Tamil Nadu's Debt Argument Is Conducted in the Wrong Units

TL;DR

Tamil Nadu’s debt is argued almost entirely in absolute rupees and in one ratio, and the ratio has two official values. The state’s budget documents put outstanding debt near 26% of GSDP and falling. NITI Aayog’s fiscal brief puts total public debt at 31.4% of GSDP in 2022-23 — above the state FRBM Act’s own 25.2% ceiling — and its debt sustainability assessment projects the ratio “on an upward and increasing trajectory, under all the scenarios” to 2026-27, with the four scenarios landing at 36.7%, 34.7%, 34.7% and 41.0%.¹ Neither figure is fabricated; they measure different things. But the framing both sides use hides the constraint that actually binds. Tamil Nadu’s total revenue receipts are 10.4% of GSDP against a median state’s 19.9%, its total expenditure 13.6% against 24.0%, and its capital expenditure 1.9% against 4.0%.¹ Its own tax effort is at the median. Its subsidies went from roughly 8% of total expenditure in 2018-19 to 37.0% in 2022-23, against a median state’s 3.7%.¹ This is a small-budget state making large transfer commitments, not a spendthrift one.

A ₹2,000 note. Public debt is argued in these units and settled in ratios. Photo: Ravi Dwivedi, CC BY-SA 4.0.²

Two Numbers, One Year

The political argument runs on absolute figures, because absolute figures always rise and always sound alarming. Outstanding liabilities roughly doubled over the DMK government’s first four years. They also roughly doubled over the preceding AIADMK term, and over the one before that, in a state whose nominal GSDP was compounding throughout. Absolute debt in a growing economy is a number that cannot go down, which makes it useless for deciding whether anything has gone wrong.

So the argument moves to the ratio, and here it gets genuinely interesting, because there are two official ratios.

The state’s budget documents report outstanding debt at roughly 26% of GSDP, below the 15th Finance Commission’s indicative ceiling, and drifting slightly downward. NITI Aayog’s Macro and Fiscal Brief: Tamil Nadu, published March 2025, reports total public debt at 31.4% of GSDP for 2022-23, describes it as “slightly higher than that of a median State,” and notes that the state’s contingent liabilities are also higher than a median state’s.¹

Five percentage points is not a rounding difference. It is a definitional one — what counts as debt, whether public account liabilities and guarantees are inside the boundary, whose GSDP series is used. Both institutions are reporting honestly within their own definition, and the state’s own FRBM Act sets a ceiling of 25.2% that the narrower measure clears and the broader one does not.¹

The practical consequence: whichever side of this argument you are on, an official document supports you, and neither side has an incentive to mention the other number. That is worth knowing before believing anybody’s chart, including the ones below.

The Constraint Nobody Argues About

Set the ratio aside and look at the flows, because this is where Tamil Nadu is genuinely unusual — and it is not in the direction the debate assumes.

All figures from NITI Aayog’s fiscal brief.¹

Read the first and fourth rows together. Tamil Nadu raises and spends roughly half what a median Indian state does relative to the size of its economy. NITI states it plainly: “The State collects much less revenue compared to a median State,” and its expenditure-to-GSDP ratio is likewise lower.¹

Now read the second row. Own tax revenue is 6.4% of GSDP against a median of 6.3% — Tamil Nadu’s own tax effort is at the median, not below it. The gap is not in what the state collects from its own economy. Transfers from the Centre run at 3.3% of GSDP and have been, in NITI’s words, “consistently lower than what a median State receives.”¹

And the deficits are better than median on both measures. This is not a picture of fiscal indiscipline. It is a picture of a state operating a comparatively small public budget relative to a large economy, with a below-median share of central transfers, and correspondingly little room.

The row that should worry everyone is capital expenditure: 1.9% of GSDP against a median 4.0%, and 14.1% of total spending against a median 17.6%.¹ Whatever the borrowing is funding, it is disproportionately not assets.

Where the Money Went

Which brings us to the part of the argument that is usually conducted with the word “freebies” and no denominator.

Subsidies reached 37.0% of total expenditure in 2022-23, up 28.7 percentage points from 2018-19, and 5.0% of GSDP against a median state’s 1.0%.¹ Tamil Nadu spends five times the median state’s share of its economy on subsidies. That is the largest single anomaly in the whole dataset, considerably larger than anything in the debt ratio.

Two honest qualifications, both of which cut against reading this as pure profligacy.

First, some of the increase is almost certainly reclassification rather than new spending. Off-budget borrowings sit at just 0.05% of GSDP against a median 0.2%¹ — unusually low, and consistent with support that other states keep off the books being carried openly here. Moving a power-utility subsidy from a state guarantee into the subsidy line makes the subsidy number rise and the honesty of the accounts improve at the same time.

