Ajuste recursivo con transformaciones invariantes y bootstrapping: El caso de una caminata aleatoria con intercepto

we use monte carlo simulations to study the performance of shin-so unit root test (dfss) under invariant transformation approaches and bootstrapping. if the alternative hypothesis is a stationary process around a linear trend, then the parametric bootstrap test is the best in terms of statistical power. however, if we transform the observations to build an invariant test, then the dfss test is the best. therefore, the recommendation is to use transformations of the invariant shin-so unit root test because its implementation is straightforward and less costly.