Notice how Keynes expected employment to fall in capital goods industries. We have no version 2.0 for an economy so heavily dependent on financial services. I also wonder, even though the US badly needs infrastructure, if any of these newfangled theories allow for how specialized labor has become. One of the reasons that employment didn’t fall sooner is that even seemingly mundane jobs now require employer specific knowledge (computer systems, internal procedures) that make it more costly to bring a new person on board and deters firing.
Put more simply, how is creating jobs in repairing infrastructure going to help unemployed bank workers? Even if they were willing, many will prove not to be able, and will also be living at a remove from where the jobs would be. In an advanced economy, labor is not terribly fungible.
I have never been a fan of Keynes for reasons that are similar to this. Only my objections have always been based around the fact that real capital is not terribly fungible. I feel that the primary cause of recessions and depression is a misalignment of the capital structure and I don’t think those types of problems can be cured by creating more “demand”.
Yves Smith is simply pointing out the human capital has similar problems. If you train too many people to be lawyers and bankers and not enough to be construction workers, you can’t fix that problem by stimulating the economy.