Increase in Households Triggered by Accommodation Closure Due to the COVID-19 Pandemic in the Historical Center of Kyoto City

Abstract

The COVID-19 pandemic has forced many accommodations to close. However, the pandemic might play an important role in providing an opportunity to achieve sustainable tourism with a good balance between housing for residents and accommodation for tourists. As the theoretical framework, this study aims to investigate the change in households triggered by accommodation closure due to the COVID-19 pandemic in Kyoto City’s historical center. Furthermore, the causes of these changes were examined by analyzing the real estate properties traded on the market. For the analysis, this study considered the COVID-19 pandemic as a natural experiment to investigate the causal relationship between the number of households, closed accommodations, and real estate properties. As a result, it was found that households increased by approximately 1.34 in neighborhood associations with closed simple accommodations. Regarding the causes of the increase, closed simple accommodation properties tend to change to short-term rentals. This study also highlighted that closed simple accommodations have significantly smaller room sizes than other property types, with only slightly higher prices. As a theoretical contribution, our findings suggest that the pandemic might have suppressed tourism gentrification, but increased the number of households.

1. Introduction

The tourism industry was severely damaged by the COVID-19 pandemic [1]. In Japan, tourism shrank, and tourism businesses closed in the early stages of the pandemic [2], including most accommodations [3]. Some of these closed accommodations were reportedly distributed in the housing market as short-term rentals [4]. For example, short-term rentals could be transformed into houses for working from home in the context of the post-pandemic labor market [5]. In addition, in the case of Beijing, the tourism gentrification before the pandemic was reported to be associated with the recovery of Airbnb listings after the pandemic, with fewer listings remaining in gentrified neighborhoods [6]. In Italy, the population living in short-term rentals was found to transform not only neighborhoods but also entire cities [7]. These results suggest that the areas that experienced tourism gentrification before the pandemic may have increased in population during the pandemic due to the shift from accommodation to short-term rentals. The increase suggests that the pandemic might have played an important role in providing an opportunity to achieve sustainable tourism with a good balance between housing for residents and accommodation for tourists [8]. The possibility means that the pandemic may have unexpectedly shown us a way to control the negative aspects of tourism. However, a research gap exists in the correlation between the number of closed accommodations and that of households during the pandemic.

This study aims to clarify the changes in the number of households triggered by accommodation closure during the pandemic in Kyoto City’s historical center. In addition, the causes of these changes were examined, based on real estate transactions in the market. As the theoretical framework, this study considered the COVID-19 pandemic as a natural experiment. A natural experiment is an observational study that takes advantage of a situation that allows for the seemingly random assignment of subjects to different groups for a given exposure. The natural experiment contributed practically to investigating the causal relationship between closed accommodations and households. For the natural experiment in this study, Kyoto City’s historical center was chosen as a study case because it is one of the most popular tourist cities in the world. Kyoto City’s historical center is a famous tourist area that has over 1200 years of history [9]. There are many famous shrines and temples in Kyoto City’s historical center. They are registered as World Heritage Sites. This study adopted the pandemic period from 31 March 2020 to 31 March 2022. In April 2020, the Japanese government declared several states of emergency [10]. However, after April 2022, the Kyoto City Government gradually eased restrictions on daily activities. After deregulation, the government gradually revived the tourism industry. For example, after April 2022, the government promoted the “Kyoto Charm Rediscovery Travel Project” to provide subsidies for accommodation [11].

The neighborhood association (NA) scale was used for the analysis. The NA scale is the basis for community government, including childcare, elderly care, and disaster prevention, in Kyoto City [12,13]. NAs are the same as the postal codes in Kyoto City’s historical center. This study considered the pandemic as a natural experiment to analyze causal relationships [14,15]. As part of the natural experiment, this study compared the changes in households in NAs with and without closed accommodations. More specifically, this study set NAs with closed accommodations as the exposure group and those without closed accommodations as the non-exposure group. In addition, in this study, the outcomes were set as changes in the number of households. The results may help achieve sustainable tourism with a good balance between housing and accommodation in the post-pandemic period.

