You are here: Eucalyptus Action Group reports » Report: Eucalyptus nitens for solid timber ›  Appendix 3: Literature review, survey methods

Eucalyptus nitens, recovery and economics of processing 15 year old trees for solid timber

Report Date: May 2015
Author: Dean Satchell, Sustainable Forest Solutions, R.D. 1 Kerikeri, Northland 0294
+64 9 4075525

Special thanks and acknowledgement go to:

  • MPI Sustainable Farming Fund
  • Neil Barr Farm Forestry Foundation
  • John Fairweather Specialty Timbers
  • North Canterbury, South Canterbury, South Otago and Southland branches of NZFFA
  • NZFFA Eucalyptus Action Group
  • NZFFA Research committee
Appendix 1: Assumptions in Discounted Cash Flow Analysis
Appendix 2: Sawn timber price estimates
Appendix 3: Literature review - Value-based survey pricing methods
Appendix 4: Literature review - Estimating profitability of growing E. nitens for solid timber production
Appendix 5: Sawmilling method
Appendix 6: Flooring price survey instrument
Appendix 7: Survey results table
Appendix 8: Survey analysis
Appendix 9: Wood physical properties, test results
Appendix 10: Glossary of terms
Appendix 11: Case study stand plot
Appendix 12: Comparison between levels of internal and surface checking
Appendix 13: Air drying experiment
Appendix 14: Sensitivity analysis

Appendix 3: Literature review on value-based survey pricing methods

The aim of this literature review was to identify social survey methods that would be suitable for pricing timber species or timber products new to the market.

Value-Based Pricing Methods

Consumer utility is preference represented as value, as judged subjectively by an individual. Estimating prices for new products based on consumer utility can be undertaken by using either cost or value-based models. Value-based models base price on consumers’ willingness to pay for the product in question, where the individual measures utility in terms of monetary values. In the absence of observable market prices that directly reveal consumer preferences (for example when pricing a new product), indirect revealed preference methods that utilise implicit market approaches, such as hedonic pricing, can be used to infer the value of the product or attributes using statistical techniques. Conjoint analysis and contingent valuation are expressed (or “stated”) value-based techniques that measure consumer preferences. These techniques utilise social surveys rather than a market approach to estimate product value or worth based on respondents’ choices.

Hedonic price analysis and conjoint analysis are both empirical applications of Lancaster’s (1971) theory implying that products are valued by consumers based on multiple features or quality attributes (Novotorova, 2007, p. 28).

Hedonic pricing utilises data from actual consumer purchases to estimate values for specific product attributes based on their impact on a product’s price (Besanko, Dranove, Shanley, & Schaefer, 2009, p. 405). Hedonic regression uses a set of hedonic variables for the attributes and sale price is a function of these (Wang & Wolverton, 2002, p. 112). Price is the dependent variable and the quality attributes are the independent variables in a multiple regression analysis.

In a hedonic regression analysis, coefficients are multiplied by the level of each characteristic. The component characteristics each contribute utility and are summed to equal product value. A simpler hedonic model, the sales comparison adjustment (or grid adjustment) method uses the sale prices of two similar properties or products to estimate the value of the one attribute that differs between them. The coefficient is multiplied by the difference in price between two products to value the level of only one attribute.


Expressed (stated) preference techniques

Contingent valuation

Contingent valuation is a stated preference method that utilises surveys to estimate consumers’ willingness to pay for a good or service (Novotorova, 2007, p. 28). In traditional contingent valuation the direct hypothetical question describes in detail the good or service requiring payment to improve and a preference is elicited from the respondent (Boxall, Adamowicz, Swait, Williams, & Louviere, 1996). Although traditionally used for valuing non-market goods (e.g. environmental goods that do not have a market price) that provide utility, contingent valuation survey methods are also applied to valuing new products (Cameron & James, 1987, p. 394; Shi, Gao, & Chen, 2014, p. 1; Wiggins Johnson, 2005). Wattage (2001, p. 1) describes contingent valuation as “amenable to use in private good markets to identify consumer preferences for a new product, however, the use of it in this context is so far very limited”. However Carson and Louviere (2011, p. 541) define goods being valued in contingent valuation questions as necessarily public because value estimates are contingent on a constructed scenario.

