1 Introduction
The fundamental definition of commercial real estate encompasses all rental properties, particularly those led by shopping centers, including office buildings, hotels, rental apartments, and logistics properties. These properties are characterized by a long-term leasing strategy as their primary means of generating profit and returns. According to the international council of shopping centers (ICSC), a shopping center is a collection of retail and other commercial spaces managed uniformly but operated independently within a commercial collective.
The COVID-19 pandemic was confirmed at the end of 2019 and became prominent in January 2020. It quickly spread across the nation, significantly restricting physical and economic activities in cities worldwide
[1]. The pandemic posed severe challenges to normal life and public health globally. The overall economic uncertainty during the pandemic had a major impact on the retail sector, especially on businesses reliant on physical stores. This led to a significant decline in consumer spending and a reduced dependency on physical stores. Consumers increasingly turned to online channels for products unavailable offline, accelerating the shift from traditional markets to e-commerce
[2, 3]. Furthermore, with the closure of physical stores, online sales for omnichannel retailers increased. While the pandemic accelerated the transition to online retail, physical stores continue to play a crucial role in retail sales
[4, 5]. Shopping center managers have a vital role in the development of contemporary shopping centers
[6]. During the pandemic crisis, they must address two major issues: 1) how to manage emergency responses and maximize business recovery in the short term, and 2) how to grasp trend changes and systematically achieve stable development of the physical economy in the medium to long term
[7]. Identifying the correct emergency measures and response strategies is essential for physical retail to seize opportunities, over-come difficulties and crises, and achieve further transformation and upgrading
[8, 9].
The impact of the pandemic on the economy is evident. Although many scholars have studied the effects of the pandemic on retail supply chains, most research focuses on the operations of online retailers
[10, 11] and consumer shopping behavior
[12, 13]. To our knowledge, there remains a gap in research specifically addressing offline physical stores, with a lack of verifiable empirical evidence. Therefore, our study focuses on the impact of the pandemic on physical retail stores. Building on previous research
[10, 12, 14, 15], this study aims to address the following questions: 1) How does the pandemic affect shopping center rental rates and lease termination rates? 2) How do rental rates and lease termination rates vary with the severity of the pandemic? 3) What are the differences in the impact of the pandemic on retail rents and lease termination rates at various stages of the outbreak? 4) How does the pandemic impact rental rates and lease termination rates based on different types of rental agreements?
Using actual operational data from 7, 010 shopping center stores between 2020 and 2022 and incorporating data on the pandemic's impact in China, we analyze the overall effect of the pandemic on retail performance from both rental and tenant perspectives. This extensive and unique dataset provides a broader and more relevant research context. Our findings indicate: First, the pandemic has a significant negative impact on shopping center rental rates and lease termination rates; second, the negative impact is greater in more severely affected areas compared to less affected areas; third, the negative impact of the pandemic on shopping center rents and lease termination rates was more significant during the lockdown phase than during the overall pandemic period. Fourth, the pandemic exerts a more substantial negative impact on rental rates and lease termination rates for stores where rent is based on sales compared to those where rent is not linked to sales. The pandemic significantly reduces rental rates, while the likelihood of tenants terminating leases early due to the pandemic is relatively low. This study enriches the empirical research on physical stores and offers practical strategies for managers during crises, such as adjusting rental policies in response to the pandemic and optimizing rental structures. They may implement differentiated operational strategies based on the severity of the pandemic and its various stages. Additionally, establishing an emergency management mechanism and strengthening tenant relationship management are crucial steps. The pandemic also presents an opportunity for transforming and upgrading operational management practices.
The structure of the remaining sections is as follows: Section 2 reviews and synthesizes relevant literature; Section 3 outlines the theoretical framework and research hypotheses; Section 4 describes the data and methodological models used; Section 5 presents the empirical analysis and interpretation of regression results; Section 6 conducts robustness checks to ensure the reliability of the results; and Section 7 concludes with a summary of findings and recommendations for future research.
2 Literature Review
The literature review of this study is divided into two main sections: The first section examines the impact of the pandemic on retail businesses, while the second section analyzes the operational management strategies of retailers.
2.1 The Impact of the Pandemic on Retail
Research indicates that the pandemic has had profound and multifaceted effects on the retail industry. Asgari, et al.
[11] analyzed changes in consumer online shopping behavior post-pandemic, revealing that consumers with more online shopping experience were more likely to continue purchasing groceries online during this period. Shi and Goulias
[12], using data from the American time use survey (ATUS) from 2019 to 2022, explored the pandemic's lasting impact on individual time allocation, travel behavior, and purchasing habits, finding a significant decline in outdoor activities and travel enthusiasm among Americans, which had not returned to pre-pandemic levels by 2022.AbdulHussein, et al.
[16] noted that as restrictions on in-person shopping were implemented, consumers increasingly regarded e-commerce as the primary means of purchasing necessities, leading to shifts in online spending patterns. Mortimer, et al.
[17] discovered that in the post-COVID era, the utilitarian and transactional values of customer experiences significantly increased, while the hedonic and social interaction values diminished. Wei, et al.
[14] super examined the impact of pandemic factors on the online shopping behavior of urban residents in China, providing empirical insights for the transformation of urban retail spaces. Furthermore, Andruetto, et al.
[13] conducted multivariate statistical analyzes to investigate the shift from in-store to online shopping among consumers in Sweden and Italy during the pandemic, clarifying changes in shopping behaviors and strategies.
