New York Negawatts

Deck: 

Balancing risks and opportunities in efficiency investments.

Fortnightly Magazine - January 2010

In June 2008, the New York Public Service Commission (PSC) established the electric energy-efficiency portfolio standards for New York’s investor-owned utilities.1 In its order, the PSC directed utilities to file three-year energy-efficiency plans which, once implemented, would generate cumulative savings of nearly 2.1 GWh from 2009 to 2011 at a total cost of about $518 million. Later that year, the PSC issued a supplemental order approving shareholder incentives for utilities successfully implementing their portfolios.2 If all goes according to plan, the six affected IOUs stand to earn about $27 million annually in performance incentives over three years (see Figure 1).

The structure of the incentive mechanism approved by the PSC presents risk factors that might affect utilities’ ability to realize the full earning potentials the mechanism offers.

The Case for Incentives

The PSC order was issued in August 2008, 21 years after the July 1989 resolution of the National Association of Regulatory Commissioners’ (NARUC), which recognized the earnings implications of conservation and least-cost planning for utilities.3 The NARUC resolution recognized the need for removing obstacles standing between the idea of least-cost planning and the practical realities of traditional ratemaking, which created a strong economic disincentive to the utilities’ implementation of it. The resolution urged NARUC member state commissioners to: consider the loss of earnings potential associated with conservation and demand-side management; adopt appropriate ratemaking mechanisms to encourage utilities to help their customers improve end use efficiency cost-effectively; and ensure the successful implementation of a utility’s least-cost plan as its most profitable course of action.4

The resolution proved popular and effective. Within four years, by the end of 1993, shareholder incentives had been approved for over 50 utilities in 20 states, including Con Edison and NYSEG.5 The hallmarks of these early incentive structures were assured recovery of prudent energy-efficiency investments, and financial incentives to offset the loss of revenues resulting from reduced sales.

However, as utilities in many regions dealt with regulatory reform and restructuring of the electric utility industry in the mid 1990s to early 2000s, these schemes were abandoned in most jurisdictions. Uncertainties related to restructuring in general and the potential risks of stranded investment resulting from open access (in particular) led many utilities to cut back, or in many cases, to altogether halt their energy-efficiency investments. The low activity levels and dwindling new investments in energy efficiency lessened the importance and, to some extent, relevance of shareholder incentive mechanisms.

The past few years have seen a significant resurgence in energy efficiency for a number of reasons, mainly due to a renewed interest in integrated resource planning and the adoption of energy-efficiency portfolio standards (EEPS) in many states given concerns over climate change or energy security. These developments have prompted many regulatory commissions to reconsider energy-efficiency incentives. Based on recent data compiled by the Edison Electric Institute, incentive formulas have been approved in 18 states and are pending approval in seven more.6

Incentive Structures

Incentive schemes in effect today fall into four primary classes: simple mark-ups, which award the utility a percentage of spending; bonuses, which reward achieved energy or energy and capacity savings; shared savings mechanisms, which reward net benefits; and rate-of-return kickers, which allow the utility to seek a higher-than-allowed return on equity for approved energy-efficiency program costs. Hybrid mechanisms also are available that combine features of these four schemes to achieve multiple goals.7

After considering alternative incentive proposals, the PSC adopted a bonus scheme, arguing that the approach provides a “reasonable balance of risk and opportunity” for utilities to engage in energy efficiency. The mechanism is structured as a three-tiered scheme, allowing rewards or imposing penalties, depending on the percent of annual saving targets achieved (see Figure 2). Utilities are penalized if their portfolio savings fall below 70 percent of the target, in proportion to their performance. No awards or penalties accrue if savings fall in the dead-band region between 70 and 80 percent of the established targets. Utilities are rewarded if their portfolio savings exceed 80 percent of the target in proportion to their performance. Performance awards are capped at 100 percent of the target, and penalties are limited to performance levels below 50 percent of the target.

The mechanism is structured as a simple bonus scheme, designed to provide a reasonable level of incentives, while limiting the regulator’s upside risk by imposing an earnings cap. It also limits the utility’s downside risk with a penalty cap. The PSC rejected an incentive mechanism based on utility expenditures, arguing such an approach could lead to utilities artificially inflating their costs and produces no incentives for innovation and efficiency in operating programs. The PSC also dismissed a shared savings structure on the grounds that it could lead to utilities pursuing least-cost measures, thus creating lost saving opportunities. The PSC didn’t entertain a possibly superior hybrid mechanism, such as that adopted in California, which combines bonus and shared-saving schemes, thus pursuing the dual goals of encouraging greater savings and higher cost-effectiveness.8

Risks in Performance

If all goes according to plan, the six IOUs stand to earn about $27 million annually in performance incentives over three years, or $38.85/MWh. Things, however, seldom go as planned. The success of an energy-efficiency program in realizing its saving target depends on multiple factors, many of which are uncertain and fall beyond the utility’s control. These factors may be classified roughly into three classes of risks, depending on whether they relate to a program’s expected or ex ante savings, gross savings, or net savings.9 These three classes include market-acceptance risks; measure-performance risks; and consumer-behavior risks.

