Urban productivity in the developing world

That is a new NBER working paper from Ed Glaeser and Wentao Xiong.  Here are a few things I learned from it:

1. “As agglomeration size doubles, wages rise by approximately five percent in the U.S. and Brazil, but the link is much larger in India and China.”

2. “Soichiro Honda began his remarkable career as a car mechanic.”

3. Per capita gdp is three times higher in Shenzhen than in the rest of China.  Bangalore per capita gdp is 2.5 times higher than the rest of India.

4. In the United States, urbanites earn 30% more, and this gap does not disappear with controls for human capital attributes.

5. The urban to rural earnings gap is 45% in China, 122% in India, and 176% (!) in Brazil.

6. In the U.S….”as area size or density doubles, wages increase by…about five percent.”  But agglomeration economies are much stronger for India or China than for the U.S. or Brazil.  Brazil is a city vs. countryside effect, not so much about size of the city per se, Sao Paulo aside.

7. “In 1961, Benjamin Chinitz argued that New York City was more resilient than Pittsburgh during the 1950s, because New York City had a culture of entrepreneurship that meant that its business leaders were good at adapting to industrial decline.  In modern language, we might describe New York as having a healthy endowment of entrepreneurial capital because its dominant industry, garment production, had limited-scale economies and few barriers to entry.  In contrast, Pittsburgh had U.S. Steel, and the steel industry had large-scale economies, which meant that Pittsburgh trained company men instead of entrepreneurs.”

Overall I found this a very good paper for stimulating thought.  There is also a new paper by Joan Hamory Hicks, Marieke Kleemans, Nicholas Y. Li, and Edward Miguel on agricultural productivity gaps, it is receiving high praise on Twitter.  I have not yet had a chance to look at it.


NYC's garment industry was dominated by Jews, who tend to be entrepreneurial, while the labor in Pittsburgh's steel industry was comprised largely of Catholic ethnics, who tended to be less so.

One way that people "become entrepreneurial" when they are barred from industry. The classic example is a junk man, the job open to anyone because no will stop you picking up trash.

NYC has benefited from waves of despised immigrants who became entrepreneurs of necessity.

A related story:


Living in KL, my hunch is that growth is infrastructure-based and human capital grows as a side effect (Learn how to build trains by building trains)

Building infrastructure for the country-side (water, roads, power, telecommunications) takes a much longer time and a good part of the gdp difference between rural and urban areas could be explained by the differences in physical capital per person.

"Pittsburgh trained company men" - so do Japan and South Korea, yet they seem to be doing just fine.

Japan and South Korea are not dominated by one industry, that sucks all talent into it.

"“Pittsburgh trained company men” – so do Japan and South Korea, yet they seem to be doing just fine."

Pittsburgh has roughly the same household income than Japan and higher than South Korea's.

Much of that probably has to do with , paraphase Garret Jones' Hive Mind, the productivity of others - not Pittsburgh's

Well why don't you just open up a steel factory then?

That logic applies to barbers and other services which are fairly similar in labour demand at $5 per capita per day income or $100. Not to heavy industry (much, likely, basically not at all).

> In the United States, urbanites earn 30% more, and this gap does not disappear with controls for human capital attributes

....but what about when you control for cost of living? You effectively become wealthier by working $8 an hour on a Texas ranch than by working $15 an hour at a San Francisco bakery.

A ranch hand is an skilled and disciplined worker, horses are thirsty and need a walk even on Sundays. In round numbers, 50% higher hourly wage when compared to fast food workers. https://www.bls.gov/oes/current/oes452093.htm#st

CCs point is that PPP matters: https://www.advisorperspectives.com/dshort/updates/2016/10/17/median-household-purchasing-power-for-the-50-states-and-dc

California is bottom 25% and NY is dead last.

This is the same guy that calculates Delaware is much happier than Hawaii


The US Census bureau also ranks California as having the highest Poverty rate in the nation when looking at PPP. The PPP Poverty rate in CA is over 20% vs 17% for Mississippi.


