Combining Discrete and Continuous Representations of Preference Heterogeneity: A Latent Class Approach

Unobserved preference heterogeneity has been widely recognized as a critical issue not only for modelling choice behaviour, but also for policy analysis. This paper examines alternative approaches for incorporating heterogeneity in recreational demand. We apply a hybrid model combining discrete and continuous heterogeneity representations of tastes to capture the defining features of both the latent class and the random parameter logit specifications. This model allows for the joint estimation of discrete segments and within segment heterogeneity providing a richer interpretation of preference heterogeneity. A database of recreational trips to forest sites in Mallorca has been used to compare the empirical performance of this hybrid approach with common specifications such as the conditional logit, the random parameter logit, and the latent class model.     

Understanding Hurricane Evacuation Decisions Under Contingent Scenarios: A Stated Preference Approach

We conduct a stated preference analysis to understand the factors that individual households take into consideration to evacuate during hurricanes. In designing the contingent scenarios for evacuation, we randomly assign varying levels of hurricane characteristics (wind speed, lead time for landfall and the height of storm surge) combined with different types of emergency management options (voluntary versus mandatory evacuation order and a voucher with varying amounts to cover evacuation expenses). Findings indicate that individual households respond, in a non-linear fashion, to the intensity of hurricanes when making evacuation decisions. Respondents are also more likely to evacuate when the storm surge reaches a certain threshold. In terms of policy interventions, mandatory evacuation orders are more effective to increase the likelihood of evacuation. The potential intervention in the form of providing evacuation vouchers to assist households to cover their expenses (e.g. for food, water, transportation and lodging) also seems effective. We discuss policy implications of our findings.     

A Joint Latent-Class Model: Combining Likert-Scale Preference Statements With Choice Data to Harvest Preference Heterogeneity

In addition to choice questions (revealed and stated choices), preference surveys typically include other questions that provide information about preferences. Preference-statement data include questions on the importance of different attributes of a good or the extent of agreement with a particular statement. The intent of this paper is to model and jointly estimate preference heterogeneity using stated-preference choice data and preference-statement data. The starting point for this analysis is the belief that the individual has preferences, and both his/her choices and preference statements are manifestations of those preferences. Our modeling contribution is linking the choice data and preference-statement data in a latent-class framework. Estimation is straightforward using the E-M algorithm, even though our model has hundreds of preference parameters. Our estimates demonstrate that: (1) within a preference class, the importance anglers associate with different Green Bay site characteristics is in accordance with their responses to the preference statements; (2) estimated across-class utility parameters for fishing Green Bay are affected by the preference-statement data; (3) estimated across-class preference-statement response probabilities are affected by the inclusion of the choice data; and (4) both data sets influence the number of classes and the probability of belonging to a class as a function of the individual’s type.     

A Nonparametric Revealed Preference Approach to Measuring the Value of Environmental Quality

We develop an approach to valuing non-market goods using nonparametric revealed preference analysis. We show how nonparametric methods can also be used to bound the welfare effects of changes in the provision of a non-market good. Our main context is one in which the non-market good affects the marginal utility of consuming a related market good. This can also be framed as a shift in the taste for, or quality of, the market good. A systematic approach for incorporating quality/taste variation into a revealed preference framework for heterogeneous consumers is developed. This enables the recovery of the minimal variation in quality required to rationalise observed choices of related market goods. The variation in quality appears as a adjustment to the price for related market goods which then allows a revealed preference approach to bounding compensation measures of welfare effects to be applied.     

Understanding Heterogeneous Preferences in Random Utility Models: A Latent Class Approach

A finite mixture approach toconditional logit models is developed in whichlatent classes are used to promoteunderstanding of systematic heterogeneity. The model is applied to wilderness recreationin which a branded choice experiment involvingchoice of one park from a demand system wasadministered to a sample of recreationists. The basis of membership in the classes orsegments in the sample involved attitudinalmeasures of motivations for taking a trip, aswell as their stated preferences overwilderness park attributes. The econometricanalysis suggested that four classes of peopleexist in the sample. Using the model toexamine welfare measures of some hypotheticalpolicy changes identified markedly differentwelfare effects than the standard singlesegment model, and provided insight into thedifferential impact of alternative policies.     