Participants are asked direct questions about their preference for a particular product attribute and thus their willingness to pay (or alternatively their willingness to accept compensation) for the attribute or change in level of the attribute (Wiggins Johnson, 2005, p. 17). Open-ended questions ask how much the respondent would pay and closed-ended questions ask “would you be willing to pay $X” (Wiggins Johnson, 2005, p. 17), a referendum question with a yes or no answer.

Conjoint analysis

Conjoint analysis is a value-based survey method (‘Value-based pricing’, 2014) often used in pricing new products (Wiggins Johnson, 2005, p. 14).

Indirect stated preference in the form of trade-off experiments has its origin in conjoint analysis (also called trade-off analysis or choice analysis). Conjoint experiments require respondents to make trade-offs between two or more options as a measure of utility. This contrasts with simple preference selection, which reveals nothing about utility. Conjoint analysis offers a hypothetical construct of choices and/or evaluations consumers might make in order to predict market attitudes and preferences. Simulated “offers” are chosen or evaluated by the respondent and the respondent’s view of the product is quantified as a value called utility. These “as if” utilities relate to individual’s behavioural responses (‘Conjoint Method Overview’, 2013).

Consumers consider features or attributes when making purchase decisions and conjoint analysis simulates this behaviour by assigning values called part-worth utilities to the levels of product attributes. Conjoint analysis applications include the assignment of economic value to different attributes that each contribute to price, along with the relative importance these attributes hold (Cerda, García, Ortega-Farías, & Ubilla, 2012). In marketing research conjoint analysis is used to estimate the impact product attribute levels (including price) have on consumers’ preference for products and their propensity to buy.

Carson and Louviere discussed terms as used in stated preference procedures and associated statistical techniques as having “ambiguous and sometimes conflicting meanings” (2011, p. 539) and thus recommended that “use of the term conjoint be abandoned for virtually all purposes”(2011, p. 544). Carson and Louviere (2011) proposed a common nomenclature, including the following classes of methods for eliciting stated preferences :

  • Matching methods where respondents select or provide a number so that they are indifferent between alternatives. Matching methods include direct, bidding game, payment card, allocation game and time trade-off elicitation techniques.
  • Discrete choice methods where respondents pick their preference from a set of options. Discrete choice experiments include binary choice, multinomial and best-worst questions, complete ranking exercise and alternative subsetting elicitation techniques.

These two survey techniques are not mutually exclusive and hybrid survey methods can combine matching and discrete choice questions, such as an open-ended question to establish a constraint, followed by choices based on that constraint (Carson & Louviere, 2011, p. 551).

Matching methods of stated preference surveys

The direct question form of the matching method asks respondents for their willingness to pay for a good, thus is equivalent to an open ended contingent valuation question. A specific response is requested rather than selecting a preference (Carson & Louviere, 2011, p. 546). The quantity being matched could be monetary.

The allocation game form of the matching method (also called constant-sum allocation) requires the respondent to divide the attributes into a constant sum.

Discrete choice methods of stated preference surveys

Choice-based conjoint experiments elicit the survey respondent to make a choice from each set of alternatives (Boxall et al., 1996). The alternatives may be varying levels of attributes within a good plus a range of prices, allowing respondents to effectively trade off product features with price. By selecting, rating or ranking a defined set of combined attribute values, survey respondents infer preferences that reveal the “part-worth utilities” of individual product attributes (Conjoint Analysis in 10 minutes - Business Performance Management, 2010). This allows the researcher to estimate utility allocated to each attribute by participants (Wiggins Johnson, 2005, p. 14). Although the effect of the attributes is considered jointly, relative preference can be calculated for individual attributes. However, where ratings measures are used these estimate utility with ordinal rather than cardinal measures and thus difference between values cannot be measured (Carson & Louviere, 2011, p. 551). Such non-comparative scaling methods do not provide for the relative ordering of attributes because these are each scaled independently of each other.