Market-acceptance risks arise from economic conditions and considerations that might prevent consumers’ willingness—and ability—to participate in programs offered by the utility and adopt energy-efficiency measures. These factors directly will affect a program’s expected or ex-ante savings. Although improved program design, enhanced marketing, and higher incentives may help overcome some of these barriers, achieving expected market penetration largely remains a function of customers’ willingness and ability to participate in programs.

Measure-performance risks are technical in nature and related to factors affecting how an energy-efficiency measure functions and delivers savings, such as poor quality, improper installation or operation, and measure interactions. These factors generally affect a program’s gross savings. Careful assessment of energy-efficiency technologies, commissioning protocols, and more rigorous measurement and verification improve measure performance. Still, as many energy-efficiency program evaluations have shown, they rarely, if ever, eliminate technical risks.

Finally, behavioral uncertainties mainly arise from potential free-ridership (i.e., savings that would have occurred without the program) and rebound or take-back effects. The combined effects of these risk factors will determine the success of the New York’s six IOUs to achieve their energy-efficiency portfolio objectives and, ultimately, earn the incentives.

Given the range of uncertainties about the actual savings of energy-efficiency programs, it’s reasonable to analyze their outcomes in probabilistic (stochastic) terms, depending on the likely range of possible values of the risk factors. For this analysis, three variables—market acceptance, measure-savings realization rates, and net-to-gross ratios—were used to analyze the potential risks that might adversely affect a utility’s ability to reach its annual targets and earn an incentive. Figure 3 shows the assumed probability distributions for the three risk variables. For each of the three stochastic variables, the table shows five points on the variable’s probability distribution. The upper and lower bounds show the possible maximum and minimum values for each variable. The 25 percent, 50 percent, and 75 percent are, respectively, the first quartile, the median, and the third quartiles of distribution. For example, the value of 0.98 for market acceptance in the residential sector means there is a 25- percent probability a residential program will reach 98 percent of its expected participants.

The numbers in Figure 3 represent likely values for typical programs. They reflect the best knowledge about the variability of these factors, based on information available from evaluation reports on energy-efficiency programs being run in several states. Changes in these assumptions clearly will affect the results.

Performance of individual portfolios of the six IOUs was simulated 200 times using the Monte Carlo simulation. The procedure generated probability distributions for savings and potential incentive earnings for each utility, by customer sector and year, for each of the 200 iterations. Values for market acceptance, measure performance, and net-to-gross were drawn for each utility and iteration from distributions in Figure 1. The resulting distributions were then aggregated across all utilities, and summarized in histograms by dividing the range of potential outcomes into bins, then counting the number of observations falling within each bin. In addition, smoothed probability density functions (in red) were generated by nonparametric kernel density estimation (see Figures 4 and 5). For each distribution, the mean, standard deviation, median, first and third quartiles, and fifth and 95th percentiles of the distributions are shown in the insets.

Figure 4 shows the probability distribution of cumulative annual energy savings between 2009 and 2011, relative to the savings target (2,082 GWh) for the six utilities. As the inset in Figure 4 shows, the utilities are likely to reach at least 92 percent of their targets on average, assuming the same performance levels historically observed for energy-efficiency programs. There’s a 5-percent probability that savings would exceed the targets, and a 5-percent probability that the utilities will achieve only 81 percent of their saving targets collectively. The results also show considerable opportunities for achieving greater-than-expected savings, as probabilities associated with both tails of distribution are significant; the upper quintile of savings lies above 2,150 GWh.

Shareholder Earnings Potential

The distributions of shareholder incentives relative to maximum-allowed earning potentials and actual amounts are shown in Figure 5 in relative and absolute terms. The results suggest utilities will, in all likelihood, earn at least 63 percent of the allowed maximum incentives, and the probability of an imposed penalty being nearly zero, assuming typical performance by the six utilities. On average, the six utilities stand to earn approximately $51 million in net present value terms over the course of their three-year plans. This figure lies slightly below the $53 million median and significantly below the $81 million maximum possible earnings.

In contrast to the distribution of energy savings, measures of the central tendency of earnings lie significantly below the maximum (target) given how the incentive mechanism is structured. While energy savings can exceed the target in any year, total annual shareholder incentives are capped at $27 million at the target. As a result, while above-target savings in one year might offset below-target savings in a later year, savings aren’t allowed to carry over to later years and be rewarded.

Distribution of potential earnings shows significant variability. The standard deviation is estimated at $16 million, which is 20 percent of mean savings. As in Figure 5, the distribution of earnings is asymmetrical and negatively skewed. In addition, the inter-quartile range of earnings lies between $41 million and $61 million. Utilities face a 5-percent probability of earnings falling below $22 million and an equal likelihood of earning more than $75 million annually. The probability of a penalty is close to zero.

Fair or Efficient?

