Three disclosed channels finance the outbound-student cohort: scheduled commercial banks (RBI sectoral aggregate), education-focused NBFCs (Avanse + Credila as the named disclosed universe), and informal household-collateral channels (NABARD + NSSO + IOM). Three publishers, three measurement universes. Each true under its own definition. Decomposed here, not adjudicated.
Avanse Financial Services disclosed total FY25 disbursement of ₹6,914 cr with overseas-education segment disbursement of ₹5,152 cr. Credila Financial Services disclosed AUM of ₹41,810 cr as on 31 March 2025. Together they account for the named-NBFC disclosed universe; the rest of the money trail is decomposed in the next layers.
Roughly 13.35 lakh Indian students were enrolled overseas in 2024 per MEA's Lok Sabha reply. Three disclosed channels finance that cohort: SCBs (RBI Sectoral Deployment captures the aggregate inside personal loans); education-focused NBFCs (CRISIL: sector AUM ≈₹43,000 cr end-FY24, forecast ₹60,000 cr FY25); and informal-collateral channels (NABARD NAFIS + NSSO AIDIS capture rural debt in aggregate but not by education-purpose × land mortgage).
Three publishers, three measurement universes. Each true under its own definition. Decomposed here, not adjudicated. This brief redrafts an earlier version that was scrapped 2026-05-25 after 3-AI review caught fabricated multi-year figures (Avanse disbursement overstated by approximately 5x versus verified annual results).
We distinguish five publisher-defined universes. Each captures a different slice of the same money trail; none captures the full cross-tab.
(a) RBI Sectoral Deployment — monthly, all SCBs, education inside personal loans. (b) Named-NBFC disclosed segments — Avanse FY25 audited financials + Credila ICRA/CARE rationales. (c) PSB NPA — Ministry of Finance via PIB, two endpoints only. (d) Outward-remittance LRS — RBI bulletin, "studies abroad" head separately reported. (e) Rural debt surveys — NAFIS 2021-22 and AIDIS NSS 77th round, neither with an education-by-land-collateral cross-tab.
The IOM India Data Needs Assessment (2024) documents structurally that no single Indian instrument captures financing channel of outbound students. The five publishers we cite are therefore additive, not interchangeable.
The RBI Sectoral Deployment of Bank Credit return is monthly. On the NBFC side, two named lenders dominate the disclosed universe: Avanse and Credila. The CRISIL sector forecast for FY25 frames where the NBFC book lands in aggregate.
The most recent RBI Sectoral Deployment release covers March 2026 (released 30 April 2026) and reports the personal-loans segment growing 16.2% YoY. Education loans sit within personal loans; the headline press release does not break out the education-loan ₹ figure, which lives in the detailed sectoral tables released alongside. A secondary source (gradsloan.com) attributes to RBI that the SCB education-loan outstanding crossed ₹1 lakh crore in 2024-25 — flagged as secondary-attribution pending direct primary verification.
On the NBFC side, Avanse Financial Services — IPO-bound, DRHP filed late-June 2024, returned by SEBI on technical grounds end-July, addendum filed October 2024 — disclosed in its FY25 results: total AUM ₹18,985 cr vs ₹13,303 cr FY24 (1.4x YoY); total disbursement ₹6,914 cr vs ₹6,335 cr FY24 (+9%); overseas-education segment disbursement ₹5,152 cr with segment AUM ₹15,275 cr; net profit ₹502 cr vs ₹342.4 cr (+46.6%); revenue from operations ₹2,347 cr (+36%).
Credila Financial Services (formerly HDFC Credila) — acquired June 2023 by a BPEA EQT-led consortium with ChrysCapital at INR 103.5 bn (~USD 1.26 bn) pre-money for 90%; HDFC retained 9.99%, consortium injected ₹2,000 cr primary capital — disclosed AUM ₹41,810 cr as on 31 March 2025 (+48% YoY) and overseas-education segment disbursement of ₹14,089 cr FY24 plus ₹12,408 cr in 9MFY25.
The two named NBFCs together account for approximately ₹61,000 cr combined AUM at end-FY25 — broadly congruent with the CRISIL FY25 sector forecast of ₹60,000 cr.
Avanse FY25 disbursement and FY24 disbursement are flow measures (annual). Credila ₹41,810 cr is a stock measure (AUM on a single date). Plotted on a common ₹-crore axis for visual scale reference only — not implying like-for-like comparability. Source: BRIEF Panel A.
Ministry of Finance via PIB: MoS Finance Pankaj Chaudhary, in a written reply to the Lok Sabha dated 15 December 2025, disclosed that gross NPA on the PSB education-loan portfolio reduced from approximately 7% (FY 2020-21) to approximately 2% (FY 2024-25). Only the two endpoints are published; no intermediate-year figure is in the release. We restrict the reading to what the publisher disclosed.
