Analysis of the Potential Economic Impact of Parking Space Comprehensive Utilization on Traditional Business District

Abstract

This paper investigates the latent classes of parking preference for drivers and the economic effects after implementing Parking Space Comprehensive Utilization (PSC) in traditional business districts (TBD), with a particular focus on the parking preferences of electric vehicle users (EVU). Firstly, Exploratory Factor Analysis (EFA) is used to reduce dimensionality and determine the latent structure. Then, based on the Latent Class Model (LCM), the customers are classified, and the proportion of each class under various latent variables is analyzed. Finally, the paper conducts a quantitative analysis of economic effects by considering different psychological factors across different customer classes. With the data obtained from revealed preference (RP) and stated preference (SP) surveys, this paper identifies the customers’ preferences for the three scenarios presented. The results show that (1) customers can be classified into four classes: core customers (CCS, 34%), potential customers (PCS, 29%), regular customers (RCS, 22%), and marginal customers (MCS, 15%), among which EVU do not show a significant preference for parking charging facilities in TBD; (2) the potential economic improvements for these four classes are: 9%, 12%, 8%, and 10%; (3) CCS has the greatest potential to increase store revenue by ¥7041, while PCS has the greatest potential to increase store customer flow by 31%. These findings provide a valuable reference for decision-making by TBD store managers.

1. Introduction

Business districts (BD), especially traditional business districts (TBD), are plagued by serious parking issues [1,2], which have a detrimental impact on the sustainable development of TBD merchants. To address this crisis, the concept of Parking Space Comprehensive Utilization (PSC) has been proposed. Consequently, it is necessary to investigate the sustainable economic benefits that PSC can bring to merchants. It can be assumed that visitors with different characteristics exhibit distinct commercial behavioral patterns. This study intends to classify and analyze these behavioral traits, aiming to identify the sustainable economy they may generate for TBD.

Based on the aforementioned considerations, shared parking has become a hot topic in the research on parking difficulties due to its advantages of low cost and high utilization rate. It holds the potential to profit from the business model of the sharing economy [3,4]. Shared parking refers to the efficient utilization of scattered idle parking resources. For example, based on the parking characteristics of residents in residential areas [5], online real-time supply–demand matching can yield higher profits [6]. Its essence is the integration and reuse of discrete parking resources to generate profits. In reality, this sharing model can be extended to BD and is better suited for the business models of the sharing economy era. It may bring sustainable profits to some declining stores by combining shared parking with the marketing models of BD shops. This is also the main focus of the present study.

Based on the characteristics of shared parking, which involve discrete resources and reuse, it provides a perspective for integrating roadside parking in TBD into a unified shared parking system. This model is referred to in this study as ‘PSC’ (Figure 1). PSC differs from previous research in three key ways. Firstly, in homogeneous parking spaces, all parking spaces are of the same type and are located within the same land-use category, such as the BD. Secondly, PSC utilizes consistent parking pricing established by the platform to prevent malevolent bidding. Thirdly, endogenous parking fees. PSC converts parking fees to expenditure, elevating the parking experience and increasing value-added expenditure. In contrast, prior shared parking models ignored this strategy of translating parking fees into expenditure by using commercial incentives policies. Once the scattered on-street parking is unified, the parking supply will increase in a limited space. As a result, the management issues associated with on-street parking [7] can be effectively addressed, and the customer parking experience can be improved. In addition, shops can also benefit from shared parking if they enter into a cooperative agreement.

On the other hand, this study focuses on customers who drive to TBD and who are the primary consumer force in offline TBD. To ensure that these customers contribute to TBD’s revenue, it is essential for them to have convenient access to parking spots. Alternatively, if shops provide parking spots at a distant location, the desire for consumption may vary among customers, either increasing or decreasing. Therefore, when exploring whether customers’ consumption levels are influenced by parking issues, it is crucial to consider their heterogeneity. Previous scholars have examined heterogeneity from various aspects, including individual economic attributes [8], psychological attributes [9], and travel characteristics [10]. Building upon previous research and considering the characteristics of TBD, this study introduces additional factors such as commercial preference (CP), preference for activities (PA), psychological factors related to hidden costs (HCP), conformity (CE), consumption habits, car types, and household attributes, aiming to conduct a more detailed analysis of latent customer classes.

Based on the aforementioned analysis of latent customer classes, this study specifically explores the potential consumption and parking characteristics of electric vehicle users (EVU). Considering the global commitment to the Paris Agreement, electric vehicles (EVs) are expected to dominate the market within the next 20 years [11]. However, due to the issue of limited driving range, customers’ willingness to purchase EV remains relatively low. Consequently, many scholars have been eager to research methods to increase customers’ willingness to buy EVs [12,13]. Nevertheless, there is currently a lack of research on the level of EV adoption and the differences in parking demands between EVU and Not-EVU. This study fills this gap and provides a foundation for future studies on the value of charging facilities for EVU.

Therefore, returning to the initial research objectives, based on parking considerations, the question of whether customers can bring sustainable benefits to non-target shops within the TBD is divided into the following three questions:

• Question 1: Based on the survey data, what are the different customer classes, and what are the heterogeneities among EVU?
• Question 2: For the different customer classes identified above, to what extent can the combination of PSC and store marketing attract customer consumption?
• Question 3: How much can the PSC within the TBD enhance the economic vitality of surrounding shops?

