Exploring the Impact of Transportation Infrastructure Construction on China's Economic Growth

1.Introduction

With the rapid growth of China's economy, domestic infrastructure construction has been newly developed. In 2008, the international financial crisis broke out under the influence of factors such as the poor operation of capital market funds, and China launched an economic stimulus plan to meet the challenges brought by that crisis; the government invested a considerable portion of funds in infrastructure construction projects, which strongly promoted the development of China's infrastructure construction. The construction of transportation infrastructure has driven the flow of goods and promoted the employment of workers. For a long time, the relationship between transportation infrastructure construction and economic development has been widely discussed, and scholars at home and abroad have also given arguments and explanations on different aspects.

1)Boartnet (1997) examined the relationship between transportation investment and economic development in all California counties from 1968 to 1988 and found that there were differences in the impact of transportation on regional economies. Some areas have benefited and some have been damaged.

2)Ming Liu and Yulin Liu found that when the level of transportation infrastructure reaches a certain extreme, continued investment will significantly reduce the pulling effect of transportation on the economy. Conversely, underinvestment in transportation infrastructure will also constrain local economic development.

3）Tan Jiangrong found that there is a lag in the impact of transportation infrastructure construction on economic development, and the role of transportation infrastructure in promoting economic development will gradually increase with the extension of time.

As China's economy enters a new stage, whether the Chinese government should continue to invest in transportation infrastructure has become a question. Due to the further economic development of each region, is there a difference in the economic boosting effect of transportation infrastructure? Is there a difference in the impact of railroad and road construction on the economy? Based on these questions, we choose the GDP and transportation infrastructure construction data from 2011 to 2020 to conduct the study.

2.Data and descriptive statistics

df = pd.read_csv("/work/data/data_new.csv") df2 = df.query('province!="China"')

My data has 320 entries, recording data on road and rail mileage and economic development in 23 Chinese provinces. The average road mileage is about 290,000 km and the average railroad mileage is about 7,400 km. In terms of economic development indicators, the GDP per capita is 55,000 RMB.

df.loc[:, df2.columns != 'year'].describe().round(2)

Railroad mileagefloat64

Road mileagefloat64

count

320

mean

7473.09

289176.29

std

20512.53

786374.69

min

461.27

12084

25%

2341.98

104007.36

50%

4011.94

157134.5

75%

5194.01

210322.95

max

146330.44

5198120.27

3.Exploratory data analysis

1. China's overall GDP from 2011-2010

The figure shows China‘s overall GDP from 2011 to 2010

px.line(df2, x='year', y='GDP', hover_name='province', line_group='province', color='region', )

px.line(df2, x='year', y='GDP', hover_name='province', color='province', facet_col= 'region', facet_col_wrap= 2, height= 800)

df_map = pd.read_csv("/work/data/data_new.csv") df2020=df_map.query("year==2020") df_map=df2020[['省份','GDP']] map = ( Map() .add('', df_map.values.tolist(), 'china',is_roam=False,is_map_symbol_show=True) .set_global_opts( # 设置标题 title_opts=opts.TitleOpts(title="China GDP Map"), # 设置标准显示 visualmap_opts=opts.VisualMapOpts(max_=110000, is_piecewise=False), ) # 系列配置项 # 关闭标签名称显示 .set_series_opts(label_opts=opts.LabelOpts(is_show=False)) ) map.render_notebook()

Through the map above, we can find that there is a regional imbalance in China's economic development, with the eastern and central provinces being more economically developed.

tl = Timeline() # 播放的速度，单位毫秒（ms） tl.add_schema(play_interval=300, symbol='emptydiamond') for i in range(2011, 2021): dfi=df2.query('year==@i') map0 = ( Map() .add('', dfi[['省份','GDP']].values.tolist(), 'china',is_roam=False,is_map_symbol_show=False) .set_global_opts( title_opts=opts.TitleOpts(title='{}National provincial GDP(Trillion)'.format(i)), visualmap_opts=opts.VisualMapOpts(max_=110000,is_piecewise=True), ) .set_series_opts(label_opts=opts.LabelOpts(is_show=False)) ) tl.add(map0, '{}年'.format(i)) tl.render_notebook()

With the above MAP, we can find that from 2011 to 2020, the economic development of the central and eastern provinces is faster.

