’s rela- the overall decline in electricity prices, since

tively accurate sales forecast, EIA projections in operation /maintenance costs exceeded the fore-the 1980s significantly overestimated electricity cast estimate, while the other major price com-costs in the future. For example, the 1982 Annual ponent — capital charges — just about equaledEnergy Outlook forecast real electricity prices to the projected number. Table 3 summarizes therise by somewhat more than 8% during 1980– relevant data.1990; in fact, real prices declined by 10%. By the Table 3 also provides a rough estimate of themid-1980s, the softening of world oil prices began breakdown in real delivered electricity price be-to be reflected in EIA’s updated forecasts, so that tween costs incurred at the electric generatingits 1984 Annual Energy Outlook (which serves as plant (so-called busbar costs) and those accountedthe basis for the observations that follow) foresawa real price decline during 1983–1995 of around

Table 3. Actual and projected electricity prices (cents /kW h,5%. The actual decline over the period was more a1983 and 1995), by major components (in 1995 dollars)than 25%.

Actual (1983)It is instructive to dissect the components ofGeneration T/D Totalthat 12-year price change. The steady and increas-

Total 6.1 3.2 9.3ingly sharp fall in world oil prices beginning inCapital 2.2 1.6 3.8

the early 1980s was of prime importance to the Fuel 3.3 – 3.3economics of electric power production, as it was O/M 0.6 1.6 2.2

to other sectors of the economy. Thus, the entireActual (1995) Forecast (1995)difference between the projected and actual

Generation T/D Total Generation T/D Totalenergy price in 1995 arises from the degree toTotal 3.6 3.5 7.1 6.4 2.4 8.8which the fuel component of the price fell short ofCapital 1.7 1.1 2.8 1.7 1 2.7Fuel 1.1 – 1.1 4 – 4O/M 0.8 2.4 3.2 0.7 1.4 2.1

21 aEIA (1998a) states that the reports published in the years Sources and notes: Total columns from EIA, Annual1983–1988 were titled as the Annual Energy Outlook Energy Outlook (various issues). The split between plant and

T/D is approximated on the basis of information in OECD/(1982)–Annual Energy Outlook (1987). In 1989, theIEA, Projected Costs of Generating Electricity: Update 1998numbering scheme changed, and that year’s report was(Organisation for Economic Co-operation and Development,titled the Annual Energy Outlook (1989). Thus, although aParis, 1998). See also: Electricity Prices in a Competitiveforecast has been published annually, there is no AnnualEnvironment (EIA, US Department of Energy, Washington,

Energy Outlook (1988). Hence, for AEO82–AEO87, the DC, August 1997); Financial Statistics of Major US Investorforecasts were based on data ending with the year iden- Owned Electric Utilities (EIA, US Department of Energy,tified in the title. For AEO89–AEO98, the forecasts were Washington, DC, 1996, 1997) and Financial Statistics ofbased on data ending with the year prior to the one Major US Publicly Owned Utilities (EIA, US Department ofidentified in the title. Energy, Washington, DC, 1997).


Fig. 12. Conventional generation cost of electricity (retail prices) by date of forecast.

22 retail prices attributable to generation and aggre-for by transmission /distribution (T/D) costs.gated these estimates into groups of 4 years each,The former is of particular interest here, since itover the period examined in EIA (1998a,b),provides the more appropriate basis of compari-Annual Energy Outlook Forecast Evaluation. Fig.son with renewable technologies, whose cost is12 indicates that the familiar fan diagram emergesmeasured at the point of electricity generationespecially clearly when viewing the projectionsrather than delivery. In contrast to the 21% gapcompared with actual value for the generationbetween projected and realized delivered electrici-portion of retail price projections.ty prices, Table 3 indicates that the actual out-

come in 1995 with respect to real generation costswas about 44% below what had been forecast 3.7. Viewing projections by author affiliationfrom a 1983 base. This larger difference is notsurprising since the fuel component (whose costs 3.7.1. Generation /market penetration. To theare highlighted in Table 3) applies exclusively to extent that NGOs have historically served asthe generation stage. In total, we estimate that advocates for renewable technologies, theseabout 51% of the cost of retail electricity costs is groups could presumably be expected to have

23attributable to generation. been most optimistic with respect to the potentialTo develop an estimate of cost projections for for renewables in general and market penetration

generation from all sources, which are predomi- in particular. We did not find this to be the case,nantly non-renewable, we relied on estimates of however.

