π Lecture 21 β Cost Analysis and Estimation (Continued 3) β Empirical Estimation of Cost Functions
π Overview: This lecture covers the practical methods for estimating cost functions using empirical data, addressing both short-run and long-run perspectives. It explains the data collection challenges, functional forms of cost curves, and the specific techniques used to estimate long-run cost functions, emphasizing their importance for managerial decisions on output pricing and plant scale.
ποΈ Topics Covered
The lecture explores the empirical estimation of short-run cost functions using regression analysis, detailing data and measurement problems such as differentiating opportunity costs from accounting costs and correcting for inflation. It then examines the functional forms of short-run cost functions (cubic, quadratic, and linear) and their shapes. Finally, it introduces three key methods for estimating long-run cost curves: cross-sectional regression analysis, the engineering technique, and the survival technique, noting the common finding of L-shaped long-run average cost curves.
π Lecture Summary
Empirical Estimation of Cost Functions
The study of cost curves began with Joel Dean, who pioneered the first managerial economics textbook. Estimating short-run cost functions helps define short-run marginal costs, which assists managers in setting output and prices. In the long run, the firmβs decision involves building the most efficient plant size, which depends on the existence of scale economies and diseconomies.
For short-run cost functions, researchers often use time series data for a specific plant over time, assuming plant size and technology do not change significantly. For long-run cost functions, which are planning functions, cross-sectional analysis is most common.
Empirical cost estimates are essential for determining the optimal level of output and price (short-run) and for planning the optimal scale of plant (long-run).
The Estimation of Short-Run Cost Functions
Data and Measurement Problem The most common method is regression analysis, where total variable costs (TVC) are regressed against output and other variables like input prices. TVC is used instead of total cost because allocating fixed costs to multiple products is difficult. The firmβs average variable cost (AVC) and marginal cost (MC) functions are then derived from the TVC function.
Empirical Estimation Data Collection Issues
Key data collection issues include:
- Opportunity Costs Must Be Extracted from Accounting Cost Data: Each input must be valued at its opportunity cost, not its historical expenditure.
- Costs Must Be Divided Among Products: Proper allocation is required.
- Costs Must Be Matched to Output Over Time: Costs should be allocated to the period when output was produced, not when costs were incurred.
- Costs Must Be Corrected for Inflation: Input prices must be deflated using appropriate price indices.
Other independent variables in the regression may include fuel and material costs, input quality, technology, weather, and product mix changes. The general regression model is:
π (1) C = f(Q, Xβ, Xβ, ... , Xβ) Where C = total variable costs, Q = output, and Xβs = other determinants.
π Definition β Opportunity Cost: The value of an input in its best alternative use, as opposed to its historical accounting cost. π Example: If a firm owns its building, the cost of using it is not zero but equals the rent the firm could earn by renting it to the highest bidder. Similarly, inventories must be valued at current market prices, not historical cost.
π‘ Why this matters: Accounting data records actual expenses, but economic decision-making requires opportunity costs. This is the most difficult problem in empirical estimation.
The Functional Form of Short-Run Cost Functions
Economic theory postulates an S-shaped (cubic) TVC curve with corresponding U-shaped AVC and MC curves.
Cubic TVC Function π Formulae: (2) TVC = aQ + bQΒ² + cQΒ³ (3) AVC = TVC/Q = a + bQ + cQΒ² (4) MC = a + 2bQ + 3cQΒ²
- Conditions for U-shape: a > 0, b < 0, and c > 0
Linear TVC Function (Approximation) π Formulae: (5) TVC = a + bQ (6) AVC = a/Q + b (7) MC = b The linear form often gives a good empirical fit over the observed output range. The AVC curve becomes flat, approaching the value of b (the horizontal MC curve).
Quadratic TC Function π Formulae: (8) TC = a + bQ + cQΒ² (9) AC = a/Q + b + cQ (10) MC = b + 2cQ
- Conditions: a > 0, c > 0 and b < 0 The quadratic function rises at an increasing rate throughout (diminishing returns).
The Shapes of Short-Run Cost Function
Three main specifications are used:
- Cubic relationship: TVC first increases at a decreasing rate, then at an increasing rate (S-shaped).
- Quadratic relationship: TVC increases at an increasing rate.
- Linear relationship: TVC increases at a constant rate.
These are the three shapes most frequently encountered in statistical studies.
