The large and persistent shifts in the current-account balance of industrial and developing countries (DCs) in the first half of the nineties, as well as the balance-of-payments turbulence affecting emerging markets in the mid- and late nineties, have prompted renewed interest in the dynamics of the current account. The issue is relevant because current-account dynamics has direct implications for the sustainability of external debt and the debate on capital mobility. Moreover, current-account imbalances are often mentioned among the triggering factors of currency and financial crises. Developing countries are of special interest in this field, because their external economic relations are monitored by multinational financial institutions which provide policy advice and may call for modifications in public conduct. According to the intertemporal approach to the current account, a swing in the current-account balance may have two different origins: first, there can be transitory fluctuations determined by the smoothing of unexpected transitory shocks. Second, shifts in the fundamentals may permanently affect the current account path. In principle, episodes of the first kind will not be relevant for a country's intertemporal borrowing constraint (provided their transient nature is correctly understood by the agents in the lending economies). We are therefore interested in explaining episodes of the second kind, i.e. "current-account reversals", defined as permanent improvement ("upward" reversals) or deterioration ("downward" reversals) of the current-account balance-to-GDP ratio (where negative values of this ratio indicate current-account deficits, i.e., an excess of domestic investment over saving to be financed by capital inflows or depletion of official reserves). Structural changes in the current account path can be related to the dynamics of the external debt-to-GDP ratio. Large negative values of the current-account balance-to-GDP ratio can lead to a sharp increase in the external debt-to-GDP ratio. Under such circumstances, agents in the lending economies may eventually lose confidence in the creditworthiness of the borrowing country. This shift in confidence will lead to a financial crisis, possibly associated with the intervention of multilateral agencies, thus forcing the borrowing economy on a more sustainable path of indebtedness. Therefore, such a sustainability crisis will show up ex post as an upward reversal, i.e., as a persistent increase in the current-account balance (a deficit reduction). Differently from other studies, we also consider downward reversal, i.e., episodes of persistent deterioration where the current-account balance is put on a decreasing path which over time will likely turn out to be unsustainable. Hence, while upward reversals are generally related to the resolution of sustainability crises, the study of downward reversals could indicate a few fundamentals that have to be monitored in order to prevent potentially unsustainable paths. Despite a recent vintage of studies, the determinants of current-account reversals in DCs have not yet been thoroughly investigated and some important questions are still unanswered. Why do reversals occur? To what extent are such permanent changes associated with corresponding shifts in fundamentals? Is the impact of macroeconomic variables the same in the case of positive and negative reversals? In our paper we provide an answer to these questions by analyzing current-account reversals and investigating their determinants in panel of 49 DCs. Our study involves two distinct stages: reversal identification and reversal explanation. In the first stage (reversal identification) we discriminate current-account reversals (i.e., turning points corresponding to permanent changes in the structure of the economy) from temporary swings in the current-account balance. Previous studies in this field make similar distinctions relying on either prior information (event study) or ad hoc criteria (e.g., by considering "permanent" the changes that exceed a given threshold). We show that these approaches lead to unsatisfactory identification and dating of the relevant episodes. Therefore, we follow a different approach by using formal tests for structural change and segmented trend analysis of the current account path. These tests are applied following a general-to-specific procedure that starts from a general structure of the data generating process (DGP), encompassing an unknown break date and allowing for both deterministic and stochastic trending components. The tests results show that 35 out of the 49 countries included in our sample display significant structural breaks in the current-account balance-to-GDP ratio, of which 24 are upward and 11 downward reversals. As was to be expected, we find that upward reversals are generally associated with the start or renewal of IMF-supported adjustment programs, thus enforcing the view that these episodes correspond to the resolution of sustainability crises. In the second stage (reversals explanation) we look for a structurally stable relationship around reversal time between the current-account balance and a number of explanatory variables suggested by the intertemporal approach to the current account: the real GDP growth, the external debt-to-GDP ratio, the rate of change of the terms of trade, and the fiscal balance (as a fraction of GDP). This is performed by estimating a panel-data regression with both country- and time-specific fixed effects, where the time dimension consists of seven observations centered on the reversal dates. We also check for parameter constancy over time and across the two kinds of reversal (upward and downward). These tests enable us to assess whether the selected explanatory variables include the major determinant of the current-account reversal, and whether they play the same role in episodes of improvement or deterioration of the current-account balance. Estimation results show that: (1) the estimated equation provides a good explanation of the current-account dynamics around the break date, with strongly significant, correctly signed regressors and satisfactory misspecification test statistics; (2) the time fixed-effects are not significant, thus suggesting that no relevant time-varying explanatory variables are omitted from the regression; (3) the Chow tests of parameter constancy over time does not reject the null: therefore, the shifts in the current-account path appear to be actually explained by corresponding shifts in the above-mentioned fundamentals; (4) on the contrary, the hypothesis of parameter invariance across the two set of reversals is rejected. The last result suggests that the selected fundamentals play a different role in upward or downward reversals. The estimation of separate regressions for the two set of reversals shows that fundamentals do actually play different roles in upward and downward reversals. While the former are associated to fiscal consolidation (i.e., increase in the fiscal balance-to-GDP ratio) and terms of trade improvement, the latter are mainly related to high real growth rates and external debt-to-GDP ratios. This suggests that some DCs have found themselves on unsustainable path of foreign indebtedness, as large foreign debts seem to have contributed to permanent worsening of the current account
