High-Low Method — Estimate cost functions using the high-low method (intro) - Cost Accounting

SECTION 1

Historical Context & Motivation

Every business faces the same fundamental question when planning and budgeting: how will total costs change as output changes? Answering that question requires separating costs into their fixed and variable components—a task that has occupied managers and accountants since the dawn of industrial cost tracking. The high-low method is one of the oldest and most intuitive approaches to this decomposition, relying on nothing more than two data points and basic algebra.

Before spreadsheets and regression software became ubiquitous, managers needed a rapid way to estimate cost behavior from limited accounting records. The high-low method filled that gap by using the periods of highest and lowest activity to anchor a straight line through the data. Although more sophisticated statistical tools now exist, the high-low method remains a staple of cost accounting curricula and a useful first-pass diagnostic in practice.

1880s–1920s • Rise of Industrial Costing

Rapid industrialization created complex factories where managers first grappled with distinguishing costs that stayed constant (rent, supervisory salaries) from those that rose with output (raw materials, direct labor).

1930s–1940s • Formalization of Cost Behavior

Cost accounting textbooks began systematically categorizing costs as fixed, variable, or mixed. Simple estimation techniques like the high-low method and the scattergraph method entered standard pedagogy.

1950s–1970s • Flexible Budgeting & CVP Analysis

Cost-volume-profit analysis became central to managerial decision-making. Accurate cost function estimation—often starting with the high-low method—was essential for break-even and target-profit calculations.

1980s–Present • Regression & Data Analytics Era

Least-squares regression replaced the high-low method as the gold standard for cost estimation, but the high-low method endures as a quick benchmark and an accessible introduction to cost function concepts.

The central question the high-low method addresses is deceptively simple: given a set of observed total costs at various activity levels, how can we quickly estimate the variable cost per unit and the total fixed cost? Understanding why two data points suffice—and where this simplification breaks down—is the gateway to deeper cost estimation techniques such as regression analysis.

SECTION 2

Core Principles & Definitions

Before applying the high-low method, you need a solid grasp of how costs behave relative to an activity driver (also called a cost driver). An activity driver is the factor—such as machine hours, units produced, or miles driven—that causes total costs to change. The high-low method assumes that total cost is a linear function of a single activity driver within the relevant range, the band of activity over which the linear relationship is expected to hold.

Variable Costs

Costs that change in direct proportion to the activity level. Total variable cost rises as output increases, but the per-unit variable cost remains constant within the relevant range. Examples include direct materials and sales commissions.

Fixed Costs

Costs that remain constant in total regardless of activity level changes within the relevant range. The per-unit fixed cost declines as volume rises. Examples include rent, insurance, and salaried management.

Mixed (Semi-Variable) Costs

Costs containing both a fixed and a variable component. A utility bill with a base charge plus a per-kilowatt-hour rate is a classic example. The high-low method's primary purpose is to decompose mixed costs into their fixed and variable parts.

Cost Function (y = a + bx)

A mathematical representation where y = total cost, a = total fixed costs, b = variable cost per unit, and x = activity level. The high-low method estimates a and b.

✦

KEY TAKEAWAY

Think of a mixed cost like a taxi fare: there is a flat drop charge the moment you sit down (the fixed component) and a per-mile rate that accrues as you ride (the variable component). If you only knew two taxi receipts—one for a short ride and one for a long ride—you could back into both the drop charge and the per-mile rate by comparing the difference in fare to the difference in miles. That is exactly what the high-low method does with cost data.

SECTION 3

Visual Explanation — Plotting Cost Behavior

The scattergraph below illustrates why the high-low method works. Each dot represents a month of observed data—activity level on the horizontal axis and total cost on the vertical axis. The method identifies the highest activity point and the lowest activity point, draws a straight line through them, and uses that line to define the cost function. The y-intercept of the line represents estimated fixed costs, and its slope represents the variable cost per unit of activity.

The low point and high point are selected based on the extreme activity levels (x-axis), not the extreme costs. The dashed amber line connecting them defines the estimated cost function, whose y-intercept is the estimated fixed cost and whose slope is the estimated variable cost per unit.

⚠

Important Distinction

Always choose the high and low points based on the activity level (x-axis), not the total cost (y-axis). A month might have the highest total cost due to an unusual event (e.g., a major repair), yet not represent the highest activity. Using cost extremes instead of activity extremes can produce a misleading cost function.

