Globalisation: The Trend, the Reversal and the Consequences

Globalisation is the rising share of what a country makes, buys, finances or staffs across a border rather than inside one. Globalisation is therefore three separate shares, not one mood, and the three move independently. The falling cost of shifting goods, money and information lifted all three. Only some of those costs stopped falling, so any turn is partial.

Two things already established carry the weight underneath that sentence. The first is comparative cost: the reason two countries both end up better off is that each gives up less to make the thing it ends up making, and being better at something has nothing to do with it. The second is output, the single total that everything a country produces in a year adds up to. A share needs something underneath the line, and output is what goes there. A share has a numerator and a denominator, and once both are named, most of what gets said about globalisation stops being a claim at all.

What is globalisation, once it has to be measured?

The awkward part comes first. Globalisation is a word that gets used as though it named a condition, the way humid names a condition. It does not. The word names a direction of travel in a set of ratios, and until the ratio is named, nothing has been said that could turn out to be right or wrong.

There are three ratios worth separating, and they are separate because they have different things on top and different things underneath.

The first is the goods and services share: the value of what crosses the border, set against the value of everything the country produced. If Sankhya, an invented country, sells onions to Marut and buys machines back, both directions of that traffic sit on top and the whole of Sankhya's output sits underneath. The second is the money share: the value of investment arriving from outside in a year, again set against output. Somebody in Marut buying a workshop in Sankhya, or lending to one, moves that ratio and touches the first one not at all. The third is the information share, and it is the one people forget. The information share counts the part of what Sankhya uses that was worked on somewhere else and delivered down a wire: the accounts kept elsewhere, the drawings made elsewhere, the helpdesk answering from elsewhere. Nothing physical crosses anything.

The three shares have different numerators, different denominators and no obligation to move together, so a claim about globalisation that does not name which of the three it means is not a claim about anything. Every difficulty in the subject follows from that one fact.

The household version runs the same way. Consider a household wanting to know whether its street has become more connected to the rest of the city. One count is how much of what the street eats is grown outside it. Another is how much of the money invested in its shops came from outside it. A third is how much of the work done on the street is done for customers who live nowhere near it. Three honest counts, three different answers, and somebody who quotes one of them while meaning all three will mislead a listener without ever saying anything false.

One term is needed before the next block. A supply chainThe ordered set of suppliers whose work has to happen before a finished thing exists: the ore before the steel, the steel before the part, the part before the machine. Each step may sit in a different country. is just the ordered list of everyone whose work has to happen before a finished thing exists, and the interesting question about any of them is how many of the steps sit outside the country doing the finishing. The count of outside steps is really a version of the goods share, looked at from inside one product rather than from above a whole economy.

What actually fell to make any of this possible?

Three costs fell, by wildly different amounts, and keeping the three apart is most of the work.

The cost of moving a physical thing fell a long way. Bigger vessels, standard boxes that never get unpacked between the field and the buyer, ports that turn a ship around in hours. Real, large, and still a cost: a quintalA unit of weight equal to one hundred kilograms, used across Indian agricultural markets. Onion volumes are given in quintals throughout, so nothing has to be inferred from a bare number. of onions is heavy, it goes stale, and no amount of engineering makes moving it free.

The cost of moving money fell further. A payment that once needed a correspondent bank, a paper instruction and a week now needs a message. Money has no weight and does not go stale, so almost the entire cost of shifting it was the cost of the paperwork around it, and that is exactly the kind of cost that collapses.

The cost of moving information fell to almost nothing, and the word almost is doing very little work. Sending a drawing, a set of accounts, a design, a conversation or a diagnosis across the world costs so close to zero that nobody meters it any more. Work that had always been thought of as immovable then turned out to be a tradeable serviceA service that can be sold to a buyer in another country without either side travelling, because what actually moves is information rather than a person or a thing. Bookkeeping and design qualify; a haircut does not. after all.