Protesting and Justifying: A Latent Class Model for Contingent Valuation with Attitudinal Data

This article develops a latent class model for estimating willingness-to-pay (WTP) for public goods using simultaneously contingent valuation (CV) and attitudinal data to identify individuals with similar characteristics, such as WTP and protest attitudes. We find evidence that the answer to the CV question influences the responses to the attitudinal questions. In our case, this influence reflects rational behavior (budget constraint issues) and justification biases related to protest reasons, such as lack of trust in institutions or fairness issues with respect to the distribution of the burden of preservation. The results from our empirical application confirm the importance of accounting for those biases.     

A Common Nomenclature for Stated Preference Elicitation Approaches

It is often difficult to determine what actually was done in work involving data collected with stated preference surveys because the terms used to describe various procedures have ambiguous and sometimes conflicting meanings. Further, terms used to describe data collection procedures often are confounded with terms used to describe statistical techniques. We call for the use of a common nomenclature to describe what was done in a data collection effort for stated preference studies, and take a first step at setting out such nomenclature. We only seek to improve clarity in the communication of research results and take no position here on appropriateness of particular procedures.     

A Meta-analysis of Hypothetical Bias in Stated Preference Valuation

Individuals are widely believed to overstate their economic valuation of a good by a factor of two or three. This paper reports the results of a meta-analysis of hypothetical bias in 28 stated preference valuation studies that report monetary willingness-to-pay and used the same mechanism for eliciting both hypothetical and actual values. The papers generated 83 observations with a median ratio of hypothetical to actual value of only 1.35, and the distribution has severe positive skewness. We find that a choice-based elicitation mechanism is important in reducing bias. We provide some evidence that the use of student subjects may be a source of bias, but since this variable is highly correlated with group experimental settings, firm conclusions cannot be drawn. There is some weak evidence that bias increases when public goods are being valued, and that some calibration methods may be effective at reducing bias. However, results are quite sensitive to model specification, which will remain a problem until a comprehensive theory of hypothetical bias is developed.JEL classifications: C9, H41, Q26, Q28     

A Latent Class Nested Logit Model for Rank-Ordered Data with Application to Cork Oak Reforestation

We analyze stated ranking data collected from recreational visitors to the Alcornocales Natural Park (ANP) in Spain. The ANP is a large protected area which comprises mainly cork oak woodlands. The visitors ranked cork oak reforestation programs delivering different sets of environmental (reforestation technique, biodiversity, forest surface) and social (jobs and recreation sites created) outcomes. We specify a novel latent class nested logit model for rank-ordered data to estimate the distribution of willingness-to-pay for each outcome. Our modeling approach jointly exploits recent advances in discrete choice methods. The results suggest that prioritizing biodiversity would increase certainty over public support for a reforestation program. In addition, a substantial fraction of the visitor population are willing to pay more for the social outcomes than the environmental outcomes, whereas the existing reforestation subsidies are often justified by the environmental outcomes alone.     

Environmental Risk and Averting Behavior: Predictive Validity of Jointly Estimated Revealed and Stated Behavior Data

We conduct predictive validity tests using revealed and stated behavior data from a panel survey of North Carolina coastal households. The application is to hurricane evacuation behavior. Data was initially collected after Hurricane Bonnie led to hurricane evacuations in North Carolina in 1998. Respondents were asked for their behavioral intentions if a hurricane threatened the North Carolina coast during the 1999 hurricane season. Following Hurricanes Dennis and Floyd in 1999, a follow-up survey was conducted to see if respondents behaved as they intended. A jointly estimated revealed and stated behavior model indicates that the hypothetical and real evacuation behavior is based on the same choice process. Using predictions from this model with a hypothetical bias correction, we find that it predicts actual evacuation behavior with a small forecast error. These results suggest that stated behavior data has some degree of predictive validity.     

