- 17 -
exceed the foreign rate by a devaluation risk premium, but different currency crises
models differ on how this premium is modeled. This literature also generates interesting
insights on what the UIP differential should depend on. In the case where there is a
positive probability, say ρ of the devaluation and where the rate of devaluation, if it
occurs is known at δ. Then, the UIP relation will be:
i = i* + ρδ * st
The first generation crises models of Krugman (1979) and Flood and Garber
(1984) the fundamentals, i.e. the ability of the central bank to defend the exchange rate
are continuously deteriorating because of a fiscal policy inconsistent with a monetary
one. The interest rate in these models is fixed until the date of depreciation when it
jumps. If you add uncertainty in the level of reserves that the central bank will commit to
the exchange rate defense and limit the mobility of capital, then one gets more interesting
parity conditions like the one suggested by Artus (1994):
i = i* + set+1 -λ*dt
where
∆dt+1 =i* - α(st -pt)+zt
Where dt is the external debt of the domestic country, λ is a constant inversely
proportional to the degree of capital mobility, pt is the price level, z is the external
balance. Other modifications of the UIP7 include adding a self-fulfilling mechanism by
specifying the UIP as follows [Flood and Marion, 1998]:
i = i* + ∆set+1 + xvar(st+1)(bt -bt* -st)
7 Discussed in Arias (2001)