Estimation processes for total shipments of retail establishments, services, or household/business moves are addressed in this section.  Data sources and estimation methods applied in generating each set of flows are presented as well. 

8.1 RETAIL

8.1.1 Data Sources for Retail Sector

Census Annual Retail Trade Survey

One of the main data sources used in generating retail sector estimates was the Annual Retail Trade Survey, published by the U.S. Bureau of Census (Census) in the table “U.S. Retail Trade Sales – Total and E-commerce: 2013-1998.”  The “2012 revised” sales estimates were extracted from this table by 3-digit NAICS code and used as national control totals for industries associated with the retail sector.  Total retail trade sales in 2012 were estimated at $4,306 billion, including $229 billion from e-commerce.  Businesses with or without paid employees are included in these Census estimates.  The Census¹⁸ defines e-commerce as “transactions sold online whether over open networks such as the Internet or proprietary networks running systems such as Electronic Data Interchange (EDI).”

2012 Economic Census Receipts

Estimates of state-level business data including sales receipts by NAICS are also published by the Census based on information collected under the 2012 Economic Census¹⁹.  Receipts are defined by the Census²⁰ as “operating revenue for goods produced or distributed, or for services provided … excludes local, state, and federal sales and other taxes collected from customers or clients.”  Because the Economic Census is conducted by the Census in a five-year interval, receipts data are available for years ending in “2” and “7” only.  Using information from the 2012 Economic Census, Census estimated a total of $4,238 billion in receipts was generated from the retail sector (NAICS 44-45).  Note that, receipts data associated with the retail sector in all states and the District of Columbia (DC) are published at the aggregated 2-digit NAICS “44-45” level.  Receipts data for retail sectors at the disaggregated 3-digit NAICS level, however, are released only for 26 states and for most of the retail sectors in Washington D.C. (where NAICS 441 and 452 data were suppressed). 

2012 County Business Patterns

Employment payroll data from Census’ County Business Patterns²¹ (CBP) is utilized in distributing state-level estimates to the CFS areas.  The way this information was used in the FAF4 process is discussed in Section 8.1.3.

8.1.2 Assumptions on Percent of Retail-Sales Receipts Involving Truck Shipments

It is expected that most retail sales be made at store locations where customers carried goods purchased to home.  Often, when delivery is made from the retail stores, the customer-purchased items are delivered via store-owned vans (or pickup trucks), or delivered by mail or package carriers.  Certain large items purchased from the retail stores, such as furniture or appliances on the other hand, are more likely delivered by truck to a customer’s home.  Due to lack of data on actual share of sales associated with goods delivered versus carried out, a set of assumptions on percent of retail-sales receipts involving truck were made for the FAF4 process.  Depending on characteristics of the industry (at 3-digit NAICS), their truck-share of receipts could be assigned with a different percentage ranging from 1% (e.g., clothing) to 70% (e.g., furniture).

8.1.3 Estimation Approach for Goods Movements in the Retail Sector

Estimating Total Retail Values Shipped by States at 3-digit NAICS-Level

As mentioned earlier, 2012 receipts data at the 3-digit NAICS for 26 states are published by the Census.  Therefore, for these states, their total retail receipts associated with industry sectors involving truck deliveries (NAICS codes 442-446, 448, and 451-453) can be estimated directly using the assumed shares.

Estimating Total Retail Values Shipped by States at the Aggregated NAICS-Level

For states that Census does not publish receipts data at the 3-digit NAICS level, where only state-level receipts for the entire retail sector as a whole (NAICS 44-45) is available, patterns from the national total are applied.  Specifically, total retail receipts in each individual state are reduced by 43% to remove the portion of receipts involving NAICS 441/447/454 sectors (i.e., to eliminate receipts from in-scope CFS sectors).  The share of truck-delivery shipments for these states was assumed at 8.4%, as calculated based on combined totals from all known states (i.e., dividing “shipped receipts” by “total in-scope receipts”).  Applying this share to adjust receipts of individual states, state-level shipment values can be estimated.

Estimating State-Level Shipment Values and Weight by Commodity

Total state-level shipment values are distributed among commodities involved in each NAICS sector considered in this OOS area.  A simple equal-share assumption was applied when more than one commodity could be involved within a specific 3-digit retail subsector.  Once the shipment values are separated by commodity code, value-to-weight ratios by commodity, as calculated based on data for domestic shipments from the 2012 CFS Public Use Microdata (PUM), are applied to derive estimates for shipment weights at the state level.

