The Influencing Mechanism of Household Food Purchasing Behavior and Household Reserve Efficiency under Non-Normal Conditions

Abstract

Family reserves are an important part of national reserves, and how to do a good job in family reserves is a common concern of the government and society. Under the non-normal conditions of major accidents and disasters, wars, plagues, social unrest, etc., urban food supply mainly depends on external supply guarantee, and urban residents’ risk perception is more sensitive. Based on the Theory of Planned Behavior and Norm Activation Model, this study constructs an analytical framework for the risk perception, perceived behavior control, and family reserve efficacy of urban residents under non-normal conditions from the perspectives of rationality and sensibility, self-interest, and altruism. The perceived behavior control of household food reserves in non-normal conditions is affected by risk perception, subjective norms, and personal norms. On this basis, countermeasures and suggestions are put forward: Urban residents should strengthen their sense of risk and responsibility for storing food at home, reserve food appropriately, and develop a good habit of family saving. On the other hand, it is necessary to pay attention to personal norms, reduce the negative impact of subjective norms on residents, and avoid excessive food storage and food waste.

1. Introduction

The context of the article is China. During the period of prevention and control of COVID-19 from 2020 to 2022, residents in some areas were restricted from traveling, facing restrictions on the demand for daily necessities, which had a direct impact on residents’ daily consumption and rigid demand for food, clothing, and housing [1]. The world is facing major changes unprecedented in a century, and various foreseeable and unforeseeable risk factors have increased significantly. Under the non-normal conditions of major accidents and disasters, wars, international trade disputes over agricultural products, plagues, social unrest, and unforeseen factors, problems such as production stagnation, food supply chain disruption, food safety panic, international trade disruption, and reduction of residents’ income are common, accompanied by food safety hazards, the sustainability of agricultural activities have been under major threat [2,3] and residents’ dietary consumption has been severely impacted [4,5]. In August 2023, affected by the discharge of nuclear sewage from Japan’s Fukushima nuclear power plant into the sea, which is harmful to the sustainable development of the environment, many people in China hoarded a large amount of edible salt. Affected by non-normal events, residents generally reserve a large number of daily necessities, resulting in rising prices and scarcity of goods, which also has a serious impact on the material supply and demand system, exacerbating the contradiction between supply and demand.

In recent years, many scholars have extensively studied the non-normal state, such as Gao Rui et al. [6] mainly focused on the special moderating role of corporate social responsibility performance under the non-normal state, and Wang et al. [7] believed that the non-normal grid governance should be carried out in response to the public crisis, but there are few studies on the food purchasing behavior of residents under the non-normal state, especially on the household reserve efficiency under the non-normal state. When a crisis occurs, families usually reserve a certain amount of emergency supplies, such as food, drinking water, evacuation, and survival equipment, so as to effectively improve the self-rescue and mutual rescue capabilities of residents, thereby improving the risk prevention level of the whole society. The construction of the “national reserve system” is related to China’s national security and social stability, which mainly includes government reserves, enterprise reserves, public welfare organization reserves, and family reserves [8]. To reconstruct the large national reserve system and improve the efficiency of reserves, a combination of government reserves and household reserves can be adopted [8]. Household reserves are an important part of national reserves, and how to improve the efficiency of family reserves is a common concern of the government and society.

Efficacy means the role and utility, reserve efficiency means the beneficial effect of the reserve behavior, and family reserve efficiency is the sense of security, satisfaction, and responsibility brought to the family by the reserve behavior of the family unit, and the family reserve efficiency reflects the purchase effect of the family in the food reserve when the non-normal event occurs, as well as the family’s ability to resist risks in the face of non-normal events. For example, Yang [9] studied the comprehensive disaster preparedness standards for household emergency supplies, which are mainly divided into medicines, food tools, and other standards, but there are few studies focusing on the effectiveness of household emergency food reserves, and the relevant concepts of family food reserve efficiency are also relatively vague. Based on the integrated Theory of Planned Behavior and the Norm Activation Model, this paper explores the influencing mechanism of urban residents’ household reserve efficiency under non-normal conditions from the aspects of self-interest and altruism, rationality and morality, and proposes measures and ways to improve the efficiency of household reserves. Since the urban food supply mainly depends on the external supply guarantee of the city, urban residents are more sensitive to the risk perception under non-normal conditions. What are the particularities of household food purchasing behavior under non-normal conditions, what are the effects of these food purchasing behaviors on the efficiency of household reserves, and what are the influencing mechanisms between the two? These are the questions that need to be answered in this paper. This study has important theoretical and practical significance for the construction of a unified and efficient “national reserve system”.

2. Theoretical Basis and Research Framework

2.1. Theoretical Basis

Based on the assumption of rational people, rational residents take a rational perspective to buy food reserves for emergencies. Under the non-normal situation, residents’ risk perception of crisis events affects their perceived behavior control, which in turn affects the efficiency of household reserves. Therefore, this study established an analytical framework of “non-normal events-risk perception-residents’ perceived behavior control-family reserve efficacy”. Therefore, this study considers the two factors of subjective norms and perceived behavior control in the Theory of Planned Behavior, takes the risk perception of non-normal events as the starting point and incorporates the “personal norms” and “responsibility consciousness” in Norm Activation Model into the analytical framework and establishes the analytical framework from the perspectives of rationality and sensibility, self-interest and altruism (Figure 1).