For wind, geothermal and biomass, NGOs werethe most conservative in their projections of

22This decomposition is crude. For instance, EIA (1997) (p. capacity and generation, and in each case they11) presents a different estimate of the share of costs were below levels actually realized. Projectionsassociated with T/D (EIA, 1997). However, we believe

by NGOs were relatively conservative with re-any bias implicit in our methodology to be relativelyspect to generation from photovoltaics as well,unimportant. We seek to find a rough baseline against

which we can compare trends, in an order of magnitude with projections below realized levels. Only in thethat is roughly comparable to the costs of generation with case of solar thermal technology were NGOsrenewable technologies. relatively high in their projections, although not23The data suggest that 61% of total capital costs are attribut-

the highest. (All groups overstated the marketable to generation and 39% to T/D. We apply 25% of totalpenetration of solar thermal to date.)O/M costs to generation, and 75% to T/D. For 1995

actual, EIA shows a ‘wholesale power cost’ of 0.4 cents / Studies sponsored or conducted by governmentkW h in addition to the other three components shown. We (more than half of our sample) and independentallocated the 0.4 cents in proportion to the other three research organizations (which include the nationalcomponents. We apply the shares of costs attributable to

laboratories) have tended toward the highestgeneration and to T/D that are estimated for the year 1995projections of production and capacity. For allto the calculations for 1983 in Table 3, and the discussion

in the remainder of this section. technologies except biomass, these studies have


issued projections that erred on the high side effect that higher energy prices would have onwhen compared with actual installed capacity and demand for energy.generation. Studies associated with research or- Beginning in the early 1970s, the rate of growthganizations were almost accurate with respect to for electricity demand deviated from historicgeothermal. These organizations also offered the trends. Indeed, the error in demand projections ledhighest projections for biomass, which turned out to an overcommitment to capacity additions withto be quite accurate. vintages of technology — both nuclear and fossil

Studies by EPRI usually, though not always, — from earlier decades that has important linger-offered the most conservative projections across ing implications for the electricity industry today.all technologies. These projections were on the The fan diagram appears prominently in projec-low side for the solar technologies and biomass tions of generation for two out of five of thecompared with actual outcomes, and fairly accur- renewable technologies we surveyed. Early pro-ate for geothermal. jections for solar photovoltaics and solar thermal

proved to be too high, and they were reduced over3.7.2. Costs. Overall, there is little systematic time by dramatic margins. It is clear that these

difference among the sponsors and authors of the renewable technologies failed to meet expecta-studies with respect to projections of costs. tions with respect to market penetration.

For wind, EPRI’s projections have been lower A modest fan diagram appears in two otherthan others, and consequently the most accurate. cases: projections of generation from the 1970s

For solar photovoltaics, NGOs were relatively were too high by significant amounts for wind,optimistic regarding trends in cost of electricity and early projections for geothermal proved to becompared with those offered by other groups, and too high by modest amounts. In both cases thethey are the only ones to have erred on the technologies have come close to meeting revisedoptimistic (low-cost) side. Independent research projections from the 1980s and 1990s.organizations also erred on the optimistic side The exception to this pattern is biomass appli-with respect to capital costs for solar photovol- cations, for which market penetration has ex-taics. ceeded previous projections.

All groups that conducted studies offered simi- However, a different picture emerges in thelar estimates for solar thermal, and these proved assessment of projections for the cost of renew-to be accurate or slightly pessimistic (high-cost) able technologies. In every case, successive gen-compared with actual outcomes. The different erations of cost projections have either agreedtypes of studies converged in their projections for with previous projections, or have declined rela-1985 and beyond for the costs of wind and solar tive to them. More important, in virtually everytechnologies. case the path of actual cost has equaled or been

24In the case of geothermal, NGO projections of below the projections for that period in time.cost were the lowest but nonetheless the closest to The story is reversed in our evaluation ofactual costs. Research groups have projected the projections of conventional technologies. Ex-highest costs and they are the only group to pectations generally were accurate with respect toproject increasing costs. generation from conventional technologies, and