The remaining sections of this paper are organized as follows. Section 2 summarizes the existing research on the sharing economy and behavior in shared parking. Section 3 designs the research scenarios and constructs the latent class model. Section 4 presents the survey results based on questionnaire data and classifies customers using the Latent Class Model (LCM). Section 5 provides the research conclusions, limitations, and policy recommendations for shop managers in TBD.

2. Literature Review

Shared parking, as a form of the sharing economy, makes use of idle resources by facilitating the matching of supply and demand. Therefore, the study of the shared parking economy and the behaviors of both the supply and demand sides form the theoretical basis of this study.

2.1. BD Parking Issues

The essence of the parking problem in BD is the imbalance between parking supply and demand. Therefore, the focus of research is on reducing demand or increasing supply. On the supply side, one approach is to construct a new parking facility in the vicinity of the BD. However, the construction of a new parking facility needs to consider the parking demand and its distribution as they determine the scale and location of the parking facility. For instance, [14] conducted a three-day survey of on-street parking demand using license plate data, while [15] considered multiple objectives to determine optimal candidate locations for constructing off-street parking facilities. However, considering the limited available land and high costs associated with constructing new facilities in BD, the research focus has shifted towards suppressing parking demand. Some scholars have explored measures such as road pricing and parking pricing [16], promoting private car diversion to public transportation [17], and implementing park-and-ride [18], which have partially alleviated parking pressure. However, these measures may reduce the attractiveness of BD in terms of customer traffic or increase the time spent searching for parking since customers tend to prefer parking spots with lower prices and shorter walking distances. This raises another research topic: how to improve parking search efficiency [19,20]. Moreover, numerous uncertainties make it challenging to determine parking demand. For example, hourly parking fees can be adjusted flexibly or demand-induced based on the duration of parking [16]. However, both supply side and demand-side measures have room for improvement. Consequently, the concept of shared parking emerged.

2.2. Shared Parking Economy

Existing research has demonstrated that shared parking can effectively reduce costs and increase revenues [21], and there is potential for a shared EV business model in the future [22].

The key to profitability in shared parking is the optimal utilization of idle resources, which involves attracting more users using network effects and the matching of supply and demand. Therefore, the stronger the willingness of participants to engage in shared parking and the larger the number of participants, the greater the economic benefits that can be generated. Thus, scholars have studied both the willingness to participate in shared parking and the design of supply–demand matching. On the one hand, there are efforts to enhance the willingness to participate. [23] measured the willingness of 1008 residents in New South Wales, Australia, to pay for shared parking, while [24] combined the Technology Acceptance Model (TAM) with the satisfaction-loyalty model to analyze the key factors influencing behavioral intentions in shared parking modes. The authors found that perceived risk (PR) is a critical variable leading to changes in parking behavior. However, research in this area is not yet mature. On the other hand, there are endeavors to design optimal supply–demand matching mechanisms to improve the success rate of matching. [25,26] designed a novel double auction mechanism to maximize social welfare, while [27] developed an allocation mechanism for shared parking spaces and an adjustment method for auctioning parking permits. Both studies have confirmed the effectiveness of the proposed mechanisms. In addition to the aforementioned research on the shared parking economy, the combination of parking and EV sharing has become a hot topic. Previous studies have examined alternative policies to incentive schemes, such as consumer preferences and the gradual phasing out of subsidies [28], as well as the dispatch optimization of EV sharing [29].

It can be observed that EVs will dominate the market in the future, highlighting the significance of studying EVU. Currently, there is a gap in understanding the value of charging facilities for EVU. Additionally, it remains to be investigated whether their parking preferences differ from those of Not-EVU.

2.3. Shared Parking Behavior

Shared parking has been studied by many scholars, from initial parking intentions to actual parking behaviors. However, there is a gap in the analysis of the economic impacts following shared parking behaviors. The study of shared parking behaviors is a progressive process, starting from various studies on shared parking behaviors, considering the differences between different providers and users, and eventually exploring the heterogeneity of each participant. In terms of various shared parking behaviors, scholars have investigated reservation behaviors in shared parking [30] (temporary reservation behaviors [31]), parking choice behaviors [4], parking search behaviors [20], and parking violation behaviors [32]. They have found that individual attributes (such as gender, age, monthly income), parking fees, pricing strategies, and walking distance play important roles in behavioral decision-making. Additionally, the influence of bounded rationality on decision outcomes should not be overlooked. Therefore, relevant scholars have explained the behavioral motivations from the perspectives of Cumulative Prospect Theory [8], Anticipated Regret [33], and other theories. Regarding supply–demand matching, there are significant differences in behaviors between providers and users. For instance, Liang [34] used a combined TAM and theory of planned behavior (C-TAM-TPB) as a theoretical framework to study shared parking. The research revealed that perceived control and self-efficacy are the most influential factors for parking space seekers, while the influence of significant others is the greatest for parking providers. In terms of individual heterogeneity research, a common approach is to combine revealed preference (RP) and stated preference (SP) surveys [35,36] to aggregate and analyze individual behaviors and identify a category of individuals with similar attributes for further analysis. In this process, LCM [37,38] provides a research approach that can classify latent classes and analyze each individual within them.

