# How much room do Indian states really have in their budgets? > In 2023-24, states spent about Rs 40 lakh crore on running expenditure and Rs 7.5 lakh crore on capital outlay. Transfers supplied about 42% of revenue, interest and pensions took about a quarter, and education's budget share was lower than in the early 1990s. **State budgets are big. Their room is smaller.** The useful question is not how large state budgets are in raw rupees. It is what is left after transfers, old bills and recurring spending. RBI State Finances shows capital outlay at about 15.7% of revenue expenditure plus capital outlay in 2023-24. It also shows what that capital line is made of, the running account still in deficit, GST changing the own-tax basket, education down as a share of revenue expenditure, health still small, and thin per-person public spending in Bihar, Uttar Pradesh and Jharkhand. ## The big budget is mostly the running budget Indian states and UTs collected about Rs 39.3 lakh crore in revenue in 2023-24. They spent about Rs 40.2 lakh crore as revenue expenditure. The asset-building line, capital outlay, was about Rs 7.5 lakh crore. That first split matters. A state budget speech can sound like roads, bridges, hospitals and schools being built. The accounts say most of the money is the running cost of government: salaries that are inside department budgets, subsidies, maintenance, police, schools already open, hospitals already treating patients, grants passed onward, interest on past borrowing and pensions for past employees. Capital outlay is still large. Rs 7.5 lakh crore is not a footnote. But it is not the budget. ## Capital outlay is not the whole budget In 2023-24 capital outlay was about 15.7% of revenue expenditure plus capital outlay. If someone says 18.6%, check the denominator: that is capital outlay divided by revenue expenditure, not by the combined direct-spending base used here. The difference is not pedantry. The same rupee amount can sound larger when the denominator is smaller. A useful budget story states the denominator before making the claim. ## The running account still leaks The all-state revenue balance was still negative in 2023-24, about -2.4% of total revenue. That means revenue receipts did not fully cover revenue expenditure before we even get to the capital account. Revenue deficit is not the whole fiscal deficit. Borrowing, loans, capital receipts and capital outlay sit elsewhere. But it is a clean warning sign. If the running account is short, then part of borrowing is not building assets; it is paying for today's running bill. ## The Centre is in every state budget In 2023-24, about 42.2% of all-state revenue came from central tax share plus grants. The remaining 57.8% was states' own revenue, mostly own taxes and own non-tax revenue. That is not a scandal by itself. India designed fiscal federalism this way. Poorer states and hill states have smaller tax bases and higher costs, so transfers are meant to equalise. The uncomfortable part is choice. A government that raises most of its money locally has more room to change course. A government that depends on transfers must wait for tax devolution, Finance Commission rules, scheme design and grant timing. ## Transfer dependence is a state-level fact The state ranking is stark. In 2023-24, central transfers were about 90% of Manipur's revenue, about 86% of Arunachal Pradesh's and Nagaland's, and about 72% of Bihar's. At the other end, Delhi was below 4%, Telangana and Haryana were near 20%, and Karnataka, Maharashtra, Tamil Nadu, Kerala and Gujarat sat far lower than the transfer-heavy states. This is why one all-India average is not enough. The Centre is present everywhere, but it is not present in the same weight. ## Old bills arrive before new promises Interest and pensions took about 25.4% of all-state revenue in 2023-24. That is before a new school, a bus depot, a welfare promise or an industrial park. ## The old bills are not evenly spread The burden is uneven. Punjab's interest-plus-pension bill was about 47.8% of revenue in 2023-24. Kerala was about 42.3%. Himachal Pradesh was about 40.1%. These are not tiny accounting lines. They are the old budget arriving at the door before the new budget gets to speak. The chart deliberately keeps interest and pensions together because both are rigid, but they are not the same thing. Interest is the price of past borrowing. Pensions are the price of past employment rules and the age structure of public employees. Salaries would make the rigidity larger, but this workbook does not separate them cleanly enough for the same all-state calculation. ## Capital outlay has its own mix Capital outlay is not one kind of asset-building. In 2023-24, roads and bridges took about 22.9% of all-state capital outlay. Irrigation and flood control took about 17.8%. Water and sanitation took about 11.2%. Education was about 4.8% and medical and public health about 4.3%. That matters because "more capex" can mean very different things. A road-heavy capital budget, an irrigation-heavy budget and a hospital-building budget all sit under the same capital-outlay label. The chart does not tell us whether the asset was completed on time or built well. It tells us what kind of claim the capital budget is making. ## The