PPF: The Rules, the Lock-In and the Arithmetic

The Public Provident Fund (PPF) is an account where the government sits on the other side. A household pays money in, the scheme adds a credit it declares rather than one a market produced, and the money stays out of reach for a term the scheme fixes. The credit is worked on the lowest balance held from the fifth of a month to its end, so the date a deposit lands changes what that month earns.

Everything about this account follows from one swap, so the swap is worth naming before the mechanics start. Every arrangement covered earlier in this sequence exposed the household to a market: something quoted, something that moved, something whose value on any given morning was whatever somebody was prepared to pay. A public provident fund account does not. The exposure here is to time, and to the household's own inability to reach the money while that time passes. The swap is a genuine trade, not a free lunch, and an account of the credit that leaves out the lock-in has described half an arrangement.

Both halves are set out below: the arithmetic, line by line, and then what a household gives up by holding money it cannot touch while its buffer stands at Rs 31,320/-, or 0.73 months of what it commits. Both halves are true at once, and which one weighs more depends on a household's position rather than on the account.

What is the Public Provident Fund, and who is on the other side?

A question people rarely ask about money they have set aside is who owes it to them. Every rupee that is not physically in hand is somebody else's obligation. Cash in a bank account is the bank's obligation. Money lent to a cousin is the cousin's obligation. A share is not an obligation at all. Nobody promised to give anything back for it, and that is precisely why its price moves.

The Public Provident FundAn account where the government is the counterparty and declares the credit that is added to the balance. answers that question differently from anything else in this sequence. Money paid into it becomes an obligation of the government, administered through the ordinary banking and post office network but not owed by the branch that took the form. The branch is the counter. The government is the counterpartyWhoever owes the money. Not the same as whoever handed over the form or stamped the passbook..

The identity of the counterparty is what makes the credit on this account a declared figure rather than a produced one. A bank pays interest out of what it earns lending the deposit onward, so a bank in trouble is a bank whose deposits are in question. A share pays what the business happens to make. The balance grows by a figure the scheme states for a stretch at a time, and that statement does not depend on how any market behaved that quarter.

The difference has a shape that can be felt. A tailoring counter in a market lane takes what the lane brings: a busy fortnight before a wedding season, a thin one in the rains, and nobody announces in advance what June will hold. A monthly rent agreement is the opposite shape. The tenant pays a stated amount on a stated date because a document says so, and the amount does not move because the lane was quiet. A credited returnOne added by the scheme's own rules rather than produced by a market. The scheme states the figure; nobody has to sell anything for it to appear. is the rent-agreement shape. The counter is the other shape. Neither is better than the other; they go wrong in different ways and at different times.

One more thing before the mechanics. The person at the window, the passbook, the branch name across the top of it, the form with its columns: none of these say who the counterparty is, and all of them are what a household actually sees. The service is what a household sees. The obligation is what matters, and the obligation is almost never printed in the largest font on the document.

How PPF Works, and what are its three moving parts?

Strip the forms away and this account has three moving parts. Not eight, not a list of conditions, three. Everything a household ever has to think about here is one of the three, and the confusions people carry about this account almost always come from mixing up which part a question belongs to.

The first part is the paying in. A household decides an amount and pays it into the account, in one go or spread across a year. The second part is the credit. Once a year the scheme adds an amount to the balance, worked out on the balance rather than on what was paid in that year. Money paid in early therefore does more work than money paid in late. The third part is the holding. The account runs for a termThe length the account is meant to run for, fixed by the scheme rather than chosen by the household. the scheme fixes, and across that stretch the money is not freely reachable.

The three parts are the entire mechanism, and only the first one is a household decision. The size of the credit is set elsewhere. The length of the holding is set elsewhere. A household controls how much goes in and when, and the when is worth more than most people expect.

Now the part that shapes everything printed below. Every actual number attached to parts two and three is a scheme rule: how much may go in during a year at the smallest and at the largest, what the credit rate is, how long the term runs, whether it can be extended, whether anything can come out early and under what circumstances, and how any of it is treated for tax. Each has been changed by the authorities before, so any text printing today's version would be quietly wrong to somebody reading it in three years.

