Mapping construction sector greenhouse gas emissions: a crucial step in sustainably meeting increasing housing demands

This paper examines the tension between needing to build more infrastructure and housing and simultaneously reduce greenhouse gas emissions (GHG) to avoid the most catastrophic impacts of climate change. This study uses an Environmentally Extended Input-Output approach to conduct a high-resolution top-down analysis of Canada’s national construction GHG emissions. Our findings highlight that Canada’s current construction practices cannot accommodate the construction required to restore housing affordability by 2030 without substantial environmental consequences. On a consumption life cycle basis, the construction sector was responsible for approximately 90 Mt CO2e in 2018, equivalent to over 8% of Canada’s total GHG emissions, while delivering less than a third of Canada’s annual housing needs. Residential construction was responsible for the largest share (42%) of total construction emissions. Overall, 84% of emissions are from material manufacturing and 35% of construction emissions are imported, underscoring the need for a comprehensive regulatory framework addressing both domestic and imported emissions. Under current construction practices (i.e. current material use patterns and emissions intensities), meeting Canada’s 2030 housing affordability and climate commitments requires an 83% reduction in GHG emissions per construction product (i.e. per home) compared to the 40% economy-wide reduction promised in Canada’s international reduction commitments. Mitigating the GHG gap between emission caps and housing demand calls for changes in the ratio of housing to other infrastructure (e.g. fewer roads, less fossil fuel infrastructure), new construction approaches (e.g. increasing material efficiency) and/or disproportionally allocating climate budget to construction. The implications of our study extend beyond Canada, offering valuable insights for other growing countries with climate goals. The results emphasize the urgency in considering and establishing sectoral GHG budgets for construction and for transformative changes in the construction sector to meet national GHG emission reduction commitments.


Introduction
Using Canada as a case study, this paper quantifies the construction sector's greenhouse gas (GHG) emission footprint to support a discussion on how to meet GHG reduction commitments amidst growing housing and infrastructure demands.Canada is currently experiencing its fastest population growth in nearly 70 years with commensurate pressure to build more housing and infrastructure.In 2023, Canada's population grew by 3.2% and has become one of the fastest-growing countries in the world (Marion and Ducharme 2024).As such, Canada is a microcosm of the challenges facing much of the globe; the Earth's population is expected to increase by 1.5 billion people through 2030, with most of the population growth in cities (United Nations Department of Economic and Social Affairs Population Division 2022).This growth requires increased construction of housing and associated infrastructure.At the same time, to limit the most catastrophic

