Pollution Accounting for Corporate Actions: Quantifying the Air Emissions and Impacts of Transportation System Choices Case Study: Food Freight and the Grocery Industry in Los Angeles

Abstract

Credible corporate commitments to environmental and sustainability outcomes build upon reasonable estimates of corporate impacts and realistic plans to ameliorate those impacts. Although many companies have already begun to account for their goods movement emissions, the vast majority of environmental, social, and governance (ESG) disclosures do not. This report creates and critically evaluates two complementary accounting mechanisms for air pollution emissions resulting from local transportation systems—for use in ESG disclosure and impact mitigation planning. These mechanisms are applied to a case study of businesses involved in food freight in Los Angeles: demonstrating the scope of local goods movement impacts on air quality and climate, and paving a path for additional analyses to follow. By quantifying the scope of impact from certain business and supply chain operations, this analysis makes the case for enhanced corporate responsibility by documenting and then reducing transportation system emissions from supply chain and logistics systems.

1. Executive Summary

While the air quality impacts of goods movement operations have generally been well studied, until now there has been no accepted method to affiliate air pollution to individual companies that rely on and pay for goods movement services, especially in a defined geography. The analysis presented herein aims to begin to fill that gap.

In 2019 and 2020, EDF collected and collated disparate sets of emissions and goods movement data to develop two methods to attribute regional supply chain air pollution impacts to major corporations involved in the food industry in the Los Angeles area. In doing so, the foundation of a Corporate Pollution Accounting (CPA) methodology was developed to advance several goals.

First, the CPA methodology enables more specific awareness of the importance of truck fleet electrification on meeting corporate sustainability goals. While many companies do not own or operate the vehicles upon which they rely (most major brands depend on third parties to ship and deliver their products), those that are concerned about ESG (environmental, social, and corporate governance) have begun to quantify emissions from these types of transportation operations in addition to direct emissions. A limited number of companies—including Nestlé, Walmart, UPS, Unilever, FedEx, Amazon, and Pepsi—have already begun to independently take responsibility for their goods movement emissions and pilot freight electrification programs [1].

Combined with other data, the CPA method may enable companies, policy-makers, investors, and advocates to define a company’s ESG risk from trucks in a more accurate and transparent way, and support companies to prioritize truck electrification for health and equity impact.

Second, the CPA mechanism can also fill a gap that presently exists in corporate reporting and mitigation planning. Current corporate sustainability reporting generally avoids delineation of contributions to local air pollution entirely and very rarely includes specific information on traditional sources of air pollution that contribute to public health impacts. As a result, most ESG commitments focus on greenhouse gas (GHG) emissions and omit local pollutants such as ozone-precursors, particular matter, and carcinogens that degrade local and regional air quality and cause human mortality and morbidity. The analysis presented here develops methods to quantify GHG and local air pollutants.

For climate emissions, corporations tend to report across entire operations and supply chains, making it difficult to isolate emissions in a specific region based on company-supplied data and disaggregated distribution services provided by third-party shippers. Regional air emissions inventories, on the other hand, provide detailed measurements of local air pollutants but do not categorize them by company or industry. The CPA method provides a way to link emissions owned or controlled by corporations in specific air sheds.

As developed, the CPA methodology involves two potential analytical approaches. The first approach utilizes is a top-down method that combines high-level commodity flows with market data to attribute freight-related emissions first to specific industries and then to companies within those industries. The second approach is a bottom-up method that would model the emissions of a single distribution network based on company-specific assumptions. The top-down approach is identified as having greater utility and overall accuracy when only public data are available.

The CPA methods presented have a key limitation in so far as they assess air pollution impacts across a region, but goods movement emissions are not spread equally throughout major metropolitan areas. Rather, they are concentrated in high-traffic areas that disproportionately harm communities of color and socioeconomically disadvantaged communities. Accordingly, this methodology provides a jumping-off point for more refined reporting and quantification of locational impacts, and further development for other sectors and geographic regions.

Notwithstanding its limitations, the analysis contained herein provides a guiding framework for companies, investors, regulators, and researchers to improve public accounting mechanisms for major economic sectors and businesses. Additional data sets, research, and analysis would likely improve the overall precision and accuracy of the case study developed and would be important for the expansion of the CPA mechanism into other geographies and industries. In taking these additional steps, goods movement emissions will be both documented and then seen as ripe for elimination as critical actions in support of ESG goals.

2. Background

While the CPA methodology described in this paper could be applied to any sector in any geographic region, this study focuses on the food sector in Los Angeles. Los Angeles is home to a high concentration of truck-based goods movement operations, the largest shipping hub in the U.S. at the Port of Los Angeles and Long Beach, and one of the largest consumer markets in the world. Serving this market requires an extensive goods movement network that receives freight from domestic and international markets and distributes it across the region. Collectively, a total volume of approximately 600 million tons of freight valued at USD 1.7 trillion moved through Southern California in 2015. Diesel trucks are the dominant mode of transportation, moving over 500 million tons of that freight [2]. These numbers are rising annually.

In addition to supporting massive economic activity, goods movement in Southern California contributes to severe air quality impacts particularly in the South Coast Air Quality Monitoring District (SCAQMD), as shown in Figure 1 and

Table 1. VOC and NOx emissions from on-road mobile sources in the goods movement sector in the South Coast Air Quality Management District for calendar year 2014 [7].

