Strategic Cost Management · Learning Curve

Learning Curve Mathematical Model Y = aX^b Explained

Updated 11 October 2026 · Fact-checked

The cumulative average time model says the average time per unit falls by a fixed percentage each time cumulative output doubles. It is written Y = aX^b, where a is the first unit's time, X is cumulative units and b = log r ÷ log 2. Find Y, multiply by X for total time.

Understand Learning Curve Mathematical Model (Y = aX^b)

When a worker or team repeats a task, they get faster. The first unit takes longest. Later units take less time as people learn the method. The learning curve puts a number on this effect so you can estimate labour hours and cost.

In the cumulative average time model, the rule is about the average. Each time cumulative output doubles, the cumulative average time per unit falls to a fixed percentage of its previous value. That percentage is the learning rate (r). An 80% curve means the average for 2 units is 80% of the first unit's time. The average for 4 units is 80% of the 2-unit average, and so on.

The doubling rule can be written as a formula: Y = aX^b. Here Y is the cumulative average time per unit for X units, a is the time for the first unit, and X is the cumulative number of units. The exponent b is the learning index. It is negative, because average time falls as X rises.

The index comes from the rate: b = log r ÷ log 2. For r = 0.8, b = log 0.8 ÷ log 2 = about -0.3219. You can use logs of any base, as long as you use the same base on top and bottom.

Total time for X units = Y × X. Time for a later batch = total time for the larger cumulative quantity minus total time already spent. Always work on cumulative figures first, then take differences.

Key rules to remember

Learning curve model: Y = aX^b; Y = cumulative average time (or cost) per unit for X units; a = time for first unit; X = cumulative units.
Learning index: b = log r ÷ log 2; r is the learning rate as a decimal (80% = 0.8). b is negative for r below 1.
Learning rate from index: r = 2^b; Use to get back the rate once b is known.
Learning rate from data: b = log(Y ÷ a) ÷ log X, then r = 2^b; Use when the first-unit time and the cumulative average for X units are given.
Total time: Total time for X units = Y × X; Time for units m+1 to n = total for n units minus total for m units.
Doubling shortcut: Average for 2X units = r × average for X units; Valid only when quantity doubles exactly, such as 1, 2, 4, 8, 16.

How to solve Learning Curve Mathematical Model (Y = aX^b) questions

Use this order for any cumulative average time question. It keeps cumulative and incremental figures apart.

1Identify the model. Check that the question says cumulative average time per unit falls with doubling output.
2Write down a (first unit time), the learning rate r, and the quantity X asked for.
3Check whether X is a doubling of the first unit (2, 4, 8, 16). If yes, apply r repeatedly and skip logs.
4If not, compute b = log r ÷ log 2. Keep at least four decimals.
5Compute Y = a × X^b using logs: log Y = log a + b × log X, then take the antilog.
6Multiply Y by X to get the total time for X units.
7For a later batch, compute the cumulative total at both quantities and subtract.
8Convert hours to cost using the labour rate, and state the answer with units.

Quickest way: Doubling table first, logs only if needed

When to use it: Use when the quantities are 2, 4, 8, 16 units or when you are asked for a batch between doublings.

Write a row: units 1, 2, 4, 8 with average time = a, a×r, a×r², a×r³.
Multiply each average by its units to get cumulative total time.
Subtract totals for the batch asked.
Use logs only for quantities such as 3, 5, 6 or 10 units.

