Ecological Economics 110 (2015) 141–153 Contents lists available at ScienceDirect Ecological Economics journal homepage: www.elsevier.com/locate/ecolecon Analysis The job generation impacts of expanding industrial cogeneration Paul Baer a, Marilyn A. Brown a, Gyungwon Kim b,⁎ a b School of Public Policy, Georgia Institute of Technology, 685 Cherry Street, Atlanta, GA 30332-0345, United States School of Public Policy, Georgia Institute of Technology, 685 Cherry Street Room #219, Atlanta, GA 30332-0345, United States a r t i c l e i n f o Article history: Received 14 March 2014 Received in revised form 19 November 2014 Accepted 19 December 2014 Available online xxxx Keywords: Employment Green job Energy-based economic development Combined heat and power Investment tax credit a b s t r a c t Sustainable economic development requires the efficient production and use of energy. Combined heat and power (CHP) offers a promising technological approach to achieving both goals. While a recent U.S. executive order set a national goal of 40 GW of new industrial CHP by 2020, the deployment of CHP is challenged by financial, regulatory, and workforce barriers. Discrepancies between private and public interests can be minimized by policies promoting energy-based economic development. In this context, a great deal of rhetoric has addressed the ambiguous goal of growing “green jobs.” Our research provides a systematic evaluation of the job impacts of an investment tax credit that would subsidize industrial CHP deployment. We introduce a hybrid analysis approach combining simulations using the National Energy Modeling System with Input–output modeling. NEMS simulates general-equilibrium effects including supply- and demand-side resources. We identify first-order employment impacts by creating “bill of goods” expenditures for the installation and operation of industrial CHP systems. Second-order impacts are then estimated based on the redirection of energy-bill savings accruing to consumers; these include jobs across the economy created by the lower electricity prices that would result from increased reliance on energy-efficient CHP systems. On a jobs-per-GWh basis, we find that the secondorder impacts are approximately twice as large as the first-order impacts. © 2014 Elsevier B.V. All rights reserved. 1. Introduction Many economic, environmental and political factors are driving a growing emphasis on the efficient and environmentally sustainable production and use of energy (Brown and Sovacool, 2011; Pollin et al., 2008). From climate change to foreign exchange, our current patterns of energy use in the United States and worldwide are severely stressing natural and social systems (Diamond, 2005; Rockstrom et al., 2009). U.S. energy demand is projected to continue to grow,1 and concerns about the security and affordability of energy supply are literally front-page news. Conflicts about the policy drivers of economic growth and job creation and anxieties about persistent structural under-employment are feeding debates over infrastructure investments and environmental policy. Regulatory policies that are feared to lead to the loss of jobs are easy political targets, uniting business owners and workers, even when health and other social benefits are large in comparison. Alternatively, regulatory or fiscal policies that can be shown to produce net job growth tend to be politically attractive. ⁎ Corresponding author. E-mail addresses: paul.baer@mac.com (P. Baer), marilyn.brown@pubpolicy.gatech.edu (M.A. Brown), joykim@gatech.edu (G. Kim). 1 The U.S. Energy Information Administration (2012) forecasted that U.S. total energy consumption would grow by 0.3% per year from 2010 to 2035. http://dx.doi.org/10.1016/j.ecolecon.2014.12.007 0921-8009/© 2014 Elsevier B.V. All rights reserved. Recent studies of “green jobs” have shown positive contributions of clean energy policy legislation to job creation and sustainable economic development (Laitner and McKinney, 2008; Pollin et al., 2008). However, these studies shed little light on the relationship between clean energy investments, energy market dynamics, and macroeconomic effects including both direct and indirect employment development. For example, analysis to date has not fully evaluated the second-order employment effects from the redirection of energy-bill savings accruing to participants in energy-efficiency programs (although in a different context, these expenditures have been considered by analysts of the “rebound effect” (e.g., Sorrell et al., 2009)). In addition, the literature has rarely examined the impact of lower energy prices economy-wide that could result from the lower energy use that occurs following energy-efficiency investments. With largescale energy efficiency, competitive markets would see lower clearing prices for energy and price-regulated markets would experience lower marginal dispatch costs — in both cases, prices would benefit from decreasing reliance on the most expensive marginal generating equipment (Kim et al., 2013; Kramer and Reed, 2012; Steinhurst and Sabodash, 2011). This “demand reduction induced price effect” (DRIPE) suggests that increased energy efficiency could reduce energy prices for all customer classes, generating jobs across the economy as the resulting savings are spent on goods and services that are more job-intensive than the capital-intensive industries associated with energy production. Two studies which have addressed these effects quantitatively include Laitner (2009) and Laitner et al. (2010). The results are not precisely 142 P. Baer et al. / Ecological Economics 110 (2015) 141–153 comparable because these papers model end-use efficiency improvements in all sectors, whereas our research addresses direct efficiency improvements in the industrial sector and calculates second-order impacts based only on price reductions in the commercial and residential sectors, but the job-creation impacts from energy-bill savings are similar. In another similar study, Rhodium Group (2013) looked at economy-wide impacts of major efforts to improve “energy productivity,” and found a net increase of 1.3 million jobs in 2030 derived from net energy bill savings of $494 billion, $151 billion of which was attributed to the reduction in energy prices driven by reduced demand. This study assesses the employment impacts and energy market dynamics of a sizeable increase in the deployment of one key energy efficient technology – combined heat and power (CHP) systems – driven by a federal investment tax credit (ITC). CHP technology is often regarded as a transformational technology with potential for significantly improving energy efficiency by productively reusing waste heat (Shipley et al., 2008); indeed, a recent executive order has set a national goal of 40 GW of new industrial CHP by 2020, targeting a broad set of stakeholders including states, manufacturers, and utilities (The White House, 2012). Our analysis recognizes that subsidies can produce changes in energy consumption, production, and prices across the economy, including the industrial, residential, and commercial sectors. By combining an Input–output (I–O) model with the projections of an energy systems model (the National Energy Modeling System (NEMS)), we develop a hybrid analytical tool to generate plausible estimates of the consequences of various policy, price, and technology scenarios. 2. Industrial CHP and ITC Policy Also known as cogeneration, CHP is the production of electricity together with economically useful heat, for use in industrial processes and for heating and cooling buildings. By capturing energy that would otherwise be wasted, the efficiency of conversion can be increased from 45% in typical thermal power plants to as much as 70% in efficient natural gas CHP facilities (U.S. Environmental Protection Agency Combined Heat and Power Partnership, 2008). In addition, while the main fuel of CHP systems is natural gas,2 CHP can often be fueled with industrial waste products or with biomass, further reducing fossil fuel consumption and carbon dioxide emissions. CHP is also a form of distributed generation, as CHP technologies allow end-users to generate electricity on site. The primary CHP technologies (so-called “prime movers”) include gas turbines, reciprocating engines, and boiler/steam turbine combinations, which are combined into systems with electrical generators and heat recovery equipment. Such systems are tailored to available fuels, plant operating costs, the difference between electricity price and fuel costs,3 and the on-site need for electrical power versus thermal energy (Sentech Inc., 2010). Deployment of CHP systems reduces electricity purchased through the grid from central utility stations and usually produces power to sell back to the grid. This onsite generation avoids energy losses from electricity transmission, and it can increase overall system resilience, as has been shown in the development of locational marginal pricing for distributed generation of all types (Lewis, 2010). These characteristics make CHP especially attractive for industrial users who want to enjoy the benefits of site-specific, strategic energy production to supply their electricity and thermal energy needs. The industrial sector is the largest consumer of energy in the U.S., accounting for 31% of total energy consumption in 2010 (U.S. EIA, 2012). According to the Annual Energy Outlook 2012, industrial energy 2 Approximately two-thirds of industrial CHP systems in the U.S. are fueled by natural gas (ICF International, 2011). 