The Theory of Planned Behavior (TPB) is a classic theory for the study of consumer behavior, which was proposed and refined by Ajzen. As a rational choice model, TPB starts from individual self-interest and assumes that individual behavior is based on cost–benefit rational choice, so the role of emotional factors and altruistic motivation in behavior shaping is ignored. On the basis of rational behavior theory, Ajzen combined the expectation-value theory and added a new research variable to explain the impact of external objective conditions on individual behavior from the perspective of information processing perceived behavior control and proposed a planned behavior theory that can explain the decision-making process of individual behavior [10]. The theory, as an improver of the theory of rational behavior, argues that behavioral intention indirectly influences behavior through all factors that may influence behavior [11].

Norm Activation Model (NAM), which considers affective factors and altruistic motivations and emphasizes the importance of moral responsibility, is a classic model for explaining pro-environmental behavior, which was introduced by Schwart in 1977 to describe altruistic behavior [12]. As an important theory for studying altruistic behavior, NAM regards personal norms as the most direct factor influencing altruistic behaviors, while the sense of responsibility affects personal norms [13]. The two important components of this theory are personal norms, which refers to the individual’s perception of moral obligation to perform a certain behavior, and responsibility attribution, which refers to the individual’s sense of responsibility recognition for carrying out a certain behavior [14,15]. Pan et al. [16] combined NAM and TPB to jointly clarify the pro-environmental willingness of forest tourism enterprises from the perspectives of managers’ environmental ethics and corporate rationality; Qin et al. [17] expanded the traditional TPB and combined NAM and TPB to clarify the pro-environmental willingness of marine aquatic enterprises from both moral and rational aspects. And Wan et al. [18] explained the self-interested and altruistic parts of public participation intention through integrated TPB and NAM so as to provide a theoretical basis for governments and enterprises to formulate policies related to public participation. However, based on the integrated TPB and NAM, it is still relatively rare to explore the reserve efficiency of urban households under non-normal situations.

Based on the above analysis, this study constructs an analytical framework for the study of residents’ food purchasing behavior and household reserve efficiency under non-normal conditions based on TPB and NAM. Under the external stimulus of non-normal events, urban residents perceive external risks, which are manifested in the aspects of rationality and emotion, self-interest, and altruism, which affect the perceived behavior control and further affect the efficiency of family reserves.

2.2. Research Hypothesis

Perceived behavior control refers to an individual’s perception of the difficulty of performing a certain behavior, reflecting that the behavior is constrained by various factors such as resources, personal ability, and opportunities [19]. In general, the easier an individual perceives to practice a certain behavior, and the more external factors there are, the more likely it is that the behavior will be realized [20]. NAM suggests that the important factors that activate personal norms are the awareness of responsibility and the consciousness of consequences [21], and when individuals perceive that they have stronger behavioral control ability, the more likely their personal norms are to be activated to implement corresponding behaviors. Hu et al. [22] studied the impact of work-family conflict on the fertility intention of urban workers based on TPB, and the study showed that perceived behavior control had a significant positive impact on the fertility intention of the second child and perceived behavior control played a complete mediating role in the relationship between work-family conflict and second-child fertility intention. Gao et al. [23] showed that in terms of green agricultural products, consumer motivation and cognitive level of agricultural product quality and safety level positively affect consumer purchasing behavior through perceived behavior control and subjective norms. This study suggests that residents perceive that the barriers to purchasing food will increase under non-normal conditions, the difficulty of purchasing food will increase compared with the normal situation, and the convenience of purchasing will decrease, so they are more inclined to buy food to store, prevent untimely needs, and improve their sense of security, satisfaction, and responsibility. As the first responder after the occurrence of a non-normal event, family members will perceive that the more obstacles to buying food, the stronger the perceived behavior control ability of residents to buy food reserves, the greater the possibility of purchasing food reserves, the stronger the sense of security, satisfaction, and responsibility to the family, and the corresponding improvement of the efficiency of family reserves. Therefore, the hypothesis is proposed:

H1. 

Perceived behavior control has a positive effect on the efficiency of family reserves.

