Systematic literature review on impacts of COVID-19 pandemic and corresponding measures on mobility

Given the unprecedented severity of COVID-19 and the extent of spread, many studies have been published within a short time frame. The review was conducted through five steps (Snyder 2019; Wolfswinkel et al., 2013; Khan et al. 2003): (1) developing research questions; (2) searching for relevant studies based on inclusion and exclusion criteria; (3) assessing studies’ quality to identify literature relevant to our research interests; (4) identifying research topics and classifying them by the subject; and (5) summarizing and synthesizing the selected studies’ results.

After the systematic review objective was defined, we conducted a literature search using search engines, including Google Scholar and three of the most recognized academic databases that cover top-notch journals, Scopus, Science Direct, and Web of Science. The search was restricted to journal articles that included selected keywords. We used comprehensive combinations of search terms; additionally, two main search strings were employed and combined using parentheses with “AND,” one specifying all the relevant keywords for “COVID-19” and “transport,” the other specifying keywords such as “mobility,” “impact,” “effect,” “travel behavior,” “restriction,” and “measurement.”

After relevant literature was initially identified based on titles, abstracts, and conclusions, we applied the following inclusion criteria to narrow the results: (1) studies published from 2020 to the present (March 2023); (2) studies with access to the full text written in English; (3) all review articles, empirical studies, conferences or proceedings, peer-reviewed journals, including quantitative or qualitative case studies; and (4) studies examining all types of human mobility, including private car, public transport (bus, railway), bicycle, and personal mobility. However, although impacts related to the aviation sector were not specifically excluded, the analysis focused on intra-country mobility as much as possible, because inter-country movement is directly affected by each country’s immigration policy. From the selected studies, we screened article relevance using the following exclusion criteria: (1) studies covering COVID-19 in general and not related to the transport sector (i.e., COVID-19 impacts not related to human mobility and activity); (2) studies that focused on personal health, pharmacological intervention, epidemiological or pathological evidence; and (3) studies on freight/marine/water transport.

Before the full text assessment, we also reviewed reference lists for relevant literature and discovered additional relevant articles through forward and backward reference tracing, adding them to the search lists to complement the literature identified through database searches. Subsequently, duplicates were removed, and the remaining studies were further screened for relevance and scope by examining each article’s abstract, introduction, and conclusion. After filtering, 364 articles remained for the final analysis. We thoroughly reviewed each study’s content and conducted thematic analyses to categorize studies based on their topics and study perspectives. This approach effectively identified each study’s purpose, data, and results, and grouped them into major topics and sub-topics. When the study subject and transport means were similar across multiple studies, those not significantly meaningful to this study’s review subject were not included in the analysis. The articles relevant to each subject were extracted and summarized in tables that included publication details (author(s) and year), study area, research objective and method, data type, and transport mode type.

Increased traffic volume increases the possibility of contact between people and thus increases potential COVID-19 spread. Extensive research investigated the link between human mobility and COVID-19 spread, and showed a positive correlation (Table 1). Because COVID-19 was first detected in Wuhan, China, many studies focused on data from China Shi and Fang 2020; Chen et al. 2020; Liu et al. 2020b; Zhao et al., 2020a). For example, Shi and Fang (2020) investigated the temporal relationship between daily outbound traffic from Wuhan to 31 Chinese provinces and COVID-19 incidence during the virus’s early spread in 2020. Outbound traffic volume was positively associated with COVID-19 incidence in all provinces, with correlation coefficients ranging from 0.22 to 0.78, and statistically significant at the 95% level. Chen et al. (2020); Liu et al. (2020b) found that correlation coefficients between population emigration from Wuhan ranged from 0.597 to 0.996 depending on regions and mobility patterns. Moreover, a significant and positive association was observed between public transportation daily frequency—including flights, buses, and trains operating from Wuhan—and the number of COVID-19 cases (Zheng et al. 2020).

Case studies outside China showed similar patterns (Jaekel and Muley 2022; Yang et al. 2021; Cintia et al. 2020). Mobility patterns derived from mobile phone data in 25 US counties showed a strong correlation, with Pearson correlation coefficients above 0.7 for 20 of the 25 counties (Badr et al. 2020; Kissler et al. 2020) observed that the mean estimated prevalence of COVID-19 infection by borough in New York City was strongly negatively correlated with reduced commuting (-0.88Iacus et al. (2020b) also confirmed that human mobility (internal and outbound movements) had a high impact on initial virus spread in case studies of France, Italy, and Spain, with between 52% and 92% in France, up to 91% in Italy, and up to 75% in Spain. Kartal et al. (2021) revealed a cointegrated relationship between mobility and pandemic indicators through the Toda-Yamamoto causality test.

There are studies examining COVID-19’s impact on each type of transport mode to evaluate the contribution of different transport modes to virus spread. Several studies found a strong correlation between air traffic (e.g., airline passenger, number of airports, and flight routes) and COVID-19 spread (Su et al. 2022a, b; Lau et al. 2020; Sokadjo and Atchade 2020; Oztig and Askin 2020) found that both domestic and international passenger volumes in China were strongly associated with domestic and international COVID-19 cases. The results indicated that adequate measures are necessary to prevent a long-term crisis, such as on-site disease detection, temporary passenger quarantine, and limited air traffic operation. Oztig and Askin (2020) employed a negative binomial regression analysis on data from 144 countries, including population density as a control variable, and found a positive association between high numbers of airports in a country and high numbers of infected patients. A strong and significant association was also found between travel volume by train and the number of confirmed COVID-19 cases (Pang et al. 2023; Zhao et al. 2020b) found that one more high-speed railway (HSR) train originating from Wuhan each day increases the cumulative number of COVID-19 cases in a city by about 10%. Zhao et al. (2020b) estimated that a 10% increase in the number of train passengers from Wuhan to major cities in China resulted in an 8.27% increase in infections. However, cars and flights were not statistically significant in the study. Conversely, Zhang et al. (2020b) presented slightly different results. Flight and high-speed train frequencies in and out of Wuhan, China were positively and significantly associated with number of confirmed cases in the destination cities at the level of 1% and 10%, respectively. In contrast, coach (inter-city bus) services were not significantly associated with imported confirmed cases, presumably because most of coach travelers use the service for relatively short trips. Therefore, the authors argued that limiting air transport from a pandemic center is the first measure to employ to reduce travel related imported infections. Another study suggests that accessibility is related to the spread of COVID-19. Carteni et al. (2021) focused on the hypothesis that areas with higher accessibility were more easily reached by the virus. Based on data from Italy, the regression model showed that transport accessibility, population, population density, and particulate matter (PM), were significantly related to COVID-19 cases. Rail-based transport accessibility (39.7% in weight) was the best predictor for number of COVID-19 infections, followed by population and population density (about 14%), and territorial and pollutant variables (9.3%).

