eco-innovations and firms’ market value: a · economic value of inventions (e.g. trajtenberg,...
TRANSCRIPT
Eco-innovations and firms’ market value: a
micro-econometric analysis of European data
Alessandra Colombelli, Politecnico of Turin & BRICK, Collegio Carlo Alberto
Claudia Ghisetti, University of Ferrara & SEEDS
Francesco Quatraro, University of Turin & BRICK, Collegio Carlo Alberto
IAERE 2016, Bologna 11° February 2016
Background Literature
Broad literature investigated market value returns of innovative activities
• R&D is an I activitity that creates an asset, intangible in nature
Market valuation of the firm should:
• i) reflect such intangible capital - «knowledge stock»
• ii) reflect the current present value of expected returns from the invention and/or R&D
(Griliches, 1984)
MV is a function of firms’ bunch of assets
𝑀𝑉 = 𝑓(𝐴𝑖𝑡 + 𝐾𝑖𝑡 + 𝐼𝑖𝑡)
All in all, strong support that knowledge stock increases firm market value, either R&D or patents
2
Background Literature (cont’d)
Whether being «green» pays for the firm is still a debated issue
Horváthová (2010) on the effects of environmental on economic performance:
• 15% negative (Cordeiro and Sarkis, 1987)
• 55% a positive (Al Tuwaijri et al. 2004; Bragdon and Marlin 1972; Dowell et al. 2000; King and Lenox, 2001,
2002; Russo and Fouts, 1997; Salama, 2005; ),
• 30% neutral (Freeman and Jaggi; 1992; Wagner et al., 2002)
Lack of empirical evidence on market valuation of green “knowledge”
Do green Technologies (GT) positively affect MV (as for standard innovations and R&D)?
• Supply: Porter Hypothesis – possible win win efficiency gains from innovation leading to
productivity growth (e.g. Marin, 2014). This should led investors to better evaluate those
firms
• Demand: the increasing stringency of regulatory frameworks is likely to raise the derived
demand of green technologies. Firms generating these technologies are likely to experience
higher sales, and hence to be better evaluated by the market
Overall expectation of a + effect
3
Empirical strategy: Data
1. Data:
‐ ORBIS – Balance sheet data (only firms listed in stock market) from 2002 to
2011
‐ REGPAT september 2015 (Through HAN) – assignement of patents and
ipc(s) to listed firms. Patent applications at EPO from 1985 to 2011
2. Sectors: Manufacturing with high propensity to patent in
environmental technologies
• FIL: Nace Rev2 CE (chemicals); CI (electronics & components); CL
(motor vehicles and trasport equipment)
• FIL2 = FIL + CG (rubber, plastic, glass, ceramic, cement, other non
metallic mineral products) + J (ICT services)
3. Countries: 5 EU countries UK, FR, DE, IT, SE
4
Empirical strategy: Core variables
Market Value (MV) = Tobin’s q index
• Market value (market capitalization) over book value (Tangible fixed assets)
PAT_R&D = Patent yield of R&D (as in Hall et al. 2005)
• Knowledge stock from patent applications at EPO since 1985 on R&D stock
• Perpetual inventory method with 𝜎 =15% (Hall, 1990)
ReD_A= R&D intensity ‐ Ratio of deflated R&D stock and value of intangible assets
‐ PIM with 𝜎 =15% and an annual growth rate of 8%
‐ Initial R&D constructed artificially
• GT= Green technologies stocks ‐ Patent applications at EPO since 1985 in «environmental» technologies
‐ Labeled as «environmental» from two merged alternative classifications: • WIPO IPC Green Inventory (GT_WIPO)
• OECD EnvTech (GT_OECD)
5
Empirical strategy: approach Firm level market value equation – hedonic pricing model
ln (𝑀𝑉)𝑖𝑡 = 𝛼 + 𝛽1𝑃𝐴𝑇_𝑅&𝐷𝑖𝑡 + 𝛽2 𝑅𝑒𝐷_𝐴𝑖𝑡 + 𝛽3𝐺𝑇𝑖𝑡 + 𝜀𝑖𝑡
The good being priced is the firm, the charachteristics its assets (tangible and
intangible)
Approximation log (1+ K/A) log (K/A) and OLS estimation (e.g. Griliches 1981; Jaffe 1986; Cockburn et al. 1988; Hall 1993a, 1993b; Czarnitzki et al 2006):
Tested through several estimation choices:
1. MV equation on only RD reporting firms (possible bias)
2. MV equation on an estimated R&D for non reporting firms (Heckman)
3. MV equation on an estimated R&D for non reporting firms (Wooldridge)
The last two alternatives are based on three equations: ‐ Firms’ decision to engage in R&D activities
‐ The determinants of the amount of R&D activities
‐ market value equation as a function of the estimated level of R&D
6
Inclusion of GT
R&D = R&D expenditures: no green R&D
Green Technologies have been included so far by augmenting the MV
equation by GT stock
𝐺𝑇𝑅&𝐷 =𝐺𝑇𝑠𝑡𝑜𝑐𝑘𝑖𝑡
𝑅𝐷𝑠𝑡𝑜𝑐𝑘𝑖𝑡
• By creating an alternative PATstock (PAT(noGT)) that excludes GTs
from the count of total patent applications
7
Empirical strategy: correcting for sample
Estimation of R&D predicted
Heckman (1979) • Decision to engage in R&D activity as a first step on ii)
• Ii) Decision on the amount of I in R&D, depending on Intangible assets Y* and Dsec*
• Exclusion restriction part of a group (GROUP) and the size
• Pooled OLS on panel data
Wooldridge (1995) specific for panel data:
• 1° step captures time-invariant effects through time averages of core variables;
• the time-invariant effects are assumed to be linked with x through a linear function on the time averages and an orthogonal error term which exhibits no variation over time
• 2° step predicts R&D investments – bootstrap se, clustered
Pwcorr RD & RD_hat: 0.6571 (Heckman) 0.9647 (Wooldridge)
1419 firms with R&D reported; 2250 final sample 37% R&D correction
8
Descriptive stats
9
n Mean S.D. Min Max n Mean S.D. Min Max n Mean S.D. Min Max
lnTQ 2330 -5.0 1.7 -10.5 3.0 879 -4.97 1.86 -10.54 3.04 1451 -5.03 1.65 -10.03 2.10
Capitalization 2330 1833.0 6750.9 0.3 85891.3 879 150.45 849.87 0.27 17422.94 1451 2852.20 8367.15 0.55 85891.31
T.Fixed Assett 2330 696647.5 3368531.0 1.0 69700000.0 879 54864.22 266095.70 1.00 4440838.00 1451 1085433.00 4216848.00 2.92 69700000.00
RD 1451 170.7 587.3 -50.3 5649.0 0 . . . . 1451 170.67 587.26 -50.31 5649.00
PATstock 2330 70.7 518.4 0.0 7952.0 879 3.05 33.69 0.00 618.82 1451 111.76 653.09 0.00 7951.96
PAT(NoGT)stock 2330 63.8 478.8 0.0 7224.7 879 2.10 19.13 0.00 332.11 1451 101.24 603.57 0.00 7224.73
Gtstock 2330 14.8 108.2 0.0 1466.4 879 1.55 22.86 0.00 429.56 1451 22.79 135.31 0.00 1466.39
RED_hat2 2165 673.8 2965.8 0.0 32584.3 815 40.27 101.62 0.17 1082.56 1350 1056.31 3703.32 0.03 32584.25
RED_hat 2266 108.8 460.7 -54.5 5649.0 815 7.93 16.05 0.04 139.15 1451 165.43 567.85 -54.47 5649.00
Full Sample R&D missing R&D reporting
Results, MV Equation no correction
10
(1) (2) (3) (4) (5) (6)
lnTQ lnTQ lnTQ lnTQ lnTQ lnTQ
PAT/R&D 0.0552*** 0.0514***
(0.0093) (0.0090)
PAT(noGT)/R&D -0.0091 -0.0078
(0.0254) (0.0234)
R&D/ASSET 11.8782*** 11.8850*** 11.8870*** 10.3562*** 10.3610*** 10.3599***
(1.9971) (2.0013) (2.0006) (1.6666) (1.6688) (1.6692)
GT/R&D 0.2252*** 0.1983*** 0.1885*** 0.2116***
(0.0727) (0.0173) (0.0171) (0.0676)
lSIZE -0.1671*** -0.1666*** -0.1664*** -0.1973*** -0.1969*** -0.1970***
(0.0166) (0.0166) (0.0165) (0.0127) (0.0127) (0.0127)
_cons -3.5733*** -3.5895*** -3.5905*** -3.7895*** -3.7939*** -3.7924***
(0.2647) (0.2647) (0.2645) (0.1944) (0.1941) (0.1942)
N 1419 1419 1419 2604 2604 2604
Dtime & Dsector incl Incl. Incl. Incl. Incl. Incl.
adj. R2 0.4421 0.4438 0.4442 0.5077 0.5087 0.5085
Sectors FIL1 FIL1 FIL1 FIL2 FIL2 FIL2
Standard errors in parentheses* p < 0.10, ** p < 0.05, *** p < 0.01
FIL1= CE, CI, CL; FIL2= FIL1 + CG + J (NACE Rev2)
Results – RD_hat Heckman .
