What is the Value of Native Data in a Market Potential Study?
Energy efficiency market potential studies (“potential studies”) are a common planning tool for energy efficiency providers, stakeholders and regulators. Potential studies date back to the 1980s, with the objective of identifying energy efficiency opportunities for a given jurisdiction, and quantifying the size of the market in terms of both energy and demand savings through technology or operational improvements. Potential study findings are often the subject of stakeholder engagement opportunities, providing information for policy making such as energy savings goal-setting, establishing funding requirements at the program or portfolio level, or as an input to integrated resource plan (IRP) modeling of demand side management (DSM) at the system level.
Indeed, several senior staff at DNV GL developed one of the first potential models called DSM AssystTM under a legacy DNV GL company (XENERGY, Inc.). DSM Assyst is a spreadsheet-based, customizable model that uses a bottom-up approach, building up defensible energy efficiency potential estimates from underlying assumptions about measure costs, savings, and applicability that are grounded in the context of a jurisdiction’s program offerings and customer base. As a bottom-up model, DSM Assyst specifies the potential energy efficiency opportunities in terms of various equipment, measures or buildings and then evaluates them based on whether they can be cost-effectively implemented given the jurisdiction’s unique market, system characteristics and forecasts. By contrast, top-down analyses assess energy savings opportunities by developing an understanding of end-use energy consumption and then predicting the corresponding current and future generation profiles.
Now is an especially opportune time to conduct a potential study, given the recent availability of secondary data sources on end use energy consumption and equipment saturation that are higher quality than has ever been. Under the 2009 American Recovery and Reinvestment Act (ARRA), the Congress authorized funding to the U.S Energy Information Administration (EIA) which in turn funded the largest and most complete end-use energy data surveys to date. EIA recently published the 2012 CBECS building characteristics (or saturation) and energy consumption data, along with 6720 records of microdata. The CBECS data publication is a very significant development, in that potential study models (and many other forecasting applications) overwhelmingly relied on 2003 CBECS energy consumption data because the 2007 CBECS data were not published due to quality issues. Additionally, ARRA funded the 2009 RECS to collect data from 12,083 records “to match the Census Bureau’s statistical estimate for all occupied housing units in 2009 derived from their American Community Survey (ACS).” Finally, 2015 RECS microdata (5600 records) and housing characteristics data are now available on EIA’s web site, with square footage estimates expected this summer and energy consumption data to become available in 2018.
While many of our competitors offer potential study modeling services, or develop and support their own models, they all do approximately the same thing in the same way. I recently reviewed a series of potential studies available on the US Department of Energy’s Catalogue of Energy Efficiency Potential Studies. The similarity between the reports and methodologies (top-down versus bottom-up notwithstanding) is significant. In many cases the source data are similar as well, relying on secondary data from RECS and CBECS, the California Database of Energy Efficiency Resources (DEER), and the Regional Technical Forum (RTF) sponsored by states in the northwestern United States to name a few.
Then why collect native saturation data for a potential study? Based on a review of recent potential study requests for proposals (RFPs) received by DNV GL, over half wanted saturation data collection included to “ground” the baseline assessment in local conditions, requested as either a core study requirement or as an optional task. Moreover, few jurisdictions ever fund data collection efforts similar in comprehensiveness and rigor as EIA. This step was preferred over relying on publicly available data resources such as RECS or CBECS.
In DNV GL’s experience, the contributions of native data can have some impact on a potential study’s findings, but not to the extent one would expect. Localized saturation data for fuel-specific end uses such as heating, water heating and air conditioning can be fairly important data to leverage; however, reasonable estimates can be developed from consumption data analyses and recent RECS or CBECS data without an extra primary data collection effort. Updated building codes and equipment efficiency standards are driving national trends toward more uniform baseline conditions for energy efficiency technology stocks nationwide. In addition, national energy efficiency campaigns from large retailers such as Home Depot and Lowes (to name a few), in combination with energy efficiency program providers, are driving higher levels of awareness of higher efficiency equipment. Indeed, national ENERGY STAR awareness in 2016 exceeded 90% according to DNV GL’s analysis. Differences across potential study findings, from our observation, are driven more by local avoided cost structures, and, indirectly related, local climate conditions which drive seasonal peaking characteristics.
If RECS and CBECS provide detailed technology and building characteristics and consumption data that is fairly complete and robust across jurisdictions, what is the value of performing native data collection? Because the time is also ripe for conducting in-depth market characterization and baseline studies, and the value of performing those in-depth end-use data collections is larger than the contribution to the potential study itself. The time is ripe for the following reasons:
- Technology for managing, sampling, collecting, and analyzing large data sets have increased tremendously in recent years.
- Universal phone coverage is waning and other survey data collection modes are proving faster and to be of higher quality.
- Digital communications and account level identification data between customers and electricity and gas providers is increasing.
- Web technology significantly reduces data collection costs while increasing total responses and data quality.
- Electronic solutions for site-level data collection have also eased the customer burden by accelerating the process while increasing the level of site detail, data quality, and speed of data aggregation.
- Non-intrusive, lower-cost end use metering is becoming a reality.
Capitalizing on the trends above, DNV GL has delivered several recent large-scale in-depth market characterization, saturation or baseline studies to our clients—well beyond the minimum requirements for populating the inputs to a potential study. Our clients have realized multiple benefits from doing so, including:
- Integrating detailed measure level data into program design
- Integrating baseline data into revenue forecasting
- Understanding changes in sub-region level consumption patterns and trends
- Communicating energy efficiency program design details to stakeholders
- System energy, peak load, and load shape forecasting
- Corporate marketing and communications
In conclusion, when considering primary data collection just to support a potential study, and what level of investment to make, also consider the uncertainty and limited value of the results. For overall goal setting and measure level assessment, RECS and CBECS are as complete and comprehensive as any secondary data sources there have ever been for populating potential studies. Otherwise, going big is your best strategy, and is worth doing irrespective of any plans you may have for conducting a potential study because of all the other valuable applications that a robust customer data collection effort can provide. By going big, your company or organization will receive much more value for your primary data collection investment, while the return on collecting data simply to populate your potential study alone will deliver less.