3 The difference between the price received by a generator for the electricity it produced and the cost of the natural gas needed to produce that electricity is called the “spark spread.” Spark spread (in $/MWh) is calculated as Price of Electricity − [(Price of Natural Gas) ∗ (Heat Rate)] = $/MWh − [($/MMBtu) ∗ (MMBtu/MWh)]. consumption is also expected to show the largest increase of any sector over the next 25 years. Therefore, improving energy efficiency in the industrial sector is a critical agenda item for policy-makers. Despite the economic and environmental attractiveness of CHP, decision-makers in the industrial sector face financial, regulatory, information, and workforce barriers to what are generally considered to be cost-saving investments. Many studies have documented a gap between optimal and actual energy efficiency (Dietz, 2010; Hirst and Brown, 1990; Jaffe and Stavins, 1994). First of all, the economic challenges of CHP investments are the greatest barrier to viability (Chittum and Kaufman, 2011); although CHP promises long-term energy-bill savings, companies often feel a greater financial risk because CHP installations have high upfront costs and long payback periods compared to traditional equipment. The current economic downturn in the U.S. has caused companies to become increasingly conservative, with even greater aversion to longer payback periods compounded by difficulties securing financing (Chittum and Kaufman, 2011). Second, utility monopoly power and utility rate structures also distort CHP economics. Many utilities discourage CHP facilities from acting as independent distributed generators who can sell excess power to nearby customers at retail or negotiated rates. In some states, utilities own and manage the transmission and distribution infrastructure and they discourage CHP users from selling their excess power back to the grid at a wholesale rate. Furthermore, utilities impose additional charges for private wire usage and for standby or back-up service (Chittum and Kaufman, 2011; Sciortino et al., 2011). These electricity rate structures reduce the money-saving potential of on-site generation. Third, the enforcement of interconnection standards and environmental regulations can be substantial barriers to CHP investments, especially for smaller CHP projects. Although many states have developed interconnection standards that ensure stable utility service, the lack of uniformity in application processes has caused unnecessary project delays and has generated high transaction costs (Shipley et al., 2008; U.S. EPA, 2012). In addition to the costs of dealing with interconnection standards, various permits and regulations—such as input-based emission standards—can also increase upfront project costs. Satisfying the conventional emission regulations based on heat input (lb/MMBtu) or exhaust concentration (parts per million) can be challenging to CHP deployment at the beginning of a project's lifespan. CHP generally increases the emissions onsite, but due to its high efficiency, reduces the overall emissions of all pollutants in a given region as well as overall fuel consumption (Chittum and Kaufman, 2011). Many CHP studies argue that the transformation from current input-based emission standards to output-based standards can capture the total regional emissions benefits of CHP development (Shipley et al., 2008; Cox et al., 2011; Sciortino et al., 2011). Lastly, as CHP has been utilized in quite varied sectors, the difficulty of effectively sharing lessons and information across industries can impede the process of diffusion and modernization of CHP projects (The Committee on Climate Change Science and Technology Integration (CCCSTI), 2009). Given the uncertainties about the benefits and risks of CHP technology over a project's whole lifespan, the information incompleteness can be a substantial barrier to expensive capital investments. Subsidies that encourage the market penetration of CHP systems and continuing technology development may mitigate these information barriers. CHP users, manufacturers, and service providers have advocated for expanding CHP-friendly tax credits to reduce market barriers to the expansion of CHP (ICF International, 2010). The federal government has established a 10 percent ITC for qualified CHP systems through 2016. The eligible system size is capped at 50 MW that exceeds 60% energy efficiency on a lower heating value basis.4 Several states are beginning to tackle current regulatory barriers. Legislative proposals have suggested 4 The Database of State Incentives for Renewable Energy, www.dsireusa.org/. P. Baer et al. / Ecological Economics 110 (2015) 141–153 increasing the ITC from 10% to 30% for highly efficient CHP technologies5 and removing the 50 MW capacity limit on qualified systems.6 Increasing the ITC to 30% for all efficient CHP systems would increase CHP market penetration, improve energy efficiency, enhance operational reliability, and provide economic savings that would improve business cost-effectiveness. In this context, we examined three ITC scenarios that apply 10, 20, and 30% subsidies and remove the 50 MW cap through 2035. Prior analysis of a 30 percent ITC estimated the deadweight losses from such a federal tax subsidy, but these losses were more than offset by the social benefit produced by addressing the negative externalities of air pollution and climate change (Brown et al., 2013). 3. Green Jobs: Key Concepts From the Literature Although much academic evidence suggests otherwise, there remains a significant perception in the U.S. of a “jobs vs. environment tradeoff” (Claussen and Peace, 2007; Goodstein, 1999). To counter this perception, much effort has gone into promoting “green jobs,” a vague term that generally refers to a wide range of economic activities aimed at mitigating environmental threats and improving energy security. Recently the Bureau of Labor Statistics introduced the following definitions of green jobs: A. Jobs in businesses that produce goods or provide services that benefit the environment or conserve natural resources. B. Jobs in which workers' duties involve making their establishment's production processes more environmentally friendly or use fewer natural resources.7 According to surveys they found about 3.4 million workers in “green goods and services” (definition A) in 2011, and about 850,000 workers who worked more than half time on “green technologies and practices” (definition B) (Bureau of Labor Statistics, 2013). Even these definitions leave lots of ambiguity. On the one hand, it is clear that wind turbine installers hold green jobs; but what about the workers in the mine that produces the iron that goes into the steel for wind turbines? Would it matter if it was all one firm? Additionally there are regulators and the workers who monitor compliance with regulation — “green jobs” by many definitions but not directly productive of goods and services, thus not necessarily what one wants to maximize. More importantly for our purposes, inasmuch as one goal of investments in ecological efficiency is to increase overall social welfare, the reduction of energy expenditures allows redirection of household income to more valued goods and services. One consequence of this is the “rebound effect” (actually a combination of price and income effects in economic terms), which offsets the initial efficiency gains to a greater or lesser extent; however, it also typically leads to employment gains as spending is redirected from the very highly capital intensive energy industries to more labor-intensive service and manufacturing industries. The jobs produced from this redirection are a benefit of efficiency improvements, and can be an important indirect consequence of environmental policies (Turner, 2009). Since the American Recovery and Reinvestment Act of 2009 (ARRA), discussions of green job creation have increasingly focused on “energybased economic development,” a term coined by Carley et al. (2011) to capture the integration of policy-driven transformations of energy systems for environmental and security goals with regional and national concerns for economic development and resilience. Domains of energybased economic development include energy technology innovation, 143 energy equipment manufacturing, installation and service, research and development, fuel economy, and electricity consumer's energy bills (Laitner and McKinney, 2008; Pollin et al., 2008; White and Walsh, 2008). Distinct from traditional economic development strategies, this approach adds a focus on clean energy to emerging sustainable economic development practices that care for both people and place by improving standards of living for all and sustaining local employment capacity (Blakely and Leigh, 2009). Reflecting these various issues, a wide range of academic and consulting studies have used different kinds of models to estimate the employment effects of environmental and climate policies, including I–O models, Computable General Equilibrium (CGE) models, and what are often called “Analytic Models” that (typically in a spreadsheet) use various “bottom up” methodologies to estimate job creation (Wei et al., 2010). Even where similar methods are used, model projections vary widely, since they are dependent on baseline assumptions and model parameterizations. Furthermore, at a large scale, policies can actually drive economy-wide changes in prices and interest rates, and comprehensive modeling efforts must account for these general equilibrium effects endogenously. Overall, energy policies promoting green jobs should be able to consider not only the employment that stems from the investment in energy technologies and R&D (the “direct, indirect and induced jobs” of conventional I–O analysis), but also the “second order” indirect and induced economic activities resulting from energy-bill savings due to price and demand changes. Using IMPLAN (IMpact analysis for PLANning) or similar I–O models, many studies have utilized an estimate of the national-scale multiplier effects of additional direct stimulus spending on energy efficiency (Geller et al., 1992; Laitner et al., 1998; Pollin et al., 2008). These studies have usually concluded a net positive return in job opportunities per installed capacity unit compared to business-as-usual. Job creation has commonly been attributed to the construction, installation, and operation of energy efficient