Distribution of State Totals to FAF Zones

The state-level estimates of values and weights are distributed to each FAF4 zone within the given state using their shares of total annual payroll amounts obtained from the 2012 CBP dataset.  Note that the calculation of shares considered only payroll information associated with the associated 3-digit NAICS codes for retail.  For simplicity, this FAF-zone level distribution is conducted uniformly over all commodity codes. 

Since most purchases at retail stores occurred in regions where the customers reside, the O-D FAF zones for the retail OOS sector are assumed the same.  Note that shipments involved under this OOS area are assumed to be transported by truck.  Based on the processes just described, it is estimated that a total of $206 billion, weighing 224 million tons, of CFS OOS retail goods were transported by truck in 2012. 

8.2 SERVICES

8.2.1 Data Sources for Estimating OOS Shipments from Services Sector

2012 Service Annual Survey Data and Report

The Census conducts the Service Annual Survey (SAS) to provide national estimates of annual revenues and expenses of establishments classified in select service industries.  The estimates published²² by the Census are developed using data from a probability sample of firms located in the United States that have paid employees (i.e., employer firms).  Consequently, Census-published estimates only include data for employer firms.  The sample is regularly updated to reflect the universe of employer service businesses and covers both taxable firms and firms exempt from Federal income taxes.  Furthermore, 2012 Service Annual Survey data are summarized by industry classification based on the 2007 NAICS (same as the 2012 CFS).  A table titled “Estimated E-Commerce Revenue for Employer Firms: 1998 through 2012” from the Annual Services Report was used and supplemented with other SAS tables from the same report to generate estimates for the OOS Services sector.  The 2012 revenue information was extracted from that table and used as the national control totals for industries associated with the services sector, by 3- to 5-digit NAICS codes.  The NAICS codes involved in this OOS component are listed in Table 8–1. 

Table 8–1 . NAICS Industries Involved in the OOS Services Sector

2012 Economic Census

Similar to the retail sector, information collected under the 2012 Economic Census was also used in the estimation process for the services OOS component.  Census estimates a total of $11.7 trillion in receipts was generated from the services sector (NAICS 51-81) in 2012.  Receipts data associated with the services sector in all states and DC are published by the Census at an aggregated 2-digit NAICS level.  Receipts data for services subsectors and industries at the 3-digit to 5-digit NAICS level are released only for a limited number of states.

Census County Business Patterns

As in the retail OOS estimation process, the 2012 CBP data was used to disaggregate state-level estimates to the FAF zones.

8.2.2 Estimation Approach for Services Sector Goods Movements

Estimating Total Value by State

As mentioned previously, 2012 receipts data at the 3- to 5-digit NAICS level are available from Census for a select number of states.  For each of these states, state-level receipts for relevant services industries (identified by NAICS in Table 8–1 ) can be estimated directly using Census-published data and adjusted with shares of truck shipment in a similar matter as for the retail sector.  

For states that Census did not publish service-sector data at the detailed NAICS level (i.e., only state-level receipts for the entire sector are available), patterns from the national total were utilized.  Specifically, each individual state’s total services receipts were adjusted to remove the portion of receipts involving NAICS industries not involving truck deliveries.  This adjustment was made with a factor calculated from all known states, where the “total receipts involving truck” is divided by the “total receipts from all in-scope industries.”  By applying this factor to adjust receipts of individual states within this group (i.e., those without detailed NAICS level data), their state-level shipment values can be estimated.

Estimation of State-Level Shipments by Commodity

As in the retail OOS component, the state-level shipment estimates obtained from above process are distributed among commodities involved in each involved-NAICS industry.  The same simple even-share approach was applied to allocate shipment values among commodities within a given NAICS.  Value-to-weight ratios by commodity, as calculated based on domestic shipments from the 2012 CFS PUM data, by service sector industry, are then applied to derive estimates for shipment weights at the state level.

Distribution of State Totals to FAF Regions

Using the same method as in the retail sector, state-level service-sector estimates of values and weights are distributed to FAF zones in the given state, based on their shares of payroll from the 2012 CBP data (considering service sector NAICS only). 

