Wiring America - The Short- and Long-Run Effects of

Wiring AmericaThe Short- and Long-Run Effects of Electricity Grid Expansion

Gaurav Doshi, PhD CandidateDept. of Ag & Applied Economics

University of Wisconsin MadisonJuly 8, 2021

Electricity generated from renewable sources in the US has beenincreasing | 1

Intro Context Short-run Long-run Conclusions References Appendix

US needs more transmission lines to “green the grid” | 2

I Availability of transmission a key factor in fully utilizing the growingwind and solar capacity in the US.

I Market integration due to grid expansion can have consequences formarket power as well (Borenstein, Bushnell, and Stoft 2000; Borenstein,Bushnell, and Wolak 2002; Joskow and Tirole 2000, 2005; Woerman 2019; Ryan2021).

I Grid expansion has significant non-market benefits - increasing valueof wind and lowering emissions (Jorgensen, Mai, and Brinkman 2017;LaRiviere and Lu 2020; Fell, Kaffine, and Novan 2021).


US needs more transmission lines to “green the grid” | 3

5/27/2021 Down to the wire: Biden’s green goals face a power grid reckoning - POLITICO

https://www.politico.com/news/2021/04/08/biden-green-goals-power-grid-480446 1/12

The U.S. will need new electric transmission lines to meet thepresident’s aim of eliminating the power sector’s net carbon pollution.But public opposition has doomed many such projects.

President Joe Biden's dream of a climate-friendlyelectric grid hangs on a slender wire: hisadministration's ability to speed the construction ofthousands of miles of power lines.

��

5/27/2021 Down to the wire: Biden’s green goals face a power grid reckoning - POLITICO

https://www.politico.com/news/2021/04/08/biden-green-goals-power-grid-480446 1/12

The U.S. will need new electric transmission lines to meet thepresident’s aim of eliminating the power sector’s net carbon pollution.But public opposition has doomed many such projects.

President Joe Biden's dream of a climate-friendlyelectric grid hangs on a slender wire: hisadministration's ability to speed the construction ofthousands of miles of power lines.

��

NEW YORK

(AP) — The

federal

government

said Tuesday

RELATEDTOPICS

Joe Biden

Environmenature

Environme

Governmepolitics

Business

Click t

Biden releases money in push tomodernize US electric gridBy CATHY BUSSEWITZ April 27, 2021

Trending on AP News

Tale of rescue after falling several floorsin Fla. collapse

US left Afghan airfield at night, didn’t tellnew commander

Ventilator shortage as Missouri virushospitalizations spike

AP NEWSTop Stories VideoTopics Listen

NEW YORK

(AP) — The

federal

government

said Tuesday

RELATEDTOPICS

Joe Biden

Environmenature

Environme

Governmepolitics

Business

Click t

Biden releases money in push tomodernize US electric gridBy CATHY BUSSEWITZ April 27, 2021

Trending on AP News

Tale of rescue after falling several floorsin Fla. collapse

US left Afghan airfield at night, didn’t tellnew commander

Ventilator shortage as Missouri virushospitalizations spike

AP NEWSTop Stories VideoTopics Listen

�� 7KHUH�,V�1RW�(QRXJK�7UDQVPLVVLRQ�WR�0HHW�86�&RUSRUDWH·V�5HQHZDEOH�(QHUJ\�'HPDQGV��5HQHZDEOH�(QHUJ\�:RUOG

KWWSV��ZZZ�UHQHZDEOHHQHUJ\ZRUOG�FRP��WKHUH�LV�QRW�HQRXJK�WUDQVPLVVLRQ�WR�PHHW�XV�FRUSRUDWH�V�UHQHZDEOH�HQHUJ\�GHPDQGV� ��

��'��9>��4��"��#��`4"#aL�2I0I

��>8�$4��

��I�%��L��

��D��I

)��2I0I��G��#��=88��G��

��D��

��I��4"#��L��

��988��>8�$4��:8:=I�1��

��=9�$4��

��L��

��I

T1��

��LU��'��(��L��

��L�4"#I�T��D�� L��D

��D��

��IU

1��4"#��L��

��

��I�#��L��D��

��L��D��<:��

��D��L��?@��

��D��I

T$*S��

��LU��/��1��L��$��*��I�T&�S�

9I::I9@

� �


What are the short- and the long-run effects of grid expansion?| 4

1. How does grid expansion affect marginal fossil fuel generators in theshort-run?I Realized markups.I Impact on marginal emissions - global (CO2) and local

pollutants (SO2 and NOx).

