Dataset Development on Commuting Efficiency by Travel Mode in Shanghai (2015)

Yue, L. Y.¹  Zhu, Y.^{1*  Li}, K. M.^2,3

1. Asian Demographic Research Institute, Shanghai University, Shanghai 200444, China;

2. Department of Architecture, Shanghai Academy of Fine Arts, Shanghai University, Shanghai 200444, China;

3. Technology Innovation Center for Land Spatial Eco-restoration in Metropolitan Area, Ministry of Natural Resources of China, Shanghai 200003, China

Abstract: This study conducts a comparative analysis of commuting efficiency across different travel modes within the excess commuting framework. It contributes to a deeper understanding of interactions between urban commuting patterns, land use, and transportation systems. This kind of study holds significant practical values for promoting green, low-carbon cities and advancing urban sustainable development. Using data by sub-district from the 1% Population Sampling Survey in Shanghai, this paper explores commuting efficiency for three travel modes (non-motorized, public transport, and cars) across educational worker subgroups. It results in a dataset of commuting efficiency by travel mode in Shanghai (2015). The dataset includes the following data of Shanghai in 2015: (1) the number of commuting flows in sub-districts; (2) descriptive statistics of travel modes for different educational worker subgroups; (3) results of commuting efficiency metrics for different travel modes; (4) observed commuting flows matrix for each education-mode subgroup. The dataset is archived in .xlsx data format, and consists of one file with data size of 15.5 MB.

Keywords: commuting efficiency; excess commuting; jobs-housing balance; mode; Shanghai

DOI: https://doi.org/10.3974/geodp.2024.04.08

CSTR: https://cstr.escience.org.cn/CSTR:20146.14.2024.04.08

Dataset Availability Statement:

The dataset supporting this paper was published and is accessible through the Digital Journal of Global Change Data  Repository at: https://doi.org/10.3974/geodb.2024.09.05.V1 or https://cstr.escience.org.cn/CSTR:20146.11.2024.09.05.V1.

1 Introduction

Commuting is an important part of urban life, and its intensity can be seen as a sign of urban vitality^[1,2]. However, excessive commuting volumes can lead to significant urban management challenges, including traffic congestion, resource depletion, and environmental pollution, all of which can severely hinder urban sustainable development^[3,4]. Urban land use fundamentally shapes residents?? commuting behaviors, and optimizing residential and   employment spatial layout is widely regarded as a key solution to addressing urban   commuting issues^[5,6]. In 2014, the National Plan on New Urbanization (2014?C2020), officially issued by the State Council of China, identified ??city-industry integration?? as a crucial strategy for resolving urban development problems, such as sleepers?? towns and ??ghost?? towns, during the rapid urbanization phase. In recent years, megacities such as Shanghai and Beijing have increasingly embraced the concept of jobs-housing balance in their urban development plans, aiming to enhance urban efficiency and livability.

Excess commuting serves as a critical paradigm for evaluating urban commuting      efficiency^[7]. However, existing studies mainly focus on the single perspective of travel cost, paying limited attention to the perspective of commuting spatial organization. Moreover, few empirical studies have explored these two perspectives within the same framework. It is limiting to assess urban commuting efficiency solely based on travel costs, as this overlooks the intricate dynamics between urban land use and commuting behavior. Additionally, previous studies have largely overlooked the interaction between travel modes and individual socioeconomic attributes within the framework of excess commuting.

Using data from the 2015 Shanghai 1% Population Sampling Survey Data, this study    examines urban modal commuting efficiency across educational worker subgroups within the excess commuting framework. It quantifies the nonlinear relationship between commuting costs and spatial organization, ultimately develops the dataset of Commuting Efficiency by Travel Mode Dataset in Shanghai (2015).

2 Metadata of the Dataset

The metadata of the Commuting efficiency by travel mode dataset in Shanghai (2015)^[8] is summarized in Table 1. It includes the dataset full name, short name, authors, year of the dataset, data format, data size, data files, data publisher, and data sharing policy, etc.

