What a degree unlocks, why families chase it, and what an NCVT certificate is worth

Internal working note (no access: public frontmatter → withheld from docs.avantifellows.org). 13 September 2026. Companion to HYPOTHESIS.md.

Three questions from the Avanti team, answered in that order:

What does not holding a degree or 3-year diploma formally cost a person — which doors close?
What does the literature say drives the demand for degrees, and how much of it is the government-job aspiration?
What good is an ITI / NCVT trade certificate: does PLFS show it paying off, or does it mainly supply industry with cheap labour?

Provenance. Warehouse figures [W] are PLFS calendar-2025, first visit, weighted, from plfs_fact_persons (queries reproduced in section 6). Literature and rules [S] come from a three-agent research pass that, like the earlier passes, could not open any web page: every [S] figure is a search-engine excerpt of the cited document at the URL given, to be confirmed before external use.

1. What the warehouse says

1.1 Who actually holds government jobs (ages 25–54)

Pathway	Of this group, in a government job	Share of ALL government jobs	Private salaried	Of salaried, formal
Engineering degree (UG+PG)	11.7%	3.6%	55.4%	94%
Medicine degree	18.8%	1.1%	34.3%	87%
Other technical degree (B.Ed-heavy)	25.7%	6.7%	27.8%	86%
Graduate-level diploma	24.2%	4.5%	31.6%	87%
Polytechnic diploma	12.1%	2.3%	42.4%	79%
Other sub-degree diploma	17.2%	5.0%	29.9%	77%
General PG	17.3%	12.4%	24.5%	82%
General graduate	10.4%	23.4%	21.4%	73%
Class 12 + ITI-like	12.0%	0.9%	26.3%	76%
Class 12, no full-time training	5.7%	13.1%	16.1%	55%
Class 10 + ITI-like	6.0%	0.2%	32.0%	55%
Class 10, no full-time training	3.8%	10.1%	15.0%	42%
Below Class 10	1.3%	16.7%	9.4%	28%

“Government” is a regular salaried job with a government or local body, a public enterprise or an autonomous body as employer (etyp_pas 05–07). “ITI-like” is the narrow PLFS proxy: no technical qualification, formal full-time vocational training of three months or more.

Three readings. General graduates are the largest single bloc in government employment, holding almost a quarter of all government jobs, yet only one in ten of them has one. The pathway that converts best to a government job is the B.Ed-heavy “other technical degree” at 26%, because school teaching is the largest government occupation open to a degree. An ITI-like certificate on top of Class 12 doubles the government-job rate against Class 12 alone (12% vs 5.7%) — the Railways technician and Group D doors are real — but from a small base.

1.2 What general graduates do for a living (employed, ages 25–34)

Occupation group (NCO major)	Engineering degree	Polytechnic	General graduate	Class 10/12 + ITI-like	Class 10/12, no training
Managers	12.7%	7.0%	7.1%	4.5%	2.2%
Professionals	56.0%	11.6%	16.1%	8.8%	2.2%
Technicians, associate professionals	9.4%	12.1%	8.1%	8.1%	2.9%
Clerical	5.7%	7.3%	10.8%	7.1%	2.7%
Service and sales	6.3%	12.3%	19.6%	13.0%	17.6%
Skilled agricultural	3.3%	10.2%	20.1%	10.4%	26.3%
Craft and related trades	3.9%	22.8%	7.3%	26.9%	16.7%
Plant and machine operators	1.5%	10.0%	4.8%	11.9%	10.7%
Elementary	1.4%	6.7%	6.2%	9.3%	18.8%

Fifty-eight per cent of employed general graduates aged 25–34 work in farming, sales, trades, machine operation or elementary jobs — occupations that do not require the credential. Only 31% are in managerial, professional or associate-professional work. The ITI-like group does what its training is for: 39% are in craft trades or machine operation, against 27% for the same school level without training, and it is a fifth as likely to be in elementary work.

1.3 The waiting curve (unemployment rate by age, principal status)

Pathway	22–24	25–27	28–30	31–34
Engineering degree	42.2%	17.6%	8.7%	5.5%
General PG	40.6%	25.9%	18.6%	8.6%
General graduate	32.2%	18.0%	10.5%	5.0%
Polytechnic diploma	24.3%	10.2%	7.4%	4.7%
Class 12, stopped	11.2%	5.8%	3.3%	2.4%

The curve is the queue: a third to two-fifths of new graduates report themselves unemployed and looking, and the rate halves every three years until it converges on everyone else’s by the early thirties. This is what Jeffrey called “timepass” and what Mangal measured as exam preparation (section 3). It is also why the page reads outcomes at 25–29 and 35–40 rather than at graduation.

