Productivity growth

One of my most nerdy hobbies for the past ~2-ish years has been following the econ Twitter debate on productivity growth. It’s a fascinating debate, because it gets to the heart of a sociological problem that has always interested me — how exactly do new ideas and inventions improve living standards? And what exactly is happening with tech and the economy today?

(For the unfamiliar, ‘productivity’ is a measure of how efficiently human time is turned into goods and services, which is what allows people to be able to work less, or consume more, at their same skill level. e.g. If you’re a caveman, it may take hours to create a loaf of burnt bread, whilst if you’re a modern day worker, it may only take half an hour to earn enough to buy an [above caveman standard] meal. Productivity growth, then, is a measure of how fast living standards are getting better. Since the only thing that improves productivity is technology, it’s also sort of a measure of how fast ‘technology’ — broadly defined — is progressing.)

For the past decade or so, something weird has been happening with productivity growth. At first it looked like productivity growth was just abnormally slow, which led to stagnation theories becoming popular. People said — all the low hanging fruit is gone, we can never reach the levels of scientific advancement that we had in the 20th century, we need a higher rate of invention just to keep up with the slowdown that comes with aging demographics, we just need to adjust to the depressing reality of The New Normal™.

One argument in favor of this theory was that productivity growth had been really sluggish since around 1973 anyway, with the exception of the dot-com years 1995–2005. From this viewpoint, living standards improved in the years 1973–2008 mostly because of things like immigration and free trade, rather than the diffusion of new technologies. It suggested to some people that the drastically life-improving technology of the last 200 years was a one-time phenomenon, and was basically already over.

But then, people realized that actually, productivity had not been slow in the last decade, amongst the very top performing firms. The top 5–10% of companies had crazy high levels of productivity growth, but didn’t show up in the statistics because the rest of the companies were pretty much stagnant. This divergence in productivity between firms is what was driving other commonly-discussed economic phenomena of the past 10 years, such as inequality and to some extent rising rents in cities.


Then the question becomes — why are we now in this weird limbo state, where some firms are much more advanced than others, and yet are not expanding & taking market share from the low productivity firms, as economic theory would predict? Why is the ‘creative destruction’ not happening? I think this is a question of central importance, because if the entire economy was moving forward at the rate of the very best firms, we would be enjoying 1990s levels of prosperity.

So why don’t top performing companies like Google expand? Luckily for us, someone asked Eric Schmidt this very question.

His answer: recruiting, real estate and regulation.

Those are all plausible candidates for limits to the expansion of top firms. Perhaps today’s high productivity companies need a larger supply of smart and educated people in order to expand, and our level of human capital is simply not high enough. Or maybe, today’s top firms depend on clusters of knowledge workers who live in one place, they are constrained by the housing supply in major cities. Or it might be that there has been a buildup of unnecessary regulation made to protect incumbents, that now makes it hard for new, even more productive, entrants to compete.

Last week, a really great paper came out that posited an alternative explanation: an abundance of so-called zombie firms being propped up by credit. These are firms that operate at a loss, but have been living off bank loans for 5 years or more. They are low-productivity, but won’t die. In most Western countries, the GDP taken up by zombie firms is around 10%, which is significant.

The paper investigates the effects of zombie firms on new companies looking to expand — and finds that there is a serious negative effect. More zombie firms in a sector led to lower investment and employment growth, and slower productivity growth. The paper even claims that a huge chunk of the productivity puzzle is explained by this phenomenon. Of course, this is only one paper, but it’s pretty compelling. It fits my intuition that there is a lot of technological progress, but it’s having a hard time making it’s way to the mainstream.

Coming to eat your … productivity !!

Finally, the last interesting thing I read on productivity growth was this revisionist economic history of the 1930s. The author argues (convincingly) that the 1930s were a actually decade of experimentation and rapid scientific progress, that paved the way for the steady growth period of 1945–1975. He argues it was a decade of figuring out use cases for, and implementing earlier inventions, which was why it was the decade with the highest productivity growth in a century.

During the depression, there was much unemployment, but very high levels of scientific employment and R&D. Theoretical materials science became nylon. Splitting the atom became nuclear power…and weapons. Much of the productivity growth came from new science, but just as much came from diffusion of new business processes. The automobile and structural engineering advancements led to suburban malls, shipping containers and depots. Factories were electrified and cumbersome machines became assembly line tools.

Look at that old-timey technology!!

He differentiates between incremental improvements, such as making a car 10% faster, and paradigm-shifting improvements.The world in 1950, of V8 engines and CRT tubes, was a totally different place technologically to the telegraph lines and steam turbines of 1920. But the high growth period that followed was mostly a time of steady but significant incremental advancement. It didn’t require any radical reorganization of society, or change in policy.

I’m tempted to draw analogies between that decade and now. Perhaps smartphones and AI will be looked upon in the same way as critical business process improvements like electrification of factories. Perhaps self-driving cars and urbanization will lead to a change of living conditions rather like the suburbanization of the 1950s. It seems that even though a lot of new stuff has been invented, it hasn’t diffused throughout the whole economy quite yet. I just wonder what exactly those barriers to diffusion are, and how to get rid of them!