Some Perspective

High headline growth has masked a measurable decline in the institutions of accountability and the income share of ordinary Indians. On cross-era measures India is worse under NDA (2014–26) than under UPA (2004–14). Democratic quality, statistical integrity, fiscal centralisation, press freedom and inequality all deteriorated. The one exception is inflation, and here is why that is not a policy result. See the computed verdict →

Two numbers that need more than a footnote

One measure on this site moves in the government's favour, and one measure leaves out most of the work women do. Neither can be explained fairly in a sentence, so each gets a page.

India's Growth Paradox

GDP up, jobs down, inequality soaring

High headline growth has masked a measurable decline in the institutions of accountability and the income share of ordinary Indians. On cross-era measures India is worse under NDA (2014–26) than under UPA (2004–14). Democratic quality, statistical integrity, fiscal centralisation, press freedom and inequality all deteriorated. The one exception is inflation, and here is why that is not a policy result. See the computed verdict →

Executive Summary: India's Growth Paradox (2004-2026)

These findings build one case: on cross-era measures India is worse under NDA (2014–26) than under UPA (2004–14). Democratic quality, statistical integrity, fiscal centralisation, press freedom and inequality all deteriorated. The one exception is inflation, and here is why that is not a policy result. See the computed verdict →

Key Findings

Formal vs Informal Employment

Formal/informal employment is shown for the PLFS survey years 2017-18 to 2023-24, each plotted at the year the survey starts. There is no comparable annual series before PLFS, and no annual table yet for 2024-25 onward.

Export Chart

The share of workers in regular wage or salaried jobs with social security fell from 11.5% in 2017-18 to 9.6% in 2022-23 while GDP grew, and recovered to 10.1% in 2023-24 (PLFS). Roughly nine in ten workers have no social-security cover.

Income Inequality Trends

Export Chart

The top 1% income share rose from 18.4% in 2004 to 22.6% in 2022-23, its highest since 1922. The bottom 50% share fell from 18.8% to a low of 13.9% in 2017 and was 15.0% in 2022-23. No annual figure is published for 2023 to 2026 (World Inequality Lab).

Your state

The national picture hides most of what is interesting. India’s richest large state has the per-capita income of its poorest, and a vote is not worth the same everywhere: seats were frozen on the 1971 census, so on the government’s own 2026 population projection the number of people behind each MP varies by between the best- and worst-represented states here, wider than the gap the 2011 census showed, because a frozen allocation keeps compounding. Pick a state.

Interactive Data Explorer

Export as PNG Export as CSV

Correlation Explorer

Three Novel Indices

These indices quantify institutional changes that are typically described only qualitatively. Between 2014 and 2026 the SSI rose and the DQI fell. The FCI peaked in 2020 and is now below its 2014 level, though its NDA-era average is above the UPA-era average.

All Three Indices (Normalized 0-10)

All three indices are computed for both eras (2004-2026) from one documented methodology (see data/compute_indices.py). SSI is a 0-10 weighted count of datable suppression events; FCI is the mean of six centralisation components, four of them published fiscal ratios min-max normalised over the sample and two coded (GST and the 2020 borrowing conditions); DQI is the geometric mean of the V-Dem Liberal Democracy Index, the Freedom House score, the RSF press-freedom rank and the V-Dem Core Civil Society Index. See Methodology.

Export Chart

Human Stories Behind the Numbers

Declaration of interests

The author is a former Manmohan Singh Fellow of the All India Professionals’ Congress (AIPC), a department of the Indian National Congress. This work began before the fellowship. This site compares the record of a Congress-led government (UPA, 2004–14) with the record of its successor (NDA, 2014–26) and concludes against the successor on eight of nine measures. That is a party interest in the conclusion, and it is declared here rather than left for someone else to discover.

Two things are also true. This research receives no funding from any institution, government, political party, foundation or commercial entity. And it is not party work: no party body commissioned it, reviewed it, approved it or saw it before publication. It was written at the author's own cost and is published under the author's own name.

A declared interest cannot be answered with a claim of neutrality. Nobody who has spent two decades arguing about Indian politics is neutral, and a site that claimed otherwise would be lying to you in its first sentence. The response offered instead is that every step between the raw data and the conclusion is open to inspection:

• The complete dataset is published as JSON, CSV and Excel, with every series traced to a named public source.
• The three constructed indices are built by a single published script, and the same script computes both eras, so no part of the gap between them can come from the two periods being measured differently.
• Where the evidence cuts against the argument it is reported anyway: inflation is lower in the second period, and the site says so and declines to explain it away.
• Corrections are published rather than made quietly, including an error that had run in this project's own favour.
• The governance finding is independently corroborated by a study this project did not produce, using a method it does not use.

None of that makes the author disinterested. It makes the author checkable, which is the only property that survives disagreement. If a figure here is wrong, it can be shown to be wrong from the published sources, and it will be corrected in public.

An independent check, by a different method

The strongest test of a finding is whether someone else reaches it without using your method, your data vintage, or your framing. Kevin and Robin Grier (Texas Tech) do exactly that in . Where this project builds composite indices from observed series, they build a synthetic control: a weighted blend of comparator countries fitted to India's path over pre-treatment years (–), then used as the counterfactual India that never elected the government it did.

Final-year gap between India and Synthetic India, as a percentage of India's own 2013 level. Bars run rightward in the direction of worse governance, so political corruption, where the raw gap is positive (+0.272), appears on the same side as the freedoms that fell. Freedom of religion is measured on an interval scale rather than 0–1, which is why its percentage exceeds 100.

