A question corpus as a data source

Parliamentary questions are the most under-used official dataset in India. They are dated, attributed to a named member, answered on the record by a named minister, and — unlike a press release — the topic is chosen by someone trying to extract something the government would not otherwise publish. Read one and you learn a fact. Read three decades and you learn what the political system thought was worth asking about, and when that changed.

The Rajya Sabha publishes its question database through a search backend that accepts a filter and returns the question, the member, the ministry, the date and, for older years, the full answer text. Filtering for fertiliser-related titles across the whole archive returns 2,606 questions answered between 4 August 1995 and 28 July 2026.

One cleaning note, because it materially affected the count. A naive title filter on the string urea also matches bureau, which pulled in 183 questions about the Narcotics Control Bureau, the Press Information Bureau, aircraft accident investigation and convention promotion. Those are removed. The figures below are on the cleaned corpus.

The shape of three decades

Rajya Sabha fertiliser questions by year of answer. Red marks years at or above 110. 2020 is the pandemic session disruption. 1995 and 2026 are partial years at the ends of the archive.

The peaks are 2000 (148), 2012 (141), 2025 (120), 2002 (116) and 2011 (109). Those cluster around exactly the moments you would expect: the decontrol debates and urea pricing reviews of the early 2000s, the nutrient-based subsidy transition and the subsidy-arrears crisis around 2011-12, and the current period of DAP shortage and nano-fertiliser controversy.

What is more interesting is the floor. Fertiliser has never fallen out of the House’s attention. The quietest full year in three decades, excluding the pandemic, is 2004 with 39 questions — still better than three a month in a chamber that also has to cover everything else.

What stayed the same, and what arrived

Classifying every question by theme and comparing three roughly equal periods shows which concerns are structural and which are recent.

Share of that period’s questions whose title or text matches the theme. Periods hold 796, 855 and 955 questions. Themes overlap — a question about shortage of imported DAP counts in three — so columns do not sum to 100. Classification is keyword-based and this article’s own; it is a measure of what was asked about, not of what was found.

Shortage is the big mover, from 4 per cent of questions in the first decade to 17 per cent in the current one. In the 1990s the House asked how much fertiliser cost and who made it. It now also asks, routinely, whether farmers in a named district could actually buy any. That is a different kind of question — retail rather than industrial — and it tracks a real shift in where the binding constraint sits.

Alternatives went from a curiosity to a category. Organic, bio and nano fertiliser questions rise from 3 per cent to 16. Nano alone accounts for a large share of the recent rise, which is consistent with a product launched in 2021 and regulated downward in March 2026. Parliament has been tracking it closely: recent sessions carry questions on nano fertiliser manufacturing in Assam, adoption of nano urea and nano DAP in Odisha, and self-reliance in the nano fertiliser sector.

Quality and soil health both roughly tripled, from a very low base. Two per cent of questions in the 1990s touched spurious or substandard product; 7 per cent do now. Soil health and nutrient imbalance went from 2 to 6 per cent. Neither is large, but both are concerns that did not exist as parliamentary categories thirty years ago.

Direct benefit transfer appears from nothing. Zero questions in the first two decades, 3 per cent in the current one. That is what a genuinely new policy instrument looks like in this data.

Namrup: a question asked in every one of thirty-one years

The single clearest finding in the corpus is not a trend. It is a plant.

Counted on any mention in the question title, question text or answer text. All six first appear in 1995, the earliest year in this archive, and all six appear again in 2026 — so the true span is at least 31 years and probably longer than the data can show.

Read the Namrup question titles in order and the effect is close to comic. Revival of sick fertilizer plants (1996). Rehabilitation of sick fertilizers units (1998). Viability of sick fertilizer units (1999). Closed fertilizer units (2000). Unviable chemical fertilizer projects (2003). Revival of public sector fertilizer plants (2010). Revival of sick fertilizer industries (2012). Revival of loss making fertilizer plants (2014). Revival of sick fertilizer plants (2015, again). Cost of revival of closed fertilizer plants (2016). Re-operationalizing closed fertilizer plants (2017). Revival of closed fertilizer plants (2020). Revival and expansion of urea manufacturing capacity (February 2026). Ammonia-Urea Plant at Namrup (July 2026).