In the case of biomass, relatively few studies the cost projections for conventional generationwere included. Government estimates of COE were unambiguously overestimated. Hence, in thewere high compared with what has been realized. case of conventional technologies, forecasts of

generation are accurate while successive revisionsof cost create the familiar fan diagram. The

4. DISCUSSIONappearance of the fan diagram in the projections

The fan diagrams that emerge in the figures of the cost of conventional generation has threefrom successive revisions in forecasts is a well- implications worth noting.known image in energy economics. It became First, projections of generation and cost, con-familiar through revisions to projections for oilprices from the 1970s to the 1990s, as well as

24The only exception appears in the case of capital costs forthrough revisions to forecast electricity and over-photovoltaics, where expectations from the 1970s andall energy demand growth in the late 1960s to the1980s were well below realized costs in the 1980s and

1970s. These projections often seemed to extrapo- 1990s. However, even in this case projections of the COElate historic rates of demand growth out into the for photovoltaics were roughly accurate in comparisonfuture, thus failing to anticipate the dampening with realized costs.


sidered in tandem, are not necessarily more renewable technologies. The declining price ofaccurate for conventional generation than for conventional generation constituted a movingrenewable generation. Since quantity and price are baseline against which renewable technologiesrelated, it is not clear that the projections of sales had to compete. Energy policy initiatives includ-for conventional generation are as accurate as ing PURPA and the deregulation of natural gas,indicated in Fig. 11. Either the sales projections oil pipelines, and rail industries complementedare built on an assumption of extremely price- technological and economic trends that directlyinelastic (i.e. not sensitive to incremental changes affected conventional technologies. Collectivelyin price) demand curves, or highly income-inelas- these regulatory, technological, and market struc-tic (i.e. not sensitive to changes in income) ture changes served to reduce generation costs for

26demand curves, or both, or else there is an conventional technologies. To a significant de-element of luck in their success. Had the projec- gree, renewable technologies were an innocenttions with respect to price been accurate, sales victim of these other successes.would likely have been less than were actuallyachieved, according to most estimates of elasticity

5. CONCLUSIONof demand. Were the errors in forecasting retailprices taken into account, a fan diagram con- This evaluation has a strong bearing on theceivably would appear in the demand projections. justification for continued public-sector supportThat is, if forecast prices had turned out to be for the development of renewable technologies.accurate, forecasts of electricity demand would Prior and sometimes faltering support has taken ahave been too high compared with actual demand, variety of forms, from research funding to sub-continuing the trend from the 1970s. sidies for investment and production. The public

Second, the rate of technological change might may well ask what the return has been on thesebe expected to be far greater for an emerging investments. We do not provide evidence ortechnology than for mature technology. However, attempt to evaluate the efficacy of public-sectorit is important to realize that such change con- programs as such, nor can we attribute thetinues for mature technologies. Technological changes in technological development to govern-progress has contributed to the decline in price ment policies in general. The changes we observeexperienced by all forms of generation (Bradley, may or may not be attributable to public-sector1997). Table 3 reports a 44% decrease in the cost programs. But the claim that renewable tech-of generation between 1983 and 1995. Changing nologies have failed to meet goals that influencedfuel prices are the dominant factor in explaining the nature of public-sector programs can bethis decline but not the only factor. In a recent addressed directly.econometric study, Carlson et al. find significant A critical question is how to measure thetechnological improvement and reductions in the success of renewable technologies. We proposecost of production (holding input prices fixed) at two such measures. One is their performancecoal-fired power plants between 1985 and 1994 relative to projections for their contribution to

25(Carlson et al., 1998). Indeed, the rate of meeting electricity demand. A second is theirimprovement in relatively mature conventional performance relative to projections of cost. Re-technologies may accelerate in the increasingly garding the first, renewable technologies seem tocompetitive environment of wholesale and retail have failed in general, at least in terms of theircompetition in the electricity industry that was contribution to grid-connected electricity genera-launched by another public policy initiative — the tion, with perhaps a couple of exceptions (notably

271992 Energy Policy Act. biomass). With respect to the second measure,Third, the advantageous changes in convention- renewables seem to be a success.

al generation created a difficult obstacle for The development of renewable technologiesobviously has not occurred while the world

25Table 3 indicates capital costs have declined while O/Mcosts have risen over this timeframe. Part of this changecould be realized by full depreciation of existing facilities

26ending capital charges accruing at those facilities. Al- Renewable technologies benefited indirectly from thesethough O/M costs would be expected to increase as trends, because construction costs through the entirefacilities age, this is offset by technological change mani- industry were reduced.

