Current Accounts: Determinants and Sustainability

Gian Maria Milesi-Ferretti

The current account is an important component of IMF operational work, both in terms of program design and in terms of macroeconomic surveillance for both industrial and developing countries. A substantial portion of research conducted at the IMF on current accounts has focused on two related questions: how to determine whether a given level of the current account deficit is “appropriate” or “excessive” and whether a particular level of the current account balance is sustainable from a longer-term perspective. A related path of research has examined the relationship between the current account and external crises. These questions involve several challenges in terms of the modeling framework, the design of empirical analysis, and the appropriate interpretation of the results. This article briefly summarizes recent research at the IMF on these and related issues.

Two key questions that arise in IMF country analysis are whether a given level of the current account deficit is “appropriate” or “excessive,” and whether the current account balance is sustainable.¹ Researchers who try to quantify these concepts face several difficulties. Establishing whether a given level of the current account is excessive requires a model or a framework that can determine what the “optimal” or “appropriate” level of the current account is—clearly not an easy task. Defining current account sustainability is also a difficult endeavor: the current account is the quintessential endogenous variable, and it is determined not just by public policy but also by private agents’ saving and investment decisions. Therefore, automatically extrapolating the current level of this variable into the future, in order to assess whether a country will be able to service its liabilities and/or secure access to external financing for future liabilities, is an exercise fraught with uncertainties.² Notwithstanding the difficulties highlighted above, a substantial portion of research at the IMF on the current account has indeed focused on these two related questions of “optimal” levels and sustainability. Some other papers have instead studied current account determinants and responses to shocks in both industrial and developing countries. A literature review of the two research approaches is presented below.

The model that is typically used to evaluate whether current account imbalances are excessive or not is a very stylized model based on the permanent income theory of consumption, in which the current account allows countries to smooth consumption over time when faced with transitory shocks. The model, initially applied to industrial countries to investigate whether capital flows were excessively volatile, has been used to evaluate current accounts in developing countries as well, and has been fairly successful in explaining current account fluctuations at business-cycle frequencies.³ Ghosh and Ostry (1997) extend the model to allow for a precautionary saving motive driven by income uncertainty (in addition to the standard consumption smoothing motive) and show that this extension contributes to better explaining cyclical fluctuations in current accounts in G-7 countries.⁴

Since the consumption-smoothing model assumes solvency and unfettered access to international capital markets at all points in time, it is best suited for examining the cyclical behavior of the current account; questions of current account sustainability are arguably connected to the trend behavior of the current account.⁵ A number of papers have assessed current account sustainability, taking into account a number of other macroeconomic and financial indicators such as the level of saving and investment, the level of the real exchange rate, the burden of external liabilities, the size of short-term debt relative to reserves, and, more generally, financial sector exposure and vulnerability.⁶ These studies are related to the more formal literature on “early warning indicators” that attempt to predict currency crises.⁷

Another research approach to current accounts has focused on analyzing current account determinants and the response of current accounts to different types of shocks.⁸ Debelle and Faruqee (1996) use a panel approach to examine the determinants of current accounts in industrial countries. Chinn and Prasad (2000) use a larger panel that includes both industrial and developing countries to examine the medium-term determinants of current accounts. They find that, in developing countries, current accounts are positively correlated with government budget balances, measures of financial deepening and initial stocks of net foreign assets.⁹ Calderon, Chong, and Zanforlin (2001) focus on the determinants of the current account in African countries, highlighting the observation that the degree of persistence in current account imbalances is smaller, and the sensitivity of the current account balance to private savings is larger, in African countries than in other developing countries. Cashin and McDermott (1998b) focus on five commodity-exporting developed countries and find evidence that terms of trade shocks induce strong income and substitution effects. Decressin and Disyatat (2000) examine the relative response of the current account and investment to productivity shocks across European countries and within regions in Canada and Italy, and find no systematic difference between the cross-country and the cross-regional evidence.¹⁰

A line of work related to both literatures on current account determinants and current account sustainability has been pursued by Milesi-Ferretti and Razin (1998, 2000).¹¹ These authors point out that “unsustainable” current account deficits need to be reversed eventually, and that recent external crises have been characterized by very large swings in the current account. They focus their empirical analysis on large and sustained reductions in current account deficits (reversal episodes) and examine which factors help predict the occurrence of a current account reversal as well as what are the implications for macroeconomic performance. Their findings suggest that reversals are more likely to occur in countries with large external imbalances, low foreign exchange reserves, and deteriorating terms of trade. Interestingly, reversals are not systematically associated with output slowdowns or currency crises. Some countries experience faster growth rates during the reversal period, while, in others, growth is slower (particularly, in countries that start out with an overvalued real exchange rate).

Also of interest are a few papers that focus on the theoretical implications of a credible disinflation program and of pension reform on the current account.¹² Finally, research is under way at the IMF to model the determinants and dynamics of net foreign assets—the stock concept associated with the external current account. Lane and Milesi-Ferretti (1999) construct a comprehensive historical database of net foreign assets for a large group of industrial and developing countries. In more recent work, these authors explore the cross-country determinants of net foreign asset positions.¹³

Fiscal Decentralization

Hamid R. Davoodi

During the last two decades, numerous economic and political developments—the worldwide trend toward democracy, the emergence of new economic federations and of decentralized fiscal structures in transition economies (Russia, in particular), a general dissatisfaction with central governments’ fiscal performances, and the creation of common currency unions—have reinvigorated an old debate about the efficacy of fiscal decentralization. Proponents on one side of the debate use a traditional, normative model to evaluate the observed patterns of fiscal decentralization. The other side emphasizes a positive approach asking whether attempts at fiscal decentralization have, in practice, produced the intended benefits, such as enhanced public service delivery, higher growth, lower poverty, better macroeconomic management, and better governance. While opinions seem to have converged regarding the normative analysis, underscoring the importance of getting the fundamentals right, the evidence appears to have diverged with regard to the empirical analysis. This summary reviews the debate and surveys research that has been done at the IMF since 1997.

