آیا کنترل سرمایه در بازار ارز موثر است؟
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|14836||2013||18 صفحه PDF||سفارش دهید||9170 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of International Money and Finance, Volume 35, June 2013, Pages 36–53
One of the reasons for governments to employ capital controls is to obtain some degree of monetary independence. In this paper we test whether capital controls can reduce the link between exchange rates fluctuations and cross border interest differentials. Recent capital control proxies are used in order to determine the date of capital account liberalization for a panel of Western European and emerging countries. Results show that capital controls have a very limited effect on observed deviations from interest parities, even when accounting for the political risk associated with capital controls.
Academic and policy debates about vices and virtues of capital controls exist for a long time and opinions on their usefulness swing like a pendulum. The argument dates back as far as the mercantilists who sought to control flows of bullion. This ideological school was subsequently denounced by Adam Smith in favor of free markets. The 20th century saw a large revival of capital controls, driven by the war effort of both world wars. Afterwards the Bretton Woods system combined capital controls with fixing exchange rates. Keynes considered capital controls as an important cornerstone to financial stability during the Bretton Woods system – an idea revived by Tobin (1978). The meltdown of the Gold Exchange Standard induced a liberalization wave that lasted through the 1990s. The aftermath of the Asian crisis made some reconsider the wisdom of the widespread liberalization and the debate was re-oriented once more. Nowadays, a significant fraction of the academic community supports controls in specific circumstances, see e.g. Krugman (1999) or Rodrik (1998), and advocates of a well-thought phasing out of capital controls. The classic argument in favor of capital controls goes back to the welfare theory of the ‘second best’: in the presence of market imperfections (incomplete markets, asymmetric information, transaction costs etc.) additional distortions such as capital controls might be welfare enhancing by offsetting some of the previous distortions' negative welfare effects. However, capital controls remain a distortion and as such should only be maintained if the benefits outweigh the costs. And the potential costs are numerous: capital controls have to be regularly revised to close loopholes; time and resources have to be expended to execute the controls; controls limit the potential for portfolio diversification and decrease the amount of risk that can be shared and diversified (Voth, 2003) and increase the cost of capital for local firms (Bekaert and Harvey, 2000); capital controls potentially increase exchange rate volatility (Glick and Hutchison, 2005) and the risk of currency crises (Bordo et al., 2001). This paper's goal is not to perform a full-fledged welfare analysis to quantify direct and indirect costs and benefits of capital controls: our ambition is rather more modest in that we want to investigate whether capital controls are able to bring more ‘monetary freedom’ in the foreign exchange market. Loosely speaking, ‘monetary freedom’ can be thought of as the desire to manage domestic monetary policy in a more independent way from the exchange rate. Monetary freedom constitutes one of the classic motivations for governments to impose capital controls (Magud et al., 2011). The potential gain in monetary freedom allows governments to use the monetary and fiscal instruments more effectively together to steer the domestic economy. Magud et al. (2011) survey the literature and conclude that inflow controls (but not outflow controls) contributed to increased monetary freedom in several well documented cases like Chile, Colombia, Malaysia and Thailand. In a broader setting, including more countries, this effect cannot be replicated: Montiel and Reinhart (1999) show that capital controls do not improve the ability of monetary policy to change the composition of capital flows (although capital controls themselves may have a direct effect on the composition); Edison and Reinhart (2001) find that capital controls do not affect the co-movement of domestic and foreign interest rates, and Miniane and Rogers (2007) show that the presence of capital controls do not diminish the impact of U.S. monetary policy shocks on the domestic economy. In this paper we take an alternative route to assess the effect of capital controls on ‘monetary freedom’. We investigate to what extent capital controls contribute to deviations from the (covered and uncovered) interest parity conditions for foreign exchange. Given the potential of capital controls to limit arbitrage and speculation, exchange rate parity conditions constitute a natural testing framework for the ‘monetary freedom’ hypothesis: the well-known Covered and Uncovered Interest Parity relations relate cross-border interest differentials to current and future (expected) price formation in foreign exchange markets in the following way: equation(1) t(f−s)=t(i−i*),(f−s)t=(i−i*)t, Turn MathJax on equation(2) Etst+1−st=t(i−i*),Etst+1−st=(i−i*)t, Turn MathJax on with Et the rational expectations operator, st and ft the natural logarithms of the nominal bilateral spot and 1-month forward exchange rate, expressed in domestic currency per unit of foreign currency, and i and i∗ domestic and foreign interest rates on monthly deposits, respectively. 