Source Category	Population	VMT (Miles/Day)	VOC (Tons/Day)	VOC (% of Total)	NOx (Tons/Day)	NOx (% of Total)
Light HD Gas Trucks-1 (8501–10,000 lb.)	274,553	11,988,596	6.58	48%	15.01	11%
Light HD Gas Trucks-2 (10,001–14,000 lb.)	29,078	1,261,404	0.66	5%	1.49	1%
Medium HD Gas Trucks (14,001–33,000 lb.)	23,181	960,000	1.18	9%	2.43	2%
Heavy HD Gas Trucks (>33,000 lb.)	1585	186,000	0.19	1%	1.02	<1%
Light HD Diesel Trucks-1 (8501–10,000 lb.)	86,598	3,679,455	0.56	4%	17.48	12%
Light HD Diesel Trucks-2 (10,001–14,000 lb.)	29,299	1,231,545	0.19	1%	5.69	4%
Medium HD Diesel Trucks (14,001–33,000 lb.)	80,061	4,101,000	0.94	7%	23.30	16%
Heavy HD Diesel Trucks (>33,001 lb.)	72,411	8,216,000	3.29	24%	76.43	54%
Total	596,766	31,624,000	13.59		142.85

. The SCAQMD includes all of Orange County and the non-desert portions of Los Angeles, Riverside, and San Bernardino Counties. It is home to over 16 million people, representing nearly half of California’s population. Despite significant progress, the SCAQMD contains multiple extreme nonattainment zones for both ground-level ozone and particulate matter under the National Ambient Air Quality Standards (NAAQS) [3,4], making Los Angeles air quality some of the worst in the nation. Furthermore, air quality is disproportionately worse in disadvantaged communities [4,5,6]. Additional information about SCAQMD emissions inventories and air quality studies is provided in Appendix A.

Goods movement contributes significantly to SCAQMD’s nonattainment status. A 2016 SCAQMD goods movement whitepaper found that on-road goods movement emitted over 142 tons of NO_x per day, accounting for 30% of the region’s emissions [7]. Specifically, Class 8 Diesel trucks contributed over half of the on-road goods movement NO_x emissions despite making up only 12% of the truck population (Table 1). Heavy-duty trucks also contribute to cancer-causing particulate matter (PM) air pollution. According to the 2015 Multiple Air Toxics Exposure Study (MATES), diesel particulate matter is a key driver of air toxics risk, accounting for 68% of the total estimated air toxics risk in the SCAQMD region [8]. The health burden is not equal. A recent study from EDF and partners found that in West and Downtown Oakland, where more than 70% of the population is people of color, up to 1 in 2 new childhood asthma cases were due to traffic-related air pollution [9].

The SCAQMD and MATES studies are still relevant today. Goods movement from heavy-duty trucks remains a significant source of smog and particulate pollution—meaning goods movement emissions reductions strategies should be prioritized to improve health outcomes [10].

3. Purpose: Accounting for Air Pollution by Sector and Company

This paper identifies two potential analytical approaches for quantifying the air pollution associated with the transportation choices of companies in the food freight in Los Angeles, with a focus on grocery. The approaches developed can serve as models for the development of similar corporate pollution accounting methods in other geographies and sectors.

The two mechanisms used in this analysis include a top-down method and a bottom-up method. Top-down refers to a method using a combination of high-level commodity-flow, geographic-distribution, market-share, and vehicle-emission-factor data to attribute freight-related emissions first to specific industries and then to companies within those industries. Bottom-up refers to a method that models the emissions of a single distribution network based on company-specific data.

Companies involved in food freight (e.g., grocers and food services) have numerous paths to reduce emissions from the transportation systems upon which they rely [11]. The use of zero-emissions vehicles has recently emerged as a viable option. As zero-emission models of medium- and heavy-duty vehicles further penetrate the market, the companies using these vehicles will decrease their supply-chain emissions and, consequently, attenuate their air quality impacts and public health ramifications.

4. Top-Down Approach

4.1. Purpose and Intent

The top-down approach to pollution accounting aims to quantify the goods movement emissions of individual companies by first determining the total regional emissions associated with goods movement by the entire food sector and then attributing those emissions to individual companies proportionally based on their share of the region’s market. By painting a comprehensive picture of the sector-wide air quality impacts of goods movement and defining each company’s emissions contributions, this approach intends to compel companies to reduce the emissions created by their distribution networks and invest in zero-emissions transportation infrastructure.

4.2. Method for Quantifying Total Emissions

1.

Data Sources

The primary source of data used in this analysis is the Freight Analysis Framework (FAF), a partnership between the Bureau of Transportation Statistics (BTS) and the Federal Highway Administration (FHWA). FAF integrates data from a variety of sources to create a comprehensive picture of freight movement among states and major metropolitan areas by all modes of transportation. The US Census’s Commodity Flow Survey (CFS) serves as the primary basis for the FAF, accounting for 70% of estimated freight by value. The remaining 30% integrates other data sources to incorporate industries out of the scope of the CFS such as farm-based agriculture, fishing, and logging. The FAF captures freight flows within and between 132 domestic regions made up of metropolitan areas and “rest of state” areas. The FAF’s “truck” mode of transportation covers class 5–8 trucks. The FAF classifies freight under 43 distinct commodities following the Standard Classification of Transported Goods (SCTG).

2.

Geographic Boundary Setting

Following the FAF, this study uses the Los Angeles-Long Beach Combined Statistical Area, henceforth referred to as the Los Angeles area, as its region of analysis. The Los Angeles-Long Beach Combined Statistical Area consists of the counties of Los Angeles, Ventura, Orange, Riverside, and San Bernardino. This area is significantly larger than the SCAQMD due to the addition of Ventura County and the desert portions of San Bernardino County and Riverside County.

3.

