Trade Barriers: Tariffs, Non-Tariff Restrictions and Who Actually Pays

A tariff is a tax on an imported good, charged at the border and handed over by the importer. Who actually pays it is a separate question, settled by which side can walk away from the deal. A non-tariff barrier restricts trade with no tax at all, through a quota, a standard, a licence or a delay. Restriction of that kind is far harder to see and to measure.

Two ideas already established do most of the work here. The first is pass-through: when a new cost lands on somebody selling something, only part of it usually reaches the price tag, and the part that never reaches the tag stays with the seller. The second is elasticity of demandhow sharply the quantity people buy falls when the price goes up. Where it is low, buyers keep buying at almost the same volume; where it is high, a small rise sends them away., which is what decides how much of the cost the buyer will tolerate before going somewhere else.

Here is the everyday version, before any arithmetic. Suppose the municipality puts a new annual charge on rented flats and sends the bill to the landlord. Whose money is it in the end? If there are twenty empty flats on the same street, the landlord cannot raise the rent by a single rupee without losing the tenant, so the landlord swallows the charge. If the flat is the only one within walking distance of the station and four people are waiting for it, the rent goes up by the full amount and the tenant swallows it. The bill has the landlord's name on it in both cases. The name on the bill and the identity of the person who ends up poorer are two different facts, and a trade barrier works exactly the same way.

What is a tariff, and where in the journey is it collected?

A tariff is a tax charged on a good because it crossed a border. Nothing more exotic than that. The charge is assessed when the goods are presented for clearance, and it is assessed against the landed valuethe value of the goods once they have reached the importing country's border, which normally means the price on the invoice plus the cost of getting them there. It is the base figure the customs authority works from. of the consignmentone batch of goods travelling together under one set of shipping documents. A single order may arrive as several consignments, and each is cleared separately.. The importer is the one standing at the counter with the paperwork, so the importer hands the money over.

Tariffs come in two shapes. The shape changes what happens when prices move, so both are worth knowing. An ad valoremLatin for according to value. A charge expressed as a percentage of the value of the thing, so the amount payable rises and falls with the price. tariff is a percentage of value, so a rise in the invoice price raises the amount payable. A specific tariff is a flat sum per unit, per tonne or per piece, so it does not move when the invoice price does. The worked case below holds the amount fixed at a flat sum per machine. Only one thing changes at a time, and the arithmetic stays honest.

The case worked all the way through is this one. The Republic of Sankhya, invented for teaching, buys machines from its trading partner Marut, also invented. One machine reaches the Sankhya border with a landed value of Rs 50,000/-. Sankhya applies a duty of twenty per cent. Twenty per cent of Rs 50,000/- is Rs 10,000/- for the machine. The Rs 10,000/- is handed over by the Sankhya dealer clearing the consignment, and the receipt has the dealer's name on it.

Who hands the money over at the border is a fact about administration, and it settles nothing at all about who is worse off once the dust clears. The dealer will try to recover the Rs 10,000/- from the buyer by charging more, and will also try to recover it from the exporter in Marut by demanding a lower invoice next time. Whether either attempt works is not decided at the counter. The market decides it, by who has somewhere else to go.

Who actually pays the tariff, once the arithmetic is worked through?

Who actually pays is the question the whole subject turns on, and it deserves numbers rather than a slogan. The share of a new cost that reaches the buyer's price is the pass-through share. The name matters less than the fact that it can sit anywhere from nothing to everything, and the same duty gives a completely different answer at each end. Working the Sankhya machine at three settings shows what moves.

At full pass-through the buyer meets the whole Rs 10,000/-. The machine that used to cost Rs 50,000/- now costs Rs 60,000/-, and the exporter in Marut still receives Rs 50,000/-, exactly what it received before. At sixty per cent, the buyer meets Rs 6,000/- of it and pays Rs 56,000/-. The exporter accepts Rs 46,000/- instead of Rs 50,000/- and has absorbed Rs 4,000/-. At nil pass-through, the buyer pays Rs 50,000/- exactly as before and notices nothing at all. The exporter receives Rs 40,000/- and absorbs the entire Rs 10,000/-.

