A New Categorization of the U.S. Economy: The Role of Supply Chain Industries in Performance* Mercedes Delgado, MIT Sloan Karen G. Mills, Harvard Business School This version: 05/23/2016 PRELIMINARY AND INCOMPLETE Abstract Supply chains have been an important part of the discussion of the American economy. However, this discussion has lacked an empirical definition of who are the suppliers and this has limited our understanding of their role in national performance. This paper introduces a new industry categorization that separates Supply Chain industries (i.e., those that sell their goods and services primarily to other businesses or governments) from Business-to-Consumer (B2C) industries (i.e., those that sell primarily to consumers). Our analysis uses the 2002 Benchmark Input-Output Accounts from the U.S. Bureau of Economic Analysis to identify Supply Chain and B2C industries. Using this categorization we examine the supply chain economy during the 1998–2013 period across several metrics: the number of firms, employment, wages, labor occupation composition, patenting, and growth dynamics. We find that supply chain industries comprise a large segment of the economy. In particular, there are many suppliers of traded services both in terms of employment and number of firms. Supply chain industries, especially traded services, have higher average wages than other industry segments. This can be explained in part by the larger relative presence of STEM (Science, Technology, Engineering and Math) occupations in supply chain industries, and in particular in the suppliers of traded services (though non-STEM occupations such as accounting, finance, managerial, and logistics are also important to service suppliers). While STEM occupations are most prevalent in suppliers of traded services, patents are primarily concentrated in manufacturing suppliers. Together, these observations suggest that much of the U.S. innovative activity happens in the supply chain economy. We also find that the employment in the supply chain economy has been evolving away from manufacturing and towards services for the period under examination (1998-2013), with suppliers of traded services experiencing high growth in employment and wages. Finally, the analysis of the growth trends during the business cycles reveals that the supply chain economy is particularly susceptible to economic crises (2001-2002; 2007-2009). Overall, our findings call for targeted policies that recognize that suppliers are a large segment of the U.S. economy, are a mix of manufacturers and service providers, have a diverse and distinct set of labor occupations, drive innovation, and are vulnerable to crises.

*

Scott Stern has contributed important insights. We also thank Alfonso Gambardella, Jorge Guzmán, Bill Kerr, and Ramana Nanda for helpful comments. We thank Christopher Rudnicki for great assistance with the data analysis and project stewardship. Author contact information: Mercedes Delgado (MIT Sloan, ISC; [email protected]; corresponding author) and Karen G. Mills (Harvard Business School; [email protected]).

1

1. Introduction

An ongoing and long academic and policy debate has focused on the role of the manufacturing capacity of a country on its economic and innovative performance (see e.g., Rosemberg, 1963; Dertouzos et al., 1989; Pisano, 1997; Pisano and Shih, 2012; Helper et al., 2012; Fuchs and Kirchain, 2010; Berger, 2013). This question has become even more relevant as the U.S. economy has shown a large decline in manufacturing employment in the last decades, in part due to increased import competition (Acemoglu et al., 2015). There is a concern that by reducing the manufacturing of goods in a country the subsequent innovation capacity may be reduced because there are external economies from the manufacturing process that improve the ability to innovate and the efficiency (speed, costs, and diffusion) of the innovation process. While countries can import innovative inputs, they cannot buy the associated externalities related to producing the inputs locally. In particular, Rosemberg (1963) highlights the crucial role of the suppliers of capital goods (“machine producers”) in the process of technological innovation. A country needs a large capital goods sector (supported by a high level of demand for capital goods) in order for suppliers of capital goods to specialize in the creation of tailored inputs for their buyers. This specialization of suppliers creates crucial externalities: “there is an important learning process involved in machine production, and a high degree of specialization is conducive to not only an effective learning process but to an effective application of what is learnt.” These learning externalities can improve the efficiency in the production of the new capital goods needed for the innovation process. While most prior work focuses on a narrow view of suppliers as manufacturers, an additional perspective posits that in today’s economy there is a broader group of suppliers that includes producers of both intermediate goods (e.g., microprocessors) and services (e.g., software). In other words, the production capacity of a country is broader than its manufacturing capacity of final goods, and includes both manufacturing and service suppliers. These suppliers are often clustered nearby their buyers, and generate externalities that contribute to the innovation capacity of a country and its regions (Marshall 1920; Porter, 1990 1998; Audretsch and Feldman, 1996; Delgado, Porter, and Stern, 2014; Delgado, 2016). 2

