We propose a new approach to develop vulnerability lines that are explicitly anchored to the idea of a sub‐set of the population at risk of falling into poverty. We suggest that lines developed in this way can also be applied for the purpose of identifying the middle class (or “secure”). We illustrate that such vulnerability lines can be straightforwardly estimated with panel data, drawing on data from the USA and Vietnam. Importantly, given the relative scarcity of panel datasets, we show further that our method can be applied to synthetic panel datasets. We demonstrate this by means of an illustration using repeated cross‐section data from India. Our results indicate that in Vietnam and India during the 2000s, the population shares that can be designated as poor and as secure have, respectively, been falling and expanding, with the vulnerable share of the population remaining fairly stable. Sharply contrasting trends are seen in the USA.