Questions used in discrete choice experiments can be either binary choice or multinomial choice. Where the number of alternatives to choose from are greater than two, the question is multinomial (Carson & Louviere, 2011, p. 548). A multinomial choice question might ask the respondent to select their preferred alternative but also rank all alternatives from best to worst (Carson & Louviere, 2011, p. 548), called rank-ordering. Another ranking elicitation technique is the best-worst choice question where the ranking exercise it to only elicit the best and worst choices from the respondent (Carson & Louviere, 2011, p. 549). Alternative subsetting is a term used to “generate a set of implied binary comparisons” from a question where the respondent picks alternatives that are acceptable (Carson & Louviere, 2011, p. 549).


Self-explicated stated preference methods

In self-explicated questions respondents are asked to provide direct estimates of utility or attribute importance. One self-explicated method used for estimating attribute importance is the constant-sum allocation approach. The constant-sum allocation method captures the trade-off between attributes (Netzer & Srinivasan, 2011, p. 2) as relative importance across all attributes. Although it may be difficult for a respondent to divide a constant sum such as 100 among many attributes (i.e. where the sum of attributes equals the constant), this avoids respondents allocating a high importance to every attribute (Netzer & Srinivasan, 2011, p. 2).

Levels within each attribute are presented to the respondent and are evaluated for desirability (Xu, 2005, p. 183). Green and Srinivasan (1990, p. 9) describe a ratings approach with a desirability scale of 0-10. Part-worth utility for the attribute level is then calculated by multiplying (i.e. “weighting”) the attribute importance estimate by the desirability of the attribute level (Green & Krieger, 1996, p. 852). These part worth values can be summed to calculate overall preference or utility for a profile.

Self-explicated approaches can use either trade-off or non-trade-off types of evaluation. The direct or subjective approach does not involve trade-offs and elicits importance or value estimates directly from respondents as ratings or graded-pair comparisons such as dollar metric comparisons (Camilleri, 2011; I.M. Crawford, 1997; Schlereth, Eckert, Schaaf, & Skiera, n.d.). Dollar metric comparisons provide interval-scaled and thus metric measurements of preference, whereas rating scales are not comparative because respondents evaluate individual stimuli one at a time.
Ranking and paired profile comparisons (conjoint questions) are indirect trade-off based approaches and thus the evaluation provides adequate discrimination among attributes. These non-metric procedures involve ordinal scaling and thus relative preference is not measurable.

The constant-sum method involves trade-offs and provides importance weights directly based on respondent’s perceptions. Trade-offs allow for discrimination between the attributes. The constant-sum method provides ratio-scaled metric measurements of preference, thus revealing relative importance of attributes.

Dollar metric graded-pair comparisons

The dollar metric method of graded-pairs comparison provides a choice from which the respondent selects their preference, along with a “how much” evaluation in which the respondent states how much extra they are willing to pay for their preferred option. Self-explicated part-worths are obtained by eliciting willingness to pay for a product feature. By specifying how much more they are willing to pay for their preferred attribute level than for the lowest valued level, the respondent directly estimates the part-worth for the preferred level of the attribute (I.M. Crawford, 1997; Leigh, MacKay, & Summers, 1984, p. 458; Srinivasan, 1988, p. 296).

Another approach is to compare two items based on their given characteristics and ask the respondent to judge the price of one (given the price of the other) in order to make the two items of equal worth (Camilleri, 2011, p. 2). This evaluation of total utility is compensatory because the respondent evaluates the bundle of attributes that make up the product and makes their judgement by mentally offsetting attributes against each other (Lambin, 2007, p. 2).

Lee et al. (2008, p. 5) suggest that consumers evaluate products as monetised values better than by judging value presented in terms of utility.

Graded-pair comparisons explicitly require a comparative evaluation rather than purely subjective judgments and produce reliable estimates of part-worth values (Leigh, MacKay, & Summers, 1981).


