The Room · 21:40–23:00 · Part Two — The Sawn-off Step, 1984–1991

A Headcount

Of the people who find work in India, about one in four lands a formal wage job. Three studies disagree about why, and the law that sets the line has just moved.

7 min read · Priyanshu Sekhar Patra

An imagined night. Its year is left blank. Every number on the desk is the latest this book could verify: 21 October 2026.

The Labour Secretary comes in with a folder already open and sets it beside the Triage Sheet in front of you. One line on the sheet is filled. Thirty-nine are bare.

“Count first,” the Secretary says. “Every argument tonight is an argument about this count.”

You take the first folder.

Folder ALabour Secretary21 OCT 2026

Who works, and where

  • Employed. 56.2 crore people aged 15 and over, July to September 2025.
  • Where. 42.4 percent in agriculture, 55.8 percent self-employed, 18.9 percent casual labour.
  • Organised factories. About 1.96 crore persons engaged in 2023–24, in the Annual Survey of Industries.
  • The gap. Since 2017–18 the working-age population has grown by nearly 90 million and jobs by about 60 million, a shortfall of 5 million a year; 10 to 12 million people enter the labour force every year.
  • The filter. Of those who find work, 1 in 4 enters the formal wage economy.
  • The law. Four labour codes, in force since 21 November 2025. By 24 September 2026 ten states had published their rules; all were expected by 31 October.

The Secretary taps the line marked THE FILTER. Of every four people who find work, one enters the formal wage economy. The rest are absorbed into informal services, self-employment or low-productivity farming.

The factory line is smaller still. Divide 1.96 crore by 56.2 crore and you get about 3.5 percent; the two figures come from different years and different surveys, so read the ratio as a size and not as a statistic. Inside that sliver the factories are lopsided. In 2023–24, 22 percent of factories employed 79 percent of factory workers, and the 77 percent of factories with fewer than 100 employees held 21 percent.

The count itself is disputed, and the Secretary says so before you ask. The government’s own press note puts employment up from 47.5 crore in 2017–18 to 64.33 crore in 2023–24, a net addition of 16.83 crore. NCAER sees a shortfall of 5 million jobs a year against the growth of the working-age population. Those two sentences do not say the same thing, and this book has not reconciled them and does not average them.

The Secretary closes A. You open B. “Now the part the economists fight about. Why are the factory rows so thin?”

Folder BLabour Secretary21 OCT 2026

Four papers, three claims

  • Besley and Burgess (2004). Across Indian states in 1958–92, amendments to the Industrial Disputes Act that favoured workers went with lower output, employment, investment and productivity in registered manufacturing, more unregistered output and higher urban poverty.
  • Bhattacharjea (2006). Earlier studies of the 1976 and 1982 amendments gave ambiguous results, and he disputes the index Besley and Burgess built.
  • Hsieh and Olken (2014). In 2011 India had 1,370 firms with 97 to 100 workers and 1,013 with 101 to 104, out of 17,177,148. They find the bunching below the line of 100 economically very small, and no discontinuity among formal firms.
  • Bertrand, Hsieh and Tsivanidis (2021). Contract workers were 20 percent of employment in firms with 100 or more workers in 2000 and 38 percent in 2015.

“Read them as three claims,” the Secretary says. You met the first two in chapter 5: the law holds factories back, and the proof of that is thin. The third is new tonight. Firms do not bunch below 100, yet they have moved to contract labour, and those two facts sit awkwardly with either side’s story.

Hsieh and Olken do not clear the law. They write that their evidence does not rule out such forces, and that costs from regulation may instead rise smoothly with firm size, because enforcement is easier in larger firms. Credit, land, power and the quality of management are the other suspects. Each belongs to another office, and the Secretary does not rule on any of them.

The Secretary turns back to the line itself. You have met it before.

