Scott Sumner  

How should we measure productivity?

Boudreaux on the Progressive M... Bret Stephens's Attack on Ron ...

The Financial Times has an article claiming that Japanese construction workers are far more productive than American workers. But the data they provide seems to suggest the exact opposite:

Reality is the other way around. Despite radically different demographics and essentially no immigration, Japan has consistently employed a much larger share of its workers in the construction industry than the US, although the share has dropped over time. Even at the peak of America's housing bubble, only about 5.5 per cent of workers were employed in construction. In Japan last year, more than 7 per cent of employees worked in construction
So how can they claim that Japanese workers are more productive than American workers? The article points out that Japan builds far more houses per capita than the US, indeed almost as many houses in total, despite a population only 40% of the US. But why is that? Given that Japan's population is falling, one might expect exactly the opposite. The article attributes the difference to cultural preferences:
My colleague Robin Harding has elegantly explained that much of the robust demand for new housing can be attributed to the Japanese preference for tearing down and replacing old homes, with the expectation those too will be replaced in short order.
Why are these old homes replaced so often? I live in a home built in 1930, which certainly does not need to be torn down and replaced, indeed any replacement would likely be of lower quality.

Here's another way of thinking about the productivity question. In both countries, the vast majority of people have homes to live in---somewhere around 99%. So then the question becomes: In which country does it require more labor to provide a flow of housing services, which keeps a roof over the head of 99% of the population? And the answer is clear---it takes much more labor (per capita) to house the Japanese population than the American population. That suggests that Japanese labor productivity is lower.

Of course we also need to consider quality. American housing is provided with far fewer workers, but perhaps Japan's housing is of far higher quality. So maybe the quality adjusted housing productivity is higher in Japan.

One important component of quality is size. Are Japanese houses far bigger than American houses? Maybe, but given that America has a reputation for enormous houses, and given that even Europeans consider Japanese houses to be tiny, I'm not willing to accept that claim without firm data.

Maybe Japanese houses are much smaller than American houses, but they are built with such jewel-like precision that the quality is higher despite being smaller. But in that case why are Japanese houses torn down just a few decades after they are built?

Screen Shot 2017-06-26 at 12.59.30 PM.png

The FT article doesn't even come close to justifying the claim that Japanese construction workers are more productive than American construction workers. There are some problems with American housing quality, particularly for homes built during the 40 years after WWII. But more recently built homes seem to be of pretty good quality, with lots of amenities such as multiple bathrooms and granite countertops. We are a long way from the Levittown ranch houses of the late 1940s. Last year I went to an open house for a 3500 sq. feet Toll Brothers house, which would probably be regarded a mansion in Japan. It counts as "one house" in the US data.

Rather than better construction productivity, the real story may be that Japan has better (more flexible) zoning rules, which makes it easier to build new homes. It's not our construction workers that have low productivity, it's our government.

BTW, A recent WSJ article reminded me of the debate over manufacturing---is the job loss trade or automation? Here's just one more piece of evidence that it's mostly automation:

Later this year, a new Chevron Phillips facility capable of producing 1.5 million metric tons of ethylene a year is coming online in Baytown, Texas. It covers a plot the size of 44 football fields and is made up of 350 miles of pipe, 40,000 tons of steel and 140,000 tons of concrete. It has taken four years to finish.

During the height of its construction, more than 4,500 construction workers and engineers were on site. Once operational, it will only take around 200 employees to run.

American manufacturing has a very bright future--just don't expect any more jobs.

The same article suggests that this plant is not a fluke:

In April, Exxon said it selected a site near Corpus Christi, Texas, for a $9.3 billion petrochemical complex it is building jointly with Saudi Basic Industries Corp. The proposed facility, the largest of its kind in the world, is expected to be done by 2021 and produce 1.8 million metric tons a year of ethylene, the main component of plastic.

"We don't see this as a bet," said Neil Chapman, president of the chemicals unit at Exxon, which is investing a total of $20 billion in such projects along the Gulf of Mexico. "You've got to pinch yourself sometimes and say 'this is the envy of the world.' "

Dow's plant in Freeport, Texas, when fully operational by the end of the year, will produce 1.5 million metric tons of ethylene annually. The company plans to export at least 20% of the plastic it makes in the U.S. and is particularly eyeing Latin America as a ripe market.

So maybe 600 jobs. Which is basically nothing.

COMMENTS (14 to date)
Lorenzo from Oz writes:

A post suggesting that housing restrictions drive much of politics in the UK.

Lorenzo from Oz writes:

And in the US. See also here.

(Neither post is by me.)

Jerry Brown writes:

Well this is very interesting to me because I am in the housing/construction industry myself. And I am fortunate enough to live in a beautiful old home built in 1910 that could not be replicated today at even three to four times its market value. I grant that things like solid oak trim and leaded glass windows come in and out of style, but the real reason I was able to afford to live here has to do with the property values in the area. If the plot of land my house sits on went up in value by many multiples maybe it would make sense to tear this down and build something else. Is that maybe what is happening in Japan? But what does any of this tell you about the productivity of construction workers like me today? I mean the house was built 107 years ago.

