Feasibility of Zero Energy Multi-Family Low Income Housing Tax Credit (LIHTC) Developments

The residential sector in the United States is in need of comprehensive policy-making reforms that concurrently address housing affordability and environmental sustainability. This study investigates the feasibility of state-wide zero-energy affordable housing by analyzing historical data on climate, energy use, and solar system costs in the Commonwealth of Virginia. The hypothesis examined is that the net present cost of implementation of rooftop residential solar systems to achieve zero-energy Low-Income Housing Tax Credit (LIHTC) buildings is lower than the discounted present cost of energy of otherwise identical conventional buildings that run without renewable energy generation systems. The authors propose a generalizable framework for analyzing the feasibility of achieving region- or state-wide zero-energy LIHTC developments. To validate the framework, the authors employ a longitudinal sample of monthly energy use data from 2013-2016 obtained from 310 residential units of 15 LIHTC developments across the state. Based on statistical regression analysis, energy simulation, and simulation-based risk analysis, the authors find that the net present value of zero-energy LIHTC investments can be positive with a low risk. The investment value varies often depending on the zero-energy building definition, weather characteristics, retail price of electricity, and incentive rate. This study can help housing policymakers and industry professionals analyze and benchmark the feasibility of innovative zero-energy housing policies and projects.


Introduction
While renewable energy is rapidly growing in worldwide adoption, approximately 89 percent of the energy consumed in the US in 2019 still came from non-renewable sources like coal, gas, and nuclear power [1]. Of that energy produced, the residential sector accounted for approximately 21 percent of consumption. Simultaneously, almost half of all renter households are cost-burdened, often due to poverty and rapidly rising housing prices in large metro areas [2]. The residential sector in the US is clearly in need of comprehensive policy-making reforms that concurrently address affordability and environmental sustainability, reducing uncertainties around the national economy, energy security, declining natural resources, and climate change.
In 1986, the US Congress enacted the Low-Income Housing Tax Credit (LIHTC) program, which provides state LIHTC-allocating agencies with approximately $8 billion per year to issue tax credits for the acquisition, rehabilitation, or new construction of rental housing for low-income households [3]. With more than 2.4 million active units [4], LIHTC is the principal federal housing production program, which incentivizes the production of a significant portion of below-marketrate multifamily rental units for extremely low-income to low-income households based on an indirect federal subsidy. Residents that qualify to live in a LIHTC unit benefit based on income and quality depending on their certified annual income and the maximum rent set by the project [5]. McClure (2019) describes the basics of the LIHTC program [6].
Energy-efficiency is increasingly considered important in LIHTC Qualified Allocation Plans (QAPs), which outline the criteria based on which state housing agencies allocate financial incentives in the form of tax credits to multi-family residential developers [7]. The basic federal criteria included in QAPs do not mandate specific energy-efficiency standards, but additional criteria that support state housing policy goals often include energy-efficiency incentives. Previous studies show QAPs have significant impacts on the location and quality of LIHTC developments [8]. As a result, QAPs could either promote or inhibit the application of innovative designs and technologies in low-income developments. Existing literature on the LIHTC program suggests that, among other co-benefits, energy-efficient LIHTC properties have generated considerable financial savings for the occupants [9]. Despite its significance as an opportunity to drive innovative solutions to address environmental sustainability and economic development in low-income housing in the U.S., zero-energy buildings have received little attention in low-income housing literature, policy, and practice. If proved to be more cost-effective, zero-energy developments benefit residents and society as a whole by increasing affordability, environmental sustainability, and efficient distribution of federal financial incentives.
This article explores the existing gap to zero energy in the context of the LIHTC program: how much should state housing agencies raise energy efficiency and renewable energy requirements to bridge the gap to zero-energy LIHTC apartments?; how much does a zero-energy LIHTC building cost to developers at the present time?; is a zero-energy LIHTC building more cost-effective than a conventional building over its lifecycle?; and what types of additional support does the construction industry need to bridge the gap to zero-energy? The main hypothesis examined is that the net present cost of the implementation of rooftop residential solar systems to achieve zeroenergy LIHTC buildings is lower than the discounted present cost of energy of otherwise identical conventional buildings that run without renewable energy generation systems. The authors propose a generalizable framework for analyzing the feasibility of achieving regional or state-wide zero energy in LIHTC developments. The framework is then validated using empirical data in a proofof-concept case study focused on the rooftop grid-connected photovoltaic (PV) system as the only 4

