1999 年 29 巻 2 号 p. 181-200
Many studies on economic time series focus much attention on unit root processes and there is some concern about time series data with missing observations. However most previous research on unit root processes was done only for complete observed data. This paper therefore considers unit root processes with missing observations, proposes two estimators for them and investigates their asymptotic properties (limiting distributions and order of convergence). The properties depend on the pattern of occurrence of missing observations. As an application of the main results, we also consider unit root tests for data with missing observations.