Repost: Distributed service-sector productivity

How the internet might still reshape industry

"Telephone operators, 1952" by Seattle Municipal Archives, CC BY 2.0

In the early days of my blog, a lot of people were about whether there would be a technology-driven productivity boom in the 2020s, and if so, which technologies might drive it. Usually we think of new innovations driving productivity — computers replacing typewriters and paper spreadsheets, etc. But sometimes, older innovations can give huge boosts to productivity when businesses figure out how to use them in new ways.

In “Distributed service sector productivity”, I explained how this happened with electricity, and I predicted that we might see something similar happen with the internet. It’s been a while now since we wired the world together with broadband and browsers and smartphones; a lot of people think that this produced only a small, temporary burst of productivity in the late 90s and early 00s. But I suspect that the internet still has a lot of productivity left to deliver, when we start using it to change the way we organize our businesses. Remote work isn’t just about avoiding commutes — it’s about allowing people to find new ways to work asynchronously and independently. Just as factories reorganized themselves physically to take advantage of electric power, businesses today might fragment and disperse themselves in order to use online collaboration to allocate tasks better and to cut out unnecessary work.

In fact, almost two years after I wrote this post, there are some signs that this is beginning to happen. The Economist’s Arjun Ramani recently wrote a very good article about how remote work is allowing businesses to increase their use of freelancing, outsourcing, and other sorts of collaboration — exactly as I hoped they would. Here is a Twitter thread where Ramani describes the data that supports this conclusion. This makes me more hopeful that the “Roaring 20s” are already becoming a reality.

Anyway, here’s what I wrote back in March 2021:

---

Everyone is talking about whether the pandemic is going to create a long-term shift toward remote work. But so far I’ve seen very few people talk about the possibility that it will change our entire system of production. The possibilities for productivity improvements — and for disruptions of existing businesses, cities, and infrastructure — could be enormous.

Since about the mid-2000s, productivity growth has slowed to a crawl. That doesn’t mean technology has stopped advancing, since lots of other things go into productivity — education, population aging, laws and regulations, international trade, etc. But it does mean that we would like technology to help boost us out of those doldrums, and make up for some of the other factors weighing down productivity growth.

I’ve written before about how I’m optimistic that cheap solar and batteries will spark a productivity boom, by making it easier to produce physical goods. Similarly, some people are very optimistic that machine learning will allow automation of many job tasks, boosting the amount that each worker can accomplish (or, in a less optimistic scenario, replacing some workers entirely).

But there’s another general-purpose technology that might spark a second, parallel boom: The internet.

Which is a weird thing for me to say, right? After all, the internet has been around for a while now. Predictions that it would lead to the “death of distance” were laughably wrong, as people piled even more into city centers. Meanwhile, the explosion of mobile apps and social media exactly corresponded to the recent productivity slowdown. If the internet was going to generate a burst of productivity growth, wouldn’t it have already done so?

Well…maybe not. Looking back at history, we see that general-purpose technologies often take a long time to start lifting productivity by measurable amounts. The reason is that when new technologies appear, you can’t always just swap them out for existing ones — you often have to entirely reorganize your systems of production around the new technology, and that’s a difficult and expensive process.

The example of electricity

There’s a famous story — well, at least famous in the world of economics — about how electricity drove productivity growth. Most of the key electrical inventions (light bulbs, generators, AC, etc.) happened in the 19th century, with a particular burst of innovation in the late part of that century. But U.S. productivity growth wasn’t very fast during that period, and accelerated only during the 1920s. To some extent, this was just a matter of building out the infrastructure; places that were lucky enough to be located near hydropower sources saw their productivity boom begin earlier. That implies there are at some ways in which you can just substitute electricity for existing energy sources and get immediate gains (lighting being an obvious one). But interestingly, factories were very slow to adopt electricity, and industries that electrified early didn’t see big productivity gains til the 1920s.

What happened? In a pair of famous papers in 1989 and 1990, economist Paul David offered an explanation (summarized here by Tim Harford). Basically, at that time, factories used centralized steam power, which was transmitted throughout the factory by a bunch of giant machinery. Simply swapping out a big electric motor for a big steam motor got you a small boost, but not much; it generally wasn’t worth the cost, so if you did this, your productivity would usually go down rather than up.

It was only once factory owners started building entirely new types of factories that they were able to realize the true gains from electricity. Basically, you could put a little motor at each workstation and power it through electric transmission lines. This meant that instead of having to keep a huge machine constantly turning, you could run each little machine only when you needed to. Not only did that save a ton of energy and make factories much nicer and safer places, it allowed workers to do things when they needed to be done instead of adjusting their workflow to the rhythm of a giant machine. That allowed all sorts of flexible production lines that you just couldn’t make with steam power. And productivity followed.