Energy-efficient low-income housing
In recent years, environmental concerns and the rising cost of energy have prompted a shift towards more energy-efficient homes regardless of end-users' finances. There are major reasons for decision-makers to consider energy-efficient housing for low-income residents. First, sustainable development requires a balanced integration of economic, environmental, and social goals.
Nonetheless, environment and climate policy sectors alone are not capable of achieving all of the objectives and thus must work with traditionally siloed policy sectors, such as low-income housing [11,12]. Second, due to access to resources and information, public officials are increasingly responsible for taking the initiative to recognize and deploy innovative policies and programs, whether or not such utilities are expressed wants of all citizens [13].
Similarly, empirical evidence suggests that energy and water efficiency significantly reduce the total cost of living for low-income households throughout the building life-cycle [14,15]. Lifecycle thinking is particularly consistent with the way LIHTC developments are financed, and research suggests energy-efficient LIHTC units built to high-performance building standards (e.g., ASHRAE 90.1, ASHRAE 90.2) and rating systems (e.g., Energy Star for Homes, LEED, EarthCraft) can be more cost-effective than conventional units for stakeholders and the society as a whole in the long term [16,17]. LIHTC-financed housing units are charged a flat rent depending on the Area Median Income (AMI), typically close to 30 percent of household income for gross housing costs, including utilities. Therefore, if a LIHTC tenant's income decreases, they will be spending more than 30 percent on the monthly rent. Resident behavior and rising energy prices can increase utility costs, and the extra amount taken out of residents' monthly income and spent on utilities impacts housing affordability and economic well-being [18,19]. Some LIHTC projects convert to market-rate units after the first 15-year compliance period due to lack of resources and funding to replace building systems in need of repair. Life-cycle savings from the adoption of durable energy-efficient building systems could help increase housing affordability, preserve the affordable units, and prevent the relocation of residents at the end of the compliance period [20].
Despite potential benefits, energy-efficient low-income housing units have positive externalities, meaning multiple factors, including split incentives, information lags, and asymmetries, risk aversion, skill shortages, and analytical failures. These externalities can lead to underinvestment in these units in the free market. Thus, government intervention is needed for the efficient distribution of social benefits [21,22]. Although the benefits of energy-efficient building for society as a whole are often larger than the average cost premium to obtain energy certifications, residential developers demand significant price premiums that are likely to affect affordability [23].
For instance, researchers have often reported up to 10 percent sales price premiums associated with energy-efficient single-family units with green building certifications in US cities [24,25].
Empirical studies suggest that financing initiatives highly affect (as much as 100 percent) the adoption of renewable energy generation systems in the residential sector [26]. Low-income housing legislation aimed at promoting sustainability and preserving housing affordability could balance energy-efficiency and renewable energy needs pending the careful assessment of the required levels of incentive and supportive financing mechanisms.

