COVID-19 Policies: Where the Wealthy Become Poor & the Poor Become Poorer

You likely have been a victim of this too.

https://www.hks.harvard.edu/faculty-research/policy-topics/poverty-inequality-opportunity

COVID-19 struck us all. Before we knew it, several sectors of society and the economy started to face severe restrictions, lockdowns were imposed, and life was no longer the same.

In this article, we share the process and insights from part of an Omdena challenge with the goal to use a data-driven analysis to answer the question: How have COVID-19 policies impacted vulnerable populations? This challenge was carried out over 8 weeks through the collaboration of a group of more than 60 domain experts, ranging from machine learning engineers and data scientists to policy and humanitarian experts, from more than 20 countries and spanning 5 continents.

But you can start asking ‘Wait, who are these vulnerable populations?’ Good question. We asked ourselves that question many times over the course of 14 days as we tried to come about the perfect definition together with experts from different fields.

There was not only one answer! We found several indicators of vulnerable populations including multi-dimensional poverty indices, critical scarcities in terms of access to food and water, healthcare and medicine, social status, access to education, access to sanitation, among many others.

After exploring some of the most prominent consequences of the COVID-19 pandemic highlighted in research and media, we decided to focus on four factors to describe the vulnerable population of interest for the scope of this project. The four factors chosen were the following:

1) Access to healthcare

2) Unemployment

3) Wage loss

4) Domestic violence

Here we stress that given the time constraints and the vast scope on the definition of vulnerability, we opted for focusing our Omenda challenge on the populations affected by these four factors. However other factors affect vulnerability and should be addressed in further research on the topic of Covid-19 impact on global societies.

Wage loss has been one of the most challenging struggles faced by populations during the COVID-19 Pandemic. The decrease in income is directly related to the fewer number of working hours, layoffs, halting informal sector operation which is heavily dependent on people’s mobility, and more.

In the continuation of this article, we focus on the effects that COVID-19 policies have shown on wage loss and how this seems to impact the most vulnerable populations. The study spans periods ranging from January to May 2020.

Why look at the impact of COVID-19 policies on wage loss?

It has been shown that the COVID-19 crisis is bringing several groups into poverty and is exacerbating the already existing poverty groups [1] — particularly those working in the informal sector of our communities. The informal sector represents the portion of economies that are not monitored and is also a significant part of the economies in developing countries [2]. The informal sector is said to encompass almost half of the world’s working population and is expanding [3].

In order to identify the impact of COVID-19 policies on wage loss, we came up with the strategy of identifying:

1. What descriptors should we use as indicators of wage loss?
2. What implemented policies were directly related to wage loss?

To answer the first question, we started our data collection task by looking at open-source datasets, reading articles, keeping an eye out on news outlets, and leveraging our personal experiences to identify descriptors that could reflect on the loss of income in a wide range of countries (more details are presented in the section on Key Insights). After that, we refined a wide range of relevant descriptors for wage loss and settled on the descriptors displayed in Figure 1.

Figure 1: Wage loss descriptors.

We then moved into our next task: determining which of the COVID-19 policies being implemented would be of interest to understand the impact on wage loss. For that, we used the COVID-19 policies presented by the Oxford Covid-19 Government Response Tracker (OxCGRT).

The OxCGRT dataset collects policies enacted in more than 160 countries containing:

• Policies on (1) overall government response and closure indices, (2) containment and health indices, and (3) economic support indices
• Dates of enactment of these policies
• Stringency indices describing the strictness of enactment of each policy

Through the analysis of the policies set, we found out that the most relevant set of policies for our problem are: (1) overall response and closure policies — which directly affect mobility, imposing severe restrictions on the workforce and their access to income, as well as (2) economic support policies — which directly fed into financial relief and/or assistance. You can find the corresponding selected policies that were selected to be studied with wage loss descriptors in Figure 2 below.

Figure 2: Selected sets of policies relevant to wage loss.

The Challenge: Strategy and Processing

Working on this challenge was a rollercoaster experience — to say the least. Every time we thought we were heading towards a great reveal, we would hit a wall and loop back to reassess our path and next steps. This is actually no surprise, given the complexity of the problem we chose to tackle, which involved the combination of two distinct worlds: data science — with all its potential and yet dependency on the quality of data — and political science — with all its intricate nuances of human behavior and society.

Nevertheless, after several discussions regarding goals, strategy, and methodology, our team found that the typical approach for data science projects combined with a bottom-up strategy was still a good way to go about this complicated problem.

Here are some highlights from our process:

1. Problem Statement: We defined a clear problem statement alongside field experts. This helped us pinpoint our desired inputs and targets.
2. Data Collection: We gathered our data from a collection of different sources (mostly: Trading Economics, International Labour Organization (ILO), and Eurostat), cleaned and aggregated them.
3. Data Analysis: We explored what the data had to show, investigated correlations, visualized patterns, and formed connections to identify cross-correlations between wage loss descriptors and enacted policies.
4. Communicating Results: We interpreted our results and summarized our findings to share them with the world in a Demo Day hosted by Omdena.

Except that this wasn’t a linear process at all: we went from 1 to 2 to 3, back to 2, back to 1, over and over again.