Common mistakes in Learning Curve Mathematical Model (Y = aX^b)

Treating the learning rate as the fall in time, so using 0.2 for an 80% curve.
The words 'rate' and 'reduction' get mixed up.
Fix: An 80% curve means average time becomes 80% of before. Use r = 0.8 in b = log 0.8 ÷ log 2.
Reading Y as the time for the Xth unit.
Students forget this model is about the cumulative average.
Fix: Y is the average for all X units. Multiply by X for total time, and subtract totals to get a particular unit or batch.
Dropping the negative sign of b.
Logs of numbers below 1 are negative, and the sign gets lost.
Fix: For any r below 1, b is negative. Sense-check: Y must be below a for X above 1.
Using the doubling shortcut for quantities that are not doublings.
The shortcut is quick and easy to overuse.
Fix: Use r only for 1, 2, 4, 8... Use Y = aX^b for other quantities.
Subtracting averages instead of totals when finding time for a later batch.
Averages look like usable per-unit figures.
Fix: Convert to total time at each cumulative level first. Batch time = larger total minus smaller total.
Mixing log bases or rounding b too early.
Calculators have both log and ln, and small rounding errors grow when raised to a power.
Fix: Use one base throughout and keep b to four decimals or more.

Worked examples

Example 1

A firm makes a new machine part. The first unit takes 100 hours. The cumulative average time follows an 80% learning curve. Find (a) the average time per unit and total time for 8 units, and (b) the time for units 5 to 8.

Show the solution

8 units is three doublings from 1 unit (1, 2, 4, 8).
Average for 2 units = 100 × 0.8 = 80 hours.
Average for 4 units = 80 × 0.8 = 64 hours.
Average for 8 units = 64 × 0.8 = 51.2 hours.
Total for 8 units = 51.2 × 8 = 409.6 hours.
Total for 4 units = 64 × 4 = 256 hours.
Time for units 5 to 8 = 409.6 - 256 = 153.6 hours.

Answer: Average for 8 units = 51.2 hours; total for 8 units = 409.6 hours; units 5 to 8 take 153.6 hours.

Example 2

The first unit of a product takes 20 hours. A 90% cumulative average learning curve applies. Labour costs ₹250 per hour. Estimate the cumulative average time, total time and total labour cost for 6 units. (log 0.9 = -0.045757; log 2 = 0.301030; log 6 = 0.778151.)

Show the solution

b = log 0.9 ÷ log 2 = -0.045757 ÷ 0.301030 = -0.1520.
Y = 20 × 6^-0.1520.
log 6^-0.1520 = -0.1520 × 0.778151 = -0.11828.
6^-0.1520 = antilog of -0.11828 = about 0.7616.
Y = 20 × 0.7616 = about 15.23 hours per unit.
Total time = 15.23 × 6 = about 91.4 hours.
Labour cost = 91.4 × ₹250 = ₹22,850.

Answer: Cumulative average time is about 15.23 hours; total time is about 91.4 hours; total labour cost is about ₹22,850.

Exam tips

Show b = log r ÷ log 2 with its value. Marks are often given for the index even if the later arithmetic slips.
Check the question for the model. This topic is the cumulative average model, so Y is an average, not the Xth unit's time.
Use the doubling table when quantities are powers of 2. It is faster and you avoid log errors.
Write the cumulative total time at each stage before subtracting for a batch or for the incremental cost.
In case-based MCQs, estimate first: Y must lie between a × r^(number of doublings rounded up) and a. Use this to reject wrong options.

Practice questions from Learning Curve

Learning Curve Mathematical Model (Y = aX^b) 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 Mathematical Model (Y = aX^b): frequently asked questions

What does b mean in Y = aX^b?

b is the learning index. It equals log r ÷ log 2, where r is the learning rate as a decimal. It is negative for any rate below 100%, because average time falls as cumulative output rises.

How do I find the learning rate from given data?

Compute b = log(Y ÷ a) ÷ log X using the first-unit time a and the cumulative average Y for X units. Then r = 2^b. For example, if 100 hours for the first unit gives a 4-unit average of 64 hours, then 64 ÷ 100 = r², so r = 0.8, an 80% curve.

Is Y the time for the last unit or the average?

In the cumulative average time model, Y is the average time per unit over all X units produced so far. Multiply by X for total time. To find the time for the Xth unit alone, subtract the total for X-1 units from the total for X units.

Can I use ln instead of log base 10?

Yes. The ratio log r ÷ log 2 gives the same b in any base, as long as you use the same base for both. Keep four decimals to avoid rounding error.