2. Literature Review

2.1. Accommodations Types

This study analyzed simple accommodations (SAs) and hotels, as they represent the most frequent types of accommodation in Kyoto City’s historical center [12]. Figure 1 shows photographs of SAs and hotels. This classification follows the Japanese Hotel Business Law [16]. SAs operate facilities that are primarily composed of common areas and shared equipment or furnishings. They also receive fees and provide accommodation for individuals [16]. Compared to hotels, SAs offer simpler services and facilities that are relatively inexpensive. In Japan, SAs are mainly used in peer-to-peer (P2P) digital platforms, including Airbnb. In Kyoto City, SAs have been studied as a cause of population decline, based on tourism gentrification [12,13]. Therefore, this study analyzed SAs and hotels as accommodation types, in accordance with Japanese law.

P2P accommodations, collectively known as the sharing economy, have emerged as alternative providers of the services that are traditionally provided by long-established industries [17]. P2P accommodations act as adjustment valves, balancing the demands for housing and accommodation. In addition, P2P accommodation differs significantly from the traditional hotel industry as it provides the opportunity to “live like a local” [18]. These studies indicate the various benefits of P2P accommodation that the traditional tourism industry could not provide.

Researchers have studied the empirical impact of P2P accommodation on neighborhoods in major tourist cities [19]. For example, in Berlin, only a few neighborhoods have large Airbnb markets, even though the Airbnb market is mainly located in the center [20]. In Barcelona, P2P accommodations expand the tourism pressure over residential areas in the center [21]. Neighborhood residents assessed the dominance of negative perceptions of the socio-economic impacts [22]. As these results show, it is necessary to analyze not only the benefits of P2P accommodation but also its socio-economic impact on the neighborhood. In addition to the socio-economic impact, tourism gentrification is one of the social issues that must be solved by urban planning in tourist cities. Therefore, the originality of this study is to clarify the social impact of tourism gentrification during the pandemic, focusing on SAs and hotels.

2.2. P2P Accommodation and Tourism Gentrification

The term “tourism gentrification” describes the process by which middle-class neighborhoods transform into relatively affluent and exclusive enclaves that are characterized by the proliferation of corporate entertainment and tourism facilities [23]. However, these days, it has been assumed that new forms of tourism, such as P2P accommodation, are thought to be the drivers of tourism gentrification [6,24,25]. The theory of tourism gentrification differs from that of ordinary gentrification, based on the rent gap [26] or the middle-class settlement [27]. The theory of tourism gentrification is related to the social changes that destinations undergo because of tourism-oriented service pressures, leading to the displacement of residents [28]. Based on the theory, tourism gentrification is thought to be an urban phenomenon in which tourism and housing are unbalanced beyond the tourism-carrying capacity.

Tourism gentrification imposes externalities at the neighborhood scale [29]. For example, tourism gentrification has financial, functional, and social impacts on neighborhood environments [30,31,32]. The social impacts of tourism gentrification include not only noise, dirt, the occupation of public spaces, and the fear felt by residents [33] but also displacement as a form of social injustice [25,34,35], which is verified by the theory of tourism gentrification. The displacement leaves the local community feeling dispossessed, angry, and frustrated [36]. However, the most severe problem is the population decline caused by displacement [5,31,37]. Therefore, before the COVID-19 pandemic, tourism gentrification was a problem in tourist cities worldwide, especially in historical centers [38,39,40,41,42]. Unlike these findings, this study hypothesizes that the pandemic may have eased the pressure for displacement by tourism gentrification during the pandemic. This study fills the research gap between the population decline due to tourism gentrification before and during the pandemic.

2.3. Tourism Gentrification in Kyoto City’s Historical Center

This study analyzed the case of Kyoto City’s historical center. Kyoto City is one of the most famous tourist cities and attracts visitors from all over the world. Kyoto’s advantages are its history and culture, which have been cultivated by its local residents for over 1200 years. In order to experience traditional culture, before the pandemic in 2019, approximately 13 million tourists stayed overnight in Kyoto City [43]. Meanwhile, during the pandemic in 2020 and 2021, only 5 million tourists stayed overnight in Kyoto City [43]. In particular, the number of visitors from abroad declined significantly, from 3.8 million in 2019 to 50 thousand in 2022 [43]. These changes suggest that the pandemic significantly decreased tourism demands, similar to other tourist cities [1,2]. The pandemic affected not only the tourism sector but also the housing sector. The Housing and Land Survey indicates that the vacancy rate in Kyoto City was 13.53% in 2018, before the pandemic, but improved to 13.18% in 2023, after the pandemic [44]. The results show the possibility of a correlation between households and accommodation triggered by the accommodation closures during the pandemic.

