Archive · 21 Aug 2026
Another Statistical Exercise in Belittling India’s Growth Story
Two researchers built an imaginary version of India using economic data from five other countries.
They used this imaginary country to estimate how India might have performed without Narendra Modi.
Their study says India’s incomes were about 10% lower in 2023 than in that comparison.
Vidhu Shekhar argues that the method also predicts a similar shortfall for the years before Modi became prime minister.
He says this suggests the method may be mistaking normal differences for a political effect.
He also tested other data sources and country combinations, which produced different results.
Some alternatives showed India doing about as well as or better than the imaginary comparison.
Shekhar concludes that the study does not identify one reliable version of an India without Modi.
The researchers’ estimate and the article’s critique therefore reach different conclusions about India’s growth.
Kevin Grier and Robin Grier estimate that India’s 2023 incomes were about 10% below a synthetic counterfactual without Narendra Modi.
Their synthetic India combined Ethiopia, China, Bangladesh, Pakistan and the Philippines, using data before 2014.
Vidhu Shekhar argues that back-testing the method produces a similar shortfall before Modi took office.
The article says results change substantially with different data vintages, GDP measures and donor-country pools.
Shekhar also disputes the paper’s reported p-values, arguing that its small samples cannot establish conventional statistical significance.
- Who
- Kevin Grier and Robin Grier produced the estimate; Vidhu Shekhar, an associate professor of finance at SPJIMR, criticized it.
- What
- A statistical synthetic-control study estimated that India’s 2023 incomes were roughly 10% below a counterfactual India without Narendra Modi.
- When
- The study evaluates the period after 2014 and reports an estimate for 2023; the article also discusses earlier data through 2004 and 2014.
- Where
- The analysis concerns India and uses economic data from selected countries, including Ethiopia, China, Bangladesh, Pakistan and the Philippines.
- Why
- The researchers used a synthetic-control method to estimate a counterfactual growth path, while Shekhar argues that the result depends heavily on methodological choices.
Researchers’ Estimate
Author’s Critique
Estimated Modi-era gap
Researchers’ Estimate
Kevin and Robin Grier’s synthetic-control model concludes that India’s 2023 incomes were about 10% below the level predicted for an India without Modi.
Author’s Critique
Shekhar argues that the estimate is not reliable because comparable synthetic countries can be built using different data, measures and donor pools.
Choice of counterfactual
Researchers’ Estimate
The study uses a five-country synthetic India—made from Ethiopia, China, Bangladesh, Pakistan and the Philippines—that closely tracked India before 2014.
Author’s Critique
Shekhar says close pre-2014 matching does not prove that the selected mixture is the unique or accurate India without Modi; alternative mixtures can match the past and diverge afterward.
Statistical certainty
Researchers’ Estimate
The paper reports highly extreme placebo-test results, described as “p = 0.00,” in support of its findings.
Author’s Critique
Shekhar says the small donor pools make such zeros insufficient for conventional 1-in-20 significance and may only show that India was the most extreme case in a limited comparison.
- Reported estimate
- India’s 2023 incomes were about 10% below the study’s synthetic counterfactual.
- Synthetic-country weights
- 38% Ethiopia, 28% China, 25% Bangladesh, 7% Pakistan and 2% Philippines.
- Treatment-period starting point
- The study treats 2014 as the beginning of the period associated with Modi.
- Alternative data result
- Using World Bank data with the same 14 countries, about six in ten well-fitting models put actual India ahead.
- Alternative GDP measure
- Using a PWT measure recommended for comparing growth rates, the alleged deficit disappears; India ends between parity and roughly 6% ahead.
- Broader donor pool
- Applying the selection logic more broadly produced 29 qualifying economies and 118,755 possible five-country baskets.
- Statistical-test concern
- Shekhar says the small donor pools limit exact-test p-values to about 0.07 for GDP and 0.08 for governance.