The Economic Recovery from Traffic Restriction Policies during the COVID-19 through the Perspective of Regional Differences and Sustainable Development: Based on Human Mobility Data in China

Abstract

In the post-epidemic era, balancing epidemic prevention and control with sustainable economic development has become a serious challenge for all countries around the world. In China, a range of interventions include detection policies, clinical treatment policies, and most notably, traffic policies have been carried out for epidemic prevention and control. It has been widely confirmed that massive traffic restriction policies effectively brought the spread of the pandemic under control. However, restrictions on the use of transportation infrastructure undermine the smooth functioning of the economy. Particularly, China has a vast territory, with provinces differing in economic development, leading industries and transportation infrastructure; economic shock varies from region to region. In this case, targeted policies are the key to sustainable development. This paper sets forth advice for the Chinese government on its measures to boost the economy by analyzing regional differences in the impact of massive traffic restriction policies, based on large-scale human mobility data. After applying the Data Envelopment Analysis model, we classify Chinese provinces into different regions from the perspective of economic gradient, degree of internationalization and level of traffic convenience, respectively. Classification results are matched with the indicators of New Venues Created and the weekly Volumes of Visits to Venues from Baidu Maps. We find that the regional differences in the recovery of investment and consumption levels are striking. Based on the findings, we suggest that the government should adjust the intensity of traffic restrictions and economic stimulus policies dynamically according to regional differences to achieve sustainable economic development.

1. Introduction

In early 2020, the COVID-19 pandemic broke out in Wuhan, Hubei Province, China. A range of interventions including detection policies, such as improved rates of diagnostic testing, clinical treatment policies, and most notably, traffic policies, such as isolation and restrictions on mobility have been carried out [1,2,3]. On 23 January 2020, the Wuhan City Epidemic Prevention and Control Department issued a large-scale traffic control notice: buses, subways, ferries and long-distance passenger transport were temporarily suspended; people should not leave Wuhan without special reasons, and the airport and railway station were temporarily closed. Then, other regional governments in China also implemented traffic restrictions policies, such as travel restrictions and public traffic suspensions. These measures for the prevention and control of COVID-19 are all proven to be successful strategies [4]. However, the implementation of various traffic control policies together with other factors related to COVID-19, such as the cancellation of business activities, closure of venues, rattled stock market and layoffs, and those unrelated, such as population change and trade disputes have dealt a major blow to macroeconomic operations and social production in China [5,6,7,8]. According to the National Bureau of Statistics of China (www.stats.giv.cn, accessed on 12 December 2021), in the first quarter of 2020, the GDP in China fell by 6.8% year-over-year; of which, the primary sector was agriculture, grazing, trapping, forestry, fishing and mining, the secondary sector was manufacturing and the tertiary sector was industries excluded by the secondary industries [9], fell by 3.2%, 9.6% and 5.2% year-over-year, respectively. Other vital parameters, such as industrial added value, index of services production, fixed investment, total retail sales of consumer goods and national financial general public budget, dropped by −13.5%, −13.0%, −24.5%, −20.5% and −9.9% year-over-year, respectively.

In particular, it has been widely confirmed that massive traffic restriction policies effectively brought the spread of the pandemic under control [10,11,12]. In China, traffic restriction policies have become the most common measurement for epidemic prevention and control. However, traffic restriction policies during the COVID-19 pandemic limit the utilization of transportation infrastructure. The relationship between transportation infrastructure and economic growth has always been the focus of economists. Quantitative research in this field mainly includes analysis of the production effect [13,14,15] and the spillover effect [16,17] of transportation infrastructure on regional economic growth, but the conclusions are not consistent. In this case, we believe it is an opportunity to adopt a new perspective, that is, traffic restriction, to explore the relationship between transportation and the economy. In addition, China is a country of 34 provincial administrative regions and different provinces are intricately interconnected; therefore, it is essential to analyze the impact of massive traffic restrictions from the perspective of regional differences.

In the research field of regional development in China, few studies have established a rigorous index system to measure the development of provinces and the bulk of them use rather general descriptions, such as ‘coastal’, ‘great northwest’, ‘southeast ‘, or by grouping adjacent provinces together for regional classification. Since the economic situation varies greatly among different provinces in China, only by adopting a set of indicators that can comprehensively reflect the level of income, productivity, technology, import/export trade and traffic infrastructure of provinces, can we classify regions objectively.

Above all, this study innovatively introduces data envelopment analysis models into the evaluation system, which is used to calculate the relative efficiency among provinces from the perspectives of economic gradient, degree of internationalization, and level of traffic convenience, to make a comprehensive ranking and classification. Next, we studied the impact of traffic restriction policies during the COVID-19 pandemic on economic recovery in classified regions based on indicators representing the level of consumption and investment, respectively, from Baidu Maps, which had over 389 million monthly active users by the end of March 2021.

Meanwhile, the pandemic presents a diverse impact on sustainable economic development. On the one hand, traffic restrictions have led to dramatic changes in the way people live and work, including the increase in telecommuting, virtual meetings and the new online economy, as well as the decrease in long-distance business trips and public transport. These changes are beneficial for sustainable development and might persist beyond the pandemic. On the other hand, the United Nations has assessed progress on 17 Sustainable Development Goals (SDGS) covered by the 2030 Agenda [18], and results show that efforts to achieve the SDGS have been hit by COVID-19, with years of progress in fields, such as poverty reduction and medical services vanished. In fact, whether it is an opportunity or a challenge depends on the government’s measures. Further, this paper combines the concept of sustainable development with regional analysis to provide policy recommendations for the government.

The rest of the paper is organized as follows: In the next section, we review the relevant literature. In Section 3, we introduce DEA models for region classification. Further, we analyze and discuss the indicators of New Venues Created and the weekly Volumes of Visits to Venues from Baidu Maps in the classified regions. In Section 5, we discuss the results in the view of economic gradient, degree of internationalization and level of traffic convenience. In Section 6, we put forward policy recommendations for the sustainable development of regional economies.

2. Literature Review

Pioneer researchers mainly conducted qualitative research on the relationship between transportation infrastructure and economic development. Development economists, such as Rodan (1943) and Rostow (1953) believed that transportation is a kind of social advance capital and an important prerequisite for the realization of economic take-off, which must be developed in priority [19,20]. Since the 1980s, increasing quantitative research is emerging in this field and it mainly includes the following aspects. On the one hand, the production effect of regional transport infrastructure is studied from the perspective of output elasticity and total factor productivity [13,14,15]. On the other hand, the spillover effects of transportation infrastructure on regional economic growth are examined and the research conclusions include positive spillover effects [16] as well as negative spillover effects [17].

From the perspective of neoclassical economic theory, sustained rapid growth in the long term must be accompanied by a rapid increase in total factor productivity [21]. Transportation infrastructure, which influences the growth of total factor productivity in the aspects of level of technology [22], allocation efficiency [23] and scale efficiency [24], is considered essential to economic sustainable development. Specifically, the development of transportation infrastructure is conducive to promoting and expanding inter-regional exchanges of personnel and goods, driving the dissemination of knowledge and technology, optimizing the allocation of resources, as well as economic agglomeration and market expansion.