1 The Covered Interest Parity (CIP) condition (1) reflects that proceeds on foreign currency deposits should equal the proceeds on equal-maturity domestic deposits after converting the foreign proceeds into domestic currency via the forward market. The Uncovered Interest Parity (UIP) relation (2) implies that (risk-neutral) investors are indifferent between equal-maturity domestic and foreign deposits provided the (time t expected) deposit returns (converted in the domestic currency via the time t spot market expectation) are the same. Empirically there is substantial evidence that CIP holds well, although small but significant deviations are sometimes observed (Clinton, 1988; Fletcher and Taylor, 1996; Akram et al., 2008). Deviations from UIP are much more severe and widespread. A large number of possible explanations for this ‘forward discount bias’ puzzle have been proposed (and empirically tested) ranging from time varying risk premia (see e.g. Fama, 1984; Cavaglia et al., 1994; Wolff, 1987). Learning (Lewis, 1989), central bank intervention (McCallum, 1994) deviations from rationality (Frankel and Froot, 1987), peso problems (Kaminsky, 1993; Flood and Rose, 1996), or monetary volatility (Moore and Roche, 2012), but all with limited success. Comprehensive surveys on the forward discount bias and its potential causes include Chinn (2006), Engel (1996), and Sarno (2005). Only a handful of papers empirically investigated whether capital controls distort parity conditions.2Francis et al. (2002) show that deviations from UIP can be partly attributed to financial liberalization (although results are highly country specific). Phylaktis (1988, 1990) also finds that capital controls explain part of observed UIP differentials in several Latin American countries. These last two studies explicitly split the effects of capital controls into direct tax effects and indirect political risk effects. The political risk premium can be seen as an indirect effect of capital controls arising from expected shifts in the severity of capital controls. Within the frameworks of the cited articles, both effects exist and can partially explain deviations from parity conditions. Dooley and Isard (1980) give evidence that capital controls significantly impacted offshore-onshore interest differentials in Germany, again both through direct tax effects and indirect political risk effects. Finally, Holmes and Wu (1997) hardly find any evidence that the financial liberalization of Europe in the 1990s affected deviations from CIP. This paper considers regression equations of the framework outlined in Equations (1) and (2), augmented with proxies for capital controls and political risk premiums. Augmenting the parity regressions with those variables enables one to determine the contributions of these variables to observed deviations from CIP and UIP. Anticipating on our results, we find some evidence that capital controls can effectively distort the covered interest arbitrage condition. However, capital controls seem unable to explain even a fraction of the observed forward discount bias in the UIP relation. This also implies that capital controls are ineffective in creating more monetary freedom for domestic monetary authorities. For some countries, the installment of capital controls even leads to an erosion of monetary freedom. Surprisingly, these results are not fundamentally altered when controlling for political risk premia. The remainder of the paper is structured as follows. Section 2 provides a detailed discussion on the used regression methodology and how the capital control proxies and resulting political risk premiums are determined. In Section 3 we present and interpret outcomes of the CIP regressions and UIP regressions augmented with proxies for capital controls and other control variables. Finally, Section 4 provides some concluding remarks.
نتیجه گیری انگلیسی
Policymakers often cite several arguments for implementing controls on the free flow of capital across borders. Insulating interest rate policies from exchange rate policies (so-called ‘monetary freedom’) by weakening the link between interest rates and exchange rates is one of the those prime motives for installing capital controls. The question arises to what extent controls are effective in enhancing monetary freedom. In a world of perfect capital mobility, perfect asset substitutability and the presence of arbitrageurs and speculators, exchange rate changes and domestic and foreign interest rates are linked together by parity conditions such as the Covered Interest arbitrage relation (CIP) and the Uncovered Interest Parity condition (UIP). Hence, these two parity conditions constitute a natural framework to measure the capacity of capital controls to driving a wedge between interest rates and exchange rates. In fact, there is abundant empirical evidence of deviations from parities, particularly deviations from UIP. We argue that if capital controls would have the potential to bring more monetary freedom, this should contribute to these parity deviations in the foreign exchange market. On the one hand capital controls seem to have a small but significant impact on observed deviations from covered interest rate parity, both for European countries as well as emerging countries. Capital controls are also found to have a significant impact on offshore-onshore differentials. These findings are in contrast to the conclusions of Montiel and Reinhart (1999), Miniane and Rogers (2007) and Edison and Reinhart (2001) who find that capital controls do not impact monetary freedom. On the other hand, capital controls do not explain much of the observed deviations from Uncovered interest Parity. The European sample hardly provides evidence that capital controls partly cause deviations from UIP. In the emerging sample, capital controls only have something to say on the deviations from UIP provided the effects of inflow and outflows are separated, but the impacts on UIP deviations – if statistically significant – often appear with the wrong sign. Somewhat surprisingly, the inclusion of a political risk proxy as an additional factor in the UIP regressions hardly alters the results: neither capital controls nor political risk seem important determinants of the observed deviations from UIP. The latter outcome somewhat contrasts with earlier studies by Dooley and Isard (1980) and Phylaktis (1988, 1990) who report an important role for political risk in explaining interest differentials. It cannot be excluded that the different methodologies used for determining political risk proxies as compared to the current paper may play some role in this.