Zero-energy buildings
Built on low-energy building research which dates back to the 1940s, zero-energy buildings have received growing attention in recent years and are being incentivized and mandated in the housing sector to help developed countries, including the US, achieve carbon neutrality [27,28]. Zeroenergy residential buildings use renewable sources of energy to generate enough energy to offset the operation-phase energy demand, which normally includes heating, cooling, ventilation, domestic hot water, fixed lighting, plug loads, and elevators. An alternative and/or complementary to on-site renewable energy generation is a utility-scale renewable energy grid, which benefits from scale economies and equalizes local peaks.
In 2010, the European Union (EU) directive on the energy performance of buildings asserted that member states should ensure all new buildings are nearly zero-energy buildings by 31 December 2020 [29]. In 2008, California's long-term energy-efficiency strategic plan adopted energy efficiency goals stating all new residential construction in California will be zero-energy by 2020, which became a mandate in 2018 [30]. Past research concludes that energy use in the operation phase roughly accounts for 70-90 percent of a building's life cycle energy use, showing the significance of zero-energy buildings in mitigating greenhouse gas emissions. Considering the embodied energy of electricity grids, energy savings achieved in the operation of zero-energy buildings is considerably larger than the increase in embodied energy when switching from passive to zero-energy houses [31].
The US Department of Energy (DOE) recognizes builders for leadership in increasing energy efficiency, improving indoor air quality, and making homes zero-energy ready through the Zero Energy Ready Home program [32]. The DOE defines a zero-energy building as "an energyefficient building where, on a source energy basis, the actual annual delivered energy is less than or equal to the on-site renewable exported energy" [33]. Accordingly, delivered energy refers to "any type of energy that could be bought or sold for use as building energy", and exported energy is "on-site renewable energy supplied through the site boundary and used outside the site boundary" [33]. In the DOE's definition, the source energy is composed of the delivered energy, plus the energy consumed in the extraction, processing, and transport of primary fuels plus energy losses in thermal combustion and in power generation plants plus energy losses in transmission and distribution to the building site [34]. An implicit assumption in this definition is that zero-energy buildings are not fully autonomous buildings and need to remain connected to existing energy grids that are able to exchange energy with the building. According to this definition, the embodied energy of building materials, construction, and recycling, may or may not be considered in zeroenergy building calculations.
Zero-energy building balance is a condition in which the sum of all generated energy is equal to or higher than the sum of all building energy loads over a period, nominally a month or a year.
Load matching refers to the temporal match between load and generation and represents the building's ability to work dynamically, in synergy with the grid. Salom et. al (2011) reviews various load matching indicators [35]. Equations (1) and (2) represent zero-energy building balance, where and stand for exported and delivered energy, and stand for generation and load, and and stand for weighting factor and energy carrier. , , , and stand for weighted exported energy, weighted delivered energy, weighted generation, and weighted load, respectively. Equation (3) represents load matching index, where is the load matching index, is the time interval used, e.g. hour, day or month, and stands for the number of data samples (12 for monthly and 8760 for hourly time interval). For guidance and information on the grid interaction index equation, see Sartori et al. (2012) [36].
To maximize a zero-energy building's operation at the balance condition over time, the demand should be reduced by the means of energy efficiency, and the supply should be increased by the means of renewable energy systems. Although zero-energy buildings are often designed to achieve energy balance on an annual basis, increasing the generation-load match in smaller time scales using optimized energy storage systems reduces zero-energy buildings' stress on the existing grid, particularly at the times of peak energy demand, and reduces energy waste and carbon-related emissions [37]. Past research has examined different equations and weights to define the zeroenergy building balance, including balances of primary energy, site energy, carbon emissions, and energy cost for different energy systems [38,39]. Data from the International Energy Agency

