Relationship between ridge regression estimator and sample size when multicollinearity present among regressors

The problem of multicollinearity is the most common problem in multiple regression models as in such cases, the ordinary least squares (OLS) estimator is inaccurately estimated. Of many methods suggested to solve the problem of multicollinearity, ridge regression method is a one of popular method. In this paper, simulation data with different level of correlation coefficient were generated using Monte Carlo techniques in SAS. The level of multicollinearity was detected by correlation matrix, variance influence factor (VIF) and condition number. The biased parameter (k) of ridge regression has been computed by using iterative method for ordinary ridge regression in different sample sizes. According to the results of this study, it was found that biased parameter (k) of ridge regression and sample sizes are significantly negatively correlated at level of 5% significance. This study would helpful to develop biased parameter table for different level of sample sizes in present of multicollinearity.