π Lecture 42 β Samuelson Multiplier-Acceleration Interaction Model
π Overview: This lecture introduces Samuelson's Multiplier-Acceleration Interaction Model, a seminal application of second-order difference equations in economics. It explores how the interaction of the Keynesian multiplier and the acceleration principle can generate endogenous cyclical fluctuations in national income, without relying on external shocks.
ποΈ Topics Covered
The lecture is structured into three main topics: the framework of the Samuelson Multiplier-Acceleration Interaction Model, its solution using second-order difference equations, and numerical examples to determine dynamic stability. The framework defines consumption, induced investment, and exogenous government expenditure. The solution derives the particular integral and complementary function, analyzing cases for distinct real, repeated real, and complex roots. Numerical examples apply the characteristic root formula to assess time-path convergence or divergence based on parameter values.
π Lecture Summary
TOPIC 192: SAMUELSON MULTIPLIER-ACCELERATION INTERACTION MODEL: FRAMEWORK
The Samuelson Multiplier-Acceleration Interaction Model, attributed to Professor Paul A. Samuelson, is an application of second-order difference equations in economics. It seeks to explore the dynamic process of income determination when the acceleration principle operates alongside the Keynesian multiplier. The model shows that the mere interaction of the multiplier and accelerator can generate cyclical fluctuations endogenously.
The model presents a self-reinforcing cycle. An initial increase in autonomous investment (βIβ) works through the multiplier to increase income (βY). The increase in income, through investors' expectations regarding consumption (MPC), triggers the accelerator, leading to an increase in induced investment (βIα΅’). This induced investment again works through the multiplier to further increase income (βY), continuing the process in a boom or recovery. The process works identically in reverse for a recession or depression, with decreases cascading through the system.
National income (Yβ) has three component expenditure streams: consumption (Cβ), investment (Iβ), and government expenditure (Gβ).
Consumption (Cβ): Consumption is a function not of current income (Yβ) but of the income of the prior period (Yβββ). For simplicity, it is assumed to be strictly proportional to Yβββ:
π Definition β Cβ: Cβ = Ξ² Yβββ where (0 < Ξ² < 1) and Ξ² is actually the MPC (Marginal Propensity to Consume), with no intercept term.
Investment (Iβ): Induced investment (Iα΅’) is a function of the prevailing trend of consumer spending (Cβ). An increase in income (βY) works through the multiplier, and then through the MPC, it triggers the acceleration mechanism in the model. It is assumed that investment is induced based on consumption patterns.
The formula for induced investment is:
π Formula β Induced Investment: Iβ = a (Cβ β Cβββ) = a (βCβββ) Plain-English Meaning: Investment in period t is equal to the acceleration coefficient (a) multiplied by the change in consumption from the previous period. Economic Significance of a: The parameter a is the Acceleration Coefficient. It is considered a fixed ratio, where a = Iβ / (Cβ β Cβββ) = Iβ / βCβ, and a > 0.
Government Expenditure (Gβ): The third component is government expenditure (Gβ), which is considered exogenous (Gβ).
Therefore, the framework of the Samuelson Multiplier-Acceleration Interaction Model is composed of the following system of equations:
- Cβ = Ξ² Yβββ
- Iβ = a (Cβ β Cβββ)
- Gβ = Gβ
π‘ Why this matters: This framework lays the foundation for modeling how consumption and investment decisions based on past behavior can create complex, cyclical dynamics in the entire economy.
TOPIC 193: SAMUELSON MULTIPLIER-ACCELERATION INTERACTION MODEL: SOLUTION
Borrowing the framework, the solution involves combining all three components in the expenditure approach equation: Yβ = Cβ + Iβ + Gβ.