Based on the existing research on shared parking behaviors, this study identifies areas for improvement. Firstly, when considering participation in shared parking, the focus has been solely on shared parking itself, without considering other economic incentive mechanisms or conducting specific quantitative research. For example, combining shared parking with parking coupons at shops, how much can shared parking behavior be influenced, and whether can it enhance economic vitality? Secondly, there is a gap in research on the shared parking behaviors of co-residing household members when studying individual heterogeneity.

3. Materials and Methods

The scenario studied in this paper is a TBD. It is characterized by a long-established street with small shops arranged along it. In the face of the overall decline of the TBD, only a few shops are able to retain customers, while the rest are facing a survival crisis. Thus, the parking spots of the shops have reached a cooperative agreement and agreed to share parking, establishing a unified parking fee standard for the entire TBD. The shops also offer additional coupons to customers who park their cars at other shops. This business model is referred to as “PSC” in this paper.

Under this research background, three key scenarios that influence the economic vitality of the TBD are designed:

(1)

To Leave or Not to Leave: When the shop you want to visit has no available parking spots, would you give up on shopping?

(2)

To Come or Not to Come: Different shared parking spots are provided at varying coupons, with corresponding compensation incentives. Which car park would you choose based on its distance and the offered incentives?

(3)

To Visit or Not to Visit: After parking your car, would you consider temporarily visiting nearby shops?

3.1. Experiment Design

The three designed scenarios are illustrated in Figure 2 as follows: (a) Before Parking, (b) Parking Choice, and (c) After Parking.

(1)

Scenario 1: Before PSC—Leave or Not

In this scenario, the purpose is to investigate the economic loss that shops may suffer due to parking issues. The paper provides five options, as shown in Figure 2a. Route 1 is to not visit the TBD at all. Route 2 is to postpone the visit to the shop until there is available parking. Route 3 is to switch to another shop with available parking. Route 4 is to find a parking space farther away and then proceed to the intended shop. Route 5 is to change the mode of transportation. These five routes can be classified into two groups based on whether the customer leaves the intention to L store.

(2)

Scenario 2: After PSC—Come or Not

In this scenario, the purpose is to explore which type of parking space customers prefer when different parking options are provided at varying distances. The paper designs six choices, as shown in Figure 2b, which is just one example. Considering the diverse forms of incentives and to ensure independence between different combinations, the paper designs 16 incentive scenarios, as outlined in

Table 1. China TBD parking study SP design specifications.

Attribute	800 m	500 m	300 m	200 m	100 m	Not Travel
Date	2 levels	2 levels	2 levels	2 levels	2 levels
Forms of discounts	4 levels	4 levels	4 levels	4 levels	4 levels
Degree of discounts	2 levels	2 levels	2 levels	2 levels	2 levels

. The scenarios cover various factors, including parking coupons (DP), parking charging discounts (DC), consumption coupons (DB), and free parking with maximum purchase (DF). The incentive levels are set differently for holidays and weekdays. For instance, for a parking spot located 500 m away, customers are provided with two levels of parking fee coupons (70% and 60%) to choose from under the given incentive scenario.

(3)

Scenario 3: After PSC—Visit or Not

In this scenario, the aim is to investigate whether customers who have parked at H store and are on their way to L store would consider making a temporary visit to H store or nearby shops. The paper intends to explore if customers with flexible parking time can be enticed to increase their expenditure via shop incentives and the influence of social conformity. The paper provides four options, as shown in Figure 2c. Category 1 is choosing to visit a nearby H store, Category 2 is choosing their preferred shop, Category 3 is choosing popular shops, and Category 4 is choosing randomly. To measure the influence of social conformity, a 5-point Likert scale is used (strongly disagree, disagree, neutral, agree, strongly agree) to explore the impact of five factors on customers’ conformity behavior: friends, relatives, onlookers, media, and experiences.

3.2. Data Survey

The paper investigates individual socioeconomic attributes, household attributes, parking and travel attributes, consumer habits, and psychological factors. Additionally, it includes measures for conformity effect (CE), perceived risk (PR), hidden cost psychology (HCP), commercial preference (CP), and preference for activities (PA). The economic and social attribute variables surveyed are outlined in

Table 2. Economic and social attribute survey.

Economic and Social Attribute
Items	Options	Items	Options
1. Gender	A: Male	8. Hometown	Blank
	B: Female	9. Household members	A: 1 Person
2. Age	A: 18~24		B: 2 Person
	B: 25~34		C: 3 Person
	C: 35~44		D: ≥4 Person
	D: 45~54	10. Number of cars	A: 0
	E: 55~64		B: 1 Car
	F: 65+		C: 2 Cars
3. Residence	Blank		D: 3 Cars
4. Occupation	A: Full-time		E: ≥4 Cars
	B: Part-time	11. Elderly	A: Yes
	C: Freelance		B: No
	D: Teacher	Children	A: Yes
	E: Student		B: No
	F: Civil servant	Females	A: Yes
	G: Retired		B: No
	H: Others	Pets	A: Yes
5. Education	A: Middle school or below		B: No
	B: High school	Disabled	A: Yes
	C: College/Undergraduate		B: No
	D: Master	12. Annual household income/yuan	A: <10,000
	E: Doctor or above		B: 10,000–100,000
6. Marital status	A: Single		C: 100,001–250,000
	B: Unmarried		D: 250,001–500,000
	C: Married		E: >500,000
	D: Others	13. Parking spot	A: Yes
7. Monthly income/yuan	A: <3000		B: No
	B: 3000–5000	14. Car type	A: Fuel-powered car
	C: 5001–10,000		B: Electric car
	D: 10,001–20,000		C: Hybrid car
	E: >20,000		D: Fuel cell car
			E: No car

