The impact of COVID-19 and related containment measures on Bangkok’s public transport ridership

The COVID-19 pandemic and related measures used to contain its spread affected public transport ridership in cities around the world. In Thailand, the government issued 41 Royal Decrees between April 2020 and December 2021 to mitigate the spread of the pandemic. In this study, we investigate how Bangkok's public transport services (bus, metro, and boat) have been affected during this period by analyzing the daily ridership data, confirmed COVID-19 cases, and aggregated travel trends by trip destinations using from Google mobility reports. The results show that public transport ridership decreased as daily COVID cases increased and the levels of restraining measures became higher. However, other factors, such as relative strictness compared to earlier measures and sequencing of the measures seems to have had an impact on the ridership. Moreover, the impact on ridership trends is unique for each of the three modes. Bus and metro ridership appear to be more sensitive to the changes in restrictions than the boats. Bus and metro ridership also shows similar changes in the travel trends concerning the place of visit. The findings reported here provide first insights into how Bangkok's public transport systems were affected and suggest the rationale of why different public transport modes were affected differently. These results can be useful for researchers and for decision-makers who plan and design policies and measures for public transport services.


Introduction
The widespread COVID-19 pandemic has significantly affected how urban public transport services are used and organized. In several cities, the use of these services is considered a high-risk activity, as limited space on public transport vehicles can make it difficult to practice the recommended physical distancing, increasing the chances of being infected. In their responses to these risks, travelers have altered their travel patterns or avoided commuting altogether by working from home. Transport providers and governments have adjusted the availability of public transport services in some cities (Shen, 2021). These significant changes in travel patterns on public transport and service availability have imposed new and important challenges for public transport policymaking and planning (Budd and Ison, 2020).
Several studies have provided insights into how the pandemic and public mitigation measures associated with the pandemic have impacted urban public transport systems. For example, Parker et al. (2021) and Liu et al. (2020) highlighted the impact in the form of reduced frequency services in the United States. Several studies have reported cases from Europe (e.g. London (Prez, 2020); (TfL, 2021), Naples, Rome, and Valencia (Schulte-Fischedick et al., 2021)). However, there is still limited research set in developing countries. For example, studies such as Mogaji et al. (2022) and Abdullah et al. (2021) were set in Nigeria and Pakistan. Previous studies have highlighted how public transport systems in developing countries are unique in their characteristics and the challenges faced, including the lack of public funding, availability of infrastructure, and affordability by users (Iles, 2005). It is therefore important to gather insights into how public transport services were affected during the pandemic in these contexts.
This study aims to contribute to the field by providing information and analysis to illustrate how the public transport services of Bangkok, Thailand, were affected during the outbreak of the pandemic. We analyzed Bangkok's public transport ridership data (metro, bus, and boat) and analyzed its correlations with daily COVID-19 cases and government measures during the period. We also examined correlations between ridership and aggregated data from  Community Mobility Reports which illustrate changing trends in trip destinations. The period covered by this study is between January 2020 and December 2021 (24 months), which covers three months before the pandemic declaration in Thailand (Jan -Mar 2020), the first wave of infection (April -June 2020), and the second wave of infection (April -Nov 2021).

Impact of COVID-19 on public transport services
Several scientific studies have examined how public transport usage has been affected by the COVID-19 pandemic. The literature included here can be broadly classified by the focus into two categories: a) studies that examined the change in public transport usage; and b) studies focusing on the changes in the travel behavior of public transport users. a) Changes in public transport usage Typically, public transport ridership is adversely affected by a pandemic. This is related to the frame of mind of its users toward any disruptive event that can impact their health and safety in populated spaces. Wang (2014) reported on how the public perceived the risk of contracting the SARS virus when traveling on the metro system. The study has relevance to the current COVID-19 pandemic, which is even more contagious than SARS. This could a reason for a significant reduction in public transportation trips observed in cities around the world (Haas et al., 2020;Jenelius and Cebecauer, 2020). In some studies, negative correlations between transit ridership numbers and daily confirmed COVID-19 cases were observed; the number of transit trips decreased at the moment when COVID-19 cases in these cities started to rise (Arellana et al., 2020;Wang and Noland, 2021). In comparison with other modes, the travel patterns of public transport users were found to be more significant than other modes of transport, in accordance with studies by Parker et al. (2021) conducted in New York, USA and Aloi et al. (2020) in Santander, Spain. However, the difference is not universal as reported by Arellana et al. (2020). They found that personal car trips decreased at a relatively higher proportion than public transport trips in Colombia. Possible explanations are the socioeconomic factors that influenced the ability to work from home and the proportion of mode-share transport in these cities.