1.4 What an ITI-like certificate buys, against the same school level without one

	25–29, ITI-like	25–29, no training	35–44, ITI-like	35–44, no training
Regular salaried	37.7%	18.4%	42.3%	20.7%
In a government job	5.2%	1.8%	14.1%	5.2%
Of salaried: written contract	51%	29%	53%	40%
Of salaried: any social security	55%	31%	63%	41%
Of salaried: in manufacturing	30%	28%	35%	27%
Casual wage labour	6.2%	11.3%	2.8%	9.4%
Median salary	₹15,000	₹15,000	₹23,100	₹15,000
90th-percentile salary	₹28,000	₹22,000	₹60,000	₹30,000

Sample: 763 and 830 ITI-like people in the two bands; population Class 10 or 12, no technical qualification. Not a causal estimate — completers differ from non-completers.

At entry the certificate buys formality, not pay. A 25–29-year-old with full-time training earns the same ₹15,000 median as one without, but is twice as likely to be salaried, twice as likely to hold a written contract and social security, and half as likely to be in casual labour. The pay arrives later: by 35–44 the median is ₹23,100 against ₹15,000, the 90th percentile ₹60,000 against ₹30,000, and the government-job rate 14% against 5%. Both halves of the “cheap labour” argument are visible in one table: industry takes trained school-leavers on formal terms at unskilled pay, and a decade later a real premium has opened up for those who stayed.

2. Which doors close without a degree or diploma [S]

The credential ladder maps almost one-to-one onto the 7th Pay Commission levels:

Door	Credential	Pay level / scale	Queue
Railways Group D, SSC MTS, SSC GD constable, Agniveer GD	Class 10 (Group D: Class 10 or ITI)	Level 1–2	~333 applicants per Group D post (1.08 crore for 32,438); ~195 per SSC GD post
Railways ALP / Technician	Class 10 plus NCVT/SCVT ITI in the trade, mandatory	Level 2	~76 per ALP post — the trade filter thins the queue
SSC CHSL, RRB NTPC undergraduate posts, police constable	Class 12	Level 2–4	UP Police: ~80 per post
Junior Engineer (SSC JE, RRB JE, state PWD and electricity boards, PSU trainee)	3-year polytechnic diploma (degree also accepted)	Group B, Level 6, ₹35,400 start
SSC CGL, IBPS PO and Clerk, RRB NTPC graduate posts, state PSC, UPSC	Graduation	Level 4–10
Graduate engineer / officer (UPSC ESE, PSU through GATE)	B.E./B.Tech	Level 10, ₹56,100 start
School teacher	D.El.Ed (primary) or B.Ed (upper primary) plus TET	Level 6–7
Registered pharmacist	D.Pharm	statutory (Pharmacy Act)
Electrical wireman permit / Class C contractor	ITI electrician or wireman	statutory	Supervisor certificate and Class A/B contractor need a diploma or degree
IT-services campus entry	any degree, typically 60% throughout
Driving, including transport	none since 2019 (GSR 681(E))

Benefits are gated at the institution, not the person: post-matric scholarships and the 2024 Model Skill Loan reach ITI and polytechnic students; PM-Vidyalaxmi’s collateral-free loan reaches only NIRF-ranked degree institutions. Under the apprenticeship schemes an ITI holder’s stipend is ₹8,200–9,600 a month with the government paying a quarter; a diploma holder’s is ₹10,900 and a graduate’s ₹12,300 with the government paying half.

Equivalence is state law, not national: Haryana and Gujarat notify the 3-year diploma as Class-12-equivalent, Assam notifies ITI pass-outs as Class 10/12-equivalent, and the NCVT’s own FAQ says an ITI certificate is not Class 12. A Railways-specific “ITI plus Class 10 equals Class 12” rule was not found. The DGT–NIOS integrated scheme now lets ITI trainees earn the Class 10 or 12 certificate alongside the trade, which is the system’s own answer to the gap.

What the pattern says. Most government vacancies do not need a degree: in 2024-25 the Railways posted about 32,000 Group D and 19,000 ALP posts against about 10,000 graduate NTPC and 8,000 JE posts, and SSC posted 47,000 GD posts against 18,000 CGL. But the posts that pay Level 6 and above do need a diploma or degree, and the doors open to a Class 10 or 12 leaver are the most crowded in the country. The polytechnic diploma is the only sub-degree credential that opens a Group B post on its own, and it also carries the AICTE route into the second year of a B.Tech.

Sources (all excerpt-level): SSC eligibility and vacancy pages via Guidely, Entri, Careers360, Adda247; RRB via Testbook and Adda247; JE pay via Made Easy and BIPE; IBPS via Adda247 and Testbook; Delhi Labour Department competency-certificate page; MoRTH GSR 681(E); NATS stipend revision 2026 via Key4Comply; myScheme PMSC page; PMO PM-Vidyalaxmi release; NCVT ITI FAQ; Assam Skill Department notification 2025; DGT–NIOS FAQ (July 2026). URLs in the research pass output filed with this note’s commit message; open them before quoting.