Fiscal decentralization has been a topic in the public finance literature for almost 40 years.¹ It is often defined as the devolution of fiscal responsibilities from the central government to subnational levels of government (e.g., states, regions, provinces, districts, municipalities) but this definition, in itself, is not complete. In fact, the basic issue in fiscal decentralization and fiscal federalism is that of assigning specific fiscal responsibilities to the proper level of government.² These responsibilities, which range from the design to the implementation of various aspects of intergovernmental fiscal relations, raise a number of questions:

• What types of spending should be conducted by what levels of government (i.e., expenditure assignment)?;

• What types of revenues should be raised and what tax rates should be set by what levels of government (i.e., revenue assignment)?;

• How should intergovernmental grants and revenue sharing be used to fill the gap between expenditures and revenues at subnational levels and to provide the right incentives to subnational governments?;

• Which level should be able to finance its spending by borrowing from domestic or external sources, private or public?;

• Which level is responsible for tax administration and the public expenditure management system?;

• Which level should design and administer regulations?; and

• How should regulations be harmonized across various levels of government?

The traditional, normative analysis for assigning these responsibilities to various levels of government rests on how these responsibilities influence the balance between the three basic functions or objectives of fiscal policy: efficiency in allocation of resources, income redistribution, and macroeconomic management.

Generally, allocative efficiency calls for assigning to lower levels of government those responsibilities where the costs and benefits are confined to a local jurisdiction and reflect the preferences of the local community (e.g., local police and garbage collection services). Income redistribution policies (e.g., pensions, social assistance) and macroeconomic policies (e.g., financing of government expenditures, monetary policy) are generally assigned to the central government or to a unique, central authority because certain tax bases (e.g., corporate income tax) are mobile and have economywide, automatic, stabilizing effects; some expenditures have large aggregate demand effects (e.g., federal transfers to states, interstate transportation networks, unemployment benefits); and price-stabilizing measures supercede the jurisdictions of subnational governments and are best left to a central monetary authority.

Recent research on fiscal decentralization at the IMF—based on country-specific studies, cross-country analyses, and policy discussions with member countries—has heightened awareness of the risks of decentralization and demonstrated the importance of “getting the fundamentals right”:³

• In summarizing the experiences of over 20 countries, Ter-Minassian (1997) concludes that, while efficiency-enhancing benefits of decentralization are well understood in theory, substantial decentralization makes it more difficult to carry out the redistributive and macroeconomic management objectives of fiscal policy. Cross-country empirical analysis has shown that decentralization is associated with lower growth, higher deficits, and larger governments. The evidence regarding the latter finding is mixed, however, when considering research conducted outside the IMF.⁴

• Expected efficiency-enhancing benefits of decentralization may not materialize because of factors such as: a limited administrative capacity at the subnational level to manage a transparent public expenditure system and a modern treasury operation; a limited ability to report and prepare subnational budgets; the presence of low factor mobility and limited, or even nonexistent, intergovernmental competition; and, perhaps, a higher incidence of corruption at the local level than at the central level.⁵ Empirical evidence on the relationship between corruption or poor governance and decentralization seems to be mixed at best and may depend on how subnational expenditures are financed.⁶ However, some studies have found, unambiguously, that countries with decentralized fiscal structures tend to be more democratic.⁷

• In countries with significant macroeconomic imbalances, a hard budget constraint should be imposed on subnational governments, including a rule of zero borrowing by subnational governments. In countries with no serious macroeconomic imbalances, borrowing may be allowed, but it should be subject to market discipline, and there should be clearly specified rules that link the subnational governments’ abilities to service their debts to their assigned responsibilities.⁸ (In this respect, for instance, Brazil’s Fiscal Responsibility Act seems to have been instrumental in restoring fiscal discipline at the subnational level.⁹)

• The success or failure of decentralization programs depends on the sequencing and the speed of reform measures, as well as the clear division of expenditure and revenue assignments.¹⁰ In particular, decentralization of revenues and transfers should not precede the devolution of expenditure responsibilities, and a “big-bang” approach to decentralization is unlikely to be successful.¹¹

Increasingly, attention has been paid to implications of the so-called, common-pool problem for fiscal decentralization. Common-pool problems arise when the same tax base is shared among different levels of government who can set their own tax rates, or when decentralized spending is financed by unconditional grants.¹² On the spending side, the common-pool problem can lead to an overspending bias at the subnational level if the benefits of public spending are concentrated in a few local jurisdictions and their costs are diffused across many jurisdictions through general taxation. In an empirical study of U.S. city governments, Baqir (forthcoming) finds support for this type of common-pool problem; a ceteris paribus increase in the size of the city legislature by one person—serving as a proxy for greater districting and the bestowal of benefits on a particular constituency—increases that city’s per capita expenditures by 3 percent.¹³ On the tax side, the common-pool problem can lead to interstate tax competition (the horizontal tax externality) and federal-state tax competition (the vertical tax externality), which would affect tax rates at both the federal and state levels. Keen and Kotsogiannis (forthcoming) show that a federation can result in an excessively high state tax rate if the vertical tax externality dominates the horizontal tax externality.¹⁴

Finally, the debate on fiscal decentralization also hinges on the role of the state in the economy, an issue which is at the heart of many transition economies. Attempts at decentralization should come only after the rationale for state intervention in a particular activity is firmly established. In the absence of a firm rationale for state intervention, privatization of any public sector activities, regardless of whether they are carried out by the central or subnational governments, should always be the preferred alternative to fiscal decentralization.