Commodity Freight Flow Data

The FAF is used to obtain total annual freight flows destined for the Los Angeles area by each region of origin and each commodity, and the analysis herein uses the Freight Analysis Framework Data Tabulation Tool with Year set to 2018, Destination to Los Angeles, and Domestic Mode to Truck. For Origin, Commodity, and Measure, Select All was chosen. The FAF returns the total tons, ton-miles, and monetary value of goods shipped by truck for each region of origin and commodity. For example, FAF estimates that 408,000 tons of “Other Foodstuffs” were shipped by truck from Phoenix, AZ, to the Los Angeles area in 2018.

For this analysis, the study identifies and utilizes commodity flow data within the FAF related to food freight in order to build an emissions impact estimate for the food industry. The FAF defines food freight as the following commodity codes: “01—Live Animals and Fish,” “02—Cereal Grains,” “03—Agricultural Products,” “05—Meat/seafood,” “06—Milled grain products”, and “07—Other foodstuffs.” This selection of commodity codes is intended to capture any food entering the region that will eventually be sold in grocery stores.

Previous studies of food freight have also included “04—Animal Feed, Eggs, Honey, and Other Products of Animal Origin.” Because this “04—” commodity code includes several non-food animal products, this study chooses to exclude it for the purposes of developing conservative estimates. An even more conservative option would be to avoid codes “01—” through “04—” and only include processed and ready-to-eat categories (“05—Meat/seafood,” “06- Milled grain products”, and “07—Other foodstuffs.”). This study however includes unprocessed categories (“01—Live animals and fish,” “02—Cereal grains,” “03—Agricultural products”) in order to capture food freight emissions across the local supply chain.

The “43—Mixed Freight” commodity code presents an additional analytical challenge because it includes relevant categories such as “items (including food) for grocery and convenience stores” as well as irrelevant categories such as “office supplies” and “miscellaneous.” This paper apportions a percentage of Mixed Freight as Food Freight based on data from the 2012 CFS that ties NAICS industries to SCTG commodity codes. In the 2012 CFS, 35.6% of Mixed Freight destined for the Los Angeles area was associated with “grocery and related product merchant wholesalers” and “food manufacturing.” These are the only industry classifications directly related to the food sector, so this study adopts 35.6% as a minimum estimate of the portion of Mixed Freight that is food freight.

Figure 2 shows the relative share of each commodity code in food freight destined for the Los Angeles area. “07—Other foodstuffs” is the largest category, accounting for almost half of food freight by weight. “03—Other agricultural products” is the second-largest category at just under 25% of food freight, followed by “43—Mixed freight,” “06—Milled grain products” and “05—Meat/s/eafood” at 10% each.

4.

Distance Estimation

To calculate emissions based on FAF freight flows, an estimate of the distance traveled within the Los Angeles area for each region of origin was performed. Distance estimates enable the conversion from tons to ton-miles—affording the application of emissions factors to generate emissions estimates.

For freight shipments originating within the Los Angeles area, this study assumes that the entire trip occurs within the region. For these intraregional trips, the study uses the ton-mile estimate directly from the FAF.

For shipments from the remaining 231 regions, the given ton-miles metric is not useful because it is unclear how many of those miles occurred within the Los Angeles area. Instead, this study estimates the average miles traveled within the Los Angeles area for each region of origin using Google Maps. It identifies five points of entry to the Los Angeles area and assumes that all shipments from a given region enter through the point of entry chosen by Google Maps routing (Figure 3). It sets the endpoint as “Central Los Angeles” as designated by Google Maps. The choice of Central Los Angeles is intended to approximate the average end point of inbound trucks. The average within-region miles for each region of origin is calculated as the distance between its Google Maps suggested point of entry into the Los Angeles area and “Central Los Angeles,” rounded to the nearest five miles.

For intraregional freight, trips averaged 46 miles. This study assumes that freight from Northern California, Washington, and Oregon enters on I5 from the North and travels an average of 75 miles. It assumes that freight from Southern California enters on I5 from the South and travels an average of 65 miles. Finally, it assumes that freight from the rest of the US enters from the East on I10, I40, or I15 and travels an average of 225 miles within the airshed.

The FAF classifies imports based on their point of entry. For example, imports arriving in the Port of Houston eventually destined for the Los Angeles area by truck are recorded in the FAF as a truck shipment originating in Houston. Imports that arrive at the Port of Los Angeles destined for the Los Angeles area count as intraregional freight. Imports that arrive at the Port of Los Angeles with destinations outside of the Los Angeles area would not be considered in this analysis, making its emissions estimates even more conservative.

In Figure 3 above, the green arrow represents international shipments entering to the Ports of Los Angeles and Long Beach and the Los Angeles International Airport. The blue arrows represent freight coming from the north or south, from the rest of California, Oregon, and Washington. The orange arrows represent freight from elsewhere in the US.

5.

Calculating Emissions

To calculate overall emissions, the total tonnage for each region of origin is multiplied by the average miles traveled within the Los Angeles area and then summed across all regions of origin to yield the total annual within-region ton-miles. Next, the analysis applies ton-mile emissions factors from EPA SmartWay to convert freight flows to air emissions.

EPA’s Smartway Program is a voluntary system for freight carriers to document and disclose fuel use and freight emissions across supply chains. Carriers input detailed fleet characteristics, operations data, and fuel use and EPA uses the MOVES2014 model to calculate emissions factors for CO₂, NO_x, and PM_2.5. Emissions factors include emissions from deadhead miles, idling (including hoteling), and transport refrigeration units (TRU) [16]. EPA releases carrier performance data each year for every carrier participating in the program.