technologies and other related services. The job estimates in these studies are not fully comparable due to geographical and sectoral differences. Nevertheless, previous reviews have compared the job estimates of earlier studies along different dimensions including efficiency scenarios (Laitner and McKinney, 2008) or technology types (Carley et al., 2011; Wei et al., 2010). For example, Laitner and McKinney (2008), reviewing 48 reports from 1992 to 2008, conclude that a 20–30 percent energy efficiency gain within the U.S. economy might lead to a net growth of 0.5 to 1.5 million jobs by 2030; the average among all studies reviewed is a net benefit of 49 job-years per TBtu of savings. A more recent study estimated that doubling U.S. energy productivity8 by 2030 could create 1.3 million jobs, while increasing GDP up to 2% (Houser, 2013). We compare the results of some of these studies with our own findings in Section 7. Despite the strengths and applicability of I–O modeling, most studies have acknowledged the inherent limitations of the method. For example, Lehr et al. (2008) used survey data to amend I–O tables by applying key inputs and intermediary goods of the renewable industry, and the potential for expected employment. Such efforts, however, are still rare in employment studies of energy-efficiency programs. A comprehensive approach to assessing jobs from energy-efficiency promotion should cover complex impacts including not only supply-side (oil and gas, coal, nuclear, and renewable fuels) but also demand-side (residential, commercial, and industrial sectors) and energy conversion impacts (electricity markets). In this research, as described next, we track these comprehensive energy market paths by combining an I–O model with NEMS. We further discuss limitations and future extensions in Section 8. 5 H.R.4751 (2010) — sponsor: Rep. Tonko, P. (Source: www.govtrack.us). H.R.4455 (2009) — sponsor: Rep. Thompson, M.; S. 1639 (2009) — sponsor: Sen. Bingaman, J.; H.R.4144 (2009) — sponsor: Rep. Inslee, J. (Source: www.govtrack.us). 7 http://www.bls.gov/green/overview.htm#Definition, accessed 2/24/2013. 6 8 Energy productivity, measured in $output/unit energy, is the reciprocal of energy intensity. 144 P. Baer et al. / Ecological Economics 110 (2015) 141–153 Investment Tax Credit on CHP Expansion of Industrial CHP Installation Efficiency Gains and Energy Savings ($) Operations and Maintenance: Non-Fuel Energy Price Dynamics Operations and Maintenance: Fuel Residential/Commercial Energy Bill Savings ($) •Change in natural gas demand •Change in coal and petroleum demand •Change in industrial electricity purchased from utility •Sales to the grid Excess Electricity Sales to the Grid (Direct/Indirect/Induced) (Direct/Indirect/Induced) Second-order Impacts First-order Impacts (General Equilibrium Adjustment) Fig. 1. Flow diagram of employment impacts. 4. Methodology: Hybrid Modeling As noted, this study aims to assess the employment impacts of an increase in the deployment of CHP systems through a federal ITC policy. To investigate the relationship between energy-efficiency investments and energy market dynamics, unlike other green job studies, we developed an analytical model to combine energy market projections derived from NEMS with sectoral employment coefficients taken from I–O modeling. 4.1. National Energy Modeling System (NEMS) Clean energy policies and investments are first modeled in NEMS, which can analyze energy consumption changes by fuel type9 along with policy scenario and energy market assumptions. Since the model is run on Georgia Tech computers, we call it “GT-NEMS.”10 NEMS uses resource supply and price data based on federal, state, and local laws and regulations in effect at the time of the analysis. The NEMS integrating module ensures general market equilibrium fuel prices and quantities across all twelve modules including supply (oil and gas, coal, and renewable fuels), demand (residential, commercial, industrial, and transportation sectors), energy conversion (electricity and petroleum markets), and macroeconomic and international energy market factors. Specifically, we derive the baseline projections of GT-NEMS from the version of NEMS that generated EIA's Annual Energy Outlook 2011, which is regarded as a reliable representation of the U.S. energy market (U.S. EIA, 2011). A “policy case” produces changes in fuel prices and resource consumption when compared with the “reference case.” 9 NEMS reports changes in electricity use and fuel used in electricity generation as well as direct fuel use. 10 Even when the same NEMS code is used on two hardware systems with the supporting software, the results could be distinct from those of the EIA. The fact that the GT-NEMS Reference case nearly duplicates the EIA's Reference case indicates that the two models are essentially identical. NEMS is well suited to projecting how alternative energy policies might impact energy markets over time, particularly with respect to CHP systems, because it has a detailed methodology for evaluating the market penetration of CHP technologies in different subsectors of industry. NEMS' “bottom-up” technology configuration enables an assessment of technology investments, energy prices, energy consumption and expenditures, carbon abatement, and pollution prevention over time and across regions of the U.S. In this study, focusing on industrial CHP end-users, three policy scenarios were evaluated by GT-NEMS. The reference case already reflects the current 10% ITC subsidy for 50 MW or less-sized CHP through 2016. Three policy cases of expanded ITC are modeled, assuming subsidies of 10, 20, and 30% from 2015 to 2035 across all type of CHP systems. The results of each scenario run provide estimates of changes in CHP capacity, natural gas consumption, electricity purchased from the grid and sales back to the grid, and energy prices by sector. The differences between the reference case and the three levels of ITC subsidy allow estimation of net jobs from installation and operation of additional CHP and the recycling of economy-wide energy-bill savings. 4.2. Input–Output Model and First Order Impacts Any employment study, whether focused on a project or a policy, has to specify the boundaries of the analysis and the pathways of employment impacts (positive or negative) that will be included. In spite of the numerous methodologies that have been used to analyze employment impacts and macroeconomic impacts more broadly, no single terminology exists for describing the relevant pathways. I–O modeling has developed a conventional language referring to direct, indirect, and induced employment, where direct employment is based on additional final demand for products from particular sectors, indirect employment is based on expenditures for intermediate goods by the sectors seeing increased final demand, and induced employment is based on the additional expenditure by persons earning wages and profits from the 145 P. Baer et al. / Ecological Economics 110 (2015) 141–153 25.0 InstallaƟon 19.8 20.0 OperaƟon 15.0 15.5 14.5 10.0 Energy ProducƟon Induced Impact 5.7 6.6 7.4 5.0 ConstrucƟon and Equipment OperaƟon & Maintenance-Non Fuel Electricity Natural Gas Coal & Petroleum Other - Energy Bill Savings, res and com Fig. 2. Employment coefficients by sector (jobs/$2009 M). additional production (Miller and Blair, 2009). We classify all of these as first order impacts, as they are based on partial-equilibrium effects in which all prices and technological coefficients are assumed to stay constant. 4.3. Second Order Impacts In addition, we consider second order impacts, in which general equilibrium effects such as changes in energy prices due to increased efficiency (that is, DRIPE effects) propagate through the economy. Models such as NEMS can calculate employment effects directly; however, because the linkages in NEMS between changes in sectoral demand and changes in employment are quite opaque, we use the changes in energy expenditures as an output from NEMS to calculate second order impacts based on I–O employment coefficients taken from IMPLAN. Further details of our methods are given below. 4.4. Subsets of First-order Impacts We model three different categories of first-order impacts: construction and equipment installation purchases, non-fuel CHP operating expenditures, and changes in industrial energy purchases (in this case, increased purchase of natural gas and decreased purchases of electricity, coal and petroleum products) (Fig. 1). These in turn are subdivided into one-time jobs in construction, installation and manufacturing (CIM), and “permanent” (or “annual”) jobs based on the operation of the new capacity and the corresponding changes in energy purchases. Ultimately we aggregate these into full-time-equivalent jobs. 4.5. Assumptions Regarding Second-order Impacts Modeling second-order impacts using NEMS' energy market projections requires a number of strong assumptions. Second-order impacts derive from redirection of energy bill savings by residential consumers, commercial businesses, and industry (Fig. 1).11 If the scale of efficiency investment is large enough, it will cause economy-wide changes in supply and demand, and thus prices, for energy. This in turn changes the expenditures of various actors. Businesses, whether in the industrial or commercial sector, could pass their energy bill savings on to customers through lower prices, or maintain prices and increase profits or wages, or some combination. We assume businesses would count their energy bill savings after amortizing new CHP investment costs. 