Subjective norms refer to the external social pressure that individuals feel when considering whether or not to perform a particular behavior [24], which has been proven to be a critical factor affecting consumers’ purchasing intentions [25]. Subjective norms reflect the views and attitudes of socially important groups towards the behavior [26]. Individuals with a stronger sense of subjective norms are more susceptible to the influence of others and ultimately behave in line with the expectations of the majority or behave in line with the majority. Shi et al. [27] studied farmers’ willingness to green production based on the extended TPB and found that subjective norms have a significant impact on perceived behavior control; that is, the green production behavior of surrounding people will help farmers improve their corresponding green production knowledge and skills. Yan et al. [28] found that subjective norms and perceived behavior control have a mutual impact on the basic pension insurance participation behavior of young people in the new business format. The subjective norms in this study refer to the influence of people around (such as relatives, friends, neighbors, and media) on the family’s ability to judge the ease of purchasing food during non-normal events. For people in non-normal events, the purchasing behavior of important social groups such as relatives, friends, and neighbors will make them show a herd mentality driven by subjective norms. Making changes in behavior or beliefs based on others is herd mentality, and one of the reasons why people want to be accepted by others is to conform to the herd [29]. Therefore, the stronger the residents’ awareness of subjective norms, the more easily they are influenced by the people around them, which in turn affects their perceived behavior control of buying food; that is, they affect their own judgment of the difficulty of purchasing food. Therefore, the hypothesis is proposed:

H2. 

Subjective norms positively affect residents’ perceived behavior control of food purchases.

A sense of responsibility bursts out from an individual’s failure to perform a specific behavior and produces adverse consequences [30], which is a good moral character from an individual and reflects an individual’s emotional attitude towards society and the environment. As a precursor to the activation of personal norms, a sense of responsibility is the manifestation of individuals actively following social norms and internalizing them into personal norms; that is, the stronger the sense of responsibility, the more likely they are to activate personal norms to implement altruistic behaviors [31]. Zhang et al. [32] found that a sense of responsibility significantly affects behavior control based on a comprehensive research framework of TPB and NAM, indicating that the improvement of a sense of responsibility helps air travelers produce stronger behavior control. Li et al. [33] found that human-land emotions can affect personal norms through the role of responsibility attribution to produce environmentally responsible behaviors. The sense of responsibility in this study refers to a concept of responsibility to the family or others in society that urban residents are aware of before or after making a purchase when a non-normal event occurs. The stronger the sense of responsibility of residents in a non-normal state, the more aware of the adverse consequences of not buying food for stockpiling will be on their families, the more eager they are to store food, and the more they can perceive the difficulty of the behavior of storing food. Therefore, the hypothesis is proposed:

H3. 

Sense of responsibility positively affects perceived behavior control.

Personal norms are an individual’s sense of moral responsibility for taking certain behaviors [11], which is the internal driving factor of an individual’s pro-environmental behavior [34], and this sense of responsibility enables individuals to associate their own behavior with moral principles and values, and to generate feelings of pride or guilt from their behavior. In this study, personal norms refer to a kind of moral concept towards the family or others in society that urban residents are aware of before or after making a purchase at the time of non-normal events, and individuals with stronger personal norms are more likely to implement behaviors that follow the requirements of their personal norms in production and life [35]. Therefore, when a non-normal event occurs, out of the residents’ own sense of morality and responsibility, the stronger the perception of responsibility and obligation to buy food, the more they can perceive the urgency of purchasing food, the stronger the personal norms, and the stronger the perceived behavior control. Therefore, the hypothesis is proposed:

H4. 

Personal norms positively affect perceived behavior control.

When a non-normal event occurs, residents perceive the risk through social media and other means. Risk perception refers to an individual’s subjective perception of objective risks from the outside world, which is closely related to decision-making [36]. Chen et al. [37] explored the influencing mechanism of risk perception on farmers’ land trusteeship intention and behavioral decision-making, and the study showed that natural risk perception positively affected land trusteeship decision-making through perceived behavior control. Meng [38] found that risk perception has a positive impact on the subjective norms of virtual information recognition of social media users. Jiang et al. [39] discussed the impact of herdsmen’s livelihood capital on the livelihood risk of livestock freezing injury based on the intermediary perspective of risk perception, and the results showed that herders’ risk perception has a significant mediating effect on the impact of livelihood capital on the livelihood risk of livestock frost injury. After the occurrence of non-normal events, families perceive risks in various ways, such as through media and a circle of friends. Families are generally composed of multiple members, and different individuals in the family members have different risk perceptions of the same event, which affects the subjective norms, sense of responsibility, and personal norms of food purchases.

H5a. 

Residents’ risk perception has a significant impact on perceived behavior control.

H5b. 

Residents’ risk perception has a significant impact on subjective norms.

H5c. 

Residents’ risk perception has a significant impact on their sense of responsibility.

H5d. 

Residents’ risk perception has a significant impact on personal norms.

This study suggests that subjective norms, sense of responsibility, and personal norms may play a mediating role in the relationship between risk perception and perceived behavior control, and personal norms may also play a mediating role in the path of sense of responsibility and perceived behavior control. For example, Zhang [40] explores the mediating role of green literacy and green consumption attitude in consumers’ green education and green consumption intentions. Based on TPB and NAM, Li et al. [41] studied the influencing factors of consumers’ environmentally friendly clothing purchasing behavior, and the results showed that there is a mediating role between the perception of responsibility and subjective norms, and the mediating role of sense of responsibility accounts for 19.97%. Ni et al. [42] used structural equation models to deeply study the impact and internal mechanism of public participation in cross-regional ecological compensation behavior, and the results showed that the risk-averse public had lower willingness to participate, more negative behavioral attitudes, greater social pressure to participate in compensation, and stronger perceived behavior control. In addition, in addition to the direct impact on perceived behavior control, risk perception may also have an indirect impact on purchasing behavior through three mediating variables: subjective norms, sense of responsibility, and personal norms; subjective norms, sense of responsibility and personal norms may play a mediating role in the path from risk perception to perceived behavior control, and sense of responsibility may affect perceived behavior control through personal norms. Therefore, the following hypotheses are proposed:

H6a. 

