The Benefits and Costs of Big Cities

Date March 1, 2023
Speaker Diego PUGA (Professor of Economics, Center for Monetary and Financial Studies (CEMFI))
Commentator & Moderator SAITO Yukiko (Senior Fellow (Specially Appointed), RIETI / Associate Professor, Faculty of Political Science and Economics, Waseda University)

Analyzing the benefits and costs of big cities by comparing large cities and small cities at firm level and at worker level, Puga posits that an agglomeration effect that benefits all productive firms and workers leads to high firm-level productivity in large, dense cities. For workers, higher earning premiums are available in a big city, which become even higher for skilled work. These are combined with an increased learning effect from remaining in such a city, which adds to the worker’s human capital. Puga also discussed the imperfect assessment of ability and the friction of migration, the role of personal networks, and the trade-off of costs and benefits within a city and between large cities and small cities in terms of accessibility, housing costs, productivity, earnings, and commuting costs. Puga's model of this trade-off examines location heterogeneity, planning regulation, housing prices, and the interconnections between cities.


Agglomeration economies or selection?

Early in human history, land fertility determined the location of cities. With the rise of manufacturing, access to routes of transportation, such as rivers, seas, and roads, became vital. The industrial revolution then scaled manufacturing and placed importance on access to energy sources and primary materials. However, while these aspects are still somewhat relevant, now, firms and people want to be located near other firms and people. This location of firms and workers is self-reinforcing, increasing the scale, density, benefits, and diversity of big cities, leading to higher productivity and boosting innovation. For workers, it allows for more opportunities to gain experience and access to goods, services, and amenities. The density of cities also encourages efficient modes of transportation. However, urban density comes with costs, such as increased costs of living, producing and moving in such cities.

As can be seen in data of French labor markets and firm productivity, there is clearly a positive relationship between urban density and firm productivity. There could be many reasons for this relationship. It could be that density itself makes firms more productive, as being located near other firms might allow for the sharing of suppliers, or that dense locations could allow for the exchange of ideas and flows information across different firms – i.e., agglomeration. Another theory is that bigger cities are simply tougher markets, and it is more difficult for less-productive firms to survive, leaving only more productive firms – i.e., selection.

By applying our method, published in Econometrica with Combes, Duranton, Gobillon and Roux, to this French data, what we see is that the productivity distribution in big cities is the result of the small city productivity distribution being right-shifted and expanded or dilated, showing that big cities create agglomeration economies. It is not case of selection pressure. In these agglomeration economies, the most productive firms benefit most, by being almost 15% more productive. Manufacturing firms and tech firms benefit from being densely located by sharing suppliers and labor force, allowing each respective firm to compete better on a national or continental scale, rather than competing on a city scale with each other.

Average earnings are higher in bigger cities

Agglomeration economies also benefit workers, as seen in Spanish data on wages. Wages in a big city like Madrid and Barcelona are on average about twice as high as in smaller Spanish cities, even after controlling for level of education and industry sector. However, despite earning more, they face higher cost of living and higher housing costs. That being said, living in a big city has the added benefit for workers of providing more learning opportunities and higher-level skills. These advantages for the workers build over time, as seen in our paper published in the Review of Economic Studies with De la Roca that compares the wage data of two graduates with the same education who go to work in a big city and a medium-sized city, respectively. The worker in the larger city gains initial increased wages and a compounding effect for staying in the larger city. Interestingly, even if the worker from the large city later relocates to a smaller city, they bring with them the valuable experience from the big city, transformed into more valuable human capital, allowing them to maintain that advantage.