. Furthermore, the paper emphasizes the analysis of the HCP associated with parking fees on customers’ shopping behavior. Typically, parking fees are charged based on the duration of parking, with fees rounded up to the nearest hour. Therefore, when customers are shopping, whether the parking fee within this flexible time window would influence customers to expedite their shopping process is examined. For example, if a customer’s parking time is approaching 2 h, would they consider completing their shopping within 2 h to avoid paying for an additional hour of parking? If such psychological factors exist, they can be leveraged inversely. For instance, if customers are provided with 3 h of free parking, would they extend their shopping time from their initial plan of 2 h? This would play a significant role in enhancing the overall economic vitality of the TBD. Latent variables that cannot be directly observed are assessed using multiple observable variables, with measurement items outlined in

Table 3. Latent variable survey.

Latent Variables	Measurement Items	Options
1. CE	CE-1: I would go to H shop if my friends or classmates want to go there.	1: Strongly disagree
	CE-2: I would go to H shop if my relatives want to go there.	2: Disagree
	CE-3: I would go to H shop if I see a lot of people gathered there.	3: Neutral
	CE-4: I would go to H shop if I read online recommendations for it.	4: Agree
	CE-5: I would go to H shop if I have frequented it before.	5: Strongly agree
2. PR	PR-1: I often worry about not finding parking spot.	1: Strongly disagree
	PR-2: I often worry about difficulty in finding parking spots.	2: Disagree
	PR-3: I often worry about high parking fees.	3: Neutral
	PR-4: I often worry about inconvenient parking access.	4: Agree
	PR-5: I often worry about parking security.	5: Strongly agree
3. HCP	HCP-1: When parking time is close to 1 h, I try to leave within that hour.	1: Strongly disagree
	HCP-2: When parking time is already over 1 h, I leave before it reaches 2 h.	2: Disagree
	HCP-3: When parking time is close to 2 h, I try to leave within that time.	3: Neutral
	HCP-4: When parking time is already over 2 h, I leave before it reaches 3 h.	4: Agree
	HCP-5: When parking is close to 3 h, I try to leave within that time.	5: Strongly agree
	HCP-6: When parking time is already over 3 h, I leave before it reaches 4 h.
4. CP	CP-1: Small to medium-sized business district.	Rating: 0–5
	CP-2: Large-scale business district.
	CP-3: Street-style business district.
	CP-4: Integrated business district.
	CP-5: Historical brand business district.
	CP-6: Emerging business district.
5. PA	PA-1 (CL): Clothing stores (clothing, jewelry...).	Rating: 0–5
	PA-2 (RT): Retail stores (supermarket, hardware...).
	PA-3 (RS): Restaurant stores (restaurant, bubble tea...).
	PA-4 (EN): Entertainment stores (cinema, KTV...).
	PA-5 (TR): Training stores (dance, art...).
	PA-6 (HE): Health stores (foot bath, skincare...).
	PA-7 (ST): Strolling on the street (rest, leisure...).

.

3.3. Methodology

The research methodology and process of this study are illustrated in Figure 3. Firstly, exploratory factor analysis (EFA) was conducted to reduce dimensionality and identify the latent structure, thus making the variables interpretable. Subsequently, based on the key factors obtained in the first step, customers were classified using LCM, and the effects of covariates on customer types were explored. Finally, using the customer classes in the second step, the study investigated which type of customers could generate higher profits for the store based on the principle of CE, as well as how to encourage customers to spend.

(1)

EFA

EFA is a statistical method used to explore and uncover the latent factor structure behind observed variables. By analyzing the common variance among observed variables, it aims to identify a few latent factors that best explain the variability in the data.

The steps and computational formulas for conducting EFA in this study are as follows:

$$g\left(T\right)=trZX$$

(1)

$$f\left(T\right)=tr{\left(Y-Z\right)}^{\prime }\left(Y-Z\right)$$

(2)

Firstly, data cleaning was performed, and the variables were standardized to obtain an m × n matrix X. Then, Principal Component Analysis (PCA) was used to extract factors. The covariance matrix or correlation matrix was computed, and eigenvalue decomposition was applied to it. Based on the magnitude of the eigenvalues, the number of factors to be extracted was determined. Subsequently, Varimax orthogonal rotation was applied to the extracted factors to enhance their interpretability.

(2)

LCM

LCM is a model used to explore the presence of latent classes in the data. It is an unsupervised learning method that does not require prior knowledge of class labels. Instead, it infers the latent classes based on the characteristics of the data itself.

