Manufacturing industry is the basis of economic development, and the industrial agglomeration brought by its benefits is generally valued. After entering the new normal state of the economy, China's economy has changed from high-speed growth to medium-high-speed growth. What are the spatial differences in the concentration of manufacturing industries and whether they are in a downward trend, and whether there is a positive correlation between the economic growth effects that follow are the points to ponder, and the different answers will bring different policy implications, which obviously have very important practical significance.
Industrial agglomeration based on the advantages of external economy, innovation benefits and competitive benefits can become a breakthrough point in economic development and structural adjustment. Industrial agglomeration and its effect on regional economic growth have also become one of the hotspots in academic research [1]. Due to the spatial distribution of the industry and its economic growth effects, the empirical research on industrial agglomeration has provided good practical guidance for scientific decision-making [2] and has attracted much attention from scholars.
Early related research started from abroad, and most of them indirectly explored the relationship between economic activity concentration and economic growth from some aspects such as labour productivity, urban-isation and market size. With the deepening of research, people have turned to the direct test of industrial agglomeration and economic growth. For example, Ciccone [3] researched the economic agglomeration of five European countries to regional economic growth and found that the two are mutually reinforcing. Geppert et al. [4] and other empirical tests on industrial agglomeration and economic growth in Germany from 1980 to 2000 showed a positive relationship between the two. In the above research, because the indicators for measuring industrial agglomeration are relatively simple, the endogenous nature of agglomeration is not considered, and the data used are mostly cross-sectional in nature; so the dynamic relationship between industrial agglomeration and economic growth cannot be examined. Since then, dynamic analysis based on panel data has gradually increased. For example, Bruhlhart and Sbergami [5] established a dynamic panel data model analysis of 105 countries during 1960–2000 and the European Union in 1975–2000 through a systematic GMM method by considering endogenous issues fully; the results show that the industrial agglomeration of these regions has promoted their economic growth.
Also, in China, there are many studies on industrial agglomeration and its relationship with economic growth. Most conclusions, such as Liu [6], show that there is a mechanism of interaction between industrial agglomeration and economic growth, which is a pair of endogenous processes. From the analysis of China's practical data, there is also a negative correlation [7], insignificant [8] or non-linear relationship [9, 10]. It can be said that different regions, industries, internal and external conditions, historical evolution, etc. will produce different economic agglomeration characteristics, and their corresponding effects on regional economic growth also have different correspondences.
In general, there are still some shortcomings in the current related research. The first is that most of the research is conducted in a static framework, and the research that reflects the dynamic characteristics of practical development has not yet become mainstream; the second is that some literature have simple setting indicators, which do not reflect the degree of industrial agglomeration well. In addition, in the empirical analysis of building data models, many literature do not pay enough attention to endogenous issues. Traditional panel data models cannot guarantee the unbiasedness of parameter estimates, which affects the acceptance of conclusions. Therefore, some literature have begun to use dynamic panel data model for analysis [11, 12].
China's economy has entered the new normal, which is consistent with the macro-economic development of other countries. And with the rise of Chinese manufacturing, the new normal has gradually changed in recent years. To reflect the situation of our country's industrial agglomeration after the economy enters the new normal state more accurately, this paper takes the 2012–2016 regional manufacturing data as the object and uses a two-stage system method to analyse the regional manufacturing industry agglomeration and its economic growth effects through the construction of EG indexes and dynamic panel data models. Because the GMM model can use the information of the difference and level equation variables to construct instrumental variables to control the endogenous problems of the explanatory variables, it can better reveal the hidden phenomena or problems in economic development.
