Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Prepublication copy, uncorrected proofs Summary The Bureau of Labor Statistics (BLS) asked the Committee on National Statistics (CNSTAT) of the National Academies to evaluate changes in the retail trade sector since the 1990s, assess measures of employment and labor productivity for the sector, and discuss the value and specifications for a satellite account to measure retail-related employment and labor productivity that would better capture the transformation. The request was motivated in part by shifts in the ways that warehouses, transportation, and delivery services are now supporting retail, which are not reflected in retail employment and labor productivity statistics. The panelâs primary information-gathering activity was to hold a workshop that provided input from researchers, industry representatives, data users, and relevant statistical agencies. The workshop supplemented the panelâs expertise on the economics and statistics of the retail sector with the expertise of additional economists who have studied the retail sector, experts who understand the details of current government statistical programs, and industry representatives. The panel took a broad approach in reviewing options for a retail-related satellite account, considering both pragmatic immediate steps and aspirational longer-term goals. It also identified ways to progress toward the aspirational goals by carrying out specific analyses, collecting additional data, and conducting case studies. TRANSFORMATION OF THE RETAIL SECTOR The retail sector has experienced a number of important changes over the past few decades in both the way it is structured and the nature of the goods and services it provides. These changes include the rise of warehouse clubs and supercenters; the rise of e-commerce; the digital transformation of some retail goods, such as books, music, and video; the increase in imports of retail goods and services; the role of large firms in driving the transformation in retail; increased product variety and the role of retail firms in presenting and organizing products; and recent changes in response to COVID-19 that have in turn heightened some longer-term trends. As a result of these changes, the cost structure of large retailers is now often quite different from that of small retailers (Conclusion 2-1). Large retailers often provide wholesale, warehousing, and transportation services directly, whereas small retailers usually purchase these services. In addition, large retailers sometimes outsource some traditional retail services, such as customer service and order fulfilment, whereas small retailers usually provide these services directly. This difference in cost structures between large and small retailers heightens the importance of using measures of employment and labor productivity that can be meaningfully compared across retailers that are structured differently. The recent transformation in the retail sector has also shifted some retail services outside the traditional definition of the sector, for example shifting videos from sales to leasing. It has 1
Prepublication copy, uncorrected proofs also brought some services into retail that were formerly outside the traditional definition, such as providing delivery services for e-commerce purchases (Conclusion 2-2). Where this has taken place, an understanding of the employment and productivity effect of the changes will require analyses that compare services inside and outside the traditional retail sector. Beyond these specific changes, the dynamic nature of the retail sector ensures that newâ and as yet unknownâchanges will regularly appear in the years ahead to challenge available measures of employment and labor productivity. This dynamic nature raises an additional challenge to efforts to track the ongoing transformation in retail and continue to adapt retail- related measures over time. MEASURING RETAIL EMPLOYMENT AND LABOR PRODUCTIVTY The U.S. statistical programs that collect retail-related data provide a framework for measuring retail employment and labor productivity, but they also have some notable constraints. Calculating labor productivity involves estimating the real output of the retail sector and dividing it by the employment in the sector. All aspects of this simple definition pose conceptual and practical measurement difficulties. Federal economic data are collected by industry according to the North American Industry Classification System (NAICS), which classifies establishments hierarchically according to their business processes. Although this system usefully groups similar business establishments together in providing data, it specifically separates wholesale, warehousing, and transportation services into their own industries, even though large retailers are now increasingly integrating these functions into their retail operations. Similarly, retail transactions that take place through leasing rather than sales or through digital products appear in entirely different industries. As a result, the way the data are currently collected makes it difficult to identify the portion of wholesale, warehousing, and transportation services, or the portion of leasing or digital transactions, that are closely related to retail trade and could be usefully analyzed as part of a broader retail-related sector. A study of a broader retail sector will require estimates of the retailed-related portion of industriesâsuch