The Surprisingly Swift Decline of U.S. Manufacturing Employment
description
Transcript of The Surprisingly Swift Decline of U.S. Manufacturing Employment
The Surprisingly Swift Decline of U.S. Manufacturing Employment
Justin R. Pierce Board of Governors of the Federal Reserve System
Peter K. Schott Yale School of Management & NBER
Disclaimer
Any opinions and conclusions expressed herein are those of the authors and do not necessarily represent 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 is disclosed.
2
Post-War U.S. Manufacturing Employment
3
12
14
16
18
20
Em
plo
ymen
t (M
illio
n)
1948 1958 1968 1978 1988 1998 2008
NBER Recessions ShadedU.S. Manufacturing Employment
Post-War U.S. Manufacturing Employment
4
12
14
16
18
20
Em
plo
ymen
t (M
illio
n)
1948 1958 1968 1978 1988 1998 2008
NBER Recessions ShadedU.S. Manufacturing Employment
Post-War U.S. Manufacturing Employment
5
12
14
16
18
20
Em
plo
ymen
t (M
illio
n)
1948 1958 1968 1978 1988 1998 2008
NBER Recessions ShadedU.S. Manufacturing Employment
-2.9 mill over 3 years
12
14
16
18
20
Em
plo
ymen
t (M
illio
n)
1948 1958 1968 1978 1988 1998 2008
NBER Recessions ShadedU.S. Manufacturing Employment
Post-War U.S. Manufacturing Employment
6
-2.9 mill over 3 years
Introduction
• The sharp decline in US manufacturing employment is closely linked to a change in U.S. trade policy granting China Permanent Normal Trade Relations (PNTR) status
• PNTR did not change actual tariff rates– They were already at the low levels offered WTO member– But this status required annual approval by Congress
• The possibility of large tariff increases before PNTR likely discouraged– US firms from offshoring– Chinese firms from expanding
7
Before Continuing…
8
200
400
800
1600
Rea
l Val
ue A
dde
d (B
illio
n)
10
12
14
16
18
20
Em
plo
ymen
t (M
illio
n)
1948 1958 1968 1978 1988 1998 2008year
BLS Employment BEA Real Value Added
Employment vs Log Real Value AddedU.S. Manufacturing
While US employment falls falls after 2001, real value added continues to rise at more-or-less the historical pace
Outline
• US-China Trade Policy
• Data
• Baseline results
• Extended results
• Other outcomes
• Conclusion
9
US NTR and Non-NTR Tariffs
• NTR = Normal Trade Relations– Synonym for Most Favored Nation (MFN)– Clinton?
• The US has two basic tariff schedules– NTR tariffs : generally low; for WTO members
– Non-NTR tariffs : generally high; for non-market economies;
: set by Smoot-Hawley (1930)
• So, how does China fit into these categories?
10
US-China Trade Policy, 1980-2001
11
1980 (February)China was granted temporary NTR status by the US Congress
Thereafter, this temporary NTR status requires annual re-approval by Congress
2000 (October)U.S. Congress grants China
PNTR, eliminating the risk that a failed vote might lead to a
jump in tariffs
2001 (December)China enters WTO
Annual renewals of MFN status were uncertain. The House tried to deny NTR every year after the Tiananmen Square incident; these efforts succeeded in 1990, 1991 and 1992, but the Senate failed to act on them in those years; the average vote against NTR from 1990-2000 is 38 percent
Measuring the Policy Change
• We define a measure of the effect of the policy as:
NTR Gap = Non-NTR Tariff – NTR Tariff
• Two useful attributes– Measures extent to which tariffs could increase prior to PNTR– Varies across industries
• We can preview the results in two simple pictures that divide US industries according to whether their NTR gaps are above or below the median
12
Preview of Findings – EmploymentPublic NBER-CES Data
13
• Split industries into two groups– Above median exposure– Below median exposure
.8
.9
1
1.1
1.2
200
1=1
1981 1986 1991 1996 2001 2006
Below Median NTR Gap Above Median NTR Gap
NBER-CES Manufacturing Industry DatabaseU.S. Manufacturing Employment
Preview of Findings – EmploymentPublic NBER-CES Data
14
• Split industries into two groups– Above median exposure– Below median exposure
• Trends are parallel before PNTR but diverge after PNTR
• (We use the gap as a continuous variable in our regression analysis.).8
.9
1
1.1
1.2
200
1=1
1981 1986 1991 1996 2001 2006
