Maximum likelihood is used to estimate a generalized autoregressive co
nditional heteroskedastic (GARCH) process where the residuals have a c
onditional stable distribution (GARCH-stable). The scale parameter is
modelled such that a GARCH process with normally distributed residuals
is a special case. The usual methods of estimating the parameters of
the stable distribution assume constant scale and will underestimate t
he characteristic exponent when the scale parameter follows a GARCH pr
ocess. The parameters of the GARCH-stable model are estimated with dai
ly foreign currency returns. Estimates of characteristic exponents are
higher with the GARCH-stable than when independence is assumed. Monte
Carlo hypothesis testing procedures, however, reject our GARCH-stable
model at the 1% significance level in four out of five cases.