SECTION 4

Mathematical Framework

The high-low method exploits the familiar slope-intercept form of a linear equation. Because we assume that total cost behaves linearly within the relevant range, estimating the cost function reduces to finding the slope and y-intercept of a line through two points. The procedure consists of two algebraic steps: first compute the variable cost rate (slope), then solve for the fixed cost (intercept).

COST FUNCTION (LINEAR MODEL)

y = a + bx

y = total mixed cost | a = total fixed costs (y-intercept) | b = variable cost per unit of activity (slope) | x = activity level (cost driver)

STEP 1 — VARIABLE COST PER UNIT (SLOPE)

b = (y_high − y_low) ÷ (x_high − x_low)

The change in total cost divided by the change in activity level. This is the classic rise over run slope formula applied to cost data.

STEP 2 — TOTAL FIXED COST (Y-INTERCEPT)

a = y_high − b × x_high (or equivalently: a = y_low − b × x_low)

Substitute b back into the cost equation using either the high or the low data point. Both should yield the same fixed cost, which serves as a useful arithmetic check.

Once a and b are estimated, the resulting equation can be used to predict total cost at any activity level within the relevant range. Extrapolation outside that range is unreliable because the linear assumption may not hold at very low or very high volumes where step costs, economies of scale, or capacity constraints alter cost behavior.

SECTION 5

Step-by-Step Procedure

The diagram below codifies the high-low method into a structured workflow. Following these steps in order ensures a consistent and error-free application, whether you are working with manufacturing costs, utility bills, or maintenance expenses.

The five-step flowchart for the high-low method. Steps 3 and 4 are the computational core. The verification box on the right reminds you to confirm the fixed cost using both data points as an arithmetic check.

💡

Outlier Alert

Before applying the method, scan the data for outliers—data points driven by unusual events such as a natural disaster, equipment failure, or strike. If the highest or lowest activity period is an outlier, the resulting cost function may be badly distorted. Some textbooks recommend excluding obvious outliers before selecting the high and low points.

SECTION 6

Worked Example

Greenfield Manufacturing wants to estimate its monthly maintenance cost function. The company believes machine hours is the primary cost driver. The following data were collected over six months.

Month	Machine Hours (x)	Total Maintenance Cost (y)
January	1,500	$36,000
February	2,500	$48,000
March	3,200	$55,200
April	2,800	$50,400
May	1,200	$32,400
June	3,800	$61,600

Greenfield Manufacturing — Monthly Maintenance Data

Estimating the Maintenance Cost Function

1
Step 1 — Identify the High and Low Activity Points

Scan the Machine Hours column (not the cost column). The highest activity is June: 3,800 hours, $61,600. The lowest activity is May: 1,200 hours, $32,400.

High: (3,800, $61,600) | Low: (1,200, $32,400)
2
Step 2 — Calculate the Variable Cost per Machine Hour (b)

Apply the slope formula: b = (y_high − y_low) ÷ (x_high − x_low) = ($61,600 − $32,400) ÷ (3,800 − 1,200) = $29,200 ÷ 2,600 = $11.23 per machine hour (rounded to the nearest cent).

b ≈ $11.23 per machine hour
3
Step 3 — Calculate the Fixed Cost (a)

Substitute b into the cost equation using the high point: a = $61,600 − ($11.23 × 3,800) = $61,600 − $42,674 = $18,926. As a check, use the low point: a = $32,400 − ($11.23 × 1,200) = $32,400 − $13,476 = $18,924. The $2 difference is due to rounding; both confirm a fixed cost of approximately $18,925.

a ≈ $18,925
4
Step 4 — Write the Cost Equation

The estimated monthly maintenance cost function is: y = $18,925 + $11.23x, where x = machine hours. For example, if Greenfield expects 2,000 machine hours next month, estimated maintenance cost = $18,925 + ($11.23 × 2,000) = $18,925 + $22,460 = $41,385.

y = $18,925 + $11.23x

SECTION 7

Strengths & Limitations

Like any estimation technique, the high-low method involves trade-offs. Its simplicity is both its greatest advantage and its most significant weakness. The table below summarizes the key strengths and limitations that managers and cost accountants should weigh when choosing an estimation approach.

