J-curve after devaluation

Why does a cheaper currency make a trade deficit worse before it makes it better?

A country's currency falls. Its exports are now cheaper for foreigners, its imports dearer at home. The textbook conclusion writes itself: exports up, imports down, trade deficit narrows.

Then the trade figures come out, and the deficit has widened. Not slightly, and not because something else went wrong — widened because of the devaluation, and it will keep widening for several quarters before it turns. The remedy works, eventually, by first making the patient worse. What is going on in those first months?

Split the trade balance into the two things it is made of: the money coming in for exports, and the money going out for imports. Each of those is a price multiplied by a quantity.

Now ask which of those four numbers a devaluation can change immediately.

The prices change on day one. That is what an exchange rate is — the moment it moves, every imported good costs more in domestic currency, and every exported good looks cheaper to a foreign buyer. Nothing has to happen in the world for that to be true.

The quantities cannot change on day one, and that is the whole story. The oil arriving next week was contracted months ago. The car factory's gearboxes come from a supplier who was chosen after a two-year qualification process. The foreign buyer who might now prefer your machine tools has a purchase order out with your competitor and a plant tooled for their spindle. Volumes are set by contracts already signed, inventories already ordered, and sourcing decisions that take quarters or years to revisit.

So run the arithmetic for month one. Import volumes: unchanged. Import prices in domestic currency: up by the full devaluation. The import bill therefore rises by nearly the full devaluation. Export volumes: unchanged. Export earnings rise only to the extent that the sales were priced in foreign currency to begin with. Put the two together and the balance moves the wrong way. The country is paying more for exactly the same imports it was buying yesterday.

Then adaptation begins, and it is slow because it is made of decisions rather than arithmetic. A foreign purchasing manager notices the price and asks for a sample. A domestic manufacturer looks for a local substitute for a component. A tourist changes destination. A supplier wins a contract, then has to add capacity to fill it. Each of those takes months; collectively they take years. Volumes drift toward the new prices, export earnings grow, the import bill shrinks, and the balance crosses back and keeps rising. Plot it and you have drawn a J.

Economists have a compact way of saying this. Whether a devaluation helps depends on how responsive quantities are to price — the elasticities of demand for exports and for imports. The Marshall-Lerner condition says the two must sum to more than one for the balance to improve. The J-curve is simply the observation that those elasticities are small in the short run and larger in the long run, so the condition fails at first and passes later. The curve is not a separate phenomenon; it is the same condition, watched over time.

How long? Empirically the trough tends to come a few quarters in and the full effect within one to two years, though the estimates vary a great deal by country and episode. The classic illustration is Britain's November 1967 devaluation, from 2.80 to 2.40 dollars to the pound: the trade figures deteriorated through 1968 and the balance did not move decisively into surplus until 1969.

Several refinements, some of which can flatten the J or remove it.

The first concerns who sets prices in which currency. The classical story assumes exporters price in their own currency, so a devaluation passes straight through to a lower foreign price. Much of world trade is instead invoiced in dollars, and those prices are sticky. Recent research on dominant currency pricing finds that for many countries a devaluation barely changes the foreign-currency price of their exports, so the export-volume channel is weak and most of the adjustment comes through imports being squeezed. The result can be a long flat bottom rather than a graceful J.

The second is that exports use imports. A manufacturer relying on imported components sees its own costs rise with the exchange rate, so its ability to undercut foreign rivals is smaller than the headline devaluation suggests, and shrinks further the more integrated its supply chain is.

The third is that a devaluation is rarely an isolated act. It usually accompanies a crisis, capital flight, or a stabilisation programme, and if domestic debts are denominated in foreign currency the revaluation of those liabilities can be contractionary enough to swamp the trade effect entirely. Untangling the J from everything else happening at the same time is genuinely hard, and the empirical literature is not unanimous that a J-curve appears in every episode.

Finally, the mechanism is about adaptation lags, not about the currency being magic. If nothing in the real economy can respond — no spare capacity, no substitutable imports, no export sector waiting to expand — the price effect is all you get, and the deficit simply stays worse.

• The Marshall-Lerner condition, and why estimated short-run elasticities are so often below one.
• Dominant currency pricing, and what it implies for whether devaluation is a useful policy tool at all.
• Contractionary devaluation, and balance-sheet effects when liabilities are in foreign currency.

Every term the collection defines is gathered in the glossary.