The goal of this paper is to quantify who are the suppliers in today’s economy and examine their role in national performance. While there is a large literature that focuses on examining the supply chain that operates from business-to-business (see e.g., Pisano, 1997; Pisano and Shih, 2012; Lessard, 2013; Gawer and Casumano, 2002), there is a lack of quantification of the size and types of suppliers in the economy. A few of these suppliers are very large and become the platform on which many other companies build their products (e.g., Intel in microprocessors and Microsoft in software; Gawer and Casumano, 2002), but many of the suppliers are unknown, small firms. To understand the role of suppliers in the U.S. economy, this paper develops a new industry categorization that identifies “supply chain” industries (i.e., those that provide inputs for businesses and government), and explores their role in the composition of nonfarm private employment and in national growth dynamics. The categorization of industries as defined by the U.S. Census Bureau’s North American Industry Classification System (NAICS) allows for descriptive insights into the economy’s composition and a greater understanding of the role of particular types of industries in economic performance. Throughout the 20th century the most influential categorization of U.S. industries was manufacturing versus services. However, manufacturing currently comprises around 10% of U.S. employment (Figure 1), and the service industries are extremely heterogeneous, varying from retail and restaurants to engineering and design. The economy’s evolution away from manufacturing then calls into question the usefulness of this framework for examining today’s dynamics of firms and employment. The traded versus local categorization introduced by Porter (2003) is rooted in the economies of agglomeration literature (Marshall, 1920; Porter, 1990). It separates traded economic activity, which is geographically clustered and sold across regions and countries (e.g., financial services and automotive manufacturing), from local economic activity (e.g., restaurants and retail). This framework has helped researchers shed new light on the role of the traded economy in national and regional competitiveness, and in particular on the role of industrial clusters in regional performance (see e.g., Delgado, Porter, and Stern, 2010, 2014). This paper introduces a new and complementary categorization that separates supply chain (SC) industries (i.e. those that sell their goods and services primarily to other businesses or 3

governments) from business-to-consumer (B2C) industries (i.e. those that sell primarily to personal consumers). Our analysis uses the 2002 Benchmark Input-Output (I-O) Accounts from the U.S. Bureau of Economic Analysis to systematically identify SC and B2C industries. This categorization allows for a better understanding of the composition of the economy and the types of industries that drive national growth outcomes, and complements prior categorizations. We find that supply chain industries compose a large and important segment of the economy (Figure 1). They accounted for 43 million jobs (37% of U.S. employment) and 2.5 million firms in 2012 (43% of all employer firms). These conservative estimates are the first comprehensive attempt to measure the size of the supply chain economy. We also separate SC and B2C industries into subcategories-- traded versus local and manufacturing versus services (Figure 2). By combining these three complementary categorizations, we can define and analyze important sub-segments of the economy. For example, there are many suppliers of traded services both in terms of employment and number of firms – a finding at odds with most prior work that focuses on a narrow view of suppliers as manufacturers. Suppliers of traded services account for almost 18 million jobs (~15% of all U.S. private employment), while suppliers of traded goods account for around 8 million jobs. There are four times as many traded supplier firms in services than in manufacturing (756,000 versus 169,000 firms). SC industries, especially those that are classified as traded services, have higher average wages than B2C industries. The examination of the labor occupational composition reveals that STEM (Science, Technology, Engineering and Math) occupations are more prevalent in SC industries than in B2C industries. Surprisingly, they are more important for SC-Traded categories in services than in manufacturing (17% versus 11% of employment is in STEM occupations, respectively). We also find that labor occupations differ between SC and B2C industries in ways besides the prevalence of STEM occupations. For example, logistics, managerial, finance, and accounting occupations are particularly important for suppliers of traded services. Our analysis suggests that patenting activity (a measure of innovation) is highly concentrated in the Supply Chain economy (more than 80% of the US utility patents granted in 2013), and in particular in manufacturing suppliers (77% of the patents). While it is not surprising

