2.3.2.1 Learning curve introduction

Achievable CMA Part 1

2. Planning, budgeting, and forecasting

2.3. Forecasting techniques

2.3.2. Learning curve analysis

Our CMA Part 1 course is currently in development and is a work-in-progress.

Learning curve introduction

6 min read

Font

Discuss

Feedback

Learning outcome statements

The learning outcome statements relevant for this section are:

demonstrate an understanding of learning curve analysis
calculate the results under a cumulative average-time learning model
identify the benefits and shortcomings of learning curve analysis

Definitions

Learning curve analysis: This is a forecasting technique used to predict how labor efficiency improves as workers gain experience in producing goods or delivering services. The concept is based on the idea that as employees repeat a task, they become more skilled, which leads to faster production times and lower costs. The more frequently a task is performed, the greater the efficiency gains.
Learning curve: A graphical representation that shows the rate at which labor costs or production times decrease as the cumulative output increases.
Cumulative average-time learning model: A model that calculates the average time or cost per unit, considering all previous units produced.
Improvement rate: The percentage by which time or cost decreases as the number of units produced doubles.

In industries where repetitive processes are common, such as manufacturing or assembly-line production, learning curve analysis can be a powerful tool to estimate future cost reductions and productivity improvements. The learning curve itself is typically expressed as a percentage, showing the rate at which labor hours or costs decrease as production doubles.

Understanding learning curve analysis

Example: Cumulative average-time learning model

A 90% learning curve means the cumulative average time per unit falls to 90% of its previous level every time cumulative production doubles. Suppose the first 100 units average 100 hours each (cumulative average = 100 hours/unit, total = $10, 000$ hours). When cumulative output doubles to 200 units, the cumulative average falls to $100 \times 0.90 = 90$ hours/unit (total = $18, 000$ hours). Doubling again to 400 units, the cumulative average falls to $90 \times 0.90 = 81$ hours/unit (total = $32, 400$ hours).

Notice that the 90% rate applies to the cumulative average per unit, not to the next batch alone: the second 100 units (101-200) took $18, 000 - 10, 000 = 8, 000$ hours in total, or 80 hours/unit on average - not the 90-hour cumulative average across all 200 units.

At a direct labor rate of $20 per hour, the incremental cost of producing units 201-400 is $32, 400 - 18, 000 = 14, 400$ hours, or $14, 400 \times $20 = $288, 000$ .

Answer: the cumulative average time per unit is 90 hours/unit at 200 units and 81 hours/unit at 400 units.

The learning effect applies to labor time and labor-driven costs, such as direct labor and labor-based variable overhead - it does not reduce direct materials cost per unit, which stays constant regardless of cumulative output.

Learning curve analysis is based on several key assumptions that explain how labor efficiency improves over time as production volume increases. These assumptions are critical for understanding how and when to apply the learning curve model.

Basic assumptions of learning curve analysis:

Increasing efficiency: As workers become more familiar with a task, they perform it more efficiently. This results in reduced time and effort needed to produce each additional unit.
Nonlinear curve: The curve follows a power (log-linear) function of cumulative output rather than a straight line: $y = a \times x^{b}$ , where $y$ is the cumulative average time per unit, $x$ is the cumulative number of units produced, $a$ is the time required for the first unit, and $b = \frac{l o g ( rate )}{l o g 2}$ is derived from the learning curve rate. Because $b$ is negative (the rate is less than 1), this produces a constant percentage reduction in the cumulative average with each doubling of output, so the biggest gains happen early in production and get smaller as cumulative output grows.
Stable production environment: The learning curve assumes that the production environment remains stable. This means there are no significant changes to technology, tools, or processes that could disrupt the learning curve pattern.
Repetitive tasks: Learning curve analysis applies best to tasks or processes that are repetitive in nature. The more often a worker repeats a task, the more efficiency is gained, leading to the observed decrease in time or cost per unit.
Cumulative output: The model assumes that efficiency is linked to cumulative output. As total production increases, the learning effect takes hold, and efficiency gains are realized with each doubling of cumulative output.

Benefits and shortcomings of learning curve analysis

Learning curve analysis is a valuable tool for cost management and performance evaluation, but like other forecasting models, it has practical limitations worth understanding before you apply it.