So the shape comes first, and then an arithmetic worked on two figures assumed for the purpose: an assumed rateA figure chosen so the sum can be followed end to end. The scheme declares its own rate, and this assumed figure is neither that rate nor a prediction of it. of 7 per cent a year and an assumed term of 15 years. Both are assumptions, and neither is a scheme figure. Both are round enough to check by hand and long enough to show the mechanism at work, and that is the only reason either was chosen.

Why bother separating the shape from the numbers so pedantically? Because the two decay at completely different speeds. The shape of this account, money in, a credit added on the balance, a stretch during which it stays put, has been the same for a very long time and will still describe the thing after the next revision. The numbers attached to it are revisited by the authorities on their own schedule. A reader who learns the shape can pick up today's numbers in ten minutes from the source; a reader who memorised the numbers has learned the one part that expires.

How is the credit actually computed, and why does the fifth of a month matter?

The credit calculation is the part of the account that surprises people, and it is the one place where a household decision that costs nothing changes the outcome.

The credit is not worked on the average balance across a month, and it is not worked on what the balance happened to be on the last day. The credit is worked on the lowest balance the account held across a stated stretch of the month, running from the fifth to the month end. That is the minimum balance ruleThe credit for a month is computed on the lowest balance the account held across a stated stretch of that month, not on the closing balance and not on an average., and everything else about deposit timing follows from it.

Work through what that does. Suppose the account is sitting at Rs 60,000/-. A household pays in Rs 1,000/- on the twentieth. From the fifth to the nineteenth the balance was Rs 60,000/-, and from the twentieth to the month end it was Rs 61,000/-. The lowest figure in that stretch is Rs 60,000/-, so the month is computed on Rs 60,000/-. The Rs 1,000/- is genuinely in the account, it is genuinely the household's money, and it is not in that month's calculation at all.

Now move the same deposit to the third of the same month. From the fifth to the month end the balance never dips below Rs 61,000/-, so the lowest figure in the stretch is Rs 61,000/- and the month is computed on Rs 61,000/-. Same money, same account, same household, seventeen days earlier, and a different month of calculation. Nothing else changed.

People hear this and assume a penalty is hidden somewhere for depositing late. There is not, and that is the point. The minimum balance rule only decides which balance the arithmetic looks at. A household that knows it gets the better side for free; a household that does not loses nothing it ever had. Depositing on the third rather than the twentieth is the difference between reaching a queue at ten past and at five to. Nobody was fined.

One extension of the same rule, because it changes how a year is read rather than how a month is. If the credit is added once a year and each month is measured this way, then money that goes in early in a year is measured across more months than money that goes in late. A household paying the whole year's amount at the start has every month of that year measured on the larger balance. A household paying at the end has almost none of them. The arithmetic below assumes the first pattern.

One assumption behind the next section gets buried easily, so it is worth restating plainly. Every figure in the next section is worked as though the year's Rs 12,000/- sat in the balance for the whole of that year. A household paying Rs 1,000/- on the twenty fifth of each month would not reach the same figures, and the shortfall would not be a fault in the arithmetic. The shortfall would be the minimum balance rule, applied twelve times.

What does the arithmetic actually look like over a long term?

One worked instance carries the rest of the arithmetic. Meghna Bhosale, of the invented Bhosale household, pays Rs 1,000/- a month into this account, or Rs 12,000/- a year. The arithmetic below runs at an assumed rate of 7 per cent a year and across an assumed term of 15 years. Both figures are assumptions of this guide, chosen so the sum can be checked by hand. Neither is a scheme figure, and neither is typical, expected or likely.

The rule for each row is one line. Take the balance carried in, add the year's Rs 12,000/-, work the assumed 7 per cent on that, round to the nearest rupee, and carry the total into the next row. One line, done fifteen times, is the whole calculation.

