Performance Management · Analytical techniques in budgeting and forecasting

Learning Curve Theory for ACCA Performance Management

Updated 11 October 2026 · Fact-checked

The learning curve says that each time cumulative output doubles, the cumulative average time per unit falls to a fixed percentage of its previous value. You use Y = aX^b, where a is the first unit's time. Find the cumulative average, multiply by units to get total time, then subtract totals to get incremental time.

Understand Learning Curve Theory

When people repeat a task, they get faster. Workers learn the job, waste less time and make fewer errors. The learning curve puts a number on this effect so you can forecast labour hours and labour cost for later units or batches.

The model works on the cumulative average time per unit. A learning rate of 80% means that when cumulative output doubles, the cumulative average time becomes 80% of what it was. If the average for 1 unit is 100 hours, the average for 2 units is 80 hours, for 4 units 64 hours and for 8 units 51.2 hours.

The curve gives an average, not the time of one unit. To find the time for a particular batch, work out the total time for the larger cumulative output (average × units), then subtract the total time already taken for the earlier units. This is the step where most marks are lost.

Learning does not go on for ever. After a point the steady state is reached, and each extra unit takes the same time. Questions usually tell you when learning stops. Time for units after that point is the steady-state time per unit multiplied by the number of units.

The model suits new, labour-intensive, complex, repetitive work with a stable workforce. It fits badly with automated or machine-paced work, very short production runs, or high labour turnover. Its results are only as good as the learning rate you assume. Management uses the forecasts in budgeting, pricing, target costing and standard setting.

Key rules to remember

Learning curve formula: Y = aX^b; Y = cumulative average time per unit for X units; a = time for the first unit; X = cumulative number of units; b = index of learning.
Index of learning: b = log r ÷ log 2; r is the learning rate as a decimal. It is negative because time falls. Use b = −0.152 for 90%, −0.322 for 80%, −0.234 for 85% and −0.415 for 75%. The formula sheet normally gives the formula; check what your exam provides.
Doubling rule: Cumulative average time after doubling output = previous cumulative average × r; Use this when output doubles exactly (1, 2, 4, 8, 16 and so on). No logs needed.
Total time: Total time for X units = Y × X; Y is the cumulative average time, not the time of one unit.
Incremental time: Time for units (m+1) to n = total time for n units − total time for m units; Always subtract totals, never averages.
Steady state: Total time = time for learning-period units + (extra units × steady-state time per unit); Use only after the point at which learning is stated to stop.

How to solve Learning Curve Theory questions

Use this order for any learning curve question. It keeps you from mixing up averages and totals.

1Identify the learning rate r and the time (or cost) of the first unit or batch, a. Check whether the rate applies to cumulative average time, which is the usual case.
2Decide the cumulative outputs you need. For example, to get the time for units 5 to 8 you need cumulative totals for 4 units and 8 units.
3If each output is a doubling of the last, multiply the cumulative average by r at each doubling. Otherwise calculate b = log r ÷ log 2 and then Y = aX^b.
4Convert each cumulative average into a total time: Y × X.
5Subtract the earlier total from the later total to get the time for the extra units.
6If learning stops, apply the steady-state time to the units produced after that point and add it to the learning-period total.
7Multiply hours by the labour rate per hour to get cost. Add any idle or other adjustments the question asks for.
8Check that the answer is sensible: the average time per unit should fall as output rises, and the incremental time for later units should be lower than the first unit's time.

Quickest way: Doubling table method

When to use it: Use when the required outputs are 1, 2, 4, 8, 16 or similar, which is common in exam questions. It avoids logs and calculator errors.

Write a small table with columns: units, cumulative average, total time.
Row 1: units = 1, average = a, total = a.
For each next row, double the units and multiply the previous average by r. Total = average × units.
Read off the totals you need and subtract to get the time for the batch.
If the output is not a doubling, switch to Y = aX^b. Keep b to at least three decimal places, or use the log form directly in your calculator.