Because of the COVID-19 pandemic and government countermeasures, all cities worldwide experienced reduced traffic volumes, which may have resulted from various causes, such as deceased voluntary outside activities owing to fear of COVID-19 infection, and/or government orders (i.e., travel restriction, social distancing policies) implemented to mitigate spread.

A range of studies examined the impact of COVID-19 on mobility using different data sources and research perspectives (Table 2). The impact of COVID-19 on transportation demand was greatest in the early stages of the outbreak, and early studies focused on this. Gonzalez et al. (2021) found that public and private mobilities at the peak of the pandemic dropped to 95% and 86% of pre-COVID-19 levels in Spain. C2SMART (Connected Cities with Smart Transportation Center) releases monthly reports on mobility changes in New York and Seattle, US as case studies to analyze transit ridership, bridge and tunnel traffic, travel time, and number of crashes during the pandemic (Gao et al. 2020a, d; Bernardes et al. 2020). After the stay-at-home order was implemented in New York, both transit ridership and general traffic volume dropped, with transit ridership severely impacted, dropping 94% in the peak period as of March 23, 2020 (Gao et al. 2020a) compared to the 2019 statistics. It remained down at 91% in April (Gao et al. 2020d), improving to 80% in the first week of July (Bernardes et al. 2020). Reduced traffic volumes owing to the stay-at-home policy resulted in a decrease of average travel times as well: dropped by 38% during the third week of February (Gao et al. 2020a). In contrast, cycling increased by 55% in a temporary mode shift, and all traffic safety indicators improved (vehicle collisions dropped up to 77%, pedestrian injury/fatality decreased 51%, and cyclist injury/fatality in crashes decreased 31%) (Gao et al. 2020a). A US city, Seattle, experienced similar COVID-19 mobility impacts (Gao et al. 2020d). Highway traffic volume in the US state, Florida, also decreased by 47.5%, compared to the 2019 statistics (Parr et al. 2020). Korea’s average daily traffic volume in early 2020 also differed substantially from the 2019 volume, decreasing from 149 million vehicles in 2019 to about 144 million vehicles in 2020, a 9.7% decrease (Lee et al. 2020a). Canada’s mobility trends showed a clear, large reduction in mobility to non-residential locations after the state of emergency was declared (Chen, 2020). COVID-19 also significantly reduced taxi trips, and affected taxi trips’ travel speed (increased by 29.4%), travel time (decreased by 22.6%), and average distance (increased by 2.4%) (Nian et al. 2020). Average daily taxi trips in February 2020 were only 11.3% of those in May 2019. Nighttime taxi trips (9 PM – 5 AM) were significantly impacted dropping to 8.5% of the normal period. The impact of COVID-19 was greatest at the beginning of the epidemic, and the next waves of the pandemic seem to be less than the initial ones (Pozo et al. 2022; Advani et al. 2021; Konecny et al. 2021; Rasca et al. 2021). For example, subway traffic in 2020 in the UK fell to 5% during the first lockdown (from April to July), recovered to 37% before the second lockdown, and then fell back to 25% during the second lockdown in November (Vickerman 2021).

While previous studies used various databases, extensive literature used aggregated location data obtained from mobile phones, including Google Community Mobility reports and Apple Mobility Trends reports, to quantify COVID-19’s impact (Askitas et al. 2020; Tirachini and Cats 2020; Carteni et al. 2020; Schlosser et al. 2020; Pullano et al. 2020; Klein et al. 2020b; Gao et al. 2020b; Yabe et al. 2020; Galeazzi et al. 2021; Santamaria et al. 2020; Iacus et al. 2020b). Many researchers emphasized location data’s usefulness for modeling disease spread (Heiler et al. 2020), providing empirical evidence of human mobility (Couture et al. 2022), investigating the effects of different types of government interventions on human mobility, and monitoring the impact of such measures on the epidemic trajectory (Pepe et al. 2020). Using mobile device location data, Lee et al. (2020b) found that US nationwide mobility trends changed rapidly around March 13, when the national emergency was declared, and daily movements in general decreased; the percentage of people staying home rapidly increased from 20% on normal days (benchmark week, Feb. 3 to Feb. 16, 2020) to 35% after the outbreak (Apr. 6 to Apr. 12, 2020); out-of-county trips decreased from 28 to 23%; average trip distance dropped from 40 miles to 23 miles; and number of trips per person decreased from 3.7 to 2.7. Based on Google Mobility Report data, even comparing two countries with different characteristics, Germany and Qatar, the impact on the transport sector (e.g., correlations between traffic volume and government measures) was found to be similar (Jaekel and Muley 2022). Using smart card and private vehicle records in Korea, Lee et al. (2023) found that trip frequency was significantly decreased during non-peak hours on weekdays and during weekends. In addition, private vehicle usage increased for shorter trip distances, while bus usage dropped regardless of trip distances. Mobile phone data and Google and Apple reports were also used for other studies to find a correlation between the outflow of people and the reported COVID-19 cases with an eight-day time lag (Heiler et al. 2020), develop daily time-series’ of different mobility metrics (Pepe et al. 2020), investigate the impact of COVID-19 on changes in community mobility and variation in transport modes during COVID-19 alert levels (Wen et al. 2021), and examine changes in population density and visualize spatial population distributions (Arimura et al. 2020).