Determination of OD Flows

To determine O-D flows for the above-estimated service sector shipments, additional processing is needed to identify destinations of service-related shipments.  Depending on the type of service industries, the destinations of shipments are allocated differently in two ways.

Intra-Zone Movements

For most industries in this OOS sector, their “services” most likely occurred in regions where the customers are located.  Therefore, it is reasonable to expect most of these activities are intra-zone movements, i.e., the O-D FAF zones would be the same.  Note that shipments involved in this OOS area are all assumed to be transported by truck.

Inter-Zone Movements

On the other hand, the industries sectors under NAICS 7111, 7112, and 71211 (Performing Art Companies, Spectator Sports, and Museums, respectively) are services requiring movements of goods to regions beyond their originating zones.  It was assumed that these sectors would only be traveling to major metropolitan areas, typically in the same or neighboring state; therefore, “Rest of State” FAF zones were not included as potential destination choices.  The share of truck shipments by average distance-range, calculated using the 2002 VIUS data, is used to estimate O-D flows for each of the involved NAICS sectors.

Based on the processes discussed above, the FAF4 estimates that a total of $119 billion weighing 71 million tons of services-associated shipments were transported by truck in 2012. 

8.3 FLOWS OF HOUSEHOLD AND BUSINESS MOVES

Trucking services provided by the household and business (HH&B) moving industry is under the NAICS 484210, Used Household and Office Goods Moving.  Note that NAICS 4842 is covered under the 2012 CFS.  However, as a shipper-based survey, shipments of household and business goods in the moving industry are not captured by the CFS.  This is because businesses that engaged in these moving activities do not typically produce freight or warehousing services.  

8.3.1 Data Sources for Estimating Shipments from the Moving and Storage Industry

The primary data sources for the HH&B OOS component are the American Community Survey (ACS) County-to-County Migration Files published by the Census, the Consumer Durable Goods Current-Cost Net Stock from the Bureau of Economic Analysis (BEA), and information obtained from various publications of the American Moving and Storage Association (AMSA). 

2008-2012 ACS County-to-County Migration Files

The ACS program combines consecutive yearly datasets to increase the sample size and provide reliable estimates for smaller geographic areas (e.g., county and Census tract).  Although ACS estimates are produced for 1-year and 3-year datasets, only the 5-year datasets provide estimates for county-to-county migration flow²³.  For this HH&B OOS component, the 2008-2012 release of ACS county-level migration data²⁴ was used.

The data provided in the ACS county-to-county flow files include county of current residence, county of residence 1 year ago, and the number of movers between the two years.  County-level total population and total number of housing units in 2012 are also obtained from the ACS.  The population total and number of housing units by county are used in FAF4 to estimate average household size (i.e., number of persons in a household) for each FAF region. 

Consumer Durable Goods, Current-Cost Net Stock

The BEA publishes national statistics on both fixed assets and consumer durables²⁵.  While fixed assets cover items that are used continuously in processes of production for an extended period, consumer durable goods are generally defined as tangible products that can be stored or inventoried that have an average life of three or more years.  The BEA’s Current-Cost Net Stock of consumer durable goods²⁶ in 2012 is used to identify commodities associated with moves and the value of goods being moved.

AMSA Published Statistics

Because the ACS data is population-based and the migration flows are estimated by considering changes in counties of residence, business moves are not captured.  Statistics released online by AMSA²⁷ were used to adjust estimates produced from ACS in order to capture missing moves.  According to 2014 statistics, AMSA stated that shipments from “corporate and other federal government” moves accounted for about 38% of total household goods shipments that occurred in that year.  Furthermore, AMSA reported that about 40% of the interstate household goods moves were carried out by consumer themselves (i.e., not by professional movers or by use of rental truck).  This percentage was used to adjust ACS-based estimates to remove unassisted moves.

CFS Value-to-Weight Factors

As mentioned earlier, current-cost net stock of consumer durable goods from BEA was used to estimate the volume of HH&B goods being moved.  Thus, estimates generated from this process reflect only values of goods, not their weights.  To estimate commodity weights for the HH&B goods, value-to-weight ratios calculated from the 2012 CFS PUM data, considering domestic shipments by commodity, are applied.  Since most HH&B goods moved are typically “used” items, as compared to “new” CFS goods, a simple depreciation rate of 30% was applied to discount the CFS-based value-to-weight factors for estimating the associated weights of HH&B goods.

