Application Of Discrete Choice Models Research Articles

A control-function correction for endogeneity in random coefficients models: The case of choice-based recommender systems

Applications of discrete choice models in personalization are becoming increasingly popular among researchers and practitioners. However, in such systems, when users are presented with successive menus (or choice situations), the alternatives and attributes in each menu depend on the choices made by the user in the previous menus. This gives rise to endogeneity which can result in inconsistent estimates. Our companion paper, Danaf et al. (2020), showed that the estimates are only consistent when the entire choice history of each user is included in estimation. However, this might not be feasible because of computational constraints or data availability. In this paper, we present a control-function (CF) correction for the cases where the choice history cannot be included in estimation. Our method uses the attributes of non-personalized attributes as instruments, and applies the CF correction by including interactions between the explanatory variables and the first stage residuals. Estimation can be done either sequentially or simultaneously, however, the latter is more efficient (if the model reflects the true data generating process). This method is able to recover the population means of the distributed coefficients, especially with a long choice history. The variances are underestimated, because part of the inter-consumer variability is explained by the residuals, which are included in the systematic utility. However, the population variances can be computed from the estimation results. The modified utility equations (which include the residuals) can be used in forecasting and model application, and provide superior fit and predictions.

Empirical properties and identification of adaptive learning models in behavioral game theory

This paper studies the empirical properties and identification for a broad class of adaptive learning models in games (e.g., experience-weighted attraction (EWA), payoff assessment, and impulse-matching models). I first show that adding a constant c≠0 to the utility function could affect players’ choices during the learning periods. This property is different from standard theories, such as expected utility theory and discrete choice models, where adding c≠0 preserves the same choice. This result has two further implications. First, the constant c added to the utility function could have a structural interpretation, and this paper discusses how to interpret it. Second, the estimation of learning models usually normalizes the utility of one action profile to be zero (i.e., c=0); for instance, the utility of zero dollars is zero, as commonly imposed in experimental economics. This normalization is innocuous in most applications of discrete choice models but could be misspecified in the context of learning. To deal with the above issue, this paper proposes to specify a player’s utility as an unknown function of the action profile, without imposing the above normalization. Equivalently, c is specified as an unknown parameter to be estimated. Under weak conditions, I obtain the joint identification of both utility and learning parameters. Finally, I illustrate the importance of c in the estimation of learning models by conducting a Monte Carlo experiment and revisiting the experimental study by Feri et al. (2010).

A hybrid discrete choice model to understand the effect of public policy on fare evasion discouragement in Bogotá's Bus Rapid Transit

• The study was carried out with a sample of actual evaders, which is unique in fare evasion studies. • Is critical to understand which observable and non-observable factors motivate individuals to evade fares. • Age, the evasion records, and some personality traits influence the fare evasion behavior. • Contributes with a novel application of discrete choice models to study fare evaders behavior. • Draw relevant policy implications for a large public transport system in South America. Fare evasion is currently a significant concern for public transport systems worldwide. Evasion causes huge economic losses to operators and cities, but also negatively influences the system’s security perceptions. Despite its importance, there are few efforts to understand which observable and non-observable factors motivate individuals to evade fares in a public transport system. Most studies dealing with fare evasion evaluate the effectiveness of several policies, quantify the amount of evasion, or correlate fare evasion with system satisfaction using aggregate data. This paper proposes a Hybrid Discrete Choice model to understand current evaders’ behavior considering sociodemographic, fare evasion records, trip characteristics, user satisfaction with the transport system, and perceptual indicators. The model was estimated using a unique stated preferences survey applied to 324 fare evaders of Transmilenio, the Bogotá (Colombia) Bus Rapid Transit system, in mid-2019. This is an interesting case study because, in 2018, the evasion rate was measured for the first time and it was discovered that evaders comprised approximately 15% of their users on weekdays. The contribution of this research is twofold. First, results highlight the importance of considering observable and latent variables to understand individual behavior in examining fare evasion in similar public transport systems. Age and the evasion records resulted in being the most important observable variables in our case study. Latent variables suggest that the more the user is satisfied with the transport system, the less likely is to evade fares. However, we found that personality traits, such as individual rule-unconsciousness, moderate the effect of satisfaction on the evasion behavior. Second, our results provide evidence that the discussion between certainty and severity of punishment on fare evasion should not necessarily be centered on whether to exacerbate one of these factors. A combination of both, however, should reduce the recidivism rate.

