View Conference Presentation - United States Association for

Report
Modeling the Impacts of State Energy Policies
Barry Rubin, Zachary Wendling, David Warren,
Sanya Carley, and Kenneth Richards
School of Public and Environmental Affairs
Indiana University
Bloomington, IN 47405
Presented at the 32nd USAEE/IAEE North American Conference,
Anchorage, Alaska, July, 2013
Evaluating the Impacts of State Energy Efficiency:
Status of State- and Sub-State-Level Energy Policy
Impact Analysis
• US energy policy is primarily driven by state
governments.
• Decisions at the state and sub-state levels are based
on preconceptions rather than detailed policy
analysis.
• There is a compelling need for tools that can help
state and local governments and business leaders
evaluate alternative energy pathways.
Evaluating the Impacts of State Energy Efficiency:
An Econometric Modeling Approach
• Over the past four years, our research team at Indiana
University has created a framework for evaluating the
impact of alternative energy policy scenarios.
• We have constructed an econometric, simultaneous, multiequation econometric model that addresses the
connection between energy consumption, energy prices,
and economic activity for Indiana.
• In this presentation, we describe the estimation and
simulation results for the sample period, and
• provide the results of applying the state model to analyze a
Demand Side Management (DSM) policy.
Evaluating the Impacts of State Energy
Efficiency: An Econometric Modeling Approach
• The model will eventually include 10 homogeneous multicounty regions. However, the research reported here
focuses on our state-level model and two regions (South
Bend and Indianapolis), each with 30 stochastic equations.
• The endogenous variables of the model include
employment and earnings for ten economic sectors, and
energy consumption disaggregated across three fuel types
and four end use categories.
• The model also addresses unemployment, nonwage
income, total personal income, and GDP per capita.
Evaluating the Impacts of State Energy
Efficiency: An Econometric Modeling Approach
• The Indiana econometric model was estimated with annual
data over the period 1977-2011.
• It is structured in a dynamic, simultaneous framework that
allows tracing the effect of changes in exogenous and
policy variables over the course of successive periods.
• The explicit linkage of diverse economic sectors via this
structure simulates the interdependent nature of the
economy, allowing the effects of energy and environmental
policies to be transmitted throughout the model.
Evaluating the Impacts of State Energy
Efficiency: An Econometric Modeling Approach
• Data sources include the U.S. Bureau of Labor Statistics,
the Bureau of Economic Analysis, the Energy Information
Administration (EIA), the Annual Survey of Manufactures
and the Indiana Business Research Center.
• Exogenous variables include population, climatic variables,
national level economic data, and, most importantly,
energy prices for natural gas, motor gas, and electricity.
• Figure 1 graphically depicts the simultaneous nature of the
model by identifying the various linkages across sectors
and equations.
Evaluating the Impacts of State Energy Efficiency
Evaluating the Impacts of State Energy Efficiency:
Stochastic Equations, Estimation and Simulation Results
• OLS estimation was used due to the tradeoff between
simultaneous equation bias and the potential propagation
of data errors with simultaneous equation estimation
techniques in regional models.
• The following tables identify the endogenous and
exogenous variables of the model, including the specific
variables in each equation.
• Mean Absolute Percent Errors (MAPEs) are provided for
each equation, as are graphs of actual and predicted values
for selected endogenous variables from the sample period
simulations.
Evaluating the Impacts of State Energy Efficiency:
Stochastic Equations, Estimation and Simulation Results
• MAPE values indicate the average error for each
endogenous variable when the model is “turned on” to
simulate the sample period values with which the
equations were fit.
• They provide an indication of the model’s performance and
predictive ability, in contrast to single equation results
which do not include the simultaneous interaction of the
endogenous variables.
• MAPE values under 5% are generally considered excellent,
between 5-10% very good, 10-15% good to mediocre, 1520% potentially problematic, and >20% poor.
