Learning-by-doing – how to judge whether supporting solar …... Outline • The case for...

Learning-by-doing – how to judge whether supporting solar PV is justified

Economic Theories and Low-carbon Transformation Policies

David Newbery, EPRGCambridge 23rd June 2017

www.eprg.group.cam.ac.uk

Outline

• The case for subsidizing renewables– drives down unit costs to benefit future deployment– benefit is not captured by producer

• How to decide whether to subsidize e.g. solar PV • The need for collective action

⇒Global Apollo Programme 2015• Factors influencing benefits

focussolar.de≈Full hrs/yr

Green is bad, red good


Doubling the irradiance halves the cost

http://geosun.co.za/solar-maps/ Roughly 1 kWp/m2


Steady growth of PV capacity

Source:Wikipedia

Cumulative growth rate 32%Logscale


Solar PV cost fall 20% as capacity x2 German wholesale prices fall 50% in

5 yrs, 40% of which due to RES

0

10

20

30

40

50

60

70

Jan 11 Jan 12 Jan 13 Jan 14 Jan 15 Jan 16

Day Ahead Auction Power price (Euro/MWh)


The case for support

• Producing renewables creates learning spill-overs• Not captured by producer

• Learning-by-doing depends on cumulative prodn.• not output from each unit once installed

•cost reduction per doubling of cum. prod• solar 20-22% over past 40 yrs, grew 28% last year• on-shore wind 12%

•Hard to disentangle R&D and production• two-factor rates attribute less to LbD• solar PV 12%, on-shore wind 9%


Dramatic fall in solar PV prices

IRENA (2016). The Power to Change: Solar and Wind Cost Reduction Potential to 2025

Projected


PV learning rates are highecons of scale important

Module learning rate18-22%BOS cost excl inverter now 60% of total


Predicting costs

• 2015 global av. module price $580/kWp for 234 GWp cum.• only 55% of installed cost of $1,050/kWp

• NREL (2016) total unit cost utility-scale tracking unit in cheapest state $1,190kWp

• Adjust for high cost of US labour => $1,050/kWp

• ITRPV (2016) 2,000 hrs (23% CF) $44/MWh– some 20yr PPAs signed in US for $40/MWh– Europe lower 1,000 hrs (11.4% CF) $87/MWh

• Capacity value depends on coincident peak–Quite high in Ca, zero in Europe


Estimating spill-over benefits

• Cost of doubling cum prod is low at 10 GW, much higher at 200 GW => early investment valuable• But cannot instantly raise low base by high amt.

– constraints on building production capacity– limits to rate of dissemination of learning– uncertainty whether past LbD is good guide to future

• Consider modest temporary increase in investment=> has a current cost but lowers all future costs• Is it worth it - is NPV positive in terms of costs?• is program as whole NPV positive compared to fossil?• What does that depend on?


Is acceleration worthwhile?Is program worthwhile?

Fossil cost


Costs depend on investment trajectory and future LbD returns


The learning model

Assume steady growth at rate g, then unit costs at date t are

Let Kt be cumulative installed capacity at t, ct be unit PV cost

The learning rate is


Current and future costs

If It is gross investment (kWp) and kt is current stock

where

Current investment cost is

The NPV of future costs is


Measuring spill-overs

The impact on future costs of increasing current investment is

=g

The spill-over benefit as a fraction of current cost at date 0 is


Learning benefit – PV of future cost reductions

At a learning rate of 22% and growth rate g = 15% learning benefit = 42% of cost

– discount 3%, asymptotic PV unit cost 25% current cost

=> higher earlier - in limit 64% = 1/(1+r/(-bg))=> good to subsidize earlier

–But uncertainty greatest then

But is it ultimately cost effective?


Global Apollo Programme

• Learning spill-overs are global• installing PV delivers global climate change mitigation=> ideally collectively support global programme• Each member subscribes in proportion to GDP

– or more progressively? In proportion to GHG emissions?• Funds allocated competitively per kWp

– e.g. premium subsidy for 20,000 kWh/kWp

=> invest where the subsidy needed is minimized• Regional coalitions investing within region more costly

– and capture smaller fraction of benefits

California shows impact of PV at one critical node (SP-15) in May sunny hours


Impact of PV on system costs


Investment costs & benefits

Let yt be PV output = htkt where ht is full hours/yr, pt is av. net value

Let τ be effective capacity contribution, P be capacity value,

Let Γ be market price of carbon rising at I, γ be social value, rising at i

The net present cost discounting at commercial rate R is

Social benefit of investment trajectory is


Social returns

Subsidy required is

Total net social benefit is S0 - F0 better expressed as (S0 - F0)/I0

Is it worth accelerating the programme:


Assumptions

• Life of PV L = 25 years, depreciates at 0-1% p.a.• Time horizon for discounting benefits T = 25 yrs• social discount rate r = 3%• rate at which market price of CO2 rises I = 0.5%• rate at which social cost of CO2 rises i = 1.5%• Value of displaced electricity p = $35 or 60/MWh

– declining 0-1% p.a.• Capacity price (de-rated) $75/kWyr (constant)• derating for PV τ = 0,5,10%• social carbon value of PV γ = $10/MWhe• market value of carbon for PV Γ = $0, 2,10//MWhe•hrs/yr h = 1,000-2,000 (declining 0-1% p.a.)• PV growth rate g = 15-18% p.a.• Learning rate λ = 18-22%


Sensitivity of NPV to parameters


Conclusions

• Benefits maximized by choosing right places– high insolation, grid scale, tracking, high cost power

• sunny countries, coal or oil-based, ideally with storage hydro

– dispersion to avoid cannibalisation (or strong T&D)• Global benefits need global support = Apollo

– regional benefits capture only part of cost fall• Results sensitive to fossil and carbon prices, PV learning and growth rates, discount rate, insolation

Current subsidy can be justifiedAccelerating investment can be justified


References

• ITRPV, (2016). International Technology Roadmap for Photovoltaic Results Seventh Edition 2016 at http://www.itrpv.net/Reports/Downloads/2016/ •Newbery, D.M. (2017) “How to judge whether supporting solar PV is justified”, EPRG Working Paper 17NN, at http://www.eprg.group.cam.ac.uk/eprg-working-papers/• NREL, (2016). U.S. Solar Photovoltaic System Cost Benchmark: Q1 2016, by Fu et al., National Renewable Energy Laboratory• King, D., J. Browne, R. Layard, G. O’Donnell, M. Rees, N. Stern & A. Turner (2015). A global Apollo programme to combat climate change, LSE, at cep.lse.ac.uk/pubs/download/special/Global_Apollo_Programme_Report.pdf

http://www.eprg.group.cam.ac.uk/eprg-working-papers/

Learning-by-doing – how to judge whether supporting solar …... Outline • The case for...

Documents

Transcript of Learning-by-doing – how to judge whether supporting solar …... Outline • The case for...