An Intercity Freight Mode Choice Model : A Case Study of High Speed Rail Link Northern Line Thailand (Bangkok – Chiangmai)

This paper presents a model capturing the intercity freight mode choice behaviour in high-speed rail system from Bangkok to Chiangmai. The model is developed based on the stated preference data collected from 800 freight operators, wholesaler, retailer, and people. The results show that, when the high-speed rail system from Bangkok to Chiangmai are developing in the future, the suitable products for high-speed rail system are types of an express mail service (EMS), air cargo, gold, jewellery, gold accessories, computer circuit boards, high prices agricultural products such as fruit, and flowers. Most of determining the selling price will fluctuate with the speed of transport and damage impairment of the product. With the application of discrete choice models, the results show that, aside from travel cost and time, loading and unloading, delays time, frequency are statistically significant. The application of model indicated that the holder and freight forwarder which in the line of high-speed rail (Bangkok – Chiangmai) tend to use rail-transport such as double-track rail is 27.71%, high-speed rail is 11.18% and the most is trucks 56.51% which is a policy development point loading and unloading, multimodal transportation efficiency and safety of the portion of the freight high-speed rail increased.

The Economically Active People in the Transport Process

Traffic-planning process, the traffic management process requires continuous collection and analysis of traffic data. The direct measurement usually determines the volume or speed of the traffic flows. In comprehensive traffic, assessment is necessary to know the causes of mobility and its characteristics, the group of habitants and purpose of the trips, above all. The article deals with the mobility characteristics of economically active people for work and their application to calculations using discrete functions. The data are based on an extensive transport-sociological survey of the mobility of the inhabitants of the Žilina region. The first part of the article is devoted to the basic survey data. The core of the article is the application of discrete choice models in the process of trips generation and modal split.

Context dependent process heuristics and choice analysis – A note on two interacting themes linked to behavioural realism

This note is written to remind us of the recognition of behavioural realism in the economic literature over 60 years ago, which seems to have been given limited recognition in the formal axiomatic development of economic theories commonly associated with the application of discrete choice models, especially where the interest is on obtaining welfare measures such as willingness to pay estimates of specific attributes for use in cost-benefit analysis. We need to be reminded from time to time of the importance of context dependency in defining a choice problem and the value that behavioural realism can add despite risks of violating often some axioms of economic rationality and utility maximisation.

A Note on Estimating Network Dependence in a Discrete Choice Model

Discrete choice model is probably one of the most popularly used statistical methods in practice. The common feature of this model is that it considers the behavioral factors of a person and the assumption of independent individuals. However, this widely accepted assumption seems problematic because human beings do not live in isolation. They interact with each other and form complex networks. Then the application of discrete choice model to network data will allow for network dependence in a general framework. In this paper, we focus on a discrete choice model with probit error which is specified as a latent spatial autoregressive model (SAR). This model could be viewed as a natural extension of the classical SAR model. The key difference is that the network dependence is latent and unobservable. Instead, it could be measured by a binary response variable. Parameter estimation then becomes a challenging task due to the complicated objective function. Following the idea of composite likelihood, an approximated paired maximum likelihood estimator (APMLE) is developed. Numerical studies are carried out to assess the finite sample performance of the proposed estimator. Finally a real dataset of Sina Weibo is analyzed for illustration purpose.