Before the COVID-19 pandemic, social issues of tourism gentrification were reported in Kyoto City’s historical center, like other tourist cities. The increase in the number of SAs caused a decrease in the population of Kyoto City’s historical center [12]. This population decline was due to tourism gentrification, caused by the displacement from housing to SAs [13]. In the central area, tourism gentrification was found to be correlated with the housing affordability risk of the private housing market [45]. Therefore, local communities in the historical center negatively evaluated the displacement due to tourism gentrification [46]. Some local communities struggled to cope with it by developing local guidelines [47]. These studies highlight the importance of the historical center as a major issue in the tourism gentrification of Kyoto City’s historical center. However, the analysis period of these studies in Kyoto City was from 2015 to 2019, which was before the COVID-19 pandemic. This means that there is a lack of studies on accommodations being closed due to the COVID-19 pandemic outbreak in Kyoto City, as well as in tourist cities worldwide. Therefore, the novelty of this study is to clarify the change in households triggered by accommodation closure due to the COVID-19 pandemic.

3. Methods

3.1. Case Study

This study analyzed the case of Kyoto City’s historical center. Figure 2 shows the location of Kyoto City’s historical center in East Asia. The historical center includes Kamigyo, Nakagyo, and Shimogyo wards. The area has an urban structure that is based on neighborhood blocks that have existed for over 1200 years. In the historical center area, many famous temples and shrines are registered as World Heritage Sites [48]. This is a feature of the historic center of Kyoto that is not found in other areas. Therefore, many people from all over the world come to visit [43]. Meanwhile, the historical center has suffered from tourism gentrification [12,13,45,46,47]. Therefore, this study analyzed the case of Kyoto City’s historical center during the pandemic.

3.2. Analysis Flow

Figure 3 shows this study’s analysis flow. The study used the accommodation list as accommodation data, the basic resident register data as household data, and the At Home Dataset as real estate data.

In Figure 3, Step 1 identifies closed SAs and hotels that closed from the end of March 2020 to the end of March 2022. Specifically, Step 1 uses the business license date variable of the accommodation list. Step 1 provides the accommodation data of accommodation that was licensed to operate between the end of March 2020 and the end of March 2022. These accommodations were categorized as SAs and hotels. Step 1 identifies accommodations that were removed from the list in March 2020 and March 2021. The excluded data included closed SAs and hotels. In addition, Step 1 identifies the NAs where closed SAs and hotels were located.

Step 2 analyses household changes in the NAs where closed accommodations were located. Specifically, household data were used to calculate household changes in each NA from the end of March 2020 to the end of March 2022. Step 2 also identifies the NAs where closed SAs and hotels were located, as identified in Step 1. An analysis of variance (ANOVA) of household changes was conducted for NAs with and without closed accommodations. As a natural experiment, ANOVA analyzes the causal relationship between the outcome (households) and the predictors (closed SAs and hotels). The ANOVA test was conducted separately for SAs and hotels. First, this study conducted the Shapiro–Wilk test to test the normal distribution of the outcome variable. Then, the Wilcoxon rank sum test was used to analyze significant differences. The Wilcoxon rank sum test is a non-parametric statistical test that is used to compare two independent samples to determine whether they come from the same distribution. This test is particularly useful when the data does not follow a normal distribution. That means the Wilcoxon rank sum test is more robust against non-normal distributions. Regarding the analysis, the significance level was set at p-value < 0.05 and 0.01.