2. Does grid expansion enhance wind investment in the long-run?I Do sites with transmission infrastructure see higher investment

in wind energy?

Contributions:I This paper is focused on grid expansion projects aimed at renewable

integration: contextually relevant to the US.I Use rich spatial and temporal data on transmission expansion.I First evidence on the long-run effects of transmission expansion.


What are the short- and the long-run effects of grid expansion?| 4

1. How does grid expansion affect marginal fossil fuel generators in theshort-run?I Realized markups.I Impact on marginal emissions - global (CO2) and local

pollutants (SO2 and NOx).

2. Does grid expansion enhance wind investment in the long-run?I Do sites with transmission infrastructure see higher investment

in wind energy?Contributions:I This paper is focused on grid expansion projects aimed at renewable

integration: contextually relevant to the US.I Use rich spatial and temporal data on transmission expansion.I First evidence on the long-run effects of transmission expansion.


Clear localization of wind farms and fossil fuel generators inTexas | 5

Austin

DallasFort Worth

HoustonSan Antonio

27.5

30.0

32.5

35.0

−104 −100 −96Longitude

Latit

ude

Nameplate Capacity

150

350

550

750

950

Fuel Type

Coal

NG

Other

Wind

Figure 1: Wind farms and fossil fuel generators in TexasIntro Context Short-run Long-run Conclusions References Appendix

The CREZ Transmission Expansion Project | 6

0.0

0.2

0.4

0.6

0.8

1.0

2010 2011 2012 2013 2014Date

Per

cent

CR

EZ

com

plet

ion

I Integrate existing and growing wind capacity in West Texas to majordemand centers.

I Total cost ∼ $7 Billion.I About 80% of the project completed in 2013.I All the lines brought in-service by Jan 2014.


Most data comes from ERCOT, EIA, and EPA | 7

I ERCOT (Texas’s grid operator) :I Hourly data on generator(s) submitting highest offers

dispatched in market clearing and prices (2011-2014).I Hourly data on wind generation and electricity demand.I Daily data on CREZ expansion - county location and timing.

I EPA CEMS: Hourly generator level heat input and emissions data.

I EIA Form 860: nameplate capacity, year of operation, location, fueltype, technology.

Detailed Data Sources


Short-run: Estimating the impact of grid expansion on fossil fuelmarkups | 8

I Theory: Solve marginal generator’s optimization problem toderive the optimal markup (p −mc) rule.

I Construct a two-step estimator from the optimal markup rule:

Step 1 Estimate the marginal effect of wind generation on realizedmarkups at each hour: αh.

Step 2 Estimate the impact of daily transmission expansion on hourlywind generation: βh.

I Impact of CREZ on generator markups : αh × βhTheory Model


Impact of CREZ on markups is strongest at the peak hours | 9

7 4.7 3.6 3.3 3.8 5.8 13 9.3 10.1 12.8 14.9 14.7 16.1 16.5 18.5 22.3 35.9 22.8 20.8 15.5 15.2 13.4 11.9 9

−1.25

−1.00

−0.75

−0.50

−0.25

0.00

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23Hour

$/M

Wh

Figure 2: Effect of CREZ completion on fossil fuel generator markups andthe 95% Confidence Intervals.(Peak hours: 7:00 - 22:00. Average markups for the sample (8/2011 - 12/2014)mentioned above the x axis.)


Semi-elasticity of markups is highest at the off-peak hours | 10

7 4.7 3.6 3.3 3.8 5.8 13 9.3 10.1 12.8 14.9 14.7 16.1 16.5 18.5 22.3 35.9 22.8 20.8 15.5 15.2 13.4 11.9 9

−9

−8

−7

−6

−5

−4

−3

−2

−1

0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23Hour

Per

cent

(%

)

Figure 3: Semi-elasticity of markups due to CREZ (crezi = 1) and the95% Confidence Intervals.(Peak hours: 7:00 - 22:00. Average markups for the sample (8/2011 - 12/2014)mentioned above the x axis.)


Short-run: Impact of CREZ Expansion on emissions frommarginal generator(s) | 11

Estimate the impact of additional wind generation on CO2emissions and local pollutants from generators at the margin (ρzh):

Ezt = ρzh · wt + f (Dzt |λ) + αzy + δhmy + εzt (1)

Ezt Zonal level emissions from marginal generator(s) at hour t:1. CO2 emissions in tons.2. Damages from SO2 and NOx (2020 $) - use county specific

damage estimates from Holland et al. (2016).wt Wind generation at hour t.Dzt Polynomial of zonal electricity load.


