FINAL Commercial New Construction Customer Quantitative Profile Project 1A New Construction Market Characterization Massachusetts Energy Efficiency Programs Large Commercial & Industrial Evaluation Prepared for: Submitted to: Massachusetts Energy Efficiency Program Administrators National Grid Burlington, Massachusetts, LCIEC i
LCIEC i
Table of Contents 1. Executive Summary... 1-1 1.1 Overview of Approach... 1-1 1.2 Key Findings... 1-3 1.2.1 New Construction Summary Statistics and Trends... 1-3 1.2.2 Program Penetration Rates... 1-7 1.2.3 Large New Construction Energy Use and Energy Demand... 1-8 1.3 Suggestions for Future Research... 1-9 2. Introduction... 2-1 2.1 Study Objectives... 2-1 2.2 Overview of Approach... 2-2 2.3 Organization of Report... 2-4 3. Data Collection and Methodology... 3-1 3.1 Data Sources... 3-1 3.1.1 Dodge Players Database... 3-1 3.1.2 PA Aggregate Program Tracking Data... 3-2 3.1.3 PA Program Tracking Data for Sampled Buildings... 3-3 3.1.4 Energy Consumption Data for Sampled Buildings... 3-3 3.2 Methodology... 3-3 3.2.1 C&I New Construction Market Trends Analysis... 3-3 3.2.2 Aggregate Program Penetration Analysis... 3-4 3.2.3 Selection and Screening of Representative Sample of Buildings... 3-5 3.2.4 Participant Level Program Penetration Analysis... 3-10 3.2.5 Energy Usage Analysis... 3-11 4. Results... 4-1 4.1 C&I New Construction Market Trends... 4-1 4.2 Aggregate Program Penetration Analysis... 4-10 4.3 Participant Level Program Penetration Analysis... 4-14 4.3.1 Electric Large New Construction Program Penetration Analysis... 4-14 4.3.2 Gas Large New Construction Program Penetration Analysis... 4-17 4.4 Energy Usage Analysis... 4-19 4.4.1 Electric Usage... 4-19 4.4.2 Gas Usage... 4-20 5. Conclusions & Suggestions for Future Research... 5-1 LCIEC i
Table of Contents 5.1 Key Findings... 5-1 5.2 Suggestions for Future Research... 5-2 Appendix A Detailed Dodge Player Database Tabulations, 1996-2009... A-1 Appendix B Detailed Dodge Players Database... B-1 Appendix C Glossary of Dodge Project Types... C-1 Appendix D Data Limitations... D-1 List of Tables: Table 1-1: Number of Large New/Addition C&I by Cohort... 1-3 Table 2-1 Research Objectives... 2-1 Table 3-1 All Large C&I New Construction by Cohort, 1996-2009... 3-7 Table 3-2 Sampled Large C&I New Construction by Cohort, 1996-2009... 3-7 Table 3-3 PAs Database Development Efforts... 3-8 Table 3-4 Verified Dodge by Cohort and Size Group... 3-9 Table 4-1 All MA C&I Construction by Year, 1996-2009... 4-2 Table 4-2 All MA C&I Construction by Project Type, 1996-2009... 4-4 Table 4-3 MA C&I Construction by Type of Work, 1996-2009... 4-4 Table 4-4 MA C&I New/Addition by Size, 1996-2009... 4-7 Table 4-5 MA C&I Alteration by Size, 1996-2009... 4-7 Table 4-6 Aggregate Electric Program Penetration, All C&I Construction,... 4-12 Table 4-7 Aggregate Electric Program Penetration, C&I Construction 2000-2009... 4-13 Table 4-8 Large Electric NC Program Penetration, by Cohort... 4-14 Table 4-9 Large Electric NC Program Penetration, by Cohort and Project Size... 4-15 Table 4-10 Large Electric NC Program Penetration, by Cohort and Project Type... 4-15 Table 4-11 Large Retrofit Program Participation for NC Program Non-Participants... 4-17 Table 4-12 Large Gas NC Program Penetration, Cohort 2006-2009... 4-18 Table 4-13 Large Gas Retrofit Program Participation, Cohorts 1996-2000 and 2001-2005... 4-18 Table 4-14 Median Current Year Electric Usage (in kwh) Per Square Foot... 4-19 Table 4-15 Median Current Year Gas Usage (in Therms) Per Square Foot... 4-21 Table 5-1 Sampled and Screened in 2006-2009 Cohort... 5-3 LCIEC ii
Table of Contents Table A-1: MA All C&I Construction by Year, 1996-2009... A-1 Table A-2: MA Large C&I Construction By Year, 1996-2009... A-2 Table A-3: MA Small C&I Construction by Year, 1996-2009... A-3 Table A-4: MA All C&I Construction by County, 1996-2009... A-4 Table A-5: MA Large C&I Construction by County, 1996-2009... A-5 Table A-6: MA Small C&I Construction by County, 1996-2009... A-5 Table A-7: MA All C&I Construction by Type, 1996-2009... A-6 Table A-8: MA Large C&I Construction by Type, 1996-2009... A-6 Table A-9: MA Small C&I Construction by Type, 1996-2009... A-7 Table A-10: MA New/Addition by Year, 1996-2009... A-8 Table A-11: MA Large New/Addition by Year, 1996-2009... A-9 Table A-12: MA Small New/Addition by Year, 1996-2009... A-9 Table A-13: MA New Addition by Type, 1996-2009... A-10 Table A-14: MA Large New/Addition by Type, 1996-2009... A-10 Table A-15: MA Small New/Addition by Type, 1996-2009... A-11 Table A-16: MA Alteration by Year, 1996-2009... A-12 Table A-17: MA Large Alteration by Year, 1996-2009... A-13 Table A-18: MA Small Alteration by Year, 1996-2009... A-13 Table A-19: MA Alteration by Type, 1996-2009... A-14 Table A-20: MA Large Alteration by Type, 1996-2009... A-14 Table A-21: MA Small Alteration by Type, 1996-2009... A-15 Table B-1: MA All C&I Construction, by PA Electric Service Territory... B-1 Table B-2: MA Large C&I Construction, by PA Electric Service Territory... B-4 Table B-3: MA Small C&I Construction by PA Electric Service Territory by Year... B-6 Table B-4: MA C&I Construction, by PA Electric Service Territory, by Cohort... B-8 Table B-5: MA Large C&I Construction, by PA Electric Service Territory... B-9 Table B-6:. MA Small C&I Construction, by PA Electric Service Territory... B-10 Table D-1: Data Limitations... D-1 LCIEC iii
Table of Contents List of Figures: Figure 1-1 Number of C&I by Market Sector, 1996-2009... 1-4 Figure 1-2 Number of C&I by County, 1996-2009... 1-5 Figure 1-3 The Share of Large C&I, 1996-2009... 1-6 Figure 1-4 C&I Construction Activity Trends by Year, 1996-2009... 1-7 Figure 4-1 All MA C&I Construction Activity Trends by Year, 1996-2009... 4-1 Figure 4-2 All MA C&I Total Construction Area by Project Type, 1996-2009... 4-3 Figure 4-3 MA C&I Construction Trends, Number of by Type of Work, 1996-2009... 4-5 Figure 4-4 MA C&I Construction Trends, Construction Area by Type of Work, 1996-2009... 4-6 Figure 4-5 Average C&I Project Size by Type of Work, 1996-2009... 4-6 Figure 4-6 MA C&I Construction by Type of Work, Large, 1996-2009... 4-8 Figure 4-7 MA C&I Construction Area by Type of Work, Large, 1996-2009... 4-9 Figure 4-8 Average Project Size by Type of Work, Large, 1996-2009... 4-9 Figure 4-9 Aggregate PA Program Penetration Trends, 2000-2009... 4-11 LCIEC iv