The three costs fell at different rates and by different amounts, and that single fact is why any reversal is uneven: what became nearly free to move has stayed nearly free to move. A policy can raise the cost of landing a container. The courier is not coming back, so nothing raises the cost of a drawing crossing a wire back to what it was in an era of couriers.

The same shape appears in an ordinary household over a generation. The vegetables come from further away than they did, and if fuel gets dearer they might come from nearer again. But the video call to a cousin in another country is not going to become expensive, whatever anyone does. Cheapness there came from the disappearance of a physical constraint, not from a subsidy.

Which parts have turned, and which have not?

Careless writing does the most damage here, so take it slowly and keep the three ratios apart.

The goods share can slow, stall or fall. Freight can get dearer. A route can close. Restrictions can be put on landing certain things, and what those restrictions are and who ends up paying for them is a subject of its own, taken up in the notes on trade barriers. A country can decide it wants a step of a chain back inside its borders, a move that goes by the name of reshoringBringing a step of production back into the country that had previously bought it in from abroad. The word describes the direction of the move and says nothing about whether the move was wise.. All of that is real and all of it presses on one ratio.

The money share moves for reasons of its own that have very little to do with containers. Inward investmentMoney arriving from outside the country to buy, build or lend into something inside it. Capital flows sets out why it comes and when it turns around. responds to what returns look like here against elsewhere, and it can turn around faster than any cargo can.

The information share is the one that has kept climbing through periods when the goods share did not. Nothing about a container ship touches a set of accounts prepared in one country for a business in another. A firm that cannot get a part shipped can still get the drawing for it, so the pressures that slow the movement of physical things push more work down the wire.

Calling the whole thing a reversal treats three trends as one, and the three can and routinely do move in opposite directions in the very same year. The distinction is not a quibble about wording. The difference is between a claim that can be tested and a mood that cannot.

Can a country gain from trade while people inside it lose?

Yes, and both halves of that answer deserve stating plainly, neither one softened.

The first half is already established. Each country concentrates on what it gives up least to make, so comparative cost showed that Sankhya and Marut can both end up with more than they started with. The result is about the country as a whole. Comparative cost states something about a total, and a total says nothing whatever about who inside it is holding which part.

Now put the second half beside it. When Sankhya starts selling onions abroad, the demand facing onion growing rises. More onions get grown, more field work gets done, and whatever Sankhya uses heavily to grow onions becomes more valuable. When Sankhya starts buying machines from Marut instead of making them, demand facing Sankhya's own machine making falls, and whatever that work used becomes less valuable. Trade raises the return to what a country uses intensively in what it sells and lowers it for whatever competes with what it buys. The two effects arrive together, from the same cause, in the same year.

Work it on the numbers rather than asserting it. One worker-yearOne person working for one year. A way of measuring how much work a task takes without needing wages, headcounts or hours. The measure suits a question about the size of a task rather than the pay for it. in Sankhya makes 100 quintals of onions or 2 machines. Sankhya buys 2,50,000 machines from Marut. Building that many at home, at 2 machines a worker-year, would have taken 1,25,000 worker-years. Buying them instead costs onions: at the 25 quintals a machine Sankhya trades on, 2,50,000 machines come to 62,50,000 quintals, and growing that many at 100 quintals a worker-year takes 62,500 worker-years. Sankhya gives up 50 quintals to build a machine and hands over 25 to buy one, so the same machines cost half the work.

So 62,500 worker-years are released. The released work is the national gain in a form that can be counted, and it is not a rhetorical gain: the country holds the same 2,50,000 machines at the end of the year and has 62,500 worker-years of effort left over to point at something else.

Now the part that a gain sentence on its own would bury. The 1,25,000 worker-years of machine building did not happen. Field work appeared instead, and the field work is smaller: Sankhya actually grew and shipped 50,00,000 quintals, or 50,000 worker-years of growing. So 1,25,000 worker-years of machine-shop work stopped and 50,000 worker-years of field work started.

Both facts are true at the same time: the country is 62,500 worker-years better off on the same machines, and 1,25,000 worker-years of machine making has gone against 50,000 worker-years of growing that arrived, and an account that states only one of those two is doing politics rather than economics.

