Economic Interconnectivity and Big Data

     The world economy is a huge set of interconnections. One type of job depends on other types of jobs; if a job type disappears it is likely to affect many other job positions. Scarcity of resources of one type can affect the prices of many cascading products. If the world does shift from fossil fuels to renewable energy sources, new jobs will appear and old ones will change or go away.
     The interconnectivity also causes great fragility as the world gets larger and there are more dependencies. Imagine, if you can, people waking up tomorrow and deciding that the Internet is no longer of interest (I can easily remember when it didn't exist) -- how many products would no longer have a market, how many people would no longer have a job, how would it affect others (advertising, for example -- and printed newspapers might surge back into dominance)?
     Once upon a time, I was interviewing with Google and, as part of the telephone interview, we discussed potential projects and interests. I put forth the idea that, since Google was well designed to integrate knowledge and had such massive data storage and access, they would be well able to create an economic model of interconnected occupations and salaries. At this point in time, I would like to also add in products and localized market prices.
     Why bother with any type of tool? Why not just make the change and see what happens? The main advantage of such a tool is to have a better ability to forecast the effects of policy changes. What really happens if minimum wage is increased to a living wage? What happens if the illegal immigrants who are largely responsible for hand harvesting of our fruits and vegetables are kept away -- what will be the effects on produce prices, truckers, grocery stores, and so forth? What jobs are affected if private transportation is minimized and public transportation maximized?
      Such a project would be impossible if every individual, unique job, discrete part, and location had to be tracked. Luckily, items can be aggregated -- 500 Blue F-150 trucks should only have a quantity value change over 1 Blue F-150 truck (but, at the same time, there needs to be a way of describing Red F-150 trucks without having a fully different item). There is a lot of work to be done and it would still be a difficult project but certainly within the capabilities of many of the larger data handling companies -- Google, Facebook, Amazon, IBM, Microsoft, ... What would be the Return On Investment (ROI) for such a project? It's really difficult to know but it would be a valuable service/project that should be of use to governments and businesses around the world.
     I would suggest architecting such  a project as an iterative accretion of data. Start with something relatively small -- a loaf of bread. The loaf of bread has a set of occupations associated with it -- baker, packers, delivery people, stockers, advertising, payroll, Human Resources, etc. It also has a set of ingredients -- flour, yeast, filtered water, possibly milk, salt, and so forth. Each ingredient has an amount which acts as a ratio of strength in the links to the bread. Each ingredient has its own delivery and production chain which each have associated costs and value. It would be considerable in itself but the greatest value would be the fact that it is still small enough to be thrown away. New links and new data structure values will be discovered to be needed as the database develops. Now do it over (iterate) with those better values and links. Do it again if needed. Now add butter to the bread and continue on.
     There are also usability concerns. The bread company may start off selling only white bread and then add rye bread -- each with their own percentage of sales. How does one substitute recipe ingredients? How do you change the dependencies and the ingredient ratios? What happens if a problem ruins the rye crop for the year? If modelling an auto, how easy is it to change the model from gasoline to electric? Not only is there a substitution of an "ingredient" but the interconnections to suppliers, dealers, raw materials (batteries, possibly lithium) change. The model must be able to be changed easily because modelling the existing situation may be interesting but comparisons are what gives the most value.
     How would you address such a problem? What do you see as specific practical benefits from such an economic model? Is there some subset of such a model already in existence that could be used as the core of expansion? How are unpaid people incorporated into the model, recognizing that the system falls apart without them -- even if they are not considered to be part of the Gross Domestic Product (GDP) or a paid occupation?
     While I find the project fascinating just from a theoretical basis, I keep finding more and more potential uses as I consider the matter.