1. Executive Summary This Executive Summary highlights the findings and recommendations from the research conducted to develop the Project 1A Commercial New Construction Customer Quantitative Profile. This Market Characterization project was completed as part of the evaluation of the large commercial and industrial (C&I) programs operated by the Massachusetts program administrators (PAs). The study addressed the following goals: 1. Develop a comprehensive characterization of the large commercial and industrial new construction market in Massachusetts, in terms of building type, size, ownership, geographic location, chain or franchise status, and energy use. 2. Assess how the trends for large commercial and industrial projects have changed over the past 15 years. 3. Characterize the presence of the Massachusetts Energy Efficiency Program Administrators (PAs) new construction projects in the market in terms of the number of projects that participated in them and the portion of floor space and energy use they represent in key commercial market segments. 4. Develop a dataset of information on large new commercial and industrial buildings, including the number, types, and locations of buildings, the energy profile of buildings, and the program participation rates of buildings for further research, including on-site visits. 1.1 Overview of Approach The Evaluation Team developed the characterization of the population of commercial new construction projects using the F. W. Dodge Players database, PA energy usage and energy demand data, and program participation data. The Players database contains retrospective information on commercial and industrial construction projects that, according to Dodge, have begun construction. Developing this information required close collaboration between the program evaluation team and the PAs, and required information from utility information systems. The principal study activities consisted of the following: LCIEC 1-1
1. Initial analysis of the Dodge Players database of commercial new construction projects for 1996 through 2009. These data allowed us to characterize the population of newly constructed buildings by market sector, geographic location, and building value and estimated size, and to develop trend data on new construction projects. This initial analysis also informed the development of the three cohorts (1996-2000, 2001-2005, and 2006-2009). 2. PAs provided aggregate New Construction program participation data for all available years. These data were used to compare overall program participation rates with overall new construction activity. This information was used to characterize the approximate program penetration rates for the state and by utility service territory. 3. Sample selection and screening. While the Dodge Players database furnishes a comprehensive source of information on new construction projects and the associated market actors; it has limited information on the buildings themselves. We selected a representative sample of large buildings that were constructed in different time periods (the three cohorts ) in the various utility service territories from the Dodge database and screened these records to verify that the construction took place and improved the information recorded by Dodge to overcome Dodge-PA data matching issues. 4. Dodge-PA records matching. PAs matched their customer records to the sampled and screened Dodge records. 5. PAs provided New Construction program data (participant level) for customers associated with sampled projects. We matched program participant records to sampled Dodge records using the PA customer records to develop a comparison between participating and nonparticipating buildings by market sector and time period. This also allowed a detailed assessment of program penetration rates for newlyconstructed large buildings. 6. PAs provided current year billing data for the sampled projects. These data were used to develop energy usage information for sampled projects and to conduct an exploratory analysis of energy usage of large C&I buildings to understand how building energy usage is related to different factors, including program participation. LCIEC 1-2
1.2 Key Findings The study answered some of the important questions proposed in the Work Plan for the Project 1A New Construction Market Characterization. 1.2.1 New Construction Summary Statistics and Trends Q1: How many commercial and industrial buildings are constructed in Massachusetts? In the last 15 years (1996-2009), the Dodge Database listed 28,587 C&I construction projects in Massachusetts. 26% of these projects were new/addition projects and 74% were alteration (R&R) projects. There were 7,544 large projects, i.e., projects with a construction area of at least 10,000 Sq Ft, 4,244 of which were new/addition projects. 1 The detailed information on these new/addition projects is presented in Error! Reference source not found. Table 1-1: Number of Large New/Addition C&I by Cohort Cohort 1996-2000 2001-2005 2006-2009 Size Group # Area (1,000s of Sq Ft) Value ($millions) 10-49 1,064 23,680 2,974 50-249 681 68,140 8,427 250+ 81 39,891 5,825 10-49 883 19,581 2,708 50-249 544 56,554 7,635 250+ 75 36,758 4,954 10-49 587 12,436 2,121 50-249 294 32,023 5,144 250+ 35 16,690 3,254 TOTAL 4,244 305,753 43,042 1 Our screening process for sampled Dodge projects showed that about 90% of records in the Dodge database were turned into actual buildings. On the other hand, Dodge database misses some construction projects because the Dodge Players data keeps track of only publicly-bid construction projects. Therefore, we believe that this information represents the construction market in Massachusetts fairly well. LCIEC 1-3
LCIEC 1-4
Q2: In what market sectors are the buildings constructed? As shown in Figure 1-1, in the last 15 years (1996-2009), about 32% of construction projects were office buildings; 21% were retail buildings; and 14% were educational buildings. Figure 1-1 Number of C&I by Market Sector, 1996-2009 Service 3% Storage 4% Other 5% Amusement 6% Education 14% Retail 21% Medical 8% Government 3% Hotel 2% Religious 2% Office 32% Q3: In what geographic locations are the buildings constructed? In the last 15 years, as Figure 1-2 shows, 26% of construction projects were in Middlesex county; 14% were in Suffolk; 13% were in Norfolk; 9% were in Essex; and another 9% were in Worcester county. LCIEC 1-5
