PSST and Fragility

I actually think this short explanation gave me a better idea of what you mean by PSST.

I think what might be relevant is how a chain is connected to the rest of the economy. For example, the financial sector is highly connected to nearly every chain so a financial crisis should cause a severe break down. Whereas if the yogurt-producing sector has a shock, I doubt this would matter much.

One the big issues we need to deal with in macro is sectoral shocks and their spillovers, just generically speaking.

"...we both see production processes as increasingly complex. However, Levine argues that this makes the economy more fragile, and I am not so sure that I agree."

This reminds me of "normal accidents" which occur because of high complexity and tight coupling.

http://en.wikipedia.org/wiki/System_accident

Banks became tightly coupled leading up to the financial crisis (I can't seem to find it quickly, but I believe Andrew Lo of MIT produced some graphs to display how banks became more interconnected over time), and CMOs became incomprehensibly complex. Supply chains are often tightly coupled, and complementary industries are too, in a way. I'm just brainstorming here...

If you are interested in peeking at other ideas from software, consider looking into Inversion of Control (IoC). It's the modern concept behind enterprise software where a central code no longer runs the business logic and instead the logic is decoupled to problem-specific components. The whole idea of IoC came from the realization that centralized code had the confounding paradoxical combination of being both extremely rigid and fragile/brittle. It was difficult to modify and often small changes could drop the whole enterprise.

So while in IoC software there are central components, they are often just stores of routing tables or registries where other components can be looked up. It lacks the hardened dependencies found in classic centralized code.

The result is that you may have two pieces of software, both composed of many modules doing identical things, but the control of how they dictate execution has profound effects on their stability, extensibility and reusability. It's a powerful concept that has taken over in the last few years (the Spring Framework is a good example).

"CMOs became incomprehensibly complex" -- false! They just became boring, agreements assuming other agreements have no risk problems so most attention spent on defining and dividing expected up side, with little focus on massive down side. (Like most legal agreements, including computer agreements that everybody presses the "I agree" button.)

Dr. Michael Burry (with Asperger's) read many of the Mortgage Obligation small print, and realized that piling high risk junk on top of junk doesn't really reduce the risk, despite the AAA ratings agency's saying it does -- and a globally rising underlying market hiding the risk in the models. (He made almost a billion in The Big Short.)

One way to reduce risk is to inter-connect more tightly. My mental model is a sky scraper with cables attached to buildings around it. The more cables, the safer -- but also the more disastrous if there is a fail. I fear those calling for "more safety" are usually calling for more cables; tho many are also calling for height (risk) limits.