Paul Krugman blogged a few days ago about the “surprising intellectual flexibility ” of the IMF in endorsing the use of capital controls to “calm volatile cross-border flows” (Krugman 2012). Indeed, a paper just released by the Fund  (IMF 2012) encapsulates, as Krugman notes, both recent practice as well as the analysis in a series of policy discussion notes published by Fund staff over the past couple of years (Ostry et al. 2010 and 2011 ). The institutional paper by the Fund recognises the important benefits that capital flows may bring, but also cautions about the risks. In light of such risks, it acknowledges that full liberalisation of the capital account may not be the right goal for all countries at all times, and it identifies circumstances where capital and prudential measures may be needed to safeguard macroeconomic and financial stability in the face of sudden surges and stops.
In thinking through the circumstances in which capital controls may be appropriate, a multilateral perspective is essential. Indeed this was a key tenet of the IMF’s founding fathers, Keynes and White. Keynes considered that managing capital flows would be much more difficult by unilateral actions than if movements of capital could “be controlled at both ends” of the transaction. White concluded that capital controls would be ineffective unless there was cooperation across countries in their implementation.
The perspective of Keynes and White is very much alive today. The G20 for example in its Coherent Conclusions on capital flow management last year urged that national policies to deal with capital flow volatility take account of potential cross-border spillovers. Many emerging market countries worry that policies in source countries are increasing the challenges of managing capital inflow cycles for them, underscoring the potential gains from cooperation.
In judging why the multilateral perspective is of the essence, four cases seem to be salient (these arguments are developed in a staff discussion note by Ostry, Ghosh and Korinek, 2012 ). The first is the possibility that capital controls can be used as a substitute for warranted external adjustment. When a country uses controls to sustain an undervalued currency, the resulting current account surplus is evidence of beggar-thy-neighbour behaviour. This is why one normally thinks that capital controls whose purpose is to frustrate external adjustment are multilaterally aberrant.
The practical implications are thornier, however, because gleaning the intent of a policy is very difficult. A country with an undervalued currency may impose capital controls to safeguard the stability of its financial system (rather than to limit external adjustment). In such cases, while one would normally want to see some currency appreciation as part of the adjustment process, one would not want to rule out the use of capital controls to protect financial stability if other less discriminatory measures lacked traction in the meantime.
The second case relates to the use of capital controls as a potential mechanism to manipulate the intertemporal terms of trade for the country – conceptually analogous to the use of tariffs to manipulate the goods' terms of trade. Our sense is that this is rarely, if ever, an issue in practice. However, it is not beyond credulity to think that some policies that affect capital flows can materially move world interest rates in a direction that benefits the country. To see this, ask yourself whether there are any large creditor countries that maintain restrictions on capital outflows; or whether any large debtor countries pursue policies that push down world interest rates. From a multilateral point of view, then, controls to manipulate the terms of trade should not be condoned.
The third case is when capital controls are used to address production externalities in the exportables sector. While production taxes-cum-subsidies would seem to be the natural response to such externalities, these measures may not be feasible if budgetary resources for subsidies are unavailable or if the sector to be taxed is informal and outside the tax net. In such a situation, a second best solution might be to pursue a policy of currency undervaluation supported by capital controls. Such a policy is second best because it distorts both consumption and production decisions, but the externality is only on the production side. As such, the larger effect on the trade balance is likely to have a beggar-thy-neighbour element to it, and thus likely to invite multilateral scrutiny along the same lines as the first and second cases considered above.
A fourth case relates to the use of capital controls for financial-stability purposes. If there are external effects from foreign borrowing (think of amplified crisis risks for the country, where the risks are not internalized by the borrower), then capital controls can act as Pigouvian taxes and constitute an optimal response at the country level, helping agents to internalize the external effects of their borrowing.
In a multi-country setting, the resulting spillovers from capital controls can amplify financial-stability risks for other countries by deflecting flows there. The existence of such spillovers is not per se an argument against the use of capital controls. If the imposition of capital controls is not itself costly for the country, a Nash equilibrium in which each country chooses a level of controls to address its own financial-stability risks will be globally efficient. More likely though, the imposition of capital controls will create costs (inter alia because the controls will imperfectly target the risky flows). Here coordination will be needed to achieve a globally efficient outcome. Coordination would likely need to involve both recipient countries (to minimise the risk of capital control wars and excessive mutual deflection of flows) and source countries (to ensure that they bear part of the cost burden when costs from controls are convex). In essence, the Nash equilibrium is no longer globally efficient because each country now has two targets (the domestic externality and the cost of using the capital control instrument), but only one instrument.
The need for coordination involving both source and recipient countries goes back to the Keynes-White notion of operating “at both ends of the transaction”. It speaks to the shared responsibility of source countries whose policies may contribute to excessive or excessively risky capital outflows. The policies in question encompass much more than capital controls, and include monetary policy that may encourage the purchase of risky foreign assets and weak prudential policies that allow excessive FX lending abroad to borrowers who are not hedged (think of the Sweden-Latvia example).
Coordination involving both source and recipient countries, while very much of the essence, will be more difficult to secure given that source countries may not always find it in their interest to bear (part of) the costs of greater financial stability in recipient countries. The purpose of coordination however is not to enjoin countries to pursue policies that are contrary to the domestic interest; rather, and in consonance with the recently adopted Integrated Surveillance Decision  by members of the IMF, it is to choose those policies that meet domestic objectives but have relatively less damaging outward spillovers for the rest of the world.
Krugman, P (2012), “The IMF and Capital Controls ”, The New York Times, 4 December.
IMF (2012), “The Liberalization and Management of Capital Flows: An Institutional View” , IMF Staff Paper, November.
Ostry et al (2010), “Capital Inflows: The Role of Controls ”, IMF Position Note, 19 February.
Ostry et al (2011), “Managing Capital Inflows: What Tools to Use? ”, IMF Staff Discussion Note, 5 April.
Ostry et al (2012), “Multilateral Aspects of Managing the Capital Account” , IMF Discussion Note, 7 September.