Post on 06-Oct-2020
Discussion of Chen, Kannan, Loungani, and Trehan
Daniel Aaronson
Chicago Fed
April 2011
Summary of paper
• Estimate shocks to sectoral reallocation using dispersion in industry stock market returns.
• Interesting result for at least two reasons:
1. Explains ½ (!!) of the historic rise in LTU (Figure 10). a. Nice check -- No impact on ST unemployment, which is less likely to be
“structural.”
2. Potential estimate of “structural unemployment” (Fig 11). Very large effects here too.
• Key assumption:
– Assumes movements in stock prices within an industry reflect permanent (structural) shifts, rather than temporary (cyclical) changes.
– But fair amount of evidence that “structural change” coincides with cyclical downturns. Makes identification difficult.
– Particularly tricky when industry cyclicality varies.
Some reasons to be skeptical of magnitudes
1. Using the current period in the estimation.
• Concern: Big outlier (dispersion, LTU) in 2008/09 much of the variation used to predict rise in LTU and structural unemployment recently.
y = 0.9199x + 0.0647R² = 0.1469
y = 0.4755x + 0.0965R² = 0.0477
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0 0.05 0.1 0.15 0.2 0.25
Shar
e o
f u
ne
mp
loye
d >
26
we
eks
Chen et al Dispersion index, lagged 8 quarters
Why is this a potential concern?
• Lots going on recently. Control for overall stock market dispersion is nice but not far enough. Much more of this would be great (broader financial conditions, within-industry dispersion).
• Need out of sample tests. Effects almost surely smaller.
-4
-3
-2
-1
0
1
2
3
4
5
6
Jan
-73
Jan
-75
Jan
-77
Jan
-79
Jan
-81
Jan
-83
Jan
-85
Jan
-87
Jan
-89
Jan
-91
Jan
-93
Jan
-95
Jan
-97
Jan
-99
Jan
-01
Jan
-03
Jan
-05
Jan
-07
Jan
-09
Jan
-11
Stan
dar
d d
ev
un
its
Chicago Fed National Financial Conditions Index
Removes variation due to economic activity (CFNAI) and inflation (PCE)
http://chicagofed.org/webpages/publications/nfci/index.cfm
Long lags between business cycle shock and the rise in
LTU are standard parts of recovery.
• But do we think that recessions always involve sectoral reallocation? Not clear (some evidence in a few slides).
0
5
10
15
20
25
30
35
40
45
50
0 2 4 6 8 10 12
Frac
tio
n o
f La
bo
r Fo
rce
Un
em
plo
yed
27
w
ee
ks +
Unemployment rate
Unemployment Rate vs share long-term unemployed, selected cycles
12/07-3/1211/73-10/777/81-6/857/90-6/943/2001-2/2005
2. Suggests alot more structural unemployment (8.5-
9%) than other approaches
Shock: 16% reduction in match efficiency (BF 2011)New Beveridge Curve
With free entrycondition (and otherassumptions),u/v must rise by 45%.
Shock to matchefficiency can only leadto an upper bound of7% SS unemployment
Barlevy (2011, forthcoming)
And its probably even lower: 1. Davis et al (2010) – Firm search intensity; 2. UI – worker search intensity
(Aaronson/Mazumder, Mazumder, Valetta, etc.)So can’t all be about mismatch (within DMP framework).
2 pp
And a lot more sector reallocation
(e.g. Sahin et al (2011))
• Economy is a large number of distinct labor markets (industries, occupations, etc) with unemployed resources and vacancies.– Kind of like a industry-specific (or occupation-specific, etc) Beveridge Curve
analysis
• Question: how much unemployment can be “fixed” by freely moving resources to job openings in other sectors. – i.e. the share of unemployment due to workers being in the “wrong” labor
market (the one without openings).
• By industry: explains 0.4 to 0.7 perc points of 5 point rise during recession and early recovery. Most of this is construction (and a little manufacturing).
• Index has a little trouble with timing because it reverts back pretty quickly in 2009-10 when UR and LTU are particularly high (unless there are long lags like in Prakash’s VARs).
