Evaluación de endogeneidad producida por omisión de variables psicosociales en modelos de elección discreta.
Loading...
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
La endogeneidad es un problema inevitable en modelos de elección discreta que causa estimación de parámetros inconsistentes. Dicha inconsistencia se puede ver reflejada en indicadores económicos que se utilizan para evaluar proyectos de transporte, como el valor subjetivo del tiempo y elasticidades. Por otro lado, se tiene evidencia del impacto que tiene la inclusión de variables psicosociales, tales como el hábito, la actitud y el afecto, en modelos de partición modal, cuya omisión podría causar endogeneidad. Estos antecedentes motivan estudiar la endogeneidad por omisión de variables psicosociales en modelos de partición modal.
Se proponen dos metodologías para evaluar endogeneidad, el Método de los Indicadores Múltiples (MIS) y el Método de Variables Latentes Integrado (ICLV). Para el primero, no se han encontrado aplicaciones con respecto a variables psicosociales. Más aún, se tienen muy pocos estudios que incluyen varias de estas variables de manera simultánea. Además, se proponen ambas metodologías interactuando multiplicativamente con el tiempo de viaje. Los datos provienen de encuestas realizadas a funcionarios, académicos y alumnos de la Universidad de Concepción y de la Universidad de la Frontera. Los modos estudiados corresponden al automóvil, la bicicleta y el bus.
Se demuestra que el afecto, la actitud y el hábito tienen un impacto significativo en la elección de modo de transporte, para todos los modelos estimados. Esto genera cambios sustanciales en los indicadores mencionados. Se tiene que el valor subjetivo del tiempo aumenta si se consideran las variables mencionadas aditivamente, pero disminuye si se consideran interactuando con el tiempo de viaje. Las elasticidades del tiempo de viaje se ven disminuidas para la bicicleta y para el automóvil al usar las metodologías propuestas, mientras que para el bus se ve aumentada. Por último, la elasticidad del costo también se ve disminuida para el automóvil y aumentada para el bus. Esto podría tener una repercusión directa en la evaluación de políticas públicas, dado que estas dependen de la partición modal, la cual podría estar sesgada.
Endogeneity is an unavoidable problem in discrete choice models that causes inconsistent estimation of parameters. This can be seen reflected in economic indicators that are used to evaluate infrastructure projects, such as the subjective value of time and elasticities. On the other hand, there is evidence of the impact of the inclusion of psychosocial variables in discrete choice models, such as affection, attitude and habit. The omission of these variables could cause endogeneity. This motivates studying endogeneity due to omission of psychosocial variables in discrete choice models. Two methodologies are proposed to evaluate endogeneity, the Multiple Indicator Solution (MIS) and the Integrated Choice Latent Variable Method (ICLV). For the first method, no applications have been found regarding psychosocial variables. Furthermore, there are very few studies that include multiple psychosocial variables simultaneously. Moreover, both methodologies are proposed interacting as a product with travel time. The data comes from surveys carried out among officials, academics and students of Universidad de Concepción and the Universidad de la Frontera. The modes studied correspond to car, bicycle, and bus. It is shown that affect, attitude, and habit have a significant impact on the choice of transport mode, for all the estimated models. This generates substantial changes in the indicators mentioned. The subjective value of time increases if the psychosocial variables are considered in an additive way but decreases if they are considered interacting with travel time. The elasticities of travel time decrease for the bicycle and for the car when using the proposed methodologies, while for the bus it increases. Finally, cost elasticity is also decreases for the car and increases for the bus. This could have a direct impact in evaluating public policies, since modal share is required for their assessment, and modal share may be biased.
Endogeneity is an unavoidable problem in discrete choice models that causes inconsistent estimation of parameters. This can be seen reflected in economic indicators that are used to evaluate infrastructure projects, such as the subjective value of time and elasticities. On the other hand, there is evidence of the impact of the inclusion of psychosocial variables in discrete choice models, such as affection, attitude and habit. The omission of these variables could cause endogeneity. This motivates studying endogeneity due to omission of psychosocial variables in discrete choice models. Two methodologies are proposed to evaluate endogeneity, the Multiple Indicator Solution (MIS) and the Integrated Choice Latent Variable Method (ICLV). For the first method, no applications have been found regarding psychosocial variables. Furthermore, there are very few studies that include multiple psychosocial variables simultaneously. Moreover, both methodologies are proposed interacting as a product with travel time. The data comes from surveys carried out among officials, academics and students of Universidad de Concepción and the Universidad de la Frontera. The modes studied correspond to car, bicycle, and bus. It is shown that affect, attitude, and habit have a significant impact on the choice of transport mode, for all the estimated models. This generates substantial changes in the indicators mentioned. The subjective value of time increases if the psychosocial variables are considered in an additive way but decreases if they are considered interacting with travel time. The elasticities of travel time decrease for the bicycle and for the car when using the proposed methodologies, while for the bus it increases. Finally, cost elasticity is also decreases for the car and increases for the bus. This could have a direct impact in evaluating public policies, since modal share is required for their assessment, and modal share may be biased.
Description
Tesis presentada para optar al grado de Magíster en Ciencias de la Ingeniería con Mención en Ingeniería Civil
Keywords
Endogeneidad, Medios de transporte