The bottom row is the spine of the whole subject and repays a slow reading. What the exporter receives plus what the authority collects always equals what the buyer hands over. Rs 50,000/- plus Rs 10,000/- is Rs 60,000/-. Rs 46,000/- plus Rs 10,000/- is Rs 56,000/-. Rs 40,000/- plus Rs 10,000/- is Rs 50,000/-. The check is easy to over-read, so it is worth being exact. The check is an identity, true by construction. The three amounts were defined so that two of them add to the third, so the check can only ever show that no rupee has been dropped or counted twice. The identity confirms the arithmetic is complete. The identity confirms nothing about who bore the cost, and it would hold just as neatly if every figure in the row were wrong in the same direction. The identity does establish something real all the same. The two sides are forced to share a fixed quantity of pain between them. Every rupee the buyer does not pay is a rupee the exporter gives up, and the reverse.

All three rows carry the same duty, the same rate, the same base and the same collection point, and the only thing that differs between them is who could afford to walk away from the deal. That is why the collection point is useless as evidence. The collection point is identical in a case where the buyer carries everything and in a case where the buyer carries nothing.

Now turn the same arithmetic the other way and look only at the Rs 10,000/-. The Rs 10,000/- is fixed. The line drawn through it is what moves, and wherever the line falls, the two parts still add to Rs 10,000/-. A fixed amount cut by a moving line is the cleanest way to see why the revenue figure a treasury reports is silent on the question everybody argues about. The authority collects Rs 10,000/- whether the buyer bore all of it, none of it, or any mixture in between, so the revenue number can never be used as evidence of who carried the cost.

What decides how the burden splits between the buyer and the exporter?

The split is not chosen by anybody. The split falls out of two facts about the market, and both of them are versions of the same question: who has somewhere else to go?

Start with the buyer. If the machine has a close substitutea different product the buyer would accept instead, closely enough that a price rise on the first one sends the buyer to the second. The closer the match, the less room the seller has to raise a price., made at home or bought from a country facing no duty, the buyer has an exit. The moment the imported machine gets dearer, the order moves. A dealer who tries to add the whole Rs 10,000/- to the price simply loses the sale, so the price cannot rise by much and the pass-through share stays low. Suppose instead that nobody inside Sankhya builds anything like the machine and no other country sells one. The buyer has no substitute and no exit. The buyer grumbles and pays.

Now the exporter in Marut. An exporter with a queue of other customers in other countries does not need to cut its price for a Sankhya buyer, and will not. The exporter lets the price rise, and the buyer carries the duty. If the exporter's factory is running below capacity and Sankhya is its main market, a lost order is a genuine loss, so it will shave its invoice to keep the machine moving. It absorbs.

The two together give four situations rather than one. No single sentence about who pays a tariff can be right for every case. A duty on something with no substitute is paid almost entirely by the buyer, whatever anybody says about who the measure was aimed at. The everyday test is the one from the flat above, which is to count the alternatives on each side before deciding whose money it is.

One particular combination is the case the rest of this guide turns on. With the duty at twenty per cent, the share dragged all the way to the top, and a home-built machine on the shelf at Rs 51,000/-, the slider still says one hundred per cent and the panel refuses to deliver it. The imported machine has to sell at Rs 51,000/- or not at all, and the machine landed at Rs 50,000/-, so the only room the duty has is the Rs 1,000/- between those two prices. The ceiling is the whole idea in one movement: the share reaching the buyer is not decided by anybody's intention, it is whatever the nearest alternative leaves room for. The panel is showing a ceiling, not a choice, and the identity readout underneath still balances because a ceiling changes who carries the duty and never how much of it there is.

What is a Non-Tariff Barrier, and how does it restrict without naming a price?

