J-Curve (Trade)

The counter-intuitive shape

Suppose a country has a chronic trade deficit and decides — explicitly through devaluation, or implicitly by letting a floating currency fall — to make its currency cheaper. The textbook intuition is clear. A weaker currency makes exports cheaper for foreign buyers, who buy more. It makes imports more expensive for domestic buyers, who buy less. Exports up, imports down, trade balance improves. Done.

What actually shows up in the data is uglier. For the first six to twelve months after the depreciation, the trade balance does not improve. It gets worse. Newspapers run "weak pound fails to boost exports" headlines. Politicians who pushed for the depreciation get blamed. Currency markets, watching the deteriorating numbers, may push the currency lower still. Only after roughly a year — sometimes two — does the trade balance begin a sustained climb back toward and through its pre-depreciation level. Plotted against time, the path looks like the letter J: a short hook downward, then a long climb upward.

This is the J-curve. It is not a curiosity — it is the default behaviour of trade balances after every major depreciation that has been studied in detail. Wilson's 1967 devaluation of sterling, the post-Plaza-Accord adjustment of the US dollar in 1985-88, the Mexican peso of 1994-95, the Russian ruble in 2014-15: each traced a recognisable J. The shape is sufficiently regular that a multi-billion-dollar industry of macro hedge funds positions around it.

Why prices move now and quantities move later

The mechanism is a timing mismatch between two ingredients of the trade balance:

TB = P_x · Q_x  −  P_m · Q_m

   where  P_x , P_m are prices in domestic currency
          Q_x , Q_m are quantities (exports, imports)

When the domestic currency depreciates by, say, 20 percent against the rest of the world's basket of currencies, the prices update instantly. Imports — priced in foreign currency by the foreign supplier — are still the same dollars or euros or yen abroad, but those foreign currencies now buy 25 percent more domestic currency, so the same imports cost 25 percent more in domestic terms. Exports, if priced abroad in foreign currency, earn more domestic currency on each unit sold. The price terms P_m and P_x jump immediately.

Quantities, in contrast, are sticky on the timescale of weeks to a year. A car manufacturer that bought a year's worth of Japanese steel under contract at the old exchange rate continues to ship that steel at the old domestic-price equivalent. The same manufacturer's export contracts for cars to Germany were priced in euros six months ago and cannot be re-negotiated mid-stream. Consumer purchasing habits — the supermarket buying the Italian olive oil it has always bought, the family booking the holiday it always books — are similarly slow to respond. Even when consumers want to substitute away from now-expensive imports, finding alternative domestic or third-country suppliers takes months. Building new export capacity to satisfy new foreign demand takes longer still.

The result: for several quarters after the depreciation, prices have moved but quantities have not. The trade balance is dominated by the price effect: imports cost much more in domestic currency, exports earn somewhat more, and the net result — for a country running a sizeable import bill — is a wider deficit. Only when Q_m finally falls and Q_x finally rises does the trade balance recover. That delayed quantity response is the rising leg of the J.

The Marshall-Lerner condition

The price effect's "short-run" deterioration is empirically obvious. The long-run improvement, however, is not automatic. It requires that demand respond strongly enough to price changes — that the quantity gain eventually outweighs the price loss. The classical statement is the Marshall-Lerner condition:

|ε_x| + |ε_m|  >  1

where ε_x is the price elasticity of foreign demand for the country's exports and ε_m is the price elasticity of domestic demand for imports. If the sum of these absolute values exceeds one, depreciation improves the trade balance in the long run — the J closes and rises through zero. If the sum is below one, the volume response is too weak to overcome the worse price terms even after adjustment, and the trade balance gets permanently worse: the J flattens out below the pre-depreciation line and never crosses it.

Empirically the long-run sum for most countries is comfortably above one. Typical pooled estimates for advanced economies put it in the range 1.5-2.5, well within the "improvement" zone. But the short-run sum is usually well below one — often around 0.4-0.6 — because consumers and firms have not yet had time to rearrange their purchasing. This dual fact — Marshall-Lerner fails in the short run, holds in the long run — is precisely the analytical statement of the J-curve.

