II. Empirical Interpretation

Transcription

1 percapitagdp and pollutionb1.nb 1 In[1]:= In[2]:= << Statistics`LinearRegression` << Statistics`ContinuousDistributions`; # dataa# In[4]:= Off@General::spell General::spell1D II. Empirical Interpretation In this section we examine if the inverted U-shaped relation between the per capita GDP and the pollution level can be depicted for the Japanese cross-section data which was made available by National Institute of Environmental Studies ( The institute has published atmospheric pollution levels for observation points in each of 47 prefectures since 1980s. Types of atmospheric pollution range from SOX NOX SPM (Suspended Particulate Matters) to Oxidant etc. In this paper for each type of atmospheric pollution the simple average of observed values is defined as the atmospheric pollution level of the prefecture. Starting from SOX for each type of atmospheric pollution we examine if the inverted U-shaped relation between the per capita GDP and the pollution level can be depicted. As for the data of per capita GDP for each prefecture in 2000 datab is borrowed from Japan Statistical Yearbook 2005 (Statistical Research and Training Institute (ed.) Statistics Bureau Ministry of Internal Affairs and Communications). In order to make a comparison the data of per capita income for each prefecture in 2000 databb is also borrowed from Japan Statistical Yearbook # datab and databb# ü 1. SOX (2000) In Japan SOX has been the most notorious pollutant. It was in the 1890s that farmers and ecologists demanded the alleviation of the pollution caused by two Japanese copper sulfide mines: Ashio and Besshi mines. This pollution was due to the discharged sulfurous acid gas from those mines (Arahata [1907]). In 1973 both mines were closed and almost all the copper ores are imported from abroad. In the 1950s MINAMATA disease became the world-notorious synonym for pollution. This disease erupted in a southern coastal city MINAMATA and it was the mercury poisoning where the mercury was discharged into the sea in producing nitrogen fertilizer. This pollution leads to the legislation of the Water Preservation Act and the Factory Drainage Regulation Act in 1958 and finally to the Water Pollution Prevention Act in At about the same time there erupted the atmospheric pollution in Yokkaichi a central coastal city. The disease caused by the discharged sulfurous acid gas from petro-chemical refineries and electric power plants was called the Yokkaichi asthma. This pollution leads to the legislations of Smoke Regulation Act in 1962 the Atmospheric Pollution Prevention Act in 1968 and finally to the Revised Atmospheric Pollution Prevention Act in After the Kyoto Protocol was concluded in 1997 whose purpose is to reduce the CO2 emission into the atmosphere to prevent the global warming the trading on CO2 emission rights has attracted a great deal of attention. It was little known however that the SO2 emission rights for electric power plants in the United States was indeed traded in 1993 (Nihon Keizai Shimbun April ). In this subsection we examine if the inverted U shaped relation between the per capita GDP and SO2 level can be depicted for the Japanese cross-section data which was made available by National Institute of Environmental Studies. According to this institute in Hokkaido (Prefecture) there were 91 observation points in 2000 for SO2 whose observation results are given by data[1]. In this paper the SO2 level for Hokkaido (Prefecture) in 2000 is defined by the simple average of data[1]:

2 percapitagdp and pollutionb1.nb 2 In[7]:= data@1d = <; In[8]:= Length@data@1DD Out[8]= 91 In[9]:= Mean@data@1DD Out[9]= In Aomori (Prefecture) there were 12 observation points in 2000 for SO2 whose observation results are given by data[2]. In this paper the SO2 level for Aomori (Prefecture) in 2000 is defined by the simple average of data[2]: In[10]:= data@2d = <; In[11]:= Mean@data@2DD Out[11]= Continuing this procedure we obtain the following data dataso21 the collection of SO2 levels for 47 prefectures. #dataso21# In what follows utilizing dataso21 a linear regression is attempted between the per capita GDP and the SO2 level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid line is the computed regression line which shows the positive relation between them. In[13]:= figso21 = ListPlot@Sort@dataSO21D PlotJoined True yso21 = Fit@dataSO21 81 x< xd; figso22 = Plot@ySO21 8x < DisplayFunction IdentityD; Show@8figSO21 figso22< AxesOrigin < AxesLabel 8"per capita GDP" "SOX"< DisplayFunction $DisplayFunctionD; SOX per capita GDP