cut falls on the flexible line Budget Estimates are promises made before the year begins. Accounts are what happened after revenue came in, grants arrived or did not arrive, and spending departments faced cash limits. From 2014-15 to 2023-24, capital outlay actuals averaged about 81.0% of Budget Estimates. Interest payments averaged about 98.9%. Grants from the Centre averaged about 75.7%. In plain English: the line that builds assets is easier to miss than the line that pays lenders. This does not prove bad faith. A state may budget a road, then face a revenue shock, a delayed grant or a land problem. But the pattern is still useful. When the accounts are written, capital outlay is less protected than interest. ## The human budget also gets revised down Education and health are politically protected words, but the accounts still matter. From 2014-15 to 2023-24, education actuals generally came in below Budget Estimates. In 2023-24, education was about 94.1% of BE. Health plus family welfare was about 90.1%. The health line has one important exception: shock years can push actuals above the original plan. That is exactly why budget realism should be read by category. Interest behaves like a hard bill. Capital outlay behaves like a flexible bill. Health can be routine in one year and emergency spending in another. ## A scatter for fiscal room Transfer dependence and committed spending do not point to the same states. Bihar and the northeastern states depend heavily on the Centre. Punjab, Kerala and Himachal carry heavy interest-plus-pension bills. Odisha and Gujarat have more capital-outlay space in this particular cut of the data. Delhi is a special case because it is a UT with legislature and a different responsibility set. The scatter is a sorting device. Rightward means more revenue is tied up in interest and pensions. Upward means more of direct state spending is capital outlay. Punjab sits low and right. Arunachal sits high and left. Gujarat and Odisha are high on the capital-outlay axis. This does not explain why. It tells you where the hard questions begin. A clean state-finance story has to resist one villain. Low capital outlay can come from debt, pension load, low revenue, transfer design, political choices, project readiness or the fact that a small state has different fixed costs. The data shows the squeeze. It does not supply a single cause. ## The capital-outlay map cuts both ways Punjab, Puducherry and Kerala sat near the bottom of the 2023-24 capital-outlay share ranking. Punjab's capital outlay was about 3.9% of revenue expenditure plus capital outlay; Kerala's was about 8.7%. At the other end, Arunachal Pradesh, Sikkim, Gujarat, Odisha, Meghalaya and Jharkhand were above 20%. That is not a simple league table of virtue. A high capital-outlay share can reflect catch-up investment, central projects, geography, a small population, or one large project year. A low share can reflect debt pressure, pension pressure, project delays or a deliberate choice to protect current spending. The chart earns its place because it shows where to investigate. ## The running budget is still mostly developmental Revenue expenditure is often dismissed as "just running costs". That is too lazy. In RBI's broad classification, developmental expenditure was about 62.3% of revenue expenditure in 2023-24. Non-developmental expenditure was about 34.7%, and grants-in-aid plus contributions were about 3.0%. The useful signal is the drift. Developmental spending was closer to 68% of revenue expenditure in 1990-91. The share is still the majority, but it is lower. Non-developmental spending has taken more room. That is where interest, pensions, administration and other non-developmental heads begin to matter. ## Education and health need their own window Education was about 16.9% of revenue expenditure in 2023-24. In 1990-91 it was about 21.6%. Health plus family welfare was about 6.1% in 2023-24, not much different from the early 1990s in share terms. This is one of the article's sharper caveats. Shares can fall even while rupee spending rises, because the whole budget has grown. But a share still reveals priority inside the running budget. If education loses share while pensions and interest remain large, the budget's future-facing claim weakens. ## Raw rupees mislead across states Uttar Pradesh will almost always look huge in rupees because it has so many people. Goa, Sikkim or Arunachal Pradesh can look tiny because they have fewer people. That is why the article joins the State Finances data to RBI Handbook denominators: GSDP for the size of the state economy, and a derived population estimate for per-person comparisons. The per-person chart changes the feeling of the story. In 2023-24, revenue expenditure per derived resident was about Rs 14,900 in Bihar, about Rs 18,100 in Uttar Pradesh and about Rs 19,300 in Jharkhand. It was above Rs 1 lakh in Arunachal Pradesh, Sikkim and Goa. Some of that is real fiscal capacity. Some of it is the arithmetic of small populations, geography and grants. ## Education rupees are thinnest where the child count is large Education revenue expenditure per derived resident was about Rs 2,900 in Uttar Pradesh, about Rs 3,000 in Jharkhand and about Rs 3,200 in Bihar in 2023-24. Sikkim and Goa