Overall, there has been a significant transformation in consumer shopping methods and behaviors before and after the pandemic. Previous studies have largely focused on the consumer and e-commerce perspectives, while this research emphasizes the operational management of retailers and brick-and-mortar stores, specifically examining the pandemic's effects on rental rates and lease termination rates.
2.2 Retail Management and Operations
In the field of retail operations management, enhancing the consumer shopping experience has become a focal point of research. Kim
[18] found that consumers' perceived value significantly influences their willingness to use location-based services (LBS), which in turn affects their shopping behaviors. Altug, et al.
[19] emphasized that retailers must identify strategies to stimulate demand and enhance customer experiences throughout the entire shopping process. Kuikka, et al.
[20] conducted a linguistic survey to analyze the factors driving customer loyalty and explored its relationship with loyalty itself.
Furthermore, Liu, et al.
[21] highlighted the transformative impact of e-commerce on the interaction patterns between sellers and consumers, noting that communication can positively influence customer engagement. Figueira, et al.
[22] demonstrated that consumers who place orders in advance provide strategic demand information, which helps reduce costs and improve customer service. Nakano
[23] showed that online demand is more concentrated on popular products compared to physical store demand, with the online shopping experience playing a moderating role in this process.
Gao and Su
[24] investigated an omnichannel service implementation strategy: Buy Online, Pick Up Store (BOPS), discovering that it can attract customer demand but may also increase return costs for retailers. Deshpande and Pendem
[25] examined the impact of logistics performance on third-party sellers within e-commerce platforms, while Roggeveen and Sethuraman
[15] highlighted the opportunities and challenges retailers face in enhancing consumer service, particularly regarding inventory and supply chain management.
Improving the consumer shopping experience is crucial, especially during crises. Retailers must accurately assess the pandemic's real effects on their operations and devise targeted response measures to enhance consumer experiences. This study fills a gap in the existing literature by conducting a heterogeneous analysis of shops based on the severity of the pandemic, its various stages, and different collaboration models, offering in-depth insights into the specific impacts of the pandemic on shopping center operational performance.
3 Theoretical Foundation and Research Hypotheses
3.1 Rent Theory
The evaluation of shopping center value is rooted in land rent theory, which emphasizes the differences in rental prices across various qualities and locations of land. Neoclassical synthesis posits that rent formation is influenced by the interplay of supply and demand. During the pandemic, changes in consumer shopping behavior led to corresponding adjustments in the demand and supply for in-store shopping, ultimately impacting rental levels. Modern rent theory suggests that market competition and asset price fluctuations directly affect rental rates, with variations in economic development and market conditions across different regions contributing to rental discrepancies. Given the economic uncertainty and short-term fluctuations experienced during the pandemic, rental prices for shopping center spaces have been impacted to some extent. Furthermore, shops operating under different collaboration models may encounter varying degrees of rental volatility. This theoretical foundation provides a framework for subsequent empirical analysis, enabling us to explore how the pandemic influenced rental prices and, consequently, the operational performance of shopping centers.
3.2 Tenant Theory
Leases for shopping centers can essentially be viewed as financial contracts, with their commercial value largely dependent on the operational capabilities of tenants and the appreciation of their intangible assets. The economic fluctuations brought about by the pandemic have significantly affected the performance of retail spaces, while the competence of shopping center managers directly influences tenants' willingness to renew their leases. To gain a deeper understanding of the pandemic's impact on the operational performance of shopping centers, subsequent empirical analyzes will examine the relationship between tenant turnover rates and the pandemic, particularly focusing on how the crisis has affected tenants' renewal intentions.
3.3 Impact of the Pandemic on Shopping Center Operational Performance
We analyze the impact of the pandemic on shopping center performance by examining key indicators such as store rent and lease termination rates. An empirical research model, as shown in
Figure 1, is developed to address four sets of hypotheses: The overall impact (H1a and H1b), the impact differentiated by severity (H2a and H2b), the impact of different phases of the pandemic (H3a and H3b), and the impact under different collaboration models (H4a and H4b).
Figure 1 Empirical research model |
Full size|PPT slide
3.3.1 Overall Impact
Based on rent theory, we argue that the pandemic has disrupted global consumer behavior, with the retail sector being particularly hard hit. The formation of rental prices is influenced by both supply and demand. During the pandemic, shifts in consumer behavior led to a decline in demand for in-person shopping, which negatively affected retail rents
[26-28]. Furthermore, modern rent theory indicates that market competition and economic uncertainty directly impact rental levels. Deteriorating investor expectations regarding future market conditions have exacerbated this effect. According to tenant theory, the operational status of retail spaces is closely linked to tenants' willingness to renew their leases. In light of the complex risks posed by the pandemic, tenants face greater challenges in assessing future market prospects and may be more inclined to maintain their current contracts
[29-31]. Therefore, we anticipate that the likelihood of tenants terminating their leases prematurely due to the pandemic is relatively low. This leads us to propose Hypothesis 1a and Hypothesis 1b.
Hypothesis 1a (H1a) The COVID-19 pandemic has a negative impact on shopping center store rents.
Hypothesis 1b (H1b) The COVID-19 pandemic has a negative impact on the lease termination rates in shopping center store.
3.3.2 Severity Differentiation Impact
According to rent theory, variations in economic development levels and market conditions across different regions can impact rental prices and tenant performance. Research by Han, et al.
[10] indicates that the intensity of the pandemic influences consumer shopping behavior, with higher levels of pandemic intensity significantly reducing e-commerce sales. Lee and Lee
[32] found that in communities with a higher number of cases, both non-resident inflow and retail spending experienced substantial declines. Additionally, Xu, et al.
