Satellite Night-Light GDP Analysis
Combined VIIRS night-light and World Bank indicators to examine whether satellite-observed brightness can complement conventional measures of national economic activity.
Period · Spring 2025
Role · Individual project
01 · Business / research question
The question
Can night-light intensity help explain GDP where official statistics are limited, and do urban population and electricity access change that relationship?
02 · Evidence
What the analysis used
03 · Analysis
How the work progressed
- 01
Integrate country-year data
Merged satellite brightness and economic indicators at the country-year level.
- 02
Prepare the variables
Applied log transformations to skewed variables and centered moderator variables.
- 03
Test the relationship
Estimated the baseline regression and added urban-population and electricity-access interaction terms.
- 04
Define responsible use
Interpreted model fit alongside infrastructure effects, missing data, and non-economic sources of light.
04 · Interpretation
Main insight
Night-light intensity explained 81.9% of GDP variation in the reported simple model (R-squared = 0.819).
Urban population and electricity access produced statistically significant interaction effects in the report.
The result is explanatory and should not be described as prediction accuracy or causal proof.
05 · Practical decision
Decision value
Night-light data can provide a complementary signal for early market screening, country-risk research, and economic monitoring where conventional reporting is limited.
06 · Validation
Limitations and next checks
- •The final country count and the 973/820/791 dataset stages require a reproducible data dictionary.
- •An earlier-year Albania record and the final brightness definition require reconciliation.
- •Fixed-effects and out-of-sample analysis are needed before making predictive claims.
07 · Visual evidence
Evidence, with boundaries
Alternative-data pipeline
Verified sourceThe diagram shows how satellite and economic indicators enter a country-year analysis. It explains provenance and process; it does not establish that night lights cause GDP.
Source · GDP project report, PPTX, XLSX, and SPSS output
SPSS-reported model fit
Reported evidenceReported R-squared
0.819
N = 791 in the reported simple-model output
The reported simple model has R-squared = 0.819. This is explanatory fit within the reported sample, not 81.9% prediction accuracy and not causal evidence.
Source · SPSS model summary embedded in the GDP presentation
Distinct dataset stages
Reported evidence973
Raw rows
820
Merged observations
791
GDP observations
973, 820, and 791 refer to raw rows, merged observations, and GDP-usable observations. They are different processing stages and must not be treated as interchangeable sample sizes.
Source · Project XLSX and HWPX methodology section
Log night-light versus log GDP scatterplot
Pending verificationChart to add after verified data export
The brightness definition, exclusions, transformations, and final 791 observations must be reconciled in one reproducible export.
The report describes log-transformed median brightness, while the available workbook contains brightness_sum and does not reproduce R-squared = 0.819 through a direct rerun. No scatter or regression line is published until the transformed dataset is exported.