as warehousingâwhere the relevant NAICS codes are only partially related to retail (Conclusion 3-1). Labor productivity measures are calculated with data provided by two different agenciesâBLS and Censusâthat use separate business registers with separate classifications of business establishments as sampling frames for their surveys to estimate output (Census), price deflators (BLS), and labor input (BLS). The differences between these sampling frames likely contribute to error in the labor productivity estimates (Conclusion 3-2). This error could be investigated and a reconciliation could be undertaken between the two business lists (Conclusion 3-3), and that in turn could be used to develop factors to adjust for the effects of any systematic differences between them (Conclusion 3-4). The ideal long-term solution would be for the federal government to develop and use a single common business register (Conclusion 3-5). The nominal output of the retail sector is defined in four different ways in the federal statistical system: (1) as total sales revenue; (2) as the difference between sales revenue and the cost of goods sold (gross margin); (3) as the difference between sales revenue and the cost of all purchased inputs (value added); and (4) as the difference between sales revenue and the cost of all inputs purchased within the sector (sectoral output). For narrowly defined sectors, the sectoral 2
Prepublication copy, uncorrected proofs output measure is effectively sales revenue, but as the scope of a sector becomes increasingly broad the sectoral output measure moves toward a value-added measure.1 A sales revenue measure of output is the simplest to produce, but it does not reflect changes in a retailerâs cost structure when additional functionsâlike warehousingâare integrated into the business. A value-added measure of output is theoretically preferred for measuring labor productivity in retail, capturing the difference between gross output and intermediate inputs. However, it requires estimating all noncapital purchased inputs, not just goods purchased. Comprehensive measures of value-added at the industry level rely on input- output accounts that have limitations in source data, including the frequency of updates. A gross margin measure of output for retail and wholesale trade reflects the value of the most important input for a retailerâthe cost of goods soldâwhile sidestepping problems related to estimating other inputs (Conclusion 3-6). For retail-supporting services that might be combined with retail trade in a broader retail-related sector, similar choices are necessary concerning which measure of nominal output to use, although the gross margin concept applies only to retail and wholesale trade. The Economic Census and the Economic Surveys provide limited data on purchases and operating expenses for computing gross margin and value-added output measures, respectively. These data limitations limit the level of industry detail and frequency for gross margin and value- added measures of retail output. They also offer limited data for estimating which support establishments in a firm (âauxiliariesâ) support its retail establishments and to what extent (Conclusion 3-7). Private-sector data could potentially provide more timely information about economic output (Conclusion 3-12). Nominal output must be adjusted by price changes to identify the real changes in the output of the sector. The price adjustment step is crucial, because price changes can accentuate or mask any real changes that are occurring, particularly during a period of rapid change when goods and services are evolving and are hard to compare over time. Conceptually, the key price adjustment that needs to take place for the retail sector itself relates to the services the sector provides, with respect to changes in the prices of those services and adjustment for changes in their quality. This differs from price adjustment related to the products the retailer sells, which focuses on the characteristics of the goods themselves. Price deflation in the retail sector needs to consider, for example, the shifts in services in moving from a traditional department store to a warehouse store to e-commerce, and these shifts involve changes related to such things as product variety and the process for identifying and obtaining goods (Conclusion 3-9). The federal statistical system collects two different types of price indices that can be used for deflation: the producer price index (PPI), which looks at changes in the prices of producer goods for a variety of inputs and at changes in margin prices for retail trade; and the consumer price index (CPI), which looks at changes in prices of consumer goods and is used to deflate sales revenue measures of output. Although the existing price indices provide a way of describing price changes that occur for the services and products provided by individual retail outlets, they do not capture the aggregate price changes that result as consumers move from one type of retail outlet to another. For example, the price indices do not reflect the change in the price and quality of retail services as consumers move from a traditional department store to a warehouse store to e-commerce, except when consumers move between outlets classified in different NAICS codes (Conclusion 3-8). Private-sector data could potentially be used to 1 The term âgross outputâ is used across sectors by the U.S. Bureau of Economic Analysis (BEA) to refer to a gross margin measure for retail and wholesale trade and a sales revenue measure for all other sectors. 3
