New Perspectives on the Decline of U. S New Perspectives on the Decline of U.S. Manufacturing Employment June 2018 Justin R. Pierce, Board of Governors of the Federal Reserve System Teresa C. Fort, Tuck School at Dartmouth & NBER Peter K. Schott, Yale School of Management & NBER

Disclaimer Any opinions and conclusions expressed herein are those of the authors and do not necessarily reflect the views of the U.S. Census Bureau, the Board of Governors or its research staff. All results have been reviewed to ensure that no confidential information has been disclosed.

US Manufacturing Employment is Declining -12% -17% -18%

Manufacturing Employment vs Real Value Added

Explanations? Import competition, especially from China in 2000s Bernard, Jensen and Schott (2006), Autor, Dorn and Hanson (2013), Acemoglu et al. (2016), Pierce and Schott (2016), Autor et al. (2016), Caliendo et al. (2017) Foreign sourcing and offshoring Harrison and McMillan (2011), Antras, Fort and Tintelot (2017), Boehme, Flaaen and Pandalai-Nayer (2017), Kovak, Oldenski and Sly (2018) Technology adoption and automation Autor, Levy and Murname (2003), Autor and Dorn (2013), Acemoglu and Restrepo (2017), Graetz and Michaels (2017)

Can We Distinguish Trade vs Technology? Some papers attempt to disentangle these two forces Goos, Manning, Solomans (2014): technology g polarization ADH (2015): trade matters most for employment loss after 2000 But technology facilitates trade and product fragmentation Bloom et al (2016), Fort (2017), Autor et al. (2017), Steinwender (2018), Juhasz and Steinwender (2018) And, trade can induce or reduce technology and R&D investment Bernard et al. (2006), Khandelwal (2013), Boler et al. (2015), Bloom et al. (2016), Bernard et al. (2018), Autor et al. (2017)

Disentangling Trade/Technology is Very Difficult Wall Street Journal 2017.3.16 “When Drew Greenblatt bought … a small Baltimore maker of wire baskets for bagel shops, he knew nothing about robotics. That was 1998, and workers made products manually using 1950s equipment….

Disentangling Trade/Technology is Very Difficult Wall Street Journal 2017.3.16 “When Drew Greenblatt bought … a small Baltimore maker of wire baskets for bagel shops, he knew nothing about robotics. That was 1998, and workers made products manually using 1950s equipment…. Pushed near insolvency by Chinese competition in 2001, he started investing in automation. Since then, Marlin has spent $5.5 million on modern equipment. Its revenue, staff and wages have surged and it now exports to China and Mexico.” Were the changes at Marlin driven by trade or technology? What about changes at Marlin’s US competitors? Did Marlin import its robots?

Industry Import Penetration Import penetration starts rising in the 1980s Chinese import penetration accelerates after 2000 Import penetration from China Industry import penetration

Firm-Level Importing Import penetration starts rising in the 1980s Chinese import penetration accelerates after 2000 Firms that import Firms importing China

Firm-Level Technology Large increase in share of plants using computers in the early 2000s Use of electronic networks to control and coordinate shipments also rising during that period Purchasing computers Using electronic networks

Trade and Technology Move Together! Over our sample period, firms’ use of trade and technology both rise! Hard to disentangle Goal in this paper is to highlight new dimensions of employment loss to gain deeper insights

We Examine Three Margins of Adjustment Industry-level Few clear examples of “sunrise” and “sunset” industries Trade shocks affect outputs AND inputs Firm-level 75% of decline occurs in continuing firms! NM growth offsets M declines before 2000! Region-level Pre-2000, manufacturing employment moves west and south After 2000, it falls everywhere

Outline Industry margins of adjustment Firm margins of adjustment Regional margins of adjustment Conclusion

Industry Margins of Adjustment

Sunset, Sunrise Industries Exist, But are Scarce

Sunset, Sunrise Industries Exist, But are Scarce Leather, Apparel: employment and output both fall In most sectors, output rises as employment falls

Computers Dominate Real Value Added (RVA) Growth

Computers Dominate Real Value Added (RVA) Growth After 2000, more sectors with employment decline and fewer with with output growth; computers remains outlier

Import Penetration is Everywhere (Not Just Apparel!) Computers/Electronics has highest growth in import penetration Suggests differential impact of import penetration on Outputs vs Inputs?