The restraint measures during a lockdown may reduce the frequency of public transport and cause suspension of services, which can reduce demand in certain cases. Some public transport services may also change their service frequency, but it is not a dominant factor that reduces ridership (Jenelius and Cebecauer, 2020). The reduction in transit user numbers resulted mainly from the changes in travel behavior by those trying to avoid the risks of infection while commuting and traveling (Zhang et al., 2021). It was found that the reduction in transit use occurred not only during the weekdays but also during the weekends: particularly journeys to and from stations with large-scale social activities nearby.
A study by Liu et al. (2020) revealed that the pattern of transit demand during each day can also be affected during a pandemic. The study reported a change in temporal dynamics of subway users in New York City: the highest period of subway usage shifted from the morning period to the afternoon of weekdays. However, this change in dynamics can also be different as revealed by Aloi et al. (2020). In their study, the bus ridership during the morning and midday peaks decreased less than during the afternoon peaks. Also, the study of Mützel and Scheiner (2021) shows that although there was a significant drop in ridership there was not much change in the trend of temporal metro usage in Taipei during the pandemic.
A physical lockdown is considered by several authorities to be an effective preventive measure to halt the spread of a virus. Government policies and restraining measures during pandemics can have a diverse impact on public transport ridership, depending on how the restraining measures were implemented and the levels of strictness imposed. Typically, government restraining measures include orders to stay at or work from home, as well as order that limit movement and travel between locations, which can significantly decrease demand for public transport (Arellana et al., 2020).
At the peak of the pandemic, public transport ridership across the world, as observed by Gkiotsalitis and Cats (2021), fell below its norm by 50 % to 90 %. Lockdown measures can keep mobility behavior at a low level for an extended period, which appears to have been an effective measure to contain and reduce the spread of COVID-19 in several cases as observed by Rasca et al. (2021).
The recovery rate of each transport mode after the full lockdown was widely discussed. Governmental restraining measures, such as staying at home, affects transit ridership numbers, but after relaxing the measures, numbers did not improve much, as reported in Wang and Noland (2021) with subway trips in New York city still showing negative effects even after government measures had been relaxed for two phases. The study of public transport demand by Arellana et al. (2020) also found a similar trend: transit demand did not recover much after the government ended the mandatory quarantine and announced reopening measures.
There are also studies on easing restraining measures and returning to (new) norms. Beck and Hensher (2020) observed key events, including governmental restraining measures and the change in travel behavior between the first and second waves of COVID-19 in Australia. Results show that mobility started to recover after the government announced nationwide guidelines for easing restrictions. The results from the survey show that working from home continues to be an important strategy in reducing travel and pressure on constrained transport networks. The change in public transport travel behavior also caused financial instability for transit operations as reported by Chang et al. (2021). Tiikkaja and Viri (2021) investigated the changes in public transport ridership, service frequencies, and average fill rates during the epidemic. The results suggested that the city center area was most affected by the pandemic as the ridership dropped more than anywhere else. b) Changes in the travel behavior of public transport users Two studies (Zhang et al., 2021;Chang et al., 2021) found that travelers with distinct demographics adjusted their travel behavior differently in the face of pandemics. Even though there is a great reduction in travel, especially on weekends, there was no significant change in patterns of movement for children (ages 3-11 years) before or during the pandemic.
A similar finding was also revealed in a study by Ahangari et al. (2020) which compared the impact on rail and bus ridership in the United States. They found that demographic and socioeconomic factors (racial background and unemployment) have correlations with rail ridership reduction. For bus ridership, Parker et al. (2021) and Fatmi et al. (2021) revealed that higher foreign-born background, public transport ridership, and unemployment appear to be associated with a higher reduction in bus ridership during the pandemic. Additionally, lower-income transit users were found to have a significantly smaller reduction in the number of trips and distance traveled than higherincome transit riders.