[33] pointed out that the locations and frequency of grocery shopping among residents may also be affected by the severity of the pandemic. Tenant theory further suggests that in areas severely impacted by the pandemic, retail operations face greater challenges, which may lead to more pronounced negative effects on rental prices and lease termination rates. This leads us to propose Hypothesis 2a and Hypothesis 2b.
Hypothesis 2a (H2a) Compared to areas with relatively mild pandemic conditions, the COVID-19 pandemic has a more significant negative impact on shopping center store rents in areas with severe pandemic conditions.
Hypothesis 2b (H2b) Compared to areas with relatively mild pandemic conditions, the COVID-19 pandemic has a more significant negative impact on lease termination rates in shopping center store located in areas with severe pandemic conditions.
3.3.3 Impact of Different Stages of the Pandemic
According to modern rent theory, short-term fluctuations during the pandemic (such as lockdown measures) have a phase-specific impact on the market. Mortimer, et al.
[17] indicated that the values and experiences sought by customers in shopping malls vary across different stages of the pandemic. Han, et al.
[10] found that more stringent containment measures have a greater negative impact on e-commerce sales. The decline in the attractiveness of retail spaces during the lockdown phase leads to changes in rental prices and lease termination rates. Therefore, we anticipate that the lockdown phase of the pandemic will exert a more significant negative effect on shopping centers, leading us to propose Hypothesis 3a and Hypothesis 3b.
Hypothesis 3a (H3a) Compared to the overall pandemic period, the negative impact of COVID-19 during the lockdown phase on rental prices for shopping center retail spaces is greater.
Hypothesis 3b (H3b) Compared to the overall pandemic period, the negative impact of COVID-19 during the lockdown phase on lease termination rates for shopping center tenants is greater.
3.3.4 Impact of Different Collaboration Models
Rental and tenant theories suggest that different cooperation models may lead to variations in rental prices and lease termination rates. The pandemic has driven consumers toward online shopping, significantly impacting sales at physical retail locations. Gupta, et al.
[34] note that the pandemic has accelerated the shift to online shopping, with pure lease and non-pure lease models facing different rental pressures. Pure leasing refers to arrangements that consider only fixed rent, while non-pure leasing includes a portion of sales revenue as part of the rent structure. Therefore, we anticipate that the impact of the pandemic on rental prices and lease termination rates will vary across these cooperation models. Specifically, stores that base rent on sales are likely to experience a greater negative impact due to their income fluctuations being more closely tied to market conditions. This leads us to propose Hypothesis 4a and Hypothesis 4b.
Hypothesis 4a (H4a) The COVID-19 pandemic has a greater negative impact on rents for stores that consider sales as rent compared to stores that do not consider sales as rent.
Hypothesis 4b (H4b) The COVID-19 pandemic has a greater negative impact on lease termination rates for tenants considering sales as rent compared to tenants not considering sales as rent.
4 Research Design
4.1 Data Description
This study uses two sets of data: One related to the COVID-19 pandemic and the other comprising actual operational data from 7, 010 shopping center stores. The pandemic data was sourced from the Wind database, while the actual operational data for shopping center stores was provided by an information technology company. The timeframe for both datasets spans from January 23, 2020, to October 2022. Initially, we collected information on 10, 936 stores, however, we excluded 2, 190 records that were not relevant to the pandemic period and an additional 1, 736 records with missing information. The remaining datasets were matched and merged using Excel, resulting in a final dataset that includes 7, 010 shopping center stores. The dataset includes operational data from shopping centers across 20 prefecture-level cities in 12 provinces of China, representing both southern and northern regions, as well as first-tier and non-first-tier cities, and various business models.
We examined the impact of the pandemic (measured by Cases_Number) on the operational performance of shopping centers from two perspectives-rental prices (Rent) and lease termination rates (Terminate). To account for potential heterogeneity issues during the estimation process, we incorporated control variables into our analysis.
Table 1 defines the variables related to shopping center stores and the pandemic, while
Table 2 presents the descriptive statistics of the main variables.
Table 1 Variables and definition |
| Variables | Variable definition |
| Rent | Monthly transacted rent for store in the shopping center on day |
| Unit_Rent | Monthly transacted unit rents for store in the shopping center on day |
| Terminate | Whether the tenant of store in the shopping center terminated the lease early on day (1 for yes, 0 otherwise) |
| Cases_Number | Cumulative confirmed cases of COVID-19 on day |
| Area | Area of store in the shopping center |
| Floor | The floor of store in the shopping center |
| Sales | Whether sales are taken into account in the cooperative model of shopping center store (1 for yes, 0 otherwise) |
| Enterprise_Level | Whether the shopping center where store is Top 500 (1 for yes, 0 otherwise) |
| City_Economy | Whether the area where the shopping center store is located in a first-tier city (1 for yes, 0 otherwise) |
| City_Location | Whether the area where the shopping center store is located in a coastal city (1 for yes, 0 otherwise) |
| Business model for shopping center store (department, education, child, home, Catering, Clothing) |
Table 2 Descriptive statistics of variables |
| variable | | Mean | P50 | Sd | Min | Max |
| Rent | 7010 | 31011 | 13895 | 67596 | 30.43 | 3581023 |
| Unit_Rent | 7010 | 333.4 | 124.2 | 1471 | 1.325 | 40000 |
| Terminate | 7010 | 0.137 | 0 | 0.343 | 0 | 1 |
| Cases_Number | 7010 | 188.5 | 155 | 202.3 | 6 | 2462 |
| Area | 7010 | 206.3 | 102 | 1103 | 0.500 | 75789 |
| Floor | 7010 | 2.271 | 2 | 4.949 | 2 | 82 |
| Sales | 7010 | 0.696 | 1 | 0.460 | 0 | 1 |
| Enterprise_Level | 7010 | 0.734 | 1 | 0.442 | 0 | 1 |
| City_Economy | 7010 | 0.573 | 1 | 0.495 | 0 | 1 |
| City_Location | 7010 | 0.077 | 0 | 0.266 | 0 | 1 |
| Department | 7010 | 0.241 | 0 | 0.428 | 0 | 1 |
| Education | 7010 | 0.036 | 0 | 0.187 | 0 | 1 |
| Child | 7010 | 0.102 | 0 | 0.303 | 0 | 1 |
| Home | 7010 | 0.223 | 0 | 0.417 | 0 | 1 |
| Catering | 7010 | 0.186 | 0 | 0.389 | 0 | 1 |
| Clothing | 7010 | 0.168 | 0 | 0.374 | 0 | 1 |
4.2 Variable Specification
We apply logarithmic transformations to the dependent variables Rent and Unit_Rent, as well as to the independent variable Cases_Number and the control variable Area, to reduce data variability. Additionally, the control variable Floor is transformed using both logarithms and absolute values to mitigate differences between the data points.