Step 3 clarifies the characteristics of closed accommodations in the private real estate market. For closed accommodations, the type of accommodation was found to be significantly different in Step 2. First, closed accommodations were identified from the address variables in the real estate data. Based on the identification, the real estate data were divided into closed accommodation properties and other real estate. Step 3 analyzes the cross-tabulations of property type, price, and room area for these two types. Property types were categorized as apartments for rent, apartments for sale, and mixed-use properties for rent. Prices were expressed in JPY and USD. For prices in USD, we used the rate of 31 March 2022, the last day of the analysis period [49]. In addition, this study conducted the Shapiro–Wilk test to test the normal distribution regarding price and room area. Then, the Wilcoxon rank sum test was used to analyze significant differences. The significance level was set at p-value < 0.05 and 0.01.

3.3. Dataset

3.3.1. Accommodation Data

The accommodation data were obtained from the accommodation list, which is licensed under the Hotel Business Law. This study used the accommodation list as of 31 March 2020, 31 March 2021, and 31 March 2022. The list contains open data that are published by the Kyoto City government [50]. This dataset is the only official list of accommodations in Kyoto City. According to the Hotel Business Law, all the operators of accommodation must apply to the municipality for a business license [16]. Therefore, the list covers all operating SAs and hotels in Kyoto City. This means that the entire population was analyzed in this study, not sample data which were selected randomly. Research on accommodation in Kyoto has previously used the accommodation list data in academic studies as government-reliable data [12,13,45].

The variables of the accommodation list are as follows: name, address, accommodation types, and business license date [50]. The types of accommodation are SAs and hotels. The business license date is when the government permitted the hotel business license. Meanwhile, the operators of closed accommodations must apply to the municipality to suspend their business under the Hotel Business Law. Closed accommodations are excluded from the accommodation list. Therefore, this study identified all closed accommodations by analyzing the difference between the 31 March 2020 and the 31 March 2021 lists, using the 31 March 2022 list as a baseline.

3.3.2. Household Data

Household data are the basic resident register data. The Kyoto City government publishes the number of households in each NA as open data, using the basic resident register data [51]. The basic resident register data are official resident-based data, according to the law [52]. There is also a dataset of household data from the national census, which is conducted every five years throughout Japan. However, only basic census data was collected and published for household data from 2020 to 2022. This study used data from 1 April 2020 to 1 April 2022. The difference was calculated to determine the changes in each NA for households from 1 April 2020 to 1 April 2022. These data differ from accommodation and real estate data by only one day. However, there was a small difference in this study. We concluded that this was not an issue in the analysis.

3.3.3. Real Estate Data

The real estate data used in this study were from the At Home Dataset. This study analyzed real estate data from 31 March 2020 to 31 March 2022. The At Home Dataset contains real estate data provided by the At Home Co., Ltd. (Tokyo, Japan) [53]. More than 60,000 real estate companies throughout Japan use this dataset to buy and sell real estate [54]. The dataset does not include transactions between individuals. However, the dataset covers real estate traded in the private real estate market. Some studies in Japan have shown these real estate data to be reliable [55,56]. In addition to the At Home Co., Ltd., there are other real estate agencies in Japan. However, the At Home Co., Ltd. provides real estate datasets to researchers, for research purposes only, via the Informatics Research Data Repository Dataset Service of the National Institute of Informatics. Therefore, the At Home Dataset was used in this study.

The variables in the dataset include the housing type, address, price, room area, duplicate building ID, and duplicate room ID. The housing types are detached houses for rent, detached houses for sale, apartments for rent, apartments for sale, mixed-use properties for rent, and land for sale. Among the six housing types, the address variables are not listed as detailed addresses of detached houses for rent, detached houses for sale, and land for sale to protect residents’ privacy [53]. Therefore, this study only analyzed apartments and mixed-use properties whose addresses could be identified.

The real estate dataset contained duplicates of the properties listed monthly (N = 171,047). Hence, these duplicates were removed. The duplicate room ID variable was used to remove the duplicates. The sample size was 35,698.

4. Results

4.1. Accommodation and Household in Each Neighborhood Association

Table 1. Basic statistics of accommodations and households.

