Why Did Europe’s Productivity Growth Catch
Download
Report
Transcript Why Did Europe’s Productivity Growth Catch
Understanding the Post-1995
Productivity Turnaround
between Europe and the US
Robert J. Gordon
Northwestern University, NBER, CEPR
14th Dubrovnik Economic Conference
Sponsored by Croatian National Bank
Dubrovnik, Croatia, June 26, 2008
What is Europe?
Throughout “Europe” refers to the EU-15, not
the EU-25
Initially the presentation compares EU-15 as a
whole to the U. S.
The second session divides up the EU-15 into
country groups divided by geography
Time periods: Initial reference to post-1970,
most of presentation compares 1980-95 with
1995-2005
Outline of the Talk
Focus on post-1995 EU Turnaround vs. the US
In which Industries has EU Productivity Faltered,
comparing country groups and EU vs. US?
Is there a Tradeoff between Employment and
Productivity?
Productivity growth has slowed down
Employment per capita growth has speeded up
Channels from policy and institutions to E/N
Cultural changes in female LFPR
Contribution of employment turnaround to productivity
turnaround
Good news vs. bad news conclusion
Session #2:
Is There an EmploymentProductivity Tradeoff ?
Two marked events in Europe after 1995
Are these connected causally or just a coincidence?
Slowdown in productivity growth to well below the U. S. rate
Increase in growth of employment per capita at well above
the U. S. rate
If connected, which way does the causation go?
Co-authored with Ian Dew-Becker
Look him up on google
Ian in SF, you can’t see “MV=PY”
The US Accelerates,
Europe Decelerates
From 1950 to 1995 EU productivity growth was faster
than in the US
But in the past decade since 1995 we have witnessed
An explosion in US productivity growth
A slowdown in EU productivity growth roughly equal in size
An explosion in research on the US takeoff and but much less
research on Europe’s slowdown
The magnitude of the shift (average EKS&GK
Groningen)
EU/US level of labor productivity (ALP)
1979
1995
2004
80%
97%
89%
Point of Departure: Post-95
Turnaround Plus New Heterogeneity
Notational identity: Y/N ≡ Y/H * H/N
This paper begins with two simple observations:
1. While European productivity (Y/H) has fallen back since 1995 relative to
the US, output per capita (Y/N) has not fared nearly as badly
►Y/H growth rate US - EU: 0.9%
►Y/N growth rate US - EU: 0.2%
2. A second theme: after 1995, we see divergence across the EU-15 in Y/H
growth
► St. Dev. 1970-1995: 0.62
► St. Dev. 1995-2005: 1.01
3. Even greater increase in standard deviation of H/N and Y/N growth
The Key Identity Suggests
the Tradeoff
Returning to our identity:
Y/N = Y/H * H/N
The level of EU relative to US Y/H was much higher than Y/N
in both 1995 and 2004.
Thus the paradox of high European Y/H and low Y/N must be
resolved by lower H/N
Also, Y/H and H/N are jointly determined
The task of this paper is going to be figure out which
direction the causation runs
We will argue that a good deal of the decline in ALP growth
is due to exogenous employment shocks
Also we will highlight the reversal of almost everything at
1995, comparing 1970-95 vs. 1995-2005
Bringing Together the Disparate
Literatures
Literature #1, why did Europe’s hours per capita
(hereafter H/N) decline before 1995? Prescott,
Rogerson, Sargent-Lundqvist, Alesina, Blanchard
High taxes, regulations, unions, high minimum wages
Europe made labor expensive
Movement up Labor Demand curve => low employment +
high ALP
Literature #1 has missed the turnaround
Since 1995 there has been a decline in tax rates and
employment protection measures; unionization earlier
Big increase in hours per capita, turnaround in both absolute
terms and relative to the US Move back down LD curve
The Employment-Productivity Tradeoff
Take any CRS production Y = F(K,H)
Intensive form Y/H=f(K/H)
As long as capital is fixed, an increase in
employment lowers labor productivity
We don’t know how fast capital adjusts; the
tradeoff may be quantitatively small
A major goal of this paper is to quantify the
tradeoff
Textbook Labor Economics
7
6
High-Cost Labor
Supply Curve
Labor Demand
Curve
5
Real Wage
4
(W/P)0
A
3
Low-Cost
Labor
Supply Curve
(W/P)1
B
2
1
0
Downward shift in labor
supply curve reduces real
wage and productivity
-1
-2
1
2
3
4
5
N0
6
Labor Input
7
N1
8
9
10
11
Pre-1995: Moving Northwest
1970-95 EU climbs to the northwest
Hours per capita decline, average labor productivity
increases
In this sense much of Europe’s 1970-95 productivity
catchup was “artificial,” propelled by policies making
labor expensive
No busboys, grocery baggers, valet parkers
Product market regulations kept stores shut tight many
hours of the day/night
All this reduced Europe’s employment share in
retail/services
Post-1995: Moving Southeast
1995-2004 EU slides southeast
Hours per capita start increasing while they decline in the US
Effects are magnified by slow reaction of capital.