Substantial Import Penetration Within Computers Semiconductors accounts for 71% percent of real value added in manufacturing over this period Computers accounts for 11% percent Both industries make use of offshoring and factory-less goods production

Firm Margins of Adjustment

Data Longitudinal Business Database,1977-2012 All private, employer, non-farm establishments Consistent Firm & establishment identifiers Consistent NAICS codes from Fort and Klimek (2016) Census of Manufactures, 1977(5)2012 All manufacturing establishments Observe trade/technology adoption measures displayed earlier Longitudinal Foreign Trade Transaction Database, 1992-2012 Transaction-level trade data at firm level

Definitions Employment Classified according to establishments All workers in a manufacturing establishment are in manufacturing We define manufacturing firm as a firm that ever has a manufacturing plant between 1977-2012 Broad definition Big firms often have both M and NM establishments Firm birth/death Follow Haltiwanger et al. (2013) Firm is a “birth” if all establishments are new Firms is a “death” if all establishments exit (forever)

Decompositions of Employment Across Firm Margins We decompose the overall change in US manufacturing employment along firm net margins of adjustment Net margins are relative to 1977 Net firm birth since 1977 Continuing firms’ estabs’ birth/death since 1977 Continuing firms’ continuing estabs net growth since 1977 DEt = (EtFirm Birth - EtFirm Death) + (EtContinuing Firm Estab Birth - EtContinuing Firm Estab Death) + (EtContinuing Firm Estab Growth - EtContinuing Firm Estab Shrink)

Decompositions of Employment Across Firm Margins We decompose the overall change in US manufacturing employment along firm net margins of adjustment Net margins are relative to 1977 Net firm birth since 1977 Continuing firms’ estabs’ birth/death since 1977 Continuing firms’ continuing estabs net growth since 1977 DEt = (EtFirm Birth - EtFirm Death) + (EtContinuing Firm Estab Birth- EtContinuing Firm Estab Death) + (EtContinuing Firm Estab Growth - EtContinuing Firm Estab Shrink)

Decompositions of Employment Across Firm Margins We decompose the overall change in US manufacturing employment along firm net margins of adjustment Net margins are relative to 1977 Net firm birth since 1977 Continuing firms’ estabs’ birth/death since 1977 Continuing firms’ continuing estabs net growth since 1977 DEt = (EtFirm Birth - EtFirm Death) + (EtContinuing Firm Estab Birth - EtContinuing Firm Estab Death) + (EtContinuing Firm Estab Growth - EtContinuing Firm Estab Shrink)

Decline is Not Even Across Margins The aggregate decline is 6.7 million between 1977-2012 We can decompose this aggregate decline into three net margins

Decline is Not Even Across Margins Continuing firm-plants account for 12% of the aggregate decline

Decline is Not Even Across Margins Continuing firm-plants account for 12% of the aggregate decline Net firm birth accounts for another 25%

Decline is Not Even Across Margins Continuing firm-plants account for 12% of the aggregate decline Net firm birth accounts for another 25% Net plant birth within continuing firms accounts for the remaining 63%!

Decline is Not Even Across Margins After 2000, net firm birth/death relatively more important Consistent with change in US trade policy towards China (Pierce and Schott 2016; Autor et al. 2013)

Decline is Not Even Across Margins Overall, 75% of the decline takes place within firms! Raises several questions Why not more reallocation due tio firm death? What is it about these legacy firms that allows them to survive?

