Skip to main content
Posted in

May

2020

Movement Analysis Based on Community Mobility Reports

by Najlaa Ramli for Young Digital Leaders

Regulation of Movement

We are all now familiar with the concept of movement restrictions and social distancing. The concepts were foreign to most of us just a few months earlier. COVID-19 outbreak has triggered various disease containment measures, including the abovementioned efforts.

Malaysia panned out Movement Control Order (MCO), effective on May 18, 2020. MCO primarily prohibits mass movements and gatherings across the country. MCO also inflicts the closure of all educational institutions, government, and private premises except those involved in essential services. Movements are limited for access to daily essentials. Numerous roadblock operations by PDRM also curbed the movements of the general public as a reminder for the public to stay at home and abide by the order.

Conditional MCO, or CMCO, was enforced beginning on May 4, 2020. Restrictions are eased out to allow certain business activities to resume operations. Movement restrictions are also lifted, except for interstate travels. At the time of writing this article, CMCO would remain in force until June 9, 2020.

Community Mobility Reports from Google and Apple

We seek to study the actual change of behavior of the general public in response to these policies. MCO compliance reports from relevant authorities are available. Nonetheless, it is worthwhile to explore alternative tools and available data that may help us to execute important decisions.

Leveraging on sets of data on users’ Location History, Google issues the COVID-19 Community Mobility Report that provides insights on how busy certain types of places are. The report demonstrates how the visits and length of stay at different places are changing compared to a baseline. The report includes relevant categories of places such as Grocery & Pharmacy, Retail & Recreation, Workplaces, Transit Stations, Parks, and Residential areas. 

Source: Google COVID-19 Community Mobility Report

Apple also shares its aggregated data on the volume of direction requests on Apple Maps. This data is used as a proxy to gauge the number of users who would be traveling from one point to another destination, thus signaling the potential volume of movements. They compare daily changes of the volume to a baseline volume in January 2020.

Source: Apple Mobility Trends Reports

Behavior during pre-MCO

The general public anticipated an official announcement for a lockdown measure on the night of March 16, 2020. This outlook prompted a large number of people to head over to grocery stores and stock up on the daily essentials. Needless to say, visits to places under Grocery & Pharmacy category recorded a surge. The highest number of users was recorded on March 17, a day before the MCO took effect.  

Source: Google COVID-19 Community Mobility Report

Behavior during MCO

As expected, visitation to public places saw a plunge on the first day of MCO implementation. Mobility trends for places under the Retail & Recreation category, which includes visits to the restaurants, shopping centers, and movie theaters, saw significant reduction. Mobility trends for Workplaces also dropped about 50% compared to the baseline. Alternately, data subjects are observed to occupy Residential areas as they adhered to the MCO’s guidelines.

Nevertheless, outcomes on the first day were not the most optimal and desirable. Confusion and misunderstandings amongst the public on the sudden MCO implementation were expected. The following few days saw more stringent monitoring, including implementation of penalties, to reduce MCO violations. The mobility trends at public places were further suppressed to about 60-80% below the baseline rate by the end of the first week of MCO.

As the ambiguity of the MCO measures cleared, the general public started to adapt their daily activities and movements according to the imposed restrictions. Data has backed this new adaptation, and we can observe stable mobility trends from week 2 of MCO onwards.

Source: Google COVID-19 Community Mobility Report

Behavior during CMCO

CMCO aims to reactivate the suppressed economy and social activities during the MCO. The public is now allowed to move freely, although certain rules and standard operating procedures (SOPs) are still applied. Mobility in all categories, except in Residentials, is depicting an upward trend. Visitation to public places has increased by about 2-9%, in comparison to the scenes during the MCO. This analysis is not taking into account visitation to Grocery & Pharmacy, which is considered as access to essentials.

