An analysis of the ENERGY STAR® program in Alachua County, Florida

Abstract

ENERGY STAR® certification, as a voluntary label, represents a residence that is designed and built to use 30% less energy than its counterparts. We examine the effectiveness of this program using monthly residential energy consumption data for residences in Alachua County, Florida between 2000 and 2013. Our sample represents about 25% of the ENERGY STAR® qualified homes in the area. We use panel models and a modified repeat sales approach to estimate energy savings of ENERGY STAR® residences relative to Non-ENERGY STAR® ones, while controlling for the bundle of house attributes, spatial and temporal fixed effects, changes in the Florida State Building Code (FBC), and household changes. Our results indicate that ENERGY STAR® residences have a long term, greater energy efficiency savings over Non-ENERGY STAR® houses. Thus, the ENERGY STAR® residential program can be seen as an environmentally conscious housing practice that addresses economic and environmental dimensions.

Introduction

In the U.S., as with many countries, energy efficiency has become a focus for consumers due to rising energy costs. Current energy consumption (electricity and gas) per household has increased by 5% from 2000, with the average monthly residential bill increasing by 30% across the same period (U.S. EIA, 2009). Energy efficiency policies, products, and practices have been initiated in an effort to reduce greenhouse emissions and energy costs, and to increase environmental protection. Policymakers throughout the world have used two strategies to encourage energy efficiency and energy savings in residential homes. The first strategy is to set minimum efficiency standards for new residential construction: these requirements have typically become more stringent over time (Locus, 2007, Jacobsen and Kotchen, 2013, Tulsyan et al., 2013). The second strategy is the use of voluntary programs such as the ENERGY STAR® new home certification program implemented by the U.S. Environmental Protection Agency (EPA) and the US Department of Energy (DOE) (Doris et al., 2009, ENERGY STAR® website, 2015c, Taylor et al., 2014). Voluntary energy certification programs have also been implemented in other countries to promote energy efficiency of residential and commercial buildings (BSI, 2008, Casals, 2006, EU (European Union), 2003, Janda and Busch, 1994, Rey et al., 2007, Yao et al., 2005). Examples are the Energy Performance of Buildings Directive (EPBD) adopted in Europe in late 2002 to provide an energy efficiency framework (Casals, 2006), the United Kingdom's Home Information Pack (HIP) for the home buying and selling process (BSI, 2008), and Australia's index of building's thermal performance standardized through the Building Code of Australia (BCA) (Greenwood, 2012).

The ENERGY STAR® certification program was introduced to the U.S. residential sector in 1995. If the homebuilders met the requirements, they could certify new homes with the ENERGY STAR® label to indicate that these residences were 30% more energy efficient than a reference home, that is, “…a designed-model home of the same size and shape as the actual home” (RESNET, 2016; see also Pigg, 2002). ENERGY STAR® homes are designed for energy efficiency, including features such as low emissivity (low-e) windows, high thermal resistance-value (R-value) insulation in the wall and attic, tankless water heaters and heating, and ventilation and air conditioning (HVAC) systems that make them more energy efficient (Smith and Jones, 2003, Pigg, 2002, ENERGY STAR® website, 2015c). To obtain the ENERGY STAR® residential certification, a home must undergo a process of inspections, testing, and verification to meet the requirements set by the ENERGY STAR® certification program. Since the beginning of the program, over 1 million ENERGY STAR® certified new homes have been built in the United States (ENERGY STAR, 2015a). Although the technical predictions indicate that homes with the ENERGY STAR® features will save energy relative to baseline homes without them, ultimately, the most important aspect is the amount of energy these features actually save. Given the increasing scale of the ENERGY STAR® certification program, it is important to increase the existing empirical evidence on its performance.

Previous attempts to evaluate the energy efficiency savings on energy consumption offer a suggestive picture that ENERGY STAR® residences consume less energy on average than Non-ENERGY STAR® residences. However, these studies consistently find that the energy savings of ENERGY STAR® residences is less than the 30% benchmark. For example, Smith and Jones (2003) find that ENERGY STAR® homes consumed on average 12% less electricity and gas than Non-ENERGY STAR® homes in Alachua County, Florida in 2000 and 2001. In follow-up studies, Jones and Vyas (2008) and Jones et al. (2010) find that ENERGY STAR® homes consume between 10 and 15% less electricity and gas than Non-ENERGY STAR® residences in the same county in 2006. Taylor et al. (2014) analyze ENERGY STAR® homes in Alachua county over 2000–2010, finding that their average energy savings decreased over time from 18% in 2000 to 7% in the period 2005–2010. Pigg (2002) also finds that, on average, ENERGY STAR® homes in Wisconsin consume 9% less natural gas than their counterparts during 1999 and 2000. However, Pigg (2002) also finds that the estimated difference in electricity use was not statistically significant.