For the food industry, this paper uses the average emissions factors across the proprietary truck fleets of a major LA-area grocery company as published by EPA SmartWay’s 2019 Carrier Performance Rankings [17]. Because this study aims to provide methodologies for corporate pollution accounting, as opposed to focusing on the emissions of any one specific company, the entity chosen will be hereby referred to as Company A. Although Company A’s truck data may differ from those of other fleets, it is one of the market leaders in the Southern California grocery industry. The use of its data is reasonably expected to approximate the industry average for the region. Company A’s average emissions factors are 87.8 g/ton-mile CO₂, 0.22 g/ton-mile NO_x, and 0.0035 g/ton-mile PM_2.5.

6.

Translating to monetary and health impacts

This paper uses the Environmental Protection Agency’s 2018 Value of Emission Impacts Study to assess the health impacts and monetary costs of a given level of NO_x and PM emissions [18]. EPA estimates societal impacts of USD 950,000 per ton of PM emitted and USD 20,000 per ton of NO_x emitted by the on-road mobile sources sector. Those costs are calculated based on valuations of reductions in premature mortality risk and reductions in the relative risk of certain health effects. One limitation of these estimates that should be noted is their low resolution: they represent an average cost per ton of emissions across the entire United States [19]. Consequently, they do not reflect the large variability in impact per ton of emissions across different locations. One would expect regions of high population density and underlying health vulnerability to incur significantly higher costs per ton of emissions than low-density regions with greater health security. Considering population density and health vulnerability often correlate to socioeconomic status, variability in cost per ton is a question that deserves further study.

4.3. Method for Apportioning Emissions to Companies

Up to this point, the methods discussed have been used to develop an assessment of emissions for all food freight entering the Los Angeles area. However, the goal of this case study is to isolate from region-wide emissions those emissions originating from the grocery industry and from individual corporations within that industry.

1.

Grocery industry and store share of food consumption

The first step to apportion emissions originating from the grocery industry and from individual corporations within that industry is to determine what percentage of food freight serves the grocery industry as opposed to the food services industry (restaurants, cafeterias, etc.) [20]. Saksena et al. (2018) found that on average Americans consume roughly 2/3rds of their calories at home, though other studies have found closer to a 50/50 split [21]. Assuming that food eaten at home represents grocery store purchases and that calories are roughly proportional to weight, this study attributes 2/3rds of food freight emissions to the grocery industry and 1/3rd to the food services industry. Of course, since the onset of the COVID-19 crisis and widespread concern and closure of restaurants for in-person dining, the food purchasing, and consumption trends of the general population have changed—though those changes have been found to have moved towards an increasing amount of food purchased in grocery stores for at-home consumption [22].

2.

Corporate attribution

For the grocery industry, responsibility for food transport emissions is allocated to retailers proportionally to their market share. Although there are differences amongst grocers’ fleets and vehicles within the expanded transportation systems upon which they rely, public information on corporate fleet efficiency and vehicle mix is generally unavailable and thus requires the use of broadly accepted emission factors for corporate attribution. For example, a grocer would typically rely on several tier-one suppliers that deliver to either their retail facility or warehouse, and these fleets may have different emission factors depending on the type and age of the vehicles used. For this assessment, all grocers are therefore assumed to have the same emissions intensity as Company A. This study uses market shares for Southern California from the Shelby Report. Company A leads the market at about 21% (

Table 2. Market shares of eight Southern California grocery companies [23].

Retailer	Market Share (2017)
Company A	21%
Company B	19%
Company C	13%
Company D	11%
Company E	9%
Company F	6%
Company G	6%
Company H	3%

).

For the food services industry, this study uses national market shares based on SEC filings, particularly an antitrust suit from 2015 when a major food services company (Company I) was prevented from buying another (Company J) because it would effectively control 75% of the food services market [24]. This paper assumes that those two companies still collectively control 75% of the market, albeit as separate entities. In 2019, Company I’s annual revenue was roughly double that of Company J. [25,26]. Assuming that the national market is representative of the Los Angeles market, this analysis attributes approximately 50% of food services emissions to Company I and 25% to Company J.

4.4. Analytical Demonstration of Top-Down Method

A formulaic expression of the top-down method is displayed in Figure 4. The same equation is used for NO_x and PM_2.5 emissions by replacing O with N and P, respectively. This uses parameters the parameters identified in

Table 3. Parameters of the top-down approach.

Variable	Definition	Value	Source
di	Distance from shipment i’s entry point to central LA	65, 75, or 225 miles	Google Maps
mi	1 for internal LA shipment, 0 otherwise		FAF
fi	1 for mixed freight, 0 otherwise		FAF
S	Food share of mixed freight	35.6%	CFS
Ti	Ton miles for shipment i		FAF
Ki	Kilotons in shipment i		FAF
G	Industry share of food shipments	Grocery: 67% Food Service: 33%	Saksena et al., 2018
A	Company’s grocery market share	Table 2 grocery market share (%)	LA Times, 2017
O	Emissions: g/ton-mile CO2	87.75	EPA SmartWay
N	Emissions: g/ton-mile NOx	0.22	EPA SmartWay
P	Emissions: g/ton-mile PM2.5	0.0035125	EPA SmartWay

to yield results of CO₂, NO_x, and PM_2.5 emissions in grams.

4.5. Results of Top-Down Approach Application

1.