8.3.2 General Assumption and Assignment of Commodity

In addition to the depreciation rate of value-to-weight factors discussed above, assumptions were applied in assigning commodity codes to the associated durable goods and in determining the share of specific commodities being moved by truck (versus items carried by the household members or left with the “old” house).  Moreover, due to data limitation, intra-county moves were assumed to be self-moves that did not involve moving trucks. 

The assignment of the commodity code is done by examining the types of consumer durable goods specified in the BEA Current-Cost Net Stock table.  This process identified five SCTG codes that were involved in this OOS area, including codes 35 (electronic equipment), 36 (motorcycles & bicycles), 39 (furniture), 40 (sporting goods), and 43 (miscellaneous). 

8.3.3 Estimation Process for Flows of HH&B Moves

The estimation process associated with flows for HH&B moves is straightforward.  A brief description of the process is given below.

Estimating Total Number of Moves

The ACS county-to-county migration flows, measured in population, are aggregated to the FAF-zone level and converted to number of household moves by applying the average household size factors generated using Census population and housing unit data for each region.  It is assumed that each household made one move as a whole.  Under this assumption, the number of migrated households would be treated as equivalent to the number of “moves” or “OD flows”.

The number of FAF zone-level moves is adjusted “upward” to include moves associated with “corporate and other federal government” and “downward” to exclude consumer non-assisted moves.  The national average taken from AMSA, discussed above, was applied to all regions. 

Estimating Total Value per Move by Commodity Code

The national total reported in the 2012 BEA Current-Cost Net Stock  can be divided by the total number of households to derive a per-household value for each of the 5 commodity codes (at the national level) after adjustment to eliminate items not likely to be transported in a moving truck.  These per-household values are then multiplied by the total number of mover-households from each region to obtain values of SCTGs associated with the HH&B OOS.  Finally, the CFS-based value-to-weight factors described above are applied to the values to estimate shipment weights by SCTG. 

National averages of per-household value by SCTG are applied to all moves regardless of the possible existence of regional differences.  This is a current data limitation that could be further examined and improved upon in the future.

Finally, it is estimated that truck shipments worth over $128 million, and weighing about 29 million tons were generated from the HH&B sector in 2012.  More than half (56%) of the total value of these shipments are for common household items of SCTG 39 (furniture) and 35 (electronic).

¹⁸ Definitions, E-Stats, U. S. Census Bureau, https://www.census.gov/econ/census/help/sector/data_topics/e-commerce.html, accessed July 2015.

¹⁹ Economic Census, http://www.census.gov/econ/census/ accessed July 2015.

²⁰ Definitions, Statistics of U.S. Business, U.S. Census Bureau, https://www.census.gov/econ/susb/definitions.html, accessed July 2015.

²¹ County Business Patterns, U.S. Census, http://www.census.gov/programs-surveys/cbp.html.

²² U.S. Census Bureau, Annual & Quarterly Services website at https://www.census.gov/services/index.html, accessed July 2015.

²³ The 2008-2012 ACS 5-Year Summary File Technical Documentation, U. S. Census Bureau, Version 2, February 2014. http://www2.census.gov/acs2012_5yr/summaryfile/ACS_2008-2012_SF_Tech_Doc.pdf

²⁴ 2008-2012 ACS County-to-County Migration Files Documentation, U. S. Census Bureau, accessed July 2015, https://www.census.gov/hhes/migration/files/acs/county-to-county/2008-2012/2008-2012%20Migration%20Flows%20Documentation.pdf

²⁵ “Detailed Data for Fixed Assets and Consumer Durable Goods,” National Economic Accounts, Bureau of Economic Analysis, U.S. Department of Commerce, see http://www.bea.gov/national/FA2004/Details/Index.html.

²⁶ Table 8.1 Current-Cost Net Stock of Consumer Durable Goods, Fixed Assets Accounts Tables, National Data, http://www.bea.gov/iTable/iTable.cfm?ReqID=10&step=1#reqid=10&step=1&isuri=1, Bureau of Economic Analysis, accessed July 2014.

²⁷ American Moving and Storage Association ,About Our Industry, About AMSA, http://www.promover.org/content.asp?pl=1&sl=61&contentid=61, accessed June 2015.