In this section, we describe the mechanisms by which the benchmarking and disclosure laws might have an effect on energy use, the limitations of the laws, and the kinds of data and statistical techniques needed to conduct an evaluation of their effect on energy use.
We see three ways in which the laws may directly lead to reductions in energy use and emissions. First, if market participants are currently inattentive to energy costs, the simple act of entering energy use and building characteristics into Portfolio Manager may bring energy issues into focus for building owners and lead to some changes in building operations to lower energy costs and changes in contract structures to address those costs. As we explained above, the extent to which inattention exists for commercial buildings is an open question, but some studies have identified it as a problem in other settings. Moreover, peer effects have also been shown to influence energy consumption (Allcott 2011; Costa and Kahn 2013, Ayres et al. 2013), thus if building owners see their energy use benchmarked against other buildings, this may reinforce the attentiveness effect.
Second, if tenants prefer to lease space in more efficient buildings and the disclosure laws provide new energy information to the marketplace, this could lead to improvements in efficiency. Prospective tenants may get value from both private and public good aspects of energy efficiency (Kotchen 2006). In terms of private benefits, tenants may prefer to rent in efficient buildings in order to lower their energy bills or because they are more comfortable. But prospective tenants may also have “green” preferences. Such preferences have been found by Kahn (2007) to exist in the market for hybrid cars and by Kotchen and Moore (2008) in the market for green electricity. Building owners would respond to these market pressures by making improvements and retrofits as a means of competing for tenants.Footnote 14 Research on commercial building certifications and real estate values suggests that both Energy Star and LEED certified buildings have higher rental values and higher sales prices than non-certified buildings (Eichholtz et al. 2010, 2013). Studies of commercial buildings in Europe have reached similar findings (Kok and Jennen 2012) and an analysis of apartment building sales in Singapore finds that green-certified properties sell at a premium (Deng and Wu 2014).
A third way that the requirements may have an effect is through investor behavior. Many commercial buildings are owned by real estate investment fiduciaries or real estate investment trusts (REITs). REITs are similar to mutual funds and are traded on public stock exchanges. Investors could prefer more efficient buildings because the lower energy costs increase net income, because of “green” preferences, or as a quality signal to prospective tenants. This increased demand by investors could drive up the value of more efficient buildings.Footnote 15 The market for REITs may already be moving in this direction. In late 2012, the National Association of Real Estate Investment Trusts (NAREIT), the US Green Building Council, and FTSE Group, a British provider of stock market indices and related services, announced a new green property index (Thomas 2012). While the index will be based on LEED and Energy Star certification, it is possible that the next step could be an index based on data from disclosure requirements.
There are several reasons to be cautious about the ability of these requirements to provide significant reductions in energy use. In some cities (Austin, Berkeley and Seattle), energy use information is not being made available to the public, but only to tenants, prospective tenants, and others involved in real estate transactions. Having the information readily available to the public, such as on a government website, is preferable. Even in these cases, though, it is not clear how helpful the information disclosed is to prospective tenants trying to choose space to lease based on expected energy costs. In New York, data by building is available in both Excel spreadsheet form and on the NYC Open Data platform and includes annual average source and site energy use intensity, along with an Energy Star score. The Energy Star score is an index useful only for comparison among similar types of buildings. Moreover, in a large building, the average energy use intensity also may not be that helpful, as it may not be representative of the particular space a prospective tenant is considering leasing. The EUI provides only a rough indicator of expected energy costs, which is the information the tenant needs for decision making. In cities that use Portfolio Manager, building owners must report energy use separately for natural gas, electricity, and other fuels, and in Washington and Philadelphia, this detailed information is included in the public disclosure. In our view, this is an improvement, as prospective tenants can use local prices to estimate costs and compare the numbers with those on their current utility bills.
In most cities, building owners are required to report whole building energy use, and in Atlanta, Berkeley, Boston, Cambridge, Chicago, Kansas City, Minneapolis, Montgomery County, New York, Philadelphia, Portland, Seattle, San Francisco and Washington, nonresidential tenants are required to provide the data to their landlords. Obtaining information from tenants can be difficult, however, and this is another reason to be concerned about the quality of the information disclosed.Footnote 16 The general issue of energy billing data access in benchmarking and disclosure requirements has been identified as a key issue for utilities and their regulators (SEE Action 2013). Washington may be ahead of some other cities in this regard. The District worked out an agreement with the local electric utility, Pepco, under which Pepco will provide building-level billing data to authorized requestors—namely, building owners and their agents—when five or more accounts are present in a building and a single account does not represent more than 80 % of total energy consumption for the building (DDOE 2014).Footnote 17 Use of this service was optional for the 2012 reporting year but is required for 2013 and beyond. Pepco is also the service provider for most of Montgomery County, Maryland, and is providing the County with building-level data in the same way as in Washington. Seattle has also facilitated and now requires automatic upload of energy use data by utilities into the PM software. In Boulder, Xcel Energy is providing automatic uploading of whole-building energy consumption data into Portfolio Manager.