The role of networks and confidence in residential mobility

Nevertheless, the distribution of productive workers in both big and small cities is similar, and fewer highly capable workers move to cities to gain such advantages and vice versa than we might expect. There are two key reasons for this. There is limited mobility, as although gross flows of migration are large, net migration flows between cities are small in comparison, do not vary much, and react very slowly to shocks. Also, migration decisions are very idiosyncratic, as people tend to base their migration decisions not only on the size of earnings, but also on housing costs, amenities, and the location of family and friends. This can be seen in our paper with Büchel, Ehrlich, and Viladecans-Marsal where we looked at anonymized cell phone data of people in Switzerland and how it changes over time. It shows that people tend to move near family and friends when deciding to migrate and that they are more likely to move out of a location if family and friends are not close by. Proximity to such contacts is itself valuable and complements other attractive location characteristics, such as local restaurants and schools, and even reduces moving costs as family and friends can help find homes.

Urban sorting also occurs due to the fact that more able people benefit more from moving to bigger cities. However, when people typically make such a decision, typically when they are young, they would not know if they were a more able person. Therefore, we looked at data from the United States on location throughout the subject's life and tests of both intelligence and self-confidence for the subjects when they were teenagers, and it demonstrates that it is the most self-confident that move to big cities, and not necessarily the most capable.

The urban trade-offs and periphery house prices

Living in a big city also means dealing with the trade-off of living centrally, with better access to jobs and places, or living further away, with less expensive housing but a longer commute. Our model states that incumbents living in the city are able to vote and set regulation determining construction of new housing; the further one is from the city center, the more one needs to travel; also, wealthier people place a higher value on their travel time; and in bigger cities traffic is more congested, so the model incorporates a congestion variable. The model shows that living further away should be cheaper, to some degree, but that incumbents, who control regulation, create, in their own self-interest, artificial barriers to further expansion of the city that increase house prices in the periphery, alongside natural barriers, such as the sea or mountains, which further increase prices.

Increasingly interdependent cities

It is also important to note that big cities are not simply scaled-up versions of small cities. They are systematically different in that bigger cities increasingly form a culture for innovation, and smaller cities specialize in production. Eventually, innovative activities in big cities migrate to smaller cities, opening up a gap for new innovation in big cities. It is important for both that this cycle continues.

Moreover, certain cities are also beginning to specialize in terms of function, with headquarters increasingly located in big cities and production facilities located in smaller cities. There are also patterns emerging where both top management and middle management are increasingly located in big cities, although this pattern differs by sector. Thus, there are important roles for big, medium-sized, and small cities, and the interactions between them keep the urban system alive.

Comment and Q&A

SAITO Yukiko:
What is the implication from COVID as commuting costs are reduced due to working from home combined with the prevalence of ICT?

Diego PUGA:
There are advantages of big cities in terms of production, but there's also advantages of big cities in terms of consumption as well. Even though people are working from home increasingly, it's mostly only a few days a week. Therefore, the effects from COVID are more in terms of where you live within the city, and not whether to move to that city or not. Increasingly, people are working a lot of time from home, but now they go to the city to connect with their network, go to bars and restaurants, and interact socially. So now, proximity to amenities may actually matter more than proximity to the job. So, one may be willing to put up with a longer commute in order to be able to live in an area of the city that is nice, meaning there may be less demand for commercial office space relative to the residential space.

SAITO Yukiko:
In Japan, the number of firms is decreasing, and firms' CEOs are aging; and these two factors are correlated. What would be the implications of this for Japan's aging society?

Diego PUGA:
Japan is aging, but Spain is not far behind. Its fertility rate is now lower than Japan's. This generates different dynamics compared to a country that has a rapidly increasing population, where relative shocks are in terms of growing a lot versus growing very little. In a country with a stable population or a shrinking population, negative shocks are more severe, because during a positive shock, a city needs to build more homes or offices, causing gradual growth, with large population changes but little change in prices initially. However, during negative shocks, cities do not destroy buildings, so there are drastic changes in prices. This means we should expect much more volatility in land and building prices in shrinking populations.

SAITO Yukiko:
What are the policy implications about urban planning and higher housing price at the edge from your new model?

Diego PUGA:
In terms of consequences for planning, one of the things the model says is that planning in very small jurisdictions tends to benefit incumbents. Therefore, there is some advantage for society from having planning regulation with a broad view, to take into account the trade-off between the benefits gained and the costs of building more.