Benefits

Improved cost forecasting: Learning curves provide an accurate way to predict how labor costs will decrease over time, supporting budgeting, pricing, resource allocation, and capacity planning decisions.
Efficiency gains: By understanding how workers improve with experience, companies can plan for productivity gains and adjust production schedules accordingly.
Strategic decision-making: Learning curves help managers make informed decisions about pricing, hiring, and scaling production.

Shortcomings

Assumption of continuous improvement: Learning curve analysis assumes that efficiency will continue to improve at a consistent rate, which may not always be true. Factors such as process changes, workforce turnover, or equipment malfunctions can disrupt the expected efficiency gains.
Data dependency: Accurate learning curve analysis requires reliable data on past production performance. Inaccurate or incomplete data can lead to flawed forecasts.
Not applicable to all industries: Learning curve analysis is most applicable to industries with repetitive processes. In highly dynamic or creative industries, where tasks are non-repetitive, the learning curve model may not provide meaningful insights.

Limitations of the learning curve percentage

A learning curve percentage must fall strictly between 50% and 100%. At exactly 50%, doubling production leaves total hours unchanged: using the example above, the first 100 units average 100 hours each for a total of 10,000 hours, and at a 50% curve, 200 units would average 50 hours each - still a total of 10,000 hours ( $200 \times 50 = 10, 000$ ) - meaning the second 100 units take zero additional hours. A rate below 50% would require total hours to decrease as output increases, which isn’t possible. At 100%, there’s no efficiency gain at all (doubling output simply doubles time or cost), and a rate above 100% would mean efficiency getting worse over time - neither is realistic. In practice, learning curve percentages typically range from 70% to 90%: a 70% curve produces a bigger drop in cumulative average time or cost with each doubling of production, while a 90% curve represents a smaller, slower rate of improvement.

Learning curve analysis basics

Forecasting technique: predicts labor efficiency gains as cumulative output grows
Learning curve: graph of decreasing labor time/cost per unit as production doubles
Improvement rate: % decrease in time/cost each time cumulative output doubles
Cumulative average-time learning model: tracks average time/cost per unit across all units produced so far

Calculating with the model

Formula: $y = a \times x^{b}$ , where $a$ = time for first unit, $x$ = cumulative units, $b = \frac{l o g ( rate )}{l o g 2}$
Rate applies to cumulative average, not to the next batch alone
Example (90% curve, first unit avg = 100 hrs):
- At 200 units: cumulative avg = $100 \times 0.90 = 90$ hrs/unit (total 18,000 hrs)
- At 400 units: cumulative avg = $90 \times 0.90 = 81$ hrs/unit (total 32,400 hrs)
- Incremental batch cost found by subtracting totals, then multiplying by labor rate
Learning effect applies only to labor time/labor-driven costs — direct materials cost per unit stays constant

Key assumptions

Efficiency increases with repetition; biggest gains occur early, shrinking as output grows
Curve is nonlinear (power/log-linear function), not a straight line
Assumes stable production environment (no major tech/process changes)
Best suited to repetitive tasks and processes
Efficiency tied to cumulative output, realized at each doubling

Benefits

Improves cost forecasting for budgeting, pricing, capacity planning
Helps anticipate efficiency gains for production scheduling
Supports strategic decisions (pricing, hiring, scaling)

Shortcomings

Assumes continuous, consistent improvement (can be disrupted by process changes, turnover, equipment issues)
Requires reliable historical data; poor data skews forecasts
Not useful for non-repetitive, dynamic, or creative industries

Valid range for learning curve percentage

Must be strictly between 50% and 100%
At 50%: total hours stop increasing (no logical lower bound below this)
At 100%: no efficiency gain (doubling output just doubles time/cost)
Typical real-world range: 70%–90%
- 70% = faster improvement rate
- 90% = slower improvement rate

Next | 2.3.2.2 Learning curve calculation

Learning curve introduction

Learning outcome statements

The learning outcome statements relevant for this section are:

demonstrate an understanding of learning curve analysis
calculate the results under a cumulative average-time learning model
identify the benefits and shortcomings of learning curve analysis

Definitions

Learning curve analysis: This is a forecasting technique used to predict how labor efficiency improves as workers gain experience in producing goods or delivering services. The concept is based on the idea that as employees repeat a task, they become more skilled, which leads to faster production times and lower costs. The more frequently a task is performed, the greater the efficiency gains.
Learning curve: A graphical representation that shows the rate at which labor costs or production times decrease as the cumulative output increases.
Cumulative average-time learning model: A model that calculates the average time or cost per unit, considering all previous units produced.
Improvement rate: The percentage by which time or cost decreases as the number of units produced doubles.