Background
The construction industry is a major global consumer of materials and associated GHG emissions (Hertwich et al 2020).In 2009, construction activities alone were already estimated to make up 23% of global emissions (Huang et al 2018).Within construction GHG emissions, building material manufacturing and construction are the largest contributor, accounting for approximately 11% of global GHG emissions and rising (Global Alliance for Buildings and Construction, I.E.A. and the U.N.E.P 2019, Hertwich et al 2020).Acknowledging the escalating demand for construction driven by population growth, changes in family formation, housing, and infrastructure needs, there is widespread recognition of the need for substantial global decarbonization within the construction sector to align with net-zero emission goals (Pauliuk et al 2021, Zhong et al 2021).However, construction sector emissions (buildings and infrastructure) are not identified in reported emission statistics like the United Nations' IPCC Nationally Determined Contributions Registry (NDC) reports (Hertwich et al 2020).The NDC reports are the main way national emissions are documented and understood, meaning the dividing of construction emissions into other sectors (e.g.energy, manufacturing) obscures the collective impact of this high resource sector.Moreover, the availability of building and construction data is lacking in resolution, making it challenging to comprehend trends and demands across regions, construction sub-sectors and other industry systems (United Nations Environment Programme 2022).
GHG emissions in construction are evaluated in the literature using bottom-up, top-down or hybrid approaches (Ward et al 2018).Bottom-up lifecycle assessment approaches that measure and estimate material process emissions and operational energy at the granular level in buildings have been extensively applied to quantify the embodied emissions in construction (Greer et al 2023).However, these approaches underestimate embodied upstream emissions ('truncation error') and expansion of this approach to a national level has proven challenging due to multiple assumptions relating to boundary setting, consistency in methods used and variations in data availability (Yu et al 2017, Ward et al 2018, Nahangi et al 2021).EEIO is a macroeconomic-based, top-down approach, which has been used to calculate the main drivers for environmental impacts of products and industries since the late 1960s, notably expanding these approaches to measure GHG emissions during the 2000s (Miller and Blair 2022).Increasing availability and higher resolution of economic and environmental impact datasets at the national, regional and city levels make the EEIO approach a relevant tool to tackle lifecycle emission mapping challenges across different sectors and provide input to GHG budgeting.
EEIO models have been employed to measure GHG emissions across sectors, including household consumption (Castellani et al 2019), tourism (Sun et al 2020), the food industry (Reutter et al 2017), and high-level intersectoral distribution within economies (Xia et al 2022).Within construction, EEIO has been used to evaluate economic and energy impacts (Chan et al 2020), environmental impact of the road industry (Agez et al 2020), and increasingly applied to calculate the total construction emissions of countries mostly on a broad level analysis.Huang and Bohne (2012) calculated Norway's construction emissions over time (2003)(2004)(2005)(2006)(2007).Acquaye and Duffy (2010) analyzed Ireland's direct and indirect emissions including five construction sub-sectors.Onat and Kucukvar (2020) compared a scope-based analysis between five countries (China, the USA, India, Japan and Canada).A limitation associated with EEIO analysis is dependency on data availability, specifically on input-output table details and lack of representation of countries, sectors and sub-sectors.(Wood et al 2014, Huo et al 2022).Consequently, most top-down GHG calculation approaches provide an aggregated representation of the construction industry.Within construction emissions studies, detailed EEIO studies typically focus on buildings (Berrill et al 2022, Zu Ermgassen et al 2022, Drewniok et al 2023).Meeting sectoral emission reduction commitments requires expanding construction sector analysis to non-building elements like infrastructure and extending the product stage analysis (e.g.A1-A3) to include more aspects of the full lifecycle (e.g.maintenance) (Soonsawad et al 2022).
Despite the importance of the construction sector to achieving net-zero emissions commitments, most countries, including Canada, have yet to comprehensively measure construction industry emissions, let alone establish a construction emission carbon budget to meet its emission reduction commitments.A notable exception is the UK, which has set a construction sector GHG budget within their Sixth Carbon Budget (2033-2037) (Climate Change Committee 2022).Carbon budgets are defined as 'The net amount of GHG that could be released in the future by human activities while keeping global warming to a specific level, accounting for other non-GHG warming contributions' (Intergovernmental Panel on Climate Change (IPCC) 2023).The global carbon budget, calculated by the IPCC, is translated into national carbon budget commitments that account for production emissions in each country.Sectoral carbon budgets are the allocation of emissions targets for specific sectors within the broader economy.These targets and indicators help policymakers generate balanced and informed policies and standards toward reaching emission commitments (Vogt-Schilb and Hallegatte 2017).
Methods for allocating GHG budget within sectors vary.One prevalent approach is based on economic sector size and proportional emissions per dollar.However, some researchers argue for sectoral adjustments based on sectors ease of decarbonization (Steininger et al 2020).Examples for relatively easy-to-decarbonize sectors are light-duty transport, heating, cooling and lighting, classified as such because they can be electrified and powered by renewable energy (e.g.wind and solar).Challenging sectors include manufacturing of carbon-intensive materials like steel and cement, aviation and long-distance transport and shipping (Davis et al 2018), many of which feature in the construction sector.In the construction sector, the centrality of material decarbonization is key in decarbonization efforts as these indirect emissions are the lion's share of total emissions, ranging from 77%-95% (Hung et al 2019).However, construction materials remain hard to decarbonize due to process emissions in the manufacturing of materials and the high temperatures required in manufacturing that have been challenging to produce without fossil fuel energy sources.Emerging hydrogen and carbon capture technologies which address these challenges have been slow to deploy and are unlikely to meet reduction goals on time (Nelson and Allwood 2021).
Another less-discussed challenge for GHG reduction in the construction sector is increase in construction demands and the growth in stock (Steininger et al 2020, Berrill et al 2022, Drewniok et al 2023).Meeting these growing demands will require both increases in material efficiency and recycling along with decreased GHG intensity of material production.Global scenarios suggest, however, that decarbonization of steel and cement (the most used construction materials globally) will rely heavily on carbon capture and other strategies that implicitly require large-scale deployment of infrastructure (Watari et al 2023).This in turn, places increasing strain on the supply of construction materials in a carbon constrained world, meaning that their feasible supply within GHG emission commitments will likely fall short of demand (Watari et al 2023).Such constraints further increase the urgency to set realistic GHG budgets for the construction sector and examine potential frictions with growing infrastructure needs.

