Understanding the Decline in Manufacturing Emplyment Susan Huseman Upjhn Institute fr Emplyment Research Presentatin prepared fr Manufacturing Frum, Ecnmic Plicy Institute, Washingtn, DC, January 12, 2018
Precipitus and unprecedented decline in U.S. manufacturing emplyment in 2000s Manufacturing Emplyment and Number f Establishments Emplyment, 1,000s 20,000 18,000 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 1947 1951 1955 1959 1963 1967 1971 1975 1979 1983 1987 1991 1995 1999 2003 2007 2011 2015 400,000 350,000 300,000 250,000 200,000 150,000 100,000 50,000 0 # establishments Emplyment Establishments Sparked debate ver causes f the decline: trade v. autmatin. In this talk: Discuss misunderstanding & misrepresentatin f mfg data Review research evidence n trade v. autmatin debate
Decline in manufacturing utput & emplyment shares especially prevalent since 1970s, track each ther. 250 But, manufacturing real 200 utput grwth kept pace 150 100 with aggregate utput 50 grwth in ecnmy 0 300 1947 1950 Real GDP: Private Industry and Manufacturing, 1977=100 1953 1956 1959 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 Private industries Manufacturing
Real GDP: Manfacturing and Private Industry; Manufacturing GDP and Emplyment Share 140 40 120 35 100 80 60 40 20 30 25 20 15 10 5 0 0 1948 1951 1954 1957 1960 1963 Index, 2009=100 Share, % 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 Private industries, real GDP Mfg GDP share Manufacturing, real GDP Mfg emplyment share Hw t recncile tw pictures Prevailing view: Prices f dmestic manufactured gds falling relative t dmestic services prices Prductivity grwth much higher in manufacturing Cnsumer demand fr manufactured prducts limited
Prevailing view: Rapid prductivity grwth explains the decline in manufacturing emplyment share Labr prductivity defined as utput (value added) per unit labr input (wrkers r hurs). It fllws that: e m e p = (y m y p ) (prd m prd p ) Where e, y, and prd are emplyment, utput, and labr prductivity grwth rates fr manufacturing and the aggregate private sectr. If the utput grwth rates in manufacturing and the private sectr are the same, then prductivity accunts fr ALL f the decline in manufacturing s emplyment share. Cnclusin: Nt a causal relatinship, but descriptive evidence cmpelling Prductivity, in the frm f autmatin, caused the relative (and abslute) decline in manufacturing emplyment The decline was inevitable
Cnclusin ften repeated as fact in mainstream media Frm an ecnmic perspective... there can be n revival f American manufacturing, because there has been n cllapse. Because f autmatin, there are far fewer jbs in factries. But the value f stuff made in America reached a recrd high in the first quarter f 2016, even after adjusting fr inflatin. Binyamin Appelbaum, New Yrk Times, 2016
Alternate (crrect) recnciliatin: Measurement issue Manufacturing s falling relative prices, rbust utput grwth, and extrardinary prductivity grwth driven by ne industry: cmputer and electrnics prducts Cmputer industry accunts fr less than 15% f manufacturing GDP Outsized effect reflects statistical adjustment f cmputers and semicnductr price deflatrs fr imprvements in prduct quality Extrardinary prductivity grwth in cmputer industry by extensin in manufacturing reflects prduct imprvements, nt autmatin Skews statistics, gives misleading impressin abut sectr s health
Manufacturing prices nt falling relative t ther industry prices utside cmputer industry 200 150 100 50 0 1947 1950 Price deflatrs, private industries and manufacturing, 1987=100 1953 1956 1959 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013 Price deflatrs fr cmputer industry rapidly declining after 1960s, reflects quality adjustment fr prduct imprvements Private ind, as published Excluding cmputer industry, manufacturing prices are nt falling relative t private sectr prices. Mfg, as published Index, 1987=100 250 200 150 100 50 Price Deflatr Indices, Private Industry and Manufacturing Omitting Cmputer Industry, Cmputer Industry 0 1947 1950 1953 1956 1959 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013 Private industry, w/ cmputers Mfg, w/ cmputers Cmputer industry
Manufacturing real GDP grwth much slwer withut cmputer industry Mfg GDP grwth 45% f private sectr GDP grwth, 1979-2000 12%, 2000-2016 Manufacturing utput less in 2016 than in 2007.