Conjoint analysis vs. self-explicated stated preference approach

The self-explicated (or benchmark) stated preference approach allows the respondent to estimate part-worth utilities directly. Direct questions about attributes elicit preferences. In contrast a choice experiment elicits preferences indirectly and respondents make trade-offs between options to infer measures of utility. Part-worth is a preference parameter and price is typically an attribute (Green & Srinivasan, 1990, p. 4)

Conjoint analysis is described as a “decompositional” (Green & Srinivasan, 1990, p. 4) or “top-down” (Sambandam, 2009) method of measuring consumer preferences or attribute importance, where the respondent evaluates a set of product profiles, each product profile a “full” set of attributes with each attribute at varying levels. The analysis reveals the relative importance (utility) of each attribute and each level of the attribute (Xu, 2005, p. 182). Attributes become explanatory variables within a multivariate linear regression model. Total utility is the sum of part utilities and utilities can be interpreted independently as stand-alone values (‘Conjoint Method Overview’, 2013; Kroes & Sheldon, 1988, p. 14).

The Self-Explicated preference model, described as “compositional” (Wilkie and Pessemier in Green & Krieger, 1996, p. 852; Green & Srinivasan, 1990, p. 9) or “bottom up” (Sambandam, 2009), allows product attributes to be evaluated individually (or in sets).

Multi-attribute utility models that combine attribute importance with attribute desirability form the theoretical basis of the self-explicated model (Xu, 2005, p. 183). Utility for a product can be modeled (Fishbein 1967 in John Roberts, 1989) as “a weighted linear function of its attributes”. By weighting attribute level desirability values with the relative attribute importance value, utility part-worth values are produced for each attribute level (‘SurveyAnalytics Self Explicated Conjoint Analysis’, n.d.). This approach does not require regression analysis (‘SurveyAnalytics Self Explicated Conjoint Analysis’, n.d.) and inherently assumes preference or total utility to be an additive function of part-worth utilities (Green & Srinivasan, 1990, p. 10; Srinivasan, 1988, p. 296).

Evaluating individual attributes one at a time does not simulate market behaviour because consumers’ view products as bundles of attributes. However, Self-explicated approaches are less expensive and less time consuming  (Sattler & Hensel-Börner, 2007, p. 2) and decompositional approaches can overload respondents with information (Srinivasan, 1988, p. 295).

The self-explicated method minimises information overload by questioning the respondent on each attribute separately (Srinivasan & Park, 1997, p. 286). The self-explicated approach has been shown to be robust in research experiments (Srinivasan & Park, 1997, p. 290).

By considering one attribute at a time, respondents might not rate important features as highly as they would in a decompositional analysis (Sambandam, 2009).  Respondents would also need to be made aware of all attributes and their levels before evaluating individual attributes or levels (Sambandam, 2009).


Stated preference surveys: Model specification, design and analysis

Attribute selection

An additive function assumes overall utility is the sum of part-worth utilities for the levels in the product being valued (Srinivasan, 1988, p. 296). Quality characteristics are required to define the total worth of the concept, they must be actionable independent factors that can be put into practice and must differentiate between products (Lambin, 2007, p. 3).

Estimating utilities

A monetised utility function allows for the monetary value representing each unit change in an attributes quality or level as the coefficient in a linear model. Assuming homogeneity of preferences within the target population, (Lambin, 2007, p. 8), because a direct monetary measure is ratio scaled, prices for levels of attributes are relative between and across attributes and can be compared and averaged across respondents (Kalish & Nelson, 1991, p. 329). The survey respondent directly estimates the maximum price the consumer is willing to pay, the reservation price. This is considered to be a more difficult task than rating or ranking (Stauß & Gaul, 2005, p. 579).

Perceived value pricing models

Perceived value is the price a customer is willing to pay for a product and its benefits (Thompson & Coe, 1997, p. 71). Customers value a product by trading off benefits that offer a value advantage over alternative products, with the product’s overall costs.

Economic value analysis (EVA) is a perceived value pricing technique that references life cycle costs and benefits of competing products in an attempt to establish a price that provides buyers with utility (Thompson & Coe, 1997, p. 71). By quantifying benefits to customers that a product provides for a market segment, pricing decisions can be made (Hinterhuber, 2004, p. 769).

Consumers reference existing products when valuing a new product (Monroe & Della Bitta, 1978, p. 415; ‘Pricing Strategy’, n.d.). A product’s price reflects the utility associated with forgone consumption from purchasing an alternative product (Reutterer & Breidert, 2007, p. 84). A product’s perceived benefit “is equivalent to the utility of paying the maximum acceptable price” (Dodds, 2003, p. 14). Thus economic value to the customer could be interpreted as the perceived value a customer assigns to a product (willingness to pay) minus the price paid for the product, resulting in the product’s utility to the customer, also called the “consumer surplus” (Hinterhuber, 2004, p. 769).

