Breaking the curse? The politics of drought in the Brazilian Northeast

How have decades of government interventions to combat the effects of drought interacted with politics and local governance in Northeast Brazil? Preliminary evidence indicated that large and sustained federal anti-drought initiatives have improved the living conditions of local populations in heterogeneous ways. While some municipalities have managed to improve considerably their indicators in comparison to non-recipient municipalities, others have failed terribly. Such reality raises questions about why some municipalities perform better than others, particularly in a region in Brazil described as being historically plagued by patronage and clientelism. To understand the relationship between drought-related aid and local politics, we applied a mix of methodologies: (i) a natural experiment that allows us to compare otherwise similar municipalities that have arbitrarily received substantially different amounts of government aid for dealing with the effects of drought, (ii) extensive field work in the region, and (iii) surveys and embedded experiments to measure the prevalence of clientelism, political manipulation of drought relief, and other aspects of local governance. The project has organized a dataset with all pairs of neighboring municipalities along the Semi-arid Zone (SAZ) border. The SAZ is the administrative region that provides its members with more aid to mitigate drought effects. The dataset includes data from Brazilian rural censuses (2006 and 2017), municipal-level data of anti-drought programs from Brazilian ministries, electoral results and weather. The project has also conducted four fieldwork trips, visiting 24 municipalities in the states of Pernambuco, Piauí, Paraíba, and Bahia. Moreover, we did a fieldwork trip to Brasília to interview top-level policymakers responsible for main anti-drought programs. In total, we interviewed more than 70 local authorities, leader unions, bureaucrats and residents. Field work was also supplemented by 20 interviews with representatives of NGOs in the region responsible for implementing a program funded by the federal government for building rainwater cisterns in the region. The surveys were conducted during October/November 2019 in 90 SAZ municipalities distributed along 7 different Northeast states. They covered both households (3600 interviews) and local elites (360 interviews). Both surveys covered topics related to clientelism, vulnerability to drought, electoral corruption, and dominance of non-party political groups (i.e. traditional families). They both included vignette and conjoint analysis experiments to test our main hypotheses about political manipulation of programs. Our analysis showed an impact of being part of the SAZ on the level of electoral competition at the municipal level. The result is that SAZ municipalities present a systematic lower level of competition than their non-SAZ neighboring municipalities. These results confirm partially our hypothesis of the negative effects of drought aid on electoral competition. Local politicians seen to take advantage of these resources to consolidate their political dominance by increasing costs for other candidates to compete in elections. However, incumbents in SAZ municipalities does not present a higher likelihood of being reelected in comparison to their non-SAZ counterparts. This result seems to be related to how programs have been implemented during the last 15 years. One important finding from fieldwork was the role played by local rural workers unions and state level bureaucrats responsible for rural assistance. These actors operate as gatekeepers for many anti-drought programs. Their influence rely on two important prerogatives. First, they have the power of issuing documents that grant small farmers access to anti-drought programs. The main document is called DAP (Declaração de Aptidão ao PRONAF). It grants access to different programs such as subsidized credit and crop insurance schemes. Second, they active participate in local councils responsible for allocating water cisterns for households. By using these prerogatives, strong rural workers unions and bureaucrats can strongly influence program implementation and local politics. Based on these findings we obtained data on issued DAPs containing information about both issuers and beneficiaries. Our analysis shows that issuers tend to become candidates, win votes and to be elected more often than other policy administrators. NGOs also played an important role in managing programs. We identified that cisterns built by NGOs produced clear electoral benefits for the federal government, but not for local politicians. Survey results corroborate these findings as programs tend to be perceived by local residents as following technical criteria instead of political ones. In sum, program design and characteristics of local civil society play an important role in determining the potential clientelistic use of drought aid.

In general terms, our project has contributed with a more nuanced description about the political dynamics involving drought aid in the Brazilian Northeast. Changes in the design of how drought-relief programs have empowered new actors, reducing discretion of local politicians over the distribution of these resources. The project also brought new evidence about the importance of civil society organizations in local politics and how other issues, such as access to healthcare services is a top priority in the region. Moreover, the project contributed to the broader academic community interested in Brazil with the development of a dataset containing the geolocation of polling stations in the country. Publications and events conveying project results to stakeholders are expected to contribute to the further improvement of drought-relief programs in the region with potential lessons for similar initiatives in other countries.

Our objective is to study how decades of government interventions to combat the effects of drought have interacted with politics and local governance in Northeast Brazil. Preliminary quantitative evidence indicates that large and sustained federal anti-drought initiatives have improved the living conditions of local populations in a very heterogeneous way. While some recipient municipalities have managed to improve considerably their indicators in comparison to non-recipient municipalities, others have failed terribly. Such reality raises questions about which characteristics of these municipalities explain why some of them perform better than others, particularly in a region in Brazil described as being historically plagued by patronage and clientelism. To understand the relationship between extreme weather-related aid and local politics, we will apply a mix of methodologies: (i) a natural experiment that allows us to compare otherwise similar municipalities that ? for arbitrary reasons ? have received substantially different amounts of government aid for dealing with the effects of drought, (ii) surveys and embedded experiments to measure the prevalence of clientelism, political manipulation of drought relief, and other aspects of local governance, and (iii) paired comparisons with selected municipalities in order to understand better the mechanisms behind the different outcomes. The generated knowledge will help us to answer important questions about how aid in general interacts with political and institutional dynamics at the local level. Going beyond the common knowledge that institutions matter, we aim to understand which scope conditions and mechanisms contribute to disaster aid actually producing a positive rather than a negative role in development and consolidation of democratic institutions at the local level.