India

Understanding Policy Effectiveness Using New Data Sources – Lessons Learned From COVID-19 in Maharashtra

The Challenge

Maharashtra is one of the wealthiest states in India and the second-most populous with close to 112 million residents. Maharashtra was also among the worst hit by COVID-19, registering nearly 140,000 deaths since the start of the pandemic. Researchers at the Development Data Partnership set out to investigate the impact of policies introduced in response to COVID-19, and how private sector data can improve evidence-based decision-making when more conventional data, such as data from polls and surveys, is not available in real-time.

The Approach

​To study the efficacy of using alternative datasets to inform policymaking, researchers at the Development Data Partnership, a collaboration between international development organisations and private sector data companies, relied on the following innovative data types:

• Mobile phone movement data, from geolocation data provider Outlogic

• Economic activity indices based on satellite imagery, provided by SpaceKnow, a provider of large-scale planetary analysis

This data was tracked throughout the peak of the pandemic in 2021 and compared with baseline levels from 2019. Given the specific dates of COVID-19 policy changes, researchers were able to reach the following conclusions:

1. People moved into or out of regions affected by COVID-19 policies in anticipation of a policy, and not when the policy itself was implemented.

2. Industrial, construction and manufacturing indices that show economic activity always took a week or two to react to a change in policy.

3. Emergence of the COVID-19 Delta variant across Maharashtra at the end of June 2021 did not seem to have any noticeable effect on movement or economic activity.

The Benefits

The research in Maharashtra exemplified two tangible benefits for policymakers. First, real-time insights can help make faster policy decisions. Second, insights became available at a level of granularity and detail that is not otherwise accessible to policymakers. For example, satellite imagery was able to distinguish between impacts to specific industries such as construction, manufacturing and automobile production.

This form of evidence driven policymaking contributes to the delivery of SDG 3 Good health and well-being as well as SDG 8 Decent work and economic growth by maximising the public health impact of policy measures while minimising its economic consequence. Moreover, by understanding those consequences in greater detail, policymakers can work to ensure no one is left behind, contributing to SDG 10 Reduced inequality.

140

Thousand

people died in Maharashtra because of COVID-19 since the start of the pandemic.

The context​

The second wave of COVID-19 hit Maharashtra between February and July 2021, with deaths peaking in April to May 2021. During this time, various measures to curb the virus were imposed by policymakers . An overview of the timeline is set out below.

Figure 1: Timeline of the second wave of COVID-19 in Maharashtra Source: Pachisia, 2022

Using private sector data to investigate the impact of public policies

Researchers analyzed private data from the following companies:

Outlogic: Provided movement data based on smartphone locations which were aggregated and anonymised to protect personal privacy. Researchers received daily aggregated datasets from Outlogic for India between December 2020 to July 2021 and obtained data for three months in 2019 (May, July, and August) to be used as a baseline prior to COVID-19.

SpaceKnow: Provided economic activity indices based on its own research, pollution indices and an analysis of satellite imagery.

Researchers used two such indices: the Satellite Activity Index which measures the level of economic activity in specific industries as compared to a baseline, and the Activity Structure Index, a qualitative metric that highlights the number of firms that are displaying higher, normal, or lower than normal levels of activity.

To calculate these indices, SpaceKnow extracts and aggregates changes from satellite images with the help of machine learning, advanced statistical methods, and industry expertise.

Analyzing the data over time and mapping changes onto COVID-19 policies

The researchers undertook a week-by-week analysis (15 weeks from March to July 2021) to understand the changes in movement and economic activity on a granular basis during the peak of COVID-19.

This methodology relied on:

  1. Movement data from Outlogic (see Figure 2)
  2. Economic activity indicators from SpaceKnow (see Figure 3)
  3. Mapping these onto the timeline of policies introduced in response to COVID-19

The data showed the social and economic impact of COVID-19 policies.

The researchers identified the following:

  • Mobility changed in anticipation of a policy, not when the policy itself was enforced. If residents felt that lockdowns and restrictions were about to be imposed, they tended to move out of the district (see Figure 2).
  • Industrial, construction and manufacturing indices always took a week or two to react to a change in policy (see Figure 3).
  • The emergence of the Delta variant across the state at the end of June 2021 (which prompted policymakers in Maharashtra to declare a high level of containment) did not seem to have had much of an effect on inward movement or economic activity (see Figure 3). Residents seemed to move into districts despite the threat posed by the new variant.
Figure 2: Weekly Changes in Mobility Patterns
Source: Pachisia, 2022
Figure 3: Weekly Changes in Activity of Specific Industries
Source: Pachisia, 2022

How can better data contribute to better policy?

The research in Maharashtra highlighted how using new data sources can provide policymakers with a real-time analysis. Real-time data is typically difficult to obtain in an emergency such as during the rapid spread of COVID-19, but it is in such circumstances where it can provide the greatest impact. Rapidly evolving crises respond poorly to static and one-size-fits-all approaches and require policymakers to have up to the minute and granular data at their disposal in order to make the most effective decisions.

Where do we go from here?

There are no follow-up plans for the specific research in Maharashtra but the research methodology is scalable and can be replicated in other contexts. The Development Data Partnership hosts a documentation hub where similar and other research is published.

Case Downloads

Understanding societal responses to policies undertaken during emergencies: Lessons from COVID-19's Second Wave in Maharashtra
Download

Further ressources

COVID-19 Google Mobility Reports
Download
explore more Use Cases