A Split-Sample Revealed and Stated Preference Demand Model to Examine Homogenous Subgroup Consumer Behavior Responses to Information and Food Safety Technology Treatments

The combination and joint estimation of revealed and stated preference (RP/SP) data approach to examining consumer preferences to relevant policy-based measures typically fail to account for heterogeneity in the data by considering behavior of the average individual. However, in policy-based analyses, where the research is often driven by understanding how different individuals react to different or similar scenarios, a preferred approach would be to analyze preferences of homogenous population subgroups. We accomplish this by developing a split-sample RP/SP analysis that examines whether homogenous subgroups of the population, based on individual health and behavioral characteristics, respond differently to health-risk information and new food safety technology. The ongoing efforts by the US Food and Drug Administration (FDA) to reduce illness and death associated with consuming raw Gulf of Mexico oysters provide an ideal platform for the analysis as the health risks only relate to a very specific consumer subgroup. Results from split-sample demand models indicate that educational information treatments cause vulnerable at-risk consumers to reduce their oyster demand, implying that a more structured approach to disseminating the brochures to the at-risk population could have the desired result of reducing annual illness levels. Also, findings across all subgroups provide strong empirical evidence that the new FDA policy requiring processing technology to be used in oyster production will have a detrimental effect on the oyster industry.     

Anomalies and Stated Preference Techniques: A Framework for a Discussion of Coping Strategies

This paper introduces a symposium on the issue of how stated preference (SP) research can best cope with ‘anomalies’ (i.e. systematic deviations from the predictions of standard economic theory) in survey responses. It proposes a framework for constructive debate, recognising (i) the legitimate aspirations of SP research, (ii) the relevance of evidence from sources other than best-practice SP, and (iii) the precautionary value of investigating strategies for coping with suspected anomalies, even if questions about the robustness of anomalies have not been finally resolved. Five alternative coping strategies, discussed in more detail in the symposium, are briefly introduced. JEL classifications: D61, D63, Q51     

A Meta-study Investigating the Sources of Protest Behaviour in Stated Preference Surveys

It is a well-known empirical finding that some percentage of respondents participating in Stated Preference surveys will not give responses that reflect their true preferences. One reason is protest behaviour. If the distribution of protest responses is not independent of respondent or survey characteristics, then simply expelling protesters from surveys can lead to sample selection bias. Furthermore, WTP estimates will not be comparable across surveys. This paper seeks to explore potential causes of protest behaviour through a meta-study based on full datasets from 38 different surveys. The objective of the study is to examine the effect of respondent specific variables as well as survey specific variables on protest behaviour. Our results suggest that some of the differences in WTP typically observed between different demographic groups, different elicitation formats and different question formats might actually be attributed to inherent differences in the propensity to protest. Our results indicate that the propensity for respondents to exhibit protest behaviour when asked a stated preference type valuation question depends on a number of specific factors, respondent specific as well as survey specific—knowledge which could be used in order to reduce protest behaviour.     

Convergent Validity of Revealed and Stated Recreation Behavior with Quality Change: A Comparison of Multiple and Single Site Demands

We consider the convergent validity of several demand models using beach recreation data. Two models employ multiple site data: a count data demand system model and the Kuhn–Tucker demand system model. We explore the role of existing variation in beach width in explaining trip choices, and analyze a hypothetical 100 foot increase in beach width. We compare these models to a single equation model where we jointly estimate revealed and stated preference trip data, and focus on a hypothetical scenario considering a 100 foot increase in beach width. In each case we develop estimates of the change in beach visits and the welfare impacts from the increase in width. The trip change estimates from two of the three models are similar and convergent valid, though the willingness to pay estimates differ in magnitude.     