11 A reviewer noted that first and second order impacts in the industrial sector in particular are not easily disentangled; this is plainly true in our calculations, but conceptually first order impacts are those stemming from industries which actually make the CHP investments, while second-order impacts are from price/expenditure responses only, primarily in industry sectors not making new CHP investments. The amortization schedule assumes a 20 year payback of new construction and equipment investments at a 3% interest rate.12 As energy bill savings recycle through the economy, additional employment impacts are expected when expenditures shift from capital-intensive sectors like utilities to more labor-intensive sectors like services, manufacturing and construction. As a simplifying assumption, we treat all energy bill savings as direct savings to consumers (assuming that changes in prices, wages, and dividends all eventually accrue to households), and that they are respent in direct proportion to the existing distribution of household expenditures.13 Furthermore, we assume that savings accrue to households in proportion to the existing distribution of household income; while this is unrealistic for a variety of reasons, the employment coefficients for household expenditures by different income brackets vary relatively modestly (about 8% between the highest and lowest). Using this procedure, we calculate a weighted employment multiplier of 15.5 jobs per million dollars of energy bill savings across all sectors in 2009 (see Fig. 2 for comparison with other sectors); as with all of our multipliers it is “discounted” over time to account for economy-wide productivity increases.14 4.6. IMPLAN Employment Coefficients To estimate employment impacts, NEMS outputs (e.g., additional CHP capacity, sectoral energy consumption, etc.) are combined with I–O employment coefficients (sometimes imprecisely referred to as “multipliers”) that are derived from IMPLAN. The I–O model is based on annual tracking of the national gross output of the transactions among diverse industries and government agencies, and then provides the estimation of direct, indirect, and induced employment coefficients between pairs of industries (Miller and Blair, 2009). The employment coefficients were calculated for six components of the CHP technology life-cycle and the associated economy-wide impacts: new construction and equipment installation (which is developed by bills of goods); non-fuel operation and maintenance (O&M); three energy sectors (electric utilities, natural gas, and the coal and petroleum sectors together); and all other sectors affected by energy bill savings in the residential and commercial sectors. 12 According to the Federal Reserve's survey of terms of business lending (November 2013), compounded average interest rates for commercial and industrial loans were a range of 1.64% to 2.46% depending on the degree of risk (Source: http://www. federalreserve.gov/releases/e2/current/#fn4). 13 The household respending multiplier is calculated by adding a unit of income to households in the IMPLAN model, but adjusting for the fact that there are no income taxes when “income” is actually savings. 14 We assume that productivity in all sectors increases at a 1.84% annual rate, the economy-wide average for the years 2007–2011. 146 P. Baer et al. / Ecological Economics 110 (2015) 141–153 Table 1 Weights of new CHP construction and installation expenditures: preliminary estimation vs. experts survey. Category Primary generation (turbine and power boiler) Construction Electrical equipment Machinery and fabricated metal Electronic components (controls) Environmental equipment Other materials Scientific and technical services Finance and insurance Other Total Respondents Ng-based company 1 Ng-based company 2 Biomass-based company 3 Biomass-based company 4 56% 11% 11% 6% 3% 3% 0% 11% 0% 0% 100% 39% 20% 6% 5% 1% 10% 2% 9% 8% 0% 100% 37% 22% 4% 11% 3% 5% 8% 7% 2% 1% 100% 36% 25% 6% 7% 3% 5% 3% 7% 8% 0% 100% Results from experts elicitation Preliminary estimates base on literature 39% 20% 7% 9% 4% 6% 3% 8% 4% 0% 100% 25% 20% 10% 15% 10% 7% 3% 5% 5% 0% 100% 4.7. Bills of Goods 5.1. Scenario Modeling Results To estimate the jobs associated directly with the construction and operation of new facilities, we identify the industrial sectors contributing to the CHP systems using the concept of a “bill of goods.” Our bill of goods for CHP systems involves selecting industrial sectors taken from IMPLAN's 440 sectors, the associated employment coefficients also taken from IMPLAN, and a set of estimated weights reflecting each sector's expenditure share. We began with a review of the literature to identify the relevant industrial sectors and their respective proportion of installation costs. We selected ten categories of industrial sectors and estimated the weights for each category. We then conducted an expert survey to validate our estimates. Four of ten experts contacted provided complete responses; two for natural gas-based systems and two for biomass-based systems. Since the fractions are fairly similar, we used the average proportion of all four responses. Table 1 includes the results of each expert's response and the average weights that we applied for the final employment coefficients calculation. Table 2 shows the final combination of bills of goods and IMPLAN employment coefficients. This analysis produced an estimate of 14.5 first-order jobs created per 1 million dollar ($2009) investment in CHP system installation and construction. We also identified industrial sectors for long-term O&M employment impacts and applied their employment coefficients from IMPLAN. Table 3 shows the employment coefficients for non-fuel and fuel sectors for operation and maintenance. Fig. 2 shows the aggregated employment coefficients for all six categories of employment market sectors. The non-fuel O&M sector would be the most labor-intensive sector of job generation throughout the life cycle of CHP systems. The second-order employment impacts that result from switching households' spending from energy bill payments to other consumption goods or services would be significant with the second highest employment coefficient, 15.5 jobs per million dollars of investment/expenditure. As a result, the deployment of CHP systems would generate significant employment impacts in the long-term, in addition to the short-term, one-time jobs created during the construction phase. The second-order impacts would be spread across a wide band of economic sectors, roughly proportional to the current distribution of household consumption spending. Major components of our GT-NEMS results are summarized in Figs. 3 and 4. They show increases of CHP capacity and generation (Fig. 3) from the three levels of ITC subsidies compared to the reference case, decreases of industrial electricity purchases from the grid (Fig. 4), and increases of electricity sales back to the grid (Fig. 4). Our reference case (modified slightly from the 2011 Annual Energy Outlook15) predicts that the nation's CHP capacity will expand at rates significantly greater than in the last few years, reaching 50 GW in 2020 and 80 GW in 2035. With ITC subsidies, CHP is estimated to grow by an additional 6.1 GW (8% above the reference case forecast for 2035) for the 10% ITC policy, 13.6 GW (a 17% increase) for the 20% ITC, and 22.5 GW (a 28% increase) for a 30% ITC (Fig. 3). As noted earlier, a recent executive order has set a national goal of 40 GW of new industrial CHP by 2020; assuming that 23% of this future capacity will be in the petroleum refining industry (as it is today), this would imply a goal of 31 GW of new capacity by 2020 in the non-refining industrial sectors that we model here.16 The reference case of NEMS forecasts that the nation's industrial CHP capacity would meet only 47% of the executive goal by 2020 (a 15 GW increase in non-refining industrial CHP from 2012 to 2020). The three ITC policies would bring the industrial sector closer to achieving the goal, though they still fall short, meeting only 53% of the goal with the 10% ITC, 61% with the 20% ITC, and 70% with the 30% ITC by 2020. The goal is achieved with the 30% ITC by 2023. The expanded industrial CHP capacity enables a significant increase in electricity generation in the industrial sector (Fig. 3). The growth rates of CHP electricity generation are 1–3% higher than the rate of CHP capacity growth, which means that industrial plants tend to utilize the CHP system to generate electricity in an efficient way, with higherthan-average “capacity factors.” Since expanded CHP capacity would allow industry to consume electricity from its own on-site generation, manufacturers would not need to purchase as much electricity from the central utility. (Even if they could meet all of their on-site electricity needs, industrial plants still benefit from being connected to the grid for standby and back-up power.) Fig. 4 shows the reduction of industrial electricity consumption purchased from the grid. The reference case shows the large decrease in purchased electricity consumption that occurred during the economic recession between 2007 and 2009, and forecasts a recovery to prior 5. Results 15 Note that our reference case projection is somewhat greater than the AEO 2011 projection of 43.5 GW of industrial capacity in 2020 due to a correction of the CHP installation cost database, which had an incorrectly high price for the largest and most efficient CHP systems. 16 The petroleum refining industry is modeled in a separate module of NEMS, and is not treated in this paper. This section further explains the estimated energy market impacts from GT-NEMS modeling and the employment impacts estimated by the hybrid energy system/I–O modeling. 