high-low method

Suppose you have a year of electricity bills and you suspect part is a fixed connection charge and part rises with how much you run the machines. You want to split the bill into 'fixed base plus so-much per machine-hour' but you don't want fancy statistics. A quick trick: just look at the busiest month and the quietest month, and see how much the bill changed for the change in hours. That two-point shortcut is the high-low method.

Precisely, the high-low method estimates a mixed cost's variable rate and fixed portion using only the highest-activity and lowest-activity data points. The variable cost per unit equals the change in cost divided by the change in activity between those two points. Then fixed cost equals total cost at either point minus (variable rate times that point's activity). For example, if the high month had 1,000 machine-hours costing 7,000 and the low month had 400 hours costing 4,000, the variable rate is (7,000 minus 4,000) divided by (1,000 minus 400), which is 3,000 over 600, equals 5 per hour. Fixed cost is 7,000 minus 5 times 1,000, equals 2,000. So the cost formula is 2,000 plus 5 per machine-hour.

This matters because it gives a fast, no-math-software way to separate mixed costs so you can budget and run CVP — and it is a staple of introductory courses. But be honest about its weakness: it uses only two points and ignores all the others, and it picks them by highest and lowest activity, not by typical behavior, so a single freak month (a storm, a strike) can badly distort the estimate. It is a rough first cut; for serious work, accountants prefer regression analysis, which uses every data point.