Specific Objectives of the Project

Develop a theoretically consistent approach for estimating the welfare effects of switching to block-rate water pricing; apply the model to a data set from a southern California water district; investigate distributional effects particularly for disadvantaged groups.

Summary of Results

During the funding cycle, the PIs worked closely with a graduate student to develop the theoretical framework and convert it to computer code for empirical estimation. We have made substantial progress and expect to have preliminary results ready for disse mination at two conferences this summer. The theoretical framework is complete and has been written-up as a draft manuscript (see below). The framework essentially modifies the DCC statistical structure so that it is applicable to a direct utility function rather than a demand function. The resulting likelihood equation is similar to but more complicated than the standard DCC likelihood equation and requires simulation techniques for evaluation. We are currently writing the computer code to implement these techniques and will then apply the model to an existing dataset to estimate the welfare effects of changes in block rate water prices.

The objectives of this research were to evaluate the impact of a water conservation program being promoted by many of the water districts in southern California—the high efficiency sprinkler nozzle program—on water use. In particular, we intended to investigate the impact of these sprinkler heads on water use at the household level and the subsequent potential impact at the district level. Furthermore, we attempted to estimate an adoption model that identified factors which influence a household's decision to adopt the technology.

In this program, households can receive vouchers for up to 25 free high efficiency sprinkler nozzle heads. The data consists of monthly water use and household characteristics over ten years on approximately 120,000 residential meters. Our analysis consists of two approaches to evaluating these potential effects. First, we compared different subpopulations average use before and after adoption; second, we estimate a discrete continuous choice (DCC) model. Preliminary results suggest that residential customers who redeemed vouchers for 25 high efficiency sprinkler nozzles typically experienced a subsequent reduction in overall water use of around 1.2%. As a fraction of total outdoor water use, the reduction is around 2.7%. This is markedly lower than the technically achievable reduction of 30% that has been estimated by the manufacturer.

In terms of what factors seem to influence adoption, we find that adoption rates were positively related to house value, income levels, average water prices, ET, household size, and landscape area (although nonlinearly); adoption rates were negatively related to distance to the nearest nozzle head distributor (which suggests travel and time costs are important factors influencing whether households redeem their vouchers). Adoption rates for those in the top, middle, and bottom terciles of water use were approximately 2.3%, 2.0%, and 0.78%. Ongoing research is currently focusing on gathering more complete information on those households that redeemed the vouchers in terms of the number of nozzles they actually installed and the degree to which they were installed correctly.

The broad objective of this work is to improve upon zonal approaches to recreation demand modeling. A standard zonal model ignores important aspects of spatial heterogeneity that are inherent in recreation demand contexts. The standard approach aggregates across groups of heterogeneous agents, and models them as homogenous points of origin for demand estimation. The standard approach also ignores that these heterogeneous agents deliberately choose their points of origin, which introduces a source of bias into the estimation. This work has produced a peer-reviewed article that addresses these shortcomings. This article, “A Latent Class Approach to Modeling Endogenous Spatial Sorting in Zonal Recreation Demand Models” (Land Economics86(4):800-816), demonstrates how a latent class count data model can control for unobserved heterogeneity that may lead to spatial sorting of recreationists. Results show that welfare estimates from this model for a southern California wilderness site are substantially smaller than for the standard approach.