Estimation under three industry structures

The further analysis between the pollution pressure and the economic growth in China will be the three industry structures being added in the estimation in our model. This haven’t been done by other previous studies in this field in China. Generally, besides GDP per capita I added primary industry value added per capita, secondary industry share14 _{and tertiary industry value added per}

capita into the equation (3.30) and make them to match the three effects I have mentioned in the previous discussion.

The reason for doing this is because China’s GDP format is mainly constructed by these three industry structures. I intend to get a result that demonstrates the relationship between pollution pressure and different industry structures respectively in China. The result will help us to understand what type of industry structure will hurt the environment most and also helps the Chinese government to issue a specific restriction under each industry. Primary industry contains mainly of agriculture, forestry, graziery and fishing, second industry, mainly contains industries (factories and plants) and manufacture industry and tertiary industry, mainly contains transportation industries and service in- dustries (Hitoshi and Satoko, 2009) Thus,

GDPChina=P rimaryIndustryV alueAdded + SecondaryIndustryV alueAdded

+ T ertiaryIndustryV alueAdded (3.31)

Therefore the income variable or an income effect can be decomposed by values of these three industrial structures. Recall equation (3.26) from Panayotou (1997) and Islam, Vincent, and Panayotou (1999), I made a little innovation on to this model. First, I switch GDP per unit of area which stands for scale effect into the added value of the primary industry per capita, secondary industry share for structure effect and the added value of the tertiary industry per capita for abatement effect. Second, I assume each effect takes proportional effect on the result of pollution pressure. Therefore, equation (3.26) now as,       Ambient P ollution Level       | {z } Pollution =       P rimary Industry of GDP       β1 | {z } Scale ×       Secondary IndustryShare of GDP       β2 | {z } Structure ×       T ertiary Industry of GDP       β3 | {z } Abatement (3.32)

where 0 < β1< 1, 0 < β2< 1, 0 < β3< 1 and β1+ β2+ β3≤ 1. If we take a logarithm from both sides

of the equation (3.32) then we have,

Ln(P ollution) = β1Ln(P rimary) + β2Ln(Secondary) + β3Ln(T ertiary) (3.33)

The scale effect on pollution is expected to be a monotonically increasing function of income by controlling other two effects fixed. The larger the scale of economic activity per unit of area the higher the level of pollution pressure is presented. It is happened at pre-industry economy stage (Panayotou, 1991). Therefore, the added value of primary industry has matched this scale effect well since the industries under primary industry structure are mainly pre-industry sectors. The structural change along with economic growth affects environmental quality by changing the composition of economic activity toward sectors of higher or lower pollution intensity. This means that at a lower income level, the economy shifts its structure from agriculture based, a lower pollution intensity to industry based, to a higher pollution intensity and results in an increase of environmental degradation. At higher income, the economy switches its structure from industry based to service based, a lower pollution intensity and results a decrease of environmental degradation. The structural effect is likely to be a non-monotonic (inverted-U) function of GDP and It happens in industry economy stage (Panayotou, 1991). This effect matches with secondary industry, which mainly contains industries under this structure. The share of industry first rises and then falls, then the environmental pollution will first rise and then fall with

income growth by holding other effects constant.

After determining scale and structure effects in my model, the added value of tertiary industry just perfectly describes the abatement effect representing pure income effects on the demand and supply of environmental quality. On the demand side, at a lower income, when income increases, people increase in demanding food and shelter, but not too much demand for environmental quality. In contrast, at a higher income level, when income increases people demand more for environmental quality since they have already reach a living standard for food and shelter. On the supply side, higher incomes cause an increase in private and public expenditures on pollution and make more resources provided on a service sector and environment regulations available to internalize pollution externalities (Panayotou, 1991). Thus the added value of tertiary industry capture this effect since individuals are rich enough and willing to invest their money in the pollution abatement effort at this stage. The abatement effect is expected to be a monotonically decreasing function of income.

As a result,from equation (3.33) I decompose the three effects based on primary, secondary and tertiary industry into equation (3.30), we estimate,

eit=α + β1pit+ β2p2it+ β3p3it+ β4sit+ β5s2it+ β6s3it+ β7(te)it+ β8(te)2it+ β9(te)3it

+ β10pop + β11open + β12ur + β13f di + ηi+ γt+ ui (3.34)

where pit, sitand (te)itare logged values of primary industry value added per capita, secondary industry

share and tertiary industry value added per capita respectively to represent the scale, structure and abatement effect and the rest of the variables remain same as from the equation (30).

The primary industry share is expected to be positive sign since, other things equal, the larger the volume of economic output, the higher level of pollution emissions are. The secondary industry share is also expected to a positive sign since it is highly correlated with energy consumption from industrialization. Having controlled for the scale and structure of the output, the tertiary industry share is expected to be a negative sign since service industries are the main sectors in tertiary industry where it is a stage of high demand for pollution abatement. By controlling all scale effect, structure effect and abatement effect GDP per capita should be negative sign since income effect kicks in from both demand and supply sides.

I estimate both models (1)-only GDP per capita and (2)-both GDP per capita and other variables by both fixed and random effects. The fixed effect treats differences in the intercepts as due to deterministic factors. The random effect treats those differences as due to stochastic factors. Whether the fixed effect is a better method, or whether the random effect is a better method, I use the Hausman test(Hausman 1978) to determine the preferred version by testing the null hypothesis of that the other variables were correlated with both year and region. The fixed effect model is preferred if the null hypothesis is

rejected at a significance level of five percent otherwise the random effect model is preferred.