energy’innovation and public · gregory nemet— energy innovation and public policy 1970 1980...

Challenges in Energy Innovation and Public Policy

Prof. Gregory Nemet March 2016

Gregory Nemet— energy innovation and public policy

Characteristics of innovation:surprise and stationarity

Emergent properties• Ex ante ignorance• Skewed outcomes• Pervasive spillovers• Combinatorial• Depreciating knowledge• Interaction w/ production

Drivers of smoothness• long lifetimes• risk aversion• incremental improvement

• aggregation

2


1. Policy and Innovation:Technology push and demand pull

Technology Push

Demand Pull

invention innovation diffusion

Source: Nemet, G. F. (2009). "Demand-pull, technology-push, and government-led incentives for non-incremental technical change." Research Policy 38(5): 700-709.

3

Market failures:Knowledge spillovers

Asymmetric informationRisk aversion


1. Policy and Innovation:invention innovation diffusion

Source: Nemet, G. F. (2009). "Demand-pull, technology-push, and government-led incentives for non-incremental technical change." Research Policy 38(5): 700-709.

4

TechnologyPush

Demand Pull

Reduces the cost of innovation

Increases payoffs for success

+ knowledge + size of market

• R&D• tax credits• education• Demonstrations• knowledge networks

• IPR• price externalities• subsidize demand• govt. procurement• tech. standards

GOVT GOALFOR PRIVATE

ACTORS


Two Policy Challenges

• Fragile demand pull

• Between technology push and demand pull

5


• Investment depends on expectations• Low-C: expectations about D-Pull policy• What happens to investment if expectations about policy are uncertain?

Do we need additional measures?

2. Fragile demand pull: governmentcommitments are not fully credible

6


1970 1980 1990 2000 2010 20200

5

10

15

milli

on b

bl/d

ay

ActualForecast (1973)Nixon Target (1974)Carter Target (1979)Forecast (2010)Obama target (2011)

Source: Nemet, G. F., P. Braden, E. Cubero and B. Rimal (2014). "Four decades of multiyear targets in energy policy: aspirations or credible commitments?" Wiley Interdisciplinary Reviews: Energy and Environment 3(5): 522-533.

2. Fragile demand pull: Evidence that commitments are not fully credible

U.S. Oil import targets

7


Source: Nemet, G. F., P. Braden, E. Cubero and B. Rimal (2014). "Four decades of multiyear targets in energy policy: aspirations or credible commitments?" Wiley Interdisciplinary Reviews: Energy and Environment 3(5): 522-533.


8

Advanced Review wires.wiley.com/wene

1970 1980 1990 2000 2010 20200

5

10

15

Mill

ion

bbl/d

ay

Oil imports

Actual

Forecast (1973)

Nixon (1974)

Carter (1979)

Forecast (2010)

Obama (2011)

1970 1980 1990 2000 2010 20200

10

20

30

40

Mile

s/ga

llon

Auto fuel efficiency

Historical

CAFE (1974)

CAFE extension

Actual

CAFE (2010)

1980 1990 2000 2010 20200

0.5

1

1.5

2

Mill

ion

bbl/d

ay

Biofuels

Historical

EPACT (2005)

EISA (2007)

Actual

1990 2000 2010 20200

0.1

0.2

0.3

0.4

Ren

ewab

le e

lec.

/all

elec

.

California renewables

Historical

RPS (2002)

RPS (2006)

RPS (2009)

Actual

1990 2000 2010 20200

0.1

0.2

0.3

0.4

Ren

ewab

le e

lec.

/all

elec

.

Germany renewables

Historical

EEG (2000)

EEG (2009)

Actual

1980 1990 2000 20100

2

4

6

8

10

12

Mill

ion

tons

/yea

r

SO2

Historical, ph. I unitsCAAA ph. IActual ph. I unitsCAAA ph. IIActual ph. I+II units

(a) (b)

(c) (d)

(e) (f)

FIGURE 2 | Comparisons of historical data, targets, and actual performance after target was set.

in models 4 and 6 using linear regressions as the ratesare continuous.