4.2.1 Dependent Variables
Rent and Terminate represent the monthly rental income and tenant termination rate for shopping center stores, respectively. The termination rate is measured by whether a tenant ends their lease early. The variable Terminate is binary, with a value of 1 indicating early lease termination and a value of 0 indicating no early termination. Unit_Rent is used as a replacement for rent in robustness checks.
4.2.2 Independent Variables
Cases_Number is the independent variable representing the cumulative number of confirmed COVID-19 cases on a given day. Areas with more than 100 cumulative cases are classified as severely affected, while those with fewer than 100 cases are considered relatively mild affected.
4.2.3 Control Variables
Area and Floor are used to control for the store's size within the shopping center and its floor level. Sales is a dummy variable used to account for whether the store's cooperation model considers sales volume. City_Location and City_Economy are dummy variables controlling for the location and economic level of the city where the shopping center is situated. Enterprise_Level is a dummy variable used to control for the influence of the shopping center. Variable controls for the store's business model, categorized into six types: Department, Education, Child, Home, Catering, and Clothing.
4.3 Model Specification
To analyze the impact of the pandemic on the operational performance of shopping centers, we constructed two benchmark regression models using a multiple linear regression model and a Logit binary choice model, based on the research questions, data, and hypotheses presented in this study. These models are represented by Equations (1) and (2), where denotes the intercept term, represents the coefficients for the independent variables, – represents the coefficients for the control variables, and denotes the error term. We employed ordinary least squares (OLS) to perform the corresponding regression analyses.
Since the data is cross-sectional, we used Equation (1) to establish a multiple regression model to analyze the effect of the pandemic (Cases_Number) on store rents (Rent). Given that the dependent variable Terminate is binary, we constructed a Logit binary choice model using Equation (2) to examine the impact of the pandemic (Cases_Number) on tenant termination rates (Terminate). In accordance with Hypothesis 2, we performed grouped regression analyzes on the sample by dividing the independent variable Cases_Number into groups based on a threshold of 100. Following Hypothesis 3, we extracted the data from the lockdown period within the overall sample and conducted regressions using Equations (1) and (2). Additionally, in line with Hypothesis 4, we conducted grouped regression analyzes on the dummy variable Sales, partitioned into 0 and 1.
5 Empirical Results
To address the issue of heteroscedasticity that could result in biased regression outcomes due to inaccurate
-value estimates, we implemented robust standard errors in our Stata 16 regressions. Columns (1)
(6) in
Table 3 present the estimation results for Equation (1), while columns (1)
(6) in
Table 4 show the results for Equation (2).