Figure 1-2 Number of C&I by County, 1996-2009 Barnstable 5% Worcester 9% Other 5% Bristol 7% Suffolk 14% Essex 9% Hampden 6% Plymouth 6% Norfolk 13% Middlesex 26% Q4: What is the distribution of buildings in terms of market value and estimated building size? As shown in Figure 1-3, large projects i.e., projects with a construction area of at least 10,000 Sq Ft, constitute only about a quarter of all projects in terms of the number of projects. However, in terms of total building size and project value, these projects account for nearly 90% of all commercial and industrial construction projects. LCIEC 1-6
Figure 1-3 The Share of Large C&I, 1996-2009 $66,103 Large 44,164 7,544 $8,534 Small 5,464 21,043 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 Value ($millions) Area (10,000s of Sq Ft) # Q5: How have these patterns changed over the last 15 years? Figure 1-4 shows that while the number of projects peaked in 1997, total construction area peaked in 2000. Both the number of projects and total construction activity declined between 2000 and 2005, and then they increased from 2005 through 2007. Total construction area declined between 2007 and 2009. The average project size increased between 1996 and 2005, with some fluctuations, and then declined between 2005 and 2009. LCIEC 1-7
Number of Total Area of Construction (1,000s of sqft) Figure 1-4 C&I Construction Activity Trends by Year, 1996-2009 3,500 60,000 3,000 50,000 2,500 2,000 1,500 1,000 40,000 30,000 20,000 500 10,000 0 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 # Area (1,000s of sqft) 1.2.2 Program Penetration Rates Q1: What share of newly constructed large commercial and industrial buildings participate in utility programs? Approximately 46% of large projects in the 2006-2009 cohort and 29% projects in the 2001-2005 cohort participated in Electric NC program. In terms of square footage, the participating projects constituted 45% of total construction area in the 2006-2009 cohort and 35% in the 2001-2005 cohort. The evaluation team also looked at participation in a retrofit program for buildings that did not participate in a NC Program at the time of construction. About 50% of sampled buildings constructed in 1996-2000 had at least one record that participated in a retrofit program between 2001 and 2010. When these findings are combined with findings for NC program participation, they indicate that energy efficiency program penetration rate for large projects was over 50% for any cohort. This research shows that the program misses about half of the large C&I new construction market. But, there is no particular sector that is overlooked. LCIEC 1-8
Gas program tracking data was only available starting 2007. Compared to the electric program, the gas program participation rate was significantly lower. Only 7 out of 94 in the 2006-2009 cohort projects were identified as participants in the gas NC program. Q2: How does that vary by market sector? The participation rates were very similar across different project types within the same cohort. The Electric NC Program did not overlook any particular market sector. Q3: How has that changed over time? Electric NC program participation was higher for projects in the 2006-2009 cohort than those in the 2001-2005 cohort. 1.2.3 Large New Construction Energy Use and Energy Demand Q1: What are the energy use and energy demand characteristics of large new commercial and industrial buildings? The study performed an exploratory analysis of energy usage of large new commercial and industrial buildings using simple regression analysis. Buildings that participated in NC or retrofit programs were found to use more energy per square foot than nonparticipant buildings after controlling for building type, building size, and cohort. This preliminary finding suggests that buildings with higher energy intensity are more likely to participate in energy efficiency programs. 2 Q2: How does that vary by market sector? While there were some differences in energy usage across different building types, many of these differences were not statistically significant, given the sample size. 2 For this project, PAs were asked to supply NC program participation data. However, some PAs also furnished program participation data for their retrofit programs. Therefore, retrofit program data is incomplete and may not cover participation in certain retrofit programs. LCIEC 1-9
Q3: How does energy use and energy demand for buildings that were constructed 15 years ago and 10 years ago compare to those built in the last 5 years? The results of simple regressions showed that once building type, building size, and program participation is controlled for, there are no statistically significant differences between cohorts in terms of energy usage. Further in-depth research is needed to better understand energy use and energy demand characteristics of large new commercial and industrial buildings, and the differences between participant and non-participant buildings. 1.3 Suggestions for Future Research We suggest moving forward with Project 1A Year 2 market characterization efforts using the 2010 work plan and this analysis as a starting point. This study yielded a rich database of information on large new commercial and industrial buildings, including the number, types, and locations of buildings, the energy usage and program participation information of buildings. One of the important tasks in characterizing the market is to conduct on-site inspections of new large commercial and industrial buildings. This database will help to enhance the process by which the sample of buildings is selected and will make the statistics and information developed from that analysis more robust. We recommend focusing the research efforts on sampled projects in the 2006-2009 cohort. The database in this cohort includes 112 large projects of various sizes. Of these projects, 51 were electric NC program participants and seven were gas NC program participants. For participant buildings, the database contains program participation information, including the description, type, and quantity of measures installed, incentive amounts provided, and participation date. The database also contains current year monthly energy usage for most buildings in this cohort. Evaluation team should meet with PAs and EEAC to discuss objectives and scope of year 2 market characterizations activities, with attention to target markets and the scope of on-site visits. These meetings will help identify overlap with other potential evaluation activities (i.e. codes and standards evaluation, future technical potential study) and lay groundwork of existing buildings market characterization and industrial building market characterization. LCIEC 1-10