The process of LCM in this study is as follows:

Step 1: Establish the base model. If there are three observed variables (A, B, and C), the LCM should include a non-conditional probability (unconditional probability) that represents the latent classes, denoted as ${\pi }_t^X$. Additionally, three conditional probabilities (conditional probability) representing the contribution of each observed variable to the latent class structure are included, forming the basic LCM. As the latent classes are assumed to be mutually independent, the sum of the conditional probabilities for each observed variable at each level (i, j, k) is equal to 1. The formula is as follows:

$${\pi }_{ijk}^{ABC}={\displaystyle \sum _{t=1}^T{\pi }_t^X{\pi }_{it}^{\overline{A}X}{\pi }_{jt}^{\overline{B}X}{\pi }_{kt}^{\overline{C}X}}$$

(3)

$${\displaystyle \sum _i{\pi }_{it}^{\overline{A}X}}={\displaystyle \sum _j{\pi }_{jt}^{\overline{B}X}}={\displaystyle \sum _k{\pi }_{kt}^{\overline{C}X}}=1$$

(4)

Step 2: Parameter estimation. Maximum likelihood estimation was used to estimate the probabilities of observed variables under each latent class. By summing up the estimated probabilities for each observed variable and each latent class, the maximum likelihood estimates of the joint probabilities can be obtained. The formula is as follows:

$${\pi }_{ijkt}^{ABCX}={\pi }_t^X{\pi }_{it}^{\overline{A}X}{\pi }_{jt}^{\overline{B}X}{\pi }_{kt}^{\overline{C}X}$$

(5)

Step 3: Evaluation of classification results. By dividing ${\pi }_{ijkt}^{ABCX}$ by ${\pi }_{ijk}^{ABC}$, the maximum likelihood probabilities of different observed variable levels in each latent class can be obtained. Finally, the posterior probabilities of each sample belonging to different latent classes are computed, and the class with the highest probability is assigned as the class to which the observation belongs. The final classification number is determined based on the CAIC and BIC criteria. The formula is as follows:

$${\pi }_{ijkt}^{ABC\overline{X}}=\frac{{\pi }_{ijkt}^{ABCX}}{{\pi }_{ijk}^{ABC}}$$

(6)

$${\pi }_{tijk}^{\overline{X}ABC}=\frac{{\pi }_{ijkt}^{ABCX}}{{\displaystyle \sum _{t=1}^T{\pi }_{ijkt}^{ABCX}}}$$

(7)

(3)

CE

CE refers to the psychological phenomenon where individuals tend to conform to the majority opinion or behavior of a social group. This phenomenon can lead to conformity behavior, where individuals choose to align with the majority in order to meet social expectations or avoid exclusion.

The quantification equation for the CE is introduced [39] to quantify the influence of customers’ conformity psychology on behavioral decisions. Here, the $\mu$ represents the importance level, indicating that a stronger CE corresponds to a larger value. The $\xi$ represents the conformity parameter, which depends on the number of individuals adopting a particular strategy. The $E_i$ and $E_i{}^{\prime }$ denote the expectations without considering the CE and the adjusted expectations considering the CE, respectively. The formula is as follows:

$$E_i{}^{\prime }=\left(1-\mu \right)E_i+\mu \xi$$

(8)

4. Results and Discussion

An online survey was conducted using Question Star starting 16 May 2023 and lasted for one week. A total of 522 responses were collected, and 483 of them were deemed valid. The number of valid questionnaires answered is 477. The valid response rate was 98.8%.

4.1. Validity Check of the Questionnaire

(1)

Reliability Test

Reliability testing of the questionnaire is conducted to assess the consistency and stability of the individual questions or items within the measurement tool. A higher reliability index, such as Cronbach’s alpha greater than 0.7, is considered acceptable. In this study, Cronbach’s alpha coefficient was used to test the reliability of the questionnaire, and the overall Cronbach’s alpha coefficient was found to be 0.824, indicating good internal consistency among the variables.

(2)

Validity Test

Validity testing of the questionnaire is performed to evaluate the effectiveness and accuracy of the measurement tool in accurately reflecting the intended concepts or constructs. Common prerequisite tests include the Kaiser-Meyer-Olkin (KMO) measure (>0.7) and the Bartlett’s test of sphericity (<0.05). In this study, the KMO value of the questionnaire was 0.791, and Bartlett’s test of sphericity yielded a value of less than 0.001, indicating the presence of intercorrelations among the variables and suitability for factor analysis.

(3)

Factor Analysis

Based on Table 3, five factors were predetermined. Using exploratory factor analysis, it was found that the first and third factors (CE, HCP) could be further subdivided into two factors each. Additionally, the first item (PA-1) of the fifth factor (PA) was found to interfere with the questionnaire’s validity. Therefore, after eliminating an ineffective item, the remaining 28 items were divided into seven factors. The scree plot (Figure 4) and the rotated component matrix (

Table 4. Rotated factor matrix.

Factor	1: CP	2: PAA	3: PR	4: HCPB	5: HCPA	6: CEA	7: CEB
CP-4	0.755
CP-2	0.720
CP-6	0.715
CP-3	0.687
CP-5	0.677
CP-1	0.657
PA-5		0.731
PA-6		0.719
PA-4		0.689
PA-7		0.614
PA-2		0.568
PA-3		0.531
PR-2			0.744
PR-4			0.739
PR-1			0.727
PR-3			0.577
PR-5			0.575
HCP-4				0.896
HCP-6				0.812
HCP-2				0.725
HCP-1					0.811
HCP-3					0.807
HCP-5					0.672
CE-2						0.724
CE-5						0.675
CE-1						0.622
CE-3							0.797
CE-4							0.771

) are shown below.