The condition assumes balanced initial trade. If the country starts with a large trade imbalance, a more general version applies (e.g. the Bickerdike-Robinson-Metzler condition); the qualitative story is unchanged but the threshold shifts.

A worked example

Consider a country with the following pre-depreciation position (annualised):

	Volume	Foreign price	Domestic value
Exports	100 units	$1.00 each	$100
Imports	120 units	$1.00 each	$120
Trade balance			−$20

Now the domestic currency depreciates by 25 percent — foreign currency rises 33 percent against domestic. Assume in the short run nothing changes except prices. Exporters keep prices in foreign currency unchanged (typical for differentiated manufactures); importers also keep foreign-currency prices unchanged. In domestic-currency terms, everything is multiplied by 1.33:

Short run (0-6 mo)	Volume	Domestic value
Exports	100 units	$133
Imports	120 units	$160
Trade balance		−$27

The deficit widens from −$20 to −$27 even though nothing "bad" has happened — the depreciation is presumably supposed to help. This is the descending leg of the J.

Now run forward 24 months. Assume long-run elasticities of 1.2 for exports and 0.8 for imports (sum = 2.0, comfortably Marshall-Lerner). A 25 percent depreciation means foreign-currency export prices effectively fall ~25 percent and domestic-currency import prices effectively rise ~33 percent (relative to a stable income level). Export volume rises by ~1.2 × 25% = 30 percent to 130 units; import volume falls by ~0.8 × 33% = 26 percent to 89 units:

Long run (24+ mo)	Volume	Domestic value
Exports	130 units	$173
Imports	89 units	$119
Trade balance		+$54

The trade balance has flipped from −$20 to +$54 — a swing of $74. The J has closed and risen well into the surplus zone. The time path was: −$20 (start) → −$27 (trough, ~9 months) → −$20 (~18 months) → +$54 (long-run, ~24-36 months). That is a textbook J.

Empirical case studies

Episode	Year	Depreciation	Trough	Recovery	Notes
UK sterling devaluation (Wilson)	1967	14% vs dollar	1968	1969-70 surplus	"Pound in your pocket" speech; current account turned in ~2 years
US dollar post-Plaza Accord	1985-87	~50% vs yen	1986-87 ("deficit widens")	1988-89 narrowing	Three-year lag from agreement to clear improvement
Mexican peso (Tequila crisis)	1994-95	~50% vs dollar	Q1-1995	1996 surplus	Sharp J — collapse in import demand from recession reinforced the volume effect
Asian crisis (Thai baht, won, rupiah)	1997-98	~40-80%	Brief or skipped	1998-99 surpluses	Recession collapsed imports so fast the descending leg was nearly absent
Russian ruble	2014-15	~50% vs dollar	Q1-2015	2015-16	Oil-price collapse + sanctions; classic J trajectory
Turkish lira	2018, 2021-22	40%+ in 2018	2018-19	Partial	Inflation pass-through eroded the real depreciation; J shallow

The cases differ in the depth and steepness of the J, but the basic shape recurs. The exceptions — the Asian-crisis economies in 1997-98 — are not really exceptions: the descending leg is short or invisible because the recession that accompanied the depreciation collapsed import demand on the same timescale as the price shock, so the price and volume effects overlapped instead of separating in time.

When the J fails to appear

Not every depreciation produces a visible J. The patterns that suppress it:

• Incomplete pass-through. If importers absorb the exchange-rate move in their margins instead of passing it on to consumer prices, domestic-currency import prices barely move and the descending leg shrinks. Pass-through into US import prices has been estimated at less than 30 percent at one year, much less than the full 100 percent assumed in textbook J-curve diagrams. The smaller the pass-through, the shallower the J.
• Pricing-to-market. Exporters may set foreign-currency prices to compete locally rather than to a fixed mark-up on domestic costs. A depreciation does not then lower foreign-currency export prices, and the export-volume response weakens.
• Currency invoicing. If a country invoices most of its imports in its own currency (rare, but partly true for the United States), depreciation does not raise the domestic-currency cost of imports at all in the short run — the descending leg disappears. Conversely, dollar-invoiced commodity imports in an emerging market accentuate the descending leg.
• Recession overlay. A crisis that combines depreciation with sharp output contraction collapses imports through the income channel as fast as through the price channel. The J becomes a V or a flat decline.
• Inflation feedback. When depreciation triggers wage-price spirals (Turkey, Argentina), the real depreciation evaporates. The nominal currency fell, but domestic prices rose enough that the relative price of foreign goods barely changed. Marshall-Lerner cannot operate on a real depreciation that never settled in.
• Perverse J. A small number of countries show a "perverse" or "inverse" J — early improvement, later deterioration. This typically reflects anticipatory effects: exporters rush shipments ahead of expected further depreciation; importers delay. Once the rush passes, the underlying mechanics reassert themselves.

Policy implications

The J-curve has three operational implications for governments and central banks managing exchange-rate policy:

• Patience with the data. A depreciation aimed at correcting a deficit will show worse, not better, trade numbers for roughly a year. Misreading the descending leg as policy failure has triggered policy reversals (re-pegging, additional intervention) that made things worse. Wilson's 1967 government was lambasted for months before sterling adjustment delivered.
• Financing the trough. The country must finance the worsened balance during the descending leg. This is what foreign-exchange reserves, IMF Stand-By Arrangements, and contingent credit lines are for. A country that depreciates without buffers can be forced into a second, panicked depreciation by markets that misread the first.
• Complementary policies. Because Marshall-Lerner is empirically usually but not always satisfied, and because pass-through and inflation feedback can erode the real depreciation, successful depreciations are paired with fiscal restraint, monetary tightening to anchor inflation expectations, and supply-side measures to make sure the export response actually materialises. Depreciation alone is not a policy — it is one piece of a coordinated adjustment package.

Extensions and variants

• S-curve. Some empirical studies (Backus, Kehoe, Kydland 1994) document a longer lead-up: trade balances show a small improvement just before depreciation (as markets anticipate it and front-load), then a descending leg, then recovery — a shape closer to a tilted S than a clean J. Largely a feature of advanced economies with deep forward markets.
• Pass-through estimation. A large empirical literature (Goldberg-Knetter 1997, Campa-Goldberg 2005) measures how much exchange-rate movement shows up in import prices at different horizons. Lower pass-through means a shallower descending leg.
• Real exchange rate vs nominal. The J runs on the real exchange rate (nominal exchange rate adjusted for relative price levels). High-inflation depreciations that erode the real adjustment never get a J because the relative-price impulse evaporates.
• Income absorption. The absorption approach (Alexander 1952) emphasises that a trade-balance change is also a savings-investment change. A depreciation that doesn't raise national savings — that is, doesn't reduce domestic absorption — cannot durably improve the trade balance regardless of elasticities.
• Twin deficits. If the trade deficit reflects a fiscal deficit (savings shortfall), depreciation alone treats the symptom. The classic example is the post-Plaza US deficit: the dollar fell sharply but the deficit did not close because the Reagan fiscal deficit kept national savings depressed.

Common pitfalls

• Reading the descending leg as policy failure. The most common political-economy error. The leg is mechanical, not behavioural — it would appear even if the depreciation were working perfectly.
• Assuming Marshall-Lerner automatically. The condition is overwhelmingly satisfied at long horizons in pooled data, but specific country-period combinations (especially short-horizon, high-inflation, low-pass-through, or commodity-dominated cases) can fail it.
• Ignoring the distinction between nominal and real depreciation. A 50 percent nominal depreciation accompanied by 50 percent domestic inflation is a zero real depreciation and cannot deliver a J.
• Forgetting the income channel. Depreciation can be expansionary (cheaper exports, switching to domestic substitutes) or contractionary (balance-sheet effects from foreign-currency debt, imported inflation hitting real incomes). Which dominates determines whether import demand falls primarily through the price channel (slow) or income channel (fast), reshaping the J's geometry.
• Comparing trade balances across regimes without controlling for terms-of-trade shocks. A commodity price collapse alongside a depreciation can mimic or mask the J entirely; the visible balance is the convolution of two impulses.