27fest through the advent of electronic controls in power As noted previously, in some niche markets apart from theplant management, as well as changing work relationships electricity grid, renewable technologies have achievedand articulation of work rules on site (Ellerman, 1998). significant penetration.


otherwise stood still. In fact, over the time REFERENCEShorizon we evaluated, dramatic changes have Bradley R. L. (1997) Renewable energy: Not cheap, not green.characterized energy markets in general. The Policy Analysis, vol. 280, Cato Institute, Washington, DC.

Carlson C., Burtraw D., Cropper M. and Palmer K. (1998)ultimate impacts of these changes have beenSO control by electric utilities: What are the gains from2mostly favorable for electricity consumers; they trade? Disc. Paper 98-44, Resources for the Future,

have also been frustratingly disappointing for the Washington, DC.Cohen B., Peabody G., Rodekohr M. and Shaw S. (1995) Afate of renewable technologies. The price of

history of midterm energy projections: A review of thegeneration from conventional technologies has annual energy outlook projections. Unpubl. Mimeo, Energyfallen precipitously, largely due to changes in fuel Information Administration (June).

Deudney D. and Flavin C. (1983). Renewable Energy: Themarkets. This has raised the hurdle over whichPower to Choose, World Watch Institute, W.W. Norton, Newrenewable technologies must pass in order to York.

penetrate the electricity market in a significant EPRI (Electric Power Research Institute) (1977) Fuel andEnergy Price Forecasts: Quantities and Long Term Margi-way.nal Prices, EPRI, Palo Alto, CA.We view the confluence of changes in energy EPRI (Electric Power Research Institute) (1982) 1980 Survey

markets as an indication of the success of an array and Evaluation of Utility Conservation, Load Management,and Solar End-Use Projects, vol. 2, Solar End-Use Projects,of changes in public policy, including, in par-EM-2193, Research Project 1940-1.ticular, regulatory reform and deregulation of key EPRI (Electric Power Research Institute)–USDOE (US De-

energy markets and related industries such as partment of Energy) (1997) Renewable Energy TechnologyCharacteristics, US Department of Energy, Washington,railroads. Ironically, the perceived failure of re-DC.newable technologies to contribute importantly to Ellerman D. (1998) Note on the seemingly indefinite extension

electricity generation today is to a significant of power plant lives. A panel contribution. Energy J. 19,129–132.degree attributable to these successes.

EIA (Energy Information Administration) (1980) AnnualThis does not imply that public-sector support Report to Congress 1979, DOE /EIA-0173(79) /3, vol. 3,for renewable technologies has been misplaced. July, Energy Information Administration, US Department of

Energy, Washington, DC.Public-sector financial incentives for renewableEIA (Energy Information Administration) (1990) Renewabletechnologies (as well as other energy tech- Energy Excursion: Supporting Analysis for the National

nologies) can be viewed as a precaution against Energy Strategy, Energy Information Administration, Officeof Coal, Nuclear, Electric and Alternate Fuels, US Depart-rising energy prices, vulnerability to disruptionsment of Energy, Washington, DC.of foreign energy supply, and environmental EIA (Energy Information Administration) (1993) Renewable

concerns. After all, fire insurance is not judged as Resources in the US Electric Supply, DOE /EIA-0561,Energy Information Administration, US Department ofsuccessful from the perspective of whether or notEnergy, Washington, DC.a person’s house actually burns down. Similarly, EIA (Energy Information Administration) (1997) Annual

the success of public-sector support for renewable Energy Outlook 1998, Energy Information Administration,US Department of Energy, Washington, DC.technologies should not be viewed from the

EIA (Energy Information Administration), Holte S. and Reiserperspective of largely unrelated outcomes. E. (1998a) Annual Energy Outlook Forecast Evaluation,The most important measure of success would Energy Information Administration, US Department of

Energy, Washington, DC.seem to us to be the cost of electricity generatedEIA (Energy Information Administration) (1998b) Annualfrom renewable technologies compared with the Energy Review 1997, July, Energy Information Administra-

expectations that served as the justification for tion, US Department of Energy, Washington, DC.ERDA (Energy Research and Development Administration)public-sector support. According to this measure,

(1975) A National Plan for Energy Research, Developmentrenewable technologies have met the goals set for and Demonstration: Creating Energy Choices for thethem, and could be considered an important Future, ERDA-48, vol. 1.