In general, results here are weaker than in theestimates of met. But the hypothesis of decliningambition does seem to hold; start year is negativeand significant in every case except model 2. Thediscretionary variable is negative and significant inall regressions, while binding is positive and onlysignificant when the dependent variable is Ambitionv1. This result does not support the notion that lackof enforcement allows more ambitious policies. Bothtiming variables are negative, though start year issignificant in nearly all regressions versus durationwhich is significant in only regression 4. Once again,geography was not a significant predictor of ambitionand the sign of its coefficient switches from negativeto positive. The other explanatory variables are notconsistently significant.

TABLE 4 Binding and Attainment of Target

Nonbinding Binding Total

Not met 5 1 6

Met in target year 0 10 10

Met all years 2 11 13

Met latest year 10 11 21

Not met in any year 4 3 7

Not yet 1 5 6

Total 22 41 63

DISCUSSIONIn this review of 63 past energy policy targets wefind that targets were met in 64–77% of cases. Thelower number uses a strict coding method and the

528 © 2014 John Wiley & Sons, Ltd. Volume 3, September/October 2014

WIREs Energy and Environment Four decades of multiyear targets in energy policy

1970 1975 1980 1985 1990 1995 2000 2005 2010 20150

0.1

0.2

0.3

0.4

>50%

Year implemented

Rat

e of

cha

nge

requ

ired

Fed.RPSEERSIntl.

FIGURE 1 | Ambition measured as absolute value of the annual rate of change required by each target, by year program began. Line is linear fitto all targets.

outcomes (Figure 2). We group the targets into fivecategories so that each target within a category can becompared using the same indicator. Where prominentex ante forecasts are available, we compare themagainst outcomes. For example, Figure 2(a) showsthree targets for oil imports, set in 1974, 1979, and2011. This time series shows the failure to meet thefirst two targets as well as the declining ambition ofthe targets over time. The slopes of the dashed targetlines represent ambition rate. We show similar figuresfor targets for fuel efficiency of cars (b), biofuels (c),SO2 (d), renewable electricity (e, f).

We answer question 1—Have long-term energypolicy targets achieved their goals?—with summariesof the Met variables. Using Metv1, of the 63 targets,44 attained their goal, 13 did not, and 6 results weredropped (Table 4). In Metv2, of the 63 targets, 35attained their goal, 12 did not, and 27 results weredropped. The short answer is thus: 64% of the timeusing the strict definition of Met and 77% of the timeusing the definition that credits targets for being ontrack to their goals.

Predictors of Target AttainmentWe regress the met variables on the target character-istics. We use probit models due to binary dependentvariables and use robust standard errors to calculatep-values. Table 5 shows six specifications, with oneas our base case. Models 2 and 3 use alternate defi-nitions of ambition; model 4 adds a dummy for stateRPS; model 5 adds a dummy for revision; and model6 uses the stricter definition of met. Note that thenames in the second row of Table 5 provide indica-tions of which variables are being adjusted in eachspecification. Further, we assess whether these results

are robust to various definitions of ambition by run-ning regressions that substitute in all 21 ambitionvariables.

A first observation is that the binding variable issignificant and positive in all six models. This result isrobust across all 21 definitions of ambition. A relatedobservation is that discretionary is always negative,although not consistently significant. Second, durationis positive and significant in almost every model. Itis also robust across multiple definitions of ambition.Third, the ambition variables are always negative butonly in some cases significant. In addition, startyear ispositive but insignificant in all but one case. RPS andrevision are insignificant. The SI includes a covariancematrix and tests of collinearity for these independentvariables. There are no strong correlations and theaggregate variance inflation factor is 1.3, well belowthe level of concern.

In order to test the robustness of binding, wecompare the coefficients and p-values of bindingand discretionary across all 21 ambition variables.The robustness results are given in Figure 3. Foradditional robustness, we run these regressions usinga logit rather than a probit and find very smalldifferences in sizes of effects and none in significance.

