Government Expenditure: Revenue Spending Versus Capital Spending

Revenue expenditure is government spending that is consumed inside the period it is made in. Capital expenditure creates something that outlasts the period. The test is not the size of the sum, the department that spent it, or how worthy the purpose was. The test is whether anything remains afterwards, and it is clean at the extremes and genuinely hard in the middle.

Governments do not keep two purses. Every rupee leaves the same treasury and lands in the same total. The two headings are separated by what happens once the calendar turns over. Some spending is gone the moment the work it paid for is done, and the government is left holding a receipt and nothing else. Other spending turns into a thing that is still standing, still running and still counted when the next period opens. Whether anything is still standing is what the whole classification is built on, and it is the reason a government can spend an identical amount two years running and be in a completely different position at the end of the second year.

What is the test that separates the two kinds of spending?

Ask one question and only one. When the period ends, is anything still there? A wedding is a useful place to feel it. A household spends Rs 5,00,000/- on the catering and Rs 5,00,000/- on a small flat for the couple. Both cheques are the same size, both were spent with love, and the day after the wedding one of them has completely vanished into a very good dinner while the other is a place somebody lives in. Nobody in that household would confuse the two, and nobody thinks the catering was a mistake. The catering and the flat are exactly the distinction a treasury draws.

The test is about what is left, not about how large the sum was and not about how worthy the purpose was. Almost every wrong answer in this area comes from quietly swapping that test for a different one. A vaccination drive is enormously worthy and it is revenue spending. A warehouse that nobody has yet found a use for is capital spending. Worthiness is a real question, and it is a different question, answered with different evidence. A worthy purpose does not move a single rupee across the line.

Notice what kind of thing must remain. The thing has to be something the government can still point at, count and put a value against once the period closes: a road, a school building, a piece of equipment, a loan it has made to somebody else and can still call in. The formal word for such a thing is an assetSomething an entity holds that is expected to be of use or of value in future periods, and that can be identified and valued on its own. Covered properly under accounting., and the length of time it is expected to keep being useful is its useful lifeThe stretch of time over which a thing is expected to keep doing its job before it is worn out, replaced or written off. Estimated, not observed.. If the government cannot point at anything at the end of the accounting periodThe fixed stretch of time an account is drawn up for, so that spending and receipts inside it can be compared with each other. Its length is set by rule, not by the nature of the spending., the spending was consumed inside it.

Capital Expenditure vs Revenue Expenditure: which side does a rupee fall on?

Take the Republic of Sankhya, an invented country, and open the spending side of its government account. Total spending for the year is Rs 4,00,000 crore. The total splits into three named lines and only three. Interest of Rs 90,000 crore is paid on borrowing taken in earlier years. Other revenue spending of Rs 2,40,000 crore covers salaries, pensions, running costs, every subsidyA payment that lowers what somebody pays for a good or service, with the government meeting the difference. Why subsidies exist and who bears them is covered separately. and the ordinary business of keeping the state working. Capital spending of Rs 70,000 crore builds and buys things. Add the first two and revenue spending is Rs 3,30,000 crore. Add the third and the total is Rs 4,00,000 crore.

Paying for borrowing already taken leaves nothing whatever behind, so interest is revenue spending in full, and in this account that one line is over a fifth of everything spent. Work it out and interest is 22.50 per cent of total spending, and 27.27 per cent of revenue spending on its own. Interest feels like a payment towards something, and the heading surprises people for that reason. Interest is not a payment towards anything. The road that the old borrowing built is already standing and was already booked as capital spending in the year it was built. The interest is the price of having borrowed, and when the year ends the government holds nothing new for it. There is no version of the test under which interest comes out the other way.

Read the shares off it slowly. Capital spending is Rs 70,000 crore out of Rs 4,00,000 crore, or 17.50 per cent, so a little under one rupee in six left something behind. The other five rupees and a bit were consumed by the year. The split reaches the receipts side too. Sankhya took in revenue receiptsThe money a government takes in that creates no obligation to repay and gives up no asset, chiefly taxes and its ordinary non-tax income. Where it all comes from is covered separately. of Rs 3,00,000 crore against revenue spending of Rs 3,30,000 crore, so ordinary income fell Rs 30,000 crore short of ordinary spending.

Why does the split matter more than the total?

Hold Sankhya total spending exactly where it is, at Rs 4,00,000 crore, and change nothing about the size. Now re-split it. Capital spending becomes Rs 1,40,000 crore instead of Rs 70,000 crore, so revenue spending is Rs 2,60,000 crore: interest is still Rs 90,000 crore and other revenue spending falls to Rs 1,70,000 crore. The headline number a newspaper would print is identical. The capital share has gone from 17.50 per cent to 35.00 per cent, and the country holds something completely different when the year closes.

A total is a size and a split is a shape, and two governments spending exactly the same total can leave behind entirely different things. Anyone who reads only the total has read one of the two numbers. The mistake is the same as judging a household by what it spends in a month without ever asking whether the money went on rent and groceries or on a deposit for a home. Both households spend Rs 60,000/- a month. One of them will have something after ten years and the other will have ten years of receipts. Neither is doing anything wrong. The two households are simply different shapes, and only the split shows the shape.

The re-split does a second thing, and it is easy to miss. With capital at Rs 70,000 crore, revenue spending of Rs 3,30,000 crore sat Rs 30,000 crore above revenue receipts of Rs 3,00,000 crore. With capital at Rs 1,40,000 crore, revenue spending of Rs 2,60,000 crore sits Rs 40,000 crore below the same receipts. The ordinary income of the state now more than covers its ordinary running. The gap between revenue spending and revenue receipts has a name and is covered separately. Somebody wanted a number that notices shape rather than size, and the gap is that number.