FEA (Federal Energy Administration) (1974a) Project In-component of an ongoing movement towarddependence, Federal Energy Administration Project In-sound energy policy. Whether the level of support dependence Blueprint, Final Task Report: Geothermal

has been adequate or should continue we leave to Energy, November, National Science Foundation.FEA (Federal Energy Administration) (1974b) Federal Energyother investigators.

Administration Project Independence Blueprint, Final TaskForce Report: Solar Energy, National Science Foundation,Washington, DC.

FEA (Federal Energy Administration) (1976) National EnergyAcknowledgements—The Renewable Energy Policy Project Outlook, A-N-75 /713, Federal Energy Administration,supported the research reported in this study (http: / /www.rep- Washington, DC.p.org). The authors are grateful to Joe Cook and Martin Flavin C. and Lessen N. (1990). Beyond the Petroleum Age:Heintzelman for assistance. Individuals too numerous to list Designing a Solar Economy, Worldwatch Paper 100,have contributed their time and perspective to help guide the Worldwatch Institute, Washington, DC.study. However, responsibility for errors and omissions remain Gipe P. (1995). Wind Energy Comes of Age, John Wiley, Newwith the authors. York.


Lovins A. B. (1977). Soft Energy Paths: Toward a Durable OTA (Office of Technology Assessment) (1995) RenewingPeace, Harper & Row, New York. our Energy Future, OTA-ETI-614, US Congress, OTA,

Management Information Services Inc. (MISI) (1998) Federal Washington, DC.Incentives for the Energy Industries, Washington, DC. PCAST (The President’s Committee of Advisors on Science

NAS-CONAES (National Academy of Sciences, Committee and Technology) (1997) Federal Energy Research andon Nuclear and Alternative Energy Systems) (1979a) Geoth- Development for the Challenges of the Twenty-First Cen-ermal Resources and Technology in the United States: tury, Washington, DC.Supporting Paper 4, Study of Nuclear and Alternative Regulatory Assistance Project (1998) Making Room for Re-Energy Systems, National Academy of Sciences, Washing- newables, Monday, June 1; http: / /www.rapmaine.org /ton, DC. room.html

NAS-CONAES (National Academy of Sciences, Committee SERI (Solar Energy Research Institute) (1981) A New Pros-on Nuclear and Alternative Energy Systems) (1979b) perity: Building a Sustainable Energy Future, BrickhouseEnergy in Transition 1985 –2010, National Research Coun- Publishing, Andover, MA.cil, National Academy of Sciences, Washington, DC. Stanford (1979) Alternative Energy Futures: An Assessment of

NAS-CONAES (National Academy of Sciences, Committee US Options to 2025, Institute for Energy Studies, Stanfordon Nuclear and Alternative Energy Systems) (1979c) University, Stanford, CA.Domestic Potential of Solar and Other Renewable Energy Stoughbaugh R. and Yergin D. (1979). Energy Future: ReportSources: Supporting Paper 6, Study of Nuclear and Alter- of the Energy Project at the Harvard Business School,native Energy Systems, National Academy of Sciences, Random House, New York.Washington, DC. USDOE (US Department of Energy) (1982) Energy Projec-

OTA (Office of Technology Assessment) (1978) Application tions to the Year 2000: July 1982 Update, DOE /PE-0029/of Solar Technology to Today’s Energy Needs, vol. 1, US 1, US Department of Energy, Washington, DC.Congress, OTA, Washington, DC. USDOE (US Department of Energy) (1983) Energy Projec-

OTA (Office of Technology Assessment) (1985) New Electric tions to the Year 2010: A Technical Report in Support of thePower Technologies: Problems and Prospects for the National Energy Policy Plan, DOE /PE-0029/2, US Depart-1990’s, OTA-E-246, July, US Congress, OTA, Washington, ment of Energy, Office of Policy, Planning, and Analysis,DC. Washington, DC.

winner, loser, or innocent victim? has renewable energy performed as expected?

Documents