Non-market valuation of water in residential uses

MISSING IMAGE

Material Information

Title:
Non-market valuation of water in residential uses
Series Title:
Florida Water Resources Research Center Publication Number 57
Physical Description:
Book
Creator:
Lewis, K. C.
Carriker, R. R.
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Notes

Abstract:
Increases in water demands were historically met through the augmentation of water supply facilities. However, the most easily developed water sources have already been tapped, and water pollution adds to the cost of developing some water sources. Given the increased cost and difficulty of water supply augmentation, more attention has been given recently to demand management. This requires the assigning of priorities to water uses, and the subsequent fulfillment of only the most highly valued needs. In 1978 the U.S. National Water Commission directed that the most highly valued needs be determined through the concept of consumers I willingness to pay for publicly supplied water. This research is designed to test the use of a non-market valuation technique to assess the residential consumers I willingness to pay for household water. It was hypothesized that the willingness to pay would be dependent upon a consumer's income, family size, the amount of water used, the presence of a well, and also upon variables representing a consumer's beliefs and attitudes concerning local water scarcity and conservation. Results indicated that indeed most of these variables did play a role in the formulation of a willingness to pay for residential water. Some inconsistencies in results, however, indicate that more research is required before non-market valuation techniques can be applied with confidence to residential water demand analysis.

Record Information

Source Institution:
University of Florida Institutional Repository
Holding Location:
University of Florida
Rights Management:
All rights reserved by the source institution and holding location.
System ID:
AA00001535:00001


This item is only available as the following downloads:


Full Text












.,-.-i-rket Valuation of Water


in Residential Uses



By


K. C. Lewis and R


R. Carriker


PUBLICATION NO. 57

FLORIDA WATER RESOURCES RESEARCH CENTER

RESEARCH PROJECT TECHNICAL COMPLETION REPORT

OWRT Project ,i,,i,-r B-036-FLA



Matching Grant Agreement Number

14-34-0001-8074


Report Submitted: April 1981







The work upon which this report is based was supported in part
by funds provided by the United States Department of the
Interior, Office of Water Research and Technology
as Authorized under the Water Resources
Research Act of 1964 as amended.









ACKNOWLEDGEMENTS

The authors wish to express their gratitude to the Office of Water
Research and Technology, United States Department of Interior, for
financial support of this work. The administrative assistance of Dr.
James Heaney, Director of the Florida Water Resources Research Center
is greatly appreciated. Thanks are also due Ms. Iry Robinson for
accounting assistance.

The authors also extend special appreciation to Dr. Richard Kilmer
for his constructive assistance in the execution of this research.
Gratitude is also extended to Dr. Beau BeauTieu, to Mr! Ray Boyd of the
Orlando Utilities Commission, and to Mr. Dave Nichols of the St.
Petersb, i Public Utilities for their advice and cooperation with the
*qwAing procedures, and to Mr. Richard Marella of the St. Johns River
Water Mana' i.- District for his technical assistance. Special thanks
also goes to Ms. Debra Linn who patiently prepared the final manuscript.











TABLE OF CONTENTS


ACKNOWLE E iC TS. .

LIST OF T .

LIST OF FI i..u . .

ABSTRACT. . .

I I l !i rf. l l '-'. IF 1-

The Problem Setting . .

The Research Problem ...

Objectives of Research . .

Method of Procedures . .

CHAPTER II .'iiUMER'S SURPLUS AND V,

Benefit Estimation . .

Neoclassical Assumptions About C

The Consumer's Surplus . .


Page

. . i

. .. i V


VALUATION


consumer B


. . . .. v

. . .. vi
. . . . Vi




. . . . 3


. . . 3





OF BENEFITS . 5

. . . 5

behavior . . 5

. . . 6


Consumer's Surplus: Quantity Change Analysis. .

CHAPTER III NON MAPI ET VALUATION: THE ITERATIVE BID..

Non Market Valuation Method. . . .

The Bid Curve and Consumer's Surplus . .

Application of the Iterative Bid . . .

Potential Bias in Iterative Bids . ..

CHAPTER IV VALU '". RESID.iTIAL WATER USE BY CONTI r--, i-
10 ,; I TECHNIQUES . .

WTP, WTA, and Consumer's Surplus for Residential
Wate Use . . .

Exposition Using Traditional Indifference Curves .

Valuation of Residential Water Use . L


7
. . 71

. . 11


. . 12

16

17


. . 2 1


.. 21

. . 23

S .. 25


Estimating Eln,.-tions


(










Page

The Estimation Procedure . . . . 32


The Ouestionnaire . ...

CHAPTER V RESULTS OF ANALYSIS .

Response to the Questionnaire. .

.i,1,,,:, -:,,i 1TP anid WTA Within Use C

Comparing WTP and WTA Between the

Results of e, .xion Analysis .

CHAPTER VI SUMMARY AND DISCUSSION .

APPENDIX I THE QUESTIONNAIRE. .

APP-.i[uIX II THE SAMPLE. . .

Community Characteristics .

Sampling Procedure . .

APPENDIX III SUPPLEMENTARY SUMMARY OF
RESULTS. . .. ..


categories.

Two 2.-*r- les.














QUESTIONNAIRE


LITERATURE CITED.


. . 32


. . . 74











LIST OF TABLES


1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.


Page

. . 35

. . 38

. . 41

.-. ^ ^ -- 42

. . 44

. .. 45

. 65

each income group 70

income level 72

. . 75

. . . 80


Criteria for rejecting surveys .

:. bids .. .

WTP regression results for St. Petersburg

WTIP regression result ts for Orlando- -. -

WTA regression results for St. Petersburg

WTA regression results for Orlando. .

Water rate schedules for sample cities .

liiler of census tracts and households in

Number of households per census tract, by

Socioeconomic data . .. .

Conservation attitude data .










LIST OF FIGURES


Pae_

1. The welfare impact of a change in the quantity of a good
X from Q" to Q' . . . . . 8

2. The total value curve for increments and decrements in
the level of provision of a service, Q, for an individual
who Tni i -' 11 i.,'s t i. I i Q- Aid the income Y". 13

3. The relationships between WTP and WTA, and Hicksian com-
pensating and equivalent measures of consumer's surplus. 15

4. Relationship of WTP, WTA and Hicksian measures of surplus:
the context of residential water use . . 22

5. Diagramatic exposition, using traditional indifference
curves, of the relationship of WTP, WTA, and Hicksian
measures of consumer surplus; the context of residential
water use . 24









ABSTRACT


Increases in water demands were historically met through the augmenta-
tion of water supply facilities. However, the most easily developed water
sources have already been tapped, and water pollution adds to the cost of
developing some water sources.

Given the increased cost and difficulty of water supply augmentation,
more attention has been given recently to demand management. This requires
the assigning of priorities to water uses, and the subsequent fulfillment
of only the most highly valued needs. In 1978 the U.S. National Water
Commission directed cWat the most: highly valued needs be determined thro..,.!,
the concept of consumers' willingness to pay for publicly supplied water.

This research is designed to test the use of a non-market valuation
technique to assess the residential consumers' willingness to pay for
household water, It was hypothesized that the willingness to pay would.
be dependent upon a consumer's income, family size, the amount of water
used, the presence of a well, and also upon variables representing a con-
sumer's beliefs and attitudes concerning local water scarcity and con-
servation. Results indicated that indeed most of these variables did
play a role in the formulation of a willingness to pay for residential
water. Some ilnconsis encies in results, however, indicate that more
research is required before non-market valuation techniques can be ap-
plied with confidence to residential water demand analysis.









CHAPTER I


INTRODUCTION

The Problem Settinr

Studies on demand reductions have been prompted nationwide by the
realization that water supply systems will be increasingly pressured both
by natural drawdowns and also by the environmental objections to a pro-
liferation of water projects which are not considered to be ecologically
sound. Measures taken by a water utility to conserve water through demand
reduet-ion._ -,.. 1 .-, ii-,_1 --,-.n The-most extensive research on consumer
response to water shortages has taken place in California following the
drought seasons of 1976-77.

The California drought of 1976-77 gave impetus to water research in
demand management. Precipitation in both years averaged less than one-
half of normal levels and the compounded effect had vast consequences for
the state's water supply systems. A :". .."iating the problem from the
supply side involved supply enhancement methods, which are generally
unsuccessful at such short notice, and interdistrict transfers which were
exceeding costly and politically unpopular. Demand reduction policies
proved to be far more effective. Accomplished principally through ration-
ing, the overall average demand reduction through the state for the summer
of 1977 was 49% (-loffman, et al., 1979).

Earlier, in 1966 a drought in the Potomac area of Washington, D.C.,
sparked studies by the Washington Suburban Sanitary Commission (WSSC)
which culminated in a conservation program. Between the years 1980 and
2000 the demand for water in the D.C. area given population growth and
per capital use rates, is expected to far outstrip the potential supply
and sewer capabilities. The pursuit of regional solutions by supply
augmentation was discouraged until a conservation program was put into
effect. The resulting program, consisting of a public education program,
a revised building and plumbing code, and a conservation rate structure,
was able to reduce residential consumption by 13.8% through the 1970's
(McGarry and Brusni ,n, 1979).

In Florida, water shortages result from intense urbanization in the
relatively water scarce parts of the state. Of Florida's nine million
residents, nearly 75% reside in the coastal zone. Compounding this
effect are the water demands caused by the state's annual 25 million
tourists who are attracted to the shoreline areas where water is far
less plentiful than inland (Bureau of Coastal Zone Management, 1979).
The subsequent groundwater drafts have caused excessive watertable draw-
downs. In some areas, such as Jacksonville and Tampa-St. Petersburg,
the resulting salt water intrusion has threatened both current and po-
tential well sources (Parker, 1975). 4.

Natural shortages are not the only source of difficulty facing ef-
forts to augment water supplies. Environmental pressure groups have lob-
bied consistently for the reconsideration offunding for water supply in-
frastructure. They contend that the costs of the environmental damage to
natural settings outweigh the benefits of fulfilling the growing water
demand of urbanization (Sierra Club, 1980). The political weight carried
by environmentalists was made manifest in 1977 when President Carter deleted









federal funds for 18 major federal water projects (Schlerger and Cerviso,
1980). This provided a catalyst for the implementation of a new national
water policy establishing stringent environmental criteria for future
projects. Announced in June, 1978, the new national water policy speci-
fically directed that water conservation be added to the Principles and
Standards of the Water Resources Council which evaluates all federally
constructed projects (Schad, 1978). Thus it was required that all feder-
al agencies with water programs advocate conservation and integrate it
into all program planning.

In response to nationwide water shortages and environmental objections
to ,.t-,-., -,,- I. .I. -I"A.-P...- I irjects, water conservation and the water
demand management that it necessitates, has become a national objective.
Residential demand for water has been the subject of several economic
analyses.1 Water demand studies have reported coefficients relating
quantity of water demanded by the residential sector to price, and to a
number of nonprice variables including household size, household income,
property size, property value, and climate (Andrews, 1974; Clouser and
Miller, 1979; Danielson, 1977; Gottlieb, 1963; Hanke, 1970; Headley,
1963; Hogarty and Mackay, 1975; Lineaweaver, et al., 1966; Lynne and Gibbs,
1976; Morgan, 1973; North, 1967; and Wong, 1972). The development of
water management policies has benefited from the knowledge gained by these
studies.

Water demand studies have typically relied upon secondary sources of
data, using statistical inference to test hypotheses about relationships
among variables associated with residential water demand. Available data
from secondary sources generally does not permit the refinement of analysis
to include effects on non-marginal price changes, differentiation among com-
ponents of household water demand, and the influence of attitudes, know-
ledge and beliefs on residential water demand and valuation.

As a consequence, residential water d',.i.l models statistically are
typically not designed to yield information on the social valuation of
residential water use. They are _therefore limited in usefulness for sever-
al types of policy evaluations:

(1) They are not well suited to evaluate policies which contemplate
dramatic changes in water rates, never before experienced by
users and therefore outside the range of existing data; or regu-
latory reductions in water allocated to the residential sector.


The word "demand" is used by different people to mean different things.
In economics, demand is a technical term referring to the amount of a
commodity that would be purchased at a given price (Lauria, 1975). In the
case of water for residential uses, price of water is a key determinant
of quantity demanded, given the influence of other variables such as per
capital household income, the preferences of people with respect to green
lawns, daily showers, swimming pools, and other variables. The point is,
the economic concept of demand views demand as a variable, associated in
predictable ways with the combined influences of other identifiable variables.
By contrast, the terms "requirements" or "needs," often used interchange-
ably with the word demand, are, in fact, mere expressions of unexplained
and unqualified desires for a commodity, conveying no information about
their determinants.









(2) They do not disaggregate household water demand beyond the
distinction between domestic (in-house) and sprinkling (out-
door) demand, and are to that extent not well suited to evalu-
ation of policies directed at influencing specific household
uses of water.

(3) They do not generally permit assessment of the manner in which
"taste-like" variables such as beliefs, knowledge, and attitudes
relate to demand for, and valuation of,,residential water use,
and are not well suited to evaluation of public education pro-
grams designed to influence rates of water use.

The Research Problem

Whether water policy emphasizes water supply development or demand
management, there exists a need for detailed information about the wel-
fare effect on people when limitations are placed on household uses of
water. The concept of a consumer's "willingness-to-pay" for residential
water was adopted by the '-tional Water Commission in 1973 to direct al-
location of water resources (Schad, 1978). The determination of this
benefit measure has not been previously researched in residential water
demand analysis.

Objectives of Research

The overall objective of this research is to adapt and apply a
methodology for eliciting consumers' valuations of nonmarket goods to
the measurement of consumers' valuations of water in residential uses.
Specific objectives include:

(1) measurement of consumers' valuations of the loss in utility
associated with specified reductions in the amount of water
permitted for specified residential uses-


valuations of water in residential uses; and

(3) quantification of the relationship between consumers'
beliefs and attitudes about water conservation, and their
valuation of water in specified residential uses.

Method of Procedure

Valuation, for ,,.:,es of benefit/cost analysis, is an attempt to
ascertain the quantity of *'., which gainers and losers from some
proposed action will consider equivalent in value to their respective
gains and losses (Randall and Brookshire, 1978). .'.... :-rket valuation
mechanisms, some of which are called contingent market valuation mechanisms,
elicit valuations of non-market goods by establishing hypothetical mar-
kets and recording the contingent decisions of individuals confronted with
special changes in these hypothetical markets. Non-market valuation
techniques differ from more conventional demand analysis in their reliance
upon primary data rather than secondary data.









A method of contingent market valuation called iterative bidding
will be used in this study. In this valuation procedure, a hypothetical
market is described and defined in detail (Randall and Brookshire, 1978).
Alternative levels of provision of the good are described. The insti-
tutional details pertaining to method of pa .t and enforcement of
terms are explained. An enumerator then poses prices to which the respondent
reacts, indicating whether he would pay the price or go without the good.
The price is varied iteratively until the price at which the respondent is
indifferent is identified. The process is repeated for several levels
of provision of the good.

For th-is s-t-wly, representative samples of residentia- water customers
will be drawn from two -' ior cities in Florida. Consumers will be visited
by an interviewer and confronted with a hypothetical situation in which
the water utility, because of growth in water demand and shortages of raw
water supplies, must either enforce a highly selective rationing plan or
else raise utility bills in order to finance expansion in water supply
capacity. Consumers are then asked, through an iterative questioning
procedure, to reveal their maximum willingness to pay to avoid the speci-
fied reduction in water use. The procedure will be repeated for contin-
gencies involving successively larger reductions in water use. In this
manner, a measure of willingness to pay to avoid each of several decre-
ments in water use will be obtained. A second set of questions present a
scenario in which the consumer is entitled to his current level of water
use and will elicit the amount of compensation necessary in order to induce
the consumer to willingly accept specified decrements in level of water
use. In this manner, a measure of willingness to accept compensation for
each of several decrements in water use can be derived. Payments and com-
pensations will be considered in the form of the total water bill. Separate
information about household income, household size, attitudes concerning
water conservation and other, similar variables will be obtained. Multiple
regression analysis will be used in order to test for statistically
significant relationships between the amount of bid (e.g., willingness to
pay) and variables normally associated with demand for residential water.










CHAPTER II


W,,l.' 'S SURPLUS AND VALUATIO' OF -F I'-Fi

Benefit Estimation


The objective of benefit-cost analysis is to direct the usage of
goods, services, or resources to their most highly valued employment.
Thus the measurement of values imputed by consumers to goods and ser-
vices hasaas is- ultimate purpose the .-.,-,W 11 ,Ffi;iEnt -.lacation
of resources.2 Non-market valuation methods have been developed for pur-
poses of estimating benefits and costs pertaining to provision of goods,
services, and resources for which established price-quantity data are
not available.

Consumer net benefits from a current or proposed resource alloca-
tion are defined and measured with the assistance of a theoretical concept
called the consumer's surplus. The concept of consumer's surplus has
theoretical underpinnings in the assumptions about consumer behavior of
the neoclassical economists.

Wi ,-oclassical Assumptions About Consumer Behavior

Utility Maximization

Consumers are presumed to make rational choices as to the level and
mix of goods and services they consume, with the objective of maximizing
their individually and subjectively perceived levels of satisfaction or
utility (Henderson and Quandt, 1971). Each consumer has, theoretically,
an indifference map for any combination of goods. He also has an income
constraint which limits the range of combinations, from the indifference
map, which he can afford to consume (given positive prices for the goods
and services). Assuming decreasing marginal rates of substitution among
goods and services, and assuming increases in consumption produce in-
creases in subjectively experienced utility, the rational consumer will
allocate his limited income among the goods and services in a manner
which maximizes-his utility.

Demand Functions

Given the assumptions about consumer behavior, the quantity of a
good demanded per time period is a function of the price of that good,



'Conceptually, economic efficiency is achieved if the allocation of
resources for production and consumption is Pareto optimal--a state
achieved if no reallocation of resources to improve the welfare of one
individuals can he made without reducing the welfare of one or more
other individuals (see Henderson and Quandt, 1971, fora summary).









the price of substitutes (or complements), the consumer's income, and
those aspects of beliefs, tastes, and preferences which underlie the
consumer's indifference map. Demand functions are mathematical formula-
tions which express the form and magnitude of the relationship between
quantity demanded, the dependent variable, and the relevant independent
or e. lanatory variables.

The Consumer's Surplus

Consumer's surplus is an important concept in the measurement of
social benefits in any social cost-benefit calculation; (Mishan, 1976).
A simple .1,-,, i -.-., -.i .- -'mI- 1 surpRluT is the maximum sum of money-
a consumer would be willing to pay for a given amount of a good, less the
amount he actually pays. By this definition, the market price in a per-
fectly competitive market is an adequate index of the value of a marginal
change in quantity of a good, but is not an adequate measure of larger
quantities of a good. In terms of the demand curve, beginning from a
given amount of the ;..... offered on the market, the corresponding point
on the demand curve indicates the maximum price the average consumer is
willing to pay for the last unit of that amount. But to each of the
total number of units purchased, as measured along the quantity axis,
there corresponds some average maximum valuation. The whole area under
the demand curve, therefore, corresponds to society's maximum valuation
for the quantity in question. Consumer's surplus is that portion of the
area under the demand curve above the price line.

In statistical estimates of the price-quantity relationship represented
by the demand curve, other variables known to influence quantity demanded
will be held (or assumed) constant (Mishan, 1976). However, if aggregate
money income is held constant in estimation of the demand function, any
fall in the price of the good raises real income. This increase in real
income will cause some increase (if the income effect is positive) in
the amount of the good purchased in addition to that increase in quantity
representing substitution of the good for some other in response to the
change in _relative prices.

Hicks (1941) redefined the concept of consumer's surplus, using an
ordinal system of indifference curves, as the amount of money--to be paid
by the consumer when the price falls; to be received by him when the price
rises--which, following a price change, leaves him at his original level
of welfare. This measure of consumer's surplus allows the consumer to
readjust the mix of goods which he consumes following the price change.

Henderson (1941) pointed out that, in general, the relevant compen-
sating variation in income would depend on whether the consumer had to
pay for the priviledge of buying the new -..,. or whether he was to be
paid for not being able to buy the good.

Subsequently, Hicks (1941) defined four measures of the
change in consumer's welfare resulting from an actual or proposed price
change. These four measures, summarized, are as follows:









(1) "Compensating variation" is the amount of compensation,
paid or received, that will leave the consumer in his
initial welfare position following the change in price
if he is free to buy any quantity of the commodity at
the new price.

(2) "',: .-:,sating surplus' is the amount of compensation,
paid or received, that will leave the consumer in his
initial welfare position following the change in price
if he is constrained to buy at the new price the
quantity he would have bought at that price in the ab-
s-eCnG of -com)pens-a4ti-o4n.

(3) "Equivalent variation" is the amount of compensation,
paid or received, that will leave the consumer in his
subsequent welfare position in the absence of the price
change if he is free to buy any quantity of the com-
modity at the old price.

(4) "Equivalent surplus" is the amount of compensation, paid
or received, that will leave him in his subsequent wel-
fare position in the absence of the price change if he
is constrained to buy at the old price the quantity he
would have bo.'ijlL at that price in the absence of compen-
sation.

The decision as to which measure is appropriate for a particular
analysis depends upon the specific nature of the proposed change for
which benefits and costs are to be measured. Hicks' four measures per-
tain to goods, priced in competitive markets, for which price and
quantity data are available. Benefit estimation in the context of
nonmarket, unpriced, or underpriced goods required adaptation of the
consumer's surplus concept to make it applicable in quantity change
analysis as well as in price change analysis.

Consumer's Surplus: Quantity Chage Anal y s i s

In benefit-cost analysis, the economist is sometimes concerned not
so much with the welfare impacts of price changes as with the welfare
impacts of changes in the bundle of goods, services and amenities
possessed, used or consumed by individuals (Randall and Stoll, 1980).
Proposed projects or programs may remove some goods from individual
opportunity sets or introduce new goods; and may decrease the quantities
of some goods while increasing quantities of others.

The goods affected by proposed programs may be divisible, exclusive,
marketed goods with observed prices. However, they may also be re-
creational or environmental amenities or other goods which are in varying
degrees indivisible, non-exclusive, and unpriced (Randall and Stoll, 1980).

Randall and Stoll (l '-'0) identify the conditions under which consumer's
surplus measures can be a'! pted to situations in which it is bundles of
goods, rather than prices, which are changed. Their diagrammatic exposi-
tion (Figure 1) proceeds as follows.











































Y*
y \















yg i


-- III


_ I*


, II


QUANTITY OF X













Figure 1.--The welfare impact of a change in the quantity of a good
X from Q" to '.


. I "










Consider a normal good X, which, depending on the pro-
gram alternatives chosen, may be provided in two different
quantities, Q' and Q", where Q" is greater and ceterus
paribus preferred. The pragmatic measures of value of these
two bundles of ..--: are willingness to pay (WTP) and
willingness to ac -.,t (ITA). In a market exchange situa-
tion these correspond, respectively, to the buyer's best
offer and the seller's reservation price; in a nonmarket
situation, they correspond to willingness-to-pay-and
willingness-to-accept compensation (Randall and Stoll, 1980).

If the program alternative under evaluation would re-
duce the quantity of X from Q" to Q', the compensating mea-
sure of welfare loss is WTAc, the compensation which would
keep the individual at his initial welfare level; and the
equivalent measure is WTPE, the loser's willingness to pay
to avoid the quantity reduction from Q" to Q' which, if
paid, would place the individual at his subsequent welfare
level (Randall and Stoll, 1980). If the proposed program
would increase the quantity of X from Q' to Q", the com-
pensating measure of welfare gain is WTPc which, if paid,
would keep the gainer at his initial welfare level; and
the equivalent measure is WTAE, the compensation which
would be needed to bring the potential gainer to his sub-
sequent welfare level in the event that the proposed pro-
gram is not implemented (Randall and Stoll, 1980).

If X were a perfectly divisible good, traded in large markets at
zero transactions costs, a program to reduce Q" to Q' while leaving the
individual's numeraire, Y (a composite of "all other goods"), at Y
would initially move the individual from point E to B (Figure 1), lowering
his welfare level from I" to I' (Randall and Stolli, ,,ii). However, the
existence of frictionless markets will permit him to trade along his
new budget line until he reaches D, achieving the welfare level of I*.
i' i U.i PT l- i ,. i n ,l' l,_-i i l I !II-'E i J-l : -. | lFl :, 1 ," i 1,- I rI' '
is BC, equal to Y'Y". Thus, WTPE is equal to WTAc.

Now, assume that X is lumpy and can only be held in the amounts
Q" and Q'. Since intermediate adjustments in commodity holdings are
not permissable, the Hicksian compensating and equivalent measures in
commodity space are analogous to the Hicksian surpluses, not the varia-
tions, defined over price changes (Randall and Stol 1980). Accordingly,
the price lines (Figure 1) become meaningless. WTP is EG which is equal
to Y9Y, and WTAc is BA, which is equal to yya, and larger in absolute
value than WTPE.

In identifying the appropriate measure of welfare change, several
distinctions among the Hicksian measures must be clarified (Brookshire,
et al., 1930):









(1) The Hicksian compensating and equivalent measures of
consumer's surplus differ with respect to the reference
level of welfare. The compensating measure, by using
the initial welfare level as the reference level, mea-
sures the welfare impact of changes as if the individual
had a ri:lt to his initial level of welfare. The equiv-
alent measure, by using the subsequent welfare level as
the reference level, treats the individual as if he had
a right only to his subsequent level of welfare.

(2) The Hicksian variations differ from Hicksian surpluses
in that variations are ca-lculated after the- consumer
has made optimizing adjustments in his consumption
set, while surpluses are calculated without first al-
lowing such adjustments.


-10-









CHAPTER III


NON MARKET VALUATION: THE ITERATIVE BID

Non-Market Valuation ii-thod

Measures of value for goods and services traded or sold in private
markets are typically derived from market-generated price-quantity infor-
mation. In the case of public goods, and publicly provided goods, valua-
tion methods must be devised which emulate the market process or in some
other manner generate information from which measures of value can be
dlerived.

Non-market valuation methods have been applied in benefit estimation
studies of environmental improvements, the creation or improvement of re-
creation sites, and the provision of wildlife. These methods fall into
two general categories: the proxy methods and the bid game methods.

Proxy methods require the choice of variables upon which quanti-
fiable observations can be made (i.e., for which data exist) and which
are hypothesized to be highly correlated with the price (or other)
variable for which data do not exist. An example of the proxy method
is the use of measurable travel costs as a proxy measure of the willing-
ness of recreationists to pay for the composite of experiences associated
with a particular recreational activity, taking into account other explana-
tory variables such as income, substitutability of other recreation sites,
and individual tastes (see, Sinden, 1973).

The bid game method was pioneered by Robert K. Davis (1963) who
used it to estimate the benefits of maintaining New England wilderness
areas. The bid game method elicits direct valuations of goods or ser-
vices from consumers without the use of intermediate variables. Typically
the bid game uses a questionnaire format. Respondents are oriented to
a hypothetical market (scenario) in which the current level of a good or
service is assumed to exist, Then it proposes a change in this level of
provision and records the respondents' valuation of that change in the
level of provision of the good. Several procedures have been used for
obtaining valuations or bids. With an open-ended response format the
respondent is unconstrained in providing bids. With a categorical for-
mat, the respondent is presented with a predetermined set of bid pos-
sibilities. The iterative bid format presents the respondent with a
series of alternate states, iteratively eliciting the respondent's valua-
tion of each change from the reference state (Adams, et al., 1980).

The use of prices distinguishes the market from the non-market ap-
proaches to valuation (Bradford, 1970). Demand functions for private
(marketed) goods quantify demand responses to price changes, and esti-
mation of demand functions is a first step toward benefit estimation.
fn the case of public goods and publicly provided goods for which pro-
duction and consumption is often divorced from consideration of individual
willingness and ability to pay, consumers do not normally have the op-
portunity to purchase as many units as they wish. They are, instead, con-
fronted with public decisions to change the quantity or quality levels of
provision without joint reference to price. The term "states" is used
to refer to current and subsequent levels of provision.


-11-








The Bid Curve and Consumer's Surplus

By soliciting valuations in terms of willingness to pay and/or
willingness to accept compensation, the bid game method is designed to
measure the consumer's surplus. Thus it has the same theoretical basis
as value estimating procedures using estimated demand functions for
marketed goods and services.

The Bid Curve

Consider an individual who currently enjoys some specified level,
i. -- i J .'i. .-..- He -also enroys-a given quantity-of the
Hicksian "all other goods" numeraire, Y, for convenience called income.
His level of utility, then, is a function of his income and the level
nf provision of the 'J..... represented by Q, i.e.,

(1) U = U(Q, Y).

The individual is at the origin (Figure 2), which defines his level of
welfare in the "without project" situation. To the right of the origin,
the level of provision of Q to the individual increases; to the left of
the origin, it decreases. From the origin, a move up the vertical axis
represents a decrease in income, while a move down the vertical axis
represents an increase in income.

A total value (TV) curve, or bid curve (Bradford, 1970), passes
throt.. the individual's initial state. It is of positive slope, given
that the individual is not satiated in the range of values under consid-
eration. For decreases in Q, the TV curve lies in the southwest
quadrant; for increases in Q, it lies in the northeast quadrant.

The TV curve is an indifference curve, that is,

(2) U(Q, Y) = U(Q Y+) = u(+Q Y ) .

Starting at the origin, Y0 -y is the individual's willingness to
pay (UTP) for an increment in the provision of good Q from QO to Q+.
Willingess to accept (WTA), i.e., Y YO, is the amount of money which
would induce the individual to accept voluntarily a decrease in the level
of provision of the service from QO to Q~. Restating equation (2),

(3) U(Qo, yo) = U(Q-, Yo + WTA)
= U(Q+, Yo WTP)

IUTP WTA, and Consumer's Surplus

To clarify the relationship between Hicksian compensating and
equivalent measures of value, WTA and WTP, and the total value curve of



'l-he discussion of the bid curve follows that in Brookshire, et al.,
( '-, )


-12-














Income P/ rice Line


Yo p(Q+ Q) / Total Value
/ Curve
S- Q


crements in Q n /

//


+ Y
.^ -__ Y+

'T/r


Increments in Q


+ P(Qo


- Q-)


Figure 2.--The total value curve for increments and decrements in the
level of provision of a service, Q, for an individual who
initially enjoys the level Q and the income Yo.


-13-


De









Figure 2, Brookshire, et al (1980) offer the following example. The
subject of benefit-cost analysis is a project which would divert a
specified area of wildlife habitat to some alternative use, effectively
destroying its usefulness as habitat. The analyst needs to know the value
of the losses which would be suffered by an individual who currently
enjoys the wildlife amenities provided by that habitat. In the initial
state the individual has utility level U(Qo, yo). His "with project"
utility level will be U(Q-, YO). The "with project" and "without
project" levels of Q are predetermined so that the individual has no
opportunity for optimizing adjustments.

O.. ..U i. i Ti-,,: welfare timpact -on tthis individual would be his
WTA to acquiesce in the proposed change., Call this WTACO 0 o 0
Q YO; QoyO; Q-o
Superscript C indicates that this is a Hicksian compensating measure of
value, the first subscript pair, QO, Yo indicates that the individual's
reference level of welfare (his presumed right or entitlement) is QO, YO.
The second subscript pair indicates that QO, yo is also his initial wel-
fare level. The third subscript, Q-, indicates the level of provision
of the good (in this case, wildlife-related services) the consumer would
enjoy after he has accepted the compensation and the change in the level
of services. If he were compensated by an amount just equal to his WTA,
his income after compensation would be YO +-WTAc. WTAc is a measure of
the individual's valuation of the reduction in wildlife-related amenities
from Qo to Q" and was derived from the individual's TV curve for
wildlife-related amenities (see Figure 3).

However, another measure of value might have been used to estimate
the individual's loss of wildlife-amenities: the Amount of money he
would be willing to pay to avoid a reduction in the provision of wildlife
amenities. This WTP to avoid a less preferred position reflects a pre-
sumption that the individual has no right (entitlement) to his current
welfare level. The reference level of welfare is not that associated
with the initial situation, but the proposed (or subsequent) welfare
ITveT f This secaondhmeasur- of the in dividuaT'-s welfare-+loss is denoted-
WTPE The superscript E indicates a Hicksian equivalent
Q-, YO; QO y Q0
measure of value. The first subscript pair indicates that the reference
level of welfare (his entitlement) is taken to be Q-, Yo. The second
subscript pair indicates the individual's initial state, Qo, yo. The
third subscript indicates that the individual, after he has paid will be
allowed to enjoy the Q, level of amenities. If he pays WTPE, his final
income will be Yo WTPC

Brookshire, et al. (1980) note that the pair of total value curves
in Figure 3 could be used to estimate the value of a project which would
increase the level of wildlife-related amenities from an initial level
Q- to a "with project" level QO. The individual's initial state is
Q-, Y0. If the individual is entitled only to his initial -welfare level,
the appropriate measure is UTP- yo. Q- yo. QO Note that WTPc equals

WTPE 11 the individual were entitled to the additional wildlife-related
amenity, the measure is T which equals WTACo 0. o ,
0 Q- y ; O


-14-












Income


YO ITPC






Yo







E/


YO + WIA


Qo4yo;Q-,YO;Q-


TV(Q-, Y-)


Q-Y; Q0 ,Y0;Q0


YO TV(QO,Y0


ff
.9
/
I
'!


A


Quantity of Q


Yo + WTAC
QO,yo;Qoyo;Q-


Figure 3,--The relationships between WITP and WTA, and Hicksian compen-
sating and equivalent measures of consumer's surplus.


Income
I I


L g








The example makes several points


(1) Equivalent measures apply to situations in which the
initial welfare level is different from the reference
level, when the individual's "entitlement" is given by
the subsequent state rather than by the initial state.

(2) Compensating measures assume that the initial state is
the reference welfare level; that the individual's
"entitlement" corresponds to the initial state.

(3) WTPE cannot be found using a TV curve passing through
the individual's initial state. It can be Found only
by using another TV curve passing through the reference
state.

(4) There is a compensating and equivalent version of WTP,
as there is of '1-\. When comparing two alternate'levels
of provision of a good, there are four relevant Hicksian
value measures: WTPc to obtain the preferred level;
WTPE to avoid the less preferred level; WTAc to accept
the less preferred level; and WTAE to forego a promised
increment to the preferred level.