Evaluating the Impacts of State Energy Efficiency:
Endogenous and Exogenous Variables
Evaluating the Impacts of State Energy Efficiency:
Endogenous and Exogenous Variables
Energy Variables
Prices
Consumption
esccd
esicd
esccb
esicb
esrcd
ngccd
esrcb
ngccb
ngicd
ngrcd
ngicb
ngrcb
natural gas
industrial end use
residential end use
mgacd
mgacb
motor gas
transportation end use
electricity
commercial end use
industrial end use
residential end use
commercial end use
Evaluating the Impacts of State Energy Efficiency:
Endogenous and Exogenous Variables
Climatic variables
Dummy variables
cdd5
cooling degree days (annual)
yrdum
Post 2000
hdd5
heating degree days (annual)
yrdum90
Post 1990
yrdum97
Post 1997
Lag prefixes
l...
one-year
l2 . . .
two-year
Exogenous variables
Evaluating the Impacts of State Energy Efficiency:
State Estimation and Simulation Results
Dependent variable
Independent variables
Industrial electricity
consumption
Commercial electricity
consumption
Residential electricity
consumption
Industrial natural gas
consumption
Commercial natural gas
consumption
Residential natural gas
consumption
Gross state product per
capita
Non-wage income per
capita
Two-year lagged own price, lagged motor gas price, lagged construction employment, lagged manufacturing
employment, lagged utilities and transportation wage rate
Own price, commercial natural gas price, wholesale employment, lagged finance employment, government wage
rate
Unemployment rate
Manufacturing
employment
Services employment
Manufacturing earnings
per employee
Services earnings per
employee
Own price, two-year lagged commerical natural gas price, finance employment, annual cooling degree days
Own price, motor gas price, lagged industrial electricity consumption, industrial sector employment, lagged
manufacturing employment, lagged utilities and transportation wage rate, lagged industrial natural gas price
Own price, lagged own price, two-year lagged commercial electricity price, annual heating degree days, two-year
lagged annual heating degree days, retail wage rate, lagged government wage rate
Lagged own price, two-year lagged residential electricity price, two-year lagged residential natural gas
consumption, residential electricity consumption, annual heating degree days, unemployment rate
Industrial electricity consumption, lagged commercial electricity consumption, lagged motor gas price, finance
wage rate, 1997 dummy
Residential electricity consumption, lagged motor gas price, lagged government employment, lagged commercial
natural gas price
Lagged motor gas price, construction employment, commercial electricity consumption, utilities and
transportation employment, 1990 dummy
Lagged industrial natural gas price, lagged industrial electricity consumption, lagged services employment,
construction employment, lagged utilities and transportation employment, 2001 dummy
Finance employment, motor gas consumption, retail employment, commercial electricity consumption, 2001
dummy
Lagged industrial natural gas price, motor gas price, industrial electricity consumption, lagged retail employment,
two-year lagged services employment
Commercial electricity consumption, motor gas price, wholesale wage rate, utilities and transportation
employment
R2 MAPE
0.95
3.34
0.98
2.67
0.98
2.20
0.83
4.49
0.85
2.90
0.83
2.52
0.98
1.75
0.97
2.35
0.84 13.42
0.85
3.53
0.99
2.13
0.95
1.44
0.97
1.35
Evaluating the Impacts of State Energy Efficiency:
South Bend Estimation and Simulation Results
Dependent variable
Industrial electricity
consumption
Commercial electricity
consumption
Residential electricity
consumption
Industrial natural gas
consumption
Commercial natural gas
consumption
Residential natural gas
consumption
Gross regional product per
capita
Non-wage income per
capita
Independent variables
R2 MAPE
Own price, industrial natural gas price, lagged manufacturing wage rate, lagged construction employment
0.90
6.59
Own price, services employment, lagged per capita regional product
0.95
5.60
Own price, summer cooling degree days, finance employment, state retail employment, construction wage rate
0.91
3.87
Lagged own price, two-year lagged commercial electricity price, industrial sector employment
0.76
6.13
Own price, annual heating degree days, retail wage rate, lagged commerical electricity price, utilities and
transportation wage rate
0.82
4.28
Own price, commercial sector employment, industrial sector wage rate
0.67
3.58
Construction employment, manufacturing earnings, commerical electricity consumption, 1997 dummy
0.95
3.88
Services employment, industrial sector earnings, residential electricity price
0.86
4.37
Unemployment rate
Manufacturing employment, 1990 dummy
0.79 14.88
Manufacturing
employment
Gross state product per capita, industrial natural gas price, motor gas price, lagged manufacturing employment,
2001 dummy
0.76
4.28
Services employment
Commercial electricity price, finance employment, 2001 dummy
0.97
4.99
Manufacturing earnings
per employee
Services earnings per
employee
