Page 75 - Read Online

Page 8 of 16 Hu et al. Carbon Footprints 2024;3:16 https://dx.doi.org/10.20517/cf.2024.28

To repeat, the cities studied were selected based on the availability of consistent and reliable data on city
carbon emissions, industrial activities, economic indicators, and other relevant factors over the period of
1999 to 2019. The availability of high-quality data across these cities was essential for applying the
Difference-in-Differences (DID) methodology and ensuring the accuracy of the results.

RESULTS
Descriptive analysis
The number of ETDZs grew from 38 in 1999 to 230 in 2024. EDIPs have increased steadily since 2008. As
Figure 2 shows, both ETDZ and EIDP are primarily located in the eastern region, and their presence in
central and western regions is less than half. In 2015 and 2016, 51 LCIPs were established, with 47.06% in
the eastern region, 25.49% in the central, and 27.45% in the western. In general, ETDZ has the longest
history, the largest amount, and the broadest distribution. EDIP has grown slowly, while LCIPs were
approved in 2015 and 2016, with no new additions since.

Benchmark results
To apply the DID method appropriately, the carbon emission intensities of the treatment and control
groups are supposed to show parallel trends before the construction of relevant industrial parks. Figure 3
confirms this, with the X-axis showing the difference between the current year and the year of construction
of the relevant industrial park, and the Y-axis exhibiting the estimated coefficients of the core independent
variables. The 95% confidence intervals of the coefficient estimates contain 0, indicating no significant pre-
construction difference in carbon emissions existed between the groups. Thus, the DID method is suitable
for evaluating treatment effects.

Table 1 presents the baseline results. ETDZ significantly reduces city carbon emission intensity, with a
coefficient of -1.1545. EIDP and LCIP demonstrate statistically significant coefficients of -0.3949 and
-0.4504, respectively, indicating they reduce carbon emissions, though less than ETDZ.

In addition, the effects of industrial parks on city carbon emissions may vary by region, resource
endowment, and the amount of parks. Heterogeneity analyses concerning these aspects were thus carried
out. As provided in the Supplementary Materials, Supplementary Table 1 exhibits the varying effects with
regard to geographical location. Supplementary Table 2 shows the effects arising from the differences in
resource endowments across cities. Supplementary Table 3 exhibits the heterogeneity in effects with respect
to the city-level number of industrial parks.

Dynamic effects analysis
Figure 4 shows the impact of three types of industrial parks on city carbon intensity over the periods of
10 years before and 15 years after, using the current year as the base group. It shows that for ETDZ,
significant negative effects on carbon intensity would appear starting from the 7th year post-construction.
For EIDP, the negative effects started in the 3rd year post-construction. While for LCIP, no significant
effects were observed within three years post-construction.

Mediating effects analysis
Table 2 shows the effects of industrial parks on city carbon emission intensity via industrial structure (IS).
The coefficients appear all significantly negative, suggesting that the presence of industrial parks can raise
the share of tertiary industry and/or reduce that of secondary industry. Additionally, the mediator variable
appears to have positive effects on city carbon emissions, implying the contributing role of secondary
industry.