In recent years, the rapid growth of electric vehicles (EVs) in China has transformed the transportation landscape, particularly in the online car-hailing sector. As a researcher focused on risk management and insurance, I have observed that the adoption of China EV models in ride-hailing services presents unique challenges for insurance products. Traditional auto insurance policies often fail to address the specific risks associated with electric vehicles, such as battery-related incidents and charging infrastructure dependencies. This study investigates the satisfaction and loyalty of consumers toward electric vehicle insurance in China’s online car-hailing industry, leveraging empirical data to analyze how consumer expectations, perceived value, and complaints influence their insurance experiences. By applying ordered Logit regression and mediation effect models, I aim to provide insights that can guide the optimization of electric vehicle insurance products, ultimately supporting the sustainable development of the EV market and green finance initiatives in China.
The proliferation of electric vehicles in China has been driven by government policies promoting green energy and technological innovation. In the online car-hailing sector, electric vehicles are increasingly favored due to their lower operating costs and environmental benefits. However, the insurance needs for these vehicles differ significantly from those of conventional internal combustion engine vehicles. For instance, electric vehicles face risks like battery fires, charging equipment failures, and specialized repair requirements, which are not fully covered by standard insurance policies. As a result, understanding consumer satisfaction and loyalty toward electric vehicle insurance is crucial for insurers to develop tailored products that meet the evolving demands of China EV users. In this study, I explore the relationships between consumer expectations, satisfaction, and loyalty, while considering the mediating roles of perceived value and complaints. Additionally, I examine the impact of purchasing additional insurance coverages, such as external grid fault loss insurance and charging pile-related policies, on consumer attitudes. The findings will help insurers and policymakers enhance electric vehicle insurance frameworks, fostering greater adoption of EVs in China’s transportation ecosystem.
To frame this research, I draw upon customer value theory and the Swedish Customer Satisfaction Barometer model, which posits that consumer expectations directly influence satisfaction, which in turn affects loyalty. In the context of electric vehicle insurance, consumer expectations refer to the pre-purchase beliefs about the insurance’s ability to cover losses, shaped by past experiences and market reputation. Perceived value represents the post-purchase assessment of whether the insurance provides adequate coverage relative to its cost. Satisfaction is the overall evaluation of the insurance experience, while loyalty manifests as repeat purchases and positive recommendations. Complaints act as a mediator, where dissatisfaction can lead to negative word-of-mouth, impacting loyalty. For electric vehicles in China, these dynamics are particularly relevant due to the nascent stage of EV-specific insurance products. My conceptual model, as illustrated in the theoretical framework, links expectations to satisfaction through perceived value and satisfaction to loyalty through complaints. Furthermore, I analyze how additional insurance coverages for electric vehicles, such as those addressing charging infrastructure risks, influence these relationships. This approach allows me to address gaps in the existing literature, which has largely overlooked the online car-hailing sector’s unique insurance needs for electric vehicles.
In terms of methodology, I designed a survey targeting online car-hailing drivers who use electric vehicles in China. The questionnaire consisted of two parts: demographic information and Likert-scale questions measuring variables like expectations, perceived value, satisfaction, loyalty, and complaints. Data were collected from 118 valid responses, after excluding incomplete surveys. The variables were measured using a five-point scale, ranging from “strongly disagree” to “strongly agree.” Key variables included consumer expectations (x1yuqi), perceived value (x2gzjz), satisfaction (myd), loyalty (y1zcd), and complaints (y22by). Additionally, I considered control variables such as gender, income, duration in the car-hailing industry, vehicle type, and accident history. For the analysis, I employed ordered Logit regression models due to the ordinal nature of the dependent variables, and I used mediation effect models to test the indirect pathways. The robustness of the results was verified through ordered Probit models and bootstrap methods. The descriptive statistics of the variables are summarized in Table 1, which provides an overview of the sample characteristics and variable distributions.
