All frameworks

Finance Explore

Activity-Based Costing

Activity-based costing assigns overhead through the activities a product actually consumes, so reported profit sits closer to real profit.

The questionWhat does each product, customer, or channel cost once overhead follows the work that causes it?

ABC. The time-driven variant is TDABC.

Formula · Robert S. Kaplan and Robin Cooper, 1980s

  1. 1
    Resources Payroll, equipment, space
  2. 2
    Activities Setups, order handling
  3. 3
    Cost drivers Setups, orders, inspections
  4. 4
    Cost objects Products, customers, channels
Simplified chain showing how overhead cost reaches products, customers, or channels.

When to use it

  • Overhead has grown relative to direct cost, and one blended rate makes simple, high-volume products carry the cost of complex, low-volume ones.
  • A pricing, product-line, or customer-profitability decision turns on margins that differ by a few points.
  • A service business wants cost per transaction, such as per claim, loan, or order, and not just cost per department.
  • Customer segments make different demands, for special orders, returns, or rush deliveries, but pay the same prices.

When to avoid it

  • Overhead is small relative to direct cost. The allocation method will not change any decision, so simple costing is enough.
  • The decision is one-off and the activity data is not captured anywhere. Building the model costs more than the decision is worth, and contribution margin with a check on which costs would really go away is quicker.
  • The question is where a process wastes time, and not what a product costs. A process map shows that directly.
  • Someone expects to run it by hand every quarter indefinitely. Heavy ABC systems have a record of being built and then abandoned.

Overview

Activity-based costing assigns overhead to products, customers, or channels according to the activities they trigger, rather than as a flat percentage of direct cost. It supports pricing, product-line, and customer-profitability choices where one blended overhead rate hides that a complex or low-volume item uses far more setups, orders, or inspections than a standard one.

The result is an allocation, not a forecast of savings. Overhead is mostly fixed in the short run, so a product that looks unprofitable under ABC leaves its overhead behind if you drop it. Use the figures to reprice, redesign, or batch, and to decide which capacity to remove.

What it is not

  • A cost-cutting method. It reports where cost sits. Cost falls only when activity volume or capacity falls, and dropping a product that looks unprofitable leaves its overhead behind.
  • An exact answer. The activities, drivers, and time estimates are all choices, and the rates move with them.

The steps

  1. Define the cost objects and the decision

    Decide what you are costing, such as products, customers, or channels, and what decision the answer feeds. This sets how fine the activity list needs to be. A pricing review needs less detail than a plant redesign.

  2. List activities and assign resource costs

    Walk the process and list the activities that consume overhead, such as setups, order handling, and inspection. Assign payroll, equipment, and space costs to those activities using time or usage estimates. Each activity now has a cost pool.

  3. Choose a driver for each pool

    Pick the measure that best tracks how much of the activity is used, and check that someone already records it. A driver that is theoretically right but not recorded will turn into estimates, and the model will rest on them.

  4. Compute the activity rates

    Divide each pool by the total volume of its driver. Reconcile the sum of all pools to the ledger, so no overhead is lost or counted twice.

  5. Assign cost and compare with the old view

    Charge each product or customer the rate times the driver units it uses, and add direct cost. Put the new cost and margin next to the old ones. Investigate the large differences, and keep business-sustaining costs that have no honest driver in a separate line.

  6. Decide who maintains it

    Name an owner and a refresh date for the rates, or state plainly that this is a one-off study and the numbers will age. Rates drift as volumes and processes change.

The formula

Product cost = direct cost + sum of (activity rate × driver units used). Activity rate = cost of the activity pool ÷ total driver volume.

Each pool of overhead is divided by the total amount of its driver to give a rate, and each product is charged that rate for every unit of the driver it uses.

  • Activity cost pool: the resource cost assigned to one activity, such as machine setups or order handling.
  • Cost driver: the measure of how much of the activity is used, such as number of setups, orders, or inspections.
  • Activity rate: pool cost divided by total driver volume, for example dollars per setup.
  • Driver units used: how many units of a driver a product or customer consumes.

In practice

Two products each carry 50 of direct cost per unit, sell for 80, and run at 10,000 units a year. The plant's costing system spreads 400,000 of overhead in proportion to direct cost, so each takes 200,000, or 20 a unit. Both cost 70 and both show a margin of 10.

The plant has two overhead pools. Setups cost 240,000 across 80 setups, which is 3,000 per setup. Inspection costs 160,000 across 200 inspections, which is 800 per inspection. Product A runs long and uses 10 setups and 40 inspections, so it carries 30,000 + 32,000 = 62,000, or 6.20 a unit. Product B is custom and uses 70 setups and 160 inspections, so it carries 210,000 + 128,000 = 338,000, or 33.80 a unit. A costs 56.20 and earns 23.80 a unit. B costs 83.80 and loses 3.80 a unit, or 38,000 a year. The decision is not yet to drop B. Only the share of B's overhead that goes away when its setups and inspections go away will leave the income statement, so the team first asks whether B can be repriced, batched into longer runs, or redesigned to need fewer inspections.

What you leave with

  • A list of activities with their cost pools and chosen drivers.
  • Activity rates that reconcile to the ledger.
  • Product, customer, or channel cost and margin under the old method and the activity-based one, side by side.
  • A decision list covering repricing, redesign, batching, changed terms, or no change, with the costs that would actually go away.
  • An owner and a refresh date for the rates, or an explicit note that the study is a one-off.

Failure points

A study that decays

It is built as a project, nobody refreshes it, and the rates go stale as volumes and processes change. Two years later the numbers disagree with the shop floor and nobody uses them.

Survey-based time estimates

When people split their working time across activities, the shares sum to 100 percent whether or not they were busy. Idle capacity gets loaded onto products, and the rates come out too high.

Too many activities

Every activity needs a pool, a driver, and data. A model with hundreds of them is expensive to build and hard to explain. Keep the list to the detail the decision needs.

Allocated cost treated as avoidable cost

A product looks unprofitable under ABC, so it is dropped. The fixed overhead stays and is spread over the remaining products, which now look worse. Check which costs leave before anyone acts on the margin.

Drivers that nobody records

A driver is chosen for being logical, but no system counts it. The team fills the gap with estimates, and the number looks measured when it is a guess.

The failures above are usually Commercial Acumen: the work was not tied to money, for the client and/or for you.

Origin

Robert S. Kaplan and Robin Cooper formalized activity-based costing in the 1980s, building on cost-accounting work already under way in US manufacturers, including studies run through the Consortium for Advanced Manufacturing-International. Their September 1988 Harvard Business Review article, "Measure Costs Right: Make the Right Decisions," made the case to managers. Their argument was that as overhead grew relative to direct labor and materials, spreading it as a percentage of direct cost distorted product costs.

Early implementations relied on interviews and surveys to estimate how people spent their time, and many were expensive to build and keep current. Kaplan and Steven Anderson proposed time-driven ABC in Harvard Business Review in November 2004, which replaces the surveys with managers' estimates of the time each activity takes.

Further reading

  • Measure Costs Right: Make the Right Decisions

    Robin Cooper and Robert S. Kaplan, Harvard Business Review, September 1988. The article that introduced activity-based costing to managers. Metered paywall.

  • Time-Driven Activity-Based Costing

    Robert S. Kaplan and Steven R. Anderson, Harvard Business Review, November 2004. The simpler variant, and a frank account of why many first-generation ABC systems were abandoned. Metered paywall.

  • Activity-based costing

    Wikipedia's overview, including the limitations and the history of the method. Free.

Last updated