key: cord-0761270-qoq7rp4k
authors: Okamoto, S.
title: State of Emergency and Human Mobility during the COVID-19 Pandemic
date: 2021-06-21
journal: nan
DOI: 10.1101/2021.06.16.21259061
sha: 48e3edefa58e4fba25cc801586dbac00b9d0686c
doc_id: 761270
cord_uid: qoq7rp4k

Background To help control the spread of the coronavirus disease 2019 (COVID-19), the Japanese government declared a state of emergency (SoE) thrice. However, these were less stringent than other nations. It has not been assessed whether soft containment policies were sufficiently effective in promoting social distancing or reducing human contact. Methods Utilising the Google mobility index to assess social distancing behaviour in all Japanese prefectures between 15 February 2020 and 12 June 2021, mobility changes were assessed by an interrupted time-series analysis after adjusting for seasonality and various prefecture-specific fixed-effects and distinguishing potential heterogeneity across multiple SoEs and time passed after the declaration. Results The mobility index for retail and recreation showed an immediate decline after the declaration of the SoE by 12.78 percent-points (95%CI: -13.61 to -11.94) and a further decline after the initial period (beta: -0.93, 95%CI: -1.11 to -0.74), but gradually increased by 0.02 percent-points (95%CI: 0.02 to 0.02). This trend was similar for mobilities in other places. Among the three SoEs, the overall decline in human mobility outside the home in the third SoE was the least significant, suggesting that people were less compliant with social distancing measures during this period. Conclusion Although less stringent government responses to the pandemic may help promote social distancing by controlling human mobilities outside the home, their effectiveness may decrease if these interventions are repeated and enforced for extended periods, distorting one's health belief by heuristics biases. By combining these with other measures (i.e. risk-communication strategies), even mild containment and closure policies can be effective in curbing the spread of the virus.

To help control the spread of the coronavirus disease 2019 (COVID-19), the Japanese government declared a state of emergency (SoE) thrice. However, these were less stringent than other nations. It has not been assessed whether soft containment policies were sufficiently effective in promoting social distancing or reducing human contact.

Utilising the Google mobility index to assess social distancing behaviour in all Japanese prefectures between 15 February 2020 and 12 June 2021, mobility changes were assessed by an interrupted time-series analysis after adjusting for seasonality and various prefecturespecific fixed-effects and distinguishing potential heterogeneity across multiple SoEs and time passed after the declaration.

The mobility index for retail and recreation showed an immediate decline after the declaration of the SoE by 12.78 percent-points (95%CI: -13.61 to -11.94) and a further decline after the initial period (beta: -0.93, 95%CI: -1.11 to -0.74), but gradually increased by 0.02 percentpoints (95%CI: 0.02 -0.02). This trend was similar for mobilities in other places. Among the three SoEs, the overall decline in human mobility outside the home in the third SoE was the least significant, suggesting that people were less compliant with social distancing measures during this period.

Although less stringent government responses to the pandemic may help promote social distancing by controlling human mobilities outside the home, their effectiveness may decrease if these interventions are repeated and enforced for extended periods, distorting one's health belief by heuristics biases. By combining these with other measures (i.e. riskcommunication strategies), even mild containment and closure policies can be effective in curbing the spread of the virus. restriction without complementary mitigation measures could become ineffective 14 . While tighter restriction can decrease mobility, 15,16 it could inevitably lead to a deeper economic downturn due to the decrease in economic activities. Moreover, apart from tighter nonpharmaceutical interventions (e.g. lockdown), milder interventions such as risk communication strategies were also found to be effective in reducing the spread of infection. 15 Therefore, an ideal stringency of non-pharmaceutical interventions should be implemented, but the adopted countermeasures by each country against the pandemic highly depend on local governance and socio-economic and cultural orientations. 17, 18 In Japan, the government responses to the pandemic are less stringent compared to other countries, 19 requesting, rather than mandating, individuals and businesses to practise social distancing and avoid nonessential activities even under a SoE. Nevertheless, human mobility was still effectively reduced in specific urban cities 9,10 . Thus, Japan's experiences with COVID-19 countermeasures should be helpful in supporting the mitigation of the spread of the virus by reducing human mobility with less strict interventions.

