Download Does foreign environmental policy influence domestic innovation

Does foreign environmental policy
influence domestic innovation?
Evidence from the wind industry
Antoine Dechezleprêtre, LSE
Matthieu Glachant, Mines ParisTech
Motivation

The empirical literature has extensively shown that
environmental regulation fosters innovation in
environmentally-friendly technologies



Jaffe & Palmer, 1997; Brunnermeier & Cohen, 2003 ; Newell et al,
1999; Popp, 2002
But most papers assess the impact of domestic regulation
on domestic innovation
Innovators are likely to be influenced not only by domestic
but also by foreign policies

The market for technologies is global: 49% of wind power patents
are filed by non-residents
The literature on cross border
effects of policies

A few descriptive papers, but a lack of
econometric analysis:



Lanjouw & Mody (1996): vehicles emissions
regulations in the US spurred innovation in Germany
and Japan
Popp (2006): Inventors of air pollution control devices
for coal-fired power plants do not respond to foreign
environmental regulation
Popp, Haffner, Johnstone (2007) : domestic and
foreign regulation influence innovation in chlorine-free
technology in the pulp and paper industry
Research questions

Does foreign wind policies encourage domestic
innovation?


If so, the empirical literature under-estimates the
overall effect of environmental policies
We distinguish between:

Demand-pull policies encouraging the deployment of
new turbines


Feed-in tariffs, renewable portfolio standards…
Technology-push policies encouraging innovation

Public support to R&D
Policy relevance


Environmental policies are also sold by
politicians as a tool to achieve technological
leadership of domestic firms
But domestic policies might also help foreign
competitors

Ex: Spain’s solar policy has strongly benefited Chinese
producers
Wind power today



160 GW of worldwide installed capacity in 2009
+30% annually since 2000
2% of global electricity generation


Some countries with high levels of penetration:
Denmark (20%), Portugal and Spain (14%), Ireland
(11%), Germany (8%)
A policy driven development




Guaranteed tariffs (feed-in tariffs)
Investment tax credits
Mandatory production quotas
R&D tax credits
Data


A panel of 28 OECD countries, 1995-2006
Innovation



Demand-pull policies


Patent data from the EPO World Patent Statistical
database (PATSTAT)
15,835 patent applications
Annual installations of new wind power capacities
(IEA)
Technology-push policies

Public R&D expenditures (IEA)
Econometric framework

We estimate the following equation:
log  N i ,t   1 log capi*,t 1   2 log cap* i ,t 1
 1 log rdi ,t   2 log rd  i ,t
  X i ,t   t   i   i ,t




Nit is the number of patents filed by inventors from country
i in year t
capi,t+1* is the expectation in year t of the capacity which
will be added in country i in year t + 1
cap-i,t+1* is the the same for capacities in the rest of the
world
rdi and rd-i are public R&D expenditures at home and
abroad
Control variables





Knowledge stocks: discounted stock of previous patents
Time trend (time dummies not possible)
Global level of IPR
GDP
Country fixed-effects
Why expected installations?



Because inventors’ decisions are based on expectations
about future demand
Problem: Expectations are not observed in the data
Solution: We assume the innovators form adaptive
expectations, so that the expected cap is current cap +
weighted average of past increases in installations :
capi*,t 1  capi ,t  (1   )  k (capi ,t  k  capi ,t  k 1 )
k 0
Endogeneity of rdi,t

Domestic R&D expenditures are inputs of the innovation
process


A simultaneity problem
We instrument rdi,t with R&D expenditures in solar power,
biomass and geothermal


uncorrelated with wind innovation because very different
technologies
correlated with wind R&D because jointness of policy decisions
Results
(1)
(2)
IV OLS
OLS
log capi*,t 1
0.0505**
(0.0224)
0.0546**
(0.0234)
0.0586***
(0.0207)
log cap *i ,t 1
0.4849***
0.4771***
0.4396***
(0.1625)
0.3853**
(0.1630)
-0.0508
(0.3669)
0.3576**
(0.1669)
(0.1556)
0.1319
(0.1052)
-0.2943
(0.3919)
0.4842***
(0.1770)
(0.1677)
0.0127
(0.0700)
-0.6209*
(0.3212)
0.3759***
(0.0985)
yes
yes
yes
336
28
yes
yes
yes
336
28
yes
yes
yes
336
28
log rdi ,t
log rd i ,t
log K i ,t
Controls for GDP & IPR
Country fixed effects
Time trend
Observations
Countries
(3)
Negative
binomial
Tests

First-stage model statistics of the IV – OLS model


Test of overidentifying restrictions



The Hansen J statistics = 0.198
Hence, the instruments satisfy the orthogonality conditions.
Test for the endogeneity with a “difference in Sargan”
statistics which is robust to clustering.