3. What drives the demand for degrees [S]

Ranked by how well each driver is evidenced.

Government-job eligibility and the exam queue — strongest. Mangal (Journal of Development Economics, 2024; Ideas for India): a Tamil Nadu hiring freeze that cut 86% of state vacancies lowered male graduate employment by about 8 points as men stayed out of work to keep preparing, and a decade later the exposed cohorts earn less — the preparation built no human capital (https://www.sciencedirect.com/science/article/abs/pii/S0304387824000804). Azim Premji University’s 2023 public-recruitment report: 300 to over 1,000 applicants per vacancy, two to five years of preparation typical. Lokniti-CSDS youth surveys: 65% preferred a government job in 2016, 55% in 2021 (CSDS-KAS, Indian Youth: Aspirations and Vision for the Future, p. 15). An earlier version of this note cited “three in five in 2023” to Lokniti’s Social and Political Barometer 2023; that is an electoral survey with no job-preference question and the citation is withdrawn (14 Sep 2026). Glinskaya and Lokshin (World Bank, 2005): public-sector wage premium of 62–102% over formal private and 164–259% over informal work, on 1990s NSS data; no PLFS-era estimate surfaced (the warehouse’s own government-versus-private medians in HYPOTHESIS.md §1 are the nearest thing: ₹35,500 vs ₹20,000 for a general graduate at 25–54, 1.8×).
The degree as a screening device in a graduate glut — well evidenced descriptively. Graduate unemployment 17.2% (2019-20) falling to 11.2% (2023-24) in PLFS annual reports, against near zero for the illiterate; ILO India Employment Report 2024: youth graduate unemployment 29.1%, educated share of the unemployed 35% (2000) to 66% (2022). “Degrees of Devaluation” (arXiv, 2026): in districts where colleges expanded fastest after 2004, the graduate premium eroded about 9%, and about 34% more for SC/ST graduates, consistent with employers falling back on group priors as the signal weakens (https://arxiv.org/abs/2606.22620). No India-specific sheepskin-discontinuity estimate surfaced.
Marriage-market return, especially for women — good causal evidence. Adams-Prassl, QJE 2025: parents in rural India hold beliefs in a large marriage-market return to girls’ schooling, which raises schooling demand and delays marriage (https://academic.oup.com/qje/article/140/3/2107/8113989). 86% of college-educated women marry college-educated men; hypogamy rising (Lin, Desai and Chen, 2020). Dowry-versus-degree causal evidence: thin.
Status, respectability and masculinity — rich ethnography, qualitative. Jeffrey, Jeffery and Jeffery, Degrees Without Freedom? (2008) and Jeffrey, Timepass (2010): education as a “contradictory resource” for lower-middle-class young men in western Uttar Pradesh, who accumulate degrees and wait for government posts. Ørberg (2018) on Kota coaching as middle-class credential strategy.
Stigma of vocational and manual tracks, caste-linked — policy consensus, little household evidence. NITI Aayog’s 2026 skilling vision names “low trust and social stigma” and “degrees pursued for prestige or government exam eligibility” (https://www.niti.gov.in/sites/default/files/2026-08/Reimagining-Skilling-for-Viksit-Bharat-2047.pdf); the Sharada Prasad Committee (2016) proposed vocational colleges and a degree-awarding TVET university to counter the stigma. Oh (NBER w28462): workers forgo sizeable wages to avoid caste-inconsistent tasks.
Employer degree inflation for Class-12 jobs — plausible, anecdotal. 50,000 graduates among applicants for Uttar Pradesh police messenger posts; about 23% of food-delivery workers hold a degree (TISS). No academic study of the “any graduate” norm surfaced.

The warehouse corroborates drivers 1 and 2 directly: the waiting curve in §1.3 is the queue, and §1.2 shows what the queue resolves into — six in ten employed general graduates in work that needs no degree.

4. What an NCVT certificate is worth [S], read against §1.4 [W]

Evidence that it has real value.

Formal vocational training carries a 12–13.5% earnings premium in Heckman-corrected PLFS estimates (Venugopal, IES 2025; the Education + Training 2023 paper); full-time training specifically, not short courses (IZA DP 15002, 2022).
Large manufacturers gate campus hiring on NCVT/SCVT: job-portal listings show Maruti at ₹28,000–44,000 CTC, Hero at ₹20,000, Tata Motors taking 1,000 apprentices at ₹17,350 (unverified listings). Tata Technologies has an agreement to convert 71 Tamil Nadu government ITIs into technology centres.
The Uttarakhand tracer (2024) reports incomes 41% above first-job pay, at about ₹19,500 a month (self-reported).
The warehouse: double the formality at entry and a 54% pay premium by 35–44 (§1.4).

