Syndicate content

Add new comment

Pricing water when the poor share: evidence from Manila: Guest post by William Violette

This is the ninth in this years' job market series
Despite large investments in piped water throughout the developing world, the share of urban households without piped water has remained stable at 5% for middle-income countries and at 20% for low-income countries over the past decade.  Given health, time-savings, and other benefits from piped water, how can water utilities set prices in order to close gaps in access while still covering costs?  Conventional wisdom is that subsidizing fixed connection fees with high marginal prices can improve access especially for the poor, but this policy can have the opposite effect when households share water connections, which is common in the developing world.  I observe that over 23% of households in Manila access piped water through a neighbor's connection.  In this context, high marginal prices weaken incentives for households to extend water access to their neighbors through sharing.  Similarly, connection fee subsidies may have limited impacts on access because sharing households already split any fixed costs with their neighbors.

Pricing with sharing
 
My Job Market Paper finds that high fixed fees and low marginal prices can ensure nearly universal access to piped water and produce large welfare gains by increasing water sharing.  The intuition is that shared connections allow low-demand households to enjoy low marginal prices (instead of substituting to costly vendors) and allow cities to save on installation and maintenance costs.  To measure welfare impacts of counterfactual pricing policies, I collected administrative data on 1.5 million water connections in Manila, which I was able to uniquely merge with detailed survey data on the number of households and people using each water connection.  Using this data, I structurally estimate household water demand across three sources: purchasing directly from the provider, sharing with a neighbor's tap, or buying from a small-scale vendor.
 
Why do households share water?
 
Water sharing allows households to tradeoff higher marginal costs from sharing against dividing fixed costs of owning a connection among many households.  Using a mobile app survey, I find that sharing households often split the upfront connection fee with each other.   At the same time, households face higher marginal costs from sharing in two ways:
 
(1) Since sharing households connect plumbing directly to the meter (61%), extend plastic pipes (25%), or fetch water with containers (14%), these households may face time costs or low pressure from sharing, which are captured by a per-unit sharing cost in the model;
 
(2) Sharing households also report evenly splitting the monthly water bill, creating a ‘’free-rider’’ incentive to over-consume, which I model directly. 
 
The magnitudes of these sharing frictions determine how much policymakers may want to set prices to disincentivize shared connections.
 
Estimating water demand and sharing costs
 
Evaluating the welfare impacts of pricing policies requires measuring water demand, costs of sharing water, and costs of supplying water.  Given supply costs from administrative data, I set out to estimate demand and sharing costs from consumption and water source choices.  The estimation takes place in three steps:
 
1. Water Demand is identified from variation in the non-linear price schedule (both across and within households) using monthly consumption data.  I find wide heterogeneity in demand with large users being much less sensitive to price, which provides scope for pricing policy to cross-subsidize small users.
 
2. Marginal Sharing Costs are recovered through a quasi-experiment where households face leaks in their water pipes between the meter and the main line, forcing them to substitute to a neighbor's connection.  The model maps the extent that these households offset their consumption to a neighbor's connection into an estimate of the marginal costs of fetching water.  I recover sizable sharing costs on the order of 61% of the average marginal water price in Manila. 
 
3. Fixed Costs of owning a connection (encompassing any application costs, monthly service fees, as well as land-tenure issues in qualifying for a water connection) are estimated from cross-sectional variation in water source choices.  I recover large fixed costs, which help explain the large proportion of households sharing connections in Manila.  With census data on households unconnected to the network, I also estimate high marginal prices and low fixed costs of local water vendors such as deepwells or tanker trucks.
 
Optimal pricing
 
The figure below compares the marginal prices for the current price schedule in Manila (which vary according to monthly usage per connection) against the marginal prices for the optimal pricing policy.  The optimal pricing policy is calculated by choosing a fixed fee and two different marginal prices (above and below 20 m3) in order to (1) maximize consumer surplus and (2) ensure that the provider just covers its costs: marginal costs per unit, maintenance costs per connection, and fixed capital costs. 
 
The current tariff in Manila uses steeply increasing marginal prices with the intention of cross-subsidizing access for low-demand, poorer households through higher bills from large users.   By contrast, the optimal price schedule has much lower marginal prices, which are offset by a much higher fixed fee of 13.4 USD relative to a current fixed fee of 8.4 USD.  This tariff leads to a new equilibrium where the percentage of households fetching from neighbors rises from 23% to 45% while the percentage using from vendors drops from 6% to 2%.  The optimal price schedule produces large welfare improvements on the order of 70% of consumer surplus or 0.6% of household income.

Figure: Current Tariff in Manila and Optimal Tariff




In additional counterfactuals, I find that subsidizing the fixed fee with high marginal prices reduces welfare especially for low-demand households; instead, low-demand households benefit disproportionately from very high fixed fees and very low marginal prices.  Compared to a counterfactual world without sharing, I also find that the presence of sharing means that complex pricing structures are ineffective at redistributing surplus from high-demand households to low-demand households.

Caveats

In focusing on water sharing behavior, this approach abstracts away from several features that may also affect optimal pricing policy.  First, I do not directly model environmental, health, or other externalities associated with piped water use.  From a policy perspective, water shortages may lead policymakers to promote conservation through higher marginal prices while positive health externalities may motivate further subsidizing access.  Second, I assume that the government regulator is able to perfectly observe production costs for the water provider, eliminating scope for prices to correct for principal-agent frictions.

Policy takeaways
 
Pricing policies can have large impacts both on how much water households use as well as where households access water: either from their own connection, a neighbor's connection, or a water vendor.  Increasing price schedules and connection fee subsidies may not increase access to piped water when households can easily share connections with their neighbors.  By sharing with their neighbors, low-demand households can benefit from low marginal prices and mitigate high fixed fees.   In this way, households act as efficient subcontractors for the water provider, extending water access to poorer, low-demand households. 
 
Instead of focusing solely on pricing, policymakers may also consider a broader range of policies to improve water use and access such as assisting with the implementation of informal sharing networks.  For example, providing engineering knowledge and appropriate materials to sharing households may help mitigate the frictions associated with transporting water and further improve welfare for poorer households.
 
William Violette is a PhD student at Brown University.