Stumbling upon Unexpected Hurdles

With the goal of this challenge being to identify policy impact on vulnerable populations, our biggest challenges were:

• Lack of up-to-date data for the duration of COVID-19. As this project was carried out in parallel with the pandemic, it involved dealing with data generated and acquired in parallel with the development of the project.
• Lack of data from developing or low-income countries — which reflect largely vulnerable populations. Developing countries frequently face constant corruption. This tends to be a factor that is enhanced during a crisis such as the COVID-19 pandemic, making data for some locations in the world unreliable or simply non-existent.
• Difficulty finding data reflecting the informal sector. Since by definition this is a sector with few regulations, this kind of data is less frequently documented.

Key Insights

After going through the process of gathering and analyzing data, we concluded that our results are heterogeneous, given the availability of data for specific descriptors only for certain countries. Such analysis was still sufficient to bring important insights into the impact of policies on wage loss.

Here are our main takeaways:

• Social distancing, lockdowns, and consequently decreased mobility have shown large negative impacts on wage loss — potentially reflecting large losses on the informal sector. This result can be seen in Figure 3, for the period of March to May 2020, for countries in South America, Africa, and Asia.
• Interestingly, the enactment of economic policies (e.g. income support) has not shown a clear impact on wage loss descriptors that were studied — neither negative nor positive.
• Lack of current, up-to-date data may reflect a lack of data-driven policymaking. Policies that are not well-studied can have severe repercussions on vulnerable populations and increase the risk of poverty.

Imposing policy measures in the early stages of the pandemic does not guarantee countries having fewer COVID-19 cases nor lower losses on wage, as illustrated in Figure 4, for South Korea and the United States.

Figure 3: Daily Wage Loss from March to May 2020. Created by Jose Mira.

Figure 4: Relation between underemployment, mobility, and new COVID-19 daily cases. Created by Sanchit Bhavsar.

In summary, we have studied 11 key descriptors to extract insights of the impact enacted closure and economic policies on wage loss. Our analysis has shown that policies enacted during the COVID-19 outbreak have had a large, direct impact on personal income. Moreover, policies are imposing restrictions that will bring several groups into poverty and exacerbate the existing poverty group — particularly in the informal sector.

Data-driven decisions in a global pandemic are crucial to ensure that policies, which are meant to save populations from a critical health crisis, do not end up creating consequent disasters.

Future Directions

1. Public transparency and availability of up-to-date data are necessary in times like these to allow for collaborative efforts in solving the global crisis we all face today. Currently, there is a lack of available data on economic support policies (e.g. income support) being enacted. For example, if a government enforces economic support policies, some questions that we need to ask are: Is the support accessible to communities with no access to banks? Are there public data to track how funds are being distributed?
2. Today, we were able to extract preliminary insights from a small set of descriptors for which data was available for a limited set of countries. Future work should take forth the developed insights and approaches to treat more recent recorded data as the pandemic matures.
3. An effort must be put into improving the mechanism of identifying vulnerable populations in order to better design and address economic support on pandemics events. For instance, what other factors aside from healthcare access, wage loss, unemployment, and domestic violence would be relevant to explore?
4. It is important to create a framework for data collection where diverse information on similar outbreaks is stored. This should be a publicly available repository containing a history of cases, fatalities, enacted policies, and economic impacts on different countries. Therefore, that information could be used by governments to make better data-driven decisions in future outbreaks (including what could be a second wave of COVID-19).
5. Further analysis is needed to evaluate the impact of the time and sequence in which policies are enacted, considering that some policies have shown to be more effective in containing the spread of cases. Consequently, governments will have better tools to select which policies should be enacted first and when it is convenient to implement them.

Acknowledgments

This project was made possible by the Omdena community and their initiative in creating and running this global challenge.

We would like to particularly thank Mauricio Calderon (Sustainable Healthy Habitats & Healthy Humans for Peace — SH4P), Branka Panic (AI for Peace), and Chris P. Lara (United Nations) for guiding our strategies and decisions throughout the process.

We would also like to thank Kritika Rupauliha, Alan Ionita, Arthur Wandzel, Rohet Sareen, Baidurja ‘Adi’ Ray, Reem Ghunaim, Kimberly King and so many others for valuable discussions and informative interactions that have helped shape the direction of the project.

Finally, the results of this article studying the impact of COVID-19 policy enactment on wage loss were made possible through the contribution of the following members over the duration of 8-weeks of the Omdena challenge: Reem Mahmoud, Jose Mira, Sanchit Bhavsar, Magda Kalbarczyk, Anju Mercian, Rosana de Oliveira Gomes, Anis Ismail, Laura Clark Murray, and Kushal Vala.

This article was written and edited by Reem Mahmoud, Jose Mira, and Rosana de Oliveira Gomes.

To join one of Omdena´s projects as a changemaker, find more information on their website.

References

[1] World Bank Blogs (June 8, 2020). Updated estimates of the impact of COVID-19 on global poverty. Retrieved from https://blogs.worldbank.org/opendata/updated-estimates-impact-covid-19-global-poverty

[2] Wikipedia (June 1, 2020). Informal economy. Retrieved from https://en.wikipedia.org/wiki/Informal_economy

[3]International Labour Organization (March 11, 2002). The Informal Sector. Retrieved from https://www.ilo.org/global/about-the-ilo/multimedia/video/video-news-releases/WCMS_074529/lang--en/index.htm