Cost-benefit analysis and simulation-based risk analysis
Researchers have examined the feasibility of utilizing renewable energy systems for residential and commercial purposes, while often in isolation [40,41]. The common approach is to evaluate the integration of one or more renewable energy systems like solar systems, wind turbines, heat pumps, district heating and cooling into new or existing developments using simulations and calculating the energy payback period or levelized cost of energy in each scenario [42]. Amasyali & El-Gohary (2018) reviewed past research on data-driven building energy consumption prediction [43]. Some have used simulation software, such as EnergyPlus, HOMER, and TRNSYS, in conjunction with daily or monthly load profile data obtained from study cases. Most studies have focused on a single building and fewer studies have considered multiple developments with similarities and differences. Cost-benefit analyses include multiple non-financial criteria, such as the availability of technology, warranty period, ease of maintenance, required roof or ground area, avoided carbon emissions, and sensitivity analyses include the effect of changing key input variables for project outcomes, such as the effects of the cost of delivered energy, the value of exported energy, the introduction of a carbon tax, and the price of power generation technology.
Fewer studies include an assessment of risk and uncertainty (e.g., the occurrence of alternative future events, into financial analysis) and most available studies rely on performing simple financial analyses [44].
As a powerful tool for evaluating the effects of policy decisions, cost-benefit analysis relies on estimates about variables that are not accurately predictable, e.g., future prices, weather patterns, occupant behavior [45]. To effectively contribute to rational decision making, an analysis (e.g., scenario analysis, sensitivity analysis, Monte Carlo simulation) of risks and uncertainties associated with observed data and modeling errors that could change the project outcome and identify the range of possible outcomes should be an integral part of any cost-benefit analysis [44,[46][47][48]. A conundrum about cost-benefit analysis is the choice of discount rate for future costs and benefits of projects. The United States (US) Government recommends deterministic cost-benefit models followed by sensitivity analyses and Monte Carlo simulations to perform a full risk analysis of costs and benefits of policy interventions. US Government guidelines for cost-benefit analysis of federal programs have required a discount rate of 7 percent, which is supported by more recent research suggesting discount rates in the range of 6-8 percent [49,50].
The National Renewable Energy Laboratory (NREL) benchmarks average U.S. solar photovoltaic (PV) system installed costs weighted by state installed capacities, accounting for all system and project development costs incurred from the perspective of the developer/installer, including profit in the cost of the hardware, along with the profit the installer/developer receives. According to NREL, residential rooftop systems, ranging from 3-10 kW in size, cost $2.71 per watt DC (Wdc) or $3.11 per watt AC (Wac) in the first quarter of the year 2018. Figure 1 presents the NRELestimated inflation-adjusted national average rooftop system costs in 2018 dollars. The gradual decrease in total cost has resulted from various drivers working simultaneously, including increases in module efficiency, labor productivity, small installer market share, and decreases in supply chain cost, permitting cost, and structural and electrical balance of system commodity pricing. According to the benchmarking analysis, the total cost of residential PV systems has decreased by 4.91 percent, from 2017-2018, and it can be expected that in 2019 and 2020, the total cost per watt DC (Wdc) has decreased to ~$2.57 and ~$2.45, respectively. Since on-site electricity storage systems significantly increase the total cost of installation, the most common approach to zero-energy building, particularly where full retail net metering is not available, is to simply rely on the electricity grid [51]. Additional data and information on residential solar photovoltaics with energy storage systems are available in NREL's installed cost benchmarks [52]. Information on major trends in the U.S. solar industry can be found in the Solar Energy Industries Association's (SEIA) solar market insight reports [53].

Analysis framework
Several US states, including Virginia, have enacted legislation demanding the production of 100 percent of electricity from carbon-free sources by 2050. The flowchart depicted in Figure 2 presents a generalizable framework for analyzing the feasibility of achieving region-or state-wide zero-energy in LIHTC developments. Since the future life-cycle values included in the feasibility analysis of any hypothesized study cases are unknown, the framework relies on values that have the greatest probability of occurring as suggested by historical data and provides reasonable estimates for modeling purposes. The authors apply and validate this framework using empirical data in a proof-of-concept case study to test the study hypothesis and answer the research questions.    [56,57]. All the required weather information -hourly, or sub-hourly with up to one-minute time resolution -for the selected cities is obtained based on the closest weather station data available in the current version of the National Solar Radiation Database (NSRDB) and added to the software.

