More on F versus t tests for unit roots when there is no trend

Rodrigues and Tremayne (2004) interpret a problematic size result in a Monte Carlo study reported in Elder and Kennedy (2001) as arising from Elder and Kennedy's use of an inappropriate testing equation. In expositing their result, Rodrigues and Tremayne inadvertently lead readers to believe that the Elder and Kennedy conclusion is in error. We clarify the Rodrigues and Tremayne contribution, putting the validity of the Elder and Kennedy result in proper perspective and underlining the important role played by the starting value in Monte Carlo analyses.