An alternative concept of economic value involves setting a price for a new product by determining the price differential that adjusts the price from a reference product’s value (Jose Briones, 16:26:54 UTC, p. 13; Monroe & Della Bitta, 1978, p. 415). Forbis and Mehta’s (in Thompson & Coe, 1997, p. 71) Economic value to the customer (EVC) pricing model is a value-based EVA pricing technique used for describing a maximum theoretical price for a product, or price ceiling a firm needs to price at or below, to sell their product. The maximum acceptable price (also called the price ceiling or reservation price), a closely related concept to willingness to pay (Stauß & Gaul, 2005, p. 578), is the price point where a consumer would ignore the difference between the new product being priced and the price of the reference product (Monroe, 1990, p. 97). Setting a price any greater than the price ceiling gives the consumer greater utility from a competitor’s alternative product, or negative utility for the product of interest (Dodds, 2003, p. 14). The price ceiling (or customer’s total economic value) for a new product is defined as the price of the nearest competitor’s product (reference value or competitive reference value) plus the price differential (also called differentiation value or incremental value) between the two products (Hinterhuber, 2004, p. 769; Monroe & Della Bitta, 1978, p. 415; Tucker, 2010).

Hinterhuber (2004, pp. 769–770) described six steps to quantify economic value:

  1. Identify the best alternative product’s life-cycle costs.
  2. Select the market segment.
  3. Identify those factors that differentiate the product from the reference product.
  4. Determine the value to the customer of the differentiating factors.
  5. Determine economic value.
  6. Estimate sales volumes for different price-points.

Total life cycle costs for both the reference product and the product being priced include the purchase price and costs incurred throughout the life of the products. These might include freight, installation and also maintenance costs that keep the product functional (Thompson & Coe, 1997, p. 72). Consumers are aware of additional costs to the purchase price of the product such as start-up costs and post-purchase costs that may be incurred over the lifetime of the product. Such costs will vary between the product being priced and the reference product (Dodds, 2003, p. 175). A company marketing a new product might consider including in its total offering elements additional to quality or performance (Dodds, 2003, p. 173).

Improvements in functionality (or value advantage) of the product being priced add to the incremental value and thus the economic value to the customer. Cost savings also increase the economic value to the customer. Pricing the product then involves deciding how value is shared between buyer and seller. To sell a product, price must be set based on value advantage and the differentiation value may be discounted to set the price of a product (Tucker, 2010).

Perceived value pricing models provide methods that could be used to assess economic value for specific products such as edge-jointed solid timber flooring and compare and contrast quality and costs with competing products such as engineered flooring. Comparing benefits of a new product along with production costs and life cycle costs with existing products form the basis for determining price that customers would be willing to pay for the new product. This type of analysis would be useful in exploring opportunities to improve sawn timber revenue from E. nitens.

Value-Based Methods and Pricing E. nitens Products

Discrete choice conjoint methods are more suitable for determining preference than utility. Self-explicated methods allow respondents to provide interval-scaled direct estimates of utility. Self-explicated methods, by eliciting respondents’ willingness to pay, are easier for respondents to understand, require fewer questions than choice-based conjoint analysis and have been shown to produce reliable estimates of part-worths provided evaluations are comparative.

Graded-pairs comparison and constant-sum allocation are both comparative self-explicated methods that provide interval-scale judgments and monetised part-worths. Constant-sum allocation evaluations capture the trade-off between attributes. Graded-pair evaluations are comparative and compensatory and simulate purchase decisions by allowing respondents to judge total utility as a bundle of attributes. These are mentally offset against one another as would happen in the marketplace.


Disclaimer: The opinions and information provided in this report have been provided in good faith and on the basis that every endeavour has been made to be accurate and not misleading and to exercise reasonable care, skill and judgement in providing such opinions and information. The Author and NZFFA will not be responsible if information is inaccurate or not up to date, nor will we be responsible if you use or rely on the information in any way.


Farm Forestry - Headlines

Article archive »