^{1}

^{1}

^{1}

^{2}

^{1}

^{2}

Efficiency and equity have always been the two points of focus of transport projects. Compared with efficiency, equity is easily overlooked in the evaluation of transport projects. Many studies emphasize that defining and operationalizing costs and benefits and the distributive principle are critical parts in the assessment of transportation equity. However, the scope and time frame of the assessment target are also critical. In this paper, we took China’s fastest urbanizing city, Shenzhen, as a case study to assess transport equity by comparing accessibility among groups. First, the public transport system was divided into bus and subway, and the residents were divided into two groups: urban village and nonurban village. Second, we adopted an enhanced potential opportunity model to measure residents’ bus and subway accessibility and summarized them as transit opportunity. Third, we used the Dagum Gini coefficient decomposition and kernel density estimation method to explore the fair distribution of transit opportunity among groups and districts from 2011 to 2020. Decade-long changes in disparity and distribution of transit opportunity gave us a clear picture. On the one hand, the development of Shenzhen public transport system had a positive effect. All populations are benefiting, and their accessibility is increasing. On the other hand, it also had a negative effect to exacerbate inequality between populations. For the absolute value of the opportunity, Shenzhen’s urban village populations do have fewer transportation opportunities than nonurban villages, and this gap between them will be wider more and more. The public transport system is more inclined to improve the population with high initial opportunity and make them higher. The results illustrated the importance of examining transportation equity over an extended period and could provide information on urban development strategies.

Public transport is an effective way to solve the problem of traffic congestion and environmental pollution in high population density metropolitan. More importantly, it provides the necessary motorized transport to access jobs and social activity needed especially for low-income people without cars [

Transport-related equity involves a wide range of topics and previous studies can be divided into four areas:

This paper is organized as follows: Section

Conducting transport equity analysis first involves conceptual issues of equity. The definition of equity has extensive discussions in all fields from philosophy to economics. It differs in different historical periods and different perspectives of research. The definition of equity used in this study is “the distribution of benefits and costs over members of society” [

Three issues need to be clear when conducting an equity assessment of transport policy or infrastructure projects [

Many variables can be used to represent the costs and benefits in transportation equity, and the most often used are transport affordability, access to transport, and accessibility to opportunities. Transport affordability measures individuals or household’s actual expenditure on public transport usually as the percent of household disposable income [

Accessibility to opportunities is related to cumulative-opportunity and potential/gravity measures which sum the number of destinations/jobs reachable within certain times by transport mode; substantial literature discusses measure [

Equity analysis needs to define a unit that can be distinguished, and units usually are groups of people/households or regions. Many studies use demographic and geographic factors categorizing people to identify transport disadvantaged people in equity evaluation [

Well-known horizontal equity measures are the Gini coefficient, Theil index, and coefficient of variation; they are expressed as ratios which are compared among groups to measure equity performance. Gini coefficient initially indicates level of equality of income distributions in economic studies; it ranges from 0 to 1; 0 means absolute equality, and 1 indicates absolute inequality; in transportation equity it is used to evaluate the degree of accessibility concentration level of different regions or groups of people and compare the level of equity before and after implementing a policy or transport infrastructure. Some argue that the Gini coefficient fails to indicate the structure of inequality; the same Gini coefficients can have different income distributions by the group. Theil index is using the information entropy concept to measure individual or interregional income inequality named. The Theil index has good decomposability as a measure of inequality when the sample is divided into multiple groups. The Theil index can measure the contribution of the intragroup gap and the intergroup gap to the total gap, so it provides more interpretation of the inequality among different groups. The primary approach of vertical equity is to evaluate transport policy or infrastructure projects according to how they affect accessibility between disadvantaged people/households or regions. It is fairer if transportation disadvantaged group benefits, like transport service improvements, favor lower-income areas, and groups, or transportation services provide more access to job opportunities and other “basic” activities.

