Chih-Hai Yang, National Central University Yu-Hsuan Tseng , National Central University

112/04/21 1

Environmental Regulations, Induced R&D, and Productivity:

Evidence from Taiwan’s Manufacturing Industries

Chih-Hai Yang, National Central University

Yu-Hsuan Tseng , National Central University

Paper presented at Thammasat University September 7, 2011, Bangkok, Thailand

112/04/21 2

Q: Do you agree with the point that the government

should enact stringent environmental regulations??

A picture taken in China

112/04/21 3

But, if it comes at a cost of a lower economic growth….

Do you still prefer stringent environmental regulations??

→ Three main issues in Environmental Economics1. Environmental Kuznets Curve (EKC) The relationship between per capita income and the

use of natural resources and/or the emission of wastes has an inverted U-shape.

→ At relatively low levels of income the use of natural resources and/or the emission of wastes increase with income. Beyond some turning point, the use of the natural resources and/or the emission of wastes decline with income.

112/04/21 4

Thailand ??

Source: Dasgupta, S., B. Laplante, H. Wang and D.Wheeler (2002), Confronting the Environmental Kuznets Curve, Journal of Economic Perspective, 16(1), 147-168.

112/04/21 5

2. Porter hypothesis Porter and van der Linde (1995, JEP) claim that well-

designed environmental policies can lead to improved competitiveness in firms through efficiency improvement and induced innovations.

→The stringent environmental regulations can reach a win-win situation that an economy can reach both goals of a cleaner environment and competitiveness simultaneously.

→ Technological change is one of predominately factors for solving long-term environmental problems and serves as the prerequisite of supporting the argument of Porter hypothesis.

3. Pollution Haven Hypothesis You can investigate foreign firms’ pollution emission compared

with local firms in Thailand.

112/04/21 6

Outlines of the Presentation

1. Motivation and Aims

2. Literature Review

3. Development of Environmental Regulations and Innovations in Taiwan

4. Empirical Specification and Data

5. Empirical Results and Discussion

5. Concluding Remarks and Policy Implications

112/04/21 7

Aims of this study

This study examines the R&D enhancement effect of environmental regulations, its consequently indirect influence on productivity, and the direct relationship between regulatory stringency and productivity in Taiwan’s manufacturing industries.

112/04/21 8

Why this study?The question regarding the influences of

environmental regulations on productivity is a long-standing issue in environmental economics, why this study is interesting??? What’s new ??

1. New evidence from (NIEs),Taiwan. Taiwan’s postwar economic miracle is a very familiar one in

development economics literature, but her rapid development of manufacturing industries along with lax environmental regulations leads to an increasingly worsened environment.

Taiwan’s government has devoted more efforts in ruling environmental regulations since the 1980s, especially since the early 1990s.

Taiwan has excellent innovation performance, e.g. R&D/GDP ratio is 2.94 % in 2009. placed 4th in U.S. patents since 2003.

Q: Is there any part of the increasing innovations attributed to the stringency of environmental regulations?

112/04/21 9

2. Previous studies generally adopt productivity as a proxy of “competitiveness” to test the Porter hypothesis

Direct effect:

environmental regulations

immediately, higher production cost ( --)

adjust the composition of inputs, use pollution control capital (+)

productivity (???)

Indirect Effect

environmental regulations Innovations productivity

Q: The measure of productivity?? (endogenous problem)

→ the total factor productivity (TFP) developed by Levinsohn and Petrin (2003)

112/04/21 10

3. The measurement of environmental regulations ??

In literature:

“Pollution Abatement and Control Expenditures” (PACE).

→ Either Capital or Operating cost

This study:

We have both PACE on capital and fees.

The PACE on capital can be identified into PACE on waste water, waste gas, wastes disposal pollution, and noise pollution capital.

→The detailed information helps obtain insightful analyses and lend policy implications.

112/04/21 11

Literature Review Testing the Porter hypothesis from various aspects, such

as regulatory effects on innovations, productivity, investment, and profits. Survey see…

Ambec and Barla (2006), and Brannlund and Lundgren (2009)

1. Environmental regulations and R&D activities Lanjouw and Mody (1996) RP, cross country evidence 1970s~1980s + Jaffe and Palmer (1997) RE&Stat panel dataset of US manufacturing industries 1973-1991 + Hamamoto (2006) ResourEE Japanese industries 1960s-1970s + OECD firm-level evidence Arimura et al. (2007), Frondel et al. (2007), and Lanoie et al.

