A Student’s Guide to Cost-Benefit Analysis for Natural Resources
Lesson 8 - Valuing Market and Non-Market Natural Resources
We said that a key concept in valuing the benefits of public policies was changes in consumers and producers surpluses.
Conceptually, in order to do this, you must know the market supply and demand curves for the good/service in question. There are econometric methods for estimating supply and demand curves, provided price and quantity market data exist.
Non-existent data is especially a problem for public projects related to governmental CBA analysis. With public projects, however, prices often do not exist and therefore the supply and demand curves are usually unknown.
The reason for no data is obvious when you think about it. Recall, the public sector is usually considering a project because of market failure. Thus, in most cases when we want to conduct a CBA for a public project, markets (and hence price/quantity data) are incomplete or non-existent.
We will discuss several methods which are used for determining prices in the absence of competitive market prices. These measures are also called shadow prices - assigned prices used when competitive market prices do not exist.
Before we discuss the various methods, first some philosophical/ethical issues about valuation:
Human-centered (anthropocentric) point-of-view:
Traditional economic valuation is done from a human-centered point of view and deal with values in the context of trade-offs. Why is this? The basic premise of economic theory is that the purpose of economic activity is to increase the well-being of the people in society through production and consumption of goods and services. It follows, then, that the basis for deriving measures of economic value is their importance to humans. The issue of trade-offs enters-in, because economists are always asking what would society be willing to sacrifice in order to obtain a good or service.
An important fact: economists recognize that non-market commodities, even non-uses, contribute to the welfare of humans. Economics is not concerned solely with marketed goods and services. Modern economics attempts to value these non-market items, but from a human perspective. Based upon human willingness-to-pay (WTP) for these items.
The economist’s point of view contrasts with religious/philosophical views (such as Deep Ecology's “ecocentrism”) where humans are said to have no moral authority to place value on other things. Furthermore, these religious/philosophical values are not values in trade. They are more-or-less values bestowed upon things by humans in a one-sided fashion. It may be one person's opinion that something is priceless. But try convincing the rest of society that this is the case.
Money as a common measure:
In order to conduct an analysis of economic efficiency, we need to compare inputs and outputs that are measured in the same units. We cannot add/subtract unlike elements ( e.g., “apples and oranges”).
We thus seek a unit of common measure. For economists, money has tended to be this common unit of measure mainly because that is the same function money plays in society. It is the common measuring rod to facilitate trade; rather than bartering.
Money valuation of environmental and natural resources strikes some people as improper, or even immoral. But keep in mind, money functions in CBA simply as a common measuring rod; it does not urge greed nor exploitation of resources. We could use some other measuring rod, but thus far, no better ones have emerged.
Finally, it may involve a value judgement to chose money as the measuring rod. But the use of non-comparable measuring units (e.g., apples and oranges) does not avoid a value judgment, it merely postpones it.
Classification of Natural Resource Values
Here, we classify natural resource values, by their uses and even non-uses. There are other such classification systems. This one comes mostly from Loomis.
1. Use Values (i.e., personal use):
Use Values are divided into market and non-market commodities:
a. Market commodities are those which are traded in competitive cash markets. Private timber is a good natural resource example of a market commodity, also some rangeland grazing permits, some minerals, some water.
Under market commodities we also include quasi-markets commodities where a cash market exists although it is imperfectly competitive; prices are administered by the government (BLM grazing allotments, public minerals leases, public timber).
b. Non market commodities involve public goods, open access or unique resources potentially constituting natural monopoly situations.
Examples:
i) outdoor recreation: The most common example of non-market commodities involve outdoor recreation: nature viewing, camping, fishing, hunting, wilderness use, water use (e.g., boating, swimming).
Some public fees may be charged, but generally these fees do not come close to measuring the full social value of the use.
ii) option value (option for personal use): another item classified under uses is option values.
Option value - willingness to pay by a person to guarantee that a resource would be available should he/she choose to use it in the future (Freeman, p. 261.) It is a personal-use type of value, but for use at some future, perhaps distant, time.
2. Non-(personal) Use Values:
Bequest value - WTP by a person so that a resource might be passed on to future generations.
Existence value - WTP by a person so that a resource will continue to exist today even if he/she never uses the resource. There is no personal use.