Although lockdowns and SoEs have been shown to be effective in reducing human mobility, [9] [10] [11] [12] [13] little is known about whether longer and repeated 'alerts' requesting citizens to avoid nonessential activities with risk communication strategies are equally effective.

Therefore, this study aims to evaluate the association between declarations of SoEs and human mobility utilising the data covering all prefectures in Japan.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (2) State of Emergency . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) In response to the drastic virus spread, the Japanese government declared an SoE thrice in prefectures wherein the spread of infection was serious and healthcare capacities were under strain, requesting these areas to refrain from nonessential activities.

The first nationwide SoE occurred from 7 April to 25 May 2020 and was declared in seven prefectures, before expanding to all other prefectures. The second SoE was longer than the first and was declared in 11 prefectures located in urban areas from 8 January to 21

March 2021. The third SoE started on 25 April 2021 in four prefectures, and then expanded to five other prefectures. The timelines of each SoE are described in Appendix Note 1.

To quantify the stringency of government responses to the pandemic, a research group from the University of Oxford published a 'stringency index', which comprises containment and closure policies and public information campaigns, and is reported to be associated with human mobility. 12 This index, in addition to the SoE, was used to evaluate its association with population behaviours in Japan.

As shown in Appendix Figure 1 , the stringency index of Japan increased during the SoEs, and generally corresponded to the waves of infection spread.

(4) Household consumption . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint As a result of stronger government interventions to control the virus spread, the effects on the economy may be worsened due to the decrease in individual economic activity. While individual-level monthly data on economic activity is not publicly available, the Japanese government publishes monthly statistics on average household consumption based from a nationwide sample of approximately 7,900 households 20 . Therefore, the monthly average of household consumption was used as an indicator of economic activities of the citizen. To adjust for the seasonality, percent changes from the past three-year averages of the same month were calculated. More details on the data used in this study are presented in Appendix Note 2.

De-trending of mobility index

The mobility index was used as the tracer indicator of the transmissibility of the coronavirus in this study. However, human mobility is largely affected by various factors, such as weather conditions, the days of the week, national holidays, and other specific events, which can differ across regions. Therefore, since simple intertemporal comparisons (e.g. pre-post and pastpresent comparisons) could generate biases, the mobility index could only be utilised after removing effects of weather conditions and location-specific seasonality. With this, the . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint mobility index was de-trended by obtaining the residuals ( , ) between the actual values and the fitted values as expressed in the equation (1):

̃, = ℎ , + + ,ℎ + , + , + ,

where Yi, t denotes the actual mobility index of prefecture i on the date t. ̃, represents the linear fitted value of the mobility index predicted by daily weather conditions (i.e. mean temperature, total precipitation, sunshine duration, total snowfall, and mean wind speed), prefecture-specific fixed-effects ( ), prefecture-by-national-holiday fixed-effects ( ,ℎ ), prefecture-by-days-of-the-week fixed-effects ( , ), prefecture-by-month-effects ( , ), and prefecture-by-year fixed-effects ( , ). Intuitively, , expresses the human mobility pattern which is not explained by weather conditions and prefecture-specific calendar effects.

To evaluate the association between a SoE and human mobility, several models were estimated in this study.

First, the average effects of the SoE were obtained by estimating the following equation (3):

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint where 0 is the constant and 0 represents the parameter to be estimated for quantifying percent changes from the baseline during the SoE-a dummy variable indicating the state of emergency periods (coded 1) or not (coded 0)-of the de-trended mobility index. To control potential changes in population behaviours due to the infectious conditions, the per-day infections confirmed in the past seven days for each prefecture was included in the model.

The same by-prefecture fixed-effects presented in the equation (2) are also included.

Moreover, the following equation (4) was estimated to assess heterogeneity in multiple SoEs:

where each 1 , 2 , and 3 represents the mobility changes in the first, second, and third SoE, respectively.

To deepen the understanding for the association between the SoE and mobility changes, an interrupted time series analysis 21 was further conducted by estimating the following equations (5):

where represents the number of days since the start of the study and represents the time since the start of the SoE. Additionally, 1 indicates the underlying preintervention trend, 2 is the level of change following the SoE, and 3 denotes the slope . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint change following the SoE. By extending equation (4) in the same manner as equation (3) and (5) , the mobility changes associated with each SoE was also evaluated by the time interrupted series analysis. To account for potential autocorrelation, prefecture-level cluster robust standard errors were estimated. Models were analysed by the panel data linear regression models with high-dimensional fixed effects 16, 22 , and were weighted by the total population of each prefecture. Furthermore, potential non-linear time trends were assessed for each model, assuming a quadratic trend.

