Ask an Indian investor for a safe, higher-yielding alternative to a fixed deposit and you will hear about corporate bonds, debt funds, or small savings schemes. You will almost never hear about State Development Loans, which are auctioned by the Reserve Bank of India, settle exactly like central government securities, and pay a measurable premium over them.
What they are
An SDL is a borrowing of an Indian state government — Maharashtra, Tamil Nadu, Uttar Pradesh and the rest. The RBI conducts the auctions on the same platform it uses for central government securities, on a published calendar, and the paper settles through the same infrastructure.
They are not centrally guaranteed. That is the technicality the spread is paid for, and it is worth being precise about it rather than waving it away.
What backs an SDL is the state’s own revenue, plus a consent mechanism: states borrow under Article 293 of the Constitution, and a state that owes the centre money needs the centre’s consent to borrow further. The RBI also administers state overdrafts and ways-and-means advances, so a state under cash stress is visible and managed well before it reaches an interest payment date. No SDL has defaulted.
So the honest framing is not “risk-free” and not “credit risk” either. It is near-sovereign paper with a legal structure that is one step removed from the centre.
What the spread actually is
Measured across our own data on a recent trading day, average spread over the central government par curve at matched maturity:
| Segment | Avg spread |
|---|---|
| Central G-Sec | ≈ 0 bps |
| STRIPS | 18 bps |
| SDL | 44 bps |
| UDAY (state discom) | 49 bps |
| Corporate | 253 bps |
That ordering is the whole story. G-Secs price on the curve by definition. Corporate paper pays 250+ for genuine default risk. SDLs sit at 44 — far too little to be compensation for meaningful default probability, and far too much to be nothing.
The two things it is actually paying for:
Illiquidity. SDLs trade much less than the central benchmark. A benchmark 10-year G-Sec can print thousands of times in a day; a given SDL may not print for weeks. If you need to sell before maturity, that is your problem, and the spread is your advance compensation for it.
Fragmentation. Every state issues repeatedly across tenors, so the SDL universe is enormous and no single line is a benchmark. Our data carries daily valuations for 6,578 distinct SDL ISINs, against 185 for the entire central G-Sec segment. Thirty-five times the instrument count, a fraction of the liquidity per line.
Why this is newly reachable
Historically an SDL was an institutional instrument — banks holding them for SLR, insurers matching long liabilities, provident funds. A retail investor’s only access was through a gilt fund, paying a management fee for the privilege.
RBI Retail Direct, launched in 2021, changed the mechanics. An individual can open an account directly with the RBI, bid in the primary auctions non-competitively, and hold government securities — including SDLs — in an RBI-maintained account, with no broker and no fund wrapper.
That does not make SDLs liquid. It makes the spread accessible, which is a different and smaller claim, and the right one.
Building a ladder on them
The template selects on spread rather than on any stored category label, and there is a specific reason.
Our security master carries a bond_category field populated from the exchange’s own instrument taxonomy. Checked against the valuation publisher’s segment classification, they disagree — around 835 government STRIPS are labelled “corporate” in the exchange taxonomy. Selecting on a label that mislabels government paper as credit is exactly how a “safe” bond screen ends up holding something else.
Spread does not have that problem, because it is measured rather than asserted. A bond yielding between 15 and 90 basis points over the sovereign curve is near-sovereign paper, whatever anyone filed it as. That band is the selection, and it is economically meaningful in a way a category string is not.
Three other filters do necessary work:
At least three years to maturity. A bond with months to run has almost no duration, so its spread stops being a yield decision and becomes a rounding error. It also matures out of the portfolio immediately, creating turnover with no thesis.
A current valuation. FBIL publishes daily valuations for government segments — the mark debt mutual funds are required to use. It exists on days the bond did not trade, which is what makes a continuous curve possible at all.
Quarterly rebalancing. The tradeable set changes slowly and every trade crosses a wide spread. Rebalancing an illiquid bond portfolio monthly is a cost programme.
Two limits worth stating plainly
The history is short. Daily SDL valuations in our data begin in February 2023. That is roughly three and a half years — one rate cycle at best. Any backtest over it is a plausibility check, not evidence. Treat a Sharpe ratio computed on this window with real suspicion.
You are being paid not to need the money. The spread is compensation for illiquidity. If you intend to hold to maturity, you collect it and the illiquidity costs you nothing — which is the correct way to own an SDL. If you intend to trade it, you will pay that spread back on the way out, possibly twice. This is a hold-to-maturity instrument that a backtest will happily pretend is a trading vehicle.
There is also a marking subtlety worth understanding. Because most SDLs do not trade on most days, the price series is largely made of published valuations rather than transactions. That is correct for marking a portfolio and wrong for assuming you could have transacted. Every bond row in our data carries a price_is_traded flag for exactly this reason, and the screens use it — a bond strategy that ignores the distinction is measuring a mark, not a trade.
Try it
The State Loan Spread Ladder template holds the 15 widest spreads inside the near-sovereign band with at least three years to maturity, equally weighted, rebalanced quarterly.
Run it against the G-Sec Liquidity Ladder over the same window. Same sovereign-class credit, different point on the liquidity curve — and the gap between the two return streams is, almost exactly, what the market charges for being able to sell in a hurry.
Further reading
- G-Secs are two markets, and only one of them is liquid on why liquidity concentrates in a handful of benchmark issues
- Corporate bonds in India: the yield is real, so is the liquidity trap for the far end of the same spread spectrum