From the preceding classification, it is clear that modern economics tries to recognize a wide range of social values. Not only non-market values but even non-use values as well.
Remember, definitions of these terms by economists are not always consistent.
Methods for Obtaining Values
Loomis lists 4 sources of prices for natural resources provided by the public sector (i.e., government):
1. private market data (adjacent private markets)
2. quasi-market data (where non-competitive markets exist, e.g., federal rangelands grazing allotments)
3. revealed through behavior (e.g., Travel Cost Method)
4. obtained from surveys of preference (e.g., Contingent Valuation Method)
Each of the following methods for obtaining WTP measures fits in one of the above categories.
Common Methods for Estimating Prices for Publicly-Provided Natural Resources:
A. Market or Quasi-Market Methods of Price Estimation:
1. competitive market transaction data - you are fortunate when you have competitive price information from actual private market transactions as a basis for valuation. Example: Private market annual timber sales data is a good example of market transaction data; when active timber private markets are near federal lands. Example: Some eastern national forests.
Timber Mart is a private timber price reporting service that can be purchased commercially.
2. residual valuation - used in situations where private markets for natural resources may not be active, hence there is not much competitive transaction data. Used for the valuation of intermediate (producer goods) rather than final (consumer) goods.
A producer good price is calculated as a residual of final good price. The extraction and processing costs are deducted from the final product price to obtain the raw material price as a residual. Example:
Final product price - extraction/processing cost = raw resource price
Lumber price - logging cost - transportation cost - milling cost = stumpage price
This method is often used to determine the price for western federal stumpage and minerals where nearby market for raw materials is weak or do not exist. Residual valuation is the most common method used to price irrigation water.
3. change in net income - another method of price estimation used when market data are weak or nonexistent. Used some for pricing water resources.
The general idea behind this pricing method is to estimate what the natural resource contributes to a producers profits (net income), and base the competitive price upon this contribution to profit.
Net income per additional unit of input = per unit value of the inputs
Example: a government water project provides 2 million gallons of water; allows ranchers to increase herd size by 10,000 cattle at $15 profit per head.
1) multiply 10,000 cows x $150 profit per cow = $1.5 million profit (additional net income, profit).
2) divide net income by gallons of water: $1.5 million/2 million gallons = 75 cents/gallon.
Thus, 75 cents/gallon is the imputed value of the water. Govt. will not likely charge this amount, but it tells us the maximum potential price of the water. That is, the price above which all producer profits would disappear.
B. Non-Market and Non-use Methods:
Our interest with non-market, non-use is in valuing total-willingness-to-pay and consumers surplus, not producers surplus, because government is the producer of these non-market outputs. No private producers are affected.
4. Travel cost method (TCM) - most widely-used method for determining the demand WTP for outdoor recreation sites. It is a revealed preference method. As such, it is an indirect method of ascertaining value (i.e., you determine values indirectly, by inferring from actual behavior).
TCM was first suggested by Hotelling and later Clawson, and has since been used extensively to estimate demand for recreation sites.
Key attributes/assumptions of TCM:
1) based upon the fact that admission fees (often low or non-existent) to recreation areas are an inadequate measure of the value of a visit to recreation site,
2)the cost of round-trip travel is a proxy measure of WTP to visit a recreation site.
3) the journey itself yields no value, and
4) opportunity cost of time is derived from wage rates
5) recreation site users will react to changes in gate fees in the same manner that they react to changes in travel cost,
Defining Travel Cost
Specifically, the total cost of a site visit consists of 4 elements:
a) direct travel expenses (gas, motels, etc),
b) time-cost of travel (opportunity cost of time)
c) cost of time spent at the site (opportunity cost of time)
d) admission fee, if any.
Key point: The TCM permits estimation of a negatively sloped demand curve because the observations on individual cost (travel cost) and number of visits varies inversely across the population of users. This variation permits estimation of a demand curve for the recreation site. This is Marshallian demand, thus it measures consumers surplus.
The steps in estimating a TCM :
1. Select a random sample of households within the market area of the recreation site. Because of the high cost of surveying households, some choose to survey visitors at the recreation site.