The reason it hurts is not the arithmetic. The 1,25,000 and the 50,000 are different people. An onion field is not a machine shop, it is very often not in the same district, and a fitter of forty five does not become a grower of onions because a national total came out positive. The mismatch is the whole of the distribution problem in one sentence, and no amount of national gain dissolves it.

One loose end, and it is worth naming rather than hiding. The prices asked for 62,50,000 quintals and Sankhya shipped 50,00,000. The missing 12,50,000 quintals are 12,500 worker-years of growing that the year did not do. The gap has a rupee name, and the share worked below gives it one.

Whether that loss is a price worth paying is a position rather than a finding, and so is the contrary view. The reason to state the arithmetic carefully is precisely so that whoever argues either side has to argue against the same numbers.

Why does the distribution decide the politics?

Because of a lopsidedness in how the two sides are felt, and it is worth putting numbers on rather than gesturing at.

Take the gain first. Sankhya's own opportunity costWhat is given up to get something, measured in the thing given up rather than in money. Comparative cost sets it out in full. of a machine is 50 quintals of onions, because a worker-year makes either 100 quintals or 2 machines. At the terms of tradeThe rate at which one country's exports exchange for its imports. The rate moves for reasons set out under the terms of trade, and the same movement is not symmetric in both directions. Sankhya actually trades on, 25 quintals per machine, it gets a machine for half of that. So the 2,50,000 machines that would have taken 1,25,000 worker-years to build cost 62,50,000 quintals to buy, or 62,500 worker-years of growing. Sankhya gets the same machines for half the work, and 62,500 worker-years are released to do something else entirely.

Now spread the two sides over the people they land on. Sankhya has about 20 crore people. The 62,500 released worker-years, spread across all of them, come to one released worker-year for every 3,200 people. Nobody feels that. Nobody could feel that. The gain arrives as machines being slightly cheaper, and through them, as whatever those machines make being slightly cheaper, in a hundred purchases nobody itemises.

The loss is arranged completely differently. The loss lands on 1,25,000 workers as one whole livelihood each. Every one of them knows exactly what happened, exactly when, and exactly what caused it. The 1,25,000 already work together, so they can find each other. For them the matter is everything, so they have every reason to spend time and money on it.

The loss is roughly 3,200 times more concentrated than the gain. The losers therefore organise, the winners never notice, and neither side is being irrational. The winner who does not campaign for cheaper machines is behaving perfectly sensibly, because the campaign would cost more than the gain is worth to any single one of them. The loser who does campaign is behaving perfectly sensibly too.

The same shape appears on a single street. A proposal to allow a large shop at the end of it spreads a small saving across every household and takes a whole livelihood from four shopkeepers. The four turn up at every meeting. The three hundred households do not turn up at any, and not one of them is being lazy or foolish. Each household stands to save the price of a few bus tickets a year, and a meeting takes an evening.

What does the trade share actually come to, worked on Sankhya?

Now put a number on the goods share, and then be honest at once about what the number is.

Sankhya sells 50,00,000 quintals of onions at Rs 2,000/- a quintal, or Rs 1,000 crore. Sankhya buys 2,50,000 machines at Rs 50,000/- each, or Rs 1,250 crore. Set those against Sankhya's output for the year of Rs 17,47,200 crore, and the goods share of that one pair is small enough to be startling.

Onions and machines are one product pair, and a real trade share counts everything a country buys and sells, so 0.1288 per cent demonstrates the arithmetic and nothing more. Sankhya does not trade in onions and machines alone, any more than a household lives on onions and machines alone. Every other traded thing is missing from that table. The number is a rounding error rather than a reading.

A share is only as complete as its numerator. Someone who computes a trade share from the two products they happen to have data on has not computed a small trade share. The number they computed is not a trade share at all, and the difference becomes invisible once the working is thrown away.

The last row of that table is the trade balance for the pair, and it is negative. The trade balance and the trade deficit set out what a negative balance does and does not mean, and why reading it as a shortfall is the standard error. Here it is a subtraction and nothing else.

