Capital Flows: Why Money Crosses Borders and When It Reverses

Money crosses a border to buy a return it cannot get at home, after allowing for what could go wrong and for what it would cost to get back out. How fast a flow can leave is what separates a useful inflow from a dangerous one, so the kind of flow matters more than the amount. A factory cannot be withdrawn on a Tuesday. A deposit can.

Two things already established are holding that up. The first is trade itself. When Marut takes 50,00,000 quintals of onions from Sankhya, both of them invented economies, at Rs 2,000/- a quintal and sends back 2,50,000 machines at Rs 50,000/- apiece, goods travel one way, money travels the other, and every one of those crossings is finished with before the year closes. The two shipments are Rs 1,000 crore going out against Rs 1,250 crore coming in on one product pair, not the whole outside trade of either economy, and the Rs 250 crore gap between them is a difference in goods before it is ever a difference in rupees. The second is the policy rateThe interest rate a central bank sets, which pulls the rates that banks and borrowers actually face along with it. Set out separately under monetary policy.. The gap between what money earns in one place and what it earns in another is one of the reasons money goes travelling. Capital flows are a different kind of traffic altogether: money that crosses not to pay for a crate of anything, but to stay for a while and earn something. And money that stays for a while is money that has to decide, eventually, whether to keep staying.

Why does money cross a border at all?

Money crosses for one reason and then survives two tests. The reason is a better return. The two tests are what could go wrong, and what it would cost to get back out again. Most people get the first two and skip the third. The third is where the trouble is.

Start with the reason, on numbers small enough to follow. A lender sitting in Marut can put money to work at home and earn 4.00 per cent a year. Lending the same money into Sankhya, where there is a shortage of money to build with and therefore a queue of people willing to pay for it, earns 9.00 per cent. The gap is 5.00 points, and a gap is the entire motive. Nothing else needs to be true. The lender does not have to admire Sankhya, believe in its future or know where it is on a map.

Now the first test. Some of that 5.00 point gap is not profit, it is payment for a chance of loss. Suppose the lender prices that chance at 2.00 points a year. The gap after the pricing is 3.00 points, still an advantage, and on that basis the money moves.

The second test is the one that gets skipped. Getting in and getting out are not the same transaction and they do not cost the same. If the lender ever has to get the money out in a hurry, and the price of doing that in a hurry is 6.00 per cent of the amount, then a holding that ran for a year at a 3.00 point advantage comes back 3.00 points short. The advantage was real. The advantage was simply smaller than the exit.

Money crosses a border on a return, stays on a price for what could go wrong, and leaves on the cost of leaving. Exit cost decides the difficult part. Hold on to that, because every kind of flow below is a different answer to the question of what the exit costs.

Here is the household version, and it is closer than it looks. A household has some savings. A cousin two towns away is expanding a workshop and offers a better return than the bank. The first test is what happens if the workshop struggles, and the household weighs it. But the decision actually turns on whether the money can be got back the month the roof leaks. Money in a bank comes back on Tuesday. Money in a cousin's workshop comes back when the workshop can spare it. The month the roof leaks is exactly when it cannot.

What kinds of flow are there, and why does the kind decide more than the amount?

Sorting the kinds by name produces a list to be memorised. Sorting them by how long it takes to get the money out derives the same list, in the right order, and removes the need to memorise it at all. Exit time gives the ordering that follows.

The slowest kind buys a real thing or a controlling share of one. Somebody in Marut builds a cold store in Sankhya, or buys enough of an existing one to run it. To reverse that, the holder has to find another buyer for a cold store, agree a price, and wait for the paperwork. There is no button. In a bad month there may be no buyer at all, or none at a price worth taking. The exit gets expensive exactly when it is wanted. Reversing this kind is measured in months and often in years.

The middle kind buys a claim that trades: a share, a bond, a unit in something. The holder never has a cold store at all, only a piece of paper saying that somebody else has one. Because that paper changes hands on a secondary marketA place where a claim that already exists changes hands between two holders. The original issuer gets nothing from the sale and often does not know it happened., the exit is a sale, and a sale can happen in an afternoon. The cold store does not move, does not close and does not notice. Only the name on the register changes, and the money walks out of the country.

The fastest kind is not sold at all. Short term money is lent for a short period, and it leaves by simply not being lent again. When money is placed in Sankhya for three months, the holder does not need a buyer, a market or a decision from anybody else. On the date the three months end, the holder does nothing, and doing nothing is the exit. A date that arrives on its own is why maturityThe date on which a loan or a deposit comes to an end and the money is due back. Short maturity means that date arrives soon and arrives often. is the single most useful thing to know about any borrowed money: it gives the date on which somebody gets to change their mind for free.

The ranking is the ordering, so putting the kinds in order of how long the exit takes gives the risk ranking without memorising anything. Slow to reverse is safer for the country receiving it, fast to reverse is not, and the reason is not that one sort of money is better behaved. The ranking is arithmetic about doors.