The New York shareholder incentive mechanism has four features with significant implications in terms of its effectiveness in encouraging energy-efficiency investments and how such investments will be rewarded. First, incentives are tied to annual, rather than cumulative, targets. Utilities qualify for incentives according to the savings level achieved each year, relative to the established annual target. Second, incentive payments are capped once savings targets are reached. Third, a dead band occurs between 70 percent and 80 percent of savings, when utilities earn nothing for their efforts. Finally, potential penalties stop to accrue once a utility’s performance falls to 50 percent of the annual target.

Basing incentives on annual targets is problematic because it doesn’t allow for smoothing random variations in savings between years. A utility might make the optimal amount of investment in energy efficiency and achieve the three-year savings target, yet fall short of the annual target in one or more years, thus earning less than the maximum due to factors beyond its control.10 Comparing the savings distribution (see Figure 4) and incentives (see Figure 5) illustrates the inherent bias in the incentive mechanism’s structure. Figure 4 shows median total incremental savings are 93 percent of targeted savings, and the distribution of savings is fairly symmetric around the median. However, the distribution of incentive payments is skewed to the left and concentrated at incentive levels significantly less than the possible maximum. The median incentive is only 65 percent of the maximum. This suggests the utilities probably will come close to achieving the cumulative savings goal, but will earn incentives significantly less than the maximum. A fairer system would reward utilities on the basis of their cumulative savings performance.

The cap on incentive payments potentially could discourage investment in energy efficiency as expected savings approach the target. The mechanism’s maximum incentive and penalty cap are intended to simultaneously reduce risks to regulators of unexpected over-performance and to limit the liability of utilities for under-performance. In structuring the incentive mechanism, the PSC intended to create a “reasonable mix of risk and opportunity.” However, while balancing risks, the cap discourages additional investment in energy efficiency as expected savings approach the target. Because of the random nature of energy savings, the savings amount may exceed the target in a year. Excess energy savings would cause further erosion of revenues, without an opportunity for additional earnings to offset them. The cap limits what utilities can earn to compensate them for investing in energy efficiency, and creates investment disincentives as expected savings approach the target. Finally, the dead band between 70 percent and 80 percent of savings discourages energy-efficiency investments.

The mechanism’s intended balance and the fairness it implies have merits. Clearly, an unreasonably high penalty cap could impose some risk on the regulator that the utility will discontinue all investments in energy efficiency once it falls within the penalty region. This analysis shows a very low probability (i.e., less than 2 percent) that three-year savings will fall below 50 percent of three-year targets. In the small number of cases where they do, they average a relatively high level of 45 percent. The cap on incentives, on the other hand, limits the earnings potential for performance significantly above expectations. The analysis results show, assuming typical performance for the six IOUs, a significantly higher probability (e.g., 25 percent) that utilities will exceed their targets, but will receive no reward for their superior performance.

 

Endnotes:

1. New York Public Service Commission Case 07-M-0548, Energy Efficiency Portfolio Standards, and Order Approving Programs, issued June 23, 2008. In May 2009, the Commission issued a similar order that established targets and standards for natural gas efficiency programs.

2. New York Public Service Commission Case 07-M-0548, Proceeding on Motion of the Commission Regarding an Energy Efficiency Portfolio Standard, Order Concerning Utility Financial Incentives, Aug. 22, 2008.

3. Resolution in Support of Incentives for Electric Utility Least-Cost Planning; issued by the NARUC Conservation Subcommittee and passed by the Executive Committee during the Summer Committee Meeting on July 27, 1989 in San Francisco.

4. For the text of the resolution and the rationale for shareholder incentives in general, see David Moskovitz, “Profits and Progress Through Least-Cost Planning,” Sponsored by the National Association of Regulatory Utility Commissioners, Washington D.C., November 1989.

5. For a description of early incentive designs and a discussion of their merits, see Stoft, S, J. Eto and S. Kito, DSM Shareholder Incentives: Current Designs and Economic Theory, Energy & Environment Division, Lawrence Berkeley Laboratory, University of California, Berkeley, California, January 1995.

6. Performance Incentives for Energy Efficiency Programs by State, The Edison Foundation, Institute for Electric Efficiency, April 2009.

7. Joe Bryson, Aligning Utility Incentives with Investment in Energy Efficiency: A Resource for the National Action Plan for Energy Efficiency, U.S. Environmental Protection Agency, November 2007.

8. For a thorough discussion of the California demand-side management incentive mechanism, see Jiong Eom and James L. Sweeney, “Shareholder Incentives for Utility-Delivered Energy Efficiency Programs in California,” Precourt Energy Efficiency Center, Stanford University, Stanford, CA, April 2009.

9. Savings from energy-efficiency programs typically are measures at three levels: ex ante, or as designed; “gross” savings, or ex ante estimates adjusted for the actual conditions; and “net” savings—gross savings adjusted for market effects such as free-ridership, spillover, and take-back.

10. As a simple example, if a utility has a 90-percent probability of achieving its annual savings target in each year, and savings in each year are determined independently, there would be only a 73-percent probability of the utility reaching its target in every year and earning the maximum award.