dfbar=df2.sort_values(by='Road mileage',ascending=True) fig = px.scatter(dfbar, #dataframe x = 'province', #x-values column y = 'Road mileage', #y-values column facet_col= 'region', facet_col_wrap= 2, animation_frame = 'year', ) fig.show()

dfbox=df2.sort_values(by='Railroad mileage',ascending=True) px.box( dfbox, x="Railroad mileage", color="region", hover_name= 'province', animation_frame = 'year' )

px.box( df2, x="Road mileage", color="region", hover_name= 'province', animation_frame = 'year' )

China's rail and road mileage has grown in the decade from 2011 to 2020.

tl = Timeline() # 播放的速度，单位毫秒（ms） tl.add_schema(play_interval=300, symbol='emptydiamond') for i in range(2011, 2021): dfi=df2.query('year==@i') sort_info = dfi.sort_values(by='Road mileage', ascending=True) map0 = ( Bar() .add_xaxis(sort_info['province'].values.tolist()) .add_yaxis('', sort_info['Road mileage'].values.tolist()) .reversal_axis() .set_series_opts(label_opts=opts.LabelOpts(position="right")) .set_global_opts( title_opts=opts.TitleOpts(title='{}National provincial road mileage'.format(i)), datazoom_opts=[opts.DataZoomOpts(is_show=True, orient='vertical')], ) ) tl.add(map0, '{}年'.format(i)) tl.render_notebook()

tl = Timeline() # 播放的速度，单位毫秒（ms） tl.add_schema(play_interval=300, symbol='emptydiamond') for i in range(2011, 2021): dfi=df2.query('year==@i') sort_info = dfi.sort_values(by='Railroad mileage', ascending=True) map0 = ( Bar() .add_xaxis(sort_info['province'].values.tolist()) .add_yaxis('', sort_info['Railroad mileage'].values.tolist(),itemstyle_opts = opts.ItemStyleOpts(color="blue"),) .reversal_axis() .set_series_opts(label_opts=opts.LabelOpts(position="right")) .set_global_opts( title_opts=opts.TitleOpts(title='{}National provincial railroad mileage'.format(i)), datazoom_opts=[opts.DataZoomOpts(is_show=True, orient='vertical')], ) ) tl.add(map0, '{}年'.format(i)) tl.render_notebook()

Transportation infrastructure development varies widely from province to province in China, but is generally growing.

4.Regression analysis

4.1Model

Based on the data I developed the following model with the independent variable of rail miles/road miles and the dependent variable of GDP.

Model:ln(GDP)=α+βln(Road mileage)+u

4.2Visualiztion

4.2.1China

df1 = df.query('province=="China"') fig=px.scatter(df1, x='Road mileage', y='GDP', log_x='true', log_y='true', hover_name='province', trendline="ols", trendline_scope="overall" ) fig.show()

According to the country's economic development plan, China is divided into the central, eastern, western, and northeastern parts.

4.2.2Road mileage and GDP

df2 = df.query('province!="China"') fig=px.scatter(df2, x='Road mileage', y='GDP', log_x='true', log_y='true', color='province', hover_name='province', trendline="ols", trendline_scope="overall", facet_col= 'region', facet_col_wrap= 2, ) fig.show()

4.2.3Railroad mileage and GDP

4.2.4Conclusion

根据散点图，可以发现，铁路里程和公路里程和GDP数据存在正相关关系在中国的各个区域。

4.3Regression analysis of GDP and road mileage

4.3.1Regression analysis of GDP and road mileage

y = df2['lnGDP'] X = df2['lnRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.138
Model:                            OLS   Adj. R-squared:                  0.135
Method:                 Least Squares   F-statistic:                     49.18
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           1.49e-11
Time:                        08:43:50   Log-Likelihood:                -411.06
No. Observations:                 310   AIC:                             826.1
Df Residuals:                     308   BIC:                             833.6
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          4.6595      0.727      6.413      0.000       3.230       6.089
lnRM           0.4355      0.062      7.012      0.000       0.313       0.558
==============================================================================
Omnibus:                       25.685   Durbin-Watson:                   1.095
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               30.572
Skew:                          -0.670   Prob(JB):                     2.30e-07
Kurtosis:                       3.755   Cond. No.                         165.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.2Regression analysis of GDP and railroad mileage

