Clientelistic politics and pro-poor targeting: Rules versus discretionary budgets

Past research has provided evidence of clientelistic politics by local governments in delivery of private good benefits and manipulation of local budgets by elected officials at upper tiers. Using household panel survey data spanning 1998–2008 in West Bengal, India, we examine the consequences of replacing the observed allocation of local government, or gram panchayat (GP), program budgets based on discretion of higher level officials, with a grant allocation determined by a formula recommended by the 3rd West Bengal State Finance Commission (SFC) based on measures of village need. We assume that the allocation of benefits within GPs continues to be delegated to elected GP officials. We use the household data to classify them as ultra-poor, moderately poor, marginally poor, and non-poor respectively, depending on the number of deprivation dimensions applicable (landlessness, illiteracy and low caste status). In the next step, we estimate within-GP targeting patterns for different programs across these four groups, and how they are affected by the program grant received by the GP from upper tiers. This allows us to predict how targeting patterns would have changed, had the observed across-GP grant allocations been replaced by the formula-based allocation. We find that targeting of anti-poverty programs was progressive both within and across GPs while the targeting of public goods was not. This pattern is consistent with clientelistic opportunism of upper level officials. The SFC-rule based formula resulted in allocations that were less progressive than the observed allocation. Moreover, alternative formulae for across-GP budgets obtained by varying weights on GP characteristics used in the formula would have marginally improved pro-poor targeting. Hence, it is unlikely that switching to SFC formula-based grants would have improved pro-poor targeting.