Energy savings can be reduced when customers use the money they save from energy-efficiency measures to purchase increased comfort or convenience. The theory is that once people can afford to use more energy in their homes, they may boost their thermostat settings or increase their lighting levels. This occurrence is commonly known as the "takeback" or "snapback" effect. By the same token, consumers who purchase energy-efficient equipment may purchase larger units or units with more features. Utilities are concerned with takeback because it reduces the savings from their demand-side management programs.

A recent study by Steven Nadel of the American Council for an Energy-Efficient Economy (ACEEE), however, found that while takeback does occur, it is not as widespread as generally believed.* After reviewing 42 empirical studies that dealt with the issue in some way, Nadel concluded that takeback is a local phenomenon largely limited to a few specific end uses.

An analysis of fifteen studies of takeback relating to residential space heating measures painted a consistent picture--takeback for these measures was generally limited. On average, customers showed little if any change in setpoint; all the studies found that some customers increased their setpoints while others decreased them.

Most of the studies attempted to quantify takeback in terms of changes in average indoor temperature or thermostat setting before and after weatherization. Twelve studies quantified estimates of changes in thermostat setpoints, while indoor temperatures were available from nine studies. Results varied from a 0.5deg.F decrease to a 0.9deg.F increase in setpoint after space heating measures were installed. The median increase was 0.25deg.F. Putting this in perspective, Nadel noted that for a city with an average 40deg.F outdoor temperature during the heating season, a 0.25deg.F increase in indoor temperature would increase annual energy use for heating by less than 1%.

A study of temperature takeback in the well-known Hood River Project supported the hypothesis that the phenomenon is more pronounced in low-income houses than in higher-income houses. Low-income households in Hood River increased interior temperatures by an average of 0.8deg.F, while higher-income households increased interior temperatures by an average of roughly 0.2deg.F after space heating measures were installed. Yet in six studies on low-income homes, including Hood River, the median increase in thermostat setpoint or interior temperature after space heating measures were installed was only 0.5deg.F, evidence that takeback is very limited in low-income homes.

Eight studies examined takeback in relation to air conditioning; three covered weatherization measures and five covered improved air conditioner efficiency. One study found no evidence of takeback, two found evidence of limited takeback, one found circumstantial evidence of takeback, and four were inconclusive. Evidence indicated that takeback might be more likely to occur in moderate climates and moderate temperature months.

Residential water heating studies primarily addressed low-flow showerheads, but two focused on rebates for efficient water heaters. Studies relied on in-home measurements, consumer surveys, and utility bills analysis carried out before and after measure installation. In two studies, there was no evidence of takeback, while a third study found limited evidence for very minor amounts of takeback. A study of 286 showerheads by the U.S. Department of Housing and Urban Renewal found that "nonconserving" and "low-flow" showerheads were used for an identical 4.8 minutes per shower, although average water temperature for the low-flow models was 1deg.F higher.

Similarly, a study of Pacific Gas and Electric's (PG&E) Showerhead Coupon Program found little or no takeback after low-flow showerheads were installed. In an on-site survey, program participants said they used their showerheads an average of 7.6 minutes per shower, compared to 8.2 minutes for nonprogram showerheads in participant households, and 8.6 minutes in nonparticipant households. Asked if replacement showerheads were used for more or less time than the ones they replaced, 15% said showers were now shorter, and 5% said they were longer. Average temperatures for all three groups were within 0.4deg.F of each other.

Six studies addressed takeback and residential lighting, but changes in lamp operating hours before and after retrofits were measured through customer surveys, not physical measurements. There was some evidence of moderate takeback, but data were inconclusive.

Results of surveys by PG&E, Boston Edison, and New England Electric Service (NEES) indicated that 8%-32% of compact fluorescent lamps (CFLs) were used more often and 2%-12% were used less often than the bulbs they had replaced. PG&E and Boston Edison collected data on lamp operating hours and calculated average increases of 5% and 12%, respectively. PG&E also collected data on lamps that were replaced when CFLs were installed. On average, 18 W and 27 W lamps displaced 74 W and 88 W lamps (incandescent lamps and CFLs have virtually the same lumen output), indicating that takeback in terms of increased lighting levels was minimal or nonexistent. Nadel found indirect support for these findings in a 1990 impact analysis of the NEES Energy Fitness Program, which installed CFLs in participating homes (see "Energy Fitness Inside View," HE Nov/Dec '92, p.5). There was some evidence of moderate takeback, but again, the data were inconclusive.

Surveys by PG&E and Wisconsin Power and Light addressed takeback in their refrigerator rebate programs. PG&E found that participants were somewhat less likely than nonparticipants to purchase refrigerators with less energy-efficient icemakers and through-the-door features (20%-29% versus 29%-52%, depending on the feature). Nonparticipants also purchased slightly larger refrigerators than participants. In comparing their new units to their old ones, a larger proportion of nonparticipants reported that they had increased their unit size or the number of features--indicating that little takeback occurred. However, 6% of participants said that due to the program, they had purchased larger refrigerators, and 10% said that the program influenced their decisions to purchase more features. In the WP&L study, 6% of program participants reported that without the rebates they would not have purchased refrigerators "of this kind," and 5% said that the rebates induced them to purchase larger refrigerators. Thus, limited data on refrigerator programs provides conflicting information, but generally indicates that if any takeback is occurring, it is very limited.

Summing it up, Nadel found little if any takeback in residential space heating; no evidence of takeback in residential water heating studies; that data were inconclusive for other end uses; that takeback was unlikely for residential refrigerators; but that it may occur in residential air conditioning.

-- Cyril Penn

* See "The Takeback Effect: Fact Or Fiction?" by Steven M. Nadel, published in the Proceedings of the 1993 Energy Program Evaluation Conference, held in Chicago.