Finding the drivers of broadband in Turkey

A new research paper shows how the well-established correlation between broadband take-up and GDP per head for whole countries can also apply at the provincial level.  A joint production by Point Topic and Data Learning, the paper identifies the main drivers for both mobile and fixed broadband take-up across the 81 provinces of Turkey.

The authors believe the results will have useful applications for broadband policy and investment even where detailed statistics are not available.  For example, the same linear-modelling approach can be used to predict the likely take-up of new broadband networks, and to assess the potential role of mobile broadband.

Turkey is one of the world’s most important developing nations. With GDP per head ranging by a factor of three between Istanbul in the northwest and the poor provinces in the south-east, it spans many of the issues of development in a single country. At the same time it produces detailed official statistics for both fixed and mobile broadband take-up, and household numbers, which are often not available from much richer countries.

As a result, the authors were able to show that, across Turkey’s 81 provinces, fixed broadband take-up per head of population correlates with GDP per head with an R2 of 76%. Mobile broadband take-up has different drivers. After testing multiple variables, the strongest correlation proved to be with a linear model combining household size and the percentage of the population with tertiary education.  This had an R2 of 71%.

One striking result was that the proportion of internet users opting for mobile broadband actually increases with falling GDP. This is because mobile plays a dominant role in providing internet access in the south-east of Turkey, where fixed infrastructure is often lacking. The gradient is not very steep but it has a high R2 of 67%. Together with the other results it offers an initial model for estimating or forecasting the role of mobile broadband in other countries where less data is available.

The study was carried out by Christophe Le Lannou, of Data Learning and Tim Johnson of Point Topic.  Point Topic’s work on mapping broadband coverage across Europe revealed the availability of good official statistics in Turkey. Data Learning specialises in extracting significant relationships from multiple datasets so this was a good opportunity to discover what Turkey’s statistics might reveal about the drivers of broadband.

Total broadband take-up as a percentage of population in Turkey (2013)


The full report can be found here.