Food freight tonnage and ton-miles

Based on the analysis conducted herein, this paper finds that 71.6 million tons of food freight were shipped by truck to the Los Angeles area in 2018, accounting for 17.1% of total truck freight destined for the region. The majority, 42.5 million tons, is intra-regional freight, originating from locations within the Los Angeles area. 15.7 million tons originate on the West Coast, entering on I5, while 13.4 million tons come from the rest of the US, entering the region in East San Bernardino County.

Food freight destined for the Los Angeles area contributes 6.3 billion ton-miles of truck traffic within the region every year. Despite contributing 60% of total tonnage, intraregional food freight contributes only 34% of total ton-mileage. Freight from other origins on the West Coast contributes 19% of total ton-mileage. Freight from the rest of the US represents the smallest portion of freight by weight but contributes almost 50% of within-region ton-mileage because it travels more than three times as far within the region.

2.

Emissions impact

Our top-down approach yields the following estimates about the Lost Angeles Area food industry annual goods movement emissions:

• 618,000 tons of CO₂,
• 1549 tons of NO_x and
• 24.6 tons of PM_2.5.

We estimate NO_x and PM_2.5 emissions from goods movement in the food industry cause local air quality health impacts of USD 54.4 million annually. Health impacts include early mortality, asthma, and strokes, among other ailments [27]. Using a conservative social cost of carbon of USD 45/ton, CO₂ emissions add an additional USD 27.8 million in social damages for a total societal cost of just over USD 82 million per year.

3.

Apportionment to individual companies

Apportionment of emissions to individual companies is based on their share of the overall market. This analysis, therefore, estimates that grocery retailers are responsible for two-thirds of the food industry’s goods movement emissions, amounting to USD 54.7 million per year of damages. Emissions impacts may be attributed to individual companies under this method by multiplying the total damage impact by each company’s market share. For example, with 20% of the market Company A is estimated to contribute 85,000 tons CO₂, 212 tons NO_x, and 3.4 tons PM_2.5, equal conservatively to an estimated total social cost of over USD 11 million per year. Company A is followed by Companies B, C, and D at USD 10 million, USD 7 million, and USD 6 million, respectively (

Table 4. Estimated annual emissions and social costs of goods movement serving the food industry in the Los Angeles area by end retailer.

Retailer	CO2 Emissions (Tons Per Year)	NOx Emissions (Tons Per Year)	PM2.5 Emissions (Tons Per Year)	Local Air Pollution Cost (USD Millions Per Year)	Total Social Cost (USD Millions Per Year)
Company A	84,769	212	3.4	7.5	11.3
Company B	76,951	193	3.1	6.8	10.2
Company C	54,730	137	2.2	4.8	7.3
Company D	46,500	117	1.9	4.1	6.2
Other Grocery	148,552	372	5.9	13.1	19.8
Grocery Total	411,501	1031	16.4	36.2	54.7
Company I	102,875	258	4.1	9.1	13.7
Company J	51,438	129	2.1	4.5	6.8
Other Food Services	51,438	129	2.1	4.5	6.8
Food Services Total	205,750	516	8.2	18.1	27.4
Grand Total	617,251	1547	24.6	54.3	82.1

).

The remaining one-third of food freight emissions is attributed to the food services industry. With 50% of the food services market, Company I can be attributed 103,000 tons CO₂, 258 tons NO_x, and 4.1 tons PM_2.5, amounting to a total social cost of USD 14 million per year. Company I’s main competitor, Company J, is responsible for half as many emissions, valued at USD 7 million per year.

5. Bottom-Up Approach

5.1. Purpose and Intent

The bottom-up approach to pollution accounting aims to quantify the regional emissions and air quality impacts of a specific company’s goods movement network. Optimally, this approach would be based on validated data quantifying regionally oriented fuel use associated with activities such as driving and idling. In the absence of such data, however, a bottom-up CPA approach seeks to quantify emissions based on public data and then infer the goods movement emissions of other companies in the sector based on their market share relative to the company studied. By grounding its analysis in a particular company’s distribution system, the bottom-up approach provides a method for comparing the air quality impacts of the goods movement networks of different companies in the food sector. The bottom-up approach complements the bird’s-eye view of the top-down approach with granular detail. In tandem, the two approaches have the power to encourage companies to commit to reducing their pollution and investing in zero-emissions vehicles.

While this study was able to apply the top-down approach and use it to draw conclusions about corporate responsibility for goods movement emissions in the Los Angeles area, there are currently too many data gaps for this paper to apply the bottom-up approach to a particular case study. Instead, this section proposes a methodology for the bottom-up approach and identifies areas of uncertainty and data gaps that need to be addressed for it to be implemented in the future.

5.2. Method for Quantifying Corporate-Level Emissions

1.

Distribution segment analytical framework

The foundation of the bottom-up approach is a map of a company’s network of manufacturing centers, distribution centers, and retail stores within a region. This paper proposes a five-part distribution segment analytical framework to quantify a company’s goods movement emissions. These segments include truck movement and idling in each of the following areas of activity:

• Transport from a good’s first point of entry to first destination in region of interest
• Pickup and transport from manufacturing and processing centers
• Pickup and transport from distribution centers
• Drop off at retail/wholesale stores
• Final drop off at customer establishment after pick-up at retail/wholesale store

Figure 5 shows that within each of these segments, the bottom-up approach estimates the amount of time spent idling, as well as the truck distance traveled. Then, by summing emissions associated with idling and driving, an emissions estimate is developed for each segment for a company for a particular year. Segments one through four are then summed to generate total emissions—emissions from the fifth segment (final drop off at customer establishment after pick-up at retail/wholesale store) are excluded because the goods are no longer under the company’s control.