Understanding learning curve analysis

Example: Cumulative average-time learning model

A 90% learning curve means the cumulative average time per unit falls to 90% of its previous level every time cumulative production doubles. Suppose the first 100 units average 100 hours each (cumulative average = 100 hours/unit, total = $10, 000$ hours). When cumulative output doubles to 200 units, the cumulative average falls to $100 \times 0.90 = 90$ hours/unit (total = $18, 000$ hours). Doubling again to 400 units, the cumulative average falls to $90 \times 0.90 = 81$ hours/unit (total = $32, 400$ hours).

Notice that the 90% rate applies to the cumulative average per unit, not to the next batch alone: the second 100 units (101-200) took $18, 000 - 10, 000 = 8, 000$ hours in total, or 80 hours/unit on average - not the 90-hour cumulative average across all 200 units.

At a direct labor rate of $20 per hour, the incremental cost of producing units 201-400 is $32, 400 - 18, 000 = 14, 400$ hours, or $14, 400 \times $20 = $288, 000$ .

Answer: the cumulative average time per unit is 90 hours/unit at 200 units and 81 hours/unit at 400 units.

Basic assumptions of learning curve analysis:

Increasing efficiency: As workers become more familiar with a task, they perform it more efficiently. This results in reduced time and effort needed to produce each additional unit.
Nonlinear curve: The curve follows a power (log-linear) function of cumulative output rather than a straight line: $y = a \times x^{b}$ , where $y$ is the cumulative average time per unit, $x$ is the cumulative number of units produced, $a$ is the time required for the first unit, and $b = \frac{l o g ( rate )}{l o g 2}$ is derived from the learning curve rate. Because $b$ is negative (the rate is less than 1), this produces a constant percentage reduction in the cumulative average with each doubling of output, so the biggest gains happen early in production and get smaller as cumulative output grows.
Stable production environment: The learning curve assumes that the production environment remains stable. This means there are no significant changes to technology, tools, or processes that could disrupt the learning curve pattern.
Repetitive tasks: Learning curve analysis applies best to tasks or processes that are repetitive in nature. The more often a worker repeats a task, the more efficiency is gained, leading to the observed decrease in time or cost per unit.
Cumulative output: The model assumes that efficiency is linked to cumulative output. As total production increases, the learning effect takes hold, and efficiency gains are realized with each doubling of cumulative output.

Benefits and shortcomings of learning curve analysis

Learning curve analysis is a valuable tool for cost management and performance evaluation, but like other forecasting models, it has practical limitations worth understanding before you apply it.

Benefits

Improved cost forecasting: Learning curves provide an accurate way to predict how labor costs will decrease over time, supporting budgeting, pricing, resource allocation, and capacity planning decisions.
Efficiency gains: By understanding how workers improve with experience, companies can plan for productivity gains and adjust production schedules accordingly.
Strategic decision-making: Learning curves help managers make informed decisions about pricing, hiring, and scaling production.

Shortcomings

Assumption of continuous improvement: Learning curve analysis assumes that efficiency will continue to improve at a consistent rate, which may not always be true. Factors such as process changes, workforce turnover, or equipment malfunctions can disrupt the expected efficiency gains.
Data dependency: Accurate learning curve analysis requires reliable data on past production performance. Inaccurate or incomplete data can lead to flawed forecasts.
Not applicable to all industries: Learning curve analysis is most applicable to industries with repetitive processes. In highly dynamic or creative industries, where tasks are non-repetitive, the learning curve model may not provide meaningful insights.

Limitations of the learning curve percentage

Achievable CMA Part 1

2. Planning, budgeting, and forecasting

2.3. Forecasting techniques

2.3.2. Learning curve analysis

Our CMA Part 1 course is currently in development and is a work-in-progress.