The keys to equity analysis of public transportation are the measure of accessibility and the equity measure of distribution. Our work complements previous research from four aspects. First, when evaluating the equity impact of public transport, only one mode of public transport is concerned in previous studies, so the result of equity evaluation could be bias. We combined subways and buses to consider equity issues in this paper and proposed an enhanced potential opportunity model to measure residents’ bus and subway accessibility considering public service reliability, attractiveness, and frequency. Second, due to the limitations of data acquisition, scholars discuss the impact on equity of transport policy or infrastructure projects during a relatively short period (before and after the implementation of the target). Our research used a long-term data to examine equity situation dynamic change, and it was helpful to capture the trends of equity influence on different groups in the development of public transport. Third, indicators of transport distribution effects were further explored and applied. We used the Dagum Gini coefficient decomposition which is more convenient than Theil measure and kernel density estimation method to investigate the fair distribution of potential opportunities. At last, existing research focused on Europe and the United States, and we took the China fastest urbanizing city, Shenzhen, as a case study to assess transport equity by comparing accessibility among social groups. The results can provide a reference for the study of the impact of transportation equity in the world.

In this study, the public transport system was including bus and subway, and we divided the residents into two groups: urban village population and nonurban village population which will be explained in Section

This research adopted cumulative-opportunity and potential/gravity measures models to measure transit-based job accessibility and made some enhancements. Given that accessibility measurement is especially important for the analysis of traffic equity, this section will detail how this research calculates accessibility.

Calculate the service range of each transit stop. The service radius of the bus stop is 500 meters, and the service radius of the metro station is 700 meters.

Calculate the population in each transit stop service area and calculate the job opportunity in each transit stop service area. In our study, job opportunity is represented by the floor area of factory, company, and government office.

For transit line

Calculate per capita service frequency:

Calculate

The access and egress times are assumed to be 5 min of walking time, which transit users are generally willing to undertake, waiting time at transit stops is assumed to be one-half of the scheduled headway when the average headway of transit service is around 10 min, and the in-vehicle travel time is calculated using the scheduled arrival and departure time that is obtained from transit service schedules.

Calculate the distance decay factor:

Calculate the job opportunity of

where

Consider the job opportunity of

where

Calculate the cumulative opportunities of i:

Convert residential area to the centroid and calculate the sum of job opportunities of bus stops in the 500-meter buffer of residential centroid and the number of metro job opportunities of metro stations in the 700-meter buffer of residential centroid (if there are more than 1 transit stops belonging to the same line in the buffer, they will be averaged as 1 transit stop). The sum of bus and metro opportunities is the transit opportunity for residential areas.

The method of decomposition of the Gini coefficient in a discrete space is proposed by Dagum [

Decomposition of the Gini coefficient not only effectively solves the source of group disparities but also describes the distribution of subgroups and solves the problem of overlap between groups (shows the structure of inequality). In this paper, the Gini coefficient of transit opportunity is calculated and decomposed; the population in Shenzhen is divided into different groups; it helps us to know if the transit opportunity gaps within groups generate the inequalities or if the transit opportunity gaps between groups engender the inequalities.

Kernel density estimation (KDE) is a nonparametric way to estimate the probability density function of a random variable in statistics [_{1},_{2}, …,_{n}) be a univariate independent and identically distributed sample drawn from some distribution with an unknown density

where

As mentioned in Section

Shenzhen is located in the Pearl River Delta region with a land area of 1996.8 km^{2} and an urban population of over 14 million in 2016. It is the first Special Economic Zone (SEZ) city after the institution of reform and the Open-Door Policy in China in 1979. In the past 30 years, the operation of a market economy has made Shenzhen’s economy develop rapidly, bringing with it a dramatic population increase and spatial expansion. In the study of transport equity, an important part is to group residents according to their socioeconomic level. In Shenzhen, detailed data on residents’ occupations and income is not easily accessible, so we use three characteristics of a resident’s residence to reflect his/her socioeconomic status. These three characteristics are average house price of residence, the average rental price of residence, and whether the residence is in the urban village; the third feature especially is an essential basis for judging a resident as disadvantaged people.