Everything so far has been about a tax. Now drop the tax and keep the restriction. A non-tariff barrier is anything that makes importing harder or dearer without being a duty: a limit on quantity, a rule the goods must satisfy, a permission somebody must grant, a wait at the port, a requirement about what the label says or where the parts came from. Non-tariff barriers are not a footnote to tariffs. In practice they are the larger part of what restricts trade, and they are far harder to argue about, precisely because none of them ever states a figure.

Five shapes are worth holding on to. A quota caps how many units may come in, and once the cap is reached the answer is simply no. A standard states what the goods must meet before they may be sold, on safety, emissions, testing or construction. A licence makes the import conditional on a permission that somebody has to issue, and whoever issues a permission can also delay it. A procedural delay holds the goods at the port while paperwork is checked, and holding goods costs money every day. A content or labelling rule requires a share of local parts, or particular markings, before the goods may enter the market.

Every one of these raises the cost of importing without ever naming a price, and the silence is exactly why they get used. A duty announces itself as a number. A testing requirement announces itself as a paragraph about safety.

Give the quota a number so the shape is not abstract. Suppose Sankhya buyers wanted 2,50,000 machines from Marut over a year. Priced at Rs 50,000/- apiece that is an import bill of Rs 1,250 crore, and now a quota caps entry at 2,00,000 machines. The permitted imports are now 2,00,000 machines, or Rs 1,000 crore, and 50,000 machines' worth of demand, a fifth of the total, has to find something else or go without. Notice what the quota did not do. The quota did not say what a machine may cost. Fewer machines are available than people wanted, so the price of a machine inside Sankhya will move. No document anywhere states by how much, and nobody signed anything that names a figure. The difference comes to one line: a duty fixes the amount and lets the quantity settle, while a quota fixes the quantity and lets the amount settle.

The everyday version is a shop during a shortage. A sign that says two packets per customer has not raised the price by a single rupee on paper. The customer still walks out with less than they came for, and the resale price on the street shows what the restriction was worth. Nobody wrote that number down either.

Why is a standard harder to see and harder to challenge than a duty?

A duty is a number sitting in a published schedule. A published number can be read, compared against the number another country applies to the same goods, and added up across a year of imports. A complaint about it points at the number. Nothing about it is a matter of interpretation.

A standard is a paragraph. The paragraph says the machine must have a particular guard, or be tested to a particular method, or carry a particular marking. Read on its own terms it may be entirely reasonable, and quite often it is exactly what it appears to be: a rule about safety written by people thinking about safety. But the same paragraph can also be written so that the machines built at home happen to satisfy it and the machines built in Marut happen not to. From the outside, those two paragraphs look identical. Whether a standard is a genuine standard or a restriction wearing a standard's clothes is a judgement about intent and effect, not something the text alone settles.

The judgement is what makes a standard hard, and the reason is worth stating precisely. The ambiguity is not a defect of a non-tariff measure. A rule that cannot be cleanly labelled a restriction is much harder to object to than a number, so to whoever imposes it the ambiguity is the feature. Which of the two a given measure really is, in any actual place, is settled by judgement and never by reading.

The everyday version sits inside an ordinary wedding. A hall says every caterer working there must hold a particular hygiene certificate. Nobody can call that unreasonable. The effect is that three caterers in the city qualify and the rest do not, and the price of catering in that hall is now set by three people. No price was ever mentioned in the rule.

What do the two routes to the same Rs 10,000/- look like side by side?

The comparison that makes the whole subject click into place is this one. Take the Sankhya machine again and forget the duty entirely. Instead, Sankhya introduces a rule: every imported machine of this kind must be tested on arrival and carry a certificate of conformity, and the paperwork takes two months at the port. Nobody has taxed anything. Now cost the rule out for one machine, with figures chosen so the total lands on a familiar number.

