Democratic accountability relies on voters accurately evaluating the government's performance on tackling the important issues of the day. However, partisan identification can bias these evaluations - partisans perceive outcomes more favourably when their party is in power, and they hold their party less responsible for bad outcomes than good outcomes. In this paper we propose two additional ways in which partisans might arrive at biased judgements of government competence. Partisans may recall the past performance of their party more positively than they perceived it at the time, and they might extrapolate from their party's past performance to future performance more positively than they would for other parties. We use a survey experiment of over 6000 respondents to test these four possible forms of motivated reasoning, using vignettes to prime respondents about the UK government's successes and failures in tackling the Covid-19 pandemic. Importantly, half of our sample were also surveyed a year prior to the experiment, allowing us to objectively measure whether the past is viewed through 'rose-tinted glasses' by government partisans.