When two policies are available to achieve the same goal why is the relatively inefficient one often observed? We address this question in the context of policies used to redistribute income towards special interest groups (SIGs) where in the first stage the constraints on policy instruments are chosen and in the second the government bargains with SIGs over the level of the available policies. Restrictions on the use of efficient policies and the use of inefficient ones reduce the surplus over which SIGs and governments can bargain but it also improves the government's bargaining position thus increasing its share of the surplus. The positive effect for the government dominates under plausible conditions. Inefficient policies are the equilibrium outcome under alternative policy selection mechanisms, e.g., election of policymakers and bargaining between SIGs and the government. The model also explains the coexistence of transfer policies. Moreover, we show why a weak government is more likely to choose the inefficient transfer and discuss how this result may be tested.