CRISIL Ratings (3 September 2024 sector note) discloses 90-days-past-due figures on a comparable book: ~0.2% NBFC education loans, ~2.0% private-sector banks, ~3.9% public-sector banks. CRISIL also discloses approximately 90% of the NBFC education-loan book is under moratorium and prepayment/foreclosure rates in moratorium are 35-45%. The CRISIL ~3.9% PSB figure (90+ DPD on the NBFC-comparable, overseas-skew book) and the PIB ~2% figure (gross NPA on ALL PSB education loans) measure different universes and are not directly comparable. The reconciliation gap is real and not bridged by a single publisher.
Two disclosed endpoints; intermediate years not in source. Top panel: PIB / Ministry of Finance disclosure (MoS Chaudhary, Lok Sabha 15 December 2025) covers ALL PSB education loans, gross NPA basis. Bottom panel: CRISIL Ratings 3 September 2024 covers the NBFC-comparable book (overseas-skew), 90+ DPD basis. The two universes are not directly comparable; both are reported as published.
Disclosed repayment-product structure (per Avanse and Credila product documentation in the rating rationales): moratorium = course duration + 6-12 months, then 10-15 year bank tenors and 12-15 year NBFC tenors. PM Vidyalaxmi (Union Cabinet 6 November 2024; budget ₹3,600 cr for FY25-FY31) adds 3% interest subvention during moratorium on loans up to ₹10 lakh for families with income ≤ ₹8 lakh, plus 75% credit-guarantee cover on outstanding default up to ₹7.5 lakh — limited to a defined "quality-higher-education-institution" list and not to overseas study.
The NABARD NAFIS 2021-22 (released 9 October 2024; one lakh rural households, 28 states + J&K + Ladakh) discloses outstanding-debt incidence rising from 47.4% (2016-17) to 52.0% (2021-22); among agricultural households, institutional-only borrowing rose from 60.5% to 75.5%; "domestic needs" was the single most popular use case at 29% (Dvara on NAFIS). The NSSO AIDIS (NSS 77th round, Jan-Dec 2019; reference date 30.6.2018) is the most recent comprehensive AIDIS round and the canonical baseline for household-debt-by-purpose; the next decadal round is not yet published.
Critical caveat — disclosure gap, not analytical: neither NAFIS 2021-22 nor AIDIS 2019 discloses an education-purpose borrowing sub-category cross-tabulated with agricultural-land collateralisation. The prior draft inferred a rupee figure here; this redraft does not. The IOM India Data Needs Assessment (final 04-03-2024, 48 pp) documents structurally that India's emigrant-tracking infrastructure (e-Migrate: 128,817 ECR clearances in 2021) covers low-skilled ECR workers; outbound students and skilled emigrants are captured only through scattered visa, passport and remittance data — no instrument discloses financing channel by destination cohort.
Disclosure gap: no publisher discloses education-purpose × land-mortgage cross-tab. NAFIS aggregates capture the rising debt-incidence and rising institutional-only share among agricultural households, but no rupee figure tied specifically to education-purpose borrowing collateralised by agricultural land is published. The IOM (2024) data-needs assessment documents this gap structurally.
What IS disclosed at cohort level: 13.35 lakh Indian students overseas in 2024 per MEA — top five Canada 427,000 · USA 337,630 · UK 185,000 · Australia 122,202 · Germany 42,997 — and RBI LRS "studies abroad" head USD 2.92 bn FY25 (down from USD 3.48 bn FY24, -16% YoY), approximately ₹24,500-25,000 cr at closing rates. This is the visible household-level outflow; the share funded by bank loans, NBFC loans, and liquidated household assets is not separately disclosed.
Four disclosed money flows finance the education-migration cohort: (1) formal bank — RBI Sectoral Deployment inside personal loans, secondary attribution reports the SCB book crossed ₹1 lakh crore in 2024-25; (2) formal NBFC — Avanse + Credila ≈₹61,000 cr combined AUM end-FY25, CRISIL sector forecast ≈₹60,000 cr FY25; (3) LRS "studies abroad" head — USD 2.92 bn FY25 (-16% YoY) per RBI bulletin; (4) informal / household-asset — present in NAFIS 2021-22 and AIDIS 2019 aggregates but NOT disclosed at the education-by-land-collateral cross-tab level by any publisher.
