A data-science study of how visible women's economic activity is: measurement model, uncertainty, sensitivity, scenarios and value of information - Niue

The Gender Economic Visibility Index (GEVI) asks whether women’s economic activity is present, measured and recognised in the systems that set policy, budgets and taxes. This paper applies the index to Niue as a data-science measurement problem: 13 inputs are converted to parity scores, aggregated into five dimensions by a weighted geometric mean, and run through a Monte Carlo model of 10,000 draws that varies indicator scores and dimension weights together.

Niue's provisional GEVI is 51.8 (90% interval 43.1 to 58.7), 6th of 15 on the point estimate but with a rank range of 5 to 8. The position is driven by one measured input: women hold 6 of 20 seats in the Niue Assembly (30%), the second highest share in the series. The other 11 of 13 inputs are imputed (blind-spot ratio 0.85), so the score is a placeholder.

The strongest dimension is job quality and seniority (0.806) and the weakest is institutional visibility (0.260); statistical and political voice takes the largest Shapley share (37%) of the 48.2-point gap to full parity. Of the 13 inputs, 1 is grade A, 1 is grade B and 11 are grade C (imputed), so the score is treated as a placeholder and its rank as unreliable. Across seven weighting schemes the rank runs from 6 to 7, and across ten alternative specifications from 4 to 11, with the score moving by up to 16.7 points. Raising data coverage to 9 of 12 rows is worth +9.1 points, and a greedy pathway of +0.10 steps on measured indicators reaches the 65-point Emerging threshold after 13 steps. Every number is reproducible from the accompanying Python scripts with a fixed random seed of 2026.