Abstract:
Generalized Ridge regression (GRR) estimators have been identified to be more efficient than
the Ordinary Least Square (OLS) estimator in the presence of Multicollinearity. In this
research work, the concept of different forms on the basis of original, minimum, maximum
and measures of locations of XIX of eigen values of Linear regression model resulted into
some proposed GRR estimators. Three new GRR estimators in their original and different
forms were also proposed. Investigation of the proposed and existing GRR estimators were
done by conducting Monte Carlo experiments 5000 times on two linear regression models (p
= 3 and p = 6) under six (6) levels of multicollinearity (= 0.8, 0.9, 0.95, 0.99, 0.999, 0.9999),
three (3) levels of standard error (σ = 1, 5 and 10) and seven (7) levels of sample sizes (n =
10, 20, 30, 50, 100, 150, 250). The estimators were compared using Mean Square Error
(MSE) criterion. Results showed that the proposed GRR estimators performed better than the
existing GRR estimators in most cases, and that only one of the existing estimators is among
the best seven (7) GRR estimators preferred. The overall best GRR estimator is a newly
proposed one.