Urban village (Cheng Zhong Cun in Chinese), some scholars preferring the term “urbanized villages” or “villages in the city” to avoid the confusion with the Western planning concept “urban village”, is an outcome of China’s rapid urbanization and its associated rural-urban migration. When urban expansion encroaches into rural land; the city government needs to acquire land rights from the rural collective to convert the rural land into urban land. The city government only expropriates the farmland of the village to avoid the costly compensation to relocate villagers, and the housing land remains in the hands of the collective. Over time, the village settlement is surrounded by urban built-up area, creating the so-called urban village.

The data supplied by the Shenzhen Urban Planning Bureau (SUPB) and the Urban Planning and Design Institute of Shenzhen (UPDIS) shows that there are 2,942 urban village residential lands and 4,683 nonurban village residential lands in Shenzhen 2009. The Municipal Building Survey 2009 provides information for all buildings in Shenzhen, including the urban villages. There are 615,702 buildings in 2009 and 333,576 (54%) in urban villages. Urban villages in Shenzhen, which are thought to accommodate approximately seven million, meet the basic needs of people, particularly poor and low-income residents [

Our study first divided the population into two groups: residents living in urban villages called the urban village group (UVG) and residents not living in urban villages called the Not urban village group (NUV). The UVG is mainly composed of low-income migrants and contains some high-income local residents. A study shows that the ratio of local residents to migrants in the urban village is 1:88 [

Average house price and average rental price of ten districts in Shenzhen in 2017.

| | | |
---|---|---|---|

| Futian | 52968 | 119.5 |

Luohu | 38143 | 94.7 | |

Nanshan | 56597 | 121.7 | |

| |||

| Baoan | 22580 | 76.9 |

Longgang | 27567 | 50.3 | |

Longhua | 36432 | 61.2 | |

Yantian | 33970 | 59.4 | |

| |||

| Dapeng | 13377 | 27.3 |

Pingshan | 12601 | 39.3 | |

Guangming | 10278 | 24.6 |

Distributions of urban village lands and nonurban village lands in 2011.

Futian, Luohu, and Nanshan are the core areas of Shenzhen. Investments have been made in the service industry and high-technology companies in the three areas, so the house price and rental price are the highest. Baoan, Longhua, Longgang, and Yantian are the subcenter of Shenzhen, and manufacturing industry provides many jobs in these three regions, so the house price and rental price are lower than in core areas. The remaining three regions are relatively far from the city center, and they have the lowest house price and rental price. For core areas, subcenters, and suburbs, there is a clear difference between house prices and rents, so NUV and UVG are each divided into three subgroups. The descriptions are shown in Table

Subgroups of NUV and UVG in Shenzhen.

| | | | |
---|---|---|---|---|

| | Residents who live in formal urban housing at core area (Futian, Luohu, and Nanshan) | 4007619 | 27.8% |

| Residents who live in formal urban housing at subcenter (Longhua, Yantian, Longgang, and Baoan) | 3505463 | 23.69% | |

| Residents who live in formal urban housing at suburb (Dapeng, Pingshan, and Guangming) | 162948 | 0.37% | |

Total | 7676030 | 51.86% | ||

| ||||

| | Residents who live in the urban village at core area (Futian, Luohu, and Nanshan) | 1402041 | 9.48% |

| Residents who live in the urban village of subcenter (Longhua, Yantian, Longgang, and Baoan) | 4724083 | 31.93% | |

| Residents who live in the urban village at suburb (Dapeng, Pingshan, and Guangming) | 995467 | 6.73% | |

Total | 7121591 | 48.14% |

This study mainly analyzes the changes in public transport opportunities in the three periods from 2011 to 2020. 2011 is a base scenario, and 2020 is analyzed as a planning scenario. Calculating transportation opportunity requires transportation network and active location information. Public transport data includes subway and bus network as shown in Figure

Maps of the transit network in three periods.