Although the stringencies of containment and closure policies differ across regions due to differences in infectious conditions, the index was only available at the national level.

Therefore, country-level analyses were conducted to assess the daily association between the stringency index and human mobility.

For consumption, given that only the monthly averages for the whole country were available, the association between the stringency index and consumption was visualised. All analyses were conducted by the Stata software version 16.1 (StataCorp LLC, College Station, USA).

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 21, 2021. <Table 1> Figure 1 shows the descriptive presentations of de-trended mobility for retail and recreation, parks, and supermarkets and pharmacies. To focus on the mobility for retail and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint recreation, which could be highly related to quarantine, a remarkable decline in mobility was observed during the first SoE, while it was less obvious in the second and third SoEs. Figure   2 also presents mobilities related to self-quarantine, such as work-from-home and stay-athome behaviours. During the first SoE, people visibly avoided using public transportation, worked from home (or stopped their businesses), and generally spent more time at home. However, the trends were found to be more unstable during the second and the third SoEs. <Figure 1> <Figure 2>

SoE and mobility Table 2 shows the mobility changes for retail and recreation, grocery and pharmacies, and parks, associated with the SoE. A similar trend was observed for mobilities for the three types of places. Overall, visits and duration of stay at these places declined during all SoE declarations, showing further drops as time passed. However, the trends (i.e. downwardconvex curves) suggested that if the SoE lasted too long, it may lose its efficacy. For instance, the mobility index for retail and recreation declined by 14.77 percent-points (95% confidence interval [CI]: -15.85 to -13.69) overall during SoEs compared to non-SoE periods. The index shows an immediate drop after the declaration of the SoE by 12.78 percent-points (95%CI: -13.61 to -11.94) and a further gradual decline (beta: -0.19, 95%CI: -0.25 to -0.12). When . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint considering the quadradic time trend, the mobility index gradually increased by 0.02 percentpoints (95%CI: 0.02 -0.02) while declines in initial periods of the SoE remained unchanged. <Table 2> Table 3 presents the mobility changes during all SoE declarations for public transport, workplaces, and residential areas, suggesting that mobilities for these places are similar to the results in Table 2 . Overall and initially, mobilities declined during the SoE, but increased as the time passed. People tended to use less public transport (beta: -17.46, 95%CI: -18.62 to -16.30), spend less time at workplaces (beta: -8.23, 95%CI: -10.03 to -6.44), and more time at home (beta: 4.99, 95%CI: 4.29 to 5.69) during the SoE compared to non-SoE periods.

The mobility index shows an immediate drop after the declaration of the SoE by 15.51 percent-points (95%CI: -16.93 to -14.08) and 5.81 percent-points (95%CI: -7.50 to -4.13) for public transport and workplaces, with a further gradual decline (beta: -0.18, 95%CI: -0.26 to -0.11; beta: -0.23, 95%CI: -0.29 to -0.17, respectively). On the other hand, there was an increase by 4.12 percent-points (95%CI: 3.54 to 4.71) for residential areas, with a further gradual increase (beta: 0.08, 95%CI: 0.05 to 0.11). When considering the quadratic time trend, the mobilities for public transport and workplaces gradually increased but decreased for residential areas.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint Mobility during the first, second, and third SoE Table 4 and 5 present the differences in mobility changes between the first, second, and third SoEs. Overall, while mobilities during the first and second SoE declined, with larger magnitudes during the first SoE, the decline in mobility during the third SoE was less significant or not observed. For instance, although drops by about 19.0 and 11.2 percentpoints in the mobility for retail and recreation were observed during the first and second SoE, respectively, a decline by only about 3.8 percent-points was found in the third SoE. Moreover, the time trends during each SoE were inconsistent. The mobility index for retail and recreation increased during the first SoE (beta: 0.07, 95%CI: 0.04 to 0.11) and more sharply during the third SoE (beta: 0.19, 95%CI: 0.17 to 0.22), but it mildly declined during the second SoE (beta: -0.04, 95%CI: -0.08 to -0.00).

