The Baumol effect: not a disease but a cure
One of the most important phenomena in economics is the Baumol effect. That is: when productivity rises in one part of the economy — often the manufacturing sector — wages rise in all parts of the economy, even in industries where productivity does not grow. This helps to explain why musicians, hairdressers, economists have seen their wages rise dramatically even if they produce no more output than they did 100 years ago.
The Baumol effect is like a tide that lifts all boats. It brings higher wages to everyone, sharing the benefits of productivity growth across the economy. It seems like an obviously good thing to me. But the Baumol effect is normally described as a “disease”, including by Baumol himself. He described it as a “cost disease”, which drives up the costs of services, drags down productivity across the economy as a whole and even hurt people on lower incomes.
So which is it — disease or magic sauce? In this piece I’ll argue that the benefits of Baumol far outweigh the downsides, and that the Baumol effect is powerful tool we should try to use more.
Why did Baumol call it a disease?
Baumol’s version of his “disease” actually combines two effects. The first is the famous finding, outlined above, that productivity gains in one sector lift wages in all sectors. The second, less positive part is that productivity growth will necessarily be slow or stagnant in some parts of the economy — typically the service sector. The combination of these effects, according to Baumol, is that the stagnant sector takes up a larger share of the economy over time, and that it gradually drags down overall productivity growth. Because services get more expensive, we have to spend more on them. Because they then make up a larger share of the economy, the economy gets less productive over time. Sounds bad.
And Baumol’s pessimistic theory matches the empirical data since 1967 (when he first published) pretty well. The rate of productivity growth has slowed down in advanced economies, and the service sector has grown as a share of the economy. The prices of many services have risen, while prices of many manufactures have fallen.
If you’re looking for an explanation as to why growth has slowed down in advanced economies, Baumol provides a simple, almost mechanical explanation, which presents the slowdown as inevitable. Maybe it’s Baumol’s world and we’re just living in it?
The magic that Baumol overlooked
But I think treating Baumol’s findings as an irreversible prophecy of decline is a mistake. In fact, the Baumol effect as a powerful tool we can use to try and grow our economies.
To see this, we need to separate out the different parts of Baumol’s prognosis. The first and more famous effect — that productivity gains in one sector lift wages in all — is a good thing. The second — that some sectors have stagnant productivity — is a bad thing, but it is not inevitable. It is something we can and should tackle. There is also a third key part of Baumol’s thesis which is often overlooked: demand. For Baumol to work, the demand for outputs from stagnant industries must rise even as prices rise.
Let’s delve into each of these three parts. For the first part — productivity gains in one sector lift wages in all — you can see the magic of the Baumol effect by considering the counterfactual: what would happen if the Baumol effect didn’t exist? What if wages matched productivity within each sector? To make this work, you have to assume some barrier that means people can only ever work in one sector — they can’t switch to another sector with higher wages.
When manufacturing productivity exploded in the 19th century, a world with no Baumol effect would have become more unequal. Workers in manufacturing would have seen their wages rise, but those in the service sector (over 30% of employment in Britain by 1861) would not. There are also implications for aggregate demand. Manufacturing workers — and the owners of capital — would have enjoyed much cheaper services (maybe the stately homes could have all survived), but wealth would be concentrated among fewer people, who would likely have had lower propensity to consume. A world without Baumol would certainly be more unequal, and quite possibly poorer as well.
This is the magic of Baumol — it takes productivity growth in one industry and spreads the benefits to the whole economy.
The second part — some sectors have stagnant productivity — is clearly a problem. It is true that many parts of the service sector seem largely immune from productivity growth. These problems are most obvious in construction, public services, personal services and recreation (Baumol’s original observation was based on the wages of musicians in orchestras). But it is not true that all parts of the service sector have stagnant productivity; in fact, the ICT sector has played an outsized role in the UK’s productivity growth over the last decade. And more importantly, nor is it inevitable that we have stagnant sectors. Many of the recent waves of technology — computers, the internet, maybe AI — have considerable scope to raise productivity in the service sector. That should be an important goal for economic policy.
There are also a couple of challenges to Baumol worth raising here. One is that we just can’t measure service productivity very well. Another is that many of the supposedly stagnant services sector may actually produce intermediate inputs for the production sector, helping to raise its productivity. These are hard for me to adjudicate, but I think one thing is clear: stagnant productivity growth in some industries is not inevitable, and we should aim to avoid it.