NPA direction on the PIB-disclosed PSB segment is improving FY21-FY25 (two endpoints only); CRISIL discloses a higher PSB 90+ DPD on its NBFC-comparable book — different universes, not reconcilable from public data alone. Named-NBFC books are reliable to the publisher-disclosed rupee. The informal-collateral channel is real and large in NAFIS aggregate, but its education-specific decomposition is a publisher-side disclosure gap, not analytical absence.
| Channel | Publisher | What is disclosed |
|---|---|---|
| (1) Formal bank · SCB book | RBI Sectoral Deployment (primary); gradsloan.com (secondary) | Personal-loans segment +16.2% YoY in March 2026; secondary attribution reports SCB education-loan outstanding crossed ₹1 lakh crore in 2024-25 |
| (2) Formal NBFC · Avanse + Credila | Avanse FY25 results; ICRA + CARE Credila rationales; CRISIL sector note | Avanse ₹18,985 cr AUM + Credila ₹41,810 cr AUM ≈ ₹61,000 cr combined end-FY25; CRISIL FY25 sector forecast ≈ ₹60,000 cr |
| (3) LRS · studies abroad head | RBI monthly bulletin (primary); Business Standard / NextIAS (secondary) | FY25 USD 2.92 bn (down from USD 3.48 bn FY24, -16% YoY); ≈ ₹24,500-25,000 cr at closing rates |
| (4) Informal / household-asset | NABARD NAFIS 2021-22; NSSO AIDIS 77th round; IOM India Data Needs Assessment 2024 | Aggregate rural-debt rise (47.4% → 52.0% incidence; 60.5% → 75.5% agricultural institutional-only) — but NO education-by-land-collateral cross-tab disclosed by any publisher |
The visible stack — RBI aggregates + named NBFCs + LRS — is publisher-anchored and quantifiable; the household-asset portion remains a structural gap in the Indian statistical instrument set as IOM (2024) documents. Each layer is true under its own publisher's definition.
The publishable discipline here is the same one Mirror Brief No. 01 applied to office-leasing definitions: report each publisher's number under that publisher's definition, mark the universe each one measures, and do not infer a rupee figure where no publisher discloses one. The prior draft of this brief inferred such figures and was scrapped. This redraft does not.
Every quantitative figure in the body prose and SVG annotations of this brief traces to a verbatim publisher disclosure quote in the companion file RECEIPTS.md (R1-R18). The sources below are listed in BRIEF.md citation order.
This Mirror Brief decomposes a single descriptive question — how Indian students overseas are financed — by stacking five publisher-defined universes against each other. The window for asset-quality, NBFC-disclosed AUM, and LRS is FY24-FY25 disclosed quarters; for the rural-debt baseline, NAFIS 2021-22 and AIDIS 77th round (reference 30.6.2018) are used. Geographic scope is India outbound; the cohort universe is the 13.35 lakh MEA-disclosed student count for 2024.
Every quantitative figure is anchored to a verbatim publisher quote in the companion RECEIPTS.md (R1-R18). Where a primary URL was unreachable during the 2026-05-25 verification cycle (RBI Bulletin, MEA PDF, PIB PRID), a secondary publisher carrying the same figure with the same attribution is cited inline. Where a figure is secondary-source-only (R12 SCB ₹1 lakh crore milestone, R13 LRS data via Business Standard / NextIAS), the body prose explicitly flags the secondary attribution.
The prior draft of this brief was scrapped 2026-05-25 after 3-AI review (Codex GPT 5.5 + Gemini Pro 2.5 + Reality Checker). The fabrications caught and removed: (i) Avanse FY-wise disbursement of ₹39,630 cr / ₹30,130 cr / ₹20,310 cr (overstated approximately 5x versus the verified annual ₹6,914 cr); (ii) a derived two-year CAGR built on those overstated figures; (iii) an SBI Global Ed-Vantage ₹15,247 cr / 52,000-student annual line that could not be verified against SBI's FY25 Integrated Annual Report; (iv) intermediate FY22 / FY23 / FY24 PSB NPA values for the PIB chart — only the two endpoints are disclosed by the publisher; (v) an inferred rupee figure for education-purpose-by-land-collateral borrowing for which no publisher discloses a cross-tab.
The three inline SVGs in this brief use only numbers that appear verbatim in BRIEF.md, restricted to the Panel under which they appear. Figure A plots Avanse FY24 and FY25 disbursement, the FY25 overseas-segment slice, and Credila end-FY25 AUM (with a "stock, not flow" tag). Figure B plots only the two PIB-disclosed PSB NPA endpoints (FY21 7% and FY25 2%) and the CRISIL Sep-2024 90+ DPD comparator strip (0.2% / 2.0% / 3.9%) — explicitly NO intermediate years for the PSB series and no smoothing. Figure C plots the two NAFIS time-points (47.4% → 52.0% rural debt incidence; 60.5% → 75.5% agricultural institutional-only) with a caption that the education-by-land cross-tab is not disclosed.
This brief is analytical commentary on publicly disclosed regulatory, rating-agency, and government-statistical material. It does not allege inaccuracy in any cited publisher, does not advocate any course of action by households, lenders, regulators, or universities, and does not forecast future education-loan volumes or outbound-student counts. The Indian Statistical instrument set does not currently disclose the education-by-land-collateral cross-tab; the IOM 2024 Data Needs Assessment documents this structurally. The Mirror Brief decomposes what is published; what is not published, it explicitly does not infer.