Subjective norms may play a mediating role in the pathway from risk perception to perceived behavior control.

H6b. 

Sense of responsibility may play a mediating role in the path from risk perception to perceived behavior control.

H6c. 

Personal norms may play a mediating role in the pathway from risk perception to perceived behavior control.

H6d. 

Personal norms may play a mediating role in the path of a sense of responsibility and perceived behavior control.

3. Study Design

3.1. Variable Design

This study is based on the non-normal events of the 2022 COVID-19 lockdown, and the questionnaire uses a seven-level Likert scale and basic information multiple-choice questions, ranging from completely disagree to complete agreement, with 1 representing complete disapproval and 7 representing complete agreement. In order to ensure reliability and validity, the items designed for each variable in the questionnaire are based on the maturity scale, and the design of the questionnaire scale and the source of the items are shown in

Table 1. Questionnaire scale design and item sources.

Variables	Item Number	Item Content	Item Sources
Risk Perception	RP1	During the lockdown, I thought I was more likely to be infected with COVID-19.	Cui X.Q et al. [43]
	RP2	During the lockdown, I felt that this round of the epidemic was expected to last for a long time.
	RP3	During the lockdown, I felt that being infected with COVID-19 would have a great impact on my studies and work.
	RP4	During the lockdown, I felt that being infected with COVID-19 would have a great impact on my life.
	RP5	During the lockdown, what I said on social media had a big impact on my perception of risk.
Subjective Norms	SN1	During the lockdown, the food-buying behavior of my relatives/friends had a great impact on my food purchase.	Jia Y.J et al. [44]
	SN2	During the lockdown, the food-buying behavior of my classmates/colleagues had a great impact on my food purchase.
	SN3	During the lockdown, the food purchase advice of authoritative media and experts had a great impact on my food purchase.
	SN4	During the lockdown period, the food purchasing behavior of the neighborhoods had a great impact on my food purchases.
Perceptual Behavioral Control	PBC1	During the lockdown, I was able to buy more food to stock up on.	Jia L. et al. [44]
	PBC2	During the lockdown, I was able to buy more food.
	PBC3	During the lockdown, I was able to spend more time buying food to stock.
Sense of Responsibility	SOR1	During the lockdown, I felt guilty for not stocking up on food in time to affect my family’s lives.	Liu J. et al. [45]
	SOR2	During the lockdown, I felt guilty about the waste of resources caused by overbuying food.
	SOR3	During the lockdown, failure to stock up on a certain amount of food is irresponsible for yourself and your family.
Personal Norms	PN1	During the lockdown, I should distribute food to those who need it most, such as the elderly, the weak, the sick and the disabled.	Jia L. et al. [44]
	PN2	During the lockdown, you should share the excess food at home with your neighbors.
	PN3	During the lockdown, you should share excess food with friends and relatives at home.
Household Reserve Efficiency	HRE1	During the lockdown, buying food gave me a sense of security.
	HRE2	During the lockdown, the food purchases gave me a sense of satisfaction.
	HRE3	During the lockdown, my food purchases have shown my sense of responsibility.

.

“Risk perception” was measured from five dimensions: “possibility of infection”, “estimated duration of the epidemic”, “impact on study or work”, “impact on life”, and “social media speech”, and “subjective norms” were measured from four dimensions: “relatives and friends”, “classmates and colleagues”, “authoritative media experts” and “neighborhoods”. “perceptual behavioral control” was measured from three dimensions: “being able to buy more food”, “having more channels” and “spending more time”, while “sense of responsibility” was measured from two dimensions: “feeling guilty” and “thinking that practices are not responsible”. “Personal norms” were measured from three dimensions: “food should be distributed to those who need it most, such as the elderly, the weak, the sick and disabled”, and the excess food at home should be shared with “neighbors” and “friends and relatives”, and “household reserve efficiency” was measured from the dimensions of “sense of security”, “satisfaction” and “sense of responsibility” brought by families who bought food during the epidemic lockdown. The content of each question item and the basis of the measurement are shown in Table 1.

3.2. Data Sources

Through the Credamo network data research platform and Wenjuanxin questionnaire research platform, this study conducted a survey on food purchasing behavior under the lockdown of the COVID-19 epidemic among 501 residents in Jiangsu, Zhejiang, Shanghai, and Anhui provinces in July 2023. This study is based on the epidemic prevention and control in China in 2023. First of all, compared with Western countries, which pay more attention to individual independence and the concept of boundaries between family members, Chinese people have a stronger family concept and attach importance to affection and ethics [46], and the effectiveness of family reserves under the non-normal situation is a hot issue in society. Second, since ancient times, Chinese people have a strong sense of distress [47] and know how to plan ahead. In addition, during the epidemic prevention and control period from 2020 to 2022, residents in Jiangsu, Zhejiang, Shanghai, Anhui, and other regions have experienced household food purchases under epidemic prevention and control, and they have a deep understanding of family reserves under non-normal conditions. This survey set two selection criteria: first, the location of the respondent’s mobile phone should be consistent with the set area, and second, the questionnaire was set with detection items to check whether the respondent answered the question seriously and the samples that did not be answered seriously have been rejected.