Theoretical, methodological and practical issues of choice models

This paper analyzes and synthesizes the fundamentals of discrete choice models. This paper alsodiscusses the basic concept and theory underlying the econometrics of discrete choice, specific choicemodels, estimation method, model building and tests, and applications of discrete choice models. Thiswork highlights the relationship between economic theory and discrete choice models: how economictheory contributes to choice modeling and vice versa.
 Keywords: Discrete choice models; Random utility maximization; Decision makers; Utility function;Model formulation

A note on estimating network dependence in a discrete choice model

Discrete choice model is probably one of the most popularly used statistical methods in practice. The common feature of this model is that it considers the behavioral factors of a person and the assumption of independent individuals. However, this widely accepted assumption seems problematic because human beings do not live in isolation. They interact with each other and form complex networks. Then the application of discrete choice model to network data will allow for network dependence in a general framework. In this paper, we focus on a discrete choice model with probit error which is specified as a latent spatial autoregressive model (SAR). This model could be viewed as a natural extension of the classical SAR model. The key difference is that the network dependence is latent and unobservable. Instead, it could be measured by a binary response variable. Parameter estimation then becomes a challenging task due to the complicated objective function. Following the idea of composite likelihood, an approximated paired maximum likelihood estimator (APMLE) is developed. Numerical studies are carried out to assess the finite sample performance of the proposed estimator. Finally a real dataset of Sina Weibo is analyzed for illustration purpose.

Intra-respondent Heterogeneity in a Stated Choice Survey on Wetland Conservation in Belarus: First Steps Towards Creating a Link with Uncertainty in Contingent Valuation

Applications of discrete choice models in environmental valuation increasingly use a random coefficient specification, such as mixed logit, to represent taste heterogeneity. The majority of applications rely on data containing multiple observations for each respondent, where a common assumption is that tastes stay constant across choices for the same respondent. We question this assumption and make use of a model developed in the transport field which allows tastes to vary over choices for each consumer in addition to variation across consumers. An empirical analysis making use of a stated choice dataset for wetland conservation in Belarus shows that superior performance is obtained by allowing jointly for the two types of heterogeneity and that recovery of these intra-respondent variations is not possible using standard approaches, such as allowing for scale heterogeneity across tasks. We show also that intra-respondent heterogeneity can be especially high for attributes which respondents are unfamiliar with, and that a failure to account for it can substantially affect welfare estimates. We interpret this as an indication that this heterogeneity relates primarily to uncertainty. Finally, we offer initial insights into the relationship between intra-respondent heterogeneity and findings on uncertainty in a contingent valuation context.

An Analysis of Residential Location Choice Behavior in Bangkok Metropolitan Region: An Application of Discrete Choice Models for the Ranking of Alternatives

There has been an increasing interest in land use and transportation interaction in the past decades. Many substantial attempts have been made to empirically investigate the connection among those impacts can be interpreted in developed countries, yet few attempts in developing countries. The paper examines residential location choice behavior using a sample of households where their workplace is in the CBD of the Bangkok Metropolitan Region. A stated preference approach will be employed, where each of a respondent is asked to imagine moving to a new location. Discrete choice models, i.e., rank-ordered logit (ROL) and rank-ordered nested logit (RONL) will be applied to treat the residential location choice behavior. The results confirm the applicability of the ROL and RONL models and support similar behavioral interpretations from other empirical studies in the residential location choice literature. This information is important for suggesting appropriate policy such as promoting urban rail transit use.

On the role of unobserved preference heterogeneity in discrete choice models of labour supply

The aim of this paper is to analyse the impact of unobserved preference heterogeneity in empirical applications of discrete choice models of labour supply. Typically, unobserved heterogeneity is estimated either with continuous or discrete mixture models. However, in order to avoid estimation difficulties, most of the empirical analysis assumes a relatively constrained mixture, standard examples being models where only few coefficients are allowed to vary with independent normal distributions or with discrete distributions with few mass points. We compare labour supply elasticities obtained with these typical specifications of unobserved heterogeneity with those from a more general model that we are able to estimate through an EM algorithm for the nonparametric estimation of mixed models. Results show that labour supply elasticities change significantly with respect to a basic model without unobserved heterogeneity only when the joint distribution of the varying tastes is left completely unspecified.