However, traffic restriction policies, which are the most common measurement for epidemic prevention and control in China during the COVID-19 pandemic limit the utilization of transportation infrastructures and lead to a recession in many industries. The tourism industry and aviation industry are at the top of the list. A study shows that traffic restriction policies have greatly influenced the tourism industry through quantitative transversal analysis [25]. Studies that also focus on sustainable tourism development during the epidemic include Marek (2021) [26], Jones et al. (2020) [27] and Romagosa (2020) [28]. Besides, Liu et al. (2020) focus on the volatility estimation of stock indexes in the Chinese Airport Shipping Set at industry-enterprise levels to explore how the aviation industry has evolved after the traffic restrictions [29].

More broadly, transportation restrictions during COVID-19 also brought sustainable development challenges to supply chains and international trade [30]. Kumar et al. (2020) [31] argue that the pandemic has broken most transport links and distribution mechanisms between suppliers, producers and customers, thus challenging sustainable production models. Among them, the food supply chains have been hit the most [32,33]. In terms of international trade, Narasimha et al. (2021) show that maritime industries account for 90% of the international trade that has been impacted. Traffic restrictions resulted in the disruptions of shipping and border shutdowns, leading to a dramatic increase in the trade costs, and ultimately having a significant impact on international trade [34].

Based on the above analysis, we propose that transportation restrictions have a certain impact on the economy. Existing literature only takes this factor as one of the simple mediators when discussing the negative impact of the epidemic. In fact, transportation as an economy-related factor should be taken independently for analysis. In addition, there are conflicting conclusions on the impact of transportation infrastructure on economic development, and we believe it is an opportunity to adopt a new perspective, using traffic restrictions to explore the relationship between transportation and the economy. Besides, China is a country of 34 provincial administrative regions and different provinces are intricately interconnected; therefore, it is essential to analyze the impact of massive traffic restrictions from the perspective of regional differences.

Most studies on China’s regional development are based on the economic gradient, which points to the gaps in economic development levels between regions [35]. For instance, Qu et al. (2017) use GDP per capita as a characteristic indicator to delineate the economic gradient of each county-level administrative region and reveal the correlation between the transition of rural settlements and economic development [36]. Qin et al. (2018) select GDP per capita and the ratio of secondary and tertiary industries as indicators to measure the economic gradient of 42 cities in the Yangtze River Delta region [37]. In addition, Liu et al. (2017) construct a population location index to explore the gradient differences in economic development levels among coastal, riverine and inland cities [38].

In addition to the impact of regional economic differences, we further notice that the more internationalized the region is, the more dependent it is on foreign import and export trade, and the more severely it is affected by the epidemic and border blockades since COVID-19 has distorted the trade network [39]. Additionally, the radiation theory of regional economies illuminates the interaction and diffusion between modernization and economic development [40]. In this case, the medium of radiation is transportation infrastructure [41], and the advantages of the medium determine the efficiency of radiation, i.e., the level of traffic convenience in a region. Specifically, the convenience of railroad, highway, shipping, and air transport network connections, determines the intensity of economic interactions between regions, and thus may also affect the economy of regions during and after the traffic restrictions.

Taken together, economic gradient, degree of internationalization and level of traffic convenience are three complementary dimensions that form a relatively complete and systematic view to measure regional development. Therefore, we set out to classify provinces of China by integrating these three dimensions, drawing on the indicators used in regional development measurements.

In particular, we innovatively introduce the DEA model in terms of classification methods. The Data Envelopment Analysis (DEA) model can effectively solve problems of relative effectiveness evaluation among departments with multiple inputs and outputs. Since Charnes and Cooper (1978) [42] proposed the DEA method for evaluating and controlling managerial behavior in public programs, the method has been popular in the management and operations research communities. In previous studies, Chen and Guan measure the efficiency of China’s regional innovation systems based on the DEA model [43]. Studies applying DEA models in the field of regional development also include topics, such as regional industrial energy efficiency [44,45,46], regional economic development [47,48] and regional transport sustainability [49,50]. Although the DEA model has been widely applied in this field, extant research mainly conducts analysis using the DEA model on regions that have already been classified, limited studies have conversely, used the DEA model for classification after obtaining regional evaluation results. Besides, instead of using absolute indicators, such as GDP per capita or regional unemployment rate to evaluate the regional economic development, the DEA model in this paper is constructed to accurately measure and compare the ratio of inputs to outputs in provinces in China from three perspectives, so that different provinces can be classified more objectively according to regional characteristics.

After classification, we adopt indicators of New Venues Created (NVC) and the Volumes of Visitors to Venues (V³) from Baidu Maps to portray the impact of the COVID-19 on different regions, and the reasons are as follows. As mentioned above, studies on the impact of traffic restrictions focus on the perspective of industries, and studies in this field primarily utilize stock prices as an indicator [51,52,53]. In this paper, we attempt to measure the impact of traffic control from a regional perspective, so the above indicator is not applicable. Notably, research highlights that individual mobility was severely influenced by traffic policies [54,55] so that V³ can intuitively measure the consumption situation of different regions since there are few opportunities for physical stores to be patronized when people are quarantined. NVC in this paper refers to new placemarks added to Baidu Maps. After the merchants apply to add new placemarks with a business license through the Merchant Center (https://bgc.map.baidu.com/lbc/merchant/enterpage, accessed on 12 December 2021), placemarks of new venues even under construction can also be added to Baidu Maps with a mark of ‘under construction’. Therefore, the NVC indicator from Baidu Map is a relative time-sensitive index of investment. Besides, the level of GDP of each region in China is proved to be significantly positively correlated with both indicators and the real-time mobility data collected from Baidu Maps have been successfully leveraged to study the impact of COVID-19 in the extant studies [56,57]. Above all, we suggest V³ and NVC can effectively portray the consumption situation and the intensity of investment in different regions, respectively. Therefore, our study analyzes the impacts of traffic restriction policies on regional economies by comparing the trends of V³ and NVC in different regions.

3. Classification Based on DEA Models

Data in this study for region classification are obtained from the China City Statistical Yearbook 2019 [58]. Starting from 1985, every December, the National Bureau of Statistics in China collects and publishes statistical data yearbooks on various aspects of social and economic development and urban construction from 31 provinces and 656 cities in China for the previous year, specifically including data on population, labor force and land resources, industry, transportation, posts and telecommunications, trade, fixed asset investment, education, culture, health, social security, etc.

In the selection of input and output indicators for the DEA model, the following indicators were selected for the measurement of economic gradient, degree of internationalization, and level of traffic convenience, drawing on the indicators used in regional economic measurement in the domestic and international literature [43].

3.1. Input Indicators

To compare the economic gradients and degree of internationalization in different regions, we assign virtual inputs as 1 to both models to follow simple logic, that is, when the inputs are the same, the economic gradient and degree of internationalization are evaluated and compared according to the output indicators. The degree of traffic convenience is related to reasonable geographic layout and the number of transportation hubs [34], which is more associated with the area of the province than the population. For example, Xinjiang province has a small population intensity while covering a large area [59]. In this case, if the input is population, transportation hubs proportional to population will be difficult to cover this vast area. Therefore, the area is a more suitable variable than the population as an input indicator, due to the mismatch between population and area in China’s provinces [60]; we select the area of the province as the input indicator from the perspective of level of traffic convenience in the model.