A few studies developed models or simulations to investigate the impact of COVID-19 on future mobility (Peng et al. 2023; Wang et al. 2020). Using MATSim, an agent-based simulation model, and assuming that the mode preference during the pandemic is maintained, Wang et al. (2020) predicted that a full reopening scenario of the NYC transportation system would result in 73% of pre-COVID transit ridership owing to changed mode preferences, while increasing car traffic as much as 142% of pre-pandemic levels. When limiting transit capacity to 50%, transit ridership would decrease by as much as 64% of pre-COVID ridership, while increasing the number of car trips to as much as 143% of pre-pandemic levels.

Other studies examining COVID-19’s impact on mobility focused on heterogeneous impacts on socioeconomic demographics or across space (Pan and He 2022; Habib et al. 2021; Guzman et al. 2021) found significant inequalities between income groups with respect to access to essential services in Bogota. Lee et al. (2020b) found that a higher income group was more likely to stay home after the national emergency declaration, and a higher density group tended to have lower trip distance after the outbreak. Ruiz-Euler et al. (2020) and Yang et al. (2021) also found different rates of reduced mobility owing to COVID-19 for high- and low- income groups, called the mobility gap. The second phase of the pandemic also showed heterogeneous changes in travel behavior according to individual attributes (e.g., age, gender, education level, marital status, income, etc.) (Jiao and Azimian 2021; Glaeser et al. 2022) estimated that total cases per capita decreased by 19% when mobility dropped by 10% in five US cities. The authors observed substantial heterogeneity across space and over time: east coast cities (i.e., NYC, Boston, and Philadelphia) had stronger effects than Atlanta and Chicago. For these differences, the authors presumed to reflect the initial infection rate rather than mobility characteristics.

The previous section confirmed that all countries worldwide experienced a pandemic related mobility drop, and public transportation was one of the most disrupted sectors (Table 3). A remarkable drop in public ridership was reported from many cities worldwide, with a 93% drop in the worst affected cities (Pozo et al. 2022; Medlock et al. 2021; Hasselwander et al. 2021; Gkiotsalitis and Cats, 2021; Aloi et al. 2020; Ahangari et al. 2020). Jiang and Cai (2022) found that for each additional local COVID-19 cumulative case within 14 days, subway ridership decreased by 0.091% in Beijing and 0.112% in Shanghai. Because public transport vehicles and stations are perceived as high risk, and fear of contagion between travelers was related to higher passenger density in a limited physical space, governments in many countries implemented restriction policies to limit or discourage public transport use, and some public transport operators reduced their services (Marra et al. 2022; Kłos-Adamkiewicz and Gutowski 2022; Jenelius and Cebecauer 2020; Tirachini and Cats 2020; Gkiotsalitis and Cats, 2021). However, public transportation is one of the most important modes of mobility, because it is sustainable and transports people on a large scale. Many transit dependent riders do not have access to a private vehicle (Pawar et al. 2020; Shakibaei et al. 2020), especially low income and historically marginalized people, who experience further loss of mobility when public transport is restricted (Suman et al. 2020; Wilbur et al. 2020; Shaheen and Wong 2020).

Regardless of the public transit transmission risk controversy, when traffic volume decreases owing to COVID-19, the reduced ridership impact is much more severe than mobility changes related to private cars. Several studies focused on this issue, examining unprecedented decline in demand and revenue, limited capacity, and social equity (Pozo et al. 2022; Shelat et al. 2022; Hasselwander et al. 2021). For example, public transport ridership decreased by about 80%, while the percentage of people using a car increased from 43 to 65%, and cycling (reduced by 23%) and bike sharing (reduced by 2%) were not significantly impacted in Budapest, Hungary (Bucsky 2020). A similar trend was reported in New York City. Subway ridership dropped 96% on April 12, 2020, compared to that before the pandemic (Kaufman et al. 2020). Commuter rail use in New York (dropped up to 97.9% compared to 2019 levels) was the most significantly affected by the pandemic, followed by subway (91.7%), buses (78.3%), and vehicle traffic volume for bridges and tunnels (65.5% by the end of May). In three regions of Sweden, which relied on recommendations instead of government mandates, public transport ridership was severely impacted (declining by 40% in Vastra and Gotland and 60% in Stockholm) (Jenelius and Cebecauer 2020). Public transit users changed their mobility patterns by switching from monthly period tickets to single tickets and travel funds (Jenelius and Cebecauer 2020; Orro et al. 2020) found that bus ridership in Coruña, Spain, was only 8–16% of 2017–2019 ridership. Lozzi et al. (2020) found that public transit dropped by 76% in April 2020 in 62 countries and 89 cities, compared to a baseline date of January 13, 2020.

Air transportation was also severely impacted by COVID-19, because many countries implemented international travel bans. Commercial flight operations were dramatically reduced worldwide, with over two thirds fewer flights than in the same period in 2019 (Falchetta and Noussan 2020). Major airline carriers’ capacity dropped by 60–80% and airline industry job loss was estimated around 7% (Sobieralski 2020). Moreover, the study estimated that recovery from the adverse effects of the current uncertainty shock will take between four and six years. Iacus et al. (2020a) forecast air traffic volume and analyzed travel bans’ impact on the aviation sector using historical air traffic data, real time flight tracks, and online booking systems data.