Australia, Japan and the OECD: GDP Rankings and Revealed Preference

Data on volumes and prices of consumption and investment are used to compare Australian real GDP for 1990 with the other OECD countries. Australian consumption patterns, including leisure, and price structure are very different from most other countries and especially from those of Japan. The Australian bundle of consumption, investment and leisure is revealed preferred to that of Japan and a number of other countries which are conventionally ranked above Australia in comparisons of real GDP per capita at international prices.     

As Time Goes By: Examination of Temporal Stability Across Stated Preference Question Formats

We evaluate the temporal stability of willingness-to-pay values obtained from two identical stated-preference surveys undertaken in 2004 and 2012. The surveys employ two discrete-choice elicitation formats to value reductions in two health endpoints related to tap water consumption, and are explicitly designed to make direct comparisons between formats. We find no significant differences in real willingness-to-pay between the two surveys for either format—a result that is robust to the structure of heterogeneity used to model respondent preferences. Findings lend support, at least within the context of tap water quality, to the temporal transfer of nonmarket benefits over extended time periods.     

Land Use Change with Spatially Explicit Data: A Dynamic Approach

Most of the economic literature that uses spatially-explicit data to estimate the determinants of land-use change is limited to static models and cross-sectional data sets. Recent attempts to move to a more dynamic analysis include using panel data sets and survival analysis. In this study, we use a discrete choice dynamic model of land-use where the agent’s choices are regarded as the solution to a dynamic optimization problem. The irreversibility of some decisions, expectations about future prices, and forward-looking behavior of the land operator can all be accounted for. Our results show that a model specification that incorporates some of the complexities of the decision process improves upon results found in the existing literature. First, prediction accuracy of land use change is superior to any of the existing models. Second, we demonstrate that models that do not account for transactions costs tend to overestimate the effects of changes in transportation costs.     

Context and the VSL: Evidence from a Stated Preference Study in Italy and the Czech Republic

We report on the results of a survey based on conjoint choice experiments that was specifically designed to investigate the effect of context on the Value of a Statistical Life (VSL), an important input into the calculation of the mortality benefits of environmental policies that reduce premature mortality. We define “context” broadly to include (1) the cause of death (respiratory illness, cancer, road traffic accident), (2) the beneficiary of the risk reduction (adult v. child), and (3) the mode of provision of the risk reduction (public program v. private good). The survey was conducted following similar protocols in Italy and the Czech Republic. When do not distinguish for the cause of death, child and adult VSL are not significantly different from one another in Italy, and the difference is weak in the Czech sample. When we distinguish for the cause of death, we find that child and adult VSLs are different at the 1% level for respiratory illnesses and road-traffic accidents, but do not differ for cancer risks. We find evidence of a “cancer premium” and a “public program premium.” In both countries, the marginal utility of income is about 20% lower among wealthier people, which makes the VSL about 20% higher among respondents with incomes above the sample average. The discount rate implicit in people’s choices is effectively zero. We conclude that there is heterogeneity in the VSL, and that such heterogeneity is primarily driven by risk characteristics mode of delivery of the risk reduction, and income, while other individual characteristics of the respondent (e.g., age and education) are less important. For the most part, our results are in agreement with environmental policy analyses that use the same VSL for children and adults, and that apply a cancer premium.     

A Revealed Preference Theory for Non-expected Utility on "Certain × Uncertain" Consumption Pairs

In this paper we consider a two-period decision problem, where the feasible set is the set of "certain × uncertain" consumption pairs. That is, the decision-maker chooses ( x , m ) in a feasible set, where x is a certain first-period consumption and m is a random second-period consumption, a Borel probability measure on the set of real numbers. The purpose of this paper is to present revealed preference theory for non-expected utility on "certain × uncertain" consumption pairs. We present necessary and sufficient conditions for the data to be consistent with some non-expected utility functions.
JEL Classification Numbers: D11, D81, D91