Benchmarking the performance of water companies for regulatory purposes to improve its sustainability

Access to clean water and sanitation is a fundamental human right, and the sustainable management of water resources is a key Sustainable Development Goal. Water companies, often operating as natural monopolies, require robust regulation to protect consumers' interests. Benchmarking methods are crucial for water regulators to compare the performance of water companies (WCs) and ensure fair competition and service quality. While various efficiency assessment techniques exist, including stochastic frontier analysis (SFA) and data envelopment analysis (DEA), DEA is frequently used. Traditional DEA models, however, assign variable weights to inputs and outputs for each WC, potentially leading to an inefficient ranking and hindering regulatory efforts. This study addresses this limitation by employing a common set of weights (CSW) DEA model, which utilizes the same weights for all WCs, enhancing the fairness and transparency of the benchmarking process and regulatory decisions.

Existing literature extensively explores water sector efficiency using frontier models such as SFA and DEA, and combinations thereof like StoNED. Debate continues regarding the optimal method for regulatory applications. While some studies combine DEA and SFA, the majority of past research on WC efficiency utilizes DEA. A key limitation of traditional DEA is its assignment of flexible weights to variables for each WC, maximizing individual efficiency scores but potentially obscuring comparative performance and leading to challenges in regulatory implementation. Alternative approaches, including cross-efficiency and common-weight evaluation methods, have been proposed to address this limitation. The cross-efficiency method, while improving discrimination, faces drawbacks related to non-Pareto optimality. The CSW approach, using a single weight set for all units, offers greater potential for regulatory use, but its application in benchmarking WCs remains largely unexplored. This gap in the literature motivates the present study.

This study employs both the traditional Charnes, Cooper and Rhodes DEA model (DEA-CCR) and the DEA-CSW model proposed by Wu et al. (2016) to evaluate the efficiency of 23 Chilean WCs. The DEA-CCR model allows flexible weights for each WC, while the DEA-CSW model uses common weights for all. The study uses four variables: operational expenditure (OPEX) and capital expenditure (CAPEX) as inputs, and quality-adjusted volume of drinking water supplied and quality-adjusted number of customers with wastewater treatment as outputs. The quality-adjusted outputs account for penalties when WCs fail to meet quality standards. Pearson correlation analysis confirmed the isotonic relationship between the input and output variables. The DEA-CSW model uses the concept of 'satisfaction degree' to determine the common weights, maximizing the minimum satisfaction degree across all WCs and ensuring fairness. The study compares the efficiency scores and rankings generated by both models, analyzing weight allocation patterns and the satisfaction degree of each WC to assess the acceptability of the results. Outlier detection was performed using a modified median absolute deviation method.

The DEA-CSW approach exhibited significantly improved discriminatory power compared to the DEA-CCR approach. While five WCs were identified as efficient using DEA-CCR, only one WC was identified as efficient using DEA-CSW, providing a clearer benchmark for comparison. The average efficiency score decreased from 0.747 (DEA-CCR) to 0.584 (DEA-CSW), suggesting that, on average, WCs could improve output by 25.3% (with same CAPEX) and 41.6% (with same OPEX). The DEA-CCR model revealed significant variability in weight allocation for each WC, with many excluding variables that did not perform well. In contrast, the DEA-CSW model provided a consistent weighting across all WCs, with the quality-adjusted volume of drinking water being the most influential variable. The satisfaction degree analysis showed that for most WCs, the difference between DEA-CSW efficiency and their highest possible efficiency (DEA-CCR) was small, suggesting good acceptance of the common weights. However, some WCs showed substantial discrepancies, highlighting the potential impact of common weights on individual rankings. The comparison of rankings derived from both methods strongly supports the use of DEA-CSW for regulatory purposes.

The findings underscore the advantages of DEA-CSW for benchmarking WCs in regulatory contexts. The improved discriminatory power, the consistent weighting scheme, and the increased acceptability of the results contribute to a fairer and more transparent regulatory process. The inherent flexibility of the DEA-CCR model, while allowing each WC to present itself in the best light, makes it less suitable for regulatory purposes where objective comparison is vital. The DEA-CSW model addresses these shortcomings by imposing common weights and promoting acceptance among the WCs. The resulting consistent and objective ranking facilitates more effective regulatory intervention and resource allocation.

This study demonstrates the superiority of the DEA-CSW approach over traditional DEA models for benchmarking WC performance for regulatory purposes. The enhanced discriminatory power, objective weighting scheme, and increased acceptability of the results make it a valuable tool for water regulators. Future research could explore the dynamic efficiency of WCs over time using DEA-CSW, incorporate additional environmental and service quality variables, and investigate the adaptability of weight allocation according to specific regulatory goals and priorities.

The study focused on a static efficiency evaluation of a sample of 23 Chilean WCs, potentially limiting generalizability. The number of input and output variables was restricted, although sufficient to capture significant aspects of WC performance. Further research could benefit from a larger sample size and inclusion of additional environmental and service quality variables.