Source: Google COVID-19 Community Mobility Report

Mobility Changes during CMCO vs. MCO

Day TypeGrocery & PharmacyRetail & RecreationWorkplacesTransit StationsParks
Weekday+15.9%+3.3%+6.3%+2.6%+1.8%
Weekend+21.1%+2.5%+8.7%+7.3%+2.2%

Source: YDL’s analysis of Google COVID-19 Community Mobility Report, comparing the median value of mobility changes during CMCO and MCO

Conclusion

Before the COVID-19 crisis, the analysis of users’ Location History data was primarily to identify when a local business tends to be the most crowded. Meanwhile, data on direction requests is a mere data product for transactions in Apple Maps. However, these types of data provide useful observation on the responses to public policies. Public adherence to policies that aim to slow down the rate of transmission is observable during this period of the pandemic. Also, these data provide insights on how the general public restarting their business-as-usuals, given that various alternatives to achieve productivity have emerged during MCO. Ultimately, these types of data enrich our understanding of the changes in behavior as we are adapting to the new circumstance.

Source: 

  1. Google COVID-19 Community Mobility Report (https://www.google.com/covid19/mobility/)
  2. Apple Mobility Trends Reports (https://www.apple.com/covid19/mobility)

Covid-19 Government Response Tracker: A Dynamic Comparative Reflection of Global Performance

by Dr Dhesi Baha Raja, Dr Adlan Suhaimi & Khairul Omar for Young Digital Leaders

1. Preamble

Wide-ranging responses have been launched by countries around the world since the onset of the Covid-19 pandemic in their bid to minimize its toll within the shortest time possible. The enemy is unseen and its true nature unknown, with some governments either throwing all resources possible on one end of the spectrum, while others opt a more liberal approach by minimally restraining and relying upon good public literacy and trusting self-regulatory social measures.

Benchmarking governmental responses which may vary in timing and stringency as exemplified by the Oxford Covid-19 Government Response Tracker (OxCGRT) attempts to provide insights on the impacts following various state interventions. Towards the end, we shall visit various state interventions in the betterment of our understanding of how we have fared in relation to others.

2. Performa of Different Countries

Our comparative analysis goes further three-fold by combining disease trends of respective nations, testing and their stringency index in our attempt to provide greater width and depth of the effectiveness of respective policies implemented.

Data for the number of confirmed cases and deaths in the following sections is obtained from John Hopkins Coronavirus Resource Centre while the number of tests performed is sourced from Our World in Data, a collaboration between University of Oxford Martin Programme on Global Development and the non-profit organization Global Change Data Lab.

2.1 South Korea: government policies vs. cases

The number of daily cases was successfully brought to a stable base-line after an initial peak. Reciprocating was a steady increase in the number of fatalities as expected with a steady decline. South Korea had a consistent Testing Policy (TP) complemented by an automated Contact Tracing Policy (CTP), both of which decisively instituted early and uninformedly. Apart from School Closure (SC) and Public Event Closure (PEC), South Korea eluded implementing stricter social measures via lock-downs. Arguably, both policies of CTP and TP were pivotal in their success without resorting to stifling public movement nor to socio-economic activities.

2.2 Sweden: government policies vs. cases

Sweden had from the onset opted for a sustainable policy, one which aims to attain Herd Immunity (HI) without overwhelming their healthcare services nor compromising the economy to much effect. Community suppression was avoided apart from school closure (kindergarten through grade nine were open). As with South Korea, businesses and restaurants remained opened. Social distancing was self-imposed where many abided. Public literacy of the latter being an asset to their success as was the case in South Korea. Tests were minimal as were other stringent social measures. Both the number of cases and deaths (reaching 3000 cases as of early May) have steadily declined. With a population of 10 million, Sweden leads other Nordic nations in the number of deaths. Their policy of waiting for natural immunity to develop remains, which is largely supported by the nation.

2.3 Singapore: government policies vs. cases

Singapore had extensive tests conducted from the onset. Similarly, as with South Korea, automated CT played an important role in containing the outbreak at the initial phase, though compounded of late by local transmissions and by the foreign worker clusters. This is reflected by the late peaks of both cases and deaths of recent. Noteworthy is their early decision to practice wide-spread testing and stringent automated CTP allowing early detection and decisive informed intervention.

