Measuring social connectedness using air travel data

ABSTRACT

Previous research has proposed a novel way of using Facebook friendship data to measure on-line social connectedness between county pairs and country pairs. Meanwhile, in-person social connectedness is equally, if not more important. In this paper, we provide a seemly promising way to measure in-person social connectedness using air travel data. We find that counties tend to be more socially connected if they are closer to each other or have similar income per capita, share of white people, or share of Obama voters. We also find that counties with stronger social connectedness tend to have higher bilateral trade and immigration but do not tend to have more cross-county patent citations.

KEYWORDS:

• Social connectedness
• air travel

JEL CLASSIFICATION:

• O41
• O43

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The Pew Research Centre, Public Perceptions of Privacy and Security in the Post-Snowden Era (2014), http://www.pewinternet.org/2014/11/12/public-privacy-perceptionshttp://www.pewinternet.org/2014/11/12/public-privacy-perceptions.

² Contact of the consulting firm is available upon request.

³ We thank these authors for kindly sharing their data with their publication.

⁴ We use the patent data compiled in berkes. We are grateful to Enrico Berkes for kindly allowing us to use this dataset.

⁵ We have also plotted heat maps using the ISCI constructed using our second method, and they have similar patterns.

⁶ More specifically, we use $log(ISCI+1)$ as some county-pairs have zero air travellers and therefore their ISCI equal zero.

⁷ The results using weighted average do not notably change. Full results are available upon request.

⁸ Using travel in economy class as an instrument, we find that the elasticity for ISCI (simple average) with full set of control variables rises from 0.058 to 0.062.

⁹ We obtain the patent data compiled by Enrico Berkes (mailto:[email protected]@osu.edu), and construct the citation index using the Stata program provided by Bailey et al. (2018a). Please refer to Bailey et al. (2018a) and its accompanying online appendix for the construction details.

¹⁰ We try to use booking data excluding economy class to partly address the reverse-causality problem, and the elasticity of ISCI (simple average) with full set of control variables rises from 0.156 to 0.165.

Additional information

Funding

This research is funded by National Natural Science Foundation of China (No. 71403168) and National Social Science Foundation of China (No. 19CLJ015).