Empirical Estimation Long-Run Cost Curves
Three methods are used:
- Cross-Sectional Regression Analysis
- Engineering Method
- Survival Technique
Estimating Long-Run Cost Functions with Cross-Sectional Data
This method uses cross-sectional data across different firms. Difficulties include firms paying different input prices and uncertainty about whether each firm is operating at the optimal point on its short-run average cost (SAC) curve. If firms are not operating at the optimal points, the estimated LAC curve will overestimate both economies and diseconomies of scale. Empirically, LAC curves are often found to be L-shaped β sharply falling at low outputs, then nearly constant at higher output levels.
Estimating Long-Run Cost Functions With Engineering and Survival Techniques
Engineering Technique: Uses knowledge of physical input-output relationships (production function) to determine optimal input combinations for various output levels. Professionals calculate required inputs and multiply by their prices.
- Advantages: Based on present technology, avoids problems of different input prices and cost-allocation accounting problems.
- Disadvantages: Deals only with technical aspects, ideal conditions, and current technology (which may become obsolete).
- Result: Confirms L-shaped LAC curves, with declining unit costs up to a point and flat costs thereafter.
Survival Technique: An alternative method used when cross-sectional data is insufficient.
β Key Takeaways
The most critical points for an exam are: First, empirical cost estimation requires converting accounting data to economic (opportunity) cost data. Second, the primary functional forms for short-run cost functions are cubic (S-shaped), quadratic, and linear, each with unique equations for TVC, AVC, and MC. Third, long-run cost curves are estimated using cross-sectional regression, the engineering technique, or the survival technique, and they typically appear L-shaped. Fourth, data collection issues such as matching costs to output over time and correcting for inflation are crucial for valid estimation. Finally, the cubic TVC function requires specific coefficient signs (a>0, b<0, c>0) to yield U-shaped AVC and MC curves.
π§ Quick Revision Questions
- Why is Total Variable Cost (TVC) typically estimated in regression analysis instead of Total Cost (TC) for short-run cost functions?
- State the equation and coefficient sign conditions for a cubic Total Variable Cost (TVC) function that produces U-shaped AVC and MC curves.
- What are the three main functional forms (shapes) of short-run cost functions discussed in the lecture?
- What is the primary difference between the engineering technique and cross-sectional regression for estimating long-run cost curves?
- Based on empirical studies, what is the typical shape of the Long-Run Average Cost (LAC) curve, and what does this imply about economies of scale?
π Lecture 22 β Linear Programming
π Overview: This lecture introduces linear programming (LP) as a powerful tool for solving constrained optimization problems, particularly in production decisions. It covers the derivation of isoquants in LP, determination of least-cost input combinations, and the formulation and graphical solution of LP models, emphasizing its application in determining optimal output mixes under resource constraints.
ποΈ Topics Covered
The lecture begins by defining production processes and deriving isoquants in a linear programming framework, using four production processes with fixed input ratios. It then explains how to determine least-cost input combinations by adding isocost curves to isoquants, followed by an analysis of optimal input combinations when resources are limited. The historical perspective of LP and its basic componentsβdecision variables, objective function, constraints, and non-negativityβare introduced. The lecture details the procedure for formulating and solving LP problems, including both maximization and minimization problems, the basic assumptions of the model, and the handling of inequality constraints and the linearity assumption. Finally, it presents a production planning example for multiple products, culminating in a graphical solution.
π Lecture Summary
PRODUCTION PROCESSES ISOQUANTS IN LINEAR PROGRAMMING
Assume a firm produces a single product, Q, using two inputs, L and K (labor and capital). Instead of continuous substitution, Q can be produced using only four input combinations, representing four different production processes. Each process uses a different fixed combination of inputs L and K. These processes are illustrated as rays in Figure 1. Process A requires 15 units of L and 1 unit of K per unit of Q. Process B uses 10 units of L and 2 units of K per unit of output. Processes C and D use 7.5 units of L and 3 units of K, and 5 units of L with 5 units of K, respectively, per unit of Q. Each point along a production ray combines L and K in a fixed ratio. Process A is very labor intensive, whereas B, C, and D are based on increasingly capital-intensive technologies. Doubling both L and K along a ray doubles the quantity of Q produced, indicating constant returns to scale. However, equal distances along different process rays do not ordinarily indicate equal output quantities.
