Publication: EL COMOVIMIENTO ENTRE LOS TIPOS DE CAMBIO Y EL PRECIO DE LAS ACCIONES EN LOS MERCADOS EMERGENTES DE AMÉRICA LATINA
Date
2017
Authors
ARAYA BARRAZA, RODRIGO ANDRÉS
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
El presente estudio tiene como fin estudiar y determinar la relación entre los índicesbursátiles y el tipo de cambio nominal para 6 países de América Latina: Chile, Brasil,México, Colombia, Perú y Argentina. Además, se agrega el volúmen de transaccionesdiarias del mercado bursátil para cada índice, con el fin de establecer efectivamente unarelación de causalidad entre los mercados bursátil y cambiario. La recopilación de datosse realizó con ayuda de la base de datos de Economática y su frecuencia es diaria, paraefectos de estudio se utilizan datos mensuales, entre un rango variable dependiendo de cadapaís, según se señala a continuación: Chile, Brasil y México 2000:01 hasta 2016:11; Perú2007:06 hasta 2015:04; Colombia 2000:01 hasta 2016:11, en dos períodos según el índicebursátil a analizar (IGBC y COLCAP); y Argentina 2002:01 hasta 2016:11. Se utilizó elsoftware Eviews para todos los análisis econométricos realizados.En primer lugar se analiza la estacionariedad de las variables a través de un Test deRaíz Unitaria y se arreglan los datos para cumplir con este requisito. Luego se evalúa laexistencia de causalidad entre las variables mediante un Test de Causalidad de Granger,el que arroja la presencia de causalidad y el sentido de la misma. Para aquellos casosen los que se visualiza causalidad se continúa con el análisis de cointegración de las variables,estudiándose la estacionariedad del error del modelo obtenido. Con lo anteriorse pudo concluir que las variables en estudio cointegraban para solo 2 de los países estudiados:Brasil y Perú. Obteniéndose un modelo de equilibrio de largo plazo para cada uno.Finalmente, para todos los países en estudio se aplica el enfoque Markov SwitchingBayesian VAR, que vincula el concepto de cadenas de Markov con los modelos de econometríaclásica. Al compararse esta última metodología con el enfoque clásico de cointegraciónse puede concluir que para todos los países existe un modelo MS-VAR que mantiene unequilibrio de largo plazo entre las variables, dada la flexibilidad del mismo.
The purpose of this study is determine the relationship between stock prices and exchangerates for 6 countries in Latin America: Chile, Brazil, Mexico, Colombia, Peru andArgentina. In addition, volume of daily transactions of the stock market for each index isadded, in order to establish a causal relationship between stock and exchange markets. Datacollection was done with the help of Economatica and its daily frequency, for study purposes,according to a variable index depending on each country, as indicated below: Chile,Brazil and Mexico 2000:01 to 2016:11; Peru 2007:06 until 2015:04; Colombia 2000:01 to2016:11, in two times according to the stock market analyst index (IGBC and COLCAP);and Argentina 2002:01 to 2016:11. Eviews software was used for all econometric analyzescarried out.First, stationarity time series is analyzed through a Unit Root Test and data is arrangedto comply with this requirement. Then the existence of causality between the variables isevaluated by Granger Causality Test, which shows the presence of causality and the senseof it. For those cases in which causality is visualized, cointegration analysis of the variablesis continued, studying the stationarity of the model error obtained. With the above it waspossible to conclude that the variables under study cointegrated for only 2 of the countriesstudied: Brazil and Peru. Obtaining a long-term equilibrium model for each one.Finally, for all countries in the study, the Markov Switching Bayesian VAR approachis applied, which links the concept of Markov chains with classical econometric models.When comparing this last methodology with the classic cointegration approach, it can beconcluded that all countries have an MS-VAR model that maintains a long-term equilibriumbetween the variables, given its flexibility.
The purpose of this study is determine the relationship between stock prices and exchangerates for 6 countries in Latin America: Chile, Brazil, Mexico, Colombia, Peru andArgentina. In addition, volume of daily transactions of the stock market for each index isadded, in order to establish a causal relationship between stock and exchange markets. Datacollection was done with the help of Economatica and its daily frequency, for study purposes,according to a variable index depending on each country, as indicated below: Chile,Brazil and Mexico 2000:01 to 2016:11; Peru 2007:06 until 2015:04; Colombia 2000:01 to2016:11, in two times according to the stock market analyst index (IGBC and COLCAP);and Argentina 2002:01 to 2016:11. Eviews software was used for all econometric analyzescarried out.First, stationarity time series is analyzed through a Unit Root Test and data is arrangedto comply with this requirement. Then the existence of causality between the variables isevaluated by Granger Causality Test, which shows the presence of causality and the senseof it. For those cases in which causality is visualized, cointegration analysis of the variablesis continued, studying the stationarity of the model error obtained. With the above it waspossible to conclude that the variables under study cointegrated for only 2 of the countriesstudied: Brazil and Peru. Obtaining a long-term equilibrium model for each one.Finally, for all countries in the study, the Markov Switching Bayesian VAR approachis applied, which links the concept of Markov chains with classical econometric models.When comparing this last methodology with the classic cointegration approach, it can beconcluded that all countries have an MS-VAR model that maintains a long-term equilibriumbetween the variables, given its flexibility.
Description
Catalogado desde la version PDF de la tesis.
Keywords
COINTEGRACION , INDICES BURSATILES , MARKOV SWITCHING , TIPOS DE CAMBIO