Objective:

1

• Objective
• Develop a water demand forecasting model that
  predicts water demand based on known drivers and
  appropriate water demand factors, with the
  understanding that this is a regional forecast
  and not a utility forecast

2
Terminology
3
Criteria for Selecting Best Water Demand
Forecasting Approach

• Goals Objectives
• What information is needed by planners and
  decision-makers?
• What type of models are needed to provide this
  information?
• Data Availability
• What is available?
• What is the quality?
• What models will the data support?
• Budget
• What are financial constraints?

Goals
Data
Budget
4
Water Demand Forecast Approaches
PerCapita
Cost Complexity
Low
High
TrendExtrapolation
UnitUse
Econometric
5
Modified Unit Use

• Approach
• Starts with unit use approach
• Then applies elasticities from other studies for
  appropriate water use factors

Recommended Approach
6
Modified Unit Use
Formula
Unit UseWater Demand c
X
WhereWF Use factor adjustment (e.g., price,
income, weather) f Future year c
Current year ß elasticity for water use
factor
7
Modified Unit Use

• Example
• How does single-family unit use demand change
  over time as a result in changes in the real
  price of water?Assume
• Price elasticity is -0.20
• Current marginal price is 2.13/ccf
• 2010 marginal price is 2.51/ccf (in real terms)
• 2020 marginal price is 2.74/ccf (in real terms)

8
Example Elasticities
The following are elasticities estimated for
water use factors from almost 200 statistical
water demand equations in the United States
Water Use Factor Winter Season
Summer Season
Marginal Price -0.050 to -0.250 -0.150 to
-0.350 Income 0.200 to 0.500 0.300 to
0.600 Household Size 0.400 to 0.600 0.200
to 0.500 Housing Density -0.200 to
-0.500 -0.400 to -0.800 Precipitation -0.010
to -0.150 -0.050 to -0.200 Temperature 0.30
0 to 0.600 0.800 to 1.500
(Paredes, 1996).
9
Selecting Use Factor Adjustments

• Data availability
• Information on elasticities that are applicable
  to the Central Puget Sound region
• Benefit in explaining regional water demand

10
Use Factor Adjustments Already Identified for
this Approach

• Weather
• Maximum daily temperature
• Rainfall
• Price (real marginal price)
• Income (real personal or household income)
• Other use factor adjustments will be explored to
  see if they add value and can be incorporated in
  a cost-effective manner

11
Benefits of this Approach

• For those use factor adjustments in which data
  are limited or a relationship has not been
  established, we can do sensitivity to describe
  the potential impact
• If subsequent demand studies in the region result
  in established relationships between other use
  factor adjustments and demands, they can be
  easily incorporated into this approach in future
  Outlooks

12

• Clarification Slides

13
Conservation

• Passive Will be estimated based on current
  plumbing codes and age of housing stock
• Existing Will be estimated based on existing
  programs of the utilities, continued into the
  future
• Potential Will push the boundaries of
  conservation

14
Housing Density

• Current demand models in the region have not
  considered housing density (so there is little
  established relationship)
• Future values of housing density are not
  available
• Housing mix (SF vs MF) will reflect some of the
  trend in density
• Scenarios of future density along with
  elasticities from other studies can be run to
  test the sensitivity in demand due to this
  variable

15
Demand Sectors

• Single-Family Residential
• Gallons per home per day
• Multifamily Residential
• Gallons per home per day
• Non-Residential
• Gallons per employee per day

16
Non-Residential Model

• Method is to take non-residential water demand
  and divide by employees to get unit use factor
  (GED)
• Biggest factor influencing GED is industry mix
  (employment in major categories)
• PSRC projects employment by major 2-digit SIC
  codes
• Non-Residential model could be tested based on
  GEDs from other studies and applied to
  employment projections at 2-digit SIC codes

17
Example GEDs from National Database
18
Model Parameters

• Model parameters will be selected based on water
  use factors collected from largest utilities
• If smaller utilities do not have such data,
  proxies from the larger utilities in closest
  geography will be applied
• Demographic projections are available for PSRC
  FAZ, and generally can be correlated to utility
  boundaries