California has become a high inequality state, with a glittering cloud of billionaires flying over the massive amount of poor inhabiting LA County and the states hinterlands.

That's a weird way to look at "the state's hinterlands," which are really farmlands, feeding the country. But some kid working at Apple is responsible, not you?

"But some kid working at Apple is responsible, not you?"

I'm certainly not responsible for the high poverty rate of California. Are you in denial that California has the highest PPP poverty rate of any state? Or are you trying to blame someone else besides Californian's themselves?

How can you not be? You depend on those salads.

Although lettuce is produced in many states, California and Arizona dominate U.S. production. California accounted for 71 percent of U.S. head lettuce production in 2013, followed by Arizona producing nearly 29 percent. These states also produce over 98 percent of the leaf lettuce in the U.S.

Agriculture is less than 2% of California's economy. Of the state's 19 million workers, only 400K are farmers.

The stereotypical poor Californian isn't a farm hand, she's a retail clerk making minimum wage in San Bernardino or Los Angeles.

"How can you not be? You depend on those salads."

So, I'm responsible for poverty in CA because I might occasionally buy CA lettuce? That's just some desperate grasping of straws.

"Or are you trying to blame someone else besides Californian’s themselves?"

Obviously, California agriculture needs to take a whole bunch of pages from Apple and double the market price of its crops, eg, head lettuce should be $2 in the supermarket, not $1, etc. If Apple can sell things for twice what the general market does, why not California farmers?

And the entire food chain pricing should double for the most labor intensive.

Trump has impacted the supply of farm workers in California, but but even $15 an hour is not getting European origin US born out into the fields. Doubling the price level of salad crops would allow wages of $25 an hour.

And the share of US household spending on food is lowest in the world.

mulp, I think you just fail to appreciate the wonders of the billion dollar curve.

And that's only based on what Samsung had to pay for infringement of the billion dollar curve.

Who knows, maybe it's a ten billion dollar curve?

I really don't think anyone looking at a county level map of the US would pick California as the most pressing problem region:


1. They are using the wrong definition of poverty in that map. You need to adjust for purchasing power parity.

2. Counties aren't the best dividing line. Many of those counties in Mississippi have maybe 10K residents. Kern County, California with a 21.9% poverty rate according to your map has over 800,000 residents.

I'm all for finer granularity, but note that when people use "purchasing power parity" for "California" they are putting Malibu and the Imperial Valley in one bucket.

"they are putting Malibu and the Imperial Valley in one bucket."

Umm yeah, that was kind of the whole point. As I said above: "California has become a high inequality state, with a glittering cloud of billionaires flying over the massive amount of poor inhabiting LA County and the states hinterlands."

The Red Queen strikes again? In that one must run faster and faster just to stay in place, as the price of (locally produced, at least) goods and services rises to accommodate the rising wages of those who produce these goods and services?

Isn't inflation sort of a Red Queen game anyway, as one runs faster and faster to spend one's cash before its value evaporates?

Presumably they benefit mainly from increased minimum socially acceptable standards, for example as seen through a higher minimum wage.

In the meantime, they've gotta be good enough to justify the price in a competitive market. When a haircut can be had for a dollar, it's hard to justify years of specialized training or much specialized equipment ...

"You effectively become wealthier by working $8 an hour on a Texas ranch than by working $15 an hour at a San Francisco bakery."

Only if the ranch provides room and board.

Actually no, just straight up cost of living comparison.

San Francisco has a Cost of Living Composite Index of 164.0

Dallas is a 91.9. So $8 per hour in Dallas is the equivalent of $14.28 per hour in San Francisco.


interesting to read alongside Gollin et al Urbanization with and without industrialization

"In countries that are heavily dependent on resource exports, urbanization appears to be concentrated in “consumption cities” where the economies consist primarily of non-tradable services."


It is (obviously) not unexpected that localised GDP (GVA) and (contingent) pay rates are higher in urban areas than in rural areas.

However it is also obvious that living costs (particularly housing) are generally higher in urban areas.