EG indexes are more commonly used in research because they can more accurately measure and reflect the degree of industrial agglomeration. Under normal circumstances,
Drawing on the practice of Lu and Tao [13], the general calculation formula for EG index is:
In Eq. (1),
In Eq. (2),
In Eq. (3),
This paper calculates the
Industrial agglomeration status of 31 provinces and cities in China's manufacturing industry from 2012 to 2016
Area | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
2012 | 2016 | Increase or decrease | Increase | 2012 | 2016 | Increase or decrease | Increase | Rank | Rank increase or decrease | |
Beijing | 0.013 | 0.016 | 0.002 | 18.92 | 0.013 | 0.016 | 0.002 | 19.08 | 13 | 0 |
Tianjing | 0.011 | 0.005 | −0.006 | −58.1 | 0.011 | 0.005 | −0.006 | −58.1 | 24 | −8 |
Hebei | 0.014 | 0.018 | 0.004 | 26.74 | 0.014 | 0.018 | 0.004 | 27.09 | 11 | 1 |
Shanxi | 0.017 | 0.008 | −0.01 | −56 | 0.017 | 0.008 | −0.01 | −56.1 | 16 | −7 |
Neimenggu | 0.004 | 0.006 | 0.002 | 59.59 | 0.004 | 0.006 | 0.002 | 59.77 | 20 | 5 |
Liaoning | 0.044 | 0.026 | −0.018 | −41.5 | 0.044 | 0.026 | −0.018 | −41.6 | 9 | −2 |
Jilin | 0.011 | 0.018 | 0.007 | 66.7 | 0.011 | 0.018 | 0.007 | 66.76 | 10 | 7 |
Heilongjiang | 0.015 | 0.005 | −0.01 | −64.1 | 0.015 | 0.005 | −0.01 | −64.1 | 23 | −12 |
Shanghai | 0.026 | 0.028 | 0.002 | 9.247 | 0.026 | 0.028 | 0.002 | 9.628 | 7 | 1 |
Jiangsu | 0.057 | 0.11 | 0.053 | 92.14 | 0.057 | 0.11 | 0.053 | 93.23 | 2 | 3 |
Zhejiang | 0.140 | 0.081 | −0.06 | −42.5 | 0.141 | 0.081 | −0.06 | −42.6 | 3 | 0 |
Anhui | 0.005 | 0.004 | −0.002 | −29.2 | 0.005 | 0.004 | −0.002 | −29.5 | 26 | −3 |
Fujian | 0.158 | 0.075 | −0.082 | −52.1 | 0.158 | 0.075 | −0.083 | −52.2 | 4 | −2 |
Jiangxi | 0.004 | 0.007 | 0.003 | 82.91 | 0.004 | 0.007 | 0.003 | 83.03 | 17 | 9 |
Shandu | 0.066 | 0.054 | −0.012 | −17.9 | 0.066 | 0.054 | −0.012 | −18 | 5 | −1 |
Henan | 0.011 | 0.027 | 0.015 | 132.8 | 0.011 | 0.026 | 0.015 | 134.2 | 8 | 10 |
Hubei | 0.013 | 0.012 | −0.001 | −9.58 | 0.013 | 0.012 | −0.001 | −9.67 | 15 | −1 |
Hunan | 0.012 | 0.007 | −0.005 | −39.5 | 0.012 | 0.007 | −0.005 | −39.5 | 18 | −3 |
Guangdong | 0.168 | 0.429 | 0.262 | 156.2 | 0.169 | 0.44 | 0.272 | 161.2 | 1 | 0 |
Guangxi | 0.009 | 0.006 | −0.002 | −25.5 | 0.008 | 0.006 | −0.002 | −25.5 | 19 | 1 |
Hainan | 0.000 | 0.000 | 0.000 | −22.6 | 0.000 | 0.000 | 0.000 | −22.6 | 30 | −1 |
Chongqing | 0.008 | 0.006 | −0.003 | −31.9 | 0.008 | 0.005 | −0.003 | −32.1 | 22 | −1 |
Sichuan | 0.017 | 0.016 | −0.001 | −5.47 | 0.017 | 0.016 | −0.001 | −5.4 | 12 | −2 |
Guizhou | 0.007 | 0.004 | −0.003 | −47.8 | 0.007 | 0.004 | −0.003 | −47.8 | 27 | −5 |
Yunnan | 0.047 | 0.044 | −0.002 | −5.2 | 0.047 | 0.044 | −0.002 | −5.19 | 6 | 0 |
Xizang | 0.000 | 0.000 | 0.000 | 2.598 | 0.000 | 0.000 | 0.000 | 2.594 | 31 | −1 |
0 | ||||||||||
Shanxi | 0.009 | 0.012 | 0.003 | 33.94 | 0.009 | 0.012 | 0.003 | 34.07 | 14 | 5 |
Gansu | 0.005 | 0.006 | 0.001 | 15.31 | 0.005 | 0.006 | 0.001 | 15.39 | 21 | 3 |
Qinghai | 0.001 | 0.001 | 0.000 | −31.4 | 0.001 | 0.001 | 0.000 | −31.4 | 29 | −1 |
Ningxia | 0.000 | 0.001 | 0.000 | 42.15 | 0.000 | 0.001 | 0.000 | 42.2 | 28 | 3 |
Xinjiang | 0.004 | 0.004 | 0.000 | 12.51 | 0.004 | 0.004 | 0.000 | 12.51 | 25 | 2 |
From Table 1, we can see that after the new normal economy, the agglomeration status of China's manufacturing industry has basically remained stable, and there has been no obvious downward trend, but it is still in a moderate agglomeration status as a whole. In the past five years, the degree of industrial agglomeration in China's manufacturing industry has increased. The average regional
In terms of index changes, most of the provinces with higher levels of industrial agglomeration have declined, while some provinces with a low degree of agglomeration have increased, and the differences in the degree of agglomeration of manufacturing between regions have slowed down, showing an overall favourable development trend.