Depending on the model, expanded employment should eventually
stimulate growth of capital, shifting the labor demand curve up and
eliminating much of the productivity decline
Literature #1 misses the turnaround in hours
Since 1995 decline in tax rates and employment protection
measures
We are unaware of much macro-level research on the
turnaround in hours
Literature #2: The EU-US ALP gap
Central Focus of Lit #2 on post-1995 turnaround in
US Productivity Growth
Jorgenson, Ho and Stiroh (2006): ’95-’00 due to ICT, ’00-’05
something else
Retail is often noted: contrast between big boxes at highway
intersections in US vs. inner city pedestrian districts in EU
Van Ark, Inklaar and McGuckin (2003)
Foster, Haltiwanger and Krizan (2002) on new establishments
Baily and Kirkegaard (2004) on product market regulations
Need to free land use restrictions
Restrictions on shop-closing hours
Fully 85% of EU productivity slowdown has its
counterpart in a speed-up of EU H/N
Europe paid for lower ALP mainly with higher
hours rather than less consumption
This runs counter to the Blanchard story about
preferences for leisure
Now we hear that they’re not lazy, just unproductive
Huge literature on different structural reasons for
EU sclerosis
Literature #3: relationship between
Y/H and H/N
There is a long line of research examining the
relationship between hours and productivity
Increases in H/N drive down Y/H
This makes sense in a single factor model or with any slow
adjustment of capital
Measuring the speed of adjustment of investment is difficult
This tradeoff idea was first proposed as an explanation of
slow productivity growth in the US during 1973-95
View today’s talk as a report on research in progress,
not the final polished word
You’ll find the complete text as a CEPR DP February 2008
Our First Look at the Data
EU-15 vs. US
Hodrick-Prescott filtered, not actual growth rates
Expanding the identity
Y/N ≡ Y/H * H/N
H/N ≡ H/E * E/N
Combine:
Y/N ≡ Y/H * H/E * E/N
Trends in Labor Productivity
Growth, 1970-2006
6
5
Percent
4
EU-15 Y/H
US Y/H
3
2
1
0
1970
1975
1980
1985
1990
1995
2000
2005
What to Notice About LP
The EU Slowdown is steady and continuous
The US post-1995 revival is looking increasingly
temporary
We created the US trend from quarterly data through 2007,
not just the annual data through 2004 used by EU-KLEMS
The fact that the US trend is turning around is
important for interpretations of what caused the post1995 US revival
That’s a separate paper. Today we primarily look inside
Europe and exclude the US from the employment and
productivity regressions
U. S. Productivity Growth Trends
Based on Data to 2007:Q4
3.5
3.0
NFPB LP
2.5
2.0
TE LP
1.5
1.0
NFPB minus TE
0.5
0.0
1955
-0.5
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
Growth Trends in Y/N and H/N,
1970-2006
4
EU-15 Y/N
US Y/N
3
Percent
2
1
0
-1
EU-15 H/N
-2
US H/N
-3
1970
1975
1980
1985
1990
1995
2000
2005
Comments about H/N and Y/N
Importance of expressing everything per capita
EU Growth in H/N strongly negative pre-1995, US strongly
positive
