Capital controls as a means of minimising speculative bubbles in real exchange rates: key features of the literature and its application to China and India

Economic Papers (Economic Society of Australia), Sept, 2003 by Craig Applegate

1 Introduction

The imposition of controls on capital flows has the potential to impose several well-known costs on an economy. The most important of these costs is a trend redaction in the rate of growth resulting from higher local interest rates (Hamada and Takeda 2001). Capital controls can also impose other costs. Residents have a reduced ability to smooth their consumption intertemporally through adjustments in the current account deficit. Residents are also less able to diversify their investment portfolios internationally or to make use of exchange rate futures and swap markets. Capital controls can also act as an inducement to corruption (Edwards 1999a).

Since the Asian financial crisis there has been a renewed focus on the possible offsetting benefits of capital controls in a world of highly mobile short-term capital flows (Bhagwati 1998). Fane (2000) outlines some arguments for imposing controls on capital flows. The four first-best arguments he proposes relate to sovereign default risk from Aizenman (1989), accounting for the effects of the tax system, seigniorage revenue and accounting for the existence of pre-existing trade restrictions.

In the context of a floating exchange rate regime, one possible benefit of controls on international capital flows is that they might be able to reduce the level of volatility in real exchange rates that results from the formation of positive or negative speculative bubbles (Hamada and Takeda 2001). Speculative bubbles can be found in the behaviour of real exchange rates just as they can affect stock or property prices. During the Asian financial crisis there was substantial overcorrection of the real exchange rates of Korea, Thailand, Indonesia, the Philippines and Malaysia (Edwards 1999b, I.M F. 2000a, Hamada and Takeda 2001).

Meng and Velasco (1999) developed a model in which a high degree of capital mobility can lead to the real exchange rate following an indeterminate path that is subject to sudden and potentially large movements. Bubbles can be considered as being a manifestation of such multiple equilibria intertemporal models (Stiglitz 1990). Frenkel, Nickel, Schmidt and Stadtmann (2001) extend the Dornbusch (1976) overshooting model to incorporate a proportion of foreign exchange traders whom are assumed to be chartists, rather than fundamentalists. In their model, the presence of a Tobin-style tax on capital movements can act to reduce the real exchange rate volatility that results from a shock.

Reduced volatility in real exchange rates is not one of the possible benefits of capital controls that were listed in Fane (2000). Fane believes that floating exchange rates would in fact be more volatile in the presence of controls on short-term capital, as such flows serve to smooth the lumpy supply of and demand for currencies that results from trade and direct foreign investment flows. While speculative flows can serve to smooth real exchange rates, this argument does not make adequate allowances for the possibility of short-term herd behaviour that leads to the formation of substantial positive or negative speculative bubbles in the real exchange rate.

In the context of a full-employment classical macroeconomic model with no nominal rigidities in either prices or wages, it is difficult to mount a case that there is anything undesirable about a real exchange rate that exhibits substantial and sustained deviation from purchasing power parity. Changes in the real exchange rate can form an important means of re-allocating resources towards industries in which they earn the highest rate of return.

In a context of a two sector New Keynesian macroeconomic model with short-run nominal rigidities in either prices or wages, such overshooting of real exchange rates can impose short-run adjustment costs on an economy as factors move back and forth between the traded and non-traded sectors of the economy. A real depreciation can cause structural unemployment in the non-tradable sector, while a real appreciation can cause structural unemployment in the tradable sector. In the context of an overcorrecting real exchange rate, the economy can be constantly undergoing costs associated with restructuring. While such adjustment costs are difficult to quantify, this paper assumes both that these costs exist and are substantial. Hamada and Takeda (2001) make a similar assumption. During the Asian financial crisis between 1997 and 1999, after the currencies of Indonesia, Korea and Thailand were floated the initial substantial real depreciations were subsequently partially unwound (Hamada and Takeda 2001). These nations also moved from having substantial current account deficits to substantial current account surpluses in response to both their substantial real depreciations and slowing domestic economies.

Thailand provides a case in point. To pick out some relatively extreme quarterly results, Thailand experienced a high current account deficit of 7% of G.D.P in quarter 2 of 1997 and this had reversed to an annualised current account surplus of 13.7% of G.D.P. in quarter 1 of 1998 (I.M.F. 2000b). Thai unemployment was highest amongst those who were previously employed in the non-tradable construction centre 11.4%, followed by the crisis hit banking and insurance sectors 4.5%. The tradable mining, and manufacturing sectors were relatively unscathed with unemployment rates of 3.8 and 3.4% respectively (Krongkaew 2001). The overall unemployment resulting from the macroeconomic recession in Thailand tended to mask the structural unemployment resulting from the over-correcting exchange rate. Part of the slump in Thai real estate prices was the consequence of the previous bubble.