The Neighborhood Impact of COVID-19, an Impetus for Innovative Testing Strategies
Many are making strong arguments in the appeal for more comprehensive coronavirus infection testing. For instance, Paul Romer, former World Bank Chief Economist, contributes greatly to the debate with his modeling and provocative spending comparisons, to encourage testing everyone in the nation once every two weeks. (His comparison with soda production is particularly provocative.) However, with the possibility that a large national testing program will remain unattainable for the foreseeable future, we see a powerful role for geospatial data analytics in shedding light on the potential for innovative testing strategies that can have sizable net positive impacts (maximize blunting of transmission and spread against deployment of resources), can protect vulnerable populations, and can be seamlessly integrated into the flow and fabric of daily routine (minimize disruption).
Our own modest contribution in this area is motivated by the inherent difficulty in maintaining social distancing practices on mass transit and the potential for prolonged exposures as defined by the CDC. Our study analyzes Bronx subway ridership composition based on home neighborhoods of riders, and asserts lower income neighborhoods may be significantly more vulnerable to economic disruption and health destabilization.
Over 80% of the U.S. population resides in urban environments and many urban residents have a significant reliance on mass transit. Furthermore, mass transit and its unrestrained use is critical for the economic vitality of large urban environments. We believe there is great merit in studying urban mobility patterns, including travel distances, dwell times, trip chains, route repetitions, anchor points, and activity spaces, and population dynamics like convergences, ephemeral clusters, and sustained pools in the context of the coronavirus, and perhaps supplemented with contact tracing data. Some patterns of mobility and population dynamics are known, such as convergences at mass transit ingress and egress points, pooling on subway cars during high-occupancy commuter periods, and clustering at mass transit interchanges. Mathematical frameworks for simulating movement, such as T. Alex Perkins et al in “Theory and data for simulating find-scale human movement in an urban environment,” or for analyzing virus spread patterns, such as Andrew J. Tatem et al in “Assessing spread risk of Wuhan novel coronavirus” could help us model coronavirus transmission and spread, and the impact to unique population groups.
At this writing the U.S. is just now reporting a consistent level of testing at or above 500,000 tests per day that some experts indicate is the bare minimum needed to track and contain (not diminish) the nation’s coronavirus outbreak. Greater emphasis on modeling mobility patterns and population dynamics could lead to some truly innovative testing strategies, that maximize impact and curb negative consequences.
How to predict covid19 impact on urban retail.
How to predict covid19 impact on urban retail. A geospatial risk evaluation model.
As soon as the covid19 crisis broke out, lockdown forced retail stores to close in order to avoid contagion. EIXOS started developing a geospatial model to predict the economic impact on urban retail. In parallel, an MIT research team published a proposal for a predictive model. MIT model assesses risk factors complementary to those assessed by EIXOS.
On 24th April, the Massachusetts Institute of Technology research team formed by Seth G. Benzell, Avinash Collis, and Christos Nicolaides published an article in the scientific journal Proceedings of the National Academy of Sciences of the United States of America (PNAS). They also propose a risk model for calculating the impact on small business.
EIXOS geospatial covid19 impact on urban retail risk model
It is based on data on the retail composition of entire cities. These cities were mapped by the EIXOS international geographical network.
The model evaluates the risk of excessive financial stress caused by covid19 lockdown, that could lead to business closure.
We measure the risk activity by activity, taking into account several factors:
– Whether or not it is essential in terms of lockdown, according to authorities.
– If it is labor intensive, which adds financial stress.
– If it stores expiring product at the point of sale.
MIT risk model
MIT model is based data captured through user surveys. It also uses mobile phone data to measure user concentration in several types of stores.
It uses two indicators:
– user’s priority for a given sort of business, according to the preferences of the users surveyed.
– probability of infection amongst users and store staff in the normal course of business.
The (weighted) sum of these factors described in both models, EIXOS and MIT research team, gives a final potential risk for each activity.
Risk level is expressed as a percentage of business closures in a given activity (e.g., “restaurant”). It indicates the number of stores that could close as a result of covid19 impact financial stress.
Restaurant is one of the activities that adds the highest risk values. The model predicts up to 60% of restaurant closures for a full impact scenario. This means that 6 out of 10 restaurants are at risk of closure as a covid19 impact direct consequence.
The case of London, Sant Cugat, Reus and Bilbao
The resulting predictive model will be applied in London first. Sant Cugat del Vallès, Reus and Bilbao will follow immediately.
Retail recovery strategy. Opportunization.
EIXOS has spent the last few years defining new strategies for recovering and boosting urban retail.
First method is the location based business opportunity detection service. It finds clear opportunities on specific locations to start a new business. Therefore, it also analyzes retail composition and availability of vacant retail spaces in order to do so. This service has been successfully deployed in several cities, such as Bilbao (Bilbaoin.com).
Second method is the so-called optimization of retail composition service. A location intelligence algorithm detects new possible combinations of retail stores in a city. Then, it proposes new configurations to improve the retail health parameters. The result is a more resilient and attractive to consumers retail scenario.
Combining both methods, opportunity detection and retail optimization, is what we call “opportunization”.