Industrial electricity price, industrial electricity consumption, lagged retail employment, utilities and
transportation wage rate
0.89
1.97
Lagged gross regional product, commercial electricity consumption
0.95
1.78
Evaluating the Impacts of State Energy Efficiency:
Indianapolis Estimation and Simulation Results
Dependent variable
Independent variables
Industrial electricity
consumption
Commercial electricity
consumption
Residential electricity
consumption
Industrial natural gas
consumption
Commercial natural gas
consumption
Residential natural gas
consumption
Gross regional product per
capita
Non-wage income per
capita
Own price, lagged industrial natural gas price, lagged motor gas price, government employment, lagged
manufacturing wage rate, lagged state utilities and transportation wage rate, lagged labor force participation rate
Two-year lagged own price, commercial natural gas price, lagged government employment, retail wage rate,
wholesale employment, annual cooling degree days
Own price, two-year lagged residential natural gas price, two-year lagged state finance employment, retail wage
rate, lagged winter heating degree days, summer cooling degree days
Own price, motor gas price, industrial electricity price, manufacturing employment, utilities and transportation
employment, lagged industrial sector employment
Own price, lagged commercial electricity price, state commercial natural gas consumption, finance employment,
lagged utilities and transportation wage rate, annual heating degree days
Own price, state residential natural gas consumption, lagged state residential electricity consumption, finance
wage rate, retail employment, labor force participation rate
Lagged industrial electricity price, lagged motor gas price, lagged commercial electricity consumption, services
employment, 1997 dummy
Lagged commercial electricity price, lagged motor gas price, government employment, lagged service
employment
Commercial electricity price, lagged motor gas price, construction employment, manufacturing employment,
lagged commercial sector wage rate, 1990 dummy
Lagged industrial electricity price, two-year lagged industrial electricity consumption, farming employment, state
manufacturing employment, retail employment, 2001 dummy
Commercial electricity price, commercial natural gas price, lagged commercial electricity consumption, lagged
non-wage income per capita, 2001 dummy
State industrial electricity consumption, motor gas consumption, lagged state retail employment, national GDP
per capita, national unemployment rate
Lagged motor gas price, lagged commercial electricity consumption, government employment, finance
employment
Unemployment rate
Manufacturing
employment
Services employment
Manufacturing earnings
per employee
Services earnings per
employee
R2 MAPE
0.91
9.95
0.97
4.55
0.98
2.50
0.79
6.44
0.92
2.92
0.92
3.20
0.99
1.29
0.97
2.40
0.86 11.76
0.78
3.96
0.99
4.00
0.97
2.28
0.97
1.63
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Manufacturing employment
(State)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Manufacturing employment
(South Bend)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Manufacturing employment
(Indianapolis)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Employment in services
(State)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Employment in services (South
Bend)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Employment in services
(Indianapolis)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Residential electricity
consumption (State)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Residential electricity
consumption (South Bend)
Evaluating the Impacts of State Energy
Efficiency: Simulation Results
Residential electricity
consumption (Indianapolis)
Evaluating the Impacts of State Energy
Efficiency: Policy Analysis
• Policy analysis results with multi-equation models are
derived via differencing a baseline forecast without the
policy implemented, with a forecast in which the policy is
present.
• The policy is translated into changes in the exogenous
variables of the model.
• The policy that we address is the Indiana Utilities
Regulatory Commission’s (IURC) Demand Side
Management (DSM) ruling, passed in 2008, which requires
all state jurisdictional electric utilities to achieve a two
percent decrease in electricity use by 2019.
Evaluating the Impacts of State Energy
Efficiency: Policy Analysis
• This policy is the only requirement for DSM in Indiana and
has not been evaluated in a systematic way with respect to
its impacts on the Indiana economy.
• It also serves as an example of how the econometric model
can be utilized for policy analysis.
• We simulated the implementation of DSM by treating the
electricity consumption variables as exogenous, setting the
2019 values two percent below the 2008 levels, and using a
linear rate of decline between 2008 and 2019.
• The following charts compare selected elements of the
baseline and policy forecasts, illustrating the percent impact
of the policy.
