What explains the persistence of unemployment? The literature on hysteresis, which is based on unit root testing in autoregressive models, consists of a vast number of univariate studies, i.e. that analyze unemployment series in isolation, but few multivariate analyses that focus on the sources of hysteresis. As a result, this question remains largely unanswered. This paper presents a multivariate econometric framework for analyzing hysteresis, which allows one to test different hypotheses about non-stationarity of unemployment against one another. For example, whether this is due to a persistently changing equilibrium, slow adjustment towards the equilibrium (persistent ‡fluctuations), or perhaps even a combination of the two. Different hypotheses of slow adjustment, as implied by theories of hysteresis, nominal rigidities or labor hoarding can also be compared. A small illustrative application to UK quarterly data on prices, wages, output, unemployment and crude oil prices, suggests that, for the period 1988 up to the onset of the …financial crisis, the non-stationarity of UK unemployment cannot be explained as a result of slow adjustment, including sluggish wage formation as emphasized by the hysteresis theories. Instead, it is the equilibrium that has evolved persistently as a consequence of exogenous oil prices shifting the price setting relation (in the unemployment-real wage space) in a non-stationary manner.