The high-low estimate of fixed costs is lower than the estimate provided by least-squares regression. The high-low estimate of the variable cost per machine hour is vice-versa. A straight line that minimized the sum of the squared errors would intersect the Y-axis at $10,090 (fixed cost-least-squares regression) instead of $4,000 (fixed cost-high-low method). Inasmuch as the estimated variable cost per unit is lower than the high-low method, it would also have a flatter slope.