Substituting the consumption and investment equations into the national income identity: Yβ = Ξ²Yβββ + a(Cβ β Cβββ) + Gβ
By introducing a lag, we get Cβ = Ξ²Yβββ and Cβββ = Ξ²Yβββ. Substituting these into the income equation yields: Yβ = Ξ²Yβββ + a(Ξ²Yβββ β Ξ²Yβββ) + Gβ Yβ = Ξ²Yβββ + aΞ²Yβββ β aΞ²Yβββ + Gβ Yβ = (Ξ² + aΞ²)Yβββ β aΞ²Yβββ + Gβ Yβ β (Ξ² + aΞ²)Yβββ + aΞ²Yβββ = Gβ
Introducing a lead of 2 on both sides transforms this into a standard second-order difference equation: Yβββ β (Ξ² + aΞ²)Yβββ + aΞ²Yβ = Gβ
This is comparable with the standard second-order difference equation: yβββ + aβyβββ + aβyβ = c. Here: y = Y, aβ = β(Ξ² + aΞ²), aβ = aΞ², and c = Gβ.
Particular Integral (Yβ): The particular integral depends on aβ and aβ. Since the sum aβ + aβ = β(Ξ² + aΞ²) + aΞ² = βΞ² β β1, we can use the standard formula:
π Formula β Particular Integral: yβ = c / (1 + aβ + aβ) Substituting the values: Yβ = Gβ / (1 β (Ξ² + aΞ²) + aΞ²) Yβ = Gβ / (1 β Ξ² β aΞ² + aΞ²) Yβ = Gβ / (1 β Ξ²) This gives the intertemporal equilibrium of national income (Y*).
Complementary Solution (Yc): The complementary solution depends on the discriminant (aβΒ² β 4aβ).
Case β I: Distinct Real Roots (aβΒ² > 4aβ) Condition: {β(Ξ² + aΞ²)}Β² > 4(aΞ²) β (Ξ² + aΞ²)Β² > 4aΞ² β Ξ² > 4aΞ² / (Ξ² + aΞ²)Β²
Case β II: Repeated Real Roots (aβΒ² = 4aβ) Condition: {β(Ξ² + aΞ²)}Β² = 4(aΞ²) β (Ξ² + aΞ²)Β² = 4aΞ² β Ξ² = 4aΞ² / (Ξ² + aΞ²)Β²
Case β III: Complex Roots (aβΒ² < 4aβ) Condition: {β(Ξ² + aΞ²)}Β² < 4(aΞ²) β (Ξ² + aΞ²)Β² < 4aΞ² β Ξ² < 4aΞ² / (Ξ² + aΞ²)Β²
π‘ Why this matters: These conditions, expressed in terms of the acceleration coefficient (a) and the MPC (Ξ²), determine the qualitative nature of the time path (e.g., monotonic, oscillatory, or cyclical) without needing to compute the full solution.
TOPIC 194: SAMUELSON MULTIPLIER-ACCELERATION INTERACTION MODEL: NUMERICALS
Under the Samuelson Multiplier-Acceleration Interaction Model, the characteristic equation is similar to the standard form of a second-order difference equation.
The reduced form is: yβββ + aβyβββ + aβyβ = 0 Using trial solutions yβ = Abα΅, we get the characteristic equation: bΒ² + aβb + aβ = 0
The characteristic roots are:
π Formula β Characteristic Roots: bβ, bβ = (βaβ Β± β(aβΒ² β 4aβ)) / 2
Recalling aβ and aβ from the Samuelson model, with aβ = β(Ξ² + aΞ²) and aβ = aΞ²: bβ, bβ = (β{β(Ξ² + aΞ²)} Β± β({β(Ξ² + aΞ²)}Β² β 4(aΞ²))) / 2
π Formula β Samuelson Model-Specific Characteristic Roots: bβ, bβ = (Ξ²(1 + a) Β± β(Ξ²Β²(1 + a)Β² β 4aΞ²)) / 2
Numerical # 1: For the Samuelson Multiplier-Acceleration Interaction Model, if a = 3.5 and Ξ² = 0.8, find the characteristic roots and judge the dynamic stability of the time-path.
bβ, bβ = (0.8(1 + 3.5) Β± β(0.8Β²(1 + 3.5)Β² β 4(3.5)(0.8))) / 2 bβ, bβ = (3.6 Β± β(1.76)) / 2 bβ, bβ = (3.6 Β± 1.3266) / 2 bβ = (3.6 + 1.3266) / 2 = 2.46 bβ = (3.6 β 1.3266) / 2 = 1.14 Interpretation: Both |bβ| > 1 and |bβ| > 1, therefore, the interaction time-path shall be divergent and hence dynamically unstable.
