Predicting the implementation effect of the municipal solid waste mandatory classification policy based on the residents’ behavior

ABSTRACT

The rapid development of China’s economy has led to a sharp increase in the municipal solid waste (MSW). China has promulgated the MSW mandatory classification policy. The implementation effect of such mandatory policy is an important concern to the government, scholars, and the public, but has not been extensively studied. This paper explores the implementation effect of such policy through assessing the waste classification behavior of residents based on the Cellular Automata model. The simulation results show that the residents will not adjust their MSW classification behavior during the time period when the mandatory classification policy is not implemented. But when the mandatory classification policy is imposed, the residents will adjust their classification behavior over time from t = 100 to t = 300. The results indicate that the extent of residents’ participation in waste classification varies by different enforcement intensities. The higher the intensity is, the more rapidly the residents’ MSW classification behavior improves. The results also indicate that the extent of residents’ participation in waste classification varies by different urban population size. The larger the urban population size is, the higher the proportion of classification participation is.

Implications: The implementation effect of the municipal solid waste mandatory classification policy is an important concern to the government, scholars, and the public, but has not been extensively studied. This paper explores the implementation effect of such policy through assessing the waste classification behavior of residents based on the Cellular Automata model. The simulation results show that the residents will not adjust their municipal solid waste classification behavior during the time period when the mandatory classification policy is not implemented. But when the mandatory classification policy is imposed, the residents will adjust their classification behavior over time. The results also indicate that the extent of residents’ participation in waste classification varies by different enforcement intensities and different urban population sizes.

Introduction

With the acceleration of China’s urbanization process and the influx of rural population flooding into cities, the amount of Municipal Solid Waste (MSW) has risen sharply (Chu, Xi, and Li 2012). In 2017, China collected and transported a staggering amount of 215.21 million tons of MSW (China Urban Construction Statistical Yearbook 2017). To address the worrisome “waste siege” problem, the Chinese government has promulgated a series of laws and regulations and systems, of which the Municipal Solid Waste Mandatory Classification Policy (MSWMCP) is the most important one. There are several reasons for invoking this mandatory classification policy, such as first, the MSWMCP can effectively reduce the occupation of city land, which is getting scarce (Liu 2011). Second, the MSWMCP can effectively curb various environmental pollution problems caused by MSW (Liu 2011). Third, the MSWMCP can achieve resources recycling (Liu 2011).

Since the promulgation of MSWMCP, the implementation effect of the policy has been an important issue of concern to the government, academia, and the public. Of particular concern is how the residents adjust their MSW classification behavior in response to MSWMCP. However, research on this issue remains understudied. Urban residents are the main consumers, users of goods, and the waste generators, and hence the residents’ MSW classification behavior plays a crucial role in determining whether the implementation of MSWMCP can achieve the expected goals (Chen 2013). Therefore, it is of great theoretical and practical significance to study the implementation effect of MSWMCP based on the residents’ behavior.

The influencing factors of the effect of MSW policy

Scholars have found that policy subjects, policy objects, and social factors all have an impact on the effect of MSW policy. Some researchers have investigated how policy subjects can influence the effect of the MSW policy. Anantanatorn et al. (2015) indicated that policy subjects (contents) can inspire people and improve the level of MSW management by providing quality services. Kirakozian (2016) pointed out that economic incentives can induce residents to change their behavior. However, Moh, Manaf, and Juahir (2014) found that the role of economic incentives in promoting waste recycling may be overstated. In addition, the design of policy should also fully consider the actual feelings of the people, and the implementation of policies that bring convenience is more conducive to improving the effect of waste collection than the implementation of punitive policies (Mueller 2013), and mixed usage of incentive and punitive policies can help achieve waste reduction (Zimmermann 1988).

