and country premia; those that relate the combined risk premium ρt or individual com-
ponents of it, i.e. ρtE or ρtD, to macroeconomic fundamentals which might reflect market
perception of risk; and those that analyze time-series properties of risk premium as defined
in equation (11).
Frankel and Okongwu (1996) and Domowitz et al. (1998) analyze currency and country
premia for Mexico in the early 1990s. They report that both the currency and country
premia are economically large, with the currency risk premium being higher and more
volatile. Accordingly, one may conjecture that deviations from the UIP condition for Mexico
during this period stem mainly from currency related risks rather than country risks. A
number of studies relate the currency and country premia with macroeconomic variables
that are expected to gauge market perception of risk such as Werner (1996), Schmukler
and Serven (2002), Rojas-Suarez and Sotelo (2007), Bratsiotis and Robinson (2007), and
Poghosyan et al. (2007).
Schmukler and Serven (2002) explore the patterns and determinants of currency pre-
mium for Argentina under the currency board. Their results suggest that imports-to-
international reserves, current account deficit-to-GDP and public deficit-to-GDP ratios,
and the liquidity positions of local banks, as well as macroeconomic conditions in other
emerging markets reflect an exchange rate risk premium, ρtE , for Argentina.
Using eurobond interest rate as a proxy for default risk, ρtD , Rojas-Suarez and Sotelo
(2007) analyze the relationship between default risk and interest rates on domestic assets
for various Latin American countries. Their results show that there is a unidirectional
Granger causality from ρtD to it and itf , implying that if ρtD is not taken into account, the
UIP regressions will produce biased estimates. In addition, redefining ρD as iEB — i^k, they
report that default risk premium is indeed related to macroeconomic fundamentals, such as
ratios of external debt-to-government revenue, government liabilities held by banks-to-total
bank assets, as well as global liquidity conditions.
Werner (1996) relates the time-varying risk premium to the relative share of foreign
currency denominated government debt through a simple portfolio model of Dornbusch
21