Application of discrete choice models to choice-based revenue management problems: A cautionary note

Choice-based revenue management algorithms directly integrate a discrete choice model of customer behavior into the optimization function. In the classic formulation by Talluri and van Ryzin (TvR), the choice set is defined to include a no-purchase alternative, and the utility of this alternative is set to zero ‘without loss of generality’. We show that only a limited number of discrete choice models can be used with the TvR formulation, and that misuse of the TvR formulation can result in significant estimation and forecast errors. Based on the Matching Shares Property of multinomial logit (MNL) models, we show that estimated choice model parameters will match probabilities in the estimation data set for MNL models that include alternative-specific constants (ASCs). MNL models that exclude ASCs may be able to match probabilities, but in this case the analyst cannot blindly set the no-purchase utility to zero. This property holds regardless of which method is used to estimate choice model parameters.

Recovery of inter- and intra-personal heterogeneity using mixed logit models

Most applications of discrete choice models in transportation now utilise a random coefficient specification, such as mixed logit, to represent taste heterogeneity. However, little is known about the ability of these models to capture the heterogeneity in finite samples (as opposed to asymptotically). Also, due to the computational intensity of the standard estimation procedures, several alternative, less demanding methods have been proposed, and yet the relative accuracy of these methods has not been investigated. This is especially true in the context of work looking at joint inter-respondent and intra-respondent variation. This paper presents an overview of the various different estimators, gives insights into some of the theoretical properties, and analyses their performance in a large scale study on simulated data. In particular, we specify 31 different forms of heterogeneity, with multiple versions of each dataset, and with results from over 16,000 mixed logit estimation runs. The findings suggest that variation in tastes over consumers is captured by all the methods, including the simpler versions, at least when sample size is sufficiently large. When tastes vary over choice situations for each consumer, as well as over consumers, the ability of the methods to capture and differentiate the two sources of heterogeneity becomes more tenuous. Only the most computationally intensive approach is able to capture adequately the two sources of variation, but at the cost of very high run times. Our results highlight the difficulty of retrieving taste heterogeneity with only cross-sectional data, providing further evidence of the benefits of repeated choice data. Our findings also suggest that the data requirements of random coefficients models may be more substantial than is commonly assumed, further reinforcing concerns about small sample issues.

Applications of Discrete Choice Models – Selected papers from the 11th World Conference on Transport Research

Applications of Discrete Choice Models – Selected papers from the 11th World Conference on Transport Research

THE PURE CHARACTERISTICS DEMAND MODEL*

In this article, we consider a class of discrete choice models in which consumers care about a finite set of product characteristics. These models have been used extensively in the theoretical literature on product differentiation and the goal of this article is to translate them into a form that is useful for empirical work. Most recent econometric applications of discrete choice models implicitly let the dimension of the characteristic space increase with the number of products (they have “tastes for products”). The two models have different theoretical properties, and these, in turn, can have quite pronounced implications for both substitution patterns and for the welfare impacts of changes in the number and characteristics of the goods marketed. After developing those properties, we provide alternative algorithms for estimating the parameters of the pure characteristic model and compare their properties to those of the algorithm for estimating the model with tastes for products. We conclude with a series of Monte Carlo results. These are designed to illustrate: (i) the computational properties of the alternative algorithms for computing the pure characteristic model, and (ii) the differences in the implications of the pure characteristic model from the models with tastes for products.