Past studies have also found that policy objects affect the effectiveness of MSW policy. Many local governments believe that residents play an influential role on the management of MSW (Trilnick and Tal 2014). Ojewale (2014) pointed out that the educational level of policy objects affects the implementation of policy, and highly educated people are more rational and reasonable in choosing the waste disposal method. Also, residents with more relevant knowledge are more willing to participate in waste collection activities (Gamba and Oskamp 1994; Radzi and Ain 2016).In addition, Ilevbare Femi et al. (2014) found that the psychological attributes or dispositions of policy objects can affect MSW management. For instance, more responsible residents are often more likely to comply with the waste policy (Indrakaran, Sharifah, and Irniza Fazizi 2014).

Some scholars pointed out that social factors also have an important bearing with the effect of MSW policy. For example, Nicolli et al. (2012) conducted an empirical study in Europe and showed that population density can affect the implementation of landfills. Thiyagarajah, Bi, and VanSickle (2016) pointed out that population density is positively correlated with the MSW recycling rate. Oyekale (2015) used a binary probability model for empirical research, and found that household income is an important factor affecting waste disposal and recycling. Similarly, Xu et al. (2017) found that social factors such as income can affect waste recycling and waste classification. Social factors have become a dimension that must be considered in the field of MSW management (Vieira 2016).

The methods of analyzing MSW policy

Scholars have adopted a variety of methods to analyze policies related to MSW. One commonly used method for the MSW policy analysis is Material Flow Analysis (MFA). Shimizu and Okayama (Shimizu and Okayama 2004) used MFA to analyze the waste management system and its impact on the environment. Muchangos, Tokai, and Hanashima (2017) used MFA to evaluate the waste management system of Maputo City. Stanisavljevic and Brunner (2014) used MFA to solve the decision-making problem in waste classification and disposal. Owens, Zhang, and Mihelcic (2011) studied the household waste disposal strategy of developing island countries through MFA.

Another commonly used method is Life Cycle Assessment (LCA), which can link the causal chain between the research process and environmental impact (Marchand et al. 2013), and is considered an effective tool for identifying and assessing the environmental impact of waste management programs (Cherubini, Bargigli, and Ulgiati 2009). Lazarevic (2013) established the rationale for applying LCA in the field of waste management. Vidal, Gallardo, and Ferrer (2001) argued that LCA can greatly improve the decision-making process and help decision-makers scientifically analyze the effect of waste policy. Cao et al. (2014) combined LCA with a cost-benefit analysis to study the benefit of MSW classification policy. Ojoawo, Bosu, and Oyekanmi (2014) analyzed MSW policy and its impact on the environment through LCA.

Full Cost Accounting (FCA) is widely used in analyzing the complete life cycle of waste policy (Association). Metin, Eröztürk, and Neyim (2003) proposed the specific application of FCA in the field of MSW management. Bakshi (2016) believed that the FCA framework helps to improve MSW management. Paul, Boorsma, and Gaudiel used the FCA to conduct research on the household waste management system in Bayawan City and offered advice to improve the system. A local government in Thailand used FCA to gain a deeper understanding of the actual cost of MSW management and to promote better implementation of the management objective (Chanchampee 2010). However, Lim (2012) found that the lack of corresponding regulations hinders effective MSW management.

The above brief review of the existing literature shows that scholars have studied the MSW policy from two aspects: exploring the influencing factors and using various methods for analysis and have achieved fruitful research results. However, most of the previous studies on the MSW policy were carried out from a static perspective; few scholars have conducted dynamic research on the implementation effect of MSW policy. This lack of dynamic research limits the offering of an effective solution to the MSW management problem. To address that gap of the literature, we construct a dynamic prediction model of the implementation effect of MSWMCP. Our paper aims to enrich the relevant research in the field of MSW management and to provide useful insight to effective MSW management.

