It is argued that standard impulse response analysis based on vector autoregressive models has a number of
shortcomings. Although the impulse responses are estimated
quantities, measures for sampling variability such as
confidence intervals sometimes are not provided. If
confidence intervals are given, they often are based on
bootstrap methods with dubious theoretical properties. These
problems are illustrated using two German monetary systems.
Proposals are made for improving current practice. Special
emphasis is placed on systems with cointegrated
variables.