Below Median NTR Gap Above Median NTR Gap
NBER-CES Manufacturing Industry DatabaseU.S. Manufacturing Employment
Preview of Findings – TradePublic Census Trade Data
15
• Divergence is also evident in trade data
• Imports of more-exposed imports from China jump after PNTR relative to the imports from rest-of-world (ROW)
01
23
200
1=1
1991 1993 1995 1997 1999 2001 2003 2005 2007year
China: Below Mean China: Above Mean
U.S. Census Bureau Foreign Trade DataU.S. Imports Excluding Natural Resources
16
• Divergence is also evident in trade data
• Imports of more-exposed imports from China jump after PNTR relative to the imports from rest-of-world (ROW)
01
23
200
1=1
1991 1993 1995 1997 1999 2001 2003 2005 2007year
China: Below Mean ROW: Below MeanChina: Above Mean ROW: Above Mean
U.S. Census Bureau Foreign Trade DataU.S. Imports Excluding Natural Resources
Preview of Findings – TradePublic Census Trade Data
Outline
• US-China Trade Policy
• Data
• Baseline results
• Extended results
• Other outcomes
• Conclusion
17
NTR Gap = Non-NTR Tariff – NTR Tariff
• Compute NTR Gap for each HS8 using the ad valorem equivalent tariff rates from Feenstra, Romalis and Schott (2003)
• NTR Gap for industry i is the mean gap over the HS8s captured by the industry
• Available for 1989-2001, we use the NTR Gap for 1999 in the regression results to follow
18
Distribution of 1999 NTR Gap Across Industries
19
01
23
4D
ensi
ty
0 .2 .4 .6 .8
Distribution of 1999 NTR Gap
Mean: 0.32Std Dev 0.15
Census Employment Data
• LBDxxxxxxxxxxxxxxxxxxxx
• CMxxxxxxxxxxxxxxxxxxxxx
• LFTTD
• Annual employment of all U.S. establishments, 1977-2009
• Employment + other attributes for all manufacturing establishments every five years, 1977(5)2007
• Transaction-level US import data– Value– HS Product– Importer ID– Foreign exporter ID
20
Census Employment Data
• LBDxxxxxxxxxxxxxxxxxxxx
• CMxxxxxxxxxxxxxxxxxxxxx
• LFTTD
• Annual employment of all U.S. establishments, 1977-2009
• Employment + other attributes for all manufacturing establishments every five years, 1977(5)2007
• Transaction-level US import data– Value– HS Product– Importer ID– Foreign exporter ID
21
Outline
• US-China Trade Policy
• Data
• Baseline results
• Extended results
• Other outcomes
• Conclusion
22
Empirical Strategy
• We use a difference-in-differences strategy to examine the link between PNTR/WTO and U.S. manufacturing employment outcomes– 1st difference: industries with higher vs lower NTR Gaps– 2nd difference: growth after 2001 versus after 1990 (prior peak)
23
Industry-Level OLS Diff-in-Diff (DID)
24
Cumulative percent change in industry i
employment d years after NBER
peak t={1990,2001}
DID TermInteraction of indicator for
post-PNTR period and continuous, time-invariant
NTR Gap
Industry and peak-year
fixed effects
Industry attributes, e.g., K/L and S/L
t = NBER peak {1990,2001} d = 1:6 years after peaki = industry
Industry-Level OLS Diff-in-Diff (DID)
25
Cumulative percent change in industry i
employment d years after NBER
peak t={1990,2001}
DID TermInteraction of indicator for
post-PNTR period and continuous, time-invariant
NTR Gap
Industry and peak-year
fixed effects
Industry attributes, e.g., K/L and S/L
t = NBER peak {1990,2001} d = 1:6 years after peaki = industry
• Two sets of results– Baseline: only control for industry capital and skill intensity– Extended: control for everything we can think of
Basic Industry-Level RegressionsBold=statistically significant at 10% level
26
Percent Change in Industry Employment
Years After NBER Peak (LBD)
1 2 3 4 5 61{post-PNTR} x NTR Gapi -0.104 -0.187 -0.332 -0.387 -0.469 -0.482 0.058 0.082 0.105 0.114 0.149 0.147
ln(K/Lit) -0.058 -0.032 0.021 0.099 0.140 0.170 0.036 0.056 0.071 0.077 0.101 0.093
ln(S/Lit) -0.048 -0.110 -0.140 -0.131 -0.087 -0.108
0.046 0.059 0.075 0.087 0.096 0.111
Observations 652 652 652 652 652 652R2 0.67 0.70 0.70 0.70 0.66 0.66Fixed Effects i,t i,t i,t i,t i,t i,tEmployment Weighted Yes Yes Yes Yes Yes Yes
Implied Impact of PNTR -0.034 -0.060 -0.107 -0.125 -0.151 -0.156
0.019 0.026 0.034 0.037 0.048 0.047
