When Indian Bonds Went Wrong: IL&FS, DHFL, Franklin and What They Cost Investors
Nothing teaches bond investing like the times it failed. India’s 2018–2021 credit cycle produced a full syllabus: AAA names defaulting, “safe” funds freezing, bank bonds written to zero while equity survived. If you invest in anything beyond sovereigns, this history is your risk disclosure. Worth reading once, properly.
IL&FS (September 2018): the AAA that wasn’t
Infrastructure Leasing & Financial Services was a systemically woven infra lender with more than 300 group entities. It defaulted on obligations against ₹90,000+ crore of group debt, weeks after carrying top-grade ratings. The default froze money markets, triggered the NBFC funding crisis, and forced a government-appointed board. Resolution has run for years, and recoveries vary widely by entity and seniority.
What it teaches: ratings lag reality, and they lag catastrophically when disclosure is poor. Complex group structures hide where the money actually sits. And “systemically important” describes contagion potential, not safety.
DHFL (2019): housing finance, AAA to D inside a year
Dewan Housing was a top-3 housing financier. It went from AAA in 2018 to default in mid-2019 amid a funding freeze and fraud findings, then became India’s first large financial company resolved under IBC, with Piramal acquiring it in 2021.
What you got back depended almost entirely on how much you held. Secured NCD holders with less than ₹2 lakh in a series recovered their full principal, though not the interest owed. From ₹2 lakh to ₹10 lakh the recovery was 43.82%, paid in cash. Above ₹10 lakh it came partly as cash and partly as Piramal debentures. Fixed deposit holders, who ranked below the secured NCDs, received about 23%, some ₹1,241 crore against their claims. The whole process took two years.
Four things follow. Retail-friendly public NCDs aren’t retail-safe by construction, so the issuer’s balance sheet is what matters. “Secured” means you recover something, eventually, rather than everything, immediately. A 9.x% coupon from a lender whose own borrowing costs are spiking is the market warning you in plain sight. And the resolution deliberately protected the smallest holders, so someone with ₹1.8 lakh in one series came out whole on principal while someone with ₹18 lakh took a haircut. That’s an argument about the size of a single claim, separate from the usual one about diversification.
Yes Bank AT1 (March 2020): the bonds junior to equity
₹8,415 crore of Additional Tier 1 bonds were written to zero in the RBI-led rescue, while equity holders kept their (diluted) shares. Much of it had been sold to retail investors as “better FDs.” The litigation reached the Supreme Court, where judgment was reserved in February 2026 and remains pending.
What it teaches: instrument structure can dominate issuer quality. Regulatory capital exists to absorb losses, and the losses it absorbs are yours. Any pitch that translates a complex instrument as “like an FD” is the moment to leave the room.
Franklin Templeton (April 2020): when the fund is the risk
Franklin wound up six debt schemes holding ₹25,000+ crore. Not because everything in them defaulted, but because the funds held illiquid lower-grade paper against a daily-redemption promise, and COVID redemptions broke the match. The Supreme Court appointed SBI Mutual Fund as liquidator, and by the time the asset sale finished in August 2023 the six schemes had paid out ₹27,548 crore, between 107% and 113% of what each was worth on the day it froze.
That reads like a happy ending and wasn’t one. Recovering 109% of an April 2020 valuation across three years is a thin return on money you could not touch, and none of it was certain while it was happening.
What it teaches: a fund’s liquidity promise is only as good as its portfolio’s liquidity. Yield-topping debt funds are yield-topping for a reason. And spreading money across schemes isn’t diversification when they share a style.
The supporting cast
SREI (2021) took twin infra lenders into insolvency. Reliance Capital and the wider ADAG group defaulted through 2019–21. Amtek Auto (2015) was the earlier warning nobody generalised from. And a run of co-operative bank failures left DICGC’s ₹5 lakh as the only floor that held.
What the history teaches
- Position sizing is the only defence that always works. Every episode punished concentration hardest. The 5%-per-issuer cap isn’t conservatism for its own sake; it’s the empirical lesson of 2018–21.
- Credit risk arrives suddenly after accumulating slowly. These issuers deteriorated for years while yields politely widened. The default itself took days. Watching spreads beats watching ratings.
- Recovery is slow even when it comes. DHFL took about two years, Franklin’s distributions ran from 2021 to 2023, IL&FS is measured in half-decades, and Yes Bank AT1 is in its sixth year of litigation. Money you might need has no business earning credit spread.
- The sovereign core exists for a reason. Through every episode above, G-secs, SDLs and T-bills paid on time. The 150–300 bps that corporate paper offers over them is the insurance premium for everything on this page.
- Panic is expensive and preparation is cheap. Franklin investors who sold side-pocket units at panic discounts lost real money on paper that ultimately paid. The alternative to panic isn’t courage. It’s sizing (rule 1) that makes panic unnecessary.
FAQ
Has a G-sec or SDL ever defaulted? No missed payments in the modern era. The SDL mechanics explain why the state record has held.
Did rating agencies get punished? SEBI tightened norms around rating-action timelines, liquidity disclosures and withdrawal rules. The structural incentive, issuer pays, is still there.
Is the system safer now? Meaningfully so: SEBI’s fund liquidity buffers and side-pocketing rules, NBFC regulation after IL&FS, and the OBPP framework for retail bond sales. Safer plumbing, same physics. Credit risk didn’t retire, it re-priced.
Educational content, not investment advice. Tax rules current for FY 2026-27 to the best of our knowledge, but verify with a professional before acting. See the full disclaimer.