The third part — demand for stagnant industries growing more quickly — sometimes gets lost in discussion of Baumol’s “disease”. If services get relatively more expensive over time and manufactured goods get cheaper, why do we choose to buy more services? If the same haircut costs so much more in San Francisco than Cleveland, why do San Franciscans spend their extra wealth on hair cuts, rather than clothes or iPhones or whatever? Baumol’s answer is that services tend to be both price inelastic (demand doesn’t fall much as prices increase) and income elastic (demand increases a lot as income increases). That is certainly true for some services — education is often cited — but it is worth thinking about why people value services in that way. You can frame it in terms of demand curves, or argue that services tend to be higher up Maslow’s hierarchy of needs, but it’s not something that should be assumed away. It seems like a good thing to me — people meeting their wants and needs, even if it doesn’t maximise productivity.
There’s another point about the shape of demand which I think is very important: the sources of demand for high and low productivity sectors tend to be different. It is possible to increase the demand for your high productivity sector through trade — a possible solution to the cost “disease”.
High productivity sectors tend to be tradable — the market for them is typically international, and this is where prices are set. Lower productivity sectors tend not to be tradeable, to be sheltered within domestic economies (there may be a connection between tradeability and productivity growth). That means the demand for each sector comes from different places. For higher productivity, tradeable sectors, demand depends primarily on how internationally competitive you are. You can’t easily redistribute workers from lower productivity sectors into higher ones if they’re operating in a global market. For lower productivity, non-tradable sectors, demand depends on the size of your domestic market. The size of your domestic market in turn depends on two things: one, how much demand you bring in to your economy from the tradables sector; and two, how big the multipliers you get from that demand are. What the Baumol effect does is increase the size of that multiplier — it distributes more income to those in the domestic sectors, who can then spend more of it.
I think the implication of this is important. Economies can grow their most productive sectors through trade, as long as they have internationally competitive industries. Doing so should raise productivity growth, but should also increase demand for local services. This is what economic growth looks like.
You can apply the logic of the Baumol effect to local economies. The most successful local economies generally have a large, productive tradable sector — usually a combination of manufacturing, resource exploitation, tech and business and financial services. Typically they make up around 30% of jobs in a successful local economy (see chart below for England). These industries bring income into the local economy and drive growth — and fuel a more localised, non-tradeable economy, covering retail, entertainment and personal and public services. The bigger and more productive your tradeable sector, the richer your local economy tends to be. But without Baumol spreading the wealth across the area, your local economy would be poorer and probably be able to meet fewer people’s wants and needs.
If you wanted to avoid the impact of Baumol’s cost disease, you might deliberately suppress your domestic sector to grow your tradables sector as much as possible. There is, of course, a rather important country that does this at present: China. China’s strategy of suppressing consumption to fuel investment has been effective in raising productivity up to a point, but it brings two problems. First, it lowers living standards for people in China relative to what they otherwise might be. Second, it relies on other countries — mainly Americans and Europeans who have reaped the benefits of Baumol — consuming more to buy all of China’s exports. This doesn’t really benefit either country, unless they are planning a destructive conflict. Ultimately, trying to fight the Baumol effect is bad for living standards and bad for global economic stability.
Putting the Baumol effect to work
I’ve made the case here that the Baumol effect is more good thing than bad thing. In fact, I think it is one of the most powerful forces in our economies, and we should try to make much more active use of it. How should we do that?
First, we should try to grow the productive, tradable sector of our economies — local economies, national economies, the global economy. These industries create wealth and have it multiplied many times over by the majesty of Baumol. We should want more of them; size as well as productivity matters in these industries.
Second, consumption is good, in moderation. Strong, healthy economies have big productive sectors and big Baumol effects, and they meet their citizens’ needs and wants in the process. The global economy would be healthier if every country leaned in to Baumol and let their people enjoy the standard of living they can afford.
Third, raising productivity in less productive sectors is still worth pursuing. It is possible to grow productivity in many supposedly stagnant sectors, and doing that could help get even more benefit from Baumol. We should park the idea that construction or retail are doomed to stagnant productivity growth. And if we’re trying to use AI to drive economic growth, one of the biggest prizes is enabling productivity growth in typically less productive sectors.
Fourth — this is something I haven’t discussed so far — we should be alert to the possibility that the Baumol effect might not be as strong as it once was. As some jobs — think of tech companies or finance — get more separate from the rest of the labour market, the impact of productivity gains in those industries might fail to raise wages across the board. This may already be underway. It may not be possible to directly fix a breakdown in the Baumol effect if there is one, but there are other policy responses that can increase wage equality.
That’s all. Get as many productive industries as you can. Spread the wealth. Try to raise service productivity. Embrace higher wages. You could base your whole economic development strategy around the Baumol effect. We shouldn’t treat it as a disease but as one of the key forces making our lives better.