y = df2['lnGDP'] X = df2['lnRRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.192
Model:                            OLS   Adj. R-squared:                  0.189
Method:                 Least Squares   F-statistic:                     73.00
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           6.10e-16
Time:                        08:43:50   Log-Likelihood:                -401.05
No. Observations:                 310   AIC:                             806.1
Df Residuals:                     308   BIC:                             813.6
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          5.0260      0.554      9.069      0.000       3.935       6.117
lnRRM          0.5866      0.069      8.544      0.000       0.451       0.722
==============================================================================
Omnibus:                        2.835   Durbin-Watson:                   1.063
Prob(Omnibus):                  0.242   Jarque-Bera (JB):                2.627
Skew:                          -0.154   Prob(JB):                        0.269
Kurtosis:                       2.670   Cond. No.                         90.3
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.3Regression analysis of GDP and road mileage in the Center

dfcenter=df2.query('region=="center"') y = dfcenter['lnGDP'] X = dfcenter['lnRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.849
Model:                            OLS   Adj. R-squared:                  0.847
Method:                 Least Squares   F-statistic:                     326.6
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           1.69e-25
Time:                        08:43:50   Log-Likelihood:                 22.679
No. Observations:                  60   AIC:                            -41.36
Df Residuals:                      58   BIC:                            -37.17
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const         -9.6443      1.093     -8.823      0.000     -11.832      -7.456
lnRM           1.6165      0.089     18.072      0.000       1.437       1.796
==============================================================================
Omnibus:                        0.150   Durbin-Watson:                   0.817
Prob(Omnibus):                  0.928   Jarque-Bera (JB):                0.216
Skew:                          -0.111   Prob(JB):                        0.898
Kurtosis:                       2.808   Cond. No.                         618.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.4Regression analysis of GDP and railroad mileage in the Center

y = dfcenter['lnGDP'] X = dfcenter['lnRRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.331
Model:                            OLS   Adj. R-squared:                  0.319
Method:                 Least Squares   F-statistic:                     28.70
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           1.52e-06
Time:                        08:43:50   Log-Likelihood:                -22.013
No. Observations:                  60   AIC:                             48.03
Df Residuals:                      58   BIC:                             52.21
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const         -0.4803      1.977     -0.243      0.809      -4.437       3.476
lnRRM          1.2603      0.235      5.357      0.000       0.789       1.731
==============================================================================
Omnibus:                       10.589   Durbin-Watson:                   2.159
Prob(Omnibus):                  0.005   Jarque-Bera (JB):               11.218
Skew:                          -1.054   Prob(JB):                      0.00366
Kurtosis:                       3.202   Cond. No.                         367.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.5Regression analysis of GDP and road mileage in the east

dfeast=df2.query('region=="East"') y = dfeast['lnGDP'] X = dfeast['lnRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.424
Model:                            OLS   Adj. R-squared:                  0.418
Method:                 Least Squares   F-statistic:                     72.18
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           2.20e-13
Time:                        08:43:50   Log-Likelihood:                -100.90
No. Observations:                 100   AIC:                             205.8
Df Residuals:                      98   BIC:                             211.0
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          4.5434      0.681      6.670      0.000       3.192       5.895
lnRM           0.5166      0.061      8.496      0.000       0.396       0.637
==============================================================================
Omnibus:                       20.578   Durbin-Watson:                   2.238
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               25.432
Skew:                          -1.124   Prob(JB):                     3.00e-06
Kurtosis:                       4.026   Cond. No.                         115.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.6Regression analysis of GDP and railroad mileage in the east

dfeast=df2.query('region=="East"') y = dfeast['lnGDP'] X = dfeast['lnRRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.373
Model:                            OLS   Adj. R-squared:                  0.367
Method:                 Least Squares   F-statistic:                     58.41
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           1.46e-11
Time:                        08:43:50   Log-Likelihood:                -105.12
No. Observations:                 100   AIC:                             214.2
Df Residuals:                      98   BIC:                             219.4
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          5.3542      0.651      8.222      0.000       4.062       6.647
lnRRM          0.6473      0.085      7.642      0.000       0.479       0.815
==============================================================================
Omnibus:                        7.945   Durbin-Watson:                   2.139
Prob(Omnibus):                  0.019   Jarque-Bera (JB):                8.264
Skew:                          -0.704   Prob(JB):                       0.0161
Kurtosis:                       2.942   Cond. No.                         72.8
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.7Regression analysis of GDP and road mileage in the west

dfwest=df2.query('region=="west"') y = dfwest['lnGDP'] X = dfwest['lnRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.592
Model:                            OLS   Adj. R-squared:                  0.589
Method:                 Least Squares   F-statistic:                     171.2
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           1.01e-24
Time:                        08:43:50   Log-Likelihood:                -115.85
No. Observations:                 120   AIC:                             235.7
Df Residuals:                     118   BIC:                             241.3
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const         -6.1390      1.166     -5.267      0.000      -8.447      -3.831
lnRM           1.2893      0.099     13.086      0.000       1.094       1.484
==============================================================================
Omnibus:                        9.438   Durbin-Watson:                   2.031
Prob(Omnibus):                  0.009   Jarque-Bera (JB):                9.782
Skew:                          -0.697   Prob(JB):                      0.00751
Kurtosis:                       3.111   Cond. No.                         237.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