(5) WTA and WTP, whether they be compensating or equivalent
measures, represent Hicksian variations if the consumer
has an opportunity to make optimizing readjustments ,in
his consumption set.

(6) WTA and WTP, whether they be compensating or,'equivalent
measures, represent Hicksian surpluses if no optimizing
adjustments in the individual's consumption set are
possible.

Application of the Iterative Bid

Application of the iterative bidding technique [also known as "con-
tingent market valuation (Randall and Brookshire, 1978)] requires identi-
fication of the distribution of rights. That is, the quantity of a good
to which a respondent is entitled must be determined before the relevant
consumer's surplus measure can be chosen. For example, if the consumer
is entitled to the quantity he currently consumes and a proposed project
would reduce the quantity available to him, the relevant consumer's sur-
plus measure would be WTAc. On the other hand, if the consumer is en-
titled only to the reduced quantity the relevant measure would be UTPE.

The contingent market or iterative bid format requires that the
respondent understand the quantity and quality characteristics of the
scenario with which he is confronted. For example, in a study of willing-
ness to pay for environmental improvements, Randall and Brookshire (1978)
used photographs to ::pict levels of provision of cleaner air to ensure
uniform perception among the respondent population.

The pa,..,'t method or vehicle must be chosen for its relevance and
feasibil ity and must be clearly specified to the respondents (Randall and


-16-








Brookshire, 1978). For example, expressions of willingness to pay for
increased recreational opportunities or wildlife-related amenities might
best be elicited in terms of willingness to pay hunting license fees or
recreational site access fees, since these vehicles of payment are famil-
iar to consumers and relevant to the specific goods or amenities for
which measures of valuation are sought.

In the iterative bidding form of contingent market valuation the
respondent reacts to prices posed by an enumerator, indicating whether
he would (in a WTP case) pay the price or go without the good. The price
is varied iteratively until the price at which the respondent is indif-
ferent is identified (Randall and Brookshire, 1978). T!,-. h pl tl tical
market thus established has the advantage of low transactions costs,
"trade" in goods which cannot be marketed in the conventional sense, the
ability to evaluate many options and perturb the components of publicly
provided packages of goods (and packages of public goods) in order to
examine the contributions of these components to the value of the package.
Contingent market approaches to valuation are inferior to actual markets
in that the bids obtained are not firm, enforceable offers, but are be-
havioral intentions given the occurrence of the hypothesized contingencies.
If the survey instrument is realistic and coherent, then the main condi-
tions under which behavioral intentions should predict actual behavior
can be met. Studies by Ajzen and Fishbein (1977) and Crespi (1971)
have been used as the psychological foundation for the economic research
in hypothetical valuations.

Potential Bias in Iterative Bids

Empirical application of the iterative bid poses three problems:

(1) the potential exists for respondents to engage in strategic
behavior;

(2) variations in responses may result from using different starting
points for the same bid; and

(3) variations in response may result from the use of different
vehicles of payment for the same bid.

Strategic Behavior Bias

Of the three possible sources of bias, respondent strategic behavior
is of the most concern and has consequently precipitated the most research.
The problem of strategic behavior arises from the very essence of the con-
tingent market valuation method: its hypothetical nature.

It is necessary to distinguish strategic behavior which may result
from paucity of information concerning procedures and purposes of the
survey, or from a marked divergence between the survey scenario and ex-
periences of real life. In other words bias can be introduced into par-
ticipant responses simply from confusion.

Strategic behavior results when the respondents in an attempt to
maximize personal self-interests, sabotage the bid game by responding dis-


-17-









honestly to the questions. It has been generally accepted that respondents
attempting to maximize the benefits which may possible accrue from a sur-
vey encounter will answer questions in a predictable manner. The two
main factors which affect the response are (Freeman, 1979):

(1) the individual 's perceived personal influence on the
outcome, and

(2) the probability that he will actually have to forfeit the
amount he states as his maximum willingness to pay (when
WPA is the desired measure).

The tendency for an overstatement of one's true willingness to pay occurs
when the individual desires a certain outcome and believes that his re-
sponses will affect that outcome but that he will not be asked to ac-
tually relinguish that bid amount. An understatement will occur when a
respondent desires an outcome but believes that the number of partici-
pants is large c-nii,-.11, to ensure that outcome regardless of the size of
his bid which he may, in fact, be required to relinguish. These two
biases are considered to be the upper and the lower bounds on bid esti-
mates and are the key targets for solutions to systematic bids in sur-
veys (Bohm, 1972).

Researchers have attempted to adjust for the effects of strategic
responses in two ways. The first approach attempts to measure, statis-
tically, the bias imposed by strategic responses and adjust recorded
responses by this calculated degree of bias (Kurz, 1974). This proce-
dure is based on the unverified assumption that all biases in a given
situation are uniform, and suffers from lack of operational format to
follow from its theoretical exposition.

The second approach to potential strategic behavior attempts to
structure questions in the survey situation to eliminate incentives for
strategic behavior (Bohm, 1971). Respondents are left uncertain
as to how their responses will affect their payment outcome. This can
be achieved by informing them that they may be called upon to pay their
bid, some proportion of their bid, or to make no payment at all. This
uncertainty will create no clear advantage to the respondent in under-
stating or overstating his true bid. This approach of "outwitting" the
respondent has been explored theoretically by Maler (1974), Kurz (1974),
and Tideman and Tullock (1976). It is the approach used in the at-
tempts at non-market valuation represented by the work of Adams, et. al.
(1980), Cicchetti and Smith (1973), Randall, et. al. (1974), and Hammack
and Brown (1974).

Elimination of incentives for strategic responses may also eliminate
incentives for giving accurate answers (Freeman, 1979 ). Care must be
taken to stress the importance of careful and thoughtful answers (Hammack
and Brown, 1974). It is also probable that respondents will, in any case,
attach a probability of their own choosing to the potential influence
of the s. : and of their bids (Freeman, 1979). These problems are
embodied in what Randall and Brookshire (1978) calls conflict between
"strategic versus hypothetical bias," the latter being the result








...not of systematic influences but rather of noise
resulting from failure to invest as much effort in
the contingent decision as would be invested in an
actual decision, presumably because the penalties
from a wrong decision in a hypothetical market are
not so tangible (Randall and Brookshire, 1978).

This discussion reemphasizes the need to distinguish between genuine
strategic behavior and the bias caused by lack of relevance, and accen-
tuates the need for conveyance of reality in the design of contingent
market scenarios.

Although no methodology exists which conclusively tests for strategic
bias, two empirical studies made by Bohm (1972) and Brookshire, et al.,
(1976) have attempted to reveal its existence. Bohm's study consisted
of five separate surveys, each intended to elicit a certain biased re-
sponse in situations which were not ',.pthetical (respondents had to ac-
tually pay their total bid or some proportion. The main result of the
test was that none of the five surveys yielded average bids which signi-
ficantly deviated from any of the others. Results of surveys which were
designed to elicit overstated bids did not differ greatly from those
which intended to elicit understated bids (based on criteria mentioned
above). The difference came with the sixth survey which was based on a
situation known to the respondents as "hypothetical." Bohm concluded
that, in any case, the behavior exhibited supports the notion that re-
spondents tend to view their impact on total demand as -important and that
understatements are neutralized by the potential threat that a good may
not be provided. Bohm makes it clear that this limited testing does not
disprove the possibility of bias but suggests that the deviation between
honest and biased response may be small.

The study by Brookshire, et al. (1976), which measured the value of
improvements in air quality of an Arizona recreation and wilderness area,
confronted survey respondents with a hypothetical situation in which the
respondents knew the survey personnel were not working in an official
it, i i_ l II.. 1-. ,_,'-, ,o.ul ni-o-be- requi-re- t-&-aet-u&a-l-y- rov-ide--for pay---
ment or receive compensation. The authors hypothesized that bias, if it
existed, would be exhibited in the following way: when asked what their
maximum willingness to pay would be to insure an improvement in environ-
mental quality, -the "environmentalists" would tend to overstate their
bids, far exceeding the mean, and "non environmentalists" bids would tend
.o be zero. The authors assumed that honest bids would be distributed
normally, and that bias of the hypothesized type, if widespread, would
tend to flatten the distribution. They concluded that strategic bias
was not prevalent in their survey because the distribution was, in fact,
highly centered about the mean bid. However, they recognized that strategic
bias may still exist since they have no way of insuring that bids were in-
fluenced by an incentive to be accurate.

In summation, strategic behavior in hypothetical :.i..1,iing cannot be
completely eliminated because it cannot be accurathly'/bFasured. At best,
its vestiges, arising in either an actual or hypoth ta') situation, can
only be hypothetisized. That the problem exists.,,h.owever, .is generally
agreed upon. The ultimate caveat is that the confidence associated with


-19-









the statistical testing of bid estimates does not indicate the reliability
with which behavioral intentions will be translated into actual behavior.

Vehicle of Payment Bias

Vehicle of pa .,,'.it bias can be said to exist when the value of mean
bids or the incidence of protest bids are significantly different across
payment vehicles (Randall and Brookshire, 1978). In such cases it is
possible that respondents are revealing a preference for paying a bid in
one form rather than another and disguising a true willingness to pay.

T,.u _tL.1i .. ,:... .j--i-fi, ..ill,' a4rse-s-ed- the po4t-ential f -or ve-h4i-c e- -o-f
payment bias in their work. Randall et. al. (1974) did not find vehicle
bias at statistically significant levels in their research, but at less
strain iL levels of significance some differences were observed between
the sales tax, the user fee, and the electricity bill forms of payment.
The authors suggested that the vehicle of pa,iint selected for any res-
pondent group should be the most germane to the issue at hand. This
appears to make the most intuitive sense. Individuals confronted with
forms of payment which they do not prefer may tend to give low bids, and,
while it may seem less likely, persons confronted with the payment form
"of their choice" may tend to give higher bids. The most likely-to-be-
employed form of pa. Inii't would assist in smoothing the extremes.

Starting Point Bias

Starting point bias exists if different bids are elicited from a
set of questions which contain, ceterus paribus, different starting
points. If a starting point bias exists, a bid game with iterations be-
ginning at $2.00, rather than at $1 00, would produce bids which are
biased upward. Randall and Brookshire (1978) suggest that starting
point bias is only a problem for situations in which respondents have
little independent basis for valuation of the good. As with vehicle of
payment bias, starting point bias may be tested for by varying the
starting point across the sample and testing whether the responses dif-
fer -si gifTcaintty between groups.


-20-









CHAPTER IV


VALUING RESIDENTIAL WATER USE BY
CONTI !-,rf MARKET TECHNIQUES

WTP, WTA, and Consumer's Surplus
for Residential Water Use

The relationship between Hicksian compensating and equivalent mea-
sures of value and WTA and WTP in the context of residential water is
illustrated by the following scenario. Hypothetically, if current
trends in population growth. continue, demand for water will exceed
existing capacity for supply in five years.4 A project to augment
water supply capacity will make it possible for current rates of use
per household (Q, in Figure 4) to be sustained in five years. Failure
to invest in increased supply capacity will require an average decrement
in Q relative to current rates of consumption.

If it is assumed that citizens are :entitled to their existing level
of utility, the benefit-cost analyst wishes to measure consumer's WTAc
(Willingness to Accept payment) for voluntarily accepting decrements in
water availability and, therefore, water use. This measure is a com-
pensating measure, denoted by superscript c, since it assumes individuals
are entitled to current utility levels.

Figure 4 depicts the situation just described. The consumer starts
at Y, Q, and is entitled to the level of utility associated with that
situation. If a decrement from Q' to Q- is inevitable, the question to
the consumer becomes, iThat amount of compensation will you accept to
voluntarily acquiesce in the decrement in Q and feel as well off as
before the decrement in Q? In Figure 4 the WTAc is depicted at Yo + WTAc"

On the other hand, if a decrement from Q0 to Q- is inevitable and the
consumer is not entitled to his existing level of utility, the appropriate
question to ask the consumer is, "How much are you willing to pay to avoid
a--decrerment i+n-water- avaalrb-ilt-ty from Q- to--( ?-" In ase, the-eve
of welfare (reference level) to which the individual is entitled is that
associated with Yo, Q- in Figure 4. The WTPE (Willingness to Pay) to
avoid the decrement in Q represents a decrement in income relative to the
initial Yo level-, leaving the consumer at income level Y WTPE. Moreover,
the contingency as proposed entails a loss of welfare for the individual,
a new reference level of Y and Q, and a different TV curve.5 The super-
script, E, denotes an equivalent measure.



4The time reference to five years is selected arbitrarily for purposes
of illustration.

5An additional point can be raised concerning interpretation of the
above scenario. The scenario posits "inevitable" reductions in Q (unless
supply capacity is augmented), and for urban water systems customers, it
is not clear whether individual customers would (or could) make subsequent
quantity adjustments. If it is assumed, for example, that customers can


-21-









Income










Yo
/
/


Income

Q ,



IJ




/./


/


4 O- WTP E


- Quantity of Q


- -- -- I Yo- + WTAC


Figure 4.--Relationship of WTP, WJTA and Hicksian measures of surplus:
the context of residential water use.









Exposition Using Traditional Indifference Curves

Figure 4 depicts the relationship of WTPE and WTAc to Hicksian equiv-
alent and compensating surpluses in terms of the total value (bid) curve
for residential water. Figure 5 allows dia .,ni-,tic exposition in terms
of traditional indifference curves [following the exposition by Randall
and Stoll (1980)].

Consider an individual whose rate of use of residential water, Q, and
an "all other goods" numeraire represented by income, Y, is currently
Qo and Yo, respectively (corresponding to the origin of Figure 4) and per-

sume that Q must be reduced to Q- from Q, but the individual is entitled
to remain at his original utility level. If no income compensation were
recei'.i by the individual he would find himself with income Yo (un-
changed) and quantity Q- (reduced from Q) permitting him to achieve only
utility level I' at point B.

Since the scenario depicts a quantity change without reference to rate
or price changes, price lines are irrelevant to this indifference curve
analysis.6 Compensation needed to return this individual to his original
utility level will be BA which equals YY', and which corresponds to WTAc
in Figure 4. In Hicksian terms, this WTAc is a compensating surplus.

Alternatively, consider an individual with income Y' and quantity of
water Q who is at point C on utility level I". Assume further that the
level of Q must be reduced from Q to Q- and that the individual is not
entitled to compensation. The individual's willingness to pay to continue
using Q of water is WTPE and will be the amount of Y he would have to give
up in order to reach point D, on utility level I', corresponding to
quantity QO. This WTPE is, in Hicksian terms, an equivalent surplus and
amounts to CD in Figure 5, equaling Y"Yo. On Figure 4, this corresponds
to WTPL. This arrangement leaves the individual on a different lower
indifference curve, I' at point D.



not resell publicly supplied water in a competitive market, the relevant
compensating and equivalent measures are Hicksian surpluses, rather than
variations. On the other hand, if it is assumed that competitive markets
for the re-sale of publicly supplied water would emerge in the wake of man-
datory supply reductions (or use reductions), then the compensating and
equivalent measures are Hicksian variations, not surpluses.

6The irrelevance of price lines also hinges on the assumption that no
"after market" for the private exchange of publicly supplied water exists.


-23-



















Y' = Y- + WTAc

,f


Y"1 = yo I


WTA <








WTPE



,]TpE .


' \

i










; \
i
1


Q- Q

Quantity


Figure 5.---Diagramatic exposition, using traditional indifference curves,
of the relationship of UTP, WTA, and Hicksian measures of
consumer surplus; the context of residential water use.


-.,,
i




i


I


-..... I


-24-








Valuation of Residential Water Use


Introduction

This study pursued three objectives:

(1) to measure consumers' valuations of the losses in utility
which correspond to reductions of water in specified resi-
dential uses,

(2) to identify the major determinants of these individual
valuations, and

(3) to quantify the relationship between those determinants
and the individual valuations.

The method employed to measure consumers' valuations was the contin-
gent market, using the iterative bid. Selection of independent variables
was based on the hypothesis that variables which affect a household's
monthly water demand will be similarly correlated to the individual con-
sumers' valuations. Multiple regression analysis was used to test for,
and to quantify, significant functional relationships between the selected
independent variables and the consumers' valuations.

The Dependent Variables

WTP and WTA were solicited by personal interviews with a sample of
residential water users and are direct measures of the consumer's sur-
plus derived from residential water use. The individual .bids were used
to compute the mean bids for the sample.

WTP and WTA bids were obtained for each of five categories of
resident ial water use:

(1) toilet flushing,

(2) bathing and showering,

(3) clothes washing,

(4) lawn watering, and

(5) total household water use.


The contingent market scenarios

Each respondent was visited personally by a survey enumerator and
asked to respond to a hypothetical situation concerning the eminent


70f these five categories, the bid functions for clothes washing and
lawn watering were not estimated because the data base contained too many
missing values in the corresponding bid and quantity figures.









availability of water for selected household activities. Specifically
they were told:

Please indicate your willingness to pay, through
charges in your monthly water/sewer bill, for water
in specific uses. Imagine that requirements are
set for water use reductions but that you may
avoid compliance if you are willing to pay to do so.

Respondents were then allowed, through an iterative procedure, to
select a monetary bid representing their willingness to pay, for example,
i... j-, _i -, ,i,,l i,., .-, I1 :n. r-. :, t-,_,,, and-a- 40% red-uctioi-, -resp-ecti-ve-y-,
in each of the five categories of household water use.8 To elicit bids for
each reduction contingency, the respondents were asked the following:

If there was a requirement to reduce bath/shower
usage in your household by 20%, how much additional
money would you be willing to pay per month to
avoid complying with this requirement?

Iterations of bids were then presented to the respondents, beginning with
the current price they pay for the amount of water represented by the re-
duction contingency. The completion of this process produced a set of
WTP bids.

The WTA bids were elicited in a similar manner, essentially using the
following scenario:

If you had decided to comply with the requirements
of water use reduction what amounts of compensation
would you require to repay you for the inconveniences
which you incur when you make these reductions?
Please indicate the minimum amount of ii,.' t' which
you could accept through reductions in your monthly
bill, for specific water uses.

Dealing with sources of bias

It was crucial to the design of the iterative bid section of the
questionnaire that the problems of bias (see Chapter III) not be built
into the data. Although strategic bias, starting point bias, and
vehicle of pa in,:it bias cannot be conclusively identified or eliminated,
efforts were made to minimize their occurrence.



-The entire questionnaire has been included as Appendix I. Reduc-
tion contingencies were expressed in different terms for different uses,
eg., in terms of hours per week for lawn watering, flushes per day for
toilet flushing, cycles per week for clothes washing, etc.









To counteract the potential for strategic bias, the bid scenario was
characterized as being hypothetical (it was emphasized that the enumerators
were not acting in any official capacity). On the other hand, enough
"reality" may well have been injected into the scenario by the fact, well
known to the consumers, that recent water price increases had been ef-
--ted in both cities from which samples were drawn.9 It was hoped that
in consequence of these two factors respondents would see no personal
benefit to sabotaging a hypothetical bid game but would recognize the
general pertinance of the scenario.

To minimize the likelihood of starting point bias, current water
rates -weyre used as-the-basis- for i1i -. q.ii, i i-d- for -all -cquesti-o4is -
Respondents were therefore not confronted with unrealistically high or
low initial bids.

Care was taken to minimize vehicle of payment bias by using the most
realistic form of payment: the water bill. Since no other form of
payment is relevant there was no need to include other payment vehicles
to test for significant discrepancies.

The personal interview method was used to minimize the likelihood
of bias resulting from misunderstanding of the purpose or procedure of
the questionnaire.

The Independent Variables

The variables which affect a household's monthly water consumption
are hypothesized to also affect that household's valuation of water
use. The logic behind this hypothesis is relatively simple: the WTA
and WTP bids are measurements of value, or more accurately, expressions
of individual perceptions of value. This value is derived from the utility
of household water which, in turn, is a function of how water is used and
how much water is used.

How water is used and how much is used by a household are items of

Some variables were expected to have positive influences on valuation and
others were expected to have negative influences.

The variabl-es were categorized in the following variable groups:

(1) the household status group

(a) number of members per household

(b) annual income



9The sampling procedure and characteristics of the cities from
which samples were drawn is included as Appendix II.








(2) the household technology group


(a) average total monthly water consumption

(b) average monthly water consumption for each water use
category

(c) the presence of a private well

(d) type of irrigation system

(3) the conservation 'o1,-

(a) belief in local water shortage

(b) expressed willingness to decrease water consumption if
the price were increased by 50%.

The rationale for selecting each variable and the hypothesized nature
of its relationship to individual bids, varies from one variable to another.

The household status group

Size of household. The number of persons in a household is hypothesized
to have a positive influence on the willingness to pay to maintain current
levels of water use. As the number of persons in a household increases the
necessary minimum rate of water use also increases. Size of household is
also expected to have a positive influence on the willingness to accept
compensation for essentially the same reason.

Annual income. Income is hypothesized to have a positive influence on
the willingness to pay an increased amount to maintain current water use
levels. Willing .ji to pay is expected to be related to ability to pay.
It may also be positively correlated with the willingness to accept compen-
sation for water use reductions.

The household technology group

Monthly average water consumption. The monthly average quantity of
water that a household uses is hypothesized to be positively correlated
with both the willingness to pay and the willingness to accept bids
It is expected that households which require relatively large quantities
of water are those which rely upon numerous water using appliances, e.g.,
more than one bath room, and/or have larger family sizes.

Monthly average consumption for each water use category. The presence
and use of specific household water using fixtures is an indication of
how much water the household is likely to feel it "needs." It is hypoth-
esized that the average rate of each water use will be positively correlated
with willingness to pay to avoid, and willingness to accept payment to
acquiesce in, reductions in those particular use rates.


-28-








Presence of private well. The presence of a well is hypothesized
to have a negative effect on willingness to pay to maintain current levels
of water use because the well is a substitute for publicly supplied water.
If the well water iis used for irrigation only it is not a perfect sub-
stitute since it is often not of potable quality and because there are
normally official restrictions on the use of wells. Consequently, the
effect of a private well on WTP and WTA will be less if the well is
used only for irrigation.

Type of irrigation system. Irrigation systems which use the most
water should have the greatest positive effect on willingness to pay to
maintain current levels. The automatic sprinkler system requires more
water to o-i.i ffi: i,-ti I, tlian does a rotating hose-andi-sprTnkler
system. Therefore, the presence of the former should be more highly
correlated with WTP and WTA than the latter.

The conservation variable group

Belief in water shortage. The belief in a current or pending water
shortage is hypothesized to have a positive effect on the willingness
to pay to maintain current levels of water use. Recognition of scarcity
implies recognition of the likely increase in cost of maintaining a
particular rate of supply to households.

However, recognition of water shortage situation may have the op-
posite effect on WTA. Recognition of a true shortage may entail a willing-
ness to acquiesce in measures to curtail use without demanding or expecting
compensation.

Willingness to reduce consumption in the event of an increase in
price of water. This variable measures a willingness to conserve water,
reflecting an attitude about the need or desirability of conservation.
It is .,ithesized that a greater willingness to reduce consumption is
positively correlated with a willingness to pay to avoid reduction in
use and negatively correlated with WTA.

Estimating Equations

Willingness to Pay

The estimating equations to explain willingness to pay to avoid
reductions in total water use are:

(1) WPWAT10 = f(NUMRESPH, INCOME, AVGCON10, HAVEWELL, KI 'L. 'S,
BELWSHRT, DECON50),

(2) l'F.:.'T30 = f(NUMRESPH, INCOME, AVGCON30, HAVEWELL, KINDSYS,
BELWSHRT, DECON50),

(3) HTWAT50 = f(NUMRESPH, INCOME, AVGCON50, HAVEWELL, KINDSYS,
BELWSHRT, DE( ii '.0),


-29-









where:


WPWAT10, WPWAT30, WPWAT50 = amount of increase in the total monthly
water bill that a consumer is willing to pay to avoid 10%, 30%,
and 50% reductions, respectively, in monthly average water con-
sumption,

NUMRESPH = number of persons residing in household,

I iCOlE = combined annual income of persons residing in household,

AVGCON 10, 30, 50-= = i ,I apr-eseenting -a negative 10%,- 30% 50%--
respectively of monthly average water consumption,

HAVEWELL = indication of whether or not household owns a private well
and whether or not it is used solely for lawn irrigation,

KINDSYS = indication of whether or not household waters a lawn and
what method, hose and sprinkler, or automatic system, is
applied,

BELWSHRT = indication of whether or not household believes that there
is a current water shortage in its locale and/or that there will
be a shortage by the year 2000,

DECOi'0.O = indication of whether or not household would be willing to
reduce its household water consumption if the price of its water
rose by 50%.

The estimating equationsto explain willingness to pay to avoid re-
ductions in bath and shower use are:

(4) .I;S20 = f(NUII'LR PH, INCOME, BATHSQ20, BELWSHRT, DECON50),

(5) WPBS30 = f(NUMRESPH, INCOME, BATHSQ30, BELWSHRT, DECON50),

(6) WPBS40 = f(.!Hi.i!iriPH, INCOME, 3ATHSQ40, BELWSHRT, DECON50),

where:

WPBS20, WPBS30, WPBS40 = willingness to pay per month to avoid 20%,
>., and 40%,reductions ,respectively, of monthly average water
use in household, for bathing and showering.

BATHSQ20, BATHSQ30, BATHSQ40 = a negative 20%, 30%, and 40%,
respectively, of monthly average water use in household, for
bathing and showering.

All other variables retain the same definitions as for equations (1),
(2), and (3).

The estimating equations for the willingness to pay to avoid reduc-
tions in toilet flushing water use are:


-30-









(7) WPTF1 = f(NUMRESPH, INCOME, TOILQ1, BELWSHRT, DECONSO),


(8) WPTF2 = f(NUMRESPH, INCOME, TOILQ2, BELJ'-.Hri-', DECON50),

(9) WPTF3 = f(NUMRESPH, INCOME, TOILQ3, BELWSHRT, DECON50),

where:

WPTF1, WTPF2, WPTF3 = monthly willingness to pay to avoid reducing
toilet flushing by an average of 1 time, 2 times, 3 times
per person per day.

TOILQ1, TOILQ2, TOILQ3 = figure representing a negative monthly
quantity of water corresponding to 1, 2, and 3 flushes of the
toilet per person per day.

All other variables retain the same definitions as in equations (1),
(2), and (3).

Willingness to Accept

The estimating equations to explain willingness to accept (WTA) com-
pensation to voluntarily accept specific reductions in total household
water use are:

(10) WAWAT = f(NUMRESPH, INCOME, HAVEWELL, KINSYS, BELWSHRT,
DECON50, AVGCONT),

where:

WAWAT = willingness of the consumer to accept compensation in the form
of a reduced monthly water bill to acquiesce in a specified re-
duction in .,,_ilhly total household water use,

AVGCONT = the quantity reduction (carrying a negative sign) in monthly


All other variables are as defined for equations (1), (2), and (3).

The estimating equation to explain WTA compensation to voluntarily
accept a specified reduction in water used for bathing/showering is:

(11) WABS = f(NUMRESPH, INCOME, BELWSHRT, DECON50, BATHST),

where:

WABS = willingness to accept compensation in the form of a reduced
monthly water bill, for a reduction in water used for bathiii'i/
showering,

BATHST = the quantity reduction (carrying a negative sign)in monthly
water use for bathing and showering to where the consumer's WTA
pertains.

All other variables are defined as in equations (1), (2), and (3).


-31-









The estimating equation to explain WTA compensation for reductions in
water used for toilet flushing is:

(12) WATF = f(NUMRESPH, INCOME, BELWSHRT, DECON50, TOILT),

where:

WATF = willingness of the consumer to accept compensation, in the form
of a reduced monthly water bill, for a specified reduction in
water used for toilet flushing,

TOILT =-the quantity redu-ction- carryingg a negative sign) in monthly
water used for toilet flushing to which the consumer's WTA
pertains.

All other variables are as defined in equations (1), (2), and (3).

The Estimation Procedure

The estimating equations for WTP and WTA postulate causal relationships
between the observed bids and each of several independent variables. The
procedure for estimating the parameters of those equations and testing for
statistical significance of the coefficients is multiple regression analysis.
A complete exposition of this procedure can be found in any standard econom-
etrics text book [for example, Wonnacott and Wonnacott (1970)].

The Questionnaire

Information on the hypothetical variables was obtained using primary
and secondary data collection methods. The WTP and WTA bids, and all data
excluding monthly water consumption data were collected through a question-
naire. Customer records of the water utilities of the two cities from which
samples were drawn, St. Petersburg and Orlando, were used to obtain the
water consumption data for the period November 1978 October 1980.10

The questionnaire was personally presented by trained enumerators.
It was suspected that the complexity of the questionnaire would have
created data problems if respondents were left solely to their own inter-
pretations of the questions.

It was anticipated that bids would differ between geographic areas
which experience, either currently or in the recent past, different water
availability conditions. To test this hypothesis two sampling groups were
chosen, St. Petersburg and Orlando. The former has a history of water
scarcity problems and the latter is located in a relatively water plentiful
area of the state. (See Appendix II for a complete discussion of the
sampling procedure).

The questionnaire (Appendix I) contained four sections designed to
provide data on:



10St. Petersburg Public Utilities in St. Petersburg and the Orlando
Utilities Commission in Orlando.









(1) the socioeconomic status of the household: income, household
size, education, etc.;

(2) the estimated amount of use of water using facilities and
appliances in the consumer's home;

(3) the actual valuation of water in specific activities (the WTP,
WTA bids)

(4) the beliefs and attitudes the consumer had concerning water
scarcity in his/her region, and concerning water conservation
-practices,

The third section of the questionnaire, discussed in a previous section,
provided the WTP and WTA bids.


-33-









CHAPTER V


RESULTS OF ANALYSIS

Four aspects of this research are of interest:

(1) the success of the questionnaire and the survey procedure in
terms of usable responses,

(2) the difference between WTP and WTA bids for the same sample
in the same water use category,

(3) the difference between corresponding mean bids for the St.
Petersburg sample and the Orlando sample,

(4) the results of the regression analysis.

These aspects will be discussed in turn in the following sections.

Response to the Questionnaire

For St. Petersburg, 182 households were contacted and 165 question-
naires were completed. Of these, 114 were usable'. Therefore, 63%
of the 182 household contacts produced usable questionnaires.

For Orlando, 130 households were contacted and 120 questionnaires
were completed. Of these, 62 were usable. Therefore, 48% of the 130
household contacts produced usable questionnaires.

The surveys rejected represented protest bidders. Protest bidding is
demonstrated through respondents' answers which reveal a failure to proper-
ly play the bidding game. Where answers indicated that a respondent did
not consider the tr it.l, process between paying (receiving) money and
acquiring (foregoing) water in a specific use when making a bid, the bids
were rejected. This generally took the form of zero bids with explana-
ti ons-of thtei-r-- unwtl -trnes-s to pay money or accept compensaon.

The following criteria were employed to determine protest bids:

(1) bidding all zeros: persons registering zeros for all thirty
WTP and WTA questions were considered unwilling to play the
bid game because it is highly unlikely that "true" responses
would be such. For St. Petersburg this eliminated 12 surveys,
or 7% of the original sample; in Orlando this eliminated 32
surveys, or 25% (see Table 1);

(2) bidding lowest amounts for WTP and highest amounts for WTA:
again it is unlikely that bids this consistent are "true."
Also it is blatantly contradictory to register the highest
value for WTA and the lowest value for WTP in the same use
category. This criteria eliminated 1 survey in St. Petersburg
and 7 surveys in Orlando.


-34-














Table I--Criteria for rejecting surveys


Criteria


Bidding all zeros

No consumption data

Zero bid for WTP, highest
bid for WTA

Not willing to reduce
or pay more

Would pay .m thing

Across the board compliance

Minimum users

Critical inconsistencies


Number (%) of surveys unuseable

St. Petersburg 182 Orlando 130


(7.0)

(10.0)


1 (0.5)


(25.0)

(22.0)


7 (5.0)


(4.4)

(0 5)

(2.0)

(3.0)

(0.5)


1 (0.8)


Not returned 17 (9.0)

Total used 114 (63.0)

Number of WTA only 9 (5.0)
WTP only 1 (0.5)

Number with both games 104 (57.0)

Number of bids:
WTP equations 1,092
WTA equations 1,150


-35-


(48.0)

(11.0)
(0.8)

(36.0)








(3) indicating an unwillingness to pay or reduce usage: persons
so bidding were not open to the tradeoff design of the bid game.
Their zero bids failed to consider the available choices. This
eliminated 8 surveys in St. Petersburg and none in Orlando.

(4) indicating a willingness to pay anything: again, the person is
not considering the tradeoff. Also, it is not realistic for
a person to be willing to pay "anything" since willingness is
a function of capability to pay. This eliminated 1 survey in
St. Petersburg and none in Orlando.

(5) indicati"Jg total compliance: persoQns who indicated across the
board compliance with water use reductions were rejected because
it is unrealistic to assume that the major reductions (e.g.,
50% of all water use) could be enacted in lieu of even the lowest
price. This eliminated 3 surveys in St. Petersburg and none
in Orlando.

(6) indicating minimum usage: persons who felt that their current
level of water consumption was at its minimum were not willing
to consider reductions for any price. This eliminated 6 surveys
in St. Petersburg and none in Orlando.

(7) critical inconsistencies: as will be explained below, some bids
were inconsistent and accepted in particular cases, however,
when all bids appearedto be consistently inconsistent with no
apparent justification, the entire survey was eliminated. This
eliminated I survey from each city.

Two other reasons for eliminating surveys existed which were unrelated
to protest bidding. First, 17 surveys were not completed and returned
from the St. Petersburg enumerators. Second, water use consumption data
was n.ivailable for a percentage of each -.mnple. In St. Petersburg 12
surveys were from respondents who were not listed as customers of the St.
Petersburg Public Utilities Company. In Orlando where utility data col-
-- tecti-o was-it-lts-si-b-l-bebecaause--custoiTe-rs' -names were- not--prov-ided, re-s porn-
dents estimates of water use had to be utilized. For 28 surveys this esti-
mate was not provided. Thus, for St. Petersburg this second criterion elim-
inated 10% of the original sample and for Orlando it eliminated 22% of the
original sample.