The financing line is worth pausing on because it is the one people forget. Twelve per cent a year on Rs 50,000/- is Rs 6,000/- a year, or Rs 500/- a month, so a two month wait costs Rs 1,000/- in working capitalthe money a business has tied up in stock, goods in transit and unpaid customer bills at any moment. It is money the business has already spent and cannot yet use for anything else. alone. Nothing was produced during those two months, nobody was paid a fee for the wait, and the money is gone all the same. The rest of the table is compliance costwhat it costs to prove a rule has been met, as distinct from what it costs to meet it. Testing fees, certificates, inspections and the paperwork behind them all sit here. in the ordinary sense.

The rule and the duty added exactly the same Rs 10,000/- to the cost of landing one machine, and the difference is that the duty produced Rs 10,000/- of revenue while the rule produced none. Under the duty, that money sits with the Sankhya authority and can be spent on something. Under the rule, the same money has gone to testing labs, storage yards and lenders, and no line anywhere in any account records it as a trade measure at all. The two mechanisms differ in where the money goes rather than in how much of it there is, and which arrangement a country should prefer is a separate question.

What does a barrier do to the price a household pays?

Two things happen, and the second one surprises people. The first is direct: whatever share of the cost was passed on shows up in the price of the imported goods. In the Sankhya case at sixty per cent pass-through, the machine goes from Rs 50,000/- to Rs 56,000/-, and anyone buying that machine pays twelve per cent more than before.

The second is indirect and it reaches much further. Think about the workshop inside Sankhya that also builds machines and was previously being undercut by the imports from Marut. The competition it faced has just got dearer. The workshop does not have to hold its price down as hard as it did before, so its own price can drift upward without losing a single customer. Nobody imposed anything on that workshop. Its price rose because the thing it was competing against rose.

Feel this in a kitchen rather than a factory. Suppose imported cooking oil gets dearer because of a duty. The locally pressed oil sitting on the same shelf did not get dearer for any reason of its own. The tin it had to undercut is now priced higher, so the local oil can be sold for more. The second effect reaches households that never buy the imported goods at all. The number of people touched by a barrier is always larger than the number of people importing anything.

How far that reaches into a general price level, into company results and into markets is a longer chain with several more links in it, traced on its own. The immediate step is the one already worked: a barrier raises the price of the imported goods by the share passed on, and it loosens the pressure on every close alternative made at home.

What does a lender ask a borrower who imports?

The mechanism stops being theory at a lending desk. Suppose a dealer inside Sankhya imports 2,500 machines a year, one machine in every hundred that crosses into the country, and asks a lender for a working capital line. A new duty of Rs 10,000/- a machine means the dealer will hand over Rs 2,50,00,000/- at the border over a year. A lender's first instinct is to ask about the duty rate. The duty rate is public, and the answer changes nothing about the dealer, so that is the wrong first question.

The right question is the one everything so far leads to: how much of the Rs 10,000/- can the dealer move off its own books, and to whom? There are three destinations, not two, once an actual business is in view. Some goes to the dealer's customers as a higher selling price. Some goes back to the exporter in Marut as a renegotiated invoice. Whatever neither of them accepts stays with the dealer.

The right column is only the middle one multiplied by the year, so read the middle column first. Rs 6,000/- plus Rs 2,000/- plus Rs 2,000/- is the Rs 10,000/- the dealer handed over, and nothing has been invented to make it balance. Each line multiplied by 2,500 machines brings the year out at Rs 2,50,00,000/-, the same duty seen from a distance.

Now put that against the dealer's own results. Its turnover on those machines is 2,500 at Rs 50,000/- each, so Rs 12,50,00,000/- a year. Put the dealer's operating profit before the duty at Rs 1,50,00,000/- a year, twelve per cent of that turnover. The Rs 50,00,000/- it could not move off its books is a third of that profit, gone. Nothing about the dealer's operations changed. The dealer sells the same machines to the same customers.

A lender assessing this borrower is not really asking about trade policy at all, it is asking how much pricing power the borrower has, and the duty simply made that question urgent. A dealer selling a machine nobody else supplies will move most of the cost along and barely notice. A dealer selling something three workshops down the road also build will move very little of it and will feel every rupee. Same duty, same country, same border, two entirely different credit stories.