A perceived risk of becoming infected in crowded and enclosed spaces can also drive travelers to avoid public transport and shift to private vehicles. The study by Labonté-LeMoyne et al. (2020) suggested commuters shifted to using their private cars instead of mass transit (including subway, bus, and commuter train) because of the perceived impacts on physical and psychological risk. The study suggests cleaning practices, mask-wearing, hand sanitizing, and physical distancing are major measures to be considered by public transport operators.
Several studies focused on assessing the impacts of the pandemic on different modes of transport. They compared ridership across different public modes, such as transit and bike-sharing against private cars. Single-occupant modes, such as private cars, bicycles, and walking, seem to be preferred (Kolarova et al., 2021).
An increase in bicycle trips was observed by Huang et al. (2020), during a period when the number of COVID-19 cases rose and public transport trips decreased. Another study by Wang and Noland (2021) also shows bicycle trips in New York city rose significantly during the period of the pandemic. The daily number of bicycle trips was higher than in the pre-COVID-19 period. However, as public transport trips started to recover, the number of bicycle trips decreased. Xin et al. (2021) studied the effect of COVID-19 on the daily ridership of urban rail transit and found that some Chinese cities had already experienced a more severe ridership reduction but a lower infection rate than others. The study found that the ridership reduction was not strongly associated with the infection rate. The urban rail transit ridership reductions are associated with the severity and duration of governmental restrictions and lockdowns: More stringent and longer lockdowns can lead to a greater ridership reduction under the assumption of similar health risks perceived by citizens.
The study by Rasca et al. (2021) observed how transit use in Vienna, Innsbruck, Oslo, and Agder was affected by different levels of government restraining measures, such as gradual restriction, sudden imposed restrictions, and relaxed measures. This study revealed a stronger decrease in public transport ridership during the early phase of the pandemic than in other periods, even though the subsequent daily COVID-19 cases increased dramatically from October to December. The finding indicates how emerging disruption can have a stronger impact on the use of public transport due to uncertainty or what was described as "fresh fear".
It appears that different modes of transport may have different patterns in how their ridership bounced back. A study by Orro et al. (2020) shows that Coruña's bus ridership recovered to its pre-COVID-19 level at a slower pace than the city's shared bicycles and private cars. Even as bus service operations and frequencies returned to normal, the ridership was at only 50-60 % compared with the pre-COVID-19 period. Bikesharing also seemed to be more resilient than public transportation during the pandemic as its use showed a quick recovery to the same level as the pre-pandemic time in 2019, soon after the first relaxation of government mitigation measures (Wang and Noland, 2021).
The brief scan of the literature above highlights that while several aspects of COVID-19 affecting the use of public transport have been investigated, there is a lack of empirical evidence on the dynamics of public transport usage related to governmental restraining measures during the global crisis that affects transport restrictions. This study attempts to fill the gap by outlining the relationship between the change in public transport use during each level of restriction and the COVID-19 situation in Bangkok to provide insights into public transport adaptation in the future.

Methodology
The methodology applied in this study is guided by the litureature (e. g. Brakewood et al., 2015;Liu et al., 2020;Quéré et al., 2020) and applied here to describe the impact of COVID-19 on public transport use under the sequence of governmental restraining measures. A descriptive analysis is include in Section 4.1 to examine the general patterns of public transport ridership in Bangkok city during the period. We then examined the correlations between the ridership data with the daily confirmed COVID-19 cases in Bangkok retrieved from OTP (2020) in Section 4.2 using Pearson's correlation coefficient analysis (Kurumida et al., 2020). Finally, we analyzed the correlation between the ridership data with aggregated trip destination data obtained from Google COVID-19 Community Mobility Reports (Google, 2021) to ascertain public transport use trends according to traveler destination types (work and leisure) during the period (Section 4.3). These analyses are combined to provide insights into how the pandemic and associated public restraining measures affected public transport travel behavior.

