June 17-18, 2003 in Washington, D.C.

Transcription

1 Learning-by-doing and the Photovoltaics case Resources for the Future workshop on Learning-by-Doing in Energy Technologies June 17-18, 2003 in Washington, D.C. Richard D. Duke

2 PV technology and markets The PV experience curve Benefit-cost analysis of buydowns Buydown implementation strategies Prioritizing scarce buydown funds APPENDIX: Buydown economics

3 CRYSTALLINE STILL DOMINATES THE 450 MWp PV MARKET IN 2002 multicrystalline 46% thin-film 7.9% ribbon crystalline 3.7% monocrystalline 43% Source: Strategies Unlimited (2003). Note that Heterojunction with Intrinsic Thin layer (HIT) modules, produced by Sanyo, are composed of a monocrystalline cell surrounded by thin amorphous-silicon layers. Total shipments for these modules are split evenly between thin-film and monocrystalline categories. HIT modules offer the highest solar conversion efficiency available for any terrestrial PV technology (19.5% cell efficiency).

4 SUBSIDIZED GRID-CONNECTED MARKETS ARE BECOMING DOMINANT Other subsidized grid plus off-grid commercial sales Subsidized residential PV in Germany Subsidized residential grid PV in Japan 300 MWp Source: Berger (2001); Weiss and Sprau (2001); Hirschman and Takano (2003); Krampitz and Schmela (2003)

5 JAPANESE RESIDENTIAL PROGRAM HAS DRIVEN SYSTEM COSTS DOWN $ US$ per Wp $30 $25 $20 $15 $10 Installation Balance of systems Modules $5 $0.69 $1.48 $3.97 $ Source: Kurokawa and Ikki (2001); NEF (2001); Hirshman and Takano (2003)

6 GERMAN RESIDENTIAL PV PROGRAM HAS HAD MIXED PRICE RESULTS Modules Balance of systems Installation $ US$ per Wp $7 $6 $5 $4 $3 $2 $0.69 $0.55 $1.43 $1.30 $4.75 $5.10 $0.54 $1.26 $5.63 $0.40 $1.08 $4.81 $1 $ Source: Krampitz and Schmela (2003)

7 PV technology and markets The PV experience curve Benefit-cost analysis of buydowns Buydown implementation strategies Prioritizing scarce buydown funds APPENDIX: Buydown economics

8 REMARKABLY CONSISTENT ALL-PV EXPERIENCE CURVE $ Wholesale price (2002$/Wp) $10 PR=0.80 R2 = $ ,000 10,000 cumulative PV production (GWp) Source: Johnson (2002) and Dunay (2003)

9 SPURIOUS MICROSTRUCTURE IN PV EXPERIENCE CURVE 100 Nitsch PR=.84 Price (USD(2000)/Wp) 10 Harmon (PR=.80) Nitsch PR=.80 Nitsch PR=.53 Nitsch PR=.79 Strategies Unlimited (PR=.80) Cumulative Sales (MW)

10 LEARNING IS THE PRIMARY DRIVER FOR THE PV EXPERIENCE CURVE Input prices may vary, but: Random price shocks do not impact long-term buydown economics Buydowns will generally drive long-term input prices lower (e.g. new dedicated PV-grade silicon factories) Scale economies are crucial, but: Learning-by-doing is integral to the process of scaling up manufacturing facilities and markets Available econometric evidence suggests learning effects dominate scale impacts for PV Supply-push research and development efforts matter, but: Learning-by-doing becomes increasingly important as products move from the lab to fullscale deployment Private research and development is often funded in proportion to sales levels Econometric efforts to distinguish supply-push from learning effects remain unsatisfactory (e.g. adding time as an additional independent variable does not substantially improve the model fit)

11 PV technology and markets The PV experience curve Benefit-cost analysis of buydowns Buydown implementation strategies Prioritizing scarce buydown funds APPENDIX: Buydown economics

12 THE CONVENTIONAL BREAK-EVEN METHOD HAS SERIOUS FLAWS Does not account for sales in niche markets under the no-subsidy scenario Does not allow estimation of the optimal sales path under buydown

13 THE OPTIMAL PATH METHOD Faster P w/ buydown Computationally determine the optimal buydown subsidy/output path to maximize NPV P t derived from experience curve given cumulative production at time t Demand schedule derived from empirical estimation Advantages estimation of transfer subsidies realistic modeling of sales growth under the NSS