3 percapitagdp and pollutionb1.nb 3 According to the following regression report on the linear regression however the positiveness of the slope is statistically not confirmed (p-value= ). In[17]:= Regress@dataSO21 81 x< x Out[17]= x RSquared AdjustedRSquared = Next utilizing dataso21 a quadratic regression is attempted between the per capita GDP and the SO2 level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid curve is the computed regression curve which shows the inverted U-shaped relation between them. In[18]:= yso23 = Fit@dataSO21 81 x x^2< xd; figso23 = Plot@ySO23 8x < DisplayFunction IdentityD; Show@8figSO21 figso23< AxesOrigin < AxesLabel 8"per capita GDP" "SOX"< DisplayFunction $DisplayFunctionD; SOX per capita GDP If the quadratic regression is conducted however the coefficients are not statistically confirmed although they are statistically confirmed if the 10% probability rule on the null hypothesis is adopted according to the following regression report on the quadratic regression. In[21]:= Regress@dataSO21 81 x x ^ 2< x Out[21]= x x RSquared AdjustedRSquared = Next in what follows the regression is conducted between the per capita income and SO2 level. First utilizing dataso22 a linear regression is attempted between the per capita income and the SO2 level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid line is the computed regression line which shows the positive slope. #dataso22#

4 percapitagdp and pollutionb1.nb 4 In[23]:= figso22 = ListPlot@Sort@dataSO22D PlotJoined True yso22 = Fit@dataSO22 81 x< xd; figso22000a = Plot@ySO22 8x < DisplayFunction IdentityD; Show@8figSO22 figso22000a< AxesLabel 8"" "SOX"< DisplayFunction $DisplayFunctionD; SOX Contrary to the regression utilizing the per capita GDP the positiveness of the slope is statistically confirmed (pvalue= ) according to the following regression report on the linear regression. In[27]:= Regress@dataSO22 81 x< x Out[27]= x RSquared AdjustedRSquared = Finally utilizing dataso22 a quadratic regression is attempted between the per capita income and the SO2 level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid curve is the computed regression curve which shows the inverted U-shaped relation between them. In[28]:= yso23 = Fit@dataSO22 81 x x^2< xd; figso22000b = Plot@ySO23 8x < DisplayFunction IdentityD; Show@8figSO22 figso22000b< AxesLabel 8"" "SOX"< DisplayFunction $DisplayFunctionD; SOX Contrary to the case of regression on the per-capita GDP case the coefficients are statistically confirmed for the per-capita income case according to the following regression report on the quadratic regression.

5 percapitagdp and pollutionb1.nb 5 In[31]:= Regress@dataSO22 81 x x ^ 2< x Out[31]= x x RSquared AdjustedRSquared = As a conclusion it may be said that the inverted U-shaped relation holds in 2000 between the per capita income and the SO2 level in Japan although the linear positive relation between them cannot be denied. ü 2. SOX (1990) In this subsection following the same procedure as abovewe obtain the data dataso23 the collection of pairs of per capita income and SO2 level in 1990 for 47 prefectures for the purpose of examining how the relation changed since 1990 until #dataso23# First utilizing dataso23 a linear regression is attempted between the per capita income and the SO2 level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid line is the computed regression line which shows the positive relation between them. In[33]:= figso23 = ListPlot@Sort@dataSO23D PlotJoined True yso23 = Fit@dataSO23 81 x< xd; figso = Plot@ySO23 8x < DisplayFunction IdentityD; Show@8figSO23 figso231990< AxesLabel 8"" "SOX"< DisplayFunction $DisplayFunctionD; SOX According to the following regression report on the linear regression the positiveness of the slope is statistically confirmed (p-value= ).

6 percapitagdp and pollutionb1.nb 6 In[37]:= Regress@dataSO23 81 x< x Out[37]= x RSquared AdjustedRSquared = Next utilizing dataso23 a quadratic regression is attempted between the per capita income and the SO2 level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid curve is the computed regression curve. In[38]:= yso24 = Fit@dataSO23 81 x x^2< xd; figso24 = Plot@ySO24 8x < DisplayFunction IdentityD; Show@8figSO23 figso24< AxesLabel 8"" "SOX"< DisplayFunction $DisplayFunctionD; SOX According to the following regression report on the quadratic regression the coefficients are statistically not confirmed. In[41]:= Regress@dataSO23 81 x x ^ 2< x Out[41]= x x RSquared AdjustedRSquared = As a conclusion on the comparison between 1990 and 2000 it may be said that the linear positive relation holds between the per capita income and the SO2 level in 1990 while the inverted U-shaped relation holds in 2000 between the per capita income and the SO2 level.