Falling level of H/N in Europe is what Prescott and others have
been trying to explain
Productivity and employment turnarounds cancel out. Growth
in Y/N almost equal 1980-2005
Average EU population growth 0.7 percent per year slower than US
EU 1.92 percent per year, US 1.97 percent per year
But EU is at only 70-75 percent of US level and is not catching
up
Growth Trends in E/N,
1970-2006
2
US E/N
1.5
Percent
1
0.5
0
-0.5
EU-15 E/N
-1
1970
1975
1980
1985
1990
1995
2000
2005
Growth Trend Turnaround in
H/E is less Dramatic, 1970-2006
1
US H/E
0.5
Percent
0
-0.5
-1
EU-15 H/E
-1.5
-2
1970
1975
1980
1985
1990
1995
2000
2005
We Need to Look at Everything
Per Capita
Population growth in EU 0.7 percent per year
slower than US over the past decade
Output per capita in the EU doesn’t look bad at
all
Post-1995 hours turnaround is a counterpart to
the Y/H turnaround
We will see that there is a similar pattern within
the EU – a strong negative correlation between
the hours and ALP turnarounds
Turnarounds in Hours and Output
Turnarounds are 1995-2006 minus 1980-1995
growth
The relative turnarounds (EU minus US) almost
cancel each other out for Y/N
Y/H + H/N = Y/N
-2.20 1.99 -0.21
1980-2005 Y/N growth is identical
But the EU is not catching up from its level ratio
of 70 percent
US vs EU E/N
0.55
1.25
1.20
0.50
1.15
1.10
0.45
1.05
EU-15
1.00
0.40
0.95
0.90
0.35
Ratio
(Right hand axis)
0.85
0.30
0.80
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
EU-US ratio
Employment-Population Ratio
US
Defining the Four Country Groups,
Pop Share and ALP Growth 1995-2006
Nordic: Denmark, Finland, Sweden
Pop Share: 17
ALP: 2.18
Continental: Benelux, Austria, France, Germany,
Portugal
ALP: 2.09
Anglo-Saxon: UK and Ireland
Pop Share: 5
Pop Share: 49
ALP: 1.75
Mediterranean: Greece, Italy, Spain
Pop Share: 29
ALP: 0.24
A closer look at the Mediterranean
Countries
Mainly driven by Spain and Italy
Spain:
►-4.43 turnaround in Y/H
►+5.04 turnaround in H/N
Italy:
►-2.28 turnaround in Y/H
►+1.16 turnaround in H/N
Had we ranked the countries according to their post1995 annual growth rates of output per capita, Spain
would be a Tiger, behind only Greece and Ireland
Making Sense of Cross-EU
Heterogeneity in Table 1
Notice the homogeneity pre-1995 and heterogeneity
post-’95. Stdev LP 0.63 to 1.0. Stdev H/N 0.46 to
1.02
The only two countries with a noticeable acceleration in
LP are Sweden, Greece and Ireland
Declines < 1% for Finland, UK, Austria, Lux, NL
Sharp declines for Belgium, Denmark, France,
Germany, Portugal, and especially Italy and Spain
We emphasize the experience of the Mediterranean
countries and their contrast with Nordic & AngloSaxon
Research Strategy
Divergence across the EU has increased
The Y/H slowdown in the Med countries is balanced
by healthy H/N growth, which mainly consists of E/N
growth
We will estimate regressions that allow us to determine
how much of the turnaround in E/N growth can be
attributed to policy/institutional variables
Then how much of the productivity slowdown can be
explained by the E/N growth and by policy variables,
separately and together?