Something similar may have happened with computers. In 1987, Robert Solow famously quipped that “you can see the computer age everywhere but in the productivity statistics”. Only a short time after, a productivity boom began, driven in part by computerization. Part of that came from simply swapping out existing tech for computers — word processors instead of typewriters, disk drives instead of filing cabinets. But computers also allowed production to reorganize itself, with the rise of oursourcing. When electronic records and documents and written communications could be transmitted easily between companies, it became easier to split supply chains into pieces and have each piece specialize in what it did best. I’m not sure this is true; (I can find lots of research on global supply chains boosting productivity, but not as much on domestic outsourcing, so this is mostly just a conjecture so far.)

Anyway, the general point here is that in order to realize really big gains from a new general-purpose technology, you often have to figure out and implement whole new ways of organizing production in the economy.

Services and productivity growth

When we think about boosting productivity growth in our economy, we have to think about service industries. Manufacturing productivity tends to grow faster than productivity in services. Here’s a chart from Ana Maria Santacreu and Heting Zhu of the St. Louis Fed for 24 OECD countries during 2000-2014:

Manufacturing was only about 17% of value added, but contributed 42% of the productivity growth!

This creates a problem for overall productivity growth if demand for services increases faster than demand for physical goods. Maybe eventually there comes a point where you just can’t keep buying a larger volume of stuff. Demand growth never vanishes, of course — you can keep buying better and better stuff, better cars and nicer TVs and so on, but eventually the growth slows down. And that means that services occupy a larger share of the economy.

In fact, you can probably see something like this happening. Consumption of services has outpaced consumption of goods in the U.S. economy for years (though COVID has produced a slight reversal):

That’s a problem for productivity growth, unless we can figure out how to improve productivity in services. And that’s where the internet might finally start to help.

Imagining the possibilities: Beyond WFH

In a recent Bloomberg post, I wrote:

The social distancing that Covid-19 forced on society might be teaching us how to use online services in a more productive way. Remote work could allow companies to distribute their workforces to low-cost locations, and could nudge them to reevaluate the necessity of many meetings and routine office tasks. A shift from brick-and-mortar business to e-commerce might cause supply chains to become less fragmented.

Robert Gordon, the strongest proponent of the idea of technological stagnation, doesn’t seem like the kind of guy who would be inclined to agree. But he does! Here’s what he had to say in a recent interview:

This shift to remote working has got to improve productivity because we’re getting the same amount of output without commuting, without office buildings, and without all the goods and services associated with that. We can produce output at home and transmit it to the rest of the economy electronically, whether it’s an insurance claim or medical consultation. We’re producing what people really care about with a lot less input of things like office buildings and transportation. In a profound sense, the movement to working from home is going to make everyone who is capable of working from home more productive…It’s very possible that the transition to working from home – once we get the rest of the economy sorted out – will give us a sizeable jump in the annual growth of productivity.

So, just for fun, I want to build on what I wrote and what Gordon said, and brainstorm some ways that production could be reorganized around remote work technologies like Zoom and Slack in ways that substantially increase productivity, especially in those difficult service industries.

1) Work-from-home

This is the obvious one, so let’s get it out of the way. Working from home, either part of the time or all of the time, means you don’t have to commute as much. That saves not just hours out of the week, but also dramatically reduces the cost of transportation — gas, car depreciation, bus fare, train tickets, and all of the physical resource costs that those pay for.

It also allows people to dual-use living space and office space with home offices, freeing up downtown office space to become housing or retail…though I suspect this effect is a lot less substantial than people think, because working out of your home gets pretty cramped and you eventually need more space. But saving on commute time and costs will be big.

2) Location arbitrage

If you can work remotely, you don’t have to live in an expensive place. Now, maybe you still want to — Manhattan is a fun place to party, not just a convenient place to work. But truly remote work will mean some people move to where the rent is cheaper. That will increase productivity by lowering housing costs. The effect will be even greater if cities learn that they can compete to get remote workers by furnishing cheap housing.

In fact, the extreme form of location arbitrage is international. Why work from a cheap-ish beach town in Florida when you can work from an insanely cheap beach town in the Philippines? Of course, this comes with an offshoring component as well; companies will be able to hire Filipinos more easily, with all of the cost savings that entails.