4.2. Descriptive Statistics of the Questionnaire

Descriptive statistical analysis was conducted on the 483 respondents (by SPSS), primarily categorized into the following five categories: scene selection statistics, parking attribute statistics, consumption level statistics, personal household economic statistics, and the sensitivity of the parking prices and walking distances.

(1)

Scene Selection Statistics (Figure 5)

In total, 88% of the respondents choose to go to BD during holidays.

In scenario one, 26% of the respondents (12% + 14%) would leave the store due to a lack of parking spaces, resulting in losses.

In scenario two, By offering coupons, it is possible to effectively attract drivers to park in this area. Specifically, using BD coupons and parking coupons, 42% and 40% of drivers can be attracted to park at a distant distance, respectively. Additionally, using parking charging coupons and free parking duration with maximum purchase, 49% and 32% of drivers can be attracted to park at a far distance. Furthermore, less than 4% of drivers would not come to park here.

In scenario three, 35% of drivers would visit H stores, indicating the potential impact of parking on surrounding businesses. Additionally, the respondents’ preference for store types is primarily focused on restaurant and leisure stores.

(2)

Parking Attribute Statistics (Figure 6)

There is a significant difference in transportation mode choice between purposeless trips and purposeful trips. Taking private cars as an example, the proportion of private car usage for purposeless trips is 31%, but it increases to 70% for purposeful trips, which is nearly twice as much. Additionally, more than half of the respondents (59%) have had parking reservations, and over half (52%) of them enjoy parking coupons occasionally. The majority (66%) prefer to be charged based on the parking duration.

(3)

Consumption Level Statistics (Figure 7 and Figure 8)

The respondents have a relatively high average consumption level. They typically spend around 20 min (42%) to 30 min (31%) traveling to BD. They usually travel in groups of 2 people (42%) to 3 people (36%) together. The average duration of their stay at BD is typically 2 h (44%) to 3 h (37%). The average spending per visit ranges from 401–500 yuan (23%). Furthermore, they visit BD a time (37%) to 2 times per week (35%). The majority of respondents have 2 BD locations (41%) to 3 BD locations (33%) near their residential areas. Their preferred departure times for BD are in the afternoon, with 2:00 PM and 6:00 PM being the top choices, followed by 8:00 AM and 12:00 PM. Most respondents consider 6:00 AM and 10:00 PM as the least likely times to visit BD.

(4)

Personal and Household Economic Statistics (Figure 9)

The overall economic attributes of the respondents are relatively concentrated, with the majority falling within the age range of 25–44 (83%). The gender ratio is balanced, with 52% being male and 48% being female. Most of them come from Beijing (48%) and Jiangxi (22%), hold a bachelor’s degree (77%), and work full-time (75%). The average monthly income is concentrated between 5000 and 20,000 yuan (71%). The majority (69%) own private parking spots and the majority of their cars are traditional fuel vehicles (51%), followed by pure electric vehicles (23%) and hybrid vehicles (20%). The majority have three or more family members (78%), including children (65%) and women (92%), but fewer elderly individuals (40%), disabled individuals (2%), and pets (46%). Most households have one car (59%). Their annual family income is mostly between ¥250,000 and ¥500,000, accounting for 38%.

(5)

Sensitivity of Parking Prices and Walking Distance.

The respondents’ minimum price (MIP), optimal price (OPP), acceptable price (ACP), and maximum price (MAP) were measured using the Price Sensitivity Measurement (PSM) model, as well as their minimum walking distance (MIW), optimal walking distance (OPW), acceptable walking distance (ACW), and maximum walking distance (MAW). Four PSM curves are depicted in Figure 10. The respondents’ MIP, OPP, ACP, and MAP are ¥9, ¥10.5, ¥11, and ¥12, respectively. The respondents’ MIW, OPW, ACW, and MAW are 325 m, 425 m, 450 m, and 525 m, respectively. However, for the respondents, the MIP and MIW should be zero. Therefore, this study does not utilize the minimum values but instead employs the subsequent three values.

4.3. Customer Class Results

(1)

Classification Quantity and Proportions

Based on the seven factors obtained from the aforementioned EFA and the non-scale questions in the survey, a correlation analysis was conducted to identify factors with correlation coefficients >0.5. Using individual household, parking attributes, and psychological variables as categorical variables, LCM was performed in this study (By LatentGOLD). The evaluation criteria for determining the number of classes were the lowest CAIC and BIC values. It is recommended to classify the respondents into four classes: Core Customers (CCS, 34%), Potential Customers (PCS, 29%), Regular Customers (RCS, 22%), and Marginal Customers (MCS, 15%). (The specific meanings will be explained in the next section) (

Table 5. Table of 6 model fit indices.

Class	CAIC (LL)	BIC (LL)	Npar	L2	Df	p-Value	Class. Err.
1-Cluster	55,467	55,291	176	48,298	303	0.000	0.00
2-Cluster	54,874	54,646	228	47,332	251	0.000	0.04
3-Cluster	54,693	54,413	280	46,778	199	0.000	0.05
4-Cluster	54,654	54,322	332	46,366	147	0.000	0.07
5-Cluster	54,714	54,330	384	46,053	95	0.000	0.06
6-Cluster	54,825	54,382	436	45,784	43	0.000	0.06

).