The minimum unit for calculating transit opportunity is the residential land unit, and transit opportunity at the different aggregate levels are calculated by population weights. Figure

Maps of transit opportunity in three periods.

We examined the average transit opportunity of two groups in each district. Table

Transit opportunity of two groups in each region.

| | | | |||
---|---|---|---|---|---|---|

| | | | | | |

Futian | 12381 | 13156 | 19592 | 21091 | 23095 | 23848 |

| ||||||

Luohu | 12477 | 13041 | 16143 | 15106 | 18823 | 17035 |

| ||||||

Nanshan | 5713 | 5886 | 8056 | 7372 | 10231 | 9035 |

| ||||||

Baoan | 3367 | 2012 | 3821 | 2185 | 4197 | 2323 |

| ||||||

Longgang | 4602 | 3337 | 4919 | 3543 | 6340 | 4820 |

| ||||||

Longhua | 4499 | 2736 | 4805 | 2848 | 6170 | 4572 |

| ||||||

Yantian | 1630 | 1285 | 2030 | 1411 | 4931 | 2858 |

| ||||||

Dapeng | 274 | 373 | 431 | 420 | 431 | 420 |

| ||||||

Pingshan | 1163 | 1107 | 1231 | 1134 | 1231 | 1134 |

| ||||||

Guangming | 1862 | 909 | 1881 | 927 | 2835 | 1297 |

Table

The average transit opportunity of two groups in the whole city.

| | | | |
---|---|---|---|---|

| | 7307 | 9959 | 11972 |

| ||||

| | 10597 | 15367 | 18249 |

| 3819 | 4172 | 5255 | |

| 1457 | 1530 | 2090 | |

| ||||

| | 4028 | 4846 | 5870 |

| ||||

| | 10825 | 14397 | 16480 |

| 2662 | 2835 | 3726 | |

| 943 | 987 | 1102 |

To analyze the horizontal equity of public transport, we calculated and decomposed the Gini coefficient using the transit opportunity of 7625 residential units; the total populations of Shenzhen was divided into the following subgroups: UVG and NUV. Table

Decomposition of the Gini coefficient between UVG and NUV.

| | | | |
---|---|---|---|---|

| | 0.5725 | 0.5916 | 0.5736 |

| ||||

| | 0.5049 | 0.5160 | 0.4920 |

| 0.6267 | 0.6480 | 0.6385 | |

| ||||

| | 0.6026 | 0.6303 | 0.6140 |

| ||||

| | 48.14% | 47.63% | 47.38% |

| 25.04% | 28.84% | 29.45% | |

| 26.81% | 23.53% | 23.17% |

As for within group inequity,

The between-group inequity (

We know that within groups inequality has the most significant impact on overall inequality in Table _{i} means the equity index of transit opportunity distribution of subgroup_{2} and G_{3}, G_{1} is the smallest and is consistently decreasing in three periods, and its decrease is also the largest. G_{2} and G_{3} are almost equal in 2011 and 2016, but the value of G_{3} will be larger in 2020. The analysis of inequality contribution of

Decomposition of the Gini coefficient of NUV between subgroups.

| | | | |
---|---|---|---|---|

| 0.5049 | 0.5159 | 0.4920 | |

| ||||

| | 0.4060 | 0.3685 | 0.3398 |

| 0.5045 | 0.5192 | 0.5131 | |

| 0.5097 | 0.5101 | 0.5458 | |

| ||||

| | 42.6% | 38.75% | 38.19% |

| 47.32% | 55.68% | 56.36% | |

| 10.8% | 5.57% | 5.45% |

Table _{4} is significantly different from G_{5} and G_{6}, and the most equitable distribution of transit opportunity is subgroup 4. From the changes in the Gini coefficient, the inequality of all UVG population has increased, and the opportunity distribution gap of UVG population in core area tends to be smaller. Subgroup 5 is the most unfair distribution of opportunities in UVG. G_{6} is slightly smaller than G_{5}. The construction of subway infrastructure from 2011 to 2020 will have a negative impact on the distribution of transit opportunity in subcenters and suburbs. From the analysis of the contribution to overall inequality of UVG, the inequality among the groups

