Catálogo de publicaciones - libros

In models with unobserved taste heterogeneity, distributional assumptions can be placed in two ways: (1) by specifying the distribution of coefficients in the utility function and deriving the distribution of willingness to pay (), or (2) by specifying the distribution of and deriving the distribution of coefficients. In general the two approaches are equivalent, in that any mutually compatible distributions for coefficients and can be represented in either way. However, in practice, convenient distributions, such as normal or log-normal, are usually specified, and these convenient distributions have different implications when placed on ’s than on coefficients. We compare models that use normal and log-normal distributions for coefficients (called models in preference space) with models using these distributions for (called models in space). We find that the models in preference space fit the data better but provide less reasonable distributions of than the models in space. Our findings suggests that further work is needed to identify distributions that either fit better when applied in space or imply more reasonable distributions of when applied in preference space.

A number of papers have recently contrasted classical inference estimation methods for logit models with Bayesian methods. It has been argued that two particularly appealing features of the Bayesian approach are its relative simplicity in estimation, and its ability to derive, individual-specific willingness to pay () measures that are less problematic than the classical approaches in terms of extreme values and unexpected signs. This paper challenges this claim by deriving both population derived measures and individual-specific values based on the classical mixed logit model, establishing the extent of unacceptable valuations. Our aim is not to estimate Bayesian contrasts per se but to show that the classical inference approach is likewise straightforward — indeed the individual-specific estimates are a by-product of the parameter estimation process. We also reveal the benefits of calculating measures from ratios of individual parameters which are behaviourally more appealing approximations to the true values of each individual, in contrast to draws from population distributions that run the risk of allocating two parameters that are poorly juxtaposed in a relative sense, resulting in extreme value estimates. Our results suggest that while extreme values and unexpected signs cannot be ruled out (nor can they in the Bayesian framework), the overall superiority of the Bayesian method appears overstated. Both approaches have merit.

We use a repeated dichotomous choice contingent valuation survey to elicit households’ willingness to pay to a void unannounced interruptions in electricity service. The data pose multiple econometric challenges including: correlated responses for a given household, heteroskedastic errors, and a willingness to pay distribution with large mass near zero. We address these issues by combining a gamma distribution for outage costs with a lognormally distributed scale parameter defined as a function of household characteristics, outage attributes, outage history, and random coefficients. The model is estimated through simulated maximum likelihood. We demonstrate that cost estimates are sensitive to the interaction of attributes of previously experienced and hypothetical interruptions.

Research in the area of discrete choice modelling can be split into two broad categories; applications accounting for the prevalence of unobserved inter-alternative correlation, and applications concerned with the representation of random inter-agent taste heterogeneity. The difference between these two is however not as clear-cut as this division might suggest, and there is in fact a high risk of confounding between the two phenomena. In this article, we investigate the potential of mixed Generalised Extreme Value (GEV) models to simultaneously account for the two phenomena, using a Stated Preference (SP) dataset for mode-choice in Switzerland. Initial results using more basic modelling techniques reveal the presence of both correlation and random taste heterogeneity. The subsequent use of mixed GEV models on this dataset leads to important gains in performance over the use of the more basic models. However, the results also show that, by simultaneously accounting for correlation and random taste heterogeneity, the scope to retrieve the individual phenomena is reduced. This shows that a failure to account for the potential impacts of either of the two phenomena can lead to erroneous conclusions about the existence of the other phenomenon. This is a strong indication that the use of mixed GEV models to jointly explain random taste heterogeneity and inter-alternative correlation in a common modelling framework should be encouraged in the case where the nature of the error-structure is not clear a priori.

The chapter presents an approach for simultaneously estimating farmers’ decisions to accept incentive payments in return for adopting a bundle of environmentally benign best management practices. Using the results of a multinomial probit analysis of surveys of over 1,000 farmers facing five adoption decisions in a voluntary program, we show how the farmers’ perceptions of the desirability of various bundles change with the offer amounts and with which practices are offered in the bundle. We also demonstrate an estimator for the mean minimum willingness to accept for adoption of a practice conditional on the cost share offers for other practices.

