There is an open debate in the literature about whether voters in a democracy benchmark policy outcomes against those elsewhere, punishing (rewarding) incumbents for national outcomes that underperform (outperform) an international or regional comparison. In this paper we argue that the effect of benchmarking upon incumbents’ accountability is contingent on the specific region against which outcomes are compared. Historical developments within nations may separate regions into rival communities. A rival community may turn into out-group community when its members persistently prompt negative emotions and sentiments among the members of the in-group. Our main hypothesis is that the effect of benchmarking will more prominent when benchmarking is against an out-group community as compared to non-out-group communities. We test this hypothesis using a benchmarking experiment conducted separately in three Spanish regions - Catalonia, Madrid and Valencia - that vary in their rivalry connections. We treat individuals in each region with information that compares negatively the capacity of their region in fighting the COVID pandemic against: a) the national average; b) a non-out-group region; c) an out-group region. We then test whether regional incumbents are more severely punished when the negative comparison is against the out-group region.