3.3. Descriptive Statistics

The sample size of the questionnaire was evenly distributed in all provinces and cities, including 126 samples in Shanghai, 125 samples in Jiangsu Province, 125 samples in Zhejiang Province and 125 samples in Anhui Province; 43.31% of the respondents were males and 56.69% were females; 86.43% were owners; the age was mainly distributed in 31~40 years old (47.31%); and the monthly household income was mainly concentrated in 1–20,000 yuan (50.70%) The majority of the respondents (96.61%) are in good physical condition; most of the people living in the same family are about 3 (44.11%); Most of the families (76.25%) have elderly or children; The people participating in this survey generally have a high level of education, with 70.46% having a college or bachelor’s degree and 19.75% having a postgraduate degree; and working in private enterprises (48.50%). From the perspective of the personal characteristics of the respondents, the sample is representative.

4. Results and Analysis

4.1. Overall Goodness-Of-Fit Analysis

In this study, AMOS26.0 software was used to fit the data to the structural equation. The overall fit is shown in

Table 2. Overall fit series table.

Indicator Category	Statistical Test Volume	Adaptation Standards or Thresholds	Test Result Data	Model Adaptation Judgment
Absolute Fit Index	RMSEA	≤0.08	0.060	Yes
	GFI	≥0.80	0.910	Yes
Enhance the Fit Index	NFI	≥0.80	0.855	Yes
	IFI	≥0.80	0.901	Yes
	TLI	≥0.90	0.881	Yes
	CFI	≥0.80	0.900	Yes
Simple Fit Index	PGFI	≥0.50	0.698	Yes
	PNFI	≥0.50	0.720	Yes
	PCFI	≥0.50	0.759	Yes
	X2/df	<3 Fine, ≤5 Acceptable	2.828	Fine

, and the indicators show that the survey data in this paper fit well with the model so that the path analysis can be continued.

4.2. Reliability and Validity Test

SPSS26.0 and AMOS26.0 software were used to test the reliability and validity of six latent variables, namely risk perception, subjective behavior norms, perceived behavior control, sense of responsibility, personal norms, and reserve efficacy, as shown in

Table 3. Confirmatory factor analysis and reliability test.

Latent Variables	Observed Variables	Factor Loading	CR	AVE	Cronbach’s Alpha	KMO	Bartlett Test
Risk Perception	RP1	0.527	0.697	0.317	0.844	0.740	489.614 (p < 0.000)
	RP2	0.607
	RP3	0.481
	RP4	0.534
	RP5	0.651
Subjective Norms	SN1	0.874	0.810	0.528		0.733	750.708 (p < 0.000)
	SN2	0.849
	SN3	0.471
	SN4	0.638
Perceptual Behavioral Control	PBC1	0.629	0.626	0.360		0.659	237.563 (p < 0.000)
	PBC3	0.517
	PBC4	0.646
Sense of Responsibility	SOR1	0.545	0.576	0.315		0.621	133.454 (p < 0.000)
	SOR2	0.479
	SOR3	0.646
Personal Norms	PN1	0.764	0.852	0.658		0.724	653.117 (p < 0.000)
	PN2	0.864
	PN3	0.802
Household Reserve Efficiency	HRE1	0.667	0.699	0.437		0.654	263.22 (p < 0.000)
	HRE2	0.704
	HRE3	0.608

. As can be seen from Table 3, the Cronbach coefficient of the six latent variables is 0.844, indicating that the reliability of the scale is very good. The KMO values of each variable are greater than 0.6, and the significance level of the Bartlett spherical test is less than 0.001, indicating that the questionnaire has construct validity. According to the results of confirmatory factor analysis, the factor loading of each observed variable basically reached 0.5, the combined reliability (CR) basically reached 0.6, and the average variation extraction (AVE) basically reached the acceptable threshold (0.36–0.5), indicating that the questionnaire had a fairly high validity as a whole.

4.3. Direct Effect Test

In this paper, the direct relationship between each variable is tested by using AMOS26.0 software to construct structural equations, as shown in

Table 4. Structural equation model estimation results.

Hypothesis	Estimate	S.E.	C.R.	P	Conclusion
H1: Perceived Behavior Control→Household Reserve Efficiency	0.550	0.072	6.661	***	Supported
H2: Subjective Norms→Perceived Behavior Control	0.190	0.068	2.135	0.033 *	Supported
H3: Sense of Responsibility→Perceived Behavior Control	0.213	0.094	1.955	0.051	Not supported
H4: Personal Norms→Perceived Behavior Control	0.136	0.050	2.313	0.021 **	Supported
H5a: Risk Perception→Perceived Behavior Control	0.236	0.119	2.062	0.039 *	Supported
H5b: Pisk Perception→Subjective Norms	0.682	0.091	10.206	***	Supported
H5c: Risk perception→Sense of Responsibility	0.582	0.105	6.645	***	Supported
H5d: Risk Perception→Personal Norms	0.264	0.100	3.204	0.001 **	Supported

Note: * means p < 0.05, ** means p < 0.01, and *** means p < 0.001.