147 P. Baer et al. / Ecological Economics 110 (2015) 141–153 Table 2 Selected IMPLAN sectors and employment coefficients for CHP installation. Table 3 Selective IMPLAN sectors and employment coefficients for operation and maintenance. IMPLAN code and industrial sector Weights (%) Jobs per $2009 M IMPLAN code and industrial sector Weights (%) Jobs per $2009 M Installation 1. Primary generation 222 Turbine and turbine generator set units manufacturing 188 Power boiler and heat exchanger manufacturing 2. Construction 35 Construction of new nonresidential manufacturing structures 3. Electrical Equipment 266 Power, distribution, and specialty transformer manufacturing 267 Motor and generator manufacturing 268 Switchgear and switchboard apparatus manufacturing 269 Relay and industrial control manufacturing 272 Communication and energy wire and cable manufacturing 275 All other miscellaneous electrical equipment and component manufacturing 4. Machinery and fabricated metal 171 Steel product manufacturing from purchased steel 174 Aluminum product manufacturing from purchased aluminum 186 Plate work and fabricated structural product manufacturing 193 Hardware manufacturing 194 Spring and wire product manufacturing 195 Machine shops 196 Turned product and screw, nut, and bolt manufacturing 198 Valve and fittings other than plumbing 201 Fabricated pipe and pipe fitting manufacturing 202 Other fabricated metal manufacturing 207 Other industrial machinery manufacturing 226 Pump and pumping equipment manufacturing 5. Electronic Components 234 Electronic computer manufacturing 235 Computer storage device manufacturing 236 Computer terminals and other computer peripheral equipment manufacturing 244 Electronic capacitor, resistor, coil, transformer, and other inductor manufacturing 6. Environmental equipment 214 Air purification and ventilation equipment manufacturing 216 Air conditioning, refrigeration, and warm air heating equipment manufacturing 250 Automatic environmental control manufacturing 7. Other materials 127 Plastics material and resin manufacturing 136 Paint and coating manufacturing 144 Plastics pipe and pipe fitting manufacturing 151 Rubber and plastics hoses and belting manufacturing 160 Cement manufacturing 8. Scientific and technical services 369 Architectural, engineering, and related services 374 Management, scientific, and technical consulting services 375 Environmental and other technical consulting services 9. Financial and insurance service 357 Insurance carriers 358 Insurance agencies, brokerages, and related activities 359 Funds, trusts, and other financial vehicles 100% 39% 14.48 12.58 11.34 13.42 18.04 18.04 Operation & maintenance — non-fuel 39 Maintenance and repair construction of nonresidential structures 385 Facilities support services 416 Electronic and precision equipment repair and maintenance 417 Commercial and industrial machinery and equipment repair and maintenance Operation & maintenance — electricity 31 Electric power generation, transmission, and distribution Operation & maintenance — natural gas 32 Natural gas distribution Operation & maintenance — coal & petroleum 21 Mining coal 115 Petroleum refineries 119 All other petroleum and coal products manufacturing 100% 19.80 20.08 20% 7% 11.56 11.23 11.23 10.76 11.50 10.02 14.62 9% 13.74 12.74 10.37 14.98 4% 13.34 14.19 18.94 15.09 12.52 13.71 14.79 15.82 12.71 11.09 8.57 11.26 13.37 16.39 6% 13.05 14.68 12.45 3% 8% 4% 14.57 11.27 9.59 11.44 11.40 13.36 11.78 22.08 22.17 20.75 23.15 14.80 11.33 20.31 15.50 levels of consumption by 2014, followed by a gradual decline over the subsequent 20 years. The policy scenarios show greater declines in purchased electricity. In 2035, industrial electricity purchases are forecast to gradually drop by an additional 30.7 billion kWh (4% of the reference case) with the 10% ITC, 66.5 billion kWh (8%) with the 20% ITC, and 105.5 billion kWh (12%) with the 30% ITC (Fig. 4). On the other hand, the CHP-generated electricity sold back to the grid grows as shown in the bottom of Fig. 4. Both the growth of on-site generation electricity sales and the reduction of electricity 21.55 17.77 19.96 100% 5.71 5.71 100% 6.64 6.64 7.43 10.83 5.12 6.82 100% purchased from the grid would lead to overall energy bill savings for industrial CHP users; however, industrial CHP users would also consume more natural gas, the fuel for approximately two-thirds of CHP systems in the U.S. that are coupled with gas turbines or gas-fueled steam turbines. NEMS forecasts that industrial natural gas consumption will grow by 4% in 2035 (relative to the reference case) for the 10% ITC policy, by 10% for the 20% ITC, and by 17% for the 30% ITC. 5.2. Investment Increases The macroeconomic analysis of the three ITC scenarios was developed in a way that converted all changes of CHP capacity and energy consumption into market investment increases and energy bill savings. These investment costs and energy savings were matched to sectoral employment coefficients derived from IMPLAN, as discussed in Section 4. The additional investment in CHP systems is proportional to the net growth of CHP capacity spurred by the ITC policy. The investment cost is calculated by converting the net growth of CHP capacity to dollar value added over the reference case, using the unit of total installation cost for typical gas turbines that is identified by Sentech Inc. (2010) and included in NEMS input files. This typical CHP system has a capacity of 25 MW, and an efficiency of 0.71 in 2010 increasing to 0.74 in 2035. The average total installation cost is the equipment cost excluding O&M and service costs. The equipment cost projections gradually decrease over time, from a high of $1080/kW in 2010 to a low of $905/kW in 2030, reflecting economies of scale, learning by doing, and R&D. Table 4 shows the estimated investment costs for the three ITC policies relative to the reference case. In the reference case, investment costs are forecast to decline over the next two decades from $2.4 billion in 2010 to $1.55 billion in 2035, reflecting both declining CHP system prices and the slightly declining rate of capacity growth shown in Fig. 3. In 2020, the total investment costs could grow by 18% above the reference case with a 10% ITC, by 42% with a 20% ITC, and by 70% with a 30% ITC. The investments in 2035 increase by 16–52% in our three ITC policy scenarios. Non-fuel O&M costs typically include operating labor, routine inspections, scheduled repairs, and preventive maintenance, which are sources of long-term job creation. According to U.S. Environmental Protection Agency Combined Heat and Power Partnership (2008), total O&M costs range from $0.004/kWh to $0.011/kWh for typical gas turbines and are less than $0.005/kWh for steam turbines. Our O&M costs are calculated by applying $0.005/kWh and an 80% capacity factor for the new CHP systems. 148 P. Baer et al. / Ecological Economics 110 (2015) 141–153 1050 Reference Case Industrial CHP Capacity (GW) 100 28% 17% 8% 30% 20yr ITC 20% 20yr ITC 80 10% 20yr ITC 60 40 20% ITC case: Capacity: 79.9->93.5 GW Elec. Price: 6.35->6.33 cents/kWh 20 0 2005 Industrial Purchased Electricity Consumption (Bill kWh) 120 2010 2015 2020 2025 2030 Reference Case 30% 20yr ITC 20% 20yr ITC 10% 20yr ITC 1000 950 900 850 750 2005 4% 8% 12% 20% ITC case: $4,388million (in 2009$) Savings 800 2010 2015 2020 2025 2030 204 2035 Industrial CHP Generation (Bill kWh) 700 31% 19% 9% 600 500 400 300 25% 15% 7% 164 144 124 104 84 64 44 20% ITC case: $1,388 million (in 2009$) gains 24 200 4 2005 100 0 2005 Sales to the Grid (Bill kWh) 184 800 2010 2015 2020 2025 2030 2035 Fig. 3. Total industrial CHP capacity and generation. 5.3. Energy Price Impacts NEMS calculates equilibrium energy prices and quantities across energy fuels and across sectors of end-use demand. Fig. 5 shows how the three ITC policies affect electricity price dynamics across all consumers. Compared with the reference case, the three policies generally lead to decreases in electricity rates, ranging from 0.001 cents/kWh to 0.1 cents/kWh. The effect is variable, however, so for example in some years a 10% ITC policy is shown to slightly increase electricity rates (particularly between 2020 and 2025). These price increases, and the nonlinear response more generally, derive from complex market responses modeled in NEMS such as rebound effects from the electricity price declines and the dynamics of the timing of coal plant retirements caused by the reduction in utility grid sales superimposed on increasing environmental regulations. Volatile electricity and natural gas prices are a sustained source of financial pain for industrial end-users. Energy-efficient CHP systems can be a strategic option to reduce such market threats. At the same time, CHP systems are increasingly cost-competitive with today's glut of shale gas and the forecast for cheap natural gas prices over at least the next several years. In 2012, US natural gas electric power prices dropped to a 10-year low of $2.79 per Mcf (thousand cubic feet) in April; then, reflecting its historic volatility, prices increased by about 50% to $4.36 per Mcf in December.17 Electricity price declines from the expansion of CHP systems can also provide wide economic benefits to residential and commercial end-users. This means that the national market would expect to see induced effects from re-spending of energy cost savings in other sectors of the economy. 5.4. Energy Cost Savings Changes in industrial energy expenditures are calculated from industrial energy consumption and energy price changes (Table 5). 17 EIA, U.S. Natural Gas Electric Power Price: http://www.eia.gov/dnav/ng/hist/ n3045us3m.htm. 2035 2010 2015 2020 2025 2030 2035 Fig. 4. Industrial purchased electricity consumption and sales to the grid. Industrial CHP users would benefit from reduced costs from purchased electricity, coal and petroleum, and from increased revenues from selling excess power to the grid. In contrast, they would spend more on natural gas, the most common fuel for industrial CHP. By comparing Tables 4 and 5, it can be concluded that a 10% and 20% ITC would generate net benefits because total energy cost savings exceed total investment costs (private and public). The 30% ITC is less cost-effective with incremental investment costs exceeding energy savings in both 2020 and 2035. However, when the subsidies are removed as a component of investment costs, reflecting the CHP developer's perspective, the return of energy savings to private investments is nearly favorable in 2020 and clearly favorable in 2035 even in the 30% ITC scenario. The industrial energy savings are less in the 30% ITC scenario because its larger natural gas consumption causes gas prices to rise more aggressively, and these additional costs are offset only slightly by the industrial sector's decreased purchase of electricity and its increased grid sales. 