Prepublication copy, uncorrected proofs estimate the price effect of consumers moving between retail outlets (Conclusion 3-11) and provide more timely estimates of price changes in general (Conclusion 3-12). Finally, employment is measured by estimating hours worked in the sector. The simple quantity of work hours should also be adjusted to reflect the different qualities of work provided by workers with different skill sets. In practice, this is done by looking at pay differences across groups of workers defined by difference in educational attainment, age, and gender. However, the retail transformation is substantially changing the workforce among some of the large retailers that are driving the biggest changes, with large increases in the number of workers with high-end programming and data analysis skills that support e-commerce (Conclusion 3-10). Private-sector data on payrolls could potentially be used to provide more timely estimates of quality-adjusted work hours (Conclusion 3-12). BLS currently develops measures of employment and labor productivity in retail that focus on the retail sector as specifically defined by NAICS, use a sectoral output measure of nominal output that is deflated by the CPI, and reflect hours worked in retail establishments that are not adjusted for labor quality. CONSIDERATIONS FOR CREATING A RETAIL SATELLITE ACCOUNT A satellite account provides a framework to explore a specific aspect of the economy that is linked to the System of National Accounts while deviating in ways that help address important questions about that aspect of the economy. These deviations may involve grouping or valuing economic activities in ways that differ from those that the national accounts use or providing more detailed statistics than are provided in the national accounts (Conclusion 4-1). There are several ways a retail satellite account might be defined to incorporate some of the related activities currently being integrated with retail services, such as wholesale, warehouse, and delivery functions. Including all establishments in these other industries would be feasible, but it would include many establishments with no relation to retail. Including only those establishments in these other industries that are part of retail enterprises would also be feasible, but that would exclude many relevant establishments simply because they are not owned by a retail enterprise. A âretail supportingâ scope for a satellite account could include all establishments in transportation, warehousing, wholesale trade, and business services that serve retail trade firms, in addition to retail trade establishments themselves (Conclusion 4-2). If a retail satellite accountâs scope is limited to only those retail-supporting establishments that are part of larger retail enterprises, it will miss aspects of the sectorâs transformation that are taking place between rather than within firms. The implementation of a retail-supporting satellite account would require estimating the portion of establishments in transportation, warehousing, wholesale trade, and business services that support retail (Conclusion 4-4). This split between retail-supporting and nonretail- supporting pieces would likely be different for the outputs of these sectors than for their labor input. It would be necessary to explore a variety of approaches for carrying out this split, including the use of alternative data sources. The input-output tables provide some information for estimating the split in output, but not the split in labor input. A collaborative effort across agencies could use microdata to explore issues related to a retail satellite account, including structural changes in firms and the role of auxiliary establishments (Conclusion 4-3). The definition of the broader retail sector for a satellite account could be developed initially by using several definitions that are each simple to implement and that together provide 4
Prepublication copy, uncorrected proofs lower and upper bounds for the included activities. A lower-bound definition could include all NAICS codes for retail establishments and for industries that are focused on supporting retail. An upper-bound definition could include all NAICS codes for industries that at least partially support retail. The range between these estimates would then indicate the potential benefit of developing careful approaches for splitting the input and output of industries that only partially support retail (Conclusion 4-5). Several existing satellite accounts developed by the Bureau of Economic Analysis (BEA) may provide useful models for developing a retail satellite account, given the measurement challenges posed by the retail transformation. The digital economy satellite account includes e- commerce and digital services, which are both important aspects of the retail transformation. The health care satellite account involves a reconceptualization of health care spending, which might suggest novel ways to reflect the changing cost structure of retail. The outdoor recreation satellite account addresses the challenge of dividing up statistics from several industries to combine some of them in a new grouping that is useful to the field. The small business satellite account addresses the challenge of identifying establishments of different sizes, which may also be an important way to