More visits and time spent at residential areas were not observed during the third SoE, although this was observed during the first and second SoEs. Furthermore, the immediate increase in stay-at-home time was the largest in the first SoE (beta: 6.66, 95%CI: 5.85 to 7.48), but showed a gradual decrease by 0.01 percent-points (95%CI: -0.03 to -0.00).

For the second SoE, a modest initial increase in stay-at-home time by 2.85 percent-points (95%CI: 1.46 to 4.23) was followed by a gradual increase (beta: 0.04, 95%CI: 0.01 to 0.07), while no immediate change (beta: 0.52, 95%CI: -0.67 to 1.71) was observed during the third . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. In addition to mobility changes associated with the SoE, the associations between mobility indices and stringency of government responses to the pandemic were assessed (Appendix Table 1 ). With tighter government responses, mobility outside the home declined while residential mobility increased.

Moreover, although the consumption data during the third SoE were not available, the per-month total consumption declined by about 13.2% and 7% in the first and second SoE, respectively, compared to the past-three-year average. Consumptions, especially for public transportation, culture and recreation, and clothing and footwear-dimensions which can be linked to activities outside the home-remarkably declined by about 20%, 35%, and 57% during the first SoE, and about 11%, 21%, and 33% during the second SoE.

Further analyses were conducted to evaluate mobility changes associated with a quasi-SoE, a semi-state of emergency was implemented by the Japanese government, reflecting the infectious conditions and their effects on healthcare capacity (Appendix Tables 2-5 ). However, . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint it was not evident whether the quasi-SoE was associated with mobility changes to reduce human contact.

Mobility changes of the population may differ across regions depending on the infectious conditions. For instance, citizens in prefectures with higher numbers of infections are more likely to engage in social distancing compared to those with lower numbers. Therefore, the prefectures were classified into two groups: High and Low. A prefecture was classified as 'High' if the cumulative number of confirmed cases in that area was above the median of all prefectures within the study period. In contrast, 'Low' prefectures were those with median or below median confirmed cases. In both groups, mobility patterns were similar, showing a decline in activities outside the home and an increase in stay-at-home behaviours after the declaration of an SoE (Appendix Tables 6 and 7) . However, the post-SoE slopes suggest conflicting trends: a gradual decrease in mobilities outside the home and an increase in residential visits and duration were observed in the 'High' group, while the opposite trend was found in the 'Low' counterpart. When considering non-linear time trends, identical results were found for both groups. Findings for mobility changes during each SoE for the 'High' group were almost identical with the pooled estimates for all prefectures (Appendix Tables 8   and 9 ).

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. 

This study aimed to evaluate mobility changes during the SoE after distinguishing potential heterogeneity across each SoE and time passed after each declaration. To summarise, three main findings were found in this study.

First, human mobility in places outside the home effectively declined while stay-athome time increased during SoEs. This was also confirmed by adopting the interrupted timeseries analysis, suggesting that individuals engage in social distancing behaviours during the initial periods of the SoE but become less compliant as time passes.

Second, when mobility changes during each SoE were distinguished, overall declines in mobilities outside the home and increases in stay-at-home time were less obvious during the succeeding SoEs. For instance, during the third SoE, immediate declines in mobilities were smaller or not found in all places, there was a post-SoE rise in mobilities outside the home, and there was a decline in stay-at-home time.

Third, under the stringent government responses to the pandemic and decline in mobilities, the consumption level-especially for activities outside the home-sharply declined, suggesting that strong public interventions may worsen the economy.

Under a state where people are requested to voluntarily engage in social distancing behaviours, it may be difficult for them to comply for extended and repeating periods.

Following the Health Belief Model, 23 individual beliefs regarding the virus (e.g. susceptibility . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint and severity), benefit of one's behaviours, barriers to make the desired behavioural changes, and self-efficacy to overcome given challenges affect an individual's actual social distancing behaviour. By being subjected to a SoE multiple times, individuals could become more optimistic to the infection due to heuristics biases 24 , which may lead individuals to engage in higher-risk behaviours.

Even though the mobility indices were de-trended, there were notable differences between the first, second, and third SoE. In particular, the third SoE occurred during one of the busiest holiday seasons in Japan (i.e. Golden Week), which could be a factor in motivating individuals to leave their homes. Thus, the mobility level may not be at the expected level for a pandemic and SoE even though it might have been well-suppressed during the prepandemic Golden Weeks. Additionally, the fear of new coronavirus variants with potentially high transmissibility rates and disease severity could have also affected population behaviours.