In comparing ENERGY STAR® and Non-ENERGY STAR® houses, the previous studies account for differences in a number of house characteristics that impact energy consumption. At the same time, differences in consumer behavior have been difficult to account for given limited information about key variables that affect residential energy consumption such as income, expenditure classifications, household demographics, etc. For instance, consumer behavior could reduce the amount of gains from new technologies that increase the efficiency of resource use. This systematic behavior has been often referred as the rebound effect or energy paradox (Alcott, 2005), which tends to offset the beneficial effects of the new technology or other measures taken.

Nevertheless, the existing empirical evidence on rebound effects is mixed. After reviewing existing estimates of the rebound effects in household energy consumption in the literature, Greening et al. (2000) find that increases in efficiency, although partially offset by increases in consumption, result in between 10 and 30% reduction in energy consumption. Hassett and Metcalf (1995), analyzing energy tax credits and energy conservation, point out that, after controlling for unobserved heterogeneity, the rebound effect is smaller than previously suggested by past research. Dubin et al. (1986) find that the use of energy-efficient appliances results in smaller energy savings relative to engineering projections. They find that actual energy conservation is as much as 13 percentage points below engineering estimates for cooling and 8 to 12 percentage points below for heating. Vassileva et al. (2012) identify that, beyond building characteristics, socioeconomic household characteristics (particularly income), usage and possession of appliances, and knowledge and attitudes towards energy consumption, can lead to a decrease of electricity consumption of 32.5% for rented apartments in Sweden. Qiu (2014) analyzes the rebound effects of the adoption of energy efficient technologies in commercial buildings and concludes that energy efficiency can reduce electricity use by about 35% and natural gas consumption by about 50%.

Another relevant aspect to consider when evaluating the ENERGY STAR® program is the concurrent effect that building codes may have on energy savings, particularly in the case of Florida, where building codes have been evolving frequently. Locus (2007) focuses on the energy efficiency improvements in new residential buildings during the Katrina reconstruction process in the U.S. states of Louisiana and Mississippi. This report examined five different building level energy efficiency alternatives, including older vintage housing, the 1995 Model Energy Code (equivalent to current practice), the 2006 International Energy Conservation Code, ENERGY STAR®, and energy tax credits. Locus finds that an energy cost savings of about 45% over that set by the 1995 Model Energy Code could be achieved by meeting ENERGY STAR® Home specifications. Such findings have also been found internationally, for example, Tulsyan et al. (2013) analyzed the potential energy savings from the Energy Conservation Building Code (ECBC) for six different type of buildings in Jaipur, India. Buildings with ECBC compliance experience energy savings that vary from 17% for governmental buildings to 42% for hospital buildings. This means that by making building codes more stringent, homes will consume energy more efficiently. For instance, after the 2001 Florida Building code change, homes use 4% less electricity and 6% less natural gas than homes built before the 2001 change (Jacobsen and Kotchen, 2013). As a result, energy savings by ENERGY STAR® houses may be eroded over time if the higher stringency of building codes is not taken into consideration.

Our paper analyzes the energy consumption of ENERGY STAR® and Non-ENERGY STAR® residences in Alachua County, Florida between 2000 and 2013. Our contribution to the energy efficiency literature is twofold. First, we use new panel data on household-specific energy consumption to examine the energy savings effects of ENERGY STAR® over time. We do so by identifying the effect of the ENERGY STAR® label on monthly energy consumption while controlling for changes in building codes, as well as for households moving in and out of ENERGY STAR® residences and their counterparts. This is important since most of the previous work on residential energy consumption assumes that the same household resides in a given residence for the entire period in which energy consumption is observed. Second, using a modified repeat sales approach, we control for time-invariant home characteristics that could potentially bias cross-sectional estimates. Most importantly, we are able to control for time-invariant household characteristics in estimating the energy efficiency savings for ENERGY STAR® houses. Our estimates contribute to the important question of whether the ENERGY STAR® label is effective in providing informational and energy savings benefits.

Computer-based building energy simulation and annual community baselines (ACB) predictions are often used to assess and develop energy labels (Rey et al., 2007, Taylor et al., 2014). In a recent and related paper addressing a similar question and using similar data on Alachua county, Florida, Taylor et al. (2014) employ annual community baselines (ACB) that consist of predictions (based on observables) of energy use by conventional homes in the same location and year, which are then compared to the predictions of energy use by ENERGY STAR® homes using the same predictive model. They find that the energy savings of ENERGY STAR® homes have decreased from 18% in 2000 to an average of just 7% in the five years spanning 2005–2010. The present work complements Taylor et al. (2014) by employing a methodology (a hedonic longitudinal-data model of energy use) that accounts not only for observable factors, but also for unobserved factors that vary over neighborhoods, time, and neighborhood-and-time (through the use of “fixed effects”). Furthermore, this study controls for changes in the Florida building codes (which have become more stringent over time), and considers a specification that holds constant the household that inhabits the house. All of these additional factors are potential sources of energy-use variability that can be erroneously attributed to the effects of ENERGY STAR®.

The rest of the paper proceeds as follows. In Section 2 we provide details on the residential ENERGY STAR® program. Section 3 describes our area of study and data. Section 4 presents our econometric methodology. Section 5 discusses the results and implications of our findings; and Section 6 presents additional discussion and conclusions.