Method

Cellular automata (CAs), originally designated by Stanislaw Ulam and John von Neumann in the 1940s, are a group of dynamic systems whose time, space, and states are all discrete (Xu et al. 2018). They have been confirmed to be useful both for constructing theoretical models to characterize chaos and complexity of non-linear dynamics and for designing specific applications in a broad range of scientific fields (Neumann 1966). In addition, Xue et al. (2013) pointed out that cellular automata would be a valuable tool for simulating human behavior. Therefore, it is scientifically valid and practical to use the cellular automata to simulate the residents’ participation in MSW classification.

Fundamental construct of the model

In the context of the cellular automata model, the residents living in the urban area are nodes in the network, and the social relationship between the residents is represented by the relationship between the nodes of the cellular automata model. Residents’ neighbors are a certain group of people who are concerned with the residents’ MSW classification behavior. The residents’ MSW classification behavior is also affected by their neighbors due to the interactions among them. The living area of all the residents constitutes a two-dimensional cellular space of the model. In the model, the residents have two possible states: the residents who participate in MSW classification/sorting are indicated by state 1, otherwise indicated as in state 0.

In the determination of the cellular space, the fixed boundary is selected as the boundary condition of the cellular space. The cells in the specified boundary are always in state 0 and do not change with the evolution of the model, so as to eliminate the uncertainty of cellular evolution in the spatial boundary. On the determination of the cells’ neighbors, the model of the $Moore$neighborhood is selected. The model of the $Moore$ neighborhood is the model in which the cell has eight neighbors. When the residents are located inside the space, they have eight neighbors. When the residents are located on the edge of the space (except for the four corners), they have five neighbors. When the residents are located on the corner of the space, they have three neighbors. The positional relationship is illustrated in Figure 1, where the black node indicates the resident, and the numbered nodes indicate the neighbors of the resident.

Figure 1. The rule of cellular neighbor

Constraints

Residents’ neighbors are able to constrain (influence) the residents’ MSW classification behavior. The behaviors of other people can make a big difference to urban residents’ classification collection intentions of household solid waste (Wang, Dong, and Yin 2018). Further, the existence of the community strengthens the information interaction and sharing among residents and their neighbors. As a result, the attitude, perception, and will of the residents are more susceptible to the influence of residents’ neighbors in the same community. Over time, their behaviors and conceptions in the same community tend to exhibit a certain positive convergence.

To quantitatively describe the constraint of residents’ neighbors on the residents’ MSW classification behavior, we set the parameter $\alpha \left(0.5\le \alpha \le 1\right)$which implies a random distribution. It is used to measure the binding force of different residents’ neighbors. The larger the neighbor’s $\alpha$, the stronger is the neighbor’s constraint. By summing all constraints of the resident’s neighbors, we can obtain the total constraint of the resident’s neighbors.

If the neighbors’ total constraint exceeds a certain limit, the residents will change MSW classification behavior. The limit is related to the psychological characteristics of the residents. It is expressed by $R_L$ and $R_H$. If the neighbors’ total constraint is greater than $R_H$, the residents who are initially unwilling to participate in MSW classification will participate in MSW classification. If the neighbor’s total constraint is less than $R_L$, the residents who have already participated in MSW classification will give up participating in MSW classification.

The sphere outside the cellular space where the residents live is considered the external environment. Even if the residents live in different cellular spaces, there may be cross-regional interaction between residents because of the complexity of information exchange between residents. The constraint from the external environment is quantitatively described by $p_0$ and $p_1$, where $p_0$ represents the probability that the residents randomly change to state 0 under the constraint from the external environment, and $p_1$ represents the probability that the residents randomly change to state 1 under the constraint from the external environment.

Transition rules

The residents’ MSW classification behavior is also constrained by their previous behavior. Therefore, the MSW classification behavior of the residents is mainly determined by four factors:

The previous MSW classification behaviors of the residents;

The constraint from the residents’ neighbors;

The residents’ ability to resist the constraint from their neighbors;

The constraint from the external environment.

