Shouldn't productivity improvements reduce GDP?

[Estimated time: 10 mins]

I was chatting the other day with a friend about AI and economic growth. In the middle of our conversation, she asked: “Why don’t productivity improvements reduce GDP?”

At first blush, this might sound like a silly question. It’s well-established that countries with higher levels of labour productivity—like the US, Norway and Switzerland—also have higher GDP on a per capita basis than countries with low levels of productivity. And the correlation is pretty damn strong. Clearly, then, productivity improvements increase GDP per capita, not reduce it.

Yet my friend’s question is a surprisingly deep one. The idea that productivity improvements should reduce GDP actually makes a lot of sense.¹ And digging into the causal mechanisms behind this relationship between productivity and GDP can be rather illuminating.

Why productivity improvements should reduce GDP

If you find some way to produce the same level and quality of output with fewer inputs, you’ll get a productivity improvement.² Usually, when people talk about “productivity” without specifying what type, they’re referring to labour productivity—i.e. how much output you get for each hour you work. I’m going to focus on labour productivity in this post, but the same principle should apply to other factors such as capital or energy productivity as well—it just means you’ll get the same output for less capital, energy, etc.

So let’s say you’re a carpenter who makes a living carving chairs. It normally takes you 10 hours to carve one chair, so you can produce 4 chairs in a 40-hour week. But you’re just a small fry in your country’s overall chair industry, which sells a million chairs each year for $100 each, contributing $100 million to your country’s GDP.

Now, assume there’s a technological breakthrough. A researcher has discovered a superior chair-carving technique that doubles your productivity. Instead of taking 10 hours to carve a chair, you can do so in just 5 hours. Over the course of a 40-hour week, you can now produce 8 chairs.

But all of your competitors also have access to this marvellous new technique and can also double their chair production. So what happens?

One possibility is that all carpenters coordinate to halve their working hours. This keeps both the number of chairs and prices unchanged. In this case, the chair industry’s contribution to GDP stays the same as before. But behind that statistic, carpenters’ lives have improved as they can now work just 20 hours a week.

Time to carve 4 chairs dropping from 40 hours per week to 20 hours

Of course, coordination (or “collusion”, from the point of view of a consumer) is pretty hard to maintain, especially when lots of carpenters are involved. It’s also usually illegal.

The more likely possibility if there are many carpenters is that chair prices will fall. If you’re still willing to work 40 hours per week, you could produce 8 chairs instead of 4. So you might try to lower your chair prices to $90 to undercut your competitors and earn $720 ($90 * 8) instead of just $400. The problem is that your competitors, thinking along the same lines, will then be tempted to lower their price to $80 to undercut you. Fierce competition among carpenters will drive down prices until they reach a competitive equilibrium price of $50. This 50% price drop matches the doubling in productivity that came from discovering the new chair-carving technique.

This might sound like bad news for carpenters, who end up working the same number of hours for the same amount of pay. Their entire productivity gain has been competed away. But it’s great news for anyone wanting to buy a chair.

What happens to GDP? Well, if the number of chairs sold each year stays the same at 1 million but the price of a chair has dropped to $50, then the chair industry only contributes $50 million to GDP, not $100 million. This means the superior chair-carving technique, which doubled productivity, will halve GDP.

Why don’t productivity improvements reduce GDP?

This may sound pretty confusing. Politicians and economists typically assume that GDP going up is a good thing, and talk about how “productivity improvements” are the key to boosting economic growth. But my chair example is suggesting that increasing productivity should lower GDP and economic growth!! So what’s going on here?

I must now confess that my chair example relies on a bunch of rather unrealistic assumptions. However, this doesn’t mean my example is garbage. Taking some time to understand these assumptions will give us a better appreciation of what’s really going on when we look at economic statistics like GDP.

So let’s unpack them.

Imperfect competition

The first assumption in my chair example that may not hold up in reality is perfect competition among carpenters. Above, I assumed that all chairs were identical and that every carpenter would have equal access to this superior new chair-carving technique.

But in the real world, productivity improvements rarely spread freely. Suppose that the superior chair-carving technique were discovered by a giant chair-making company—let’s call them Big Chair—which serves 80% of the chair market. Big Chair will guard this superior technique as a trade secret. Employees who learn of the technique will have to sign NDAs and non-compete agreements so they don’t leak the technique to rival chairmakers.

In this case, Big Chair might decide not to lower prices at all. Since it can produce chairs at a lower cost than its competitors, the superior technique will increase Big Chair’s profits even if it doesn’t change its prices. People buying chairs are no better off, but Big Chair can fully enjoy the benefits of its productivity improvement.