Wind added from CREZ prevented damages due to marginalcarbon emissions | 12

Houston North South West

0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20

−200

−150

−100

−50

0

50

Hour

CO

2 em

issi

ons

(ton

s/G

Wh)

Figure 4: Impact on wind generation on carbon emissions

Converting these estimates to damages avoided from CREZ (2020 $):

=⇒∑

z

24∑h=0

44× βh × ρzh ≈ $54, 000/day


Damages avoided from local pollutants largely concentrated toWest Zone | 13

Houston North South West

0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20 0 4 8 12 16 20

−20

0

20

Hour

Dam

ages

from

SO

2 an

d N

Ox

(100

0 $)

Figure 5: Impact on wind generation on damages from SO2 and NOx

Converting these estimates to damage avoided from CREZ (2020 $):

=⇒∑

z

24∑h=0

βh × ρzh ≈ $87, 000/day


Long-run: Did counties with CREZ expansion see higher windinvestment? | 14

I ∼ 3600 miles of 345kV CREZ lines were constructed between newand existing electrical substations.

I I observe the county specific location of these substations -treatment indicator.

Figure 6: A typical electrical substation


Long-run: Did counties with CREZ transmission see higher windinvestment? | 15

yit = β · crezi + X′Π + εit (2)

yit total wind capacity/total turbines/mean project capacity in county iin year t (T = 2012 - 2019).

crezi CREZ indicator for a county.X vector of controls: time trend, wind resource quality, project costs,

regulatory controls, county demographics, and ERCOT Zone FE.Endogeneity of crezi : Counties with already high levels of wind capacitysited CREZ substations.


Use designation of Renewable Energy (RE) Zones as aninstrument for selection into CREZ | 16

(a) RE Zones in 2007

27.5

30.0

32.5

35.0


Latit

ude

County Designation

Non−RE ZonesRE Zones

(b) RE Zones and CREZ counties(hatched)

I Exclusion Restriction: Conditional on X, RE Zones unlikely to affectwind investment post 2012 other than through CREZ counties.

I Threat to identification: Counties within RE Zones are inherentlydifferent than other counties - common support problem.


Use designation of Renewable Energy (RE) Zones as aninstrument for selection into CREZ | 16

(a) RE Zones in 2007

27.5

30.0

32.5

35.0


Latit

ude

County Designation

Non−RE ZonesRE Zones

(b) RE Zones and CREZ counties(hatched)

I Exclusion Restriction: Conditional on X, RE Zones unlikely to affectwind investment post 2012 other than through CREZ counties.

I Threat to identification: Counties within RE Zones are inherentlydifferent than other counties - common support problem.


Coarsened Exact Matching to construct control and treatedgroup | 17

Variables Post-MatchingMeans Treated Means Control p-value

[CREZ = 1] [CREZ = 0]Wind Capacity as of 2008 (MW) 5.581 4.264 0.138Wind Speed (m/s) 7.887 7.891 0.619Capacity Factor 0.437 0.439 0.949Avg. Land Price (2007-2010) 228.424 231.216 0.929Median Land Acreage 360.746 351.736 0.161Avg. Farm Size in 2007 1, 183.140 1, 262.035 0.118Median Income in 2007 35, 789.190 35, 574.620 0.837Avg. Population (2007-2010) 28, 917.870 20, 612.030 0.026Total Units 104 240

Notes: This table presents balance test of key pre-treatment observable characteristics of acounty. Each unit is a county-year observation. Exact matching on discrete variables: ERCOTZone and Power Curve designation of a county based on wind resource.