3 Methods

3.1 Data Source

This paper uses 2015 Shanghai 1% Population Sampling Survey Data, sourced from the Municipal Bureau of Statistics in Shanghai^[10], to construct urban commuting flow matrix. The raw data used in this study consists of samples with unequal proportions. All data are adjusted and weighted to account for the sampling proportions. Regarding residents?? socioeconomic attributes, educational levels are categorized into three simplified groups: low (elementary school or below), middle (junior or senior high school), and high (college or above). Travel modes are classified into three categories: non-motorized transport (walking and biking), public transport (buses and subways), and private cars. Following the existing research method^[11], commuting cost matrices are constructed using road network distances derived through ArcGIS network analysis.

Shanghai comprises 17 districts, including Huangpu, Xuhui, Changning, Jing??an, Putuo, Zhabei, Hongkou, Yangpu, Minhang, Baoshan, Jiading, Pudong, Jinshan, Songjiang, Qingpu, Fengxian, and Chongming, covering a total area of approximately 6,340 km². Since Chongming is an isolated island, this study focuses on the main urban area of Shanghai, excluding Chongming, and adopts the sub-district as the geographical spatial analysis unit. The dataset includes various commuting-related data, such as the number of residents and workplaces by

Table 1  Metadata summary of the Commuting efficiency by travel mode dataset in Shanghai (2015)

sub-district, mode splits for three educational worker subgroups, and the commuting flow matrix in Shanghai.

3.2 Algorithm

Within a given urban jobs-housing spatial structure, the disparity between actual commuting () and theoretical minimum commuting () is referred to as excess commuting^[12], which originates from the concept of wasteful commuting proposed by Hamilton in 1982^[13]. Based on commuting spectrum theory, this paper adopts theoretical minimum commuting as the lower benchmark and random commuting as the upper benchmark^[14]. A lower theoretical minimum commuting indicates a greater degree of jobs-housing intermixing, whereas a higher value suggests greater spatial segregation. It is highly sensitive to the spatial allocation of housing and employment and is often used as a planning tool to assess jobs-housing balance at the local level^[15]. Random commuting, by contrast, represents a commuting  pattern in which residents are indifferent to travel costs (i.e., the friction coefficient equals zero), thereby reflecting regional jobs-housing imbalance.

To calculate commuting benchmarks, this study employs the Transportation Problem in Linear Programming to estimate the theoretical minimum commuting () and its corres-

ponding commuting flow matrix. Additionally, the doubly constrained spatial interaction model is applied to compute the theoretical minimum commuting entropy () using the Newton-Raphson iterative algorithm.

Based on the observed jobs-housing distribution by sub-district and observed commuting flow matrix, the observed average commuting distance () and entropy () are  calculated. Finally, random commuting distance () and random entropy () are  derived by setting the decay coefficient (??) to zero. We evaluate and compare modal  commuting efficiency across educational worker subgroups using commuting benchmarks and excess commuting indicators, as summarized in Table 2.

Table 2  The indicators of commuting benchmarks and excess commuting

Note:denotes the number of commuters living in zone i and working at zone j; anddenote the total numbers of workers at zone i and jobs at zone j, respectively; C denotes the average commuting distance;denotes the commuting distance between zone i and zone j;anddenotehe associated unknown Lagrangean multiplier;anddenote the balancing factors^[16].

4 Data Results

4.1 Data Composition

This dataset includes following data of Shanghai in 2015: (1) the number of commuting flows by sub-district; (2) descriptive statistics of travel modes for different educational worker subgroups; (3) results of commuting efficiency metrics for different travel modes; (4) observed commuting flows matrix for each education-mode subgroup.