Learning curve introduction

6 min read

Font

Discuss

Feedback

Learning outcome statements

The learning outcome statements relevant for this section are:

demonstrate an understanding of learning curve analysis
calculate the results under a cumulative average-time learning model
identify the benefits and shortcomings of learning curve analysis

Definitions

Learning curve analysis: This is a forecasting technique used to predict how labor efficiency improves as workers gain experience in producing goods or delivering services. The concept is based on the idea that as employees repeat a task, they become more skilled, which leads to faster production times and lower costs. The more frequently a task is performed, the greater the efficiency gains.
Learning curve: A graphical representation that shows the rate at which labor costs or production times decrease as the cumulative output increases.
Cumulative average-time learning model: A model that calculates the average time or cost per unit, considering all previous units produced.
Improvement rate: The percentage by which time or cost decreases as the number of units produced doubles.

Understanding learning curve analysis

Example: Cumulative average-time learning model

A 90% learning curve means the cumulative average time per unit falls to 90% of its previous level every time cumulative production doubles. Suppose the first 100 units average 100 hours each (cumulative average = 100 hours/unit, total = $10, 000$ hours). When cumulative output doubles to 200 units, the cumulative average falls to $100 \times 0.90 = 90$ hours/unit (total = $18, 000$ hours). Doubling again to 400 units, the cumulative average falls to $90 \times 0.90 = 81$ hours/unit (total = $32, 400$ hours).

Notice that the 90% rate applies to the cumulative average per unit, not to the next batch alone: the second 100 units (101-200) took $18, 000 - 10, 000 = 8, 000$ hours in total, or 80 hours/unit on average - not the 90-hour cumulative average across all 200 units.

At a direct labor rate of $20 per hour, the incremental cost of producing units 201-400 is $32, 400 - 18, 000 = 14, 400$ hours, or $14, 400 \times $20 = $288, 000$ .

Answer: the cumulative average time per unit is 90 hours/unit at 200 units and 81 hours/unit at 400 units.

Basic assumptions of learning curve analysis:

Increasing efficiency: As workers become more familiar with a task, they perform it more efficiently. This results in reduced time and effort needed to produce each additional unit.
Nonlinear curve: The curve follows a power (log-linear) function of cumulative output rather than a straight line: $y = a \times x^{b}$ , where $y$ is the cumulative average time per unit, $x$ is the cumulative number of units produced, $a$ is the time required for the first unit, and $b = \frac{l o g ( rate )}{l o g 2}$ is derived from the learning curve rate. Because $b$ is negative (the rate is less than 1), this produces a constant percentage reduction in the cumulative average with each doubling of output, so the biggest gains happen early in production and get smaller as cumulative output grows.
Stable production environment: The learning curve assumes that the production environment remains stable. This means there are no significant changes to technology, tools, or processes that could disrupt the learning curve pattern.
Repetitive tasks: Learning curve analysis applies best to tasks or processes that are repetitive in nature. The more often a worker repeats a task, the more efficiency is gained, leading to the observed decrease in time or cost per unit.
Cumulative output: The model assumes that efficiency is linked to cumulative output. As total production increases, the learning effect takes hold, and efficiency gains are realized with each doubling of cumulative output.

Benefits and shortcomings of learning curve analysis

Learning curve analysis is a valuable tool for cost management and performance evaluation, but like other forecasting models, it has practical limitations worth understanding before you apply it.

Benefits

Improved cost forecasting: Learning curves provide an accurate way to predict how labor costs will decrease over time, supporting budgeting, pricing, resource allocation, and capacity planning decisions.
Efficiency gains: By understanding how workers improve with experience, companies can plan for productivity gains and adjust production schedules accordingly.
Strategic decision-making: Learning curves help managers make informed decisions about pricing, hiring, and scaling production.

Shortcomings

Assumption of continuous improvement: Learning curve analysis assumes that efficiency will continue to improve at a consistent rate, which may not always be true. Factors such as process changes, workforce turnover, or equipment malfunctions can disrupt the expected efficiency gains.
Data dependency: Accurate learning curve analysis requires reliable data on past production performance. Inaccurate or incomplete data can lead to flawed forecasts.
Not applicable to all industries: Learning curve analysis is most applicable to industries with repetitive processes. In highly dynamic or creative industries, where tasks are non-repetitive, the learning curve model may not provide meaningful insights.

