Incentives for Small PV

Transcription

1 in France Anna Creti 1 MarÌa Eugenia Sanin 2 1 Univ. Dauphine and Ecole Polytechnique 2 EPEE, UniversitÈ d Evry Val d Essonne and Ecole Polytechnique

2 Questions and modelling strategy Do we really need Feed In Tari s (FIT) for French residential consumers? Comparison with electricity price: grid parity? Is the trend in installation prices exogenous, and if not, which is the level of learning achieved in the small PV market? Which is the optimal path for installed capacity to reach a predetermined target for 2030? What we do: 1 We estimate Photovoltaic (PV) demand by linking proötability of investment with annual installed capacity. panel of small PV installations across France 2 We simulate the optimal path and Önd the FIT that induces such path.

3 Motivation Technology with a fast market increase. Policy debate on various incentive measures (feed-in, R&D subsidies, tradable green certiöcates, renewable portfolio standard, net metering) coupled with the European 2030 target (27% share of renewable energy consumption) and in our case study, with the 2030 objectives of the French "Transition Energetique" (40% of energy consumption must be renewable). At the end of 2010, the moratorium froze the photovoltaic market in France by suspending subsidies and connections to grid network of ongoing projects except for small installations with power below 3 kwp. We study household PV development, particularly present in Spanish debate after Januaryís proposal in Congress of the Ley de Fomento del Autoconsumo Electrico versus "Impuesto al Sol". Projections in France

4 French context: ups and downs in Feed-in-Tari s (FiT)

5 French context (II): zoom in PV smaller than 3 kwp

6 Summary of the related literature ProÖtability of PV investment: values of investments in di erent countries in Europe, Japan, Germany in presence of incentive schemes (Dusonchet and Telaretti, 2010, Zhang and Hamori, 2011, etc...). Limits: no link with long term targets on terms of installed capacity. Deployment: di usion of innovation are often described by logistic functions or "S-curves" (the Bass model-gerowski, 2000, Guidolin and Mortarino, 2010) Limits: these models do not take into account the incentive policies such as subsidies and feed-in tari s. Learning curve: PV prices decrease as the amount of technology installed increases (seminal works: Shae er et al. 2004, Nemet, 2006; Pillai 2015, Rubin 2015) Limits: sensitivity to data, complexity of the technology and factors of cost reduction other than learning.

7 An overview of the Database (I) 180 observations spread over 22 French regions: yearly observations on residential PV installations -quantity and price both of materials and installation- generating less than 3 kwp. France has added MWp of solar capacity from residential sector during Further description

8 An overview of the Database (II): cost di erences

9 An overview of the Database (III): number of installations

10 An overview of the Database (IV): installations per year

11 Comparing the FiT with PV costs FiT (granted 20 years) for PV integrated to buildings started in 2006, in 2009 reached 60,17 euro-cents per KWh and then unevenly decreased (starting in 2011 suspended for bigger PV installations). Limited proötability overall (Fit and Price very close). Grid parity is not achieved since Price is much higher than the electricity price throughout the period.

12 Comparing the FiT with PV costs (II) Regional di erences

13 Our Regionalization

14 The Model We deöne the demand q it as the probability of adoption P it in each region (now NE, SE, NW, SW, Bretagne) times the potential market sizes M t at each point in time: q it = M t.p it where P it = exp(v it) 1 + exp(v it ) V it = NPV it + l it where N! 1 NPV it = FiT t.e it. Â k =1 (1 + d) k! p Xit it and l it = log M t where d is capital depreciation rate, N is the life length of a panel, E is the amount of electricity generation per unit of installed capacity over its life-cycle, p it is the unit price of the installation and X it is the cumulated installed capacity per region. "

15 The Model: empirical version Using logit equation, after assuming that q it /M t < 1andM t = M, and applying a log transformation gives log(q it )=NPV it, that is log(q it )=a 1. log(npv it ). + a 2 l it + a 3i + # it,

16 Estimation Results and interpretation Model (2) both the FiT and the di usion e ect remain signiöcant but the coe cient values attached to these variable change dramatically due to decrease in demand after Model (4) shows that indeed the decrease in the development of the PV market after the moratorium is di erent among regions: Bretagne is particularly a ected, followed by NW and SW.

17 Optimal FIT trajectories Assume 3 alternative expansion objectives: 1 Current growth rate until 2030: we continue the trend after 2013.(at national level adds 203 units per year, 7810 in total). 2 Doubling capacity in 2030 as compared to 2013 (256 per year nationally) 3 Constant growth rate until 2030: 5% per year (exponential expansion path with 9991 installations in total by 2030). Both in national and regional cases. For each scenario we determine a path for the level of adoption at each period until Using the adoption level determined by each scenario, we calculate the corresponding net present value per period. Then, considering learning coe cient estimated, we are able to calculate the (decreasing) price of panels and consequently the optimal FiT needed to reach the net present value calculated.

18 Is learning-by-doing explaining the price path? We estimate two alternative equations for the evolution of PV price per region and nationally (as well as a combination of them)... An exogenous time trend: p it = a 1 + a 2 t + a 3 t 2 + a 4 nörms it A standard learning curve (after log-linearization):! "!bi xit p it = p i0. where b x i = log(1! h i ) i0 log 2 where b is an elasticity coe cient, h is the learning rate. In average, coe cient is b =! and h = : the average system price is reduced by 21% each time that installed capacity doubles.

19 Current growth scenario by region Trajectories

20 Current growth scenario by region FiT

21 Current growth scenarioís interpretation For SW to follow this path, a very high FiT of nearly 1 e/kwh is required in the short run quickly decreasing to 25 ce/kwh in NE requires a FiT of 57 ce/kwh Örst, gradually reducing to 20 ce/kwh in In contrast, NW and SE regions where solar panels are the least and the most developed, respectively, this scenario allows the government to gradually lower the FiT rates from about 25 ce/kwh to about 12 ce/kwh. Bretagne shows slow growth rate in recent years so continuing in this trend only requires a low FiT of 5-6 ce/kwh.

22 Conclusion Notorious slow down in the adoption of small PV. Waves of adoption that alternate in di erent regions. Thin proötability and no grid parity: veriöed by the estimation results where the episode in 2011 has a signiöcant impact: highly dependant on incentive policies.. Very ambitious PV development objectives may need an increase in FiT but most realistic scenarios show that, thanks to learning e ects, optimal FiTs are generally decreasing. Regional market di erences should be considered to Öx optimal FiTs.

23 The End