3.2. Output Indicators

The output indicators in this study are classified into three categories, namely the economic gradient, the degree of internationalization, and the level of traffic convenience of each province.

Table 1. Variables and units for DEA models.

		Dimensions
		Economic Gradient	Degree of Internationalization	Level of Traffic Convenience
Input Indicators	$\mathrm{X}$	1	1	Area of province
Output Indicators	${\mathrm{Y}}_1$	Per capita disposable income (yuan)	Ratio of exports to GDP	Number of bus trams (104)
	${\mathrm{Y}}_2$	Per capita GDP in the secondary sectors (104 yuan)	Ratio of imports to GDP	Length of operating routes by bus trams (kilometer)
	${\mathrm{Y}}_3$	Per capita GDP in the tertiary sectors (104 yuan)	Number of foreign enterprises per 104 people	Total number of passengers transported by trams (104)
	${\mathrm{Y}}_4$	Per capita sales revenue of new products (104 yuan)	Ratio of foreign direct investment to GDP	Number of rail vehicles (104)
	${\mathrm{Y}}_5$	Per capita number of patent applications (104 yuan)	Ratio of foreign enterprise exports to GDP	Length of operating routes by rail vehicles (kilometer)
	${\mathrm{Y}}_6$	Reciprocal of unemployment rate	Ratio of foreign enterprise imports to GDP	Number of passengers transported by rail vehicles (104)
	${\mathrm{Y}}_7$		Ratio of international tourism to GDP	Number of cabs (104)
	${\mathrm{Y}}_8$			Number of passengers (104)
	${\mathrm{Y}}_9$			Passenger turnover (108 passengers per kilometer)
	${\mathrm{Y}}_{10}$			Freight volume (104)
	${\mathrm{Y}}_{11}$			Cargo turnover (108 tons per kilometer)
	${\mathrm{Y}}_{12}$			Length of railroads (104 km)
	${\mathrm{Y}}_{13}$			Length of roads (104 km)
	${\mathrm{Y}}_{14}$			Length of inland rivers (104 km)

presents the variables and units for each model.

Zhang (2021) indicates that the most common way to measure the regional economic gradient is to adopt the indicator of per capita disposable income [61]. Besides, in regional economic development theory, the main differences between high-gradient and middle-gradient regions are reflected in industry structure and regional innovation capacity [35]. To evaluate the development of industry structure, we select per capita GDP in the secondary and tertiary industries of provinces as indicators in this dimension. In terms of regional innovation capacity, Acs et al. (2002) argue that patents and revenue from new product sales are the intermediate output and the final outcome of innovation, respectively [62]; therefore, we also select indicators of per capita sales revenue of new products and per capita number of patent applications. Middle-gradient and low-gradient regions differ mainly in indicators, such as unemployment rate [35], which reflect productivity level and economic operation of regions. As a result, we also adopt reciprocal of unemployment rate in this dimension. In sum, our study establishes a systematic and complete index system by comprehensively measuring the economic development strength of provinces from the perspectives of economic development speed, productivity development level and technology level in the dimension of the economic gradient. The above six indicators conform well to the principles of science, comprehensiveness, refinement, and integration with provincial conditions. Results of DEA models on economic gradients are shown in

Table A1. DEA model on economic gradient.

DMU	X	Y1	Y2	Y3	Y4	Y5	Y6	θ
Beijing	1.0	62,361.2	2.6	11.4	19,204.4	9.6	71.4	1.0
Shanghai	1.0	64,182.6	4.0	9.4	40,415.5	12.1	28.3	1.0
Tianjin	1.0	39,506.1	4.9	7.1	24,715.8	9.6	28.5	1.0
Jiangsu	1.0	38,095.8	5.1	5.9	35,306.2	20.5	33.7	1.0
Zhejiang	1.0	45,839.8	4.1	5.4	40,627.8	17.5	38.5	1.0
Guangdong	1.0	35,809.9	3.6	4.6	34,704.8	21.3	41.5	1.0
Fujian	1.0	32,643.9	4.4	4.1	13,450.6	8.0	27.0	0.9
Shandong	1.0	29,204.6	3.3	3.8	15,175.2	6.1	29.9	0.7
Shaanxi	1.0	22,528.3	3.1	2.7	5262.3	2.6	31.2	0.7
Hubei	1.0	25,814.5	2.9	3.2	14,978.8	4.7	39.2	0.7
Inner Mongolia	1.0	28,375.7	2.7	3.4	4057.9	1.5	27.9	0.6
Chongqing	1.0	26,385.8	2.7	3.4	13,592.2	5.8	33.8	0.7
Liaoning	1.0	29,701.4	2.3	3.0	10,453.7	2.9	25.4	0.6
Anhui	1.0	23,983.6	2.2	2.1	15,073.3	8.9	35.3	0.6
Ningxia	1.0	22,400.4	2.4	2.6	7015.2	3.2	25.7	0.6
Jilin	1.0	22,798.4	2.4	2.8	4983.3	1.2	28.9	0.6
Jiangxi	1.0	24,079.7	2.2	2.1	9706.9	5.7	29.1	0.6
Henan	1.0	21,963.5	2.3	2.3	8004.4	2.9	33.1	0.6
Hunan	1.0	25,240.7	2.1	2.7	11,039.6	3.8	27.9	0.5
Hebei	1.0	23,445.7	2.1	2.2	6920.2	2.2	30.3	0.6
Qinghai	1.0	20,757.3	2.1	2.2	2044.2	1.4	33.7	0.6
Xinjiang	1.0	21,500.2	2.0	2.2	1740.5	1.4	42.4	0.7
Shanxi	1.0	21,990.1	1.9	2.4	5221.4	1.5	30.7	0.5
Sichuan	1.0	22,460.6	1.8	2.5	4287.7	3.2	28.8	0.5
Hainan	1.0	24,579.0	1.2	2.9	1127.5	0.6	43.5	0.6
Tibet	1.0	17,286.1	1.8	2.1	526.4	0.1	35.3	0.6
Guangxi	1.0	21,485.0	1.6	1.9	3722.3	1.3	42.7	0.6
Heilongjiang	1.0	22,725.8	1.1	2.5	1487.9	0.7	25.1	0.4
Guizhou	1.0	18,430.2	1.6	1.9	2075.0	1.7	31.6	0.5
Yunnan	1.0	20,084.2	1.4	1.7	1923.0	1.3	29.4	0.5
Gansu	1.0	17,488.4	1.1	1.7	1043.4	1.3	36.0	0.5

in Appendix A.

