As you know, the principal-agent problem stress the effect of asymmetric information between the principal and the agents on economic outcomes, the equilibrium in a principal-agent relation, under incomplete information can be characterized by the inefficiency relative to a situation of full information. Note that this inefficiency does not in itself signify the effect of special interest groups on policy process and the outcome.
While comparing a standard complete information social planner with the real world policy process, we must compare both of them on the same environment of incomplete information. When we do this we find that delegation of decisions under imperfect information can have significant implications only if Principal and agent do not have identical preferences. Otherwise i.e., when they have identical preferences there will be no problem of mechanism and process design. In policy process the great majority of relationships involve asymmetric information. The paradigm which public economics follows, takes the social planner as maximizing the welfare of the representation individual or of a weighted sum of individuals of different preferences where weights are predetermined. Hence this paradigm assumes no problem of agency. The key factor is the heterogeneity which gives rise to the agency problem, and makes the principal-agent paradigm relevant.