2. Introduction This report presents the results of the data collection and analysis conducted to develop the Commercial New Construction Customer Quantitative Profile for Project 1A New Construction Market Characterization. This Market Characterization project was completed as part of the evaluation of the large commercial and industrial (C&I) programs operated by the Massachusetts program administrators (PAs). The Evaluation Team developed the characterization of the population of commercial new construction projects using the F. W. Dodge Players database, PA energy usage data, and program participation data. 2.1 Study Objectives The overarching objective of the Market Characterization study, as defined by the Request for Proposal, is the following: To define the attributes of a specific market area in enough detail that the program planners and administrators can use the information for improving program implementation. The principal research objectives of the Market Characterization s Customer Quantitative Profile are provided in Table 2-1. Table 2-1 Research Objectives # Primary Objective 1 Develop a comprehensive characterization of the large commercial and industrial new construction market in Massachusetts, in terms of building type, size, ownership, geographic location, chain or franchise status, and energy use. 2 Assess how the trends for large commercial and industrial projects have changed over the past 15 years. 3 Characterize the presence of the Massachusetts Energy Efficiency Program Administrators (PAs) new construction projects in the market in terms of the number of projects that participated in them and the portion of floor space and energy use they represent in key commercial market segments. 4 Develop a rich database of information on large new commercial and industrial buildings, including the number, types, and locations of buildings, the energy profile of buildings, and the program participation rates of buildings for further research, including on-site visits. LCIEC 2-1
2.2 Overview of Approach The Evaluation Team developed the characterization of the population of commercial new construction projects using the F. W. Dodge Players database, PA energy usage and energy demand data, and program participation data. The Players database contains retrospective information on commercial and industrial construction projects that, according to Dodge, have begun construction. The Customer Quantitative Profile made use of the Dodge Players database, utility energy usage data, and program participation data to furnish the following information: New Construction Summary Statistics and Trends. How many commercial and industrial building are constructed in Massachusetts? In what market sectors are the buildings constructed? In what geographic locations are the buildings constructed? What is the distribution of buildings in terms of market value and estimated building size? How have these patterns changed over the last 15 years? Large New Construction Energy Use and Energy Demand. What are the energy use and energy demand characteristics of large new commercial and industrial buildings? How does that vary by market sector? How does energy use and energy demand for buildings that were constructed 15 years ago and 10 years ago compare to those built in the last 5 years? Program Penetration Rates. What share of newly constructed large commercial and industrial buildings participates in utility programs? How does that vary by market sector? How has that changed over time? What are the implications in terms of program maturity? What are the implications in terms of missed opportunities? Developing this information required close collaboration between the program evaluation team and the PAs, and required information from utility information systems. The principal analytical steps, including descriptions of the data provided by the PAs, for this project consisted of the following: 1. Initial analysis of the Dodge Players database of commercial new construction projects for 1996 through 2009. These data allowed us to characterize the population of newly constructed buildings by market sector, geographic location, and building value and estimated size, and to develop trend data on new construction projects. Using these statistics, we worked with the project team to identify the specific targets for this analysis LCIEC 2-2
by identifying the market sectors that have the greatest energy saving and demand modification potential and by defining what is considered to be a large building in each of the targeted market sectors. Furthermore this initial analysis informed the development of the three cohorts. 2. PAs provided aggregate New Construction program participation data for all available years. These data were used to compare overall program participation rates with overall new construction activity (Step 1). This information was used to characterize the approximate program penetration rates for the state and by utility service territory. 3. Creation of cohorts. The composition of the cohorts was determined based on the results of Step 1 and Step 2. 4. Sample selection and screening. While the Dodge Players database furnishes a comprehensive source of information on new construction projects and the associated market actors; it has limited information on the buildings themselves. We selected a representative sample of buildings that were constructed in different time periods in the various utility service territories from the Dodge database and screened these records to verify that the construction took place and improved the information recorded by Dodge to overcome Dodge-PA data matching issues. 