Stata format | Comma-separated flat file
This data was originally published in the Unesco Statistical Yearbooks (various years). The variable "students" in the files corresponds to Unesco's "Inbound Internationally Mobile Students by Country of Origin". The variable "istudents" is the linear interpolation of "students". This students variable includes both short-term and long-term international students, that is, both those going abroad for their entire course of study and others who may be on exchanges or from immigrant families. The data make no distinction between the two types of students.
I digitized this data in 2005 by scanning pages of the yearbooks and using optical character recognition (OCR) software. This software frequently made errors, which I have for the most part corrected, but some errors in the data may remain. Another source of error is in the reporting of the data itself. The data come from country reports to the UN. Countries may vary in how accurately they counted their international student populations.
I initially used this data for a minor instrumental variable in my 2012 paper Skilled Immigration and Innovation: Evidence from Enrolment Fluctuations in U.S. Doctoral Programmes. More recently I used it as the dependent variable in my 2019 paper The Effects of Social Networks on the Flow of International Students.
The complete panel of data for that 2019 paper is here:
Note: In the variables' labels, country 1 refers to host country, country 2 refers to origin country.
Information about the data is given in the paper and appendix.
The panel above omits the counts of PhD students obtained from the National Science Foundation's Survey of Earned Doctorates. The NSF permits access to the microdata from the survey under strict confidentiality restrictions. So, when I was done with the project I relinquished my access to the PhD student data in the panel. It could not be included in the panel even aggregated to the field/university level. If you have access to the SED data and want to reconstruct my panel from it, the following Stata "do" file will help. Field number/name correspondence is listed in the appendix to our paper.