.

Perceived behavior control positively affected household reserve efficiency (β = 0.550, p < 0.001), and H1 was verified. This conclusion shows that the stronger the perception of the difficulty of purchasing food, the stronger the perceived behavior control, the greater the possibility of purchasing food reserves, and the greater the resilience of family food reserves, the more they can bring a sense of security, satisfaction, and responsibility to family members, and household reserve efficiency will be greatly improved. This conclusion is helpful for the establishment of a “national reserve system”, especially the construction of a family-based reserve system.

Subjective norms significantly positively affect perceived behavior control at the 5% level (β = 0.190), and H2 is true. This conclusion shows that based on herd psychology and demonstration effect, the stronger the subjective norms of residents, the more susceptible they are to the influence of others. Influenced by the social groups around them, the residents’ perceived behavior control to carry out purchasing behavior will be enhanced, and the possibility of purchasing behavior will also be improved, which cannot explain the social phenomenon of panic buying by the whole people once there is a non-normal event.

Sense of responsibility did not significantly affect residents’ perceived behavior control (p > 0.05), and H3 was not valid. The possible reason is that residents’ perception of the difficulty of implementing food purchasing behavior is affected by many objective factors, such as various factors such as channels, resources, opportunities, and income, although families have a strong sense of responsibility and willingness to buy food to ensure family food security, but restricted by objective factors, it will not objectively affect the difficulty of residents’ purchasing behavior, so the sense of responsibility cannot significantly affect perceived behavior control.

Personal norms significantly positively affect perceived behavior control at the 1% level (β = 0.136), and H4 is validated. This suggests that the stronger the personal norms, the stronger the perceived behavior control. When a non-normal event occurs, out of concern for the survival and development of the whole family, the residents’ sense of morality and responsibility is stimulated to a large extent, and the stronger their perception of the moral obligation to buy food, the more they can perceive the urgency of buying food and the difficulty of purchasing food.

From the estimation results of the structural equation model in Table 4, it can be seen that risk perception has a significant impact on perceived behavior control, subjective norms, sense of responsibility, and personal norms, and H5a, H5b, H5c, and H5d are all verified. In the hypothesis test of the path relationship in this study, risk perception significantly positively affects perceived behavior control at the 5% significance level (β = 0.236), which indicates that in the non-normal situation, the greater the residents’ perception of the risk of non-normal events, the more they will realize that the event may affect the channels, time, and opportunities of residents’ food purchases, and the difficulty of residents’ food purchasing behavior will increase. Risk perception has a significant positive impact on subjective norms (β = 0.682, p < 0.001), which indicates that under the non-normal situation, residents not only have a risk perception of the event but also people around them, such as family, friends, colleagues, and classmates, also realize that the event has a risk to the food purchasing behavior, and the difficulty of buying food will also increase. Risk perception has a significant positive impact on the sense of responsibility (β = 0.582, p < 0.001); the possible reason is that, based on the sense of responsibility to the family or others in society, when a non-normal event occurs, the stronger the sense of responsibility of residents, the more aware of the insecurity of the family caused by not buying food reserves, and thus the more eager to store food, through the purchase of food reserves to ensure the normal life of other family members. Risk perception has a significant positive impact on personal norms at the 1% significance level (β = 0.264), which indicates that when non-normal events occur, the stronger the residents’ personal norms, the more residents perceive the risk, but the more they can stimulate their own sense of morality and responsibility.

4.4. Mediator Effect Test

This paper not only examines the direct effects but also examines the possible mediating effects of subjective norms, sense of responsibility, and personal norms in the control path from risk perception to perceptual behavior, as well as the possible mediating effects of personal norms on the path from sense of responsibility to perceived behavior control. Based on AMOS26.0 software, this paper uses the Bootstrap mediation test method to test the specific indirect effects in each mediation path. The Bootstrap sampling number is set to 2000, the non-parametric percentile method with bias correction is selected, the confidence interval is set to 95%, and the test results of the mediation effect are shown in

Table 5. Results of the mediating effect test.