PRODUCTION ISOQUANTS
Joining points of equal output on the four production process rays creates a set of Isoquant curves. Figure 2 illustrates Isoquants for Q = 1, 2, 3, 4, and 5. Each Isoquant represents combinations of input factors L and K that can be used to produce a given quantity of output. Production Isoquants in linear programming are composed of linear segments connecting the various production process rays. Each of these Isoquant segments is parallel to one another. Points along each segment of an Isoquant between two process rays represent a combination of output from each of the two adjoining production processes. Consider point X in Figure 2, which represents production of 4 units of Q using 25 units of L and 16 units of K. That combination is possible by producing part of the output with process C and part with process D. In this case, 2 units of Q can be produced using process C and 2 units using process D.
LEAST-COST INPUT COMBINATIONS
Adding Isocost curves to a set of Isoquants permits one to determine least-cost input combinations. This is shown in Figure 3 under the assumption that each unit of L costs $3 and each unit of K costs $10. The Isocost curve illustrated indicates a total expenditure of $150. The tangency between the Isocost curve and the Isoquant curve for Q = 3 at point B3 indicates that process B, which combines inputs L and K in the ratio 5 to 1, is the least-cost method of producing Q. For any expenditure level, production is maximized by using process B. Alternatively, process B is the least-cost method for producing any quantity of Q, given the assumed prices for L and K.
OPTIMAL INPUT COMBINATIONS WITH LIMITED RESOURCES
Firms with limited inputs often find it optimal to use inputs in proportions other than the least-cost combination. Assume only 20 units of L and 11 units of K are available, and the firm seeks to maximize output. These constraints are shown in Figure 4. The horizontal line at L = 20 indicates the upper limit on L; the vertical line at K = 11 indicates a similar limit on K. Production possibilities are determined by operating within the area bounded by production process rays A and D. Combining production possibilities with input constraints restricts the firm to operation within the shaded area on 0PRS, known as the feasible space.
π Definition β feasible space: The area in a programming problem where all constraints are satisfied simultaneously.
HISTORICAL PERSPECTIVE OF LINEAR PROGRAMMING
The founders of the concept are Leonid Kantorovich, a Russian mathematician who developed linear programming problems in 1939, George B Dantzig, who published the simplex method in 1947, and John von Neumann, who developed the theory of the duality in the same year.
INTRODUCTION TO LINEAR PROGRAMMING
A Linear Programming model seeks to maximize or minimize a linear function, subject to a set of linear constraints. The model consists of: a set of decision variables, an objective function, a set of constraints, and a set of non-negativity constraints. Constraints include production capacity, time, money, raw materials, budget, space, and other restrictions. These can be viewed as inequality constraints. A "linear" programming problem assumes a linear objective function and a series of linear inequality constraints.
PROCEDURE USED IN FORMULATING AND SOLVING LP PROBLEMS
LP is a mathematical modeling technique used to determine a level of operational activity to achieve an objective, subject to restrictions called constraints.
For a maximization problem: Maximize: Ο = cβxβ + cβxβ Subject to: aββxβ + aββxβ β€ bβ, aββxβ + aββxβ β€ bβ, aββxβ + aββxβ β€ bβ Non-negativity: xβ β₯ 0, xβ β₯ 0.
For a minimization problem: Minimize: C = cβxβ + cβxβ Subject to: aββxβ + aββxβ β₯ bβ, aββxβ + aββxβ β₯ bβ, aββxβ + aββxβ β₯ bβ Non-negativity: xβ β₯ 0, xβ β₯ 0.
Compact Form Maximization: Maximize cα΅x, Subject to Ax β€ b, x β₯ 0. Compact Form Minimization: Minimize cα΅x, Subject to Ax β₯ b, x β₯ 0.
BASIC ASSUMPTIONS OF THE MODEL
β The parameter values are known with certainty. β The objective function and constraints exhibit constant returns to scale. β There are no interactions between the decision variables (the additivity assumption). β The Continuity assumption: Variables can take on any value within a given feasible range.
INEQUALITY CONSTRAINTS
Many production or resource constraints are inequalities. Constraints often limit the resource employed to less than or equal to some fixed amount available, or specify that output must be greater than or equal to some minimum requirement. Linear programming handles such constraint inequalities easily.