Those surveyed are often assigned to zones of origin of travel, hence the Zonal Travel Cost Method. Zones may be concentric circles, or other geographic divisions.
2.Conduct a survey of households/visitors asking:
a) how often did you visit the site each year?
b) how far did you travel round trip (what zone do you live in)?
c) what did the trip cost you?
d) what are substitute sites & cost to visit them?
e) personal data that may influence demand (income, age).
3. Perform a regression analysis on the data:
Visits/pop./ year = f(Pt, Ps, Y, Z)
where: Pt = travel cost; Ps = travel cost to substitute sites; Y = income, Z = personal characteristics
OLS was once used; now other methods of estimation are used to estimate the regression model.
4. From regression, you derive a recreation WTP (demand) function for the site.
V/Pop./Year = f(Pt)
What can you do with this demand function?
1. Freeman says that: the area under the demand curve tells you what total annual value the public places on the recreation site.
2. Determine the maximum amount the public would be WTP for creating such a site.
3. Predict annual attendance at various gate fees.
4. Compute consumers surplus.
5. Examine the effect on consumer surplus resulting from the imposition of gate fees.
6. Compute Average Consumer Surplus
Total CS/Total RVDs = the CS for an average RVD
Problems with TCM:
1. Basic TCM measures demand for visits to the site as a whole, rather than for specific site features (e.g., water quality). Thus, with basic TCM there is no way to value improvements in site quality.
2. TMC requires significant variation in TC. Thus, if all residents are nearby the site, there is probably too little variation in TC for use of TCM.
3. TCM assumes that people respond to TC regardless of its composition; travel cost is the same as gate fees. 4. Problems with estimating opportunity cost of time:
opportunity cost of travel time & time at site typically = 1/3 to 2/3 x wage rate
under-estimating the opportunity cost of time will bias benefit estimates downward
5. Problem allocating fixed costs to specific trips, e.g., the cost of camping gear
6. TCM cannot effectively value multiple purpose trips. For example, if one goes to a campsite, then goes river rafting elsewhere. Possible solutions: exclude multiple purpose trips or allocate costs to various trips.
7. The trip itself is assumed to yield no utility; if the trip itself yields value, then the TCM will overstate site benefits.
8. Truncated (limited) dependent variable generated by sampling only those who are at the campsite and not those who do not camp. A statistical issue; OLS is not be appropriate estimation method. Maximum likelihood estimation procedures are recommended.
9. TCM is expensive to conduct because of the cost of surveys.
Using TCM to Measure Demand for Specific Site Attributes:
Traditional TCM measures the demand for the recreation site as a whole. Many times we want to know: how does the demand for a site change with changes in sight quality/site attributes?
Examples of empirical measures of site attributes (they can be natural or management-elated attributes):
1. Fishing quality: (angler success rates, fish populations)
2. water quality (DO, oil, bacteria, turbidity, fish kills, algae)
3. skiing (annual snow fall, lifts, terrain)
4. Campground facilities (toilets, potable water, hiking, etc)
5. vegetation/land use type (spruce, fir, clearcut)
The attribute demand model:
In general we want to derive an inverse D-function that tells:
Px= f(Qx)
where: Px = price of attribute x; Qx = attribute x;
Knowing this function, we can calculate the change in WTP associated with increases or decreases in attribute x. However, the problem is, we do not know WTP (i.e., Px ) for attribute x.
Two methods for measure economic value of site quality:
1) Hedonic TCM
Define Hedonic method - the Hedonic price method is used to estimate the value of an attribute (e.g., site quality) of a good (e.g., recreation site). In other words, the hedonic model, thru multiple regression, decomposes the total value of the good into the value of its several attributes. Hedonic price = implicit price. Hedonic method assumes that the value of the attribute is capitalized into the total value of the good.
Housing price example from Boardman: the price of a house comprises what factors?
P=f(D,S,V,N)
where: P= price of house; D=distance from town, S=size of house; V = view from house; N = neighborhood. The Regression coefficients( (β)d = MP/MD) tell the partial contribution of an attribute (e.g., distance) to total value; it is an implicit price. Hedonic price is WTP for a marginal increase in an attribute. Hedonic models are mostly regressed on market-determined prices (i.e., as the dependent (Y) variable).