Some studies investigated COVID-19’s heterogeneous impact on public transport users with different socioeconomic-demographic characteristics. A study shows that older people and female travelers are more likely to be conscious of COVID-19, while those who report using the train more often tend to be indifferent to infection (Shelat et al. 2022; Almlof et al. 2021) found that public transport use decreases were associated with income levels, house ownership, and high employment levels. Similarly, Liu et al. (2020c) found uneven impacts on transit systems and social groups in an analysis of 113 public transit systems in US communities. The study showed higher levels of transit demand during the pandemic in areas with higher proportions of essential workers, vulnerable populations (African American, Hispanic, female, and people over age 45), and more coronavirus Google searches. In a case study of Nashville and Chattanooga, TN, US, fixed-line bus ridership dropped by 66.9% and 65.1%, respectively, with a significant impact on low-income groups (Wilbur et al. 2020; Nikolaidou et al. 2023; Ahangari et al. 2020) investigated factors affecting public transport ridership, including the cleanliness of public transport, income inequality index, unemployment rate, poverty, education, and percentage of foreign-born residents.

The COVID-19 pandemic also led to changes in public transit services, where some public transport operators reallocated their services and provided minimum operations to meet essential travel demands, while considering government regulations and maintaining a safe transport mode (Limsawasd et al. 2022; Tiikkaja and Viri 2021; Meena 2020; Tirachini and Cats 2020; Ahangari et al. 2020). For example, Milan and Barcelona reduced vehicle occupancy to a maximum of 25% and 50%, respectively. Catalonia provided app users with bus occupancy levels in real time. The city of Hamburg adopted flexible bus routes to increase service on the busiest routes and reduce service frequency on lower demand routes (Lozzi et al. 2020). Gkiotsalitis and Cats (2022) and Suman et al. (2020) developed optimization models to redesign public transport services such as optimal service frequencies.

As reviewed in previous sections, most research involved on case studies of the impact on personal and public transportation. Relatively few studies have examined the impacts of COVID-19 on other transport modes, such as bicycles and shared mobility. The usage behavior of these modes shows inconsistent results in each city, probably because the factors of decrease (e.g., decreased numbers of trips and increased working from home) and increase (e.g., effects of short-distance travel shifting from public transit) are mixed together. For example, during the pandemic, bike-sharing use decreased in London (Li et al. 2021b; Heydari et al. 2021), Lisbon (Teixeira et al. 2022), Bangkok (Sangveraphunsiri et al. 2022), and Slovakia (Kubal’ák et al. 2021), remained moderately stable in Korea (Choi et al. 2023), and increased in Singapore (Song et al. 2022) and Washington DC (Chen et al. 2022b). Ten cities in Germany also showed inconsistent results; the bicycle traffic volume decreased where the ratio of bicycle means was high and increased where the ratio of means was low, while pedestrian traffic decreased with higher local infectiousness and government measures (Mollers et al. 2022). In the case of London, shared bicycle usage immediately decreased due to the effect of the first lockdown but bicycle use increased during the lockdown period and showed a much larger increase after the first lockdown was lifted (Li et al. 2021b). Interestingly, morning peak travel and short-time travel by public bicycles in London maintained a low level of use during the lockdown and easing periods but were significantly higher at other times of the day and travel with middle and long duration. According to the study on the change in the travel behavior of bicycle sharing in Bangkok, shared bicycles were mostly used for business travel during the morning and afternoon peak hours on weekdays and leisure on weekends before COVID-19 (Sangveraphunsiri et al. 2022). However, the number of bicycle trips connecting subway stations in major university districts increased significantly after the pandemic.

There were also a few studies on changes in shared transportation and micro-mobility, and it was found that ridership was mostly decreased due to COVID-19 (Li et al. 2021c, d; Teixeira and Lopes 2020). For example, shared mobility ridership decreased by about 35% compared to normal in India (Meena 2020). In the case of Beijing, the overall share of shared mobility was kept constant between 36% and 38% both before and after COVID-19, but the proportion of ride-sharing decreased by 4.5% after COVID-19, while that of ride-hailing, car sharing, and bike sharing increased by 3.11%, 2.02%, and 0.89%, respectively.

Studies that examined COVID-19’s impact on environment and safety demonstrated that travel restrictions and reduced travel activities owing to COVID-19 resulted in improved air quality and safety (Llaguno-Munitxa and Bou-Zeid 2023; Nian et al. 2020; Muley et al. 2020; Cui et al. 2020; Sasidharan et al. 2020). Many studies have shown significant reductions in vehicle fuel consumption and emissions (Fischedick et al. 2021; Aloi et al. 2020). Vehicle emissions were estimated to decrease by 88.4% in 2020 and 48.6% in 2021 in Slovakia (Harantová et al. 2022) and by 14% in India (Advani et al. 2021). GHG emission was also estimated to decrease by 64% during the lockdown in Canada (Alama et al. 2022). COVID-19 and travel restriction policies have had a positive impact on traffic accidents, dropping by 67% in Santander, Spain (Aloi et al. 2020), 41% during the first month of COVID-19 in Greece, and 76% during the lockdown (March 16 – April 26, 2020) compared to 2018–2019 in Spain (Saladie et al. 2020).

The literatures reviewed in the previous chapter were mainly studies based on observed data. It is necessary to survey to analyze changes in personal travel behavior due to COVID-19 or the corresponding government’s measures. In this chapter, we review studies on this topic that were not revealed in aggregated data. Travel restrictions are effective tools for controlling infectious disease spread at the initial stages, while behavioral changes are important to limiting spread at a later stage (Muley et al. 2020). In addition to unprecedented total mobility reductions, the pandemic drastically impacted activity patterns and travel behavior through government implemented travel restrictions and individuals’ perceptions of safety and health (Table 4). These changes include transport mode choice (i.e., preferring more active and non-motorized modes), travel patterns (i.e., reducing non-essential trips and increasing work from home), and activity behavior (i.e., reducing outdoor activity and increasing online shopping) (Jou et al. 2022; Puelo, 2022; Nikiforiadis et al. 2022; Bhaduri et al. 2020; de Vos 2020; Moslem et al. 2020; Campisi et al. 2020; Borkowski et al. 2021; Tan and Ma 2021; Shamshiripour et al. 2020). Trip purpose, distance traveled, and trip frequency also changed (Abdullah et al. 2020). Significant predictors of mode choice included gender, car ownership, employment status, travel distance, primary purpose for travel, and pandemic-related underlying factors (Abdullah et al. 2020).