Where the two projects agree

Note that these are not the same quantity. Their figures are gaps against a counterfactual; this project's are observed changes over time. What matters is that two different questions point the same way. The numbers themselves are not interchangeable.

Where they part company

Synthetic India for the income test is built mostly from . The democracy test draws on a separate pool of middle-income democracies, weighted afresh for each outcome (for polyarchy, 61.1% Argentina, 14.5% Sri Lanka and 13.4% Brazil): .

, (). Governance data: . Income data: . Figures transcribed from the authors' published summary; nothing on this page is recomputed by this project.

Cite & download

Every chart on this site has a ⤓ button that exports it as a PNG with the source and citation baked in. The full dataset and code are open.

Every series, its source, and how it was produced. All indicator values are observed (2014–2026); the Economic Trajectory tab holds the scenario projections.

Glossary

Plain-language definitions for the terms used across the site. Dotted-underlined words elsewhere are tappable for the same pop-up.

No terms match “”.

Methodology

SSI_t = Σ weight_i × severity_it   (0-10)

Weights: census 2.5, mortality 2.0, consumption 1.5,
GDP 1.5, institutional 1.5, employment 1.0
Severity: withheld or postponed 1.0, disputed 0.7,
scheduled delay 0.3, with decay after resolution

FCI_t = mean of the components published for year t

C1: Cess and surcharge share of gross tax revenue
C2: Tax devolution share of gross tax revenue (inverted)
C3: States' own revenue share of revenue receipts (inverted)
C4: Non-FC grants share of Union transfers
C5: Borrowing conditions (coded 0, 0.5, 1)
C6: GST centralisation (coded 0-1)
C1-C4 are min-max normalised over the years published.

DQI_t = (LDI × FH/100 × (180-RSF)/180 × CSI)^(1/4)

LDI: V-Dem Liberal Democracy Index
FH:  Freedom House aggregate score
RSF: RSF press-freedom rank
CSI: V-Dem Core Civil Society Index
A low value on any input pulls the score down.

International Comparisons & Implications

Democratic Backsliding: International Context

Export Chart

Era Comparison: UPA vs NDA

Both columns are period averages computed live from the data. Only consistent single-source series (no definitional breaks) are compared here. Unemployment and formal employment have no comparable UPA-era series; see the scorecard below.

Scenario Lab: "What If?" Simulator

Move the sliders above. The Sankey shows how each lever flows into the three indices. Band width shows each lever's scenario contribution.

State Explorer

India's states are diverging in income, in representation, and in demographics. Pick a metric; states are ranked and coloured by region.

Takeaway:

Deep Analysis

Computed from data.json. Every value here is observed (2004–2026); scenario projections live in the Economic Trajectory tab.

UPA vs NDA: Computed Scorecard

Averages over each regime period, computed live from data.json: UPA from the 2004-2013 series and NDA from 2014 onward. SSI, FCI and DQI are this project's constructed indices for both eras.

The Decoupling: Growth Without Jobs

Cumulative real GDP growth since 2014 (left axis) plotted against the formal employment share (right axis), which PLFS measures from 2017-18. If growth brought formal jobs, the two lines would rise together. Between 2017-18 and 2022-23 GDP grew and the formal share fell from 11.5% to 9.6%, and it was 10.1% in 2023-24.

Cumulative GDP series compounds annual growth from a 2014 base of 100. Formal employment is the share of all workers who are regular wage or salaried employees with any social-security benefit (PLFS, 2017-18 to 2023-24).

Peer Trajectories (2014 → 2026)

India alongside four other countries whose democracy scores have moved since 2014. V-Dem classifies India and Turkey as electoral autocracies. Click a header to sort.

What Moved in the Latest Year

Change between the previous and latest observed years for each headline indicator.

Supplementary Data Series

Three series kept as CSV files in data/, each holding only the years with a published figure and the source for each value. Years with no published figure are left empty. Click Export CSV on any chart to download its data.

What Next? Action Items

Research Package

Full paper, data & code

Technical Appendix

Index formulas & methodology

Guided Walkthrough

18 slides, live charts

Research Paper

Full working paper & PDF

Where is India's economy heading?

This tab shows where the numbers have been and where they are heading. Each indicator below is projected from 2026 to 2030 by extending its recent trend, then applying one of three explicit scenarios. The further out the line, the wider the band, because beyond the near term the number depends more on the assumption than on the economy.

Indicator forecast to 2030

Black line = observed (2014–2026). Dashed amber = central projection. Shaded band = ±1 scaled standard error, widening with the horizon. Method: OLS trend over recent years + scenario annual adjustment.

How to read a forecast (and why we resist calling it a prediction)

• • A trend can break. Each central path extends the recent OLS trend of an observed series. Trends break.
• • Scenarios are assumptions. The reform/stress adjustments are deliberate, documented choices. Change them in data.json and the future changes with them.
• • The band shows the uncertainty. It widens with the square root of the horizon: 2030 is far less certain than 2027.
• • The index depends on its weights. It averages eight indicators with transparent good/bad anchors; weight them differently and the headline moves.

Reading the Indian Economy

How the economy is measured, what the numbers say, and who decides what counts. Eleven sections, following the lecture from A (what an index even is) to K (how we know any of it). Pick a section below.