Another concern, pointed out by Stavins et al. (2013), is the veracity of the information disclosed. One problem in this regard is the estimate of building size that is used to calculate the EUI. In some cities, such as Minneapolis, the ordinances provide no guidance on what to use for size. In others, such as Chicago, the ordinance is very specific, listing exactly which areas to include.Footnote 18 Ordinances in Montgomery County, Berkeley, Boulder and Seattle are also very specific about what to include in the floor space calculation. However, it still is not clear that all building owners will calculate square footage in the same way, and periodic independent verification may not be enough to adequately maintain a consistent standard for this measurement. In Washington, only 12 % of buildings reported exactly the same square footage as what is recorded in the tax records (DDOE 2014). The numbers reported in the disclosure requirements are generally larger than those in the tax records. Without further information, it is not clear which numbers are more accurate.
Kontakosta (2013), who has carefully studied the New York program, also argues that manual input of the energy disclosure data leads to significant errors. His analysis using PM data from the New York City benchmarking program identifies some common data entry problems such as a frequent misallocation of energy consumption data when two buildings on separate parcels share the same meter (Kontokosta 2014).
Despite these concerns, the laws provide an important source of information on building energy use that was previously unavailable. This is particularly true for the confidential data on building characteristics and use that feed into Portfolio Manager and that will make possible more detailed analysis of how building features and use affect energy use intensity. In some cities, the data that these laws provide are also being used by utilities and other entities that operate energy efficiency programs, such as the DC Sustainable Energy Utility, to target investment of rate payer and public dollars into buildings where the data suggest there are large unrealized opportunities for energy savings.Footnote 19
From a broader policy perspective, these new laws may be serving as useful real-world experiments. Information provision is widely touted as something that will be necessary to overcome the energy efficiency gap. A careful evaluation should be able to shed light on whether benchmarking and disclosure laws are serving that role.
Data needs for evaluation
The data that building owners are required to report are useful for understanding the components and drivers of building energy use, but they are not sufficient for assessing the full effects of the policy. Performing such an assessment requires a representation of what energy demand would have been in the absence of the program, which by definition is unobservable. As a substitute, analysts need data for a control or comparison group that approximates energy use under baseline conditions in buildings that are subject to the policy (SEE Action 2012). Energy use in affected buildings before the policy takes effect (which is required for reporting in some cities, including Washington) is a potential baseline. However, because other factors that affect energy use, such as weather or economic conditions, also change over time, the pre-policy data are generally insufficient, and an analysis that compares the use of energy in affected buildings before and after the policy takes effect could confound the effects of the policy with other factors, thereby producing a biased estimate of the program effects (Angrist and Pischke 2009, 2010). A better comparison group is one that allows the analyst to capture the effects of other factors that change over time and distinguish those effects from the effects of the policy.
The inclusion of building size thresholds in the design of benchmarking and disclosure laws creates a natural experiment that provides a well-defined control group for assessing program effects. Buildings that fall just short of the minimum size threshold are similar to those just above the threshold. Thus, one could compare energy use before and after the policy takes effect between these two groups of buildings, controlling for other factors such as weather. This should provide an unbiased estimate of the energy savings resulting from the policy. A regression discontinuity approach enables such an evaluation (Imbens and Lemieux 2008). Another possibility is to compare buildings in cities with benchmarking and disclosure laws before and after adoption of the laws with buildings in other cities. A difference-in-differences regression approach could be employed (Meyer 1995; Angrist and Pischke 2009).
Conducting either of these analyses requires energy consumption data beyond that collected under the policy. In a regression discontinuity approach, data would be needed for buildings that lie below the minimum size threshold in cities that have benchmarking and disclosure laws. In a difference-in-differences model, data from buildings in cities that have not passed these laws would be needed. These data are typically in the possession of utilities and subject to strict confidentiality requirements. However, overcoming this hurdle is paramount to a full evaluation of how well the policy is working.Footnote 20
Independent sources of data may be available in some cases to carry out one of these approaches. In recent research, Palmer and Walls (2015a) use a national dataset on investor-owned commercial office buildings to assess the impact of disclosure and benchmarking requirements.Footnote 21 The study employs a difference-in-differences regression model to compare utility expenditures per square foot in office buildings in cities with and without benchmarking policies, before and after the initial reporting deadlines in each of four early adopter cities: Austin, New York City, San Francisco, and Seattle.Footnote 22 The results indicate that disclosure laws have a statistically significant negative effect on utility expenditures after the first reporting deadline. In the central specification, which includes property-level fixed effects and thus controls for many unobserved building-level characteristics, the results show that, all else equal, utility expenditures per square foot are approximately 3 % lower after the laws’ reporting requirements take effect in office buildings covered by the laws.Footnote 23