Second, “subsidy” is a budget category, not a judgement. It contains free bus travel for women, agricultural power, and food. The Chief Minister’s Breakfast Scheme is the clearest case: launched as a Madurai pilot in September 2022, scaled within a year to nearly 30,000 government schools and 18.5 lakh children, now past 20 lakh students at around ₹600 crore a year, with participation above 90% in rural areas and about 85% in urban ones.³ The state’s Revenue Secretary reports that it “has not only increased student attendance but also improved learning levels and health indicators.”³ That is a departmental claim rather than an independent evaluation, and I have not found one. But note the units: ₹600 crore a year, against a debt stock argued in lakhs of crore. The most-cited “freebie” in the debate is three orders of magnitude smaller than the thing it is supposed to explain. A transfer with a measurable outcome and a transfer with an electoral one look identical in the accounts — and neither is where the money is.

The Power Utility, More Carefully Than Usual

TANGEDCO is routinely described as the fiscal albatross, and on accumulated losses that is fair. But NITI’s assessment contains a fact that almost never appears in the argument.

Tamil Nadu’s Aggregate Technical and Commercial losses — the standard measure of how much power a distribution utility fails to bill or collect for — are lower than the national average, and “have reduced consistently since 2011-12, driven by improved billing efficiency and collection efficiency.”¹ The state signed the MoU for the Ujwal DISCOM Assurance Yojana, which the Government of India launched in November 2015, on both performance-efficiency and financial-turnaround targets.¹

So the utility is not operationally incompetent and is getting better on the metric the sector uses. Its losses come overwhelmingly from a tariff structure that does not cover the cost of supply — which is a political decision about who pays, taken repeatedly and by every party. Calling that a management failure misplaces the responsibility.

What the Argument Should Be About

The debt-versus-welfare framing is close to unfalsifiable: any level of transfer can be called irresponsible, and any level of borrowing defended as investment. Three questions are answerable.

Is the ratio rising or falling? On the state’s definition, falling. On NITI’s, rising under all four scenarios.¹ Anyone quoting one without the other is not arguing in good faith, and the definitional question — which liabilities belong inside the boundary — is the actual disagreement.

Is borrowing buying assets? Mostly not. Capital expenditure at 1.9% of GSDP against a median 4.0% is the single most legitimate criticism available, and it is almost never the one made.

Is the revenue base adequate to the commitments? Own tax effort is at the median, transfers are below it, and total receipts are barely half the median. A state cannot run median-state transfer commitments on half a median-state revenue share indefinitely, regardless of who is at fault for the transfers being below median.

That last question is the same shape as an argument this site has made about uniform rules and unequal capacity: the rule looks even-handed and the outcome depends on the stock of resources each party brought to it. A devolution formula applied identically across states does not land identically on them.

Where I’d Hold This Loosely

Five limits, and the first two are corrections to this post rather than caveats about Tamil Nadu.

An earlier version of this post asserted that the debt-to-GSDP ratio “remains remarkably stable” near 26% and “consistently remains below” a 28% ceiling, and concluded that fiscal discipline was being maintained. It cited the NITI Aayog brief while reporting figures that contradict it, and it did not mention that the government’s own sustainability assessment projects the ratio rising under every scenario. That was the reverse of the honest reading and I have rewritten the post around the source it was already citing.

Second, that version embedded four charts hotlinked from a third-party AI code-interpreter bucket. All four return 403 and had been broken for readers. They have been replaced with charts built from figures I can point to in a document. It also carried claims sourced to a Reddit thread, an Instagram post, two X posts, a LinkedIn post, an exam-coaching site and a partisan blog; those claims and those citations are gone. Where I could not verify a figure in a readable source, I removed the figure rather than re-cite it.

Third, almost everything here is 2022-23, because that is the most recent year in the NITI brief. Both the subsidy share and the debt ratio may have moved since, in either direction, and the sustainability scenarios are projections rather than outcomes. A 2026 reader should treat the levels as dated and the structural comparison as more durable.

Fourth, “median state” is doing a lot of work and hiding heterogeneity. Tamil Nadu is a large, urbanised, high-income state; comparing its revenue-to-GSDP against a median that includes very different economies is informative about magnitude and unreliable about causation. A high-GSDP denominator mechanically depresses every ratio in the table.

Fifth, and against my own emphasis: a low capital-expenditure share is not automatically a failure. A state that has already built much of its physical infrastructure and is spending on health, schooling and transfers is making a defensible allocation, and per capita social expenditure did rise by ₹7,670 over the decade to 2022-23.¹ I have flagged capex as the strongest available criticism, not as a proven one — and the honest position is that nobody in the public argument is making it, on either side.