Identical set of counties based on observable dimensions | 18

27.5

30.0

32.5

35.0


Latit

ude

County Type

ControlTreatedNone

Figure 8: Control and treated counties obtained using Coarsened Exact Matching


CREZ counties saw higher wind investment post transmissionexpansion announcement in 2008 | 19

Dependent Variable:Total Capacity Total Turbines Avg. Project

(MW) Capacity (MW)2SLS CEM 2SLS CEM 2SLS CEM

CREZ

62.181∗∗ 72.640∗∗∗ 30.285∗∗ 39.419∗∗∗ 14.089 32.756∗(26.786) (26.499) (13.596) (13.075) (16.136) (19.093)

First Stage F-Stat

12.842 − 12.842 − 12.842 −

Controls

X X X X X X

Mean Dep. Variable

32.939 35.907 15.866 16.067 19.911 26.951

R2

0.216 0.467 0.205 0.476 0.195 0.426

Observations

2,024 344 2,024 344 2,024 344

Notes: This table reports results of 2SLS regression and OLS regression on matching sample obtainedusing Coarsened Exact Matching. Controls include cubic polynomial for time trend, wind resource qual-ity, project costs, regulatory controls, county demographics, and Zone or group FEs. Robust StandardErrors clustered at county level reported in parenthesis. Significance: ***p<0.01;**p<0.05;*p< 0.1





CREZ 62.181∗∗ 72.640∗∗∗

30.285∗∗ 39.419∗∗∗ 14.089 32.756∗

(26.786) (26.499)

(13.596) (13.075) (16.136) (19.093)

First Stage F-Stat 12.842 −

12.842 − 12.842 −

Controls X X

X X X X

Mean Dep. Variable 32.939 35.907

15.866 16.067 19.911 26.951

R2 0.216 0.467

0.205 0.476 0.195 0.426

Observations 2,024 344

2,024 344 2,024 344






CREZ 62.181∗∗ 72.640∗∗∗ 30.285∗∗ 39.419∗∗∗

14.089 32.756∗

(26.786) (26.499) (13.596) (13.075)

(16.136) (19.093)

First Stage F-Stat 12.842 − 12.842 −

12.842 −

Controls X X X X

X X

Mean Dep. Variable 32.939 35.907 15.866 16.067

19.911 26.951

R2 0.216 0.467 0.205 0.476

0.195 0.426

Observations 2,024 344 2,024 344

2,024 344






CREZ 62.181∗∗ 72.640∗∗∗ 30.285∗∗ 39.419∗∗∗ 14.089 32.756∗(26.786) (26.499) (13.596) (13.075) (16.136) (19.093)

First Stage F-Stat 12.842 − 12.842 − 12.842 −Controls X X X X X XMean Dep. Variable 32.939 35.907 15.866 16.067 19.911 26.951R2 0.216 0.467 0.205 0.476 0.195 0.426Observations 2,024 344 2,024 344 2,024 344



Evidence of market and non-market impacts in the short- andthe long-run | 20

I CREZ led to a drop in fossil fuel markups across all hours withhighest semi-elasticity at the off-peak hours.

I Approximately $51 million (2020 $) worth of total damages avoidedfrom marginal emissions annually.

I Counties with transmission expansion saw substantially higher levelsof wind investment and bigger wind projects in the long-run.

I Investment in transmission could have substantial gains inaddressing the transmission hold-out problem especially in locationswith high wind quality.


Thank you!

Any Comments, Suggestions, Ideas:[email protected]


mailto:[email protected]

References | 22Borenstein, Severin, James Bushnell, and Steven Stoft. 2000. “The competitive effects of transmission capacity in a

deregulated electricity industry.” The Rand Journal of Economics 31 (2): 294.

Borenstein, Severin, James B Bushnell, and Frank A Wolak. 2002. “Measuring market inefficiencies in California’srestructured wholesale electricity market.” American Economic Review 92 (5): 1376–1405.

Fell, Harrison, Daniel T Kaffine, and Kevin Novan. 2021. “Emissions, transmission, and the environmental value ofrenewable energy.” American Economic Journal: Economic Policy 13 (2): 241–272.

Holland, Stephen P, Erin T Mansur, Nicholas Z Muller, and Andrew J Yates. 2016. “Are there environmentalbenefits from driving electric vehicles? The importance of local factors.” American Economic Review 106(12): 3700–3729.

Jorgensen, Jennie, Trieu Mai, and Greg Brinkman. 2017. Reducing wind curtailment through transmission expansionin a wind vision future. Technical report. National Renewable Energy Lab.(NREL), Golden, CO (UnitedStates).

Joskow, Paul L, and Jean Tirole. 2000. “Transmission rights and market power on electric power networks.” TheRand Journal of Economics, 450–487.

. 2005. “Merchant Transmission Investment.” The Journal of Industrial Economics 53 (2): 233–264.

LaRiviere, Jacob, and Xueying Lu. 2020. “Transmission constraints, intermittent renewables and welfare.” WorkingPaper.