Many studies conducted preference surveys and demonstrated significant travel behavior changes with different perspectives and impacts on sociodemographic groups (Ferreira et al. 2022; Javadinasr et al. 2022; Szczepanek and Kruszyna 2022; Downey et al. 2022; Zhou et al. 2022; Currie et al. 2021; Echaniz et al. 2021; Abdullah et al. 2020; Morita et al. 2020; Tan and Ma 2021; Shakibaei et al. 2020; Ghader et al. 2020; Przybylowski et al. 2021; Pan et al. 2020). All survey results indicated that the pandemic impacted mode choice behavior, with people avoiding crowded places to maintain social distance, which resulted in significantly reduced public transport and shared mobility demand owing to health concerns, and increased dependence on private vehicles (Oestreich et al. 2023; Nian et al. 2020). In Santiago (Astroza et al. 2020), overall trips were reduced by 44% (with the highest reduction in metro (55%), ride-hailing (51%), and bus (45%)). Transport modes relatively less affected by COVID-19 were motorcycle (28%), auto (34%), and walking (39%). While 77% of workers from low-income households had to go out to work, 80% of workers from high-income households worked from home. In the UK, 81.9% of private commuters responded that they would continue to use their car even when restrictions are lifted, while only 3.6% and 6.5% said they could switch to walking and biking, respectively (Harrington and Hadjiconstantinou 2022). On the other hand, public transportation users from diverse locations in the world were 31.5, 10.6, and 6.9 times more likely to change their commuting mode than car users, motorcycle users, and pedestrians, respectively (Dingil and Esztergár-Kiss 2021; Bhaduri et al. 2020) analyzed the effect of traveler’s sociodemographic characteristics on travel mode choice. About 95% of respondents said that both their daily commute and discretionary travel behavior were affected by the pandemic. Meena (2020) analyzed the impact of COVID-19 on travel patterns during normal, pre-lockdown, and post lockdown periods and found that private car use increased during pre-lockdown (21%) and was expected to increase more significantly during the post lockdown period (31%), compared to the normal situation (17%). Although there were differences in degree, mode choice changes were similarly observed in other surveys, including Palermo and Catania in Italy (Moslem et al. 2020), Istanbul in Turkey (Shakibaei et al. 2020), Gdansk in Poland (Przybylowski et al. 2021), and China (Tan and Ma 2021). The COVID-19 pandemic has also changed the factors influencing mode choice (Das et al. 2021). Prior to COVID-19, factors such as travel time saving, safety, security, and comfort (Zubair et al. 2022) and factors including distance and duration of travel (Mussone and Changizi 2023) were important influencing factors in the choice of modes, but during the pandemic, concerns about infection, social distancing, wearing a mask, and worry about using public transport became important influencing factors (Zubair et al. 2022; Mussone and Changizi 2023). Studies also show that comfort and frequency of transit (Costa et al. 2022), the availability of vaccines, and the obligation to wear a mask onboard (Mashrur et al. 2022) affect the choice of public transit. Mancinelli et al. (2022) investigated a change in travel patterns departing from airports and ports. They found that before COVID-19, about 73% of respondents used public transportation as an accessibility mode, but the proportion was less than 50% during the pandemic, and the intention to use public transportation after COVID-19 surveyed to be about 56% (Mancinelli et al. 2022).

While many studies focused on mode choice behavior changes related to the pandemic, other studies investigated travel characteristics, such as travel distance. In Switzerland, compared to 2019, the travel distance of all means of transportation decreased by 50% at the beginning of the outbreak, and when the first restriction was implemented, the travel distance of public transportation decreased by more than 90% (Hintermann et al. 2023; Marra et al. 2022; Meister et al. 2022). Using a survey distributed in various countries, Abdullah et al. (2020) found that the percentage of respondents who traveled for a short trip (a distance less than 10 km) dropped from 71% before the pandemic to 45% during the pandemic. The average work trip distance was 3.6 km and 2.6 km before and during the pandemic, respectively. In fact, these numbers are much smaller than expected, probably due to the analysis of diverse countries, including underdeveloped countries. Travel distance differences before and during the pandemic were also reported by other studies (Borkowski et al. 2021; Bounie et al. 2020; De Haas et al. 2020).

Travel behavior is a complex issue, influenced by various factors such as sociodemographic and personal characteristics (Simovi´c et al. 2021; Abdullah et al. 2021; Jiao and Azimian 2021; Borkowski et al. 2021). When investigating the mode choice behavior before and during the pandemic, Abdullah et al. (2022) found that during the pandemic, monthly household income and epidemic-related factors were important predictors for short-distance (i.e., < 5 km) mode choice, whereas gender, car ownership, and monthly household income were significant predictors for longer distances (i.e., > 5 km). In a survey administered in Sicily, Italy, women were 1.5 times more likely to reduce walking frequency than men (Campisi et al. 2020; De Haas et al. 2020) found that about 80% of respondents in the Netherlands panel data reduced their outdoor activities. In particular, older people tended to reduce activities more than before the pandemic. Travel behavior changes in terms of out-of-home travel activities, activity purposes, and travel differences by income level were also observed in Canada (Fatmi 2020). Respondents in Lagos, Nigeria showed a positive correlation between transportation influenced by COVID-19 and its impact on economic (correlation coefficient of 0.442), social (0.313) and religious (0.274) activities (Mogaji 2020).