A complete set of thirty bids were usable from only a tiny portion
of all respondents. Individual bids (from usable surveys) were eliminated
for the following reasons:

(1) bidding all zeros on one game eliminated the bids from that game.
In St. Petersburg 9 respondents played only a WTA game and one
played only a WTP game. In Orlando 14 respondents played only a
WTA game and one played only a 1.TP game;

(2) bids were eliminated if they were irrelevant to the household's
water use level, e.g., if the household did 2 cycles of laundry
per week, their bid for 3 cycles was considered irrelevant (for
the most part, respondents replied "does not apply" to these
(!,. _'Lions); and









(3) bids were eliminated if the respondent was indecisive or if '
the individual bid was otherwise rendered unusable by the
aforementioned criteria which eliminated whole surveys.

The decision toeliminate bids was difficult because the initial
survey design did not account for an exploration of the zero bid or the
inconsistent bid. Yet neither could be dismissed indiscriminately be-
cause valid reasons could exist for both. Some respondents bid posi-
tive amounts for retaining lower levels of water use and zero for higher
levels. It would appear to be inconsistent to, say, bid $3.00 to retain
10% of one's total water use and zero to retain 50%. Yet some respondents
stated-th-at-they would seek- an alternative source of s-upply--if h-gh-pr-
centages of their water use were threatened with price increases. Others
stated that they could not afford the increases at higher levels. Both
reasons were considered valid zero bids by the theoretical definition of
value measurement that this study adapts.

Zero bids at lower levels of use were accepted because they usually
indicated a willingness to comply with a reduction requirement only at that
level. Unlike the surveys indicating only total compliance, this explana-
tion is realistic and reasonable. Bids at the other extreme were too many
standard deviations from the mean bids calculated for each equation and
constituted outliers. As bids tended to be uniformly conservative this
criteria eliminated very few bids. As expected most of the extremely
high bids were in the WTA game.

In general, bids which could be explained and subsequently rendered
reasonable were included in the sample. As stated the problem existed
primarily with zero bids. Fortunately, with the St. Petersburg sample,
bids were well explained and this aided in the process of determining
whether or not a bid was usable. With the Orlando sample bids were
not as well explained but proportionately fewer zero or inconsistent
bids occurred.

Comparing WTP and WTA Within
Use Categories

St. Petersburg Sample

Mean bids for each water use category increased as the amount of
reduction avoided (or compensated for) increased (see Table 2). These
results are consistent with the concept of a bid curve passing from
the southwest quadrant through the origin into the northeast quadrant
of Figures 2 and 3 (Chapter II).

i0i thin the St. Petersburg sample the mean WTA bids for the total
household water (w) category averaged 220% of their corresponding WTP
bids. For the bath/shower (BS) category the mean IWTA bids averaged
196% of the mean WTP bids. For the toilet flushing category the mean
WTA bids averaged 113% of the corresponding mean WTP bids. These results
are consistent with a hypothesis that WTA will normally be greater than
,rTP.


-37-











Table 2.--Iean bids


Number o Percentage bidding
Use observations Current price Mean bid current price

WTP ITA T TP WTA WTP WTA
St. Petersburg


W

w
U
U


Orlando


1.90
3.84
5.76

1.56
2.34
3.12

1.98
5.94
9.90



2.16
4.32
6.36

1.75
2.63
3.50

2.22
6.66
11.10


2.80
5.46
8.42

1.76
2.87
4.29

2.41
6.97
11.81



2.00
4.30
6.71

2.16
3.06
4.34

2.52
6.58
10.66


3.09
5.80
10.38

3.15
5.32
8.97

5.16
15.10
26.53



4.34
9.07
15.31

5.95
8.91
13.60

8.23
21.67
46.10


BS = Bathing/sh wearing; W


TF = Toilet flushing;


= Total household water use








The Orlando Sample

Mean bids in the Orlando sample also increased as the amount of re-
duction avoided (or compensated for) increased (Figure 2).

Within the Orlando sample, mean WTA bids for the total water use
ca'.ei,:-j.- averaged 362% of the corresponding mean WTP bids. For the bath
and shower category, the mean WTA bids averaged 293% of: the corresponding
mean WTP bids. For the toilet flushing category the mean WTA bids
averaged 219% of the mean WTP bids.

Comparing WTP and WTA
Between the Two Samples

It was hypothesized that the bids, for corresponding categories,
would be higher in St. Petersburg because those respondents are familiar
with water availability problems (Appendix II). However the sampled
Orlando residents had a much higher mean income than the St. Petersburg
group and this would tend to offset the difference in bids between the
groups.

With respect to the total water use category (w), WTP, mean bids
from St. Petersburg were roughly the same as for Orlando. However, for
WTA, the Orlando mean bids averaged 15;" of the corresponding St.
Petersburg mean bids (Table 2).

With respect to the bathing and showering category, WTP mean bids
from Orlando averaged 110% of those from St. Petersburg. But WTA mean
bids from Orlando were about 169% of those from St. Petersburg.

With respect to the toilet flushing category of use, WTP bids
from St. Petersburg averaged about 131% of those from Orlando. However,
WTA mean bids from Orlando averaged about 147% of those from St. Petersburg.

In general, then, WTP bids were roughly comparable between the two
samples--,b-ut-iTA--bids-we re -cns-i -ste-t!4y-muc--h4-ihe r- m--t e -r-r 1----o
s ampl e.

Results of Regression Analysis

4illi nqness to Pay Equations

For the total water use category, the WTP dependent variables are
WPWATIO, 'IPIPT30, and WPWAT50; for bath/showering use, WPBS20, WPBS30
and WPBS40; and for toilet flushing use, WPTF1, WPTF2, WPTF3. Each WTP
dependent variable represents the amount the consumer is willing to pay
through his monthly water bill to maintain current levels of household
water for specific water-using activities, rather than experience a
specified reduction in water use.

The independent variables in the WTP equations are NUMRESPH, INCOME,
HAVEWELL, BELWSHRT, DEOI'.d, and the quantity variables, AVGCONIO, AVGCON30,
A'.Ni!'.., TOILQ1, TOILQ2, TOILQ3, BATHSQ20, BATHSQ30, BATHSQ40, each of
which denotes the quantity reduction to which the bid pertains. The
',..thesized relationship between these variables is discussed in Chapter








Four. Because of a high correlation between the variable, NUMRESPH,
symbolizing the number of persons residing in a household, and the quantity
variables, equations for both possibilities were presented.

Explanation of Results of WTP Regressions

The results of the regression analysis for WTP are summarized in Tables
3 and 4. Income was expected to be highly significant because it would
appear intuitively to have a substantial influence on a consumer's
willingness to pay. This variable was significant, in fact, in five
equations in the St. Petersburg sample and in nine equations in the
Orlando sample and in neither case did income play_ an imortantrole in
the WTP equations for total household water use. In St. Petersburg,
income was significant in both WPBS20 and both WPBS40 equations. In
Orlando income was significant in one WPTF1 equation, in both WTPF2
equations and in all WPBS equations. It appeared with the anticipated
sign (positive) in all but two equations. The differences in the two
samples may be attributed to the differences in mean income levels
between the two city samples. In the St. Petersburg sample the mean
income approaches $10,000, while in the Orlando sample the mean income
level approaches $20,000.

NUMRESPH was also expected to be a highly significant variable. In
St. Petersburg it appeared as significant in all equations in which it
appears except WPBS40. In Orlando it did not appear as significant in
any equation. The St. Petersburg sample had a higher mean household size
than in Orlando and this may account for the difference, however, the
differential is not large (2.5 in Orlando and 2.9 in St. Petersburg).
NUMRESPH appeared with its expected sign (positive) in all St. Petersburg
equations and in all but two, one WPWAT30 and in one WPWAT50, in the
Orlando sample.

Originally HAVEWELL and KINDSYS were included in the total water
use equations. KINDSYS was eliminated because it was not significant
at any test level in any of the equations. HAVEWELL was retained but was
significant in only o equation, WPWAT30,-i- Pthe Or ando sample, where
it appeared with a positive sign (its hypothesized sign was negative).
In four out of six Orlando equations the HAVEWELL sign was positive.
In the St. Petersburg sample HAVEWELL appeared consistently with the
hypothesized sign but was not significant in any equation.

This variable was difficult to interpret and this may be the reason
for its performance. Since most sampled households in St. Petersburg
had a well its effect on the willingness to pay should be negative
since this implies the existence of a substitute water system. However,
this effect is apparently not substantial for the sample. On the other
hand, the Orlando households, for the most part did not have a well and
that fact had a positive effect on the willingness to pay for water. As
with the St. Petersburg sample the presence of a well did not appear to
have a substantial effect on the WTP.

The quantity variables were expected to have a negative effect on the
WTP because they were entered into the equation as negative values, re-
presenting quantity reductions. (".-tually the amount of water a household
used was h .',thesized to be a positive influence on the willingness to
pay). They appeared in all equations (both samples) with the hypothesized


-40-













Table 3.--iTP rea session results for St. Petersburg


WTP
model

WPHAT10 (a)

WPWAT10 (b)

WIPAT30 (a)

lPUWAT30 (b)

WPWAT50 (a)

WPWAT50 (b)

HPTF1 (a)

WPTF1 (b)

'PTF2 (a)

WPTF2 (b)

WPTF3 (a)

WPTF3 (b)

WPBS20 (a)

WPBS20 (b)


WPBS30 (a) .

WPBS30 (b)

WPBS40 (a)

WPBS40 (b)


RESPH INCOME HAVEWELL BELUSHRT

180 .122 .299 --064
.185) (.164) (.447) ( -607)
.119 -.305 -.043
(.163) (.442) (.603)
333 .377 -.689 -.049
.247) (.219)** (.595) (.808)
.380 -.723 -.0004
(.218) (.588) (.804)
20 .537 -.107 2.80
(.633)* (.562) (1.53) (2.07)*
.666 -1.52 3.16
(.568) (1.53) (2.09)*
.624 .039 -.008
.198)*** (.179) (.666)
.016 -.022
(.179) (.663)
1.37 .202 1.23
(.353).** (.329) (1.23)
.247 -1.21
(.329) (1.22)
1.61 .234 2.30
(.670)*** (.610) (2.25)
.179 2.27
(.610) (2.25)
.152 .161 -.408
( 114)* (.103)* (.385)
.166 -.390
( .104)* (.386)

.395 .215 .553
(.195) (.177) (.659)
.206 .545
(.176) (.657)
.297 .456 .766
(.232) (.210)** (.784)
.430 .712
(.208)** (.776)


AVGCON10 AVGCON30 AVGCON50


TOILQI TOILQ2 TOILQ3 BATHSQ20 BATHSQ30 BATHSQ40


-.867
( .736)


DECON50

-.252
(.652)
-.203
(.654)
-.362
(.868)
-.306
(.873)
-3.13
(2.23)*
-3.39
(2.27)
.755
(.714)
+.754
(.709)
.035
(1 .31)
.020
(1 .31)
-.812
(2.41)
-.823
(2.41)
-.319
(.412)
.275
(.411)

-.597
(.706)
-.685
(.699)
-.819
(.8"39)
-.860
(.827)


-.416
(.511)


-.003
(.001)***


-.004
(.0009)***


-.003
(.001)**


-.0004
(.004)


-.001
(.0005)**


-.0008
(..0004)*


Significant at .10 level.
Significant at 05 level.

Significant at .01 level .

(a)odel with NiUMRESPH.

(b)l del with quantity variable.


-.464
(.327)













Table 4.--iOP regression results for Orlando


WP AT1O (a)

!UPWT10 (b)

IPF AT30 (a)

WPIAT30 (b)

WPW1AT50 (a)

WPWAT50 (b)

WPTF1 (a)

UPTF1 (b)

-1WPTF2 (a)

UPTF2 (b)

NPTF3 (a)

WPTF3 (b)

UPBS20 (a)

I:PBS20 (b)

3PBS30 (a)

IJPBS30 (b)

UIPBS40 (a)

W.PBS40 (b)


HAVEWELL BELWSHRT


.283
(.793)
.308
(.803)
1 .57
(1 .37)
1.93)
(1 .37)*
1 .34
(2.68)
2.15
(2.68)


!U'RRESPH INCOME

.061 .076
.312) (.166)
.086
(.154)
-.167 .260
(.542) (.287)
.194
(.263)
-.697 .160
(1 06) (.560)
-.049
(.514)
.077 .232
(.249) (.132)
.231
(.133)**
-.100 .456
(.445) (.235)**
.428
(.239)**
-.222 .245
(.762) (.403)
.176
(.409)
.122 .196
(.249) (.132)*
.199
(.132)
.187 .328
(.326) (.173)**
.338
(.173)
.027 .386
(.484) (.256)**
.362
(.256)**


DECON 0 AVGCOlN10 AVGCON30 AVGCO40 TUILQI


.161
(.790)
.230
(.827)
.602
(1 .37)
-.38
(1.41)
-1 .77
(2.67)
.968
(2.76)
.784
(.625)
.779
(.625)
.550
(1.12)
.528
(1.12)
-1 .28
(1.92)
1 .31
(1.92)

.671
(.626)
.679
(.626)
1.06
(.819*
1 .07
(.820)
1.59
(1 .22)*
1:59
(1.21)


TOILQ2


TOIL03 BATIHSQ20 BATHS030 BATHSQ40


- .'I09
( .428)


-.333
(.243)0


-.418
(.285)


-.791
(.84S)
-.792
(.834
-3.68
(1 .47)***
-3.14
(1 .42)**
-4.21
(2.87)*
-2.82
(2.781
-1 .13
(.676)*
-1 .13
(.674)
-1 .34
(1 .21|)
-1 .24)
(1.21)
-.75!3
(2.071)
-.51
(2.07)

-1.151
( .67|7)**
-1.19
(.665)
-1.81
(.887)**
-1.87
(. 8710)
-2.27
(1 .32)**
-2.21!
(1 1-


-.00008
(.001)


-.0002
(.001)


-.0005
(.001)


.0004
(.001)


.0003
( 01)


Significant at .10 level.

Significant at .05 level.

Significant at .01 level.
(a) ode, l with N(UMRESPH.

(13,,el with quantity variable.


.0004
( .001)








sign and were significant in all St. Petersburg equations in which they
appeared except for WPWATTO, WPWAT50, and WPBS20. In Orlando they were
significant in WPWAT30 and WPWAT50 only.

The conservation variables performed differently than expected.
BELWSHRT, the variable which registered belief in a water shortage was
anticipated to be negative and appeared so in seven equations in both
the St. Petersburg and Orlando samples. In eleven equations BELWS'il:T
was positive and this included the two cases in the St. Petersburg
sample and the four cases in the Orlando sample in which the variable
was significant. In Orlando BELWSHRT was significant in both WPBS30
and in both WPBS40 equations. In St. Petersburg, BELWSIIRT was signi-
H i,.,,it i Lr..I.. i- F'l J,",.T ,,i : t..i io ns .

DECON50 appeared as negative (its hypothesized sign) in twelve
cases and as positive in 6 cases in the St. Petersburg sample. It was
significant in the WPWAT50 equation where BELWSHRT was also significant.
In the Orlando sample DECONO50 had the expected sign in all equations
except one, WPTF2, and was significant in eleven equations, both WPWAT30
equations, WPWAT50O, both WPTF1 equations, and all WPBS equations.

Willingness to Accept Equations

In the WTA group of equations the dependent variables are WAWAT, WATF,
and WABS which were creating from combining the WTA data sets of the total
water WTA bid group, WAI.ATIO, WAWAT30, WAWAT50; the toilet flushing WTA
bid group, WATFI, WATF2, WATF3; and the bath/showering WTA bid group,
WABS20, WABS30, WABS40.

The independent variables are the same as in the WTP estimation with
the exception of the quantity variables. The quantity variables, AVGCONT,
TOILT and BATHST were created from the combination of their respective
separate WTP data sets.

Description of Results of WTA Regression

The results of regression analysis for WTA equations are summarized
in Tables 5 and 6. In the Orlando sample, income has the expected sign
(positive) in all but one of the WATF and WJABS equations. It was sig-
nificant in only one I..I..T equation for the Orlando sample. In the St.
Petersburg sample income was positive and highly significant in all
equations.

I.MRESPH appeared in three equations for Orlando and each time with
a negative sign (its hypothesized sign was positive). It was not sig-
nificant in any equation which is similar to the results for this
variable in the WTP equations for Orlando.

For the St. Petersburg sample, NUMRESPH was significant in two of the
three equations in which it appeared and it was positive, as expected,
in all three cases.

The quantity variables in the Orlando sample had the expected sign
(negative) and significancein the three Orlando equations in which they
appeared. Likewise they were negative and significant for all the rele-
vant equations in the St.. Petersburg sample.


-43-










Table 5.--WTA regression results for St. Petersburg


NUMRESPH INCOME


HAVEWE'L


BELWSHRT DECON50 AVGCONT


1.14
(.819)


1.70
(.732)***,


'- .T (b) 1 .4b
(.663)***


.679
(. 281)***


.143
(.260)


-1 .99
(1.95)

-.723
(1 .77)


.640
(.257)***

.573
(.245)**

.601
(.238)***


.495
(.233)


2.74
(2.69)

2.25
(2.46)

1.23
(.950)

1.03
(.909)

1.04
(.880)

.855
(.860)


.035
(2.96)

2.20
(2.72)

.459
(1 .04)


-5.06
(.662)***


-.003
(.0005):***


.113
(.964)

.278
(.942)


.038


.184


.053


.127


.030


-.002
(.0006)***


*Significant at .
Significant at .10 level.
**
Significant at .05 level.
Significant at .01 level.
(a)Model with NI';,rESPH.

(b)Model with quantity variable.


WAWAT (a)


TOILT


BATHST


WATF (a)


WATF (b)


: (a)


WABS (b)


i -











Table 6.--UTA regression results for Orlando


NUMRESPH I COME


HAVEWELL BELWSHRT DECON50 AVGCONT


12.04
(5.97)


17.66
(5.97)***

5.50
(1.98)***

5.27
(1 .93)***

3.12
1 .61)**

3.09
(1 .58)**


-10.53
(6.42)**

-5.48
(5.89)

-3.43
(2.15)*

-2.09
(2,04)

-2.65
(1 .75)*

-1.74
(1 .66)


-3.30
(.706)***


-.004
(.001)***


WAWAT (a)


'-T (b)


WATF (a)


WATF (b)


E-AB (a)


WABS (b)


*
Significant at


.10 level.


**
Significant at .05 level.
***
Significant at .01 level.
(a)Model with N .i!'[,PH.

(b)Model with quantity variable.


WTA
model


-3.44
(5.99)

.065
(5.69)


1 .67
(1 .25)*

1 .07
(1 .09)

.117
(.417)

-.293
.393)

.311
(.339)

-.025
(.324)


TOILT


-1 .34
(2.36)


-.274
(.789)


-.166
(.642)


BATHST


-.004
(.002)***


.035


.138


.053


.099


.030


.072


I









The HAVEWELL variable in the Orlando sample was not significant in
either equation in which it appeared and it appeared as a negative influ-
ence in one WAWAT equation and as a positive influence in the other
equation.

fn the St. Petersburg sample HAVEWELL was not significant in either
of the total water WTA equations in which it appeared, however, it did
appear with the expected sign (negative) in both cases.

The conservation variables for the WTA equations for the Orlando
sample performed very much as they did for the WTP equations. BELWSHRT
-was consistently a. positive variable and it was significant in all
equations. DECON50 was consistently a negative variable and was signi-
ficant in three equations, one WAWAT, WATF and WABS equation. In each of
these equations NUlIN'i',1H, rather than the corresponding quantity variable,
was entered into the equation.

In the St. Petersburg sample BELWSHRT and DECON50 had positive signs
in all equations and wereconsistently insignificant throughout.








CHAPTER VI


SUIh1A.FY AND DISCUSSION

This study adapted a non-market valuation technique and applied it
to the measurement of consumers' valuations of water in residential uses.
Representative samples of single-family residential water customers were
drawn from two major cities in Florida, and data for the study was ac-
quired through the use of personal interviews following a specially de-
signed questionnaire. The iterative questioning procedure of the survey
was structured to elicit consumers' maximum willingness to pay to avoid
specified reductions in residential water use, and, in addition, to
elicit the minimum compensation necessary to induce consumers to volun-
tarily accept specified reductions in residential water use. Multiple
regression analysis was used to test hypotheses concerning the role of
selected independent variables as determinants of consumers' valuations
of water in residential uses.

The results must be considered preliminary, since the initial analysis
did not exhaust all possibilities for specifying the regression equations.
Several conclusions can be drawn from the study:

(1) response of consumers to the contingent market, iterative
bidding, context of the questionnaire indicated that a
substantial percentage of the respondents understood the
purpose of the questionnaire and attempted to honestly
assess their willingness to pay and willingness to accept,

(2) mean bids for both samples, both measures (WTP and WTA),
and all water use categories consistently demonstrated
higher bids for greater contingent reductions in water
use--a pattern which would produce bid curves passing from
the southeast quadrant, through the origin into the north-
east quadrant of a graph depicting a total value (or bid)
curve,

(3) willingness to accept (WTA) bids consistently exceeded
willingness to pay (WTP) bids for corresponding use
reduction contingencies, and

(4) regression results suggest that a substantial portion of
the hypothesized functional relationships between observed
bids and selected explanatory variables were, in fact,
statistically significant.

Additional investigation is needed in several areas:

(1) there is a need to further explore the ability of consumers
to formulate valuations of water in household uses when
such considerations are being made for the first time.
Can "preparation" be provided without biasing consumers
perceptions?


-47-








(2) The potential for other forms of bias must be carefully
examined. While starting point bias was not expected
to be a factor, it may well have existed. An alternated
high and low starting point for bids within each sample
could be applied to test for the presence of starting
point bias.

(3) It was hypothesized that the variables which influence
the willingness to pay for water would be the same
as those which influence the amount of use. The theoret-
ical underpinnings of these hypotheses need to be
developed with greater care.

(4) Empirical estimation of a bid function must require a zero
intercept in order that the estimated function retain the
properties inherent in its definition.



































APPENDIX I

THE QUESTIONNAIRE










Part One: Socioecomonic Data


How many persons presently live in this household?____

How many members of your household, including yourself, are in each
age group?


0-10 years 11-20 years __

--- -60 years- over hO years

Do you own or rent your home? own rent

If own, what is the market value of your property?_
what is your rent/month?

How long have you lived in the St. Petersburg area?

How old is your home?

Place of previous residence? cit

Do you live in Florida year-round?


21-40 years





If rent,


___years



y/ state


yes

no

If no, what months do you spend in Florida?

from to

-9~ s-.tre head of the uus-eliold- employed, retired or unemployed?

unemployed

retired

unemployed

If employed, what is his/her occupation?

10. What are the occupations of other working members of the household?



11. What is the highest grade of school completed by the head of the household?

less than high school
high school
some college or technical
B. A. or B. S.
Masters or Ph.D.
_RF_








12. What is the number of overnight guests in your home per year?

0-10

S11-20

21-30

over 30

Their average length of stay is


13.


14.


15.


I --day

2-7 days

1-2 weeks

2-4 weeks

over 1 month

low many of the following water using appliances does your household have?

bathtubs

--____toilets

washing machines

dishwashers

showers (with and without bathtubs)

sinks

garbage disposals

other (hot tubs, jacuzzi, etc.)

Do you have a swimming pool?

yes

_____no

Do you use bottled water?

yes

no

If yes, how many gallons per month?

__ gall ons


-51-









16. Do you have a home water softener?

yes

no

17. Do you have a septic tank?

yes

no





yes

yes, but only for lawn sprinkling

19. What kind of system do you use to water your lawn or garden?

none

hose and sprinkler(s)

automatic sprinkler system

20. Estimate the size of your property:

Less than 1/10 acre

1/10 acre

1/5 acre

1/4 acre

1/2 acre

3/4 acre

1 acre or more

21. How often do you wash your car(s) at home?

less than once a month

1 to 3 times a month

more than 3 times a month

22. What is your total monthly water bill in dollars:

for water only

for sewer/wastewater only

-52-









23. What is an estimate of how much water your household uses each month?

gallons

do not know

24. Indicate in which period your water bill is the highest:

January March

April June

______ July September

October December

25. What is the combined income of your household?

less than 5,000.

5,000 to 9,999.

S10,000 to 14,999.

15,000 to 19,999.

20,000 to 24,999.

25,000 to 29,999.

30,000 to 34,999.

35,000 to 39,999.

over 40,000.







5


The following tables provide a water use estimate for your household. Please
indicate the frequency for each use category (how many times it is done) in
column 2. Column 3 provides approximate amounts of water used by each item.


Item


2
Use per week


3
Gallons per item


--I -I I-


a. Automatic
dishwasher 12

b. Hand dishwashing 8

c. Clothes washing
(cycles or loads) 50

d. Hours of lawn watering
(see below) 500

e. Garbage disposal
(2 minute use period) 6

Subtotal

Use per day

f. Shower 33

g. Bath 30

Subtotal

No. of persons Gallons per day

h. Toilet flushing 32

i. Cooking and drinking 3

Subtotal

Total


Please indicate the months within which you water your lawn at least
once a week:


-54-


4
Total









Part Three
Iterative Bid

(Note to Surveyors:
THIS SCENARIO IS HYPOTHETICAL! NO SUCH REQUIREMENTS ARE PENDING FOR
CotlSiU ERS).

Supplying water in the future may require higher than present costs.
Imagine that a water planning agency has to pay higher costs to provide
enough water, and is interested in finding out their consumers' willingness
to pay these higher costs for current use levels.

Please indicate your willingness to pay through changes in your monthly
w.-,te L'-, hil11 for water in specific uses. Imagine that requirements are
set for water use reductions but you may avoid compliance if you are willing
to pay to do so. (The first amount presented is the current price for the
amount of water given).


A. Lawn watering

1. If there was a requirement to reduce lawn watering by 1/2 hour a
week, how much additional money would you be willing to pay to
avoid complying with this requirement?

$1.98 $3 $4 $6 $8 $12 $16 $24
other

2. To avoid reducing lawn watering by 1 hour?

$3.96 $6 $8 $12 $16 $24 $32 $48
other

3. To avoid reducing lawn watering by 3 hours?

$11.88 $16 $24 $32 $48 $72 ,'.: $95 $143
; .. other

B. Toilet Flushing

1. If there was a requirement to reduce toilet flushing by 1 time per
person per day, how much additional money would you be willing to
pay to avoid complying with this requirement?

$1.90 $3 $4 $6 $8 $11 $15 $23
other

2. To avoid reducing toilet flushing by 2 times per person?

#3.84 $6 $8 $11 $15 $23 $31 $46
other


-55-








3. To avoid reducing toilet flushing by 3 times per person per day?

$5.76 $9 $12 $18 $23 $35 $46 $70
other
C. Bath/Shower

1. If there was a requirement to reduce bath/shower usage in the house-
hold by 20%, how much additional money would you be willing to pay to
avoid complying with this requirement?

--- a-20% decrease-i-s-equi-val ent- to-a person-taking- six-showers -a -week-
instead of seven, or reducing a 10 minute shower to 8 minutes).

$1.56 $3 $5 $6 $9 $12 $19
other
2. To avoid reducing bath/shower usage by 30%?

$2.34 $4 $5 $7 $9 $14 $19 $28
other
3. To avoid reducing bath/shower usage by 40%?

#3.12 $5 $6 $9 $12 $19 $25 $37
other

D. Clothes Washings

1. If there was a requirement to reduce clothes laundry use by 2 cycles
or loads per week, how much additional money would you be willing to
pay to avoid complying with this requirement?

$.79 $2 $3 $5 $6 $9
other

2. To avoid reducing laundry use by 3 cycles or loads per week?

$1.19 $2 $4 $5 $8 $10 $14
other

3. To avoid reducing laundry use by 4 cycles or loads per week?

$1.58 $3 $4 $6 $10 $13 $19
other


-56-









E. What are you willing to pay to avoid a reduction of 10%
usage for your household?


of all water


$1.98 $3 $4 $5 $8 $12 $16


other


2. To avoid a reduction of 30% of all water usage for your household?


$5.94 $9 $12 $18


$24 $36 $48


$96


other


3. To avoid-a reduction of i-0' of all ,.iter :.,'j f.ir 01or hou-i.ehold?


$9.90 $15 $20 $30


$40 $60 $80


$120 $160


If you had decided to comply with requirements of water
what amounts would you consider necessary to compensate you
veniences which you incur when you make these reductions?


use reduction
for the incon-


Please indicate the minimum amount of money which you would accept,
through reductions in your monthly water bill, for specific water uses.
(The first amount presented is the current price for the amount of water
given).

A. Toilet Flushing

1. If you were required to reduce toilet flushing 1 time per person per
day, what amount would fully compensate you for your loss?


$1.90 $3 $4 $7 $9 $13 $17


2. For a loss of 2 times per person per day?

$3.84 $6 $8 $11 $15 $23 $31


3. For a loss of 3 times per person per day?


other


other


$5.76 $9 $12 $18


$23 $35 $46


other


B. Lawn Watering


1. If you were required to reduce lawn watering by 1
amount would fully compensate you for your loss?

$1.98 $3 $4 $6 $8 $12 $16


/2 hour a week, what


other


2. For a loss of I hour a week?


$3.96 $6 $8 $12 $1:6 $24 $32


other


$46


other








B. (continued)

3. For a loss of 3 hours a week?


$11.88 $16


$24 $32 $48 $72 $95 $143


C. Bath/Shower

1. If you were required to reduce bath/shower usage in the household by
20%, what amount would fully compensate you for your loss?

(a 20% decrease is equivalent to a person taking six showers a week
instead of seven; or reducing a 10 minute shower to 8 minutes).


$1.56 $3 $5 $6 $9 $12 $19


other


2. For a loss of 30% of your current bath/shower use?

$2.34 $4 $5 $7 $9 $14 $19


3. For a loss of 40% of your current bath/shower use?


$3.12 $5 $6 $9 $12 $19 $25


$37


other


other


D. Clothes Washings

1. If you were required to reduce your laundry use to 2 cycles or loads
a week, what amount would compensate you for your loss?


$.79


$2 $3 $4 $5 $6 $8 $10


other


2. For a loss of 3 cycles or loads a week?

$1.19 $2 $4 $5 $8 $10 $15


3. For a loss of 4 cycles or loads a week?

$1.53 $3 $4 $6 $10 $13 $19


other


other


E. If you were required to reduce all of your household water usage by 10%
what amount would fully compensate you for your loss?


$1.98 $3 $4 $5 $8 $12 $16


$36


other


other









E. (continued)

2. For a loss of 30% of all your household water usage?

$5.94 $9 $12 $18 $24 $36 $48 $72 $96


3. For a loss of 50% of all your household water usage?

$9.90 $15 $20 $30 $40 $60 $80 $120 $160


other


-59-


other







Part four: CAS
1. During the recent years, the media has been reporting the existence
of water shortages in many areas of the United States. These shortages
place pressure on urban water supply systems which provide water for
residential home use.

Do you believe a water shortage exists in the St. Petersburg area?

No

Yes

-f- yes,- -how serious- wutld yoiu s-ay the water-sho-rtage- is in St. P&ters-burg
at present?

very serious
fairly serious
not serious
do not know

2. Whether or not you believe that there currently is a water shortage in
your area, do you believe that there will be such a problem in the
future? Please indicate by what year you think a water shortage may
become a problem or continue to be a problem?

1981
1935
2000 or beyond
never

3. If you believe a water shortage exists or will exist in St. Petersburg,
to what extent do you believe each of the following to be a cause of
the problem?
Great Some Not at
(a) Natural causes: Extent Extent all

1) lakes, rivers drying up
2) lack of rainfall
3) depleting groundwater
sources (includes salt
water intrusion).

(b) Man made causes:

1) growth of population in
St. Petersburg area
2) growth of population
in other areas
3) increased water usage
per person due to in-
creased use of water-
using home appliances
(i.e. hot tubs)








Great Some Not at
3. (continued) Extent Extent all

4) increased water
due to wasteful
practices
5) the heavy use of
water for com-
mercial (industry,
mining)
6) the heavy use of
water for agricul-
-tural-at vi-ty ___ -
7) pollution of
water supplies

(c) Institutional causes:

1) Water utilities have
not taken the neces-
sary steps to provide
for enough water for
residential use.

4. In some areas throughout the country the public has been asked to practice
water conservation in order that water demand may be reduced. Water con-
servation can be achieved in numerous ways. Following is a list of
possible conservation practices. Please indicate to what extent you
believe each practice may result in conserving water:

Great Some Not at
Extent Extent all

a) Filling the bathtub
only one-fourth full

b) Turning off the water
while brushing your
teeth and shaving

c) Taking showers in a
shorter amount of
time

d) Using the dishwasher
only when it is full

e) Cutting back on lawn
sprinkling time

f) Cutting back on times
the toilet is flushed

g) Capturing and reusing
shower water for non
drinkable use (e.g.,
plant watering)


-61-







Great Some Not at 1
Extent Extent all
4. (continued)

h) Placing a brick in
the toilet tank

i) Use of the following Are not
water saving devices: Great Some Not at familiar
Extent Extent all with

1) pressurized showerheads
2) suds-saving washing
machine
3) shallow trap water
saving toilets
4) chemical toilets
5) low volume dishwasher
6) dual flush toilet tank
7) washerless faucets

5. Which of the following conservation practices are you actually using
to reduce the amount of water your household uses? Please indicate
whether you are making large efforts, medium efforts, small efforts, or
no effort in each of the areas listed below:

Large Medium Small No Does Not
Effort Effort Effort Effort Apply
a) turning off the water
while brusing your
teeth and shaving
b) using the dishwasher
only when it is full
c) taking showers in a
shorter amount of
t i me
d) cutting back on lawn
sprinkling time
e) filling the bathtub
only one-fourth full
f) cutting back on times
the toilet is flushed
g) capturing and reusing
shower water for non-
drinkable use (e.g.,
plant watering)
h) placing a brick in the
toilet tank
i) use of the following
water saving devices: Yes No
1) pressurized showerheads
2) suds-saving washing
machi nes
3) shallow trap water
saving toilet
4) chemical toilet
5) low volume dishwasher
G) dual flush toilet tank
7) washerless faucets






14
6. In some areas the price of water (per 1,000 gallons or per cubic feet)
has been increased with the belief that water usage will decrease as a
result. In your area has the price of water/sewer recently (last few
years) been increased?
No
Yes

If yes, has this price increase led to a reduction of water usage in your
household?
No
Yes

7. Do you anticipate an increase in the price of water/sewer for your area
in the near future (within the year)?
No
____ Yes

8. Do you think that your household would decrease its water usage if the
price of water/sewer was raised by 50%?
No
Yes

If yes, by approximately what percent would your household decrease its
water usage:
0-10%
S11-20%
21-30%
over 30%

9. Do you think that your household would decrease its water usage if the
price of water was raised by 100%?
No
Yes

If yes, by approximately what percent would your household decrease water
usage?
0-10%
S11-20%
21-30%
over 30%

10. If you would decrease your water usage would you do so with the instal-
lation of water saving devices?
_Yes, I would consider immediate installment.
Yes, but only when I needed a new toilet, sink, etc.
No.


