Instructional Goals: You will understand:

  • The difference between opportunity costs and accounting costs.
  • The importance of sensitivity analysis.
  • How to derive long run cost curves from production functions by minimizing long run costs using both marginal and incremental analysis.
  • How to derive short run cost curves from short run production functions.
  • How to perform a shut down analysis.
  • How to use Break-even analysis as a rule of thumb.

And the relevance of these concepts to operational decision making. These concepts will also have relevance to more complex questions of marketing policy and strategy.

Opportunity costs vs. accounting costs

Costs are bad things endured or good things lost. Cost always means cost to do something. You cannot have a cost without a cost objective. Most of the confusion about costs reflects a failure to be clear about cost objectives. Nevertheless, where economists and accountants are concerned, there is a second and equally critical source of confusion about costs: economists and accountants use the term "cost" to mean different although related things.

Economists define cost in terms of opportunities that are sacrificed when a choice is made. Hence, economic costs are simply benefits lost (and, in some cases, benefits are merely costs avoided). Economic costs are subjective -- seen from the perspective of a decision maker not a detached observer -- and prospective. Moreover, economic cost is a stock concept -- economic costs are incurred when decisions are made. Economic cost estimates are used for making decisions about pricing, output levels, buying or making, alternative marketing tactics/strategies, product introductions and withdrawals, etc.

Accountants define cost in terms of resources consumed. Hence, from an accountant’s standpoint, costs are objective -- seen from the perspective of a detached observer -- and retrospective. Accountants usually define costs as flows. Accounting costs reflect changes in stocks (reductions in good things, increases in bad things) over a fixed period of time. Accounting cost measures are used in the evaluation of managerial performance (usually together with information on income) and as a basis for economic cost estimation.

There are two kinds of mistakes you can make when you use accounting costs to estimate economic costs: you can include cost measures that should be ignored; or, you can ignore costs that should be included. You should ignore costs that will not vary as a result of your decision; you should include all costs that will vary as a result of your decision.

Example of including costs that should be ignored: sunk costs.

The DOE developed a cyclotron to enrich uranium. It spent billions on research and development, and almost had a fully operational machine (a rare "success") but they never brought it on-line because Congress required them to charge a high enough price for enriched uranium to recover the cost of capital. If they brought it on-line, they would have "priced themselves out of the market."

Examples of ignoring costs that should be considered: the "hidden cost" problem.

opportunity costs of capital (like all costs, opportunity cost depend on the question being asked)
opportunity cost of office space.


Dan Connor, Artie Zimmer, and Bob Bruss each earn $25 per hour (including vacations, social security, and other benefits) as automobile mechanics at a car dealership. They are considering opening a shop that specializes in fast-service oil changes. The projected annual cost of the building and equipment is $60,000. On average, an oil change requires $6 of materials (oil and oil filter). The average price of an oil change is $20 in the shops that currently provide this service. Dan, Artie, and Bob think that they could each perform as many as six oil changes per hour and could charge at least $25 per oil change if they could guarantee to return the customer's car in fifteen minutes or less.

What are the accounting and economic (or opportunity) cost functions for this oil-changing business?
Opportunity costs are the costs of the foregone or next best alternative.
One should consider opportunity costs, not accounting costs when making decisions.


You are trying to determine whether it makes sense to buy a house or to continue renting for $1100. You start looking at houses but decide that you will not purchase a house unless it is more profitable than renting. What price house is equivalent to $1100 in rent? Make several assumptions:

1. Assume that any transactions costs incurred in buying and selling the house will exactly be offset by the capital gains on the house

2. Assume that you borrow the purchase price at 7.25%. Your investments also earn 7.25%.

3. Assume that you do not pay any principle on the mortgage, only interest.

HINT: the annual rent equivalent for a $350,000 house is $25375.

Now you think that you will make enough money to put yourself in a 33% tax bracket. Redo the calculation above, assuming a 33% tax bracket, and explain in words how this affects your willingness to pay for a house.

Suppose that you think there is a 75% chance that you will make enough to put yourself in a 33% tax bracket, and a 25% chance that you will stay in the 0% tax bracket. Compute the expected value of monthly rent equivalent payment on a $250,000 house.

Note this is a version of what is known as sensitivity analysis. When you are uncertain about the future, it is important to redo your forecasts, computations, or spreadsheets, using a variety of scenarios: a best case, middling, and worse case scenario.

Long run production functions and cost functions

Q = f(K, L). Quantity is a function of the inputs used to produce it: in this example capital and labor. Quantity is measured as a rate of production (flow) as are capital and labor, e.g. the amount of cars washed per day is a function of the amount of labor and capital used each day.

  • The production function specifies a technically efficient use of labor and capital necessary to produce output, i.e. no resources are "wasted."
  • The cost function specifies an economically efficient use of resources, i.e. the firm chooses the least cost combination of inputs, to produce a given output.
  • This yields the long run cost function: total costs (C) = g(Q), which depends on the prices of inputs. This function can be a pretty good proxy for the opportunity cost of delivering Q, at least where we measure costs in units of present value: i.e., the change in present value or owner’s equity caused by some specified action (and where for purposes of measurement the attendant increase in wealth is excluded from the computation of equity).

The following long-run functional relationships traditionally obtain in the single product case: Cost varies as a function of total production volume (V), the rate of output (x), the date of first delivery (T), and the date of completion of the full production run (m), where x(t) denotes the rate of output at moment t. Moreover, as the total quantity of units produced increases, the cost of future output tends to decline because production-knowledge increases as a result of production experience (this proposition is known as the ‘learning’ or ‘progress’ curve"). That is: dC / dT | x = x0 , V = V0 , < 0. This relationship probably holds for most products produced in large batches using traditional mass-production methods. Moreover it is usually assumed that:

  1. dC / dx(t) | T = T0, V = V0 > 0
  2. d2C / dx(t)2 | T = T0, V = V0 > 0
  3. dC / dV | x = x0, T = T0 > 0
  4. d2C / dV2 | x = x0, T = T0 < 0

Many of these functional relationships have been attenuated by the rise of computer assisted design and manufacturing technology and modern information.

