Home | Business News | Browse by Publication | T | The Energy Journal

Demand-side management and energy efficiency revisited.

Article Excerpt
The key finding of Loughran and Kulick (2004) is that utilities have been overstating electricity savings and underestimating costs associated with energy efficiency demand-side management (DSM) programs. This claim is based on point estimates of average DSM-related savings and costs implied by an econometric model of residential electricity demand. We first argue that the chosen test statistics bias results in favor of rejecting the null hypothesis that utility-reported savings reflect true values. We also note that utility estimates of average program savings and costs are rejected based on point estimates alone. We use the same data and econometric model to estimate the appropriate test statistics. We then construct nonparametric bootstrap confidence intervals. These intervals are quite large; we fail to reject the average electricity savings and DSM costs reported by utilities. Our results suggest that the evidence for rejecting utility estimates of DSM savings and costs should be re-interpreted.

1. INTRODUCTION

As public concerns about climate change and air quality escalate, there is increasing political pressure to find ways to reduce the environmental impacts of energy use. One approach currently being pursued by policymakers involves increasing support for "demand-side management" (DSM) programs. Since the 1970s, utilities in the United States have been implementing DSM programs designed to reduce residential and commercial electricity demand through information dissemination programs, subsidies, free installation of more efficient technologies, and other conservation related activities. Whereas program evaluations routinely find that these utility-sponsored DSM programs are highly cost effective (EPRI, 1984; Eto et al., 1996; Eto et al., 2000; Fickett et al., 1990; Jordan and Nadel, 1993; Nadel, 1992; Nadel and Geller, 1996), in the past some economists have viewed these results with skepticism (Joskow and Marron, 1992; Nichols, 1995). (1) A more recent paper by Loughran and Kulick (2004) has refueled this debate.

The stated objective of the Loughran and Kulick (LK) paper is "to test whether DSM expenditures during the 1990s succeeded in increasing the electricity efficiency of the U.S. economy" (p 21). LK fail to reject this hypothesis, however they do conclude that "DSM (has) had a much smaller effect on retail electricity sales than estimates reported by utilities themselves" (p. 19). This claim has attracted considerable attention. In the two years since its publication, this paper has been cited in a wide range of contexts, including utility revenue requirement hearings (B.C. Commission, 2005), academic papers (Gellings et al., 2007; Gillingham et al., 2006; Metcalf, 2006), policy briefs (Geller and Attali, 2005), and partisan position papers (Crane and Boaz, 2005).

Several authors have pointed out shortcomings of methods used to calculate DSM savings and costs, including the potential for free riding, unmeasured positive spillovers, and moral hazard issues. (2) LK derive their result from a novel approach to addressing the problem of free riders (that is, beneficiaries of a utility DSM program who would have saved energy even in the absence of a DSM program). While questions could be raised about whether LK have successfully dealt with the free rider issue, we do not examine such questions in this response. Instead, our objective is to demonstrate that the empirical evidence provided by the authors is consistent with (rather than contradicts) the findings of past DSM program evaluations. We use a simple hypothesis testing framework to show that DSM savings estimates reported by utilities to the Energy Information Administration (EIA) cannot be rejected even when the data and estimation approach used by LK are taken at face value.

This response proceeds as follows: Section 2 restates the question addressed by LK in terms of a hypothesis test; Section 3 uses the data and econometric models used by LK to estimate the appropriate test statistics; Section 4 reports the results of hypotheses testing; Section 5 concludes.

2. FORMULATING THE NULL HYPOTHESIS

In the past, studies demonstrating the cost effectiveness of DSM programs have relied heavily on cost and savings estimates that the utilities are required to report annually to the Energy Information Administration (EIA). Each year, utilities are not only required to report their annual DSM expenditures (denoted EE) and electricity sales (kWh), but also to estimate the annual savings (s). LK use these data from 324 utilities over the period 1989-1999 to estimate several models of DSM electricity savings. (3)

The first aspect of the LK paper we take issue with is the statistic used to test the stated null hypothesis. In order to test the hypothesis of whether DSM expenditures increased the energy efficiency of the US economy, one needs to consider the percent change in aggregate US electricity consumption due to aggregate expenditure on energy efficiency DSM. We have verified that LK use the average percent change in electricity consumption due to energy efficiency DSM expenditures across utilities and years as their indicator. As we will show below, this choice of test statistic results in an underestimation of percent savings and an overestimation of costs, assuming that we are interested in measuring economy wide savings and costs.

A simple example helps to illustrate this point. Suppose utility A spends $1 on DSM and saves 20 kWh, producing 980 kWh instead of 1,000 kWh in the counterfactual. Now consider utility B, which...

View this article FREE - Now for a Limited Time, try Goliath Business News
Free for 3 Days!