4

that patenting is concentrated in manufacturing, it is insightful that most patenting is taking place in the suppliers of intermediate goods versus in the producers of consumer goods. Note that there is a big difference between the STEM content and patenting of the different segments of the Supply Chain economy. While STEM occupations are most prevalent in suppliers of traded services, this sub-segment only accounts for 2% of all the patents. This suggests that the innovativeness of the suppliers of traded services is much higher than predicted based on patenting. We need new measures of innovation for the suppliers of services that take into account their STEM content and their input-output links with high patenting industries. Finally, we examine the growth trends of the supply chain economy at the national level. We find that the employment composition of the supply chain economy has been evolving away from manufacturing and towards services for the period under examination (1998–2013), with suppliers of traded services experiencing high growth in both employment and wages. This compositional change reflects the evolution that some innovative firms have experienced over the past few decades (e.g., IBM), and is consistent with the increasing importance of some service industries, like enterprise software and business services (Gawer and Casumano, 2002; Low, 2013). The growth analysis during the 1998-2013 business cycles also reveals that the Supply Chain economy is very cyclical. It experiences larger contractions in employment and wage growth than the B2C economy in both crises periods (2001-2002 and 2007-2009). The vulnerability of suppliers calls for targeted policies during crises. Our new categorization also helps better understand the performance of the local economy, the largest part of the economy in terms of employment (more than 60% of all jobs). We break out the local economy into Healthcare and “Main Street.”

We then use our

categorization to divide Main Street into two segments: B2C-Main Street represents the traditional consumer facing shops and services (e.g., restaurants, dry cleaners, and car repair), while SC-Main Street represents the business-to-business local firms (e.g., commercial real estate and landscaping services). The B2C-Main Street segment has the lowest average wages of any industry segment. It grew faster in terms of employment than the total economy during 19982013, but this category was the only one that experienced a decline in real wages. In contrast, 5

SC-Main Street experienced low employment growth, but a positive wage growth during the period. Healthcare registered high employment and wage growth, and has been a significant growth factor in the economy. Overall, our findings suggests that well paid jobs have been created in the supply chain economy, especially in traded services. SC traded services is the segment of the traded economy experiencing the highest growth in employment and wages. Our findings call for targeted policies that recognize that suppliers are a large part of the economy, are not just manufacturers, encompass both STEM and other set of labor occupations, drive innovation, and are cyclical. The remainder of the paper is organized as follows: Section 2 describes the prior literature on the role of suppliers in firm, regional, and country competitiveness. Section 3 presents our empirical method to define SC and B2C industries. Section 4 explains the composition of the supply chain economy in terms of employment, number of firms, wages, and labor occupations. The employment, wage, and patenting growth trends of the SC and B2C industry categories is examined in Section 5. A final section concludes and offers policy implications and remarks regarding the applicability of our categorization for future research. 2. Is Prior Work Underestimating the Importance of Suppliers? Suppliers provide inputs for businesses and governments and have played a central role in the economic and strategy literature. In the economic growth literature, the production of intermediate goods (capital goods) has been linked to the innovation capacity of a country (see e.g., Rosemberg, 1963; Dertouzos et al., 1989). There are learning externalities from producing specialized intermediate goods that improve the efficiency of the innovation process. In economic geography, the co-location of suppliers and buyers has been identified as a driver of agglomeration economies (Marshall, 1920; Chinitz, 1961; Porter 1990, 1998; Audretsch and Feldman, 1996; Feldman and Audretsch, 1999; Glaeser and Kerr, 2009; Delgado, Porter, and Stern, 2014, 2016; Feldman et al., 2014). When suppliers are clustered together and located near their buyers, they can create agglomeration benefits through shared pools of skills, technologies, knowledge, and specialized inputs. Prior work highlights the importance of nearby suppliers (often measured by the presence of upstream industries) for fostering innovation and entrepreneurship in manufacturing industries (Glaeser and Kerr, 2009). The collaboration 6

