Estimated Demand Function for Water
Attributes
Medium: Water
Country: Philippines
Analytical Framework(s): Contingent Valuation, Market Price of Output
Study Date: 1995
Publication Date: 1998
Major Result(s)
Functional Transfer: The following demand equation (vended water alone Dw1; R2 = 0.418, DW = 1.893, intercept se = 1.57) for (including free public faucets) in double-log form was estimated by two-stage least squares (2SLS):
Dw1 = 7.092 - 2.092 APRICE + 0.254 INCOME + 0.411 HHSIZE - 2.464 HOSEF - 1.063 HOSEC - 2.016 PKUPMS - 0.257 PKUPDW - 8.845 PLFAUCET - 0.074 DISTANCE + 0.094 TRBDY + 0.891 TASTE,while the demand equation (Dw2; R2 = 0.818, DW = 1.937, intercept se = 3.17) for water based on households using both vended water (including free public faucets) and water distributed by Metro Manila's water agency, Metropolitan Waterworks and Sewerage System (MWSS) in double-log form and estimated by 2SLS is shown by the following:
Dw2 = 2.153 - 0.492 APRICE + 0.173 INCOME + 0.351 HHSIZE - 0.851 HOSEF - 0.419 HOSEC - 0.642 PKUPMS - 0.155 PKUPDW - 2.329 PLFAUCET - 0.085 DISTANCE + 0.059 TRBDY + 0.073 TASTE,where APRICE is the average price of water (Dw1 significant at 10%, se = -1.71; Dw2 significant at 1%, se = -7.58), INCOME is the monthly household income in thousand pesos (Dw1 significant at 10%, se = 3.17; Dw2 significant at 1%, se = 3.03), HHSIZE is the household size (Dw1 significant at 1%, se = 3.93; Dw2 significant at 1%, se = 4.33), HOSEF is = 1 if by hose at fixed charge and 0 otherwise (Dw1 significant at 10%, se = -1.90; Dw2 significant at 1%, se = -6.73), PKUPMS = 1 if picked up from public faucet and 0 otherwise (Dw1 significant at 10%, se = -1.85; Dw2 significant at 1%, se = -4.49), PKUPDW = 1 if picked up from households using tubewells and 0 otherwise (Dw1 se = -1.17; Dw2 se = -0.85), PLFAUCET = 1 if picked up from public faucet and 0 otherwise (Dw1 significant at 10%, se = -1.76; Dw2 significant at 1%, se = -9.05), DISTANCE is the distance from source in meters (Dw1 se = -1.63; Dw2 significant at 1%, se = -2.19), TRBDY = 1 if without particles and 0 otherwise (Dw1 se = 0.79; Dw2 se = 0.69), and TASTE = 1 if water has good taste and 0 otherwise (Dw1 se = 1.34; Dw2 se = 0.38). Average price, household income, and household size coefficients were all expectedly very significant. The dummy variables representing the mode of vending water were also statistically significant.
Study Note: The two demand models were estimated using 2SLS to address simultaneity problems that were encountered and to avoid biased and inconsistent estimates with ordinary least squares regression. For the first case, a price equation was estimated in the first stage, followed by specifying the predicted price with other explanatory variables in the demand equations for the second stage. In the second case, two equations where the marginal price and the difference between the actual water bill and what the bill would have been had all the water been bought at the marginal price were first estimated, followed by the inclusion of the predicted marginal price and difference in the demand equation for the second stage.
Study Details
Summary: The urban water problem has been made a central policy issue by widespread water shortages in Metro Manila. The MWSS inefficiency is attested to by Metro Manila's unfavorable record on ratio of non-revenue water, hours of water availability, and number of personnel per 1000 connections. In order to determine optimal pricing policies, demand and supply functions for water were estimated. The study tried to characterize the household sources of water supply, quality of water service, cost of water, and levels of water demand in relation to household income, and to draw some policy implications based on cross-section household survey data. Results showed that the poor are paying much higher water prices (two to five times more) than the better-off households who typically have MWSS connections, simply because the poor do not have access to MWSS water connections. The study recommended that in setting water prices efforts must be made to determine the price level that will equate the demand for water to the supply of water produced considering the marginal cost of extracting/delivering good water quality, cost of water scarcity, and other externalities involved in the production and consumption of water.
Site Characteristics: A 1995 survey on 506 households in Metro Manila covered 95 barangays in 11 major cities and municipalities. The barangays were representative of the different types (private water vending, individual tubewells, private waterworks, MWSS) and qualities (for example water pressure and time availability) of water service, as well as the various household income levels (low, middle and high income groups) within each municipality. The barangays were selected following a brief reconnaissance survey, an examination of the water pressure map of the MWSS, and a review of various profile studies on urban poverty. Based on the information from barangay captains regarding the sources of water supply, nature of private water vending, and characteristics of households in their respective jurisdictions, about five to ten households in each barangay were selected and interviewed.
Comments: The 2SLS estimates provided superior results compared to the OLS ones due to the greater number of statistically significant variables. Also, an unbiased and consistent estimate of price elasticity of water demand (-2.1 for the vended water sample, and -0.5 for the pooled sample) was provided. The estimated price elasticity for the pooled sample fell within the range of estimates for other countries. The results indicated a highly responsive demand function to changes in price, commonly found in low income households dependent on vended water. Such an elastic price response suggested that pricing can be an effective means of more efficiently managing the allocation of limited water supply. On the survey data, it should be noted that the sampling procedure among high income households was not strictly followed in some cases because interviews with such households could only be conducted through personal relationships. As such, fewer households were interviewed in high-income barangays.