14 OUTWARD SHIFTS IN PV DEMAND UNDER THE OPTIMAL PATH METHOD break-even value of modules ($/Wp) $5.00 $4.50 $4.00 $3.50 $3.00 $2.50 $2.00 $1.50 $1.00 $0.50 $0.00 X(50)=(25/(1+exp(-0.09*50+2))*P^-2 X(10)=(25/(1+exp(-0.09*10+2))*P^-2 X(1)=(25/(1+exp(-0.09*2+2))*P^ GWp per year This figure compares the logistically shifting all-market PV demand schedule used in the optimal path method analysis with the annual OECD residential PV demand schedule (i.e. the breakeven schedule for PV demand in new U.S. homes increased by roughly a factor of 4 to account for retrofits as well as markets in Europe and Japan). In the first year of the analysis (2003) annual demand falls short of the breakeven schedule, but the two schedules overlap by year 10 after markets have matured. After 50 years, demand shifts out by another factor of four due to growth in commercial buildings and developing country markets. At the price floor of $0.50/Wp, this yields a mature sales rate of 100 GWp/y. Even in the first year, the all-market demand schedule exceeds the OECD residential PV schedule for sales levels below 1 GWp/y because it includes off-grid PV markets for which the willingness to pay is much greater than for distributed grid-connected PV. Similarly, the all-market demand schedule extends beyond 10 GWp/y (at which point the OECD residential PV demand schedule tapers off) because other markets open up at these low prices including commercial buildings and a full range of grid-connected applications in developing countries.

15 SNAPSHOT OF THE FIRST YEAR OF AN OPTIMAL PV BUYDOWN $10 $9 $8 quantity demanded w/o year-one subsidy Year One quantity demanded w/ year-one subsidy $7 $6 $/Wp $5 consumer surplus current price $4 $3 $2 free riders failure to price discriminate minimum possible subsidy cost TMC $1 $ GWp/y True marginal cost (TMC) exceeds the $0.50/Wp price floor since r > 0 Transfer subsidies are modest

16 SNAPSHOT OF THE TWENTIETH YEAR OF AN OPTIMAL PV BUYDOWN $2.0 $1.5 consumer surplus Year 20 quantity demanded w/o year-20 subsidy quantity demanded w/ year-20 subsidy current price $/Wp $1.0 free riders failure to price discriminate minimum possible subsidy cost TMC $0.5 $ GWp/y TMC approaching the $0.50/Wp price floor Transfer subsidies are potentially more severe

17 APPLYING THE OPTIMAL PATH METHOD TO GLOBAL PV MARKETS Optimal path requires tripling current PV subsidies and sustaining support for 43 years $ billions per year (r=0.05) $4 $3 $2 $1 $ buydown year minimum subsidies + transfer subsidies minimum subsidies $235 buydown benefits - $144 NSS benefits - $37 minimum possible cost $54 billion NPV (r = 0.5) PV provides 5% of OECD electricity by 2030 vs. <1% under NSS GWp per year buydown year buydow n NSS Excludes environmental benefits $5 $4 Annual cost < 0.5% of OECD electricity expenditures $/Wp $3 $2 $1 $ buydown year buydown price buydown net price NSS price

18 SENSITIVITY OF THE NPV ESTIMATES FOR AN OPTIMAL PV BUYDOWN Transfer subsidies valuation is controversial NPV is $54 billion with marginal excess burden estimate of zero NPV drops to $44 billion assuming MEB = $0.33 but no transfer subsidies NPV drops to $27 billion assuming MEB = $0.33 and government has no ability to reduce transfer subsidies by 1) excluding free riders or using price discrimination when doling out subsidies; 2) re-optimizing the buydown to account for the MEB of transfer subsidies Minimizing transfer subsidies is therefore crucial if MEB > 0 and/or buydown funding is constrained by politics rather than benefit-cost criteria Progress ratio and discount rate assumptions are also crucial

19 PV BUYDOWNS MITIGATE MULTIPLE MARKET FAILURES Current prices do not reflect the full value of PV electricity A $120/tC tax would increase the cost of natural gas electricity by $0.01/kWh and coal electricity by $0.03/kWh Local air pollution externalities (primarily health impacts from particulate emissions) range from about $0.02/kWh in California (natural gas and hydro) to $0.13/kWh for coal-fired Midwestern electricity PV electricity reduces peak demand, alleviates strain on the transmission and distribution system, and reduces price volatility Demand-side market failures from cognitive biases/limitations Consumers tend to under-invest in highly cost-effective energy efficiency technologies possibly due to the high cost of processing information about the available alternatives Risk-averse homeowners/businesses may worry about individual system performance variability Supply-side market failures from spillover of learning-by-doing Manufacturing spillovers System spillovers

20 PV technology and markets The PV experience curve Benefit-cost analysis of buydowns Buydown implementation strategies Prioritizing scarce buydown funds APPENDIX: Buydown economics

21 PROMISING BUYDOWN IMPLEMENTATION STRATEGIES Loosely coordinated regional buydowns Reduce transfer subsidies by allowing better market segmentation to reduce free riders and allow price discrimination when distributing subsidies Bypass the collective action problem Reduce the risk to manufacturers of a global market slow down Foster learning-by-buying down Quantity mandates (e.g. renewable portfolio standard with a PV tranche) Reduce information burden Fair-share of global PV buydown Subsidy cap contains risk to obligated parties Well-suited to long-term optimal buydown