| Variable | Description | Mean | Standard Deviation | Min | Max |
|---|---|---|---|---|---|
| c1gender | Gender | 1.390 | 0.490 | 1 | 2 |
| c2income | Monthly Income | 2.280 | 0.836 | 1 | 4 |
| c3time | Time in Industry | 1.729 | 0.854 | 1 | 4 |
| c4vehicle | Vehicle Type | 1.669 | 0.472 | 1 | 2 |
| c5shigu | Accident History | 1.703 | 0.459 | 1 | 2 |
| x1yuqi | Expectations | 3.508 | 1.211 | 1 | 5 |
| x2gzjz | Perceived Value | 3.432 | 1.230 | 1 | 5 |
| myd | Satisfaction | 3.525 | 1.217 | 1 | 5 |
| y1zcd | Loyalty | 3.441 | 1.181 | 1 | 5 |
| y22by | Complaints | 1.272 | 1 | 5 | |
| c6fj1 | External Grid Insurance | 2.737 | 1.270 | 1 | 5 |
| c7fj2 | Charging Pile Loss Insurance | 2.703 | 1.276 | 1 | 5 |
| c8fj3 | Charging Pile Liability Insurance | 2.797 | 1.298 | 1 | 5 |
The empirical analysis began with baseline regression models to examine the direct effects. For the first stage, I used an ordered Logit model to regress satisfaction on expectations, controlling for demographic factors. The model is specified as:
$$ \text{logit}(P(\text{myd} \leq i)) = \beta_0 + \beta \cdot \text{x1yuqi} + \sum \beta_j \cdot C_j $$
where \( \text{myd} \) represents satisfaction levels, \( \text{x1yuqi} \) is consumer expectations, \( C_j \) denotes control variables, and \( \beta \) are coefficients. The results, shown in Table 2, indicate that expectations have a significant positive effect on satisfaction (coefficient = 0.929, p < 0.01). This suggests that higher pre-purchase expectations for electric vehicle insurance in China lead to greater satisfaction among online car-hailing drivers. Similarly, for the second stage, I regressed loyalty on satisfaction using the same model:
$$ \text{logit}(P(\text{y1zcd} \leq i)) = \gamma_0 + \gamma \cdot \text{myd} + \sum \gamma_j \cdot C_j $$
The results also show a significant positive relationship (coefficient = 0.833, p < 0.01), implying that satisfied consumers are more likely to remain loyal to their electric vehicle insurance providers. The marginal effects analysis revealed that as expectations increase, the probability of being “generally satisfied” first rises and then falls, indicating a nonlinear relationship. Similarly, satisfaction’s impact on “generally loyal” follows a comparable pattern, highlighting the complexity of consumer behavior in the China EV market.
| Variable | Satisfaction Coefficient | Loyalty Coefficient |
|---|---|---|
| Expectations (x1yuqi) | 0.929*** (0.165) | – |
| Satisfaction (myd) | – | 0.833*** (0.148) |
| Control Variables | Yes | Yes |
| Cut-point 1 | 1.544 (1.509) | -0.139 (1.611) |
| Cut-point 2 | 2.937** (1.470) | 1.340 (1.567) |
| Cut-point 3 | 4.413*** (1.502) | 3.186** (1.594) |
| Cut-point 4 | 5.977*** (1.536) | 4.551*** (1.641) |
| Log-likelihood | -158.5 | -159.3 |
Next, I investigated the mediating effects using a stepwise regression approach. For the first stage, I tested whether perceived value mediates the relationship between expectations and satisfaction. The models are:
$$ \text{logit}(P(\text{myd} \leq i)) = \beta_0 + \beta_1 \cdot \text{x1yuqi} + \sum \beta_j \cdot C_j $$
$$ \text{logit}(P(\text{x2gzjz} \leq i)) = \lambda_0 + \lambda_1 \cdot \text{x1yuqi} + \sum \lambda_j \cdot C_j $$
$$ \text{logit}(P(\text{myd} \leq i)) = \gamma_0 + \gamma_1 \cdot \text{x1yuqi} + \gamma_2 \cdot \text{x2gzjz} + \sum \gamma_j \cdot C_j $$
The results, presented in Table 3, show that the direct effect of expectations on satisfaction remains significant (coefficient = 0.769, p < 0.01), while the indirect effect through perceived value is also significant (coefficient = 0.310, p < 0.10). This confirms that perceived value partially mediates the relationship, meaning that consumers’ expectations for electric vehicle insurance in China enhance satisfaction by increasing their perception of the insurance’s value. For the second stage, I examined whether complaints mediate the satisfaction-loyalty relationship:
$$ \text{logit}(P(\text{y1zcd} \leq i)) = \beta_0 + \beta_1 \cdot \text{myd} + \sum \beta_j \cdot C_j $$
$$ \text{logit}(P(\text{y22by} \leq i)) = \lambda_0 + \lambda_1 \cdot \text{myd} + \sum \lambda_j \cdot C_j $$
$$ \text{logit}(P(\text{y1zcd} \leq i)) = \gamma_0 + \gamma_1 \cdot \text{myd} + \gamma_2 \cdot \text{y22by} + \sum \gamma_j \cdot C_j $$
The mediation analysis reveals that satisfaction has a direct positive effect on loyalty (coefficient = 0.769, p < 0.01), and complaints serve as a significant mediator (coefficient = 0.474, p < 0.01). This indicates that when consumers are dissatisfied with their electric vehicle insurance, they are more likely to complain, which negatively affects their loyalty. These findings underscore the importance of managing consumer perceptions and addressing complaints promptly in the China EV insurance market.