Across the whole corpus, 319 questions use the language of revival, rehabilitation, reopening or restarting — 59 in the first decade, 142 in the second, 118 in the third.

What the answers said, decade by decade

The answers are on the record too, and they have their own rhythm.

In December 2005, the Minister of Chemicals and Fertilizers told the House that a proposal envisaged “setting up of brownfield gas based fertilizer plants at the existing sites of Gorakhpur and Sindri units of Fertilizer Corporation of India Ltd”, and that GAIL “has indicated that they have drawn up a plan”.

In August 2011, on Ramagundam, the answer was that “pursuant to the decision of the Cabinet, an Empowered Committee of Secretaries (ECOS) under the Chairmanship of Secretary (F) was constituted to evaluate all options for revival of closed units” of FCIL and HFCL.

In March 2020, the answer had become concrete: the government “is reviving five closed fertilizer plants” — Talcher, Ramagundam, Gorakhpur and Sindri of FCIL, and Barauni of HFCL — “by formation of Joint Venture Companies of nominated PSUs for setting up new Urea Ammonia plants of 12.7 Lakh metric tonne per annum capacity each”.

Fifteen years from a plan GAIL had drawn up to five joint ventures with a stated capacity. That is slow, but it is movement, and the 2020 answer is the first in the series that names a number rather than a committee. By July 2026 the House is asking about the fourth ammonia-urea plant at Namrup and about production capacity added by the revived Ramagundam, Gorakhpur, Sindri, Barauni and Talcher units — which is the question you ask when the earlier answer has finally become checkable.

The archive closes in 2020

One finding is about the data rather than the fertiliser, and it matters for anyone who wants to repeat this exercise.

Answer text is present and machine-readable for 1,959 of the 2,606 questions, or 75 per cent. The coverage is essentially complete from 1999 to 2019 — every year at 100 per cent — partial at the 1995–1998 start, and then it stops. From 2021 onward the answer field is empty in every single record. The answers still exist; they are attached as PDFs. But the text is no longer in the database.

Counts here are on the pre-clean corpus for the yearly breakdown and so are approximate against the 2,606 clean total; the 2021–2026 zero is exact and was checked record by record.

This is not a conspiracy, and it is probably a platform migration. But the practical effect is that the last five years of ministerial answers on fertiliser — covering the nano-urea launch, the DAP shortage, the subsidy spike and the plant revivals actually completing — cannot be searched in bulk the way the previous twenty-five can. Anyone wanting to analyse what the government has said since 2021 has to download and parse several hundred PDFs.

What this corpus is good for

• Dating a policy concern. DBT appears from zero. Nano appears from nothing. The year a theme enters the question stream is a reasonable proxy for when it became politically live.
• Testing whether a problem was solved. A question asked in 1996 and again in 2026 in nearly identical words is evidence about outcomes that no annual report will give you.
• Finding what the government will not publish. Members ask precisely because the data is not in a release. The answers frequently contain state-wise or plant-wise tables that exist nowhere else.
• What it cannot do. Question volume measures political attention, not severity. A spike may reflect a genuine crisis or simply a member with a constituency interest. And keyword classification of the kind used here is a blunt instrument — the bureau problem is a reminder of how blunt.

The honest summary

Three decades of questions describe a policy area where the framing has been remarkably stable and the specifics have moved a great deal. The House has always wanted to know what fertiliser costs the exchequer and why India does not make more of it. It has more recently added whether the stuff reaches the farmer, whether it is genuine, whether the soil can take it, and whether something else might work instead.

And it has asked about Namrup, Talcher, Gorakhpur, Sindri, Barauni and Ramagundam in every year the archive covers. Five of those six now have joint ventures and stated capacities, which is more than could be said at any earlier point in the series. Whether that ends the questions is the thing to watch: the 2026 questions are no longer asking whether the plants will be revived, but what the revival produced. On this evidence that is what progress looks like in the Indian fertiliser sector — not the question disappearing, but its tense changing.

Sources and caveats