-63-









APPENDIX II


THE SAMPLE

This study required the drawing of two separate samples, one from the
city of Orlando, Florida and one from the city of St. Petersburg, Florida.
The qualifications for the population were: a) it was confined to the
geographic areas under the jurisdiction of the water utility which served
the largest number of customers. The reasons for this were the need to
deal with one rate and the need to represent as large an area as possible
within the city itself. The two water companies used, Orlando Utilities
fommi-s-s-ior -{h-)--in- Ortarrdo- aTd-St. -PetLers&rrg- PubtTi cUttties i-r S- -- --
Petersburg, both had one residential rate structure which extended to the
city limits (see Table 7); b) it was confined to single family dwellings
with 5/8 inch water meters. This was because apartment residents often do
not pay their own water bills and because meters of different size follow
differing rate structures. Although the construction of multi-unit
dwellings is increasing, the majority of housing structures in both
Orlando and St. Petersburg are single-family units.

This appendix contains two sections. The first section provides an
introduction to the communities selected, including,a brief description of
their water supply systems, the status of their water reserve capabilities,
and a discussion of the demographic characteristics which influence the
demand for residential water in both communities. In the second section of
this appendix, a detailed review of the sampling procedure is presented.

Community Characteristics

St. Petersburg (Pinellas County)

Demographic character stics

Pinellas county is the most densely populated of all Florida counties.
Current density in developed resideltia-are-as of Pin e1as cou nty i-s---1-4
persons per acre or 5.83 units per residential acre (Board of County
Commissioners, 1973). The density for the total county is 3.6 persons per
acre with a unit density of 1.5 units per acre. If the county continues
to grow at the-rate set since the Census of 1970, the estimated population
in year 2000 would be over 1.8 million.

From 1940 to 1979 the population increased at the annual average rate
of 72% (Pinellas Planning Council, 1978). St. Petersburg accounted for
about 39% of the county's total population in 1970 and the growth rate of
its population has been recently somewhat lower than for the rest of the.
county. The immigration rate is largely responsible for the increase,
adding to the population about six times faster than the birth rate.
During the sixties immigration averaged a moderate rate of 14,800 persons
annually. Between 1972-1975 the annual rate jumped to 32,600 and recently
averages about 12,000 persons a year. Most of the immigrants are in the
older age groups. More than one-third of the current population have lived
in Pinellas County for less than five years.


-64-









Table 7.--Water rate schedules for sampled cities.


ST. PETERSBURG WATER RATE SCHEDULE
(for 5/8 inch meter within city limits)


Service/Price Base Rate Price per 1,000 gallons



Water $1.75 $.72
Wastewater $2.95 $.79*






Maximum charge for wastewater to single family residences is $14.85.





ORLANDO WATER RATE SCHEDULE
(for 5/8 inch meter within city limits)




Service/Price Base Rate Cost per 1,000 gallons

FEirs0t D 0 10rd0 00f0 0-var-lf 10 000

Water $1.88 $.44 $.37
Wastewater $1.75 $1.45 $1.45 $1.45


-65-








The characteristics of the incoming population mirror the character-
istics of the general population (Bureau of Economic and Business Research,
1980). In Pinellas County the percentage of persons in specific groups
have increasingly been weighed toward the older categories. In 1950 the
percentage of the population over age 65 was 19%; in 1975 that percentage
nearly doubled to 34%.

The large component of older persons in Piinellas county accounts for
several other demographic attributes. Median education levels tend to be
low, according to the 1970 census, 77% of the population completed equal to
or less than a high school education (Pinellas Planning Council., 1978).
Income averages had kept pace with the national averages until 1974 but
have been growing more s -owTy Vn te Ta-sa five years. Per capT!~ incom-e i
1975 for St. Petersburg was $5,817 (Pinellas Planning Council, 1977).
Household income which averaged $12,395 in 1976 for the county is expected
to increase by 48% to $18,318 by 1981.

The older population also helps explain the average household size and
type of housing structure statistics. Both a rising death rate and a high
rate of immigrating retirees contribute to an average household size which
is below the state average by 16% and the national average by 22%. In
1950 the average household size in Pinellas county was 2.71, in 1975 it was
2.43. There is a high percentage of one and two person households with one
person households alone accounting for 11% of all county households. Other
trends which contribute to this are the declining birth rate, increasing
divorce rates and the tendency for young adults also to form single-person
households.

There has been a more intense development of multi-unit dwellings to
accommodate the demand for single person units. In 1970 in the city of St.
Petersburg alone, the construction of multi-unit dwellings exceeded the
single-unit dwellings by 228%. In 1975 this trend was reversed but again
in 1979 the multi-unit percentage exceeded the single-unit by 300%. In
1977 the inventory of housing indicated that single-unit residences accounted
for 56%, multi-unit residences accounted for 32%, and mobile homes accounted
for 12% of the total housing structures. The a-verage facility sieof
apartments tends to be small, 63% of the total number have one bathroom
(Pinellas Planning Council, 1978).

Water supply and-demand characteristics

St. Petersburg receives its water from its own municipal system which
draws from the Floridan aquifer through well fields in Pasco and Hillsborough
counties. The number of wells has increased from 24 in 1963 to 35 in 1980.11
Average daily pumpage rates have been increasing from 20 million gallons in
1956 to 35 million gallons in 1975. Demands upon the system have been
growing, serving a population of 250,000 in 1970 and 283,000 in 1980. Per
capital daily use (GPCD) has risen from 130 gallons in the late sixties to


Telephone communication with Dean Hughes, St. Petersburg Public
Utilities, St. Petersburg, Florida, November 6, 1980.








nearly 140 gallons in 1975 (Healy, 1977). Projections for Pinellas county
indicate that by the year 2000 the GPCD will increase to 151, with the
country water systems requiring more than twice the amount of water that
they now distribute (Southwest Florida Water Management District, 1978).
The months of highest demand have been consistently April-May and the low
periods are December-February (Healy, 1977).


The remainder of Pinellas county is served by the Pinellas County Water
System. Together with the St. Petersburg Public Utilities these systems
have had shortage problems through the 1970's. The principal causes for
_ he scarcity are 1 since 1961 the geographic area _f the_ SWFWMD _had
experienced a critical lack of rainfall (Parker, 1975). Consequently
the aquifer was losing its chief source of recharging water. This re-
sulted in 2) excessive pumping of the available groundwater reserves
further lowering the area's water table. Excessive pumping was the natural
result of growing demand pitted against a depleting supply. Rising demands
were caused by a large immigrating population and also the prodigious water
requirements of the citrus and phosphate industries. The next problem in
the cycle is 3) the lowering of fresh water preserve yielded to salt water
intrusion into groundwater sources.

In 1973 the Southwest Florida Water Management District (SWFWMD) regu-
lated St. Petersburg's wells to provide more water to the Pinellas County
system until the latter could augment their capacity. Both systems were
severely taxed and an official water crisis with restrictions was de-
clared (Board of County Commissioners, 1973).

Since then the Pinellas County Water System has increased its number
of wells from 45 to 64. St. Petersburg has increased their total supply
to 35 production wells.

Orlando (Orange County)

Demographic characteristics

Orlando which is located in Orange county has a high density of 400
persons per square mile. Its suburban county, Seminole, has a density of
over 700 persons ,per square mile (Bureau of Economic and Business Research,
1980). The population of Orange county estimated for 1979 is 441,337 with
the city of Orlando accounting for 28% of the total population. The popu-
lation of Orlando has been increasing since 1940 at an average annual rate
of 61%. Orange county has experienced a net migration average rate of
about 4,300 persons annually between 1960 and 1970. Persons in age groups
of 10-60 years account for 75% of this figure. The age group of 60+ years
accounts for 19% of net migration. In contrast to Pinellas county new
residents are primarily from the younger age categories (Bureau of Economic
Analysis, 1979).

The distribution has not changed significantly in the past twenty years
in either Orange or Seminole counties however, the number of persons in the
0 14 years of age has been declining. In 1978 in Orange County 45% of the
population fell into the 15-44 age category, 22% fell into the 45-64 age
category and 11% fell into the 65+ category (Bureau of Economic and Business
Research, 1980).


-67-








According to the 1970 Census 57% of the Orange County residents had
completed a high school education. Due to the increased urbanization since
that time this percentage has most likely increased (Orlando Chamber of
Commerce, 1978). Median family income in 1969 was $8,880 and had increased
by 22% to $10,832 by the mid seventies. Per capital income in 1976 was
$5,123. Average household income was $14,719 in 1976 and it is expected
to increase by 50% to $22,062 by 1981.

Housing characteristics in Orange County differ from Pinellas although
not too drastically. Due to the younger population the average household
size is larger than in Pinellas County, in 1970 it was 2.06 and in 1975 it
was 3.02 persons (Orlando Chamber of Commerce, 1978), In a 1970 housing
inventory of lOOOO00 households, 76% were single-unit residences and 17%
were multi-unit. In 1975, the number of households increased to 140,000
with the percentage and multi-unit percentage increasing to 30%. In both
years the percentage of mobile homes averaged 5.8% of the total. Between
the years 1971-1974 construction starts for multi-unit structures for
outweighed single unit dwellings. However, since 1975 the construction
of single-unit structures has been greater. In absolute figures the con-
struction activity of both periods differs substantially: between 1971-1974
single-unit construction averaged 381 units annually while multi-unit con-
struction averaged 2,756 units annually. Between 1975-1977 single-unit
construction averaged an annual 289 units and multi-unit construction
averaged an annual 30 units.

Water supply and demand characteristics

Orlando receives its water from its own municipal system which draws
from the Floridan aquifer through well fields throughout Orange County
(Healy, 1980). The number of wells has increased from 18 wells in 1970
to 23 wells in 1980. Until 1956 surface water from nearby lakes was
utilized; since then the supply has been exclusively groundwater. Average
daily ,miii'.ge rates have increased from 15 million gallons in 1956 to 41
million gallons in 1975. Demands have been increasing; population served
in 1970 was 175,000 and in 1980 it was well over 200,000. Per capital
daily water use Jias increased-fram-15 -gallon is-i-n--197-to-20-8-ga-11-ons--in -
1975. Generally the period of the highest seasonal demand is April-May,
the lowest demand occurs during February.

Orange County is, for the most part, a water abundant area. The
effect of increasing urbanization, however, has since the 1960's begun to
change the quantity, and the quality of the available reserves. The
major source of aquifer recharge is rainfall, less of which is reaching
the groundwater supplies due to manmade surfaces. Also, contact with such
surfaces is causing an increase of pollutants recharged into the aquifer.
Urbanization trends also create an increase in demand. It is expected that
by 1985 water withdrawals will equal the water recharged. By 1990 it i~s
anticipated that the former may exceed the latter by 10%, i.e., water will
be "mined" in Orange County. The result of this could be severe contamina-
tion of the aquifer through intruding salt water (East Central Florida
Regional Planning Council, 1977; and St. John's River Water Management
District, 1977).


-68-








The inland areas of Florida have water shortage problems which cannot
compare in severity with those of the coastal cities. Despite the above
caveats, Orlando itself is situated close to several of the most productive
recharge areas in the Floridan aquifer (East Central Florida Regional
Planning Council, 1977). So far water shortages which warrant limitations
on residential consumer consumption have not existed.

Sampling Procedure

The population of the two communities, when adjusted to include
only single family dwellings with 5/8-inch water meters was approximately
74,900 households in St. Petei-sbii,i, and 25,000 in Orlando, Due to the high
cost of collecting observations the sample sizes were limited to a total
of 312 sampling units, or households.12

The procedure followed was a combination of stratified and cluster
sampling (Scheaffer, et al., 1979, Chapters 5 and 7). The reason for
using both was to insure the lowest cost for the greatest amount of varia-
bility in the major independent variables. The personal interview,
selected as the mode of data gathering, is very expensive. Travel costs
were cut by "clustering" the sample respondents into various separate
areas of the sample geographic area. The sample was stratified into three
groups characterized by high, medium and low levels. This insured that
the cluster of the respective incomes were chosen from all three in the
proportion in which they exist in the population.

The census tract was used as the primary cluster unit and the census
block as the secondary cluster unit. Census tracts were categorized into
income groups according to that group which constituted the highest per-
centage of the households in the tract.13 The results of this procedure,
listed in Table indicated that in St. Petersburg 5% of the sample would
be drawn from 3 census tracts, 32% from 15 tracts, and 63% from 38 tracts.
In Orlando, 1% would be drawn from 1 tract, 66% from 15 tracts, and 33%
from 6 tracts. Census tracts are not of uniform size. In St. Petersburg
the tract sizes range from 377 to 1901 households, and in Orlando they
-range from--1-1 -t~--2, 099--householdds.

The sample required 100 households from Orlando and 150 from St.
Petersburg. The marketing research firms responsible for executing the



12Personal interviews for each sampling unit cost $9.25 in St.
Petersburg and $9.34 in Orlando. The interviews were conducted by private
marketing research firms.

13Household counting was made prior to adjusting the population by the
qualifications indicated by (b) on page one. Census data was used from
1970. Updates of census information for income and household counts was
available for St. Petersburg from the Economic Base Study, Pinellas County,
1977, and the Demographic Study, Pinellas County, 1978; and for Orlando
from the Orange County Economic Data, 1979.


-69-













Table 8.--Number of census tracts and households in each income group.


ST. PETERSBURG


No. of No. of
Income groups tracts % Households


High over $15,000 3 5 2,863

Medium $7,000 $15,000 15 32 19,374

Low under $7,000 38 63 37,740

TOTAL 56 100 59,977





ORLANDO


No. of No. of
Income groups tracts % households


High over $15,000

Medium $7,000 $15,000

Low under $7,000

TOTAL


175

11,491



17,448









survey requested that "around 500 households be provided in each city.
To keep proportions even, 500 were provided in Orlando and 66 were pro-
vided in St. Petersburg. To maintain the proper proportions the 500
households (in St. Petersburg) were divided as follows: 25 from the
high income group, 160 from the medium income group, and 315 from the
low income group. In Orlando the 500 households were divided as fol-
lows: 5 from the high income group, 330 from the medium group, and 165
from the low income group.

Census tract maps were provided by the city planning departments of
St. Petersburg and Orlando. To use every census tract would not have
- -alleviatead the ravel -cos-t probl-e. ---Therefore the- number of- tracts- us-ed-
in St. Petersburg was divided by three and in Orlando it was divided by
two (see Table 8). This decrease in the number of tracts selected
caused the size of each cluster (tracts) actually used to increase,
Usually this is not desirable if homogeneity within a cluster is anticipated.
In this case, however, variability is insured because clusters (tracts)
were chosen independently in the three different income groups. The
proportions of income groups in the population was maintained by dividing
each income group of census tracts by the same number. The required
number of tracts per income group was then selected by simple random
sampling.

The next step was to select blocks from within the chosen census
tracts. The geographic boundaries of each tract were demarcated on a city
street map. The blocks were listed and selected through simple random
sampling. Within each income group the number of households was divided
by the number of tracts yielding the number of households required within
each tract (see Table 9). This number divided by the number of house-
holds residing on a block in each tract would give the number of blocks
required per tract. There is certainly no uniformity among city blocks,
however, on a tract by tract basis, this resulted in a range of 2-4 blocks
per tract which were to be canvassed. 1

The average tract in St. Petersburg was more dense than its counter-
part in Orlando although the range of the household count per tract is
broader in the latter. Consequently the tract number is less in St.
Petersburg. While it would seem likely that the number of households per
tract would be higher in St. Petersburg, the differential between house-
hold counts in the high income group accounts for a higher number of
households per tract in Orlando.


14Accurate counts of total housing units and their types (e.g., single-
family, multi-family, etc.) can be obtained for each block in a census
tract through the Block Statistics, 1970 Census of Housing. However this
data is computerized and was prohibitively expensive to obtain. Once a
first block in a tract was chosen its size was used to determine whether
another block was needed (after accounting for the presence of apartments
and commercial establishments). In all cases the tracts required at least
two blocks. Four blocks was the most that any tract required.


-71-












Table 9.--Number of households per census tract, by income level.


ST. PETERSBURG



Number of
Number of Number of households
Income level- -a-ets households per- tract



High 1 25 (33) 25 (33)

Medium 5 160 (213) 32 (43)

Low 13 315 (420) 24 (32)

Total 19 500 (666) '






ORLANDO



Number of
Number of Number of households
Income level tracts households per tract



High 1 5 5

Medium 14 330 47

Low 7 165 41


Total 22 500


-72-








Blocks were selected by simple random sampling. Chosen blocks were
located in the 1979 city directories of both cities. All names (of the
head of the household) were listed with the corresponding street ad-
dresses. A member of each of these households then became a potential
respondent.

Since many more households were provided than were required, each tract
had to be designated with the number necessary for the actual sample. This
was acquired by taking the number of total households necessary in each
tract group and dividing it by the number of tracts in that group. In
each group a percentage increase was given to account for responses which
cannot be used. For St. Petersburg an extra 32 were provided and for
OrTando an extra 30 surveys were provided for this purpose.


-73-














APPENDIX III


SUPPLEMENTAL SUMMARY OF QUESTIONNAIRE RESULTS


The questionnaire which was administered in order to elicit individual
-valuations- of residential -water-use--aso-contained --upptlemenilta6y questions
about socioeconomic characteristics of the sampled households, and about
the attitude, beliefs, and practices with regard to the water use of
households. Discussion of formal research results did not include much of
this information. It is the purpose of this appendix to summarize those
supplementary questionnaire results.


-74-










Table O1.--Socioeconomic data


Percentage in
response category


Response
Question category Orlando St. Petersburg


1. NUMRESPH: number
of persons in a
household









2. AGOTO1: household
members of age
0 to 10 years






3. AG11T20: household
j__embrs of-age
11 to 20 years


0


2


74


8


4. AG21T40: household
members of age
21 to 40 years


-75-










Table 10.--Socioeconomic data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


5. AG41T60: household 0
ImI'-.ibil ; of age 1
41 to 60 years
2


6. AG60: household 0
members of age 1
60 and over
2
3


7. SYRSCOM: highest
level of education
attained by head
of household
less than 12 yrs.
12 yrs.
some college/tech
B.A.,B.S.
M.S./Ph.D.


8. OWNRENT: own
or rent home


own


rent


9. II'IITVAL:
market value
of home


0 $30,000
31 $50,000
51 $80,000
over $80,000
no answer/do
not know


-76-










Table 10.--Socioeconomic data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersurg


10. HOMEAGE: age
of home in
years
0 10 11 15
10 20 37 16
20 30 36 38
30 50 10 17
over 50 0 12
no answer/
do not know 6 2


11. SIZEAREA:
estimated size
of property
less 1/5 acre 8 26.
1/5 3/4 acre 62 64
3/4 1 acre 2 7

26 2


12. INCOME:
yearly household
annual income

less $5,000 5 0
$5 $9,999 16 16
$10 14,999 13 21
$15 19,999 18 27
$20 24,999 13 11
$25 29,999 10 15
$30 34,999 10 4
$35 39,999 3 4
over $40,000 13 2


-77-










Table O1.--Socioeconomic data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


13. OCCUPATION:

professional, technical 13 12
managerial, administrative 16 12
sales 10 4
clerical 12 14
craftsmen 2 5
operative, laborer, service 8 4
farm personnel 2 4
retired 33 41
unemployed 3 4


14. WASHCAR:
frequency of
carwashing
less once monthly 47 51
1 3 times monthly 39 30
more 3 times monthly 15 5
no answer 13


15. SWIMPOOL:
presence of
swimming pool
yes 16 10
no 84 90


16. BOTTWAT:
use of
bottled water
yes 95 7
no 5 93










Table O1.--Socioeconomic data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


17. SEPTICT:
presence of
a septic
tank yes 5 4
no 95 96


18. HAVEWELL:
presence of a
private well
yes 3 8
yes, for lawn
watering only 19 61
no 77 31


19. KINDSYS:
kind of
irrigation
system
do not water 5 7
hose and sprinkTer -69 80
automatic system 26 13


20. HIGHPERD:
periods of
highest water
bills
Jan March 3 7
April June 19 8
July Sept 48 59
Aug Dec 0 3
always same 19 21
do not know 10 4


-79-










Table 11.--Conservation attitude data


Percentage in
response category


Response
Question category Orlando St. Petersburg


1. BELWSHRT:
belief in a
water shortage


yes
no


2. PROBYR: year
by which a
water shortac
will continue
or develop







3. DEGSEV: degr
of severity
o f-pre-sean--twa
shortage


1981
1985
2000 or beyond
never


'ee

ter ----- -
very serious
fairly serious
not serious
do not know
no shortage exists


Causes of present of
future water shortage:

4. LARDRY: lakes,
rivers drying up


great extent
some extent
not at all
no answer/
do not know


-80-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


5. LACKRAIN: lack
of rainfall
great extent 36 31
some extent 38 40
not at all 26 22
no answer/
do not know 2 7


6. DEPLGRW:
depleting
groundwater great extent 34 35
some extent 43 36
not at all 15 17
no answer/
do not know 7 12


7. POPGRLOC:
local
population
growth
great extent 70 69
some extent 28 20
not at all 0 4
no answer/
do not know 2 6


8.--POPGROTH:
population
growth in
other areas
great extent 49 57
some extent 38 30
not at all 9 7
no answer/
do not know 4 6
-81-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


9. INCWATUS:
increased water
usage due to
technical change
great extent 43 47
some extent 42 37
not at all 13 8
no answer/ 2 8
do not know

10. ICWASTE:
increased water
usage due to
waste
great extent 51 47
some extent 42 38
not at all 6 8
no answer/ 2. 7
do not know

11. COMINDUS:
increased water
usage due to
commercial-and
industrial needs
great extent 43 29
some extent 43 51
not at all 6 12
no answer/ 6 7
do not know


-82-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


12. AGRIUSE:
increased water
usage due to
agricultural
needs


13. POLLUTE:
increased
contamination
of water
supplies


great extent
some extent
not at all
no answer/
do not know


great extent
some extent
not at all
no answer/
do not know


14. WATUTIL:
water utilities have
not taken tbe
necessary steps
to provide
enough water great extent

some extent
not at all
no answer/
do not know


-83-









Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


Indicate which of these
conservation methods you
practice and to what
extent:

15. TEESHAVC: turning
off the water while
brushing teeth and
showering
large effort 10 44
medium effort 19 26
small effort 39 14
no effort 25 15
does not apply 0 1


16. DISHWASC: using
dishwasher only
when full
large effort 23 15
medium effort 2 4
small effort 6 0
no effort 2 3
does not apply 68 79


17. SHOWERC: taking
shorter showers
large effort 19 41
medium effort 19 27
small effort 31 11
no effort 15 12
does not apply 16 8


-84-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


18. WATERLGC: cutting
back on lawn
watering time
large effort 23 16
medium effort 16 12
small effort 19 15
no effort 15 5
does not apply 27 52


19. BATHC: filling
bathtub only
1/4 full
large effort 16 25
medium effort 13 10
small effort 26 9
no effort 13 18
does not apply 32 40


20. TOILTC: cutting
back on times
the toilet is
flushed
large effort 6 19
medium effort 21 20
small effort 24 19
no effort 45 39
does not apply 3 3


-85-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


21. CAPREUSE:
recycling
shower water
large effort 10 3
medium effort 1 0
small effort 8 4
no effort 68 86
does not apply 13 8


22. TTBRICC:
placing a brick
in the toilet
tank
large effort 2 7
medium effort 2 4
small effort 8 4
no effort 60 75
-does not apply 29- -0


Indicate which of these
water saving devices is
presently installed in
your home:


23. PRESHDC:
pressurized yes 39 29
showerheads
no 61 71


24. SSWASMC:
suds-saving yes 16 12
washing matching no 82 87


-86-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


25. SHTRWSTC:
shallow trap
water saving
toilet


26. CHEMTTC:
chemical toilet



27. LOWVDISC:
low volume
dishwasher


28. DUFLTTC:
dual flush
toilet tank


29. WLESSFAC:
washerless
faucets


30. RECINCR: a
recent local
water price
increase


31. DECONO: this
price increase
led to a water
use reduction


yes
no



yes
no



yes
no
does not apply


yes
no



yes
no



yes
no
do not know


yes
no
does not apply


-87-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


32. FUTINCR:
there is an
anticipated
water price yes 69 88
increase in
the near future
do not know 5 2


33. DECON50:
you would de-
crease your water yes 74 71
use if the price no 26 29
rose by 50%

34. HOWDE50:
you would de-
crease your water
use (if the price
rose by 50%) by
how much
1 10% 19 35
11 20% 34 18
21 30% 15 11
over 30% 2 9
does not'apply 26 2
do not know 3 25


35. DECON100:
you would
decrease your
water use if yes 89 81
the price rose by no 11 19
100%


-88-










Table 11.--Conservation attitude data--Continued


Percentage in
response category


Response
Question category Orlando St. Petersburg


36. HOWDE100:
you would decrease
your water use
(if price rose by
100%) by how much

1 10%
11 20%
21 30%
over 30%
does not apply
do not know


37. INSTWADV:
you would decrease
your water usage
by installing water
saving devices
yes, -immediate-
installment
yes when needed
a new fixture only
no
do not know


-89-










LITERATURE CITED


Adams, R. C., Currie, J. W., Hebert, J. A., and Shikiar, R., The Visual
Aesthetic Impact of Alternative Closed Cycle Cooling.Systems, Report
No. CR-0989, U.S. Nuclear Regulatory Commission, April, 1980.

Ajzen, I., and Fishbein, M., "Attitude-behavior Relations: A Theoretical
Analysis and Review of Empirical Research," Psychological Bulletin,
Vol. 84, No. 5, September, 1977, pp. 888-918.

Andrews, D P. A Etimaiot i- Ln of Pesidenti-l Dermand for UaLtr in Dad e
County, Florida, M.S. Thesis, Department of Food and Resource
Economics, University of Florida, 1974.

Bureau of Coastal Zone Management, Florida Department of Environmental
Regulation, The Florida Coastal Zone Management Program, 1979.

Bureau of Economic Analysis, Florida Department of Commerce, Orange
County of Economic Data, January 1979.

Bureau of Economic and Business Research, University of Florida, Florida
Estimates of Population, 1979, February, 1980.

Board of County Commissioners, Position Statement No. 2: Resource Needs
and Managed Growth for Pinellas County, Clearwater, Florida, October
30, 1973.

Position Statement: Water Resource Needs
of Pinellas County, July 30, 1973.

Bohm, P., "An Approach to the Problem of Estimating Demand for Public
Goods," Swedish Journal of Economics, Vol. 63, No. 3, June 1971,
pp. 94-105.

---,-"Estimating Demand-for Publie-Goods-: -A Experiment,"
European Economic Review, Vol. 3, No. 2, March, 1972, pp. 111-130.

Bradford, D., "Benefit-Cost Analysis and Demand Curves for Public Goods,"
Kylos, Vol. 23, No. 4, 1970, pp. 775-791.

Brookshire, D. S., Ives, B. C., and Schulze, W. D., "The Valuation of
Aesthetic Preferences," Journal of Environmental Economics and
Management, Vol. 3, No. 4, December, 1976, pp. 325-346.

Brookshire, David S., Randall, Alan, and Stoll, John R., "Valuing Incre-
ments and Decrements in Natural Resource Service Flows," American
Journal of Agricultural Economics, Volume 62, Number 3, August 1980,
pp. 478-488.

Bruvold, W. H., "Residential Response to Urban Drought in Central California,"
Water Resources Research, Vol. 15, No. 6, December, 1979, pp. 1297-
1304.


-90-










Cicchetti, C. J., and Smith, V. K., "Congestion, Quality Deterioration,
and Optimal Use: Wilderness Recreation in the Spanish Peaks Primitive
Area," Social Science Research, Vol. 2, No. 1, March, 1973, pp. 15-30.

Clouser, R. L., and Miller, W. L., Household Demand for Water and Policies
To Encourage Conservation. Technical Report No. 124, Water Resources
Center, Purdue University, August 1979.

Crespi, I., "What Kinds of Attitude Measures are Predictive of Behavior?"
Public Opinion Quarterly, Vol. 35, No. 3, Fall, 1971, pp. 327-334.

Currie, J. M., Murphy, J. A., and Schmitz, A., "The Concept of Economic
Surplus," Economic Journal, Vol. 81, No. 324, December, 1971, pp. 1297-
1304.

Danielson, L. E., Estimation of Residential Water Demand, EconomicsResearch
t1.,'rt No. 39, Department of Economics and Business, North Carolina
State University, October, 1977.

Davis, 0. A., and Whinston, A. B., "On the Distinction between Private and
Public Goods," American Economic Review Papers and Proceedings, Vol. 57,
No. 2, May, 1967, pp. 360-373.

Davis, R. K., The Value of Outdoor Recreation: An Economic Study of the
Maine Woods, Ph.D. Thesis, Department of Economics, Harvard University,
1963.

East Central Florida Regional Planning Council, Policy Alternatives in
Water Recharge Areas, July 1974,

Freeman, Myrick, The Benefits of Environmental Improvement, Johns Hopkins
University Press for Resources for the Future, Baltimore: 1979.


-Gehm, H Land-reagman-,J.-- e-d()-,-Handbook--of-- Wate- Resourc-se-and
Pollution Control, Van Nostrand Reinhold Company, New York: 1976.

Gottlieb, M., "Urban Domestic Demand for Water: A Kansas Case Study"
Land Economics, Vol. 39, No. 2, May, 1963, pp. 204-210.

Hammack, J., and Brown, G. M., Waterfowl and Wetlands: Toward Bioeconomic
Analysis, JohnsHopkins University Press for Resources for the Future,
Baltimore: 1974.

Hanke, S. H., and Boland, J. J., "Water Requirements or Water Demands?"
Journal of the American Water Works Association, Vol. 63, No. 11,
November, 1971, pp. 677-681.

,_ and Davis, R. K., "Demand Ianagement through Responsive
Pricing," Journal of the American Water Works Association, Vol. 63,
No. 9, September, 1971, pp. 555-560.

"Demand for Water Under Dynamic Conditions," Water Resources
Research, Vol. 6, No. 5, October, 1970, pp. 1253-1261.


-91-











"Some B'havioral Characteristics Associated with Residential
Water Price Changes," Water Resources Research, Vol. 6, No. 5, October,
1970, pp. 1383-1386.

Headley, J. C., "The Relation of Family Income and Use of Water for
Residential and Commercial Purposes in the San Francisco-Oakland
Metropolitan Area," Land Economics, Vol. 39, No. 4, November, 1963,
pp. 441-449.

Healy, H. G., Public Water Supplies of Selected Municipalities in Florida,
195U, U. S -Geological-N.u..L-W .a i 1ldo1aw- V1. rida, 1-9-77.

Henderson, A., "Consumer's Surplus and the Compensating Variation,"
Review of Ecouomic Studies, Vol 8, No. 2, February, 1941, pp. 117-121.

Henderson, J. M., and Quandt, R. E., Microeconomic Theory, McGraw-Hill
Book Company, New York: 1971.

Hicks, J. R., "The Four Consumer's Surpluses," Review of Economic Studies,
Vol. 11, No. 1, Winter, 1943, pp. 31-41.

"The Rehabilitation of Consumer's Surplus," Review of
Economic Studies, Vol. 8, No. 2, February, 1941, pp. 108-116.

Hirschleifer, J., DeHaven, J. C., and Milliman, J. W., Water Supply:
Economics, Technology, and Policy, University of Chicago Press,
Chicago: 1960.

Hogarty, T. F., and Mackay, R. J., "The Impact of Large Temporary Rate
Changes on Residential Water Use," Water Resources Research, Vol. 11,
No. 6, December, 1975, pp. 791-794.

Hoffman, M., Glickstein, R., and Liroff, S., "Urban Drought in the San
Francisco Bay Area: A Study in Institutional and Social Resiliency,"
tJournal of-TiheAmerican Water Works Association, Vol. 7Y, No. 7,
July 1979, pp. 356-363.

Howe, C. W., and Linaweaver, F. P., "The Impact of Price on Residential
Water Demand and Its Relation to System Design and Price Structure,"
Water Resources Research, Vol. 3, No. 1, First Quarter, 1967, pp. 13-32.

Kurz, M., "An Experimental Approach to the Determination of the Demand for
Public Goods," Journal of Public Economics, Vol. 3, No. 4, November,
1974, pp. 329-348.

Lattie, J., 'Public Education for Water Conservation," Community Water
Management for Drought and Beyond: A Handbook for Local Government,
California Office of Emergency Services, Sacramento, California, 1979.

Lauria, D. T., "Water Demand Forecasting--Some Concepts and Techniques,"
in McJunkin, F. E. (ed.). The State of America's Drinking Water,
North Carolina State University, April 1975, pp. 235-258.


-92-











Linaweaver, F. P., Geyer, J. C., and Wolff, J. B., Final Summary Report
on the Residential Water Use Research Project, Department of En-
vironmental Engineering Science, John Hopkins University, July, 1966.

Loehman, E. Consumer Surplus and Cost-Benefit Comparisons for Collective
.Goods, Center for Economic Policy Research, Stanford Research Insti-
tute International, Menlo Park, California, November, 1978.