Strengths	Limitations
Requires only two data points—fast to compute with a calculator or even mental math.	Ignores all data points between the high and the low, potentially discarding valuable information.
Easy to understand and explain to non-accountants such as plant managers or executives.	Highly sensitive to outliers; if the high or low point is anomalous, the entire estimate is distorted.
Provides a quick sanity check or first approximation before running more complex analyses.	Assumes a perfectly linear relationship—no accommodation for step costs, curvilinear behavior, or multiple cost drivers.
Useful when only limited data are available (e.g., a new product line with few months of history).	Produces no measure of goodness of fit (e.g., R²), making it impossible to assess how well the line explains the data.

High-Low Method — Strengths vs. Limitations

✦

WHEN TO USE (AND WHEN NOT TO)

Use the high-low method when you need a quick, rough estimate—for instance, in a budget meeting where a back-of-the-envelope figure is needed before a formal analysis can be run. Think of it like using a ruler to estimate the slope of a hiking trail from a map: it gives you a reasonable ballpark, but a GPS survey (analogous to regression analysis) would be far more precise. Whenever you have enough data and access to software, least-squares regression is the preferred technique because it uses all observations and provides statistical diagnostics.

SECTION 8

Connection to Advanced Cost Estimation

The high-low method is best understood as the simplest member of a family of cost estimation techniques, all of which attempt to fit a linear model to observed data. The table below contrasts the high-low method with two more rigorous alternatives: the scattergraph (visual-fit) method and least-squares regression. Understanding these differences prepares you for more advanced treatment of cost functions in intermediate and graduate-level courses.

Feature	High-Low Method	Scattergraph Method	Least-Squares Regression
Data Points Used	Only 2 (highest and lowest activity)	All points (line drawn visually)	All points (mathematically optimized)
Objectivity	Objective—same data always yields same answer	Subjective—depends on the analyst's eye	Objective—minimizes sum of squared errors
Goodness-of-Fit Metric	None	None (visual inspection)	R², standard error, p-values
Software Required	None—calculator sufficient	Graph paper or basic charting tool	Spreadsheet or statistical software
Best For	Quick estimates, limited data, exam settings	Detecting non-linearity and outliers	Formal cost function estimation in practice

Comparison of Cost Estimation Methods

In practice, a thoughtful analyst often combines these methods. A scattergraph reveals the overall pattern and flags any obvious outliers; the high-low method provides a quick benchmark; and regression delivers the statistically rigorous estimate used for decision-making. As you progress in your cost accounting studies, you will learn to evaluate regression output—coefficients of determination, confidence intervals, and residual plots—to assess whether a linear model adequately captures cost behavior or whether a more complex specification is warranted.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

When applying the high-low method, a student selects the two months with the highest and lowest total costs rather than the highest and lowest activity levels. Explain why this approach is incorrect and describe a scenario in which it would produce a materially different cost function than the correct approach.

Show answer

The high-low method estimates the slope of the cost function by measuring how cost changes per unit change in the activity driver. To isolate that rate, we need the maximum range of the independent variable (activity). If we select months based on extreme costs instead, we may pick a month whose cost was inflated by an unusual, non-recurring event (e.g., an emergency equipment overhaul) rather than by high production volume. For instance, suppose March had 2,000 machine hours but $70,000 in costs due to a one-time repair, while June had 3,500 machine hours and only $60,000 in costs. Choosing March as the 'high' point would overstate the slope and understate fixed costs, producing a cost function that systematically mispredicts at normal activity levels.

Question 2: BASIC CALCULATION

Baxter Company reports the following data for delivery costs: Month with highest deliveries: 4,000 deliveries, total cost $52,000. Month with fewest deliveries: 1,500 deliveries, total cost $31,250. Using the high-low method, estimate the variable cost per delivery and the total fixed delivery cost per month.

Show answer

Variable cost per delivery (b) = ($52,000 − $31,250) ÷ (4,000 − 1,500) = $20,750 ÷ 2,500 = $8.30 per delivery. Fixed cost (a) using the high point = $52,000 − ($8.30 × 4,000) = $52,000 − $33,200 = $18,800. Verification with the low point: $31,250 − ($8.30 × 1,500) = $31,250 − $12,450 = $18,800. Both yield the same result, confirming the cost function: y = $18,800 + $8.30x.