Working from home (WFH) increased, emerging as one of the government policies during the pandemic (Hensher et al. 2022, 2023; Ecke et al. 2022; Mouratidis and Peters 2022; Balbontin et al. 2022; Beck and Hensher 2020a, b). About 71% of Chicago US respondents reported that they had not experienced working from home before the pandemic, while about 63% reported that they did experience working from home during the pandemic (Shamshiripour et al. 2020). The value of travel time has changed due to the WFH policy, increasing by 12.55% compared to before the pandemic in Australia (Hensher et al. 2021). Using GPS tracking data in Switzerland, Huang et al. (2023) found more significant reductions of trip distance, travel time, travel frequency, morning peak hours trips, and trips to the CBD among the WFH group. Promoting WFH also decreased traffic congestion, especially during morning peak hours, in Hong Kong (Loo and Huang 2022). The main factor influencing WFH during the lockdown period in the Netherlands was job characteristics; office workers and teaching staff were more likely to spend more time working from home (Kalter et al. 2021). In a study analyzing WFH patterns using data from eight countries, the results show that the role of socioeconomic characteristics differs from country to country (Balbontin et al. 2021). In South America, for example, older adults and women are more likely to have WFH compared to other countries analyzed, and income has a positive effect on the number of WFH days in Australia and Chile. However, an issue of inequity was revealed as low-income and low-educated people were mainly unable to WFH and did not have flexible working hours (Ecke et al. 2022).

Fear of contagion and perceived risk also significantly impacted travel patterns (Airak et al. 2023; Navarrete-Hernandez et al. 2023; Zavareh et al. 2022; Aghabayk et al. 2021; Przybylowski et al. 2021; Abdullah et al. 2020). Awareness of overcrowding during the COVID-19 pandemic is about 1.04 to 1.23 times higher than before the pandemic (Cho and Park 2021). Women tend to be more sensitive than men to fear of infection and the use of face masks on public transport (Basnak et al. 2022; Schaefer et al. 2021). On the other hand, younger and low-income people are relatively less sensitive to overcrowding (Basnak et al. 2022). When exploring risk perception effects on human mobility for 58 countries using Global Preferences Survey data, Chan et al. (2020) found that regions with risk-averse attitudes were more likely to adjust their mobility behavior in response to the WHO declaration of a pandemic even before official government lockdowns. Przybylowski et al. (2021) found that willingness to use public transport depended mostly on perceived comfort and safety during the pandemic. Parady et al. (2020) examined pandemic related factors affecting behavioral changes in non-work-related activities in Japan, which focused on the effects of risk perception and social influence. Yuksel et al. (2020) conducted a case study in Canada that examined behavioral parameters of change in mobility and sentiment that reflected people’s beliefs about how contagious the disease is on the level of compliance with public orders. Mode choice behavior changes might be maintained for a long time owing to concerns about infection risk (Nian et al. 2020). Although Hotle et al.‘s (2020) survey was not conducted during the COVID-19 pandemic, the authors found that a recent personal experience with influenza symptoms resulted in higher risk perception at mandatory and medical trip locations in women, while men were not likely to change their travel patterns in response to potential virus spread or increasing exposure. Interestingly, high perceived workplace risk did not significantly reduce individuals’ travel to their workplaces. In addition, when Pawar et al. (2020) investigated the impact of COVID-19 on mode choice during the transition to a lockdown period in India, they found that commuters’ safety perceptions did not have a significant effect on transportation mode choice.

During the pandemic, most cities around the world are experiencing a decrease in traffic volume, resulting from a combination of restrictive policies rather than the impact of the outbreak itself (Jaekel and Muley 2022), because many epidemic prevention and control policies involve travel and activity restrictions. Global statistical data indicated that the restriction policy has substantially reduced transport demand. In particular, China’s city lockdown policy is unprecedentedly strong in the world, showing that it has the effect of controlling traffic and preventing the spread of COVID-19. For example, the Wuhan lockdown reduced inflows by about 77%, outflows by about 56%, and within-Wuhan movements by about 56% (Fang et al. 2020b). In addition, without the Wuhan lockdown, it was estimated that the number of positive COVID-19 cases would be 105% higher (Fang et al. 2020b). Although not as strong as China, several other countries have implemented city lockdowns and have shown effectiveness in controlling traffic (Jaekel and Muley 2022; Mars et al. 2022; Hadjidemetriou et al. 2020): that is, lockdown restrictions reduced (1) human mobility by 65% in France (Pullano et al. 2020), (2) mobility rate by 74.2% (21.7 trips/week before the pandemic vs. 5.6 trips/week during the lockdown) in Spain (Mars et al. 2022), and (3) long-distance travel in Germany (Schlosser et al. 2020).

However, several case studies show that a lockdown is not the only effective means of reducing traffic volume. According to the study analyzing public transport demand in Chile using smart cards (Gramsch et al. 2022), when the first measures (e.g., schools suspended in-person classes) were implemented at the beginning of the pandemic, the demand for public transport decreased by 72.3% compared to the year 2019, while it decreased by 12.1% with the dynamic lockdown implemented by each city. In particular, the effect of the lockdown decreased five weeks after its implementation, suggesting that the lockdown policy effectively controls the traffic volume in a short period of time. In this sense, the results of Dahlberg et al. (2020)’s study analyzing Sweden’s less restrictive policy are interesting. When using mobile phone data to investigate COVID-19 causal effects, they found that even less restrictive and mild public recommendations convince people to comply with social distancing and avoid unnecessary travel (i.e., residential area daytime population increased by 64%; industrial and commercial area daytime population decreased by 33%; travel distance decreased by 38%; share of short trips less than one kilometer from home increased by 36%; and mobility change effects did not differ across socioeconomic and demographic characteristics). In addition, when comparing lockdown measure effects on mobility patterns in France, Italy, and the UK, Galeazzi et al. (2021) found that their mobility patterns differed in response to the travel restrictions owing to differences in existing infrastructure characteristics and initial mobility structure.

