Guide · Bonds and fixed income · Updated 2026-09-23

The BHARAT Bond index selects credit by who owns the issuer, which is a different risk from the one its AAA describes

A common description of these indices, repeated on pages still being written in 2026, calls the bonds government-backed. The exchange's methodology says the opposite in one line of its eligibility rules: a bond backed or serviced by the Government of India cannot enter. What the government provides here is ownership, and ownership is a weaker promise than a guarantee. It can be sold, diluted or left unused, and it obliges the owner to nothing on a coupon date.

That difference decides what a holder is exposed to, and it can be measured: in the rules, in the rating agencies' own criteria, in the concentration the caps allow, and in what the universe has actually been paid over government bonds. How a target maturity index behaves as its date approaches is the subject of the guide to target maturity funds; this page is about the credit inside it.

Ownership draws the universe, and a guarantee disqualifies a bond

The methodology applies its tests in a fixed order, and the order is the argument. First the owner: a central public sector enterprise on the lists kept by the Department of Public Enterprises and DIPAM, a Maharatna, Navratna or Miniratna company, a public financial institution owned and managed by the Government of India, a statutory body created by an Act of Parliament, or another issuer DIPAM names. The Department of Public Enterprises counts as a CPSE any company in which the Centre holds more than half the equity, with that company's Indian subsidiaries, and leaves banks and insurers out, as the CAG's Report No. 7 of 2020 records. Ownership therefore passes down a chain: a company owned by another state-owned company qualifies without the government holding a share of it directly.

Second the instrument: a plain rupee bond with a fixed coupon and a fixed maturity, listed and rated, and none of tax free, floating, perpetual, partly paid, convertible, optioned or stepped. It must also not be backed or serviced by the Government of India. Such paper exists: the Ministry of Finance's status paper on government debt describes extra-budgetary resources raised by public sector undertakings through Government of India fully serviced bonds, whose principal and interest the government pays from the Union Budget. That is the public sector paper closest to a sovereign claim, and it is precisely what the rule keeps out.

Third the rating, where the lowest of the agencies' ratings must be AAA at every review. Fourth, size: more than ₹100 crore of eligible bonds maturing in the final twelve months. Only then come weights, by amount outstanding, with no issuer above 15 per cent at a review.

Weighting by amount outstanding hands the largest weights to the heaviest borrowers until the cap stops them. A debt weighted index is a register of who has borrowed most, and in this universe the largest borrowers are the state's own lenders to power, infrastructure and rural credit.

The AAA is partly the owner's rating

A domestic rating is an opinion on timely payment, and for a state-owned issuer it is built in two layers. One of the largest domestic rating agencies sets out the method in its criteria for factoring parent, group and government linkages, February 2025 edition: a standalone view of the entity, notched up for the probability and extent of distress support from the Government of India. Entities fall into four classes. Commercial companies with a limited policy role, whose default would cost the government little, get no uplift; those whose default would be politically or economically costly get an uplift sized to that cost. Policy institutions the government has propped up with equity and loans in the past get a limited one, because that support kept them solvent without necessarily keeping their debt service on time. Policy institutions whose failure would choke funding to sectors the government wants funded are equated with the government's own rating, AAA. What a rating does and does not forecast is in the guide to credit ratings.

The fund built on the April 2033 index states the consequence in its own risk factors. Its scheme information document, dated 28 November 2025, says most CPSE securities carry a “higher credit rating essentially due to government ownership and implied government support”, and that a change in government control or shareholding could bring a downgrade.

Set those two layers beside the index rule and the filters stop being independent. Step one selects owners; step three reads ratings that are, in large part, a reading of the same owner. The April 2030 index's seventeen issuers diversify seventeen standalone businesses, but they hold one assumption seventeen times: that the owner keeps standing behind companies it has chosen not to guarantee. The credit risk therefore does not fall with the issuer count the way it would across unrelated private companies. An owner that let one of them miss a payment would tell the market something about every other one on the same day. And because the rule reads only the final rating, an issuer that is AAA on its own balance sheet and one that is several notches lower before support look identical to it.