The bottom row of the table above, read against its top row: Rs 1,80,000/- went in over fifteen years, Rs 1,42,660/- was added on top of it, and the two sum to Rs 3,22,660/-. The credit is 79.3 per cent of what was paid in. The same account read at year five stands differently: Rs 60,000/- paid in, Rs 13,841/- credited, a total of Rs 73,841/-, and the credit is 23.1 per cent of what was paid in.

The rate was 7 per cent in both readings, and the share tripled from 23.1 per cent to 79.3 per cent purely because the number of years tripled. Nothing about the rate did that. Long accounts of any kind are misread in exactly this way. People argue about the rate, a figure nobody outside the scheme can set, and treat the term as background, though the term is the part that actually decided the answer.

There is a moment inside the table that makes the point better than any of the totals. Look at the credit column in year 10, Rs 11,606/-, and then in year 11, Rs 13,258/-. Somewhere between those two rows the account crosses over: from year 11 onwards, what the scheme adds in a single year is larger than the Rs 12,000/- the household paid in that year. From that point on the balance is being built more by the arithmetic than by the household, and the household is still paying in exactly the same Rs 1,000/- a month it paid in the first year.

Read the first and last rows of the credit column together and the shape is complete. Year one adds Rs 840/-. Year fifteen adds Rs 21,109/-, more than twenty five times as much, at the same assumed rate on the same Rs 1,000/- a month. The account did not get better. It got older.

Two things are worth doing with that control before reading on. Drag the years down to 5 and watch the lime part almost vanish, then drag it to 25 and watch it overtake the dark part entirely. Then leave the years at 15 and click through the four rates. The balance moves, of course, but it moves far less than dragging the years did. The two drags make the whole claim in one experiment: over a long holding, length is the larger lever, and length is the one the household controls.

The lower half of that diagram is the part households feel and calculators never show. Rs 31,320/- is 0.73 months of what this household commits, so the buffer bar barely leaves the left edge. The locked bar passes it before the end of the first year and keeps going. By year fifteen there is more than seven and a half months of commitments sitting in an account that cannot be opened. Both bars are the same household's money on the same afternoon.

What does the lock-in actually do to the money?

A lock-inA stretch of time across which the money cannot be freely taken out, whatever the holder would like to do with it. is usually described as a restriction, and the word makes it sound like small print in a form. A lock-in is better understood as a change in what kind of thing the money is.

Household Resilience drew the line this sequence keeps returning to. Money the household holds splits into what can be turned into rupees on the afternoon it is wanted, and what cannot. The split is not a judgement about quality. Gold at the household's own estimate of Rs 1,40,000/- is real and takes a trip and a haircut to become money. A two-wheeler at Rs 38,000/- is real and becomes money only if somebody buys it. And a balance behind a lock-in is real and becomes money on a date the scheme picked, not a date the household picked.

The lock-in does not make the money smaller; it moves the money out of the column the household can act on and into the column it can only count. The Rs 84,000/- already sitting in this household's public provident fund is exactly as real as the Rs 41,887/- in its two bank accounts. One of them can pay a hospital deposit on Tuesday afternoon and one of them cannot, and no amount of arithmetic makes them the same kind of rupee.

Call the second kind unreachable moneyAn amount that genuinely belongs to the household and cannot be used today. It counts towards what the household has and not towards what it can do.. Unreachable money belongs on the statement of what the household has. It does not belong in any sentence about how long the household could manage.

What does the lock-in cost a household, and what does it protect?

Now the two readings, each with equal room.

Read as a cost, it is straightforward. Something happens in June. A parent needs a deposit paid at a hospital counter before a procedure is scheduled. The household has Rs 3,22,660/- in an account with the government on the other side of it, and it cannot be touched. So the money comes from a card, a loan against something, an amount borrowed from relatives, or three things that get postponed. Each has a price, and it is being paid by a household that is not, on paper, short of money. The price of using a card or a loan instead is a real cost, worth naming as one rather than as an inconvenience.