R4.3.8egression analysis of GDP and railroad mileage in the west

dfwest=df2.query('region=="west"') y = dfwest['lnGDP'] X = dfwest['lnRRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.575
Model:                            OLS   Adj. R-squared:                  0.571
Method:                 Least Squares   F-statistic:                     159.6
Date:                Sun, 14 Aug 2022   Prob (F-statistic):           1.17e-23
Time:                        08:43:50   Log-Likelihood:                -118.33
No. Observations:                 120   AIC:                             240.7
Df Residuals:                     118   BIC:                             246.2
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.6418      0.672      0.955      0.341      -0.689       1.972
lnRRM          1.0508      0.083     12.631      0.000       0.886       1.216
==============================================================================
Omnibus:                       22.441   Durbin-Watson:                   1.126
Prob(Omnibus):                  0.000   Jarque-Bera (JB):                6.347
Skew:                           0.213   Prob(JB):                       0.0419
Kurtosis:                       1.957   Cond. No.                         92.3
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.9Regression analysis of GDP and road mileage in the Northeast

dfnortheast=df2.query('region=="Northeast"') y = dfnortheast['lnGDP'] X = dfnortheast['lnRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.000
Model:                            OLS   Adj. R-squared:                 -0.035
Method:                 Least Squares   F-statistic:                   0.01252
Date:                Sun, 14 Aug 2022   Prob (F-statistic):              0.912
Time:                        08:43:50   Log-Likelihood:                -6.4961
No. Observations:                  30   AIC:                             16.99
Df Residuals:                      28   BIC:                             19.79
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const         10.0842      3.133      3.218      0.003       3.666      16.503
lnRM          -0.0299      0.267     -0.112      0.912      -0.577       0.517
==============================================================================
Omnibus:                        6.634   Durbin-Watson:                   2.643
Prob(Omnibus):                  0.036   Jarque-Bera (JB):                2.910
Skew:                           0.478   Prob(JB):                        0.233
Kurtosis:                       1.811   Cond. No.                         652.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

4.3.10Regression analysis of GDP and railroad mileage in the Northeast

dfnortheast=df2.query('region=="Northeast"') y = dfnortheast['lnGDP'] X = dfnortheast['lnRRM'] X_withconst = sm.add_constant(X) OLS = sm.OLS(y, X_withconst).fit() print(OLS.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  lnGDP   R-squared:                       0.056
Model:                            OLS   Adj. R-squared:                  0.022
Method:                 Least Squares   F-statistic:                     1.651
Date:                Sun, 14 Aug 2022   Prob (F-statistic):              0.209
Time:                        08:43:50   Log-Likelihood:                -5.6434
No. Observations:                  30   AIC:                             15.29
Df Residuals:                      28   BIC:                             18.09
Df Model:                           1                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          5.6827      3.153      1.802      0.082      -0.776      12.142
lnRRM          0.4703      0.366      1.285      0.209      -0.279       1.220
==============================================================================
Omnibus:                        6.833   Durbin-Watson:                   2.660
Prob(Omnibus):                  0.033   Jarque-Bera (JB):                3.702
Skew:                           0.646   Prob(JB):                        0.157
Kurtosis:                       1.862   Cond. No.                         499.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

5.Concluding remarks

Through regression analysis, I got the following conclusions.

1. There is a regional imbalance in China's economic development, with the eastern region being the most developed and the western region is less developed. This regional imbalance also exists in transportation infrastructure development.

2. In general, transportation infrastructure construction (railroad construction and road construction) positively affects economic growth without considering the effects of other control variables.

3. Overall, railroad construction impacts China's economic development more than road construction.

4. In northeastern China, road construction hurts economic growth.

5. There are regional differences in the impact of transportation infrastructure on economic development, with transportation infrastructure construction having a more substantial effect on economic development in the central and western regions.