Predictors of Target AmbitionWe also identify predictors of ambition—with particu-lar interest in whether the apparent decline in Figure 1is robust to the inclusion of omitted variables. Table 6shows the results for six models. Models 1–4 use vary-ing definitions of ambition for the dependent variable.Model 5 adds RPS to model 1 and model 6 adds RPSto model 4. We regress ambition indices (models 1–3and 5) on controls using negative binomial regressionsas the indices are counts. We regress the ambition rates

Volume 3, September/October 2014 © 2014 John Wiley & Sons, Ltd. 527


Koch, N., G. Grosjean, S. Fuss and O. Edenhofer (2015). "Politics Matters: Regulatory Events as Catalysts for Price Formation Under Cap-and-Trade." Available at SSRN.


9

2011 2012 2013 2014

EUA

year

-ahe

ad D

ecem

ber f

utur

es p

rice

(EUR

/tCO

2)

4

6

8

10

12

14

16

18EP ENVI set-aside proposal

EP ITRE agrees set-aside

Council EE directive w/o set-aside

EC plan to backloading

EC backloading proposal

EP ITRE against backloading

EP ENVI in favor of backloading

EP ENVI no speedy backloading

EP negative vote

EP ENVI amended backloading

EP positive vote

EP+Council confirm support

EP ENVI fasttrack backloading

Backloading in legislation

2008 2009 2010 2011 2012 2013 2014

EUA

year

-ahe

ad D

ecem

ber f

utur

es p

rice

(EUR

/tCO

2)

5

10

15

20

25

30

Compromise 2020 package

Adoption 2020 package Moving beyond 20% I Moving beyond 20% II

Cap for 2013 I

Cap for 2013 II

Proposal low-carbon roadmap 2050

Council on low-carbon roadmap I

Proposal energy roadmap 2050

Council on low-carbon roadmap II

EP on low-carbon roadmap 2050

Green Paper 2030 framework

Proposal 2030 framework

EP resolution 2030 framework

Council on 2030 framework


Thompson Reuters (2015) Carbon Market Survey.


10

carbon price (eur/tCO2)0 5 10 15 20 25 30 35 40

dens

ity

0

0.05

0.1

0.15

0.2

EU Emissions Trading System (ETS) 2020Western Climate Initiative (WCI) 2020Regional Greenhouse Gas Initiative (RGGI) 2016


Nemet question to venture capitalist:

How do you value the benefits of policy in your decisions to invest in start-up companies?

“We ignore it. What the government giveth, it can taketh away.’’

- venture capitalist in energy sector

“It does not affect profit projections. It goes below the line.”– energy finance professional

2. Fragile demand pull: consequences of not fully credible commitments

11


How did solar PV get cheap?

• Full answer: a combination of factors

• Short answer: expectations

2. Fragile demand pull: importance of credibility to incentives

12


80 years of PV prices

13


80 years of PV prices

14

How did solar get cheap?


$3.68

5%2%2%3%12%

30%

43%

3%

$25.30

$0

$5

$10

$15

$20

$25

20

02

$/W

1979

price

2001

price

efficiencyplant

size

Si

price

wafer

size

Si

used

yield poly-

x-stal

un-

explained

2. Fragile demand pull: importanceHow did solar get cheap?

Biggest reason: Economies of scale

Nemet, G. F. (2006). "Beyond the learning curve: factors influencing cost reductions in photovoltaics." Energy Policy 34(17): 3218-3232.

15



…expectations of future demand

Biggest reason: Economies of scale

Nemet, G. F. (2006). "Beyond the learning curve: factors influencing cost reductions in photovoltaics." Energy Policy 34(17): 3218-3232.

16



…expectations of future demand17

1. Technology Push in 1970s-80s

2. German FIT, 2004-present – pure demand pull– high credibility– Other countries too.

3. Chinese scale up

4. Cheap PV everywhere.


• allow for policy experimentation• recover from policy mistakes• make use of new information• respond to unexpected events• account for changes in social priorities

2. Fragile demand pull:is there a case for flexibility?

18


Approach: look at how other policy areas have done it:

- monetary policy- fiscal policy- trade policy

2. Fragile demand pull:How do we navigate the trade off

between commitments and flexibility?

19


2. Fragile demand pull:Addressing credibility problems

Learning from other sectors

Nemet, G., M. Jakob, J. Steckel and O. Edenhofer (in preparation). "Addressing credibility problems in climate policy.”