| Variable | Model 1: Satisfaction | Model 2: Perceived Value | Model 3: Satisfaction with Mediator | Model 1: Loyalty | Model 2: Complaints | Model 3: Loyalty with Mediator |
|---|---|---|---|---|---|---|
| Expectations (x1yuqi) | 0.929*** (0.164) | 0.977*** (0.168) | 0.769*** (0.185) | – | – | – |
| Perceived Value (x2gzjz) | – | – | 0.310* (0.171) | – | – | – |
| Satisfaction (myd) | – | – | – | 0.833*** (0.156) | 0.392*** (0.141) | 0.769*** (0.160) |
| Complaints (y22by) | – | – | – | – | – | 0.474*** (0.150) |
| Control Variables | Yes | Yes | Yes | Yes | Yes | Yes |
| Cut-point 1 | 1.544 (1.346) | 1.061 (1.375) | 1.960 (1.367) | -0.139 (1.337) | -1.145 (1.289) | 0.951 (1.387) |
| Cut-point 2 | 2.937** (1.340) | 2.279* (1.385) | 3.351** (1.361) | 1.340 (1.335) | 0.150 (1.264) | 2.505* (1.395) |
| Cut-point 3 | 4.413*** (1.370) | 3.475** (1.395) | 4.863*** (1.396) | 3.186** (1.365) | 1.406 (1.267) | 4.458*** (1.439) |
| Cut-point 4 | 5.977*** (1.418) | 5.552*** (1.445) | 6.464*** (1.449) | 4.551*** (1.388) | 2.443* (1.283) | 5.895*** (1.469) |
| Log-likelihood | -158.5 | -157.1 | -156.8 | -159.3 | -175.7 | -154.2 |
In addition to the main analysis, I explored the impact of purchasing additional insurance coverages for electric vehicles on satisfaction and loyalty. Using ordered Logit models, I regressed satisfaction and loyalty on three types of additional insurance: external grid fault loss insurance (c6fj1), self-use charging pile loss insurance (c7fj2), and self-use charging pile liability insurance (c8fj3). The models are specified as:
$$ \text{logit}(P(\text{myd} \leq i)) = \beta_0 + \beta \cdot F + \sum \beta_j \cdot C_j $$
$$ \text{logit}(P(\text{y1zcd} \leq i)) = \gamma_0 + \gamma \cdot F + \sum \gamma_j \cdot C_j $$
where \( F \) represents the additional insurance variables. The results, summarized in Table 4, show that all three types of additional insurance have significant negative effects on both satisfaction and loyalty (p < 0.01). For example, the coefficient for external grid insurance on satisfaction is -0.763, indicating that consumers who purchase this coverage report lower satisfaction levels. Similarly, charging pile liability insurance has the strongest negative impact on loyalty (coefficient = -0.873). This suggests that while these add-ons are designed to address specific risks of electric vehicles, they may not meet consumer expectations or could be perceived as overly costly, leading to dissatisfaction. This finding is critical for insurers in China, as it highlights the need to refine additional coverage options for electric vehicles to align with consumer needs in the online car-hailing sector.
