Twenty-four years of India's off-budget borrowing, by instrument and government: from Vajpayee's oil pool bonds to the Modi-era FCI-NSSF and NHAI stack.
Off the books
India's hidden fiscal stack, 2002–2026.
The BJP campaigned in 2014 against the UPA's oil bonds — and the UPA had used both oil and fertilizer bonds, peaking at roughly ₹1.88 lakh crore in FY10. The argument that the principal would burden a future government was correct. What followed wasn't a clean-up, it was a larger, more diversified version of the same playbook: FCI-NSSF, NHAI bonds, IRFC, fertilizer arrears, bank recap, Air India holding co., Sovereign Gold Bonds. The total off-budget exposure peaked at ~₹13.6 lakh crore in FY21 — about seven times the UPA peak. The CAG flagged the disclosure gap year after year. In FY22 the government brought ~₹4.27 lakh crore of FCI's NSSF liabilities onto the books, but ~₹9.5 lakh crore is still off the books today — the cleanup absorbed FCI, rotated everything else.
Vajpayee NDA
2002
₹9,000 cr
oil pool deficit bonds — first instance
Issued via RBI to liquidate 80% of the oil pool deficit. The mechanism that subsequent governments would scale.
UPA-1 + UPA-2
2005–2010
₹1.88 lakh cr
oil + fertilizer bonds, peak FY10
Oil bonds (~₹1.48 L cr) plus fertilizer bonds (~₹0.40 L cr), occasionally also FCI. All cleared by March 2026 with ~₹3.5 L cr total outflow including interest. FM Sitharaman's own figure: ₹1.9 L cr off-the-books, FY06–FY10.
Modi NDA
2014–present
₹13.6 lakh cr
peak total exposure, FY21
FCI ₹6.7 + NHAI ₹3.17 + recap ₹2.5 + IRFC/AIAHL/NABARD/fert. ~₹2 + residual UPA oil bonds ₹1.3 + SGBs ₹0.4 L cr (mark-to-market). Still ~₹10 L cr today (FY26) — including ~₹2.2 L cr SGB-MTM that has grown as gold has rallied. The cleanup absorbed FCI but rotated, not removed, the overall stack.
Stock of off-budget liabilities by instrument
The stack, year by year
Total outstanding off-budget exposure by major instrument, ₹ lakh crore. The 2021 step-down on the orange band is the Centre absorbing ~₹4.27 lakh crore of FCI's NSSF loans onto the union balance sheet. The tan stripe at the top is Sovereign Gold Bonds shown at mark-to-market by default — the actual rupee liability at current gold prices. Toggle to issue price for book-value comparison with the other instruments.
Source: Union Budget receipt documents (Annexure 6E), CAG audit reports, parliamentary Q&A (Ministry of Road Transport, Ministry of Finance), PRS Legislative Research. NHAI series begins FY16 when borrowing scaled. Approximations where mid-year disclosures missing. SGB mark-to-market uses end-FY gold prices.
"Government has increasingly resorted to off-budget financing for revenue as well as capital spending… food subsidy bills of FCI through borrowings… irrigation schemes through NABARD… railway projects through IRFC… power projects through PFC are outside the budgetary control."
— Comptroller and Auditor General of India, Report No. 20 of 2018, on FY 2016–17 compliance with FRBM
The instruments
What hides where
Each instrument routes spending off the union budget. The government's argument is that these are productive borrowings backed by revenue-generating assets (tolls, food sales, freight). The CAG's repeated finding is that the principal and interest are ultimately serviced by the union, the obligations are not disclosed under government liabilities, and the reported fiscal deficit is therefore understated.
A different kind of off-budget
Sovereign Gold Bonds: borrowing in gold
The other instruments above borrow rupees and repay rupees. Sovereign Gold Bonds borrow rupees and repay rupees at the prevailing gold price — an unhedged sovereign short on gold. The government collected ₹72,274 crore between 2015 and 2024. Gold has roughly quintupled since the first tranche. The outstanding liability now sits several times higher than what was raised, and a major redemption wave hits in 2029 — under whoever forms the next government. The scheme was paused in February 2024 after the Economic Affairs Secretary called it "a high cost method of borrowing."
Raised at issue
₹72,274 cr
across 67 tranches, Nov 2015 – Feb 2024 · 146.96 tonnes of gold
Outstanding at issue price
₹67,322 cr
~130 tonnes still out · MoS Finance reply to parliament, March 2025
Outstanding at current gold price
~₹2.2 lakh cr
implied loss ~₹1.5 lakh cr · S.C. Garg, former Finance Secretary, Oct 2025
Source: RBI Annual Report 2024, parliamentary Q&A (MoS Finance, March 2025; MoS Finance, December 2025), S.C. Garg analysis (Quint, Oct 2025), India Infoline estimates. Mark-to-market liability uses average closing gold price for each fiscal year; FY25–FY26 projection assumes flat current price.
The first redemption set the precedent
The first tranche (Nov 2015, ₹2,684/g) matured in Nov 2023 at a 128% premium. Including the 2.5% annual interest paid over eight years, the total outflow was 148% above what was raised. Every subsequent tranche carries the same mechanism.
The 2029 wave
SGBs issued during the 2021–24 surge (the largest tranches, at gold prices of ~₹4,800–6,200/g) — roughly 60 tonnes — mature in 2029. At today's prices (~₹12,500/g), the redemption cost would be ~₹75,000 cr against ~₹30,000 cr originally raised. Higher gold prices, higher cost.
A partial hedge — kind of
The RBI accumulated 321 tonnes of gold between 2015 and 2025, generating ~$20 billion in mark-to-market gains. This is not a formal hedge against SGB redemption — the gold sits on RBI's balance sheet, not the union's — but it does soften the macro picture even if not the fiscal one.
