The presence of severe Multicollinearity among regressor variables leads to:

When the regressors are highly correlated, it becomes difficult for OLS to separate the individual effect of each one. For the coefficient of regressor X_j, Var(beta-hat_j) = σ² / [TSS_j × (1 - R_j²)] where R_j² is the R² of the regression of X_j on the other regressors. As R_j² approaches 1, the variance, and hence the standard error, becomes very large. The variance inflation factor is VIF = 1/(1 - R_j²). Large standard errors mean small t-statistics, so the individual coefficients appear statistically insignificant. The typical symptom is a high overall R² and a significant F-statistic, combined with insignificant individual t-ratios. Option (A) is wrong, as the estimates remain unbiased. Option (C) is wrong, since TSS depends only on Y and cannot be negative. Option (D) is wrong, because so long as the collinearity is not perfect, the Gauss-Markov assumptions are still satisfied. Option (E) is wrong, as the overall R² is not reduced and often remains high. Hence the correct option is (B).