The specific behavior algorithm can be expressed as the following:

(1) $s_i\left(t+1\right)=\sum {\alpha }_k{s}_k\left(t\right)+R_i+E_i+s_i\left(t\right)$(1)

where $s_i\left(t+1\right)$ – – the MSW classification behavior of the resident at the next time point, with a value of 0 or 1

$s_k\left(t\right)$ – – the MSW classification behavior of the neighbor at the current moment, with a value of 0 or 1

$R_i$ – – resident’s ability to resist the constraint from the resident’s neighbors, measured by $R_L$ and $R_H$

$E_i$ – – the constraint from the external environment on the resident, measured by $p_0$ and $p_1$

$s_i(t)$ – – the resident’s MSW classification behavior at the current moment, with a value of 0 or 1.

The transition rules of the residents’ MSW classification behaviors can be summarized as follows.

1. If the neighbors’ constraint is not inside the threshold interval $\left[R_L,R_H\right]$, the residents will change the behavior.

2. The residents are constrained by the external environment to randomly change the behavior with the probability of $p_0$ or $p_1$.

3.In other cases, the residents maintain their behavior.

According to the two-dimensional synchronous cellular automaton theory, the above transition rules can be summarized as follows.

(2) $s_i(t+1)=\left\{\begin{array}{c}0,\sum {\alpha }_k{s}_k(t)\le R_{L}or{p}_0\\ 1,\sum {\alpha }_k{s}_k\left(t\right)\ge R_{H}or{p}_1\\ s_i(t),otherwise\end{array}\right.$(2)

When the MSWMCP is implemented, it will cause some changes in the parameter values. First, the MSWMCP introduces assessment, rewards and punishments, and other auxiliary means to further increase the binding force of government supervision on the residents’ MSW classification behavior. This will cause changes in the parameters by lowering the thresholds $R_L$ and $R_H$. Second, when the MSWMCP is implemented, relevant units involved will increase social participation through the construction of MSW collection stations and MSW recycling education demonstration bases, and strengthening the role of the media, civil organizations, and volunteers. As a result, the residents will be more likely to participate in MSW classification. The change is expressed by a lower $p_0$ and a higher $p_1$.

Results and discussion

In order to predict the implementation effect of MSWMCP, we set numerical values and perform simulation to visually show the changes in the implementation effect over time. The size of the residential population in urban communities should be contained to between 7000–15,000 people, and the layout should take into account the road network structure (Code of Urban Residential Areas Planning & Design 2002). In order to adequately and reasonably represent the general characteristics of current Chinese cities, it is assumed that the simulation object is a cell with 10,000 inhabitants, the spatial layout is the most common square format, and the residents are evenly distributed in the two-dimensional rectangular space. Because MSW classification incurs costs, before the implementation of MSWMCP, only a small percentage of residents are able and willing to actively participate in MSW classification. It is assumed here that the proportion of such residents is 10%. And the unit of time “t” is the day.

Without the implementation of the MSWMCP

We first simulate the residents’ MSW classification behavior without the implementation of MSWMCP. In this case, the governmental unit or department does not have a strong supervising power over the residents’ MSW classification behavior. It is assumed that if, on average, less than three or more than seven out of the eight neighbors participate in MSW classification, the residents will change their behavior. We thus set$R_L$ to 2.5 and $R_H$to 5. Without the implementation of the MSW compulsory classification policy, the degree of social participation is weak. So the parameter$p_0$is high and$p_1$is low. We set$p_0$to 0.05 and set$p_1$to 0.1. Figure 2 shows the evolution over time of the proportion of residents who participate in classifying MSW when the mandatory policy is not implemented.