From the figures given, it is not clear whether differing living costs are taken into account, for example by using GDP(PPP) adjusted for localised costs. I expect not, but am not certain - the original paper is behind a pay-wall.

It would be good to know, and also the extent of the actual difference if one takes into account PPP - especially for housing costs.

Best regards

Tyler summarized "Per capita gdp is three times higher in Shenzhen than in the rest of China. Bangalore per capita gdp is 2.5 times higher than the rest of India."

I looked up that Shenzen's economy is $400 billion dollars PPP s with their population of 11 million that comes to a GDP per capita of about $36,000. China's GDP per capita is $14,000 so Shenzen's GDP per capita is 2.6 times higher than the rest of China.

That is only 20% higher than Beijing and Shanghai. At any rate, it looks like the authors used PPP for China so probably did for the U.S. as well.

The driver of productivity improvements is specialisation and associated economies of scale. Take an example of someone building a car by making all the parts themselves vs assembling a car from multiple suppliers of parts who are making them for all the other car manufacturers. The widescreen wiper manufacturer can optimise his processes and his R&D around making the best and cheapest windscreen wiper manufacturers, plus spread his costs across many car manufacturers. The buyer of the windscreen manufacturer also benefits, now there is competition driving innovation in windscreen wiper manufacture.

Cities are factories for services in the same way. They allow increasing economies of scale and specialisation in services, driving productivity growth. As the city grows in size, the amount of specialisation can increase. So we should see an increasing trend from country to city, and from smaller cities to bigger cities, which is exactly what we see.

I was once in a small startup that made external disk drives for Macintoshes. In those years it was a simple business. You bought an SCSI drive, some cables, and an enclosure. Our sheet metal guy was in Santa Ana, California. One day he said "you shouldn't build your enclosures like that, you should build them like this." Why? "That's the way another customer makes his, it's much quicker."

That's the kind of benefit Shenzhen gets now, x100.

In the computer industry, engineers get together all the time to develop standards for increasingly complex components. Thus you can source components from multiple manufacturers.

That mindset is contrary to almost all engineering because a high priority given the engineers is capturing the customer: make sure the customer can not switch suppliers.

"Take an example of someone building a car by making all the parts themselves vs assembling a car from multiple suppliers of parts who are making them for all the other car manufacturers"

If reality were only like you describe...

What is maddening is how a factory producing parts for multiple manufacturers ships out several times the unique parts as car makers supplied for the same functional part. Even when GM tried to cut costs by creating common platforms for multiple models, each model still put model unique parts on the common platform instead of pulling parts for an established list of standard parts.

The lower cost of outsourcing did not come from economies of scale, but by paying workers less.

Elon Musk has found the process of sourcing parts he hasn't built the production line for to be frustrating. If one supplier has problems delivering a common functional part, you can't source the part from a different supplier, so you can't ship any cars.

7. "Culture of entrepreneurship" may be misleading. The NYC economy is mostly a service economy, whereas Pittsburgh's economy was heavily industrial. A service economy tends to have many small businesses, each complementing and servicing the others, whereas an industrial economy tends to have one or a few large businesses that employ many. Of course, Pittsburgh has transformed itself into a service economy, with health care and education the main industries today. I suspect that "entrepreneurship" has greatly increased as a result. Does that mean a service-based economy is better than an industrial-based economy, as this study seems to imply? I've resided most of my adult life in a large urban area that is mostly service-based, with few large employers and many "entrepreneurs". The jobs are mostly low skill, low pay, where "middle class" is identified by numbers rather than income. It is a rapidly growing area. It is the future of America for most.

NYC circa 1960 was still heavily manufacturing employing millions of workers, more than steel in Pittsburg.

These numbers could mostly underscore just how undeveloped the rural part of the developing world is. There is a larger gap in infrastructure between the U.S. Corn Belt and the Amazon than there is between Chicago and Rio.