(2007) → In sum, positive relation

112/04/21 12

Literature Review2. Environmental regulations and

productivity ◎ Direct test Lanoie et al. (2001) JPA: Canada case + Berman and Bui (2001) RE&Sta. LA case + Managi et al. (2005) Land E, Gulf of Mexico + Barbera and McConnell (1990) JEEM five American emission-intensive industries – Gray and Shadbegin (2003) JEEM U.S. pulp and

paper mills -- ◎ Indirect Effect Hamamoto (2006) ResourEE Japanese industries +

112/04/21 13

Development of Environmental Regulations and Innovations in Taiwan

◎ Development of Environmental Regulations 1. economic take-off in the late-1960s. 2. The mid-1970s, industrial structure transformed

to capital-intensive and as energy-consuming industries, such as metal, petrochemical, synthetic fiber, and electronics industries.

→ lax environmental regulations, resulting in a rapid increase in emission of various pollutions, worsening the environmental quality.

→ debate on the relationship between economic growth and environmental quality emerged in the late 1970s (Shaw, 1994).

112/04/21 14

◎ Development of Environmental Regulations The government began to enact environment

statutes 1974: Water Pollution Control Act Waste Disposal Act 1975: Air Pollution Control Act 1983: Noise Control Act 3. 1980s, the enforcement of laws is weak a. economic growth remained as the primary

national target for Taiwan in the 1980s b. each of the aforementioned environmental

statutes only have been minor revised one time or none in the 1980s.

112/04/21 15

4. The 1990s: toward stringent environmental regulations

Turning point: RCA event was disclosed in 1989 (this event lasts

…) Mon (2002) argues that this event can be

attribute to … the failure of government regulation, lack of corporate self-regulation, lack of public concern about corporate crime, the low self-control tendency of corporate

manager. → This event deeply awakened the environmental

awareness for both the public and government.

112/04/21 16

◎ What is RCA event Another important issue in Environmental Economics: → Pollution Haven Hypothesis development of this event1. American electrical appliances company (RCA) was

approved to invest in Taiwan in 1966 and began to operate in 1968.

2. The award-winning “model factory” in Taiwan. 3. RCA was merged by GE in 1986 and then resold to

Thomson (France) in 1988 but the contract: only buy factory, not pollution Both of them know something inside…..??? 1989, Thomson Inc. dug wells to draw water and soil

to examine ….. Shock the Thompson and it decide to sell the land During the 1968-1988, RCA emitted toxic chemical

pollutions.

112/04/21 17

4. 1998 EPA announced that that area can not be rebuilt “permanent polluted area”

That region began to have piped-water in 1986 in the entire region

so… the water they drank during 1968-1988 the rice people ate during that period… In 1998, some former RCA employees organized

the “self-help union” and reported victim statistics (RCA employees):

1059 people got cancer, 216 people died, and…. 2002, this union sought for international help in

U.S. 2004, this union sued RCA for compensation… Nobody gets compensation until now

112/04/21 18

◎ Development of Environmental Innovations The categories of five environmental friendly

technologies is referred to Hascic et al. (2008): five categories

112/04/21 19

◎ Discussions 1. The number of issued patents for various technologies

increased sharply since the early 1990s. 2. waste disposal and noise control technologies place the

first and second respectively in terms of patent numbers, whereas issued patents of environmental monitoring technology remained as least.

3. Waste Disposal Act was revised in 1991, 2001, and 2004. 4. Noise Control Act was revised in 2003. Urbanization, noise technology can be widely adopted in

various electronics instrument. 5. Air Pollution Control Act was modified in 1999, 2002, and

2006.

