It has been a while since my last post. What does a golf cart, AI and growth have in common?

I will begin talking about Golf carts. In Florida, as well as in regions where people enjoy playing golf, golf carts are very high on demand. One can buy a golf cart nearly at the price of a car, or even at a higher price. For reference, a 4-seats golf cart can be around $10k usd, while a used BMW 325ci of the year 2004 can be somewhere around that price. Of course, one wonders how substitutable the goods are: a car is sturdy, heavier, has AC, travel longer distances, and in general has a significantly higher marginal cost to produce (trivially). The golf cart, on the other hand, is smaller, comfortable (when it is not raining), easy to use, and can be used off-roads (like on the sidewalk or a golf range) where a car is otherwise prohibited. Of course, the reason why the golf cart is so expensive is the result of supply and demand. In a nutshell, there are not that many companies producing golf carts, and the demand is sufficiently high. Anecdotally, Chinese brands have been trying to sell golf carts recently in the US, but they struggle with reputation, as it is arguably complicated and expensive to repair and get the parts once the golf cart eventually breaks down. How does this relate to growth? More on that below.

A topic that has become relevant in the last years has been the question of growth, as shown in this article.

I have written about growth in the past. There’s at least two main reasons why the question on growth is relevant. In the short run, because inflation went up for the first time since many years. More importantly, in the long run, because it is not so clear whether or not AI and technology will be able to support growth in the years to come. This is coupled with the very low world-wide fertility rate that puts pressure on the economy. I am jumping to AI as the main factor of productivity because that is where all the attention (and resources) has been placed lately.

There has been a lot of debate on the future of AI. For instance, AEI has written here on the topic. Notably, Daron Aceomglu recently wrote a paper estimating the potential 10 year run impact of AI. He is not very optimistic.

The purpose of this blog entry is not to give a definite answer, but simply to point at an important component: AI is a natural monopoly. In what sense? It is non-rivalrous and excludable. Similar to golf carts. It is not easy to start a company to sell golf carts. It is certainly non-trivial to start a company that produces AI. Most people expect the future of AI to be monopolised by a few. This puts pressure on the already ever-increasing levels of inequality.

The question on how to go about monopolies is one of the toughest to answer. Most economists are very cautious, because of the potential distortions that fixing them could bring about. Destroying monopolies goes against free-market economies, insofar they by definition require government intervention beyond protection of property rights and maintaining law and order. However, it guarantees a free-market economy insofar it creates a level field for competition where there are barriers to entry. To make matters worse, competition within countries has pushed the world closer to what was lived during Cold War.

Different scholars have taken different approaches on the issue, but there is no clear answer. The questions are twofold: will AI support growth? And do we have to do anything to prevent monopolies? From the economic perspective those are the bigger ones. The smaller ones, like labor allocation or UBI will come as a consequence of the first.