Methods
Our method was split into two phases (1) data preparation and ( 2) analysis (figure 1).In the data preparation phase, we used multi-regional EEIO tables for Canada to generate life cycle inventory data for the entire construction sector, generating datasets representing different cross sections of disaggregated GHG results (e.g. by geography, by life cycle stage,) to understand the construction sector's footprint within the economy.In the analysis phase, we (i) analyzed the emission profile of the construction sector to categorize current sources and drivers of emissions; (ii) explored per-m 2 GHG emissions for housing and the challenges of top-down versus bottom-up accounting and policy setting; and (iii) discussed the implications of construction emissions on infrastructure and housing policy going forward.

Environmentally extended input-output (EEIO) models
Input-output (IO) tables map the financial interrelationships between industry production and consumption in a specific geographic region (national state, county, or metropolitan scale).Multi-regional input-output (MRIO) builds on single region IO tables by representing additional economic linkages between several regions or countries, including the trade between regions.The environmental extension of IO connects sectoral monetary transactions to their environmental consumption or releases using pollution generation coefficients to convert dollars to emissions using aggregated average emission coefficients for each industry (Miller and Blair 2022).
IO models are constructed from industry transaction records, which are usually obtained from national statistical data agencies (de Bortoli and Agez 2023), or international organizations like the OECD (2023).These records are converted into matrices and vectors representing the flow of goods and services in industries (e.g.wood, steel, architectural and engineering services) through intermediate outputs (i.e. which goods are sold to what industry), final demands (the sales of each good to its final markets such as personal consumption or government use), and total outputs (the sum of the intermediate and final demands).In IO tables, all transactions are measured in monetary units such as euros or dollars.
IO models can be used to measure environmental impacts like GHG emissions, water, and air quality.This is done by augmenting the standard IO model with pollution generation or abatement coefficients, which are gathered through industry surveys and reporting and are compiled by environmental agencies, academic research, and government agencies (Miller andBlair 2022, de Bortoli andAgez 2023).This provides a cradle-to-gate mapping of the measured pollution impact throughout the economy for a given region.See the supplementary information for the fundamental equations underlying IO and EEIO models.
In this research, we use OpenIO-Canada, an EEIO model developed by The International Reference Center for Life Cycle Assessment and Sustainable Transition (CIRAIG) (Agez 2023) OpenIO-Canada combines data from two sources: Statistics Canada (2022), and Exiobase 3 (Stadler et al 2018).The resulting EE MRIO datasets cover 492 distinct Canadian commodities with regional-specific data for the ten provinces and three territories that make up Canada.The dataset encompasses additional 200 commodities originating from 49 international jurisdictions (44 countries and five global regions).The final EE MRIO compilation encompasses a collection of 16 196 commodities (e.g.492 commodities × 13 provinces, + 200 commodities × 49 international jurisdictions), with a specific focus on three key GHG: carbon dioxide (CO 2 ), methane (CH 4 ,), and nitrous oxide (N 2 O), Short term global warming potential (IPCC 2014, Agez 2023).
Particularly relevant for this paper, the resulting EE MRIO model includes 4175 distinct material manufacturing commodities and construction services from Canada and 8199 internationally out of the 16 192 total commodities of the model.Examples of relevant commodities and their data source are included in table S1 and table S2 in the SI.The approach to biogenic carbon emissions in OpenIO-Canadafollows the 0/0 method (Andersen et al 2021).In other words, no credit is applied for CO 2 uptake during plant growth, and no emissions are applied to the release of biogenic carbon at end of life, with the implicit assumption that the two balance out in the long run.The 0/0 approach is appropriate for this analysis as this is a cradle-to-gate evaluation, and any other approach would require more analysis of the temporal dynamics of GHG emissions at the end-of-life of natural products like wood.

Data preparation
The EE MRIO model includes final demands for all gross fixed capital formation (GFCF) construction-sub sectors in Canada, according to the NAICS Canada ( 2021) definition.This includes constructing, repairing, and renovating building and engineering work, as well as subdividing and developing land.Examples of GFCF construction sectors include transportation and warehousing, commercial buildings, oil and gas extraction, municipal government services and residential buildings (54 sectors in total).For our purposes, we classified these sub-sectors into three main types: (1) residential buildings, (2) infrastructure, (3) government services education and healthcare buildings.The full list of GFCF construction sectors (n = 54) is given in the SI (SI, table S3).
We calculate the construction industry emission impact by using the EE MRIO final demands generated by OpenIO-Canada (Agez 2023).The monetary input for the final demand was obtained from Statistic Canada's IO tables (domestic final demands) and the Exiobase 3 tables (international final demands) and resolved to the GFCF-Construction vectors in the final demand table.The calculations were executed for each province separately and for Canada's economy as a whole.
The Python code used to run the models can be found on GitHub (https://github.com/hyoffe/CA_Const_EEIO/tree/main).In our preparation for analysis, we code the matrices calculations to generate emission datasets grouped by (1) GFCF construction sector across each province and territory, (2) spatially using both consumption and production approaches, (3) on a cradle-to-gate basis as well as a point-of-emissions basis analysis, and ( 4) with a breakdown of material, transport, and onsite emissions.

