Digital Infrastructure & Economic Development in India

Just my personal research project looking at how telecom infrastructure affects economic growth across Indian states. Been working on this for a few months now using data from 2018-2023.

What I'm Working On

I got curious about whether building more cell towers and digital infrastructure actually leads to better economic outcomes. So I gathered data from various Indian states and started digging into the relationships between things like:

• Telecom tower density
• Wireless/wireline user numbers
  
• GDP per capita
• Digital payment adoption
• School internet access

The results have been pretty interesting - there's definitely some strong patterns emerging.

Files in This Repo

I've organized my analysis into different notebooks as I worked through various approaches:

notebooks/01_spatial_infrastructure_analysis.ipynb - Started here just calculating averages for neighboring states to see if there were any spatial patterns. Pretty basic stuff but helped me understand the data.

notebooks/02_telecom_correlation_analysis.ipynb - Dove deeper into correlations between telecom variables. Found some surprising relationships, especially with porting requests.

notebooks/03_infrastructure_relationships.ipynb - Refined the correlation analysis after I realized I needed better data cleaning. This one has the state-by-state breakdowns that were really eye-opening.

notebooks/04_network_spatial_mapping.ipynb - Got into NetworkX to map out state relationships geographically. The visualizations turned out pretty cool.

notebooks/05_panel_causality_analysis.ipynb - Attempted some serious econometric analysis here. Had to install linearmodels and figure out panel data methods. Still not 100% sure I got everything right.

notebooks/06_temporal_causality_testing.ipynb - Looking at UPI transactions vs wireless users over time. The Granger causality tests were tricky to set up properly.

notebooks/07_gdp_prediction_models.ipynb - Built a linear regression model to predict GDP from telecom variables. Actually got decent results (R² around 0.82).

notebooks/08_comparative_regression_analysis.ipynb - Comparing different infrastructure types. Still working on this one.

Running the Code

You'll need Python 3.8+ and Jupyter. I've been using conda but pip should work too:

bash git clone https://github.com/SuhaanCoding/digital-infrastructure-india.git cd digital-infrastructure-india conda env create -f environment.yml conda activate digital-infra jupyter lab

Main Findings So Far

• Wireline users have a much stronger correlation with GDP than wireless users (was not expecting this!)
• States with high tower density don't necessarily have better economic outcomes
• There's definitely some spatial spillover effects between neighboring states
• UPI adoption seems to follow wireless user growth by about 4-6 months

Data Sources

Mostly pulled from: - TRAI reports (telecom data) - Government statistical yearbooks - RBI payment system data - Various state government websites

Notes

This is ongoing research, so some of the analysis might change as I refine my methods. Also, I'm still learning some of the more advanced econometric techniques, so feedback is welcome!

The data cleaning was honestly the most time-consuming part - lots of inconsistent formatting and missing values to deal with.

License

MIT License - feel free to use this for your own research.