were near Rs 20,000, and Arunachal Pradesh was above Rs 16,000. This is not the perfect education denominator. The cleaner denominator would be school-age children or enrolled students. But per resident is still useful because it shows the fiscal floor behind a state's education system. The big-population states are trying to run large school systems with thin rupees per resident. ## Health spending needs rupees, not only shares Health shows the same warning in a sharper way. Health and family welfare revenue expenditure per derived resident was about Rs 850 in Bihar and about Rs 1,035 in Uttar Pradesh. Arunachal Pradesh and Goa were around Rs 10,000 per derived resident. That is not the same as health outcomes, and it excludes private and out-of-pocket spending. But it is a useful floor check on the public budget. ## Capital outlay needs an economy denominator Capital-outlay share answers one question: how the budget is split between running spending and asset-building. Capital outlay as a share of current-price GSDP asks a different question: how much state investment effort sits against the size of the state economy. Punjab and NCT Delhi were around 0.6% of GSDP in 2023-24. Puducherry was about 0.9%, Kerala about 1.2%. At the other end, Arunachal Pradesh was about 21.9%, Meghalaya 8.5%, Nagaland 7.8% and Manipur 6.3%. Odisha and Sikkim were both around 5.4%. Do not read this as an infrastructure-quality ranking. A high ratio can come from a small economy, transfer-heavy budgets, geography, or a lumpy project year. The chart earns its place because it separates a state's investment intensity from its budget mix. ## Revenue deficit is the pre-capex warning light In 2023-24, Punjab's revenue balance was about -3.7% of GSDP. Andhra Pradesh, Himachal Pradesh and Rajasthan were around -2.6% to -2.7%. These states began with a running-account shortfall before capital spending entered the picture. The surplus side is also telling. Odisha, Nagaland, Tripura, Meghalaya and Jharkhand had positive revenue balances as a share of GSDP, while Arunachal Pradesh was very high because its transfer-heavy budget is large relative to a small economy. A revenue surplus is not proof of better services. It says the running account is not the immediate leak. ## GST changed what own revenue means The own-revenue line is not stable under the hood. Before GST, sales tax and VAT were the dominant state tax handle. In 2016-17, sales tax and VAT were about 64.3% of own tax revenue. SGST was zero because it did not exist yet. By 2017-18, SGST was about 31.3% of own tax revenue and sales tax/VAT had fallen to about 35.5%. By 2023-24, SGST was about 43.1% and sales tax/VAT about 20.8%. State excise and stamps remain important, but GST changed the composition of what "own tax" means. This is why state fiscal room cannot be read only as a revenue-effort morality play. The tax base, the rules of a shared GST system, the Centre's transfers and state choices all sit in the same account. ## How to read these numbers This article uses RBI's State Finances: A Study of Budgets 2025-26 e-STATES workbook as the main source. Account values are treated as actuals. 2023-24 is the latest full actual year in this workbook. 2024-25 is Revised Estimate and 2025-26 is Budget Estimate, so they are not used as observed spending. The all-state line has a scope break. RBI notes that the All States/UT aggregate excludes UTs from 1990-91 to 2016-17, except that NCT Delhi is included from 2000-01 to 2004-05. From 2017-18 onward, it covers all states and UTs. Long-run all-state charts should be read with that break in mind. Derived shares are computed from the workbook's published rows. Central transfer share is share in central taxes plus grants from the Centre, divided by total revenue. Interest-plus-pensions share is interest payments plus pensions, divided by total revenue. Capital outlay share uses capital outlay divided by revenue expenditure plus capital outlay unless the chart says otherwise. Actual-to-budget is Account divided by Budget Estimate for the same fiscal year and the same budget head. Cross-state per-person and per-GSDP charts join State Finances to RBI Handbook of Statistics on Indian States denominators. Per-person metrics convert the State Finances rupee-crore value into rupees and divide by population derived from current-price NSDP divided by per-capita NSDP in the Handbook snapshot. It is not a fresh Census count. GSDP metrics divide State Finances rupee-crore values by current-price GSDP from the same Handbook snapshot. The denominator charts exclude the All States/UT aggregate because the State Finances aggregate and Handbook UT coverage do not match exactly. One source-quality issue is preserved rather than "fixed": Karnataka's top-level revenue components do not reconcile to total revenue for 2021-22 Account, 2022-23 Revised and 2023-24 Budget in the source workbook. The article avoids making a Karnataka composition claim from those problematic rows. ## Sources - [object Object] - [object Object] - [object Object] - [object Object] --- Source: [This Indian Life](https://thisindianlife.today/articles/how-much-room-state-budgets-have/) · Updated 2026-07-08. Licensed CC BY 4.0. Please cite as "This Indian Life — https://thisindianlife.today".