In the aspect of degree of internationalization, we select trade indicators of the ratio of exports/imports to GDP and financial indicators of the ratio of foreign direct investment to GDP from the Maastricht Globalization Index [63], as well as an indicator of the ratio of international tourism to GDP from KOF Globalization Index [64]. The ratio of exports to GDP reflects the degree of dependence of a country’s economy on foreign trade and the ratio of imports to GDP reflects the degree of openness of a country’s market. The ratio of foreign direct investment measures the involvement of multinational corporations in an economy, which is regarded as an important indicator of the ability to attract foreign investment, and the ratio of international tourism indicates the degree of openness of a country’s society. Besides, to further examine the dependence on foreign trade of regions, we also select the indicators of the ratio of foreign enterprise exports/imports to GDP. Meanwhile, the more international a region is, the more foreign companies it attracts, as a result, we select the indicator of the number of foreign enterprises per 10,000 people. Combined with the city internationalization index system proposed by the Second United Nations Conference on Human Settlements [65], the above seven indicators can adequately measure the degree of internationalization of each province. Results of the DEA model on the degree of internationalization are shown in

Table A2. DEA model on degree of internationalization.

DMU	X	Y1	Y2	Y3	Y4	Y5	Y6	Y7	θ
Guangdong	1.0	0.5	0.3	0.002	0.5	0.2	0.1	0.2	1.0
Hainan	1.0	0.1	0.2	0.0	1.0	0.0	0.1	0.2	1.0
Shanghai	1.0	0.4	0.6	0.0	1.2	0.3	0.4	0.2	1.0
Beijing	1.0	0.1	0.2	0.0	0.7	0.0	0.1	0.2	0.6
Jiangsu	1.0	0.3	0.2	0.0	0.4	0.2	0.1	0.1	0.6
Hunan	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.5
Zhejiang	1.0	0.4	0.1	0.0	0.3	0.1	0.0	0.0	0.4
Henan	1.0	0.1	0.0	0.0	0.1	0.0	0.0	0.0	0.4
Tianjin	1.0	0.2	0.3	0.0	0.7	0.1	0.1	0.1	0.3
Fujian	1.0	0.2	0.1	0.0	0.3	0.1	0.1	0.1	0.3
Anhui	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.1	0.2
Shandong	1.0	0.1	0.2	0.0	0.2	0.0	0.0	0.0	0.1
Guizhou	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.1
Liaoning	1.0	0.2	0.2	0.0	0.6	0.1	0.1	0.1	0.1
Hubei	1.0	0.1	0.0	0.0	0.1	0.0	0.0	0.1	0.1
Chongqing	1.0	0.1	0.1	0.0	0.2	0.1	0.0	0.1	0.1
Yunnan	1.0	0.0	0.1	0.0	0.1	0.0	0.0	0.2	0.1
Guangxi	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.1	0.1
Jiangxi	1.0	0.1	0.0	0.0	0.2	0.0	0.0	0.0	0.1
Shaanxi	1.0	0.1	0.1	0.0	0.2	0.1	0.0	0.1	0.1
Sichuan	1.0	0.1	0.1	0.0	0.1	0.1	0.0	0.0	0.1
Hebei	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.0	0.1
Gansu	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.1
Jilin	1.0	0.0	0.1	0.0	0.1	0.0	0.0	0.0	0.0
Inner Mongolia	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.1	0.0
Heilongjiang	1.0	0.0	0.1	0.0	0.1	0.0	0.0	0.0	0.0
Xinjiang	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.1	0.0
Ningxia	1.0	0.0	0.0	0.0	0.2	0.0	0.0	0.0	0.0
Tibet	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.2	0.0
Shanxi	1.0	0.1	0.0	0.0	0.1	0.0	0.0	0.0	0.0
Qinghai	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0

in Appendix A.

In previous studies, Guo (2010) used the indicators of road freight volume, waterway freight volume, number of road passengers, and number of waterway passengers to measure the regional transportation convenience in Zhejiang Province [66]; Liu and Zhao (2005) evaluated the traffic convenience in different provinces and cities by calculating the transportation network density using data on road length, railroad length, and actual freight volume [67]. In this study, representative indicators from previous studies were selected and supplemented. The public transportation system in China mainly consists of bus trams, rail vehicles and cabs [68]. Therefore, we select indicators of the number of bus trams, the length of operating routes by bus trams and the total number of passengers transported by bus trams to characterize the situation of operation of public tram transportation; the number of rail vehicles, the length of operating routes by rail vehicles and total number of passengers transported by rail vehicles to characterize the situation of rail operations and the number of cabs to characterize the situation of cab operations. Additionally, we follow Jiang et al. (2017) to select the number of passengers, passenger turnover, freight volume, and cargo turnover to measure the efficiency of transportation in the regions [69]. Besides, we select indicators of the length of railroads, roads, and inland rivers which are considered to have important impacts on the level of traffic convenience [70]. Results of the DEA model on the level of traffic convenience are shown in

Table A3. DEA model on level of traffic convenience.