Numerical # 2: For the Samuelson Multiplier-Acceleration Interaction Model, if a = 0.2 and Ξ² = 0.9, find the characteristic roots and judge the dynamic stability of the time-path.
bβ, bβ = (0.9(1 + 0.2) Β± β(0.9Β²(1 + 0.2)Β² β 4(0.2)(0.9))) / 2 bβ, bβ = (1.08 Β± β(0.4464)) / 2 bβ, bβ = (1.08 Β± 0.6681) / 2 bβ = (1.08 + 0.6681) / 2 = 0.87 bβ = (1.08 β 0.6681) / 2 = 0.21 Interpretation: Both |bβ| < 1 and |bβ| < 1, therefore, the interaction time-path shall be convergent and hence dynamically stable.
β Key Takeaways
The Samuelson Multiplier-Acceleration Interaction Model demonstrates that the dynamic interaction between the multiplier (driven by MPC) and the accelerator (driven by the investment-consumption relationship) can generate endogenous business cycles. The model is formalized as a second-order difference equation, Yβββ β (Ξ² + aΞ²)Yβββ + aΞ²Yβ = Gβ, with the intertemporal equilibrium income given by Y* = Gβ/(1βΞ²). The dynamic stability of the time path is determined by the characteristic roots; if the absolute values of both roots are less than one, the path converges to equilibrium (stable), and if greater than one, it diverges (unstable). The numerical examples confirm that higher values of the accelerator (a) and MPC (Ξ²) tend to produce divergence, while lower values lead to convergence.
π§ Quick Revision Questions
- What are the three main equations that form the framework of the Samuelson Multiplier-Acceleration Interaction Model?
- How is the induced investment function (Iβ) expressed in terms of consumption (Cβ) and the acceleration coefficient (a)?
- What is the particular integral (Yβ) in this model, and what does it represent economically?
- What condition on the absolute values of the characteristic roots (bβ and bβ) indicates a dynamically stable time path?
- In Numerical #1 (a=3.5, Ξ²=0.8), what were the computed characteristic roots, and what was the conclusion about the time-path's stability?
π Lecture 43 β SECOND ORDER DIFFERENCE EQUATION WITH VARIABLE TERM
π Overview: This lecture extends second-order difference equations from constant-term to variable-term cases, introducing the Method of Undetermined Coefficients. It covers both exponential and power function variable terms, then extends analysis to higher-order difference equations and introduces Schur Theorem for testing dynamic stability without solving characteristic equations.
ποΈ Topics Covered
The lecture begins with second-order difference equations with variable terms, specifically exponential functions (cβα΅) and power functions (ctβΏ). It then extends to higher-order linear difference equations (n>2), covering standard forms, complementary functions with repeated and distinct real roots, and hybrid solutions. Finally, it introduces dynamic stability analysis through Schur Theorem determinants and provides numerical examples for convergence testing.
π Lecture Summary
SECOND-ORDER DIFFERENCE EQUATION WITH VARIABLE TERM
When the right-hand side of a second-order difference equation contains a variable term {f(t)} instead of a constant c, we use the Method of Undetermined Coefficients. This method requires that the variable term and its successive differences together contain only distinct types of expressions.
π Definition β Method of Undetermined Coefficients: A technique for finding the particular integral of a difference equation with a variable term by assuming a trial solution of the same functional form as the variable term, then determining coefficients by substitution.
For exponential terms like 7α΅:
- yβ = 7α΅
- Ξyβ = 6(7α΅)
- ΞΒ²yβ = 36(7α΅)
TOPIC 196: SECOND-ORDER DIFFERENCE EQUATION WITH VARIABLE TERM (CTN)
When the variable term is a power function (ctβΏ) rather than an exponential, the method adapts accordingly. For quadratic terms tΒ²:
- yβ = tΒ²
- Ξyβ = 2t + 1
- ΞΒ²yβ = 2
π Pattern recognition: For exponential terms, assume yβ = B(mα΅); for power functions of degree n, assume a polynomial of degree n.
