3 ms·

The FANG effect in Silicon Valley, where salaries have skyrocketed over the past 10 years, has nothing to do with current productivity and everything to do with

The FANG effect in Silicon Valley, where salaries have skyrocketed over the past 10 years, has nothing to do with current productivity and everything to do with effective monopolies, whether due to network effects, or data, or whatever gives each dominant company it's lead. Google's core search/ads is hyperprofitable, not based productivity of the current workers, but instead the accumulated R&D that created the great product that it is today. This hyperprofitabity means Google can pay high salaries to many thousands of very talented people, most of whom are not working on core search/ads. Many of whom seem to be "working from the bus" when you stop by Google at 4 in the afternoon. One can speculate on the average productivity. This high pay at Google (and Facebook and Apple) has raised the pay for all talented developers even at early stage startups (where productivity is zero, there's no product yet shipping to anyone. Like my company, please don't take this as a dig against startups). EDIT: Don't have a gripe, it's just an observation

mrfredward 7y ago
Baumol's effect isn't meant to explain the cost of tech workers. Software has an incredibly variable ratio of labor in to economic value out, so it is the thing that causes "cost disease" in other industries, not an example itself.
rahimnathwani 7y ago
"Google's core search/ads is hyperprofitable, which has nothing to do with productivity of the current workers and everything to do with accumulated spectacular R&D that created the amazing product that it is today." Your gripe seems to be that the definition of labour productivity (output, i.e. revenue, divided by the number of people) is flawed. Intuitively you are correct, because a layperson wouldn't consider either of the following to be true: 1) An engineer becomes 10x as productive when they move from an early stage startup to a FAANG job. 2) A labourer doing job X in a company became twice as productive from one week to the next, because 50% of his colleagues (doing job Y) were just fired and replaced with machines. I share the same dissatisfaction with the label (although it's a useful measure, it's something different from how it sounds). However, this dissatisfaction doesn't have any bearing on whether Baumol's cost disease exists. As you say, it does. If labour is the dominant cost for most companies in the economy, then labour productivity is a natural thing to measure, and natural thing to optimise. If, on the other hand, labour is a small proportion (e.g. due to automation, monopoly/network effects, etc.) then it starts to look special, and more just like any other input (e.g. electricity or copper).

mrfredward 7y ago

Baumol's effect isn't meant to explain the cost of tech workers. Software has an incredibly variable ratio of labor in to economic value out, so it is the thing that causes "cost disease" in other industries, not an example itself.

rahimnathwani 7y ago

"Google's core search/ads is hyperprofitable, which has nothing to do with productivity of the current workers and everything to do with accumulated spectacular R&D that created the amazing product that it is today." Your gripe seems to be that the definition of labour productivity (output, i.e. revenue, divided by the number of people) is flawed. Intuitively you are correct, because a layperson wouldn't consider either of the following to be true: 1) An engineer becomes 10x as productive when they move from an early stage startup to a FAANG job. 2) A labourer doing job X in a company became twice as productive from one week to the next, because 50% of his colleagues (doing job Y) were just fired and replaced with machines. I share the same dissatisfaction with the label (although it's a useful measure, it's something different from how it sounds). However, this dissatisfaction doesn't have any bearing on whether Baumol's cost disease exists. As you say, it does. If labour is the dominant cost for most companies in the economy, then labour productivity is a natural thing to measure, and natural thing to optimise. If, on the other hand, labour is a small proportion (e.g. due to automation, monopoly/network effects, etc.) then it starts to look special, and more just like any other input (e.g. electricity or copper).