Besides strong restriction measures, such as city lockdowns and travel restrictions, several studies investigated the effects of less restrictive or non-compulsory policies. When examining the mobility impact of different non-pharmaceutical countermeasures for 41 cities worldwide, Vannoni et al. (2020) found that the decrease in mobility is 18% due to closing public transport, 15% due to workplace closures, 13.3% due to restricting internal movements, 10% due to school closures, and 7.09% due to canceling public events. In Tokyo, non-compulsory policies, such as remote working with private companies and school closures, reduced human mobility and social contact by about 50% and 70%, respectively (Yabe et al. 2020). Similarly, government interventions reduced overall mobility by about 50% in several major US cities (Klein et al. 2020b) and reduced all station ridership by about 40.6% in Seoul, Korea (Park 2020). National social distancing measures were effective for intra-city vehicle movement, particularly at night, but not for inter-city movement in Korea (Sung 2022). Stay-at-home orders in the U.S. and Japan also showed a positive effect in reducing mobility (Liu and Yamamoto 2022; Gao et al. 2020c). Analysis using mobile phone data showed that counties without stay-at-home orders in the U.S. reduced mobility by 52.3%, while counties with stay-at-home orders experienced a slightly larger mobility drop at 60.8% (Dasgupta et al. 2020).

The social distancing policy of maintaining at least six feet between people emerged as a widely accepted non-pharmaceutical intervention to mitigate the pandemic (Chen et al. 2022a; Liu et al. 2022; Vichiensan et al. 2021; Zhang et al. 2020b; Morita et al. 2020). Although social distancing might negatively affect subjective well-being and limit physical activity, many studies supported the positive effects on travel behavior to prevent social contact and COVID-19 spread (De Vos 2020; Fang et al. 2020b). The public geo-located US Twitter data showed a significant 61.83% overall travel reduction after social distancing policies were implemented (Xu et al. 2020). In particular, larger reductions were found in states that were early adopters of social distancing practices, whereas smaller changes were found in states without such policies. Analysis using Google Community Mobility reports in the US showed that state-of-emergency declarations had only a modest effect on mobility (about a 10% decrease), but implementing one or more social distancing policies resulted in an additional 25% mobility decrease (Wellenius et al. 2021).

The implications of the studies discussed above are: (1) the effects of coercive policies are effective in the short-term (Liu et al. 2022); (2) policy announcement and implementation timing are important because unexpected anomalous behaviors can occur (Pullano et al. 2020; Liu et al. 2020b); and (3) it is more effective to introduce a combination of different types of control policies (Wang et al. 2022; Chinazzi et al. 2020; Anzai et al. 2020; Wellenius et al. 2021).

Governments implemented control and prevention policies to decrease traffic volume and person-to-person contact, which ultimately lead to reduced disease spread. Various studies examined the effectiveness of measures developed to limit COVID-19 spread and found that travel restrictions and social distancing directly affected travel behaviors, thus, effectively slowing COVID-19 spread, but at different levels (Chen et al. 2022a; Wang et al. 2022; Manzira et al. 2022; Espinoza et al. 2020; Zhang et al. 2020b; Chen and Pan 2020). The number of daily COVID-19 cases in Italy was directly associated with trips taken three weeks before (Carteni et al. 2020). In addition, the population outflow distribution significantly influenced the spatiotemporal distribution of confirmed COVID-19 cases in Wuhan, China, and the authors argued that the effect of quarantines on mobility to limit COVID-19 transmission was obvious (Jia et al. 2020).

There have also been studies showing that local travel restrictions were effective for controlling COVID-19 infection. In China, where the spread of COVID-19 was most severe, the suspension of high-speed rail and air connectivity with Wuhan reduced the number of daily new confirmed cases by 18.6% and 13.3%, respectively (Zhu and Guo 2021). According to the scenario simulation results using data from China, if lockdown and decreased population mobility policies were not implemented, the total number of infectious cases would have reached 138,824 in February 2020, corresponding to 4.46 times the actual case number (Wei et al. 2021). Travel restrictions implemented by local cities outside Hubei also decreased confirmed cases by 22.4% in the first two weeks after the Wuhan lockdown (Liu et al. 2020b). Without intra-city travel restrictions, the confirmed cases were estimated to increase by 33.1%. Based on these results, the authors asserted that if travel restrictions were implemented in advance in the entire Hubei province, the number of confirmed cases might have decreased by another 10.5%, emphasizing the importance of a timely and coordinated response to mitigate the pandemic. Another study found that travel restrictions may have reduced expected cumulative incidence by 39% in Wuhan by February 29, 2020 (Shi and Fang 2020). Staying in the same county also effectively limited COVID-19 cases and deaths. US data showed that staying in the same county reduced total weekly COVID-19 cases by 139,503, and deaths by 23,445 (Yilmazkuday 2020).

Other studies suggest that simply implementing one restriction measure does not have a significant effect on decreasing new infections. According to Chinazzi et al.’s (2020) transmission model, 90% travel restrictions to and from mainland China only modestly affected pandemic spread, delaying it for two weeks at best, unless it was combined with a strong reduction (i.e., 50% or higher) in community transmission. China and worldwide data analyzed with the susceptible-exposed-infectious-recovered (SEIR) model showed that more rigorous government control policies were associated with a slower infection rate, and isolation and quarantine procedures were less effective for controlling the pandemic (Fang et al. 2020a). When estimating the impact of travel restrictions, including lockdown in Wuhan, China, on COVID-19 incidence, Anzai et al. (2020) found that the estimated delay was smaller than the authors expected depending on the scenario. Therefore, they argued travel restriction decisions, such as a complete lockdown, should be carefully applied by comparing the resulting estimated epidemiological impact and predicted economic outcomes. Pan et al.’s (2020) analysis of mobile phone location data showed a similar trend. The authors proposed a social distancing index, which indicated that both government orders and local outbreak severity were significantly associated with the strength of social distancing.

Since the COVID-19 outbreak, many studies have examined and explained the effectiveness of government control and prevention measures. However, the effectiveness of diverse measures has been a subject of debate (Anzai et al. 2020) because studies report inconsistent results or findings showing less effectiveness than expected, and identify controversial issues, such as control strategy side effects. Moreover, it is difficult to quantify and distinguish measure effects from other potential contributing factors (Liu et al. 2022; Fang et al. 2020b). It may be unreasonable to draw one conclusion based on a single standard in this study because the timing and method of government policy implementation, citizen compliance, and analysis data and methodologies are different in each city. Nevertheless, it is of great significance to review the research results published so far and to learn some lessons. Accordingly, this section reviews the related issues and conflicting research results.