After classifying the respondents into four classes, the corresponding proportions of each class for Scenario 2 and Scenario 3 are shown in Figure 11 and Figure 12, respectively.

In Scenario 2, there are significant differences among the four classes in their choices between distant walking distances and near distances, while the differences in choices for other walking distances are not significant. PCS prefers distant walking distances in exchange for greater coupons, while CCS, on the other hand, prefers near walking distances. However, there is no significant difference between RCS and MCS. Both classes show a preference for using consumption coupons and parking coupons to attract them to park at locations within walking distance.

In Scenario 3, there are significant differences among the four classes in their choices between temporarily going to shops near the parking spot and going directly to their destinations, while other aspects show no clear differences. CCS and PCS have a higher proportion of choosing to visit shops near the parking spot, with CCS accounting for nearly half of them (41%). In contrast, RCS and MCS prefer to go directly to their destinations.

Therefore, PCS is the class that is most likely to bring revenue to the TBD. When they drive to the TBD and find no parking spots, they are generally more willing to park at a distant location by obtaining coupons, and they are also more willing to visit surrounding shops near the parking spot. This not only avoids losses for the shops but also generates additional revenue for the TBD as a whole.

(2)

Characteristics of the Four Classes (Figure 13)

The characteristics of the four classes are analyzed based on the outputs of LCM:

• Class 1: CCS. This class primarily consists of young to middle-aged males, mostly from Beijing, engaged in full-time jobs with high monthly incomes. They are married with children, have experience with parking reservations, and own parking spaces. Their households own 1–2 cars, and they have a high annual family income. Additionally, their consumption level mostly exceeds ¥400, and they have a higher CEB. Relatively, they prefer parking spots within a close walking distance.
• Class 2: PCS. This class mainly comprises young males, primarily from non-Beijing areas, with a college/university education and low monthly income. They have experience with parking reservations and own parking spots. Their households own 1–2 cars, and they have a low annual family income. Furthermore, their consumption level mostly ranges from ¥200 to ¥500, and they have a higher CEB, HCP, and PR. Relatively, they prefer parking spots with greater coupons.
• Class 3: RCS. This class is primarily composed of young to middle-aged females from Beijing. Most of them are married and have high monthly incomes, but they lack experience with parking reservations and do not own parking spots. Their households own one car, and they have a high annual family income. However, most of these households do not have elderly or pets. Moreover, their consumption level mostly ranges from ¥100 to ¥500, and they exhibit significant PAA. Relatively, they prefer parking spots near their destination stores.
• Class 4: MCS. This class primarily consists of young females from non-Beijing areas, including students, civil servants, and retirees, with a master’s degree. They are single with low monthly income, and have no experience with parking reservations or parking spaces. Most of their households own one car, and they have a low annual family income. Additionally, their consumption level mostly ranges from ¥100 to ¥400, and they have a higher HCPA. Relatively, they prefer parking spots with longer parking coupons.

(3)

EVU (Figure 14)

CCS (46%) and RCS (33%) are the primary users among EVUs. They mainly consist of young and middle-aged women from Beijing who work full-time and have a college/university education. They are married with children and have a high monthly income. Additionally, they have a higher CE, HCPA, and PR. Comparing the data from the overall survey respondents to those who are Not-EVUs, it is observed that the BD coupons attract more EVUs to park at distant distances (8%). Furthermore, parking and charging incentives attract more EVU (4%) to park at far distances. It can be inferred that EVU and Not-EVU have different preferences for parking and charging facilities, although the difference is not significant. This might be due to the smaller sample size of EVUs (23%) or the relatively short duration of visits to TBD (81% within 2–3 h).

4.4. Covariate Analysis

This section mainly analyzes the respondents who did not generate revenue by leaving the store in Scenario 1, accounting for 26% of the total sample. A covariant analysis is conducted using three significant psychological variables: CE, HCP, and PA.

Among the 124 respondents who left the store, the proportions of the four identified classes are as follows: 35%, 31%, 19%, and 15%. The comparison between the leaving and staying respondents is illustrated in Figure 15. Relative to the staying respondents, those who left the store showed a higher preference for restaurant and retail stores. On the other hand, the staying respondents often had higher expectations for their intended destinations. Among the staying respondents, there was a notably higher proportion of RCS, which aligns with their significant PA.

(1)

CE (Figure 16)

Overall, the four classes exhibit relatively high CE, averaging around 3.9 out of 5. When ranked from highest to lowest, the order is as follows: PCS, CCS, RCS, and MCS. Based on the four cumulative probability distribution charts mentioned above, using a cumulative probability of 0.7 as a dividing line, it can be observed that all four classes have high CE-2 and CE-5 ratings but low CE-3 ratings. In relative terms, CCS and RCS show more dispersed ratings in CE, while PCS and MCS are relatively concentrated. Among them, CCS and RCS have lower CE-3 and CE-4 ratings, while PCS demonstrates a more concentrated CE distribution. Additionally, CE-3 has the lowest rating among all classes.