	Total (N)	Average (N/NA)
(Accommodation)
Operating SAs:	1325	0.816
Closed SAs:	388	0.239
Operating Hotels:	351	0.216
Closed Hotels:	28	0.017
Change in household:	138	0.085

shows the basic statistics of accommodation and households in each NA. In addition, Figure 4 shows the locations of NAs with closed SAs or hotels. In Figure 4, the dot points indicate the locations of closed SAs (a) and hotels (b). These accommodations are shown separately as operating and closed accommodations. These data are all populations that the government identifies. As shown in Table 1, there were 1713 SAs. Among the SAs, there were 388 closed SAs, accounting for 23% of all of the SAs. Figure 4a shows that the SAs were located entirely in Kyoto City’s historical center. In addition, SAs were randomly located and not clustered in specific areas.

As shown in Table 1, there were 379 hotels. This shows that there were 4.5 times more SAs than hotels. Among the hotels, 28 were closed. Figure 4b shows that the hotels were concentrated in the southeastern part of Kyoto City’s historical center. The southeastern part is the downtown area with transportation terminals. However, the operational and closed hotels were located randomly and were not concentrated in specific areas in the southeastern region.

Figure 5 shows the changes in the number of households. Figure 5b shows the time series of total household changes from March 2015 to March 2022 in Kyoto City’s historical center. This graph shows a gradual increase in the number of households before the pandemic (2015–2019). However, this increasing trend stopped during the pandemic (2020–2022).

Next, this study focused on the changes in each NA for households from 2020 to 2022. Figure 5a shows that there were no clusters of NAs with increasing or decreasing numbers of households in particular areas. This means that NAs with increasing or decreasing household changes were randomly located, even in the southeastern area, where the hotels were concentrated. These results suggest that household changes from 2020 to 2022 may have been influenced by the location of accommodations, rather than by topography or other factors.

4.2. Households and Closed Accommodations

Figure 6 shows the average change in the number of households in NAs with and without closed accommodations. Figure 6a shows the change in SAs, and Figure 6b shows the change in hotels. First, the Shapiro–Wilk test showed that the data did not follow a normal distribution (p-value < 0.01). Therefore, significant differences were analyzed using the Wilcoxon rank sum test.

In Figure 6a, the average number of households increased by 1.34 in NAs with closed SAs. This result is reasonable because closed SAs allowed one or two households to return to NAs with closed SAs. Meanwhile, the average household number decreased by 0.19 in NAs without closed SAs. This result is also reasonable because SAs caused a population decline before the pandemic [12]. In addition, NAs with closed SAs showed a statistically significant increase in households compared with those without closed SAs (p-value < 0.05). These results suggest that closed SAs are statistically related to the number of households.

Next, this study focused on hotels. In Figure 6b, the average number of households for NAs with closed hotels increased by 2.23. By contrast, the average number of households increased by 0.05 for NAs without closed hotels. No statistically significant differences were found between the number of households in NAs with and without closed hotels. These results suggest that closed hotels are not statistically related to the number of households. Based on these results, Section 4.3 focuses on closed SAs, which statistically relate to the number of households.

4.3. Closed Simple Accommodations and Real Estate

4.3.1. Real Estate Type

This section focuses on closed SAs, which were found to have statistically significant differences. Section 4.3.1 clarifies the characteristics of closed SA properties in the private real estate market.

First, closed SA properties were identified from the real estate trade in the private market using the address variable (n = 801). Closed SA properties accounted for 2.24% of the total number of properties in the market. Notably, many closed SAs rent out multiple rooms in an apartment building. This is because the number of closed SAs identified in Section 3.1 differs from the number of closed SAs analyzed in Section 3.3. Properties other than closed SAs were classified as other real estate (other-RE) (n = 34,896). This section compares the closed SA properties (n = 801) with those of other-RE (n = 34,896).

Table 2. Closed SAs and real estate types.

	(Total)	Apartments for Rent	Mixed-Use Property for Rent	Apartments for Sale	p-Value
Closed SA	801 (2.24%)	776 (2.17%)	25 (0.07%)	0 (0.00%)	<0.01
Other RE	34,896 (97.76%)	31,657 (88.68%)	1055 (2.96%)	2184 (6.12%)

presents a cross-tabulation analysis of the property types. Table 2 shows statistically significant differences (p-value < 0.05). Regarding the closed SA properties, there were 776 apartments for rent and 25 mixed-use properties for rent. However, no apartments were available for sale (n = 0). This absence of apartments for sale was a characteristic of closed SA properties that differed from those of other-RE. This suggests that closed SA properties were transformed into short-term rentals.