| Variable | Satisfaction: Model 1 | Satisfaction: Model 2 | Satisfaction: Model 3 | Loyalty: Model 1 | Loyalty: Model 2 | Loyalty: Model 3 |
|---|---|---|---|---|---|---|
| External Grid Insurance (c6fj1) | -0.763*** (0.146) | – | – | -0.532*** (0.144) | – | – |
| Charging Pile Loss Insurance (c7fj2) | – | -0.772*** (0.148) | – | – | -0.798*** (0.156) | – |
| Charging Pile Liability Insurance (c8fj3) | – | – | -0.781*** (0.173) | – | – | -0.873*** (0.154) |
| Control Variables | Yes | Yes | Yes | Yes | Yes | Yes |
| Cut-point 1 | -3.549** (1.445) | -3.888** (1.601) | -3.907** (1.541) | -3.602*** (1.319) | -4.972*** (1.321) | -5.729*** (1.543) |
| Cut-point 2 | -2.295* (1.389) | -2.604* (1.541) | -2.590* (1.510) | -2.239* (1.270) | -3.544*** (1.268) | -4.266*** (1.463) |
| Cut-point 3 | -0.913 (1.393) | -1.187 (1.537) | -1.211 (1.517) | -0.624 (1.268) | -1.769 (1.249) | -2.444* (1.413) |
| Cut-point 4 | 0.633 (1.419) | 0.360 (1.561) | 0.317 (1.526) | 0.647 (1.283) | -0.350 (1.260) | -0.969 (1.395) |
| Log-likelihood | -161.9 | -161.6 | -161.9 | -167.5 | -159.5 | -156.5 |
To ensure the robustness of my findings, I conducted several checks. First, I re-ran the baseline models using ordered Probit regression, which yielded similar results: expectations significantly positively affect satisfaction (coefficient = 0.549, p < 0.01), and satisfaction positively influences loyalty (coefficient = 0.487, p < 0.01). The marginal effects patterns remained consistent, with probabilities for “generally satisfied” and “generally loyal” showing an initial increase followed by a decrease. Second, I performed Sobel tests and bootstrap methods for the mediation effects. For the first stage, the Sobel test indicated a significant indirect effect of expectations on satisfaction through perceived value (z = 1.854, p = 0.06), and the bootstrap confidence interval (0.0035, 0.2281) did not include zero, confirming mediation. For the second stage, the Sobel test for complaints mediation was significant (z = 2.139, p = 0.03), and the bootstrap interval (0.0169, 0.1541) supported the indirect effect. Finally, the ordered Probit models for additional insurance coverages also showed significant negative effects, reinforcing the initial conclusions. These robustness checks validate that the relationships hold across different statistical methods, enhancing the reliability of my results for the China EV insurance context.
In discussion, my findings align with existing literature on customer satisfaction but extend it to the niche of electric vehicle insurance in China’s online car-hailing industry. The positive effect of expectations on satisfaction underscores the importance of managing consumer pre-purchase beliefs, which can be influenced by marketing and word-of-mouth. The mediating role of perceived value suggests that insurers should emphasize the value proposition of their electric vehicle insurance products, perhaps by highlighting coverage for unique EV risks like battery fires. The satisfaction-loyalty link, mediated by complaints, indicates that insurers need to invest in customer service and claims handling to mitigate negative feedback. The negative impact of additional insurance coverages on satisfaction and loyalty may reflect a mismatch between product offerings and consumer needs; for instance, policies related to charging infrastructure might be perceived as unnecessary or overpriced. This is particularly relevant for China EV users in the online car-hailing sector, who often operate under tight margins and may prioritize cost-effectiveness. My study contributes to the growing body of research on green insurance and electric vehicles by providing empirical evidence from a key user group, offering practical insights for insurers aiming to tailor products for the evolving China market.
In conclusion, this research demonstrates that consumer expectations for electric vehicle insurance directly enhance satisfaction, with perceived value acting as a mediator, while satisfaction boosts loyalty, influenced by complaints. Additionally, purchasing additional insurance coverages for electric vehicles tends to reduce both satisfaction and loyalty. These findings have important implications for insurers, policymakers, and industry associations in China. Insurers should focus on aligning electric vehicle insurance products with consumer expectations by offering customizable policies that address specific risks, such as those related to batteries and charging. Improving perceived value through competitive pricing and comprehensive coverage could enhance satisfaction. Moreover, streamlining claims processes and addressing complaints proactively can foster loyalty. For additional coverages, insurers should reassess their design and pricing to ensure they meet the practical needs of online car-hailing drivers using electric vehicles. From a policy perspective, regulatory bodies could support the development of standardized electric vehicle insurance frameworks that promote transparency and consumer trust. Future research could expand on this work by exploring regional variations in China or comparing different EV segments, such as private versus commercial use. Overall, optimizing electric vehicle insurance is essential for supporting the sustainable growth of the EV industry in China and advancing green finance goals.