Is big tech bleeding our productivity?

I’m probably not the first to think of this, but what if the big tech companies are the source of, not solution to, our productivity woes?

At the Atlanta Fed’s conference on AI a couple of weeks back, one of the panellists made the point that big firms like Google can afford to hire top talent – MIT computer science PhDs being the quintessential example – and not care whether these hyper-productive people actually contribute more to the firm’s bottom line than they take out in their hefty pay check.

Google is happy because Amazon isn’t employing the talent, and the talent is happy because they’re left to their own devices to research whatever they please, and to take home a suitably enormous salary. Occasionally a PhD comes up with a ‘moonshot’ that makes Google a ton of cash, and at any rate, Google’s market power means it has masses of spare profits it can spend on draining its competitors of talent.

It’s often noted that in advanced economies afflicted with productivity problems, there is a small group of highly productive firms and a long ‘tail’ of unproductive ones. I suspect part the problem with much of the tail is its inability to attract top computer scientists, so it can’t make full use of major advances such as AI that the FAANGs can exploit at will.

We saw something similar in the run up to 2008 with all the PhDs being hoovered up by Goldman Sachs and the like. Economists at the Bank for International Settlements have posited a version of this hypothesis to explain how financial booms can cause productivity-harming misallocations of resources in the real economy. I would submit that Goldman et al are likely still partly to blame, but big finance has now been joined by big tech.

If some all-powerful god of bureaucracy were to spread MIT computer science PhDs evenly throughout the economy, it seems highly likely that the productivity advantage of the FAANGs would disperse into the unproductive tail. Competition would improve, growth would rise, and the productivity puzzle would be alleviated if not completely solved.

A further nefarious consequence of the FAANGs taking all the talent is that regulators can’t keep up. Those who can afford to pay (read: Wall St and big tech) will always be several steps ahead. Financial regulators in the US have openly admitted they don’t have the resources to examine every machine learning model in detail, even while fretting that these could be adding to financial instability.

I admit it’s easier to point to the problem than to fix it. I’m still thinking about that part.