Limitations of the learning curve percentage

Learning curve analysis basics

Forecasting technique: predicts labor efficiency gains as cumulative output grows
Learning curve: graph of decreasing labor time/cost per unit as production doubles
Improvement rate: % decrease in time/cost each time cumulative output doubles
Cumulative average-time learning model: tracks average time/cost per unit across all units produced so far

Calculating with the model

Formula: $y = a \times x^{b}$ , where $a$ = time for first unit, $x$ = cumulative units, $b = \frac{l o g ( rate )}{l o g 2}$
Rate applies to cumulative average, not to the next batch alone
Example (90% curve, first unit avg = 100 hrs):
- At 200 units: cumulative avg = $100 \times 0.90 = 90$ hrs/unit (total 18,000 hrs)
- At 400 units: cumulative avg = $90 \times 0.90 = 81$ hrs/unit (total 32,400 hrs)
- Incremental batch cost found by subtracting totals, then multiplying by labor rate
Learning effect applies only to labor time/labor-driven costs — direct materials cost per unit stays constant

Key assumptions

Efficiency increases with repetition; biggest gains occur early, shrinking as output grows
Curve is nonlinear (power/log-linear function), not a straight line
Assumes stable production environment (no major tech/process changes)
Best suited to repetitive tasks and processes
Efficiency tied to cumulative output, realized at each doubling

Benefits

Improves cost forecasting for budgeting, pricing, capacity planning
Helps anticipate efficiency gains for production scheduling
Supports strategic decisions (pricing, hiring, scaling)

Shortcomings

Assumes continuous, consistent improvement (can be disrupted by process changes, turnover, equipment issues)
Requires reliable historical data; poor data skews forecasts
Not useful for non-repetitive, dynamic, or creative industries

Valid range for learning curve percentage

Must be strictly between 50% and 100%
At 50%: total hours stop increasing (no logical lower bound below this)
At 100%: no efficiency gain (doubling output just doubles time/cost)
Typical real-world range: 70%–90%
- 70% = faster improvement rate
- 90% = slower improvement rate

Next | 2.3.2.2 Learning curve calculation

Learning curve introduction

Learning outcome statements

The learning outcome statements relevant for this section are:

demonstrate an understanding of learning curve analysis
calculate the results under a cumulative average-time learning model
identify the benefits and shortcomings of learning curve analysis

Definitions

Learning curve analysis: This is a forecasting technique used to predict how labor efficiency improves as workers gain experience in producing goods or delivering services. The concept is based on the idea that as employees repeat a task, they become more skilled, which leads to faster production times and lower costs. The more frequently a task is performed, the greater the efficiency gains.
Learning curve: A graphical representation that shows the rate at which labor costs or production times decrease as the cumulative output increases.
Cumulative average-time learning model: A model that calculates the average time or cost per unit, considering all previous units produced.
Improvement rate: The percentage by which time or cost decreases as the number of units produced doubles.

Understanding learning curve analysis

Example: Cumulative average-time learning model

A 90% learning curve means the cumulative average time per unit falls to 90% of its previous level every time cumulative production doubles. Suppose the first 100 units average 100 hours each (cumulative average = 100 hours/unit, total = $10, 000$ hours). When cumulative output doubles to 200 units, the cumulative average falls to $100 \times 0.90 = 90$ hours/unit (total = $18, 000$ hours). Doubling again to 400 units, the cumulative average falls to $90 \times 0.90 = 81$ hours/unit (total = $32, 400$ hours).

Notice that the 90% rate applies to the cumulative average per unit, not to the next batch alone: the second 100 units (101-200) took $18, 000 - 10, 000 = 8, 000$ hours in total, or 80 hours/unit on average - not the 90-hour cumulative average across all 200 units.

At a direct labor rate of $20 per hour, the incremental cost of producing units 201-400 is $32, 400 - 18, 000 = 14, 400$ hours, or $14, 400 \times $20 = $288, 000$ .

Answer: the cumulative average time per unit is 90 hours/unit at 200 units and 81 hours/unit at 400 units.