Instruments are relevant. The Shea partial R-square is 0.2085
and the cluster-robust F-statistics of joint significance of the
instrument variables is 5.33.
Endogeneity hypothesis is weakly rejected (the p-value of the
Chi-square test is 0.1303).
Autocorrelation

Use robust standard errors
Results
(1)
(2)
IV OLS
OLS
log capi*,t 1
0.0505**
(0.0224)
0.0546**
(0.0234)
0.0586***
(0.0207)
log cap *i ,t 1
0.4849***
0.4771***
0.4396***
(0.1625)
0.3853**
(0.1630)
-0.0508
(0.3669)
0.3576**
(0.1669)
(0.1556)
0.1319
(0.1052)
-0.2943
(0.3919)
0.4842***
(0.1770)
(0.1677)
0.0127
(0.0700)
-0.6209*
(0.3212)
0.3759***
(0.0985)
yes
yes
yes
336
28
yes
yes
yes
336
28
yes
yes
yes
336
28
log rdi ,t
log rd i ,t
log K i ,t
Controls for GDP & IPR
Country fixed effects
Time trend
Observations
Countries


(3)
Negative
binomial
Local & foreign regulation influence innovation
Foreign public R&D has no impact
Marginal effects of demand




+1 MW at home
+1 MW abroad
A factor of 28
= + 0.0223 invention
= + 0.0008 invention
There exist barriers to technology diffusion
which discourage innovation

Policies to increase technology transfer also increase
innovation
The total effect of 1 MW



Our data set includes 28 countries
Hence 1MW induces 0.0008 * 27 ≈ 0.022
foreign invention worldwide
Demand-pull policies have a comparable
aggregate impact on domestic and foreign
innovation
Marginal effects of domestic R&D

Marginal effects of domestic public R&D
expenditures

+1 million USD = + 0.82 invention
Tech-push vs demand-pull

The innovation impact of 1 million USD:
Effect of 1 million USD spent
on:
Public R&D
expenditures
New capacities
Home
0.82 invention
0.03 invention
Abroad
0
0.03 invention
Global
0.82 invention
0.06 invention
Assumptions: Installing 1 MW costs 0.65 million USD relative to conventional electricity

Innovation has to be seen as an ancillary benefit of
demand-pull policies
Conclusion


Foreign demand policies influence innovation
The marginal effect is 28 times less than the effect of
domestic installations


But the total effect of foreign installations is the same as
that of domestic installations



Removing barriers to international diffusion is key
Existing studies strongly underestimate the overall impact of env’l
policies on innovation
Foreign public R&D has no significant impacts
1 million USD spent in subsidizing new turbines induces
14 times less patents than 1 million USD of domestic
R&D
Robustness checks (1)
log capid,t *
log capif,t *
log rdid,t
log rdi,ft
log Ki ,t
GDP, IPR
Country FE
Time trend
Observatio
ns
Countries
(1)
(2)
(3)
(4)
(5)
λ = 0.65
λ = 0.85
 = 0.15
 = 0.2
Drop Japan
0.0491**
(0.0210)
0.4601***
(0.1560)
0.3835**
(0.1628)
-0.0885
(0.3711)
0.3582**
(0.1671)
0.0523**
(0.0235)
0.5112***
(0.1694)
0.3895**
(0.1627)
-0.0205
(0.3641)
0.3563**
(0.1665)
0.0473**
(0.0216)
0.4739***
(0.1590)
0.3440**
(0.1535)
-0.0730
(0.3566)
0.3919***
(0.1428)
0.0444**
(0.0211)
0.4617***
(0.1561)
0.3182**
(0.1456)
-0.0798
(0.3478)
0.3998***
(0.1225)
0.0442*
(0.0227)
0.4714***
(0.1626)
0.2937*
(0.1590)
-0.0486
(0.3749)
0.3839**
(0.1737)
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
336
336
336
336
324
28
28
28
28
27
Robustness checks (2)
log capid,t *
log capif,t *
log rdid,t
log rdi,ft
log Ki ,t
GDP, IPR
Country FE
Time trend
Observatio
ns
Countries
(1)
(2)
(3)
(4)
(5)
R&D t-1
R&D t+1
t2
log(t)
add K-i
0.0535**
(0.0252)
0.4733***
(0.1475)
0.0858
(0.1882)
0.1362
(0.4079)
0.5267***
(0.1753)
0.0516**
(0.0243)
0.5131***
(0.1616)
0.3434**
(0.1497)
0.3247
(0.3524)
0.3215*
(0.1708)
0.0505**
(0.0224)
0.4846***
(0.1624)
0.3853**
(0.1630)
-0.0491
(0.3676)
0.3577**
(0.1669)
0.0505**
(0.0224)
0.4851***
(0.1626)
0.3852**
(0.1630)
-0.0526
(0.3661)
0.3576**
(0.1669)
—
—
—
—
0.0475**
(0.0214)
0.3853**
(0.1856)
0.3779**
(0.1577)
0.1774
(0.4594)
0.3539**
(0.1683)
-0.9120
(0.9673)
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
336
331
336
336
336
28
28
28
28
28
Thank you