7 percapitagdp and pollutionb1.nb 7 ü 3. NOX (2000) NOX as well as SOX cause pollution such as acid rain or respiratory diseases and the government has attempted to reduce them. Compared with SOX however the reduction of NOX is more difficult. The origin of NOX is same as SOX. NOX such as NO or NO2 is generated by the combustion of petroleum. For the purpose of distinguishing the origin of NOX and SOX let us restrict it to the one from automobiles. SOX is generated inside the engines of automobiles: i.e. sulfur in petroleum is oxidized. Meanwhile NOX is generated outside the engines of automobiles: i.e. nitrogen in the atmosphere is oxidized by the heat of engines. In other words if the automobile manufacturers attempt to improve the efficiency in combustion of fuels usually the attempt succeeds in the reduction of SOX but it fails in the reduction of NOX. In 1992 the Automobile NOX Reduction Act was enacted. In this subsection following the same procedure as above we obtain the data datanox1 the collection of pairs of per capita GDP and NOX level in 2000 for 47 prefectures. #datanox1# Utilizing datanox1 a linear regression is attempted between the per capita GDP and the NOX level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid line is the computed regression line which shows the positive relation between the per capita GDP and the NOX level in In[43]:= fignox1 = ListPlot@Sort@dataNOX1D PlotJoined True ynox1 = Fit@dataNOX1 81 x< xd; fignox2 = Plot@yNOX1 8x < DisplayFunction IdentityD; Show@8figNOX1 fignox2< AxesLabel 8"per capita GDP" "NOX"< DisplayFunction $DisplayFunctionD; NOX per capita GDP According to the following regression report on the linear regression the positiveness of the slope is statistically confirmed (p-value= ). In[47]:= Regress@dataNOX1 81 x< x Out[47]= x RSquared AdjustedRSquared =

8 percapitagdp and pollutionb1.nb 8 If the quadratic regression is conducted the coefficients are not statistically confirmed according to the following regression report on the quadratic regression. In[48]:= Regress@dataNOX1 81 x x ^ 2< x Out[48]= x x RSquared AdjustedRSquared = If the per capita GDP is modified to the per capita income what would happen? In what follows utilizing data- NOX2 a linear regression is attempted between the per capita income and the NOX level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid line is the computed regression line which shows the positive relation between the per capita income and the NOX level in #datanox2# In[50]:= fignox21 = ListPlot@Sort@dataNOX2D PlotJoined True ynox2 = Fit@dataNOX2 81 x< xd; fignox22 = Plot@yNOX2 8x < DisplayFunction IdentityD; Show@8figNOX21 fignox22< AxesLabel 8"" "NOX"< DisplayFunction $DisplayFunctionD; NOX According to the following regression report on the linear regression the positiveness of the slope is statistically confirmed (p-value= ). As is clear from the improvement in the R 2 the fitness for this case is superior to the case between the per capita GDP and the NOX level in In[54]:= Regress@dataNOX2 81 x< x Out[54]= x RSquared AdjustedRSquared =

9 percapitagdp and pollutionb1.nb 9 If the quadratic regression is conducted the coefficients are not statistically confirmed either according to the following regression report on the quadratic regression. In[55]:= Regress@dataNOX2 81 x x ^ 2< x Out[55]= x x RSquared AdjustedRSquared = As a conclusion it may be said that the linear positive relation holds between the per capita GDP (or income) and the NOX level in Quadratic relation does not hold between the per capita GDP (or income) and the SPM level in 2000 let alone the inverted U-shaped relation. ü 4. NOX (1990) In this subsection it is examined how the relation changed since 1990 until In the same way as in the case of datanox1 datanox1990 the collection of pairs of income and NOX level in 1990 for 47 prefectures is constructed in what follows. #datanox1990# Utilizing datanox1990 first a linear regression is attempted between the per capita income and the NOX level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The thick solid line is the computed regression line for 1990 which shows the positive relation between the per capita income and the NOX level in The solid line in this figure is the computed regression line for It may be said from the comparison between 1990 and 2000 that since 1990 until 2000 Japanese society became more pollution-intolerant. In[57]:= fignox19901 = ListPlot@Sort@dataNOX1990D PlotJoined True ynoxa2 = Fit@dataNOX x< xd; fignox31990 = Plot@yNOXA2 8x < PlotStyle Thickness@0.01D DisplayFunction IdentityD; Show@8figNOX22 fignox19901 fignox31990< AxesLabel 8"" "NOX"< DisplayFunction $DisplayFunctionD; NOX