A formulaic expression of the bottom-up method is displayed in Figure 6. Like the equation in Figure 4 for the top-down method, Figure 6 can be used to calculate NO_x and PM_2.5 emissions by replacing O with N and P, respectively. This uses parameters the parameters identified in

Table 5. Overview of parameters for bottom-up approach.

Variable	Definition
dx	Truck ton-miles associated with segment “x”
tx	Idling time associated with segment “x”
O	Driving emissions: g/mile CO2
N	Driving emissions: g/mile NOx
P	Driving emissions: g/mile PM
Oh	Idling emissions: g/h CO2
Nh	Idling emissions: g/h NOx
Ph	Idling emissions: g/h PM

to yield results of CO₂, NO_x, and PM_2.5 emissions in grams.

The equations in Figure 6 model the emissions associated with transporting one truckload of food from its entry into a region of interest to its arrival at a retail store.

2.

Scaling to single-truck-level emissions:

The above calculations account for emissions from one truckload of food being transported into an area of interest and eventually making its way to a store shelf. From entry to a region through drop-off at retail stores, a truckload would be carried the sum of d₁ + d₂ + d₃ + d₄. To scale the emissions of one truckload to the average annual emissions of a single truck, divide the average annual mileage of a truck by the average distance of a single truckload of food. The average truck in California, for example, travels 20,039 miles per year [28], meaning that one of a company’s trucks may emit about $\frac{\mathrm{20,039}\mathrm{miles}}{{\mathrm{d}}_1+{\mathrm{d}}_2+{\mathrm{d}}_3+{\mathrm{d}}_4\mathrm{miles}}$ times more than the emissions calculated for a single truckload.

3.

Scaling to corporation-level emissions:

Scaling the emissions from moving one truckload of food from its entrance into a region until its arrival at a grocery store to a corporation’s annual emissions within the geographic area requires the use of several company-specific data variables to arrive at a calculation of the number of truckloads received by each store each year. The needed variables are listed in

Table 6. Variables needed to calculate corporation-level emissions.

Variable	Definition
r	Annual revenue
s	Number of stores in the US
c	Average cost of food freight
P	Truck payload capacity
W	Percent food waste [29]
M	Percent markup [30]

.

The number of truckloads delivered per store per day can be calculated as follows:

4.

Translating to monetary and health impacts

As with the top-down approach, this analysis can use the Environmental Protection Agency’s 2018 Value of Emission Impacts Study to assess the health impacts and monetary costs of a given level of NOx and PM emissions. As discussed in the top-down method section above, EPA estimates societal impacts of USD 950,000 per ton of PM emitted and USD 20,000 per ton of NOx emitted. Those costs are calculated based on valuations of reductions in premature mortality risk and reductions in the relative risk of particular health effects.

5.3. Areas of Uncertainty and Data Gaps

Uncertainty in this approach stems primarily from limitations in the data available concerning distance traveled and truck idling time. Driving and idling emissions factors contain inherent uncertainty as well. The following list breaks down the areas of uncertainty and proposes a method for the estimation of each.

1.

Distance travelled

An accurate estimate of distance traveled for each segment must determine the average distance a truck travels to and from each site.

For segment 1, the analysis must determine the average distance a truck travels from point of entry to its first destination in the region of interest. To do so, it would be helpful to have a statistical account of a company’s freight flows into a region.

For segment 2, the analysis must determine the fraction of goods that skip manufacturing centers and ship directly to distribution centers. It also must determine the origin of a company’s trucks that pick up the goods from manufacturing and processing centers. Next, it must evaluate the average distance between manufacturing centers and distribution centers.

For segment 3, the analysis must ascertain the average distance a truck travels from the site of origin to the distribution center. It must then determine the average distance between a distribution center and a retail store.

For segment 4, the analysis must evaluate the average distance a truck travels from retail store back to its storage site.

2.

Idling time

Another area of uncertainty is the amount of time trucks spend idling during each segment. Certain companies may track this data. If not, it could be ascertained by surveys of a company’s manufacturing/processing centers, distribution centers, and retail stores.

3.

Emissions factors

Driving emissions factors will vary depending on the type of truck a company uses. One way to estimate the emissions factors for a company’s truck fleet would be to obtain a profile of their fleet and generate a “statistically average truck” based on the numbers of each type of truck they use. EPA SmartWay emissions factors could be used as a reference point for driving emissions factor estimates.

Idling emissions factors will also vary depending on the type of truck a company uses. As for driving emissions, with a profile of a company’s fleet, a bottom-up analysis could generate a company’s “statistically average truck” and determine idling emissions factors accordingly. Khan et al.’s 2009 study estimated the idling emission factors for diesel trucks to be 8350–10,100 g/h CO₂, 34.63–161.41 g/h NO_x, and <5 g/h PM [31].

4.

Scaling

This paper provides a method for scaling the emissions of a single truck to a corporation’s annual goods movement emissions in the absence of data providing the number of truckloads received by each store each year. The method relies on assumptions about the average cost of food freight and percent food waste to generate its estimate. However, if a bottom-up analysis were granted access to a corporation’s data tracking the number of truckloads received by each store each year, it could produce a much more accurate estimate of that corporation’s annual goods movement emissions.

6. Results

6.1. Summary of Analytical Findings

1.

Top-down method

The top-down approach developed for this study produces estimates of the local air pollution impacts and attributes responsibility to market-leading corporations based on market share. It found the total social cost of goods movement emissions generated by the food sector in Los Angeles to be USD 82.1 million per year, with Company A individually costing USD 11.3 million in air quality impact. Quantifying company- and industry-specific air pollution impacts represents an unprecedented step towards attributing responsibility for goods movement emissions to corporations.