5. Dodge-PA records matching. PAs matched their customer records to the sampled and screened Dodge records. 6. PAs provided New Construction program data (participant level) for customers associated with sampled projects. We matched program participant records to sampled Dodge records using the PA customer records to develop a comparison between participating and nonparticipating buildings by market sector and time period. This also allowed a detailed assessment of program penetration rates for newlyconstructed large buildings. 7. PAs provided current year billing data for the sampled projects. These data were used to develop energy usage information for sampled projects and to conduct an exploratory analysis of energy usage of large C&I buildings to understand how building energy usage is related to different factors, including program participation. The entire process yielded a rich database of information on large new commercial and industrial buildings, including the number, types, and locations of buildings, the energy LCIEC 2-3
profile of buildings, and the program participation rates of buildings. The constructed database can serve the evaluation in a number of different ways, including: Market Characterization. The primary purpose of the database was to furnish information on the number and types of buildings that are being constructed in Massachusetts, and to enrich that analysis with information on how energy usage and energy demand varies within and across market sectors, as well as by building vintage. On-Site Sampling (Year 2 ). One of the important tasks in characterizing the market is to conduct on-site inspections of new large commercial and industrial buildings. This database will help to enhance the process by which the sample of buildings is selected and will make the statistics and information developed from that analysis more robust. Other Market Actor Analysis. Similarly, the year 2 studies detailed in the Project 1A 2010 Work Plan include research with other market actors. Concrete information on building characteristics and energy usage will enhance the discussions with those market actors by putting design and specification decision-making in the context of specific buildings and specific energy use patterns. 2.3 Organization of Report The remainder of the report is organized as follows: Section 3 - Data Collection and Methodology provides a detailed description of the methodology and data collection activities. Section 4 - Results presents the findings of the study, including new construction trends, market penetration, and energy usage. Section 5 - Conclusions and Suggestions for Future Research presents study conclusions and suggestions for future research. Appendix A - Detailed Dodge Players Database Tabulations, 1996-2009 presents detailed information on non-residential construction trends in Massachusetts from 1996 through 2009, as reported by Dodge. LCIEC 2-4
Appendix B - Detailed Dodge Players Database by PA Electric Service Territory, 2000-2009 presents detailed information on non-residential construction trends in Massachusetts from 2000 through 2009 by PA service territory. Appendix C - Glossary of Dodge Project Types contains a glossary of Dodge project types used in the study. Appendix D Data limitations summarizes the limitations associated with the data sources used in this study. LCIEC 2-5
3. Data Collection and Methodology Section 3 describes the data and methodology used to develop the Commercial New Construction Customer Quantitative Profile. The data sources used in the evaluation appear first. Then we present the methodology used for the C&I new construction trends analysis, aggregate program penetration analysis, sample and screening of buildings, participant level program penetration analysis, and the energy usage analysis. 3.1 Data Sources The evaluation team used the Dodge Players Database as the main data source to measure commercial and industrial new construction market activity in Massachusetts. The team relied upon the program tracking and billing data from the Program Administrators to assess Energy Efficiency Program penetration rates and to develop the energy profile of a representative sample of buildings that were constructed in different time periods in Massachusetts. 3.1.1 Dodge Players Database The project team has acquired and analyzed the entire F. W. Dodge Players Database for nonresidential construction projects in Massachusetts for the years 1996 through 2009. The Dodge Players database contains retrospective information on commercial and industrial construction projects that, according to Dodge, have begun construction. Dodge attempts to compile data on the following: Identification of the owner and project principals Identification of the architect and general contractor Principal uses Building size Estimated project costs These data are developed by field reporters employed by McGraw Hill who rely on contacts throughout the new construction industry to identify potential new projects and who regularly visit construction permitting offices to identify newly permitted building projects. The field LCIEC 3-1
reporters also make follow-up contacts with project managers to collect information on the status of the construction and the businesses that are involved with the construction. The evaluation team used the Dodge Players Database as a measure of commercial and industrial new construction market activity in Massachusetts. Some shortcomings of the Dodge data need to be acknowledged: the Dodge data only tracks information for project starts in a given time period and only track publicly-bid projects (whereas some projects do not go to public bid). In addition, the Dodge data may include retrofits, i.e., projects that consist entirely of retrofit measures in the alterations category, that do not qualify for the PAs' New Construction Program, and do not report the square footage of renovation/remodeling projects. However, despite those limits, the Dodge data furnishes useful information on the overall level or scope of the C&I new construction market. 3.1.2 PA Aggregate Program Tracking Data The Evaluation Team requested aggregate electric and gas program tracking data on participating new construction projects for all available years from all Program Administrators. The electric PAs were able to provide electric energy efficiency program tracking data for 2000-2009. 