Data
The daily public transport ridership of Bangkok city was obtained from the Ministry of Transport (MOT, 2021) and the number of daily COVID-19 cases was obtained from the Department of Disease Control of Thailand (DDC, 2021). Both sets of data were publicly accessible. Additionally, we collected place visit data from  Community Mobility Reports (Google, 2021). The Bangkok public transport ridership data consists of the ridership of metro, bus, and boat services for which the modal split accounted for 46 %, 52 %, and 2 % respectively. The Bangkok Metropolitan Administration (BMA) has five metro lines (Blue Line, Purple Line, Light-Green line, Dark-Green line, and the Airport Rail Link). This study excludes the Gold Line which started operations during the pandemic due to incomplete ridership data for comparison with the other public transport ridership data used in this study. The bus services included the official public buses operated by the Bangkok Mass Transit Authority (BMTA) and affiliated bus operators. The use of the informal minibus and vans is not included in this study due to unavailable data. The public boat services included are express boats with routes on the Chao Phraya river and canal boats. These public transport trips faced approximately a 90 % drop in ridership during the first lockdown to mitigate COVID-19 in April 2020.
The Google mobility data set illustrates the aggregated trends in the travel behavior of Android phone users during the period, which accounted for approximately 75 % of all smartphone users in Thailand (around 54 millions or 78 % of the population). The data is classified by trip purpose into two categories: a) leisure activity visits, which include retail and park visits, and b) workplace visits only. The classification was made to highlight the differences in leisure and work-based trips during the pandemic. The Google data illustrates the relative changes in trip purposes and destinations with a reference to the baseline value (average value of the 5-week pre-COVID-19 period from 3rd January to 6th February 2020).
The details of the restraining measures announced by the Thai government to mitigate the pandemic were obtained from the government website.

Physical restraining measures
The Thai government issued a total of 41 Royal Decrees to enforce physical limitations on its citizens within the study period (between 1st January 2020 and 31st December 2021). Decrees are proposed by the government and endorsed by the head of the state, the King, to provide a legalized order to enforce the physical movement limitation of the population. Each decree is unique in detail as they were crafted in accordance with the severity of the pandemic at the time. For this study, we clustered the decrees into ten periods (called a decree period) and classified them by their levels of restriction (levels 1 to 4).
A level 4 decree is the most restrictive imposing a full prohibition of any social activities and gatherings. Schools and universities were suspended and limited to online classes. Access to parks and other public venues also ceased and a street curfew was set from 8 pm to 4 am. Access to restaurants was only allowed for take-out meals. There was limited coverage and frequency of public transport services. Registration with the police would be required to travel across different administration areas or when crossing a high-risk area. A level 3 decree imposed similar restrictions to level 4 on social activities but it would allow access (with a controlled number of people) to public and communal spaces, such as schools, universities, restaurants, and sports complexes. Street curfews were still enforced but with an extended time that coincided with the operations hours of public transport systems (approximately between 4 am and 11 pm). A level 2 decree provided restrictions on entertainment activities and large social gatherings, such as large concerts and pubs, but without a night curfew nor shortened hours of operation for public transportation. Finally, a level 1 decree resulted in an official warning to ensure personal hygiene practices, such as wearing masks and washing hands, being enforced in closed spaces and public buildings. The decree periods, including duration and restriction level, are explained in Table 1.
The relative strictness of the containment measures during the previous period is indicated by stating Step-up (+) to indicate the change to stricter measures and Step-down (− ) to indicate the change to lesser measures. This aims to explain the situation of physical restraining measures that changed regarding governmental enforcement.

Descriptive analysis
The average daily ridership of the three public transport modes (metro, bus, and boat) are presented as a bar chart superimposed with the average daily number of reported COVID-19 cases in Fig. 1. The data is clustered by the decree period (see Table 1). Ridership appears to have a negative correlation with the restriction level of the decree, which is assumed to be determined by the number of daily COVID-19 cases. The overall net changed trends suggest that transit ridership decreased most during a period with a higher level of restriction. Considered from the baseline (Decree period 1), level 1 caused a public transport ridership reduction of between 18 and 53 %, level 2 -between 37 and 77 %, level 3 -between 48 and 85 %, and level 4 -between 64 and 91 %. There also seems to be a variation in how ridership was affected across different modes of transport. For example, during Decree period 8 the average number of boat trips decreased the most (91 %) from the baseline. In the same period, the observed ridership of metro systems decreased by 74 % and of the bus by 72 %.
