An application of VAR and Almon Polynomial Distributed Lag models to insurance stocks: Evidence from KSE

In the time series data, a regressand may respond to regressors with a time lag. This study employs dynamic methodology of Almon Polynomial Distributed-Lag (PDL) model as an application to the stocks of 13 selected insurance companies, using daily data for the period from 1996 to 2008. Realizing the importance of causality in economics and finance, this study focuses on the causal relationship between investment, growth in returns and market uncertainty. The study also employs VAR which is of non-structural approaches amongst the a-theoretic models. In this study I have constrained the coefficients on the distributed lag to lie on a third degree polynomial with the satisfactory test results of near and the far end points of the lag distribution. Generating the series of risk variable through GARCH (p, q) is also an academic contribution of this study. The Almon PDL model may also be considered as an alternative to the lagged regression models. For the PDL avoids the estimation problems associated with the autoregressive models. This study is in a way an attempt to invite researchers and practitioners for the maximum application of these very important dynamic models in economics, business and finance. The results of this study reveal mixed causality among the three variables. The Almon Polynomial Distributed Lag results support the theory of adaptive expectations.