Loehman, E., Ben-David, S., and De, V. H., Measuring Demand and Political
Acceptability for Nonmarket Goods: A Case Study of Health Effects
P eatIdt to Air Oualitr, Center for- Economic Policy- Resear-ch, St-anfo-rd
Research Institute International, Menlo Park, California, October, 1978.

Lynne, G., and Gibbs, K., Demand and Pricing Policy for Residential Water,
Economic Report 83, Institute of Food and Agricultural Sciences,
University of Florida, 1976.

Maler, K.G. Environmental Economics: A Theoretical Inquiry, Johns
Hopkins University Press for Resources for the Future, Baltimore; 1974.

McGarry, R. S.,and Brusnighan, J. M., "Increasing Water and Sewer Rate
Schedules: A Tool for Conservation," Journal of the American Water
Works Association, Vol. 71, No. 9, September, 1979, pp. 474-479.

Milne, M., Residential Water Conservation, Report No. 35, Water Resources
Center, University of California at Davis, March, 1976.

Mishan, E. J., Cost-Benefit Analysis, Praeger Publishers, New York: 1976.

Morgan, W. D., "Residential Water Demand: The Case from Micro Data,"
Water Resources Research, Vol. 9, No. 4, August, 1973, pp. 1065-1067.

North, R. M., Consumer Responses to Prices of Residential Water, Journal
Series No. 185, Georgia Agricultural Experiment Station, University
of Georgia, 1967.

Orlando Chamber of Commerce, Statistical Data, Orlando, Metropolitan Area,
July, 1978.

Parker, CG. G., "Water and Water Problems in the Southwest Florida Water
in _,.ement District and Some Possible Solutions," Water Resources
Bulletin, Vol. 11, No. 1, February, 1975, pp. 1-20.

Pinellas Planning Council, Demographic Study, Pinellas County, Clearwater,
Florida, April, 1978.

Economic Base Study, Pinellas County, Clearwater,
Florida, 1977.

Randall, A., and Brookshire, D. S., Public Policy, Public Goods, and
Contingent Valuation Mechanisms, Staff Paper No. 68, Department of
Agricultural Economics, University of Kentucky, June, 1978.


-93-




Full Text

PAGE 1

Non-Market Valuation of Water in Residential Uses By K. C. Lewis and R. R. Carriker PUBLICATION NO. 57 FLORIDA WATER RESOURCES RESEARCH CENTER RESEARCH PROJECT TECHNICAL COMPLETION REPORT OWRT Project Number B-036-FLA tchi ng Grant Agreement Number 14-34-0001-8074 Report Submitted: April 1981 :'" The work upon whi ch th; s :report is based vIas suppor,ted in part by funds provided by the United States of Interior, Office of Water Research and Technology as Authorized under the Water Resources Research Act of 1964 as amended. : .. ;

PAGE 2

ACKNOWLEDGEMENTS The authors wish to their gratitude to the Office of Water Research and Technology, United States Department of Interior, for financial support of this work. The administnative assistance of Dr. James Heaney, Director of the Florida Water Resources Research Center is -greatly appreciated. Thanks are also due Ms. Mary Robinson for accounting assistance. The authors also extend special appreciation to Richard Kilmer for his constructive assistance in the execution of this research. Gratltude 1S also extended to Dr. Beau Beaulieu, to Ray Boyd of the Orlando Utilities Commission, and to Mr. Dave Nichols of the St. Petersburg Public Utilities for their advice and cooperation with the sampling procedures, and to Mr. Richard Marella of the St. Johns River Management District for his technical assistance. Special thanks also goes to Ms. Debra Linn who patiently prepared the final manuscript.

PAGE 3

ACKNOWLEDGEMENTS. LIST OF TABLES. LIST OF FIGURES ABSTRACT. CHAPIER I INIRODOCIION. The Problem Setting. The Research Problem TABLE OF CONTENTS Objectives of Research Method of Procedures .. CHAPTER II CONSUMER'S SURPLUS AND VALUATION OF BENEFITS Benefit Estimation ... Neoclassical Assumptions About Consumer Behavior The Consumer's Surplus. Consumer's Surplus: Quantity Change Analysis. CHAPTER III NON MARKET VALUATION: THE ITERATIVE BID. ,a; fl':,., Page --, i iv v vi 3 3 3 5 5 5 6 7 11 Non Market Valuation Method. 11 The Bid Curve and Consumer's Surplus 12 Application of the Iterative Bid 16 Potential Bias in Iterative Bids 17 CHAPTER IV -VALU1NG RESIDENTIAL WATER USE BY CONTINGENT MARKET TECHNIQUES .. ...... ...... 21 WTP, WTA, and Consumer's Surplus for Residential Wa ter Use. . . . . 21 Exposition Using Traditional Indifference 23 Valuation of Residential Water Use '. 25 Estimating Equations .. 29 i i

PAGE 4

The Estimation Procedure The Questionnaire .... CHAPtER V -RESULTS OF ANAL YS I S J. Response to the Questionnaire. Comparing vlTP and WTA lJ;th;n lise Categories Comparing WTP and WTA Between the Two Samples. Results of Regression Analysis CHAPTER VI -SUMMARY AND DISCUSSION APPENDIX I -THE QUESTIONNAIRE. APPENDIX II -THE SAMPLE. ... Community Characteri st i cs. Samp 1 i ng Procedure . APPENDIX III SUPPLEMENTARY SUMMARY OF QU[SIONNAIRE RESULTS. LITERATURE CITED. . . . . iii Page 32 32 34 34 37 39 39 47 49 64 64 64 74 90

PAGE 5

LIST OF TABLES 1. Criteria for rejecting surveys. 2. Mean bi ds . . 3. 4 5. 6. 7. 8. WTP regression resul ts lHP regress; on resII]ts WTA regression res ul ts \lJTA regress i on resul ts Water rate schedules for Number of census tracts for St. Petersburg for Orl ando. for St. Petersburg for Orlando. sample cities and households in each i nco'me group 35 38 41 42 44 .. 45 65 70 9. Number of households per census tract, by income level. 72 10. Socioeconomic data. 75 11. Conservation attitude data. 80 iv

PAGE 6

LIST OF FIGURES 1. The welfare impact of a change in the quantity of a good X from Q" to Q I. 8 2. The total value curve for increments and decrements in the level of provision of a service, Q, for an individual who lnltlally enJoys the level QO and the lncome yo. .. 13 3. The relationships between WTP and WTA, and Hicksian com-pensating and equivalent measures of consumer's surplus. 15 4. Relationship of WTP, WTA and Hicksian measures of surplus: the context of res; denti alit/a ter use . 22 5. Diagramaticexposition, using traditional indifference curves, of the relationship of WTP, WTA, and Hicksian measures of consumer the context of residential water use. . . . . . . 24 v

PAGE 7

ABSTRACT Increases in water demands were historically met through the augmenta tion of water supply facilities. However, the most easily developed water sources have already been tapped, and ter poll uti on adds to the cos t of developing some water sources. Given the increased cost and difficulty of water supply augmentation, more attention has been given recently to demand management. This requires the assigning of priorities to water uses, and the subsequent fulfillment of only the most highly valued needs. In 1978 the U.S. National Water Commlsslon dlrected that the most highly valued needs be determined through the concept of consumers I willingness to pay for publicly supplied water. This research is designed to test the use of a non-market valuation technique to assess the residential consumers I willingness to pay for household water. It was hypothesized that the willingness to pay would be dependent upon a consumer's income, family size, the amount of water used, the presence of a well, and also upon variables representing a consumer'sbe1iefs and attitudes concerning local water scarcity and conservation. Results indicated that indeed most of these variables did playa role in the formulation of a willingness to pay for residential water. Some inconsistencies in results, however, indicate that more research is required before non-market valuation techniques can be ap plied with confidence to residential water demand analysis. vi

PAGE 8

CHAPTER I INTRODUCTION The Problem SettinR Studies on demand reductions have been prompted nationwide by the realization that water supply systems will be increasingly pressured both by natural drawdowns and also by the environmental objections to a proliferation of water projects which are not considered to be ecologically sound. Measures taken by a water utility to conserve water through demand reduction ale relatively recellt. The IllOSt extensive research on consumer response to water shortages has taken place in California following the drought seasons of 1976-77. The California drought of 1976-77 gave impetus to water tesearch in demand management. Precipitation in both years averaged less than onehalf of normal levels and the compounded effect had vast consequences for the state1s water supply systems. Accommodating the problem from the supply side involved supply enhancement methods, which are generally unsuccessful at such short notice, and interdistrict transfers which were exceeding costly and politically unpopular. Demand reduction policies proved to be far more effective. Accomplished principally through rationing, the overall average demand reduction through the state for the summer of 1977 was 49% (Hoffman, et al., 1979). Earlier, in 1966 a drought in the Potomac area of Washington, D.C., srarked studtes by the ngton Suburban Sanitary Commi ss ion which culminated in a conservation program. Between the years 1980 and 2000 the demand for water in the D.C. area given population grbwth and per capita use rates, is expected to far outstrip the potential supply and sewer canabilities. The pursuit of regional solutions by supply augmentation was discouraged until a conservation.program was put into effect. The resulting program, consisting of a public education program, a revised building and plumbing code, and a conservation rate structure, was able to reduce residential consumption by 13.8% through the 1970ls (McGarry and Brusnighan, 1979). In Florida, water shortages result from intense urbanization in the relatively water scarce parts of the state. Of Florida1s nine million residents, nearly 75% reside in the coastal zone. Compounding this effect are the water demands caused by the state1s annual 25 million tourists who are attracted to the shoreline areas where water is far less rlentiful than inland (Bureau of Coastal Zone Management, 1979). The subsequent groundwater drafts have causeu excessive watertable drawdowns. In some areas, such as Jacksonville and Tampa-St. Petersburg, the resulting salt water intrusion has threatened both current and po-tential well sources (Parker, 1975). .' Natural shortages are not the only source of difficulty facing efforts to augment water supplies. Environmental pressure groups have lob bied consistently for the reconsideration offunding for water supply infrastructure. They contend that the costs of the environmental damage to natural settings outweigh the benefits of fulfilling the growing water demand of urbanization (Sierra Club, 1980). The political weight carried by environmentalists was made manifest in 1977 when President Carter deleted -1-

PAGE 9

federal funds for 18 major federal water projects (Schlerger and Cerviso, 1980). This provided a catalyst for the implementation of a new national water policy establishing stringent environmental criteria for future projects. Announced in June, 1978, the new national water policy specifically directed that water conservation be added to the Principles and Standards of the Water Resources Council evaluates all federally constructed projects (Schad, 1978). Thus it was required that all feder al agencies with water programs advocate conservation and integrate it into all program planning. In response to nationwide water shortages and environmental objections Lo wa LeY's upply developlllent projects, water conservati on and the wa ter demand management that it necessitates, has become a national ,objective. Residential demand for water has been the subject of several economic analyses.l Water demand studies have reported relating quantity of water demanded by theresiderlti:al sector to price, and to a, number of nonprice variables including household size, household income, property size, property value, and climate (Andrews, 1974; Clouser and Miller, 1979; Danielson, 1977; Gottlieb, 1963; Hanke, 1970; Headley, 1963; Hogarty and Mackay, 1975; Lineaweaver, et al., 1966; Lynne and Gibbs, 1976; Morgan, 1973; North, 1967; and Wong, 1972). The development of water management policies has benefited from the knowledge gained by these studies. demand studi es have typi cally rel ied upon secondary sources of data, using statistical inference to test hypotheses about relationships among variables associated with residential water demand. Available data from secondary sources generally does not permit the refinement of analysis to include effects on non-marginal price changes, differentiation among components of household water demand, and the influence of attitudes, knowledge and beliefs on residential water demand and valuation. As a consequence, residential water demand models statistically are typically not designed to yield information on the social valuation of residential water use. They are therefore limited in usefulness for several types of policy evaluations: (1) They are not well suited to evaluate policies which contemplate dramatic changes in water rates, never before experienced by users and therefore outside the range of existing data; or regulatory reductions in water allocated to the residential sector. lThe word "demand" is used by different people to mean different things. In economics, demand is a technical term referring to the amount of a commodity that would be purchased at a given price (Lauria, 1975). In the case of water for residential uses, price of water is a key determinant of quantity demanded, given the influence of other variables such as per capita household income, the preferences of people with respect to green lawns, daily showers, sVJimming pools, and other variables. The point is, the economic concept of demand views demand as a variable, associated in predictable ways with the combined influences of identifiable variables. By contras t, the terms "requi rements" or "needs," often used i nterchangeably with the word demand, are, in fact, mere expressions on unexplained' and unqua 1 ifi ed des i res for a commodity. convey; ng no i nforma ti on about their determinants. -2-

PAGE 10

(2) They do not disaggregate household water demand beyond the distinction between domestic (in-house) and sprinkling (out door) demand, and are to that extent not well suited to evaluation of policies directed at influencing specific household uses of water. (3) They do not generally permit assessment of the manner in which IItaste-likell variables such as beliefs, knowledge, and attitudes relate to demand for, and valuation of, ,residential water use, and are not well suited to evaluation of public education programs designed to influence rates of water use. The Research Problem Whether water policy emphasizes water supply development or demand management, there exists a need for detailed information about the welfare effect on people when limitations are placed on household uses of vJater. The concept of a consumer's IIwillingness-to-pay" for residential water was adopted by the National Water Commission in 1973 to direct allocationofwater resources (Schad, 1978). The determination of this benefit measure has not been previously researched in residential water demand analysis. Objectives of Research The overall objective of this research is to adapt and apply a methodology for eliciting consumers' valuations of nonmarket goods to the measurement of consumers' valuations of water in residential uses. Specific objectives include: (1) mea s urement of cons umers' va 1 ua t ions of the los sin ut il ity associated with specified reductions in the amount of water permitted for specified residential uses: (2) identification of the major determinants of consumers' valuations of water in residential uses; and (3) quantification of the relationship between consumers' beliefs and attitudes about water conservation, and their valuation of water in specified residential uses. Method of Procedure Valuation, for purposes of benefit/cost analysis, is an attempt to ascertain the quantity of money which gainers and losers from some proposed action will consider equivalent in value to their respective gains and losses (Randall and Brookshire, 1978). Non-market valuation mechanisms, some of which are called contingent market valuation mechanisms, elicit valuations of non-market goods by establishing hypothetical mar-kets and recording the contingent decisions of individuals confronted with special changes in these hypothetical markets. Non-market valuation techniques differ from more conventional demand analysis in their reliance upon primary data rather than secondary data. -3-

PAGE 11

A method of contingent market valuation called iterative bidding will be used in this study. In this valuation procedure, a hypothetical market is described and defined in detail (Randall and Brookshire, 1978). Alternative levels of provision of the good are described. The institutional details pertaining to method of payment and enforcement of terms are explained. An enumerator then poses prices to which the respondent reacts, indicating whether he would pay the price or go without the good. The price is varied iteratively until the price at which the respondent is indifferent is identified. The process is repeated for several levels of provision of the good. For this study, representative samples of residential water customers will be drawn from two major cities in Florida. Consumers will be visited by an interviewer and confronted with a hypothetical situation in which the water util ity, because of growth in water demand and shortages of raw water supplies, must either enforce a highly selective rationing plan or else raise utility bills in order to finance expansion in water supply capacity. Consumers are then asked, through an iteiative questioning procedure, to reveal their maximum willingness to pay to avoid the specified reduction in water use. The procedure will be repeated for contingencies involving successively larger reductions in water use. In this manner, a measure of willingness to pay to avoid each of several decrements in water use will be obtained. A second set of questions present a scenario in which the consumer is entitled to his current level of water use and will elicit the amount of compensation necessary in order to induce the consumer to willingly accept specified decrements in level of water use. In this manner, a measure of willingness to accept compensation for each of several decrements in water use can be derived. Payments and compensati ons wi 11 be consi dered in the form of the total water bill. Separate information about household income, household size, attitudes concerning water conservation and other, similar variables will be obtained. Multiple regression analysis will be used in order to test for statistically significant relationships between the amount of bid (e.g., willingness to pay) and variables normally associated with demand for residential water. -4-

PAGE 12

CHAPTER II CONsm1ER I S SURPLUS /\ND VALUA nON OF BENEFITS Benefit Estimation The objective of analysis is to direct the usage of j goods, services, or resources to their most highly valued employment. Thus the measurement of values imputed by consumers to goods and services has as its ultimate purpose the economically efficient allocation of resources. 2 Non-market valuation methods have been developed for pur poses of benefits and costs pertai ning to provisi on of goods. services, and resources for which established price-quantity data are not available. Consumer net benefits from a current or proposed resource .allocation are defined and measured with the assistance of a theoretical concept called the consumer's surplus. The concept of consumer's surplus has theoretical underpinnings in the assumptions about consumer behavior of the neoclassical economists. Neoclassical Assumptions About Consumer Behavior Utility Maximization Consumers are presumed to make rational choices as to the level and mix of goods and services they consume, with the objective of maximizing their individually and subjectively perceived levels of satisfaction or utility (Henderson and Quandt, 1971). Each consumer has, theoretically, an indifference map for any combination of goods. He also has an income constraint which limits the range of combinations, from the indifference map, which he can afford to consume (given positive prices for the goods and servlces). J\ssumlng decreaslng marglna1 rates or'substitution alllong aoods and servi ces, and assum;:ng increases in consumpti on produce i ncreases in subjectively experienced utility, the rational consumer will allocate his limited income among the goods and services in a manner which maximizes#his utility. Demand Functions Given the assumptions about consumer behavior, the quantity of a good demanded per time period is a function of the price of that good, ') LConceptually, economic efficiency is achieved if the allocation of resources for production and consumption is Pareto optimal--a state achieved if no reallocation of resources to improve the welfare of one -individual can be made without reducing the welfare of one or more other individuals (see Henderson and Quandt, 1971, fora summary). -5-

PAGE 13

the price of substitutes (or complements), the consumer's income, and those aspects of tastes, and preferences which underlie the consumer's indifference map, Demand functions are mathematical formulations which express the form and magnitude of the relationship between quantity demanded, the dependent variable, and the relevant independent or explanatory variables. The Consumer's Surplus Consumer's surplus is an imp0rtant concept in the measurement of social benefits in any social cost-benefit calculationi (tishan, 1976). A simple definition of COlisUiller' 5 surplus is. tile IIlaximuliI sum of Iliolley a consumer would be willing to pay for a given amount of a good, less the amount he actually pays. By this definition, the market price in a perfectly competitive market is an adequate index of the value of a marginal change in quantity of a good, but is not an adequate measure of larger quantities of a good. In terms of the demand curve, beginning from a given amount of the good offered on the market, the corresponding point on the demand curve indicates the maximum price the average consumer is 0illing to pay for the last unit of that amount. But to each of the total number of units purchased, as measured along the quantity axis, there corresponds some average maximum valuation. The whole area under the demand curve, therefore, corresponds to society's maximum valuation for the quantity in question. Consumer's surplus is that portion of the area under the demand curve above the price line. In statistical estimates of the price-quantity relationship represented by the demand curve, other variables known to influence quantity demanded will be held (or assumed) constant 1976). However, if aggregate money income is held constant in estimation of the demand function, any fall in the price of the good raises real income. This increase in real income will cause some increase (if the income effect is positive) in the amount of the good purchased in addition to that increase in quantity representing substitution of the good for some other in response to the change in relative prices. Hicks (1941) redefined the concept of consumer's surplus, using an ordinal system of indifference curves, as the amount of money--to be paid by the consumer when the price falls; to be received by him when the price rises--which, fOllowing a price change, leaves him at his original level of welfare. This measure of consurner's surplus allows the consumer to readjust the mix of goods which he consumes following the price change. Henderson (1941) pointed out that, in general, the relevant compensating variation in income would depend on whether the con9umer had to pay for the priviledge of buying the new good or whether he was to be paid for not being able to buy the good. Subsequently, Hicks (1941) defined four measures of the change in consumer's welfare resulting from an actual or proposed price change. These four measures, summarized, are as follows: r -0-

PAGE 14

(1) "Compensating variation" is the amount of compensation, paid or received, that will leave the consumer in his initial welfare position following the change in price if he is free to buy any quantity of the commodity at the nevI pri ce. (2) "Compensating surplus" is the amount of compensation, paid or received, that will leave the consumer in his initial welfare position following the change price if he is constrained to buy at the new price the quantity he would have bought at that price in the absence of compensation. (3) "Equivalent variation" is the amount of compensation, paid or received, that will leave the consumer in his subsequent welfare position iQ the absence of the price change if he is free to buy any quantity of the com modity at the old price. (4) "Equivalent surplus" is the amount of compensation, paid or received, that will leave him in his subsequent wel fare position the absence of the price change if he is constrained to buy at the old price the he would have bought at that price in the absence of compensation. The decision as to which measure is appropriate for a particular analysis depends upon the specific nature of the proposed change for which benefits and costs are to be measured. HIcks' four measures pertain to goods, priced in competitive markets, for which price and quantity data are available. Benefit estimation in the context of nonmarket, unpriced, or underpriced goods required adaptation of the consumer's surplus concept to make it applicable in quantity change analysis as well as in price change ,analysis. Consumer's Surplus: Quantity Change Analysis In benefit-cost analysis, the economist is sometimes concerned not so with impacts of price changes as with the welfare impacts of changes in the bundle of goods, services and amenitites possessed, used or consumed by individuals (Randall and Stoll, 1980). Proposed projects or programs may remove some goods from individual opportunity sets or introduce new goods; and may decrease the quantities of some goods while increasing quantities of others. The goods affected by proposed programs may be divisible, exclusive, marketed goods with observed prices. However, they may also be recreational or environmental amenities or other goods which are in varying degrees indivisible, non-exclusive, and unpriced (Randall and Stoll, 1980). Randall and Stoll (1980) identify the conditions under It/hich consumer's surplus measures can be adapted to situations in which it is bundles of goods, rather than prices, which are changed. Their diagrammatic exposition (Figure 1) proceeds as follows. -7-

PAGE 15

QI Q* Q" QUANTITY OF X Figure 1 .--The v,Jelfare impact of a change in the quantity of a good X from Q" to QI. -8-

PAGE 16

Consider a normal good X. which, depending on the program alternatives chosen, may be provided in two different quantities, Q' and QII, where QII is greater and ceterus paribus preferred. The pragmatic measures of value of these two bundles of goods are willingness to pay (WTP) and will ingness to accept (InA). In a market exchange situation these correspond, respectively. to the buyer's best offer and the seller's reservation price; in a nonmarket s i tua ti on, they correspond to VIi n i ngness-to-pay ",and willingness-to-accept compensation (Randall and Stoll', 1980). If the program alternative under evaluation would reduce the quantity of X from QII to Q', the compensating measure of welfare loss is WTAc the compensation which would keep the individual at his initial welfare level; and the equivalent measure is WTpE, the loser's willingness to pay to avoid the quantity reduction from QII to Q' \,Ihich, if paid, would place the individual at his subsequent welfare level (Randall and Stoll, 1980). If the proposed program would increase the quantity of X from Q' to QII, the compensating measure of welfare gain is WTpc which, if paid, would keep the gainer at his initial welfare level; and the equivalent measure is WTAE the compensation which would be needed to bring the potential gainer to his sub sequent welfare level in the event that the proposed program is not implemented (Randall and Stoll, 1980). If X were a perfectly divisible good, traded in large markets at zero transactions costs, a program to reduce Q" to Q' while leaving the individual's numeraire, Y (a composite of lIall other goods"), at Y would initially move the individual point E to B (Figure 1), lowering his welfare level from III to I' (Randall and Stoll, 1980). However, the existence of frictionless markets will permit him to trade along his new budget line until he reaches D, achieving the welfare level of 1*. Given this adjustlllent. his 'vJTpE is EP, equal to '" Y', while his WTAc Now, assume that X is lumpy and can only be held in the amounts QII and Q'. Since intermediate adjustments in commodity holdings are not permissable, the Hicksian compensating and equivalent measures in commodity space are analogous to the Hicksian surpluses, not the variations, defined over price changes (Randall and 1980). Accordingly, the price lines (Figure 1) become meaningless. WTP is EG which is equal to yg9, and WTAc is BA, which is equal to '{Va, and larger in absolute value than WTPE. In identifying the appropriate measure of welfare change, several distinctions among the Hicksian measures must be clarified (Brookshire, et a 1 ., 1930): -9-

PAGE 17

(1) The Hicksian compensating and equivalent measures of consumer's surplus differ with respect to the reference level of welfare. The compensating measure, by using the initial welfare level as the reference level, measures the welfare impact of changes as if the individual had a right to his initial level of welfare. The equivalent measure, by using the subsequent welfare level as the reference level, treats the individual as if he had a right only to his subsequent level of welfare. (2) The Hicksian variations differ from Hicksian surpluses in that variations are calclllated after the consllmer has made optimizing adjustments in his consumption set, while surpluses are calculated without first allowing such adjustments. -10-

PAGE 18

CHAPTER III NON MARKET VALUATION: THE ITERATIVE BID Non-Market Valuation Method of value for goods and services traded or sold i,n private markets are typically derived from market-generated price-quantity information. In the case of public goods, and publicly provided goods, valuation methods must be devised which emulate the market process or in some other manner generate information from which measures of value can be den ved. valuation have been applied in benefit estimation studies of environmental improvements, the creation or improvement of recreation sites, and the provision of wildlife. These methods fall into two general categores: the proxy methods and the bid game methods. Proxy methods require the choice of variables upon which quantifiable observations can be made (i .e., for which data exist) and which are hypothesized to be highly correlated with the price (or other) variable for which data do not exist. An example of the proxy method is the use of measurable travel costs as a proxy measure of the willingness of recreationists to pay for the composite of experiences associated with a particular recreational activity, taking into account other explanatory variables such as income, substitutability of other recreation sites, and individual tastes (see, Sinden, 1973). The bid game method was pioneered by Robert K. Davis (1963) who used it to estimate the benefits of maintaining New England wilderness areas. The bid game method elicits direct valuations of goods or vices from consumers without the use of intermediate variables. Typically the bid game uses a questionnaire format. Respondents are oriented to a hypothetical market (scenario) 1m which the current level of a good or service is assumed to exist. Then it proposes a change in this level of provision and records the respondents I valuation of that change in the level of provision of the good. Several procedures have been used for obtaining valuations or bids. With an open-ended response format the respondent is unconstrained in providing bids. With a categorical for mat, the is presented with a predetermined set of bid possibil ities. The iterative bid format presents the respondent with a series of alternate states, iteratively eliciting the respondent's valuation of each change from the reference state (Adams, et al., 1980). The use of prices distinguishes the market from the non-market ap proaches to valuation (Bradford, 1970). Demand functions for private (marketed) goods quantify demand responses to price changes, and estimation of demand functions is a first step toward benefit estimation. In the case of public goods and publicly provided goods for which pro duction and consumption is often divorced from consideration of individual willingness and ability to pay, consumers do not normally have the opportunity to purchase as many units as they wish. They are, instead, confronted with public decisions to change the quantity or quality levels of provision without joint reference to price. The term "states" is used to refer to current and subsequent levels of provision. -11-

PAGE 19

The Bid Curve and Consumer's Surplus By soliciting valuations in terms of willingness to pay and/or willingness to accept compensation, the bid game method is designed to measure the consumer's surplus. Thus it has the same theoretical basis as value estimating procedures using estimated demand functions for marketed goods and services. The Bid Curve Consider an individual who currently enjoys some specified level, Q, of a good Or' service.3 He also enjoys a given qualltity of the Hicksian "all other goods" numeraire, Y, for convenience called income. His level of utility, then, is a function of his income and the level of provision of the good represented by Q, i.e., (1) U = U(Q, Y). The individual is at the origin (Figure 2), which defines his level of welfare in the "without project" situation. To the right of the origin, the level of provision of Q to the individual increases; to the left of the origin, it decreases. From the origin, a move up the vertical axis represents a decrease in income, while a move down the vertical axis represents an increase in income. A total value (TV) curve, or bid curve (Bradford, 1970), passes through the individual's initial state. It is of positive slope, given that the individual is not satiated in the range of values under consideration. For decreases in Q, the TV curve lies in the southwest quadrant; for increases in Q, it lies in the northeast quadrant. The TV curve is an indifference curve, that is, Starting at the origin, yO -Y-is the individual's willingness to pay OHP) for an increment in the provision of good Q 'from QO to Q+. Willingess to accept (tHA), i.e., y+ yO, is the amount of money which would induce the individual to accept voluntarily a decrease in the level of provision of the service from QO to Q-. Restating equation (2), (3) U(Qo, yo) = U(Q-, yO + WTA) U(Q+, yo vJTP). WTP, WTA, and Consumer's Surplus To clarify the relationship between Hicksian compensating and equivalent measures of value, WTA and WTP, and the total value curve of 3The discussion of the bid curve follows that in Brookshire, et al., ( 1 980) -12-

PAGE 20

Income IQDecrements .....-yo / -------y+ / Q+ Increments .. + P(Qo q-) in Q Price Line Total Value Curve Figure 2.--The total value curve for increments and decrements in the level of provision of a service, Q, for an individual who initially enjoys the level QO and the income yo. -13-

PAGE 21

Figure 2, Brookshire, et al. (1980) offer the following example. THe subject of benefit-cost analysis is a project which would divert a specified area of wildlife habitat to some alternative use, effectively destroying its usefulness as habitat. The analyst needs to know the value of the losses which would be suffered by an individual who currently enjoys the wildlife amenities provided by that habitat. In the initial state the individual has utility level U(Qo, yO). His "with project" utility level be U(Q-, yo). The "with project" and "without project" levels of Q are predetermined so that the individual has no opportunity for optimizing adjustments. One lIIeasuy'e of tile wel fare impact on this individual would be his to acqui esce in the proposed change., Call thi s HTAc o 0 Q,y;Q,Y;Q. Superscript C indicates that this is a Hicksian compensating measure of value, the first subscript pair, QO, yO indicates that the individual's reference level of welfare (his presumed right or entitlement) is QO, yO. The second subscript pair indicates that QO, yO is also his initial welfare level. The third subscript, Q-, indicates the level of provision of the good (in this case, wildlife-related services) the consurT)er wouli.d enjoy after he has accepted the compensation and the change in the level of services. If he were compensated by an amount just equal to his WTA, his income after compensation would be yO +WTAc. HTAc is a measure of the individual's valuation of the reduction in wildlife-related amenities from QO to Q-and was deri ved from the i ndi vi dua lis TV curve for wildlife-related amenities (see Figure 3). However, another measure of value might have been used to estimate the i ndi vi dua l' sloss of wil dl i fe-ameniti es: the amount of money he would be willing to pay to avoid a reduction in the provision of wildlife amenities. This WTP to avoid a less preferred position reflects a pre sumption that the individual has no right (entitlement) to his current wel fare level. The reference level of wel fare is not that associated with the initial situation, but'the proposed (or subsequent) welfare level. This secoild llieasure of tile ilidividual 's we1 Far'l= loss ;s denoted WTpE The superscript E indicates a Hicksian equivalent Q-, yO; QO, yO; Q O measure of The first subscript pair indicates that the reference level of welfare (his entitlement) is taken to be Q-, yO. The second subscript pair indicates the individual's initial state, QO, yO. The third subscript indicates that the individual, after he has paid will be allowed to enjoy the QO_level of amenities, If he pays WTpE, his final income will be yo -WTpE. Brookshire, et al. (1980) note that the pair of total value curves in Figure 3 could be used to estimate the value of a project which would increase the level of wildlife-related amenities from an initial level Q-to a "with project" level QO. The individual's initial state is Q-, yO, If the individual is entitled only to his initial ,welfare level, the appropriate measure is yO; Q-, yo; QO, ,.Note that vJTpc equals WTpE, If the individual were entitled to the additional wildlife-related amenity, the measure is yO; Q_, yO, Q_ which equals yO; QO, yO, Q-. -14-

PAGE 22

Income Income Quanti ty of Q --;Figure 3.--The relationships between WTP and WTA, and Hicksian compensating and equivalent measures of consumer's surplus. -15-

PAGE 23

The example makes several points. (1) Equivalent measures apply to situations in which the initial welfare level is different from the reference level, when the individual's "entitlement" is given by the subsequent state rather than by the initial state. (2) Compensating measures assume that the initial state is the reference welfare level; that the individual's lIentitlementli corresponds to the initial state. (3) WTpE cannot be found using a TV curve passing through the individual's initial state. It can be found only by using another TV curve passing through the' reference sta te. (4) There is a compensating and equivalent version of WTP, as there is of WTA. When comparing two alternate'levels of provision of a good, there are four Hicksian value measures: WTpc to obtain the preferred level; WTpE to avoid the less preferred level; WTAc to accept the less preferred level; and WTAE to forego a promised increment to the preferred level. (5) \iJTA and vJTP, whether they be compensating or equivalent measures, represent Hicksian variations if the consumer has an opportunity to make opti mi zi ng readj ustments ,i n his cdnsumption set. (6) HTA and WTP, whether they be compensating or:equ'ivalent measures, represent Hicksian surpluses if nO'6ptimizing adjustments in the individual's consumption are possible. Application of the Iterative Bid Application of the iterative bidding technique [also known as ,lIcontingent market valuation (Randall and Brookshire, 1978)] requires identification of the distribution of rights. That is, the quantity of a good to whi ch a respofldent is entitl ed must be determi ned before the rel evant consumer's surplus measure can be chosen. For example, if the consumer ;s entitled to the quantity he currently consumes and a proposed project would reduce the quantity available to him, the relevant consumer's surplus measure would be WTAc On the other hand, if the consumer is entitled only to the reduced quantity the relevant measure would be UTpE. The contingent market or iterative bid format requires that the respondent understand the quantity and quality characteristics of the scenario with which he is confronted. For example, in a study of willingness to pay for environmental improvements, Randall and Brookshire (1978) used photographs to depict levels of provision of cleaner air to ensure uniform perception among the respondent population. The payment method or vehi cl e must be chosen for its rel evance and feasibility and must be clearly specified to the respondents (Randall and -16-