Table 3 Estimated results for rents |
| | (1) | (2) | (3) | (4) | (5) | (6) |
| Variables | Rent | Rent | Rent | Rent | Rent | Rent |
| lnCases_Number | 0.234*** | 0.190*** | 0.104*** | 0.331*** | 0.310*** | 0.157*** |
| | (14.72) | (5.42) | (3.68) | (14.22) | (15.48) | (6.64) |
| lnArea | 0.673*** | 0.687*** | 0.651*** | 0.657*** | 0.585*** | 0.790*** |
| | (57.05) | (46.42) | (39.05) | (38.30) | (41.57) | (48.77) |
| lnFloor | 0.205*** | 0.127*** | 0.363*** | 0.207*** | 0.236*** | 0.198*** |
| | (13.58) | (6.30) | (16.87) | (10.37) | (12.19) | (7.88) |
| Sales | 0.630*** | 0.772*** | 0.462*** | 0.593*** | | |
| | (28.34) | (27.88) | (12.57) | (17.33) | | |
| Enterprise_Level | 0.397*** | 0.102** | 0.998*** | 0.279*** | 0.415*** | 0.224*** |
| | (10.44) | (1.99) | (17.64) | (4.91) | (6.49) | (4.50) |
| City_Economy | 0.879*** | 1.514*** | 0.754*** | 1.056*** | 0.880*** | 0.721*** |
| | (29.62) | (37.85) | (18.10) | (24.22) | (25.46) | (13.32) |
| City_Location | 1.154*** | 1.712*** | 0.881*** | 1.269*** | 1.187*** | 0.495*** |
| | (32.92) | (23.00) | (21.34) | (26.61) | (26.56) | (8.13) |
| Department | 0.229*** | 0.055 | 0.082 | 0.124* | 0.480*** | 0.118 |
| | (4.12) | (0.79) | (0.90) | (1.67) | (5.99) | (1.40) |
| Education | 0.514*** | 0.118 | 0.357*** | 0.374*** | 0.406*** | 0.792*** |
| | (7.20) | (1.36) | (3.56) | (3.66) | (4.32) | (7.83) |
| Child | 0.166*** | 0.259*** | 0.226** | 0.120 | 0.354*** | 0.202** |
| | (2.91) | (3.55) | (2.48) | (1.61) | (4.28) | (2.40) |
| Home | 1.035*** | 0.165* | 0.667*** | 1.096*** | 1.325*** | 0.312*** |
| | (19.02) | (1.90) | (7.23) | (14.84) | (17.27) | (2.91) |
| Catering | 0.567*** | 0.184*** | 0.541*** | 0.412*** | 0.797*** | 0.531*** |
| | (10.92) | (2.93) | (6.10) | (6.37) | (10.36) | (7.07) |
| Clothing | 0.449*** | 0.215*** | 0.365*** | 0.370*** | 0.692*** | 0.311*** |
| | (8.14) | (3.18) | (4.05) | (5.01) | (8.74) | (3.53) |
| Constant | 6.586*** | 6.327*** | 6.835*** | 7.037*** | 7.798*** | 5.632*** |
| | (66.95) | (38.36) | (31.26) | (52.31) | (66.34) | (38.31) |
| Observations | 7010 | 3773 | 3237 | 1288 | 4882 | 2128 |
| -squared | 0.698 | 0.830 | 0.647 | 0.733 | 0.658 | 0.673 |
| Notes: -values in parentheses, 0.1, 0.05, 0.01. |
Table 4 Estimated results for tenants |
| | (1) | (2) | (3) | (4) | (5) | (6) |
| Variables | Terminate | Terminate | Terminate | Terminate | Terminate | Terminate |
| lnCases_Number | 0.280*** | 1.106*** | 0.924*** | 0.299*** | 0.370*** | 0.199*** |
| | (5.41) | (8.47) | (7.63) | (6.16) | (4.50) | (2.65) |
| lnArea | 0.065* | 0.070 | 0.089 | 0.174*** | 0.104** | 0.075 |
| | (1.86) | (1.39) | (1.53) | (3.52) | (2.38) | (1.13) |
| lnFloor | 0.188*** | 0.277*** | 0.136 | 0.152* | 0.203** | 0.238** |
| | (3.22) | (2.84) | (1.37) | (1.85) | (2.56) | (2.43) |
| Sales | 0.055 | 0.692*** | 0.122 | 0.062 | | |
| | (0.62) | (4.70) | (0.80) | (0.47) | | |
| Enterprise_Level | 0.295* | 1.428*** | 1.269*** | 0.268 | 0.444 | 0.317 |
| | (1.85) | (7.30) | (4.80) | (1.31) | (1.48) | (1.63) |
| City_Economy | 0.399*** | 0.525** | 0.876*** | 0.379** | 0.095 | 0.378* |
| | (3.59) | (2.25) | (5.71) | (2.24) | (0.63) | (1.96) |
| City_Location | 0.109 | 0.518* | 0.659*** | 0.138 | 0.277 | 1.297 |
| | (0.73) | (1.81) | (3.03) | (0.68) | (1.48) | (1.28) |
| Department | 1.278*** | 0.426* | 1.252*** | 0.868*** | 1.791*** | 0.024 |
| | (6.43) | (1.64) | (3.51) | (3.44) | (6.27) | (0.06) |
| Education | 0.403* | 0.289 | 0.435 | 0.066 | 1.137*** | 0.775** |
| | (1.68) | (0.81) | (1.09) | (0.19) | (3.12) | (2.00) |
| Child | 0.092 | 0.318 | 0.268 | 0.349 | 0.711** | 0.957*** |
| | (0.48) | (1.10) | (0.77) | (1.41) | (2.49) | (2.63) |
| Home | 0.408* | 1.743*** | 0.087 | 1.148*** | 1.327*** | 2.099*** |
| | (1.95) | (4.92) | (0.24) | (3.69) | (4.65) | (5.32) |
| Catering | 0.214 | 0.066 | 0.050 | 0.033 | 1.020*** | 1.123*** |
| | (1.21) | (0.30) | (0.15) | (0.15) | (3.85) | (3.36) |
| Clothing | 1.063*** | 0.206 | 1.320*** | 0.390 | 1.660*** | 0.031 |
| | (5.38) | (0.79) | (3.69) | (1.55) | (5.99) | (0.08) |
| Constant | 0.246 | 3.472*** | 3.476*** | 1.115** | 1.201*** | 1.540*** |
| | (0.75) | (5.85) | (3.72) | (2.31) | (2.80) | (2.61) |
| Observations | 7, 010 | 3, 773 | 3, 237 | 1, 288 | 4, 882 | 2, 128 |
| Notes: -values in parentheses, 0.1, 0.05, 0.01. |
Column (1) in
Table 3 indicates that the coefficient for the independent variable (
0.234,
0.000) is negative and statistically significant. Specifically, an increase in the cumulative number of confirmed COVID-19 cases significantly reduces store rent, negatively impacting stores in shopping centers. This supports Hypothesis 1a, thus validating it. Regarding tenant early terminations, the data shows that 957 stores terminated their leases early, accounting for 14% of the total sample, while the remaining 6, 053 stores did not, representing 86% of the total sample. The result in column (1) of
Table 4 shows that the coefficient for the independent variable (
0.280,
0.000) is negative and significant. This suggests that the likelihood of early lease terminations by tenants is low as the number of cumulative confirmed cases increases, thereby validating Hypothesis 1b.