22 PV technology and markets The PV experience curve Benefit-cost analysis of buydowns Buydown implementation strategies Prioritizing scarce buydown funds APPENDIX: Buydown economics

23 RATIONALE FOR LIMITING SUPPORT TO THE CLEAN ENERGY SECTOR Non-learning public benefits Slow diffusion rates Severe system spillovers Strong experience curve effects in some important cases Benefits are easy to estimate based on the market price of incumbent substitutes Energy is a commodity with thin profit margins making forwardpricing particularly difficult

24 CRITERIA FOR SELECTING SPECIFIC CLEAN ENERGY TECHNOLOGIES Technology picking is essential for buydowns Buydowns are too expensive to permit a shotgun approach There is relatively good information available at the deployment stage when buydowns are relevant Governments should restrict buydowns to clean energy technologies characterized by: 1) Strong non-learning public benefits 2) A competitive (high spillover) industry 3) Strong experience curves and low expected price floors 4) Slow current sales relative to a large long-term market potential 5) The technology is at least as promising as foreseeable substitutes

25 PV IS AT LEAST AS PROMISING AS FORESEEABLE SUBSTITUTES Low risk from incumbent or emerging substitutes for PV EIA predicts nearly flat U.S. residential electricity rates through 2020 Fuel cells not a major threat Unlikely to compete well in individual residences Struggling to get past demonstration phase Fuel cell vehicles plugged in at work could reduce value of commercial PV but concept unproven and residential markets are the focus of this analysis Advanced energy efficiency would still leave most homes needing 4 kwp PV systems Other renewables not a serious threat for the distributed residential market The optimal path method would need to be modified to consider buydown candidates with overlapping markets (e.g. advanced biomass and wind electricity) Must analyze both/all simultaneously Demand for each defined as total market potential minus sales of the other Conduct joint optimization to determine simultaneously the buydown subsidy/output path for each that maximizes total NPV Increasing returns to adoption likely to suggest buying down one or the other If markets only partially overlap or government wants to diversify buydown performance risk then simultaneous buydowns could be optimal

26 PV technology and markets The PV experience curve Benefit-cost analysis of buydowns Buydown implementation strategies Prioritizing scarce buydown funds APPENDIX: Buydown economics

27 THE TECHNOLOGY POLICY TRIAD

28 MYOPIC MONOPOLIST CASE [firm s r = infinity, social r = 0, zero spillover] Learning/experience Experience curve curve True Marginal Cost True marginal cost (TMC) = cost floor (since must produce initial high cost units sooner or later) Current profit shown by the lightly-shaded box Welfare loss equals the darkly-shaded triangle (TMC would be higher and the welfare loss would be lower if the social discount rate exceeded zero)

29 FORWARD-PRICING MONOPOLIST CASE [firm s r = 0, social r = 0, zero spillover] Learning/experience Experience curve curve True Marginal Cost Forward-pricing monopolist suffers current loss (light shading) but maximizes profit over technology lifecycle by reducing costs more quickly along its experience curve Output closer to optimum than with myopic monopolist, but monopoly welfare losses remain

30 ECONOMIC RATIONALE FOR BUYDOWNS [Perfect spillover/competition, r = 0] Learning/experience Experience curve curve True Marginal Cost Perfect spillover implies: Perfect competition is possible Price = unit cost (including a fixed competitive profit margin) There is no welfare loss from market power but there is a welfare loss from the failure of firm s to forward-price High spillover is common and buydowns can eliminate the associated welfare losses

31 MARGINAL EXCESS BURDEN OF TAXATION (MEB) Traditional literature [Pigou (1928); Harberger (1964); Browning (1976)] Social costs of public spending include welfare loss from distortionary taxes MEB estimates range as high as $1.65, i.e. projects need BCR > 2.65 (Feldstein, 1997) Parry (1999) surveys literature and provides central estimate of MEB = $0.33 Recent revolution Kaplow (1996) shows MEB = 0 a better default estimate MEB exactly zero if finance a public good with income tax adjustments that offset the benefits of the public good (e.g. progressive taxes used to finance environmental improvements that the rich value more than the poor) MEB > 0 implies taxes and associated spending reduce regressivity and vice-versa Must consider the combined social value of the efficiency and distributional impacts Kaplow (1998) offers withering response to critique by Browning and Liu (1998) Ng (2000) argues MEB < 0 in some cases, e.g. due to benefits of pollution taxes Slemrod and Yitzhaki (2001) agrees that both distributional and distortionary impacts are important and there is a tendency for more progressive tax schemes to be more distorting but points out that there are exceptions