the legal and actual intensity of controls. Restrictions are not always binding (i.e. effective) or some indirect
restrictions are simply not taken into account. As a result it may be a mistake to conclude that the market
is segmented. The private sector is extremely creative in finding ways to move capital internationally. In
countries with restrictions on capital mobility, the private sector has typically resorted to leads and lags
in average payments terms for exports and imports to evade legal controls on capital flows.4 Moreover,
the Eurocurrency market has played an important role in evading capital controls. So, one needs to go beyond
legal restrictions in assessing the extent of capital mobility.
The measure we use takes account of short-term financial integration because forward exchange rates exist
only for short horizons generally not exceeding one year.5 In the investigation design, the underlying financial
assets differ only with respect to currency of denomination and country-specific regulation (e.g. capital
controls and tax treatment), rather than with respect to asset-specific types of risk (e.g. default risk and
liquidity risk) or other risk characteristics.6
We argue that closed nominal interest rate differentials are most suited to capture differences in the intensity
of capital controls. This follows from the decomposition of onshore covered nominal interest rate parity
in Table 1. This may be demonstrated by distinguishing between covered nominal interest parity in onshore
markets (for comparable assets in different political jurisdictions and restrictions on cross-border capital
flows) and covered nominal interest parity in offshore markets (for comparable assets in the same political
jurisdiction and no restrictions on cross-border capital flows). Each of the components φDomestic, φForeign and
φEuro provides information on the source of the onshore covered nominal interest rate differential. φDomestic
measures the extent to which domestic controls are the cause of a non-zero onshore covered nominal interest
rate differential. Similarly, φForeign measures the extent to which foreign controls contribute to a non-zero
onshore covered nominal interest rate differential. φEuro measures deviations from covered interest arbitrage
in offshore markets. We assume that interest arbitrage ensures that differences from covered interest rate
parity in offshore markets (φEuro) are negligible. Since banks in Euromarkets set Euro-interest rates of the
domestic country (say the Euro-Sterling rate) equal to foreign Euro-interest rates (say the Euro-Deutsche
Mark rate) adjusted for the forward premium (discount) on the domestic currency, offshore covered nominal
interest parity will always hold in Euromarkets. Deviations in the Euromarket are largely due to technical
factors and/or transactions costs. Under this assumption, the domestic onshore-offshore interest rate differential
may be approximated by the adjusted domestic covered nominal interest rate differential (see Giavazzi and
Giovannini, 1989, p. 172).
4 Therefore, Milesi-Ferretti (1995) and Grilli and Milesi-Ferretti (1995) use a current account restrictions dummy variable to proxy
for the intensity of capital controls.
5 Although currency swaps allow us to calculate deviations from covered interest rate parity for longer horizons, they are not available
over a sufficiently long time horizon (see Popper, 1993).
6 Preferably, one might also want to disentangle the effect of interest rate withholding taxation. Interest withholding taxes importantly
affect the pre-tax gross return demanded by international investors. Huizinga (1994) adjusts the interest rate parity condition for
the effect of nonresident interest withholding taxation. Since the effects of interest rate withholding taxation are difficult to grasp,
we leave this aspect for further research.