Wednesday, March 1, 2017

Do Energy Efficiency Investments Deliver at the Right Time?

Abstract:
Electricity cannot be cost-effectively stored even for short periods of time. Consequently, wholesale electricity prices vary widely across hours of the day with peak prices frequently exceeding off-peak prices by a factor of ten or more. Most analyses of energy-efficiency policies ignore this variation, focusing on total energy savings without regard to when those savings occur. In this paper we demonstrate the importance of this distinction using novel evidence from a rebate program for air conditioners in Southern California. We estimate electricity savings using hourly smart-meter data and show that savings tend to occur during hours when the value of electricity is high. This significantly increases the overall value of the program, especially once we account for the large capacity payments received by generators to guarantee their availability in high-demand hours. We then compare this estimated savings profile with engineering-based estimates for this program as well as a variety of alternative energy-efficiency investments. The results illustrate a surprisingly large amount of variation in economic value across investments.
...
Before smart meters and other advanced metering infrastructure, it was impossible to measure policy impacts at the hourly level.... Meters were only read once per billing cycle. This situation is rapidly changing. Today in the United States more than 40% of residential electricity customers have smart meters, up from less than 2% in 2007. ...
Savings are strongly correlated with the value of electricity, making the program 48% more valuable than under a naive calculation ignoring timing. As we demonstrate, including capacity payments is important in this calculation. Most of the value of electricity in ultra-peak hours is captured by forward capacity payments to generators to guarantee their availability
...
Across six major U.S. markets, we find that air conditioning investments are on average 29% more valuable than under a naive calculation ignoring timing. 
For commercial and industrial heat pumps and chillers the "timing premiums" are 29% and 
25%, respectively. Other investments like refrigerators and freezers have timing premiums 
below 5% because savings are only weakly correlated with value. Lighting also does surprisingly poorly, reflecting that savings occur disproportionately during evening and winter 
hours when electricity tends to be less valuable.
...
Electric utilities in the United States, for example, spent $36 billion on energy-efficiency programs between 2006 and 2015, leading to more than 1.5 million gigawatt hours in reported total electricity savings. In addition, the U.S. Federal government has spent $12 billion since 2009 on income tax credits for residential energy-efficiency investments
...
Nine new standards promulgated by the DOE in 2016 achieve a total present value of $76 billion in energy cost savings, vs. $28 billion in avoided CO2 emissions and $5 billion in avoided NOx emissions.7 That is, more than two-thirds of the benefits come from private energy cost savings. Moreover, the hourly variation in external costs is small relative to the hourly variation in electricity prices and capacity values. Private value varies across hours by a factor of ten or more, while emission rates vary only by about a factor of two between fossil-fuel plants. 
...
Our empirical application is an energy-efficiency rebate program offered by Southern California Edison (SCE), a major investor-owned utility.... known as the Quality Installation Program. It provides incentives of up to $1,100 to households that install an energy-efficient central air conditioner.... Air conditioning is responsible for 10% of average residential electricity use and 15% of average commercial electricity use in California... Air conditioning is projected to be one of the fastest growing uses of electricity worldwide ....


The state utility commission compensates SCE for running the program by allowing the utility to pass on costs to ratepayers in the form of higher electricity prices. The Quality Installation Program includes an additional focus on proper installation of the new subsidized central air conditioner, which can further improve energy performance. 

...
The event study figure for summer shows a sharp decrease in electricity consumption in the year in which the new air conditioner is installed. The magnitude of the decrease is about 0.2 kilowatt hours per hour... As expected, winter consumption is essentially unchanged after the new air conditioner is installed.
...
On mild days, between 50 and 70 degrees Fahrenheit, estimated energy savings are zero or not statistically distinguishable from it.... From 70 to 100+ degrees, there is a steep, continuous relationship between temperature and energy savings, as expected from a new air conditioner. Air conditioner usage is largest on the hottest days, so energy-efficiency gains have the largest effect on these days. There is also a small decrease in consumption on days below 50 degrees following air conditioner replacement. This may be explained by improvements to ductwork, insulation, thermostats, or other HVAC-related upgrades that could in some cases occur as part of a central air conditioner replacement
...
During July and August there are large energy savings, particularly between noon and 10 p.m. Savings reach their nadir in the summer at 6 a.m. which is typically the coolest time of the day. During non-summer months savings are much smaller, less than 0:05 kilowatt hours saved on average per hour, compared to 0:2 to 0:3 kilowatt hours saved on average per hour during July and August.... The implied annual savings per household are 375 and 358 kilowatt hours per year, respectively.... Prior to installing a new air conditioner, program participants consumed an average of 9,820 kilowatt hours annually, so this is a 4.5% decrease in household consumption. A typical central air conditioner (3 ton, 13 SEER) in this region uses about 4,237 kilowatt hours per year, so the savings represent a 10% decrease in annual electricity consumption for air conditioning. This is broadly similar to, but slightly less than, what would be expected based on a simple engineering prediction. For example, a Department of Energy calculator shows that ignoring rebound and other factors a typical central air conditioner upgrade in Los Angeles saves 565 kilowatt hours per year.... Using hourly microdata, household by month-of-year by hour-of-day fixed effects, hour-of-sample by climate zone fixed effects, and the sample exclusions the estimate of annual program savings is 442 kilowatt hours per year.  16
...
Capacity payments are made to electricity generators to remain open and available, thereby ensuring desired reserve margins. Capacity costs are zero or close to it during off-peak hours because electricity demand can be easily met by existing inframarginal generators (plants that are not close to the margin between staying in the market and exiting). However, during peak hours large capacity payments are required to ensure desired reserve margins. 