Feasibility analysis
The discounted cash flow (DCF) analysis tests whether the net present cost of the implementation of rooftop residential solar systems to achieve zero-energy LIHTC buildings would be lower than the discounted present energy cost of an otherwise identical building with no renewable energy generation systems. The first approach to feasibility analysis is to obtain the base case deterministic Net Present Value (NPV) of investment in a solar system that its generated energy exceeds the  Table 1 Table 2 presents descriptive statistics of historical data for state-wide retail prices of electricity for residential use in Virginia from 1990 to 2018. City-specific mean retail prices of electricity are used for financial analysis. Table 3 introduces descriptive statistics of the sample data.    A1). When adjusting the price using the Consumer Price Index (CPI) of electricity for urban consumers in US cities, the real price of electricity has, on average across the US, decreased by 0.30 percent per year during the same period [58]. Whether this descending trend is going to remain in place is not quite clear, while an increasing trend in the price means PV system users would accrue more financial benefits than initially estimated. The US EIA has developed three different scenarios, suggesting that the national average price of 10.53 cents/kWh in 2018 can range anywhere from 9.7-11.6 cents/kWh in 2050 depending on economic growth, oil and gas resource prices, and renewable energy generating technology adoption rate [59]. These scenarios are considered in the current feasibility study.   Figure 4 depicts the predicted energy load (left) and residuals (right) for each observation.

Load profiles
Together, the regression models and the scatter plots suggest uncontrolled variables have large impacts on energy load in LIHTC apartments. Obtained from the conditional mean model, Figure   5 shows the predicted correlation of HERS score with monthly energy load with a 95 percent confidence interval, alluding to the practical significant role of energy-efficiency measures in reducing the operation phase energy use and carbon emissions. Holding all variables at the mean, each one-unit change in HERS score changes energy load by ~4.61 kWh per month.  Table 4.
Similarly, HERS and unit size categories correspond to the first, second, and third quartiles of the corresponding variables. Table A2 presents Table 5 and Table 6 present the selected PV module and inverter specifications currently available in the market and the design characteristics of the PV system at the reference location. The energy generated by the system exceeds the annual energy load of a new family-occupied apartment at the median unit size and the median HERS score during the first 30 years of operation. The annual balance partially restores energy loss in the infrastructure system, but to restore all the estimated energy loss (~32.31 percent) discussed previously, each apartment's array size must increase from 6.5kW(dc) to 7 kW(dc), which increases each module area from 28 m 2 to 42 m 2 . The 0.5 kW increase in capacity and 50 percent increase in area can increase the investment profit but may not be practical due to limited rooftop space on the 2-story multi-family building. Figure 7 shows a monthly pattern of energy generation and load in the reference location.   The red bars represent negative monthly balance, e.g., generation shortage to fully offset the energy load.

Discounted cash flow analysis
The study case is a two-story 60-unit LIHTC building comprising small (693 sf), medium (910 sf), and large (1087 sf) apartments in equal numbers. The total residential load is estimated from the sum of predicted loads for all given apartment sizes multiplied by the number of apartments of each size (20) in the building. Table 8    The analysis suggests that with a long-term financing mechanism available, the net present cost of the zero-energy LIHTC apartments in Richmond, VA is lower than that of otherwise identical conventional apartments that have no renewable energy generation systems. Table 9 presents the results of the 30-year Monte Carlo cash-flow analysis of the six climatic divisions based on the same assumptions presented in Table 8 with randomized location-specific energy price, load, generation. Except for Norton, where the combination of low solar radiation, low ambient temperature, and low energy price reduces the value of investment, the NPV of investment in other cities is positive at zero or low risk of loss. Table A3 describes the two-way sensitivity of the simulated NPVs to installation cost and price of exported energy in Richmond, VA and Table 10 describes the sensitivity of the simulated NPVs to installation cost assuming equal prices for   Table A4 presents the results of a statistical two-sample t-test with equal variances for Richmond, VA at 26 percent incentive, suggesting that the NPV obtained from subtracting the net present cost of the implementation of the rooftop residential PV systems from the discounted present cost of energy of the same apartments without PV systems is statistically significant.