Is Adoption of Trade, Technology Easier for Incumbents? Estimate size and productivity advantages by year ln(Attributeft) = a + btActivityft + hjt + eft Attributes Employment, productivity Activities Importing, purchasing computers, importing robots, usingelectronic networks Estimate separately by year

Technology and Trading Premia Firms using computers are much larger than firms that don’t before 1997 After that, computers are become ubiquitous Log size differential of computer purchaser in 1977

Technology and Trading Premia Firms that import directly are also larger than firms that don’t In contrast to computer use, their premia are relatively flat over time Log size differential of firm importing directly in 1997

Technology and Trading Premia Firms that import industrial robots directly have the highest premia Robots Electronic Network

Plant Closure and Trade versus Technology Adoption Estimate probability that plant i in firm f and industry j exits over next 5 years Pr(Deathijft:t+5=1|Xijft) = a + bActivityijft + gln(Eijft) + hf + dt +eijft:t+5 Activities Purchasing computers, using electronic networks, concurrent changes in industry import penetration Estimate separately for pre and post 2000 Control for plant size

Extensive Margins Relate to Trade and Technology

Extensive Margins Relate to Trade and Technology

Extensive Margins Relate to Trade and Technology

Extensive Margins Relate to Trade and Technology

Eight Gross Margins of Adjustment Decompose each margin in previous figure into Gross job creation (e.g., expanding continuing firm-plant) Gross job destruction (e.g., shrinking continuing firm-plant) Add additional margin Establishments that switch into or out of manufacturing (By definition this has to be along the intensive margin) Here, reset the margins each decade I.e., 1977, 1990, 2000

Decline is Not Even Across Margins Small contribution of intensive margin masks considerable gross churn along that margin Source: LBD.

Job creation and destruction within continuing firm-plants dominates the other margins

Job creation and destruction within continuing firm-plants dominates the other margins Over time, gross job creation margins contribute less

Job creation and destruction within continuing firm-plants dominates the other margins Over time, gross job creation margins contribute less Especially after 2000 Within firms, establishments are shutting down, shrinking

Manufacturing Plants and Employment/Plant Can also see this trend in data showing overall number of US manufacturing establishments, and their average size, over time

Geographic Margins of Adjustment

We Examine Census Regions

Manufacturing Firm Employment Across US Regions NE and MA decline throughout the 1977-2012 sample period Some regions grow in the 1990s Domestic “offshoring” from north and east to south and west? All regions decline in the 2000s 1977-1989 1990-2000 2000-2012

Manufacturing Employment Margins Vary Across Regions In Northeast and Mid-Atlantic, the majority of employment loss is from death Mountain and Pacific have net employment gains from firm births

Manufacturing Firms’ Non-Manufacturing Employment

Manufacturing Firms’ Have Non-Manufacturing Employment Manufacturing firms manufacturing employment falls from 17.8 to 11.1 million

Manufacturing Firms’ Non-Manufacturing Employment Manufacturing firms non-manufacturing employment rises from 13.0 to 23.6 million Manufacturing firms manufacturing employment falls from 17.8 to 11.1 million

Manufacturing Firms’ Non-Manufacturing Employment Manufacturing firms total employment rises from 30.7 to 34.7 million Total Manufacturing firms non-manufacturing employment rises from 13.0 to 23.6 million Non-manufacturing Manufacturing Manufacturing firms manufacturing employment falls from 17.8 to 11.1 million

NM Employment at M vs NM Firms Manufacturing Firms create NM employment via new ESTABLISHMENTS Non-Manufacturing Firms create NM employment via new FIRMS At M firms, 80% of NM growth is within firm, via net establishment birth At NM firms, NM growth is driven by new firms

What Kind of NM Workers are M Firms Adding? Retail NAICS 44-45 Professional NAICS 51: IT 52-3: finance, insurance 54: engineering, technical 55: headquarters services 56: admin support, waste Other Services NAICS 11: agriculture 21: mining 22: utilities 23: construction 42: wholesale 48-9: transportation 61: educationer 62: healthcare 71: arts, entertainment 72: accommodation, food 81,92: other, public admin Other Professional Retail In contrast to manufacturing firms’ M employment, their NM employment recovers after 2000

Where are M Firms’ Adding NM Employment? Growth across almost all regions in the 1980s and 1990s More variation after 2000

Conclusion

Conclusions/Lingering Questions Manufacturing employment is declining; real output is not Most of decline is within firms But these firms are adding non-manufacturing activities Trade and technology are inter-related, hard to separate Adopters are bigger and more productive Adoption premia seem to decline over time, especially for tech Tech and trade relate to establishment survival Need further exploration of heterogeneity in response to trade and technology shocks Which firms take advantage, which are only hurt?

Thanks!