Hedonic Travel Cost Model:
The HTC model is estimated in 2 stages:
Stage 1. Hedonic cost function - measures the cost of traveling to several sites each with different characteristics. A different regression is performed for each place of origin:
Cij = f(Qjk,)
where: Cij = travel cost for individual i to site j.
Qik = a vector or k site characteristics variables at site j; characteristics may be continuous (i.e., number of campgrounds) or binary (i.e., potable water or not).
The Regression coefficients MCij/MQjk tell the partial contribution of a kth characteristic at site j to total site travel cost for individual i. This coefficient is the implicit price per unit of that attribute.
Stage 2. Separate inverse [P= f(Q)]demand functions for various attributes prices
Pik = f(Qki, Yi, Zi)
where: Pik = price of characteristic k by i
Qki = quantity of characteristic k demanded by individual i
Yi, = income of individual (demand shifter)
Zi = personal characteristics (demand shifter)
From this, you can measures changes in WTP given changes in site attributes.
However: Hedonic TCMs have come under severe criticism because they do not have market determined prices like regular (i.e., housing) Hedonic models; thus there is no market mechanism that brings the various recreation site attributes into a market clearing relationship. As a result, Hedonic TCM sometimes results in unexplainable negative implicit prices for attributes.
2) Random Utility Models
RUMs are used to measure the economic impacts of changes in recreation site quality. RUMs are a type of TCM, i.e., the value of the recreation experience is determined by travel cost. But, rather than modeling the number of visits to a site over time (as, per year), RUMs model an individual's decision process when choosing among alternative recreation sites with different attributes.
An example of a RUM situation:
Consider that a person can fish at many possible streams.
Each stream is a different distance hence travel cost varies.
Each stream has different characteristics (water quality, fish populations) which will effect choice. These are independent variables.
The basic RUM states that the probability of angler i visiting a site j is as follows (Freeman 1993):
where: Prji = probability of angler i visiting site j (0 to1)
Yi = money income for angler i
Cji = travel cost for angler i visiting site j
Qkj = a vector of k characteristics of site j
Zi = a vector of socioeconomic characteristics of angler i
Form of the RUM Regression model:
RUMs use the form of a multinomial logistical regression model (i.e., logit model) where the predicted dependent variable (i.e., probability of a visit given values of independent variables) is restricted to lie between the unit interval, 0 to 1. The model’s predicted dependent variable is restricted to a cumulative logistic probability function.
3. Contingent valuation method
Many important valuation problems are non-observable; i.e., no value measures can be derived from observing individual choices in a market. Some of these situations involve potential rather than actual policy changes. Such cases call for value measurement methods which use hypothetical, constructed markets.
The most common technique of this type of measurement is the Contingent Valuation Method (CVM). The name contingent derives from asking people what they would be willing to pay contingent upon some hypothetical change in the state of the world.
For example, consider the following CVM question: “Suppose the management of the San Juan River is stocked so that the trout population during the summer months is increased by an average of 1000 fish per mile. What is the maximum amount you would be willing to pay for this change?”
Basic Concepts Behind CVM:
The basic idea behind CVM is to create a hypothetical but realistic market situation for nonmarket/non-use resources, communicate this to a person, and have them respond in an honest and realistic manner.
CVM is a stated preference technique used to estimate WTP for nonmarket and non-use public resources. It is a direct method of estimating value. You ask people directly to tell you their WTP.
CVM, is the only method available for valuing bequest, existence and option values, however it is used for other applications as well.
6 Key Methodological Choices When Devising a CVM Study:
1. identify population and sample - is the issue local, regional national? Get a representative sample of the population.
2. resource definition - a realistic description of the public resource, i.e., clean air, water, new park, wildlife, etc.(use words, photos, maps).
3. payment vehicle - a means for making simulated payment in exchange for the resource (invoices, taxes, higher site fees). The payment vehicle should be realistic, plausible and a non-controversial means of financing public projects.
4. value elicitation procedure - what value are they WTP, annually usually?
5. supplement/personal data - to be used as demand shifters & for use in projecting results across the population.
6. method of statistical analysis - methods for regressing WTP (dependent variable) on independent variables to obtain a WTP function.