COVID-19’s spatial distribution in China was well explained by human mobility data in the early stages of the pandemic (until February 10, 2020) outside of Wuhan, China (Kraemer et al. 2020). However, after control measures were implemented, this correlation dropped and pandemic growth rates became negative in most China locations. The authors asserted that travel restrictions may have effectively reduced the flow of case importations from Wuhan in the early stages of the pandemic. However, restrictions may have been less effective once the outbreak was more widespread, thus other local mitigation measures may have been more important to mitigating spread. Another study using data from China also found that the fastest and most widespread way to prevent the spread of COVID-19 infection is to control the route connected to the epicenter in the early stages of the epidemic (Lu et al. 2021). If the virus is widespread, implementing restrictions in hub cities is much more efficient than imposing the same travel restriction across the country (Lu et al. 2021). Another interesting simulation determined when mobility restrictions effectively reduced the pandemic’s size within and between heterogeneous neighboring communities, including one with a high infection risk and another with a low infection risk (Espinoza et al. 2020). The study found that the number of secondary cases increased with the level of mobility, increasing the overall final pandemic magnitude. However, the cordon sanitaire did not always minimize the overall number of infected individuals. Accordingly, the authors argued that mobility restrictions may not always effectively contain disease spread that is evaluated by overall final pandemic size.

A few studies compared the effectiveness of various measurements. Martin-Calvo et al. (2020) used a SEIR model to evaluate the impact of different social distancing strategies under various what-if-scenarios for control and mitigation in Boston, US. The results showed that passive social distance strategies were not enough to contain the pandemic, while active strategies (i.e., large scale testing, remote symptom monitoring, isolation, and contact tracing) are needed. In addition, full confinement was not feasible and did not solve the problem without active measures in place after confinement in case a new outbreak occurred. Askitas et al. (2020) conducted a similar study that examined the impact of various non-pharmaceutical interventions on COVID-19 incidence and mobility patterns for 135 countries. The findings showed that canceling public events and restricting gatherings had the largest effects on limiting the pandemic. Workplace and school closures and stay-at-home requirements also had an effect, but it was not as large. Conversely, internal movement restrictions, public transport closures, and international travel controls did not have a significant impact on reducing new infections.

City lockdowns and travel bans are also controversial and do not always successfully control COVID-19 infections. Based on the susceptible-infection-recovery (SIR) model, Zhang et al. (2020a) argued that lockdown measures, for example those adopted by China, have a severe social-economic cost and may not be a feasible solution for other countries, because there was no strong connection between population flow and cross-regional infection except at the very early stage of the outbreak. The authors asserted that non-lockdown-type measures may have outcomes similar to lockdowns if the measures are quickly prepared and strictly executed. Muller et al. (2020) also claimed that a single restriction strategy (i.e., a complete removal of infections in childcare, primary schools, or workplaces) is not sufficient to control infection dynamics. In addition, the estimated delay of contagion was smaller than expected depending on the model scenarios (Anzai et al. 2020). Therefore, even if the results show positive effects, some measures do not work everywhere (Arellana et al. 2020).

Two interesting studies discussed unexpected effects of social distancing. US state and local government interventions decreased daily mobility by between 45% and 55% as of late April 2020, and person-to-person contact events decreased further by 65–75% on average (Klein et al. 2020a). However, after social distancing guidelines expired on April 30, 2020, mobility and contact patterns increased slightly by 14% as of early May 2020. Ghader et al. (2020) observed a similar trend when examining COVID-19 and social distancing policy effects on human mobility in the US. They found that when COVID-19 cases first emerged (i.e., early- to mid-March, 2020), social distancing statistics (i.e., percentage staying home, number of trips per person, trip distance, etc.) began to improve, regardless of government social distancing orders. However, these statistics stopped improving after about two weeks, despite continuously increased COVID-19 cases and government stay-at-home orders. The authors called this unexpected mobility and COVID-19 case trend “social distancing inertia.” This phenomenon was universal throughout US states, despite different COVID-19 case timelines and government orders in each state. The authors concluded that: (1) those able to follow social distancing orders had already done so before government intervention was adopted, and (2) there is a natural behavior inertia on social distancing, which limits improvement related to social distancing (Ghader et al. 2020).

Although many governments discouraged public transit to limit COVID-19 spread, whether public transport actually spreads the virus is another debate, because there is currently a lack of comprehensive research or scientific evidence on that (Liu et al. 2022; Zhang et al. 2021c; Bucsky 2020; Wielechowski et al. 2020) found a negative but insignificant relationship between human mobility changes in public transport and the number of confirmed COVID-19 cases in Poland, although the strength and statistical significance of the correlation varied substantially across regions. However, there was a strong, negative, and significant correlation between public transport mobility changes and the stringency of government anti-COVID-19 policies. Therefore, the authors argued that forced lockdowns effectively enforced social distancing in public transport, and government travel restrictions contributed to decreased mobility. However, Borsati et al. (2022); Schwartz (2020b) concluded that there is no direct correlation between urban public transit ridership and excess mortality or COVID-19 transmission. When comparing 418 policy measures from six developed countries (Australia, Canada, Japan, New Zealand, the UK, and the US Zhang et al. (2021c) also found that none of the measures in public health and transport is associated with a reduction of either cumulative deaths or cumulative infection cases. Based on US case studies, Schwartz (2020b) emphasized that COVID-19 cases are primarily associated with local community spread, rather than public transit ridership rates. Furthermore, Musselwhite et al. (2020) argued that, although infectious diseases spread in dense public transport vehicles, this does not support the effectiveness of restricting public transport services to prevent spread. As with the influenza case, infection in the subway is very rare (Cooley et al. 2011), while the risk of infection within household contact may be greater (Williams et al. 2010).