between nearby buyers and suppliers in manufacturing has been associated with greater innovation (Helper, MacDuffie, and Sabel, 2000). A recent study finds that firm innovativeness (i.e., the development of products new to the market) is facilitated especially by the collaboration with suppliers and to a much lesser extent by the collaboration with end-users (Arora, Cohen, and Walsh, 2014). This suggests that, to the extent that this collaboration is facilitated with geographical proximity, the presence of suppliers can play a crucial role on the innovation capacity of a region. Suppliers also provide important inputs to the government. The U.S. government has a long tradition of contracting with businesses in many sectors (e.g., defense, energy, utilities, healthcare, and research labs).1 Significant policy initiatives have focused on helping small suppliers, including efforts led by the Small Business Administration to meet the small businesses contracting goals of 23 percent; and the Obama Administration’s QuickPay initiative in 2011 aimed to reduce the working capital costs of small suppliers to the government (Helper, Nicholson, and Noonan, 2015). In the strategy literature, decisions regarding value-chain design and internal versus external sourcing of inputs are the cornerstone of firm strategy and performance (Porter, 1996; Pisano, 1997; Cohen and Levinthal, 1990; Helper, MacDuffie, and Sabel, 2000; Alcacer and Delgado, 2016). These choices will depend on the availability of nearby suppliers. International business studies have paid close attention to the organization of global supply chains, and managerial practices to reduce the coordination costs and risk inherent in outsourcing inputs from multiple locations (Lessard, 2013). Furthermore, the organization of a firm’s value chain will also depend on whether the firm is business-to-business (i.e., a supplier) or business-to-consumer (B2C). For example, prior studies find that the adoption of some operational practices, like electronic selling (McElheran, 2015), can vary for B2B and B2C firms. Overall, the prior literature suggests that suppliers are important for the competitiveness of firms, regions, and countries, but who are the suppliers? Most prior work has focused on manufacturing suppliers. This does not capture the increasingly important role of service inputs 1

Government contracts in 2014 totaled over $450 billion with approximately $90 billion going to smaller suppliers.

7

(e.g., business services, accounting services, financial services, engineering, software publishers, and design) in the value of final products (Ali-Yrkkö et al., 2011). Some of these service suppliers, like Microsoft in software, become the platform on which many other companies build their products (Gawer and Casumano, 2002). To identify and characterize the suppliers in the U.S. economy, this paper distinguishes between supply chain (SC) industries that primarily sell goods and services to businesses or to the government and business-to-consumer (B2C) industries that focus on selling to consumers. In the next Section we explain how we identify SC and B2C industries. Using this new categorization, we then examine the importance of the supply chain economy across several metrics: the number of firms, employment, wages, labor occupation composition, patenting, and their growth dynamics. Prior studies focus on the capacity to manufacture final goods as a key driver of the innovativeness and economic performance of a country (e.g., Dertouzos et al., 1989; Pisano, 1997; Berger, 2013). Our characterization of the supply chain economy suggests that we should focus on a broader definition of the production capacity of a country that includes both manufacturing and service suppliers. As we explain below, the majority of traded suppliers are in services industries. This segment of the supply chain economy has the highest wages, a very large presence of STEM occupations, and has experienced a high growth.