3 The evaluation team used the gas program tracking data that was provided for Project 4 Custom Gas Impact Evaluation and Project 5 Prescriptive Gas Impact Evaluation. Consistent with the Project 4 & 5 data request, the gas program tracking data was only provided for June 2008 through June 2010. 4 In the data request, new construction program participants were defined as commercial and industrial customers that received financial or technical assistance from the PAs for the construction of new buildings or major renovations of existing facilities. Major renovations include the physical expansion of existing facilities (i.e. additional square footage) and major conversions of existing facilities to accommodate new occupants or uses/functions of the facility. Customers that had replaced failed equipment (eligible for 2010 Mass Save New Construction Program) were not considered new construction program participants for purposes of this research. 3 Cape Light Compact's program participation data was for 2002-2009 and Unitil's electric program participation data was for 2003-2009. 4 An analysis of aggregate gas program penetration was not possible with these data. LCIEC 3-2
Instead of providing aggregate data by year, PAs chose to furnish participant level electric and gas program tracking data. In the preferred format for the participant level program tracking data, the evaluation team requested information on project (building) type, project square footage or total project value. None of the PAs were able to provide total project value because they tracked incentive amounts rather than total project construction costs. While some PAs were able to provide square footage information for some projects, none was able to provide it for all projects. 3.1.3 PA Program Tracking Data for Sampled Buildings Following the selection and screening of a representative sample of newly constructed C&I buildings, the evaluation team collaborated with the PAs to match the Dodge records to the PA participation and customer records. Once the matching records were identified, the PAs reported whether the record has participated in their New Construction or retrofit programs. For the participants, PAs provided information on the type, description, and quantity of measures installed, the time of participation, the program type, and the incentive amounts provided. 3.1.4 Energy Consumption Data for Sampled Buildings The PAs provided current year electric and gas consumption data for all customer records associated with the screened and sampled buildings. The evaluation team has aggregated energy consumption for all active accounts associated with the building to get the total current year energy consumption of the building. 3.2 Methodology In this section we present the methodology used for the Dodge NC trends analysis, aggregate program penetration analysis, sample and screening of buildings, participant level program penetration analysis, and the energy usage analysis. 3.2.1 C&I New Construction Market Trends Analysis The evaluation team used the Dodge Players Database as the main data source to measure commercial and industrial new construction market activity in Massachusetts. In the analysis of commercial and industrial new construction market trends in the last 15 years, two adjustments have been made to the Dodge data: Project values have been adjusted for inflation to reflect 2009 dollars. LCIEC 3-3
For alteration (renovation/remodeling) projects, project square footage has been imputed using information on square footage per dollar for new construction projects for the same project type. After reviewing the initial Dodge data tabulations, the evaluation team defined a project to be large, for the purposes of this research, if it has a construction area of at least 10,000 square feet. The evaluation team assigned Dodge project records to PA service territories based on the electric and gas service zip codes provided by the PAs to analyze nonresidential construction trends by PA service territory. 5 3.2.2 Aggregate Program Penetration Analysis Instead of providing aggregate data by year, PAs chose to furnish participant level electric and gas program tracking data. The participant level program tracking data was at the measure level for all PAs. The evaluation team first collapsed measure records at the construction project level and then aggregated the data by year to assess aggregate program penetration rates by year. There were certain limitations associated with the program tracking data PAs have provided. These include: New Construction vs. Retrofit Some PAs have not historically tracked new construction projects separately from retrofits to existing buildings. New Construction vs. Major Renovation The Dodge data distinguish new construction from major renovation projects. Except for data for National Grid, it was not possible to distinguish new construction projects from major renovation in the PA program data. Square Footage While some PAs were able to provide total square footage for some projects, others could not provide it at all. Therefore, aggregate program penetration 5 New construction trends in Massachusetts from 2000 through 2009 by PA service territory are provided in Appendix B - Detailed Dodge Players Database by PA Electric Service Territory. LCIEC 3-4