Application of Discrete Choice Models to Predict White Crappie Temperature Selection in Two Missouri Impoundments

AbstractTemperature selection by white crappie Pomoxis annularis over 24 h was determined in two Missouri impoundments, Little Dixie Lake and Rocky Fork Lake, on six and five dates, respectively, during June‐August 2001 using temperature‐sensitive radio transmitters. Temperature and dissolved oxygen (DO) profiles showed that the habitable portions of the upper water columns associated with fish locations varied by as much as 6°C in temperature. The habitable portions of water columns were subdivided into three strata (warm, moderate, and cool) such that each stratum covered an equivalent temperature range (≤2°C). Discrete choice models (DCMs) were developed for each impoundment to predict the relative probabilities of white crappies selecting the different strata in response to stratum height, DO concentration, average temperature throughout the habitable water column (AVG), and fish weight. For each impoundment, a distinct DCM containing AVG as the sole predictor variable was most supported by the data. Cross‐verification indicated that the DCMs accurately predicted white crappie thermal strata selection within the impoundments for which they were developed but that they were inaccurate when applied across the impoundments. The DCM for Little Dixie Lake predicted that white crappies mostly select warm strata at AVGs less than 27°C and mostly select cool strata at AVGs of 27°C or more. In contrast, the DCM for Rocky Fork Lake predicted that white crappies continue to select warm strata until AVGs reach 30°C, select moderate strata at AVGs of 30–32°C, and select cool strata only when AVGs exceed 32°C. The predicted mean daily thermal experiences of the white crappies in each impoundment, obtained by multiplying the relative selection probabilities for the different strata by the average temperatures in those strata, remained within 0.5°C of the mean of observed temperatures experienced by white crappies. These findings indicate that discrete choice modeling provides a novel and effective approach for predicting the daily thermal experiences of fish.

Impacts of Paddy-Field Consolidation Projects on Farmland Rental Transactions: Application of Discrete Choice Model

In order to reallocate farmland to efficient farmers, improvement in Farmland Rental Transactions (FRT) cooperating with public investment, Paddy-field Consolidation (PC) projects, is one of the most important policy issues in Asian monsoon regions as well as in Japan. Unfortunately, there are few empirical studies on FRT due to a lack of flexible data under regulated markets. This study aims to analyze FRT and to evaluate the effect of PC projects by modeling both supply and demand sides of farmland renting with micro-data from a discrete choice type questionnaire. Empirical results show that (i) effects of PC project appear as a remarkable increase in the rental rate and a moderate increase in the rental agreement level, (ii) there is an economic inefficiency in the Japanese rental market, but such an inefficiency can be reduced by PC project implementation, and (iii) regional differences in project effects are caused by differences in agricultural and social situations.

THE APPLICATION OF DISCRETE CHOICE MODELS TO WILDLIFE RESOURCE SELECTION STUDIES

This paper introduces the theory and application of discrete choice models to resource selection studies. Discrete choice models calculate the probability of an individual selecting a resource as a function of the attributes of that resource and all other available resources. The data for these attributes may be continuous or categorical. When availability is the same for all individuals and only two resources are available, the multinomial logit discrete choice model reduces to the logistic model. Discrete choice models and advances in GIS technology give the researcher flexibility in defining resource availability separately for each individual over time and space. The output of the discrete choice approach also provides managers with a tool to explore the effects of potential management actions and provides researchers with new hypotheses deserving of further investigation. To illustrate the application of discrete choice models to resource selection studies, we present a case study of summer diurnal bed site selection by elk (Cervus elaphus) in Custer State Park, South Dakota, United States. The results demonstrate the importance of factors relating to thermal regulation, hiding cover, and potentially forage, in elk bed site selection in this region.

Modeling Methods for Discrete Choice Analysis

This paper introduces new forms, sampling and estimation approaches fordiscrete choice models. The new models include behavioral specifications oflatent class choice models, multinomial probit, hybrid logit, andnon-parametric methods. Recent contributions also include new specializedchoice based sample designs that permit greater efficiency in datacollection. Finally, the paper describes recent developments in the use ofsimulation methods for model estimation. These developments are designed toallow the applications of discrete choice models to a wider variety ofdiscrete choice problems.