In sum, it seems to suggest a positive relationship between the stringency of environmental regulations and innovations

112/04/21 20

Empirical Specification and Data ◎ Empirical Specification The two-step approach developed in Hamatomo (2006). First-step: environmental regulation influences R&D

expenditure

PAF: pollution abatement fees PACE: Pollution Abatement and Control Expenditures on

capital CR4: 4-firm concentration ratio; FOR: foreign share (??) GROWTH: sale growth (%); SIZE: industry size (number of

emp.) CAP: capital to labor ratio; PROFIT: profitability (%) TI: technology imports; EXPR: export ratio (%)

ititititi

titititi

titititiit

uEXPRTIPROFIT

CAPSIZESIZEGROWTH

FORCRPACEPAFRD

1,111,101,9

1,82

1,71,61,5

1,41,31,21,10

ln

lnlnln

4lnlnln

112/04/21 21

◎ Empirical Specification second-step: induced R&D on productivity

RD1: induced R&DRD2: total R&D – RD1DH: dummy variable which equals unity if an industry

belongs to energy-intensive and/or pollution-intensive industries

TFP: Levinsohn and Petrin (2003). Their semi-parametric approach can well deal with problems of endogeneity and unobservable heterogeneity in determining inputs.

, 1 1ˆ1 / and 2 = 1it PAF it t it it it it itRD PAF PAF RD RD RD RD

itiitit

iitititit

TIMEDCRTI

DHRDRDRDTFP

4ln

)1(ln2ln1lnln

54

3210

112/04/21 22

Econometric strategy

Panel data model

Fixed effect model

112/04/21 23

◎ Data 1. the manufacturing-plant surveys (hereafter

MPS), 1997-2000 and 2002-2003. 2. the MPS included slightly more than 81,000

plants in 1997-2000 and over 73,000 plants in 2002-2003, accounting for about 55% of total firms, on average.

3. 234 four-digit industries for 6 years, or a sample of 1,404 observations. Due to the use of one-year lagged variable, the used observation number reduces to 1,170.

4. export intensity and foreign ownership ratio obtained from the Industrial and Commercial Census in years 1996 and 2001.

112/04/21 24

Table 1 Definitions and summary statistics of all variables

Variable Definitions Mean S.D.

RD industrial R&D expenditures (NT$ million) 540.278 2325.684

PAF Pollution abatement fees: costs and expenditures related to abatement operation, maintenance, supervision, test

and inspection (including personnel expenses), and also pollutant emission fees, etc. (NT$ thousand)

62215.09 196914.5

PACE Total pollution abatement capital expenditures: the sum of waste water, waste gas, wastes disposal, and noise

pollution abatement capital expenditures. (NT$ thousand)

54943.04 248579

Water Waste water pollution abatement capital expenditures (NT$ thousand) 23469.24 104710.9

Gas Waste gas pollution abatement capital expenditures (NT$ thousand) 22424.84 147410.4

Wastes Wastes disposal pollution abatement capital expenditures ((NT$ thousand) 7052.679 31580.03

Noises Noise pollution abatement capital expenditures (NT$ thousand) 1871.542 9045.661

CR4 Four-plant concentration ratio 0.433 0.240

FOR Foreign ownership ratio (%) 3.816 5.094

GROWTH Growth rate of industry sales (%) 23.330 128.023

SIZE Number of employees in an industry (person) 9456.615 13957.5

CAP Capital intensity: Ratio of capital to labor employed (NT thousand/employee) 220.7783 380.5595

TI Technology import expenditures (NT$ million) 183.213 1185.539

EXP Export intensity: ratio of exports to industrial sales (%) 24.170 20.656

PROFIT Profitability: (Sales - Operating expenditures)/Sales (%) 6.764 9.819

TFP Total factor productivity: measured using methods in Levinsohn and Petrin (2003)

(NT$ thousand )

14358.34 12551.49

VA Value added: measured by output minus intermediate inputs (NT$ million) 11885.57 32733.45

DH Dummy variable: energy- and pollution- intensive manufacturing industries defined by Taiwan EPA (Iron and

steel, petrochemicals, electronics, textiles, pulp and paper, and cement)

