Advances in Social Science, Education and Humanities Research, volume 176 2nd International Conference on Management, Education and Social Science (ICMESS 2018) Analysis of Spatial-temporal Dynamics and Influencing Factors of County Economic Disparity: A Case Study of Gansu Province Yan Han, Yuan Zhang, Meiling Deng Department of Economics and Management Lanzhou Jiaotong University Lanzhou China Abstract—This paper uses the analytical framework of [3].The above studies provide useful analytical approaches for exploratory spatial-temporal data analysis (ESTDA) to analyze understanding regional economic differences. the spatial-temporal dynamics and influencing factors of county- level economic disparities in Gansu Province from 1995-2015. The A large number of empirical studies have shown that spatial results are as follows: Through the analysis of spatial-temporal dependence and heterogeneity are ubiquitous and not an evolution, it is found that the regional economy of Gansu Province exception. Economic development differences themselves has a local spatial differentiation while the spatial agglomeration coexist with spatial patterns of temporal dynamics and spatial area is expanding. The phenomenon of club convergence has patterns of temporal behavior. On the other hand, more and emerged in regional economic development, and the regional more scholars have noticed that the temporal and spatial economy has obvious path dependence characteristics. The spatial properties of data are equally important, but the existing regression model was used to analyze the factors affecting regional research methods focus on only one of them, namely spatial type economic differences in Gansu Province. Among them, the degree analysis (formal analysis) and time series analysis (process of decentralization and marketization have greater impact on analysis) is often separated [11-12]. However, the fact that spatial differentiation. cannot be ignored is that the dynamic process of regional economic development in empirical research includes two Keywords—Exploratory spatial-temporal data analysis; dimensions of time and space. Based on the effective integration Agglomeration; Spatial Markov chain; Influencing factors of EDA (integration of spatial elements) and ESDA (integration I. INTRODUCTION of time factors), Rey proposed the framework of exploratory spatial data analysis (ESTDA), and the coupling and visual As a geographical phenomenon with unbalanced display of spatial-temporal correlation through the introduction development, regional economic differences are not only an of graph theory is a more successful exploration. important topic for academic research, but also problems that cannot be ignored in the process of regional economic Based on the ESTDA analysis framework, this paper takes development. The research on regional economy can be traced Gansu Province as the research object and takes the county area back to the 1950s, and gradually formed a theory of neoclassical as the study scale to analyze the spatial-temporal evolution and theory, inverted U-shaped development theory, growth pole change process of regional economic differences, and adopts theory, core edge theory, imbalance theory, etc. The subject three models: linear regression model, spatial lag regression perspective is based on the multidisciplinary comprehensive model and spatial error regression model, to analyze the factors intersection of regional economics, econometrics, and to provide useful help for the coordinated development of environmental science, and geography [1-6]; the research scale regional economy. includes global, national, provincial (state), municipal, county, etc. The provincial scale is the main trend, and it is gradually II. OVERVIEW OF THE STUDY AREA becoming smaller. This is because the differences of small-scale Gansu Province is located in the northwest fortress of China spatial units are more sensitive to fluctuations in the economic and is an important channel. As of the end of 2015, Gansu environment; research methods range from single indicators to Province has a land area of 453,700 square kilometers. It has multiple indicators, from measurement models to jurisdiction over 14 prefecture-level cities and 58 counties with comprehensive analysis, and from single scales to Multi-scale a total population of 25,995,500 people and a urbanization rate comprehensive transformation, spatial measurement methods of 43.19%.This paper uses 87 counties (districts) of Gansu have gradually become the mainstream research methods; the Province as the basic research unit for the analysis of regional main factors affecting the differences are globalization, policy economic differences and spatial pattern evolution, and divides tilt, decentralization and urbanization [9], resource and the entire province into four regional units1: Hexiregion, the environment, industry basis, location factors [7-10], spatial central region of Gansu, Southeast Yunnan, and the plateau neighbors Effects, special factors of the provincial border area margin according to geographical conditions. This paper is sponsored by Gansu Science and Technology Agency Soft Science Project. (Number: 1504ZKCA017-3); Lanzhou Science and Technology Plan Project (number: 2014-1-249) Copyright © 2018, the Authors. Published by Atlantis Press. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). 1198 Advances in Social Science, Education and Humanities Research, volume 176 III. RESEARCH METHODS probability matrix Decomposes a conventional k×k Markov matrix into k k×k conditional transition probability matrices. A. Differences and agglomeration indicators For the k-th conditional matrix, the element (k) represents the This paper uses the Global Theil and Moran’s I indices to region at t The spatial delay type k of the year is a measure the relationship between regional economic differences background condition, and the year belongs to the space and spatial dependence over time. The Global Theil coefficient, transition probability that the type i is transferred to the type which measures global relative differences, is calculated as j in the next year. follows: 푚(1,1|1) ⋯ 푚( | ) 푛 1 1, 푘 1 T = ∑푖=1 푆푖,푡 [log 푆푖,푡 − log( ⁄푛)] (2.1) 푀1 = ⌈ ⋮ ⋱ ⋮ ⌉ ⋯ 푀푘 = The degree of spatial agglomeration is measured using the 푚( | ) … 푚( | ) 푘, 1 1 푘, 푘 1 푘∗푘 global autocorrelation Global Moran’ I. The formula is as 푚 ⋯ 푚 follows: (1,1|푘) (1, 푘|1푘) ⌈ ⋮ ⋱ ⋮ ⌉ (2.4) 푛 푤푖,푗(푥푖,푗−푥̅̅푡̅)(푥푗,푡−푥̅̅푡̅) 퐼 = ∑ ∑ (2.2) 푚( | ) … 푚( | ) 푗푖 2 푘, 1 푘 푘, 푘 푘 푘∗푘 푆0 (푥푖,푡−푥̅̅푡̅) C. Spatial Regression Analysis where S ,ti is the proportion of the province's GDP in the In order to better reveal the dynamic mechanism of regional t year i region, N = 87, x for the county i in the t year economic development and differences in Gansu Province, this ，ti paper adopts the traditional linear regression model and spatial of GDP per capita and the ratio of the province's per capita regression model analysis to conduct quantitative research, and GDP, S0 is the sum of all elements in the weighting matrix compares the significance of several regression models so as to better conduct the influencing factors. Quantitative analysis. for the space, xt is the mean of a random variable xt for t IV. SOURCES OF DATA years, and x , x the observed value of county iand jin ，ti ,tj The original data of this paper comes from the <<Statistical the t year, wij for spatial weight matrix elements; I ]1,1[ , Yearbook of Gansu Province>> (2002-2016), and the data in <<China City Statistical Yearbook>> in the corresponding The closer to 1 indicates that the positive space years. The 15-year period from 2002-2016 was selected as the agglomeration is stronger, and vice versa. time series spectrum, 87 spatial units, and 1305 observations. The local spatial autocorrelation coefficient measures the The GDP per capita is used to represent the level of economic degree of dispersion of each spatial unit and identifies the development in the county; taking into account the availability spatial form of the agglomeration area to fully reflect the of data, five indicators are selected, namely, financial changing trend of regional economic spatial differences. expenditure (FE), total retail sales of consumer goods (MAR), Calculated as follows: total fixed-asset investment (FIX), and proportion of industrial output(IND) and(SAVING)for urban and rural residents )1( zwzn respectively represent government decentralization, degree of j , jjii Ii (2.3) marketization, regional policy tilt, industrial development level, z2 and residents' storage capacity. j j V. RESEARCH RESULTS The values of 퐼푖 and 푍푖 can indicate the clustering type of the location i and its neighbors, respectively, which are A. Evolution of County Economic Disparity and high clustering, low concentration, high and low Agglomeration concentration, and low concentration. Since 2002, the economic development of Gansu Province B. Spatial Markov chain has been rapidly increased, and the per capita GDP has increased from 4,768 yuan to 27,643 yuan from 2002 to 2016. Markov chain is an effective method to analyze the The average annual increase has been large, but there has been convergence of regional economic growth clubs. It an uneven regional economic development. As can be seen from discretizes the data of continuous attributes of geographical the figure, the Global Theil value of county per capita GDP in phenomena at different times, converts them into k types