3. The Supply Chain Categorization of the U.S. Economy In this section we first explain prior categorizations: Manufacturing versus Services, and Traded versus Local. We then describe the method to define SC industries and B2C industries. A detailed explanation of the methodology is offered in the Appendix. 3.1 Prior Categorizations of the U.S. Economy The manufacturing versus services categorization of industries has been broadly used in economics and policy since the creation of the Standard Industrial Classification (SIC) by the U.S. Government in 1937. The current NAICS code industry classification (as well as the prior SIC code) separates out manufacturing and services industries based on what is produced. The 2-digit NAICS codes 31-to-33 correspond to manufacturing industries, and the rest correspond to 8

services industries. There are 1,088 6-digit NAICS (2007 definition) industries (excluding farming and some government activity) in the US Census Bureau’s County Business Pattern data (CBP). These industries are classified into 472 manufacturing and 616 service industries. In 2012, manufacturing industries accounted for less than 10% of total U.S. employment (Figure 1). A related categorization used in the I-O accounts separates goods-producing industries from service-producing industries. Goods-producing industries includes both manufacturing goods and other goods in the following sectors: Agriculture, Forestry, Fishing, and Hunting (NAICS code 11); Mining (NAICS code 21); and Construction (NAICs code 23). Goods-producing industries represented 15% of the total employment in 2012. While the “Services” category that we use includes all industries not classified as manufacturing, our sensitivity analysis considers an alternative definition for the Service category that excludes all durable goods (what we refer to as the “Services (no goods)” category). The traded versus local categorization developed by Porter (2003) represents a more recent classification of industries based on their patterns of spatial competition (selling across regions versus within regions). Specifically, traded industries are geographically concentrated and sell their goods and services across regions and countries. Local industries sell primarily in the local market, and they are geographically dispersed (i.e., their employment in a region is proportional to the size of the population). We use the most recent traded categorization defined in Delgado, et al. (2016b). Building on Porter (2003), they identify traded industries using the employment specialization and concentration patterns of each industry across U.S .regions. Using the 1,088 6-digit NAICS-2007 industries in the CBP data, they identify 778 traded industries and 310 local industries. In 2012, traded industries account for 36% of total U.S. employment (Figure 1). Throughout our analysis, we separate the local economy into industries related to Healthcare and those industries related to the ‘Main Street’ economy. This distinction allows us to shed some light on the dynamics of the local economy. Healthcare includes 35 industries.2 The other 275 local industries are defined as Main Street industries.

2

Healthcare industries are those within the local Health Services cluster identified in the US Cluster Mapping Project (USCMP). It includes local health care establishments and services such as hospitals, medical laboratories, home and

9

3.2 New Categorization: Supply Chain and Business-to-Consumer Industries To identify and characterize the suppliers in the U.S. economy, we distinguish between SC and B2C industries. Conceptually, SC industries primarily sell goods and services to businesses and the government. Thus, most of the firms operating in these industries will be suppliers (i.e., they sell mainly to other businesses or to the government). In contrast, B2C industries primarily sell final goods and services for personal consumption. Note that the concept of SC industry is different from upstream industry. The latter is a relative term that refers to an industry that provides inputs to another industry.3 To measure the extent to which industries sell to personal consumption, we use the 2002 U.S. Benchmark Input-Output (I-O) Accounts of the Bureau of Economic Analysis (BEA). The I-O Accounts allow researchers to capture input and output flows between industries and output flows from each industry into final use for personal consumption (see e.g., Feser, 2005; Feldman et al., 2014; McElheran, 2015; Delgado et al., 2016a). To our knowledge, we offer the most systematic and comprehensive classification of industries into SC and B2C. Building on prior work, we classify all 6-digit NAICS industries into SC or B2C based on the output sold to Personal Consumption Expenditure (PCE). PCE is a final use item in the I-O Accounts that captures the value of the goods and services that are directly purchased by households, such as food, cars, and college education (see detailed explanation in the Appendix). We identify industries as SC if they sell less than one-third of their output to PCE, and the rest are classified as B2C. Thus in our definition those industries that sell most of their goods and services to other businesses or the government (i.e., more than two-thirds sold outside PCE) will be classified as SC. We explored additional cut-offs to define SC industries (PCE scores