(2)

HCP (Figure 17)

Overall, the HCP ratings for the four classes are generally moderate, averaging around 3.5 out of 5. When ranked from highest to lowest, the order is as follows: PCS, RCS, MCS, and CCS. From the graph, it can be observed that CCS has almost identical ratings for each HCP sub-item, while the other classes show clear distinctions between odd-numbered and even-numbered sub-items. Among them, MCS shows the most distinct pattern, with lower HCPB ratings and higher HCPA ratings. Additionally, the largest gap is observed between HCP-3 and HCP-4.

(3)

PA (Figure 18)

Overall, the four classes show relatively low ratings for PA, averaging around 2.3 out of 5. When ranked from highest to lowest, the order is as follows: PCS, CCS, RCS, and MCS. From the graph, it can be observed that all four classes have a higher preference for restaurants and strolling around the streets while showing a lower preference for training and health stores. In relative terms, CCS and PCS have a higher preference for training and health stores. Additionally, RCS exhibits the most distinct pattern in terms of PA.

(4)

Increased Customer Flow Revenue (Figure 19)

The analysis incorporates the increased customer flow resulting from the PSC and its corresponding proportion with the three standardized psychological levels mentioned above. The graph above illustrates this analysis. According to the quantification equation of CE, the proportions of consumer flow generated by the four classes are as follows: 24%, 31%, 28%, and 29%. It can be observed that the proportion of increased consumer flow is linearly correlated with the three psychological levels. The higher the psychological level, the greater the proportion of consumer flow that can be attracted. Based on the surveyed consumer levels, the economic benefits generated are calculated and presented in the graph. Although CCS has the lowest proportion of consumer flow, it exhibits the highest level of economic improvement (¥7041). The next highest economic effects are seen in PCS and RCS, while MCS shows the lowest economic impact (¥2153).

5. Conclusions and Recommendations

This study utilizes data from both RP and SP surveys to analyze the latent customer classes and their economic effects in TBD under the PSC. Using the EFA-LCM-CE model, an analysis was carried out concerning the sustainable economic benefits that diverse potential classes can offer to merchants. The following conclusions are drawn:

(1) By considering economy and society, parking, and psychological variables as the primary classification variables, the respondents are classified into four classes: Core Customers (CCS, 34%), Potential Customers (PCS, 29%), Regular Customers (RCS, 22%), and Marginal Customers (MCS, 15%). Additionally, EVUs are mainly comprised of CCS and RCS, who are generally high-income individuals, but their preference for charging facilities in TBD is not significant.

(2) By offering parking coupons, a total of 26% of customers can be attracted to visit the stores. Offering BD coupons and parking coupons can attract drivers to park at a distance (42%) while providing parking charging coupons and free parking coupons can attract drivers to park at a far distance (49%). The proportions of the four classes attracted are as follows: 24%, 31%, 28%, and 29%. CCS shows a higher consumer expenditure and prefers parking spots that are closer in walking distance. PCS exhibits a higher CEB, HCP, and PR and prefers parking spots with greater discounts. RCS shows significant PAA in TBD and prefers parking spots near their destination stores. MCS demonstrates a higher HCP and prefers parking spots with longer parking duration coupons.

(3) Under the quantification equation of CE, the economic effects brought by the four classes are as follows: ¥7041, ¥6000, ¥2769, and ¥2153. Compared to the before PSC, the level of economic improvement is 9%, 12%, 8%, and 10%. It can be observed that PCS exhibits the highest economic impact.

(4) Policy recommendations for the four classes:

General Recommendations: It is suggested that the radius of collaboration among stores in PSC within TBD be set at 500 m, and a unified parking price of ¥10–12 per hour is recommended. Additionally, it is advised that merchants use parking coupons and free coupons to convert drivers’ parking expenses into consumer expenditures. This approach not only enhances the parking experience but also further promotes the economic development of businesses. By dividing the customer market and promoting cooperation between merchants, once-autonomous businesses change into connected corporations, modifying the operational environment of the complete TBD. This collaborative approach helps in fulfilling harmonious and economic development initiatives within the established business district. The enhancement of the economic conditions of the TBD ensures the sustainable growth of urban areas. Specific recommendations for four classes to contribute to store economic development are as follows:

CCS: Implement a parking membership system. Based on the customer classification from market research, reserve a certain proportion of parking spaces (in this study, 34%) exclusively for CCS. Additionally, provide special parking coupons and benefits to CCS, ensuring accessibility and sufficient parking supply in nearby parking spots.

PCS: Offer various convenient incentives. Provide multiple incentive conditions for PCS, attracting them to park at parking spots within walking distance (300–500 m in this study). Collaborate with businesses to introduce parking coupons, install charging facilities, and improve parking convenience.

RCS: Strengthen store brand image. Stores should clearly define their unique positioning and characteristics, offer high-quality products and services, and ensure customer satisfaction and word-of-mouth spread. It is recommended for stores to determine suitable store types based on market research data (in this study, leisure stores).

MCS: Extend parking coupons duration. Collaborate with surrounding businesses to provide joint coupons, allowing customers to enjoy free parking for a certain duration after reaching a specific expenditure. Special parking coupons for students and an extension of parking duration can be implemented (in this study, 3 h of free parking).

This study did not discuss the cost of incentive combinations or the distribution ratio of increased benefits between businesses and parking. In the future, the introduction of platforms and big data analysis technologies can enable effective quantitative analysis of all customers, thereby attracting merchant cooperation and ultimately achieving the sustainable economic and socio-environmental development of all merchants.