In addition, the ranking of the degree of manufacturing agglomeration in each region has changed greatly, but the regions with a higher degree of agglomeration still continue to present a ‘one pole, two domains’ pattern, that is, a leading pattern with ‘Guangdong as the pole and supported by the Pearl River Delta and the Yangtze River Delta’. In terms of ranking, the top six provinces have not changed, but the overall ranking has changed significantly. The largest increases were in Henan and Jiangxi, whereas the largest declines were in Heilongjiang, Tianjin and Shanxi. Among the declining provinces, one reason was the rational adjustment made due to the high degree of industrial agglomeration, such as in Fujian and Zhejiang; the other was that the old industrial base was affected by factors such as industrial structure transformation and upgrading, zombie corporate governance policies, and slow industrial transformation and upgrading. Third, the backward manufacturing areas are subject to a variety of internal and external factors, and the aggregation effect of resource elements is low. On the whole, the degree of industrial agglomeration in the Yangtze River Delta and the central and western regions has increased. This situation is basically consistent with the status of regional economic development.
To reveal the spatial changes of manufacturing agglomeration more intuitively, according to the changes of manufacturing industry index of 31 provinces and cities in 2016, a map of manufacturing agglomeration changes in each region was made. It can be seen that the region with a higher concentration of manufacturing industries still presents a ‘one pole, two domains’ pattern; this situation of continuous leading has been in less change.
To reflect the dynamic characteristics of practical development and the endogenous problems of data processing, this paper analyses through the construction of a dynamic panel data model and the application of a two-phase system GMM method.
Regional economic growth is the result of the combination of multiple variables such as factor inputs, innovation clusters and natural conditions. Based on the Cobb-Douglas production function, this paper adds agglomeration variables that reflect spatial factors to reflect the basic conditions of regional economic growth from the perspective of time and space. The model is set as follows:
Among them, the dependent variable
To reduce the estimation bias caused by missing variables, the three corresponding control variables are added to the model. (1) Fixed assets investment (assets): It is an important input factor for manufacturing enterprises in production. Its size reflects the scale and strength of the industry. It is a necessary condition for production and can be used to measure the degree of input of internal factors in manufacturing. (2) Government expenditure (ge): On the one hand, the government directly supports the local manufacturing funds, and on the other hand, the government stimulates demand through expenditure, stimulates production, and promotes the development of manufacturing. The general budgetary expenditures of local governments are generally used to measure the degree of government expenditure and the degree of market pull. (3) Dummy variables: Add dummy variables for year and province to control fixed effects of time and region.
For model's robustness and endogenous problems, to overcome the measurement errors caused by different units, initially, all the variables with economic units were used with their logarithmic values. In addition, to consider the inertia and path dependence of China's economic growth, referring to the practice of Sun et al. [14], without using manual setting of instrumental variables, a systematic GMM method was selected to estimate the effects of economic growth and industrial agglomeration to control endogenous problems. Therefore, based on Eq. (4), the specific model of this study is set as:
In Eq. (5), in addition to using the lag period value of the explanatory variable, relevant dummy variables are added to control the fixed effects of time and region. At the same time, the dimensions of the independent and dependent variables are different, and natural logarithmic processing is performed on all continuous variables.
To more clearly show the situation after the new normal of the economy, the data used in this article are from the China Statistical Yearbook 2012 and the China Industrial Economic Statistical Yearbook 2012–2017, and the samples were checked with the National Bureau of Statistics and other websites. Due to the lack of individual data in Tibet, the arithmetic mean of the previous and subsequent years was used in the study to complete the data to ensure the integrity of the data.
Simple descriptive statistics of each variable are shown in Table 2.