Figure 2. The evolution over time of the proportion of residents who participate in classifying MSW when the MSWMCP is not implemented

Figure 3 shows the resident status, absent of the implementation of MSWMCP. The black nodes represent residents who participate in classifying MSW and the white nodes represent other residents. In different periods of evolution, the MSW classification behavior of each resident is not the same, and the spatial distribution is irregular. Figure 2 shows the evolution over time of the proportion of residents who participate in classifying MSW when the mandatory policy is not implemented. As the evolutionary period increases, the proportion of residents who participate in classifying MSW fluctuates only slightly around the initial value of 0.1. The maximum value is 0.1090, and the minimum value is 0.0977. Thus, the proportion of residents who participate in classifying MSW does not change significantly over time. The simulation result shows that the residents will keep their original behavior when there is no mandatory policy.

Figure 3. The residents’ status without the implementation of MSWMCP

The reason for no discernable change in the residents’ waste classification behavior is that MSW mandatory classification is accompanied by external costs, including the hassle and inconvenience of doing the classification. To say the least, residents need to spend resources and time to learn how and what to do with classification and incur costs from changing the original MSW throwing habits. According to Adam Smith’s economic man hypothesis, each individual’s behaviors are motivated by economic incentives, and the individuals behave to maximize their self-interests. Therefore, it is rational for most people not to sacrifice their own personal interests to participate in MSW classification when they are not mandated to do so. As a result, the number of residents who participate in classifying MSW is maintained at the original level and it is difficult to motivate people to change their MSW classification behavior.

With the implementation of the MSWMCP

We simulate the residents’ MSW classification behavior with the implementation of the MSWMCP by changing the evolution parameters. In order to visualize the impact of MSWMCP on the residents’ MSW classification behavior, the MSWMCP is set to be implemented at the time point when t = 100.

Predicting the implementation effect of MSWMCP over time

The parameters underlying our model would start to change following the implementation of MSWMCP. First, the implementation of MSWMCP increases the government’s supervision power over the residents’ MSW classification behavior. This would result in decreasing the value of $R_L$ and $R_H$. Second, the implementation of MSWMCP is expected to gradually improve the atmosphere, attitude, or perception in favor of MSW classification, prompting the residents to be more likely to participate in MSW classification. Consequently, the value of$p_0$ decreases, and the value of$p_1$increases. The resulting specific values of the model parameters are $R_L$ = 2.1, $R_H$ = 4.5, $p_0$ = 0.028, and $p_1$ = 0.122. Figure 4 shows the simulation result.

Figure 4. The evolution over time of the proportion of residents who participate in classifying MSW with the implementation of MSWMCP at t = 100

According to the evolution, the implementation of MSWMCP is divided into two stages. When $0\le \mathrm{t}\le 100$, the MSWMCP is not implemented from t = 0 to t = 100. When $100<t<300$, the MSW compulsory classification policy is implemented. The proportion of residents who participate in classifying MSW is increasing, and the velocity of increase experiences growth first and then decline. This stage is called the first stage after the implementation of MSWMCP. When $\mathrm{t}\ge 300$, the proportion of residents who participate in classifying MSW plateaus, and does not change further beyond that point of time. This is the second stage of the implementation of MSWMCP.

The plateauing of the policy effect can be understood in two aspects. First, a policy loss will eventually occur in the process of public policy implementation. When a policy is implemented (or promulgated) by the central government, the local government is driven by the motive to maximize the local or personal interests (Cao 2016). Thus, the local government will take advantage of information asymmetry and modify the policy formulated by the central government and the government at a higher level to suit their local interests (Cao 2016). This leads to the divergence between the expected social benefit of the original policy intent and the actual social benefit of the public policy (Cao 2016). After the implementation of the MSWMCP, the problem of policy loss will inevitably arise. It will lead to a gradual increase in the policy failure rate. As a result, the effectiveness of the public policy toward the end of the policy implementation process will have a larger gap compared with the policy effect at the medium term of the implementation process. Second, after the MSWMCP has been implemented for some time, most residents would have already adapted their MSW classification behavior according to the policy requirements. If the internal and external conditions remain unchanged, the implementation effect of the MSWMCP will eventually run through its course and stabilizes (stagnates) at a peak.

