Municipal water consumption forecast accuracy

Abstract

[1] Municipal water consumption planning is an active area of research because of infrastructure construction and maintenance costs, supply constraints, and water quality assurance. In spite of that, relatively few water forecast accuracy assessments have been completed to date, although some internal documentation may exist as part of the proprietary "grey literature." This study utilizes a data set of previously published municipal consumption forecasts to partially fill that gap in the empirical water economics literature. Previously published municipal water econometric forecasts for three public utilities are examined for predictive accuracy against two random walk benchmarks commonly used in regional analyses. Descriptive metrics used to quantify forecast accuracy include root-mean-square error and Theil inequality statistics. Formal statistical assessments are completed using four-pronged error differential regression F tests. Similar to studies for other metropolitan econometric forecasts in areas with similar demographic and labor market characteristics, model predictive performances for the municipal water aggregates in this effort are mixed for each of the municipalities included in the sample. Given the competitiveness of the benchmarks, analysts should employ care when utilizing econometric forecasts of municipal water consumption for planning purposes, comparing them to recent historical observations and trends to insure reliability. Comparative results using data from other markets, including regions facing differing labor and demographic conditions, would also be helpful.

Long-Term Water Demand Forecasting

“…The development of this model type is reflected by an abundant scientific literature (for a review, see Espey et al 1997;Arbués et al 2003;Dalhuisen et al 2003). The main weakness of statistical models for long-range forecasting is their out-of-sample predictive capacity (Fullerton & Molina 2010).…”

“…Many water consumption prediction models have been developed and used in a deterministic context despite the presence of uncertainty in assumed model structures and parameters. It is now widely recognised that water use forecasts, regardless of the timeframe or the forecast method employed, are likely to be always highly inaccurate (Osborn et al 1986;Fullerton & Molina 2010). It is thus crucially important to give consideration to model uncertainties.…”

Evaluating the performance of feature selection techniques and machine learning algorithms on future residential water demand

“…A powerful method as it examines the effect of a set of independent variables on per capita consumption (dependent variable) and shows a statistical relationship between them (Fullerton & Molina 2010).…”

Municipal Water Consumption and Urban Economic Growth in El Paso

“…Given its prominence as a growing metropolitan economy located in a semi-arid desert setting, El Paso water consumption has been dissected in a number of empirical studies. Among the topics examined in these efforts are short-range consumption reactions to weather and economic stimuli [24], surface usage rights transfers [25], long-term historical forecast accuracy [26], and short-term consumption prediction accuracy during business cycle downturns [27]. Of course, El Paso still faces important policy questions regarding the provision of municipal water services.…”