2.4 United Kingdom: government policies vs. cases

Initially opting for HI as with Sweden, the UK made a late decision to impose stringent policies via lock-down and testing were ramped up in tandem. Clearly indecisiveness had resulted in the escalation of cases and with reciprocal deaths. The UK continues to struggle with healthcare facilities being overwhelmed with the economy tail-gaiting in the aftermath.

2.5 Malaysia: government policies vs. cases

Strict public control measures were implemented in three phases from the onset. The number of official cases and deaths peaked gradually attaining an almost plateau-like crescendo with a decline following. Tests were incrementally instituted over time. Manual Contact Tracing is an on-going exercise in limiting the spread of the epidemic. The government’s recent policy in gradually retracting the Movement Control Order (MCO) and its subsequent results will be much anticipated over time both on effects on health and on the social-economic front.

2.6 Ranking of Stringency Index: Malaysia vs other countries

Malaysia was ranked at 68.4 as of 7 May 2020 on the Stringency Index ahead of developed nations such as Sweden and Austria. The total number of cases and deaths have been kept minimal throughout these measures.

3. Discussion

Various measures have been instituted by different states in their efforts to contain the disease with different results. The outcome decided from the onset being crucial on the course of action being decided by different countries. Arguably, many nations had opted for stringent measures as an immediate effect to prevent disease transmission and to reduce mortality. Some had decided for sustainability and had bitten-the-bullet to achieve Heard Immunity through community exposure at the expense of higher morbidity and mortality.

The role of Automated Contact Tracing has emerged as an important tool for early informed intervention, which is a hallmark of efficient outbreak management. Countries such as South Korea, Singapore and Taiwan have demonstrated that technology can effectively assist in outbreak management.

Big Data should be an essential tool of the future in this respect. It allows rapid data integration and analysis at the disposal of policy makers to execute important decisions. The dynamics of newly emerging diseases not only makes available valuable information for clinicians but also to policy makers and researches alike.

In this respect, issues pertaining to greater data transparency, data confidentiality and data sharing is invaluable. The time has come to adopt a pragmatic approach towards data sharing especially in addressing global pandemics where the speed of communicating information affords early and pre-emptive action to be taken which, ultimately could save millions of lives.

Extensive testing with rapid Lab Turn Around Time (LTAT) has also shown to be pivotal in arresting chains of disease transmissions when implemented timely. Timeliness in attaining real-time-data as opposed to ‘lag-time’ data does not only allow one to assess the disease burden but also to allow early interventions. Dynamics of newly emerging diseases can also be analysed with the availability of timely data. Ultimately, test results represent the most important variable which determines policy making.

The success of less stringent policies undertaken appear to be possible where communities had a high literacy on self-imposed social restrictions where, public compliance was crucial. It allowed society to function with almost normality despite certain social restrictions. Socio-economic activities transpired with minimal stifling.

Various policies have been adopted by Malaysia with variable degrees of stringencies. Guided by the disease burden, the outcome has been favourable to date. The decision to gradually lift social constraints was made to balance both health and non-health determinants. The next course of weeks will decide how effective these decisions are in sailing through the uncharted waters of the Covid-19 pandemic.

4. Conclusion

Benchmarking stringencies as a performa of favourable policies can serve as an important indicator in our quest to seek the best solutions. There is no ‘Magic Silver Bullet’ policy to date for comparisons or benchmarking to learn from. It should serve as a tool for continuous improvement on a journey to the finishing line of ending this pandemic.

Estimating Economic Impacts Due to Impaired Workforce Productivity (PART 1)

by Najlaa Ramli for Young Digital Leaders

Updates of MCO in Malaysia

Movement Control Order (MCO) in Malaysia has been exercised since March 18, 2020. The order requires civilians to show solidarity by enforcing social distancing for the safety of all. On the flip side, this measure has deprived business-as-usual for many economic and social activities. This measure has also impacted the optimal productivity of the current workforce in most economic sectors. However, Malaysia is starting to lift some restrictions allowing a large number of economic activities to restart on May 4, 2020.