0.28205 0.45016

112/04/21 25

Empirical ResultsVariable Model 1

(FE) Model 2

(FE) Model 3

(FE) Model 4

(FE) Model 5

(FE) Model 6

(FE) lnPAF 0.098*** 0.101*** 0.098*** 0.097*** 0.094*** 0.097*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) lnPACE -0.003 -0.010 (0.874) (0.663) lnWater -0.011 (0.548) lnGas 0.006 (0.744) lnWaste 0.031* (0.097) lnNoise 0.018 (0.292) CR4 0.910* 0.887* 0.922* 0.895* 0.852 0.900* (0.087) (0.096) (0.082) (0.091) (0.107) (0.089) FOR -0.030 -0.029 -0.030 -0.030 -0.028 -0.029 (0.119) (0.128) (0.116) (0.120) (0.134) (0.127) MARKET 0.003*** 0.003*** 0.003*** 0.003*** 0.003*** 0.003*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) lnSIZE 2.783*** 2.758*** 2.771*** 2.799*** 2.881*** 2.830*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) [lnSIZE]2 -0.113*** -0.113*** -0.112*** -0.115*** -0.123*** -0.118*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) lnCAP 0.096 0.097 0.096 0.087 0.077 0.088 (0.223) (0.224) (0.219) (0.264) (0.324) (0.259) lnTI 0.032* 0.036** 0.032* 0.031* 0.031* 0.031* (0.061) (0.039) (0.058) (0.065) (0.067) (0.068) EXPR -0.002 -0.001 -0.002 -0.002 -0.001 -0.002 (0.776) (0.839) (0.781) (0.771) (0.805) (0.784) PROFIT -0.0090 -0.009 -0.009 -0.009 -0.009 -0.009 (0.182) (0.198) (0.170) (0.188) (0.191) (0.188) Time dummy No Yes No No No No R-square 0.577 0.580 0.5776 0.5761 0.5729 0.5739

Hausman test 173.34*** 168.46*** 177.44*** 172.62*** 178.95*** 179.90*** F-test all α i =0 5.22*** 5.19*** 5.21*** 5.22*** 5.25*** 5.22***

Observations 1170 1170 1170 1170 1170 1170

112/04/21 26

Findings1. PAF variable is positive and statistically significant at the 1 %

statistical level which is consistent with findings in Brunnermeier and Cohen (2003) and Gray and Shadbegian (2003) that stronger environmental regulations do stimulate firms to increase their R&D investment. .

2. PACE is not statistically significant, implying that no significant evidence reveals that more pollution abatement capital expenditures (PCAE) can induce R&D expenditure. This finding contradict findings for the U.S. (Jaffe and Palmer, 1997) and Japan (Hamamoto, 2006).

Why fees rather than capital?? owing to the Statute for Upgrading Industries (SUI) → Article 6, investment in pollution control equipment may

credit 7% of the equipment expenditure against the amount of profit-seeking enterprise income tax payable for the current year.

Imported pollution control equipment is also be exempt from import duties.

Note: happens generally one-time for a long period, Fees every year

112/04/21 27

3. Estimates in models (3) to (6) of Table 2 show that the coefficient of (lnWASTE) is positive and statistically significant, implying that stringency environmental regulation in waste disposal seems to effectively induce more R&D activity.

→ This finding echoes the development depicted in Figure 1

4. the coefficient for TI is positive and significant in all estimates, suggesting a complementary relationship between technology imports and in-house R&D

112/04/21 28

◎Impact of Environmentally Induced R&D on productivityTable 3 displays the ratios of environmentally induced

R&D to total R&D over the period of 1998-2003

1. the corresponding shares to R&D spending range from 1.141% to 6.135%, reaching an average of 4.267%.

2. Inducement effect is high or low???? No study R&D tax credit in 2001-2005: 5.30% to 13.91%

year RD1 / RD (%)

1998 4.205

1999 5.886

2000 6.135

2002 1.141

2003 3.967

112/04/21 29

Table 4 shows a series estimates of the influence of induced R&D

on industrial productivityVariable Model 1 (FE) Model 2 (FE) Model 3 (FE)

lnRD1 0.007*** 0.009*** 0.007***

(0.004) (0.002) (0.009)

lnRD2 0.021*** 0.022*** 0.018***

(0.000) (0.000) (0.000)

lnRD1*DH -0.006 -0.006

(0.169) (0.152)

lnTI 0.027***

(0.000)

CR4 -0.011

(0.920)

YEAR-99 0.060** 0.061** 0.058**

(0.027) (0.025) (0.027)

YEAR-00 0.135*** 0.133*** 0.123***

(0.000) (0.000) (0.000)

YEAR-02 0.179*** 0.179*** 0.149***

(0.000) (0.000) (0.000)

YEAR-03 0.215*** 0.213*** 0.191***

(0.000) (0.000) (0.000)

R-square 0.286 0.276 0.434

Hausman test 7863.56*** 9549.04*** 135.98***

F-test all δ i =0 15.25*** 15.27*** 13.70***

Observations 1170 1170 1170

112/04/21 30

Findings1. the coefficients of lnRD1 is positive and significant at the

1% statistical level in models (1) to (3), implying that

environmentally induced R&D indeed contributes to promote productivity in Taiwan’s manufacturing industries.

