Local politicians can function as crucial intermediaries between voters and party bosses in a clientelistic network. We study their role by matching data on 300 million welfare payments in the Indian state of West Bengal to village-level election returns. Local politicians systematically misallocate resources based on party loyalty and successfully deliver votes to their national co-partisans. Politicians are compensated for successful mobilization through a performance bonus immediately after the national election. The (promise of) increased compensation from government funds induces opposition candidates to switch to the ruling party in strategically important local councils, bringing them under its control.

Political checks and balances make different organs of government complementary in the exercise of power. We study how a party organization can seek and abuse these comple-mentarities to expand its influence. We combine data from the Indian state of West Bengal on elections across different levels of government. These are matched to 300 million pay-ments from a welfare scheme that requires approval from both state and local governments. Using a multidimensional close election design, we study the consequences of co-partisan alignment between these two tiers. The state government gives disproportionate funding to co-partisan local officials, who target core supporters to raise votes for co-partisan national candidates. Local officials are rewarded through diverted welfare payments, including a per-formance bonus immediately after the national election. The ruling party expands its power by recruiting opposition candidates in strategically important local councils, bringing even greater control over public funds.

Governments and NGOs have invested heavily in fighting corruption by designing anti- poverty programs that maximize transparency and accountability. We analyze whether corruption is still widespread in the context of one such program, a massive make-work scheme in India where every job spell is posted publicly online. Linking millions of administrative job records to local election outcomes, we measure how many jobs they self-deal. In the year after the election, winners of close elections receive 3 times as many workdays as losers and typical villagers, a sum equal to two-thirds their official stipend. We find that corruption persists because of a gap between de jure and actual transparency. Only when citizens have tools to access information in a timely manner does corruption eventually vanish.