Although this method presents a step forward, the analysis required the use of assumptions and generalizations that introduce uncertainty in the quantification of impact. Relative to a bottom-up approach, the top-down analysis paints in much broader strokes. It does not account for the unique profile of each individual company’s goods movement network; it operates in averages. The uncertainty could be reduced by refining several key assumptions: commodity selection, distance estimation, emissions factors, and the social cost of emissions. Additionally, since the FAF data used for this method is based on the Los Angeles-Long Beach Combined Statistical Area (which includes uninhabited areas and covers several air basins), refining freight and market data to isolate South Coast Air Quality Management District-based freight operations would allow for more localized comparison to regional emissions inventories and higher accuracy. A more refined spatial scale would also enable the analysis to identify potential inequities in air quality impact on disadvantaged communities.

Notwithstanding the analytical uncertainty that reduces the precision of the model output, this analysis proposes and validates a methodology for attributing goods movement emissions to consumer-facing corporations that can be easily expanded to other regions and industries.

2.

Bottom-up method

The bottom-up approach presented provides a foundation for developing spatially detailed analyses of the transportation emissions associated with the operations of individual companies within a particular sector and region. By using company-specific data for factors such as business locations, idling times, and truck characteristics, the bottom-up method creates a more granular and tailored assessment of corporate emissions.

Like the top-down method, the bottom-up method introduces several layers of uncertainty. Although much of the company-specific data can be inferred, without specific knowledge of private operations, the quality of the analytical output for this paper was found to be too low to present results. However, by centering the analytical framework on known manufacturing/distribution centers for individual companies, this method presents a clear connection between the operations of individual businesses and the emissions of the supply chain and logistics suppliers.

Ideally, corporations will provide the company-specific data for future bottom-up analyses. Companies could then use these studies to document and report their truck emissions reductions to customers, shareholders, and regulators.

6.2. Discussion: Further Areas of Analysis for Accuracy and Scaling

To improve the methods discussed in this paper and create scaling opportunities to move into other sectors, several areas of additional study and analysis are identified:

• First, a comprehensive analysis of goods movement emissions in the Los Angeles area that expands the top-down approach beyond the food sector and then compares observations with the findings of this paper.
• Second, a truck-level analysis that quantifies the air quality benefits from a single zero-emissions truck and the damages of a single diesel truck, depending on VMT and truck type.
• Third, community-level proximity mapping that highlights the disproportionate risk from goods movement emissions on communities of color and provides a detailed spatial assessment of where individual corporations can prioritize investments in local communities.
• Fourth, emissions analysis that evaluates the appropriateness of current emissions factors used for vehicles and ensures those factors accurately represent in-use truck emissions

This four-pronged approach would provide data and observations from multiple angles to expand and refine quantification of the impact caused by the diesel-powered goods movement associated with consumer-facing corporations.

1.

Comprehensive Goods Movement Emissions Analysis

Having established an analytical methodology for attributing goods movement emissions from the food sector in the Los Angeles area, the CPA method could expand its analysis beyond the food sector. The methodology employed here can easily be translated to assess the emissions impact of any SCTG commodity or combination of commodities. One challenge of this approach would be translating from SCTG commodity codes to industries of interest. An input-output modeling approach should be utilized to identify which industries use which commodities.

A 2012 analysis conducted by ICF for EDF entitled U.S. Freight Emissions Segmented by BCO Industry provides a framework for attributing FAF commodity codes to industries by beneficial cargo owner (BCO) [32]. Modifying ICF’s approach to the Los Angeles area would yield estimates of emissions by the NAICS industry. While NAICS industries are still broad for these purposes, they could be broken down further using sales data or generalized assumptions. For example, ICF found that the retail industry was responsible for 20% of nationwide freight emissions. Determining the freight emissions of the retail industry in the Los Angeles area would be a significant and worthwhile step. Going a step further to determine emissions by specific retail sectors would be an even greater achievement.

2.

Truck-Level Analysis

The analysis conducted in this study used the EPA SmartWay emissions factors to estimate vehicle emissions. A more robust yet time-consuming analysis would be to use the California EMFAC2017 database to estimate truck emissions. This database can be used to estimate the average emissions impact of several high-priority categories of diesel trucks. EMFAC2017’s air emissions inventory also provides an estimate of truck populations at the county or air district level. The inventory includes 16 distinct sub-categories for class 8 tractors. For each subcategory, EMFAC estimates daily and annual emissions of NO_x, PM_2.5, CO₂, and seven other pollutants.

For each vehicle category, dividing emissions by truck population yields the average air emissions impact of a hypothetical vehicle traveling exclusively within that district. (If a truck drives 75% of its miles in one air district and 25% in a different air district, EMFAC2017 splits the vehicle proportionally between the two districts when estimating population). Back-of-the-envelope analysis for the Port of Los Angeles Drayage Trucks category suggests that on average each truck emits 75 tons of CO₂, 0.261 tons NO_x, and 2855 g PM_2.5 each year. A similar analysis could be conducted for the remaining 15 categories of class 8 trucks. Truck categories can also be broken down by model year to account for the age of the truck being replaced [17].

3.

Proximity Mapping

Goods movement emissions are not spread equally throughout the Los Angeles area; rather, they are concentrated in high traffic areas that disproportionately harm communities of color and socioeconomically disadvantaged communities. Distribution centers and other truck-attracting businesses should represent an important unit of analysis for quantifying and highlighting the adverse risks from goods movement in local communities.