Question 3: INTERMEDIATE

Metro Electric recorded the following utility cost data over eight months: Jan: 10,000 kWh, $4,500 | Feb: 12,000 kWh, $5,100 | Mar: 15,000 kWh, $5,900 | Apr: 18,000 kWh, $6,700 | May: 20,000 kWh, $7,200 | Jun: 22,000 kWh, $7,900 | Jul: 25,000 kWh, $8,500 | Aug: 8,000 kWh, $4,100. Use the high-low method to derive the cost function, and then predict the utility cost for a month with 16,500 kWh of usage.

Show answer

The highest activity month is July (25,000 kWh, $8,500) and the lowest is August (8,000 kWh, $4,100). Variable cost per kWh (b) = ($8,500 − $4,100) ÷ (25,000 − 8,000) = $4,400 ÷ 17,000 ≈ $0.2588 per kWh. Fixed cost (a) = $8,500 − ($0.2588 × 25,000) = $8,500 − $6,470 = $2,030. The cost equation is y = $2,030 + $0.2588x. For 16,500 kWh: y = $2,030 + ($0.2588 × 16,500) = $2,030 + $4,270.20 = $6,300.20, or approximately $6,300.

Question 4: APPLIED

Pacific Logistics operates a fleet of delivery trucks. The fleet manager suspects that fuel and maintenance costs are mixed costs driven by miles driven. Over the past year, the highest-mileage month was October (120,000 miles, $84,000 total fleet cost) and the lowest was February (40,000 miles, $44,000). (a) Estimate the cost function using the high-low method. (b) If Pacific is evaluating a new contract that would add 15,000 miles per month, estimate the incremental monthly fleet cost. (c) Explain one managerial decision this information could support.

Show answer

(a) b = ($84,000 − $44,000) ÷ (120,000 − 40,000) = $40,000 ÷ 80,000 = $0.50 per mile. a = $84,000 − ($0.50 × 120,000) = $84,000 − $60,000 = $24,000. Cost function: y = $24,000 + $0.50x. (b) The incremental cost of 15,000 additional miles is $0.50 × 15,000 = $7,500 per month, because fixed costs do not change within the relevant range. (c) This estimate helps Pacific set the minimum price for the new contract. The contract should be priced to cover at least $7,500 in incremental variable costs; any price above that contributes to covering fixed costs and generating profit. The information also supports fleet budgeting and make-or-buy decisions regarding outsourcing deliveries.

Question 5: CRITICAL THINKING

Suppose a company's six-month data set contains an outlier in the lowest-activity month—a plant shutdown for retooling that compressed production into fewer days, inflating per-unit costs. A colleague argues that the high-low method should still be applied because 'the math is the math.' Critically evaluate this argument. What alternative approaches could the analyst take within the high-low framework, and how might the analyst determine whether the resulting cost function is reasonable?

Show answer

The colleague's argument is flawed because the high-low method's validity depends on the two selected points being representative of normal operating conditions. An outlier driven by abnormal circumstances (a shutdown that concentrated fixed costs into a smaller output base) violates this assumption. The resulting slope would overstate the variable cost rate, and the intercept would understate fixed costs. Within the high-low framework, the analyst could exclude the outlier month and instead use the next-lowest normal-activity month as the low point. Alternatively, the analyst could plot all data on a scattergraph to visually confirm or reject the outlier before computing. To test reasonableness, the analyst could plug each of the remaining months' activity levels into the estimated equation and compare predicted costs to actual costs; large and systematic deviations would signal a poor fit. Ultimately, if enough data exist, regression analysis is superior because it uses all observations, minimizes total prediction error, and provides an R² statistic and residual analysis to formally assess fit.

SUMMARY

Lesson Summary

Summary

The high-low method is a straightforward technique for estimating a linear cost function of the form y = a + bx. It works by selecting the periods with the highest and lowest activity levels (not costs), computing the variable cost per unit (b) as the change in cost divided by the change in activity, and then solving for the fixed cost (a) by substituting back into the cost equation.

While valued for its speed and simplicity, the method carries important limitations: it uses only two data points, ignoring the rest of the data set, and is vulnerable to outliers. It provides no statistical measure of fit, unlike least-squares regression, which uses all observations and produces diagnostics such as R². Treat the high-low method as a first-pass estimate and a pedagogical stepping stone toward more rigorous cost estimation tools.