Tuesday, April 14, 2020

Modularity, Reconfigurability and Institutional Robustness: A post-COVID19 model

COVID-19 crisis has revealed the vulnerabilities of our interdependent and globalized economies, where highly specialized and focused manufacturing industries are distributed across the world and supply chains that are managed through numerous tenuous links, where the resultant structure is only as strong as its weakest link. At this time of emergency, incentives to reconfigure businesses to produce essential items did not work as efficiently as we generally credit private enterprises: governments had to step in with war-time powers to compel businesses; broken supply chains had to be restored through intergovernmental airlifts. We are rudely awakened to limitations of private enterprises for rising to the challenges of a globalized emergency, and a renewed appreciation of efficient governance is in order.

How might such governance be organized?

There are two previous models where nationalism provided the impetus: wartime efforts in the west and in the east. The US and UK used war powers acts to first subdue and to ultimately kindle the awesome power of private industries manufacturing automobiles and railways to make tanks, battleships, fighter planes and bombs; the Soviet dictatorship did the same, perhaps even more effectively, to convert lumbering shipyards and steam engine manufacturing factories to make tanks and rifles. The problem was that at the end of the war both had to keep the war-time factories working for a while so as to feed the nations—so we and the Soviets had our Korean war and our Vietnam.

Is there a less destructive way out of war-time diversion of resources? Or must we always be burdened by our momentum? Is there a way that minimizes the retooling for reconfiguration of industries to enter and then to leave the war-economy, be it due to human conflict or a future pandemic?

A recent concept in evolutionary biology is ‘modularity’—a term borrowed from engineering and systems science. Like Lego blocks, are there basic building blocks of gene circuits that are reconfigured by evolution to produce the bewildering diversity in nature? Moreover, such modularity is thought to provide evolutionary robustness—the niche vacated by the extinction of a species is quickly replaced by organisms that evolve through reconfigured genetic modules. Modules reconfigured perform novel functions that their previous ensembles didn’t. There are lessons here to be learned.

What we need is an abstraction, a conceptualization of modularity of manufacturing industries, and of supply chain Lego pieces. A high-level government agency would need to examine each industry to identify the modularity and reconfiguration strategies for natural (pandemic, earthquake, global-warming) or man-made (foreign or civil war) catastrophes. They will be the intelligence gatherers, systems modelers, and will develop scenario-specific contingency plans based on data and model.
They will interface with FEMA, the NAS, the Congressional Budget Office, will be overseen directly by the Congress, and will work in direct consultation with a similar structural entity established through the UN. An agency for the analysis management and design for systemic robustness.

Thursday, March 26, 2020



Estimated Loss to the US Economy due to COVID-19, If the Disease were to Take its Course

Here I attempt a back of the envelope estimation of the total GDP loss to the USA alone if COVID-19 is allowed to run its course without any containment measures in place.

This assumes a conservative figure (30%) of the oft-repeated range(25 – 75%) of the total incidence rate of COVID-19 if no quarantine or stay-at-home types of segregation measures are taken.

I have used the 2010 demography numbers of the US population by age and sex and used the life-expectancy table of 2010 for the US population from the US census.

Here are some broad stroke assumptions I made for this rough estimation:
1.     2010 US census age and sex distributions
2.     2010 US life expectancy distributions
3.     Infection rates follow a normal distribution (this is the epidemiological standard)
4.     Males and females are equally infected (though not equally affected)
5.     30% of the population are ultimately infected within 1 year (the range provided by epidemiologists are from a low of 25% to a high of 70%)
6.     All age groups are equally infected (though not equally affected)
7.     I have used a sliding scale of treatment weight for various age groups, assuming 100% treatment rates for all age groups, except the above 65 years, in which I assumed 80% treatment rates (an ad hoc assumption but something that is seen in Italy where the resources are saturated).
8.     Used a sliding scale of death rate (0.01 – 0.035) per age group. The upper range is a low estimate. Current mean rate according to WHO is 0.045 (which is widely considered to be an over-estimate, but current >75 years age group mortality rate is ~14.8%, so I have erred in favor of an under-estimation)

Using these parameters, I have calculated the DALY lost due to COVID-19 in one year. DALY, Disability-Adjusted Life Years is an economic measure used to estimate the loss to the economy due to a particular health issue.

DALY = Number of cases x disease duration x Disability weight + YLL

Here, the number of cases was estimated according to the assumptions stated above; disease duration was assumed to be 2 weeks for below 65 years and 3 months for >65 years; disability weight was 0 for all age groups except for 65+ yeas which on average was assumed to be 20%.

YLL is more complex, and is automatically calculated by the R-package “DALY” within R, and is defined below (ref: https://www.ncbi.nlm.nih.gov/pubmed/23927817)



Below is a snapshot of the input data:


And below is the resulting distribution of DALY estimated with the above parameters:

The mean value of DALY loss turns out to be: 41,056,030 years.  This is the estimated mean loss of DALY to the economy due to COVID-19.

To translate that to the loss to the economy, we will need to multiply the per capita annual GDP of the US by this number:

The total estimated loss to GDP in one year due to COVID-19 if no specific additional measures to minimize the natural course of COVID-19 were in place

= 41,056,030 years x (per capita GDP of 2019) per year
= 41,056,030 x $65,116
= $2.67 x 1012
=$2.67 Trillion

Assuming again the US GDP of 2019, this means ~13% expected drop in GDP in 2020 relative to the previous year's GDP.