How would you interpret a regression coefficient that has a high R-squared value but a statistically insignificant t-statistic?