Although the estimation of flexible multinomial discrete choice models generally needs the incorporation of simulation methods, their application is recently common in environmental and resource economics such as in many other economic disciplines (e.g. transportation economics). Based on a firm level data set of the German manufacturing sector, this paper examines determinants of environmental innovations by comparing the estimation results in flexible multinomial probit models and restrictive multinomial logit and independent probit models. The analysis of the two latter models implies that some specific environmental organizational measures, technological opportunities, and market pull factors have a significantly positive effect on both environmental product and process innovations. Taking this into consideration, the flexible multinomial probit model analysis provides few new insights since the simulated maximum likelihood estimations are rather unreliable as a consequence of the sole inclusion of firm-specific characteristics as explanatory variables. In this respect, the incorporation of simulation methods into the maximum likelihood estimations is not crucial since the problems do not decrease if the number of random draws in the considered Geweke-Hajivassiliou-Keane simulator rises. Furthermore, the difficulties grow if the number of choice alternatives increases. It can therefore be concluded that the applicability of these flexible multinomial discrete choice models without the incorporation of choice-specific attributes as explanatory variables is rather limited in practice.

The use of a joint normal distribution for partworths is computationally attractive, particularly with Bayesian MCMC procedures, and yet is unrealistic for any attribute whose partworth is logically bounded (e.g., is necessarily positive or cannot be unboundedly large). A mixed logit is specified with partworths that are transformations of normally distributed terms, where the transformation induces bounds; examples include censored normals, log-normals, and distributions which are bounded on both sides. The model retains the computational advantages of joint normals while providing greater flexibility for the distributions of correlated partworths. The method is applied to data on customers’ choice among vehicles in stated choice experiments. The flexibility that the transformations allow is found to greatly improve the model, both in terms of fit and plausibility, without appreciably increasing the computational burden.

In this chapter we summarize recent advances with Kuhn-Tucker demand system approaches to non-market valuation. Over the past five years, simulation-based estimation and welfare calculation strategies have been developed that enable the Kuhn-Tucker framework to address policy-relevant valuation questions in applications with many quality-differentiated goods. We review these modeling innovations in the context of three generic Kuhn-Tucker specifications that differ in terms of their ability to account for unobserved preference heterogeneity. For illustration, we apply the alternative specifications to Canadian moose hunting data and present parameter and welfare estimates. We conclude the chapter by suggesting important areas for future research within the Kuhn-Tucker framework.

We present, pedagogically, the Bayesian approach to composed error models under alternative, hierarchical characterizations; demonstrate, briefly, the Bayesian approach to model comparison using recent advances in Markov Chain Monte Carlo (MCMC) methods; and illustrate, empirically, the value of these techniques to natural resource economics and coastal fisheries management, in particular. The Bayesian approach to fisheries efficiency analysis is interesting for at least three reasons. First, it is a robust and highly flexible alternative to commonly applied, frequentist procedures, which dominate the literature. Second, the Bayesian approach is extremely simple to implement, requiring only a modest addition to most natural-resource economist tool-kits. Third, despite its attractions, applications of Bayesian methodology in coastal fisheries management are few.

Bayesian econometric approaches to modeling non-market valuation data have not often been applied, but they offer a number of potential advantages. Bayesian models incorporate prior information often available in the form of past studies or pre-tests in Stated Preference (SP) based valuation studies; model computations are easily and efficiently performed within an intuitively constructed Markov chain Monte Carlo framework; and asymptotic approximations, unreasonable for the relatively small sample sizes seen in some SP data sets, need not be invoked to draw (posterior) inferences. With these issues in mind, we illustrate computationally feasible approaches for fitting a series of surveys in a sequential manner, and for comparing a variety of models within the Bayesian paradigm. We apply these approaches to a series of SP surveys that examined policies to conserve old growth forests, northern spotted owls, and salmon in the U.S. Pacific Northwest.