DMU	X	Y1	Y2	Y3	Y4	Y5	Y6	Y7	θ
Guangdong	1.0	0.5	0.3	0.002	0.5	0.2	0.1	0.2	1.0
Hainan	1.0	0.1	0.2	0.0	1.0	0.0	0.1	0.2	1.0
Shanghai	1.0	0.4	0.6	0.0	1.2	0.3	0.4	0.2	1.0
Beijing	1.0	0.1	0.2	0.0	0.7	0.0	0.1	0.2	0.6
Jiangsu	1.0	0.3	0.2	0.0	0.4	0.2	0.1	0.1	0.6
Hunan	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.5
Zhejiang	1.0	0.4	0.1	0.0	0.3	0.1	0.0	0.0	0.4
Henan	1.0	0.1	0.0	0.0	0.1	0.0	0.0	0.0	0.4
Tianjin	1.0	0.2	0.3	0.0	0.7	0.1	0.1	0.1	0.3
Fujian	1.0	0.2	0.1	0.0	0.3	0.1	0.1	0.1	0.3
Anhui	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.1	0.2
Shandong	1.0	0.1	0.2	0.0	0.2	0.0	0.0	0.0	0.1
Guizhou	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.1
Liaoning	1.0	0.2	0.2	0.0	0.6	0.1	0.1	0.1	0.1
Hubei	1.0	0.1	0.0	0.0	0.1	0.0	0.0	0.1	0.1
Chongqing	1.0	0.1	0.1	0.0	0.2	0.1	0.0	0.1	0.1
Yunnan	1.0	0.0	0.1	0.0	0.1	0.0	0.0	0.2	0.1
Guangxi	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.1	0.1
Jiangxi	1.0	0.1	0.0	0.0	0.2	0.0	0.0	0.0	0.1
Shaanxi	1.0	0.1	0.1	0.0	0.2	0.1	0.0	0.1	0.1
Sichuan	1.0	0.1	0.1	0.0	0.1	0.1	0.0	0.0	0.1
Hebei	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.0	0.1
Gansu	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.1
Jilin	1.0	0.0	0.1	0.0	0.1	0.0	0.0	0.0	0.0
Inner Mongolia	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.1	0.0
Heilongjiang	1.0	0.0	0.1	0.0	0.1	0.0	0.0	0.0	0.0
Xinjiang	1.0	0.1	0.1	0.0	0.1	0.0	0.0	0.1	0.0
Ningxia	1.0	0.0	0.0	0.0	0.2	0.0	0.0	0.0	0.0
Tibet	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.2	0.0
Shanxi	1.0	0.1	0.0	0.0	0.1	0.0	0.0	0.0	0.0
Qinghai	1.0	0.0	0.0	0.0	0.1	0.0	0.0	0.0	0.0
DMU	X	Y8	Y9	Y10	Y11	Y12	Y13	Y14	θ
Guangdong	1.0	17,450	276.5	2240.5	52,221	1153	88	16,257	1.0
Hainan	1.0	15,845	218.7	28,299.9	106,983	466	2091	13,106	1.0
Shanghai	1.0	58,935	254.4	1034.2	20,873	1264	0	22,256	1.0
Beijing	1.0	60,587	493.1	3597.9	128,491	2326	4352	157,483	0.9
Jiangsu	1.0	67,443	1289.6	10,052.2	354,019	6336	1117	275,642	0.8
Hunan	1.0	110,421	1775.1	8982.1	259,884	5410	1403	268,589	0.8
Zhejiang	1.0	63,347	1163.7	11,803.7	406,761	4324	5641	208,826	0.7
Henan	1.0	120,612	1539.3	8969.3	233,157	3062	24,380	158,729	0.7
Tianjin	1.0	98,350	1258.9	6675.5	204,307	4341	8470	275,039	0.7
Fujian	1.0	142,144	2085.6	28,338.3	416,389	4524	12,112	217,699	0.6
Anhui	1.0	98,380	1103.7	11538.1	269,083	2813	9761	120,662	0.6
Shandong	1.0	106,680	1463.1	4386.6	229,957	5070	11,496	240,060	0.5
Guizhou	1.0	93,025	798.7	1797.9	102,537	3565	3740	196,908	0.5
Liaoning	1.0	47,346	1289.2	13,873.0	249,323	7362	0	193,252	0.5
Hubei	1.0	71,343	938.8	10,654.5	223,346	6525	413	122,974	0.5
Chongqing	1.0	14,383	130.5	875.8	22,040	1033	343	35,023	0.5
Yunnan	1.0	60,686	993.7	4528.6	174,285	4278	5638	161,941	0.5
Guangxi	1.0	23,837	393.9	4489.4	211,497	5441	467	143,326	0.5
Jiangxi	1.0	48,105	600.0	7646.2	136,947	3514	3245	108,901	0.4
Shaanxi	1.0	71,583	798.0	4024.9	173,245	5002	1146	177,128	0.4
Sichuan	1.0	98,569	878.3	2946.1	187,385	4950	10,818	331,592	0.3
Hebei	1.0	31,956	427.3	1704.7	52,156	5043	1456	105,399	0.3
Gansu	1.0	41,484	431.6	1971.9	140,670	3848	4024	252,929	0.3
Jilin	1.0	47,931	816.6	4983.8	190,652	5202	5707	125,449	0.3
Inner Mongolia	1.0	6137	88.3	627.7	38,916	1373	130	35,405	0.3
Heilongjiang	1.0	31,568	433.8	1601.3	55,190	6894	5098	167,116	0.2
Xinjiang	1.0	42,185	634.7	2609.9	70,386	4672	911	143,228	0.2
Ningxia	1.0	13,268	337.1	5596.0	232,525	12,766	2403	202,641	0.1
Tibet	1.0	6443	141.0	551.4	18,905	2349	674	82,137	0.1
Shanxi	1.0	21,204	405.8	2483.9	97,498	5959	0	189,050	0.1
Qinghai	1.0	1399	46.8	150.1	2433	785	0	97,785	0.0

in Appendix A.

By reasonably ranking the relative efficiency of each province on the three dimensions, 31 provinces are classified into three categories on each dimension. The classification result is shown in

Table 2. Classification results based on DEA models.

	Dimensions
	Economic Gradient		Degree of Internationalization		Level of Traffic Convenience
	DMU	θ	DMU	θ	DMU	θ
High	Beijing	1.0	Guangdong	1.0	Tianjin	1.0
	Shanghai	1.0	Hainan	1.0	Shanghai	1.0
	Tianjin	1.0	Shanghai	1.0	Beijing	1.0
	Jiangsu	1.0	Beijing	0.6	Chongqing	0.9
	Zhejiang	1.0	Jiangsu	0.6	Shandong	0.8
	Guangdong	1.0	Hunan	0.5	Henan	0.8
	Fujian	0.9	Zhejiang	0.4	Anhui	0.7
	Shandong	0.7	Henan	0.4	Jiangsu	0.7
	Shaanxi	0.7	Tianjin	0.3	Hubei	0.7
	Hubei	0.7	Fujian	0.3	Guangdong	0.6
Intermediate	Inner Mongolia	0.7	Anhui	0.2	Zhejiang	0.6
	Chongqing	0.7	Shandong	0.1	Hunan	0.5
	Liaoning	0.6	Guizhou	0.1	Guizhou	0.5
	Anhui	0.6	Liaoning	0.1	Hebei	0.5
	Ningxia	0.6	Hubei	0.1	Liaoning	0.5
	Jilin	0.6	Chongqing	0.1	Hainan	0.5
	Jiangxi	0.6	Yunnan	0.1	Jiangxi	0.5
	Henan	0.6	Guangxi	0.1	Shanxi	0.5
	Hunan	0.6	Jiangxi	0.1	Fujian	0.4
	Hebei	0.6	Shaanxi	0.1	Shaanxi	0.4
	Qinghai	0.6	Sichuan	0.1	Sichuan	0.3
	Xinjiang	0.6	Hebei	0.1	Jilin	0.3
	Shanxi	0.6	Gansu	0.1	Yunnan	0.3
	Sichuan	0.6	Jilin	0.0	Guangxi	0.3
Low	Hainan	0.5	Inner Mongolia	0.0	Ningxia	0.3
	Tibet	0.5	Heilongjiang	0.0	Heilongjiang	0.2
	Guangxi	0.5	Xinjiang	0.0	Gansu	0.2
	Heilongjiang	0.5	Ningxia	0.0	Inner Mongolia	0.1
	Guizhou	0.5	Tibet	0.0	Qinghai	0.1
	Yunnan	0.5	Shanxi	0.0	Xinjiang	0.1
	Gansu	0.4	Qinghai	0.0	Tibet	0.0

.

4. Analysis of NVC and V³ Based on Classification Results

From the third week of 2020 (January 23), the outbreak led to the massive traffic restrictions in Wuhan and gradually impacted the national economy. On 8 April 2020, the outbreak was largely under control, and Wuhan was unblocked. In this part, we adopt indicators of New Venues Created (NVC) and the Volumes of Visitors to Venues (V³) to conduct a regional analysis of the impact of traffic restriction policies during the COVID-19 pandemic.

First of all, we draw the trend charts of NVC and V³ from the first week to the 22nd week in 2020, in regions with high, intermediate and low economic gradient, regions with high, intermediate and low degrees of internationalization and regions with high, intermediate and low levels of traffic convenience, respectively. For better comparative analysis, we also plotted the trend for the corresponding periods of 2018 and 2019. The trends are shown in Figure 1, Figure 2 and Figure 3.