Where does the classification get genuinely hard?

At the extremes the test is clean, and a reader who stops there carries away more confidence than the subject supports. In the middle the test is not clean, and three large categories of spending sit near the line rather than on one side of it. The three are not errors in the system. Each is a place where the underlying question, does anything remain, genuinely has more than one defensible answer.

Take maintenance first. A government spends Rs 900 crore resurfacing roads that already exist. Nothing new has been created. The road count is the same the day after as the day before. And yet without that spending the road would have stopped being usable in a few years, so the money has bought years of continued service from a thing the government already holds. Is that consumption inside the period or is it a top up of an asset? Accounting has a partial answer, in that spending which merely stops depreciationThe steady writing down of a thing's recorded value as it is used up or wears out over its life. A recording convention, covered under accounting. from biting is treated differently from spending which genuinely extends a life. The distinction is a real answer, and the boundary still has to be judged, item by item, by somebody.

Then take transfers. Sankhya sends money to a state government or to a public body as a grantMoney handed over with no expectation of repayment and no asset taken in return by the giver, usually tied to a purpose. A loan is repayable and a grant is not.. The money leaves the central account and nothing remains in the central account. But the receiving body may spend every rupee of it building a school, in which case something certainly remains, just not in the hands of the government that paid. Whose balance sheetA statement of what an entity holds and what it owes at a single moment in time. The place where a thing that remains would show up. Covered under accounting. the resulting thing lands on decides the answer, and a classification made at the moment of payment cannot see where the money will end up.

Then take training. A government spends Rs 300 crore training nurses. The nurses are better nurses for the rest of their careers, so something has plainly lasted well beyond the period. But nobody can put the improvement on a shelf, count it, value it or sell it, and if the nurses leave the service it goes with them. An account has to record things it can identify, so economists have a phrase for the improvement and treasuries mostly do not use it.

Maintenance, transfers and training are real ambiguities and not errors, and a reader who thinks the line is always obvious will trust a published split further than it deserves. The three cannot be settled by assertion without misrepresenting how the accounts work. The right response is a different one: when a split moves sharply and the movement matters, the thing to examine is what sits in the lines that moved, and whether the underlying spending changed or the treatment of a borderline item did.

What does the split refuse to say?

Put two payments side by side. A teacher is paid a year of salary in a school with full classrooms and children who can read at the end of it. A bridge is built across a river that almost nobody needs to cross, and it stands there being crossed four times a day. The salary is revenue spending. The bridge is capital spending. Both are correctly classified. Apply the test and there is no wobble at all: at the end of the period the teaching is done and consumed, and the bridge is still there.

A salary for a teacher is revenue spending and a bridge to nowhere is capital spending, and the classification says nothing whatever about either one. The test asks what remains. The test does not ask whether what remains was worth having, whether what was consumed was well consumed, or which of the two did more good. Somebody who reads a rising capital share as a government doing better, and a rising revenue share as a government wasting money, has picked up an instrument that measures shape and is reading a verdict off it. The split is the wrong instrument for that job, and it will be wrong in both directions: it will praise an unused bridge and it will condemn a classroom.

The right way to hold both ideas at once is to keep the questions separate and answer each with its own evidence. What does this spending leave behind? The split answers that, subject to the hard cases above. Was this spending worth making? The answer comes from what the spending actually bought, who it reached, what it displaced and what it cost to run afterwards. The second question is harder, slower and cannot be answered from an account at all. The split is not a shortcut to it and was never built to be one.

Which line comes down first when spending has to be reduced?

Now watch what happens to the split when a government is under pressure and total spending has to fall. The two headings do not behave the same way at all, and the reason is entirely about who feels the cut and when. Reducing a revenue line means somebody is paid less, or a payment somebody was receiving stops, and it happens this year, visibly, to a named person. Reducing a capital line usually means a project moves more slowly. The road still gets built. The road arrives in the year after next instead of next year. Nobody is paid less this year.

Because a project can slow without anybody being paid less this year, the capital line is almost always the easier one to move, and that makes the split informative about pressure in a way the total is not. This is a direction, not a judgement, and whether any government should have done it is a separate question. A falling capital share in a year when total spending held roughly steady is very often the visible trace of pressure that arrived somewhere else in the account. The revenue lines are made of contracts, entitlements and people. Interest is owed on borrowing already taken, so it does not move at all.

What does an analyst actually do with the capital share?

Somebody who reads government accounts for a living treats a single period's capital share as a starting point rather than a finding, and the reason is practical. A capital line in any one year is as much about what happened to be ready to spend as about what anybody intended. Land was not cleared, a tender ran late, a contractor stopped work, monsoon arrived early. All four of those push a capital line down in a year when nothing about the intention changed at all. So the number they want is the direction over five or six consecutive periods, not the level in one.

A single period shows where the capital share is and almost nothing about where it is going, so the useful reading is a run of periods rather than a point. A lender sizing a loan to a government does the same thing for the same reason. So does a household, in miniature, when it works out whether it is actually getting anywhere: one month in which nothing was saved says nothing, and five months in which the saving shrank every month says a great deal. The second reading is the one worth having, and it is only available if somebody kept the earlier months.

Two habits go with this. First, one large borderline item reclassified will move a share without any spending changing, so when a share moves sharply, examine what sits inside the lines that moved before believing the movement. Second, a revenue line that is mostly interest behaves very differently from one that is mostly salaries, so the capital share is read alongside what the revenue lines are made of. The salary line is at least in principle something a government decides each year, and the interest line is not.