Transcript Does Work First Work? Long-Term Consequences of Temp

Sharing my worries regarding the
affects of offshoring on REMI
regional impact estimates
George Erickcek & Susan Houseman
W.E. Upjohn Institute for Employment Research
Presentation prepared the Annual REMI Users Conference Santa Fe,
New Mexico October, 2008
Production and Employment in Manufacturing
120
20000
110
18000
100
16000
90
14000
80
12000
70
10000
60
8000
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Year
Federal Reserve Board Production Index (Manufacturing)
U.S. Manufacturing Employment
U.S. Mfg Employment (000)
Production Index 2002=100
It ain’t necessarily so.
Areas of concerns

Recently revealed errors in BEA’s pricing of
imports may result in the overestimation of
output and productivity gains.

I worry that, if true, these errors could result in
REMI’s forecast for manufacturing goods and
some services to be too robust.

These problems are compounded by time lags
in the data especially in the construction of the
national input-output accounts.
Why is this a problem?

Global offshoring has become “standard”
practice for many manufacturers and possibly
a growing number of service functions.

The hollowing out of a region’s supply chain
will have immediate impact on the size of the
“multiplier impact”
 Regional output flow are lessen
 Income earnings are reduced
 Resulting consumption expenditures
are diminished.
Federal Reserve Board and BEA Studies Show Substantial
Growth in Outsourced and Imported Intermediate Inputs

Domestic providers of outsourcing services –
significant growth in share of GDP 1982-2006:
7% to 12%

Domestic outsourcing esp. strong durable
manufacturing

Substantial substitution imported inputs for
domestic inputs in production of goods &
services 1997-2005

Growth imported intermediates esp. strong
manufacturing – accelerated 1997-2005
Yuskavage, Strassner, Medeiros (2008); Kurz & Lengermann (2008)
Growing Importance of Imports from
Developing Countries


Imports as percent
GDP grew from 10.8%
in 1989 to 17.0% in
2007
Developing
countries—growth
non-oil imports:
o
o

56% 1989-2000
70% growth 2000-2007
Growth imports from
China especially
dramatic:
o
o
13% growth non-oil
imports 1989-2000
39% growth 2000-2007
Imports and Exports as
Percent of GDP, 1989-2007
30
Imports + Exports
25
20
Imports
15
Exports
10
5
0
1989
1992
1995
1998
2001
2004
2007
Problem Outsourcing and Offshoring Pose for
Price Measurement

Changes in sourcing of intermediate inputs
generally motivated by lower prices

Price drops associated with changes in sourcing
generally not captured in US price statistics

Surveys implicitly assume patterns of sourcing
are stable or change slowly over time
Implications of not measuring input price
drops with changes in sourcing




Price indexes used to compute real value added at
industry and sector levels—GDP at aggregate
levels;
These real output measures, in turn, used to
compute many productivity statistics
When price index growth is overstated
o Real value of offshored input understated
o Real domestic output and/or sector value-added
growth overstated,
o Aggregate and/or sectoral productivity growth
overstated
Output and productivity statistics unreliable,
especially in industries that have experienced
rapid changes in offshoring.
Example: Producer Switches from Domestic to LowerCost Foreign Supplier of Intermediate Input

Price indexes constructed as average of changes across
periods, zero in this example
Producer Buys
Intermediate Input
Input A
US supplier
$2
Period 2
% change = 0
Input A
US supplier
$2
Period 1

Input A
foreign supplier
$1
Period 2
% change = 0
Input A
foreign supplier
$1
Period 1
If producer switches from US supplier in period 1 to foreign
supplier in period 2, input price drops by 50%, but no price
drop measured in official statistics
A simple illustration of the problem:
A Simple Example: A single company
Value of Shipments
-Labor
-Interest
-Profits
-Intermediate goods
Domestic
Prices
International
Output (VA)
Period Two
Period One (unadjusted)
100
110
30
15
20
20
10
35
25
1.00
15
60
10
Producers
price
1.00
30
70
This is very similar to the “Walmart” effect on the CPI.
Period Two
(adjusted)
110
15
20
5
10
Buyers
price
0.50
60
40
Timeliness of BEA benchmark I-O tables

Latest I-O table used to compute GDP by
industry and sector productivity measures
based is based on 2002 data.
o Imports are accounted in the final demand.



This means that the interindustry transactions
do not capture the share of the an industry’s
demand that is met through imports.
So if imports grow and the inter-industry
transaction estimates are not adjusted, then the
tech. coefficients are too high and so are the
multipliers.
Imports are valued at “foreign port value” plus
freight and duties.
Standard input-output model
Inter-industry Matrix
Industry 1
Industry 2
a11 a12 a13 a14 ….
a21
Final
Demand
a1n
c1 i1 g1 x1
mi
Total
aij
a31
a41
an1……
Labor
Profits
Interest
ann
(Value Added)
Total
Where: a11 = X11/X1total, mi = M1/Xtotal
Note:
There are no estimates of industry 11’s
demand for industry’s 21 product that is
imported
Real Growth in U.S. Imports in REMI 9.5
Industry
Apparel
Chemical
Furniture
Motor vehicles
Electrical equip, appliance
Non-auto transp equip
Machinery
Miscellaneous mfg
Plastics
Fabricated metal products
Computer, electronic products
Paper
Primary metals
Real Growth in
Imports 20062014
19.2%
16.6%
15.0%
6.7%
6.7%
3.4%
2.7%
1.7%
-0.4%
-3.5%
-5.0%
-11.3%
-13.0%
How is the growing use of global supply
chains captured in the REMI model?


Not immediately clear. Highly focused on
the impact of inter-regional comparative
costs: Market Share.
REMI captures the trend in offshoring in
two separate ways:
o The national inter-industry coefficients are
adjusted for imports. But how?
o The sector’s regional presence is based on
employment – If offshoring is occurring then
the supply industries’ employment base
would be declining lowering the model’s
RPC.
REMI modeling
Output for industry i in region k.
Qki= ski DDi + ski Xi
DD = domestic demand for i based on the models
inter-industry estimates (somehow adjusted for
imports)
Xi = International exports
ski = f(rel. prod. cost, geog, industry mix, export share)
REMI modeling
Domestic demand for industry i in region k.
DDki= ( ∑aijQj + ∑ ciCj + ∑ kiIj + ∑ giGj) ski
∑aij Qij
= standard
input-output inter-industry demand
estimation – the amount of i purchased to
generate a dollar of j in the region k.
However, REMI adjusts this by factoring in its
estimate of the intermediate input access.
REMI modeling
Domestic demand for industry i in region k.
DDki= ( ∑aijQj + ∑ ciCj + ∑ kiIj + ∑ giGj) ski
The amount of i purchased to meet the
final demand of consumption, private
investment and government.
Ski = The domestic share of
k’s demand for i.
Ski = f(hist. import trends, rel. prod. costs)
Closing thoughts and concerns.




The possible errors in the nation’s output estimates may
have significant impacts on the output growth rates in the
REMI model.
Given business reports that offshoring is still alive and well
even in the era of high transportation costs (Mexico’s
advantage), the necessary 5 to 10 year lag in the
construction of the national input-output model is
worrisome.
It is unclear how imports (a final demand component) are
allocated in the interindustry transactions.
In short, all of these concerns make me think that
multipliers generated by REMI and all other regional models
may be too high.