LINEARITY ASSUMPTION
(1) Constant prices for outputs (as in a perfectly competitive market). (2) Constant returns to scale for production processes. (3) Each decision variable has a non-negativity constraint.
Typical decision problems that can be solved using LP involve revenue and cost functions and their composite, the profit function. Each must be linear; as output increases, revenues, costs, and profits must increase linearly. For revenues to be linear, product prices must be constant. For costs to be linear, returns to scale and input prices must be constant. To illustrate, suppose an oil company must choose the optimal output mix for a refinery with a capacity of 150,000 barrels per day. The company is justified in basing its analysis on the $25-per-barrel prevailing market price, regardless of purchase or sale volume, because within the feasible output range, prices will be approximately constant.
π‘ Why this matters: The linearity assumption is critical because it allows the use of LP; if functions are non-linear, LP cannot be applied, and more complex optimization techniques are needed.
PRODUCTION PLANNING FOR A MULTIPLE PRODUCTS
This problem, faced by a host of companies, is readily solved with LP. LP problems can be solved using graphical techniques, algorithms using matrices, or software (e.g., Fore Profit, LINDO, MS Excel Solver). In the graphical technique, each inequality constraint is graphed as an equality constraint. The Feasible Solution Space is the area which satisfies all inequality constraints. The Optimal Feasible Solution occurs along the boundary of the feasible solution space, at the extreme points or corner points.
OBJECTIVE FUNCTION SPECIFICATION Assume the firm wishes to maximize total profits from two products, X and Y. If per-unit profit contribution is $12 for product X and $9 for product Y, the objective function is: (1) Maximize Ο = $12Qβ + $9QΚ
CONSTRAINT EQUATION SPECIFICATION Table 1 specifies available input quantities and usage for X and Y. 32 units of input A are available, requiring 4 units per X and 2 per Y. 10 units of input B are available, requiring 1 unit per X or Y. 21 units of input C are available, requiring 3 units per Y.
Constraint A: 4Qβ + 2QΚ β€ 32 Constraint B: 1Qβ + 1QΚ β€ 10 Constraint C: 3QΚ β€ 21
NON NEGATIVITY REQUIREMENT To prevent economically meaningless results (e.g., negative output), a non negativity requirement must be introduced: Qβ β₯ 0 and QΚ β₯ 0.
GRAPHIC SPECIFICATION AND SOLUTION
The decision problem is expressed as: (1) Maximize Ο = $12Qβ + $9QΚ Subject to: (2) Input A: 4Qβ + 2QΚ β€ 32 (3) Input B: 1Qβ + 1QΚ β€ 10 (4) Input C: 3QΚ β€ 21 Where Qβ β₯ 0 and QΚ β₯ 0.
STEPS FOLLOWED IN SOLVING A LP PROBLEM ARE:
- Express the objective function as equations and the constraints as inequalities.
- Graph the inequality constraints as equations on a set of coordinates in a plane.
- Identify the feasible solution space on the graph where all constraints are satisfied simultaneously.
- Graph the objective function as a series of isoprofit or isocost lines to find the point on the boundary of this space that maximizes or minimizes the value of the objective function.
- A solution is called feasible when it satisfies all the constraints. The non-negativity constraints are shown by the positive quadrant.
β Key Takeaways
The most critical points from this lecture are that linear programming is a method for optimizing a linear objective function (maximizing profit or minimizing cost) subject to linear inequality constraints representing resource limitations. Production isoquants in LP are composed of linear segments connecting different production process rays, with each ray representing a fixed input ratio. The optimal solution in a graphical LP problem always occurs at a corner point (extreme point) of the feasible region, not in its interior. The linearity assumption requires constant prices and constant returns to scale, which is a key limitation but often a reasonable approximation for short-run decisions. Finally, formulating an LP problem correctlyβspecifying the objective function, all constraints, and non-negativity conditionsβis the most difficult but crucial step in the analysis.
π§ Quick Revision Questions
- In the linear programming context, what does each production process ray in Figure 1 represent?
- How is a production isoquant derived in a linear programming framework, and what shape does it have?
- Why might a firm with limited resources (constraints on L and K) not use the least-cost input combination found by tangency with the isocost curve?
- State the three key components of any linear programming model and explain the purpose of the non-negativity constraint.
- In the production planning example for products X and Y, what does the constraint "1Qβ + 1QΚ β€ 10" represent, and why is the optimal solution always found at a corner point of the feasible region?