20

Example1.*Design*of*rules

Rules&on&future&targets M F interest&rate&targetsConditional&rules M F T fiscal&rulesDiscretion&within&rules M F safety&valvesPeriodic&review&of&targets O 5=year&stocktakeCounter=cyclical&mechanisms M F O target&over&bus.&cycle

2.*Transparency*and*trustMonitoring&and&verification M F national&accountsIndependent&authority M T WTOReputation&and&experience T tough&on&inflation

3.*Political*economy*and*distributionCompensate&losers F T O grandfatheringCreate&new&winners F T O exportersTwo=level&game T trade&liberalizationPolicy&windows O ozone&treaty

4.*RobustnessMultiple&instruments F O social&insuranceDecentralized&policy&making T O tariff&setting

Policy*Area


2. Fragile demand pull: A taxonomy for improving incentives in climate policy

21


Advantages for credibility Disadvantages and risks Climate example1. Design of rules

Rules on future targets

Can be legally binding Too inflexible for a dynamic problem

Binding emissions limits

Conditional rules Account for multiple goals Can increase emissions Intensity targetsDiscretion within rules

Respond to new information and shocks

Vulnerable to political expediency

Safety valves

Periodic review Enables ratcheting; revise on predictable schedule

Possible to weaken; incentive to delay

5-year global stocktake

Counter-cyclical mechanisms

Avoids amplifying cycles; cyclical opportunities (low interest rates)

Difficult to implement, e.g. in defining a cycle.

Public investment when energy/carbon prices low

2. Transparency and trustMonitoring and verification

Provide accountability; enable cooperation

Conflicts with sovereignty; institutional capacity needed

UNFCCC Paris Agreement: INDCs

Independent authority

Can exercise discretion independently of politics

Regulatory capture; unresponsive to electorate

Carbon market efficiency board

Reputation and experience

Can create strong incentives without rules

Can be reversed without deliberative process

UNFCCC process; leader nations


2. Fragile demand pull: A taxonomy for improving incentives in climate policy

22


Advantages for credibility Disadvantages and risks Climate example

3. Political economy and distributionCompensate losers Avoid efforts by powerful

interest groups to weaken credibility

Inefficient; prone to gaming; fairness; hard to know how much to compensate;

Grandfathering; free allocation of emissions permits

Create new winners Can create support for strong policy

Benefits too dispersed; subsidies not 1st best

Carbon pricing; new tech. subsidies

Two-level game Mechanism in which international commitments reinforce domestic ones

Can work in opposite direction if international coordination or domestic support is weak

UNFCCC in combination with bottom-up nationally determined contributions

Policy windows Politically infeasible targets may rapidly become feasible

Credibility can be reversed when crisis fades

Pursue efforts to limit to 1.5C as a possible future target

4. RobustnessMultiple instruments Incentives for investment

maintained if one policy is abandoned

Inefficient; complexity may weaken incentives; perverse incentives

Carbon prices and regulations; multi-level governance

Decentralized policy making

Enables policy innovation; incentives may be robust to changes in one jurisdiction

Inefficient; may be poorly coordinated; may not enable cooperation

Sub-national climate policies; co-benefits


3. In between Tech Push and Demand Pull:the “Valley of death”

those making an investment. Because knowledge about perfor-mance may have high value, but may also be nonexcludable,social returns to investment at this stage may far exceed privatereturns. A lack of investment by both the public and privatesector has been a typical result. Successful models exist; thisstage of the technology innovation process is particularly ame-nable to cost sharing between governments, private firms, andindustrial consortia. Investment by the public sector is madedifficult however by the need to concentrate substantial fundsin a small number of projects. The concentration of funds hasmade investments at the demonstration stage vulnerable toshifting political support and, conversely, prone to regulatorycapture that may excessively prolong programs and funding.

Other Sources of Market Failure

Additional market failures exist. The term ‘information asym-metry’ is sometimes used to describe a mechanism very similarto the knowledge spillover mechanism described above toexplain the low adoption of energy-efficient end-use devicesand behaviors. Consumers may not have access to informationabout the benefits and risks of adopting a new, energy-efficienttechnology. Early adopters may reveal information about thesecharacteristics, for example, convenience or reliability. Theymay also discover ways to use devices that improve perfor-mance, a process sometimes called learning by using. Earlyadopters thus create positive externalities for those who ob-serve them.