By grouping the independent variable Cases_Number in Equation (1) using a threshold of 100 cases, we have 3, 773 stores in areas with more than 100 cases and 3, 237 stores in areas with fewer than 100 cases. We conducted a seemingly unrelated regressions (SUR) test for differences in coefficients between these groups, obtaining a significant
-value (0.000), which leads us to reject the null Hypothesis and thus confirm that the coefficients can be compared. Columns (2) and (3) in
Table 3 display the estimated results for rent after grouping. The coefficients for the independent variable are negative and significant in both cases. Specifically, the coefficient for areas with more than 100 cases (
0.190,
0.000) is less than that for areas with fewer than 100 cases (
0.104,
0.000). This indicates that the impact of the pandemic on stores rent varies by severity, with a higher negative impact in more severely affected areas, thus validating Hypothesis 2a.
Columns (2) and (3) in
Table 4 present the results for tenant early terminations after grouping. The coefficients for the independent variable are negative and significant. The coefficient for areas with more than 100 cases (
1.106,
0.000) is lower than that for areas with fewer than 100 cases (
0.924,
0.000). This shows that the impact of the pandemic on lease termination rates varies with severity, with a higher negative effect in more severely affected areas, thus validating Hypothesis 2b.
To examine the impacts at different stages of the pandemic, we selected data from 1, 288 stores during the lockdown phase based on publicly available information, allowing for a comparative analysis against the overall effects of the pandemic. We used a Seemingly Unrelated Regressions (SUR) test to compare the coefficients between these groups, yielding a significant
-value (0.000), leading us to reject the null Hypothesis and confirm that the coefficients are comparable. The results show that the coefficients for the independent variables are all negative and statistically significant. Column (4) of
Table 3 presents the estimated effects of the lockdown phase on rental prices, revealing that the coefficient during this phase (
0.331,
0.000) is lower than that of the overall pandemic phase (
0.234,
0.000). This finding supports Hypothesis 3a, confirming its validity.
Additionally, Column (4) of
Table 4 illustrates the estimated effects on tenants during the lockdown phase. The coefficient for the lockdown phase (
0.299,
0.000) is also lower than that for the overall pandemic phase (
0.280,
0.000). Specifically, the negative impacts on rental prices and lease termination rates during the lockdown are greater than those observed during the overall pandemic phase, thus validating Hypothesis 3b and confirming its legitimacy.
By grouping the control variable Sales in Equation (2) into two categories, those that consider sales as part of the rent and those that do not. We have 4, 882 stores that consider sales in their rent and 2, 128 that do not. We used a seemingly unrelated regressions (SUR) test to compare the coefficients between these groups, yielding a significant
-value (0.000), leading us to reject the null Hypothesis and confirm that the coefficients are comparable. Columns (5) and (6) in
Table 3 present the estimated results for rent after grouping. The coefficients are negative and significant in both categories. The coefficient for stores that include sales in their rent (
0.310,
0.000) is less than that for stores that do not (
0.157,
0.000). This indicates that the negative impact of the pandemic on rent is greater for stores that consider sales in their rent, thus validating Hypothesis 4a.
Columns (5) and (6) in
Table 4 show the estimated results for tenant early terminations after grouping. The coefficients for the independent variable are negative and significant. The coefficient for stores that include sales in their rent (
0.370,
0.000) is less than that for stores that do not (
0.199,
0.000). This implies that the pandemic's negative impact on lease termination rates is greater for stores that factor in sales as part of the rent, thus validating Hypothesis 4b.
6 Robustness Test
6.1 Using Unit Rent as the Dependent Variable
In the baseline regression model, we used Rent as the dependent variable and controlled for the size of the shopping center stores using the control variable Area. For the robustness check of Equation (1), we replaced the dependent variable in the baseline regression model with Unit_Rent, also taking its logarithm, and performed the regression after removing the control variable Area. We established a new regression model as shown in Equation (3).
The results from Equation (3) are presented in columns (1)
(6) of
Table 5, where the coefficients of the independent variables remain negative and statistically significant. The coefficient for regions with more severe pandemics is lower compared to regions with milder conditions, the coefficients of the independent variables during the lockdown phase are lower than those during the overall pandemic phase, and the coefficient for stores considering sales volume as rent is lower than for those that do not. These results are consistent with previous findings, further validating Hypotheses 1a, 2a, 3a, and 4a, and confirming the robustness of the results.