Evaluating the Impacts of State Energy
Efficiency: Economic Impacts
State impacts, percent difference versus
8.5%
the baseline forecast, 2019:
5.9%
3.2%
2.7%
2.0%
0.3%
Total
employment
Total earnings
Total earnings
per employee
GDP per capita
Services
employment
Manufacturing
earnings per
employee
Evaluating the Impacts of State Energy
Efficiency: Economic Impacts Vary by Location
Percent difference in total employment
versus the baseline, 2019:
9.1%
5.5%
3.2%
State
South Bend
Indianapolis
Evaluating the Impacts of State Energy
Efficiency: Economic Impacts Vary by Location
Percent difference in total earnings per
employee versus the baseline, 2019:
6.7%
4.9%
2.7%
State
South Bend
Indianapolis
Evaluating the Impacts of State Energy
Efficiency: Economic Impacts Vary by Location
Percent difference in employment in
services versus the baseline, 2019:
7.8%
6.0%
0.3%
State
South Bend
Indianapolis
Evaluating the Impacts of State Energy
Efficiency: Natural Gas Consumption Forecasts
Percent change in industrial
natural gas consumption versus
the baseline forecast, 2019
23.5%
11.5%
3.5%
State
South Bend
Indianapolis
Evaluating the Impacts of State Energy
Efficiency: Natural Gas Consumption Forecasts
Percent change in
commercial natural
gas consumption versus
the baseline, 2019
6.9%
0.9%
0.1%
State
South Bend
Indianapolis
Evaluating the Impacts of State Energy
Efficiency: Natural Gas Consumption Forecasts
Percent change in
residential natural gas
consumption versus
the baseline, 2019
4.1%
1.8%
State
1.7%
South Bend
Indianapolis
Evaluating the Impacts of State Energy
Efficiency: Economic Impact Results Summary
Implementation of the IURC DSM order is projected to have a
significant impact on the state economy by 2019, with varying
magnitudes by region:
• Total state employment is expected to rise by 107,000 (3.2%) as
compared to the baseline forecast. South Bend and Indianapolis
are projected to gain 25,000 (9.1%) and 55,000 (5.5%) jobs,
respectively.
• Earnings per employee are projected to rise by about $1,200
(2.7%). Gains of $2,500 (4.9%) and $3,900 (6.7%) are forecast for
South Bend and Indianapolis.
• GDP per capita goes up $3,300 (8.5%), with an increase of $6,700
(20.7%) and $2,400 (5.2%) forecast for South Bend and
Indianapolis.
Evaluating the Impacts of State Energy
Efficiency: Energy Impact Results Summary
As for energy consumption, as compared to the baseline forecast:
• The state is forecast to increase industrial, commercial, and
residential natural gas consumption by 11.5%, 0.1%, and 1.8%,
respectively.
• Again, impacts in South Bend and Indianapolis vary in magnitude,
with modest increases in the three natural gas end use sectors for
South Bend (3.5%, 6.9%, and 4.1%). Indianapolis is projected to
see small increases in commercial and residential natural gas
consumption (0.9% and 1.7%) but a large increase in industrial
natural gas consumption (23.5%).
Evaluating the Impacts of State Energy
Efficiency: Conclusions
This research represents an advance in the ability to identify
potential impacts of alternative energy policies by:
• producing useable policy analysis results and thus serving as a
“proof of concept” for the econometric modeling approach,
• disaggregating the potential impacts of critical policy
alternatives by economic sector, fuel type, and end use sector,
• having the potential to inform public and private sector
decision-makers as to impacts on the state economy and
specific industries, and
• assisting energy utilities in planning for the future.
No other policy analysis framework provides such an extensive
range of impact analysis results, nor are other policy analysis tools
as customizable to specific state or local economies.
•
Evaluating the Impacts of State Energy
Efficiency: Extensions
Planned extensions/further research:
• addition of eight more sub-state, multi-county
regions interacting with state-level model,
• linking the econometric model to MARKAL for
addressing energy technology and more accurately
portraying alternative policies as translated into
exogenous policy variables,
• integrating GIS to incorporate land use and
transportation impacts and feedback, and
• eventually extending to other state and subregions.
Evaluating the Impacts of State Energy
Efficiency: Limitations
Limitations:
• the model is based on historical data and
relationships, implying that forecasts are sensitive
to structural shifts,
• this extensive a modeling effort requires significant
resources to construct and test the model, and
• the accuracy of forecasts and policy analysis
depends on the quality of state-level data.
Evaluating the Impacts of State Energy
Efficiency: Interesting Development
We recently completed, in tandem with the
Indiana Business Research Center, an
analysis of a proposed coal gasification plant
for a major Indiana utility using a previous
version of our state model.

similar documents