Another set of market failures has to do with firms’ appetitefor risk and their preferred timing of payoffs to investments.Firms may be risk averse and reluctant to take on the riskassociated with investing in a new technology that, like allnew technologies, may not ultimately succeed, whether tech-nically or because potential customers reject it. In many cases,firms’ aversion to risk and need for returns within a few years issocially beneficial in that it helps prioritize investments andkeeps firms sufficiently profitable so that they can continue toinvest. However, there may be reasons why society wouldprefer that firms take on more risk. For example, aggregationof risk across firms may reduce the adverse societal conse-quences of unfavorable investment outcomes. Similarly, inpart because of the long lifetimes of capital stock in the energysector, the time involved with bringing a technology from theR&D phase to adoption in the marketplace at a scale sufficient

to pay back investments may be longer than companies, oreven venture capital investors, find acceptable. It is possiblethat society employs a longer time horizon than those in theprivate sector making investment decisions. In combination,private-sector risk aversion and reduced time horizons may beparticularly problematic for climate change policy if there isimperfect foresight as to the state of future policies. Expectedfuture payoffs may not stimulate private-sector R&D becausefuture markets for climate-related technology may be consid-ered too uncertain, especially because demand for them istypically heavily influenced by policy decisions, which canchange and thus make markets volatile and investment evenmore risky.

Policy Instruments to Address Market Failures

The need for technological innovation in an environment char-acterized by multiple market failures creates a challenge fortechnology policy: how should public resources be allocatedacross a diverse set of policy instruments, for an unknownnumber of technological options, over a multiple-decade timescale? As a general response, governments can improve the in-centives that innovators face by implementing policy instru-ments addressing each of the externalities involved in theinnovation process.

Addressing Environmental Externalities

Foremost, governments can address the environmental exter-nality by implementing policies that would make pollutingentities confront the costs of the future damages that will resultfrom GhG emissions.

Price signals, in the form of prices for GhG or carbonemissions, raise the cost of carbon-intensive energy technolo-gies and make low-carbon alternatives more attractive as sub-stitutes; the expected future demand for low-carbontechnologies increases with the stringency of the policy,whether via an emissions constraint or a price. Investors ininnovative low-carbon technologies will expect higher payoffsand thus increase their investment as expectations about thestringency of future policy rise. Climate policy can thus induceprivate-sector efforts to invest in developing low-carbon tech-nologies and thus can reduce the costs of reducing emissions.In general, emissions fees provide an advantage over technol-ogy standards since they reward performance that is better thanthe standard, which is especially important in the context oftechnological change. For example, under emissions fees, apolluter that reduces its emissions below what the standarddesignates continues to benefit from the pollution abatementin the form of lower emissions-fee payments.

Uncertainties about both future damages and future coststo avoid those damages introduce a tradeoff in policy design.By imposing a tax, policy makers can set a limit on the eco-nomic costs of climate policy, while leaving future environ-mental damages unconstrained. Alternatively, by imposing aquantity-based constraint on the amount of pollution allowed,they can set firm limits on environmental damage, albeit withunknown future costs. If the costs of abatement are expectedto start to rise steeply relative to damages, a price-based

Basicresearch

Rat

e of

ret

urn Social

Private

Appliedresearch

Demon-strations

Earlydeployment

Diffusion Saturation

Figure 4 Illustrative social and private rates of return on investment atstages in the process of innovation.

112 Markets/Technology Innovation/Adoption/Diffusion | Technological Change and Climate Change Policy

Author's personal copy

Encyclopedia of Energy, Natural Resource, and Environmental Economics, (2013), vol. 1, pp. 107-116

Nemet, G. F. (2013). Technological change and climate-change policy. Encyclopedia of Energy, Natural Resource and Environmental Economics. J. Shogren. Amsterdam, Elsevier: 107--116.

