Discursive institutionalism is an approach that looks at policy discourse as a legitimacy and consensus building resource. Two of its limitations are that it does not clarify the relationship between policy discourse and policy outcomes, and that it does not explain how discourse should be studied empirically. The paper claims that in order to overcome these limitations we need to look deeper into the mechanisms of the discursive interactions and study discourse in the specific form it takes. Drawing on Stone (2002) as well as Jones and McBeth (2010) the paper looks at the narratives and the narrative strategies the key policy actors develop during the policy process so that new political alliances are created, empowered and legitimated to fix a policy problem in a certain way. The paper understands narratives as causal stories with a beginning, middle and end, which aim at attributing cause, blame and responsibility over a policy problem (Stone 2002; Roe 1994). Conceptually the paper shows that narrative analysis is a fruitful methodological tool for discursive institutionalism. Empirically, it looks at the policy narratives that framed the Greek case of pension reform during 1990-2002. The empirical evidence shows that the trade unions’ narratives empowered them as the only actors who could be entrusted with reform by means of a strong normative component, and the of political alliances with policy experts and employers’ associations.