The Prime Minister conceded that the country records economic losses for each day the MCO is being enforced. Up until now, the country has recorded losses amounting to RM63 billion, or approximately RM2.4 billion each day. Taking the complexity of the economic environment into consideration, we are exploring a quantitative evaluation to understand the interplay between workforce productivity and the extent of the economic losses. 

Model for Epidemic-induced Workforce Losses

Santos. et. al developed a dynamic model capable of generating sector-disaggregated economic losses based on different magnitudes of workforce disruptions. The model was used as an instrument to analyse post-2009’s H1N1 pandemic. It is an extension of the Inoperability Input-Output Model (IIM) developed by Jiang and Haimes (2001).  

The model by Santos. et al. took into consideration the rate of absenteeism and working hours lost during the pandemic. These were used as measures of workforce disruptions. Meanwhile, economic losses give information on the monetary value to the impairment to the overall production of goods and services. The complete framework of their model is presented below:

Source: Santos et.al (2012) Risk-Based Input Output Analysis of Influenza Epidemic Consequences on Interdependent Workforce Sector

In the period of social distancing, non-essential services experience sharp declines due to decreased demand. On the supply side, productivity is impaired because of restricted business operations and labor supply. It creates productivity impairment to the economic sectors to produce at its maximum potential. Due to inter-sectoral linkages, adverse impact in one economic sector cascades into other sectors, whether they are directly or indirectly related to the initial perturbation.

YDL’s Inoperability Input-Output Model

Therefore, we at Young Digital Leaders (YDL) are commencing our attempt to work on a simple Inoperability Input-Output Model, which combines the inoperability level of the sectors and economic impact assessment. The dependence of economic sectors on workforce productivity directly impacts the inoperability level. Meanwhile, the input-output (I/O) framework offers a neat way to describe cascading economic effects and the economic system’s sensitivity through inter-sectoral linkages. This model is based on preceding models.

Sample output:

The I/O framework

Department of Statistics Malaysia (DOSM) published the Supply and Use Tables that are essential in the construct of this I/O framework. We are also leveraging on Gross Domestic Product (GDP) and employment data points from DOSM in this model.

Our model currently estimates the impacts for two measures – Output and Value Added in the Malaysian economy. Output is the broadest measure of economic activity that considers total gross value in production. Meanwhile, Value Added refers to the additional value over the cost of inputs used in production (i.e. the difference between revenues and expenses on intermediate inputs). Value added is regarded as a more meaningful measure of economic impact as it avoids double-counting during each round of impacts. 

The I/O framework allows us to estimate total economic impact through three different categories – direct, indirect, and induced impacts. Direct impacts result from changes associated with the specified sector – in our case, impairments to full sectoral productivity. Indirect, or second-round impacts, result from impacts to the suppliers of goods and services to the specified sector. Induced impacts result from the displaced income of workers in direct and indirect sectors, in impacting further rounds of household spending.

Next steps

After the establishment of a framework for this model, we will continue to perform our analysis with data points related to impaired sectoral productivity in Malaysia. As mentioned earlier, we hope this model will be helpful to illustrate cascading effects of certain changes that are taking place in our economic environment. Especially with the recent restrictions lift, it is important to understand the trade-offs that we have been and continue to experience in regard to the economy.

Source: 

  1. Santos et.al (2012) Risk-Based Input Output Analysis of Influenza Epidemic Consequences on Interdependent Workforce Sector. National Center for Biotechnology Information
  2. Santos JR. Inoperability Input-Output Modelling of Disruptions to Interdependent Economic Systems. Systems Engineering. 2006;9(1):20–34
  3. Haimes YY, Jiang P.  (2001) Leontief-Based Model of Risk in Complex Interconnected Infrastructures. Journal of Infrastructure Systems. 2001;7(1):1–12
  4. Congressional Budget Office, U.S. Congress (2006)A potential Influenza Pandemic:Possible Macroeconomic Effects and Policy Issues