Data for Change
What Open Data can tell you about COVID-19 in Southeast Asia (besides number of cases)
#Data for Social Impact: Data Dashboard and Analysis for COVID-19 Responses in Southeast Asia
By Hua Deng, Online Volunteer with the UNV online volunteering service for Regional Innovation Centre, UNDP Asia-Pacific.
Overview
Since the onset of COVID-19, there have been a great number of dashboards that help understand the pandemic progress, mostly with strong focus on public health data. However, the pandemic is not just a health crisis. It also poses socioeconomic risks on people’s wellbeing, ranging from mobilities to employment. In fact, we are seeing growing amounts and wider varieties of open data sources that also have great potential to reveal those aspects, which can deepen our understanding of COVID-19 impact. Combining different open data sources together, we are able to provide rich content and extract useful insights.
We are intrigued by the question of how COVID-19 impacts people’s mobility; especially, under lockdown policies. This guides us to review Google Mobility Data, which is about how actively people are moving at different types of locations including workplaces, transit stations, groceries and other places over time. And we associate the mobility patterns with containment policies and pandemic progressions. Besides, since the effective implementation of policies is also affected by people’s willingness to comply and change behaviors, we also look at survey data on population’s behavior and attitude shifts due to COVID-19. To contextualize this analysis, we make case studies on Philippines, Vietnam and Thailand, where different government responses have resulted in different consequences.
This analysis utilizes multiple open data sources to deliver an interactive visualization dashboard in Tableau. The dashboard could help domain experts to get access to open data and gain insights easily, by offering flexibility in slicing the data and drilling into different geographical granular levels. Here is the link to the dashboard and we hope you can enjoy playing around with it.
Note: The following analysis was made mainly based on conditions before Sep 2020.
Insights: Quarantine Policy — When, Where, and How
Quarantine has been a commonly enforced policy to reduce COVID-19 transmission. But there is no “one size fits all” in terms of when, where, and how the quarantine policy should be implemented. The quarantine policies need to be contextualized and be agile enough to respond to the fast-changing circumstances. Here, we will take a deep dive into the quarantine policies of three Southeast Asia countries — — Philippines, Vietnam, and Thailand, as case studies, to discuss their good practices and lessons to learn.
Philippines: Prolonged quarantine policy becomes less effective at keeping mobility level low.
Philippines’s quarantine policies are conducted at different phases and locations with different stringency levels. Here we focus on the National Capital Region (NCR), as it is the most severely struck region in the Philippines.
There were four types of quarantine policies in the Philippines, with stringency level from high to low: Enhanced Community Quarantine (ECQ) > Modified Enhanced Community Quarantine (MECQ) > General Community Quarantine (GCQ) > Modified General Community Quarantine (MGCQ). So far, the quarantine policy at NCR could be summarized into five phases, and you could find the timeline in detail on the graph below.
From the graph above, when we compare Phase 2 vs Phase 1 (ECQ) and Phase 5 vs Phase 3 (MECQ), we should notice that the mobility level usually fails to be kept low in the latter period compared to the former period, even under the same quarantine policies.
Although we could not make causal inference here because there might be other factors influencing the mobilities, we suspect that the effect of quarantine diminished as the quarantine time was prolonged. First of all, the longer the quarantine, the higher the economic cost, and people do not have enough savings to afford staying at home for too long. According to the data from the World Bank, the Philippines has the lowest Gross Domestic Savings compared with other Southeast Asia countries, including Indonesia, Singapore, Malaysia, Thailand, and Vietnam. It was also why NCR was put under MECQ in August, though healthcare workers called for ECQ. “Only a few people have savings good for… a rainy day? Well, our savings is just good for a drizzle,” President Duterte said. The condition of Manila’s slum under lockdown is also worrying, where people are concerned if they would die from hunger rather than the virus.
Second, the population has gradually lost confidence in government handling, which is supported by YouGov Covid-19 tracker data (see chart below, sample size ~1000 per country per survey time frame). The survey results show that fewer and fewer people think the government is handling coronavirus very or somewhat well, compared to other Southeast Asia countries. The population might be less and less incentivized to follow the strict quarantine policies, because they only saw uncertain benefits of doing so.
According to recent research[1], “a suitable combination of non-pharmaceutical interventions is necessary to curb the spread of the virus,” and “less disruptive and costly non-pharmaceutical interventions can be as effective as more intrusive, drastic, ones (for example, a national lockdown).” Indeed, with very stringent lockdown in the first few months, the Philippines seems to have less room to adjust policies reacting to new conditions in the latter period. From June 1st to August 3rd, NCR was placed under GCQ for over 2 months. And with record-breaking 2000+ new cases in mid-July, there were no changes in terms of quarantine policy until August 4th, when the healthcare sector warned hospitals were almost reaching full capacity and called for a stricter quarantine. This time, although suggested to put NCR under ECQ, the government could only end with MECQ. And indeed, for some people, “even if it is just for two weeks it will be very hard for us,” and “we might go hungry,” according to a report from Reuters.
