Today, when reading about multicollinearity – one of the most dreaded problems in regression analysis, I came across the statement of Blanchard [8]: “Multicollinearity is God’s will, not a problem with OLS or statistical techniques in general.” Immediately, I thought: “If it is really God’s will, we should keep the problem within our grasp, but not rely on God.”.

Findings – As the results, by using 5 steps which are: test for the statistical significance, test for heteroscedasticity, test for multicollinearity test the assumption of normality and test for autocorrelation, a positive relationship is found between Vietnam gold price and USD/VND exchange rate. In detail, the inflation in Vietnam and the nominal interest rate in Vietnam have no statistically significant with the gold price. In statistical significance test, the coefficient for CPI and USD/VND exchange rate are statistical significant (...) but there exists the multicollinearity between CPI and USD/VND exchange rate. To ensure the result, CPI is chosen to remove when comparing R-squared between two regression equations. (shrink)