Section 4.3.2 and Section 4.3.3 analyze the apartments for rent and the mixed-use properties for rent that were traded for closed SAs on the market.

4.3.2. Real Estate Price

Figure 7 shows the changes in real estate prices for closed SA properties and other-RE. Figure 7a shows the results of the ANOVA for apartments for rent, and Figure 7b shows those for mixed-use properties for rent. First, the Shapiro–Wilk test showed that the data did not follow a normal distribution (p-value < 0.01). Therefore, the Wilcoxon rank sum test was used to identify significant differences.

As shown in Figure 7a, regarding the apartments for rent, statistically significant differences were found between the real estate prices of closed SA properties and those of other-RE (p-value < 0.01). However, the price of closed SA properties was only slightly higher than that of other-RE (JPY 1938/USD 17). In detail, the average price of closed SA properties was JPY 67,852 (USD 558). Meanwhile, the average price of other-RE was JPY 65,814 (USD 541). These results indicate that there were no problems in terms of housing affordability.

In Figure 7b, regarding the mixed-use properties for rent, statistically significant differences were also established between the real estate prices of closed SA properties and those of other-RE (p-value < 0.05). The results for mixed-use properties were not statistically significantly different from those of apartments for rent. The average price of closed SA properties was JPY 325,414 (USD 2673), whereas that of other-RE was JPY 383,481 (USD 3150). This result suggests that the price of closed SA properties was significantly smaller than that of other-RE (JPY 58,067/USD 477).

4.3.3. Room Area

Figure 8 shows the changes in the room areas for closed SA properties and other-RE. Figure 8a shows the results of the ANOVA for apartments for rent, and Figure 8b shows those for mixed-use properties for rent. First, the Shapiro–Wilk test showed that the data did not follow a normal distribution (p-value < 0.01). Therefore, the Wilcoxon rank sum test was used to analyze significant differences.

Regarding the apartment for rent, Figure 8a shows a statistically significant difference between the room areas of closed SA properties and those of other-RE (p-value < 0.05). The average room area for closed SA properties was 26.22 m², while that of mixed-use properties for rent was 29.23 m². This result suggests that closed SA properties had a significantly smaller room area than other-RE (3.01 m²).

In Figure 8b, regarding the mixed-use properties for rent, no statistically significant differences were identified between the room area of closed SA properties and that of other-RE. This result differs from that of apartments for rent. The average room area for closed SA properties was 84.54 m², whereas that for other-RE was 84.61 m². This result suggests that the room area of closed SA properties was not significantly different from that of marketable properties.

5. Discussion

5.1. Increase Households Through Closed Accommodations

This study investigated the changes in households triggered by accommodation closure during the COVID-19 pandemic in Kyoto City’s historical center. This study also examined the causes of such changes, based on the real estate traded on the private market. In particular, this study considered the pandemic as a natural experiment to analyze causal relationships. Specifically, this study compared the changes in households in NAs with closed accommodations as the exposure group and those without closed accommodations as the non-exposure group. This result suggests a causal relationship between closed SAs and households from March 2020 to March 2022 through natural experiments based on the temporal procedure. The key finding was that the change in the number of households in NAs with closed SAs was higher than in NAs without closed accommodation. Specifically, the average number of households in NAs with closed SAs increased by 1.34. The results suggest that closed SAs enabled one or two households to move up to NAs with closed SAs. The results are supported by the previous study on approximately eight population declines due to tourism gentrification before the pandemic [12]. The increase in households is a significant finding because the population increased during the pandemic. Before the pandemic, tourism gentrification due to tourism led to displacement and a decrease in households in Kyoto City’s historical center [12,13], which has also been reported in other tourist cities [31,37,57]. Therefore, our findings suggest that the pandemic contributed to a population increase in NAs with closed SAs, which stopped the displacement caused by tourism gentrification before the pandemic.