4.2 Data Results

Based on calculations and statistical analysis, Shanghai??s total permanent employed population is 12.47 million, with 12.04 million living and working within the main urban areas (excluding Chongming). Among them, 50.22% do not reside in the same street as their workplace, while 49.78% of them live and work within the same sub-district. Inter-zonal commuting flows are predominantly concentrated in five new towns and the central urban area. Figure 1 illustrates the spatial distribution of the employed population and employment opportunities (i.e., commuting volumes at origins and destinations). The employed population exhibits a decentralized spatial pattern, whereas employment opportunities display a combination of concentration and dispersion.

From the perspective of commuting modes, non-motorized travel, car travel, and public transport account for 55.73%, 19.08% and 25.19% of total trips, respectively. This indicates that, compared to Western countries, urban commuting in China remains dominated by  public transport and non-motorized travel, with car usage constituting a relatively small proportion. Examining modal differences across educational worker subgroups (Figure 2), the majority of lowly educated workers (92.34%) rely on non-motorized travel. In contrast, the proportion of non-motorized travel among highly educated workers is significantly lower, at only 22.85%. Instead, car travel and public transport make up 30.46% and 46.69% of their commutes, respectively. For workers with a middle level of education, car travel and public transport account for 13.25% and 13.42% of their commuting trips, respectively.

Figure 1  Maps of workers and employment by sub-district in Shanghai

Figure 2  Mode split across different educational subgroups

Table 3 and Figure 3 present commuting efficiency by travel mode and educational level. For theoretical minimum commuting (), low-educated workers experience the longest commuting distances (4.17?C6.56 km), whereas high-educated commuters travel the shortest distances (3.02?C5.06 km), followed by those with a middle level of education (3.94?C5.26 km). A similar trend is observed for random commuting (), where the commuting distance for low-educated workers is approximately 10 km longer than for high-educated workers. These findings indicate that job accessibility and jobs-housing balance are lowest for poorly educated workers at both local and regional scales. Although Chinese cities do not exhibit the racial discrimination and residential segregation found in American cities, the excessive separation of residence and employment imposes substantial time and economic burdens on disadvantaged groups, such as low-educated and low-skilled workers. Previous studies have demonstrated that job-housing separation can undermine the information-seeking and mobility capabilities of vulnerable groups, restricting their access to opportunities in urban housing and labor markets. Such disparities may exacerbate broader urban social and spatial issues^[17].

Observed commuting distances () range from 4.35 km to 15.91 km. For car travel, low-educated workers commute shorter distances (8.70 km) compared to high-educated workers (12.87 km). Some studies^[18] identify two distinct commuting patterns by car in Shanghai: long-distance and short-distance commuting. This suggests that the short-distance commuting pattern may be associated with the habits and preferences of newly affluent  individuals in China, for whom driving is seen as a form of conspicuous consumption^[19]. For public transport, low-educated workers commute longer distances (14.14% farther) than high-educated workers. This can be attributed to the spatial distribution of jobs and housing in Shanghai, which forces low-educated commuters to travel, on average, 22.87% farther than their high-educated counterparts. Another possible reason is that low-educated   commuters are more willing to endure longer commutes in exchange for lower housing costs. High rents and property prices make it difficult for low-educated workers to find affordable housing near their workplaces, forcing them to live in suburban areas far from their     employment. Previous studies have emphasized the significant role housing pressure plays in the jobs-housing relationship and residents?? subjective well-being in Chinese cities^[20].

Entropy is used to reflect the orderliness of commuting spatial organization, with higher entropy indicating a more organized commuting pattern. Among the different modes, the observed entropy for non-motorized travel is the lowest (5.62?C6.98), while public transport exhibits the highest entropy (8.46?C9.27). As shown in Table 3, there is a clear relationship between observed entropy and observed commuting distance. For example, low-educated and high-educated workers using public transport have observed average commuting   distances () of 15.91 km and 13.66 km, respectively, with corresponding entropy values () of 8.46 and 9.23. This suggests that, compared to high-educated commuters, low-educated commuters using public transport endure longer commuting distances but  experience more organized commuting patterns. Similarly, for non-motorized commuters, those with low and high education levels have observed average commuting distances () of 4.72 km and 4.35 km, respectively, while their entropy values () are 5.62 and 6.98, respectively. This indicates that, compared to non-motorized commuters with higher education, those with lower education levels have longer, yet more organized commuting patterns.