10 percapitagdp and pollutionb1.nb 10 The regression report for datanox1990 in what follows confirms the positiveness of the slope statistically. As is clear from the improvement in the R 2 the fitness for this case is superior to the case between the per capita income and the NOX level in Japan for In[61]:= Regress@dataNOX x< x Out[61]= x RSquared AdjustedRSquared = As shown in the previous subsection the coefficients on the quadratic regression on NOX for 2000 were not statistically confirmed. If the quadratic regression for 1990 is conducted on NOX the coefficients are not statistically confirmed either according to the following regression report on the quadratic regression for In[62]:= Regress@dataNOX x x^2< x Out[62]= x x RSquared AdjustedRSquared = ü 5. SPM (2000) Until 1999 SPM (Suspended Particulate Matters) was almost unknown to the Japanese general public. In 1999 Governor of Tokyo Shintaro Ishihara abruptly announced the revision of Metropolis of Tokyo Pollution Prevention Act in a press conference showing the reporters the black powder in a bottle at the conference. The powder small enough to float in the air causes respiratory diseases or lung cancer. In Japan 43% of SPM in the air is generated by automobiles especially trucks with diesel engines (87% among automobiles). The revision which took effect in 2003 made stricter regulations on the transport through trucks. Nationwide the Automobile NOX Reduction Act enacted in 1992 was revised into the Automobile NOX-PM Reduction Act in Regulation by Metropolis of Tokyo is stricter than the national one. In this subsection following the same procedure as abovewe obtain the data dataspm1 the collection of pairs of per capita GDP and SPM level in 2000 for 47 prefectures. # dataspm1# Utilizing dataspm1 a linear regression is attempted between the per capita GDP and the SPM level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid line is the computed regression line which shows the positive relation between the per capita GDP and the SPM level in 2000.

11 percapitagdp and pollutionb1.nb 11 In[64]:= fig1 = ListPlot@Sort@dataSPM1D PlotJoined True y1 = Fit@dataSPM1 81 x< xd; fig2 = Plot@y1 8x < DisplayFunction IdentityD; Show@8fig1 fig2< AxesLabel 8"per capita GDP" "SPM"< DisplayFunction $DisplayFunctionD; SPM per capita GDP According to the following regression report on the linear regression the positiveness of the slope is statistically confirmed (p-value= ). In[68]:= Regress@dataSPM1 81 x< x Out[68]= x RSquared AdjustedRSquared = If the quadratic regression is conducted the coefficients are not statistically confirmed according to the following regression report on the quadratic regression. In[69]:= Regress@dataSPM1 81 x x ^ 2< x Out[69]= x x RSquared AdjustedRSquared = If the per capita GDP is replaced by the per capita income what would happen? In what follows a linear regression is attempted between the per capita income and the SPM level in Japan. In the following figure the actual distribution of pairs of data is connected by the dashed line. The solid line is the computed regression line which shows the positive relation between the per capita income and the SPM level in Japan. #dataspm1a#

12 percapitagdp and pollutionb1.nb 12 In[71]:= fig1 = ListPlot@Sort@dataSPM1AD PlotJoined True y1 = Fit@dataSPM1A 81 x< xd; fig2 = Plot@y1 8x < DisplayFunction IdentityD; Show@8fig1 fig2< AxesLabel 8"" "SPM"< DisplayFunction $DisplayFunctionD; SPM According to the following regression report on the linear regression the positiveness of the slope is statistically confirmed (p-value= ). As is clear from the improvement in the R 2 the fitness for this case is superior to the case between the per capita GDP and the SPM level in In[75]:= Regress@dataSPM1A 81 x< x Out[75]= x RSquared AdjustedRSquared = If the quadratic regression is conducted the coefficients are not statistically confirmed either according to the following regression report on the quadratic regression. In[76]:= Regress@dataSPM1A 81 x x^2< x Out[76]= x x RSquared AdjustedRSquared = As a conclusion it may be said that the linear positive relation holds between the per capita GDP (or income) and the SPM level in Japan. Quadratic relation does not hold between the per capita GDP (or income) and the SPM level in 2000 let alone the inverted U-shaped relation.