π Example 1 (Exponential): Find particular integral of yβββ + yβββ β 3yβ = 7α΅
- Assume yβ = B(7α΅)
- Then yβββ = B(7α΅βΊΒΉ) and yβββ = B(7α΅βΊΒ²)
- Substituting: B(7α΅βΊΒ²) + B(7α΅βΊΒΉ) β 3B(7α΅) = 7α΅
- B(49 + 7 β 3)(7α΅) = 7α΅
- 58B(7α΅) = 7α΅ β B = 1/58
- Therefore, yβ = (1/58)(7α΅)
π Example 2 (Quadratic): Find particular integral of yβββ + 5yβββ + 2yβ = tΒ²
- Assume yβ = BβtΒ² + Bβt + Bβ
- Generate yβββ and yβββ, substitute, and compare coefficients of tΒ², t, and constants
- Result: Bβ = 1/8, Bβ = β7/32, Bβ = 13/256
- Therefore, yβ = (1/8)tΒ² β (7/32)t + 13/256
π‘ Why this matters: The complementary function (yβ) remains unchanged as it depends on bΒ² + aβb + aβ = 0; only the particular integral changes with the variable term.
TOPIC 197: HIGHER-ORDER LINEAR DIFFERENCE EQUATION
A linear, non-homogeneous higher-order difference equation with n constant coefficients has the standard form: aβyβββ + aβyβββββββ + aβyβββββββ + ... + aβββyβββ + aβyβ = c
Key properties:
- Order: nα΅Κ° (due to nα΅Κ° difference)
- Degree: Linear/First degree
- An nα΅Κ° order implies n number of characteristic roots
- A hybrid complementary solution can exist combining repeated real roots and complex roots
π Example (Third-order): Find general solution of yβββ β (7/8)yβββ + (1/8)yβββ + (1/32)yβ = 9
- Test Case I: aβ + aβ + aβ = β7/8 + 1/8 + 1/32 = β23/32 β β1
- Particular integral: yβ = c/(1 + aβ + aβ + aβ) = 9/(1 β 23/32) = 32
- Characteristic equation: bΒ³ β (7/8)bΒ² + (1/8)b + 1/32 = 0
- Roots: bβ = 1/2, bβ = 1/2 (repeated), bβ = β1/8
- Complementary function: yβ = Aβ(1/2)α΅ + Aβt(1/2)α΅ + Aβ(β1/8)α΅
- General solution: yβ = Aβ(1/2)α΅ + Aβt(1/2)α΅ + Aβ(β1/8)α΅ + 32
π Definition β Dynamic Stability: For convergence, all characteristic roots must be less than 1 in absolute value. Here |bβ| = |bβ| = 1/2 < 1 and |bβ| = 1/8 < 1, confirming convergence.
TOPIC 198: CONVERGENCE AND THE SCHUR THEOREM
The Schur Theorem provides a qualitative method for testing dynamic stability without solving for all n roots. Determinants are developed using coefficients in the difference equation.
π Definition β Schur Theorem: For an nα΅Κ° order difference equation to be dynamically stable (convergent), all determinants ββ, ββ, ..., ββ must be positive.
For second-order difference equation: ββ and ββ must both be > 0 for convergence.
π Determinant formulas for second-order:
- ββ = |aβ aβ; aβ aβ|
- ββ = |aβ 0 aβ aβ; aβ aβ 0 aβ; aβ 0 aβ aβ; aβ aβ 0 aβ|
π Example 1 (Divergent): yβββ + 3yβββ + 2yβ = 12
- aβ = 1, aβ = 3, aβ = 2
- ββ = |1 2; 2 1| = 1 β 4 = β3 < 0
- ββ = 0
- Since ββ < 0 and ββ = 0, the system is DIVERGENT
- Verification: roots are β1 and β2, both with absolute value β₯ 1
π Example 2 (Convergent): yβββ + (1/6)yβββ β (1/6)yβ = 2
- aβ = 1, aβ = 1/6, aβ = β1/6
- ββ = |1 β1/6; β1/6 1| = 1 β 1/36 = 35/36 > 0
- ββ = 49/54 > 0
- Since both determinants are positive, the system is CONVERGENT
β οΈ Limitations: (1) Dynamic stability is investigated but characteristic roots remain unknown; (2) For n > 2, determinants become huge, contrary to expectation of ease.