Analysis
After generating the underlying emission datasets, we aggregate emitting sectors by geography, subsector or life cycle stage.For example, life cycle stages and materials are determined by matching the 16 192 subsectors to one of 18 classes (e.g.concrete, metals, transport, and direct emissions) for ease of interpretation.The full classification mapping, which includes the class definition and examples of prominent commodities in each class, is provided in the SI.(SI Sheet2, Sheet3).The mapping was developed using MasterFormat (Construction Specifications Institute and Construction Specifications Canada 2016), NAICS Classification (Statistics Canada 2021) and lifecycle stages (BSI 2011).We then create a set of visualizations to elucidate key drivers of construction sector emissions, grouped by the country/province where the GHG was emitted (e.g.location of steel manufacturing plant), construction subsector that created that demand (e.g.housing, road construction), and province in Canada where the construction took place.
Further to this analysis, we compared the residential construction differences between provinces, both emission per capita and emission per-constructed m 2 , and considered the relative impact of new construction versus renovations/refurbishment. Emissions per m 2 are an imperfect functional unit, and prior work has suggested other functional units like emissions per bedroom (Arceo et al 2023).Such data are unavailable at provincial and national scales, and so are out of scope for the present work.The data used to develop these estimates was the EE MRIO construction GHG emissions (section 3.2), 2018 demographic population data (Statistics Canada 2018) to normalize our results for the per capita estimates, 2018 new construction and renovation data (Statistics Canada 2023a)which gives information of the relative dollar amount spent on new construction/renovation, and new heated floor area built in 2018 (Natural Resources Canada 2023).The residential construction per m 2 calculations rely on estimates of Canada's National Energy Use Database (NEUD).The NEUD m 2 data includes high uncertainty for the estimated built floorspace in 2018, and is subject to change (e.g. the government sometimes updates/changes the numbers without public explanation) but is the best source for such data in Canada.
We present GHG emission per m 2 of new housing calculated using four different boundary assumptions.Due to the floor area data availability, the three Canadian Territories were combined into one category.All per-unit calculations used 2018 values.We looked at four boundaries: 1. Full consumption-based emissions kgCO 2 e m −2 (consumption).We calculated all residential construction emission operations, including repair, renovation, and land development.The emissions include both territorial and offshore emissions.This is the total residential emissions in each province, output from the EE MRIO model divided by the total completed m 2 in that province in 2018.provinces.The regional share was calculated using the EE MRIO model.This indicates the level of emissions within the control of the Provincial governments who have constitutional jurisdiction over land use and building codes.
Finally, looking forward, we use the 2018 EE MRIO results, and particularly the percent of total national emissions due to the construction sector, to estimate future allowable GHG emissions from construction assuming a linear allocation of GHG emissions from Canada's international commitments.In this calculation we focused on territorial emissions, which are lower than consumption emissions calculated in early sections, because international commitments are focused on these territorial emissions only.Canada has committed to a 40% reduction in GHG emissions below 2005 by 2030; as such, total allowable emissions in Canada in 2030 are 443 MtCO 2 e (Environment and Climate Change Canada and Public Inquiries Centre 2022).Simultaneously, to meet housing affordability needs and demand, the Canada Mortgage and Housing Corporation (2022) has determined that 5.8 million new homes must be constructed in Canada by 2030, which is approximately 725 000 homes annually in Canada (between 2022 and 2030), a 3.625 multipleincrease of housing construction in 2018.To bound the problem, we estimated the business-as-usual GHG emissions from this increase in housing by linearly multiplying sectoral emissions by the rate of housing increase.This implies a continuation of business-as-usual construction including the ratio between housing and other types of construction and no reduction in the GHG intensity of material manufacturing.This linear assumption has several limitations.For example, it assumes the ratio of territorial and offshore emissions stay the same and it does not capture economic changes resulting from a tripling of the construction sector.It also assumes that the construction portfolio will repeat the same mix of residential, commercial, and infrastructure projects observed in 2018 and that new infrastructure construction would follow new housing construction proportionally.While inexact, this provides a useful framing for the scale of emissions likely in the construction sector should housing construction be rapidly scaled up without large changes in the planning and construction regimes in Canada.In this first scaled scenario, we make no adjustments to the GHG intensity of construction activities.Accordingly, this extrapolation should be taken as an upper bound for framing.
The 2022-2030 budget below also assumes that economic activities in the construction industry in 2022 are similar to 2018 (738 MtCO 2 e) given the COVID-19 economic slowdown and related GHG emissions drop and recovery between 2020 and 2021 (652-670 MtCO 2 e) (Climate Action Tracker 2022, Environment and Climate Change Canada 2023).
Working forward from the base case scenario we consider the likely impact of improvements in material manufacturing and associated GHG by 2030.For example, by 2030, 40%-45% reductions in GHG emissions are promised for recycled steel in Ontario (ARC Energy Research Institute 2022) and for a 25% reduction in of cement (Cement Association of Canada 2021, Innovation Science and Economic Development Canada 2022).Considering Canada's action plan for decarbonization of material construction (Environment and Climate Change Canada 2022b), we optimistically assumed a general GHG emission reduction for construction material by 25% for this calculation.In discussing the changes needed in the construction industry, we used this 25% to represent the potential impact of decarbonization in material manufacturing through 2030.
In summation, we project three scenarios for 2030: 1. GHG emission from construction assuming business as usual 2. GHG emission from construction assuming a rapid increase in housing construction to meet Canada's goal of building 5.8 million new homes by 2030 3. Scenario 2 (5.8 million new homes), now including 25% savings in material emissions from changes in material manufacturing This provides a framework for considering the range of trajectories facing Canada's housing and infrastructure system over the coming decade as efforts are simultaneously made to build more while emitting less.We highlight the degree of reduction per-housing unit that will be needed to both reach GHG reduction and increased housing construction targets.

