Evaluation of Panel Data Models and an Applıcatıon on Ise30

Abstract

Panel data is the intersection of time series and cross-section data. The panel data defines change between units or the change in time for each unit, explains these variabilities by some other variables and estimates each unit in terms of the relevant variables. The purpose of this study is to give information about the models to be done when panel data is obtained and to show this on a specific application. There are commonly known four panel data models, they are the pooled regression model, the most commonly estimated models are probably fixed effects and random effects models and in more complex data sets, a mixed effect model. There are several factors to make choice among panel data models which are pooled regression, fixed effects, random effects, and mixed effects model. In this study, it is aimed to investigate the situation of the ISE30 (Istanbul Stock Exchange-30) in the period 2015-2019 by establishing a panel data regression model consisting of the returns on the stock market shares and the variables affecting these returns. As a result of the study, the model of 37 variables has established and a model when the ones that could not be calculated with the ones missing from the annual 115 financial indicators in the Matriks data program which is a platform about the financial sector, were excluded. Firstly, 5 explanatory variables were determined by stepwise regression from 37 explanatory variables. The relationship between these five explanatory variables and return was investigated by panel data models. The most suitable model was selected by paired comparisons with various tests and the pooled regression model was selected. In all of these models, the rate of explaining the return response variable of within 5 explanatory variables is around 64-74%. As a result; it can be said that stocks, such as the Chande Momentum Oscillator (CMO), Swing Index, Momentum (MOM), Price Earnings Ratio (PER) and Stochastic Fast, affect of the return in the stock exchange as opposed to the popular indicators.