PAGE 24

Brookshire, 1978). For example, expressions of willingness to pay for increased recreational opportunities or wildlife-related amenities might best be elicited in terms of willingness to pay hunting license fees or recreational site access fees, since these vehicles of payment are familiar to consumers and relevant to the specific goods or amentities for which measures of valuation are sought. In the iterative bidding form of contingent market valuation the respondent reacts to prices by an enumerator, indicating whether he would (in a WTP case) pay the price or go without the good. The price is varied iteratively until the price at which the respondent is indifferent is identified (Randall and Brookshire. 1978). The hypothetical market thus established has the advantage of low transactions costs, "trade" in goods \t,Jhich cannot be marketed in the conventional sense, the ability to evaluate many options and perturb the components of publicly provided packages of goods (and packages of public goods) in order to examine the contributions of these components to the value of the package. Contingent market approaches to valuation are inferior to actual markets in that the bids obtained are not firm, enforceable offers, but are be havioral intentions given the occurrenceof the hypothesized contingencies. If the survey instrument is realistic and coherent, then the main conditions under which behavioral intentions should predict actual behavior can be met. Studies by Ajzen and Fishbein (1977) and Crespi (1971) have been used as the psychological foundation for the economic research in hypothetical valuations. Potential Bias in Iterative Bids Empirical application of the iterative bid poses three problems: (1) the potential exists for respondents to engage in strategic behavior; (2) variations in responses may result from using different starting points for the same birl; and (3) variations in response may result from the use of different vehicles of payment for the same bid. Strategic Behavior Bias Of the three possible sources of bias, respondent strategic behavior is of the most concern and has consequently precipitated the most research. The problem of strategic behavior arises from the very essence of the contingent market valuation method: its hypothetical nature. It is necessary to distinguish strategic behavior which may result from paucity of information concerning procedures and purposes of the survey, or from a marked divergence between the survey scenario and ex periences of real life. In other words bias can be introduced into participant responses simply from confusion. Strategic behavior results when the respondents in an attempt to maximize personal self-interests, sabotage the bid game by responding dis--17-

PAGE 25

honestly to the questions. It has been generally accepted that respondents attempting to maximize the benefits which may possible accrue from a survey encounter will answer questions in a predictable manner. The two main factors which affect the response are (Freeman, 1979): (1) the i ndi vi dua 1 IS percei ved personal i nfl uence on the outcome, and (2) the probability that he will actually have to forfeit the amount he states as his maximum willingness to pay (when WPA is the desired measure). The tendency for an overstatement of one's true willingness to pay occurs when the individual desires a certain outcome and believes that his re sponses will affect that outcome but that he will not be asked to actually relinguish that bid amount. An understatement will occur when a respondent desires an outcome but believes that the number of participants is large enough to ensure that outcome regardless of the size of his bid which he may, in fact, be required to relinguish, These two biases are considered to be the upper and the lower bounds on bid estimates and are the key targets for solutions to systematic bids in surveys (Bohm, 1972). Researchers have attempted to adjust for the effects of strategic responses in two ways. The first approach attempts to measure, statistically, the bias imposed by strategic responses and adjust recorded responses by this calculated degree of bias (Kurz, 1974). This proce dure is based on the unverified assumption that all biases in a given situation are uniform, and suffers from lack of operational format to follow from its theoretical exposition. The second approach to potential strategic behavior attempts to structure questions in the survey situation to eliminate incentives for strategic behavior (Bohm, 1971). Respondents are left uncertain as to how their responses will affect their payment outcome. This can be achieved by informing them that they may be called upon to pay their bid, some proportion of their bld, or to make no payment at all. This uncertainty will create no clear advantage to the respondent in understating or overstating his true bid. This approach of loutvJitting" the respondent has explored theoretically by Maler (1974), Kurz (1974), and Tideman and Tullock (1976). It is the approach used in the attempts at non-market valuation represented by the work of Adams, et. al. (1980), Cicchetti and Smith (1973), Randall, et. al. (1974), and Hammack and Brown (1974). Elimination of incentives for strategic responses may also eliminate i ncenti ves fo r gi vi ng accurate answers (Freeman, 1979 ). Care mus t be taken to stress the importance of careful and thoughtful answers (Hammack and Brown, 1974). It is also probable that respondents will, in any case, attach a probability of their own choosing to the potential influence of the survey and of their bids (Freeman, 1979). These problems are embodied in what Randall and Brookshire (1978) calls conflict between "strategic versus hypothetical bias,11 the latter being the result -18-

PAGE 26

... not of systematic influences but rather of noise resulting from failure to invest as much effort in the contingent decision as would be invested in an actual decision, presumably because the penalties from a wrong decision in a hypothetical market are not so tangible (Randall and Brookshire, 1978). This discussion reemphasizes the need to distinguish between genuine strategic behavior and the bias caused by lack of relevance, and accentuates the need for conveyance of reality in the design of contingent market scenarios. --Although no methodology exists which conclusively tests for strategic bias, tvlO empirical studies made by Sohm (1972) and Brookshire, et a"'., (1976) have attempted to reveal its existence. Bohm's study consisted of five separate surveys, each intended to elicit a certain biased response in sHuations which were not hypothetical (respondents had toactually pay their total bid or some proporti.on. The main result of the test was that none of the five surveys yielded average bids which ficantly deviated from any of the others. Results of which were designed to elicit overstated bids did not differ greatly from those which intended to elicit understated bids (based on criteria mentioned above). The difference came with the sixth survey which was based on a sHuation known to the respondents as "hypothetical." Bohrn concluded that, in any case, the behavior exhibited supports the notion that respondents tend to view their impact on total demand as and that understatements are neutralized by the potential threat that a good may not be provided. Bohm makes it clear that this limited testing does not the possibility of bias but suggests that the deviation between honest and biased response may be small. The study b.y Brookshire, et aL (1076), which measured the value of iii!" quality of an /\rizona recreat'ion and \vilderness area, confronted survey respOl1dentswith a hypothetical situation in which the respondents knew the survey personnel were not working in an official ----u 1 d n CTt-he-l"eft-tt+red to a ctu-ally----p-r-e-vi-cie--f--o-r--p-ay=----l!lent or receive compensation. The authors hypothesized that bias, if it existed, would be exhibited in the following way: when asked what their maximum willingness to pay be to insure an improvement ill environmenta"1 quality, -the "environmentalists" vvould tend to overstate their bids, far exceedinq the mean, and "non bids would tend to be The duthors l1sSUI1I(:;ci that honest bids would be clistr.ibuted normany, and that b"las of the hypothesized type, if widespread, would tend to flatten the distribution. They concluded that strategic bias was not prevalent in their survey because the distribution was, in fact, h"ighly centered about the mean bid. Hovvever, they recognized that strategic bias may still exist since they have no way of insuring that bids were influenced by an incentive to be accurate. In summation, strategic behavior in cannot be comp"'etely eliminated because it cannot be accuratJY:.qfeasured. At best, its vestiqes, aris"inq in either ail actual or situation, can. '-, t ., only be hypotiletisized. That the problem exists", h,ow-e!ver, .is generally UfJOIL The ultimate caveat is that the cotlfidence associated with -19-

PAGE 27

the statistical testing of bid estimates does not indicate the reliability with which behavioral intentions will be translated into actual behavior. Vehicle of Payment Bias Vehicle of payment bias can be said to exist when the value of mean bids or the incidence of protest bids are significantly different across payment vehicles (Randall and Brookshire, 1978). In such cases it is possible that respondents are revealing a preference for paying a bid in one form rather than another and disguising a true willingness to pay. Two stlldies have specifically addressed the potential for vehicle of payment bias in their work. Randall et. al. (1974) did not find vehicle bias at statistically significant levels in their research, but at less stringent levels of significance some differences were observed between the sales tax, the user fee, and the electricity bill forms of payment. The authors suggested that the vehicle of payment selected for any respondent group should be the most germane to the issue at hand. This appears to make the most intuitive sense. Individuals confronted with forms of payment which they do not prefer may tend to give low bids, and, while it may seem less likely, persons confronted with the payment form "of their choicell may tend to give higher bids. The most likely-to-beemployed form of payment would assist in smoothing the extremes. Starting Point Bias Starting point bias exists if different bids are elicited from a set of questions which contain, ceterus paribus, different starting points. If a starting point bias exists, a bid game with iterations be ginning at $2.00, rather than at $1.00, vJOuld produce bids which are biased upward. Randall and Brookshire (1978) suggest that starting point bias is only a problem for situations in which respondents have little independent basis for valuation of the good. As with vehicle of payment bias, starting point bias may be tested for by varying the starting point across the sample and testing whether the responses dif-fer significantly between groups. -20-

PAGE 28

CHAPTER IV VALUING RESIDENTIAL WATER USE BY CONTINGENT MARKET TECHNIQUES WTP, WTA, and Consumer's Surplus for Residential Water Use The relationship between Hicksian compensating and equivalent measures of value and WTA and WTP in the context of residential water is illustrated by the following scenarto. Hypothetically, if current trends in population growth continue, demand for water wil'] exceed existing capacity for supply in five years.4 A project to augment water supply capacity will make it possible for current rates of use per household (Q, in Figure 4) to be sustained in five years. Failure to invest in increased supply capacity will require an average decrement in Q relative to current rates of consumption. If it is assumed that citizens are :entitled to their existing level of util ity, the benefit-cost analyst wi shes to measure consumer' s (Willingness to Accept payment) for voluntarily accepting decrements in water availability and, therefore, water use. This measure is a compensating measure, denoted by superscript c, since it assumes individuals are entitled to current utiltty levels. Figure 4 depicts the situation just described. The consumer starts at yo, QO, and is entitled to the level of utility associated with that situation. If a decrement from QO to Q-is inevitable, the question to the consumer becomes, "I'Jhat amount of compensation will you accept to voluntarily acquiesce in the decrement in Q and feel as wel!l off as before the decrement in Q? In Figure 4 the WTAc is depicted at yo + WTAc. On the other hand, if a decrement from QO to Q-is inevitable and the consumer is not entitled to his existing level of utility, the appropriate question to ask the consumer is, "How much are you willing to pay to avoid !'l decreillerit ill water availability frolll QO to Q-?" In this case, the level of welfare (reference level) to which the individual is entitled is that associated with yo, Q-in Figure 4. The WTpE (Willirigness to Pay) to avoid the decrement in Q represents a decrement in income relative to the initial yo leveh leaving the consumer at income level Y IHPE. Moreover, the contingency as proposed entails a loss of welfare for tbe individual, a new reference level of Y and Q, and a different TV The superscript, E, denotes an equivalent measure. 4The time reference to five years is selected arbitrarily for purposes of illustration. 5An additional point can be raised concerning interpretation of the above scenario. The scenario posits "inevitable" reductions in Q (unless supply capacity is augmented), and for urban systems customers, it is not clear whether individual customers would (or could) make subsequent quantity adjustments. If it is assumed, for example, that customers can -21-

PAGE 29

Income Income I Quantity of Q Figure 4.--Relationship of WTP, WTA and liicksian measures of surplus: the context of residential water use. -22-

PAGE 30

Exposition Using Traditional Indifference Curves Figure 4 depicts the relationship of WTpE and WTAc to Hicksian equivalent and compensating surpluses in terms of the total value (bid) curve for residential water. Figure 5 allows diagrammatic exposition in terms of traditional indifference curves [following the exposition by Randall and Stoll (1980)J. Consider an individual whose rate of use of residential water, Q, and an Iiall other goodsll numeraire represented by income, Y, is currently QO and yo, respectively (corresponding to the origin of Figure 4) and per mitting the individual to achieve utility lel/e1 1 at point C. NO'.'I, as sume that Q must be reduced to Q-from QO, but the individual is entitled to remain at his original utility level. If no income compensation were received by the individual, he would find himself with income yo (un changed) and quantity Q-(reduced from QO) permitting him to achieve only utility level II at point B. Since the scenario depicts a quantity change without reference to rate or price changes, price lines are irrelevant to this indifference curve analysis.6 Compensation needed to return this individual to his original utility level will be BA which equals yoyl, and which corresponds to WTAc in Figure 4. In Hicksian terms, this WTAc is a compensating surplus. Alternatively, consider an individual with income yo and quantity of water Q who is at point C on utility level 1. Assume further that the level of Q must be reduced from QO to Q-and that the individual is not entitled to compensationt The individual IS willingness to pay to continue using QO of water is WTP and will be the amount of Y he would have to give up in order to reach point D, on utility level II, corresponding to quantity QO. This WTpE is, in Hicksian terms, an equivalent surplus and amounts to CD in Fi gure 5, equal i ng y"yo. On Fi gure 4, thi s corresponds to WTpE. This arrangement leaves the individual on a different lower indifference curve, II at point D. not resell publicly supplied water in a competitive market, the relevant compensating and equivalent measures are Hicksian surpluses, rather than variations. On 'the other hand, if it is assumed that competitive markets for the re-sale of publicly supplied water would emerge in the wake of mandatory supply reductions (or use reductions), then the compensating and equivalent measures are Hicksian variations, not surpluses. The irrelevance of price lines also hinges on the assumption that no "after market" for the private exchange of publicly supplied water exists. -23-

PAGE 31

QJ E o u C I--i y l'JTAc I I 1------I I Quantity II Figure 5.--Diagramatic exposition, using traditional indifference curves, of the relationship of HTP, WTA, and Hicksian measures of consumer surplus; the context of residential water use. -24-

PAGE 32

Valuation of Residential Water Use Introduction This study pursued three objectives: (1) to measure consumers' valuations of the losses in utility which correspond to reductions of water in specified residential uses, (2) to identify the major determinants of these individual valuations, and (3) to quantify the relationship between those determinants aQd the individual valuations. The method employed to measure consumers' valuations was the contin gent market, using the iterative bid. Selection of independent variables was based on the hypothesis that variables which affect a household's monthly water demand will be similarly correlated to the individual con sumers' valuations. Multiple regression analysis was used to test for, and to quantify, significant functional relationships between the selected independent variables and the consumers' valuations. The Dependent Variables WTP and WTA were solicited by personal interviews with a sample of residential water users and are direct measures of the consumer's sur plus derived from residential water use. The were used to compute the mean bi ds for the sample. .' t'. f" WTP and WTA bids were obtained for each of five of residential water use: (1 ) toilet flushing, (2 ) bathing and ( 3) clothes washing, (4) lawn watering, and (5 ) total household water use.l The contingent market scenarios Each respondent was visited personally by a survey enumerator and asked to respond to a hypothetical situation concerning the eminent 70f these five categories, the bid functions for clothes washing and lawn watering were not estimated because the data base contained too many missing values in the corresponding bid and quantity figures. -25-

PAGE 33

availability of water for selected household activities. Specifically they were to 1 d: in each of the five categories of household water use. S To elicit bids for each reduction contingency, the respondents were asked the following: If there was a requirement to reduce bath/shower usage in your household by 20%, how much additional money would you be willing to pay per month to avoid complying with this requirement? Iterations of bids were then presented to the beginning with the current price they pay for the amount of water represented by the reduction contingency. The completion of this process produced a set of WTP bids. The WTA bids were elicited in a similar manner, essentially using the following scenario: If you had decided to comply with the requirements of water use reduction what amounts of would you require to repay you for the inconveniences which you incur when you make these reductions? Please indicate the minimum amount of money which you could accept through reductions in your monthly bi 11, fo r s pecifi c water uses. Dealing with sources of bias It was crucial to the design of the iterative bid section of the questionnaire that the problems of bias (see Chapter III) not be built into the data. Although strategic bias, starting point vehicle of payment bias cannot be conclusively identified or eliminated, efforts were made to minimize their occurrence. -----entire questionnaire has been included as Appendix I. Reduction contingencies were expressed in different terms for different uses, eg., in terms of hours per week for lawn watering, per day for toilet flushing, cycles per week for clothes washing, etc. -26-

PAGE 34

To counteract the potential for strategic bias, the bid scenario was characterized as being hypothetical (it was emphasized that the enumerators were not acting in any official capacity). On the other hand, enough "reality" may well have been injected into the scenario by the fact, well known to the consumers, that recent \AJater price increases had been effected in both cities from which samples were drawn. 9 It was hoped that in consequence of these two factors respondents would see no personal benefit to sabotaging a hypothetical bid game but would recognize the general pertinance of the scenario. To minimize the likelihood of starting point bias, current water rates were used as the basis for the opening bid for all questions. Respondents were therefore not confronted with unrealistically high or low initial bids. Care was tak'en to minimize vehicle of payment bias by using the most realistic form of payment: the water bill. Since no other form of payment is relevant there was no need to include other payment vehicles to test for significant discrepancies. The personal interview method was used to minimize the likelihood of bias resulting from misunderstanding of the purpose or procedure of the questionnaire. The Independent Variables The variables which affect a household's water consumption are hypothesized to also affect that household's valuation of water use. The logic behind this hypothesis is relatively simple: the WTA and WTP bids are measurements of value, or more accurately, expressions of individual perceptions of value. This value is derived from the utility of household water which, in turn, is a function of how water is used and how. much water is used. How water is used and how much is used by a household are items of .information which the independent variables were designed Lo reflect. Some variables were expected to' have positive influences on valuation and others were expected to have negative influences. The were categorized in the following variable groups: (1) the household status group (a) number of members per household (b) annual income 9The sampling procedure and characteristics of the cities from which samples were drawn is included as Appendix II.

PAGE 35

(2) the household technology group (a) average total monthly water consumption (b) average monthly water consumption for each water use category (c) the presence of a private well (d) type of irrigation system (3) the conservation group (a} belief in local water shortage (b) expressed willingness to decrease water consumption if the price were increased by 50%. The ratidnale for selecting each variable and the hypothesized nature of its relationship to individual bids, varies from one variable to another. The household status group Size of household. The number of persons in a household is hypothesized to have a positive influence on the willingness to pay to maintain current levels of water use. As the number of persons in a household increases the necessary minimum rate of water use also increases. Size of household is also expected to have a positive influence on the willingness to accept compensation for essentially the same reason. Annual income. Income is hypothesized to have a .influence on the willingness to pay an increased amount to maintain current water use levels. Willingness to pay is expected to be related to ability.to pay. It may also be positively correlated with the willingness to accept compensation for water use reductions. The household technology group Monthly average water consumption. The monthly average quantity of water that a uses is hypothesized to be positively correlated with both the willingness to pay and the willingness to accept bids It is expected that households which require relatively large quantities of v-Iater are those which rely upon numerous water using appli'ances, e.g., more than one bath room, and/or have 1 arger family sizes. '. Monthly average consumption for each water use category. The presence and use of specific household water using fixtures is an indication of how much water the household is likely to feel it "needs." It is hypoth esized that the average rate of each water use will be positively correlated wi til wi 11 i ngness to pay to avoi d, anJ .wi 11 i ngne?J5. to accept payment to acquiesce in, reductions in those particular use rates. -28-

PAGE 36

Presence of private well. The presence of a well is hypothesized to have a negative effect on willingness to pay to maintain current levels of water use because the well is a substitute for publicly supplied water. If the well water i:s used for irrigation only it is not a perfect substitute since it is often not of potable quality and because there are normally official restrictions on the use of wells. Consequently, the effect of a private well on WTP and WTA will be less if the well is used only for irrigation. Type of irrigation system. Irrigation systems which use the most water should have the greatest positive effect on willingness to pay to maintain current levels. The automatic sprinkler system requires more water to operate efflclently than does a rotat,ng hose-and-sprln[ler system. Therefore, the presence of the former should be more highly correlated 0ith WTP and WTA than the latter. The conservation variable group Belief in water shortage. The belief in a current or pending water shortage is hypothesized to have a positive effect on the willingness to pay to maintain current levels of water use. Recognition of scarcity implies recognition of the likely increase in cost of maintaining a particular rate of supply to households. However, recognition of water shortage situation may have the op posite effect on WTA. Recognition of a true shortage may entail a willingness to acquiesce in measures to curtail use without demanding or expecting compensation. Willingness to reduce consumption in the event of an increase in price of water. This variable measures a willingness to conserve water, refl ecti ng an atti tude about the need or des i rabil ity of conserva ti on. It is hypothesized that a greater willingness to reduce consumption is positively correlated with a willingness to pay to avoidreduGtion in use and negatively correlated with WTA. Estimating Equations Willingness to Poy The estimating equations to explain willingness to pay to avoid reductions in total water use are: (1 ) WPWAT10 = f(NUMRESPH, INCOME, AVGCON1O, HAVEWELL, KI i'WSYS, DECON50)" (2 ) WPWAT30 = I NCm,1E, AVGCON30, HAVEWELL, BELl'/SHRT" UECON50), ( 3) WTWAT50 = f(NUMRESPH, INCOME, AVGCON50, KI NDSYS, DECON50), -29-

PAGE 37

where: WPWATl 0, l.JP\lJAT50 = amount of increase in the total monthly water bill that a consumer is willing to pay to avoid 10%, 30%, and 50% reductions, respectively, in monthly average water con sumption, fWMRESPH = number of persons res i di'ng in househo Td, INCOME = combined annual income of persons residing in household, A\/GCON ]0, 30, 50 ::: fig'lre representing a negative 10%, 30%, 50% respectively of monthly average water consumption, HAVEWELL = indication of whether or not household owns a private well and whether or not it is used solely for lawn irrigation, KINDSYS = indication of whether or not household waters a lawn and what method, hose and sprinkler, or automatic system, is applied, BELWSHRT = indication of whether or not household believes that there is a current water shortage in its locale and/or that there will be a shortage by the year 2000. DECON50 = indication of whether or not household would be willing to reduce its household water consumption i,f the price of its water rose by 50%. The estimating equations to explain willingness to pay to avoid re-ductions in bath and shower use are: (4) \;JPBS20 = f ( NUt/1RES PH, INCOME, BATHSQ20, BELWSHRT, DECON50), (5) WPBS30 f( INCOME. BATHSQ30, BELWSHRT, DECON50) (6 ) l'JPBS40 = f( I NCQI'1E, BATHSQ40, BELWSHRT, DECON50) where: WPBS20, WPBS30, WPBS40 = willingness to pay per month to avoid 20%. 30%. and 40%, reductions ,respectively, of monthly average w*ter use in household, for bathing and showering. BATHSQ20, BATHSQ30, BATHSQ40 = a negative 20%, 30%, and 40%, respectively, of monthly average water use in household, for bathing and showering. All other variables retain the same definitions as for equations (1), (2), and (3). The estimating equations for the willingness to pay to avoid reductions in toilet flushing water use are: -30-

PAGE 38

(7) WPTFl = f(NUMRESPH, INCOME, TOILQ1, BELWSHRT, DECON50), (8) WPTF2 = f(NUMRESPH, INCOME, TOILQ2, HELWSHRT, DECON50), (9) WPTF3 = f(NUMRESPH, INCOME, TOILQ3, BELWSHRT, DECON50), where: WPTF1, WTPF2, WPTF3 = monthly willingness to pay to avoid reducing toilet flushing by an average of 1 time, 2 times, 3 times per person per day. TOILQ1, TOILQ2, TOILQ3 = figure representing a nebative monthly quantity of water corresponding to 1, 2, and 3 flushes of the toilet per person per day. All other variables retain the same definitions as in equations (1), (2L and (3). Willingness to Accept The estimating equations to explain willingness to accept (WTA) compensation to voluntarily accept specific .reductions in total household water use are: (10) WAWAT = f(NUMRESPH, INCOME, HAVEWELL, KINSYS, BELWSHRT, DECON50, AVGCONT), where: = wi 11 i ngness of the consumer to accept compensati on in the form of a reduced monthly water bill to acquiesce in a specified re duction in monthly total household water use, ,lWGCONT = the quantity reduction (carrying a negative sign) in monthly total water use to WlllCN the consumer's WTA pertains. All other variables are as defined for equations (1), (2), and (3). The estimating equation to explain WTA compensation to voluntarily accept a specified reduction in water used for bathing/showering is: (11) WABS = f(NUMRESPH, INCOME, BELWSHRT, DECON50, BATHST), where: WABS = willingness to accept compensation in the form of a reduced monthly water bill, for a reduction in water used for bathing/ showering, BATHST = the quantity reduction (carrying a negative sign)in monthly water use for bathing and showering to where the consumer's WTA pertains. All other variables are defined as in equations (1), (2), and (3). -31-

PAGE 39

The estimating equation to explain WTA compensation for reductions in water used for toilet flushing is: (12) WATF = f(NUMRESPH, INCOME, BELWSHRT, DECON50, TOILT), where: = will i ngness of the consumer to accept compensation, in the form of a reduced monthly water bill, for a specified reduction in water used for toilet flushing, TOILT = the quantity reduction (carrying a negative sign) in monthly water used for toilet flushing to which the consumer's WTA perta i ns All other variables are as defined in equations (1), (2), and (3). The Estimation Procedure The estimating equations for WTP and WTA postulate causal relationships between the observed bids and each of several independent variables. The procedure for estimating the parameters of those equations and testing for statistical Significance of the coefficients is multiple regression analysis. A complete exposition of this procedure can be found in any standard econometrics text book [for example, Wonnacott and Wonnacott (1970)J. The Questionnaire Information on the hypothetical variables was obtained using primary and secondary data collection methods. The WTP and WTA bids, and all data excluding monthly water consumption data were collected through a questionnaire. Oustomer records of the water utilities of the two cities from which samples were drawn, St. Petersburg and Orlando, were used to obtain the water consumption data for the period November 1978 October 1980.10 The questionnaire was persOnnal1y presented by trained enumerators. It was suspected that the complexity of the questionnaire would have created data problems if respondents were left solely to their own interpretations of questions. It was anticipated that bids ItJOu1d differ between geographic areas which experience, either currently or in the recent past, different 0ater availability conditions. To test this hypothesis two sampling groups were chosen, St. Petersburg and Orlando. The former has a history of water scarcity problems and the latter is located in a relatively water plentiful area of the state. (See Appendix II for a complete discussion of the sampling procedure). The questionnaire (Appendix I) contained four sections designed to provi de data on: lOSt. Petersburg Public Utilities in St. and the Orlando Utilities Commission in Orlando. -32-

PAGE 40

(1) the socioeconomic status of the household: income, household size, education, etc.; (2) the estimated amount of use of water using facilities and appliances in the consumer's home; (3) the actual val:uation of water in specific acti'vlties (tlie IHP,. bids) (4) the beliefs and attitudes the consumer had concerning water scarcity in his/her region, and concerning water conservation practices. The third section of the questionnaire, discussed in a previous section, provided the WTP and WTA bids. "'; 1 -33-

PAGE 41

CHAPTER V RESULTS OF ANALYSIS Four aspects of this research are of interest: (1) the success of the questionnaire and the survey procedure in terms of usable responses, (2) the difference between WTP and WTA bids for the same sample in the same "'later use category, (3) the difference between corresponding mean bids for the St. Petersburg sample and the Orlando sample, (4) the results of the regression analysis. These aspects will be discussed in turn in the following sections. Response to the Questionnaire .' For St. Petersburg, 182 househol ds were contac:t:ed and 165 questi on naires were completed. Of these, 114 were usable!.' Therefore, 63% of the 182 household contacts produced usable questionnaires. For Orlando, 130 households were contacted and 120 questionnaires were completed. Of these, 62 were usable. Therefore, 48% of the 130 household contacts .produced usable questionnaires. The surveys rejected represented protest bidders. Protest bidding is demonstrated through respondents' answers reveal a failure to properly play the bidding game. answers indicated that a respondent did not consider the tradeoff process between paying (receiving) money and acquiring (foregoing) water in a specific use when making a bid, the bids were r'ejected. This generally took the form of zero bids with explana. t i OilS a f the i r unw ill i nglle>ss to pay money or accept compensa tl on. The fo 11 O\"i ng criteri a were employed to determi ne protest bi ds: (1) bidding all zeros: persons registering zeros for all thirty WTP and WTA questions were considered unwilling to play the bid game because it is highly unlikely that "true" responses would be such. For St. Petersburg this eliminated 12 surveys, or 7% of the original sample; in Orlando this eliminated 32 surveys, or 25% (see Table 1); (2) bidding lowest amounts for WTP and highest amounts for WTA: again it is unlikely that bids this consistent are "true." Also it is blatantly contradictory to register the highest value for WTA and the lowest value for WTP in the same use category. This criteria eliminated 1 survey in St. Petersburg and 7 surveys in Orlando. -34-

PAGE 42

Table l.--Criteria for rejecting surveys Criteria Bidding all zeros No consumption data Zero bid for WTP, highest bid for WTA Not willing to reduce or pay more Would pay anything Across the board compliance Minimum users Critical inconsistencies Not returned Total used Number of WTA only WTP oJlly Number with both games Number of bids: WTP equations WTA equations Number (%) of surveys unuseable St. Petersburg 182 -35-12 (7.0) 19 (10.0) 1 (0.5 ) 8 1 3 6 1 17 114 9 1 104 1,092 1,150 (4.4) (0.5) (2.0) (3.0) (0.5) (9.0) (63.0) (5.0) (0.5) (57.0) Orlando 130 32 (25.0) 28 (22.0) 7 (5.0) 1 (0.8) 62 (48.0) 14 (11. 0) 1 (0.8) 47 (36.0) 655 777

PAGE 43

(3) indicating an to payor reduce usage: persons so bidding were not open to the tradeoff design of the bid game. Their zero bids failed to consider the available choices. This eliminated 8 surveys in St. Petersburg and none in Orlando. (4) indicating a willingness to pay anything: the person is not considering the tradeoff. Also, it is not realistic for a person to be willing to pay "anything" since willingness is a function of capability to pay. This eliminated 1 survey in St. Petersburg and none in Orlando. (5) indicating total compliance: persons who indicated across the board compliance with water use reductions were rejected because it is unrealistic to assume that the major reductions 50% of all water use) could be enacted in lieu of even the lowest price. This eliminated 3 surveys in St. and none in Orlando. (6) indicating minimum usage: persons who felt that their current level of water consumption was at its minimum were not willing to consider reductions for any price. This eliminated 6 surveys in St. Petersburg and none in Orlando. (7) critical inconsistencies: as will be explained below, some bids were inconsistent and accepted in particular cases, however, when all bi ds anpeared to be cons is tently i ncons i stent with no apparent justification, the entire survey was eliminated. This eliminated 1 survey from each city. Two other reasons for surveys existed which were unrelated to protest bidding. First, 17 surveys were not completed and returned from the St. Petersburg enumerators. Second, water use consumption data was unavailable for a percentage of each sample. In St. Petersburg 12 surveys were from respondents who were not listed as customers of the St. Petersburg Public Utilities Company. In Orlando where utility data col lection was !lot possible because customel s I names weI e not PI I espon-. dents estimates of water use had to be utilized. For 28 surveys this estimate was not provided. Thus, for St. Petersburg this second crUerion elim inated 10% of the original sample and for Orlando it eliminated 22% of the original A complete set of thirty bids were usable from only a tiny portion of all respondents. Individual bids (from usable surveys) were eliminated for the following reasons: (1) bidding all zeros on one game eliminated the bidsfronT/that game. In st. Petersburg 9 respondents played only a WTA game and one played only a WTP game. In Orlando 14 respondents played only a WTA game and one played only a WTP game; (2) bids were eliminated if they were irrelevant to the household's water use level, e.g., if the household did 2 cycles of laundry per week, their bid for 3 cycles was considered irrelevant (for the most part, respondents replied "does not apply" to these questions); and ,)r -,)0-

PAGE 44

(3) bids were eliminated if the respondent was indecisive or if, the individual bid was otherwise rendered unusable by the aforementioned criteria which eliminated whole surveys. The decision toeliminate bids was difficult because the initial surey design did not account fpr an exploration of the zero bid or the inconsistent bid. Yet neither could be dismissed indiscriminately be cause valid reasons could exist for both. Some respondents bid posi-tive amounts for retaining lower levels of water use and zero for higher levels. It would appear to be inconsistent to, say, bid $3.00 to retain 10% of one1s total water use and zero to retain 50%. Yet some respondents stated that they would seek an alternative source of supply if high percentages of their water use were threatened with price increases. Others stated that they could not afford the increases at higher levels. Both reasons were considered valid zero bids by the theoretical definition of value measurement that this study adapts. Zero bids at lower levels of use were accepted because they usually indicated a willingness to comply with a reduction requirement only at that level. Unl ike the surveys indicating only total compliance, this explanation is realistic and reasonable. Bids at the other extreme were too many standard deviations from the mean bids calculated for each equation and constituted outliers. As bids tended to be uniformly conservative this criteria eliminated very few bids. As expected most of the extremely high bids were in the WTA game. In general, bids which could be explained and subsequently rendered reasonable were included in the sample. As stated the problem existed primarily with zero bids. Fortunately, with the St. Petersburg sample, bids were well explained and this aided in the process of determining whether or not a bid was usable. With the Orlando sample bids were not as well explained but proportionately fewer zero or inconsistent bids occurred. Comparing WTP and WTA Within Use Ca tegorl es St. Petersburg Sample Mean bids for each water use category increased as the amount of reduction avoided (or compensated for) increased (see Table 2). These results are consistent with the concept of a bid passing from the southwest quadrant through the origin into the northeast quadrant of Figures 2 and 3 (Chapter II). Within the St. Petersburg sample the mean WTA bids for the total househo1cl water (w) category averaged 220% of their corresponding lHP bids. For the bath/shower (BS) category the mean bids averaged 196% of the mean WTP bids. For the toilet flushing category the mean WTA bids averaged 113% of the corresponding mean WTP bids. These results are consistent with a hypothesis that WTA will normally be greater than toJTP. -37-

PAGE 45

Table 2.--Hean bids Number of Percentage bidding Use observations Current price Hean bid current price HTP IITA HTP VJTA IITP WTA St. Petersburg TF 96 104 1. 90 2.80 3.09 55 47 TF 97 101 3.84 5.46 5.80 54 49 TF 92 100 5.76 8.42 10.38 61 48 BS 99 108 1.56 1. 76 3.15 54 34 BS 97 107 2.34 2.87 5.32 53 36 I BS 102 105 3.12 4.29 8.97 59 30 w 0:; I W 103 107 1. 98 2.41 5.16 56 30 I'J 86 98 5.94 6.97 15.10 63 38 H 95 101 9.90 11.81 26.53 45 30 Orlando TF 51 60 2.16 2.00 4.34 31 33 TF 51 59 4.32 4.30 9.07 35 32 TF 45 60 6.36 6.71 15.31 44 32 BS 51 64 1. 75 2.16 5.95 29 17 BS 51 59 2.63 3.06 8.91 45 17 BS 50 60 3.50 4.34 13.60 42 17 H 51 52 2.22 2.52 8.23 31 15 H 53 59 6.66 6.58 21.67 32 15 W 53 59 11.10 10.66 46.10 30 15 TF Toilet flushing; BS Bathing/sh wering; W = Total household water use