3) Task outsourcing

Right now, outsourcing happens at the level of business functions — you outsource payroll, or IT, or logistics, etc. But a shift to remote work might allow outsourcing at an even more micro level. Need someone to brush up your presentation before the big meeting? Need someone to run a regression or wrangle some data for you? Want someone to pore over a company’s financials and report key metrics? With remote work tools — and more importantly, with everyone getting used to a standardized package of remote work tools! — this sort of micro-outsourcing might become possible, enabling even greater specialization and flexibility in the production of office tasks. (Update: Adam Ozimek of Upwork calls this practice “hybrid teams”.)

4) Improved outsourcing management

Outsourcing is very common, but one drawback is that it’s difficult to monitor what a contractor is doing. This creates a natural tradeoff between companies doing things in-house and contracting out. But with remote work tools, it gets a lot easier to check on what your contractor is doing — monitor progress, give suggestions, observe results, etc. This works for manufacturing industries too, of course; if you contract out to a factory, it’s far cheaper to tour that factory remotely than to fly all the way to Bangladesh. In fact, remote work blurs the line between companies, making within-firm and between-firm cooperation more similar.

5) Efficient time management

When you’re at the office, you’re there for a block of time. If you finish all your tasks, maybe you can find something else to do, but that won’t be very productive. Or maybe you just goof around on Twitter. Studies show that people are unproductive for a large fraction of the time they’re at the office.

But when you work remotely — either at home or in a co-working space or in a cafe or wherever — you manage your own time, and so as soon as you finish your work tasks, you can go do something else. You can take care of the kids, or clean the house, or go shopping, or watch Netflix, or goof around on Twitter, or work on your side gig or your startup.

(Note that this will increase measured productivity even more than actual productivity. Time spent goofing off at the office is now counted as labor hours, even though it’s actually a form of leisure. But when remote workers respond to surveys, they probably will only count time on task.)

In addition, remote work is more likely to be asynchronous. That could force management teams to learn how to cut down on unproductive meetings.

6) Telehealth and distance education

Two sectors where productivity improvements have been notoriously slow are health care and education. (Caveat: It’s hard to measure productivity in health care, because it’s hard to measure changing quality of care.) Obviously telehealth and distance education aren’t going to totally replace in-person versions; there are lots of things you need to be in a room for in both cases. But on the margin, doctors communicating with patients virtually and students taking some classes online will help defray the costs of these big-ticket service items.

Telehealth reduces commuting time, and cuts down on needed office space (this latter being why psychotherapy was cheaper over Zoom even before COVID).

Distance education offers the even more tantalizing possibility of economies of scale — for some kinds of education you need small groups or one-on-one interaction, but sometimes one lecturer can address 200,000 people as easily as 200, and online course and homework management software also has obvious cost savings. Distance education hadn’t really caught on before COVID, and the pandemic has demonstrated how 100% online learning is clearly inferior. But there’s plenty of hope for marginal changes that will improve educational productivity.

Productivity and society

Add all these things up, and the benefits to productivity could be huge. Reduced commutes, efficient time management, and much cheaper housing will batter down the main costs of service industries — human beings’ time and space. Asynchronous management and the blurring of boundaries between companies will allow production processes to be radically reorganized, perhaps to the degree that electricity allowed factories to be broken up into little independent workstations. And remote work also allows partial remote provision of services that previously required 100% on-site interaction. I’m sure there are more opportunities I haven’t thought of; in fact, it’s entrepreneurs’ job to ferret out those savings and build business models around them.

But of course, these productivity gains don’t come without social disruptions and costs to some people (the advent of electricity was certainly disruptive in its day!). Commercial real estate companies may have to pivot from renting offices to renting co-working spaces and retail, or even to residential real estate. Transit systems will suffer from reduced ridership, gas stations from fewer cars on the road. Cities built around commutes may even have to change shape, putting retail closer to homes instead of offices. Workers who suddenly find themselves in direct competition with people in India or the Philippines — to an even greater extent than in the 2000s offshoring boom — could see downward pressure on their wages. Schools might hire fewer teachers and administrators; medical providers might hire fewer receptionists and other staff. And of course the companies that fail to adapt to the new trends and technologies will lose out to those that successfully adapt.

But this has always been how productivity growth has worked. Growth isn’t just increase; it is also change. The economy of 1950 looked very little like the economy of 1850; why should we expect the economy of 2050 to preserve our jobs and our lifestyles in amber? The only alternatives in this world are stagnation and the unknown. I’ll take the latter 9 times out of 10.

Update: McKinsey agrees with my prediction and gives some quantitative estimates. A recent paper by Barrero, Bloom, and Davis focuses just on the shift to WFH, and estimates it will boost U.S. productivity by 5% total. This Bloomberg article gives a good overview of the emerging optimism.

---