There are two findings in the above charts. Firstly, both V³ and NVC in the year 2020 bottomed out in week 7 and then started to recover in all nine regions. Therefore, the observation week of the subsequent analysis is set from week 7 to week 22. Secondly, the shape of V³ curves of all regions are similar in 2018, 2019, and 2020, respectively, that is, V³ approximately increases or decreases in the same week and same year of all regions. However, NVC is volatile with more peaks and troughs and shows a more complex change from year to year, which seems to suggest that investment activities are more irregular than consumption activities in a short period. Therefore, we conduct a regional analysis of V³ based on weekly comparisons, while the analysis of NVC is conducted by indicators of the maximum, minimum, and average value of the entire observation week from an overall perspective instead.

In addition, it is notable that due to the differences in the development abilities of different regions, the subsequent analysis is based on a ratio of the actual value to the supposed value of different regions instead of depending on the actual value in 2020 solely. The actual value refers to the actual V³/NVC in 2020, while the supposed value refers to the supposed V³/NVC in 2020 calculated according to the actual V³/NVC in 2019 and the growth rate of V³/NVC in 2019 relative to 2018. This ratio indicates the approximation between the actual level of consumption/investment, and the supposed level of consumption/investment which would have been achieved based on the development abilities of each region, assuming there are no COVID-19 and traffic restrictions.

4.1. Analysis of V³

The Wilcoxon signed-rank test is used to compare weekly V³ trends in different regions. The Wilcoxon Signed-Rank test is a nonparametric test procedure used for the analysis of matched-pair data or for the one-sample problem [71]. Compared to the paired t-test, it is applicable to the case where the difference of paired samples does not follow a normal distribution. It has been widely used in the field of economics [72,73,74]. For example, Watson et al. (2004) utilize the Wilcoxon Signed-Rank Test to determine whether there is a significant difference between the financial performances of the EMS adopter versus the non-EMS adopter [75]. Next, we test the ratio of the actual V³ of each week to the supposed V³ in pairs for regions with high and low economic gradient, high and low degree of internationalization, and high and low levels of traffic convenience in order to explore whether there is a significant difference in consumption situations between regions. The supposed V³ of each week in 2020 is calculated according to the V³ of the same week in 2019 and the growth rate of V³ of the same week in 2019 relative to 2018 as follows:

$${\mathrm{V}}_{\mathrm{s},\mathrm{i}}^3={\mathrm{V}}_{2019,\mathrm{i}}^3\times \left[1+\frac{{\mathrm{V}}_{2019,\mathrm{i}}^3-{\mathrm{V}}_{2018,\mathrm{i}}^3}{{\mathrm{V}}_{2018,\mathrm{i}}^3}\right]\left(\mathrm{i}=7, 8, 9,\dots , 22\right)$$

(1)

where ${\mathrm{V}}_{2018,\mathrm{i}}^3$ and ${\mathrm{V}}_{2019,\mathrm{i}}^3$ refers to the V³ of week $\mathrm{i}$ in 2018 and 2019, respectively. In sum, ${\mathrm{V}}_{\mathrm{s},\mathrm{i}}^3$ implies the supposed value of the V³ of week $\mathrm{i}$ in 2020, assuming the same growth rate as the week of the previous year with no COVID-19 and traffic restrictions.

The paired ratios of V³ for each week in 2020 to the supposed V³ for 2020 in regions with high and low economic gradients, regions with high and low degrees of internationalization, and regions with high and low levels of traffic convenience are shown in

Table 3. The paired ratios of V³ in 2020 to the supposed V³ of regions.

Week	Economic Gradient		Degree of Internationalization		Level of Traffic Convenience
	High	Low	High	Low	High	Low
7	0.174	0.181	0.179	0.174	0.185	0.183
8	0.306	0.378	0.321	0.262	0.273	0.323
9	0.341	0.402	0.365	0.300	0.317	0.386
10	0.415	0.477	0.441	0.356	0.358	0.455
11	0.441	0.545	0.463	0.447	0.397	0.537
12	0.529	0.592	0.546	0.566	0.488	0.613
13	0.625	0.735	0.647	0.712	0.586	0.735
14	0.421	0.504	0.444	0.495	0.400	0.510
15	0.953	1.027	1.003	0.926	0.899	1.021
16	1.076	1.174	1.140	1.018	0.960	1.167
17	1.309	1.395	1.359	1.238	1.264	1.353
18	0.745	0.758	0.804	0.594	0.675	0.739
19	1.172	1.308	1.229	1.027	1.118	1.234
20	1.148	1.147	1.214	0.983	1.095	1.109
21	1.128	1.182	1.172	0.899	1.094	1.046
22	1.626	1.647	1.682	1.351	1.574	1.598

.

In pairs, the ratios of different regions in the three dimensions are examined by the Wilcoxon Signed-Rank test and the results are shown in

Table 4. Wilcoxon Signed-Ranks Test.

	Economic Gradient	Degree of Internationalization	Level of Traffic Convenience
Z	−3.464 1	−2.792 2	−3.258 1
Asymp. Sig. (2-tailed)	0.001	0.005	0.001

¹ Based on negative ranks. ² Based on positive ranks.

. In all three dimensions, differences are calculated by the ratio of high-level regions minus the ratio of low-level regions in the observation week, and none of them follow a normal distribution, which satisfies the premise of the test.

Results show that the differences between high-level regions and low-level regions are significant in all three dimensions. Specifically, ratios in regions with high economic gradients are significantly lower than regions with low economic gradients; ratios in regions with high degrees of internationalization are significantly higher than regions with low degrees of internationalization; ratios in regions with high levels of traffic convenience are significantly lower than regions with low levels of traffic convenience. These results indicate that the recovery of consumption in regions with low economic gradient, regions with high degrees of internationalization and regions with low levels of traffic convenience are better than the other side in each dimension.

4.2. Analysis of NVC

In this section, the analysis of NVC is based on the ratios of the actual maximum, the actual minimum, and the actual average values of the NVC during the whole observation week in 2020 to the supposed maximum, the supposed minimum, and the supposed average values of NVC during the whole observation week in 2020 in pairs for regions with a high and low economic gradient, a high and low degree of internationalization, and a high and low level of traffic convenience. The supposed maximum, the supposed minimum, and the supposed average values of NVC are calculated similarly to the supposed V³ in Section 4.1. For example, the supposed maximum NVC of week 7 to week 22 in 2020 is calculated as follows.

$$NV{C}_{s, max}=NV{C}_{2019,max}\times \left[1+\frac{NV{C}_{2019,max}-NV{C}_{2018,max}}{NV{C}_{2019,max}}\right]$$

(2)

where $NV{C}_{2018,max}$ and $NV{C}_{2019,max}$ refer to the maximum of NVC for week 7 to week 22 in 2018 and 2019, respectively. The calculation results are shown in

Table 5. The paired ratios of NVC of regions.

	Economic Gradient		Degree of Internationalization		Level of Traffic Convenience
	High	Low	High	Low	High	Low
Max	0.238 1	0.436 1	0.258	0.394	0.268	0.387
Min	0.018	0.051	0.019	0.042	0.024	0.055
Mean	0.159	0.179	0.165	0.207	0.191	0.209

¹ For example, 0.238 is the ratio of actual maximum of NVC to supposed maximum of NVC during week 7 to week 22 in the high economic gradient region while 0.436 is the ratio of actual maximum of NVC to supposed maximum of NVC during week 7 to week 22 in the low economic gradient region.