Long run cost minimization: marginal analysis

Here we will assume that the firm can choose any level of capital and labor to produce output, Q. More capital leads to more output; less capital to less output. More labor leads to more output; less labor to less output. This is known as a variable proportions production technology because labor can substitute for capital, and vice-versa, in production.

  • The marginal product of labor is the additional output from one extra unit of labor.
  • The marginal product of capital is the additional output from one more unit of capital.

The cost minimization rule for producing a given quantity: choose labor and capital such that (MP of labor)/(Price of labor) = (MP of capital)/(price of capital).

Proof: dividing the marginal products of each input by the price of the input tells you how much output you can produce for a dollar. If it costs more to produce output using labor than it does using capital, then sell labor and buy capital. This allows you to produce the same amount at lower cost. Only when the costs of production using each input are the same are no further cost savings possible.

  • If (MP of labor)/(Price of labor) is greater than (MP of capital)/(price of capital), sell capital, buy labor.
  • If (MP of labor)/(Price of labor) is less than (MP of capital)/(price of capital), sell labor, buy capital.
Long run cost minimization: incremental analysis

Incremental analysis considers large discrete changes in input mix, whereas marginal analysis considers small continuous changes in input mix.

Example: 1985: John Deere acquisition of Versatile. John Deere had an old fashioned production line for making farm tractors. Very high fixed costs, but low marginal costs. Versatile had a "garage" style production facility with much lower fixed costs, but higher marginal costs.


  • It is easy to see that if production is less than or equal to 6 units, then Versatile has lower costs of production. If production is larger than or equal to 7 units, John Deere has lower costs of production.
  • The long run decision between these two production processes would depend on how many units you thought you would sell. If you anticipated selling 7 or more units, use the John Deere production process, if you thought you would sell 6 or fewer units, use the Versatile production process.
Sensitivity analysis

Suppose you were uncertain about how many units you anticipated selling. Sensitivity analysis allows you to build in uncertainty to your analysis by determining the costs of various scenarios.

Suppose you thought that your uncertainty was best described by a trinomial random variable:

  1. with p1 = .4, Q = 7
  2. with p2 = .3, Q = 10
  3. with p3 = .3, Q = 4; note that p3 = 1-p1-p2

What's the best technology to choose?

ANSWER: The expected output is 7 (.4*7+.3*4+.3*10 = 7), the average cost of the expected output is not the same as the expected average cost, because Versatile has a large advantage at small outputs, while Deere has a small advantage at high outputs. The table below shows how to compute expected average costs, which is the usually right criterion to use for deciding which technology to adopt.

Short run production functions

A long run production function relates the output produced to the inputs used, e.g. Q = f(capital, labor). In the short run, some inputs cannot be varied, so the firm does not have as much flexibility as in the long run. In this case, the short run production function is a function of only the inputs that can be varied. Suppose that capital is fixed in the short run. Then Y = g(labor)

The "usual" shape of the short run production function:

1. In the short run, output at first increases at an increasing rate with increases in labor (increasing returns to labor)

2. Then output increases at a constant rate with increases in labor (constant returns to labor).

3. Finally, output increases at a decreasing rate with increases in labor (diminishing returns to labor).

Short run cost functions are larger than long run cost functions because, in the short run, fixed inputs can not be varied. In the long run, all inputs can be varied, and this greater flexibility allows you to achieve lower costs, i.e. h(Q) is greater than or equal to g(Q).

A Numerical example

Assume, the following short run production function: Fixed costs = $20/hour, Labor costs $5/hour * The marginal productivity of labor increases, then is constant, and then decreases.

Suppose that the output sells for $1 per unit and that the firm can sell all it wants at a price of $1 (infinite elasticity, or a perfectly competitive firm). How much labor should the firm hire?

ANSWER: Keep hiring as long as the benefit of hiring another worker is greater than the cost of another worker. The benefit equals the marginal revenue of the worker (price times the marginal production), Benefit = $1*(marginal production); The cost is the wage. Cost = $5.

  • It is easy to see that profits are maximized with 11 laborers.
    • Can we derive the optimal production decision (how much labor to produce) from the cost curves instead of the production curves?

    Yes: keep producing as long as the benefit of producing another unit ($1) exceeds the cost of producing another unit. Looking at the graph, you can see that $1 intersects the marginal cost curve somewhere between 10 and 11 laborers (between 75 and 84 units). To determine whether to produce at 75 or 84, you must look at the spread sheet.

    Shut down analysis

    In the long run, stop producing if economic profits are negative.

    In the short run, stop producing if revenue is less than total variable cost. In the short run, you don't have to cover your fixed costs, but you must make enough to cover your variable costs. If not, then you can shut down. You will still have to pay your fixed costs, but at least you can avoid paying your variable costs (which are greater than revenue).

    Break-even analysis

    Assumptions: constant price, constant average variable cost. (P-AVC) is sometimes called "contribution margin" because it represents profit per unit sold (ignoring fixed costs).

    Set profits equal to zero to solve for how much output would be required to "break even." Another way of asking the same question is to ask how much quantity would be required to produce enough profit to cover fixed costs. revenue-variable costs-fixed Costs = 0 P*Q-AVC*Q -fixed Costs = 0 (P-AVC)*Q-fixed Costs = 0 Q = fixed Costs/(P-AVC)