11
(1) (2) (3) (4) (5) (6) lnTQ lnTQ lnTQ lnTQ lnTQ lnTQ
PAT/R&D_hat 0.0712* 0.0201*
(0.0418) (0.0107)
PAT(NoGT)/R&D_hat -0.0190 0.0529*
(0.0363) (0.0313)
R&D_hat/Asset 1.2400* 1.2408* 1.2412* 1.6025*** 1.6013*** 1.6038***
(0.7166) (0.7169) (0.7169) (0.5161) (0.5157) (0.5163)
GT/R&D_hat 0.5233*** 0.4576*** 0.0194** -0.0125
(0.1780) (0.1080) (0.0088) (0.0185)
lSIZE -0.2119*** -0.2102*** -0.2102*** -0.2118*** -0.2121*** -0.2114***
(0.0153) (0.0153) (0.0153) (0.0112) (0.0112) (0.0112)
Constant -2.9701*** -3.0095*** -3.0088*** -3.0536*** -3.0455*** -3.0666***
(0.2393) (0.2389) (0.2389) (0.2623) (0.2623) (0.2625)
N 2250 2250 2250 5420 5420 5420 R2 0.275 0.277 0.277 0.388 0.387 0.388
adj. R2 0.2708 0.2732 0.2735 0.3860 0.3857 0.3863
Dsec & Dtime Incl Incl Incl Incl Incl Incl Sample FIL FIL FIL FIL2 FIL2 FIL2
R&D_hat Heckman Heckman Heckman Heckman Heckman Heckman
Results – RD_hat Wooldridge.
12
(1) (2) (3) (4) (5) (6)
PAT/R&D_hat2 0.0534 0.0107**
(0.0350) (0.0052)
PAT(NoGT)/R&D_hat2 -0.0140 0.0356
(0.0326) (0.0218)
R&D_hat2/Asset 2.0665* 2.0653* 2.0660* 2.6631** 2.6619** 2.6643**
(1.1984) (1.1976) (1.1975) (1.0878) (1.0874) (1.0882)
GT/R&D_hat2 0.4387*** 0.3887*** 0.0105*** -0.0101
(0.1634) (0.1034) (0.0034) (0.0121)
lnSize -0.2085*** -0.2072*** -0.2072*** -0.2164*** -0.2166*** -0.2160***
(0.0166) (0.0166) (0.0166) (0.0130) (0.0130) (0.0130)
Constant -3.0438*** -3.0760*** -3.0760*** -2.8883*** -2.8837*** -2.8986***
(0.2562) (0.2560) (0.2559) (0.2790) (0.2789) (0.2790)
N 2165 2165 2165 5075 5075 5075
R2 0.291 0.293 0.293 0.400 0.400 0.400
adj. R2 0.2870 0.2888 0.2891 0.3984 0.3983 0.3986 Dsec & Dtime Incl Incl Incl Incl Incl Incl
Sample FIL FIL FIL FIL2 FIL2 FIL2
R&D_hat Woold. Woold. Woold. Woold. Woold. Woold.
Extensions - Citation
Account for the number of citations received by PAT and GT (forward
citations) technological importance of the patent and reflects the
economic value of inventions (e.g. Trajtenberg, 1990; Hall, et al., 2005).
• To limit truncation problems we use the OECD forward citation index (5 years)
(Squicciarini et al. 2013)
• Limitation: 18 months on avg for patent to be granted + 5 y to be cited
• last valid year would be 2004 our dataset would span from 2002-2004.
• No significant result found
13
Extensions - Regulation
Account for the role environmental regulation can have in driving Gt
positive effects on MV
• The hypothesis tested is that market positively evaluates GT because of the
existence of stringent policy target firms should commit to
• Inclusion of a policy stringency variable POL and the interaction between POL
and GT (POL*GT)
• Data: WIOD – Environmental accounts
• Proxy sector-country- year log(CO2/VA) in t-1
• We loose 2 years of observations due to lack of data on CO2 emissions for
2011 and 2012
14
Extensions – Policy (cont’d)
15
(1) (2)
PAT(NoGT)/R&D_hat2 -0.0347* -0.0367
**
(0.0185) (0.0174)
POLlag -0.1646***
-0.1548***
(0.0531) (0.0550)
POLlag*GT -0.2180
(0.1960)
R&D_hat/Asset 10.8413***
10.8531***
(1.7441) (1.7519)
GT/R&D_hat2 0.6217***
0.4113*
(0.1281) (0.2316)
lnSize -0.1642***
-0.1647***
(0.0180) (0.0180)
Constant -3.3051***
-3.2776***
(0.3147) (0.3175)
N 951 951
R2 0.410 0.411
adj. R2 0.4021 0.4019
Sample FIL FIL
Year T< 2010 T<2010
R&Dpredict Wool Wool
Preliminary conclusion
Confirmed hypothesis on the expectations of a market which
positively evaluates «environmental» knowledge stock
• Environmental innovations might be self-sustained in the market – even in the
absence of a policy stimulus
Patent quality seems not to play any role in market valuation of
firms environmental attitude
Environmental regulation plays a role but with an unexpected sign
• Can it be the wrong policy? Policy on sectors with which high inter-
relatedness can better explain it? (Future extension)
16