The relative aggregated changes by trip destination of all modes obtained from Google are presented similarly (See Fig. 2). We include three main types of trip: retail, park, and workplace visits. The overall number of trips, and the leisure trips (retail and park visits) declined significantly more than for the workplace. The percentage of trip changes was also found to be correlated with the public transit data according to the level of restriction imposed. This also seems to be a variation in how ridership is affected across different trip purposes. For example, during Decree period 2 -level 4, the leisure trip disruption is the most significant. Park visit trips decreased by 45 %, retail trips declined by 45 %, and workplace trips decreased by 29 %.

Correlation analysis
The correlation between ridership and daily COVID-19 cases is examined according to the level of containment strictness. As mentioned in Section 3.2, the decree is a means for the government to mitigate the spreading of the COVID-19 pandemic by restraining the physical movement of its citizens. It is assumed here that the strictness of the decrees is determined by decision-makers who observed and analyzed the trends in COVID-19 infections. Thus, the strictness level of the decree is a direct reflection of the relative change between the number of COVID-19 daily cases and the use of public transportation. This can be illustrated by the number of average daily COVID-19 cases. The results are presented in Table 2.
The results indicate that there is generally a negative correlation between the daily COVID-19 cases and public transport ridership (i.e. a relatively stricter physical constraint will result in a drop in ridership and vice versa). It appears that strong negative correlations between daily COVID-19 cases and public transport ridership can be observed in periods with the highest level of strictness (level 4) and the lowest level of strictness (level 1). Besides the above periods, no significant correlation can be observed. This may be due to several reasons, including the partial restriction, strong encouragement to work from home, time leads effects of policies (Bian et al., 2021). Another interesting observation is a rise in strictness disrupts the use of public transport. For instance, the mild restriction during periods 4 to 6 when citizens were allowed to travel freely, there was a short increase in COVID-19 infections which made the government implement partial travel-control measures (period 5). We can conclude that the surveillance periods are shown in the periods before and after curfew measures were used (from 2 April 2020 to 12 June 2020 -period 2 to 4) as the ridership of all modes indicates no relationship to the daily COVID-19 cases. However, there are exceptions. A step down of restriction in Period 9 (for bus and metro ridership) indicates a strong relationship to the change in COVID-19 daily cases. This could be explained by the time lag effects of policies studied by (Bian et al., 2021).

Aggregated trip purpose analysis
In this section, we examine the correlations between public transport ridership and the relative changes of different trip purposes: workplace trips and leisure activities (retail visitor and park visits) trips. We focus on these two trip types because they constitute 99 % of the total trips. The data is presented for each decree period in Table 3.

Leisure activities visits
We analyzed the two types of leisure trips: retail visit trips and park visit trips. Public transport ridership and leisure activities show strong correlations during the early stages of the pandemic (Decree periods 2 and 3). The use of public transport and venue visits changed relatively. This presents the impact of travel constraining measures that discouraged people to travel and the compulsory closing of public venues. This relationship became uncorrelated in a period of mild restriction and the travel situation returned to almost normal (levels 1 and 2). This shows the irregular pattern of leisure trip travel behavior during the period of uncertainty. The results demonstrated that the relationships between leisure activity visits to bus and metro ridership are more closely linked compared to boat ridership, especially in retail visit trips. For instance, in the second wave (periods 7 to 10) the boat ridership shows a correlation only in period 9 while the bus and metro ridership shows correlations throughout periods 7 to 9.
Park visits and the use of public transport were also found to be strongly related throughout the observation of this study. It shows a positive correlation in times with high restrictions and a negative correlation in times with fewer restrictions. After the government lifted the first curfew in Decree period 4, all modes of public transport reported Table 1 Details of level restriction and decree period to enforce physical limitations of its citizens and average COVID-19 cases within the study period. Note: Decree period 1 is the period before WHO declared a pandemic which no royal decree contained.
negative correlations. Although this study classified the retail and park visits as the same period of activity, boat use shows a correlation with retail visits in contrast to park visits.