20,000 Real GDP: Private Industry and Manufacturing less Cmputers, Cmputer Industry, 1977=100 18,000 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 1947 1949 1951 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 Private industries, less cmputers Manufacturing, less cmputers Cmputers Despite driving apparent rbust manufacturing grwth, dmestic manufacturing f cmputer & electrnic prducts lsing cmpetitiveness Huseman, Bartik, Sturgen (2016)
Grwth in dmestic cnsumptin fr manufactured prducts has been greater than fr services recently Since 2000, dmestic real cnsumptin f manufactured prducts (excluding cmputer and electrnics prducts) grwn at 2.8% per year, cmpared t 1.8% fr services Grwth in dmestic cnsumptin f furniture especially high at 3.5% Cnsistent with fld f cheap imprted prducts increasing cnsumer demand
Manufacturing prductivity grwth nt higher, r nly smewhat higher, than that in private industry Withut cmputer and electrnics prducts industry, labr and multifactr prductivity has been apprximately the same in manufacturing and private industry, 1987-2011 (Baily and Bswrth 2015) If prductivity grwth the same, it can accunt fr NONE f the decline in manufacturing s emplyment share Lwer utput grwth accunts fr all f the decline in manufacturing s emplyment share Alternate calculatins, using histrical emplyment data back t 1977: Slwer utput grwth accunts fr the majrity f the decline in manufacturing s emplyment share histrically. N prima facie case that prductivity grwth caused relative emplyment declines in manufacturing
Interpreting prductivity grwth Prductivity grwth autmatin Measured prductivity grwth may be capturing many things besides autmatin: Prduct imprvements, as in cmputers and semicnductrs Changes in the cmpsitin f prducts prduced: Prductivity grwth may be the cnsequence f trade cmpetitin if labr intensive prcesses mve verseas. Glbal supply chains and slicing up f the value chain may alter the stages f prductin dne in U.S., raising prductivity grwth Internatinal cmpetitin may spur autmatin Prductivity grwth des nt, per se, cause emplyment declines Strng argument that decline in manufacturing emplyment inevitable based n misreading f accunting identities Accunting identities and ther descriptive evidence cannt be interpreted as causal
Research finds large effects f internatinal trade n manufacturing jb lsses in 2000s Exchange rates, industry subsidies and taxes, tariffs, nn-tariff barriers affect relative cmpetitiveness f manufacturing in U.S. Glbalizatin may reduce dmestic manufacturing utput grwth by: Increasing grwth in real imprts Slwing grwth in real exprts cmpanies chse t relcate r expand verseas N study captures all aspects glbalizatin and its effects n manufacturing emplyment but cllectively research findings pint t large adverse effects, perating thrugh varius mechanisms.
Selected study findings f effects f trade n manufacturing emplyment lsses in 2000s Imprts frm China can accunt fr 25% f decline in manufacturing emplyment (Autr, Drn, Hansn 2013) Changes in trade plicy after 2000 caused increase in imprts frm China (bth frm Chinese exprters and U.S. based imprters) and significant decline in manufacturing emplyment (Pierce and Schtt 2016) Shift t less labr-intensive als led t significant emplyment declines Imprts frm China have als depressed investment, R&D, and sales in dmestic manufacturing (Autr et al. 2017, Pierce and Schtt 2017) Appreciatin f dllar and higher labr csts can explain mre than tw-thirds f decline in manufacturing emplyment in early 2000s, as well as declines in 1980s (Campbell 2016).
Research fails t find a causal link between technlgy and cllapse f manufacturing emplyment in 2000s Little evidence that IT investment and investment in advanced manufacturing technlgies increased labr prductivity, especially after 1990s (Acemuglu et al. 2014) Where crrelatins exist, assciated with declines in utput and even faster declines in emplyment (suggesting smething besides technlgical displacement ging n). Industries experiencing increased cmpetitin frm China had emplyment declines; industries susceptible t cmputerizatin f rutine tasks experienced plarizatin f ccupatins, NOT emplyment declines (Autr, Drn, Hansn 2015) Industrial rbts have ptential t displace many wrkers in future, but effect n manufacturing emplyment t date minimal (Acemglu and Restrep 2017) Rise f markups since 1980s and ffshring f labr intensive prcesses (nt capital investment) accunt fr rise f capital share (De Lecker and Eeckhut 2017; Elsby, Hbijn, Sahin 2013) incnsistent with large technlgy shck causing emplyment declines, rise in capital share
Why manufacturing matters Number f wrkers engaged in manufacturing still large: Abut half f wrkers needed in manufacturing nw emplyed utside the sectr Large spillver effects n lcal and natinal ecnmy Plant clsures have large adverse and lasting effects n wrkers and cmmunities: Dislcated wrkers experience large, persistent earnings lsses Large adverse shcks can send reginal ecnmies int dwnward spiral, depress regins fr decades (Dix-Carnair and Kvak, 2017) Cllapse f manufacturing emplyment cntributed t decline in emplyment rates, especially amng less-educated men and wmen (Charles et al. 2016, Autr et al. 2016) Lss f prductin leads t lss f R&D: Cnclusin f MIT Prductin in the Innvatin Ecnmy prject (Berger 2013)