Evidence that it mainly feeds cheap labour.

DGT’s national tracer under STRIVE: 31.6% of pass-outs in wage employment, 7.9% self-employed, 9.5% in apprenticeship, 51% unemployed but available, and a quarter of the employed working outside their trade (self-reported; https://dgt.gov.in/sites/default/files/National%20Tracer%20Study%20-%20STRIVE.pdf). Puducherry’s tracer: 80% of private-ITI alumni unemployed.
Contract workers were 42% of organised manufacturing employment in 2023-24 (ASI), the highest in at least 25 years and double the share in 2000. Maruti is reported to have 17% of plant workers permanent, the rest temporary, apprentice, trainee or contract (activist sources).
NEEM, the trainee scheme, was withdrawn in December 2022 after AICTE warned facilitators against “re-routing existing employees”; Haryana NEEM stipends of about ₹9,000 sat against permanent wages of ₹80,000–90,000 in the same plants. Foxconn Chennai had trainees at ₹2,955 a month (IndustriALL, older).
NAPS stipends of ₹6,800–12,300 a month sit near unskilled minimum wages; 18 lakh apprentices a year. PMKVY placements run 7–41% of the certified depending on source (CAG 2025: 23.2 of 56.1 lakh, 41%; live dashboard August 2026: about 7%).
ILO’s efficiency study: 30–40% of state ITI completers employed or self-employed. Mehrotra: the skilling system as a “numbers game”. State of Working India 2026: ITIs up threefold since the 2010s, mostly private, quality falling.

Reading both columns together. The two are not in conflict; they describe the same labour market at two points. At entry an NCVT certificate is what gets a school-leaver onto a formal payroll — contract, provident fund, a Maruti or Hero line — at the same ₹15,000 the untrained earn, which is exactly what “cheap but formal labour” looks like from the employer’s side. The premium appears in the second decade, for the minority who stay in wage employment; the tracer studies’ 50% unemployed and the 42% contract share say most do not get that far. Whether the certificate “works” therefore depends on completion and on the first employer, which is why the Puducherry private-ITI result and the Maharashtra government-ITI occupancy figure in HANDOFF_PASS2.md point in opposite directions for the same credential.

5. What this does to the overall argument

The team’s framing: the state invests at the very top (IITs, NITs) and the very bottom (ITIs), not enough in polytechnics, so it multiplies short-term NCVT certificates whose value is doubtful.

The demand side is documented. Degree demand is driven first by government-job eligibility and the exam queue, second by the degree’s screening value in a glutted market, third by marriage and status returns. The warehouse shows the consequence: general graduates are the biggest bloc in government employment and the least likely group to hold a government job, six in ten work in jobs that need no degree, and a third are unemployed at 22–24.
The rules reward exactly the credential the state funds least. The polytechnic diploma is the only sub-degree credential that opens a Group B government post (Junior Engineer, Level 6) and the lateral route into a degree. It gets 100 slots in a ₹4,200 crore scheme; ITIs get ₹60,000 crore.
NCVT certificates are useful but not a substitute for a diploma. They buy formality at entry and pay later, for completers; they open Level 1–2 doors that are the most crowded in the country; and the system that issues them has half its seats empty and a 42% contract share waiting on the other side. “Of no use” is wrong; “industry looking for cheap labour” is half right and PLFS shows which half.
The open question is the middle. Nothing in either pass explains why a third of government engineering and polytechnic seats are empty when the credentials they issue are the ones the rules reward. Fees do not explain a government polytechnic at ₹8,000 a year. Location, quality perception, the lateral-entry drain and the pull of the general degree’s exam eligibility are the candidates, and none is measured here.

6. Queries

All on avantifellows.external_data_sources.plfs_fact_persons, release_id = 'calendar_2025', visit = 'V1', weight_annual IS NOT NULL. Segment CASEs as in sql/00_bucket_case.sql, with the ITI-like proxy voc = '1' AND voc_typ = '3' AND voc_dur IN ('3','4','5','6').

§1.1: ages 25–54; government job = pas = '31' AND etyp_pas IN ('05','06','07'); formal = job_pas IN ('2','3','4') OR ssec_pas NOT IN ('8','9').
§1.2: ages 25–34, employed (pas IN ('11','12','21','31','41','51')), SUBSTR(ocu_pas,1,1) for the NCO major group.
§1.3: ages 22–34 in three-year bands; unemployment rate = pas LIKE '8%' over pas NOT LIKE '9%'.
§1.4: tedu_lvl = '01' AND gedu_lvl IN ('08','10'), bands 25–29 and 35–44; manufacturing = ind_pas_div 10–33; medians exact weighted.