Four exits, each on its own clock

Ownership selects the universe, and changes of ownership are also how an issuer leaves it. The methodology names four routes out, and each runs on its own clock.

A downgrade below AAA removes the issuer within 30 calendar days under the August 2026 methodology, which cites SEBI's Master Circular for Mutual Funds of 27 June 2024. That clock has moved more than once, as the timeline shows. SEBI's 2019 norms, quoted in the exchange's 2019 white paper on the series, gave a fund five working days; the July 2020 methodology kept an issuer still rated investment grade until the next quarterly review; the April 2022 edition removed any issuer cut below AAA within five working days; and SEBI's circular of 23 May 2022 (SEBI/HO/IMD/DOF2/P/CIR/2022/69) gave funds 30 calendar days and let a bond cut below investment grade be segregated. The master circular the methodology cites was itself replaced from 1 April 2026 by SEBI's master circular of 20 March 2026, issued with the SEBI (Mutual Funds) Regulations 2026.

A privatisation removes an issuer at the next rebalancing once the disinvestment is completed, whatever its rating then. At the launch of each index, entities for which the government had given in-principle approval to sell its stake were left out altogether. This is the exit a rating does not see coming: an AAA issuer can be sold out of the index, and the fund tracking it must sell at whatever the market pays for bonds whose support has just changed hands.

A merger resets weights. When the equity shareholders of a constituent approve a scheme of arrangement, the combined issuer is held to one 15 per cent weight from the next rebalance. The case is live: after the Union Budget for 2026-27 proposed restructuring the public sector's non-banking finance companies, the boards of two state-owned power sector lenders, one the majority owner of the other, approved a scheme of amalgamation in June 2026, as reported at the time, with shareholder and tribunal approvals still to come. SEBI's 25 per cent group limit for debt index funds does not apply to public sector issuers, so a parent and its subsidiary can each sit at the cap; at launch this pair held 30.03 per cent of the April 2023 index. Once their shareholders approve, the pair's combined weight in every live index is held to 15 per cent from the next quarterly rebalance, and any excess is spread across the other issuers in proportion.

Falling below eight issuers brings in eligible issuers until there are eight again, the minimum SEBI sets for a debt index. The April 2031 index launched in July 2020 exactly at that floor, with 21 bonds from 8 issuers, according to the exchange's constituent file for that month.

What a 15 per cent cap permits

A cap on the largest weight is not a floor under diversification. Six issuers fit at 15 per cent, leaving 10 per cent for everyone else, and the most concentrated index the rules allow puts that 10 per cent in a seventh issuer and slivers in the rest to reach the minimum count. Its Herfindahl index, the sum of the squared percentage weights, is 1,450, an effective count of 6.9 issuers however many names the constituent list carries. Eight equal issuers would score 1,250, an effective eight.

Two limits that bind most debt index funds do not bind this one. SEBI caps a debt index at 25 per cent in any one sector, but the limit excludes government securities, Treasury bills, state loans and AAA securities of public sector undertakings, public financial institutions and public sector banks; its 25 per cent group limit excludes public sector issuers entirely. Both exclusions are in the May 2022 circular and were carried into the June 2024 master circular. An index of state-owned issuers can therefore be as concentrated by sector as its borrowers are.

The launch weights show what that means in practice. In the exchange's own 2019 white paper the April 2030 index began with 12 issuers, 3 of them at the cap and 44.97 per cent in the top three. Power generation and transmission alone took 38.43 per cent, above the 25 per cent sector limit an ordinary debt index would face, and 57.72 per cent once the lenders to power are added. The April 2023 index began with 73.73 per cent in lenders, one sector on any classification, and 30.03 per cent in one parent and its subsidiary. On 31 August 2026 the April 2030 index held 99 bonds from 17 issuers, according to the exchange's constituent file, which lists no weights; the arithmetic above is what the rules allow those 17 to look like.