GHG emission from construction in Canada in 2018
The EEIO model for Canada's construction sector GHG emissions showed that the construction sector was responsible for 90 MtCO 2 e of emissions using a consumption based accounting approach.This calculation, including all GFCF construction sub-sectors, showed that residential buildings is the highest emitting construction sub-sector (37.7 MtCO 2 e), followed by 'Other municipal government services' (7.2 MtCO 2 e), 'Transportation and warehousing' (7.2 MtCO 2 e), and 'Oil and gas extraction' (7.1 MtCO 2 e) (figure 2).The combined distribution across development types shows Residential (42%) and Infrastructure (39%) developments each are roughly double that of the service sector (19%) (figure 2).
Of these 90 MtCO 2 e, 35%, are imported as illustrated in figure 3. The top countries of origin include China (9.78 MtCO 2 e), the United States of America (USA) (8.61 MtCO 2 e), India (1.02 MtCO 2 e) and Mexico (0.9 MtCO 2 e).If material sourcing norms continue, Canada's construction sector emissions reductions will be partially dependent on climate commitments and follow-through in other countries, particularly China and the United States.Individual international sectors with the highest direct emissions in the Canadian construction sector supply chain include Electricity by coal (2.9 MtCO 2 e) used for off-shore manufacturing, Iron and Steel (1.6 MtCO 2 e, process emissions), and Cement, lime and plaster (1 MtCO 2 e, process emissions) from China; as well as Crude oil extraction processes (1.5 MtCO 2 e), Electricity by coal (1.3 MtCO 2 e) and Gas (0.6 MtCO 2 e) from the USA (figure 3).
At a provincial level, residential construction dominates construction GHG emissions in most provinces and is particularly prominent in Prince Edward Island, Nova Scotia, Quebec, Ontario and British Columbia, where it exceeded 50% of the total (table 1).In Newfoundland and Labrador, Saskatchewan, Alberta, and the three territories, oil and gas infrastructure is the largest sectoral contributor to construction GHG emissions.These findings are in line with population growth trends in Canada and the regional investment in oil and gas infrastructure.For example, Alberta-Canada's largest oil and natural gas producer-had construction emissions for oil and gas in 2018 equal to 83% of British Columbia's residential construction and 93% of Quebec's.In New Brunswick, infrastructure/buildings for government services are notably large contributors, though as the data provides a one-year snapshot and it is not clear if New Brunswick always builds a lot of government buildings/services or if a large project in 2018 influenced these results (figure 4).