Key results and comparison with previous research
Based on the 30-year discounted cash-flow analysis, the net present cost of the implementation of rooftop residential solar systems to achieve zero-energy LIHTC apartments in Virginia can be lower than the discounted present cost of energy of otherwise identical apartments that have no renewable energy generation systems, and the value of cost savings can be statistically significant.
The investment value often depends on the zero-energy building definition, weather characteristics, retail price of electricity, and incentive rate. The DOE's definition of zero-energy building implies that these buildings should progress the restoration of the 32.31 percent loss (based on data from the 2010s) to the infrastructure system. A potential problem with this requirement in Virginia, where the solar potential is lower than that of states located in the South East, South West, and Although zero-energy LIHTC buildings can be more cost-effective than otherwise identical LIHTC buildings in the long run, the costs associated with PV systems are not evenly distributed in the state, leaving some areas like the Southeast more receptive to adoption of distributed generation systems than others like the Southwest. Differences in solar radiation, ambient temperature, energy prices, and diminishing prospects of the federal solar tax credit can affect consumers' willingness to invest in PV systems. Analyzing these variations and the significance of carbon emissions avoided as a result of renewable energy generation is a key to support the construction industry in bridging the gap to zero-energy buildings, particularly in low-income housing.
Achieving annual balance of primary energy is a required step toward performance improvement, but zero-energy LIHTC design and upgrade considerations can include an optimized combination of power storage systems, mechanical and electrical operation control systems, and additional energy-efficiency measures that help buildings achieve balance in all individual months, reduce stress on existing power infrastructure, reduce greenhouse gas emissions, and increase the building's economic value [64,65]. The analysis of HERS score presented here suggests that further reductions in the embodied energy are achievable by applying construction methods, systems, and components that are more energy-efficient than conventional ones. Virginia's solar potential is not as abundant as that of other US states located at lower latitudes, and achieving a cost-effective balance that compensates for losses in primary to final energy conversion may not be widely implemented soon using distributed generation but the potential to reduce energy demand by improving design and construction standards exists. Energy storage systems could increase the potential value of residential systems at the building level (e.g., through allowing for back-up power in the event of a grid outage and reducing residential demand charges through load shifting) and at the grid level (e.g., allowing for voltage and frequency regulation, deferred infrastructure investment, and resource adequacy) [52].

Zero-energy LIHTC opportunities and barriers
Nonetheless, PV-plus-storage systems are still expensive to purchase, the permitting process for installing and operating storage devices are complicated, and the interconnection and net metering is complex and costly. Finally, the lack of longitudinal performance data and regulatory nature of state QAPs creates uncertainty for developers and limits their willingness to test innovative technologies and approaches toward zero-energy LIHTC developments.