The results were validated because closed SA properties are more likely to switch to being apartments for rent than other-RE properties are. Additionally, closed SA properties accounted for 2.24% of the total real estate in the market. These findings are supported by previous studies that focused on other tourist cities, where short-term rentals were distributed in the housing market during the pandemic [4,5,7]. Moreover, the findings of the current study validate those of previous studies that clarified the impact on the entire city [7]. However, our findings newly provide a specific value of 1.34 households. In addition, the price of closed SA properties was only slightly higher than that of other-RE. These results rule out the possibility that the new residents during the pandemic were more likely to have higher incomes than previous residents, causing tourism gentrification. Japan’s housing policy, which protects its residents, may be helping to maintain prices [56]. Our findings suggest that closed SAs did not experience housing affordability problems during the pandemic. However, this study also shows that closed SA properties had smaller room sizes than other-RE. The reason for this could be related to the characteristics of SAs in terms of room type. Namely, they are less expensive than hotels, and some facilities are often shared with guests. These characteristics suggest that many single households moved into closed SAs, which changed to short-term rentals with small rooms.

5.2. Theoretical and Practical Implications

As a theoretical contribution, our results suggest the possibility that the pandemic might have suppressed tourism gentrification. The possibility indicates that the rapid decline in tourism demand due to the pandemic increased the number of households in NAs with closed SAs. The results of this research were unable to clarify the rent gap [26] or the middle-class settlement in tourism gentrification [27]. However, our results indicate the suppression of the population decline associated with displacement in tourism gentrification. This means that the pandemic halted the pre-pandemic displacement that was caused by tourism gentrification. This social change, seen as an increase in households, paradoxically confirms the theory of tourism gentrification, resulting from tourism-oriented service pressure leading to the displacement of residents from tourist destinations. In other words, tourists and residents may have coexisted during the pandemic. This coexistence would help to clarify sustainable tourism with a balance between housing for residents and accommodation for tourists. It is not possible to deliberately create and simulate such extreme situations prior to the formulation of urban policy. This theoretical contribution was serendipitously discovered through the natural experiment of the pandemic as an extreme event.

This suppression suggests the practical implications of regulating the total number of SAs and suggests that regulation would be an effective urban policy, as in Spanish tourist cities [58]. Zoning, with total volume control for SAs, is the appropriate urban policy for sustainable tourism with a good balance between housing and accommodation. We need to artificially generate a rapid decline in the tourism demand through urban policies, as unexpectedly occurred due to the pandemic. For the change during the pandemic, P2P accommodation, namely SAs in this study, played an important role. Specifically, P2P accommodations quickly acted as adjustment valves, balancing the demand for housing and accommodation. This rapid change may only be possible with P2P accommodation and may not be possible with traditional accommodation such as hotels.

5.3. Research Limitations

There are two research limitations in this study. The first limitation is related to the adjustments for confounding factors. This study is valuable because it is an all-inclusive survey of the government’s accommodation lists. Meanwhile, data on income level and the standard of living could not be collected on the NA scale, which could be confounding factors. Future research should obtain NA scale data and adjust for confounders such as propensity score matching. The second limitation is related to the generalizability of our results in the case of Kyoto City, Japan. Given the generalizability of the occurrence of tourism gentrification before the pandemic, it is possible that the results of this study can be generalized to other tourist cities during the pandemic. However, government support and urban recovery during the pandemic varied from country to country. In addition, studies on P2P accommodation during the pandemic are just beginning to be published. Therefore, future research on a meta-analysis review will likely demonstrate the generalizability of this study’s findings.

6. Conclusions

In conclusion, this study clarified that the average number of households increased significantly in NAs with closed SAs during the COVID-19 pandemic in Kyoto City’s historical center. Before the pandemic, SAs caused population decline through displacement. The current study has emphasized the importance of household growth. The results were validated because closed SA properties have significantly more rental units than other-RE. This indicates that closed SA properties have been converted into short-term rentals. In terms of short-term rental apartments, closed SA properties had significantly smaller room sizes than the other-RE, although they were only slightly more expensive. The results of this study confirmed the similarity between the results of many studies and the case of Kyoto City. As a theoretical contribution, our findings suggest that the pandemic suppressed tourism gentrification. P2P accommodation played an important role in this suppression as it quickly balanced housing and accommodation needs.