Comment of the Day: Kaleberg: Paul Krugman: Trade and Manufacturing Employment: "The general rule has been that when a robot costs $75K to replace a worker, that worker goes...
...Robots keep getting cheaper and more versatile. That $75K is going to be maybe $50K in a few years. Manufacturing still has a way to go down as a percentage of employment, though it seems to have bottomed out for now.
The thing is that robots mainly do manufacturing, but as they get cheaper and more versatile they are going to start taking away service jobs, like working in fast food joints. Look at McDonald's and those kiosks that eliminate order takers. McDonald's has tried to spin this as a reaction to the rising minimum wage, but it is rather obvious that, at best, keeping the minimum wage low would delay kiosk profitability for a year or two.
So, who is going to buy manufactured goods or eat at a fast food joint? Has anyone updated Leontief's input output model of the economy? Is that kind of analysis just too outdated to bother with? Leontief produced a model of inputs and outputs for the various sectors of the economy to help with war production planning during World War II. The model was basically a matrix, and Leontief's team heroically inverted it using only human calculators. We could see where things were going if we had an updated matrix. We could find the eigenvalues and eigenvectors in seconds, and see just how many of those eigenvalues are above one. It's a rather obvious exercise.
Judging from just about every discussion of trends in the current economy, just about every eigenvalue outside of the prestige goods marketplace is less than one. A robot replaces a worker because it is cheaper. That takes money away from consumers and increases corporate profits. That's a vicious circle of less than one eigenvalues.