To put numbers on the ordering, take it throughout that of money that bought real things, 2 per cent could realistically be out inside one quarter; of money in traded claims, 60 per cent; and of short term lending, 80 per cent. The three figures are assumptions chosen so the arithmetic stays visible, and a real market would put every one of them somewhere else.

What actually makes a flow turn around?

Three things, and they are worth keeping separate because a country can do something about the first two and nothing whatever about the third.

The first is a better return appearing somewhere else. Nothing has to change in Sankhya at all. If the return available in Marut rises, the gap that pulled the money into Sankhya narrows, and at some point it closes. The money then leaves not because Sankhya got worse but because the alternative got better. From inside Sankhya the two look identical.

The second is a worse risk appearing here. The chance of loss gets repriced upward, the 2.00 points that the lender was charging for it becomes 4.00, and the advantage disappears from the other end. People expect this kind of reversal, because the country receiving the money did something, or had something done to it.

The third is the one that surprises people. The money is needed at home, for reasons that have nothing to do with Sankhya, nothing to do with Marut, and nothing to do with the return on either side. A holder facing losses somewhere entirely different sells what can be sold to cover them, and what can be sold is not the failing holding but whatever has a buyer. So the well run holding in the quiet country goes first, precisely because it is the easiest to convert into money without a fight. Ease of sale has a name worth having, liquidityHow quickly something can be turned into money without having to accept a much worse price for hurrying. A thing can be valuable and still be hard to sell, and the two are separate questions., and it cuts both ways: the same quality that made the holding easy to buy makes it the first one sold.

The third reason for a reversal never asks anything about the country it is leaving, so a country can run its affairs impeccably and still watch money leave. That is not fatalism. The third reason is why the assessment below looks at the shape of what has arrived rather than at how well anything is being run.

The third reason turns up in an ordinary street. A wedding is coming in one household and the money has to be found. So they call in the small sum they lent to a neighbour six months ago. The neighbour has done nothing wrong, the loan was never in trouble, and the money still goes. A neighbour trying to work out what he had done to deserve it would be looking in entirely the wrong house.

Why is the same money both the funding and the exposure?

Because it is one flow, looked at on two different days, and the two days are usually not far apart.

On the day it arrives, money from outside pays for things Sankhya could not otherwise pay for. Sankhya saves a certain amount each year and can build only what that saving covers. Money arriving from elsewhere lifts the ceiling. A cold store gets built that was not going to be built, a road gets financed, a workshop buys the machine it has been putting off. A lifted ceiling is not a technicality. The ceiling is the entire reason a country with more ideas than saving wants the money at all, and an inflow read purely as a hazard has been read wrong.

On the day it leaves, the same money is a claim being exercised. The cold store still stands, but the money that paid for it is being asked for back, and Sankhya has to find it from somewhere. Here is the part that makes it bite: the moment when a large amount wants to leave is very rarely a calm one. Money leaves when something has gone wrong, either here or somewhere else, and something going wrong is exactly the moment when the money is hardest to replace.

Funding and exposure are not two flows, one helpful and one harmful; they are one flow seen at two points in time, and the same feature that made it useful on arrival is what makes it dangerous on exit. The feature is that it belongs to somebody else and it can be recalled.

A household knows this shape without any of the words. Borrowing to buy a scooter gets the borrower to work for two years, and the loan is not a mistake. But the month the work dries up is the same month the instalment is hardest to pay, and the lender does not adjust the date to suit the borrower. The scooter was funding. The instalment is exposure. Nobody took out two loans.

What does Rs 5,000 crore look like when the same amount arrives two different ways?

Take one inflow into Sankhya of Rs 5,000 crore for the year and examine it twice. Everything about it is identical across the two readings except the kind, and the kind is the only thing that ends up mattering. For scale, a year of Sankhya output runs to Rs 17,47,200 crore, so the inflow is about 0.29 per cent of it. A share that small is nobody's headline, and it is still large enough to hurt if it left in a week.

Reading one. The whole Rs 5,000 crore arrives as money buying real things: cold stores, a stake in a machine works, land with a shed on it. On the assumption used here that 2 per cent of that kind could realistically be out inside one quarter, the amount that could leave in a quarter is Rs 100 crore. For the other Rs 4,900 crore to go, somebody has to be found who wants to buy a shed in Sankhya, and finding that somebody takes what it takes.

Reading two. The whole Rs 5,000 crore arrives as short term lending, placed for three months at a time. On the assumption used here that 80 per cent of that kind falls due inside the quarter, the amount that could leave in a quarter is Rs 4,000 crore. Nobody has to be found. No market has to open. On the maturity date, the holder declines to lend again, and Rs 4,000 crore is due back.

The amount was identical to the last rupee and the exposure was forty times larger in one reading than the other. Read the mix, not the total. A headline that says Rs 5,000 crore arrived says nothing that distinguishes the two rows above, and the two rows above are not small variations on each other. The two rows are different situations wearing the same number.

