**3. Data**

For the empirical analysis, this study mainly uses data from the Survey on the Mining and Manufacturing Industry from 2004 to 2015 provided by the Korea National Statistical O ffice's (KOSTAT) Microdata Integrated Service (MDIS). The mining and manufacturing surveys cover establishment level data for firms with 10 or more employees. The surveys include the industry code at the five-digit level and financial information such as assets, capital stock and flow, output, value added, number of employees, wages, and the regional code.

Korea's Ministry of Environment classifies environmental industries at a five digit-level; this is in accordance with OECD's environment industry standards and we mainly call it the green sector in this

paper. The classification of environmental industries in Korea is consistent with 'the Environmental Goods and Services Industry' as defined by the OECD. The definition of the environmental industry is: activities to design, manufacture, and install environmental facilities and measuring equipment for environmental conservation and managemen<sup>t</sup> such as climate, air, water, environmental restoration, environmental safety, health, resource circulation, sustainable environmental resources or services for environmental techniques [33].

The environmental industry (or the green sector) is matched with the Korea Standard Industry Code (KSIC9); KOSTAT provides the matching code. The environmental industry is a comprehensive concept and is di fferent from an eco-friendly or less-polluting industry. By definition, an industry belongs to the environmental sector if its production is environment related. This is why this paper uses two standards to divide establishments. For example, a plant that produces air purifiers may generate harmful substances but it is classified as an environmental industry because the air purifiers provide environmental services. Under this classification, retreatment of rubber tires (code 22112) and manufacture of other industrial glass—waste glass products (code 23129)—become green and polluting industries. Therefore, manufacturers who belong to the environmental industry are subject to regulations. However, as the degree of regulations increases, the demand for environmental goods and facilities also increases and this can lead to an improvement in a firm's performance.

To identify a polluting industry that would be sensitive to regulations, this paper uses CO2 emissions data from the National GHG (Green House Gas) Emission Total Information System. The data is constructed at the 3-digit level and is only available for 2012–15. Because of this data limitation, we define polluting industries as those industries which are above the median of green gas emissions every year during the period. The criteria, 2nd and 3rd groups in terciles, are also used to see whether the main results are robust when the definition varies.

Summary statistics of the data are presented in Table 1. The results of the mean equality test (t-test) show that the mean variables between the green and non-green sectors in each industry di ffered statistically. The average size of green businesses such as output, capital, and age was larger than that of non-green businesses. In terms of the number of establishments, the businesses in the green sector accounted for about 21 percent of the manufacturing industry. Although the number of green establishments is relatively small, the average number of workers is similar or higher than in the non-green sector. The average wage of each group in the green sector is also higher. This higher salary reflects higher human capital and production technology levels.

This paper uses the capital variable as capital stock values at the beginning and end of the year. For calculating capital stock, we generally use the year-early data. However, there are some limitations with the 2010 and 2015 surveys since capital stock was omitted at the establishment level (financial and capital assets were surveyed at the firm level, so establishment level data was not reported). Hence, because of this missing data for 2010 and 2015 this paper uses capital stock at the end of 2009 and 2014 respectively. All nominal variables such as sales, value-added, and wages are converted to real values. In the case of sales and value added, the paper uses GDP deflators calculated by KOSTAT. We use the capital deflators and the consumer price index (CPI) obtained from the Bank of Korea to transform capital assets and wages to real values. The paper measures productivity using labor productivity, which is real value-added divided by the number of workers. Workers include full-time and irregular employees except dispatched employees. Age of the firm is measured in years from its date of establishment. The concentration index is measured as the Herfindahl Hirschman Index (HHI)—a square sum of shares of all establishments in the five-digit industries. The paper uses HHI for controlling the level of competition in the industry that a ffects firms' performance such as employment and labor productivity.


**Table 1.** Summary statistics of the data (by green sector and polluting industries).

**B. Polluting Industries**


Source: The Mining and Manufacturing Survey in Korean Statistics and National GHG Emission Total Information System. Notes: Employment is in units of people, the amount is one million won, and CO2 is 1,000 tons. The asterisks \*\*\* and \*\* indicate significant at the 1% and 5% levels of significance respectively. Standard errors are in parenthesis and HHI is the Herfindahl Hirschman Index.

In particular, the production and employment in automobiles accounts for a large portion of the green sector such as manufacture of passenger motor vehicles (hydrogen, hybrid, photovoltaic, and natural gas automobiles) (code 30121), manufacture of other new parts and accessories for motor vehicles (exhaust gas reduction devices, DPF, diesel oxidation catalysts, and DOC devices) (code 30399), and manufacture of parts and accessories for motor engines (code 30310). This is consistent with the fact that domestic automobile manufacturing, which is highly dependent on exports increases investments in eco-friendly product development such as electric vehicles due to the strengthening of environmental regulations by importing countries. In the non-green sector, manufacture of semi-conductors and electronic components (codes 26110 and 26211), building ships and boats (codes 31114 and 31111), manufacture of plastic products for fabricating machines (code 22240), and manufacture of parts and accessories for motor vehicle bodies (code 30320) have soaring employment and output levels.