PAGE 46

The Orlando Sample Mean bids in the Orlando sample also increased as the amount of reduction avoided (or compensated for) increased (Figure 2). the Orlando sample, mean InA bids for the total water uS.e category averaged 362% of the corresponding mean WTP bi ds. For the bath and shower category, the mean IHA bi ds averaged 293% of the correspondi ng mean \HP bids. For the toilet flushing category the mean WTj\ bids averaged 219% of the mean WTP bids. ri n9 IHP and WTA Between the Two Samples It was hypothesized that the bids, for corresponding categories, would be higher in St. Petersburg because those respondents are familiar with water availability problems (Appendix II). However the sampled Orlando residents had a much higher mean income than the St. Petersburg group and this would tend to offset the difference in bids between the groups. \iJith respect to the total water use category (w), Inp, mean bids from St. Petersburg were roughly the same as for Orlando. However. fior WTA. the Orlando mean bids averaged 158% of the corresponding St. Petersburg mean bids (Table 2). l'Jith resrect to the bathing and showering category, tnp mean bids from Orlando averaged 110% of those from St. Petersburg. But WTA mean bids from Orlando were about 169% of those from St. Petersburg. vJith respect to the toilet flushing category of use, vJTP bids from St. Petersburg averaged about 131% of those from Orlando. However, WTA mean bids from Orlando averaged about 147% of those from St. Petersburg. In general, then, WTP bids were roughly comparable between the two s amp 1 e Results of Regression Analysis l'LLlU.!1 q t:!e s s ___ qua t ion For the total water use category, the WTP dependent variables are WPWAT10, WPWAT30, and WPWAT50; for bath/showering use, WPBS20, WPBS30 and WPBS40; and for toilet flushing use, WPTF1, WPTF2, WPTF3. Each WTP dependent variable represents the amount the consumer is willing to pay through his monthly water bill to maintain current levels of household water for specific water-using activities, rather than experience a specified reduction in water use. The independent variables in the WTP equations are NUMRESPH, INCOME, IIAVE\JELl, BELUSHRT, DECON50 and the quantity variables, AVGCON10, AVGCON30, AVGCON50, TOIlQ1, TOILQ2. TOIlQ3, BATHSQ20, BATHSQ30, BATHSQ40. each of \t,dl'ich denotes the quantity reduction to which the bid pertains. The hypothesized relationship between these variables is discussed in Chapter -39-

PAGE 47

Four. Because of a high correlation between the variable, NUMRESPH, symbolizing the number of persons residing in a household, and the quantity variables, equations for both possibilities were presented. Explanation of Results of WTP Regressions The results of the regression analys{s for WTP are summarized in Tables 3 and 4. Income was expected to be highly significant because it would appear intuitively to have a substantial influence on a consumer's willingness to pay. This variable was significant, in fact, in five equations in the St. Petersburg sample and in nine equations in the Orlando sample and in neither case did income play an important role in the WTP equations for total household water use. In St. Petersburg, income was significant in both WPBS20 and both WPBS40 equations. In Orlando income was significant in one WPTFl equation, in both WTPF2 equations and in all WPBS equations. It appeared with the anticipated sign (positive) in all but two equations. The differences in the two samples may be attributed to the differences in mean income levels between the two city samples. In the St. Petersburg sample the mean income approaches $10,000, while in the Orlando sample the mean income level approaches $20,000. NUMRESPH was also expected to be a highly significant variable. In St. Petersburg it appeared as significant in all equations in which it appears except WPBS40. In Orlando it did not appear as significant in any equation. The St. Petersburg sample had a higher mean household size than in Orlando and this may account for the difference, however, the differential is not large (2.5 in Orlando and 2.9 in St. Petersburg). NUMRESPH appeared with its expected sign (positive) in all St. Petersburg equations and in all but two, one WPWAT30 and in one WPWAT50, in the Orl ando sampl e. Originally HAVEWELL and KINDSYS were included in the total water use equations. KINDSYS was eliminated because it was not significant at any test level in any of the equations. HAVHJELL was retained but was significant in only one equation, WPWAT30, in the Orlando sample, where it appeared with a positive sign (its hypothesized sign was negative). In four out of six Orlando equations the HAVEWELL sign was positive. In the St. Petersburg sample HAVEWELL appeared consistently with the hypothesized sign but was not significant in any equation. This variable was difficult to interpret and this may be the reason for its performance. Since most sampled households in St. Petersburg had a well its effect on the willingness to pay should be negative since this implies the existence of a substitute water system. However, this effect is apparently not SUbstantial for the sample. On the other hand, the Orlando households, for the most part did not have a well and that fact had a positive effect on the willingness to pay for water. As with the St. Petersburg sample the presence of a well did not appear to have a substantial effect on the WTP. The quantity variables were expected to have a negative effect on the WTP because were entered into the equation as negative values, re present; ng quantity reducti ons Mctua lly the amount of water a household used was hypothesized to be a posi'tive influence on the willingness to pay). They appeared in all equations (both samples) with the hypothesized -40-

PAGE 48

Table 3.--IHP regression results for St. Petersbul'g -----WTP model NLJllRESPH INCOflE HAVEWEll BElWSHRT DECON50 AVGCON10 AVGCON30 AVGCON50 TOIlQl TOIlQ2 TOW)3 BIITIISQ20 BATHSQ30 BATHSQ40 R2 WPWATlO (a) .180 .122 .299 -.064 -.252 ( .185) ( .164) ( .447) ( .607) ( .652) .028 HPWAT10 (b) .119 -.305 -.043 -.203 -.867 ( .163) (.442) ( .603) ( .654) ( .736) .031 \OlPWA no (a) .333 .377 -.689 -.049 -.362 ( .247)* ( .219)** (.595) ( .808) ( .U6S) .075 HP,lA no (b) .380 -.723 .0004 -.306 -.464 ( .218) ( .588) ( .804) ( .873) ( .327) .076 WPWIIT50 (a) 1.20 .S37 -.107 2.30 -3.13 ( .633)* ( .562) ( 1.53) (2.07)* (2.23)* I .099 WPWAT50 (b) .666 -1.52 3.16 -3.39 -.416 ( .56B) (1.53) (2.09)* (2.27) (.511 ) .075 I-1PTF1 (a) .624 .039 -.008 .755 (.198)*** ( .179) ( .666) ( .714) '.093 (b) .016 -.022 +.754 -.003 ( .179) ( .663) (.709) ( .001)*** .102 WPTF2 (a) 1.37 .202 1.23 .035 (.363)*** ( .329) (1.23) (1.31 ) .131 14PTF2 (b) .247 -1.21 .020 -.004 (.329) ( 1.22) (1.31 ) (.0009)*** .138 l,PTF3 (a) 1.61 .234 2.30 -.812 ( .670)*** ( .610) (2.25) (2.41) .073 WPTF3 (b) .179 2.27 -.823 -.003 ( .610) (2.25) (2.41) (.001)*1* .071 WPBS20 (a) .152 .161 -.408 -.319 ( .114)* ( .103)* (.385) ( .412) .050 IIPBS20 (b) .166 -.390 .275 -.0004 ( .104)* (.386) (-':411) (.004) .043 l4PBS30 (a) .395 .215 .553 -.597 .081 (.195) ( .177) (.659) (.706) \JPBS30 (b) .206 .545 -.685 -.OOl .087 ( .176) ( .657) ( .699) (.0005)** WPBS40 (a) .297 .456 .766 -.819 .095 ( .232) ( .210)** (.784) ( .8'39) l4PBS40 (b) .430 .712 -.860 -.0008 (.208)** ( .776) ( .827) ("0004 )* Significant at .10 level. ** Significant at .05 level. *** Significant at .01 level. (a)Model with NUMRESPH. (b)Model with quantity variable.

PAGE 49

Table 4.--IJTP regression results for Orlando .------------------HTP TOILQ2 model NUI\RESPH I NCOI,lE BELWSIIRT DECON 0 AVGCONlO AVGCON30 AVGCOri40 TOILQ1 BATHSQ20 BATHSI)30 BATHSQ40 R2 HPHAT10 (a) .061 .076 .233 .161 -.791 ( .312) (.166 ) (.793) (.790) (.8T 0''', IJPWAT10 (b) .086 .308 .230 -.79 .109 ( .154) ( .803) ( .827) (.83 ) ( .428) .02, 1,m-IAnO (a) -.167 .260 1.57 -.602 -3.68 ( .542) ( .287) (1.37) (1.37) (1.47r** .129 HPI'IAT30 (b) .194 1.93) .38 -3.14 -.333 ( .263) (1.37)* (1.41 ) (1.42 ** ( .243)* .15,_ \ HPIJAT50 (a) -.697 .160 1. 34 -1.77 -4.21 (1,06) ( .560) (2.68) (2.67) (2.El7r .046 WPHAT50 (b) -.049 2.15 .968 -2.82 -.418 (.514 ) (2.68) (2.76) (2.78 ( .285) .074 IJPTFl (a) ,077 .232 .784 -1.13 ( .249) ( .132) ( .625) (.6T* .106 IlPTF1 (b) .231 .779 -1.13 .0004 I ( .133)** ( .625) (.67 ) ( .ool) .106 IJPTF2 (a) .100 .456 .550 -1.34 I ( .445) ( .235)** (1.12) (1 .21 .073 IJPTF2 (b) .428 .528 -1.24 ) -.00008 ( .239)** ( 1.12) (1.21 ) ( .001) .072 HPTF3 (a) -.222 .245 -1.28 -.753 (.762) ( .403) ( 1.92) (2.07) .018 WPTF3 (b) .176 -1.31 .51 -.0002 ( .409) ( 1.92) (2.0n (.001 )1 .018 IJPBS20 (a) .122 .196 .671 i ( .249) ( .132)* ( .626) .100 IlPBS20 (b) .199 .679 -1.19 -.0005 ( .132) ( .626) I (.001 ) ,09E lJPBS30 (a) .187 .328 1.06 ( .326) (.173)** ( .819* .141 IJPBS30 (b) .338 1.07 -1.8 .0004 ( .173) ( .820) (.3 nl (.001 ) .1:: HPllS40 (a) .027 .386 1.59 -2.2, ( .484) (.2'56)** (1.22)* (1.3 )** .09 HPBS40 (b) ,362 ]:59 -2.21 .0003 ( .256)** (1.21) (1.2 )** (.001 ) ,092 --------------* Significant at .10 level. .* Significant at .05 level. *** Significant at .01 level. (a)Mode1 with NUMRESPH. (b)Mode1 with quantity variable.

PAGE 50

sign and were significant in all St. Petersburg equations in which they appeared except for WPWAnO, and WPBS20. In Orlando they were significant in WP\tJAT30 and only. The conservation variables performed differently than expected. BELWSHRT, the variable which registered belief in a water shortage was anticipated to be negative and appeared so in seven equations in both the "St. Petersburg and Orlando samples. In eleven BELWSHRT was positive and this included the two cases in the St. Petersburg sample and the four cases in the Orlando sample in which the variable was significant. In Orlando BELWSHRT was significant in both WPBS30 and in both WPBS40 equations. In St. Petersburg, BELWSHRT was signi-f i car;1 till bo th WPWAT50 equa ti ons DECON50 appeared as negative (its hypothesized sign) in twelve cases and as posttive in 6 cases in the St. Petersburg sample. It was significant in the WPWAT50 equation where BELWSHRT was also significant. In the Orlando sample DECON50 had the expected sign in all equations except one, WPTF2, and was significant in eleven equations, both WPWAT30 equations, WPWAT50, both WPTFl equations, and all WPBS equations. Willingness to Accept Equations In the WTA group of equations the dependent variables are WAWAT, WATF, and WABS which were creating from combining the WTA data sets of the total t'Jater \lJTA bid group, WMJATlO, HAlJAnO, the toilet flushing bid group, WATF1, WATF2, WATF3; and the bath/showering WTA bid group, WABS20, WABS30, WABS40. The independent variables are the same as in the WTP estimation with the exception of the quantity variables. The quantity variables, AVGCONT, TOILT and BATHST were created from the combination of their respective separate WTP data sets. Description of Results of IHA Regression The results of regression anal'ysis for lHA equations are summarized in Tables 5 and 6. In the Orlando sample, income has the expected sign (positive) in all but one of the WATF and WABS equations. It was si"gnificant in only one equation for the Orlando sample. In the St. Petersburg sample income was positive and highly significant in all equations. NUMRESPH appeared in three equations for Orlando and each time with a negative sign (its hypotilesized sign was positive). It was not significant in any equation which is similar to the results for this variable in the WTP equations for Orlando. For the St. Petersburg sample, NUMRE$PH was significant in two of the three equations in it appeared and it was positive, as expected. in all three cases. The quantity variables in the Orlando sample had the expected sign (negative) and significancein the three Orlando equations in which they appeared. Likewise they were negative and significant for all the relevant equations in the St. Petersburg sample. -43-

PAGE 51

Table 5.--WTA regression results for St. Pete sburg \HA model NUMRESPH L BEU'JSHRT DECON50 AVGCONT TOILT BATHST R2 (a) 1.14 1. 70 -1 .99 2.74 .035 .038 ( .819) ( 732) ***. (1. 95) (2.69) (2.96) WAWAT (b) 1.4b -.723 2.25 2.20 -5.06 .184 (.663)*** (1.77) (2.46) (2.72) (.662)*** (a) .679 .640 1.23 .459 .053 ( .281) *** ( .257)*** (.950) (1 .04) WATF (b) .573 1.03 -.003 .127 (.245)** ( .909) ( .0005) 1*** \lJABS (a) .143 .601 1.04 .113 .030 (.260) (.238)*** ( .880) ( .964) WABS (b) .495 .855 .278 -.002 ( .233) ( .860) (.942) (.0006)*** Significant at .10 level. ** Significant at .05 level. *** Significant at .01 level. (a)Mode1 with NUMRESPH. .. (b)Model with quantity variable.

PAGE 52

::" :n I Table 6.--WTA regression results for \HA model I (a) -1 .34 1.67 -3.44 (2.36) (1 .25) (5.99) (b) 1.07 .065 (1 .09) (5.69) \
PAGE 53

The HAVEWELL variable in the Orlando sample was not significant in either equation in which it appeared and it appeared as a negative influence in one WAWAT equation and as a positive influence in the other equation. In the St. Petersburg sample HAVEWELL was not significant in either of the total water WTA equations in which it appeared, however, it did appear with the expected sign (negative) in both cases. The conservation variables for the WTA equations for the Orlando sample performed very much as they did for the HTP equations. was consistently a positive variable and it was significant in all equations. DECON50 was consistently a negative variable and was signi-fi cant in three equati ons, one WAvJAT, and vJABS equation. In each of these equations NUMRESPH, rather than the corresponding quantity variable, was entered into the equation. In the St. Petersburg sample BELWSHRT and DECON50 had positive signs in all equations and wereconsistently insignificant throughout. -46-

PAGE 54

CHAPTER VI SUMMARY AND DISCUSSION This study adapted a non-market valuation technique and applied it to the measurement of consumers valuations of water in residential uses. Representative samples of single-family residential water dustomers drawn from two major cities in Florida, and data for the study \lIas ac quired through the use of personal interviews following a specially de signed questionnaire. The iterative questioning procedure of the survey was structured to elicit consumers maximum willingness to pay to avoid specified reductions in residential water use, and, in addition, to elicit the minimum compensation necessary to induce consumers to voluntarily accept specified reductions in residential water use. Multiple regression analysis was used to test hypotheses concerning the role of selected independent variables as determinants of consumers valuations of water in residential uses. The results must be considered preliminary, since the initial analysis did not exhaust all possibilities for specifying the regression equations. Several conclusions can be drawn from the study: (1) response of consumers to the contingent market, iterative bidding, context of the questionnaire indicated that a substantial percentage of the respondents understood the purpose of the questionnaire and attempted to honestly assess their willingness to pay and willingness to accept, (2) mean bids for both samples, both measures (WTP and WTA) and all water use categories consistently demonstrated higher bids for greater contingent reductions in water use--a pattern which would produce bid curves passing from the southeast quadrant, through the origin into the northeast quadrant of a graph depicting a total value (or bid) curve, ( 3) ( 4) willingness to accept (WTA) bids consistently exceeded willingness to pay (WTP) bids for corresponding use reduction contingencies, and regression results suggest that a sUbstantial portion of the hypothesized functional relationships between observed bids and selected explanatory variables were, in fact, statistically significant. Additional investigation is needed in several areas: (1) there is a need to further explore the ability of consumers to formulate valuations of water in household uses when such considerations are being made for the first time. Can "preparation" be provided without biasing consumers percepti ons? -47-

PAGE 55

(2) The pptential for other forms of bias must be carefully examined. While starting point bias was not expected to be a factor, it may well have existed. Art alternated high and low starting point for bids within each sample could be applied to test for the presence of starting point bias. (3) It was hypothesized that the variables which influence the willingness to pay for water would be the same as those which influence the amount of use. The theoretical underpinnings of these hypotheses need to be developed with greater care. (4) Empirical estimation of a bid function must require a zero intercept in order that the estimated function retain the properties inherent in its definition. -48-

PAGE 56

APPENDIX I THE QUESTIONNAIRE ,.

PAGE 57

Part One: Socioecomonic Data 1. How many persons presently live in this household? 2. How many members of your household, including yourself, are in each age group? -----0-10 years _____ 11-20 years _____ 21 -40 years _____ 41-60 years over 60 years ----3. Do you own or rent your home? __ own rent -4. If own, what is the market value of your property? If rent, ----what is your rent/month? 5. How long have you lived in the St. Petersburg area? _---"ye a r s 6. How old is your home? 7. Place of previous residence? city state -------------------8. Do you live in Florida year-round? ____ ----'yes no ------If no, what months do you spend in Florida? from to --------------9. Is the head of tile houseltol d elllpl oyed, 1 e tired 01 unelllployed? _____ unemployed retired ------unemployed ---If employed, what is his/her occupation? -------------------------10. What are the occupations of other working members of the household? 11. What is the highest grade of school completed by the head of the household? less than high school ---------------------high school some college or technical --------______ 8. A. or B. S. Masters or Ph.D. --------

PAGE 58

12. What is the number of overnight guests in your home per year? ____ 0-10 ____ 11-20 ____ 21-30 over 30 ---Their average length of stay is ____ day ____ 2-7 days 1-2 weeks ---2-4 weeks ___ --c over 1 month ---2 13. How many of the following water using appliances does your household have? bathtubs ---toil ets ----____ w.ashing machines dishwashers ----____ .showers (wi th and wi thout bathtubs) sinks ---_____ garbage disposals (hot tubs, jacuzzi, etc.) ---14. Do you have a swimming pool? ___ ------'ye s ____ no 15. Do you use bottled water? ___ ------'yes ____ no If yes, how many gallons per month? ga 11 ons ----' -51-

PAGE 59

16. Do you have a home water softener? ____ yes no ----17. Do you have a septic tank? ____ yes ____ no 18. Do you have a private well? no ----____ yes ----yes, but only for lawn sprinkling 19. What kind of system do you use to water your lawn or garden? none -------hose and sprinkler(s) ---automatic sprinkler system 20. Estimate the size of your property: Less than 1/10 acre ---1/10 acre 1/5 acre ----____ 1/4 aC.re 1/2 acre ----3/4 acre ---1 acre or more ----21. How often do you wash your car(s) at home? less than once a month ----1 to 3 times a month ---more than 3 times a month ----22. IIJhat is your total monthly water bill in dollars: for water only ----for sewer/wastewater only -----52-3

PAGE 60

23. What is an estimate of how much water your household uses each month? ____ ga 11 ons do not know ---24. Indicate in which period your water bill is the highest: ----January ____ Apri 1 -June ----July September October -December ._--25. is the combined income of your household? less than 5,000. 5,000 to 9,999. 10,000 to 1Il,999. 15,000 to 19,999. 20,000 to 24,999. 25,000 to 29,999. 30,000 to 34,999. 35,000 to 39,999. over -----53-4

PAGE 61

5 The following tables provide a water use estimate for your household. Please indicate the frequency for each use category (how many times it is done) in column 2. Column 3 pr6vides approximate amounts of water used by each item. r 2 3 4 Item Use per week Gallons per item Total a. Automatic dishwasher 1/ b. Hand dishwashing 8 c. Clothes washing (cycles or loads) 50 d. Hours of lawn watering (see below) 500 e. Garbage disposal (2 minute use period) 6 Subtotal Use per day f. Shower 33 g. Bath 30 Subtota 1 No. of persons Gallons per day h. Toil et fl us hi ng 32 i. Cooking and drinking 3 Subtotal Total Please indicate the months within which you water your lawn at least once a week: -54-

PAGE 62

(Note to Surveyors: Part Three Iterative Bid THIS SCENARIO IS HYPOTHETICAL! NO SUCH REQUIREMENTS ARE PENDING FOR CONSUMERS). Supplying water in the future may require higher than present costs. Imagine that a water planning agency has to pay higher costs to provide enough water, and is interested in finding out their consumers' willingness to pay these higher costs for current use levels. 6 Please indicate your willingness to pay through changes in your monthly for water in specific uses. Imagine that requirements are set for water use reductions but you may avoid compliance if you are willing to pay to do so. (The first amount presented is the current price for the amount of water given). A. Lawn watering 2. 3. B. 1. If there was a requirement to reduce lawn watering by 1/2 hour a week, how much additional money would you willing to pay to avoid complying with this requirement? $1.98 $3 $4 $6 $8 $12 $16 $24 other To avoid reducing lawn watering by 1 hour? $3.96 $6 $8 $12 $16 $24 $32 $48 other To avoid reducing lawn watering by 3 hours? $11 .88 $16 $24 $32 $48 $72 .:.: .$95 $143 .. """f": other l v', '", ... .. Toilet Flushing 1. If there was a requirrement to reduce toilet flushing by 1 time per 2erson per day, how much additional money would you be willing to pay to avoid complying with this requirement? $1.90 $3 $4 $6 $8 $11 $15 $23 other 2. To avoid reducing toilet flushing by 2 times per person? #3.84 $6 $8 $11 $15 $23 $31 $46 other -55-

PAGE 63

7 3. To avoid reducing toilet flushing by 3 times per person per day? $5.76 $9 $12 $18 $23 $35 $46 $70 other C. Bath/Shower D. 1. If there was a requirement to reduce bath/shower usage" in the house hold by 20%, how much additional money would you be willing to pay to avoid complying with this requirement? (a 20% decrease is equivalent to a person taking six showers a week instead of seven, or reducing a 10 minute shower to 8 minutes). $1 .56 $3 $5 $6 $9 $12 $19 other 2. To avoid reducing bath/shower usage by 30%? $2.34 $4 $5 $7 $9 $14 $19 $28 other 3. To avoid reducing bath/shower usage by 40%? #3.12 $5 $6 $9 $12 $19 $25 $37 other Cl othes ngs l. If there was a requirement to reduce clothes laundry use by 2 cycles or loads per week, how much additional money would you be willing to pay to avoid complying with this requirement? $.79 $2 $3 $5 $6 $9 other 2. To avoid reducing laundry use by 3 cycles or loads per week? $1 .19 $2 $4 $5 $8 $10 $14 other 3. To avoid reducing laundry use by 4 cycles or loads per week? $1 .58 $3 $4 $6 $10 $13 $19 other -56-

PAGE 64

E. What are you willing to pay to avoid a reducti on of 10% of all water usage for your household? $1.98 $3 $4 $5 $8 $12 $16 $24 $36 other 2. To avoid a reduction of 30% of all water usage for your household? $5.94 $9 $12 $18 $24 $36 $48 $72 $96 other 3. To avoid a reduction of 50% of all water .usage fon your household? $9.90 $15 $20 $30 $40 $60 $80 $120 $160 other If you had decided to comply with requirements of water use reduction what amounts would you consider necessary to compensate you for the incon veniences which you incur when you make these reductions? Please indicate the minimum amount of money which you would accept, through reductions in your monthly water bill, for specific water uses. (The first amount presented is the current price for the amount of water given). A. Toilet Flushing l. If you were required to reduce toilet flushing 1 time per person per what amount would fully compensate you for your loss? $1 .90 .$3 $4 $7 $9 $13 $17 $26 other 2. For a loss of 2 times per person per day? $3.84 $6 $8 $11 $15 $23 $31 $46 other 3 For a loss of 3 times per person per day? $5.76 $9 $12 $18 $23 $35 $46 $70 other B. Lawn Hatering l. If you were required to reduce lawn watering by 1/2 hour a week, what amount would full y compensate you for your loss? $1 .98 $3 $4 $6 $8 $12 $16 $24 other 2. For a loss of 1 hour a v.Jeek? $3.96 $6 $8 $12 $15 $24 $32 $46 other -578

PAGE 65

B. (continued) 3. For a loss of 3 hours a week? $11.88 $16 $24 $32 $48 $72 $95 : $143 other C. Bath/Shower 1. If you were required to reduce bath/shower usage in the household by 20%, what amount would fully compensate you for your loss? 2. 3. D. 1. 2. 3. E. (a 20% decrease is equivalent to a person taking six showers a week instead of seven; or reducing a 10 minute shower to 8 minutes). $1.56 $3 $5 $6 $9 $12 $19 other For a loss of 30% of your current bath/shower use? $2.34 $4 $5 $7 $9 $14 $19 $28 other For a loss of 40% of your current bath/shower use? $3.12 $5 $6 $9 $12 $19 $25 $37 other Clothes Washings I f you ,,,ere requi red to reduce your 1 aundry use to 2 cycles or loads a week, what amount would compensate you for your loss? $.'79 $2 $3 $4 $5 $6 $8 $10 other For a loss of 3 cycles or loads a week? $1.19 $2 $4 $5 $8 $10 $Hi other For a loss of 4 cycles or loads a week? $1.53 $3 $4 $6 $10 $13 $19 other If you were required to reduce all of your household water usage by 10% what amount would fully compensate you for your loss? $1.98 $3 $4 $5 $8 $12 $16 $24 $36 other -58-9

PAGE 66

10 E. (continued) 2. For a loss of 30% of all your household water usage? $5.94 $9 $12 $18 $24 $36 $48 $/12 $96 other 3. For a loss of 50% of all your household water usage? $9.90 $15 $20 $30 $40 $60 $80 $120 $160 other -59-

PAGE 67

Part four: CAS 11 1. During the recent years, the media has been reporting the existence of water shortages in many areas of the United States. These shortages place pressure on urban water supply systems which provide water for residential home use. Do you believe a water shortage exists in the St. Petersburg area? No -----Yes -----If yes, how set'ious would you say tile water stlortage is in St. Petersburg at present? -----very serious -----fairly serious not serious -----do not know -----2. Whether or not you believe that there currently is a water shortage in your area, do you believe that there will be such a problem in the future? Please indicate by what year you think a water shortage may become a problem or continue to be a problem? 1981 ----1985 -----2boo or beyond' _____ never 3. If you believe a water shortage exists or will exist in St. Petersburg, to what extent do you believe each of the following to be a cause of the problem? (a) Natural causes: 1) lakes, rivers drying up 2) lack of ralnfa11 3) depleting groundwater sources (includes salt water intrusion). (b) tan made causes: 1 ) growth of population in St. Petersburg area 2) growth of population in other areas 3) increased water usage per person due to increased use of waterusing home appliances (i .e. hot tubs) Great Extent -60-Some Extent Not at all

PAGE 68

3. (continued) 4) increased water due to wasteful practices 5) the heavy use of water for com-merci a 1 (industry, mining) 6) the heavy use of water for agricul-tural act; vity 7) poll uti on of water suppl ies (c) Institutional causes: 1) Water utilities have not taken the neces sary steps to provide for enough water for residential use. 12 Great Some Not at Extent Extent all --4. In some areas throughout the country the public has been asked tp practice water conservation in order that water demand may be Water conservation can be achieved in numerous ways. Following is a list of possible conservation practices. Please indicate to what extent you believe each practice may result in conserving water: Great Some Not at Extent Extent all a) Filling the bathtub only one-fourth full b) Turning off the water while brushing your teeth and shaving c) Taking showers in a shorter amount of time d) Using the dishwasher only when it is full e) Cutting back on lawn sprinkling time f) Cutting back on times the toilet is flushed g) Capturing and reusing shower water for non drinkable use (e.g., plant watering) -61-

PAGE 69

4. (continued) h) Placing a brick in the toilet tank i) Use of the following water saving devices: 1 ) pressurized showerheads 2) suds-saving washing machine 3) shallow trap water saving toilets 4) chemical toi 1 ets 5) low volume dishwasher 6) dual flush toilet tank 7) washer1ess faucets Great Extent Great Extent Some Extent Some Extent Not at all Not at all 13 Are not familiar with 5. Which of the following conservation practices are you actually using to reduce the amount of water your household uses? Please indicate whether you are making large efforts, medium efforts, small efforts, or no effort in each of the areas listed below: a) turning off the water while brusing your b) teeth and shaving using the dishwasher only when it is full c) taking showers in a shorter amount of time d) cutting back on lawn sprinkling time e) filling the bathtub only one-fourth full f) cutting back on times the toilet is flushed g) capturing and reusing shower water for non-drinkable use (e.g., plant watering) h) placing a brick in the toilet tank i ) use of the following water saving devices: 1) pressurized showerheads 2) suds-saving washing machines 3) shallow trap water saving toilet 4) chemi ca 1 toil et 5) low volume dishwasher G) dual flush toilet tank 7) washerless faucets Large Effort um Effort Yes Small Effort No No Does Not Effort Apply

PAGE 70

6. In some areas the price of water (per 1,000 gallons or per cubic feet) has been increased with the belief that water usage will decrease as a result. In your area has the price of water/sewer recently (last few years) been increased? No ---Yes ----14 If yes, has this price increase led to a reduction of water usage in your household? No ---Yes 7. Do you anticipate an increase in the price of water/sewer for your area in the near future (within the year)? No ---Yes ----3. Do you think that your household would decrease its water usage if the price of was raised by 50%? No ---Yes ----If yes, by approximately what percent would your household decrease its water usage: 0-10% ---11-20% ----21-30% ---over 30% 9. Do you think that your household would decrease its water usage if the price of water was raised by 100%? No ---Yes ----If yes, by approximately what percent would your household decrease water usage? 0-10% ---11-20% ----21-30% ----over 30% ----10. If you would decrease your water usage would you do so with the installation of water saving devides? Yes, I would consider immediate installment. -------Yes, but only when I needed a new toilet, sink, etc. No. ----63-

PAGE 71

APPENDIX II THE SM1PLE This study required the drawing of two separate samples, one from the city of Orlando, Florida and one from the city of St. Petersburg, Florida. The qualifications for the population were: a) it was confined to the geographic areas under the jurisdiction of the water utility which served the largest number of customers. The reasons for this were the need to deal with one rate and the need to represent as large an area as possible within the city itself. The two water companies used, Orlando Utilities CommissiOlI (DUe) ill Orlalldo alld SL. Petef'sbUY'g Publ ic Utilities in St. Petersburg, both had one residential rate structure which extended to the city limits (see Table 7); b) it was confined to single family dwellings with 5/8 inch water meters. This was because apartment residents often do not pay their own water bills and because meters of different size follow differing rate structures. Although the constrDction of multi-unit dwellings is increasing, the majority of structures in both Orlando and St. Petersburg ane single-family units. This appendix contains two sections. The first section provides an introduction to the communities selected, including,a brief description of their water supply systems, the status of their water reserve capabilities, and a discussion of the demographic characteristics which influence the demand for residential water in both communities. In the second section of this appendix, a detailed review of the sampli!ng procedure is presented. Community Characteristics St. Petersburg (Pinellas County) Demographic characteristics Pinellas county is the most densely populated of all Florida cOl!nties. Current density in developed residential areas of Pinellas county ;s 14 persons per acre or 5.83 units per residential acre (Board of County Commissioners,1973). The density for the total county is 3.6 persons per acre with a unit density of 1.5 units per acre. If the county continues to grow at the
PAGE 72

Table 7.--Water rate schedules for sampled cities. Service/Price Water Wastewater ST. PETERSBURG WATER RATE SCHEDULE (for 5/8 inch meter within city limits) Base Rate $1. 75 $2.95 Price per 1,000 gallons $.72 $.79* Maximum charge for wastewater to single family residences is $14.85. ORLANDO WATER RATE SCHEDULE (for 5/8 inch meter within city limits) Service/Price Base Rate Cost per 1,000 gallons First ,. 000 2 000 J 00, 000 over ] 00, 000 Water $1.88 $.44 $.37 Wastewater $1. 75 $1.45 $1.45 $1.45 -65-

PAGE 73

The characteristics of the incomtng population mirror the characteristics of the general population (Bureau of Economic and Business Research, 1980). In Pinellas County the percentage of persons in specific gnoups have increasingly been weighed toward the older categories. In 1950the percentage of the population over age 65 was 19%; in 1975 that percentage nearly doubled to 34%. The 1 a rge component of older persons ; n Pi!ne 11 as county accounts for several other demographic attributes. Median education levels tend to be low, according to the 1970 census, 77% of the population completed equal to or less than a high school education (Pinellas Planning Council., 1978). Income averages had kept pace with the national averages until 1974 but have been grow1ng more slowly 1n the last t1ve years. Per cap1ta income 1n 1975 for St. Petersburg was $5,817 (Pinellas Planning Council, 1977). Household income which averaged $12,395 in 1976 for the county is expected to increase by 48% to $18,318 by 1981. The older population also helps explain the average household size and type of housing structure statistics. Both a rising death rate and a high rate of immigrating retirees contribute to an average household size which is below the state average by 16% and the national average by 22%. In 1950 the average household size in Pinellas county was 2.71, in 1975 it was 2.43. There is a hi gh percentage of one and two person households with one person households alone accounting for 11% of all county households. Other trends which contribute to this are the declining birth rate, increasing divorce rates and the tendency for young adults also to form single-person households. There has been a more intense development of mUlti-unit dwellings to accomodate the demand for single person units. In 1970 in the city of St. Petersburg alone, the construction of multi-unit dwellings exceeded the single-unit dwellings by 228%. In 1975 this trend was reversed but again in 1979 the multi-unit percentage exceeded the single-unit by In 1977 the inventory of housing indicated that single-unit accounted for 56%, multi-unit residences accounted for 32%, and mobile homes accounted for 12% of the total hOllsing strllctllres The average facility size of apartments tends to be small, 63% of the total number have one bathroom (Pinellas Planning Council, 1978). Water supply and"demand characteristics St. Petersburg receives its water from its own municipal system which draws from the Floridan aquifer through well fields in Pasco and Hillsborough counties. The number of wells has increased from 1963 to 35 in 1980.1 1 Average daily pumpage rates have been increasing from 20 million gallons in 1956 to 35 million gallons in 1975. Demands upon the system have been growing, serving a population of 250,000 in 1970 and 283,000 in 1980. Per capita daily use (GPCD) has risen from 130 gallons in the late sixties to llTelephone communication with Dean Hughes, st. Petersburg Public Utilities, St. Petersburg, Florida, November 6, 1980. -66-