13 percapitagdp and pollutionb1.nb 13 ü 6. SPM (1990) In this subsection it is examined how the relation changed since 1990 until In the same way as in the case of SPM (2000) dataspm11990 the collection of pairs of income and SPM level in 1990 for 47 prefectures is constructed in what follows. #dataspm11990# Utilizing dataspm11990 first a linear regression is attempted between the per capita income and the SPM level in In the following figure the actual distribution of pairs of data is connected by the dashed line. The thick solid line is the computed regression line for 1990 which shows the positive relation between the per capita income and the SPM level in Japan. The solid line in this figure is the computed regression line for In[78]:= fig11990 = ListPlot@Sort@dataSPM11990D PlotJoined True y11990 = Fit@dataSPM x< xd; fig21990 = Plot@y x < PlotStyle Thickness@0.01D DisplayFunction IdentityD; Show@8fig11990 fig21990 fig2< AxesLabel 8"" "SPM"< DisplayFunction $DisplayFunctionD; SPM The regression report for dataspm11990 in what follows confirms the positiveness of the slope statistically. As is clear from the improvement in the R 2 the fitness for this case is superior to the case between the per capita income and the SPM level in Japan for The comparison between 1990 and 2000 shows that since 1990 until 2000 Japanese society became more pollution-intolerant. Note however that there is a report saying that the "SPM environmental achievement rates" in the observation points consecutively fell since 1999 until 2002 (Nihon Keizai Shimbun Sep ). In[82]:= Regress@dataSPM x< x Out[82]= x RSquared AdjustedRSquared =

14 percapitagdp and pollutionb1.nb 14 As shown in the previous subsection the coefficients on the quadratic regression for 2000 were not statistically confirmed. If the quadratic regression for 1990 is conducted the coefficients are not statistically confirmed either according to the following regression report on the quadratic regression for In[83]:= 81 x x^2< x Out[83]= x x RSquared AdjustedRSquared = Conclusions The aim of this paper was to examine the inverted U-shaped relation between the per-capita GDP and pollution level discovered by Grossman and Krueger [1995]. Colstad [1999] stressed the income effect. As in the traditional microeconomic argument on the backward-bending labor supply function it is explained that when a society is poor it endures the rising pollution level to become rich while as it becomes affluent it prefers less pollution when it becomes more affluent due to the income effect. Theoretically this argument assumes that the utility function is invariant as the society becomes rich or in the process of economic growth. To explain the inverted U-shaped relation in the economic growth however an alternative explanation is also possible. Thus assuming that in the process the hard experience of pollutions such as MINAMATA disease causes the shift of utility function towards more pollution-intolerant the alternative explanation goes as follows: i.e. although for each utility function the positive relation holds between the per-capita GDP and pollution level the locus of the society's selection of them in the economic growth may be of the inverted U-shaped. In Section I of this paper it was first examined if the inverted U-shaped relation between the per-capita GDP and pollution level can be depicted assuming the invariant utility function. It was shown that when Cobb-Douglas (invariant) utility function is assumed the positive relation holds and under assumption of CES (invariant) utility function it is possible to depict the inverted U-shaped curve for some specification of parameters. It was also shown that under assumption of CES (invariant) utility function it is possible to depict the linear positive line for other specification of parameters. Thus theoretically the above two explanations are possible. In Section II it was examined if the inverted U-shaped relation between the per-capita GDP and pollution level holds for specific years utilizing Japanese cross-section data for 1990 and In this section three types of pollution were dealt with: i.e. SOX NOX and SPM. It was ascertained that the replacement of per-capita GDP by per capita income gives rise to superior regression results for all the examinations. The inverted U-shaped relation between the per-capita income and pollution level does exist for SOX case in It is the case however only for SOX case in Note furthermore that even for this SOX case in 2000 statistically it is also possible to assert that the positive relation exists between the per-capita income and SOX level in In 1990 only the positive relation exists between the per-capita income and SOX level: i.e. the inverted-u shaped relation is statistically denied. For all the other cases on NOX and SPM only the positive relation exists while the inverted U-shaped relation is statistically denied. Furthermore since 1990 until 2000 the Japanese society becomes more pollution-intolerant. From the above results it appears that the explanation in terms of the shift of utility function towards more pollutionintolerant is more persuasive in Japan.

15 percapitagdp and pollutionb1.nb 15 References Arahata K. [ ] The History of the Extinction of Yanaka Village (in Japanese) Heimin Shoboh Tokyo (Republished by Iwanami Tokyo in 1999). Colstad C.D. [1999] Environmental Economics Oxford University Press. Fukiharu T. [1991] "A General Equilibrium Approach to Environmental Problems-Enlarged Version" Kobe University Economic Review 37 pp Grossman G.M. and A.B. Krueger [1995] "Economic Growth and the Environment" Quarterly Journal of Economics 110 pp Pigou A.C. [1920] The Economics of Welfare Macmillan London. Sumi K. [1989] Reality of Official Development Assistance (in Japanese) Iwanami Tokyo.