I can report to you honestly that the typical house built 30 years ago could be built today with less labor, so the real productivity of that labor involved has increased. But the market value of the finished product also fluctuates considerably and whether it gets torn down and rebuilt has a lot more to do with the local market than the real productivity of the carpenters at the time of construction. And how can anyone make an international productivity analysis unless the same structures of the same quality are built on very similar pieces of land?

Todd Kreider writes:
Why are these old homes replaced so often? I live in a home built in 1930, which certainly does not need to be torn down and replaced, indeed any replacement would likely be of lower quality.

I lived in Japan for 15 years in a very old historical neighborhood and accumulated a little knowledge about housing there.

Anyone living in an American house built in 1930 has a great, sturdy house but was for the upper 5% when built.

Most houses built in Japan until maybe the 80s and especially 90s were crap. Keep in mind, Japanese were poor until the 1970s. A friend took an old house and remodeled it but saw the original building was anything but precise. This hasn't been true for many years, though. (I don't know about productivity measurements, though.)

Thaomas writes:

Unless "productivity" is being measured as # of "houses" built, I do not wee how the difference between the sizes and ages of houses constructed and deconstructed has much bearing on the productivity of construction workers. Surely the meaning ought to be something like value of construction/worker.

The number of workers to maintain a flow of housing services would be quite idiosyncratic, like measuring productivity of auto workers by the number needed to maintain a flow of auto-use services.

AbsoluteZero writes:


This is just to answer your question "Why are these old homes replaced so often?"

First, Todd Kreider is right. Many Japanese houses, particularly those built before the 1980s, were not of high quality. Some were, many were not. As usual it's a mix. The later ones are usually better, mostly because people could afford better material. Indeed, there is a grade of lumber called J-grade, where the J stands for Japan.

The main reason houses are replaced so often has to do with how people view buying a house. While some people do buy ready-built homes (there's even a word for them, 建て売り住宅 (tateurijuutaku), literally build-sale-house/home), most people or families don't buy houses. They buy the land and have a house built to their spec. They do so partly because most of the cost is in acquiring the land. Another reason is, the interior layout is almost never to their liking. But that's in turn because the last family had it custom built to their spec. And the cycle continues.

On a per capita basis, Japan has more architects than pretty much any other country. But, unlike in many countries, the vast majority of these architects do nothing other than small, normal, residential homes. This is why Japanese houses seem so unimpressive on the outside. Externally they all seem very similar, almost identical, but their interiors are often quite different.

Gordon writes:

I rooted around the internet about this topic and came across a Freakonomics podcast about Japan's home depreciation and building craze:

And this was linked to an article from an architect who says that even though home quality has improved dramatically over the last 20 years in Japan, there is still a mindset to tear down and rebuild:

And this article in turn was linked to an article about how the timber for Japanese homes is turned out quickly in a robotic factory and can be assembled on site in a day:

I also recall seeing on tv once that whole rooms are prefabricated and are fit together quickly on site in Japan.

Rob Fox writes:

Construction productivity should be measured strictly in quantitative terms but the article sadly gets entangled in qualitative conjecture. This detracts from the value of any information that may be offered. Instead of measuring percentage of workers vs numbers of homes built and then wondering about cultural and qualitative attributes, it would be more meaningful to measure square footage of homes per FTE-hr., then you'd have something to chew on.

Scott Sumner writes:

Everyone, Thanks, lots of interesting comments.

Thaomas, You said:

"The number of workers to maintain a flow of housing services would be quite idiosyncratic, like measuring productivity of auto workers by the number needed to maintain a flow of auto-use services"

That's exactly how I'd measure the productivity of autoworkers, indeed I can't even imagine any other way that makes sense.

Andrew_FL writes:

I imagine Thaomas thinks we measure the productivity of autoworkers by the manhours necessary to produce a given number* of cars

*effective number, adjusted for different market prices of different makes and models. Of course in this case the prices are implicitly weighting cars by their discounted lifetime service value.

Jerry Brown writes:

Andrew FL, "I imagine Thaomas thinks we measure the productivity of autoworkers by the manhours necessary to produce a given number* of cars".

That is my concept of what the labor 'productivity' of autoworkers is attempting to measure. Is that mistaken?

Andrew_FL writes:

@Jerry-in theory there should be no difference between the concept of the market value per manhour of produced automobiles and the value of the stream of automotive services per manhour used to produce them, which those automobiles deliver over a lifetime. The time discounted value of those services should be implicit in the market price of the automobile.

Jose writes:

On why the Japanese rebuild houses, I remember reading somewhare that because of frequent earthquakes, the Japanese build homes that are easy to rebuild, with lighter materials, therefore their houses are expected to have shorter useful lives when compared to those in other countries.

Alec Fahrin writes:

I am a number of other commenters made our exasperation known in the article's comment section. The biggest flaw was considering American and Japanese homes to be at all similar in size or quality.

I actually believe that is not true. They have built most residential structures since the 1980's with earthquake survivability in mind. Japanese just have a really inane collusion cycle between construction companies, the local governments, and cultural norms.
Japanese in the 1950's didn't rebuild their houses every decade. That's just a product of this collusion post-1970's (when the construction companies gained massive political power).

Comments for this entry have been closed
Return to top