Predicting the policy effect of MSWMCP with different implementation intensities

To further explore the impact of implementation intensity on the policy effect of MSWMCP, we simulate the implementation effect of MSWMCP with different implementation intensities. In this simulation exercise, the implementation intensity is divided into four levels: low intensity, medium low intensity, medium high intensity, and high intensity. Table 1 shows the simulation parameters of different implementation intensities.

Table 1. Simulation parameters of different implementation intensities

Implementation intensity	$R_L$	$R_H$	$p_0$	$p_1$
Without implementation	2.50	5.00	0.050	0.100
Low intensity	2.20	4.60	0.030	0.120
Medium low intensity	2.15	4.55	0.029	0.121
Medium high intensity	2.10	4.50	0.028	0.122
High intensity	2.05	4.45	0.027	0.123

Figure 5 shows the simulation of mandatory classification policy effect with different implementation intensities. When $t=100$(when the policy is first imposed), regardless of the intensity of the MSWMCP, the residents’ MSW classification behavior all shows improvement, with the greatest improvement of residents’ MSW classification behavior occurring under the high-intensity policy. After some time of evolution, regardless of the intensity of MSWMCP, the implementation effect eventually reaches the same stable state. However, the higher the implementation intensity, the shorter time is needed to reach the stable state, implying that the residents’ MSW classification behavior improves more rapidly under higher implementation intensity.

Figure 5. Simulation of the policy effect of MSWMCP with different implementation intensities

This result can be explained as follows. First, under high-intensity MSWMCP, the government exerts much stronger or stricter enforcing or supervision power on the residents’ MSW classification behavior, and the wider participation of residents further reinforces the residents’ MSW classification behavior. It prompts more residents to be willing to sacrifice their own interests to participate in MSW classification. As a result, the number of residents who participate in MSW classification grows at a faster rate under more intensive enforcement. Second, the implementation of the high-intensity MSWMCP accelerates the compliance of the law. Because of the compulsory nature of the law, when the law is intensively or forcefully enforced, the vast majority of residents would choose to comply with the policy regardless of whether or not they are willing to participate in MSW classification.

Predicting the implementation effect of MSWMCP under different urban population sizes

To explore whether there is a link between urban population size and the implementation effect of MSWMCP, we simulate the policy effect of MSWMCP under different urban population sizes. Assuming medium high implementation intensity, we conduct simulations with five different sizes of residential areas (A, B, C, D, E). Table 2 shows the simulation parameters of different population sizes.

Table 2. Simulation parameters of different population sizes

Area	Population size	Population distribution
A	3 600	60*60
B	6 400	80*80
C	10 000	100*100
D	14 400	120*120
E	19 600	140*140

Figure 6 presents the simulation results of the implementation effect of MSWMCP under different urban population sizes. When MSWMCP is not implemented, the proportion of residents who participate in classifying MSW fluctuates around the initial value of 0.1 in each urban size. It indicates that when the policy was not implemented there is no significant difference in the proportion of residents who participate in classifying MSW across different district sizes. In the first stage of MSWMCP implementation, the situation of residents’ MSW classification significantly improved and there is a slightly higher rising profile of the residents’ MSW classification behavior over population sizes. When $\mathrm{t}=250$, the implementation effect of MSWMCP tends to be stabilized. It means that the implementation of MSWMCP enters the second stage. It also appears that the MSWMCP exhibits a better implementation effect (higher proportion of classification participation) in larger population areas.

Figure 6. Simulation of the implementation effect of MSWMCP under different urban population sizes

The cause for this result is that the change in residents’ MSW classification behavior is influenced and realized through the constraint of the environment. When there are only some isolated residents in the city, since they only know fairly few neighbors, they will not be influenced by these few neighbors to change their MSW classification behavior. On the other hand, when the population size of the area is large, the relationship network of the residents is more complex, and the neighbors exert greater influence on residents’ behavior. Therefore, the isolated or spotty residents tend to have a lower proportion of participation in MSW classification in the area, and the larger population area results in a better implementation effect.

