I recently approached to the study of time-series unit root tests with multiple breaks. The most applied test in this case - as far as I know - is the one proposed by Clemente, Montanes, and Reyes (1998). In STATA it is implemented through the commands clemio2 clemao2 for innovative and additive outliers, respectively. My question is specifically related to the output provided by STATA and the proper interpretation of it.

My variable is the following one: log(agricultural area/land area) for Costa Rica, a variable which span from 1961 up to 2015 in my database (source FAOSTAT). I first ran the following unit root tests: ADP, PP, DF-GLS, and KPSS. Only the PP test concluded for an I(1) variables while the other tests considered higher order of integration to obtain a stationary variable. Despite those results I conducted the Clemente-Montanes-Reyes test of unit roots with two structural breaks. Testing for additive outliers, the p-value for the two structural breaks resulted positive while the t-statistic resulted to be inferior to the critical value reported by STATA (5%), actually it resulted lower even by the value of 2,5% (reported in Table 1 of the author’s paper).

Thus, my question is the following: did I interpret the results in a proper way or I am irreparably wrong? Furthermore, according to these results can I conclude that my time-series with the two structural breaks provided by the test is stationary?

In attachment there is an image of the output provided by STATA for the command clemio2.

Thank you in advance.