Common mistakes in Learning Curve Theory

Treating the time for the second unit as a × r, or treating the average as the time of the last unit.
The curve is defined on cumulative average time, but students read it as the time of each unit.
Fix: Remember Y is an average. Multiply by X to get a total, then subtract totals to get the time for specific units.
Subtracting cumulative averages instead of cumulative totals to find the time for later units.
It looks quicker and the averages are the numbers you have just calculated.
Fix: Always convert to totals first. For units 5 to 8, use total for 8 units minus total for 4 units.
Using 0.8 as the multiplier for an 80% curve when the question gives a 20% learning effect (or the reverse).
Wording varies between 'learning rate' and 'rate of improvement'.
Fix: A learning rate of 80% means the average becomes 80% of the previous. A 20% reduction also means multiplying by 0.8. Convert the wording to a decimal multiplier before you start.
Applying the curve beyond the steady-state point.
Students apply the formula to every unit without reading the note that learning stops.
Fix: Underline the point where learning ends. Use the curve up to that output only, then use a constant time per unit for the rest.
Rounding b too early, or entering the formula wrongly in the calculator (for example, X^−0.152 as X^0.152).
The negative index is easy to drop, and rounding to two decimal places changes the answer noticeably.
Fix: Keep b to three or more decimals, or use log r ÷ log 2 directly. Check that the average is less than a for X above 1.
Forgetting that the result is an estimate and giving no comment on assumptions in written parts.
Students focus on the numbers.
Fix: When asked to discuss, mention that the rate is assumed, learning may not continue, workforce stability and repetitive work are needed, and the model is poor for automated production.

Worked examples

Example 1

A new product has a first unit time of 100 labour hours. An 80% learning curve applies. Labour costs ₹250 per hour. Calculate the total labour time and cost to make units 5 to 8.

Show the solution

Output doubles from 4 to 8, so use the doubling table.
1 unit: average 100 hours, total 100 hours.
2 units: average 100 × 0.8 = 80 hours, total 160 hours.
4 units: average 80 × 0.8 = 64 hours, total 64 × 4 = 256 hours.
8 units: average 64 × 0.8 = 51.2 hours, total 51.2 × 8 = 409.6 hours.
Time for units 5 to 8 = 409.6 − 256 = 153.6 hours.
Cost = 153.6 × ₹250 = ₹38,400.

Answer: Units 5 to 8 take 153.6 labour hours, costing ₹38,400.

Example 2

A firm makes a batch of 10 units of a product. The first unit takes 20 hours and a 90% learning curve applies to the first 4 units only. After the fourth unit, learning stops and each further unit takes 12 hours. Labour costs ₹300 per hour. Calculate the total labour time and cost for the 10 units.

Show the solution

Learning period: 4 units, which is two doublings.
1 unit: average 20 hours.
2 units: average 20 × 0.9 = 18 hours, total 36 hours.
4 units: average 18 × 0.9 = 16.2 hours, total 16.2 × 4 = 64.8 hours.
Steady-state units: 10 − 4 = 6 units at 12 hours each = 72 hours.
Total time = 64.8 + 72 = 136.8 hours.
Total cost = 136.8 × ₹300 = ₹41,040.

Answer: The 10 units need 136.8 labour hours, costing ₹41,040.

Exam tips

Look for the words 'cumulative average'. If the question gives only the time of a batch or a total, work out what a is before you start.
Set out your doubling table on the workings pad. It earns method marks in constructed response questions and reduces errors in objective questions.
In objective tests, one wrong step makes the whole answer wrong, so check that you have subtracted totals when the question asks for later units only.
In written parts, link the learning curve to its uses (budgeting, pricing, target costing) and to its limits (rate assumed, learning may stop, needs stable, repetitive, labour-intensive work).
Read when learning stops. Questions often combine the curve with a steady-state time or with a labour rate that changes.

Practice questions from Analytical techniques in budgeting and forecasting

Learning Curve Theory in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Learning Curve Theory: frequently asked questions

What does an 80% learning curve mean?

Each time cumulative output doubles, the cumulative average time per unit falls to 80% of the previous average. It does not mean each unit takes 80% of the time of the one before. It is a statement about averages.

How do I find the time for the nth unit only?

Calculate the total time for n units and subtract the total time for n − 1 units. For units that are not powers of 2 you need Y = aX^b. Keep enough decimal places so the difference is accurate.

What is the steady state in learning curve theory?

It is the point after which no further learning occurs, so each unit takes the same time. Use the learning curve up to that point and a constant time per unit after it. Questions normally tell you when this happens.

What are the limitations of the learning curve?

The learning rate has to be assumed, and learning may stop or may not follow a smooth pattern. The model works best for new, manual, repetitive work with a stable workforce. It is unreliable for automated processes, short runs or when staff turnover is high.