Read as a protection, the same sentence is doing something else entirely. Fifteen years is a very long time to leave money alone. Across fifteen years a household meets a wedding it wants to do properly, a vehicle that would make the commute easier, a business opportunity a cousin is confident about, four festivals a year, and roughly forty moments where a large round balance would have been extremely convenient to dip into. The reason the table above reaches Rs 3,22,660/- is not that the household was disciplined. It is that the account would not open.

Both readings are the same rule, described twice, and neither description is the true one hiding behind the other. The lock-in stops the household reaching the money in June, and the lock-in stops the household reaching the money in June. Which of those two sentences matters more to a household is a fact about its buffer, its income steadiness and who depends on it, not a fact about the account.

There is a version of this that has nothing to do with money. A household that keeps its emergency cash in a tin at the back of a high shelf has built a small lock-in by hand. The tin is not secure and nothing stops anybody reaching it. Since the money is not in a pocket at the vegetable market, the shelf buys friction and nothing else. Every household that has done this understands the trade instinctively. The friction is annoying on the day the tin is wanted at short notice, and the friction is the reason there is anything in the tin at all.

A lock-in is the same trade with a rule in place of the shelf, and a rule cannot be reached by pulling up a chair. The rule makes the protection stronger and the cost sharper, in exactly the same proportion.

Where does the arithmetic go wrong in practice?

The practical fix is embarrassingly small, which is what makes it worth doing. The household sheet gets a second column rather than a second sheet: everything held, then a tick against the rows that could become rupees this week. The Rs 41,887/- across two accounts gets a tick. The recurring deposit at Rs 64,000/- gets a tick with a note about what breaking it costs. The gold, the two-wheeler and this account do not. The resilience sentence is then built only from ticked rows, and the mistake becomes hard to make rather than something to remember.

Who carries the risk here, and which risk is left?

Ask most people where the risk sits on an account like this and the answer is nothing, it is the government. The answer is half right, and the half it gets wrong is the half that matters over fifteen years.

Split the risk into its two honest halves. The first is that the money does not come back at all, or comes back late, or comes back smaller because whoever held it could not pay. Not being paid is a question about the counterparty, and it is the one everybody thinks of first because it feels like danger. Here the risk sits with the government, about as far as that risk can be pushed inside this country's own currency.

The second is the risk that the money comes back exactly as promised and does less when it arrives. Rs 3,22,660/- fifteen years from now is Rs 3,22,660/- fifteen years from now. Whether that pays for a year of what this household needs, or four months of it, is a question about prices, and prices are not the counterparty's department. A government on the other side removes the risk of not being paid and removes nothing whatsoever about what the payment will buy.

The separation between a certain amount and a safe amount was drawn under Household Resilience, and this account is where that separation earns its keep. Certainty here is genuine and it is narrow. It is certainty about a number of rupees, not certainty about an outcome, and those two things feel identical on a statement and behave nothing alike over a decade and a half.

How large the second risk is depends on an inflation figure, and the Reserve Bank of India publishes the official series.

What does this arrangement not protect against?

Four things, worth listing because the account is often talked about as though it protected against everything.

The account does not protect against prices rising, for the reason just given. The number is fixed in rupees and rupees are not fixed in what they buy.

The account does not protect against a household needing the money. The lock-in read as a cost is not a defect. The arrangement is working as designed, on a day the household would rather it did not.

The account does not protect against the rules changing while the money is inside. Every scheme figure is set by the authorities and every one of them has moved before. A household that paid in on the strength of a particular figure has not bought that figure for the whole term.

And the account does not protect against the arithmetic simply being too small for the goal. The fourth is the quiet one. Rs 3,22,660/- after fifteen years, on the assumptions used here, is a real and certain-in-rupees amount, and it is a fraction of what the earlier retirement costing in this sequence worked out this household would need for a retirement it costed at Rs 30,000/- a month in today's money. Being certain and being enough are two entirely separate tests, and passing the first says nothing at all about the second.

What should a household weigh before putting money in?

The questions can at least be put in an order that makes the decision a household one rather than a product one. Only the last of the four is about the account at all.