Use LBD to examine outcomes 1:6 years after the 2001 vs 1990 peak
Basic Industry-Level RegressionsBold=statistically significant at 10% level
27
Absolute magnitude of DID coefficient rises over time; i.e., relative losses are persistent
Percent Change in Industry Employment
Years After NBER Peak (LBD)
1 2 3 4 5 61{post-PNTR} x NTR Gapi -0.104 -0.187 -0.332 -0.387 -0.469 -0.482 0.058 0.082 0.105 0.114 0.149 0.147
ln(K/Lit) -0.058 -0.032 0.021 0.099 0.140 0.170 0.036 0.056 0.071 0.077 0.101 0.093
ln(S/Lit) -0.048 -0.110 -0.140 -0.131 -0.087 -0.108
0.046 0.059 0.075 0.087 0.096 0.111
Observations 652 652 652 652 652 652R2 0.67 0.70 0.70 0.70 0.66 0.66Fixed Effects i,t i,t i,t i,t i,t i,tEmployment Weighted Yes Yes Yes Yes Yes Yes
Implied Impact of PNTR -0.034 -0.060 -0.107 -0.125 -0.151 -0.156
0.019 0.026 0.034 0.037 0.048 0.047
Basic Industry-Level RegressionsBold=statistically significant at 10% level
28
Multiply DID estimates by average NTR gap (0.34) to compute implied impact
Post-2001 growth is 3-16 percent lower than after 1990
Percent Change in Industry Employment
Years After NBER Peak (LBD)
1 2 3 4 5 61{post-PNTR} x NTR Gapi -0.104 -0.187 -0.332 -0.387 -0.469 -0.482 0.058 0.082 0.105 0.114 0.149 0.147
ln(K/Lit) -0.058 -0.032 0.021 0.099 0.140 0.170 0.036 0.056 0.071 0.077 0.101 0.093
ln(S/Lit) -0.048 -0.110 -0.140 -0.131 -0.087 -0.108
0.046 0.059 0.075 0.087 0.096 0.111
Observations 652 652 652 652 652 652R2 0.67 0.70 0.70 0.70 0.66 0.66Fixed Effects i,t i,t i,t i,t i,t i,tEmployment Weighted Yes Yes Yes Yes Yes Yes
Implied Impact of PNTR -0.034 -0.060 -0.107 -0.125 -0.151 -0.156
0.019 0.026 0.034 0.037 0.048 0.047
Basic Industry-Level RegressionsBold=statistically significant at 10% level
29
Results are robust to including industry capital and skill intensity
Later, we allow these relationships to differ in the pre- and post-periods
Percent Change in Industry Employment
Years After NBER Peak (LBD)
1 2 3 4 5 61{post-PNTR} x NTR Gapi -0.104 -0.187 -0.332 -0.387 -0.469 -0.482 0.058 0.082 0.105 0.114 0.149 0.147
ln(K/Lit) -0.058 -0.032 0.021 0.099 0.140 0.170 0.036 0.056 0.071 0.077 0.101 0.093
ln(S/Lit) -0.048 -0.110 -0.140 -0.131 -0.087 -0.108
0.046 0.059 0.075 0.087 0.096 0.111
Observations 652 652 652 652 652 652R2 0.67 0.70 0.70 0.70 0.66 0.66Fixed Effects i,t i,t i,t i,t i,t i,tEmployment Weighted Yes Yes Yes Yes Yes Yes
Implied Impact of PNTR -0.034 -0.060 -0.107 -0.125 -0.151 -0.156
0.019 0.026 0.034 0.037 0.048 0.047
Baseline ResultsDID Controlling for Industry K/L and S/L Only
30
After 7 years, cumulative employment growth is 15.6 percent lower than 7 years after the 1990 peak
0-.
1-.
2-.
3R
elat
ive
Gro
wth
2002 2003 2004 2005 2006 2007
Note: Dashed lines indicate 90% confidence interval.
Baseline SpecificationImplied Impact of PNTR
After 1 year, cumulative employment growth is 3.4 percent lower than 1 year after the 1990 peak
Outline
• PNTR
• Data
• Baseline results
• Extended results
• Other outcomes
• Conclusion
31
Alternate Explanations?
• Alternate explanations must explain:
– Timing: why do U.S. employment fall and Chinese imports rise with PNTR?
– Variation across industries: Why are changes in employment an imports greater for industries most affected by the policy change?
• Maybe the employment declines are related to changes in Chinese policy?– Not obvious why they would be correlated with the U.S. NTR gap,
but we can check…
32
Alternate Explanations
• Changes in Chinese Policy – Reduce barriers to foreign investment – Nunn (2007)– Lower import tariffs – Brandt et al. (2013)– Eliminate export licensing requirements – Bai et al. (2007)– Eliminate production subsides – Girma et al. (2007), Bown (2012)
• Union Resistance in the US – unionstats.org
• Bursting of the US tech bubble – IT dummy; control for prior growth
• Rising Chinese competitiveness – control for capital and skill intensity
• End of Textile and Clothing Quotas – Khandelwal et al. (2013)
33
Implied Impact of PNTRBaseline vs Extended
34
Results are similar even after accounting for many alternate explanations
If we use different DID specifications, we also get similar results.
