
 Start

2183
 Prefix

A direct implication of the GDH is that the
interest rates of other EMS member countries are cointegrated with the German
interest rate, with the German interest rate playing the leading role.
However, Karfakis and Moschos (1990), Katsimbris and Miller (1993),
 Exact

Hassapis et al. (1999), and
 Suffix

Caporale et al. (1996), among others, report evidence that shortterm
interest rates in EMS countries are not cointegrated with the German interest rate.1 The
absence of a common trend in the bivariate systems of EMS and German interest rates
refutes the monetarypolicy objectives of the EMS, and suggests the absence of
convergence of European monetary policies.
 (check this in PDF content)

 Start

3591
 Prefix

Katsimbris and Miller (1993) estimate trivariate errorcorrection models
including the U.S. interest rate in the system and, based on Granger causality tests,
report negative evidence for the GDH, finding that EMS interest rates respond to each
other as well as to the U.S. interest rate.
 Exact

Hassapis et al. (1999)
 Suffix

apply the Johansen
cointegration methodology to systems of EMS interest rates extended to include the
U.S. interest rate. They find that EMS interest rates are cointegrated with the U.S.
interest rate but not with the German interest rate.2 They identify shortrun intraEMS
linkages but, in most cases, they report that German interest rates are caused by, rather
than cause, the interest rates of t
 (check this in PDF content)

 Start

7679
 Prefix

Given the empirical evidence that the spot returns series t+1s−ts is a
martingaledifference process (see, inter alia, Meese and Singleton (1982) and Baillie and
Bollerslev (1989)), nonstationary exchange rate returns are unlikely. Therefore,
nonstationary interest rate differentials should reflect nonstationary foreign exchange
risk premia.
In contrast to this prediction, studies by
 Exact

Fama (1984),
 Suffix

Hansen and Hodrick
(1980), Hodrick and Srivastava (1984), Korajczyk (1985), and Wolff (1987), among
others, find evidence consistent with stationary timevarying currency risk premia.6
More recently, Shively (2000) provides emphatic evidence supportive of timevarying
risk premia evolving as stationary processes in the postBretton Woods era.
 (check this in PDF content)

 Start

7703
 Prefix

evidence that the spot returns series t+1s−ts is a
martingaledifference process (see, inter alia, Meese and Singleton (1982) and Baillie and
Bollerslev (1989)), nonstationary exchange rate returns are unlikely. Therefore,
nonstationary interest rate differentials should reflect nonstationary foreign exchange
risk premia.
In contrast to this prediction, studies by Fama (1984), Hansen and
 Exact

Hodrick (1980),
 Suffix

Hodrick and Srivastava (1984), Korajczyk (1985), and Wolff (1987), among
others, find evidence consistent with stationary timevarying currency risk premia.6
More recently, Shively (2000) provides emphatic evidence supportive of timevarying
risk premia evolving as stationary processes in the postBretton Woods era.
 (check this in PDF content)

 Start

13916
 Prefix

In a targetzone regime, the interest rate differential can be written as
it−t
i*=
Ett+1c−tc()+tEt+1x−tx(),(6)
where tc is the central parity and tx is the proportionate deviation from the central
parity. tx, informally referred to as the exchange rate within the band, is theoretically
meanreverting
 Exact

(Krugman 1991) and
 Suffix

it empirically exhibits strong autocorrelation
(Rose and Svensson 1995). Should the exchange rate within the band exhibit long
memory, the interest rate spread may inherit such stochastic behavior. Fractionally
differenced intraEMS interest rate spreads could also possibly reflect the presence of
strongly dependent risk premia of the EMS currencies relative to the Deutsche mark.
3.2 Estimation o
 (check this in PDF content)

 Start

13991
 Prefix

In a targetzone regime, the interest rate differential can be written as
it−t
i*=
Ett+1c−tc()+tEt+1x−tx(),(6)
where tc is the central parity and tx is the proportionate deviation from the central
parity. tx, informally referred to as the exchange rate within the band, is theoretically
meanreverting (Krugman 1991) and it empirically exhibits strong autocorrelation
(Rose and
 Exact

Svensson 1995).
 Suffix

Should the exchange rate within the band exhibit long
memory, the interest rate spread may inherit such stochastic behavior. Fractionally
differenced intraEMS interest rate spreads could also possibly reflect the presence of
strongly dependent risk premia of the EMS currencies relative to the Deutsche mark.
3.2 Estimation of the Fractional Error Correction Model
The finding that the interest rat
 (check this in PDF content)