rate by square footage could not be estimated. While penetration rates by the number of projects may be small, they may be larger by square footage if the program reaches relatively larger buildings. Project Type While some PAs were able to provide information on the construction project type (facility type) for some projects, others could not provide it at all. Project Value While PAs kept track of measure costs, no information was recorded on the total value of the construction project. The provided PA electric program tracking data allowed an analysis of program penetration by number of projects only. Since the program tracking databases for most PAs did not clearly identify new construction projects, the aggregate program analysis could not be performed exclusively for new constructions projects. Missing total project size and project value in PA program tracking data limited the usefulness of the aggregate program penetration analysis. 3.2.3 Selection and Screening of Representative Sample of Buildings The evaluation team purchased all 14 years of Dodge data instead of 9 years (three 3-year cohorts) as proposed in the work plan. The team made an attempt to use the full population by merging all 14 years of Dodge records to PA program tracking records to assess market penetration rates and to identify participant and non-participant projects. However, this attempt proved not to be successful because of imprecise information in Dodge records and PA program tracking databases. Therefore, consistent with the approach proposed in the Work Plan, the evaluation team proposed to construct a representative sample of buildings from the Dodge Players Database instead of working with the entire set of records in the Dodge database to obtain billing data and to perform Dodge-PA records matching. The sampling approach was employed to overcome misalignment of information records and produce more reliable information, including: Validity of Dodge Data PAs expressed concerns on the validity of information presented by Dodge. The evaluation team screened all the records in the sample, verified that the construction actually took place, and updated information on the size and value of the project if necessary. This approach helped alleviate the concerns of the PAs on the validity of Dodge data in representing the new construction market in Massachusetts. LCIEC 3-5
Dodge-PA Data Matching The initial Dodge-PA records matching work by the evaluation team showed that there is not enough specific project/address information in either Dodge or PA records to allow for a clean and precise automated or manual match. By working with a smaller list of projects, the team was able to obtain better address and occupant information and work with billing data contacts identified by the PAs to obtain customer records for these sampled buildings. The Dodge data and PA program participation records were then merged by unique customer record identifiers. Program Penetration Rates PAs and EEAC Consultants were interested in program penetration rates at a detailed level. The sampling approach allowed for detailed and more accurate program penetration analysis including penetration rate by square footage and project type for newly constructed buildings. The proposed sample size was 450 large buildings, consisting of three cohorts based on project start years: Cohort 1 150 Buildings from 1996-2000 Cohort 2 150 Buildings from 2001-2005 Cohort 3 150 Buildings from 2006-2009 Each cohort was divided in three groups based on project square footage: Group 1 Buildings with a square footage of 10,000 to 49,999 Group 2 Buildings with a square footage of 50,000 to 249,999 Group 3 Buildings with a square footage of 250,000 or more The number of projects sampled in each group was determined by the total square footage the group represents in the cohort. The sample was picked using a proportional-to-size (PPS) sampling design, in which, larger buildings had a greater likelihood of being selected. Table 3-1 presents information on the number of new construction projects and their total square footage for each cohort/group, derived from the Dodge database, which served as the sample frame for sample selection. The project records located in zip codes known to the evaluation team as being served by the municipal utilities only were excluded from the sample frame. LCIEC 3-6
Table 3-1 All Large C&I New Construction by Cohort, 1996-2009 Cohort 1996-2000 2001-2005 2006-2009 Group # Area (1000s of Sq Ft) % Area 10-49 695 15,688 7% 50-249 527 53,193 22% 250+ 60 31,858 13% 10-49 629 13,847 6% 50-249 416 44,273 18% 250+ 61 31,044 13% 10-49 461 9,732 4% 50-249 224 24,823 10% 250+ 31 15,287 6% TOTAL 3,104 239,745 100% Source: F.W. Dodge Players Database Table 3-2 presents information on the projects that were included in the sample and their total square footage for each cohort/group. While the sampling rate was about 15% in terms of the number of projects, it was about 43% in terms of square footage. Table 3-2 Sampled Large C&I New Construction by Cohort, 1996-2009 Cohort 1996-2000 2001-2005 2006-2009 Group # Area (1000s of Sq Ft) % Area 10-49 23 649 1% 50-249 79 9,249 9% 250+ 48 28,309 28% 10-49 23 644 1% 50-249 75 8,447 8% 250+ 52 28,351 28% 10-49 34 921 1% 50-249 85 11,068 11% 250+ 31 15,287 15% TOTAL 450 102,925 100% Source: F.W. Dodge Players Database After a building was screened and verified, the evaluation team collaborated with the PA evaluation and billing contacts to match the Dodge records to the PA customer and billing records. Since Dodge provides limited information on the projects, the record matching required LCIEC 3-7
significant effort by PAs. The evaluation team and the PAs had a close and continuous collaboration, and this collaboration was crucial in completing this task. The PA customer records were used to match program participation data and energy usage information to sampled, verified Dodge records. Table 3-3 shows the list of tasks, the role and contribution of PAs, and the timeline the team followed to finish the task. Table 3-3 PAs Database Development Efforts Task PAs' Role Start Date Completion Date Selection of the sample 2/21/11 2/25/11 Screening of records in the sample Evaluation team and PAs collaborated on the Dodge-PA record matching Dodge-PA program participant records matching based on customer records Billing data retrieval Detailed Program Penetration Analysis Match customer records to Dodge records Furnish past program participation data for records identified for screened buildings Furnish billing data for the matching records 2/28/11 4/22/11 3/7/11 5/15/11 3/7/11 5/17/11 3/7/11 5/17/11 5/13/11 6/3/11 Billing data analysis 5/13/11 6/3/11 In the screening process, information available in the Dodge database was used to perform an internet search to find more information on these construction projects. If no or insufficient information was generated from the internet search, the team called the contact telephone numbers listed in the Dodge records for the project owner to obtain the necessary information. The team first attempted to verify that the construction took place. If the building existed, the team obtained the current owner/tenant information and the full address. The team then LCIEC 3-8