Table 5 Results of robustness checks on rents |
| | (1) | (2) | (3) | (4) | (5) | (6) |
| Variables | Unit_Rent | Unit_Rent | Unit_Rent | Unit_Rent | Unit_Rent | Unit_Rent |
| lnCases_Number | 0.218*** | 0.123*** | 0.085*** | 0.320*** | 0.287*** | 0.154*** |
| | (12.99) | (3.48) | (2.81) | (12.38) | (14.42) | (8.37) |
| lnArea | 0.289*** | 0.229*** | 0.418*** | 0.305*** | 0.350*** | 0.221*** |
| | (16.60) | (9.54) | (17.02) | (12.58) | (16.59) | (8.64) |
| lnFloor | 0.633*** | 0.684*** | 0.490*** | 0.644*** | | |
| | (27.19) | (24.30) | (12.55) | (18.47) | | |
| Sales | 0.536*** | 0.093 | 1.143*** | 0.420*** | 0.579*** | 0.303*** |
| | (13.54) | (1.62) | (19.20) | (7.10) | (10.58) | (5.48) |
| Enterprise_Level | 0.823*** | 1.571*** | 0.728*** | 1.032*** | 0.850*** | 0.691*** |
| | (25.75) | (32.98) | (15.97) | (21.19) | (22.30) | (15.58) |
| City_Economy | 1.097*** | 1.192*** | 0.925*** | 1.205*** | 1.143*** | 0.620*** |
| | (30.91) | (13.55) | (21.76) | (24.44) | (24.61) | (6.28) |
| City_Location | 0.601*** | 0.382*** | 0.607*** | 0.386*** | 0.675*** | 0.537*** |
| | (11.06) | (4.85) | (7.07) | (5.25) | (8.53) | (7.34) |
| Department | 0.750*** | 0.370*** | 0.731*** | 0.515*** | 0.531*** | 1.036*** |
| | (10.21) | (3.86) | (7.38) | (4.70) | (5.14) | (11.40) |
| Education | 0.544*** | 0.084 | 0.796*** | 0.380*** | 0.608*** | 0.557*** |
| | (9.47) | (0.99) | (9.16) | (4.91) | (7.30) | (7.01) |
| Child | 1.287*** | 0.478*** | 1.110*** | 1.229*** | 1.389*** | 0.672*** |
| | (23.14) | (4.75) | (12.60) | (16.06) | (17.85) | (7.01) |
| Home | 0.893*** | 0.573*** | 1.025*** | 0.640*** | 0.981*** | 0.869*** |
| | (17.02) | (7.68) | (11.94) | (9.63) | (12.84) | (12.03) |
| Catering | 0.796*** | 0.644*** | 0.867*** | 0.587*** | 0.864*** | 0.731*** |
| | (14.68) | (8.24) | (10.09) | (8.02) | (10.99) | (9.07) |
| Clothing | 4.863*** | 4.330*** | 4.772*** | 5.334*** | 5.819*** | 4.386*** |
| | (58.52) | (29.98) | (23.59) | (44.57) | (54.71) | (43.31) |
| Constant | 4.863*** | 4.772*** | 4.330*** | 5.334*** | 5.819*** | 4.386*** |
| | (58.52) | (23.59) | (29.98) | (44.57) | (54.71) | (43.31) |
| Observations | 7, 010 | 3, 773 | 3, 237 | 3, 049 | 4, 882 | 2, 128 |
| -squared | 0.552 | 0.668 | 0.565 | 0.605 | 0.506 | 0.214 |
| Notes: -values in parentheses, 0.1, 0.05, 0.01. |
6.2 Regression Using Subsamples
The full sample was used for estimating Equation (2). To avoid the possibility that tenants with expiring con-tracts might choose not to terminate early, we excluded the 2, 758 samples with contract end dates before January 1, 2023. The regression was then conducted on a subsample of 4, 252 stores, with results shown in column (1) of
Table 6. The coefficient of the independent variable remains negative and statistically significant, further validating Hypothesis 1b.
Table 6 Results of robustness checks on tenants |
| | (1) | (2) | (3) | (4) | (5) | (6) | (7) |
| Variables | Terminate | Terminate | Terminate | Terminate | Terminate | Terminate | Terminate |
| lnCases_Number | 0.353*** | 0.157*** | 0.619*** | 0.507*** | 0.443*** | 0.203*** | 0.113*** |
| | (5.73) | (5.59) | (8.32) | (8.15) | (5.04) | (4.76) | (2.74) |
| lnArea | 0.047 | 0.027 | 0.024 | 0.045 | 0.088*** | 0.047** | 0.055 |
| | (0.99) | (1.43) | (0.85) | (1.44) | (3.18) | (1.98) | (1.52) |
| lnFloor | 0.169** | 0.094*** | 0.140*** | 0.066 | 0.073 | 0.095** | 0.125** |
| | (2.36) | (2.89) | (2.64) | (1.27) | (1.58) | (2.21) | (2.29) |
| Sales | 0.215** | 0.025 | 0.359*** | 0.030 | 0.037 | | |
| | (1.97) | (0.53) | (4.63) | (0.36) | (0.53) | | |
| Enterprise_Level | 0.152 | 0.174** | 0.813*** | 0.617*** | 0.117 | 0.253* | 0.210* |
| | (0.72) | (2.18) | (7.15) | (4.77) | (1.08) | (1.91) | (1.88) |
| City_Economy | 0.387*** | 0.206*** | 0.289** | 0.476*** | 0.236*** | 0.054 | 0.168 |
| | (2.90) | (3.48) | (2.35) | (5.67) | (2.61) | (0.68) | (1.61) |
| City_Location | 0.705*** | 0.079 | 0.269 | 0.344*** | 0.061 | 0.156 | 0.632 |
| | (0.99) | (1.58) | (3.31) | (0.55) | (1.60) | (1.58) | |
| Department | 0.572** | 0.697*** | 0.197 | 0.678*** | 0.494*** | 0.994*** | 0.028 |
| | (1.99) | (6.44) | (1.31) | (3.61) | (3.48) | (6.55) | (0.14) |
| Education | 0.003 | 0.238* | 0.187 | 0.263 | 0.028 | 0.647*** | 0.428** |
| | (0.01) | (1.78) | (0.92) | (1.24) | (0.14) | (3.30) | (2.05) |
| Child | 0.318 | 0.061 | 0.207 | 0.157 | 0.195 | 0.406*** | 0.554*** |
| | (1.13) | (0.56) | (1.23) | (0.84) | (1.35) | (2.61) | (2.86) |
| Home | 0.097 | 0.247** | 0.936*** | 0.009 | 0.581*** | 0.730*** | 1.241*** |
| | (0.27) | (2.22) | (4.74) | (0.04) | (3.66) | (4.86) | (5.74) |
| Catering | 0.167 | 0.119 | 0.008 | 0.022 | 0.004 | 0.565*** | 0.650*** |
| | (0.62) | (1.19) | (0.06) | (0.12) | (0.03) | (3.95) | (3.66) |
| Clothing | 0.571* | 0.594*** | 0.082 | 0.734*** | 0.234 | 0.932*** | 0.064 |
| | (1.95) | (5.44) | (0.54) | (3.87) | (1.62) | (6.25) | (0.30) |
| Constant | 0.317 | 0.040 | 1.832*** | 1.944*** | 0.460* | 0.558** | 0.967*** |
| | (0.66) | (0.22) | (5.32) | (4.15) | (1.74) | (2.40) | (3.04) |
| Observations | 4, 252 | 7, 010 | 3, 773 | 3, 237 | 1, 288 | 4, 882 | 2, 128 |
| Notes: -values in parentheses, 0.1, 0.05, 0.01. |
6.3 Verification Using Alternative Models