Conclusion

This paper employs the cellular automata model to explore the implementation effect of MSWMCP. By changing the parameter values of the influencing factors of environmental characteristics and individual characteristics, the simulations yield the following results.

First, the residents’ MSW classification behavior does not improve over time without the implementation of MSWMCP. When the MSWMCP is not implemented, the proportion of residents who participate in classifying MSW fluctuates slightly around the original level over time. The evolutionary changes in behavior are so small that is practicably negligible. This underscores the necessity of imposing MSWMCP if changes in residents’ behavior are desired. After the implementation of MSWMCP, the proportion of residents who participate in classifying MSW continues to rise over time and eventually stabilizes at a significantly higher level. It is evident that the residents will not change their MSW classification behavior voluntarily over time, unless being nudged by a mandatory policy.

Second, following the imposition of MSWMCP with different enforcement intensities, the residents’ MSW classification behavior changes over time and can reach more or less the same optimal state. This simulation result reveals the pattern and relationship between the enforcement intensity of MSWMCP and the corresponding implementation effect of the policy. After the policy has been implemented for a certain period, the implementation effect of the policy is fully manifested, and the proportion of residents who participate in classifying MSW reaches the same stable state. In other words, when the policy effect is fully realized, the proportion of residents who participate in classifying MSW does not differ by implementation intensity of the policy. However, the lower the implementation intensity of the policy, the longer it takes to achieve the full implementation effect. This suggests that if the time horizon of policy implementation is not sufficiently long to fully demonstrate its role, the low-intensity policy is likely to face the risk of decline in effectiveness. And if the policymakers want to improve the residents’ MSW classification behavior in a short time period, then a high-intensity MSWMCP will need to be adopted.

Third, after the implementation of MSWMCP, there may be differences in the evolution over time of the residents’ MSW classification behavior in areas of different population sizes. This simulation exercise reveals that the population size has an impact on the implementation effect of MSWMCP. The network of residents’ relationships is often more complex in a larger population area. So the residents’ MSW classification behavior is more susceptible to be influenced by others’ behavior in a larger population area. As a result, the MSWMCP achieves a better implementation effect in the area with a larger population.

Although this study reveals some general rules about the implementation effect of MSWMCP, further studies are needed. Because it is costly and difficult to conduct long-term follow-up investigations on a large number of residents, the model of assessing and predicting the implementation effect of MSWMCP needs to include more empirical research on residents’ behavior in the future.

Additional information

Funding

This paper stems from research projects supported by the Major Project of Philosophy and Social Sciences Research, Ministry of Education [grant number 17JZD026]; the Leading Research Project of Shanghai Jiao Tong University—Think Tanks [grant number ZXYJ-2020017]; the National Nature Science Foundation of China (NSFC) Project [grant number 71473056]; the Fundamental Research Funds for the Central Universities [grant number HEUCFP201823 and GK2090260158]; and the Ph.D. Student Research and Innovation Fund of the Fundamental Research Funds for the Central Universities [grant number HEUGIP201719].

Notes on contributors

Yanxin Li

Yanxin Li, School of Marxism, Tongji University, Shanghai, People’s Republic of China.

Zhujie Chu

Zhujie Chu, The School of International and Public Affairs, Shanghai Jiao Tong University, Shanghai, China; China Institute for Urban Governance, Shanghai Jiao Tong University, Shanghai, People’s Republic of China

Hongda Zhang

Hongda Zhang, Faculty of Information Technology, Beijing University of Technology, Beijing, People’s Republic of China

Wei-Chiao Huang

Wei-Chiao Huang, Department of Economics, Western Michigan University, Kalamazoo, Michigan, USA.

Feiren Liu

Feiren Liu, Jiangxi Province Development and Reform Research Institute, Nanchang Jiangxi, People’s Republic of China.