attempted to obtain the square footage of the building and compared this figure against the square footage reported by Dodge. It was most challenging for the team to verify Dodge building records for the 1996-2000 cohort. Some Dodge records contained insufficient project description and incomplete address information that prevented the team from verifying whether the construction took place. Dodge record keeping appeared to have improved over time and newer project records contained more description and better address information. Table 3-4 shows the results of the screening process by cohort. In the beginning of the research project, PAs expressed concerns that the Dodge may overestimate the new construction activity, as it may contain project records that were never constructed. The team verified that 91% of Dodge records were, in fact, turned into actual buildings. Table 3-4 Verified Dodge by Cohort and Size Group Cohort 1996-2000 2001-2005 2006-2009 Group # # Verified % Verified Sampled 10-49 23 20 87% 50-249 79 71 90% 250+ 48 41 85% 10-49 23 22 96% 50-249 75 68 90% 250+ 52 50 96% 10-49 34 33 97% 50-249 85 78 92% 250+ 31 26 84% TOTAL 450 409 91% In the verification process, the evaluation team and/or the PAs identified that 12 (3%) projects were either constructed or converted into residential living facilities. These projects have been excluded from the program penetration and energy usage analysis. Moreover, 14 (3%) projects were found to be parking garages only. These projects were also excluded from the analysis, as it was difficult for PAs to identify customer records specific to these parking garages. It was not possible to verify the square footage for all sampled project records. However, square footage information for project records that the team was able to obtain closely matched the LCIEC 3-9
figure reported by Dodge. For non-matching records, no systematic bias was found with regard to Dodge under- or over-estimating the actual size of the building. 3.2.4 Participant Level Program Penetration Analysis Large New Construction C&I program penetration analysis was restricted to the sampled Dodge projects for which PAs were able to identify customer records. PAs faced certain challenges in identification of customer records, including: Multiple records/buildings The project address contained multiple customers and buildings at the specified address and it was not possible to pinpoint the records(s) associated with the building of interest. Service Territory Some building addresses were identified by the PAs as served by Municipal Utilities. For some other building records, PAs were not able to identify the PA that served that particular building. In some other cases, the PA identified the building being in its service territory but could not locate the customer information. PA Program participation data was matched to sampled Dodge records using unique customer identifiers. A building/project was considered to participate in the New Construction Program if there was at least one account associated with the building that participated in the New Construction Program. Looking at the year the construction started and the PA program types and participation dates, the team has determined whether the program participation was for the New Construction program at or around the time of construction or for a retrofit program at a later time since construction. Most Electric PAs started to track their Electric NC Program participation in 2001. 6 Because of this data limitation, the evaluation team could not analyze the Electric NC program penetration rates for the 1996-2000 cohort. For this cohort, the analysis looked at participation in a retrofit program since construction. Gas PAs, on the other hand, started to track their Gas NC Program participation in 2007. With the available program tracking data, Gas NC Program penetration analysis could only be 6 Cape Light Compact's program participation data was for 2002-2009 and Unitil's electric program participation data was for 2003-2009. LCIEC 3-10
partially performed for the 2006-2009 cohort. For the other cohorts, participation in a retrofit program has been analyzed. 3.2.5 Energy Usage Analysis Energy consumption analysis was an exploratory analysis to generate preliminary information on the energy consumption profile of the buildings and whether there were differences among the buildings by program participation status and year of construction. The energy consumption analysis was restricted only to the sampled Dodge records for which PAs were able to identify customer records and provide usage data on. PAs were asked to report electric consumption and demand data and gas consumption data for the last 12 months for all identified accounts for the sampled buildings. All PAs, except for Columbia Gas, provided current year billing data for the study. The evaluation team aggregated the monthly consumption data for the last 12 months for all active accounts for a building to get the total current year energy consumption for the building. Energy use per square foot was computed by dividing the total current year energy consumption for a building by its square footage. The evaluation team excluded buildings for which energy consumption per square foot was lower than the 10 th percentile or higher than the 90 th percentile. Very low energy usage could indicate vacant or partially vacant buildings. Very high usage could indicate a problem with the square footage information reported by Dodge for the buildings. In some occasions, a new building is added to an existing facility and the Dodge reported square footage only for this new building but the energy usage information was for the entire facility. In the analysis, the evaluation team first compared median energy consumption across cohorts and size groups. Second the team used simple regression models to analyze how energy consumption is associated with various factors including building vintage and energy program participation after controlling for the effects of other factors. The Work Plan also called for an analysis of energy demand. While some PAs did not track energy demand information, some tracked it only for very large users. Therefore, an analysis of energy demand could not be performed. LCIEC 3-11