-30
-20
-10
0
10
20
30
Construction
-30-20-10
010203040
Durable manufacturing
-20-15-10
-505
1015
Nondurable manufacturing
-15
-10
-5
0
5
10
15
Transportation and utilities
-10
-5
0
5
10
15
Wholesale Trade
-10
-5
0
5
10
15
Retail Trade
-10
-5
0
5
10
Finance, insurance, and real estate
-6-4-202468
10
Services
-4-202468
10
Government
Employment has fallen broadly
-30
-20
-10
0
10
20
30
Construction
-30-20-10
010203040
Durable manufacturing
-20-15-10
-505
1015
Nondurable manufacturing
-15
-10
-5
0
5
10
15
Transportation and utilities
-10
-5
0
5
10
15
Wholesale Trade
-10
-5
0
5
10
15
Retail Trade
-10
-5
0
5
10
Finance, insurance, and real estate
-6-4-202468
10
Services
-4-202468
10
Government
Employment has fallen broadly (and fairly cyclically)
Statistical model of Rissman (2010): Noncyclical reallocation
barely budged
0
1
2
3
4
5
6
7
8
1950 1954 1958 1962 1966 1970 1974 1978 1982 1986 1990 1994 1998 2002 2006 2010
Important caveat: Limited categorization (9 industries). Mismatch may be within industry or across other labor market segmentations.
Rissman, Ellen, 2010, Economic Perspectives.
Aaronson et al (2010)
• Oaxaca-Blinder decomposition of long-term unemployment changes over last 30 years.
• Virtually no evidence that across-industry (again, broad) changes can explain much.
• Within-industry changes are actually much more useful. Maybe this is picking up some of Prakash’s results. – Easy test: play with the level of aggregation.
• Rob V has better evidence on how unusual the current period is, after adjusting for demographics.
0.00
0.01
0.02
0.03
0.04
0.05
1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009
3. Is there Unusual Demand in Certain Sectors? Standard
wage data doesn’t suggest it.Dispersion of private industry hourly earnings(year-over-year)
Highest industry less averageStandard deviation of industries
Average hourly earnings, by private industry (CES). Excludes mining and logging. Red line = highest industry less average industry. Blue line = standard deviation of industry average hourly earnings.
Industry wage dispersion in the SIPP. Mixed evidence?
• Pro: Can deal with some compositional biases (not completely done yet…).
• Con: Very Preliminary, not many industries (yet), smallish samples
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
19
84
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
Dispersion of private industry wage growth , SIPP
Job switchers
Spell of U
Job stayers
*Thru Mar 2010
Real wage growth among hires from U is falling in
tandem
-0.10
-0.05
0.00
0.05
0.10
0.15
19
84
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
Log
wag
e c
han
ge
Annual Real Wage Growth, Individuals experiencing E-U-E
Goods-producing
Finance
Construction
Change, 2008-2010 vs 1984-2007 growth
Finance -0.099Construction -0.044Goods prod -0.037Other serv -0.018Gov +0.029
*Thru Mar 2010
Hiring is broadly bad. Some sectoral reallocation likely
but evidence is far from clear that its substantial.
10.0%
15.0%
20.0%
25.0%
30.0%
35.0%
40.0%
45.0%
Jan
-05
Jan
-06
Jan
-07
Jan
-08
Jan
-09
Jan
-10
0-4 weeks
5-14 weeks
15-26 weeks
27 plus weeks
0.0
2.0
4.0
6.0
8.0
10.0
12.0
1976 '81 '86 '91 '96 '01 '06
TotalCollege*
Unemployment rates(total; college degree or more, aged 16-24, SA percent)
* Recent college grads = No UI, no homes, highly mobile (geographically, sector). Yet UR rose to similar heights.
Fraction that move from U to E, by initial unemployment duration(total unemployed in t-1; SA percent)
Hiring bad at every duration spell.
Extra slides
Wage changes among LTU (SIPP)
-0.15
-0.1
-0.05
0
0.05
0.1
19
84
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
Log
Wag
e C
han
ge
Annual Wage Growth by Spell Length (with 90% confidence intervals)
1 to 6 month spell
7 to 12 month spell
*Thru Mar 2010