Incorporating capacity values substantially increases the value of electricity during peak periods. In California during August, for example, capacity values increase the value of electricity during peak evening hours to between $200 and $600 per megawatt hour. And, overall, the pattern is very similar across the four approaches for allocating capacity value across hours....
The value of electricity in Texas surges in August to $200+ during the late afternoon, considerably higher than the marginal cost of any generator.
...
The calculations which follow also account for line losses in electricity transmission and distribution. In the United States, an average of 6% of electricity is lost between the point of generation and the point of consumption ..., so 1.0 kilowatt hour in energy savings reduces generation and capacity requirements by 1.06 kilowatt hours. Line losses vary over time by an amount approximately proportional to the square of total generation. We incorporate these losses explicitly following Borenstein (2008) and, in practice, they range from 4.4% during off-peak periods to 8.4% during ultra-peak periods. Incorporating line losses thus further increases the variation in economic value between off-peak and peak  
...
The predicted savings when energy prices are $55/MWh are twice as large as predicted savings at $35/MWh. The fitted line for winter, however, is essentially flat. This energy-efficiency investment delivers essentially zero electricity savings in all hours during the winter, so there is little correlation between savings and prices
...
Under this calculation that accounts for timing, the average value of savings is $45 per megawatt hour. This is 17% higher than the naive value estimate using average annual prices and ignoring timing.
..
Incorporating capacity values significantly increases the value of air conditioning investments to $68 per megawatt hour. This reflects the positive correlation between electricity savings and peak hours. Air conditioning investments save electricity during the hours-of-day and months-of-year when large capacity payments are needed to ensure that there is sufficient generation to meet demand. The naive calculation that ignores timing greatly understates these capacity benefits. The naive estimate ... increases only modestly from $39 per megawatt hour to $46 per megawatt hour after including capacity value. This reflects the fact that most hours have zero capacity value, so while a few peak hours have capacity values well above $100 per megawatt-hour the average across the year is only about $6.50 per megawatt hour.

Exactly how we account for capacity values has little impact, changing the estimated timing premium only slightly.
...
Savings profiles for eight different investments ... are remarkably diverse. The flattest profile is residential refrigeration, but even this profile is not perfectly flat. Savings from residential lighting investments peak between 8 p.m. and 9 p.m. all months of the year, while savings from residential heat pumps peak at night during the winter and in the afternoon during the summer. The non-residential profiles are ... quite different from the residential profiles. Whereas savings from residential lighting peak at night, savings from commercial and industrial lighting occur steadily throughout the business day. Commercial and industrial chillers and air conditioning follow a similar pattern but are much more concentrated during summer months.... Savings from commercial and industrial heat pumps are assumed to peak only in the summer, unlike the residential heat pumps for which the engineering estimates assume both summer and winter peaks.
...
Air conditioning investments in California and Texas have the highest timing premiums. This is true regardless of whether the econometric or engineering estimates are used, and reflects the relatively high value of electricity in these markets during summer afternoons and evenings. In other U.S. markets air conditioning has a timing premium greater than zero, but nowhere else is the value as high as in California and Texas.
...
Other investments also have large timing premiums. Commercial and industrial heating and cooling investments, for example, all return premiums of about 25%, reflecting the relatively high value of electricity during the day.  

The timing premiums of other investments, like refrigerators and freezers, are much lower. The savings from these investments are only weakly correlated with system load. Lighting, as well, does surprisingly poorly as the savings occur somewhat after the system peak in all U.S. markets and disproportionately during the winter, when electricity tends to be less valuable. This could change in the future as increased solar generation moves net system peaks later in the evening, but for the moment both residential and non-residential lighting have timing premiums of about 10% or below in all markets.

by Judson P. Boomhower and Lucas W. Davis
National Bureau of Economic Research (NBER) www.NBER.org
NBER Working Paper No. 23097; Issued in January 2017

No comments:

Post a Comment