CityReads│Cities and Ideas: Bigger Is Better?
Cities and Ideas: Bigger Is Better？
What’s the relationship between agglomeration and innovation? By examining the U.S. patent granted during 1836-2010, we find that a larger city size provided a considerable advantage in inventive activities during most of the 20th century but that in recent decades this advantage has eroded.
Packalen, M., & Bhattacharya, J. (2015). Cities and ideas. NBER Working Papers No. 20921.
Source：http://econpapers.repec.org/paper/nbrnberwo/20921.htm
Picture source: http://www.citylab.com/work/2013/06/cities-are-innovative-because-they-have-more-ideas-steal/5883/
Faster technological progress has long been considered a key potential benefit of agglomeration. Physical proximity to other creative people is often a key factor in invention. It enables inventors to adopt the best new ideas faster in their own work. In larger cities, the costs of encountering new ideas are lower and inventions ‘have their merits promptly discussed’. The debate on new ideas is important because ideas are initially raw and poorly understood. Thus, every idea, if it is to develop from a germ into a substantial innovation, needs to be tried out and discussed by many people including those not responsible for originating the idea.
Modern communication technologies may have undermined or even eliminated any advantage that physical proximity to others may once have provided in terms of adopting new ideas, though there are advantages to face-to-face communication even in an internet-connected world, especially in the sharing of tacit knowledge.
On the other hand, It is also possible that agglomeration actually hinders the adoption of new ideas. For instance, suppose that the presence of large incumbent firms leads to the development of large cities. In that case, inventors in big cities will have strength in incumbent technologies and may be tilted against adopting new ideas that might lead to radical inventions and devalue existing competencies.
From a theoretical perspective, agglomeration may thus have a positive, a negative, or no impact at all on the adoption of new ideas as inputs to the inventive process. This paper used empirical evidence to measure how the tendency to adopt new ideas as inputs in invention varies by the size of the city of invention.
Approach and Data
The aim of the study is to uncover the empirical relationship between agglomeration and use of new ideas in invention.
City size is measured by population density. Density should capture the scope for frequent interactions with others better than other city size measures, though we also perform a robustness check in which city size is measured by total population.
The new idea input in inventions is measured by multiple steps. First, we index all words and 2- and 3-word sequences in each patent. We refer to these words and word sequences as concepts. For each concept we then determine the year it first appeared in the patent database. We refer to that year as the cohort year of a concept. The age of concept is defined as the number of years elapsed since the cohort year of the concept.
Then, for each patent, we calculate a summary measure Age of Idea Inputs, defined as the age of the newest idea input mentioned in it. A lower value for the measure indicates that the invention built on newer ideas. And we employ as the dependent variable an indicator measure that captures whether idea inputs for a patent are among the top 5% newest based on the Age of Idea Inputs measure.
This paper analyzes US patents granted during 1836–2010. For 1976–2010, patent documents are available as ASCII data, whereas for 1836–1975 patents are available as scanned images.
The time series on the number of patenting shows the familiar steady increase in patenting over the past century.
Figure 1 Number of US patents granted, the vertical axis depicts the logarithm of the number of patents.
During the early years of the sample the measured idea input ages start low and increase quickly. This is unsurprising given how we determine whether an idea is new – we measure how recently the concept first appeared in patents – and given the small number of patents granted during the early years of the sample. We address this issue by ignoring mentions of concepts that belong to pre–1870 cohorts and by limiting the analysis of the agglomeration–use of new ideas link to cover the period from 1880 to 2005. We exclude years 2006–2010 from the regression analyses because for the most recent idea cohorts it is not yet known which ideas will stand the test of time and represent the cohorts’ “best ideas”.
Agglomeration and Innovation
The results in Figure 1 indicate that agglomeration did indeed have a considerable positive impact on the adoption of new ideas as inputs to the inventive process throughout most of the 20th century. It also indicates that the advantage of larger cities may now be declining.
To gauge the importance of agglomeration in the use of new ideas relative to other factors, we conduct two additional sets of analyses. In one set of analyses we examine the effect of collaboration on new idea adoption: we regress the outcome variable Top 5% by Age of Newest Idea Input on a dummy variable measuring whether a patent lists multiple inventors or a lone inventor. In the other set of analyses we examine the effect of being located in the US on new idea adoption: we regress the preferred outcome variable on a dummy variable measuring whether the first inventor was located in the US or in a foreign country.
In Figure 2, a positive estimate implies that patents by a team of inventors employ newer ideas than patents by a lone inventor. We find that teams of inventors are much more likely to apply fresh knowledge than lone inventors. This may be the result that each inventor in a team brings in their own knowledge of existing ideas to the team and to which extent the result arises because inventors working in teams can solve the mysteries of new ideas faster than lone inventors through a vigorous debate on the new ideas’ merits.
In Figure 3, a positive estimate implies that patents which first inventor lives in the U.S. employ newer ideas than patents which first inventor lives elsewhere. The results show that patents developed by inventors living in the U.S. are more likely to use newer ideas than patents developed by inventors living elsewhere. This result provides direct evidence on the extent and evolution of US leadership in technological invention.
Taken together our results suggest that in the late 20th century agglomeration has become less important to invention in both absolute terms and relative to other factors – like collaboration– that predict the use of newer ideas.
Conclusion
The empirical findings indicate that during the 20th century inventions in large US cities built on recent advances much more often than comparable inventions in smaller US cities. The findings also indicate that during the most recent decades this advantage of large cities has waned. The advantage of locating R&D resources in large cities over locating the same resources in smaller cities thus seems to be much smaller now than it has been in the past.
It is interesting to speculate why the convergence occurred. One possibility is that the spread of the internet and other communication technologies has made new ideas available more cheaply to all, regardless of where they were developed. Another possibility is that these new technologies have expanded the sizes of the communities that examine and debate the merits of each a new idea from a local to a wider scale. At the same time, our results on US leadership relative to other countries suggest that the communities have not yet become truly global.
Related CityReads:
6.Life in the City Is Essentially One Giant Math Problem？
11.Why So Many Emerging Megacities Remain So Poor？
12.How Economists Study Cities？
（Enter Official Account and Reply number“6"“11”or "12”）