0-.
1-.
2-.
3
2002 2004 2006 2002 2004 2006
Baseline Baseline + Alternate Explanations
Rel
ativ
e G
row
th
Note: Each panel displays implied impact plus 90% confidence interval.
By SpecificationImplied Impact of PNTR
Outline
• US-China Trade Policy
• Data
• Baseline vs Extended Results
• Other results– Other Countries– Margins of Adjustment– Participation in trade by U.S. and Chinese firms– Plant-level
• Conclusion
35
Other Countries?
• The EU granted permanent NTR status to China in 1980– So, no annual renewals
• Dissimilarity of US and EU results indicates that the outcomes we observe in the U.S. are unlikely to be driven by – A productivity shock in China– A technology shock in industries that happen to have high gaps
• We can use employment from a different source – UNIDO – to compare the US and the EU
36
Overview of U.S. versus EUEmployment Data from UNIDO
37
.75
.85
.95
1.05
1.15
1996 2000 2004 1996 2000 2004
EU US
All Indusries
200
1=1
Source: INDSTAT4 available at http://www.unido.org/resources/statistics/statistical-databases/.
Manufacturing Employment
38
Overview of U.S. versus EUEmployment Data from UNIDO
.75
.85
.95
1.05
1.15
1996 2000 2004 1996 2000 2004
EU US
NTR Gap<Median NTR Gap>Median
200
1=1
Source: INDSTAT4 available at http://www.unido.org/resources/statistics/statistical-databases/.
Manufacturing Employment
DID regression analysis reveals that the post-PNTR differences are statistically significant for the US but not the EU
The dissimilarity of US and EU results indicates that US outcomes are not driven by: (i) a productivity shock in China (ii) a technology shock in industries that happen to have high NTR gaps
Outline
• US-China Trade Policy
• Data
• Baseline results & Alternate explanations
• Additional results– Other Countries– Margins of Adjustment– Participation in trade by U.S. and Chinese firms– Plant-level
• Conclusion
39
Margins of Adjustment
• Job Destruction (JD)– PC: plant contraction at continuing firms– PD: plant death at continuing firms– FD: firm death
• Job Creation (JC)– PE: plant expansion at continuing firms– PB: plant birth at continuing firms– FB: firm birth
40
-.15
-.1
-.05
0
Cha
nge
fro
m P
eak
Yea
r
2001-2 2001-3 2001-4 2001-5 2001-6 2001-7
By Job Creation vs DestructionDecomposition of Implied Impact of PNTR
Job Desctruction (PC+PD+FD) Job Creation (PE+PB+FB)
Implied Impact of PNTR
41
Contribution of exaggerated job destruction
Contribution of anemic job creation
JC contributes 17 to 41 percent across 2001-2007
Relates to research by Faberman (2012) on “jobless” recoveries and the end of the “great moderation” after 2001
Outline
• US-China Trade Policy
• Data
• Baseline results & Alternate explanations
• Additional results– Other Countries– Margins of Adjustment– Participation in trade by U.S. and Chinese firms– Plant-level
• Conclusion
42
Trade DID
• Trade data not available till 1990s
• So, amend DID to compare 2001-2005 to 1997-2001– 1st difference: products with higher vs lower NTR Gaps– 2nd difference: imports from China vs other U.S. trading partners
• Examine:– Value– Number of US importers– Number of Chinese exporters– Number of US-Chinese importer-exporter pairs
43
44
0.1
.2.3
.4G
row
th R
elat
ive
to R
OW
1. Value 2. US Importers 3. CHN Exporters 4. Pairs
Implied Impact of PNTR on US Trade
PNTR and U.S. Imports from ChinaImplied Growth 2001-2005 vs 1997-2001
Outline
• US-China Trade Policy
• Data
• Baseline results & Alternate explanations
• Additional results– Other Countries– Margins of Adjustment– Participation in trade by U.S. and Chinese firms– Plant-level
• Conclusion
45
Outline
• US-China Trade Policy
• Data
• Baseline results & Alternate explanations
• Additional results– Other Countries– Margins of Adjustment– Participation in trade by U.S. and Chinese firms– Plant-level
• Conclusion
46
Conclusions
• Strong link between US manufacturing job loss, US trade and PNTR
• U.S. imports from China surge in same industries where employment declines occur, along with number of U.S. importers, Chinese exporters and importer-exporter pairs
• Results are robust to inclusion of proxies for wide array of alternate explanations, as well as alternate specifications
• PNTR associated with both increased job destruction and decreased job creation
47
Thanks
48