An Evaluation of the Policy Response to Drought in the City of São Paulo, Brazil: An Election Cycle Interpretation of Effectiveness

Abstract

Drought can have large, negative impacts on livelihoods and development outcomes in low- and middle-income countries. This highlights the need for drought response policies that can mitigate these impacts. We evaluate the policy response to the 2014–15 drought in Brazil that reduced the supply of water to the largest city in South America, São Paulo, by approximately one-third. Using microdata on household water consumption and a difference-in-difference design, we find that a penalty-based instrument induced household conservation behaviour but that a reward-based instrument did not. We examine why the reward-based instrument, which was both ineffective and expensive, was implemented at all. Our suggested explanation lies in political budget cycle theory. Exploratory tests imply that the reward-based instrument increased the share of votes to the incumbent governor. Penalty-based instruments are the technically effective drought response, but water sector decision makers in developing countries may need to contend with the distortionary effect of electoral cycles to implement them.

Acknowledgements

Claudio Lucinda acknowledges support from CNPQ grant 304446/2016-5. ORCID: 0000-0002-2190-9497. The survey data used in this research are available by request from the Foundation Institute for Economic Research at the University of São Paulo. The data and code used in this research are available from the authors on request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplementary Materials are available for this article which can be accessed via the online version of this journal available at https://doi.org/10.1080/00220388.2020.1786061.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1. Incentive Program for the Reduction of the Consumption of Water (Programa de Incentivo à Redução do Consumo de Água).

2. A ‘bonus’ system was also put in place in March 2004 – another electoral year, but for mayoral elections. However, in that year the drought was less severe, and with the resumption of rainfall, water levels rose to a point that the bonus system was called off without any need for other measures.

3. The Cantareira system is a massive network of reservoirs, tunnels, canals, and pumping stations that abstracts water from rivers in the interior of São Paulo state and transports it 80 km to the MRSP where it serves about half of all households there.

4. The regulatory ruling that implemented the contingency tariff stated that the bonus had not been as effective as hoped at reigning in household water consumption. It stated that while the bonus had been paid to many households, ‘24 per cent of users [had] increased their consumption and exceeded the pre-drought mean consumption level previously defined under the [bonus] Program, despite a public campaign calling for the rational use of water and the widespread water shortage’ (ARSESP, 2015a).

5. In the final minutes of the last televised candidate debate before election day, the incumbent governor stated that the drought was the most severe in the history of the state, that billions of Reals had been invested in the state’s water and sanitation sector during his governorship, and that the ‘bonus’ program had been implemented to promote water conservation ‘on the account of SABESP’ (RBA (Rede Brazil Atual), 2014). The challenger candidates argued that the governor’s administration had been negligent in planning and maintaining the state’s water supply infrastructure, which caused the crisis, and that the governor’s response to the crisis had been late and ineffective.

6. Other information gathered by the institute included the household’s street address; the water system serving the household; household income; whether the household owns a washing machine, dishwasher, shower, and/or other appliances; whether the household owns or rents the dwelling it occupies; the number of rooms in the dwelling; and the number of occupants. In the data, these household characteristics are observed only in the cross section. They are relevant to predicting household water consumption, but most are made redundant in the estimation strategy by household fixed effects.

7. The institute designed the sample to represent 95 per cent of the earnings distribution of the households in São Paulo municipality. The sampling frame was the complete customer registry of AES Electropaulo, an electric power company with approximately 6 million customers in the MRSP. The institute used a random systematic sampling criteria to select the households from the frame into the sample. Household electricity consumption was used as a proxy for household income. The institute then conducted a single interview with each sampled household during which it collected information about the social and demographic characteristics of the dwellers and the structural and material characteristics of the dwelling. The household’s SABESP account information was also recorded and used to download the household’s monthly water bill data from SABESP’s online billing portal.

8. Hospitals, first aid stations, health clinics, police stations, prisons, and detention centres were also exempt.

9. Available on request.

10. We also tried dealing with the endogenous movers problem through the propensity score matching procedure. We created a dummy variable for whether the household crossed the 10 m³ consumption threshold, in either direction, at any time during the 12 months preceding the drought (MOVER_i – see Table 1). Then, we used this variable and others to estimate the propensity of the household being in the high-consuming treatment group during the drought period. The results were similar to those presented in the next section. This approach helps balance the characteristics of households across control and treatment groups, but it does little to mitigate the problem of households actually moving between treatment groups. The only real remedy to this later problem seems to be dropping the mover households.

11. We also tried restricting the sample to the households that, prior to the start of the contingency tariff, did not experience a consumption fall of 20 per cent or more. The approach in doing so was that, if a household did not change its behaviour in line with the bonus before the contingency tariff, then it was unlikely to do so during the contingency tariff. The contingency tariff estimate remained within the reported range.

12. For example, water authorities in Colorado and California used rebates to manage demand during drought, but in both cases in conjunction with drought pricing or use restrictions, and neither case produced strong evidence that rebates reduce consumption (Kenney et al., 2008; Renwick & Green, 2000).

13. All values are from the Brazilian Statistical Office (IBGE). Unfortunately, socioeconomic data by electoral precinct are not available, only municipality level data.