Recommendations
Beyond variability in consumption and generation rates, metering infrastructure, (e.g., sub or master-metered), impacts the value of investment in zero-energy LIHTC developments. In submetered developments each apartment's utility usage is measured individually and tenants pay through accounts they themselves establish with utility companies (Figure 9a). In master-metered developments ( Figure 9b) the owner pays for all the utility costs and passes the costs on to individual tenants within their rent costs [65]. Within each LIHTC application, the developer identifies if the utilities will be master or sub-metered. There is little incentive to invest in energyefficiency and renewable energy systems in master-metered developments since the developers cannot distribute additional costs on tenant rents due to the competitive nature of many rental assistance or housing production programs. a) b) Figure 9 Common zero-energy LIHTC metering approaches a) sub-metered approach b) master metered approach Submetering may to some extent address the so-called "split-incentive" problem in energyefficiency transactions, at least by allowing individual residents to control the use of energy. In many cases, developers are allowed to sub-meter common areas at commercial rates, which helps reduce costs of operation and maintenance but decreases the value of investment in renewable energy systems for common areas. Therefore, variations in metering infrastructure and rates and the resulting effects should be considered in the financial analysis of individual zero-energy LIHTC buildings. In practice, housing programs have started to increase incentives where developers can demonstrate that tenants can benefit from energy-efficiency or renewable energy generation systems [65].
It is recommended that state housing finance agencies evaluate new methods for metering developments. For example, in addition to a single utility provided master meter, zero-energy LIHTC developments can employ Circuit-level Owner Metering ( Figure 10) to better manage development performance. The approach provides real-time, high-resolution data on unit-level and circuit-level energy consumption, as well as solar system performance. Extant research suggests occupants are generally poor at managing systems with lags in information and delayed evaluation loops [68]. Current utility-provided metering technologies fail to provide customers with data that is usable or timely in managing system performance, improving occupant experiences, or meeting zero-energy goals. With this approach, developers install metering hardware in each apartment's circuit panel and measure energy use in 1-second intervals. The accompanying software allows easy analysis of performance at circuit, unit, and development levels and could be leveraged for predictive analytics (e.g., identifying HVAC system maintenance issues before they occur).
Circuit-level meters are commercially available starting at ~$300 per system and have been employed in pilot zero-energy developments [69]. The authors also recommend analyzing the net present values of incentivizing distributed generation using PV systems for specific stakeholders, such as private sector interests, the economy, and the government budget. Incentivizing PV systems should include an analysis of state-wide variations in solar radiation as QAP selection criteria and allocate additional scores for developments located in climatic divisions with less-than-average solar potential. Since the existing federal solar tax credit is expected to gradually decrease in the coming years, it is essential for housing agencies to identify financing mechanisms that help fill the gap to maintain the feasibility of zero-energy LIHTC developments in the 2020s.
As evidenced by past research and the energy use data analysis presented here, a significant part of variability in energy consumption is attributable to human-building interaction. Since the extra spending on utilities impacts housing affordability and economic well-being of low-income residents, empowering final users to have an active role in reducing energy consumption, particularly for heating and cooling purposes, is becoming increasingly important [70]. Such activities can be included in the state's free energy efficiency and appliance testing and repair services that help create energy savings by reducing energy bills for qualified low-income, elderly, and disabled residents in rental and owner-occupied houses [71]. To further reduce energy demand, architects and engineers can design zero-energy LIHTC projects to maximize energy-efficiency and include monitoring and control technologies that reduce user-imposed inefficiencies and contribute to cost-effective operation of new or existing buildings [72,73].

Conclusion
This study investigated the feasibility of state-wide zero-energy affordable housing by analyzing historical data on climate, energy use, and energy system costs in the Commonwealth of Virginia, US. Addressing the knowledge gap in the literature on solar panel adoption in low-income housing, the study contributes to the body of knowledge through the examination of generalizable statistical energy load prediction methods and simulation-based risk analysis, which have essential applications in planning and policy analysis but are less explored in existing feasibility studies.
The study suggests that, in most areas in Virginia, the net present life-cycle cost of the implementation of rooftop residential solar systems to achieve zero-energy LIHTC buildings is currently lower than the discounted present cost of energy of otherwise identical conventional buildings that have no renewable energy generation systems, and the value of cost savings is statistically significant. The annual electricity generated using the PV system in the 60-unit zeroenergy LIHTC study case is ~146,060 kWh/yr. The greenhouse gas emissions avoided by the PV system per year is equivalent to ~103 metric tons of carbon dioxide. This avoided carbon dioxide in 30 years amounts to nearly 348,600 gallons burnt of gasoline or 7,687,620 miles driven by an average passenger vehicle, suggesting the presence of great potentials for emission reductions, thus, contributing to economic, environmental, and equity effects in Virginia through promoting zero-energy LIHTC developments in the 2020s [74]. Based on the presented analysis, conventional LIHTC apartments built in the 2010s, which are more energy-efficient than those included in the study sample, can be converted to zero-energy with potentially higher than currently estimated investment rates of return. Since buildings represent nearly 40 percent of global energy use and 30 percent of global greenhouse gas emissions, investments in distributed renewable energy generation in buildings can significantly enhance environmental sustainability and climate change mitigation and, concurrently, address housing affordability and economic development.  Figure A1 Average retail price of electricity for residential use in Virginia