One issuer's trouble, and the trouble no cap reaches

The cap fixes the arithmetic of a single failure. It cannot diversify the one assumption every constituent shares. The table prices both on stated bonds: a par bond with a 7 per cent coupon at a 7 per cent yield, paying half-yearly and maturing half a year before each index's date, which gives a modified duration of 2.66 for the April 2030 index and 4.83 for the April 2033 index as at 18 September 2026. The spread moves and the recovery rate are illustrative inputs, not forecasts.

The single-issuer rows stay small until they are not. A cut to AA+ with a 50 basis point widening costs the April 2030 index 0.20 per cent. A default at the cap with half the money recovered costs 7.50 per cent, slightly more than a full year of the 7.25 per cent a year that index has returned since its base date. The pair row is the group exemption at work: two related issuers can fail together at twice the cap. The last row is the one the cap cannot touch. If the market reprices the support assumption for every issuer at once, a 100 basis point widening costs the whole duration, 4.69 per cent on the April 2033 index, and seventeen issuers help not at all.

Timing matters as much as size. The index has 30 calendar days to drop a downgraded issuer, and a fund tracking it has the same 30 days to sell into a market that knows it must. A bond cut below investment grade can instead be moved into a segregated portfolio, the mechanism that stops early sellers leaving the loss with everyone slower; the scheme document for the April 2033 fund records a written policy for it, approved by the trustees.

What the universe was paid over government bonds, measured

The fair test of a credit index is what it returned over the government bonds a holder could have bought instead, over the same dates. The exchange's daily files carry the four target maturity indices only from 16 January 2025, but each starts at a published base value of 1000 on a published base date, and the government indices are in the files on each of those dates. Checked before use, the base value reproduces the exchange's own since-inception return on 31 August 2026 for all four indices to within 0.01 percentage points. Two government total return indices bracket each index's maturity path: the 4 to 8 year index, built from the three most traded government bonds with four to eight years left, and the 8 to 13 year index.

The answer depends on the start date, and the sign flips. The April 2030 index, from November 2019, is ahead of both government indices, by 0.66 and 1.16 percentage points a year; the April 2031 index, from June 2020, by 0.42 and 0.86. The April 2032 index is 0.13 points behind the 4 to 8 year index and 0.15 ahead of the 8 to 13 year one. The April 2033 index, from November 2022, trails both, by 0.58 and 0.47 points a year. At every quarterly review every constituent of every index was AAA from every agency that rated it, because the rule admits nothing else.

Over a holding period the extra return of a credit index is, to a first approximation, the yield spread it was bought at less the price effect of any widening since. The files carry index levels, not yields, so the two parts cannot be separated here; the one entry yield on record is 7.66 per cent for the April 2030 index at launch, with a Macaulay duration of 6.91 years, in the 2019 white paper. The pattern across the four is what a spread that narrowed after 2019 and 2020 and widened after 2022 would produce, and the fund's own scheme document names one cause of widening: heavier issuance by some CPSEs has pushed their spreads out relative to others. The rating test did not change and neither did the ownership test. The spread did.

The daily window adds volatility and sensitivity, over the 415 sessions from 16 January 2025 to 18 September 2026.

Matched by measured sensitivity rather than by label, the four indices earned between -0.38 and +0.10 percentage points a year over their nearest government index in this window, and two of the four earned less. For twenty months the universe was paid roughly nothing for its credit and its illiquidity.

A valuation is calmer than a trade

On daily moves every target maturity index looks calmer than the 4 to 8 year government index: 2.25 per cent a year for the April 2033 index against 2.32. On 21-session moves the order reverses for the two longer indices, 3.74 and 3.88 per cent against 3.12 and 3.54. Sensitivity to the government index rises the same way, from between 0.39 and 0.48 over one session to between 0.87 and 1.15 over 21 sessions, where the April 2033 index moves with the government curve as much as the 8 to 13 year index does. With 19 non-overlapping blocks, each 21-session volatility carries a sampling error of about a sixth of its value: the reversal is a direction, not a precise size.