Vaccinated individuals with greater immunity may be less willing to engage in social distancing behaviours. However, COVID-19 vaccination in Japan only began in February 2021, so only about 4% of the whole population has been fully vaccinated by mid-June 2021. 25 Therefore, vaccination may not be a reason why people reacted differently to the succeeding SoEs.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint Until the majority of the population is fully vaccinated allowing for herd immunity, it is still vital to promote individual social distancing behaviours, such as wearing a mask, disinfection, and avoiding mass gatherings. Considering that stringent and widescale public health interventions could damage the economy, and the fact that repeating the same interventions could become ineffective as shown in this study, other appropriate measures need to be enforced in order to curb the virus spread. It has been shown that even less costly interventions, such as risk-communication strategies, can be considered as highly effective. 15 Therefore, by utilising behavioural insights 26 , methods to deliver directives to the population should be carefully designed to maximise their effectiveness.

Despite the significant findings of this study, there are still some limitations that should be considered. First, the Google mobility index used in the study is limited since it does not count an absolute amount of human mobility. Although the mobility index was detrended in this study, relative changes from the baseline given narrow time range may be inappropriate, making interpretation difficult. Also, the Google mobility index does not include information on smaller units of regions (e.g. city-level), time (e.g. daytime and night), and comparisons by sex and age. Second, the generalisability of the findings should be carefully considered. As the government responses to the pandemic could reflect cultural and institutional differences, 17,18 the findings from this study may not be applicable in other . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Ethics: Ethical approval was not required as this study was based on secondary analysis of publicly available data.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. Note) The table presents coefficients with confidence intervals estimated from robust standard errors in parentheses. ** p<0.01, * p<0.05. All models include, prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixed-effects, and prefecture-by-year fixed-effects, and are weighted by population size. Note) The table presents coefficients with confidence intervals estimated from robust standard errors in parentheses. ** p<0.01, * p<0.05. All models include, the per-day confirmed infections in the past seven days, prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixed-effects, and prefecture-by-year fixed-effects, and are weighted by population size. In addition, days since the start of the observation is included in Model (2), and additionally its squared term in Model (3). Full results are available upon request. SoE = State of Emergency 

1st SoE -21.57** -24.12** -18.12** -11.59** -11.55** -5.74** 6.44** 6.66* Note) The table presents coefficients with confidence intervals estimated from robust standard errors in parentheses. ** p<0.01, * p<0.05. All models include, the per-day confirmed infections in the past seven days, prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixed-effects, and prefecture-by-year fixed-effects, and are weighted by population size. In addition, days since the start of the observation is included in Model (2), and additionally its squared term in Model where the declaration took effect in at least one prefecture.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint • Appendix Table 3 . Quasi-SoE, SoE, and human mobility: Public transport, workplaces, and residential

• Appendix Table 4 . Quasi-SoE, 1st, 2nd, and 3rd SoE, and human mobility: Retail and recreation, grocery and pharmacies, and parks

• Appendix Table 5 . Quasi-SoE, 1st, 2nd, and 3rd SoE and human mobility: Public transport, workplaces, and residential

• Appendix Table 6 . SoE and human mobility for retail and recreation, grocery and pharmacies, and parks: Comparison by COVID-19 infections

• Appendix Table 7 . SoE and human mobility for public transport, workplaces, and residential: Comparison by COVID-19 infections

• Appendix Table 8 . 1st, 2nd, and 3rd SoE and human mobility: Retail and recreation, grocery and pharmacies, and parks in high-COVID-19 infections prefecture

• Appendix Table 9 . 1st, 2nd, and 3rd SoE and human mobility: Public transport, workplaces, and residential in high-COVID-19 infections prefecture . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint

This index was obtained from Google LLC 1 , which indicates relative changes from a baseline in visits and length of stay at different places. The index contains missing when there isn't enough data to ensure anonymity, and thus the sample sizes for each place can vary.

The mobility index at the following six places were contained:

(a) Retail and recreation: restaurants, cafes, shopping centres, theme parks, museums, libraries, and movie theatres.

(b) Supermarkets and pharmacies: grocery markets, food warehouses, farmers markets, specialty food shops, drug stores, and pharmacies.