Workplace visits
For workplace visits (Table 3), the results suggest that bus and metro ridership have a similar correlation, while boat ridership is different. Bus and metro have strong positive correlations in the mildest restriction period (level 1) including Decree periods 1, 4, and 6. Decree period 5 is the only time that bus and metro both have negative correlations. In this period (level 2), a sudden rise in daily COVID-19 cases after a long period with mild restrictions can be observed. The change in public transport ridership during period 5 shows the ridership had a drop with regard to the stricter measures issued by the government to mitigate the spread of COVID-19. The correlation between all modes of public transport observed in this study and the workplace visits was reported as negative. The results also show that travelers avoided going to work using public transport during this time.   level 4 -0.56** -0.09 -0.63** Period 9 (− ) level 3 -0.38** -0.14 -0.38* Period 10 (− ) level 2 -0.14 -0.12 -0.12 Note: (+) the relative strictness to the containment measures during the previous period. *p < 0.05. **p < 0.01.

Discussion and conclusion
In this empirical study, we explore how the spread of COVID-19 and the physical containment measures enforced by the Thai government affected the ridership of public transport services (metro, bus, and boat services) in Bangkok city. The study includes descriptive and correlation analyses of the ridership data of these modes, Google Mobility aggregated travel trends, and the daily confirmed COVID-19 cases (January 2020-December 2021). This study is unique in its approach to clustering the containment measures by strictness and how it integrates the data sets mentioned.
We highlight three main findings of the study here. First, there is a negative correlation between the number of daily COVID-19 cases and public transport ridership which is associated with the level of restraining measures imposed as well. Transit ridership appears to decline more significantly in periods with a higher level of restriction, and increases in periods with a lower level of restriction. For instance, in the first wave of the spread in 2020, the ridership dropped significantly when the government issued the first lockdown measures which included a curfew and strict venue closures. In the following period, the ridership recovered as the daily COVID-19 cases were under control and the government lifted restricting measures.
The significant drop in the first period could be due to 'fresh fear' or reaction in the face of uncertainty as observed by (Rasca et al., 2021) that illustrates the sensitivity of transit users toward an unknown threat, which in this case was the pandemic. The sensitivity is apparent when comparing the fresh fear period (Decree period 2) with the wider spreading period (Decree period 8). In the latter period, the relative changes in transit ridership and the levels of restriction between the two periods are similar even though the number of daily infections in the latter period is significantly higher. On the other hand, the most obvious ridership recovery was seen in Period 6 (level 1) when the number of daily COVID-19 cases was under control and the government relaxed measures and re-opened venues. The transit ridership recovered significantly and its correlation to the rate of place visits was almost identical to the pre-announcement period. This shows that people were traveling in normally at this stage.
Second, there are differences in how the service ridership was affected. The bus and metro ridership showed similar trends throughout the study period, both in the periods with strict restrictions (decreases) and in the periods with fewer restrictions (increases) when the mitigation measures were lifted. For instance, the ridership of the two modes shows significant and strong correlations during level 1 of restriction throughout the observation time of this study. Also, bus and metro were both strongly significant compared to the number of COVID-19 cases during the level 4 restriction while boat ridership was not found to have any significant difference in this period. This happened in both waves of the spread of COVID-19 disease in 2020 and 2021. A similar pattern of bus-metro similarity was also found in work-based trips. It shows the significance of workplace visits in the mild restriction periods, especially between the first and second waves of the spread. However, the boat ridership data illustrates different patterns with both daily COVID-19 cases and workplace visit rates compared with other modes. Boat ridership was significantly disrupted during the period. Its ridership declined the most and recovered at the slowest rate. The relationship between boat ridership and daily COVID-19 cases is only related to the long period with mild restrictions in Decree period 6 as displayed in Table 2. In contrast, the relationship between boat ridership and workplace visit rates is mostly related throughout the pandemic by having a strong positive correlation in times of relaxed control with partial lockdown as shown in Table 3. It had a lower correlation in full lockdown except during the first lockdown which is considered a fresh fear that does not correlate. The different pattern of impacts observed suggests that the ridership of boat services may be different from the other two modes, which leads to different sensitivity and recovery patterns. We discuss this point concerning the change in work-based trips observed in the next paragraphs.
