ABSTRAK. Model autoregressive conditional heteroskedasticity (ARCH) dan generalized autoregressive conditional heteroskedasticity (GARCH) merupakan model runtun waktu dengan variansi residu tidak konstan. Nilai tukar euro terhadap rupiah dapat dipandang sebagai satu fenomena data runtun waktu dengan volatilitas tinggi.

                Penelitian ini dimaksudkan untuk memodelkan dan meramalkan nilai tukar euro terhadap rupiah menggunakan data nilai tukar dari tanggal 28 Januari sampai 28 Desember 2007.

                Berdasarkan kriteria Akaike Info (Akaike Info Criterion) dan kriteria Schwarz (Schwarz Criterion) dan melalui metode BHHH yang diperkenalkan oleh Berndth, et al., diperoleh model heteroskedastisitas bersyarat yang paling sesuai adalah model GARCH (1,1) dengan model AR (1) pada model rataan bersyaratnya. Ramalan kurs jual euro terhadap rupiah untuk tanggal 2, 3, 4, 7, 8 dan 9 Januari 2008 berturut-turut adalah 14.492, 61, 14.502, 16, 14.511, 19, 14.520, 25, 14.529, 31 dan 14.538, 38 rupiah.

Kata kunci : runtun waktu, heteroskedastisitas, volatilitas, ARCH, GARCH, Metode BHHH

ABSTRACT. Autoregressive conditional heteroskedasticity (ARCH) and generalized autoregressive conditional heteroskedasticity (GARCH) models are time series models with non constant of residual variance. Exchange rate of Euro to rupiah can be considered as a time series data phenomenon with high volatility.

This research is aimed to model and forecast the exchange rate of euro to rupiah using the exchange rate data from 28 January to 28 Desember 2007.

Based on Aike Info and Schwarz Criterion and through BHHH method proposed by Bertnat et al., it can be concluded that the most appropriate model for this phenomenon is GARCH (1,1) and its conditional mean model is AR (1).  The forecasting of exchange rate for 2, 3, 4, 7, 8, and 9 January 2008 is 14.492, 16, 14.502, 16, 14.511, 19, 14.520, 25, 14.529, 31 and 14.538, 38 rupiah, respectively.

Key words : time series, heteroscedasticity ,volayility, ARCH , GARCH, BHHH method.