But look at the same subtraction in quintals. Counting in quintals closes the loose end left above, and is the more instructive way round. Rs 50,000/- divided by Rs 2,000/- is 25, so the prices say a machine is worth 25 quintals. At that price, 2,50,000 machines should have been paid for with 62,50,000 quintals. Sankhya actually shipped 50,00,000 quintals for those 2,50,000 machines, or 20 quintals a machine delivered against 25 quintals a machine priced. The gap is 5 quintals on every machine.

Five quintals short on each of 2,50,000 machines is 12,50,000 quintals, and 12,50,000 quintals at Rs 2,000/- is Rs 250 crore, the balance on the last row of the table to the rupee. That is worth pausing on, because it means the deficit is not only a rupee figure. The deficit is a quantity of onions that the prices asked for and the year did not grow. At 100 quintals a worker-year those quintals are the 12,500 worker-years left dangling above. How the gap gets settled, and by what, is the external account.

What separates a pause from a reversal?

Three tests, in order, and each one throws out a different kind of false alarm.

The first test asks whether the share is falling or merely growing more slowly. Falling and slowing are constantly confused and they are not close to the same thing. A share that goes 11.00, 11.30, 11.50, 11.60 is rising throughout and rising by less each period. Nothing has reversed. Something has decelerated. Deceleration is a fact about the second difference and not about the direction. If the words falling and slowing get used interchangeably, half the argument on this subject is people disagreeing about which one they saw.

The second test asks whether it is falling in one of the three or in all three. A fall in the goods share alongside a rise in the money and information shares is a change in the shape of a country's connection to the outside, not a retreat from it. Work that used to travel as a crate now travels as a file. Only a fall across all three supports the sentence people usually want to write.

The third test asks whether the fall survives a cycle. Trade volumes move hard with the state of the economy, and the things that cross borders most, machinery, materials, parts for other things, are exactly the things whose purchase is easiest to postpone in a bad year. So a downturn produces a falling goods share all by itself, and a recovery produces a rising one, and neither is evidence of anything structural. Which phase an economy is in, and how to tell, is set out in the notes on the business cycle, and it is the correct thing to check before calling any single weak year a trend.

One weak year inside a cycle is not a trend, and a fall in one of three shares is not a fall in the thing, so a claim that survives all three tests is a much rarer claim than the ones usually made.

What does an analyst actually check when a route closes?

Watch someone competent handle this and there is nothing sweeping about it. She does not ask whether globalisation is reversing. She asks two narrow questions about one business, and both have answers.

The first: which of this business's inputs cross a border at all? Not the finished goods, the inputs. A firm can look entirely domestic on its sales side and still stop dead if one component comes from one place. She works through the list of what has to arrive before anything can be made and marks each line as arriving from inside the country or outside it.

The second, and the one that does the work: for each line that crosses a border, how substitutableHow easily one supplier can be swapped for another without redesigning, re-testing or re-approving the thing being made. High substitutability means a route closing is an inconvenience; low means it is a stoppage. is it? One overseas source and five overseas sources are not a difference of degree, they are two different businesses. When a route closes, the second one telephones the other four and loses a week. The first one stops, and how long it stops for depends on how long it takes somebody else to be approved to make that part. For anything precise, approval is measured in quarters rather than days.

Exposure is decided not by how much of a business supply crosses a border but by how easily each crossing line could be replaced. The count of alternatives settles it, not the size of the flow.

A food stall makes the same point without any of the vocabulary. Two stalls both buy tomatoes from outside the city. One has a single supplier who brings a crate every morning. The other knows four traders in the wholesale market and buys from whoever is cheapest. The trade share of the two stalls is identical. On the morning the road floods, one is trading and one is not.

A lender runs the same test from the other side of the table. Asked to fund a business whose margins depend on a part from one distant supplier, the useful question is not what the tariff situation is. The useful question is what happens to twelve months of cash flow if that one line stops for a quarter, and whether the borrower has ever had to find out.

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