Since the dependent variable is binary, we employed a Probit model to test the robustness of Equation (2), with results shown in columns (2)
(7) of
Table 6. The coefficients of the independent variables remain negative and significant. The coefficient for more severely affected regions is lower compared to less affected regions, the coefficients of the independent variables during the lockdown phase are lower than those during the overall pandemic phase, and the coefficient for stores that consider sales volume as rent is lower than for those that do not. These results are consistent with previous findings, further validating Hypotheses 1b, 2b, 3b, and 4b, and reinforcing the robustness of our regression approach.
7 Discussion
7.1 Conclusions
Our findings indicate that the pandemic has had a negative impact on both shop rental rates and lease termination rates, with the effects being more severe in regions that experienced greater pandemic intensity. The negative impacts of the pandemic are more pronounced during the lockdown phase. We also found that the pandemic's adverse effects on rental rates and lease terminations were more pronounced for shops where rent is linked to sales compared to those where rent is not sales-based. The pandemic significantly reduced shop rental rates, and the likelihood of tenants terminating leases early due to the pandemic is relatively low.
The innovation of this study lies in its provision of new empirical evidence from the perspective of brick-and-mortar retail. Unlike previous research that predominantly focuses on the responses of e-commerce, this study specifically analyzes the direct impact of the pandemic on the operational performance of physical stores, thereby filling a gap in the existing literature. By incorporating rental prices and lease termination rates as indicators of shopping center performance, we systematically explore the effects of the pandemic on physical retail. Furthermore, we conduct a heterogeneity analysis from multiple dimensions, revealing differences in the impact of pandemic severity, various phases of the pandemic, and different collaboration models on rental prices and ten-ants. This provides significant empirical support for relevant theoretical research.
7.2 Managerial Implications
First, adjust rental policies to address the impacts of the pandemic. Research indicates that the pandemic has had a greater effect on sales-based rent structures. Therefore, it is recommended that shopping center managers consider implementing flexible rental arrangements, such as hybrid models, to mitigate the impact on rental income
[8]. Additionally, during periods of severe pandemic conditions, implementing rent reductions or concessions can help maintain tenant stability
[35, 36].
Second, establish effective emergency management mechanisms to enhance tenant relationship management. Although the risk of early lease termination by tenants is relatively low during the pandemic, managers should still develop comprehensive emergency management plans to bolster their response capabilities. Proactively communicating with tenants and providing supportive measures can enhance their trust and satisfaction, ensuring the stability of shopping center operations.
Third, implement differentiated operational strategies. Studies have found that the pandemic has varying impacts across different regions and stages. Managers should conduct detailed risk assessments based on regional characteristics and adopt appropriate measures to achieve optimal resource allocation and risk management. In areas severely affected by the pandemic and during lockdown phases, greater rental reductions and support may be necessary, while in less affected regions and other stages, support can be scaled back accordingly.
Fourth, leverage the pandemic as an opportunity to drive operational transformation and upgrade management practices. In light of the challenges posed by the pandemic and the rise of new retail formats, retailers should accelerate their digital transformation efforts and explore omnichannel innovations to adapt to the rapidly changing market environment
[37-40]. Developing sustainable growth strategies and optimizing the consumer experience will be crucial for the future success of retailers
[41-45].
7.3 Limitations and Future Research
This study has several limitations. First, due to data constraints, the analysis focused solely on rental rates and lease termination rates, without adequately considering factors such as foot traffic. Future research could incorporate a more comprehensive set of operational performance indicators. Second, this study conducted a heterogeneity analysis from only three perspectives, lacking an exploration of the differences in e-commerce infrastructure across various regions. Subsequent studies could delve deeper into this impact. Lastly, given the disruptions in logistics during the pandemic, investigating how logistics affect online shopping and its implications for the operational performance of physical stores would be an intriguing area for future research.
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
PDF(196 KB)
Figure 1 Empirical research model
Table 1 Variables and definition
