The Four Pillars
Legislature, executive, judiciary, press — the four estates of the republic, measured not by what they are meant to be but by what the official numbers show they do. Read as a full-period time series, because institutional drift is a seventy-year, every-government story, and showing it whole is what keeps this non-partisan.
In shortHow the four estates actually perform, on official numbers, across the whole period — not a party scorecard. The forms all still stand; it is the substance that has thinned, under every government.
The Lok Sabha meets 55 days a year, down from 135, and sends 16% of Bills to committee, down from 71%. Nearly half of new MPs declare criminal cases. Courts carry 5.4 crore pending cases with a third of High Court seats empty. And the state keeps no count at all of the journalists it charges.
Every figure below is from an official or first-party source — PRS and the government's own Statistical Handbook, candidates' sworn affidavits (ADR), the National Judicial Data Grid, NCRB Prison Statistics, the Law Commission, Parliament replies. Composite indices (press-freedom rankings) are graded below official data and shown only with their methodological critique and the government's rebuttal — never laundered into fact. Where a datapoint names the era or government of office, that is dated fact, not a scoreboard: nothing here is aggregated into a party's net record. The full row-level data is the vault's data/institutions.csv; every claim is tagged in the Claims Ledger.
1 · The Legislature — a House that meets less, and checks less
The first duty of Parliament is to sit, deliberate and scrutinise. On the government's own record, it does all three less than it used to — and the decline runs unbroken from the Nehru era.
| Lok Sabha | Sitting days / year |
|---|---|
| 1st (1952–57) | 135 |
| 4th (1967–71) | 123 |
| 15th (2009–14) | 71 |
| 16th (2014–19) | 66 |
| 17th (2019–24) | 55 |
In 2020 the Lok Sabha sat just 33 days — the fewest of any year — while the President promulgated 15 ordinances. Source: PRS; MoPA Statistical Handbook 2023.
| Lok Sabha | Bills sent to committee | Floor time lost to disruption | Government of record |
|---|---|---|---|
| 14th (2004–09) | 60% | 20.9% | UPA (Manmohan Singh) |
| 15th (2009–14) | 71% | 50.3% | UPA (Manmohan Singh) |
| 16th (2014–19) | 25% | 21.3% | NDA (Narendra Modi) |
| 17th (2019–24) | 16% | 25.6% | NDA (Narendra Modi) |
This is the non-partisan picture in one table. The 15th Lok Sabha (UPA) referred the most Bills to committee — 71%, the high-water mark of scrutiny — yet lost half its floor time to disruption, the worst on record. The 16th and 17th (NDA) lost far less time but scrutinised least, sending just 25% then 16% of Bills to committee, with about 80% of the budget voted through without discussion. Neither coalition is spared, and neither is the whole story: the sitting-day decline predates both, running straight from 135 days a year under Nehru to 55 today. Documented
2 · The Executive — and the watchdogs that don't bark
The clearest admission-against-interest in Indian public life is a candidate's own sworn affidavit. Across five elections and both coalitions, those affidavits show the same rising line.
| Lok Sabha election | MPs with criminal cases | …with serious cases |
|---|---|---|
| 2004 | 23% | — |
| 2009 | 30% | 14% |
| 2014 | 34% | 21% |
| 2019 | 43% | 29% |
| 2024 | 46% | 31% |
2024: 251 of 543 MPs, 27 of them convicted. Source: ADR analysis of candidates' sworn affidavits. "Serious" = murder, rape, kidnapping and the like.
The share of MPs declaring criminal cases has risen every election since 2004, under both alliances, to a record 46% — and the reason it keeps rising is structural: a candidate with a criminal case was about three and a half times likelier to win in 2024 than a clean one (15.3% vs 4.4%). Meanwhile the bodies meant to check the executive are starved: the Lokpal, legislated in 2013 after a national anti-corruption movement, got its first chairperson only in 2019 and in five years ordered 24 probes and just 6 prosecution sanctions; the Central Information Commission runs at 3 of 11 posts with a 3.2-lakh RTI backlog, despite Supreme Court orders to fill the seats. Weakening a watchdog by simply not appointing to it is a technique available to every government, and used by them. Documented
3 · The Judiciary — justice, but not in time
The courts are independent on paper and overwhelmed in fact. The capacity gap is not a recent slide; it is measured against a benchmark Parliament's own Law Commission set in 1987.
| Metric | Then | Now |
|---|---|---|
| Cases pending, all courts | 2.9 crore (2018) | 5.4 crore (2025) |
| Judges per million people | 10.5 (1987) | ~21 (2025) |
| High Court judge vacancy | — | ~32% (355/1,122) |
| Undertrials, share of prisoners | 66% (2005) | 73% (2024) |
The Law Commission's 1987 recommendation was 50 judges per million; the ~21 today (on sanctioned strength and the 2011 census) is nearer 15 on working strength and current population. Undertrials peaked at 77% in 2021; prisons run over capacity every year. Sources: NJDG; Law Commission 120th Report; NCRB Prison Statistics.