Table 3  Results of commuting efficiency metrics based on different travel modes

In terms of average commuting distance, non-motorized commuters with low educational level experience the least excess commuting (EC=11.65%) and exhibit the highest commuting efficiency. In contrast, high-educated car commuters face the greatest excess commuting (EC=64.51%) and the lowest commuting efficiency. However, when considering the upper benchmark, high-educated public transport commuters demonstrate the lowest commuting efficiency (NEC=57.42%), which is attributed to the centralized layout of public transport, particularly rail transit. From the perspective of spatial organization, except for low-educated public transport users and high-educated car users, the rankings of normalized excess entropy (NEH) and normalized excess commuting (NEC) are nearly identical. Moreover, long commuting patterns are not necessarily inefficient (e.g., low-educated public transport users), and short commuting patterns are not always efficient (e.g., high-educated public transport users). This highlights that commuting efficiency is determined by the  relative position of observed commuting within the commuting spectrum.

Figure 3  Modal excess commuting indicators across different educational subgroups

5 Discussion and Conclusion

Based on data from the 1% Population Sample Survey in Shanghai, this paper computes commuting benchmarks and excess commuting indicators for different travel modes, quantifies the nonlinear relationship between average travel distance and the spatial organization of commuting, and provides insights and references for studies on commuting efficiency. The key findings are as follows: (1) non-motorized travel accounts for 55.73%, while travel by car and public transport accounts for 19.08% and 25.19%, respectively. Compared to Western countries, the proportion of motorized travel in Chinese cities remains relatively low; (2) significant differences in commuting efficiency are observed across travel modes. From both the average commuting distance and spatial organization perspectives, commuting efficiency is ranked as follows: non-motorized > car > public transport. This indicates that public transport travel has the lowest commuting efficiency and the greatest potential for optimizing the jobs-housing relationship. The current urban land use patterns in Shanghai are not particularly conducive to public transport use; (3) an interaction exists between travel mode and residents?? socioeconomic attributes regarding commuting efficiency. In cross-dimensional analysis, a long commuting pattern (e.g., low-educated commuters using public transport) is not necessarily inefficient or disordered, while a short commuting pattern (e.g., high-educated commuters using public transport) is not inherently efficient or orderly; (4) commuting efficiency exhibits a curvilinear relationship from both the average commuting distance and spatial structure perspectives. The organization of commuting reflects the orderliness and structure of urban commuting patterns and is closely related to road congestion and traffic demand management. For two cities with the same average commuting distance, the city with a more disorganized commuting pattern typically experiences more chaotic and random commuting flows, which are more likely to lead to serious traffic congestion. Thus, it is essential to study commuting efficiency from both the average commuting distance and commuting organization perspectives.

This dataset captures the jobs-housing relationship and commuting efficiency for different travel modes in Shanghai. It provides data support for optimizing urban spatial structure and promoting low-carbon travel. Future research can use this dataset to analyze the evolving characteristics of commuting efficiency in Shanghai and conduct intercity comparison studies on commuting efficiency.

Author Contributions

Zhu, Y. and Yue, L. Y. designed the algorithms of dataset. Yue, L. Y. and Li, K. M. contributed to the data processing and analysis. Yue, L. Y., Zhu, Y. and Li, K. M. wrote the data paper.

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]     Niedzielski, M. A. A spatially disaggregated approach to commuting efficiency [J]. Urban Studies, 2006, 43(13): 2485‒2502.

[2]     Liu, X. T., Chen, X. M., Zhou, J. P. Study on jobs-housing spatial relationship and commuting efficiency: evaluation potential of excess commuting [J]. Urban Transport of China, 2018, 16(2): 10‒18.