5.5. Jobs Estimation Fig. 6 presents the comprehensive results of the I–O-based jobs analysis of the three ITC policy scenarios. The sectors of construction and CHP equipment installation, non-fuel O&M, and natural gas are all sources of job creation. While the number of one-time jobs in construction and CHP installation slows over time, the number of jobs in O&M and the natural gas sector increase with the expansion of CHP capacity. Furthermore, the potential job creation from energy cost savings in the residential and commercial sectors and industrial cost savings would be sources of substantial benefits for the national economy. These second-order impacts broadly track electricity price changes.18 In general, electricity prices are lower in all sectors in all three of the ITC policy cases compared to the reference case, though there is considerable variability over time. In the 10% ITC policy scenario, electricity prices exceed those in the reference case between 2020 and 2025 18 More precisely they track energy bill savings (or increases) after amortizing new CHP investment costs, but because the income elasticity of energy consumption is low in NEMS, and electricity prices change more than gas or other fuel prices, the electricity price changes dominate the second-order job impacts. 149 P. Baer et al. / Ecological Economics 110 (2015) 141–153 Reference 10% ITC - Private - Gov. subsidy Difference from reference % Growth 20% ITC - Private - Gov. subsidy Difference from reference % Growth 30% ITC - Private - Gov. subsidy Difference from reference % Growth 2010 2020 2035 2419 1776 2102 1682 420 326 18% 2526 2021 505 750 42% 3020 2416 604 1244 70% 1550 1796 1257 539 246 16% 1975 1383 593 425 27% 2363 1654 709 813 52% (Fig. 5), leading to negative second-order impacts in those years; in the 20% and 30% ITC policy scenarios, electricity prices are essentially identical to the reference case in the 2020–2025 period and are significantly below the reference case in other time frames, leading to the secondorder job gains shown in Fig. 6. In contrast, the electric utility sector and (to a much smaller extent) the coal and petroleum production and distribution sectors would experience job losses resulting from enhancing industrial energy efficiency as well as switching fuel consumption to natural gas. Overall, however, these effects are much smaller than the job creation in other sectors; as a result, the net annual increase in jobs (averaged between 2030 and 2035) is estimated to be 10,700 with a 10% ITC, 27,600 with a 20% ITC, and 36,500 with a 30% ITC. 6. Demand Reduction Induced Price Effects (DRIPE) The results presented in Section 5 are based on three subsidy scenarios and on the energy market dynamics specific to those scenarios. In order to provide a more general and scale-independent analysis of the impacts of CHP deployment, and in particular of the second-order impacts related to the DRIPE effect, we developed a statistical analysis relating the quantity of new capacity and generation to the changes in spending patterns and associated employment impacts. This allows us to estimate the employment impacts of new CHP in the baseline, as well as to compare the results of our analysis to similar assessments of the job-creation associated with the deployment of renewables and efficiency technologies. The statistical analysis is based on assessing the relationships between new CHP capacity and the reduction in utility power generation and increased industrial natural gas consumption, and the associated changes in gas and electricity prices across the economy.19 At a theoretical level, the relationships can be described quite simply. Reduction in purchases of utility electricity by industrial consumers, increases in sales back to the grid, and increased purchases of natural gas lead to price changes for gas and electricity in all sectors (Fig. 5). These in turn lead to changes in energy demand and energy expenditures, taking into account the feedbacks in a general equilibrium system. In the real world, or even a model of the real world such as NEMS, the impacts would be much more complex, as industrial electricity purchases are typically based on long term contracts, residential and commercial rates are governed by a wide range of regulatory structures, and the underlying system is based on discrete physical infrastructure such as power plants, transmission lines and pipelines. 19 There are also changes in industrial consumption of coal and petroleum but they are so small in these scenarios that we ignore them. Average Electricity Price to All Users (2009 cents/kWh) 9.6 Table 4 Annual investment cost increases in $2009 M. Reference Case 30% 20yr ITC 20% 20yr ITC 10% 20yr ITC 9.4 9.2 9 8.8 8.6 8.4 2005 2010 2015 2020 2025 2030 2035 Fig. 5. Average electricity price to all users. For our purposes, however, we abstract away from these complexities, and treat the variation from the reference case for each year for each variable as independent data points in a simplified economic model. To provide additional comparisons with the reference case, we ran three scenarios in which symmetrical price increases of 10%, 20% and 30% were applied to the capital costs ($/kW) of new CHP installations. While plainly this is an imperfect statistical treatment – each scenario is a time series with its own auto-correlation, for starters – in the context of a modeling analysis, we believe it provides a more comprehensive basis for a scenario-independent estimation. The results of the statistical analysis are shown in Table 6. Many of the correlations are quite tight; the least explanatory correlation is between changed natural gas consumption and changed natural gas pricing. Using the statistical relationships calculated from the NEMS output and the employment coefficients derived from the IMPLAN data, we are able to generate estimates of the per unit employment impacts of new CHP capacity and generation. CHP facilities are assumed to operate at 80% of their nameplate capacity and have a 20-year operating life, with a non-fuel O&M cost of 0.5 cents/kWh. As in the ITC scenario analysis, changes in industrial costs and revenues and commercial energy bills are assumed to be passed on to households as price reductions or Table 5 Annual industrial energy savings in $2009 M. Annual energy costs 2010 2020 2035 Reference - Purchased electricity - Sales to the grid - Natural gas demand - Coal & petroleum demand 10% ITC - Purchased electricity - Sales to the grid - Natural gas demand - Coal & petroleum demand Annual energy savings % Savings 20% ITC - Purchased electricity - Sales to the grid - Natural gas demand - Coal & petroleum demand Annual energy savings % Savings 30% ITC - Purchased electricity - Sales to the grid - Natural gas demand - Coal & petroleum demand Annual energy savings % Savings 81,135 56,825 −1732 18,975 7068 93,741 57,849 −4080 30,793 9179 93,230 57,068 −4237 31,225 9174 511 0.5% 92,916 56,275 −4498 31,963 9176 825 0.9% 93,139 55,529 −4790 33,224 9176 602 0.6% 98,696 54,315 −9384 44,249 9517 98,212 52,388 −10,082 46,410 9497 484 0.5% 97,896 49,926 −10,772 49,265 9477 800 0.8% 98,537 47,483 −11,691 53,300 9443 160 0.2% 150 P. Baer et al. / Ecological Economics 110 (2015) 141–153 20% ITC 10% ITC 30% ITC 70000 36500 35900 60000 50000 27600 27600 20400 Number of Jobs 40000 30000 21500 20000 2000 0 17700 17700 2015-2019 2020-2024 10700 10700 20000 8300 1500 7000 10000 0 -10000 -20000 -30000 2015-2019 2020-2024 2025-2029 2030-2034 2025-2029 2030-2034 2015-2019 2020-2024 2025-2029 2030-2034 Fig. 6. Estimated employment impacts by scenarios. increases in dividends. To generalize over the period during which the investments and operation take place, we take average expenditures in each sector over the 2015–2035 period as the base against which to calculate changed expenditures due to price changes; we use the midpoint (2025) level to estimate CHP capital costs for our prototype system ($948/kW) and productivity increases in the various sectoral employment coefficients. Based on these relationships and assumptions, we can estimate the changes in expenditures in the industrial, residential and commercial sectors, and the employment impacts of those expenditure changes. The cost changes in the industrial sector generate what we call firstorder jobs, although they include direct, indirect and induced jobs in the jargon of I–O analysis. The net change in industrial costs is then included as a change (increase) in household income. Changes in commercial energy costs are also assumed to be passed to households; and household energy bills change directly through changes in the prices of natural gas and electricity in the residential sector. As noted previously, we assume that the respending of these savings is proportional to average household spending across all income groups, as reported by IMPLAN. Table 7 shows that for every billion kWh of new generation electricity by industrial CHP, the net expenditure change is about a four million dollar decrease in the industrial sector, about a 13 million dollar decrease in the commercial sector, and about a 7.5 million dollar decrease in the residential sector, leading to a net decrease in expenditure/increase in income of about 24 million dollars to all households. Using these figures with the IMPLAN coefficients described above, we can calculate the net employment impacts. As shown in Table 8, around 0.09 first order jobs (full-time, 20-year-equivalent) are created per GWh of new generation, and about 0.23 second order jobs from the respending of household energy bill savings and changes in industrial/commercial costs or dividends passed through to households. Job losses are concentrated in the electricity industry, while job gains accrue in the gas industry and in the remainder of the economy across which consumer purchases are spread. Note that the net second-order impacts are roughly twice as large as the net first-order impacts. 