The benefits in these agglomeration economies, tend to be especially at a larger scale. The costs, such as noise from construction or more congestion, tend to be much more local. So, local planning regulation-based choices emphasize cost over benefits. So, more aggregate planning is suitable and actually something that's being implemented by some countries.

SAITO Yukiko:
I want to move to discuss the selection and agglomeration effect. At firm level, it is very interesting that you identify the effect by comparing the shape of the productivity distribution between a large city and a small city. But I wonder whether it is consistent with the firm dynamics. It would be interesting to see each contribution from firm dynamics sorted by the ratio of entrant firms and the productivity distribution of entrant firms.

Likewise for selection and accommodation, it would be interesting to see the ratio of exiting firms and the productivity distribution of existing firms, and the evolution of productivity for incumbent firms, respectively. And at worker level, you analyzed the evolution by year worked, but we want to see how it evolved by firm age. In Japan the exit ratio is relatively small, but it is higher in urban areas. So, it would be interesting to see whether it is the agglomeration base that is dominant or not in firm dynamics.

Diego PUGA:
Looking at the dynamics is certainly very interesting. I should say that this method does not really look for truncation. It really looks at what the probability is that the firm that is there with some productivity level in a small city is not there in a big city, and whether it is more likely if the firm has lower levels of productivity than higher levels of productivity. But certainly, this idea of looking at firm dynamics and exit and entry is, I think, a very interesting complementary way to look at them.

SAITO Yukiko:
Regarding firms and the establishment locations of firms, I'm curious about the firm borders and how do you interpret firm-level productivity and establishment-level productivity? At the end of your presentation, you were talking about the independence of cities, and maybe it might be related. But at firm level, we found that headquarters are more localized than establishments in Japan. And also, we found that for establishment export status, the location of headquarters is more important than the location of the establishment. So, the headquarters location might be important, but we also know that the location of the local labor market for establishments also matters. So, I want to know what this firm level and establishing level means and how it relates to how firms decide to locate their establishments.

And finally regarding the mobility of workers between and within cities and firms, it might affect the interpretation of the earnings premium. I want to know what the ratio of workers who changed jobs within the same city is. Is the time working in the same city similar to the time working in the same firm, or not? So, it might be that the slope might be the system wage by firm. And also, what is the ratio of workers who work for the same firm but change their residence as their working locations change to different establishments within the same firm?

Diego PUGA:
In terms of headquarters and establishments, the data that we use is at the establishment level. But, of course, we have the workforce at establishment level and the composition in terms of skills of the workforce. But the capital, we have at the firm level. So, we need to make some assumptions about how we allocate that capital to the different establishments. But we are estimating the productivity at the establishment level and that distinction is important. And yet, some of the things that you suggest are things that we are now starting to look at, in terms of how firms organize across space, across different establishments, what sort of activities they put in different places, what levels of management, and how they manage. And that's very interesting and that's something we're starting now with Portuguese data.

The Japanese government interprets the low total fertility rate in Japan as a cost of urban agglomeration, as fertility rates are lower in bigger cities. Because of this, the Japanese government wants to reverse the urban agglomeration into Tokyo. Is there a comparable situation in Spain?

Diego PUGA:
Certainly, it is also a looming problem in Spain, coupled with the pension crisis. The benefits of big cities are from learning, which is especially accrued while the inhabitants are young, but at the same time, raising children in cities is also much more difficult and expensive. This leads to a postponement of childbearing and an aging of the population, so it is also relevant in Spain.

Regarding the mechanism that underpins the tendency to agglomerate or not within a large city, what factors can change this tendency?

Diego PUGA:
The tendency to agglomerate in cities has always existed. However, research from Kyoto University has shown that there is concentration within cities, but a flattening of activity spread across a city. This is a pattern seen in a number of places, as inhabitants balance different trade-offs of housing prices and lower transportation costs.

SAITO Yukiko:
We often think about model-centric ideas, but it is interesting to take into account these real-world aspects.

*This summary was compiled by RIETI Editorial staff.