[3]     Horner, M. W. Extensions to the concept of excess commuting [J]. Environment and Planning A: Economy and Space, 2002, 34(3): 543‒566.

[4]     Liu, W. B., Yan, X. P., Fang, Y. P., et al. Related characteristics and mechanisms for excess commuting in Guangzhou [J]. Acta Geographica Sinica, 2008, 63(10): 1085‒1096.

[5]     Zhu, W., Liang, X. M., Gui, C., et al. The inter-generational differences in the effects of job-housing optimization in Shanghai [J]. Acta Geographica Sinica, 2020, 75(10): 2192‒2205.

[6]     Han, H. R., Yang, C. F., Song, J. P. Impact factors and differences in commuting efficiency between public transit and private automobile travel: a case study on the Beijing metropolitan area [J]. Geographical Research, 2017, 36(2): 253‒266.

[7]     Yue, L. Y., O??Kelly, M. E. Commuting performance in Shanghai: efficiency and orderliness [J]. Travel   Behaviour and Society, 2023, 31: 223‒231.

[8]     Yue, L. Y., Zhu, Y., Li, K. M. Commuting efficiency by travel mode dataset in Shanghai (2015) [J/DB/OL]. Digital Journal of Global Change Data Repository, 2024. https://doi.org/10.3974/geodb.2024.09.05.V1. https://cstr.escience.org.cn/CSTR: 20146.11.2024.09.05.V1.

[9]     GCdataPR Editorial Office. GCdataPR data sharing policy [OL]. https://doi.org/10.3974/dp.policy.2014.05 (Updated 2017).

[10]  Shanghai Municipal Bureau of Statistics. 2015 Shanghai 1% Population Sampling Survey Data [M]. Beijing: China Statistics Press, 2017.

[11]  Frost, M., Linneker, B., Spence, N. Excess or wasteful commuting in a selection of British cities [J]. Transportation Research Part A: Policy and Practice, 1998, 32(7): 529?C538.

[12]  Zhang, H., Xu, S., Liu, X., et al. Near ??real-time?? estimation of excess commuting from open-source data: evidence from China??s megacities [J]. Journal of Transport Geography, 2021, 91(4): 1‒14.

[13]  White, M. J. Urban commuting journeys are not wasteful [J]. Journal of Political Economy, 1988, 96(5): 1097‒1110.

[14]  Yang, J. W., Ferreira, J. Choices versus choice sets: a commuting spectrum method for representing job-housing possibilities [J]. Environment and Planning B: Planning and Design, 2008, 35(2): 364‒378.

[15]  Horner, M. W., Murray, A. T. Excess commuting and the modifiable areal unit problem [J]. Urban Studies, 2002, 39(1): 131‒139.

[16]  O??Kelly, M. E., Niedzielski, M. A. Are long commute distances inefficient and disorderly? [J]. Environment and Planning A: Economy and Space, 2009, 41(11): 2741‒2759.

[17]  Liu, Z. L., Wang, M. J. Job accessibility and its impacts on commuting time of urban residents in Beijing: from a spatial mismatch perspective [J]. Acta Geographica Sinica, 2011, 66(4): 457‒467.

[18]  Sun, B. D., Dan, B. Impact of urban built environment on residential choice of commuting mode in Shanghai [J]. Acta Geographica Sinica, 2015, 70(10): 1664‒1674.

[19]  Li, W., Dan, B., Sun, B. D., et al. The influence of rail transit accessibility on the shift of travel modal choice: empirical analysis based on the micro survey of the 1980s generation in Shanghai [J]. Geographical Research, 2017, 36(5): 945‒956.

[20]  Zhan, D. S., Kwan, M. P., Zhang, W. Z., et al. The impact of housing pressure on subjective well-being in urban China [J]. Habitat International, 2022, 127(3): 1‒10.