→ consistent with the Japan case in Hamamoto (2006) →supports the Porter hypothesis of a possible “win-win”

situation.2. the estimated magnitude of RD1 coefficient is substantially

lower than that of RD2, suggesting that the environmentally induced R&D doe not effectively contribute to promoting productivity compared with the scheduled R&D.

3. It indicates no significant evidence is found that energy-/pollution- intensive industries benefit more on promoting productivity brought about by environmentally induced R&D expenditures.

112/04/21 31

◎The direct effect of environmental regulations on productivity

Referring to Gray and Shadbegian’s (2003) specification and

equation (2)

Two measures of TFP1. Levinsohn and Petrin (2003) TFP.2. simple, value-added

itiitit

titiitit

TIMEDCRTI

PACEPAFRDTFP

4ln

lnlnlnln

54

1,31,210

112/04/21 32

Table 5 Impacts of PACE on productivityDep. Variable lnTFP (FE) lnTFP (FE) lnVA (FE) lnVA (FE)

lnRD 0.067*** 0.061*** 0.138*** 0.122***

(0.000) (0.000) (0.000) (0.000)

lnPAF_1 0.019*** 0.018*** 0.037*** 0.031***

(0.000) (0.000) (0.000) (0.000)

lnPACE_1 0.010** 0.008* 0.024*** 0.022***

(0.033) (0.055) (0.000) (0.001)

lnTI 0.022*** 0.049***

(0.000) (0.000)

CR4 0.134 -0.559***

(0.215) (0.000)

YEAR-99 0.064** 0.063** 0.041 0.034

(0.014) (0.012) (0.291) (0.357)

YEAR-00 0.129*** 0.125*** 0.156*** 0.134***

(0.000) (0.000) (0.000) (0.000)

YEAR-02 0.170*** 0.149*** 0.071* 0.024

(0.000) (0.000) (0.067) (0.528)

YEAR-03 0.251*** 0.230*** 0.176*** 0.120***

(0.000) (0.000) (0.000) (0.001)

R-square 0.211 0.242 0.242 0.315

Hausman test 612.99*** 640.37*** 654.32*** 671.26***

F-test all δ i =0 10.43*** 10.61*** 12.64*** 12.27***

Observations 1170 1170 1170 1170

112/04/21 33

Findings1. The estimated coefficients for both one-year lagged

pollution control fees (lnPFA) and capital (lnPACE) are positive and significant at a conventional statistical level in all estimates.

→ supporting the Porter hypothesis that a stringent environmental regulation is positive related to industrial productivity.

2. the estimated productivity elasticity of PACE on capital is much smaller than that of pollution control fees in all estimates.

3. The coefficient of technology imports is much higher than that of R&D, because importing desirable technologies can direct promote productivity.

112/04/21 34

Concluding Remarks and Policy

Implications Concluding Remarks1. Various estimates confirm a significantly positive relationship

between pollution abatement fees and R&D expenditures, whereas there is no evidence to support the R&D-inducement effect brought about by pollution abatement capital expenditures (PACE).

2. as the PACE is further classified into various capital costs of water, gas, waste, and noise control, we find a significant R&D inducement effect brought about by cost of waste control capital.

3. environmentally induced R&D significantly contributes to industrial productivity.

4. the alternative estimation that directly examines the influence of pollution control expenditures on industrial productivity show that both cost of pollution control capital and pollution abatement fees has a significantly positive impact on promoting productivity.

---- Porter Hypothesis is possibly supported in Taiwan

112/04/21 35

Policy Implications 1. whether the win-win situation can be reached depends

on well-designed environmental regulations →tax incentives for investments in pollution control

equipments? → rethink the necessity of tax credit for PACE on capital. → providing a more appealing incentive to

environmental R&D activities

2. no nationwide statistics of environmental R&D in Taiwan.

→ The government can devote more efforts to conduct general and up-to-date surveys of environmental R&D activities. More accurate definitions of environmentally-friendly technologies in Taiwan are also needed.

112/04/21 36

Thank you !! ขอบคุ�ณคุรั�บ!!