Basic assumptions of learning curve analysis:

Increasing efficiency: As workers become more familiar with a task, they perform it more efficiently. This results in reduced time and effort needed to produce each additional unit.
Nonlinear curve: The curve follows a power (log-linear) function of cumulative output rather than a straight line: $y = a \times x^{b}$ , where $y$ is the cumulative average time per unit, $x$ is the cumulative number of units produced, $a$ is the time required for the first unit, and $b = \frac{l o g ( rate )}{l o g 2}$ is derived from the learning curve rate. Because $b$ is negative (the rate is less than 1), this produces a constant percentage reduction in the cumulative average with each doubling of output, so the biggest gains happen early in production and get smaller as cumulative output grows.
Stable production environment: The learning curve assumes that the production environment remains stable. This means there are no significant changes to technology, tools, or processes that could disrupt the learning curve pattern.
Repetitive tasks: Learning curve analysis applies best to tasks or processes that are repetitive in nature. The more often a worker repeats a task, the more efficiency is gained, leading to the observed decrease in time or cost per unit.
Cumulative output: The model assumes that efficiency is linked to cumulative output. As total production increases, the learning effect takes hold, and efficiency gains are realized with each doubling of cumulative output.

Benefits and shortcomings of learning curve analysis

Learning curve analysis is a valuable tool for cost management and performance evaluation, but like other forecasting models, it has practical limitations worth understanding before you apply it.

Benefits

Improved cost forecasting: Learning curves provide an accurate way to predict how labor costs will decrease over time, supporting budgeting, pricing, resource allocation, and capacity planning decisions.
Efficiency gains: By understanding how workers improve with experience, companies can plan for productivity gains and adjust production schedules accordingly.
Strategic decision-making: Learning curves help managers make informed decisions about pricing, hiring, and scaling production.

Shortcomings

Assumption of continuous improvement: Learning curve analysis assumes that efficiency will continue to improve at a consistent rate, which may not always be true. Factors such as process changes, workforce turnover, or equipment malfunctions can disrupt the expected efficiency gains.
Data dependency: Accurate learning curve analysis requires reliable data on past production performance. Inaccurate or incomplete data can lead to flawed forecasts.
Not applicable to all industries: Learning curve analysis is most applicable to industries with repetitive processes. In highly dynamic or creative industries, where tasks are non-repetitive, the learning curve model may not provide meaningful insights.

Limitations of the learning curve percentage

Key points

Learning curve analysis basics

Forecasting technique: predicts labor efficiency gains as cumulative output grows
Learning curve: graph of decreasing labor time/cost per unit as production doubles
Improvement rate: % decrease in time/cost each time cumulative output doubles
Cumulative average-time learning model: tracks average time/cost per unit across all units produced so far

Calculating with the model

Formula: $y = a \times x^{b}$ , where $a$ = time for first unit, $x$ = cumulative units, $b = \frac{l o g ( rate )}{l o g 2}$
Rate applies to cumulative average, not to the next batch alone
Example (90% curve, first unit avg = 100 hrs):
- At 200 units: cumulative avg = $100 \times 0.90 = 90$ hrs/unit (total 18,000 hrs)
- At 400 units: cumulative avg = $90 \times 0.90 = 81$ hrs/unit (total 32,400 hrs)
- Incremental batch cost found by subtracting totals, then multiplying by labor rate
Learning effect applies only to labor time/labor-driven costs — direct materials cost per unit stays constant

Key assumptions

Efficiency increases with repetition; biggest gains occur early, shrinking as output grows
Curve is nonlinear (power/log-linear function), not a straight line
Assumes stable production environment (no major tech/process changes)
Best suited to repetitive tasks and processes
Efficiency tied to cumulative output, realized at each doubling

Benefits

Improves cost forecasting for budgeting, pricing, capacity planning
Helps anticipate efficiency gains for production scheduling
Supports strategic decisions (pricing, hiring, scaling)

Shortcomings

Assumes continuous, consistent improvement (can be disrupted by process changes, turnover, equipment issues)
Requires reliable historical data; poor data skews forecasts
Not useful for non-repetitive, dynamic, or creative industries

Valid range for learning curve percentage

Must be strictly between 50% and 100%
At 50%: total hours stop increasing (no logical lower bound below this)
At 100%: no efficiency gain (doubling output just doubles time/cost)
Typical real-world range: 70%–90%
- 70% = faster improvement rate
- 90% = slower improvement rate