Projects per year
Abstract
BACKGROUND:
Short-term forecasts of international migration are often based on data that are incomplete, biased, and reported with delays. There is also a scarcity of migration forecasts based on combined traditional and new forms of data.
OBJECTIVE:
This research assessed an inclusive approach of supplementing official migration statistics, typically reported with a delay, with the so-called Big Data from Google searches to produce short-term forecasts (``now-casts'') of immigration flows from Romania to the UK.
METHODS:
Google Trends data were used to create composite variables depicting general interest of Romanians in migrating into the UK. These variables were then assessed as predictors and compared with benchmark results by using univariate time series models.
RESULTS:
The proposed Google Trends Indices related to employment and education, which exhaust all possible keywords and eliminate language bias, match trends observed in the migration statistics. They are also capable of moderate reductions in prediction errors.
CONCLUSIONS:
Google Trends data have some potential to indicate up-to-date current trend of interest in mobility, which may serve as useful predictors of sudden changes in migration. However, these data do not always improve the accuracy of forecasts. The usability of Google Trends is also limited to short-term migration forecasting and requires understanding of contexts surrounding origin and destination countries.
CONTRIBUTION:
This work provides an example on combining Google Trends and official migration data to produce short-term forecasts, illustrated with flows from Romania to the UK. It also discusses caveats and suggests future work for using these data in migration forecasting.
Short-term forecasts of international migration are often based on data that are incomplete, biased, and reported with delays. There is also a scarcity of migration forecasts based on combined traditional and new forms of data.
OBJECTIVE:
This research assessed an inclusive approach of supplementing official migration statistics, typically reported with a delay, with the so-called Big Data from Google searches to produce short-term forecasts (``now-casts'') of immigration flows from Romania to the UK.
METHODS:
Google Trends data were used to create composite variables depicting general interest of Romanians in migrating into the UK. These variables were then assessed as predictors and compared with benchmark results by using univariate time series models.
RESULTS:
The proposed Google Trends Indices related to employment and education, which exhaust all possible keywords and eliminate language bias, match trends observed in the migration statistics. They are also capable of moderate reductions in prediction errors.
CONCLUSIONS:
Google Trends data have some potential to indicate up-to-date current trend of interest in mobility, which may serve as useful predictors of sudden changes in migration. However, these data do not always improve the accuracy of forecasts. The usability of Google Trends is also limited to short-term migration forecasting and requires understanding of contexts surrounding origin and destination countries.
CONTRIBUTION:
This work provides an example on combining Google Trends and official migration data to produce short-term forecasts, illustrated with flows from Romania to the UK. It also discusses caveats and suggests future work for using these data in migration forecasting.
| Original language | English |
|---|---|
| Article number | 40 |
| Pages (from-to) | 1219-1254 |
| Number of pages | 36 |
| Journal | Demographic Research |
| Volume | 45 |
| DOIs | |
| Publication status | Published - 2 Dec 2021 |
Access to Document
Fingerprint
Dive into the research topics of 'Now-casting Romanian Migration into the United Kingdom by using Google Search engine data'. Together they form a unique fingerprint.Projects
- 2 Finished
-
Estimating within EU migration flows with new temporal definitions using big data
Wisniowski, A. (PI)
1/12/19 → 30/11/22
Project: Research
-
Statistical Modelling
Onah, C. N. (PI), Shlomo, N. (PI), Chandola, T. (PI), Schoch, D. (PI), Cuitún Coronado, J. (PI), Aparicio-Castro, A. (PI), Thestrup, S. (PI), Olsen, W. K. (PI), Cernat, A. (PI), Shryane, N. (PI), Wisniowski, A. (PI), Smith, D. (PI), Aparicio-Castro, A. (PI), Shafie, T. (PI), Hannemann, T. (PI), Morales-Gómez, A. (PI), Mellon, J. (PI), Troncoso Ruiz, P. (PI), Taub, J. (PI), Murphy, J. (PI), Guest, E. (PI), Kim, J. (PI), Watson, J. (PI), Pashazadeh, F. (PI), Hoór, D. (PI), Jones, P. (PI) & Gosling, Z. (Support team)
1/08/19 → 30/09/22
Project: Research
Datasets
-
Data and code for: Now-casting Romanian migration into the United Kingdom by using Google Search engine data
Wisniowski, A. (Creator) & Avramescu, A. (Contributor), Zenodo, 2 Dec 2021
DOI: 10.5281/zenodo.5508213, https://github.com/a-wis/GT-migr-forecasts
Dataset
Activities
- 4 Invited talk
-
Estimation and forecasting of migration with traditional and novel data sources
Wisniowski, A. (Speaker)
23 Jul 2024Activity: Talk or presentation › Invited talk › Research
-
Measuring and forecasting migration with traditional and novel data
Wisniowski, A. (Speaker)
28 Nov 2022Activity: Talk or presentation › Invited talk › Research
-
Now-casting Romanian Migration into the United Kingdom by Using Google Search Engine Data
Wisniowski, A. (Speaker)
25 Nov 2022Activity: Talk or presentation › Invited talk › Research