Employment Regressions
Cover 1980-2003 EU-15, N=320, population weighted
All variables are rates of changes, not levels
Explanatory Variables:
Output Gap
Product Market Regulation (PMR)
Union Density
Employment Protection Legislation (EPL)
Average Replacement Rate (ARR)
Corporatism Dummy
Tax wedge
Dummies for time shift and for each country
Previous literature – a subset of these variables, levels vs. growth
rates
OECD Product Market Regulation Index
7
Continental
6
Mediterranean
5
4
Anglo-Saxon
Nordic
3
2
1
0
1980
1985
1990
1995
2000
Employment Protection Legislation
4
3.5
Mediterranean
3
Continental
2.5
2
Nordic
1.5
1
Anglo-Saxon
0.5
0
1980
1985
1990
1995
2000
Unemployment Benefits
50
45
Continental
40
35
Nordic
30
25
Mediterranean
20
Anglo-Saxon
15
10
5
0
1980
1985
1990
1995
2000
Taxes in Europe
41
45
Nordic
39
40
37
35
35
Continental
30
Mediterranean
33
25
31
Anglo-Saxon
(right hand axis)
20
29
15
27
25
10
1980
1985
1990
1995
2000
Employment Regression Results
Output Gap
0.52 ***
(0.05)
Product Market
Regulation
-0.44
(0.55)
Union Density
-0.46 ***
(0.10)
Employment
Protection Legislation
0.86
(0.79)
Unemployment
Benefits (ARR)
-0.18 ***
(0.05)
High Corpratism Dummy
-2.04 **
(0.98)
Tax Wedge
-0.28 ***
(0.07)
Post-1995 Dummy
0.94 ***
(0.15)
R2
RMSE
N
0.52
1.18
320
Our tax wedge coefficient is
consistent with what others have
found, -0.3 to -0.45
EPL and PMR seem to have no
significant effects
Everything else has the correct
sign – regulations and taxes
reduce employment
The post-1995 dummy is
substantial
Growth in the employment
rate rose by 1% after ’95 for
unexplained reasons
Employment Regression Results
Robustness
Results are the same if population weights are
dropped or year dummies are added
Dropping the Mediterranean countries or Spain
does not affect the size of the post-1995 dummy
Interpretation of Time Shift Dummy
In mid-1980s there was an enormous disparity in E/N
for females across European countries, ranging from 30
percent in Spain to 70 percent in Scandinavia
Gradually, but especially after 1995, there has been
entry of females into the labor force, esp. in Southern
Europe
A separate literature documents these facts and links
them to changes in cultural attitudes and social norms.
Post-1995 immigration has also contributed to the post1995 time-shift dummy
Employment vs. productivity effects
Employment Regression Results
With all of our dummies, we need to determine
the effects of the policy/institutional variables
holding constant the country and time dummies.
To calculate effects of the policy/institutional
variables, we run counter-factual simulations.
We plot predicted values with policy fixed at its
1995 level
The output gap and dummies are still allowed to
vary
Note: These plots convert growth rates to levels
Female Employment
47
Effect of the
Policy variables (1.75%)
45
Predicted
43
Fixed Policy
41
No Post-1995 Dummy
39
Effect of the
post-95 dummy (2.38%)
37
35
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
Male Employment
75
Effect of the
Policy variables (1.47%)
70
65
Predicted
60
Fixed Policy
Effect of the
post-95 dummy (6.32%)
55
No Post-1995 Dummy
50
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
Productivity Regressions
Suppose we are in a Cobb-Douglas world. What
coefficient would we expect on employment? (Here we
neglect the distinction between H and E and use lower
case letters for logs)
y = 0.33*k + 0.67*h
(y-h) = 0.33*(k/h)
If capital is fixed, the coefficient will be -0.33 and if
capital adjusts, it would be smallerIf labor is not
homogenous it could be larger
The last people to enter the labor force are likely the least
skilled and experienced. This is especially true when
unskilled immigration occurs
Productivity Regressions
Our aim is to regress changes in productivity on
changes in employment per capita and on our
policy/institutional variables.