Decomposition of the Gini coefficient of UVG between subgroups.

| | | | |
---|---|---|---|---|

| 0.6266 | 0.6480 | 0.6385 | |

| ||||

| | 0.4333 | 0.3748 | 0.3567 |

| 0.5441 | 0.5525 | 0.5760 | |

| 0.5424 | 0.5463 | 0.5572 | |

| ||||

| | 32.76% | 28.85% | 31.53% |

| 59.51% | 65.66% | 62.46& | |

| 7.73% | 5.49% | 6.01% |

Table

The Gini coefficients and transit opportunities of all subgroups.

| ||||||||
---|---|---|---|---|---|---|---|---|

| | | ||||||

| | | | | | | | |

| 0.406 | 10597 | | 0.3685 | 15367 | | 0.3398 | 18249 |

| 0.4333 | 10825 | | 0.3748 | 14397 | | 0.3567 | 16480 |

| 0.5045 | 3819 | | 0.5101 | 1530 | | 0.5131 | 5255 |

| 0.5097 | 1457 | | 0.5192 | 4172 | | 0.5458 | 2090 |

| 0.5424 | 943 | | 0.5463 | 987 | | 0.5572 | 1102 |

| 0.5441 | 2662 | | 0.5525 | 2835 | | 0.576 | 3726 |

Based on the transit opportunity of 7825 settlements, we used kernel density estimation to get the probability of transit opportunity and plotted the probability density distribution map.

Figure

Distribution of transit opportunity probability of NUV and UVG population.

The top of Figure

Figure

Comparisons of the probability distribution of the two groups in the whole city.

Figure

Comparisons of the probability distribution of the two groups in core areas.

Comparisons of the probability distribution of the two groups in subcenter.

Comparisons of the probability distribution of the two groups in the suburb.

Decade-long changes in disparity and distribution of transit opportunity gave us a clear picture, and the results illustrated the importance of examining transportation equity over a long period.

Our research is beneficial for providing information to adjust the planning of future Metro routes and urban development strategies in Shenzhen. Since social status and spatial location are factors of impact equity, Shenzhen should strengthen public transport services in subcentral areas, especially Baoan and Longgang. The two regions have a large population for both NUV and UVG, and improvement of public transport in these areas will be the most effective way for improving public transport accessibility and fair distribution. Our study also has some limitations. First, we used the same bus network for the 10-year analysis period, and this operation would cause deviations in the calculation results of transit opportunities. So, the analysis of equity ignored the changes in fairness brought about by the improvement of bus service level. Second, transportation equity not only is an infrastructure issue but also involves land use planning as well. The transit opportunity of a resident is related to the public transport system and to the distribution and size of the opportunities. The distribution of opportunities is related to urban planning, especially land use. This study does not explore the impact of land use on transportation equity. Third, it is not only the distribution of access to destinations that matter, but also in some cases the absolute level of access for those who are worse. We only discussed the changes of the residents’ opportunities and the changes of disparity and distribution of transit opportunity, and we do not discuss whether transit opportunities meet the needs of different groups of people and their satisfaction of transportation opportunities. In the future research, those are our next research direction.

The population distribution data, employment distribution data, and transit data of Shenzhen used to support the findings of this study were supplied by Shenzhen Urban Planning Bureau (SUPB) and the Urban Planning and Design Institute of Shenzhen (UPDIS) under license and so cannot be made freely available. Requests for access to these data should be made to Qingfeng ZHOU, zhouqingfeng@hit.edu.cn.

The authors declarethat they have no conflicts of interest.

The authors gratefully acknowledge financial support from the China Scholarship Council. The authors gratefully acknowledge support from Program on Chinese Cities of Center for Urban and regional Studies, University of North Carolina at Chapel Hill.