Has the drought in Catalonia changed what households spend on water?
Using internal data and big data techniques, we analyse the water consumption of Catalan households between 2021 and 2024, following the declaration of drought.
With autumn coming to an end, Catalonia remains on alert due to the persistent drought affecting the provinces of Barcelona, Girona and part of Tarragona. The state of the internal river basins in this area of Catalonia, which are mostly fed by the waters of the Ter and Llobregat rivers, contrasts with those found in the territory fed by the Ebro river. The water shortages began in 2021 and the situation reached a critical point in 2023, when most municipalities in the internal basins were declared to be in a state of exceptional drought. This state is activated when water reserves fall below 25%, indicating a severe water shortage (see first chart). In this article, we analyse the extent to which Catalan households took measures to address the water shortages following the declaration of drought in their municipalities between 2021 and 2024. Using detailed data from water bills paid by direct debit, we have observed that Catalan households did reduce their water expenditure in response to the drought. The reduction in water expenditure also varied depending on the severity of the water emergency and it persisted over time.
For this analysis, we use duly anonymised internal data from water bills paid by direct debit from CaixaBank accounts. These charges provide information on the amount paid in each bill, the supplier company, as well as characteristics of the account holders such as age, income and place of residence. This information is combined with data related to the water supply (water network and basin, state of the drought, price of water) and climate data (temperature, precipitation, humidity level) for the municipality where they live in order to study the determining factors of Catalan households’ water consumption.1 In this way, we obtain an innovative source of information for studying how households modified their water expenditure in response to the drought.
The median water bill is around 48 euros in the internal basins and 40 euros in the Ebro basins (i.e. half of all households pay lower bills and the other half, higher). However, water expenditure varies widely (see second chart) and there are other factors that affect water expenditure besides the state of the drought, such as household structure and water prices.2 To analyse whether households changed their expenditure depending on the state of the drought at any given time, we must take these other factors into account. To this end, we estimated an econometric model that incorporates these factors,3 as well as the state of the drought at all times. Our results reveal that a more severe state of emergency in the internal basins is associated with a reduction in water expenditure of between 1.5% and 4%, after adjusting for variations in prices (see third chart).4 These results coincide with other studies,5 which estimated the impact of another episode of drought (2004-2012) in the Barcelona metropolitan area and show that households reduced their consumption in line with the water shortages.
- 1. Specifically, the data sample includes all private customers in Catalonia who paid their bimonthly or quarterly water bills by direct debit between March 2021 and September 2024. The information on the prices of each water supply company comes from the Agència Catalana de l’Aigua (ACA). The information on the state of the drought at the operating unit level comes from the ACA for the internal basins and from the Confederación Hidrográfica del Ebro for the basins of the River Ebro. Climate information at the municipal level is taken from the Copernicus Climate Change Service (C3S) Climate Data Store (CDS).
- 2. A. Reynaud (2015). «Modelling Household Water Demand in Europe: Insights from a Cross-country Econometric Analyses of EU-28». Luxembourg: Publication Office of the European Union. 249 p.
- 3. Specifically, we estimated a linear regression with high-dimensional fixed effects (HDFEs) that includes household characteristics (income, age of the account holder, number of bills and amount of the previous bill), meteorological information (humidity, temperature and precipitation in the municipality), prices and the severity of the drought. In addition, the specification also includes fixed effects related to the household, municipality and month of the year.
- 4. In the basins of the Ebro, this association is weaker, since the state of the drought during the period in question did not reach such a severe level as it did in the internal basins during the same period.
- 5. V. Bernardo, X. Fageda and M. Termes (2015). «Do droughts have long-term effects on water consumption? Evidence from the urban area of Barcelona». Applied Economics, 47(48), 5131-5146.
However, this methodology can offer a distorted view of the impact of the drought by grouping together municipalities that have just entered a given state of drought and others that have been in a similar situation for much longer. In order improve the estimation of the impact of the declaration of drought on water consumption, we used a Difference-in-Differences (DiD) estimation,6 which allows us to assess the impact of an intervention by comparing the changes between a group subject to a water shortage in a given period and another control group. In this case, a water shortage is defined as a state of alert or worse being declared in a particular municipality, while the control group is the municipalities where no such situation has occurred.
The results show that, following the declaration of drought in a given municipality, its inhabitants did indeed reduce their water expenditure over the following months. The impact during the 12 months following the declaration of drought is around 5% in terms of expenditure. As we can see in the chart, this impact is gradual and peaks six months after the state of drought is first declared, with a reduction of almost 10%. After six months, the effect of the intervention is reduced. This result is also maintained if we take into account other factors such as the average income in the municipality, water prices and the month of the year in which the consumption took place.
- 6. B. Callaway and P.H. Sant’Anna (2021). «Difference-in-differences with multiple time periods». Journal of Econometrics, 225(2), 200-230.
In conclusion, the use of innovative data that combine information on water expenditure with information related to the water supply has allowed us to confirm that Catalan households have significantly reduced their water consumption during the current episode of drought.
Climate change & green transition
What polices can be implemented to stop climate change? What are the implications of shifting towards a more sustainable economy?