3 Methods of Conducting CVM Surveys:
1. personal interviews - expensive, high response rates, best for presenting realism, interviewer may create bias.
2. telephone - cheaper, moderate response rate, difficult to create realism, pressure on respondents
3. mail - cheapest, lowest response rates, pictures used for realism, less pressure in respondents.
Kerry Smith says: the current state-of-the-art recommends a mix of focus groups, interviews, pe-tests with information, visual aids and circumstances to frame the CV questions.
5 Possible Procedures for CVM Value Elicitation:
1. open-ended query regarding maximum annual WTP (how much would you pay to protect the bald eagle in Alaska?) or WTA (how much would you be willing to accept as compensation for the loss of bald eagle in Alaska?) This method has resulted in high rates of non-response or relatively large numbers of implausible valuations. It is now recognized that this method puts the respondent into a unfamiliar situation.
2. iterative bidding - starting price is given the respondent, it is raised by questioner until the point of refusal. Or, alternatively, if the initial response is no then reduce the price until yes.
Problem: It has been statistically shown that the starting point influences the final response. Hence, starting point bias.
3. payment card - an alternative to iterative bidding, respondents are shown an array of possible bids ranging from zero to very large and ask to select one number. Respondents are given comparative taxes and prices of goods to help put responses into perspective. Payment cards reduce starting point bias.
4. dichotomous choice (referendum) method - from a possible full array of prices, randomly selected prices is presented to a respondent who says yes or no to the price. Repeated trials allow a cumulative probability function to be constructed: probability of yes = f(price).
Generally, the function indicates that the lower the price the higher the probability of payment. To find aggregate WTP for population: multiply mean WTP by population. This method is subject to yes-saying bias: people tend to agree with interviewers.
5. contingent ranking - various combinations of price and resource outputs are presented to respondents. Example: low price-low water quality, high price-high water quality. Respondents rank the combinations from most to least preferred.
Criticisms of CVM:
1. CVM is hypothetical. There are problems of meaning, context and familiarity that unfavorably influence CVM results.
2. CVM questionnaires are not always neutral; they may influence the response
3. Judgement bias - respondents may make errors in judgement regarding their own WTP
4. WTP vs. WTA: WTP is for obtaining a good for the first time while WTA is compensation for loss of a good (or environmental quality). WTP should approximate WTP theoretically for the same good/situation. However studies indicate that generally WTA>WTP by a large amount due to loss aversion. Evidence indicates that experienced respondents reduce WTP/WTA differentials.
5. Strategic bidding behavior - people may tend to strategize in their bids through: a) free riding, b) over bidding.
6. Scope problem - does the individuals WTP vary directly with the scope of the proposed project? In other words, if you propose to save 1/4, 2, or all of eagle habitat, do you get proportional differences in WTP among people?
A question: Why not have them just mail in a real check rather than make a hypothetical WTP bid?
Precisely because the situation is hypothetical and may never happen. How do you get someone to give you a check for bequest value or existence value What do you do tell them that you will do with the checks for hypothetical proposed change that may never occur? Surely their actual payment would be far below their true WTP simply for not knowing where the check will go. Thus the entire exercise must be hypothetical.
How Reliable is CVM?
The 2 biggest problems with CVM reliability relate to:
1) hypothetical nature (i.e., unrealized financial consequences), and
2) strategic bidding.span>
Unfortunately, there is no unambiguous way to address these questions. Most studies indicate that strategizing does not appear to be a general problem; it is more a problem with issues of local importance.
A number of studies have been done to measure CVM accuracy and consistency.
1. Accuracy: CVM, when a true comparison exists, tends to be higher than actual, but often not significantly so.
2. Consistency: Over time, individuals tend to be consistent in their responses.
NOAA Panel chaired by Kenneth Arrow and Paul Samuelson recommends 4 reliability tests; these are based more or less on logical relationships:
1. CV choices should respond to the scope of the proposed project
2. CV choices should respond to economic variables hypothesized to be important to actual choices
3. CV estimates of WTP for partsof a project, when summed, should add to the whole project for which WTP is estimated separately.
4. CV estimates for important objects should have significantly greater WTP than that for less important objects.
The NOAA panel concluded: ...CV provides estimates reliable enough to be starting points for a (legal) process of damage assessment.
Links: