V ARIABLE M EASUREMENT

The dependent variable in this study is the binary variable, ISO certification, which represents whether a firm becomes ISO certified or not.

ISO certification is based on the unit of plants or sites instead of a firm. Given that only firm-level instead of plant-level information is available, ISO certification is coded 1 for firms adopting ISO 14001 certification for at least one plant or site, and 0 otherwise. Among the 332 certified firms in the sample, only 20 firms have adopted ISO 14001 certification for more than one plant or site. Therefore, we do not distinguish the firms with more than one plant or site from those with only one plant or site in the regressions.

For each of the sample firms, firm-specific variables are defined below.

Export ratio represents the percentage contribution of exports to total sales revenue. However, the pressure from multinational customers may vary with destination markets. To account for this variation, this study classifies the destination markets into two groups, one with a higher environmental standard for imports while the other has a lower standard. Although no unique criterion was used for the categorization, developed countries are generally believed to require a higher environmental standard for production activities.

The EU, Japan and the U.S thus comprise the higher standard group. This study then weights each firm’s export ratio by using the ratio of exports in the

firm’s industry accounted for by the higher standard group. This study calculates these ratios for each industry utilizing the trade statistics from the governments.

The dummy variable DPP governor indicates the party affiliation of local governors, and is coded as 1 if the county magistrate or city mayor for the jurisdiction where the firm’s certified sites or main sites are located is a member of the DPP, and 0 if they are a member of some other political party.

Furthermore, Subsidy, a binary dummy variable, is coded 1 for certified firms that received government subsidies to offset the costs of implementing ISO 14001, and 0 otherwise. The government reduced the maximum amount of subsidy to ISO 14001 certification from a previous amount of US$16,000 to

$6,000 since 2001. This exogenous policy change provides a natural experiment for investigating whether the government’s smaller subsidy would reduce a firm’s decision to become ISO certified. Therefore, we include in the regressions the variable of Subsidy since 2001, which equals 1 if a firm received any subsidies over 2001-2004.

Asset denotes the logarithm of a firm’s assets and is used as a proxy for firm size. Firm profitability, Profitability, is calculated as the ratio of firm before-tax profits to total sales revenue for the current year. Previous studies have used an averaged measure to account for the volatility in profitability from year to year (Chu et al., 2005). Hence, this study also presents the estimation results based on a 4-year average profitability. To assess whether the estimates are affected by different measures of profitability, this study also utilizes return on asset and return on equity to estimate the effect of profit on

ISO 14001 adoption. Returns on assets are based only on current year data.

Meanwhile, firm financial structure, Debt ratio, represents the ratio of current debts to total assets. Additionally, R&D, a proxy for relative size of firm knowledge capital, is calculated as firm R&D expenditure divided by total sales revenue. HSIP is a dummy variable used to indicate whether a firm is located in the Hsinchu Science-based Industrial Park (HSIP). HSIP is coded 1 if the firm is located in the HSIP, and 0 otherwise.

The regressions also include various control variables. This study investigates whether firms with different durations may have different likelihoods of implementing ISO 14001 EMS. Firm age is defined as the number of years between its establishment and 2004. Urban-rural disparity and income inequality exist among geographical regions and thus residents’

demand for environmental standards may vary. Therefore, it is necessary to control for possible effects owing to firms’ location in urban or county areas.

Urban represents a binary dummy and is coded 1 if the firm’s main sites are located in any one of the seven urban areas in Taiwan (Taipei-Keelung, Kaohsiung, Taichung-Changhua, Jhongli-Taoyuan, Tainan, Hsinchu and Chiayi), and 0 otherwise.

Industrial pollution characteristics influence the propensity of companies to seek ISO 14001 certification. Firms in highly polluting industries or those with older technologies frequently are involved in a constant battle to reduce emissions incrementally to match increasingly stringent environmental regulations (Bansal, 2002). ISO 14001 certification should provide firms increased latitude to deal with regulators. Because the plastics industry is

generally believed to be one of the most polluted industries, it serves as the baseline industry in this study and dummies for the other 12 industries are included to control for the industry effects in the estimations. Additionally, we also include year dummies in the regressions to control for the year effects.

Table 3 lists the summary statistics for the final sample of 6,692 observations (see Apendix 1) from 982 firms and the subsample of certified firms. As noted above, among the sample of 982 firms, 332 firms have obtained ISO 14001 certification, comprising 5% of the total observations.

Table 3 indicates that the average export ratio is 0.427. This high export ratio supports the observation that the output of Taiwan is heavily export-oriented. To explain varying pressure from customers in different destination countries, this study categorizes export destinations into two groups, among which the EU, Japan and the US represent the countries requiring higher environmental standards for imports. The average weighted ratio of exports to the EU, Japan and the US equals 0.170. The firms located in the jurisdiction of DPP local governors and receiving subsidies for ISO 14001 certification comprise 61.9% and 1% of the total observations.