(c) Parks: national parks, public beaches, marinas, dog parks, plazas, and public gardens.

(d) Public transport: public transport hubs such as subway, bus, and train stations. The per-day weather conditions were obtained from the Japan Meteorological Agency 2 . This included mean temperature, total precipitation, sunshine duration, total snowfall, and mean wind speed. Data for each prefectural capital (where the Automated Meteorological Data Acquisition System is located) were used in this study. For Saitama prefecture and Shiga prefecture, data in Kumagaya city and Hikone city were used since the system was not equipped in their prefectural capitals.

The number of confirmed cases of COVID-19 infections each day for each prefecture were obtained from the Japan Broadcasting Corporation. 3

The stringency index of the Japanese government responses to the pandemic was obtained from the COVID-19 Government Response Tracker published by the research team of the University of Oxford. 4 The index is composed of indicators for school closing, workplace closing, cancellation of public events, restrictions on gathering size, closure of public transport, stay-at-home requirements, restrictions on internal movement, restrictions on international travel, and public information campaign.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. ; https://doi.org/10.1101/2021.06.16.21259061 doi: medRxiv preprint Consumption data were obtained from the Statistics Bureau of Japan. 5 Monthly average consumptions of households with two or more members up to April 2021 were available.

To obtain the relative changes in consumption from the averages of the past three year, the consumption for each item during the periods (January 2017 -April 2021) were adjusted by the consumer price index. The consumer price index for each consumption item was obtained from the Statistics Bureau of Japan 6 .

The latest population data (2019) for each prefecture were obtained from the Statistics Bureau of Japan. 7

In the study periods, the following dates were coded as national holidays:

Year 2020 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 21, 2021. from 0~100, which indicates that higher values is more stringent. All models included the per-day confirmed infections in the past seven days, holiday fixed-effects, weekdays fixed-effects, month fixed-effects, and year fixed-effects, and were weighted by population size.

Appendix Note: The table presents coefficients with confidence intervals estimated from robust standard errors in parentheses. ** p<0.01, * p<0.05. All models included prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixed-effects, and prefecture-by-year fixed-effects, and were weighted by population size.

Appendix Note: The table presents coefficients with confidence intervals estimated from robust standard errors in parentheses. ** p<0.01, * p<0.05. All models included the per-day confirmed infections in the past seven days, prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixedeffects, and prefecture-by-year fixed-effects, and were weighted by population size. SoE = State of Emergency Appendix Note: The table presents coefficients with confidence intervals estimated from robust standard errors in parentheses. ** p<0.01, * p<0.05. All models included prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixed-effects, and prefecture-by-year fixed-effects, and were weighted by population size. A prefecture was classified as 'High' if the cumulative number of confirmed cases in that area was above the median of all prefectures within the study period. SoE = State of Emergency

The effect of large-scale anti-contagion policies on the COVID-19 pandemic

Global supply-chain effects of COVID-19 control

Japan Broadcasting Corporation. Special site for new coronavirus

COVID-19 government response tracker

Ministry of Internal Affairs and Communications. Family income and expenditure survey

Consumer price (-1.60 -1.26) (-0

Retail and recreation Supermarkets and pharmacies Parks

All models included prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixed-effects, and prefecture-by-year fixed-effects, and were weighted by population size. A prefecture was classified as 'High' if the cumulative number of confirmed cases in that area was above the median of all prefectures within the study period; 'Low' prefectures were those with median or below median confirmed cases. SoE = State of Emergency

All models included prefecture-level fixed-effects, prefecture-by-holiday fixed-effects, prefecture-by-weekdays fixed-effects, prefecture-by-month fixed-effects, and prefecture-by-year fixed-effects, and were weighted by population size. A prefecture was classified as 'High' if the cumulative number of confirmed cases in that area was above the median of all prefectures within the study period

• Appendix Table 1 . Stringency index and mobility• Appendix Table 2 . Quasi-SoE, SoE, and human mobility: Retail and recreation, grocery and pharmacies, and parks Appendix Table 3 . Quasi-SoE, SoE, and human mobility: Public transport, workplaces, and residential area Appendix Table 6 . SoE and human mobility for retail and recreation, grocery and pharmacies, and parks: Comparison across regions by COVID-19 confirmed cases Appendix 1st SoE