This allows us to determine which variables affect
productivity directly, and which others only indirectly
through their effects on employment
We can’t simply regress productivity on employment
A shock to productivity affects wages and hence
employment. Alternatively, a technology shock that
raises productivity may lead to layoffs of workers who
are no longer needed
Instrumental Variables in the
Productivity Regressions
Identification using Instrumental Variables
We want variables that affect employment but not
productivity
The tax wedge is our best candidate
We also consider using the post-1995 dummy and
union density
Pragmatism
This gives more power and passes identification tests, but
raises the question as to what caused the post-1995 change as
quantified by the dummy
Productivity Regressions
Employment Rate
-0.64 ***
(0.20)
Output Gap
0.68 ***
(0.11)
Product Market
Regulation
0.56
(0.45)
Union Density
0.03
(0.12)
Employment
Protection Legislation
1.66 ***
(0.65)
Unemployment
Benefits (ARR)
0.14 ***
(0.05)
High Corpratism Dummy
-0.49
(0.94)
Post-1995 Dummy
-0.14
(0.24)
R2
RMSE
N
0.63
0.95
320
Coefficients on policy/inst variables
on productivity are expected to be
positive, the opposite of the negative
coefficients in the employment
regressions
Tax wedge is the only instrument in
this version
Coefficient on employment is twice
what we would expect
EPL and ARR have independent
positive effects on productivity
We can drive the SE on employment
down to 0.10, but the result remains
the same
Not dependent on the Med group of
countries
Level of Labor Productivity
102
100
Fixed Policy
98
96
Policy Effect
– Lowered growth by .25%
per year
94
92
– cumulates to 2.5% decline
in the level
90
– 1/3 of the total shortfall
Predicted
88
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
Effects of the Policy and
Institutional Variables
Assuming hours per employee is stable, E/N +
Y/H = Y/N
Policy has effects on both employment and
productivity
We just add these effects up
Effects of Policy & Institutions
Shock Size
0.9
Employment
-0.14
(0.24)
Productivity
0.35
(0.25)
Output Per Capita
0.21
(0.22)
Union Density
23.32
-7.93
(1.17)
5.07
(1.23)
-2.85
(1.07)
Unemployment
Benefits (ARR)
11.31
-0.90
(0.34)
1.37
(0.31)
0.47
(0.25)
Employment
Protection Legislation
0.87
0.74
(0.36)
0.23
(0.37)
0.97
(0.31)
1
-1.02
(0.48)
0.65
(0.33)
-0.37
(0.21)
9.21
-2.67
(0.64)
1.71
(0.53)
-0.96
(0.4)
Product Market
Regulation
High Corpratism Dummy
Tax Wedge
Tax wedge and union density lower Y/N
ARR and EPL have positive effects
Driven by their direct effects on productivity
Effects of Government Policy
Why would ARR and EPL raise productivity and
output?
Acemoglu and Shimer on reservation wages and
matching
Match quality may improve
More incentive to create job-specific human capital
The New Results in this
Paper at the Industry Level
We aggregate productivity growth by industry in a way
that allows us to determine the relative role of
productivity and shares
The “productivity” effect is just the difference in
productivity growth in a given industry
The “share” effect is the addition or subtraction from
growth as shares shift within industries.
Example: Ireland shifts to high tech manufacturing, this
comes out as a “share” effect within manufacturing
The industry analysis examines both EU vs. US and as
well contrasts among the four EU country groups
Contributions, Productivity vs. Share
Effects, in EU-US, 1995-2003
Manufacturing is nearly as important
as retail
Real estate
Prod
Share
Comm.
Serv.
Finance
Trans.
Retail/wholesale
Non-durables prod
Non-durables share
Manufacturing
ICT prod
Const./utilities
Non-ICT share
Non-ICT prod
But ICT is tiny
Only ~2% hours share
ICT share
Farms/mining
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
ALP growth multiplied by nominal shares
Real Estate
U.S.
Communications
Services
Finance
E.U.
Transportation
Retail/Wholesale
Manufacturing
US acceleration is widespread, not just in retail
and manufacturing.
Construction Utilities
Farms/Mining
EU weakness is also widespread
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
Cross-Industry Correlation of
Y/H and E/N Turnarounds
15
▲ = ICT or Communications
E/N Turnaround (Percent)
10
■ = Med. (non-ICT or Comm)
5
0
-15
-10
-5
0
-5
-10
Y/H Turnaround (Percent)
5
10
15
Table 12: Regressions of LP Turnaround* on E/N
Turnaround*
Exclude ICT
and Comm.
