Keywords

JEL Classifications

A depreciation of the domestic exchange rate is expected to improve a country’s trade balance. It has been observed, however, that in reality the trade balance often worsens immediately after depreciation. Only in the longer term, if at all, does it improve. The combination of a negative short-run effect together with a subsequent positive long-run effect of a devaluation of the exchange rate on the trade balance is referred to as the ‘J-curve’ due to the similarity of the current-account behaviour to a ‘J’.

To assess the existence of a J-curve adjustment path is relevant, since, under most circumstances, the J-curve may induce dynamic instability in exchange rates (see Beenstock 1990; Levin 1985). If depreciation worsens the trade balance in the short run, the exchange rate may fall further (depreciate more). Although export volumes may rise and import volumes fall, import values tend to grow faster than export values and the exchange rate instability persists. This instability may be neutralized by speculators having rational expectations. In this case, agents know the dynamics of the J-curve and allow for it in their speculative behaviour, thereby eliminating the potentially destabilizing influence of the J-curve itself.

The J-curve is the description of an empirical phenomenon, first discussed after the 1967 devaluation of the pound sterling in NIESR (1968) and analysed in a seminal paper by Magee (1973). Theoretical models have been developed, building on some kind of frictions, such as pre-existing contracts, asymmetric use of domestic currency and foreign (international) currency, sluggishness in adding new productive capacity and sunk costs.

The lags in the adjustment can be ascribed to trade prices adjusting faster than trade volumes to changes in the exchange rate. The currency in which imports and exports are denominated, which is likely to be determined by the relative market power of traders, plays a crucial role. When both import and export contracts are expressed in domestic currency, following an unexpected devaluation, the value of existing imports rises due to the increased cost of an unchanged quantity of imports, while the value of existing exports remains constant (the price of exports in domestic currency does not change). The existence of lags on consumers’ and producers’ side induces stickiness and a worsening of the trade balance, until higher export and lower import volumes eventually, and on the assumption that import and export demand elasticities are sufficiently elastic to exchange rate changes so that their sum is higher than 1 (that is, the Marshall–Lerner conditions hold), generate a favourable trade balance response (that is, a J-shaped path). When both import and export contracts are expressed in foreign currency, if the value of contracts denominated in foreign currency is higher for imports than for exports, the J-curve will always ensue. On the other hand, the trade balance would improve immediately after devaluation if import contracts are denominated in domestic currency and export contracts in foreign currency.

Quantities may not adjust rapidly to exchange rate changes, since domestic demand for imported goods may be fairly inelastic due, for instance, to a reputation or brand and/or domestic supply may be a poor substitute for imported goods. Furthermore, producers may not be able to reallocate expenditure between foreign and domestic goods since most import and export orders are placed in advance (before depreciation). In the long run, however, quantities tend to adjust and, if the value of elasticity satisfies Marshall–Lerner, the trade balance improves.

Several models have suggested not mutually exclusive reasons for a short-run deterioration of the trade balance following depreciation. Knetter (1993, p. 473) maintains that: ‘sellers reduce markups to buyers whose currencies have depreciated against the seller, thereby stabilising prices in the buyer’s currency relative to a constant markup policy’, that is, follow a ‘local currency price stability’ and differentiate between markets (pricing to market). Some models rely on the small open economy hypothesis and emphasize intertemporal substitution. Bacchetta and Gerlach (1994) challenge the view of a rapid pass-through of exchange rates to import prices, showing that the J-curve can arise even if import prices are sticky. In an intertemporal framework, an anticipated rise in future import prices after depreciation provides agents with an incentive to decrease their current expenditure and therefore revise their future purchases, eventually displaying a J-shape dynamics of trade balance.

In a continuous time open-economy framework, Mansoorian (1998) shows that a trade balance deterioration following depreciation can be due to persistence of consumption habit, on the assumption that the utility function depends not only on current consumption but also on standard of living. Similar conclusions have been recently reached by Cardi (2005) who derives a J-curve due to the ‘strength’ of consumption habits and capital investment ‘inertia’, following an unanticipated terms of trade deterioration.

Tivig (1996) solves an intertemporal maximization problem in a dynamic oligopoly contest. In a two-period model of duopolistic competition without entry, regardless of the degree of capital mobility, she provides sufficient (at least) conditions on import demand elasticity for short-run ‘perverse’ price reactions, following temporary changes in the exchange rates, so that a temporary devaluation may initially worsen and later improve the trade balance.

Tornell and Lane (1998) use political economy considerations to explain the J- curve pattern; their model with ‘voracity effects’ suggests that a positive real trade shock may exacerbate the common pool problem, leading to a more than proportional increase in public transfers and then to a social loss. Because of this effect, the impact of an unexpected improvement in the terms of trade may lead to deterioration in the current account.

Recent literature in dynamic general equilibrium depicts an S-curve as a dynamic response of the trade balance to technology shocks. Backus et al. (1994) find that the trade balance is negatively correlated with current and future movements in terms of trade, but positively correlated with past movements. Using a two-country version of Kydland and Prescott’s (1982) closed economy model, in which each country produces imperfectly substitute goods with capital and labour, they claim that, after a once and for all positive shock to technology, domestic output increases, its relative price falls and domestic investment increases, inducing a fall in net exports. With time, the rise in investments decreases and the trade balance moves into a surplus. This dynamic gives rise to an S-curve consistent with the J-curve, since the initial deterioration and subsequent improvement of the trade balance may well deliver also an S-shaped cross-correlation function of the trade balance and the terms of trade. Senhadji (1998) extends the Backus, Kehoe and Kydland analysis to document business cycle features of several developing countries. Within a set-up with a downward sloping export demand function and limited access to international financial markets for capital formation, he shows results completely supportive of the findings of Backus, Kehoe and Kydland.

Finally, the J-curve can be due to hysteresis (cf., for instance, Baldwin 1988; Dixit 1994). In the presence of sunk costs, to export is an ‘option’ and consumers value the alternative of ‘wait and see’ before reacting to exchange rate changes. The presence of a threshold induces non-standard behaviour of the trade flows when the exchange rate depreciates.

Over the years, an extensive empirical literature has emerged. Results are not conclusive. As stated by Bahmani-Oskooee and Artatrana (2004, p. 1389), ‘the general consensus is that the short run response of the trade balance to current depreciation does not follow a specific pattern’ but, if a J-curve exists, the perverse effect has a duration of one to three years (see Junz and Rhomberg 1973; Baldwin and Krugman 1987; Spitaller 1980; Moffett 1989). Koray and McMillin (1999) use a structural VAR model to show that the pattern of the trade balance after a negative monetary shock exhibits the traditional J-curve behaviour. Support for the J-curve hypothesis has been found also by Bahmani-Oskooee and Alse (1994), Marwah and Klein (1996) and Hacker and Abdulnasser Hatemi (2003). Leonard and Stockman (2002) find a weak positive evidence of the J-curve, but document strong violations in the distributional assumptions that underline previous works. More interestingly, their evidence on the J-curve is inconsistent with traditional theoretical explanation (real business cycle models included), since they present evidence that current account surpluses are usually associated with low real GDP.

Other studies have challenged the existence of a J-curve: Rose and Yellen (1989) and Rose (1990,1991) maintain that, if import prices adjust slowly to exchange rate changes, the initial negative effect embodied in the J-curve may not occur: the value of imports does not increase and, ceteris paribus, the trade balance does not worsen. More recently, Demeulemeester and Rochat (1995), Hsing and Savvides (1996), Shirvani and Wibratte (1997) using different methodology, different data-set and estimating over different periods, found no evidence of a J-curve in the data.

See Also