The cause is in how the index is priced. End-of-day values of every fixed income index in the family come from valuations supplied by NSE Data and Analytics, the exchange group's data company, and most corporate bonds are valued each day rather than traded. A valuation absorbs news in steps, which shows as positive correlation between one session's move and the next: 0.20 to 0.27 for the target maturity indices, against 0.00 for the 8 to 13 year government index. Daily volatility and single-session sensitivity therefore understate the credit index's rate risk; a holding sized on daily figures moves more on the monthly statement than the daily numbers promised. The guide to target maturity funds measures how that sensitivity falls as each index nears its date.

Two wrappers around the same bonds

Each live index has an exchange traded fund that tracks it and a fund of funds that holds that ETF; on the index maturity date the ETF terminates and pays out its net assets. The scheme information document of the ETF on the April 2033 index, dated 28 November 2025, sets out the mechanics. Units trade on NSE and BSE in lots of one. Only authorised participants (from ₹25 lakh), the fund of funds (from ₹1 crore) and large investors (from ₹25 crore since 1 November 2022) can create or redeem units with the fund, in cash or in kind; everyone else trades on the exchange, at a price that can sit away from net asset value, against at least two market makers. The fund samples the index rather than holding every bond. SEBI's May 2022 circular caps a debt ETF's annualised tracking difference, averaged over a year, at 1.25 per cent, and its categorisation circular of 26 February 2026 requires an index fund or ETF to keep at least 95 per cent of its assets in its index's securities, and a fund of funds 95 per cent in its underlying fund.

The fund of funds exists for the investor without a demat account: it buys the ETF and adds a small layer of its own costs. Cost has changed since launch too. The ETF's expense ratio was fixed at 0.0005 per cent in the fund house's December 2018 bid to the government; by a letter received on 4 December 2024 the government let it follow SEBI's regulations instead, with the management fee at zero but operational and statutory expenses chargeable. The current figure is in each scheme's own disclosures.

The tax case the index was launched on no longer exists

The exchange's 2019 white paper sold the structure partly on tax: its worked example put the ETF at 6.94 per cent a year after tax against 5.74 per cent for a deposit taxed at 30 per cent, a 120 basis point gap that came entirely from indexation of long-term gains taxed at 20 per cent. None of it applies to money invested now.

Section 50AA of the Income-tax Act 1961, inserted by the Finance Act 2023, deems a gain on units of a specified mutual fund acquired on or after 1 April 2023 to be short term whatever the holding period, so it is taxed at the holder's slab rate. The Finance (No. 2) Act 2024 rewrote the definition from assessment year 2026-27 (clause 21 of the Bill, as the government's memorandum explains it): a specified mutual fund is now one that invests more than 65 per cent of its proceeds in debt and money market instruments, or a fund that invests 65 per cent or more in units of such a fund. The ETF sits inside the first limb and its fund of funds inside the second, and the Income-tax Act 2025 carried the rule across as its section 76 from 1 April 2026. Units bought before 1 April 2023 are outside it: they turn long term after 12 months for the listed ETF and 24 months for the unlisted fund of funds, taxed at 12.5 per cent without indexation on transfers from 23 July 2024. Every debt fund route is compared in how debt funds are taxed after the change.

What survives is deferral. The index reinvests every coupon, and the scheme document leaves any income distribution to the trustees' discretion; with none paid, a holder is taxed only on selling or on the payout at maturity. The same bonds held directly pay coupons taxed at slab rates every year, but a listed bond held more than 12 months turns long term at 12.5 per cent, because section 50AA does not reach listed bonds; the guide to buying government bonds directly works through the sovereign version of that route.

What the index is for

Read by its rules, the index is a low-cost, transparent way to hold a ladder of state-owned credit to a known date, weighted by who borrowed most and capped by issuer but not by sector. Its credit risk is one policy assumption held across all its issuers; its measured premium over government bonds has run from +1.16 to -0.58 percentage points a year depending on when an index started; and its daily prices understate how far it moves over a month. None of that is visible in the letters AAA.

Reading a methodology as a set of selection rules, then measuring what the rules delivered against the plain alternative, is method rather than an opinion about a product, and it is the way fixed income is taught here. The pricing underneath every figure on this page is in the guide to bond price and yield and the yield curve read as a price list.