Replay of an earlier scene: Chapter V-B: the line at 300, from 5 March 1976 Replay of an earlier scene: Lowered to 100 by the Act of 1982, in force from 21 August 1984

The next entry is this year’s. The Industrial Relations Code, passed on 22 and 23 September 2020, puts prior permission for lay-off, retrenchment and closure at 300 workers again, where the 2019 Bill had said 100. The four codes took effect on 21 November 2025, and states may set the line higher. In most states, an establishment with 100 to 299 workers needed the government’s permission to lay off, retrench or close under the old Act; under the Code it does not, unless the state has set a higher line.

The Secretary slides a third folder across to you, thinner than the others and stamped with a country that is not yours. You take the third folder.

1996Korea
Folder CLabour Secretary21 OCT 2026

Korea, 1996 to 1998

  • April 1996. The government sets up a Commission for Reform of Industrial Relations, the first tripartite body to let democratic unions take part. The talks centre on an exchange: fewer restrictions on unions, a more flexible labour market.
  • 26 December 1996. In the early morning the ruling party passes a new labour law in a secret session of the National Assembly, in the absence of the opposition parties. It allows redundancy dismissal, labour dispatch, substitute workers during disputes and flexible hours.
  • 26 December 1996 to 10 March 1997. The first nationwide coordinated general strike in Korea since 1948. The law is returned to the Assembly and amended: multiple unions are allowed at national and industry level, and redundancy dismissals carry a two-year moratorium.
  • February 1998. After the financial crisis of 1997, the leadership of the Korean Confederation of Trade Unions signs a pact that trades legal recognition of the confederation for relaxed rules on redundancy dismissal and irregular employment. Delegates later reject the agreement; the revision goes ahead.

The folder does not say what to do. It says what happened, in order. A rule change passed over unfinished talks met a general strike; in 1998, in the middle of a financial crisis, a rule change that arrived inside a pact went ahead, though delegates later rejected the pact. India’s codes had their own table: the Economic Survey says they follow tripartite meetings of government, employers, industry and unions from 2015 to 2019. Ten central trade unions condemned the notification of November 2025 as unilateral, and the RSS-affiliated Bharatiya Mazdoor Sangh welcomed it. Both statements can be on the desk at once, and they are.

Now the choice, and it costs something whichever way you take it.

The codes carry both halves already. The same four codes that lift the permission line to 300 also set a re-skilling fund of 15 days’ wages for every retrenched worker, credited to the worker’s account within 45 days, brings gig and platform workers into social security, and makes fixed-term employees eligible for gratuity after one year. What the Secretary asks is which half you will make real first, because the rules are being published state by state.

Put the protection first. That means paying: payouts that reach a worker within 45 days, and gig and contract workers actually inside the provident-fund and insurance schemes, before you push anyone to loosen anything further. Then, so that the next move on the line is decided by evidence that both sides can read, publish firm-size counts for every state, by size band, every year. They will show your opponents whether the line is working, and you will not choose what they show.

Two rules go onto the sheet.

The first is the one the permission line has earned for you. The Code now in force says the threshold may only be raised, and by notification, a power that the central and state governments each hold. Design every instrument as if your opponent will inherit it. A line drawn under emergency conditions should be one you would let your opponent hold. R3, the 1976 Rule.

The second is a seat, not a rule. A standing table where government, employers and unions meet on a calendar, with the state firm-size counts on it as the first paper. Korea had a table in 1996 and the law was passed over its unfinished talks; it had one again in 1998. The Secretary writes it on the sheet in pencil. The standing tripartite table.

Below the sheet is a second page you had not seen: a column headed FIGHTS. The Secretary writes the first name on it, in ink, and does not underline it. It is the ten central trade unions. “A fight you have named is one you can plan for,” the Secretary says.

Triage sheet: 3 of 40 lines filled Door D2 · closedWhy are there so few factory jobs?

The Secretary takes the folders and leaves the sheet. The headcount is still unreconciled and the line is at 300. The door asks why India has so few factory jobs, and the folders hold five suspects: the line, credit, land, power and management. Korea’s folder is the only one that came with a country attached. Nine hours to the microphone.