23

• High spillovers• Large capital required• High technology risk• Uncertain market


3. In between Tech Push and Demand Pull:the “Valley of death”

Anadon, L. D. and G. F. Nemet (2014). The U.S. Synthetic Fuels Corporation: Policy Consistency, Flexibility, and the Long-Term Consequences of Perceived Failures. Energy Technology Innovation: Learning from Historical Successes and Failures. A. Grubler and C. Wilson. Cambridge, Cambridge University Press: 257—273.

U.S. Synfuels Corp. (1979-86) $5bGoal: 2m bbl/day by 1992 (33%)

24

$0.5bThe “Technology Pork Barrel”


3. In between Tech Push and Demand Pull:the “Technology Pork Barrel”

25

Cohen, L. R. and R. G. Noll (1991). The Technology Pork Barrel. Washington, Brookings.

“American political institutions introduce predictable systematic biases to R&D programs so that on balance, government projects will be susceptible to performance underruns and cost overruns.” – Cohen and Noll

“government should not pick winners”

...but what if scale, spillovers, and market uncertainty force a choice?


3. In between Tech Push and Demand Pull:Bridging the “Valley of death” whileavoiding the “Technology Pork Barrel”

26

Nemet, G., K. Neuhoff and V. Zipperer (in preparation). "The valley of death and the technology pork barrel: support for radical low-carbon innovation in the materials sector.”

1. On what factors does the case for government intervention rest?

2. How to maximize the effectiveness of government support?



27


1. On what factors does the case for government intervention rest?•Appropriability•Radicalness•Scale•Markets



28


2. How to maximize the effectiveness of government support?• Synthetic fuels• CCS• Solar thermal electricity• Nuclear• Steel• Cellulosic biofuels• Wind power


3. In between Tech Push and Demand Pull:

29


Motivation (stated)

18

24

24

29

29

Others

Production

Scaling7up

Proving7technology

Creating7knowledge

Projects)in)sectors:)Solar,)CCS,)Wind,) Synfuels,)Biofuelsn=43 projects


0%

20%

40%

60%

80%

100%Company(Share(of(P

roject(Costs

Biofuels


30


Percentage of Total Project Costs Funded by Company

CCS

Steel

Synfuels

CSP



31


ADD SCALE SLIDE

1980 1985 1990 1995 2000 2005 2010 2015 2020

Siz

e p

er

pla

nt

(MW

)

0

50

100

150

200

250

Solar thermal electricity

SEGSComm‘l scaleComm. plantsDemo plants

Figure 5: Scale of demonstration projects relative to full commercial scale.

1985 1990

Eff

icie

ncy

0

0.1

0.2

0.3

0.4


Le

veliz

ed

co

st o

f e

lect

rici

ty (

$/M

Wh

)

0

100

200

300

400

Figure 6: Performance of demonstration projects.

6



32


ADD PERFORMANCE SLIDE

1980 1985 1990 1995 2000 2005 2010 2015 2020

Siz

e p

er

pla

nt (M

W)

0

50

100

150

200

250


SEGSComm‘l scaleComm. plantsDemo plants

Figure 5: Scale of demonstration projects relative to full commercial scale.

1985 1990

Eff

icie

ncy

0

0.1

0.2

0.3

0.4


Le

veliz

ed

co

st o

f e

lect

rici

ty (

$/M

Wh

)

0

100

200

300

400

Figure 6: Performance of demonstration projects.

6



33


1970 1980 1990 2000 2010 2020

Dem

onst

ratio

n p

lants

, in

vest

ment ($

m)

0

1000

2000

Synfuels

Price

of oil

(2014$/b

bl)

0

50

100

oil price1983 forecastdemonstrationcontract begunoperationalshut down

Figure 7: Markets for demonstration projects in synfuels.

7


In summary...• Technology Push and Demand Pull– both needed, interactions

• Policy credibility and expectations– Crucial for private incentives– 2nd best solutions likely needed...many exist

• Valley of Death and Technology Pork Barrel– Public support needed for some technologies– But implementation difficult...is it impossible?– Scale-up and iterative learning– Decisions coming on this

34

energy’innovation and public · gregory nemet— energy innovation and public policy 1970 1980...

Documents