Descriptive statistics of variables
Variable | Observations | Meaning | Unit | Mean | Standard deviation | Minimum value | Max | LLC inspection |
---|---|---|---|---|---|---|---|---|
155 | Industrial added value | 100 million yuan | 8.592 | 1.241 | 4.014 | 10.394 | −12.186 | |
155 | Index | - | 0.033 | 0.075 | 0.000 | 0.474 | −71.246 | |
155 | Exponent square | - | 0.007 | 0.033 | 0.000 | 0.225 | −53.053 | |
155 | Manufacturing fixed asset investment (excluding farmers) | 100 million yuan | 7.947 | 1.364 | 3.362 | 10.060 | −4.206 | |
155 | Fiscal General Budget Expenditure | 100 million yuan | 8.226 | 0.568 | 6.762 | 9.506 | −19.127 | |
155 | Time-controlled variable | year | 2014 | 1.419 | 2012 | 2016 | - |
To avoid the non-stationary pseudo-regression of the data variables, the LLC method is used for testing. Each variable passes the unit root test at the level of 1% (the last column of Table 2). The variables are
Examining the scatter point relationship between the main variables (see Figure 1), it is found that there is a nonlinear relationship between the industrial added value and the
Model regressions such as inter-group effects, random effects, and fixed effects were performed on
Estimated results of
Explained variable | |||||
---|---|---|---|---|---|
Explanatory variables | Fe model | Re model | Be model | A model | B model |
0.83*** | 0.85*** | ||||
14.8 | 14.02 | ||||
−0.10 | −0.04 | 0.97 | 1.07 | 2.86*** | |
−0.38 | (−0.13) | −1.04 | 0.84 | 3.05 | |
−5.2** | |||||
−2.18 | |||||
0.61*** | 1.01*** | 0.94*** | 0.16*** | 0.15*** | |
5.01 | 9.45 | 5.17 | 2.96 | 2.13 | |
0.21*** | 0.30*** | 0.54*** | 0.14*** | 0.13*** | |
6.00 | 8.53 | 8.03 | 3.25 | 3.26 | |
106.66*** | 196.12*** | −3.49*** | 47.44*** | 48.40*** | |
4.40 | 9.08 | (−3.06) | 6.17 | 5.27 | |
Time control | −0.05*** | −0.10*** | (omitted) | −0.02*** | −0.02*** |
−4.17 | (−8.90) | (−6.13) | (−5.20) | ||
Area control | Yes | Yes | |||
1326.05 | 1269.25 | ||||
0 | 0 | ||||
7.7003 | 9.0495 | ||||
0.3598 | 0.2491 | ||||
0.0483 | 0.0494 | ||||
0.9076 | 0.8012 |
(2) Autocorrelation test uses Arellano-Bond second-order autocorrelation test;
(3) The over-identification test uses the Sargan test. The null hypothesis is that all moment conditions in the model are valid.
Further, to eliminate the possible explanatory variables of the dynamic panel model and its related endogenous problems with random error terms, especially the heteroscedasticity, autocorrelation and individual effects that may exist in the ‘large N small T’ panel model, based on the existing results achieved in the literature, this paper uses a two-stage system dynamic GMM method to estimate the model. The specific operation is realised by the STATA15.0 tool. On the one hand, difference equations are used to eliminate fixed effects; on the other hand, the lag term of the difference term is used as the instrumental variable of the horizontal term to increase the number of instrumental variables and solve the problem of weak instrumental variables of the horizontal lagging term. The estimation results are shown in the column ‘A model’ in Table 3.
In terms of robustness test, since the lag term of the difference variable is introduced as a new instrumental variable, in order to prevent transitional identification caused by improper selection of instrumental variables, a
The
This paper uses the calculation of the The degree of manufacturing agglomeration in various regions of China has changed greatly, but the overall situation has remained stable and is in a state of moderate agglomeration. The regions with higher levels of agglomeration still present a ‘one pole, two domains’ pattern. After entering the new normal state of the economy, the economic growth effect brought by the agglomeration of manufacturing industries in various regions of China is not a simple linear relationship, and there is still a threshold effect.
It can be seen from Table 3 that after introducing the nonlinear explanatory variable
There is a dynamic continuation effect of regional economic growth. It has a positive linear relationship with fixed asset investment, etc., and the government fiscal expenditure pull is higher than the fixed asset investment pull.
Table 3 shows that the coefficient of
These conclusions can bring some important policy implications:
After the economy enters the new normal, China's manufacturing industry agglomeration status has a large gap between regions, and regional changes are also rapid. This requires the overall control of the reasonable flow of industries between regions. The continuous flow and pooling of resources (such as capital, technology), especially high-quality resources, to developed and more developed regions will inevitably cause the industrial economy to intensify in terms of the degree of geographical agglomeration. Industry “manufacturing agglomeration. Under the influence of many factors such as declining demand in the foreign market, accelerating the adjustment and upgrading the domestic industrial structure and increasing environmental pollution, the problems of the decline in the concentration of manufacturing industries in the Northeast and the accumulation of industries in the backward areas in the West should be paid attention to. The threshold effect of economic growth brought by the agglomeration of manufacturing industries is worth vigilant. For economically underdeveloped regions, the effective allocation of various factors and resources is needed to drive the growth of industrial agglomeration, whereas for economically developed regions, more consideration must be given to the high-quality development of industries under the new economy. In the process of stimulating industrial economic growth, regional authorities should give full play to their own micro-policies, such as the government's active fiscal policy, the vigorous support of technological innovation and the establishment of a healthy economic environment. It can be said that China's economic growth has obvious path dependence, and it is necessary to dig out the ways and methods to solve the structural contradictions in restricting economic transformation and upgrading the problems of uneven and insufficient regional economic development.
At present, with the promotion of innovation-driven role and the rapid growth of new industries, new economies and new kinetic energy in economic growth, China's economy is expected to accelerate the development of high quality and continue to maintain its stable operation development trend, manufacturing under the new normal. Industry agglomeration will also exert its positive effects due to the continuous adjustment and upgrading of the industrial structure.