**5. Results**

This analysis focuses on examining how a plant responds to the adoption of new environmental regulations through the interaction term between the policy period's indicator and industry variables. Table 2 shows the results of labor productivity and employment estimations for polluting industries. Columns 1–3 give the results comparing six years before and after LCGG and Columns 4–six compare four-year effects. The DD terms in the table show significantly negative effects with the magnitude of the effects becoming larger when we consider a longer period. Labor productivity decreased about 5–6 percent and employment decreased 3 percent during the six years. These significantly negative effects were consistent when 2nd and 3rd groups in terciles were used instead of median to define the polluting industries. This implies that the increased costs induced by strengthened regulations led to a decrease in employment and productivity and this was not a temporary effect. This is consistent with previous studies [3,8,18].

**Table 2.** Estimation results based on labor productivity and employment as dependent variables for polluting industries.


Notes: The robust standard errors are in parenthesis. Five-digit-level industry, region, and year fixed effects are used for all columns. Columns 1–3 give the results comparing the 2004–2009 and 2010–2015 periods and Columns 4–6 compare 4 years before and after LCGG. HHI is the Herfindahl Hirschman Index and K/L is capital intensity. Post indicates period after 2010 and Poll means polluting industry at the 2–3 digits level. The asterisks \*\*\* and \*\* indicate significant at the 1% and 5% levels of significance respectively.

In Table 3, we divide establishments into green and non-green sectors since we expect a positive demand shift to the green sector. This classification focuses on the product a plant supplies while polluting industries are defined as a sensitive group by the policy because of the emissions in their production processes. The estimated coefficients of interaction *Postt* × *Greeni*∈*<sup>S</sup>* are positive and significant except for Column 6. This means that the green sector relatively experienced enhanced productivity and employment as compared to the non-green sector after LCGG.

Table 4 gives the results of the estimation of Equation (1). In terms of labor productivity, the DD terms show highly statistically significant effects. The green and non-green sectors in the polluting industries responded differently to the imposed regulations considering labor productivity. An establishment in a polluting industry had a decrease of about 6–7 percent in productivity after the regulations (Columns 1 and 2). However, these negative effects were off-set if the establishment was included in the green sector in the polluting industries. It also seems that there were a few productivity gains in this establishment. These positive effects on labor productivity in the green sector came from demand effects because we control the substitution between capital and labor using capital intensity. Hence, our results confirm that environment related industries are a key to understanding the consequences of environmental regulations. Although polluting industries face a higher burden because of the relatively higher compliance costs of the regulations, however, if they produce environment related goods or use environment related processes in their production, then these negative impacts could be canceled out.


**Table 3.** Estimation results based on labor productivity and employment as dependent variables for the green sector.

Notes: The robust standard errors are in parenthesis. Five-digit level industry, region, and year fixed effects are used for all columns. Columns 1–3 give the results comparing the 2004–2009 and 2010–2015 periods and Columns 4–6 compare four years before and after LCGG. HHI is the Herfindahl Hirschman Index and K/L is capital intensity. Post indicates the period after 2010 and Green means environmental industry at the five-digit level. The asterisks \*\*\* and \*\* indicate significant at the 1% and 5% levels of significance respectively.


**Table 4.** Estimation results based on labor productivity and employment as dependent variables for the green sector in polluting industries.

Notes: The robust standard errors are in parenthesis. Five-digit level industry, region, and year fixed effects are used for all columns. Columns 1–3 give the results comparing the 2004–2009 and 2010–2015 periods and Columns 4–6 compare 4 years before and after LCGG. HHI is the Herfindahl Hirschman Index and K/L is capital intensity. Post indicates the period after 2010 and Green means environmental industry at the five-digit level. Poll is dummy of polluting industry at the 2–3 digit level. The asterisks \*\*\* and \*\* indicate significant at the 1% and 5% levels of significance respectively.

In Columns 3 and 6, we use total employment as the dependent variable. The estimated coe fficients of the DD terms are negative and not significant. This means that the green and non-green sectors in the polluting industries are not di fferent considering the intensity of the regulations' e ffects. Plants in polluting industries reduce their employment but this reduction is not recovered within the industry and jobs increase in plants, including those in green and less polluting industries. This evidence implies a trade-o ff between green and non-green sectors within the brown industries only in labor productivity. And establishments in the green sector in less-polluting industries have a marginal increase in employment (Column 3). The main results are consistent when another criterion is used for a Poll indicator, but the o ff-set to a loss captured by DDD terms diminishes.