Finally, differences between service ridership observed and the aggregated mobility trends suggest how these services may be utilized by different user groups, each of which was affected by the restraining measures differently. During the strong restriction period when curfew and working-from-home measures were applied, people were discouraged from unnecessary travel. The use of public transport declined significantly and it showed no pattern of work-based trips as there was Table 3 Correlation between public transportation ridership and aggregated trips by the purpose for each decree period. Note: (+) the relative strictness to the containment measures during the previous period. *p < 0.05. **p < 0.01. Decree period 1 considered data from 15th January 2020 to 1st April 2020.
no relationship between workplace visit rates and the use of bus and metro. This means there are numbers of workers who commute by bus and metro still traveling to the workplace during the high risk of COVID-19 infection by shifting to other modes of transport such as private vehicles. While workers who need to be at the workplace during lockdown were still using the bus and metro, those numbers may be large enough to affect the results of our analysis. The discussion on mode shift also associated with the modal shift survey during the COVID-19 pandemic conducted by Das et al. (2021) stated that during the pandemic public transport users tend to shift to cars and other forms of transport, such as motorized two-wheel vehicles. External factors, such as the percentage of the public who have been vaccinated or the perceived effectiveness of the vaccine by the public, may also influence the appearance of the uncorrelated work-based trips. There were initial delays in the public vaccination program and the confidence level of vaccination effectiveness was low. These factors may affect how individuals made their travel plan during the time.
Contrary to the other two services, boat ridership was not highly affected except in the first wave of the pandemic (fresh fear). The ridership also correlates with the level of restriction: it dropped in high restriction periods (negative correlation) and recovered in mild restriction periods (positive correlation). This infers that workers who use boats as their main mode of transport may be able to practice working from home but have limitations in shifting to other modes of public transport or private vehicles to access the workplace during a time that has a high risk of COVID-19 infection. Additionally, those that utilize boat services may not enjoy the flexibility in working arrangements like bus and metro users. There is an apparent difference between the leisure and workplace trips made by the bus and metro users in the second wave of the spread (Decree periods 7-9).
We discuss the point that some groups of workers have become used to working from home or teleworking. This does not need to be at the workplace and there are many of these types of workers. Therefore, there was no correlation between bus and metro ridership to workplace visits in this period because some of them do not need to be at the workplace but they are still using bus and metro to access leisure activities (retail and park). It can be assumed from the correlation between boats and buses that reported strongly positive relative to leisure activities as shown in data displayed in Table 3. Note that this positive correlation may be caused by the travel behavior of students as well.
This study illustrates the impact of transit ridership that changed according to the government restraining measures to mitigate the spread of COVID-19 in Bangkok. It provides evidence of how public transit ridership would change in each circumstance depending on the level of restriction as well as its relationship to COVID-19 cases and place visits. In addition, the effectiveness and consequences of travel restrictions on boat users differed from bus and metro. These findings suggest that the mitigation measures used in public transportation need to consider the characteristics of the mode of each user as well. The one-size-fits-all approach implemented in Bangkok may have been effective in preventing the spread of the pandemic in the initial stages. However, as we demonstrate here, different transport services are affected by the measures differently. The findings suggest a need to find appropriate measures for different public transportation services during a pandemic. The approach may require wider participation from different stakeholders in policymaking, but a balance should be made to ensure a timely response to an emerging event, such as Covid-19. Moreover, there is a need to have a balance between preventing the spread of disease and facilitating daily activities with the least disruption to the city's economy possible. This has been one of the biggest challenges for transport policy during the pandemic.
A limitation of this study is that we analyze data from secondary sources to explain the impact of COVID-19 and related government restraining measures on Bangkok's public transport demand. Therefore, it cannot identify the impacts caused by socioeconomic factors. Neither can this study identify any travel behavioral changes from gaining experience that influenced self and organization adaptation to deal with the government restraining measures. In addition, data is not available for analysis of other modes of paratransit that accounted for several work-based trips in Bangkok city. We recommend that further studies include the paratransit data, such as minibus transit, taxi, and bike-taxi, which play a big role in the transit desert areas in Bangkok to provide a comprehensive explanation of the impact of governmental restraining measures on public transport demand.