Hypothesis	Effect	Estimate	Lower	Upper	P	Test Results
H6a: Risk Perception→Subjective Norms→Perceptual Behavioral Control	Indirect Effects	0.141	0.024	0.285	0.014 *	A Partial Mediating Effect
	Direct Effect	0.262	0.049	0.526	0.019 *
	Total Effect	0.402	0.231	0.629	0.001 ***
H6b: Risk Perception→Sense of Responsibility→Perceptual Behavioral Control	Indirect Effect	0.061	−0.040	0.203	0.227	No Mediating Effect
	Direct Effects	0.329	0.135	0.562	0.001 ***
	Total Effect	0.390	0.231	0.581	0.001 ***
H6c: Risk Perception→Personal Norms→Perceptual Behavioral Control	Indirect Effect	0.047	0.009	0.123	0.008 **	A Partial Mediating Effect
	Direct Effects	0.361	0.187	0.608	0.001 ***
	Total Effect	0.408	0.225	0.648	0.001 ***
H6d: Sense of Responsibility→Personal Norms→Perceptual Behavioral Control	Indirect Effect	0.043	0.011	0.114	0.003 **	A Partial Mediating Effect
	Direct Effects	0.200	0.022	0.497	0.027 *
	Total Effect	0.243	0.054	0.565	0.012 *

Note: * means p < 0.05, ** means p < 0.01, and *** means p < 0.001.

. With subjective norms as the mediating variable in the path from risk perception to perceived behavior control, the confidence interval of the Bootstrap test for the total effect is (0.231, 0.629), the confidence interval of the Bootstrap test for the indirect effect is (0.024, 0.285), and the confidence interval of the Bootstrap test for the direct effect is (0.049, 0.526). In the above results, the confidence intervals of the total effect and the indirect effect do not contain 0, and the confidence interval of the direct effect does not contain 0, indicating that subjective norms have a partial mediating role between risk perception and perceived behavior control, and H6a is true. Therefore, it can be seen that risk perception can affect the perceived behavior control by influencing the mediating variable of subjective norms. The possible explanation is that when non-normal events occur, important social groups around residents, such as neighbors and friends, have a demonstration effect, which affects the residents’ perceived behavior control of food purchasing behavior.

With a sense of responsibility as the mediating variable in the path from risk perception to perceived behavior control, the confidence interval of the Bootstrap test for the total effect is (0.231, 0.581) and the confidence interval for the Bootstrap test for the indirect effect is (−0.040, 0.203). The results showed that the indirect effect contained 0 and the indirect effect was not significant, which indicated that there is no mediating role between risk perception and perceived behavior control, and H6b cannot be verified. The possible explanation is that the sense of responsibility comes from the individual’s responsibility for the family and the face of risks; even if the residents have a strong sense of family responsibility, they are subject to objective factors, which will not objectively affect the difficulty of residents to implement purchasing behavior, that is, it will not affect the objective factors that restrict residents from implementing purchasing behavior.

With personal norms as the mediating variable in the path from risk perception to perceived behavior control, the confidence interval of the Bootstrap test for the total effect is (0.225, 0.648), the confidence interval for the Bootstrap test for the indirect effect is (0.009, 0.123), and the confidence interval for the Bootstrap test for the direct effect is (0.187, 0.608). The confidence intervals of the total effect, indirect effect, and direct effect in the above results do not contain 0, which indicates that there is a partial mediating role of personal norms between risk perception and perceived behavior control, and H6c is true. This conclusion suggests that risk perception can affect perceived behavior control by influencing the mediating variable of personal norms. The possible reason is that when a non-normal event occurs, the residents’ perception of risk will stimulate the residents’ own sense of morality and responsibility. Out of their own sense of morality and responsibility, the stronger the residents’ perception of responsibility and obligation to buy food, the more they can perceive the urgency of buying food, the stronger the personal norms, and the stronger the perceived behavior control.

With personal norms as the mediating variable in the path from a sense of responsibility to perceived behavior control, the confidence interval of the Bootstrap test for the total effect is (0.054, 0.565), the confidence interval of the Bootstrap test for the indirect effect is (0.011, 0.114), and the confidence interval of the Bootstrap test for the direct effect is (0.022, 0.497). The confidence intervals of the total effect and indirect effect do not contain 0, and the confidence interval of the direct effect does not contain 0, indicating that there is a partial mediating role of personal norms between a sense of responsibility and perceived behavior control, and H6d is true. This conclusion suggests that a sense of responsibility can affect perceived behavior control by influencing the mediating variable of personal norms. The possible explanation is that the stronger the sense of responsibility, the more it can activate personal norms to implement altruistic behaviors, and when non-normal events occur out of their own sense of responsibility and morality, the stronger the residents’ sense of responsibility and personal norms for purchasing food, the more likely they are to carry out altruistic behaviors, and the stronger the perceived behavior control.

4.5. Robustness Test

In order to evaluate the reliability and validity of the analysis results, a robustness test has been carried out. In this paper, the least squares regression method with 5000 repetitions of bootstrapping and the feasible generalized least squares method is used, and the analysis results are shown in

Table 6. Robustness test with perceived behavior control as the dependent variable.

Independent Variables	OLS	OLS (Robust)	OLS (Bootstrap)	FGLS
Risk Perception	0.161 **	0.161 *	0.161 *	0.178 **
	−0.062	−0.067	−0.068	−0.062
Subjective Norms	0.204 ***	0.204 ***	0.204 ***	0.189 ***
	−0.055	−0.056	−0.056	−0.054
Sense of Responsibility	0.054	0.054	0.054	0.071
	−0.041	−0.045	−0.045	−0.043
Personal Norms	0.161 ***	0.161 ***	0.161 ***	0.146 ***
	−0.043	−0.048	−0.047	−0.043
_cons	1.852 ***	1.852 ***	1.852 ***	1.839
	−0.344	−0.39	−0.392	−0.338
N	548	548	548	548
F	21.874	16.423	-	22.906

Note: * means p < 0.05, ** means p < 0.01, *** means p < 0.001. Standard errors in the model are reported in parentheses.

and

Table 7. Robustness test with household reserve efficiency as the dependent variable.

