The identification of trends and cycles is often a challenging task under sizeable changes in economic conditions. We solve this problem with a flexible unobserved components model, featuring an (unobserved) evolving trend inflation drift to cope with distinct inflationary periods and data-driven low frequency movements to partly influence ex ante key trend components. In the long run the model displays a balanced growth path, in addition to other standard restrictions (e.g. nil output and labour market slacks). We estimate the model with Bayesian techniques using two datasets, one for the euro area and another for Portugal, two economies displaying distinct macroeconomic environments over the last four decades, and conclude that Portugal witnessed (i) a steeper deceleration of potential output, since the 1990s; (ii) a pervasively higher volatility in labour and product markets; and (iii) a long-lived interruption in convergence trends after the 2000s. Results are robust to sensitivity analyses. Parameter uncertainty is, nevertheless, significant.