Abstract
The simultaneous estimation of missing obseroations with the parameters of a model is compared to separately interpolating existing data by application of related series to estimate missing obseroations prior to estimation of model parameters. Missing obseroations are simulated on a complete time series. The basis of the comparison is the deviations between estimates of model parameters when no obseroations are missing and estimates of the same model parameters when the simulated missing obseroations are estimated. No support for favoring the simultaneous approach over the two-step interpolation approach is found.
Get full access to this article
View all access options for this article.