The first question is whether this money can be out of reach for the whole term. Not whether it probably will not be needed, the question everybody actually asks. Whether it can be gone. For a household whose buffer covers 0.73 months, that question has real weight, and an honest answer may mean a smaller amount, a later start, or the buffer first. None of those is a failure and none ranks above the others.

The second is whether the amount can be kept up in a bad year as well as a good one. A tailoring counter has thin months. A freight office can restructure. An amount that is comfortable in a normal month and impossible in a difficult one turns a plan into a monthly source of guilt, and there is no arithmetic reason to choose an amount that does that.

The third question asks how much of this money is already spoken for. Rs 71,594/- is owed. School fees arrive in terms. The recurring deposit at Rs 64,000/- has its own maturity. Money can only be committed once, and a rupee put behind a lock-in is a rupee that is not available for the payment that is already scheduled.

The fourth, and only now, is what the scheme itself currently says: the rate, the smallest and largest amounts, the length, what happens at the end, and every condition attaching to any of it. Most people meet this decision in the reverse order: the first three questions are answered by looking at the household itself, and only the fourth by looking at the scheme.

Who actually reads this arithmetic, and what do they do with it?

Four kinds of people meet these numbers in ordinary life, and each of them is asking a different question of the same balance.

The first is the household choosing an amount. The table above already answered what this will grow to, on stated assumptions. The useful question is what happens in the month I cannot pay it. The table only works if the rows actually happen, so an amount that survives a thin month is worth more than a larger amount that gets abandoned in year three. The best amount is the largest one that would still have been paid in the worst month of the last three years.

The second is a lender looking at a household's papers. A locked balance is genuinely reassuring about the kind of household a lender is looking at. No payment can be made out of it, so it is not a repayment source. Somebody assessing capacity separates the two the way the failure block above did, and a household that has already made that separation on its own sheet finds the conversation easier.

The third is somebody with no payslip at all: most self-employed work in this country, and Ashok Bhosale's position exactly. There is no employer adding a second contribution beside his, nothing is deducted before the money reaches him, and no statement arrives to remind him of anything. Every rupee that goes in is a decision he has to make again, twelve times a year, sometimes against a week the lane was quiet. The mechanism is identical for him and the difficulty is not, and any account of this scheme that leaves that out has described only the easier half of the country.

The fourth is whoever eventually has to find the paperwork. An account like this outlives jobs and moves, and the practical work is unglamorous: knowing where the passbook or the login is, keeping the nomination current, and being able to say what the last statement showed without hunting. Doing it takes an afternoon; not doing it can cost months at the worst possible time.

One last observation about why a goal like this gets postponed belongs to Richard Thaler. People weigh outcomes far in the future far more lightly than outcomes close to hand, and that is ordinary human attention rather than carelessness. A fifteen year account is exactly the sort of thing that loses that contest to a bill due on Friday, every single week. A written sum does not fade with distance the way an intention does, so written arithmetic is the only reliable counter.

What kind of figure can never be stated as fact, and why?

The rate is the clearest case. A rate is declared for a period, so a rate written down anywhere outside that period is a figure with no expiry date on it, and today's, last year's and an average of any stretch of them are equally unsafe to carry. The 7 per cent above is an assumption, labelled as one everywhere it appears, including at every setting of every control in the simulation.

The same holds for the term, the lock-in period, the smallest or largest amount that may go in, whether the account can be extended, what may be taken out early, and how any of it is treated for tax. Each of those figures is what a household acts on, and each one moves. Naming the authority that publishes them is the only version that stays correct.

Reading a statement line by line is a separate skill from understanding the account, and is covered separately. A statement carries dates, an opening balance, each deposit with the date it was credited, the year's credit and a closing figure.

And no arithmetic says whether this household, or any other, should put money in. The mechanism, one worked arithmetic on stated assumptions, both readings of the lock-in left unranked and four questions in an answerable order are what the decision has to be made with, and the decision itself belongs to the household.

The Bhosale household, Meghna Bhosale, Ashok Bhosale, Ira Bhosale and Sahyadri Freight Services Private Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.