PAGE 74

nearly 140 gallons in 1975 (Healy, 1977). Projections for Pinellas county indicate that by the year 2000 the GPCD will increase to 151, with the country water systems requiring more than twice the amount of water that they now distribute (Southwest Florida Water Management District, 1978). The months of highest demand have been consistently April-May and the low periods are December-February (Healy, 1977). The remainder of Pinellas county is served my the Pinellas County Water System. Together with the St. Petersburg Public Utilities these systems have had shortage roblems through the 1970's. The principal causes for;, the s a ci t a' experienced a critical lack of rainfall (Parker, 1975). Consequently the aquifer was losing its chief source of recharging water. This resulted in 2) excessive pumping of the available groundwater reserves further lowering the area's water table. Excessive pumping was the natural result of growing demand pitted against a depleting supply. Rising demands were caused by a large immigrating population and also the prodigious water requirements of the citrus and phosphate industries. The next problem in the cycle is 3) the lowering of fresh water preserve yielded to salt water intrusion into groundwater sources. In 1973 the Southwest Florida Water Management District (SWFWMD) regulated St. Petersburg's wells to provide more water to the Pinellas County system until the latter could augment their capacity. Both systems were severely taxed and an official water crisis with restrictions was de clared (Board of County Commissioners, 1973). Since then the Pinellas County Water System has increased its number of wells from 45 to 64. St. Petersburg has increased their total supply to 35 production wells. Orl ando (Orange County) Demographic characteristics Orlando which is located in Orange county has a high density of 400 persons per square mile. Its suburban county, Seminole, has a density of over 700 persons square mile (Bureau of Economic and Business Research, 1980). The population of Orange county estimated for 1979 is 441,337 with the city of Orlando accounting for 28% of the total population. The population of Orlando has been increasing since 1940 at an average annual rate of 61%. Orange county has experienced a net migration average rate of about 4,300 persons annually between 1960 and 1970. Persons in age groups of 10-60 years account for 75% of this figure. The group of 60+ years accounts for 19% of net migration. In contrast to Pinellas county new residents are primarily from the younger age categories (Bureau of Economic Analysis, 1979). The distribution has not changed significantly in the past twenty years in either Orange or Seminole counties however, the number of persons in the o 14 years of age has been declining. In 1978 in Orange County 45% of the population fell into the 15-44 age category, 22% fell into the 45-64 age category and 11% fell into the 65+ category (Bureau of Economic and Business Research, 1980). -67-

PAGE 75

",,' ,'If l According to the 1970 Census 57% of the Orange County residents had completed a high school education. Due to the increased urbanization since that time this percentage has most likely increased (Orlando Chamber of Commerce, 1978). Median family income in 1969 was $8,880 and had increased by 22% to $10,832 by the mid seventies. Per capita in )976 was $5,123. Average household income was $14,719 in 1976 and it is expected to increase by 50% to $22,062 by 1981. Housing characteristics in Orange County differ from Pinellas although not too drastically. Due to the younger population the average household size is larger than in Pinellas County, in 1970 it was 2.06 and in 1975 it was 3.02 persons (Orlando Chamber of Commerce, 1978) In a 1970 housing lnventory of 109,000 households, 76% were single-unit residend:es and 17% were multi-unit. In 1975, the number of households increased to 140,000 with the percentage and multi-unit percentage increasing to 30%. In both years the percentage of mobile homes averaged 5.8% of the total. Between the years 1971-1974 construction starts for multi-unit structures for outweighed single unit dV.Jellings. However, since 1975 the construction of single-unit structures has been greater. In absolute figures the construction activity of both periods differs substantially: between 1971-1974 single-unit construction averaged 381 units annually while multi-unit constructi on averaged 2,756 units annually. Between 1975-1977 s i ngl e-unit construction averaged an annual 239 units and multi-unit construction averaged an annual 30 units. Water supply and demand characteristics Orlando receives its water from its own municipal system which draws from the Floridan aquifer through well fields throughout Orange County (Healy, 1980). The number of wells has increased from 18 wells in 1970 to 23 wells in 1980. Until 1956 surface water from nearby lakes was utilized; since then the supply has been exclusively groundwater. Average daily pumpage rates have increased from 15 million gallons in 1956 to 41 million gallons in 1975. Demands have been increasing; population served in 1970 was 175,000 and in 1980 it was well over 200,000. Per capita daily water lIse has increased from 185 gallons in 1970 to 208 gallons in 1975. Generally the period of the highest seasonal demand is April-May, the lowest demand occurs during February. Orange County is, for the most part, a water abundant area. The effect of increasing urbanization, however, has since the 1960's begun to change the quantity, and the quality of the available reserves. The major source of aquifer recharge is rainfall, less of which is reaching the groundwater supplies due to manmade surfaces. Also, contact with such surfaces is causing an increase of pollutants recharged into the aquifer. Urbanization trends also create an increase in demand. It is expected that by 1985 water withdrawals will equal the water recharged. By 1990 it is anticipated that the former may exceed the latter by 10%, i.e., water will be "minedll in Orange County. The result of this could be severe contamination of the aquifer through intruding salt water (East Central Florida Regional Planning Council, 1977; and St. John's River Water Management District, 1977). -68-

PAGE 76

The inland areas of Florida have water shortage problems which cannot compare in severity with those of the coastal cities. Despite the above caveats, Orlando itself is situated close to several of the most productive recharge areas in the Floridan aquifer (East Central Florida Regional Planning Council, 1977). So far water shortages which warrant limitations on residential consumer consumption have not existed. Sampling Procedure The population of the two communities, \'Jhen adjusted to include' only single family dwellings with 5/S-inch water meters was approximately 74,900 households in St. Petersburg and 25.000 in Orlando. Due to the high cost of collecting observations the sample sizes were limited to a tot*l of 312 sampling units, or households.12 The procedure followed was a combination of stratified and cluster sampling (Scheaffer, et al., 1979, Chapters 5 and 7). The reason for using both was to insure the lowest cost for the greatest amount of variability in the major independent variables. The personal interview, selected as the mode of data gathering, is very expensive. Travel costs were cut by "clusteringll the sample respondents into various separate areas of the sample area. The sample was stratified into three groups characterized by high, medium and low levels. This insured that the cluster of the respective incomes were chosen from all three in the proportion in which they exist in the population. The census tract was used as the primary cluster unit and the census block as the secondary cluster unit. Census tracts were categorized i:nto income groups according to that group which constituted the highest per centage of the households in the tract.13 The results of this procedure, listed in Table ,indicated that in St. Petersburg 5% of the sample would be drawn from 3 census tracts, 32% from 15 tracts, and 63% from 38 tracts. In Orlando, 1% would be drawn from 1 tract, 66% from 15 tracts, and 33% from 6 tracts. Census tracts are not of uniform size. In St. Petersburg the tract sizes range from 377 to 1901 households, and in Orlando they range from 111 to 2,099 households. The sample required 100 households from Orlando and 150 from St. Petersburg. The marketing research firms responsible for executing the 12personal interviews for each sampling unit cost $9.25 in St. Petersburg and $9.34 in Orlando. The interviews were conducted by private marketing research firms. 13Household counting was made prior to adjusting the population by the qualifications indicated by (b) on page one. Census data was used from 1970. Updates of census information for income and household counts was available for St. Petersburg from the Economic Base Study, Pinellas Cciunty, 1977, and the Demographic Study, Pinellas County, 1978; and for Orlando from the Orange County Economic Data, 1979. -69-

PAGE 77

Table 8.--Number of census tracts and households in each income group. ST. PETERSBURG No. of No. of Income groups tracts % Households High over $15,000 3 5 2,863 Medium $7,000 $15,000 15 32 19,374 Low -under $7,000 38 63 37,740 TOTAL 56 100 59,977 ORLANDO No. of No. of Income groups tracts % households High over $15,000 1 1 175 Medium -$7,000 -$15,000 15 66 11,491 Low under $7,000 6 33 5,782 TOTAL 22 100 17,448 -70-

PAGE 78

survey requested that "around 500 households be provided in each city. To keep proportions even, 500 were provided in Orlando and 66 were pro vided in St. Petersburg. To maintain the proper proportions the 500 households (in St. Petersburg) were divided as follows: 25 from the high income group, 160 from the medium income group, and 315 from the low income group. In Orlando the 500 households were divided as follows: 5 from the high income group, 330 from the medium group, and 165 from the low income group. Census tract maps were provided by the city planning departments of St. Petersburg and Orlando. To use every census tract would not have alleviated the travel cost problem. Therefore the number of tracts used in St. Petersburg was di vi ded by three and in Orl ando it was di vi ded by two (see Table 8). This decrease in the number of tracts selected caused the size of each cluster (tracts) actually used to increase. Usually this is not desirable if homogeneity within a cluster is anticipated. In this case, however, variability is insured because clusters (tracts) were chosen independently in the three different inoome groups. The proportions of income groups in the population was maintained by dividing each income group of census tracts by the same number. The required number of tracts per group was then selected by simple random sampling. The next step was to select blocks from within the chosen census tracts. The geographi c boundari es of each tract were demarcat.ed on a city street map. The blocks were listed and selected through simple random sampling. Within each income group the number of households was divided by the number of tracts yielding the number of households required within each tratt (see Table 9). This number divided by the number of house holds residing on a block in each tract would give the number of blocks required per tract. There is certainly no uniformity among city blocks, however, on a tract by tract basis, resulted in a range of blocks per tract VJhich were to be canvassed. ; The average tract in St. Petersburg was more dense than its counterpart in Orlando altllougll tile lallge of the Iiousehold count per tf'act is broader in the latter. the tract number is less in St. Petersburg. While it would seem likely that the number of households per tract woul d be hi gher in St. Petersburg, the differenti a 1 between house hold counts in the high income group accounts for a higher number of households per tract in Orlando. 14Accurate counts of total housing units and their types (e.g., singlefamily, multi-family, etc.) can be obtained for each block in a census tract through the Block Statistics, 1970 Census of Housing. However this data is computerized and was expensive to obtain. Once a fitrst block in a tract was chosen its size was used to determine whether another block was needed (after accounting for the presence of apartments and commercial establishments). In all cases the tracts required at least two blocks. Four blocks was the most that any tract required. -71-

PAGE 79

Table 9.--'Number of households per census tract, by income level. ST. PETERSBURG ... --Llt.omelevgl----High Hedi.um Low Total ORLANDO Income level High Hedium Low Total Number of tract-s--. 1 5 13 19 Number of tracts 1 14 7 22 -72Number of oousehe-Hs 25 (33) 160 (213) 315 (420) 500 (666) Number of households 5 330 165 500 .. ,.-,. .-.. '\ .. .. '," \ Numb6:r of households .. pertra-ct-25 (33) 32 (43) 24 (32) Number of households per tract 5 47 41 .---

PAGE 80

Blocks were selected by simple random sampling. Chosen blocks were located in the 1979 city directories of both cities. All names (of the head of the household) were listed with the corresponding street addresses. A member of each of these households then became a potential respondent. Since many more households were provided than were required, each tract had "to be designated with the number necessary for the actual sample. This was aoquired by taking the number of total households necessary in each tract group and di vi di ng it by the number of tractsi n that group. In each group a percentage increase was given to account for responses which cannot be used. For St. Petersburg an extra 32 were provided and for Orlando an extra 30 surveys were provided for this punpose. -73-

PAGE 81

APPENDI X II I SUPPLEMENTAL SUMMARY OF QUESTIONNAIRE RESULTS The questionnaire which was administered in order to elicit individual va+ua t ffi-rts ofresi-dentta-l ats-rr containe-d supp1 ernenta riy quesfions about socioeconomic characteristics of the sampled households, and about the attitude, beliefs, and practices with regard to the water use of households. Discussion of formal research results did not include much of this information. It is the purpose of this appendix to summarize those supplementary questionnaire results. -74-

PAGE 82

Table 10.--Socioeconomic data Percentage in response category Response Question category Orlando St. Petersburg l. NUMRESPH: number 11 18 of persons in a 2 37 32 household 3 16 18 4 24 13 5 11 10 6 0 7 7 0 2 2. AGOT10: household 0 68 77 members of age 1 15 15 o to 10 years 2 13 4 3 13 2 4 5 2 3. AGllT20: household 0 74 64 members of age 1 15 19 11 to 20 yea rs 2 8 11 3 3 3 4 2 5 1 4. AG21T40: household 0 58 51 members of age 1 13 20 21 to 40 years 2 27 22 3 2 7 -75-

PAGE 83

Table 10.--Socioeconomic data--Continued Question 5. AG41T60: household members of age 41 to 60 years 6. AG60: household members of age 60 and over 7. SYRSCOM: highest level of education attained by head of household Response category 0 1 2 0 1 2 3 less than 12 yrs. 12 yrs. some college/tech B.A.,B.S. M.S./Ph.D. 8. OWNRENT: own or rent home own rent 9. : market value of home o $30,000 31 $50,000 51 $80,000 over $80,000 no answer/do not know -76Percentage in response category Orlando St. Petersburg 55 74 18 13 27 13 66 54 19 24 15 22 1 15 17 35 42 21 22 16 16 13 4 89 91 11 9 20 24 22 42 22 15 4 4 32 15

PAGE 84

Table 10.--Socioeconomic data--Continued Question 10. HOMEAGE: age of home in years 11. SIZEAREA: estimated size of property 12. INCOME : yearly household annual income Response category o 10 10 20 20 30 30 50 over 50 no answer/ do not know 1 ess 1/5 acre 1/5 3/4 acre 3/4 -1 acre less $5,000 $5 $9,999 $10 14,999 $15 19,999 $20 24,999 $25.29,999 $30 34,999 $35 39,999 over $40,000 -77Percentage in response category Orlando St. Petersurg 11 37 36 10 o 6 8 62 2 26 5 16 13 18 13 10 10 3 13 15 16 38 17 12 2 26. '6'4..'."(.: :.' 7 2 o 16 21 27 11 15 4 4 2

PAGE 85

Table 10.--Socioeconomic data--Continued Percentage in response category Response Question category Orlando St. Petersburg 13. OCCUPATION: professional, technical 13 12 'manageri a 1 administrative 16 12 sales 10 4 clerical 12 14 craftsmen 2 5 operative, laborer, service 8 4 farm personnel 2 4 retired 33 41 unemployed 3 4 14. WASHCAR: frequency of carwashing less once monthly 47 51 -3 times monthly 39 30 more 3 times monthly 15 5 no answer 13 15. SWIMPOOL: presence of swimming pool yes 16 10 no 84 90 16. BOTTWAT: use of bottled water yes 95 7 no 5 93 -78-

PAGE 86

Table 10.--Socioeconomic data--Continued Percentage in response category Response Question category Orlando St. Petersburg 17. SEPTICT: presence of a septic tank yes 5 4 no 95 96 18. HAVEltJELL: presence of a private well yes 3 8 yes, for lawn watering only 19 61 no 77 31 19. KINDSYS: kind of ; rrl gat; on system do not water 5 7 hose and spn nk I er 69 80 automatic system 26 13 20. HIGHPERD: periods of highest water bi 11 s Jan -March 3 7 April June 19 8 July Sept 48 59 Aug -Dec 0 3 always same 19 21 do not know 10 4 -79-

PAGE 87

Table ll.--Conservation attitude data Question l. BELWSHRT: belief in a water shortage 2. PROBYR: year by which a water shortage will continue or develop 3. DEGSEV: degree of severity of present water shortage no Causes of present of 2000 or Response category yes no 1981 1985 beyond never very serious fairly serious not serious do not know shortage exists future water shortage: 4. LARDRY: lakes, ri vers dryi ng up great extent some extent not at all no answer/ ,do not know -80Percentage in response category Orlando St. Petersburg 29 33 71 67 2 6 42 45 39 28 18 20 3 4 19 14 3 5 5 15 69 61 32 11 40 46 26 34 2 8

PAGE 88

Table ll.--Conservation attitude data--Continued Percentage in response category Response Question category Orlando St. Petersburg 5. LACKRAIN: lack of rainfall great extent 36 31 some extent 38 40 not at all 26 22 no answer/ do not know 2 7 6. DEPLGRW: depleting groundwater great extent 34 35 some extent 43 36 not at a 11 15 17 no answer/ do not know 7 12 7. POPGRLOC: 1 oca 1 population growth great extent 70 69 some extent 28 20 not at all 0 4 no answer/ do not know 2 6 8.--POPGROTH: population growth in other areas great extent 49 57 some extent 38 30 not at all 9 7 no answer/ do not know 4 6 -81-

PAGE 89

Table ll.--Conservation attitude data--Continued Percentage in response category Response Question category Orlando St. Petersburg 9. : increased water usage due to technical change great extent 43 47 some extent 42 37 not at all 13 8 no answer/ do not know 2 8 10. ICWASTE: increased water usage due to waste great extent 51 47 some extent 42 38 not at all 6 8 no answer/ 2 7 do not know 11. COMINOUS: increased water usage due to commercial industrial needs great extent 43 29 some extent 43 51 not at all 6 12 no answer/ 6 7 do not know -82-

PAGE 90

Table ll.--Conservation attitude data--Continued Percentage in response category Response Question category Orlando St. Petersburg 12. AGRIUSE: increased water usage due to agricultural needs great extent 19 15 some extent 60 58 not at all 15 18 no answer/ 5 7 do not know 13. POLLUTE: increased contamination of water supplies great extent 49 39 some extent 45 39 not at all 4 15 no answer/ 2 7 do not know 14. WATUTIL: water util iti es have not taken tbe necessary steps to provide enough water great extent 17 25 some extent 66 45 not at all 13 18 no answer/ 4 12 do not know -83-

PAGE 91

Table 11 .--Conservation attitude data--Continued Percentage in response category Question Response category Orlando St. Petersburg Indicate which of these conservation methods you practice and to what extent: 15. TEESHAVC: turning off the water while brushing teeth and showering 16. using dishwasher only when full 17. : taking shorter showers 1 arge effort medi um effort small effort no effort does not apply large effort medium effort sma 11 effort no effort does not apply large effort medium effort small effort no effort does not apply -84----------------10 19 39 25 o 23 2 6 2 68 19 19 31 15 16 --------44 26 14 15 15 ---4 0 3 79 41 27 11 12 8 --------------------

PAGE 92

Table ll.--Conservation attitude data--Continued Percentage in response category Response Question category Orlando St. Petersburg 18. WATERlGC: cutting back on lawn watering time large effort 23 16 medium effort 16 12 small effort 19 15 no effort 15 5 does not apply 27 52 19. BATHC: filling bathtub only 1/4 full large effort 16 25 medium effort 13 10 sma 11 effort 26 9 no effort 13 18 does not apply 32 40 20. TOllTC: cutting back on times the toilet is flushed large effort 6 19 medium effort 21 20 sma 11 effort 24 19 no effort 45 39 does not apply 3 3 -85-

PAGE 93

---------Table llo--Conservation attitude data--Continued Question --_ .. ------2l. CAPREUSE: recycli ng shower water 1 arge medium small no does not 220 TTBRICC: placing a brick in the toil et tank 1 arge Response category effort effort effort effort apply effort medium effort sma 11 effort no effort -----------. -does Indicate which Qf these water saving devices is presently installed in your home: 230 PRESHDC: pressurized yes showerheads no 24. SSWASMC: suds-saving yes washing maching no -86Percentage in response category Orlando 10 1 8 68 13 2 2 8 60 ----. 29 .-. ---39 61 16 82 St. Petersburg 3 o 4 86 8 7 4 4 75 --10------29 71 12 87

PAGE 94

Table ll.--Conservation attitude data--Continued Percentage in response category Response Question category Orlando St. Petersburg 25. SHTRWSTC: shallow trap water saving yes 3 4 toilet no 97 96 26. CHEMTTC: chemical toilet yes 2 no 98 98 27. LOWVDISC: low volume yes 8 6 dishwasher no 90 91 does not apply 2 3 28. DUFLTTC: dual flush yes 3 3 toilet tank no 97 97 29. WLESSFAC: washerless yes 42 42 faucets no 58 57 30. RECINCR: a recent local yes 66 91 water price increase no 26 5 do not know 8 4 3l. DECONO: this price increase yes 24 37 led to a water no 42 55 use reduction does not apply 33 8 -87-

PAGE 95

Table ll.--Conservation attitude data--Continued Percentage in response category Response Question category Orlando St. Petersburg --32. FUTINCR: there is an anticipated water price yes 69 88 increase in no 26 10 the near future do not know 5 2 33. DECON50: you would decrease your water yes 74 71 use if the price no 26 29 rose by 50% 34. HOVJDE50: you would decrease your water use (i f the pri ce rose by 50%) by how much 1 10% 19 35 ----------------------------------------.----------1120% 34 18 21 -30% 15 11 over 30% 2 9 does not'apply 26 2 do not know 3 25 35. DECON1OO: you would decrease your water use if yes 89 81 the price rose by no 11 19 100% -88-

PAGE 96

Table ll.--Conservation attitude data--Continued Question Response category ----------_._-36. 37. HOWDE100: you would decrease your water use (if price rose by 100%) by how much 10% 11 -20% 21 30% over 30% does not apply I : you would decrease your water usage by installing water saving devices do not know ---yes, -i mme El-iat e installment yes when needed a new fixture only no do not know -89Percentage in response category Orlando St. Petersburg 13 30 19 12 21 18 34 23 11 18 2 0 ------------13 25 24 33 60 41 2

PAGE 97

LITERATURE CITED Adams, R. C., Currie, J. H., Hebert, J. A., and Shikiar, R., The Visual Aesthetic Impact of Alternative Closed Cycle Cooling Systems, Report No. CR-0989, U.S. Nuclear Regulatory Commission, April, 1980. Ajzen, 1., and Fishbein, H., "Attitude-behavior Relations: A Theoretical Analysis and Review of Empirical Research," Psychological Bulletin, Vol. 84, No.5, September, 1977, pp. 888-918. An Estimation of Residential Demand __ for Water in Dade __ County, Florida, M.S. Thesis, Department of Food and Resource Economics, University of Florida, 1974. Bureau of Coastal Zone Management, Florida Department of Environmental Regulation, The Florida Coastal Zone Management Program, 1979. Bureau of Economic Analysis, Elorida Department of Commerce, Orange County of Economic Data, January 1979. Bureau of Economic and Business Research, University of Florida, Florida Estimates of Population, 1979, February, 1980. Board of County Commissioners, Position Statement No.2: Resource Needs and Managed Growth for Pinellas County, Clearwater, Florida, October 30, 1973. Position Statement: Water Resource Needs -------------------------------of Pinellas County, July 30, 1973. Bohm, P., "An Approach bo the Problem of Estimating Demand for Public Goods," Swedish Journal of Economics, Vol. 63, No.3, June 1971, pp. 94-105. -'-'Est-imating-])e-mandfe-r--Pllblic B0Dtlgl AnBxperimen-t,-l-'-----European Economic Review, Vol. 3, No.2, March, 1972, pp. 111-130. Bradford, D., "Benefit-Cost Analysis and Demand Curves for Public Goods," Kylos, Vol., 23, No.4, 1970, pp. 775-791. Brookshire, D. S., Ives, B. C., and Schulze, D., "The Valuation of Aesthetic Preferences," Journal of Environmental Economics and Management, Vol. 3, No.4, December, 1976, pp. 325-346. Brookshire, David S., Randall, Alan, and Stoll, John R., "Valuing Increments and Decrements in Natural Resource Service Flows," American Journal of Agricultural Economics, Volume 62, Number 3, August 1980, pp. 478-488. Bruvold, H., "Residential Response to Urban Drought in Central California, II Resources Research, Vol. 15, No.6, December, 1979, pp. 12971304. 90-

PAGE 98

Cicchetti, C. J., and Smith, V. K., "Congestion, Quality Deterioration, and Optimal Use: Wilderness Recreation in the Spanish Peaks Primitive Area," Social Science Research, Vol. 2, No.1, March, 1973, pp. 15-30. Clouser, R. L., and Miller, W. To Encourage Conservation. Center, Purdue University, L., Household Demand for Water and Policies Technrucal Report No. 124, Water Resources August 1979. Crespi, L" "What Kinds of Attitude Measures are Predictive of Behavior?',' Public Opinion Quarterly, Vol. 35, No. \3, Fall, 1971, pp. 327-334. Currie, J. M., Murphy, J. A., and Schmitz, A., "The Concept of Economic Surplus," Economic Journal, Vol.' 81, No. 324, December, 1971, pp. 1297..: 1304. Danielson, L. E., Estimation of Residential Water Demand, EconomicsResearch Report No. 39, Department of Economics and Business, North Carolina State University, October, 1977. Davis, O. A., and Whinston, A. B., "On the Distinction between Private and Public Goods," American Economic Review Papers and Proceedings, Vol. 57, No.2, Hay, 1967, pp. 360-373. Davis, R. K., The Value of Outdoor Recreation: An Economic Study of the Maine Woods, Ph.D. Thesis, Department of Economics, Harvard University, 1963. East Central Florida Regional Planning Council, Policy Alternatives in lvater Recharge Areas, July 1974, Freeman, Myrick, The Benefits of Environmental Improvement, JohnsHopkins University Press for Resources for the Future, Baltimore: 1979. Gehm, H. W., and Bregman, ] I. (ed), Handbook of Water Resources and Pollution Control, Van Nostrand Reinhold Company, New York: 1976. Gottlieb, M., "Urban Domestic Demand for Water: A Kansas Case Study" Land Economics, Vol. 39, No.2, May, 1963, pp. 204-210. Hammack, J., and Brown, G. H., lvaterfowl and Wetlands: Toward Bioeconomic Analysis, JohnsHopkins University Press for Resources for the Future, Baltimore: 1974. Hanke, S. H., and Boland, J. J., "Water Requirements or Water Demands?" Journal of the American Water Works Association, Vol. 63, No. 11, November, 1971, pp. 677-681. ______ and Davis, R. K., "Demand 11anagement through Responsive Pricing," Journal of the American Water Works Association, Vol. 63, No.9, September, 1971, pp. 555-560. "Demand for Water Under Dynamic Conditions," Water Resources ------Research, Vol. 6, No.5, October, i1970, pp. 1253-1261. -91-

PAGE 99

___ -:___ "Some Behavioral Characteristics Associated with Residential Hater Price Changes," Water Resources Research, Vol. 6, No.5, October, 1970, pp. 1383-1386. Headley, J. C., "The Relation of Family Income and Use of Water for Residential and Commercial Purposes in the San Francisco-Oakland Hetropolitan Area," Land Economics, Vol. 39, No.4, November, 1963, pp. 441-Lf49. Healy, H. G., Water Supplies of Selected Hunicipalities in Florida, -------75 r-U.-&.Geolo-g.iga1 Su :t'\1.@cy, T al1a{1a&see Fle-EicEi-a 19 7-7.Henderson, A. s Surplus and the Compensating Variation," Vol. 8, No.2, FE,br:uary. 1941, pp. 117-121. Henderson, J. M., and Quandt, R. E., Microeconomic Theory, McGraw-Hill Book Company, New York: 1971. Hicks, J. R., "The Four Consumer's Surpluses," Review of Economic Studies, Vol. 11, No.1, Winter, 1943, pp. 31-41. _________ "The Rehabilitation of Consumer's Surplus." Review of Economic Studies, Vol. 8, No.2, February, 1941, pp. 108-116. Hirschleifer, J., DeHaven, J. C., and Milliman, J. W., Water Supply: Economics, Technology, and Policy, University of Chicago Press, Chicago: 1960. Hogarty, T. F., and Mackay, R. J., "The Impact of Large Temporary Rate Changes on Residential Water Use, II Water Resources Research, Vol. 11, No.6, December, 1975, pp. 791-794, Hoffman, M., Glickstein, R., and Liroff, S., "Urban Drought in the San Francisco Bay Area: A Study in Institutional and Social Resiliency," -------JOllrna----t-dr tl1eAlnlErlc-art lts-s-ociati6n----;--V61 .. 7, -------July 1979, pp. 356-363. Howe, C. W., and Linaweaver, F. P., "The Impact of Price on Residential Water Demand and Its Relation to System Design and Price Structure," Hater Resources Research, Vol. 3, No.1, First Quarter, 1967, pp. 13-32. Kurz, H., "An Experimental Approach to the Determination of the Demand for Public Goods," Journal of Public Economics, Vol. 3, No.4, November, 1974, pp. 329-348. Lattie, J., "Public Education for Hater Conservation," Community Water Management for Drought and Beyond: A Handbook for Local California Office of Emergency Services, Sacramento, California, 19790 Lauria, .D. To, "Water Demand Forecasting--Some Concepts and Techniques, If in HcJunkin, FoE. (ed.). The State of America's DrinKing liJater, North Carolina State University, April 1975> pp. 235-258.

PAGE 100

Linaweaver, F. P., Geyer. J. C., and Wolff, J. B., Final Summary Report on the Residential Water Use Research Project, Department of Environmental Engineering Science, John Hopkins University, July, 1966. Loehman, E. Consumer Surplus and Cost-Benefit Comparisons for Collective Goods, Center for Economic Policy Research, Stanford Research Institute International, Menlo Park, California, November, 1978. Loehman, E., Ben-David, S., and De, V. H., Measuring Demand and Political Acceptability for Nonmarket Goods: A Case Study of Health Effects .. Belated t Q .. AiL_Quality .. Gen ter. fo r.Economi cJ>.o.lic..:y. Rese.arch,.a.t anfurd Research Institute International, Menlo Park, California, October, 1978. Lynne, G., and Gibbs, K., Demand and Pricing Policy for Residential Water, Economic Report 83, Institute of Food and Agricultural Sciences, University of Florida, 1976. Haler, K.G. Environmental Economics: A Theoretical Inquiry, Johns Hopkins University Press for Resources for the Future, Baltimore: 1974. McGarry, R. S., and Brusnighan, J. H., "Increasing Water and Sewer Rate Schedules: A Tool for Conservation," Jommal of the American Water Works Association, Vol. 71, No.9, September, 1979, pp. 474-479. Milne, 1'1., Residential Water Conservation, Report No. 35, Water Resources Center, University of California at Davis, March, 1976. Hishan, E. J., Cost--Benefit Analysis, Praeger Publishers, New York: 1976. Morgan, W. D., IIResidential Hater Demand: The Case from Micro Data," Vol. 9, No. August, 1973, pp. 1065-1067. North, R. H.,. Consumer Responses to Prices of Residential \.,Jater, Journal Series No. 185, Georgia Agricultural Experiment Station, University ---o-f--G-e-or-g-t-a-;----1-967. --------------Orlando Chamber of Commerce, Statistical Data, Orlando, Metropolitan Area, July, 1978. Parker, G. G., "Hater and Water Problems in the Southwest Florida Water Hanagement District and Some Possible Solutions," Water Resources Bulletin, Vol. 11, No.1, February, 1975, pp. 1-20. Pinellas Planning Council, Demographic Study, Pinellas County, Clearwater, Florida, April, 1978. Economic Base Study, Pinellas County, Clearwater, Florida, 1977. Randall, A., and Brookshire, D. S., Public Policy, Public Goods, and Contj.ngent Valuation Hechanisms, Staff Paper No. 68, Department of Agricultural Economics, Un:Lversity of Kentucky, June, 1978. -93-

PAGE 101

---------. Ives, B. C., and Eastman, C., IIBidding Games for Evaluation of Aesthetic Environmental Improvement,1I Journal of Environmental Economics and Management, Vol. 1, No.2, August, 1974, pp. 132-149. -------, and Stoll, John R., "Consumer's Surplus inCommodity Space,1t The American Economic Review, Vol. 70, No.3, June 1980, pp. 449-455. St. Johns River Water Management District, Water Resource Management Plan, Phase I, November, 1977. Schad, T. M., liThe National Water Commission Revisited,1I Water Resources Bu]JeUu, VoJ. 14, No.2, April 1978, pp. 302-312. Scheaffer, R. L., Mendenhall, W., and Ott, L., Elementary Survey Sampling, Duxbury Press, North Scituate, Mass.: 1979. Schlerger, D. L., and Cerviso, T. W., ItWater Conservation Rationales: Are There Parallels?1t Journal of the American Water Works Association, Vol. 27, No.1, January, 1980, pp. 11-38. Sierra Club, Gainesville Chapter, Newsletter, March, 1980. Sinden, J. A., Utility Analysis in the Valuation of Extra-Market Benefits with Particular Reference to lvater Based Recreation, Paper No. 17, Resources Research Institute, Oregon State University, March, 1973. Southwest Florida Water Management District, District Water Management Plan, Brooksville, Florida, 1978. Tideman, T. N., and Tullock, G., itA New and Superior Process ;Iior Making Social Choices1 Journal of Political Economy, 'vol. 84, No.6, December, 1976, pp. 1145-1159. Watkins, G. A., The Scaling of Attitudes Towards Water Conservation, M.S. Thesls, Department of Soclology, University of Florida, 1968. Willig, ,R. D., It Consumer s Surplus without Apology, II American Economic Review, Vol. 66, No.4, September, 1976, pp. 589-597. 1,Jong, S. T., "A Model on Municipal Water Demand: Northeastern Illinois,1I Land Economics, Vol. 1972, pp. 35-44. A Case Study of No.1, February, Wonnacott, Ronald J., and Wonnacott, Thomas H., Eco.nometrics, New York: John Wiley and Sons, 1970. -94-