Emission sources by category and lifecycle stage
While figure 2 focused on sectors responsible for direct emissions (point of emission basis), table 2 presents results aggregated across the supply chain for each major input to the final construction process (i.e. on a cradle-to-gate basis).In this cradle-to-gate emission analysis of Canada's construction sectors (consumption) for the product and construction stages, we found that 84% of emissions are from the product stage, 5% for transportation to site, and 11% are direct emissions of the construction stage: 10% onsite operation, 1% design services and margins.This reinforced the imperative for efforts to reduce construction sector emissions to focus on reducing material use and the GHG intensity of materials.Nationwide, the most dominant construction materials are Concrete (14%), metals (12%) and wood (9%) (table 2).For the full definitions of emission sources including each class and examples of prominent commodities see the SI (SI Sheet 2, Sheet 3).A further breakdown of these same life cycle stages and emission types is available in SI table S4 for each construction sub-sector (residential, mining, etc), and in table S5 for each province/territory.In residential construction, the top three emitting materials are wood (19%), concrete (12%), metals (9%).The inter-province analysis shows that the more urbanized provinces (Ontario, Quebec, British Columbia and Alberta) see a larger share of emissions from concrete (12%-18%) than more rural provinces (Manitoba, Saskatchewan, New Brunswick and the Territories) (1%-8%).Normalizing per capita identifies big ranges in      3). Figure 5 summarises this section by illustrating the emission flow through Canada's construction sector to emission source, construction subsector and consuming province.

Residential construction per-m 2 GHG emission
In 2018, Canada completed 200 262 new units of housing (Statistics Canada 2023b).Overall, that equates to 188 369 KgCO2 e per unit completed nationally on a consumption basis, including the impact of renovations on total housing emissions.To compare emissions intensity of residential construction across the country, we normalize our province and territory residential emission results (table S5) by built floor area in 2018.Results are adjusted to account for different emission boundaries for residential buildings, including consumption and territorial emissions, total, new construction and renovations.The emissions associated with new residential floor-area varies between provinces.Prince Eduard Island reported the lowest GHG intensity (391 KgCO 2 e m −2 ) and Alberta the highest (1251 KgCO 2 e m −2 ).The monetary share of new construction (construction lifecycle phases A1-A5) compared to renovation or refurbishment (phases B2-B5) ranges between 64% new construction in New Brunswick to 85% new construction in British Columbia and Prince Edward Island.The largest amount of new construction was reported in Ontario, representing ∼43% of all new floor area in Canada in 2018.We apply the monetary relationship to emissions 1-1 in the absence of better data on the different GHG intensity of each dollar of renovation or new construction.The share of emissions that were emitted locally (e.g. in the same province as construction) ranges from 14% in the territories to 50% in Quebec and 61% in Alberta (figure 6)-raising important questions about how national GHG budgets can/should be allocated regionally and again highlighting the difference in production versus consumption-based accounting.
Published literature on bottom-up GHG accounting finds that cradle-to-grave embodied emissions in residential buildings vary greatly between 179 and 1050 kgCO 2 e m −2 (Chastas et al 2018).Regulations in Canada and around the world are increasingly based on per-m 2 benchmarks.For example, Vancouver, Canada now sets a limit of 400 kgCO 2 e m −2 (City of Vancouver 2023); Toronto, Canada's Green Standard sets a limit of 350 kgCO 2 e m −2 for new residential buildings (A1-A5) (City of Toronto 2022).Though broadly aligned with past results, our top-down accounting skews toward the upper end of previously reported bottom-up ranges, especially when considering the full consumption-based accounting.Although these results are highly aggregated and cannot easily be adapted to evaluate any single building project, they suggest meaningful truncation errors that could undermine the success of bottom-up emission limits if those  limits fail to capture all activities and emission sources within the sector (e.g.renovation).However, using this top-down number to set regulation would be much too broad if bottom-up calculation including truncation were used to meet them.The gap between bottom-up regulation and top-down budgeting is an area that needs further research, thought and policy.

Future construction emission estimation
Finally, we compared Canada's territorial 2018 construction emissions (58 MtCO 2 e) to three scenarios which contrast allowable annual emissions with prospective future scenarios for the construction sector.Based on proportionally allocating emissions, in 2030 Canada's construction sector can emit 34 MtCO 2 e.We compare this to an upper bound 'meeting housing demand' emissions scenario if business as usual is scaled up (3.625) and ratios between housing and infrastructure construction hold with no changes in material use or manufacturing.We add a 25% material emission decarbonization while meeting 2030 housing demand to consider the likely reductions in material GHG intensity by 2030.The results (figure 7) highlight that in a business-as-usual scenario, to meet 2030 commitments, Canada would need to reduce its territorial emissions by 83% for every existing unit of construction.Even after factoring in 25% material decarbonization, a 78% reduction in GHG emissions would still be needed-requiring substantial new efficiencies to be found.These reductions concern only the territorial emissions under Canada's international reduction commitment, the 35% of emissions imported are excluded (and shown in hashed grey).