Vietnam: Targeted lockdown based on epidemiological evidence, minimizing the affected range and maximizing the containment effects.
According to YouGov’s survey, Vietnamese demonstrate very unique behaviors and attitudes towards quarantine. If advised by public health authorities to self-isolate, Vietnamese are very willing to, and feel it is very easy to do so. However, regarding how many people they come into physical contact with closely and how many times they leave their homes per day, Vietnamese represent less cautious manners. This is the opposite case for the Philippines, who are less willing and feel it hard to self-isolate, while being very cautious to meet fewer people and go outside less frequently. This is very counter-intuitive, isn’t it?
To explain this contrast, we should put it under the context of their different quarantine policies and overall government responses. Vietnam government did targeted lockdown with other measures to contain the pandemic effectively, and they were nearly back to normal very soon. Even though there was a little resurgence in August, it was managed well under control within just one month. Thus, people in Vietnam did not suffer much from the quarantine and have built trust in the government handling. There is already no need to be too nervous in terms of meeting people and going outside, but they are willing to and feel it easy to self-isolate if necessary. At the same time, the opposite holds true for people in the Philippines.
Vietnam indeed is an exemplar in the battle against COVID-19, and Our World In Data already gave a comprehensive summary of Vietnam’s story. It is worth noting that Vietnam’s targeted quarantine was based on epidemiological risks, rather than symptoms or regional level cases data. Thus, they could restrict their lockdown to the street, village, commune, and district level, affecting as few people as possible, but effectively managing the risk. Vietnam had only one nationwide lockdown, which was initially set for 15 days and was then extended to 21 days in 28 out of 63 provinces.
Thailand: “Agile” in policymaking — curfew hours adjusted from 10–4 to 11–4, then 11–3, and finally ended for a 15-day trial.
There have also been a lot of discussions on what made Thailand an exemplar on dealing with the pandemic. Here we wanted to mention a specific example about how Thailand designed the policy on curfew to demonstrate its agile policymaking.
To briefly introduce the timeline of how night curfew was enforced in Thailand:
- April 3rd: 10 p.m. — 4 a.m.
- May 17th: 11 p.m. — 4 a.m.
- June 1st: 11 p.m. — 3 a.m.
- June 15th: lift curfew for 15 days as a trial.
It is like testing the water by gradually evaluating public reactions and pandemic situations, which reduces the risks, allows for adjustments, and helps the population with a smooth transition.
We could discuss from a broad view of policymaking pace, using data from Oxford’s COVID-19 Government Response Tracker. In this data source, government responses across countries are classified into different groups and quantified by the number and stringency level of measures taken. Here, let’s look at “Containment and Health Index”, which includes containment and closure policies and health systems policies. On the graph, Philippines’s Containment and Health Index jumped sharply to nearly 100 within only a few days. At the same time, Thailand was always making small adjustments swiftly to seek the best policy package for the time. It is also interesting to look at the curve of Vietnam, which right aligns with what we discussed in the last section — the government started to respond early, and attempted to manage the size of affected population; the nationwide lockdown was enforced to clear out the risks in April, but again was limited to an affordable and manageable period, much shorter than Philippines’s lockdown.
Summary
By doing case studies on the Philippines, Vietnam and Thailand, we learnt some lessons on when, where and how for quarantine policies.
- When — The period should not be too long to exceed the affordability of the population, or the effects of the policy would diminish as time goes by.
- Where — Nationwide lockdown without other accompanying measurements would not reach expected benefits but cause harm widely, while targeted lockdown would be more effective and cause less negative impact.
- How — Being agile in policymaking, with adaptive planning and continual improvement. It is better not to abruptly enforce too stringent policy, as citizens may not be ready and may react chaotically, which may lead to a worse situation.
Thank you for reading ! We hope our analysis could inspire more questions and analysis, and we highly recommend exploring the dashboard by yourself for more insights!
Special thanks to Shumin Liu, Data and Impact Management Consultant at UNDP Regional Innovation Centre, for supervising the project and giving me tremendous support!
Appendix: Data
Private Sector
NGO
Academic Institution
- Oxford, COVID-19 Government Response Tracker: Country level indicators by day on containment and closure policies, economic policies, and health system policies, from which aggregated as overall government response index, containment and health index, economic support index, and stringency index.
- Imperial College London and YouGov, COVID-19 Behaviour Tracker: Survey results on people’s attitudes and behaviors shift in response to COVID-19, covering 29 countries around the world.
Government
- Philippines Department of Health (DOH), COVID-19 Tracker: Philippines sub-regional data in detail.
Appendix: Reference
[1] Haug, N., Geyrhofer, L., Londei, A. et al. Ranking the effectiveness of worldwide COVID-19 government interventions. Nat Hum Behav (2020). https://doi.org/10.1038/s41562-020-01009-0
Note from the editors: Towards Data Science is a Medium publication primarily based on the study of data science and machine learning. We are not health professionals or epidemiologists, and the opinions of this article should not be interpreted as professional advice. See our Reader Terms for details. To learn more about the coronavirus pandemic, you can click here.