#2. Honda started by manufacturing small engines and motorcycles. When they announced plans to enter the auto market the Ministry for Trade and Industry tried to discourage them saying Japan already had enough auto companies (big two were Toyota and Datsun (now Nissan) and there were a couple of other smaller ones). Sometimes government industrial policy gets it wrong.

Most developing countries have rules that explicitly limit movement to cities or else limit the ability of leading sectors to grow and to invest in non-established areas. China of course, has their famous hukou system that penalizes those who move to the big city without formal registration. That in turn was modeled on the Soviet propiska which was only removed in Russia very recently. Other countries have policies that subsidize low wage agriculture, impose strict minimum wages on large commercial firms in cities but not farms, limit the ability of small farmers to buy or consolidate land, and tax or otherwise regulate investment that would allow companies to invest in poorer areas and develop competing regions. This is on top of lack of proper infrastructure plus centralized rent-seeking that make it suicide for growing companies not to have presence and political contact in the dominant areas. And of course, foreign firms that could do the most to jump start growth in those areas are either not permitted or heavily constrained. As Rodrik has noted, export oriented sectors tend to be more similar in productivity worldwide. It is the non-export oriented, heavily agricultural rural areas that underperform in the Third World and most institutions are structured so as to slow or impede wage/investment equalization across regions.

I mentioning the taxes paid on investments in poorer areas, it is not mentioned that these taxes would generally be paid in either case. It could almost sound like poor areas are targeted for special taxation, which I do not believe is often the case.

"4. In the United States, urbanites earn 30% more, and this gap does not disappear with controls for human capital attributes."

Does it disappear with controls to actual capital and family holdings? How much is this a case of built up capital assets that contribute to income? Client accounts, taxi medallions, networking, family owned businesses and real estate holdings, etc.

Strike networking from the list. Cities generate dense networks. The question is how much of the value comes from the networks. Is it all of the 30% gap due to networking? Or is a significant chunk of the gap due to long standing capital assets that contribute to the wealth of the area?

There are some obvious examples. But I don't know if they are significant to the overall numbers. A taxi medallion is an obvious case. If a taxi driver holds a medallion worth $500K in NYC and earns $20K more per year in income than a taxi driver operating in a small city, their effective incomes are the same if you assume that $500K in assets will reliably generate you at least $20K in income anywhere.

It's not obvious whether you should expect capital income to be a higher share of income in urban areas compared to rural areas.

Are you imputing rental values on a $5 million property? Because something 90% cheaper in a rural area may actually be better housing.

A number of issues arise. But income generating activities tend to be fairly capital intensive in rural America. For example farming.

There is a US map of successful Kickstarter projects in the bottom half of this page:


The challenge, for urban skeptics, is why a "locationless" tool like Kickstarter would favor New York and Los Angeles across so many categories. As the page notes, location isn't "necessary," but NY and LA are there, in every map.

That link doesn't seem to support the idea of NY and LA being powerhouses. Except in theater and film.
Whether Kickstarter is a good proxy for inventiveness is...questionable...

5 seems unlikely without some statistical magic. Are they comparing urban farmers and rural farmers? And urban construction labour to rural construction labour?

The percentage of some occupations in the employment market is very high in rural areas compared to urban areas. So comparing apples and apples is a very misleading way to discuss urban/rural disparities (except in some fairly specific applications). The farmer will not likely get a job as a computer salesman in a rural area, for example.

Or, there could be aggregation issues at the sub-municipal level that could contribute to that.

Also, regional issues may dominate rural/urban issues. A larger number of relatively lower income cities in the west of the country could drag down urban earnings a lot, whereas many rural people in the east may have higher income. There should be a province or at least 5 or 6 different regional controls or this will be very misleading because mainly other things are at play.

The urban/rural gap in Guangdong, for example, alongside the urban/rural gap in some western provinces, would be incredibly more informative than to have an aspect to the comparison where the income of Guangdong farmers is being compared to the urban income in some unimportant cities much further from the sea and in the mountains.

Paper is gated, but I'm guessing basically about that.

Comments for this post are closed