A Note about Detection of Additive Outliers with Fractional Errors
Perron and Rodríguez (2003) claimed that their procedure to detect for additive outliers (_ d) is powerful even when we have departures from the unit root case. In this note, we use Monte-Carlo simulations to show that Td is powerful when we have ARFIMA (p; d; q) errors. Using simulations, we calculate the expected number of additive outliers found in this context and the number of times that the approach Td identifies the true location of the additive outliers. The results indicate that the power of the procedure Td depends of the size of the additive outliers. When we have a DGP with big sized additive outliers the percentage of time that Td detects correctly the location of the additive outliers is 100.0%. A comparison between Td and the procedure TRAMO-SEATS is also included.
Keywords
Additive Outliers, Detection of Additive Outliers
JEL Classification
C2, C3, C5