The historical irony
The BJP's repeated case against UPA's oil bonds was that the principal would have to be repaid by future governments. Sovereign Gold Bonds were entirely a Modi-era creation. The biggest redemption tranches mature in 2029, falling on whoever wins. The accusation was correct; it now applies to its accuser.
The full cost
Ledger: not just principal
The timeline chart shows outstanding principal at face value. That understates the actual cost in two ways: it ignores interest paid over the life of each instrument, and for Sovereign Gold Bonds it uses issue price rather than mark-to-market liability. The ledger below shows the full lifecycle: how much was raised, how much principal has been repaid or absorbed, interest paid to date (approximate where not disclosed line-by-line), and what's still outstanding. All figures in ₹ lakh crore.
Sources: Union Budget receipt documents (Annexure 6E), CAG audit reports, parliamentary Q&A (Petroleum, Finance, Road Transport ministries 2018–2025), RBI Annual Report 2024, PRS Legislative Research, S.C. Garg analysis. Approximate values (~) where line-item disclosure is absent or partial; aggregate "IRFC/AIAHL/NABARD/fert." reflects this. UPA total uses ₹1.88 L cr — sum of oil + fertilizer bonds, consistent with FM's own ₹1.9 L cr "off-the-books" figure (statement, May 2024). Total outflow = repaid principal + interest paid to date; does not project future interest still owed on outstanding stock.
UPA total cost: ~₹3.5 L cr over 21 years
₹1.88 L cr raised, ₹1.44 L cr in interest paid, all cleared by March 2026. The interest bill grew almost as large as the principal — exactly the burden the BJP correctly identified in 2014.
NDA outflow so far: ~₹12.6 L cr
Including ~₹9.3 L cr of principal repaid or absorbed onto budget books, plus ~₹3.3 L cr in interest paid to date. With ₹9.5 L cr still outstanding, the eventual full cost will substantially exceed this.
Interest matters more than headlines suggest
For UPA, interest paid was ~97% of principal — the bonds essentially cost twice what they raised. For NDA, interest paid is currently ~35% of principal-and-absorbed, with much still to come. The longer the rollover, the worse the ratio gets.
SGBs are uniquely costly
For ₹0.72 L cr raised, the outstanding mark-to-market liability alone is already ~₹2.2 L cr — a 3x escalation before counting interest or future gold-price moves. No other instrument compounds against the government this way.
The accounting gap
Reported deficit vs. CAG-adjusted
The CAG's audit reports periodically reconstruct what the union's fiscal deficit would look like if off-budget borrowings serviced from the Consolidated Fund of India were counted in. The gap is not theoretical: every rupee of FCI-NSSF, NHAI, and IRFC borrowing is ultimately serviced by future budgets. The chart below uses CAG's own restated numbers where available, ThePrint's reconstructions where the CAG did not publish a full restatement, and budget documents thereafter.
Source: CAG Report No. 18 of 2022 (FRBM 2019-20); CAG Report No. 20 of 2018 (FRBM 2016-17); Union Budget receipt documents; PRS Legislative Research. FY21 adjusted figure includes one-time COVID-period spending; FY22 reflects partial on-book absorption of FCI liabilities.
Pay it down
How long to clear the stack?
The current off-budget stock that the union has not yet absorbed sits at roughly ₹9-10 lakh crore — NHAI residual, IRFC, recap bonds still amortising, AIAHL, fertilizer arrears, and the resumed FCI exposure. At what pace does this get cleared? Adjust the assumptions and see. The calculator amortises the balance with interest accruing at the chosen rate, repayment applied annually.
Model: simple annual amortisation. Balance × (1 + rate) − repayment, repeated until zero. Does not model GDP growth, refinancing, or absorption shocks. Repayment over 30 years is capped to keep chart readable; if the balance is not declining, the chart will show divergence and the output will say so.
Is this actually new?
What the data actually supports
The defensible critique
Every union government since Vajpayee has used off-budget instruments. Chidambaram introduced oil bonds in 2005–06; the practice predates UPA's scaling of it. UPA also used fertilizer bonds, occasionally FCI. The honest finding is not "the NDA invented this" — it's that the NDA, after campaigning explicitly against UPA's bonds, built a substantially larger and more diversified off-budget stack across nine different mechanisms (FCI, NHAI, IRFC, recap, AIAHL, NABARD, fertilizer arrears, SGBs, oil bond residuals), and that CAG repeatedly noted the disclosure gap until partial absorption began in FY22.
What the data does not support
A direct headline comparison of "₹1.88 lakh cr UPA vs ₹13.6 lakh cr NDA" without GDP normalisation overstates the gap, since GDP roughly tripled between 2010 and 2024. The right comparison is share of GDP, instrument diversity, and disclosure quality. On the first metric, the NDA's peak off-budget exposure (~₹13.6 lakh cr in FY21, or ~6.7% of FY21 GDP) materially exceeds UPA's peak (~₹1.88 lakh cr, ~3.3% of FY10 GDP) — a real expansion, ~7x in rupees but ~2x in GDP terms. On instrument diversity and disclosure quality, the gap is wider.
The partial cleanup
The FY22 absorption of ~₹4.27 lakh crore of FCI's NSSF debt onto the union balance sheet is real, was substantive, and deserves credit. So does NHAI's halt to new borrowing from FY24 and ₹86,000 cr in prepayments. These are not cosmetic moves. The question raised by FCI's December 2025 exploration of fresh off-budget bonds is whether the cleanup was a structural reform or a cyclical pause.