Furthermore, the EPA estimates of cost per ton of emissions used by this paper represent an average for the entire United States and do not reflect variability in cost by location. The communities most densely located beside sources of goods movement emissions, or with underlying health conditions exacerbated by air pollution, may sustain significantly higher costs per ton than the national average. A more granular geographic assessment of risks and costs from emissions would more accurately attribute those costs to the facilities and roadways where the emissions are concentrated.

Government agency efforts such as the California Air Resources Board’s AB 617 program, as well as EDF’s work in Oakland, Houston, New Jersey, and London, provide a starting point for mapping vulnerability around major transportation corridors, industrial centers, and truck attracting businesses like distribution centers. In these examples, mapping and analysis of the co-location of pollution hot-spots and major business operations are helping identify the impact of transportation emissions on local communities.

One example of the relevance of air impact mapping can be seen in Figure 7, showing San Bernardino, California, an area east of Los Angeles that is increasingly known for sprawling warehouse complexes that rely on heavy-duty truck pickups and deliveries to support regional goods movement. As part of efforts to overlay local pollution sources with community health impacts, numerous truck-attracting business locations and the impacts of their emissions have been identified and are the focus of community air impact assessments [33].

In the Los Angeles area, the statewide CalEnviroScreen tool establishes a foundation for conducting community vulnerability assessments. CalEnviroScreen ranks census tracts based on the sensitivity of their population and pollution burden. During the research to develop the CPA methods identified above, several distribution centers owned by top grocery retailers were found to operate in disadvantaged communities in the Los Angeles area. Embedding additional information on air emissions inventories and air toxins maps for several highly burdened communities in Los Angeles, including the area directly surrounding the San Pedro Ports, would be helpful for ensuring the accuracy of emissions impact estimates. In tandem with CalEnviroScreen, proximity mapping tools, and public health impact assessment analyses, the CPA mechanism can help companies decide which of their truck routes to prioritize for electrification based on its relative pollution burden to the surrounding community. By linking goods movement emissions estimates to local health impacts, the CPA mechanism can emphasize the high stakes for continuing business as usual and further underscore the health benefits of truck fleet electrification.

4.

Emissions Factor Analysis

Sectoral analysis of goods movement emissions included in this paper indicates that several different vehicle use applications may contribute to the overall emissions burden of any particular company. These include delivery by sleeper cabs coming from far away distances, regional transport via day cabs, truck idling at depots and drop-off establishments, refrigeration unit emissions, and deadheading. As a result, the emissions factors extracted from a single company may provide overly broad conclusions that increase uncertainly of CPA results. Evaluating and comparing the in-use emissions of trucking segments to emission factors would therefore improve the precision and overall accuracy of results.

6.3. Relevance to Corporate Actors and Investors

The CPA methodological approaches developed in this paper offer corporations and investors valuable insight into the significance of air quality impacts in goods movement networks and a mechanism with which to independently quantify supply chain emissions and compare them to those of industry peers.

Furthermore, the proposed CPA methodology provides a publicly accessible framework for quantifying a corporation’s Scope 3 emissions that have generally eluded reporting requirements. The EPA greenhouse gas emissions inventory sorts emissions into three categories: Scope 1, Scope 2, and Scope 3. Scope 1 includes all direct emissions from the activities of a corporation, Scope 2 includes all indirect emissions from electricity purchased and used by a corporation, and Scope 3 includes all other indirect emissions from assets not owned or controlled by a corporation. The EPA only requires companies to quantify and report Scope 1 and 2 emissions, yet the majority of a company’s emissions often fall under Scope 3. Goods movement emissions in particular are often categorized as Scope 3 because trucks often belong to third-party shipping companies.

One way industry leaders can reduce the air quality impacts of their supply chains is by pursuing the use of zero-emissions vehicles in their distribution networks. Companies can accomplish this in numerous ways: chiefly, by directly purchasing new zero-emissions trucks or by encouraging the companies and haulers with which they contract to do the same.

Companies that are committed to sustainability can use the methods described in this paper to pursue independent quantification of their emissions while taking steps to reduce them across their supply chain. Then, as companies share their findings with the public and peers, including the developed strategies for reducing those emissions, decisive action to eliminate the emissions associated with all goods movement networks can result.

Investors committed to sustainability may find the top-down method useful since it enables comparisons of portfolio companies. As companies run their own analysis and disclose, investors’ ability to distinguish leaders and laggards will grow.

6.4. Relevance to Academic Institutions and Researchers

This paper provides one of the first public attempts the authors could find to develop an analytical framework for quantifying and attributing goods movement emissions to companies within a sector or region. Similar efforts identified included analysis of the emissions from a single company. Academic institutions and researchers focused on carbon accounting, business sustainability and disclosure, and other fields may be interested in this as a foundation from which to build accounting tools for other sectors and geographies. With a more robust effort, the academic community could theoretically flesh out a comprehensive map of goods movement emissions across the US and categorize emissions by sector and source company. In addition, researchers can investigate the areas of uncertainty identified in this paper to increase the confidence of future findings.

7. Conclusions

Using publicly available data, this study evaluated two approaches to quantify the emissions and air pollution impacts generated by the business operations of individual companies and their goods movement networks in the food sector in the Los Angeles area. Both methods found evidence of significant pollution created by transportation activities and documented mechanisms to attribute responsibility for that pollution to individual companies based on their share of the region’s market. While both methods have uncertainties, they provide an analytical foundation upon which to move forward with pollution accounting for goods movement. They also offer a mechanism with which to encourage corporations to take responsibility for the air quality impacts of their distribution networks.