The Internet Was an Economic Disappointment

My last newsletter was about the potential economic consequences of artificial intelligence. I argued that although they may seem large, history shows that they will be slower to materialize than most people think.

People came back to me, citing a 1998 prediction that internet growth would slow down and that it would be clear that the internet has had no greater impact on the economy than the fax machine. In a throwaway article I wrote for The Red Herring, I actually said that. I don't recall writing it, but I suppose I was trying to provoke.

Evidently, I was wrong about the internet slowing down and have now admitted it. So it goes. If an economist claims to never have made a wrong prediction, I will show you an economist who is either dishonest or unwilling take intellectual risks.

How wrong was I about the internet's economic effects? This shouldn't be about me. Or, has the past decade generally vindicated visionaries, who believed that information technology would transform everything? Oder have they vindicated technoskeptics such as Robert Gordon, an economist who claimed in a 2016 book the innovations of late 20th century and early 21st centuries were less fundamental than those made between 1870-1940.

The numbers show that the skeptics have the upper hand.

We looked at the 10-year rate of growth in labor productivity. This indicated that although information technology did produce an increase in economic growth between mid-1990s to mid-2000s, it was quite modest and short-lived. Let me try something a little different today.

The Bureau of Labor Statistics provides historical estimates of labor productivity going back to 1948. This estimate of total factor productivity is an estimate of the productivity of all inputs including labor and capital. It is widely used by economists to measure technological progress. These measures should be able to grow steadily if there is truly fundamental technological innovation, particularly total factor productivity.

Let's now look at the 25-year rates for change in these two measures. I chose 25 years because it's about one generation and partly because my bad prediction was 25 years ago. Here's the labor productivity growth for the 25 years that followed each date on this horizontal axis.

Image

Here's the total factor productivity growth in the same way.

Image

You see the huge productivity boom that followed internet's rise? You don't either.

There are two possible explanations for the difference between the numbers and the hype. The first is to claim that the internet did great things for the economy but that these were offset by negative factors such as a declining work ethic or a mysterious drop in construction productivity.

Another is that the official statistics on economic growth do not capture many of the invisible benefits. The ability to stream live music on YouTube is a great pleasure. However, they are not included in gross domestic product. Official economic growth is a poor indicator of true human progress. This has been true for a while.

Before the internet, the benefits of improved sanitation or the dramatic improvement in air quality since 1970 didn't get captured by official economic data. These invisible gains are bigger than ever before. It's doubtful.

The key point is not that the internet is useless. It has certainly contributed to economic growth. Instead, the argument is that the internet's benefits were not as great as those of other less popular technologies. In the 1920s, about one fifth of American households had washing machines. By 1970, however, nearly everyone had one or had access to one. You don't think it made much of a difference. Is it possible that it did not make a bigger difference than having widespread broadband access?

The fact is, while information is important, most of our lives are still based on physical things or services. These have not been greatly affected by the internet.

It's possible that artificial intelligence or, at the very least, what we call artificial intelligence, whatever their merits, will be a major thing. However, one thing we should have learned about the history of information tech is that even things that look very glamorous don't necessarily make them particularly useful.