How to Assess Capital-Flow Risk in Emerging Markets: what is actually looked at?

Assessing capital-flow risk is a method, and only a method. The method is general. Every reading that fills it in belongs to a particular market and a particular date. A method names the four things to go and find, and what each one would mean once found. Filling them in is the analyst's work, on figures taken from a body that publishes them, on a date recorded at the time.

The first thing to look at is the mix, not the total. Split what has arrived into the three kinds and compute what share of the whole could leave inside a quarter. The derived share carries more than the headline does. The same Rs 5,000 crore gave Rs 100 crore and Rs 4,000 crore depending only on the mix.

The second is how much of it is short term, counted by when it falls due rather than by what it is called. Money lent for three months is short term whoever lent it and whatever the paperwork calls it. Maturity is a question about dates on a calendar, and it has an exact answer.

The third is what the money paid for. Money that built a cold store leaves behind a cold store, and the cold store goes on earning after the money that paid for it has gone. Money that paid for a year of consumption leaves behind nothing to earn from, and the claim still has to be met. Same inflow, same repayment, and one of them has an asset standing next to it.

The fourth is what is held against a sudden outflow. A country holding a stock of foreign money can meet a wave of departures out of that stock instead of out of whatever it can grab. The stock matters only against the amount that could go in a quarter, and the word for the stock is reservesA stock of foreign money and similar assets held by a country's central bank, kept so that payments abroad can be met when they are demanded rather than when they are convenient.. Related, and separate, is whether what has been borrowed is owed in a different money from the money the borrower earns. Borrowing in one money and earning in another is a currency mismatchOwing money in one currency while earning it in another, so that a move in the exchange rate changes the size of the debt without anybody borrowing anything more. , and it turns an exchange rate move into a bigger debt without any new borrowing.

Running those four produces a reading. A reading is not a verdict. The step from a reading to a verdict is where all the argument lives and none of the arithmetic does, so anybody who takes that step quickly is worth suspicion.

What can a country do about flows that move this fast?

Three broad things, and every one of them costs something.

A country can hold a stock of foreign money against the possibility of a wave of departures. The stock works in the way a cash box in a shop works: the money is there when it is needed. A stock held for a bad day is also money not being used on a good one, so it costs, and somebody pays for the difference every single year.

A country can lengthen the dates on what it has borrowed, leaving less of it to fall due in any one quarter. Borrowing for five years instead of three months moves the exit date away, and that is exactly the point. Lenders charge more to be locked in longer, so lengthening costs too, and a country that lengthens everything pays a premium every year for a reversal that may not come.

A country can restrict some flows, either on the way in or on the way out. Restriction is the most argued about of the three, and the arguments are not silly on either side. Restrictions do reduce the amount of money that can leave in a hurry. Restrictions also reduce the amount that arrives in the first place, push some of what does arrive into forms that are harder to see, and are difficult to remove once put in place.

Each of the three works, each of the three costs, and reasonable people disagree about all three. Picking between them is a judgement about what a society is willing to pay for, not a result arithmetic hands over.

What does a lender watch once the money has already arrived?

A credit officer reading one of these works remarkably narrowly. She is not interested in the inflow. She is interested in a calendar.

Her question is which amounts have to be found again, and on what dates. Nobody comes and asks for a cold store back on a Tuesday, so money that bought one never has to be found again. Short term lending has to be found again on every maturity date, over and over, and each of those dates is an opportunity for somebody to say no. The act of finding it again is refinancingPaying off borrowing that has come due by borrowing again, usually from the same lender. Refinancing looks like nothing happening, so it is easy to forget that somebody decides it each time., and it looks like nothing happening right up until the moment it does not happen.

So she builds a calendar. Take the second reading of the Sankhya headline, the one where Rs 500 crore bought real things, Rs 1,800 crore sits in traded claims and Rs 2,700 crore is lent short term. Of that Rs 2,700 crore, on the assumption used here, Rs 2,160 crore falls due inside the first quarter, and the remaining Rs 540 crore is spread over the rest of the year as Rs 300 crore, Rs 140 crore and Rs 100 crore. If seven rupees in every ten get renewed and three do not, Sankhya has to find Rs 648 crore in that first quarter, from somewhere, at whatever it then costs.

A reversal is not felt as money flying out of a window; it is felt as a refinancing that does not happen, on a date that was in the calendar all along. Which is why the useful question is never how much arrived, but how much has to be agreed to again, and when.

One detail on that calendar catches even people who have learned to read maturity dates. The Rs 1,800 crore in traded claims has no date at all. Traded claims do not wait for a quarter to end. A sale goes through on any morning somebody decides to sell. Dated money at least says in advance when it is going to ask, and in that one sense the undated money is worse.

The Republic of Sankhya and Marut are invented.
Educational material. Not advice on any investment, tax, budget or market position.