Thirty-eight years after the benchmark was set, India runs its courts at under half the recommended bench strength, and pendency has nearly doubled in a decade to 5.4 crore — one in ten district cases now older than ten years. The human face of that delay is the prison: for twenty years straight, two-thirds to three-quarters of everyone in an Indian jail has been an undertrial — not convicted of anything, merely waiting. This is the same machinery the scams register shows swallowing the biggest cases for decades — capacity starvation that no single government created and none has fixed. Documented
4 · The Press — leverage, not censorship
Outside the Emergency's open censorship, pressure on the press is subtler and must be shown, not asserted — through the money the state controls and the cases it files.
| Financial year | Central government ad spend (₹ cr) | Government of record |
|---|---|---|
| 2002-03 | 45 | NDA (Vajpayee) |
| 2013-14 | 1,036 | UPA |
| 2017-18 (peak) | 1,336 | NDA (Modi) |
| 2022-23 | 154 | NDA (Modi) |
Ad spend rose under both alliances then fell; the leverage was never its budget scale (~0.02%) but the power to withhold it. See The Government Ad Bill. Sedition (IPC 124A) cases rose from 35 (2016) to 93 (2019) at a ~3% conviction rate — the process is the punishment; the section was paused by the Supreme Court in 2022 and returns as BNS §152. The state keeps no count of journalists it charges, and no central register of internet shutdowns (which India leads the world in) — that absence is itself the finding, read through the court record (Anuradha Bhasin, 2020). Documented
India's rank on the RSF World Press Freedom Index fell from 133/180 (2016) to 159/180 (2024). That points the same way as the official data — but it is a perception-weighted composite index, its methodology was overhauled in 2022 (breaking comparability), and the Government of India formally rejects it on the parliamentary record (NITI Aayog's Deciphering the WPFI, citing small sample size and opaque method). So it sits here beside the official figures, graded below them, never in their place. Contested
Read together, the estates tell one story in four voices. The legislature still sits, the executive still faces watchdogs, the courts still hear cases, the press still prints — every form stands; it is the function inside each that has thinned. And it has thinned across seventy years and every government: the House that meets least and the House that lost half its time to disruption belong to different coalitions; the criminal-MP line rises through both; the judicial shortfall predates all of them; the press pressures recur from the Emergency onward. That is the definition of a system the rest of this archive uses — the same behaviour regardless of who governs — which is exactly why the fix the project argues for is structural, not electoral. Documented
- Official data first; indices last. Press-freedom rankings are composite, perception-weighted indices shown with their critique and the government's rebuttal — graded below the official series, never laundered into settled fact.
- Affidavit figures are floors. The ADR criminal-case shares are candidates' own sworn declarations — admissions against interest, so the real figure is plausibly higher, not lower.
- Some series are shorter than others. Committee-referral and time-lost data are cleanest from the 14th Lok Sabha; NJDG pendency is a live dashboard (cite with a date); a few per-Lok-Sabha sitting-day averages are unavailable and omitted rather than guessed.
- Government-of-record, not a scoreboard. Where a term or government is named it is a dated fact; nothing here is aggregated into a party's net record, and every pillar's failing spans administrations.
Sources
- Legislature: PRS Legislative Research (Vital Statistics; Functioning of the 17th Lok Sabha) and the Ministry of Parliamentary Affairs Statistical Handbook 2023 (Tables 16, 18, 20, 30–30C).
- Executive: Association for Democratic Reforms analyses of candidates' sworn affidavits; the Lokpal (via The Hindu); Satark Nagrik Sangathan Report Card of the Information Commissions 2023-24.
- Judiciary: National Judicial Data Grid and Ministry of Law & Justice replies; Law Commission of India, 120th Report (1987); NCRB Prison Statistics India.
- Press: DAVP/CBC spend (RTI, via FACTLY, and Rajya Sabha replies); NCRB Crime in India (sedition); Anuradha Bhasin v Union (SC, 2020); the Shah Commission (1978); RSF WPFI shown with NITI Aayog's rebuttal.
- Row-level data in the vault's
data/institutions.csv; every figure tagged in the Claims Ledger.