7. Jobs per GWh Comparison With Other studies Direct comparison with other studies is difficult for a variety of reasons. Like installation and operation of other electrical generating equipment, investment in CHP leads to jobs changes in the directly affected sectors. However, precisely because CHP also produces efficiency gains, it generally leads to lower projected prices and thus to energy bill savings and household and business re-spending. Most other studies of jobs from adding renewables do not attempt to estimate these secondorder effects. Thus, while comparison of CHP with renewables would seem to be more appropriately based on only the first-order jobs, which both types of studies generally include, this excludes the efficiency benefits which motivate CHP in the first place. Table 6 Regression analysis of new CHP generation. Bivariate relationships Slope R-square Change in industrial electricity purchases per unit of new CHP −0.002 0.997 generation (Quad Btu/Billion kWh) Changes in industrial electricity sales to grid per unit of new 0.222 1.000 CHP generation (Billion kWh/Billion kWh) Change in industrial gas consumption per unit of industrial 0.007 0.999 CHP generation (Quad Btu/Billion kWh) Change in industrial electricity price per unit of new CHP −0.001 0.549 generation ($ per Million Btu/Billion kWh generation) Change in industrial gas price per unit change of industrial 0.032 0.223 gas consumption ($ per Million Btu/Quad Btu) Change in residential electricity price per unit change of 1.648 0.910 industrial electricity price Change in commercial electricity price per unit change of 1.191 0.952 industrial electricity price Change in residential gas price per unit change of industrial 1.356 0.807 gas price Change in commercial gas price per unit change of industrial 1.325 0.828 gas price Note: Based on F- and t-statistics, with 1 and 154 degrees of freedom, the R-squared values are significant, as is each estimated slope, at a 0.001 level of significance. The 154 degrees of freedom derive from the 26 years of data for each of six scenarios. 151 P. Baer et al. / Ecological Economics 110 (2015) 141–153 Table 7 Expenditure changes (in $2009 M per billion kWh of new CHP generation). Amortization O&M Gas Electricity purchased Electricity sales Net annual Industrial Commercial Residential 8.0 5.0 40.1 −43.4 −13.6 −3.9 1.0 −13.8 1.4 −8.0 −12.8 −7.5 Table 9 Previous studies. Net household Sources CIM O&M and fuel processing Total (job-years/GWh) Laitner and McKinney (2008) Simons and Peterson (2001) −24.2 Note: Net annual expenditures in each sector are summed to provide net change in household income/expenditures (negative numbers are decreases in expense or increases in income). Kammen et al. (2006) Similarly, comparing job estimates from CHP with estimates from the deployment of efficiency-improving technologies (typically in jobs/GWh saved) is complicated because capacity and generation from new CHP is what is reported, not the energy directly saved through efficiency gains. However, since the redirection of spending from energy bill savings is a fundamental driver of employment changes from efficiency investments, comparable studies of efficiency do typically address what we call second-order jobs. Table 9 shows summary data from a variety of previous studies, based on a similar table in Carley et al. (2011) and on Wei et al. (2010). Typical values for renewables range from a low of 0.03 jobs/ GWh to over 1 or even 2 jobs/GWh of generation. 8. Discussion and Conclusions Our goal in this paper was to develop a reproducible analytic method and a practical toolkit for estimating the job impacts of policies aimed at improving the efficiency of energy production and use. More specifically, we sought to examine the job impacts of a sustained federal investment tax credit for combined heat and power. We estimate that a 30% ITC would increase industrial CHP capacity by 22.5 GW in 2035, compared with the reference case, which represents a 28% growth of the total CHP capacity forecast by the reference case in that year. Such a policy would not quite meet the 2012 executive order goal for expanded industrial CHP in 2020, but would meet it by 2023. These policy effects on industrial energy efficiency would be technologically transformational and economically broad. While direct fuel expenditures would rise and more capital would be required for these energy-efficiency upgrades (on the order of $1 billion each year), the purchase of less electricity from the central utility would deliver more than $800 million in additional energy bill savings in 2035 with a 20% ITC; this benefit would drop to $160 million in 2035 with a 30% ITC because gas prices rise significantly, and these additional costs are offset slightly by the industrial sector's decreased purchase of electricity and its increased sales to the grid. In both the reference case and with an ITC policy, our analysis indicates that energy consumption in industrial plants would continue to grow, but the efficiency of CHP systems would result in a slower growth Table 8 Employment coefficients and net employment impacts (first order, second order and total) per GWh of additional CHP generation. Employment Industry Household and Total coefficients (first order) commercial (jobs per $ million (second order) expenditure) Construction 0.6 (converted to FTE) O&M 17.4 Electricity 5.0 Natural gas 5.8 Consumer respending 13.6 Total Technology 0.08 0.09 −0.33 0.25 – 0.09 0 0 −0.11 0.014 0.33 0.23 0.08 0.09 −0.45 0.26 0.33 0.31 Moreno and Lopez (2008) Energy efficiency Wind Geothermal Biomass Solar thermal Solar PV Small hydro PV1 PV2 Wind1 Wind2 Biomass-high estimate Biomass-low estimate Coal Gas Wind Solar PV Biomass-electric 0.17 0.03 0.01 0.01 0.07 0.16 0.03 0.71 0.66 0.05 0.29 0.05 0.05 0.03 0.03 0.17 0.79 0.01 0.09 0.21 0.21 0.06 0.07 0.33 0.14 0.55 0.03 0.03 0.28 0.04 0.08 0.08 0.07 1.54 0.02 0.13 0.22 0.22 0.13 0.23 0.35 0.85 1.21 0.08 0.32 0.32 0.09 0.12 0.11 0.23 2.33 0.03 in energy consumption with an ITC. Employment growth would be significantly higher in the ITC scenarios: by about 36,500 FTEs in 2035 relative to the reference case with a 30% ITC. Furthermore, these employment impacts include significant second-order impacts, which are often overlooked in estimates of employment impacts. The declining average electricity prices that result from the deployment of more CHP enable energy bill savings in the residential, commercial, and industrial sectors, which generate induced jobs. The job estimates per GWh of new generation provide interesting insights to assess the cost-effectiveness of ITC policies by sectors. The ITC investment would provide positive sources of job creation in the first-order from construction and installation (0.08 FTE/GWh) and operations and maintenance (0.09 FTE/GWh) required for the increased CHP generation. It would reduce jobs from centralized plant generation (− 0.45 job-years/GWh) and increase jobs from the consumption of more natural gas (0.26 job-years/GWh), and would create secondorder jobs from household and commercial respending (0.33 jobyears/GWh). Many of the limitations of this type of study are well known, and for these reasons the results should not be taken as firm predictions. Plainly it would be desirable to perform a wide range of sensitivity and uncertainty analyses, as the results are dependent both on the models used (NEMS, IMPLAN) and the particular parameter choices used. In our analysis, the uncertainty of expected growth of CHP deployment would be constrained by the inputs embedded in NEMS, which are developed asynchronously and do not necessarily reflect the most recent empirical data or projections (Wilkerson et al., 2013). Our job estimates also rely on the variation of exogenous inputs—such as energy prices, job coefficients, and productivity. Such uncertainties have been acknowledged as inherent limitations of the I–O modeling method, as we reviewed in Section 3. To address the uncertainty from scenario assumptions, we applied NEMS's sensitivity analyses on industrial CHP deployment. Industrial CHP deployment could vary by system costs, energy prices, and especially natural gas prices, regardless of the level of tax incentive. Given the historic volatility of natural gas prices, we examined a range of gas price forecasts. AEO 2011 provides a range of natural gas supply scenarios with 50% higher or lower recovery of shale gas relative to the reference case. To model the sensitivity and uncertainty of CHP deployment associated with natural gas prices, we combined our 20% ITC scenario with these AEO side cases. Fig. 7 shows the resulting changes of CHP generation and natural gas price. With a 50% higher recovery of shale gas, gas 152 P. Baer et al. / Ecological Economics 110 (2015) 141–153 900 13 Reference Case 20% 20yr ITC High Shale Gas High Shale Gas+20% 20yr ITC Low Shale Gas Low Shale Gas+20% 20yr ITC 11 10 9 700 600 8 7 500 30% 16% 40% 400 8% 300 6 5 200 4 100 3 2005 Reference Case 20% 20yr ITC High Shale Gas High Shale Gas+20% 20yr ITC Low Shale Gas Low Shale Gas+20% 20yr ITC 800 Bill kWh 2009$/Thousand Cubic Feet 12 2010 2015 2020 2025 2030 2035 0 2005 2010 2015 2020 2025 2030 2035 Fig. 7. Industrial natural gas price (left) and industrial CHP generation (right). would be cheaper, declining from $7.2/thousand cubic feet (Mcf) to $5.6/Mcf in 2035. In this case, electricity generation from CHP in 2035 could grow by 16% relative to the reference case, and by 30% with the addition of a 20% ITC support. Industries are more motivated to install CHP systems when the natural gas fuel prices are lower. In contrast, when natural gas price increase with lower shale gas recovery, there is 40% less CHP generation in 2035, but with the introduction of a 20% ITC, some of that loss is recovered. From this sensitivity analysis, we illustrate that exogenous factors can strongly influence investments in CHP capacity and generation, but we have shown that the ITC has a predicable direction of influence under a range of natural gas conditions. Further work is needed to more fully understand the consequences of diverging future natural gas supplies on energy prices, utility bills, and thus employment. Acknowledgments The authors gratefully acknowledge support for this research that was provided by Oak Ridge National Laboratory (Melissa Lapsa, Project Manager, contract number: 4000105765), the and Georgia Institute of Technology's Institute of Paper Science and Technology (IPST). The advice of Norman Marsolan, Director of the IPST was particularly valuable in addressing our case study of the pulp and paper industry and combined heat and power. The views expressed in this paper, and any errors, are attributable entirely to the authors. References Blakely, E.J., Leigh, N.G., 2009. Planning Local Economic Development: Theory and Practice. 