Ryan, Nicholas. 2021. “The Competitive Effects Of Transmission Infrastructure In The Indian Electricity Market.”American Economic Journal: Microeconomics 13 (2): 202–242.

Woerman, Matt. 2019. “Market size and market power: Evidence from the Texas electricity market.” EnergyInstitute Working Paper 298.


Detailed Data Sources: Short-Run | 23

I ERCOT:

I Generator level hourly data on wind generation, electricitydemand, wholesale prices (August 2011 - December 2014)

I Daily data on CREZ expansion.

I Generator level heat input and emissions data from EPA’sCEMS.

I EIA Form 860: Generator location and nameplate capacity.

I Fuel and emission allowance pricesI Data on Coal prices (Powder River Basin) from EIAI Data on NG prices (Henry Hub Spot price) from QuandlI NOx and SO2 permit allowance prices from S&P Global MI


Detailed Data Sources: Long-Run | 24

I EIA Form 860 DataI Wind generator data - nameplate capacity (MW), year of

operation, and location.I Fossil fuel generator data - nameplate capacity (MW), year of

operation, location, fuel type, technology.I Texas A&M University Real Estate Center data - land price,

median land acerage.I NREL Wind Toolkit - wind resource quality, capacity factor,

power curve data.I Lawrence Berkeley Wind Tech Report - annual wind project

cost.I 2012 and 2017 USDA Ag Census - average farm size.I WINDExchange - wind ordinance data.


Theory Model in Math | 25

Setup:I Consider two geographically distinct regions: Region W with wind

generation and Region S with fossil fuel generators and inelasticmarket demand DS .

I Transmission lines K enable import of electricity from wind qw toRegion S.

I Generator i maximizes its expected profit function πi (p) to find pricep.

I It faces uncertainty over offer schedules S−i = (b−i , q−i ) fromcompetitor fossil fuel generators in S.

I Generator i ’s residual demand curve:Dr

i (p, qw ; K) = DS − qw − Qf (qw , p).I Market clears when Qi (p) = Dr

i (p, qw ; K)


Solving generator i ’s optimization problem | 26

Optimization problem of generator i

maxbi

ES−i

[p(Qi (p)− QF

i ) + pF QFi − Ci (Dr

i (p,K ))]

(3)

Denote Qi (p, qw )− QFi as Qnet

i (p, qw ). Taking first order condition withrespect to bi and rearranging,

=⇒ ES−i

[∂p∂bi

(Qnet

i (p, qw ) + ∂Dri (p, qw )∂p [p − C ′i (Dr

i (p, qw ))])]∣∣∣∣

p=bi

= 0

(4)

Assuming constant marginal cost ci and full information on othergenerators’ strategy, optimal markup rule:

p − ci = − Qneti (p, qw )

∂Dri (p, qw )/∂p (5)


Comparative statics | 27

Differentiating Equation 5 wrt K and rearranging:

1p − ci

·∂(p − ci )∂K =

[1

Qneti (p, qw ) ·

∂Qneti (p, qw )∂K

]︸︷︷︸

∆Production

−[

1∂Dr

i /∂p ·∂2Dr

i (p, qw )∂p∂qw

]︸︷︷︸

∆Elasticity(6)

∆Production∂Qnet(p, qw )

∂K = ∂Qnet(p, qw )∂qw

· ∂qw∂K (7)

∆Elasticity∂2Dr

i (p,K)∂p∂K = − ∂ηf

∂qw· ∂qw∂K (8)

where, ηf = ∂qf (qw ,p)∂p (> 0) denotes the slope of competitor (marginal)

fossil fuel generators supply curve.


Model Predictions | 28Substituting the expressions for the the two effects from 7 and 8 in 6 andsimplifying yields:

1p − ci

· ∂(p − ci )∂K =

[1

Qneti· ∂Qnet

i∂qw

+ 1∂Dr

i /∂p ·∂ηf∂qw

]· ∂qw∂K (9)

∂(p − ci )∂K = ∂(p − ci )

∂qw︸︷︷︸R0

· ∂qw∂K︸︷︷︸>0

(10)

1. Transmission expansion leads to an increase in import of electricitygenerated from wind.

2. Addition of wind due to transmission would lead to inward shift ofi ’s residual demand curve =⇒ lower markups.

3. Firm exit could incentivize i to submit steeper offer curves =⇒higher markups.


Documents

Wiring America - The Short- and Long-Run Effects of