Alternatively, Big Chair could lower prices slightly to capture an even larger share of the chair market. Let’s say it lowered prices to $95. In my earlier example, when everyone had access to the new chair-carving technique, a price war broke out. But this time, the other carpenters don’t have access to the new technique, so may not be able to lower their prices. Any carpenter that just broke even at $100 per chair would be forced out of the market.

Consumers would be slightly better off, as they still save $5 on chairs. Just not as much as if all carpenters had this superior chair-carving technique.

So this illustrates how, without competitive pressure, producers can pocket the gains from productivity improvements. But imperfect competition only explains why productivity improvements might not reduce GDP. It doesn’t tell us why productivity improvements seem to instead increase GDP. To understand that, we need to look at how productivity affects demand.

Induced demand in the productive sector

Up to now, I’ve assumed that the demand for chairs remained constant even as prices fell from $100 to $50. But in real life, lower prices tend to increase demand—and therefore sales. This phenomenon is known as induced demand.

One of the most famous examples of induced demand is Jevons paradox, named after the English economist William Stanley Jevons in the midst of the Industrial Revolution. In his 1865 book The Coal Question, Jevons observed that when technology improved the efficiency of coal, energy prices fell, and demand surged.

It’s easy to underestimate the power of induced demand because, at a personal level, we might not feel very price-sensitive. Even if chair prices drop by 50%, you’re probably not going to race out and double the number of chairs in your home. Similarly, even if the price of energy halves, you’re not going to start doubling the length of your showers or cranking your thermostat up to 40°C.

But the power of induced demand is easier to appreciate when we zoom out. Imagine someone who had been tossing up whether to replace the 6 chairs in their old dining room set. They might have found the $600 cost to be too steep, but $300 may well have been enough to tip the scales. So the price drop causes them to buy 6 more chairs than they otherwise would’ve. The same applies to energy—as the price of energy drops, society as a whole will find more uses for it.

The power of induced demand is especially noticeable when we have international trade. Although my chair example has assumed a closed economy for simplicity, the global market can absorb a lot of chairs. If the superior chair-carving technique is not spread throughout the entire world, carpenters with access to the technique will be able to win market share off the carpenters without access.

Induced demand in other sectors

There are, however, limits to induced demand. While society can plausibly find plenty of uses for cheap energy, it may not be able to find as many uses for cheap chairs. At some point, society may reach some chair saturation point. Everyone, everywhere, has all the chairs their hearts desire.

By this point, further productivity improvements in chair manufacturing can no longer increase demand for chairs. Chair scarcity has already been eliminated! Still, if those productivity improvements continue to lower chair prices, consumers will have more money to spend on other goods and services. They can use their cost savings to buy more tables, or food, or clothing, or whatever.

You might think there’s a limit to how much demand for other goods and services can increase. After all, aren’t our desires limited? Unfortunately, as I’ve explained previously, a single zero-sum game like land or status can absorb unlimited amounts of money. So competition over zero-sum goods has the potential to negate productivity gains. This dynamic helps explain why our “wants” never seem to wane, not matter how productive we get.

Conclusion

My friend was right to question why productivity improvements don’t seem to reduce GDP. In theory (admittedly with some strong assumptions), it should. In practice, it does not.

Part of the reason is that imperfect competition lets producers capture the benefits of productivity improvements while GDP stays flat. The other part is that when competition drives down chair prices, it triggers induced demand in some form, which absorbs the cost savings and cause GDP to rise. That’s what we always seem to observe. That is why politicians talk about boosting productivity and economic growth as if they were the same thing.

To date, GDP per capita has been reasonably strongly correlated with many other measures of welfare. So even though we know money isn’t everything, many people still use GDP per capita as a rough welfare measure. However, the era of AI is likely to put some pressure on that historical correlation between GDP and welfare. I’ll explain this more in a later post.

Let me know what you think of this post in the comments below!

Images in this post were created using icons by Lima Studio in CC Attribution License via SVG Repo.

If you enjoyed this blog post, you may also like:

1
I’m mostly going to focus on GDP, but the same should also apply to GDP per capita if population stays constant. It is possible that productivity improvements may affect population sizes—such as by changing fertility rates—but I’m not going to get into that.
2
You can also get a productivity improvement if you produce higher quality outputs with the same amount of inputs. Since quality is much harder to measure, I’ll be assuming constant quality throughout this post.

Shouldn’t productivity improvements reduce GDP?