.

Results show that the ratio of maximum, the ratio of minimum, and the ratio of average values of regions with a low economic gradient, a low degree of internationalization, and a low level of traffic convenience are higher. These results indicate that the recovery of total investment in regions with a low economic gradient, a low degree of internationalization and a low level of traffic convenience is better than the other side in each dimension.

Above all, analysis results of V³ and NVC in different regions are summarized in

Table 6. Analysis results of V³ and NVC in different regions.

	Economic Gradient	Degree of Internationalization	Level of Traffic Convenience
Consumption (V3)	high < low 1	high > low	high < low
Investment (NVC)	high < low 2	high < low	high < low

¹ The consumption level of a region with low economic gradient is better than that of a region with high economic gradient. ² The investment level of a region with low economic gradient is better than that of a region with high economic gradient.

.

4.3. Cases Analysis

In this paper, the time scale is always aligned to the Wuhan lockdown. However, the intensity and influence of traffic restrictions on provinces may decrease as the distance from Wuhan increases. Therefore, in this section, we perform the following supplementary analysis on the paired provinces near Hubei and the paired provinces far away from Hubei, respectively.

As shown in Figure 4, Chongqing, Hunan and Jiangxi are selected as the provinces adjacent to Hubei, while Shanxi, Shandong, and Yunnan are selected as the provinces that are relatively far away from Hubei.

Specifically, the paired provinces of Hunan-Chongqing and Shanxi-Yunnan are the test groups of the indicator V³ due to the fact that Hunan enjoys a higher degree of internationalization than Chongqing while its economic gradient and level of traffic convenience are lower than Chongqing, and Shanxi enjoys a higher degree of internationalization than Yunnan while its economic gradient and level of traffic convenience are lower than Yunnan. According to the results in Table 6, there are supposed to be significant differences in the V³ of these paired provinces, and V³ in Hunan is supposed to be higher than Chongqing, while V³ in Shanxi is supposed to be higher than Yunnan. The paired ratios of paired provinces according to the calculation Formula (1) in Section 4.1 are shown in

Table 7. The paired ratios of V³ in paired provinces.

Week	Provinces Adjacent to Hubei		Provinces Far from Hubei
	Hunan	Chongqing	Shanxi	Yunnan
7	0.223	0.209	0.151	0.139
8	0.447	0.410	0.283	0.324
9	0.508	0.490	0.352	0.382
10	0.474	0.461	0.373	0.480
11	0.530	0.500	0.436	0.527
12	0.644	0.640	0.549	0.601
13	0.730	0.734	0.701	0.741
14	0.499	0.529	0.421	0.547
15	1.256	1.193	0.940	1.001
16	1.343	1.274	1.032	1.029
17	1.652	1.768	1.445	1.377
18	0.837	0.777	0.516	0.791
19	1.536	1.390	1.072	1.337
20	1.506	1.298	1.064	1.133
21	1.334	1.266	1.051	1.292
22	1.837	1.899	1.553	1.696

and the results of the Wilcoxon Signed-Rank Test are shown in

Table 8. Wilcoxon Signed-Ranks Test of paired provinces.

	Provinces Adjacent to Hubei (Hunan–Chongqing)	Provinces Far from Hubei (Shanxi–Yunnan)
Z	−3.464 1	−2.947 1
Asymp. Sig. (2-tailed)	0.001	0.003

¹ Based on positive ranks.

.

Results of the Wilcoxon Signed-Rank Test indicate that the significant differences of V³ between regions with different characteristics in economic gradient, degree of internationalization and level of traffic convenience exist both in paired provinces adjacent to Hubei and paired provinces far from Hubei. In addition, provinces with a lower economic gradient, a higher degree of internationalization and a lower level of traffic convenience show a higher V³ level in paired provinces.

Similarly, the paired provinces of Chongqing-Jiangxi and Shandong-Shanxi are the test groups of the indicator NVC due to the fact that the scores of Chongqing are higher than Jiangxi in all three dimensions while the scores of Shandong are higher than Shanxi in all three dimensions in the DEA models. According to the results in Table 6, there are supposed to be significant differences in the NVC of these paired provinces, and the ratios of descriptive statistic indicators of NVC in Jiangxi are supposed to be higher than in Chongqing while the ratios of descriptive statistic indicators of NVC in Shanxi are supposed to be higher than in Shandong. The paired ratios of NVC of paired provinces according to the calculation Formula (2) in Section 4.2 are shown in

Table 9. The paired ratios of NVC of paired provinces.

	Provinces Adjacent to Hubei		Provinces Far from Hubei
	Chongqing	Jiangxi	Shandong	Shanxi
Max	0.262	0.568	0.241	0.259
Min	0.027	0.049	0.025	0.041
Mean	0.172	0.291	0.165	0.246

.

Results indicate that the differences in NVC between regions with different characteristics in economic gradient, degree of internationalization and level of traffic convenience exist both in paired provinces adjacent to Hubei and paired provinces far from Hubei. In addition, provinces with a lower economic gradient, a lower degree of internationalization and a lower level of traffic convenience show a higher NVC level in paired provinces.

Therefore, we suggest that the conclusion of Section 4.1 and Section 4.2 still stands if we shift the time scales for each region to the declaration of ‘quarantine’ of the provincial capital, since each province is its own nearest province.

5. Discussion

5.1. Economic Gradient

The economic gradient development theory classifies regions into high, intermediate and low gradients, with different levels and characteristics of economic development. The leading industries in high-gradient regions are primarily thriving industries in the innovation stage, while the leading industries in low-gradient regions are primarily industries in the late maturity stage or in decline. The densely populated and economically developed provinces generally belong to the high-gradient regions, which are characterized by a long history of development, superior geographical location, high cultural quality of workers, strong technical force and a solid industrial and agricultural base, and play a leading role in the overall economic development. Provinces in China can usually be classified into three major gradient regions, namely, the eastern coastal region, the central region and the western region. In this study, a more refined evaluation of the regional economic gradient is made by taking provinces as units and calculating regional economic efficiency in the DEA model based on relevant input-output indicators. By comparing with indicators of the same period in previous years, we find that regions in different economic gradients suffer significantly different impacts from massive traffic restrictions. The leading industries in regions with higher economic gradients are high-tech industries, which have a greater demand for high-quality talents. At the same time, they have a greater dependence on the scientific and technological innovation departments, such as universities and research institutions. The transportation policies during COVID-19 prevented the majority of students from returning to their campuses, and prevented scientific researchers from gathering, which hindered the work process of inventions and new technology development, and weakened the collaboration between universities, research institutions and enterprises, thus reducing the efficiency of transforming scientific research results into productivity and economic benefits.