divide the data for the retail sector (Conclusion 4-6). One approach to constructing a retail satellite account would be to create a central account with modules for experimentation and exploration. This would allow it to reflect the current consensus in the central account while identifying areas where new information and further research are needed for a consensus to emerge. The modules might address issues such as alternative output measures and deflators; alternative aggregations and classifications of retail- related industries or inputs; experimental price indices that might better reflect new retail services; integrated analyses of retail products that cross the boundary between goods and services or between physical and digital goods; and alternative ways of measuring and allocating productivity gains. RECOMMENDATIONS FOR A RETAIL SATELLITE ACCOUNT The panel endorses the creation of a satellite account to study the transformation in retail trade. Such an account would be an appropriate and useful vehicle for BLS to use to study the impact on employment and productivity of the transformation in retail trade and to develop exploratory measures that describe that transformation (Conclusion 5-1). Given the distribution of data and expertise across government agencies, BLS should develop a satellite account for an expanded retail trade sector in collaboration with the BEA and the Census Bureau. Such a team could be formed under the Evidence Based Policy Act to facilitate administrative and collaborative efforts (Recommendation 5-1). The team developing the retail satellite account should solicit input and advice from industry and academia, with a special focus on collaboration with industry (Conclusion 3-13). Government statistics need input to ensure that the concepts being measured are relevant and keep up with the rapid pace of change in industry (Recommendation 2). In implementing a satellite account, BLS and its partners should adopt an iterative and modular approach, starting with feasible options that draw upon the BEA industry account and the BLS-BEA integrated industry-level production account to see what insights these might provide about the sector and about feasible fixes. The modular approach should include a set of estimates in a central module, with a set of submodules to investigate important side questions or 5
Prepublication copy, uncorrected proofs alternative measures and a set of studies to carry out over time to investigate relevant questions (Recommendation 3). The satellite account should cover all retail and retail-supporting establishments, identifying these by combining available information from existing and enhanced data. The retail-supporting establishments should encompass all establishments supporting the distribution of retail goods to the consumer, but excluding their manufacturing and importing (Recommendation 4). The satellite account should examine multiple measures of output, price deflators, and labor input in order to support comparisons that lead to informed decisions. Output measures should include gross sales and gross margins for trade industries, gross sales/revenues for other industries, and value-added for all industries. Deflators should include current margin deflators and new options that capture the changing characteristics of retail trade. Labor input measures should include simple hours worked and quality-adjusted hours worked to capture the changes in workforce quality. Modules should also be used to evaluate alternative approaches to estimating the split between retail-related and nonretail-related for both output and input (Recommendation 5). The modules could also address more specialized issues that contribute to understanding the transition in retail trade, such as (1) international trade and global value chains; (2) digitization; (3) labor quality; and (4) providing real-time and subsector analyses. Over time, the central module would incorporate improvements developed in the submodules and in new data collection (Recommendation 6). Given the errors introduced by the separate business registers used by BLS and Census, measures should be taken immediately to facilitate the reconciliation of business lists across agencies. This will require changes to be enacted by Congress or implemented by the Treasury Department to modify the relevant IRS regulations (Recommendation 7). BLS and Census should establish an interagency task force, potentially including other relevant agencies, to develop a plan for implementing a consolidated business register to use as the sample frame for all business surveys (Recommendation 8). Developing a retail-related satellite account will require considerable effort to acquire and use data and to address data gaps in existing data. Individual projects include: Filling data gaps in the Economic Census and Economic Surveys that relate to the calculation of gross margins, value added, and the contribution of auxiliaries; identifying data to estimate the split in hours worked between retail-related and nonretail-related for retail-related service industries; correcting for differences in numerator and denominator of productivity caused by the use of different business registers and classifications; and exploring the use of private-sector data to improve the timeliness and detail provided in the account. Some of these efforts are best accomplished by a team with access to the Census Bureauâs economic microdata (Recommendation 9). 6