4th ed. Sage Publications, Inc. Brown, M.A., Sovacool, B.K., 2011. Climate Change and Global Energy Security. MIT Press (Chapter 6). Brown, Marilyn A., Cox, Matt, Baer, Paul, 2013. Reviving manufacturing with a federal cogeneration policy. Energy Policy 52, 264–276. Bureau of Labor Statistics, 2013. Green Goods and Services News Release. Department of Labor, U.S. Carley, S., Lawrence, S., Brown, A., Nourafshan, A., Benami, E., 2011. Energy-based economic development. Renew. Sust. Energ. Rev. 15 (1), 282–295. http://dx.doi.org/10. 1016/j.rser.2010.08.006. Chittum, A., Kaufman, N., 2011. Challenges Facing Combined Heat and Power Today: A State-by-State Assessment (Vol. 20045). Retrieved from. http://www.uschpa.org/ files/public/ie111.pdf. Claussen, E., Peace, J., 2007. Energy myth twelve — climate policy will bankrupt the U.S. economy. In: Sovacool, B.K., Brown, M.A. (Eds.), Energy and American Society — Thirteen Myths. Springer, pp. 311–340. Cox, M., Brown, M.A., Jackson, R., 2011. Regulatory reform to promote clean energy: the potential of output-based emissions standards. Proceedings of the ACEEE Summer Study on Energy Efficiency in Industry. American Council for an Energy-Efficient Economy (ACEEE), Niagara Falls, NY, pp. 1-57–1-67. Diamond, J., 2005. Collapse: How Societies Choose to Fail or Succeed. Penguin Books. Dietz, T., 2010. Narrowing the US energy efficiency gap. Proc. Natl. Acad. Sci. U. S. A. 107 (37), 16007–16008. http://dx.doi.org/10.1073/pnas.1010651107. Geller, H., DeCicco, J., Laitner, S., 1992. Energy Efficiency and Job Creation: The Employment and Income Benefits from Investing in Energy Conserving Technologies. Washington, D.C. Goodstein, E., 1999. The Trade-off Myth: Fact and Fiction About Jobs and the Environment. Island Press, Washington, D.C. Hirst, E., Brown, M., 1990. Closing the efficiency gap: barriers to the efficient use of energy. Resour. Conserv. Recycl. 3 (4), 267–281. http://dx.doi.org/10.1016/09213449(90)90023-W. Houser, T., 2013. American Energy Efficiency: The Economic, Environmental and Security Benefits of Unlocking Energy Efficiency. ICF International, 2011. Combined Heat and Power Installation Database. Retrieved from. http://www.eea-inc.com/chpdata/. I.C.F., International, 2010. Effect of a 30 Percent Investment Tax Credit on the Economic Market Potential for Combined Heat and Power. Jaffe, A.B., Stavins, R.N., 1994. The energy-efficiency gap: what does it mean? Energy Policy 199422 (10), 804–810 (Retrieved from http://www.hks.harvard.edu/fs/rstavins/ Papers/TheEnergyEfficiencyGap.pdf). Kammen, D.M., Kapadia, K., Fripp, M., 2006. Putting Renewables to Work: How Many Jobs Can the Clean Energy Industry Generate? University of California Berkeley, Renewable and Appropriate Energy Laboratory. Correction of 2004 original Retrieved from http://rael.berkeley.edu/sites/default/files/old-site-files/2004/Kammen-RenewableJobs-2004.pdf Kim, G., Baer, P., Brown, M.A., 2013. The Statewide Job Generation Impacts of Expanding Industrial CHP. Proceedings of the American Council for an Energy Efficient Economy (ACEEE) Summer Study on Energy Efficiency in Industry. American Council for an Energy Efficient Economy, Niagara Falls, NY. Kramer, C., Reed, C., 2012. Ten Pitfalls of Potential Studies. Energy Future Group. Laitner, J.A. “Skip,”, 2009. Climate change policy as an economic redevelopment opportunity: the role of productive investments in mitigating greenhouse gas emissions. ACEEE Report E098. American Council for an Energy-Efficient Economy, Washington, DC. Laitner, J.A. “Skip”, McKinney, V., 2008. Positive Returns: State Energy Efficiency Analyses Can Inform U.S. Energy Policy Assessments. Washington, D.C. Laitner, S., Bernow, S., Decicco, J., 1998. Employment and other macroeconomic benefits of an innovation-led climate strategy for the United States. Energy Policy 26 (5), 425–432. Laitner, J.A. “Skip”, Gold, R., Nadel, S., Langer, T., Elliott, R.N., Trombley, D., 2010. The American power act and enhanced energy efficiency provisions: impacts on the U.S. economy. ACEEE Report E103. American Council for an Energy-Efficient Economy, Washington, DC. Lehr, U., Nitsch, J., Kratzat, M., Lutz, C., Edler, D., 2008. Renewable energy and employment in Germany. Energy Policy 36 (1), 108–117. http://dx.doi.org/10.1016/j.enpol.2007. 09.004. Lewis, G.M., 2010. Estimating the value of wind energy using electricity locational marginal price. Energy Policy 38, 3221–3231. Miller, R.E., Blair, P.D., 2009. Input–Output Analysis: Foundations and Extensions. 2nd ed. Prentice Hall, Cambridge. Moreno, B., López, A.J., 2008. The Effect of Renewable Energy on Employment. The Case of Asturias (Spain). Renewable and Sustainable Energy Reviews 12 (3), 732–751. http:// dx.doi.org/10.1016/j.rser.2006.10.011. Pollin, R., Garrett-Peltier, H., Heintz, J., Scharber, E., 2008. Green Recovery: A Program to Create Good Jobs and Start Building a Low-carbon Economy. Center for American Progress. Rhodium Group, 2013. American Energy Productivity: The Economic, Environmental and Security Benefits of Unlocking Energy Efficiency. Prepared on Behalf of the Alliance to Save Energy. February 2013, available at. https://www.ase.org/sites/ase.org/files/rhg_ americanenergyproductivity_0.pdf. Rockstrom, J., Steffen, W., Noone, K., Persson, A., Chapin, F.S., Lambin, E.F., Lenton, T.M., et al., 2009. A safe operating space for humanity. Nature 461, 472–475. Sciortino, M., Neubauer, M., Vaidyanathan, S., Chittum, A., Hayes, S., Nowak, S., Molina, M., et al., 2011. The 2011 State Energy Efficiency Scorecard (Vol. 20045). Retrieved from. http://www.aceee.org/sites/default/files/publications/ researchreports/e115.pdf. P. Baer et al. / Ecological Economics 110 (2015) 141–153 Sentech Inc., 2010. Commercial and Industrial CHP Technology Cost and Performance Data Analysis for EIA. Washington, D.C. Simons, G., Peterson, T., 2001. California Renewable Technology Market and Benefits Assessment. Palo Alto, CA and Sacramento, CA: Electric Power Research Institute (EPRI) and California Energy Commission (CEC). Retrieved from http://www.energy.ca.gov/ 2001publications/CEC-500-2001-005/CEC-500-2001-005.PDF. Shipley, A., Hampson, A., Hedman, B., Garland, P., Bautista, P., 2008. Combined Heat and Power: Effective Energy Solutions for a Sustainable Future. Oak Ridge, TN. Retrieved from. http://www1.eere.energy.gov/manufacturing/distributedenergy/pdfs/chp_ report_12-08.pdf. Sorrell, S., Dimitropoulos, J., Sommerville, M., 2009. Empirical estimates of the direct rebound effect: a review. Energy Policy 37, 1356–1371. Steinhurst, W., Sabodash, V., 2011. The Jevons Paradox and Energy Efficiency. Synapse Energy Economics, Inc., Cambridge, MA. The Committee on Climate Change Science and Technology Integration (CCCSTI), 2009. Strategies for the Commercialization and Deployment of Greenhouse Gas Intensityreducing Technologies and Practices. Washington, DC. Retrieved from. http://www. energetics.com/resourcecenter/products/studies/samples/Documents/strategies_ greenhouse_report.pdf. The White House, 2012. Executive Order — Accelerating Investment in Industrial Energy Efficiency. Office of the Press Secretary (Retrieved from http://www.whitehouse.gov/ the-press-office/2012/08/30/executive-order-accelerating-investment-industrialenergy-efficiency). 153 Turner, K., 2009. Negative Rebound and Disinvestment Effects in Response to an Improvement in Energy Efficiency in the UK Economy. Energy Economics 31 (5), 648–666. http://dx.doi.org/10.1016/j.eneco.2009.01.008. U.S. Energy Information Administration (EIA), 2011. Annual Energy Outlook 2011: With Projections to 2035. Available at http://www.eia.gov/forecasts/archive/aeo11/. U.S. Energy Information Administration (EIA), 2012. Annual Energy Outlook 2012 with Projections to 2035. Available at http://www.eia.gov/forecasts/archive/aeo12/. U.S. Environmental Protection Agency (EPA), 2012. Combined Heat and Power: A Clean Energy Solution. Retrieved from http://www1.eere.energy.gov/manufacturing/ distributedenergy/pdfs/chp_clean_energy_solution.pdf. U.S.Environmental Protection Agency Combined Heat and Power Partnership, 2008. Catalog of CHP Technologies. http://www.epa.gov/chp/documents/catalog_chptech_full. pdf. Wei, M., Patadia, S., Kammen, D.M., 2010. Putting renewables and energy efficiency to work: how many jobs can the clean energy industry generate in the US? Energy Policy 38 (2), 919–931. http://dx.doi.org/10.1016/j.enpol.2009.10.044. White, S., Walsh, J., 2008. Greener pathways — jobs and workforce development in the clean energy economy. Retrieved from. http://www.cows.org/pdf/rp-greenerpathways.pdf. Wilkerson, J.T., Cullenward, D., Davidian, D., Weyant, J.P., 2013. End use technology choice in the national energy modeling system (NEMS): an analysis of the residential and commercial building sectors. Energy Econ. 40, 773–784.