Independent Variables	OLS	OLS (Robust)	OLS (Bootstrap)	FGLS
Perceptual Behavioral Control	0.124 ***	0.124 **	0.124 **	0.104 ***
	−0.031	−0.039	−0.039	−0.031
_cons	2.378 ***	2.378 ***	2.378 ***	2.34
	−0.254	−0.318	−0.318	−0.268
N	548	548	548	548
F	40.252	26.101	-	45.274

Note: ** means p < 0.01, *** means p < 0.001. Standard errors in the model are reported in parentheses.

. The analysis results show that the coefficients of the key variables all point in the same direction, and the statistical significance is basically the same. Therefore, the robustness test confirms the validity and reliability of the results of this study.

5. Conclusions

Previous studies have shown that public health emergencies, such as COVID-19, have a significant impact on residents’ food purchasing intentions [48]. Based on TPB and NAM, this paper constructs an analytical framework for residents’ risk perception, perceived behavior control, and family reserve efficacy under non-normal conditions. Studies have shown that the stronger the perceived behavior control of residents, the higher the efficiency of household reserves. The stronger the perceived behavior control of residents’ purchase of reserve food, the greater the possibility of purchasing reserve food and the greater the efficiency of household reserves. Under the non-normal situation, the perceived behavior control of the family-based purchase of food reserves reflects the purchasing efficiency of the family in food reserves and the anti-risk ability of the family reserves in the face of non-normal events. The perceived behavior control of household food reserve under non-normal conditions is affected by risk perception, subjective norm, and personal norm, and the degree of influence from high to low are risk perception, subjective norm, and personal norm, respectively. The improvement of the efficiency of household reserves depends on the mutual adaptation and coordination between perceived behavior control, risk perception, subjective norms, and personal norms, which is the effective unity of security, satisfaction, and responsibility. This conclusion is helpful for the establishment of a national strategic material reserve system based on a “national reserve system”, especially the family-based reserve system.

5.1. Theoretical Implications

(1) From the perspective of research based on the integrated TPB and NAM, this paper explores the influencing mechanism of urban residents’ household reserve efficiency in the non-normal context from the aspects of self-interest and altruism, rationality and morality and proposes measures and ways to improve the efficiency of household reserves. (2) In terms of research methodology, based on the structural equation model, the analytical framework of “non-normal events-risk perception-residents’ perceived behavior control-family reserve efficiency” was verified. The least squares regression method is used to verify the stability and reliability of the analysis conclusions. (3) In terms of research content, the concept of “family reserve efficiency” is innovatively proposed, which enriches the research on the efficiency of household emergency food reserves to a certain extent and has important theoretical and practical guiding significance for the construction of a unified and efficient “national reserve system”.

5.2. Policy Implications

(1)

Under the non-normal state, residents’ household food purchasing behavior is constrained by resources, opportunities, personal ability, and other factors, and the more controllable factors that residents have, the greater the possibility of realizing residents’ food purchasing behavior and the greater the efficiency of household reserves in perceived behavior control. Therefore, residents should be vigilant in peace time, strengthen the awareness of family food reserves, take precautions, reserve food and other means of living in an appropriate amount, and develop a good habit of family savings so as to improve the family’s ability to resist risks in non-normal situations, so as to improve the sustainability of the survival and development of family members;

(2)

When non-normal events occur, there are not only rational and self-interested motives but also emotional and altruistic motives in household food purchasing behavior. Subjective norms can play a mediating role between risk perception and perceived behavior control, which indicates that when non-normal events occur, not only can they perceive the risk, but also family and friends can perceive the existence of risk, which has an impact on family purchasing behavior. Therefore, residents often reserve too much food due to the influence of their family members and friends, which requires residents to strengthen their awareness of social responsibility, reduce the negative impact of subjective norms on residents, improve the ability to think rationally and screen information, correctly judge the possible harm caused by risks and the measures that should be taken, avoid excessive food reserves, avoid food waste, and promote sustainable environmental development;

(3)

The government should strengthen the publicity of the importance of family reserves and let residents know how to make family reserves more efficiently through the distribution of brochures so as to contribute to the construction of a unified and efficient “national reserve system”.

5.3. Limitations and Future Research Directions

There are some limitations in this study, which need to be further explored in subsequent studies. First, this study takes the COVID-19 lockdown as the entry point of the questionnaire survey to explore the impact mechanism of urban residents’ reserve efficiency under non-normal conditions, but there is a time gap between the questionnaire survey and the time of the lockdown, so the respondents’ memory may be slightly biased. Second, the COVID lockdown has passed, and attention has declined.