Coefficient
T-Statistic
N
R2
RMSE
All
No
-0.45
-4.35
179
0.10
3.000
All
Yes
-0.54
-5.94
149
0.19
2.495
Mediterranean Only
No
-0.82
-4.19
36
0.34
2.920
Mediterranean Only
Yes
-0.83
-5.60
30
0.53
2.140
Countries
* Turnaround equals 1995-2004 average growth minus 1980-1995 average growth
Comparing Nordic with EU-15
Nordic
ICT
Mfg.
0.2
Nondurable
Retail/
Mfg
Wholesale
Business S ervices
Real Estate
Trans
Construction/
Utilities
Communication
0.1
Finance
0.05
GHI
Non-ICT Durable
0.10
-0.05
Mfg.
-0.15
Ag./Mining
Comparing Anglo-Saxon with EU-15
Anglo-Saxon
0.25
Finance
Business
S ervices
0.15
Communication
Non-ICT Durable Mfg.
Retail/
0.05
Wholesale
Nondurable Mfg.
Real Estate
0.05
Trans
GHI
Construction/Utilities
ICT Mfg.
-0.10
Ag./Mining
Comparing Continental with EU-15
Continental
Ag./Mining
0.10
Retail/
Communication
Wholesale Non-ICT Durable
Nondurable Mfg.
Real Estate
Construction/
Utilities
Business
S ervices
Mfg.
ICT Mfg.
0.05
Trans
-0.05
GHI
-0.15
Finance
Comparing Med with EU-15
Mediterranean
GHI
Business
0.15
Finance
S ervices
Communication
0.05
-0.10
ICT
Retail/
Construction/
Utilities
Mfg.
Wholesale
Trans.
-0.15
Real Estate
Non-ICT Durable
Nondurable Mfg
Mfg.
-0.25
0.05
0.10
Ag./Mining
Comparing US with EU-15
Real Estate
0.35
Finance
Retail/
0.25
US TFP
Wholesale
Business
S ervices
ICT
Non-ICT Durable
Mfg.
Mfg.
0.05
Communication
GHI
Nondurable
Mfg
45º line
-0.05
Trans.
0.05
0.10
-0.05
Ag./Mining
Construction/
Utilities
EU TFP
Conclusions from Employment
and Productivity Growth Regressions
Growing heterogeneity with EU-15 in employment and productivity growth
after 1995.
There is a strong negative correlation between growth in Y/H and E/N
evident in the data, emerging from our regressions, and also in the crossindustry data displayed at the end
At least in short run, lower taxes and looser regulations raise employment
growth and reduce productivity growth
The novelty in our framework is to show that policy changes widely endorsed
in Europe as desirable (Lisbon agenda) may boost E/N at the cost of
reducing Y/H, thus leaving ambiguous effects on growth in output per capita
(Y/N)
A 1% increase in employment only raises output by 0.36% in the short-run
Summary of effects
Unions reduce output per capita
EPL and unemployment benefits raise output per capita
PMR and the tax wedge have roughly no effects
Further Conclusions from CrossIndustry Results
Differences across Europe are in part reflected
in industries that are “national champions”.
Compared to EU average, LP turnaround
reveals
Nordic strong in ICT manufacturing
Anglo-Saxon strong in finance and business services
Continental average as would be expected
Mediterranean weak across the board, consistent
with a broad-based macro explanation rather than an
industry-specific explanation
Final Qualification
The E/N and Y/H regression analysis is static and does not
trace further dynamic adjustment
Negative effect of policy reforms on K/H should in many models be
followed by faster growth in K
This has not happened (yet) in much of Europe
There are fundamental differences in industry performance
between the US and EU that have widely accepted structural
explanations
Wholesale and retail trade, big boxes vs. inner-city pedestrian walking
districts (role of land-use planning as another policy reform)
Other industries, such as finance and business services, require further
study and may involve data comparability issues.