Skip to content
← All indicators

Infrastructure · indicator 19

Daily metro ridership

Modi ahead

Rahul India

110.4m

synthetic 2023

Modi India

180.4m

actual 2023

Difference

+63%

vs counterfactual

Placebo rank

5/31

p = 0.16

What this measures

Daily metro ridership is passengers flown by registered carriers in a year. A rising number is an improvement.

India stood at 75.6m in 2013, before the two paths separate. By 2023 Modi’s current India had risen to 180.4m, while Rahul’s India — the blend of comparison countries that tracked India before 2014 — reached 110.4m.

That leaves Modi’s current India 63% higher than Rahul’s India — better than where comparable economies ended up. But Rahul’s India missed India’s own pre-2014 path by 15.7%, so this difference should not be read as an effect.

Both paths, 1984–2023

Weights fitted on 1984-2013 only; the counterfactual is projected across 2014-2023.

Modi India · actualRahul India · syntheticModi ahead

Divergence

Modi India minus Rahul India, in passengers.

What Rahul’s India is made of

Donor weights, solved on pre-2014 data. 30 countries in the pool; those not listed carry zero weight.

  • Brazil
    34.6%
  • Ethiopia
    24.4%
  • Indonesia
    21.9%
  • Uganda
    10.7%
  • China
    8.5%

Method & reliability

Measure
air_passengers
Source type
Cross-country proxy
Estimation window
1984-2013
Projection window
2014-2023
Donor pool
30 countries
Pre-2014 fit (RMSPE)
15.7% of pre-period mean
Placebo rank
5/31
Standardised p-value
0.161
BH-adjusted q across tier 1
—

The q-value applies a Benjamini–Hochberg correction across all 20 indicators with a counterfactual. We report p as the primary figure because each indicator is treated as a separate question rather than one family of hypotheses — but no result on this site survives q ≤ 0.10, and a reader who prefers the family view should read the p-values accordingly.

Pre-2014 fit is 15.7% of the pre-period mean. The method could not reproduce India’s own path before 2014, so the post-2014 difference is not interpretable as an effect.

metro ridership -> air passengers carried (weak)

How this was calculated

Nothing here is chosen by hand. The comparison countries are not picked for being like India — they are whatever mix best reproduces India’s own path before 2014 on this particular measure.

  1. 1Describe India before 2014

    India's daily metro ridership is summarised over 1984-2013 — the value at five evenly spaced years starting 1984, plus three background controls averaged across the whole window: GDP per capita, urbanisation and life expectancy.

  2. 2Find the blend that matches it

    Each of the 30 comparison countries is described the same way. An optimiser then searches for the weighted mix of them whose pre-2014 path sits closest to India's. Weights cannot be negative and must sum to 100%, so the result is always a real combination of real countries — never an extrapolation.

  3. 3Check the match is good enough

    The blend misses India's actual pre-2014 path by 15.7% on average. That is too loose to interpret — the post-2014 difference is not reported as an effect.

  4. 4Project it forward and compare

    The weights are frozen and applied to 2014-2023, years the optimiser never saw. That projection is Rahul's India. The difference against what actually happened is the gap shown above.

  5. 5Test it against the other countries

    The whole procedure is re-run pretending each comparison country was the one treated in 2014. India's break ranks 5/31 against those placebos, giving p = 0.161 — well within what chance produces, so this gap is not distinguishable from noise.

Why Brazil?

Brazil carries 35% of the weight here — not because it resembles India economically or politically, but because its daily metro ridership moved like India’s did between 1984 and 2013. On 23 of the other 27 indicators on this site it carries no weight at all; where it does appear, the largest is defence production at 100%. The blend is re-solved from scratch for every indicator, which is why the same country can dominate one estimate and be absent from the next.

5 of 30 comparison countries carry any weight at all. The rest are assigned zero by the optimiser.

The counterfactual is 'India without the 2014 structural break', estimated from comparator countries. It is not a forecast of any individual's policies. Sources: World Bank WDI API (2026-07-13); IRENA via Our World in Data.