The average ratio of before-tax profit to sales equals 0.026 while the average ratio of return on asset and the average ratio of return on equity equal 0.1 and 0.096. The four-year average ratio of profit to sales equals 0.014, which is below the profitability for the current year and indicates that firms made smaller profits or incurred losses in the earlier years of the panel period.

The average values of debt ratio and R&D ratio equal 0.407 and 0.046, respectively. The mean value of R&D ratio equals 0.046 and suggests that most Taiwanese firms allocated limited budgets to R&D. The average length of firm establishment is 23 years. Most of the sampled firms’ factories are located in urban areas while most of the sampled firms are in the electronics industry.

The certified firms generally have higher mean values of weighted export, asset, and age than other observations. Compared to The whole sample, they also have higher percentages of firms receiving subsidy, and located in HSIP. A high ratio of certified firms became certified in 2003. In contrast, the percentages of certified firms located in jurisdictions of DPP governors and urban areas are lower than the other observations. It is also noteworthy that the four-year average ratio of profit to sale is negative for the certified firms and its variation is quite large.

Table 3: Descriptive statistics

The whole sample The certified firms

Variables Mean S.D. Min. Max. Mean S.D. Min. Max.

Asset (in logarithm) 14.412 1.306 9.14 19.99 15.441 1.343 12.55 19.61 Profitability

Current year ratio of

profit to sales 0.026 0.463 -9.66 4.12 0.044 0.366 -5.87 0.48 4-year averaged ratio

of profit to sales ^{0.014 0.376 -4.97 2.55 0.048} 0.191 -1.68 0.48 Current year return on

asset ^{0.100 0.120 -1.16 1.15 0.098} 0.103 -0.53 0.45

Current year return on

equity ^{0.096 0.302 -8.88 1.64 0.090} 0.181 -1.12 0.61

Rubber 0.010 0.101 0 1 0.024 0.154 0 1

Automobile 0.003 0.052 0 1 0.015 0.485 0 1

Electronics 0.644 0.479 0 1 0.623 0.485 0 1

Year effect

1996 0.085 0.279 0 1 0.042 0.201 0 1

1997 0.098 0.298 0 1 0.054 0.227 0 1

1998 0.112 0.315 0 1 0.102 0.304 0 1

1999 0.118 0.322 0 1 0.087 0.283 0 1

2000 0.122 0.327 0 1 0.060 0.238 0 1

2001 0.124 0.329 0 1 0.105 0.308 0 1

2002 0.123 0.328 0 1 0.102 0.304 0 1

2003 0.119 0.324 0 1 0.398 0.490 0 1

2004 0.099 0.299 0 1 0.048 0.214 0 1

Observations 6,692 332

Notes:

1. A firm’s weighted export ratio is obtained by weighting the firm’s export ratio by the share of its industry products exported to the EU, Japan and the US.

2. The statistics for a 4-year averaged ratio of profit to sales are based on a subsample of 6,627 observations after observations with larger losses are excluded.

We also provide the summary statistics for each of the six industries with observations more than 200 in Table 4. Table 4 shows that the percentages of certified firms are similar among the six reported industries. The machinery industry and the electronics industry have higher ratios of export while the chemical industry exports a smaller proportion of its products. A higher percentage of firms in the machinery industry receive government subsidy. The firms in the chemical industry and the metals industry adopted ISO 14001 certification in the earlier years of the panel period while the firms in the plastics industry adopted it in the later years of the panel period. Since the HSIP was established to encourage high-tech firms’ investment, only firms in the chemical industry and the electronics industry are located in the HSIP. Moreover, the firms in the industry have a shorter period of establishment than those in other industries.

Table 4: Descriptive statistics by industry

The plastics industry The textile industry

Variables Mean S.D. Min. Max. Mean S.D. Min. Max.

Asset (in logarithm) 14.91 1.504 12.58 19.61 14.92 1.060 12.44 18.16 Profitability

Observations 223 474

The machinery industry The chemical industry Binary dependent

Asset (in logarithm) 14.36 0.842 12.29 17.26 14.12 0.012 0.11 0.18 Profitability

Current year ratio of _{0.073 0.130 -1.10 0.47 0.010} 0.661 -8.73 2.15 4-year averaged ratio _{0.074 0.096 -0.38 0.37 0.012} 0.443 -4.20 1.15

Current year return _{0.106 0.087 -0.30 0.58 0.097} 0.101 -0.58 0.61

Observations 422 526

The metals industry The electronics industry Binary dependent

Asset (in logarithm) 15.22 1.117 12.4 19.12 14.19 1.301 9.14 19.99 Profitability

Observations 246 4,308

Notes:

1. To save space, we report the summary statistics for only the industries with more than 200 observations.

2. A firm’s weighted export ratios is obtained by weighting the firm’s export ratio by the share of its industry products exported to the EU, Japan and the US.