Discussion
This study calculated national construction GHG emissions in high resolution using a top-down EEIO approach to inform the discussion of the tensions between meeting future construction needs, in light of national GHG reduction commitments, using Canada as a case study.Our findings show that the status quo of how Canada is currently building cannot be scaled up to 5.8 million homes by 2030 without disproportionate impacts on the environment, and that big changes in how Canada builds will be needed to simultaneously meet housing and climate commitment.To meet the 2030 climate and housing commitments in parallel, an 83% GHG reduction per product (e.g. per home) is needed to achieve the 40% sector-wide emissions implied by Canada's international climate commitments.Our study showed that in Canada, material production is responsible for the majority share (84%) of the construction sector's consumption-based GHG emissions, which aligns with the international 77%-95% range (Hung et al 2019).The main challenges associated with these emission reductions include slow decarbonization of material intensity, which are expected to only rapidly reduce after carbon capture and storage technology is broadly employed.However, the delivery pace of such technology has been much slower than expected (Watari et al 2022, Climate Change Committee 2023), suggesting an urgent need to rely more on other approaches like material efficiency.Another challenge will be negotiating offshore and territorial emissions.We found that 35% of construction emissions are imported; a regulatory framework to tackle the construction sector GHG will need to address both domestic production and imports.On a transnational level, this highlights the importance of fostering global collaborations for low GHG material manufacturing, resource efficiency and transborder monitoring of material flows and environmental impacts (International Resource Panel 2022).On the national level, care needs to be taken to deter offshoring of emissions to create the appearance of territorial GHG reductions.
The GHG mitigation gap between building more and a 40% sectoral GHG reduction for construction can be tackled in a few ways: 1) Changing the ratio between housing and other infrastructure.For example, building more residential buildings while minimizing new roads, water or extraction infrastructure.Reducing the material investment in future infrastructure will allow for a higher proportion of construction GHGs to be directed towards housing.This can be accomplished, for example, by increasing the percentage of new housing built within the existing urban boundary, reducing the need for new transport and water infrastructure.New neighbourhoods could also be built with less infrastructure, for example, narrower and straighter roads.At the national scale choices can be made between infrastructure types.Within Canada, 13% of construction emissions are related to the extraction industry, particularly oil and gas.This means Canada may face a choice of whether to 'spend' future GHG emissions budgets on resource extraction or on housing-or else search for other ways to offset these emissions.).These approaches save on structural-related and embodied emissions like cement, and steel in new construction by reusing elements in situ.Though it is important to note this is not a panacea as Canada, like many fast-growing countries, currently builds much more new buildings/infrastructure than there are end-of-life resources available for repurposing.4) Allocating more GHG budget to construction from other sectors (e.g.require faster mitigation in the transportation or energy sectors).This paper has worked with the assumption that the construction sector must reduce its GHG emissions by 40%, in line with Canada's economy-wide commitments.
However, Canada could choose to allocate more GHG budget to construction in line with its economic importance and difficulty in decarbonizing.However, allowing slower reductions in the construction sector would demand faster reductions elsewhere.Further, construction sector emissions are poised to increase rapidly, so even with the reallocation of emissions from other sectors, large changes will be needed in how and what gets built.
While this paper focused on Canada, the findings here are informative for other countries which need to build more while polluting less.Construction will need much faster per-product GHG reductions than the overall mitigation goals in countries experiencing construction growth (nearly all of the world), which will need to make key decisions and more consideration on what gets built (e.g.housing versus vs. transport vs. extraction infrastructure).
The GHG associated with each m 2 of new construction is much higher when taking a top-down approach compared to the bottom-up approach, where the narrower boundaries exclude upstream impacts like services and indirect energy.Existing embodied GHG regulations generally apply only to new buildings, narrowly focusing on per m 2 metrics, which do not consider the total amount being constructed.For example, they overlook up to one-third of the emissions coming from renovations in Canadian provinces required supporting infrastructure.A top-down approach could set a total budget and then divide by the total expected amount of construction to set an annual GHG limit per m 2 .However, starting with the wider boundaries of a top-down approach like EEIO would mean results which are skewed high.The truncation error of top-down EEIO to bottom-up LCA and hybrid approaches is between 30% and 80% (Ward et al 2018).Designers and builders submit documentation to meet regulations based on bottom-up accounting, so a top-down to bottom-up adjustment would be needed.

