Time-series analysis of weekly initial claims over 1967–2012 reveal the following: (1) Initial claims are highly seasonal and cyclical, but do not follow a specific trend. Seasonality follows a “W” pattern over the 52 week period. (2) Initial claims are subject to conditional volatility and volatility clustering. The EGARCH and CGARCH specifications provide reasonable representations of the conditional volatility. The former suggests the existence of asymmetries in conditional variance. The latter implies that both permanent and transitory shocks affect volatility, but the effect of permanent shocks is more pronounced. (3) Both models perform well in terms of forecasting as well as within- and out-of-sample model selection criteria.