Suhani Bhatia
← Work

Independent analysis · 2026

Who counts as empowered?

Measuring women’s agency beyond scheme enrolment

Independent research · NFHS-4 and NFHS-5 · descriptive statistics · OLS

Financial inclusion rose sharply between NFHS-4 and NFHS-5. The change in the household decision-making measure was considerably smaller.

+25.6 ppWomen 15–49 with a bank or savings account they themselves use. NFHS-4 53.0% to NFHS-5 78.6%.

+4.7 ppCurrently married women who usually participate in three household decisions. 84.0% to 88.7%.

If scheme dashboards treat the first movement as evidence of the second, they are reading the wrong series. Own-use mobile phones moved 8.1 points. House or land ownership moved 4.9. Paid cash work moved 6.3. Accounts are the outlier.

Financial inclusion increased substantially faster than the agency and paid-work measures between NFHS-4 and NFHS-5.
Bank or savings account she herself uses53.078.6Participates in three household decisions84.088.7Mobile phone she herself uses45.954.0Owns house or land, alone or jointly38.443.3Worked in the last 12 months and was paid in cash17.223.5

Pale bar: NFHS-4 (2015–16). Dark bar: NFHS-5 (2019–21). Women 15–49. Decision-making is currently married women who usually participate in three household decisions. Source: IIPS and MoHFW, NFHS India reports.

The question this raises

Do the indicators Indian welfare uses as stand-ins for women’s empowerment move with the indicators the National Family Health Survey uses as stand-ins for agency? An op-ed with Prof. Namrata Chindarkar, JSW School of Public Policy, IIM Ahmedabad, is in progress on the monitoring half of that question. This page is the empirical companion.

Conceptual frame

Naila Kabeer’s distinction still earns its keep: resources, agency, achievements. An account is a resource. Decision-making, in the NFHS sense, is a thin measure of agency: currently married women who usually participate in own health care, major household purchases, and visits to family or relatives. Scheme dashboards typically count the resource (Jan Dhan, SHG membership, women as a share of MGNREGA persondays) and then speak in the language of the agency.

Data and variables

IIPS and MoHFW, NFHS-5 India report and Phase-II state factsheets; NFHS-4 India report for the earlier round. Published totals for women 15–49. The state analysis uses eleven Phase-II states that carry the three indicators on the same factsheet. It is an ecological cross-section.

StateDecisions %Bank %Mobile %
Arunachal Pradesh87.078.276.4
Chhattisgarh92.780.340.7
Haryana87.573.650.4
Jharkhand91.079.649.0
Madhya Pradesh86.074.738.5
Odisha90.286.550.1
Punjab91.481.661.2
Rajasthan87.779.650.2
Tamil Nadu92.892.274.6
Uttar Pradesh87.675.446.5
Uttarakhand91.080.260.9

Cross-section

Correlation between own-use accounts and decision-making: 0.72. Between own-use mobiles and decision-making: 0.22. OLS of decisions on accounts: intercept 63.37, slope 0.326, R² 0.52. Adding mobile phones takes R² to 0.55 and returns a mobile coefficient of −0.038. A 10-point difference in accounts predicts about 3 points in decision-making. The national time series is stricter. The same slope applied to a 25.6-point jump in accounts would have predicted about 8 points of decision-making. The country recorded 4.7.

Across eleven Phase-II states, a 10-point difference in own-use accounts predicts about 3 points in decision-making, not 10.
ARCGHRJHMPODPBRJTNUPUKWomen with a bank or savings account they themselves use (%)Participates in three household decisions (%)

Fitted line: decision-making = 63.4 + 0.33 × bank account (R² = 0.52). n = 11 Phase-II states. Ecological cross-section. Source: NFHS-5 Phase-II factsheets.

What monitoring would have to ask

MGNREGA’s gender number is women’s share of persondays, about 58% in FY 2024–25: labour days billed. DAY-NRLM counts members, loans, insurance covers. PMMVY counts portal instalments. Jean Drèze and Reetika Khera, using RTI figures in The Hindu (March 2025), put effective PMMVY coverage at about 9% of births in 2023–24. A short add-on to concurrent evaluation could ask whether she decided to apply, whether she knows the amount, whether the money landed in an account she operates, and whether she could refuse.

Limits of the evidence

Eleven states. Published factsheet totals, so ecological. No controls for wealth or education: adding them on n = 11 would dress the model without identifying anything. Decision-making is asked of currently married women; bank and mobile of all women 15–49. The national comparison mixes Phase-I and Phase-II in the published all-India figures. The slope is not a causal estimate of what opening an account does to a woman. It is a description of how two published series fail to travel together.

Sources

  1. IIPS and MoHFW, NFHS-5 (2019–21) and NFHS-4 (2015–16) India reports; NFHS-5 Phase-II factsheets.
  2. Ministry of Rural Development, MGNREGA dashboard, women’s persondays, FY 2024–25.
  3. Jean Drèze and Reetika Khera, “A leap backward for maternity entitlements,” The Hindu, 11 March 2025.
  4. Naila Kabeer, “Resources, Agency, Achievements,” Development and Change, 1999.

Stata do-fileCSVDictionaryResearch slides