5.2. Degree of Internationalization

Compared with the data of the same period in previous years, it is found that regions with different degrees of internationalization are affected differently. Among them, regions with a high degree of internationalization show a decline primarily in investment, while regions with a low degree of internationalization show a decline primarily in consumption. Import and export trades drive the economy, and regions with a high degree of internationalization (such as the southeastern coastal region) are highly dependent on foreign trades. During the period of massive traffic restrictions, most of the factories and enterprises engaged in foreign trade were in a state of shutdown, as a result, the net export value declined significantly. Although the epidemic was largely under control in a short period in China, it had outbreaks in countries around the world. At that time, factories in China had to face the situation of ‘no one asking for goods’, a large number of trade orders were canceled and even signed orders could not be completed normally, which also forced factories engaged in foreign trade to stop working on a large scale. The total amount of international trade has declined significantly and the investment behavior has largely been reduced. On the other hand, poor export sales force goods to be sold locally, which may mitigate the impact of supply shortages caused by the epidemic on rising product prices. Since consumption is primarily influenced by income and prices, in this case, the consumption level of a region with a higher degree of internationalization may be less affected than that of a region with a lower degree of internationalization.

5.3. Level of Traffic Convenience

Convenient transportation conditions can effectively improve the location advantages of the regions, increase investment and consumption opportunities, drive the development of advantageous resources in the regions and form advantageous industries. A well-developed transportation system brings about lower costs for the freight of regional products and higher accessibility to product markets all over the country. Besides, convenient transportation makes the division of labor more specialized and the cooperation more concentrated in the regional economic system, and can also enhance the openness of the regional economic system. A comparison with data from the same period in previous years shows that regions with different levels of traffic convenience are hit significantly differently in both aspects of investment and consumption. The economy of regions with higher levels of traffic convenience is severely shocked by massive traffic restrictions due to the higher demand for transportation.

6. Policy Implications

In sum, we first establish a complete and systematic set of indicators on the three complementary dimensions that are, economic gradient, degree of internationalization and level of traffic convenience, to portray the characteristics of each province in China. After applying DEA models, we classify 31 provinces into different categories of each dimension, respectively. Next, we match the classification results to data that include human mobility information for each province within China from Baidu Maps rather than sampling data. For longitudinal comparison, this paper considers the status quo of past years for a more objective analysis. Finally, we analyze and compare the impact of the massive traffic restriction policies in different regions.

According to the Sustainable Development Report 2021 [18], there has been a reversal in global poverty reduction, with the global extreme poverty rate rising in 2020 for the first time in more than 20 years. About 124 million people worldwide returned to extreme poverty in 2020. If current trends continue, the global poverty rate is projected to be 7% by 2030, failing to achieve the goal of ending poverty by 2030. In terms of eradicating hunger, the number of hungry people is likely to increase by 83 million to 132 million in 2020 due to the disruption of the food supply chain. The report also shows that basic medical services have been hit by COVID-19 and health services have suffered setbacks. Besides, the pandemic has exacerbated existing inequalities both within and between countries, with the Gini coefficient, which measures the fairness of income distribution, rising in some developing countries.

After the outbreak of the epidemic, governments aimed to create jobs and promote economic recovery, but they are faced with two choices: either a high pollution, high carbon, inefficient and unsustainable development mode for short-term economic benefits, or push industrial transformation, develop low carbon economy and sustainable transportation system and rationalize energy prices, to create long-term economic and social benefits. The latter is the wise way to go.

The epidemic has created opportunities for sustainable development. For example, new infrastructure plays an even more important role in supporting the realization of the SDGS as traffic restrictions have become regular methods for epidemic control. Manufacturing, services, artificial intelligence and the internet are more closely linked. People are telecommuting, hosting meetings both online and offline, hundreds of millions of kids are taking classes online, and hospitals are taking telemedicine. These lifestyle changes have a significant and profound impact on energy consumption and the mitigation of climate change. In sum, governments should make full use of the post-COVID-19 recovery to advance the 2030 Agenda for Sustainable Development and fulfill their commitments. Guided by the SDGS, countries should promote fundamental changes globally, including strengthening social protection systems, increasing investment in science and technology innovation, developing a green economy and promoting sustainable food systems.

Combining the concept of sustainable development with our analysis above, we put forward several policy implications.

(1)

Government needs to weigh up the relationship between traffic restriction policies and economic development, and implement targeted policies.

Massive traffic restriction policies are effective to achieve the macro-control purposes of government, and the trade-off is economic decline. Therefore, the government should pay extra attention when it adopts a traffic restriction policy. The appropriate timing and extent of restriction policies should be implemented specifically according to different regions. For example, we argue that although regions with a higher degree of internationalization suffer more severe in terms of investment, their decline in consumption is relatively moderate, which indicates that the government’s policy of stimulating consumption should be appropriately tilted to the regions with a low degree of internationalization. For depressed investment in the more internationalized regions, the government should impose preferential policies on taxation, loans, and plant land rent on SMEs to reduce their financial pressure and set up multi-level enterprise relief funds to compensate for credit losses and contract default losses

(2)

For regions with a higher economic gradient, the ‘new infrastructure’ can be used to promote their economic radiation to the surrounding areas and mitigate the impact of the epidemic on their economic development.

‘New infrastructure’ refers to the construction of 5G networks, big data centers, artificial intelligence and other new technology infrastructure. As regions with a high economic gradient are more severely affected by investment and consumption, large-scale investment can be driven from the demand side, and also from the supply side to empower production, consumption, social governance and other aspects through the ‘new infrastructure’. According to a report by the China Academy of Information and Communication Technology, the total economic output indirectly driven by the commercialization of 5G in China is expected to be about 24.8 trillion yuan from 2020 to 2025. The ‘new infrastructure’ connects regions through the construction of infrastructure, deepening the sharing and cooperation of information and data, and promoting economic, cultural, and financial and trade interactions between regions in a more environmentally-friendly way. At the same time, the online office platform and cloud office model emerged in the epidemic to provide a reliable way for the digital development of the regional sustainable economy.

(3)

For those regions that have suffered a sharp drop in consumption level, such as regions with a high level of traffic convenience, it is necessary to strengthen the role of government macro-regulation.

Allowances with time limits, shopping mall cards, and supermarket coupons can be issued to boost domestic demand and stimulate consumption in a short period of time. In addition, enterprises should be encouraged to develop new sustainable marketing models. For example, CCTV news host Guangquan Zhu and livestream star Jiaqi Li sold 40.14 million yuan worth of Hubei products through live streaming in April 2020 [76]. Additionally, platforms, such as Freshhema, Gome, Suning, and Carrefour, and offline stores around the country have opened up a “For Hubei” sector, specifically selling agricultural and sideline products from Hubei [77]. Since then, manufacturers in Hubei Province have also seized the opportunity to open online live, with great success. Thus, accelerating the development and application of new e-commerce models, such as cloud-based shopping and live streaming can promote regional economic recovery and provide new possibilities for future industrial structure change.

(4)

Promoting the balanced development of different regions.

Due to the superiority of geographical location, a large amount of capital and labor are concentrated in a few cities with high transportation convenience, and the dense population leads to a higher risk of epidemic transmission. To form a regional economic layout with complementary advantages and sustainable development, it is necessary to promote the balanced development of different regions. The epidemic provides an opportunity for China to push the regional economy to break the traditional pattern, to further improve the regional industrial structure to become more intelligent, balanced and sustainable.