Governance endowment and water-loss performance in Italian water services: fsQCA dataset and replication code
Description
This deposit contains the calibrated dataset and the full replication code for a fuzzy-set qualitative comparative analysis (fsQCA) of governance configurations and water-loss performance in the Italian integrated water service. The unit of analysis is the gestione-per-ambito (operator-within-Optimal- Territorial-Area), the level at which the Italian regulator (ARERA) sets objectives and assigns premia and penalties; an operator active in several ambiti contributes one observation per ambito. The workbook holds 78 matched managements, of which 77 form the analytical frame — Siciliacque S.p.A., a wholesale supplier without its own distribution network, is excluded by the replication script. The outcome is a four-valued fuzzy score derived from ARERA's assessment of macro-indicator M1 (water losses) for the 2022–2023 biennium under deliberazione 225/2025/R/idr, within the technical-quality regime established by deliberazione 917/2017/R/idr. Four conditions are calibrated: institutional scale, operational capacity (served population), sustainability-governance maturity (a five-item index) and network embeddedness (a two-item index). Calibration anchors and coding rules are documented in README.md. The data were assembled by matching ARERA's assessment to firm-level administrative sources: MEF Dipartimento del Tesoro open data on public shareholdings; ISTAT resident population; ACCREDIA and EMAS registers; company sustainability reporting; the Science Based Targets initiative register; Utilitalia membership; and participation in national recovery-plan (PNRR) water programmes. All underlying sources are public and concern organisations, not individuals. Replication_fsQCA.py reproduces every reported result: descriptive statistics; necessity analysis with consistency, coverage and Relevance of Necessity for each condition and its negation, against both the outcome and its negation; the complete truth table including logical remainders; parsimonious and conservative sufficiency solutions for the outcome and its negation, with solution-level consistency, PRI and coverage and term-level raw and unique coverage; and a robustness suite covering the declared-capacity subsample, quartile recalibration of the capacity condition, a three-condition model omitting capacity, a variant retaining only source-verified indicators, consistency thresholds of 0.75, 0.80 and 0.85, and a frequency threshold of two. The two researcher-imposed coding rules are stated explicitly in README.md so that they can be relaxed in replication. Requires Python 3 with numpy, pandas and openpyxl.
Files
Institutions
- Roma Tre UniversityLazio, Rome