Conclusion
In this study, we use an EEIO approach to provide the first comprehensive mapping of Canada's construction sector emissions, identifying a total of 90 MtCO 2 e in 2018.Residential construction emerged as the highest emitting sub-sector, accounting for 42% of these emissions.This sets a baseline for understanding Canada's potential future emissions as it aims to rapidly scale up construction, particularly housing construction.Meeting both housing demand and rapid decarbonization goals will require changing Canada's construction portfolio (e.g. more housing, less oil and gas, more housing, fewer roads), being more efficient and/or transferring GHG budget from other sectors to construction.As a third of Canada's construction GHG emissions are imported, this is not a problem Canada will be able to solve alone without large scale onshoring.In the next decade, Canada's construction sector emissions are likely to continue to depend in part on manufacturing choices in other countries, particularly the United States and China.While focusing on Canada, this study also provides insights relevant to all other countries struggling with increasing construction demands.It highlights the urgency for transformative changes in the construction sector to align with emission reduction goals and the tension between choices on what to build (e.g.housing or other infrastructure) within a limited emissions budgets.

Figure 1 .
Figure 1.Overview of methods and workflow.The CIRAIG model generates the underlying EE MRIO model and associated matrices (step 1), which we use in conjunction with final demand for gross fixed capital formation from construction (GFCFC) in each province to generate datasets providing GHG results with different dimensions of disaggregation and classifications (step 2), and subsequently apply to analyze construction sector emissions and tensions between emission targets and increasing housing needs (step 3).

Figure 3 .
Figure 3. GHG emissions in the Canadian construction sector 2018 supply chain by country of origin (values in MtCO2e) Top 6 countries where direct emissions occur (top).Top 25 sectors (by direct GHG emissions) in the 2018 Canadian construction sector supply chain, by source type and country of origin (values in MtCO2e) (bottom).For reference, total consumption-based emissions for the Canadian construction sector were 90 Mt CO2e.For the Canadian province and territory abbreviations see SI Sheet 1.

Figure 5 .
Figure 5.GHG emission flow through Canada's construction sectors(MTCO2e, 2018, consumption).Commodity type for all construction activities, including new construction and renovations (manufacturing process, transportation to site and direct emissions).Construction sector (by GFCF construction) and consuming Canadian province or territory.

New developments residential structures (consumption).
2. The residential sector data in Canada's IO tables does not differentiate between new construction and refurbishment.Using building permit data from Statistics Canada, we assessed the proportion of GHG emissions associated with new construction versus renovation/refurbishment.We excluded the renovation emissions on a proportional basis based on monetary expenditure of all permits issued in 2018 (SI, tableS7).Renovations' definition includes alterations and improvements-any construction work for the purpose of improving or modifying existing structures; Conversions-modification of existing buildings involving the gain of dwelling units; and Garage and Carports permits (Statistics Canada 2023a).3. Domestic emissions (Canada, territorial).Calculated similarly to 2 (New developments residential structures) but excludes emissions outside of national boundaries and, therefore, not subject to national production-based GHG reduction commitments.4.

Local emissions (from province, territorial).
Excluding emissions imported from other Canadian

Table 1 .
Residential construction emissions by province (roughly arranged from East to West) and territories in comparison to total GHG emissions in the province or territory (Consumption.MtCO2e, 2018).
residential emissions.The lowest emission per capita are in Nunavut (249 kgCO 2 e/capita) and Newfoundland and Labrador 540 (kgCO 2 e/capita) and the highest were British Columbia (1525 kgCO 2 e/ capita) and Alberta (1336 kgCO 2 e/capita).For context, the fastest-growing provinces/territories (on a
percent basis) in 2018 (Statistics Canada 2018) were (in order) PEI (1.8%), Ontario (1.8%), Alberta (1.5%),British Columbia (1.4%), Nunavut (2.2%) and The Yukon (2.1%), which aligns but only loosely with the top residential construction emissions per capita (table 2) Changing the amount of new material used to build and or choosing lower GHG materials.e.g.more efficiently designed structures, building smaller, more efficient reuse of existing structures and materials, selecting low GHG materials.Tackling material efficiency focuses on the larger share of materials and is immediately deployable without waiting for improvements in material manufacturing and/or carbon capture.Material efficiency and material choice options are discussed in more length in other work(Hertwich et al 2019, Berrill et al 2022, Arceo et al 2023).