β Key Takeaways
The Method of Undetermined Coefficients is the essential technique for solving second-order difference equations with variable termsβexponential terms require assuming B(mα΅) while power functions require polynomials of matching degree. For higher-order equations (n > 2), the solution structure extends from second-order: particular integrals follow the same case formulas, complementary functions involve n characteristic roots, and hybrid solutions combine repeated real roots with distinct or complex roots. The Schur Theorem offers a powerful qualitative shortcut for testing dynamic stability without solving for all n rootsβif all determinants are positive, the time path convergesβthough this method becomes impractical for very high orders due to determinant size.
π§ Quick Revision Questions
-
What is the precondition for applying the Method of Undetermined Coefficients to a variable-term difference equation?
-
For the equation yβββ + 2yβββ β 5yβ = 3α΅, what trial solution form should be assumed for the particular integral?
-
In Schur Theorem testing for a second-order difference equation, what must be true about ββ and ββ for the system to be convergent?
-
How does the complementary function (yβ) for a third-order difference equation differ when there are two repeated real roots and one distinct real root?
-
For the third-order equation yβββ β (7/8)yβββ + (1/8)yβββ + (1/32)yβ = 9, what is the value of the particular integral?
π Lecture 44 β LINEAR AND NON-LINEAR PROGRAMMING
π Overview: This lecture introduces linear and nonlinear programming techniques for optimization in economics. It covers the three-step process of mathematizing, graphing, and selecting optimal solutions for both profit maximization and cost minimization problems, and extends to nonlinear problems using Kuhn-Tucker conditions.
ποΈ Topics Covered
The lecture covers four main topics: linear programming for profit maximization using a glassware making example, linear programming for cost minimization using an insurance company employment problem, nonlinear programming for utility maximization using Kuhn-Tucker conditions, and nonlinear programming for cost minimization. It demonstrates the full process from problem formulation to solution selection, including the need for integer programming when dealing with whole units like workers.
π Lecture Summary
TOPIC 199: LINEAR PROGRAMMING: GLASSWARE MAKING
Linear programming addresses real-life problems of allocating finite resources such as labor, machine time, or raw materials to maximize profits or minimize costs. The process involves mathematizing given information into Decision Variables, an Objective Function, and (multiple) Constraints. It only deals with linear objective and constraint functions and is generally done in three steps.
Numerical Example: A glassware firm produces bowls (x) and plates (y). Profit per bowl is $150, per plate is $100. Resources are limited: glassblowing time (70 hours/week, bowls take 2 hours, plates take 1 hour), annealing time (130 hours/week, bowls take 4 hours, plates take 1 hour), and silica sand (45 kg, each product uses 1 kg). Production cannot be negative.
π Definition β Decision Variables: Quantities to be determined in the optimization problem (x = number of bowls, y = number of plates). π Formula: Objective Function β Ο = 150x + 100y (to be maximized) π Formula: Constraints: 2x + y β€ 70 (glassblowing); 4x + y β€ 130 (annealing); x + y β€ 45 (silica sand); x β₯ 0; y β₯ 0 π Example: After plotting constraints and finding the feasible region, corner points are evaluated: (0,0) profit=$0; (0,45) profit=$4,500; (25,20) profit=$5,750; (30,10) profit=$5,500; (32.5,0) profit=$4,875. The optimal solution is (x, y) = (25, 20)** producing maximum profit of $5,750.
TOPIC 200: LINEAR PROGRAMMING: EMPLOYMENT COST MINIMIZATION
Numerical Example: An insurance company employs full-time staff (x) who work 40 hours/week for $800/week, and part-time staff (y) who work 20 hours/week for $320/week. Part-time staff must not exceed one-third of full-time staff. Total work required is at least 900 worker-hours per week.
π Definition β Objective Function: C = 800x + 320y (to be minimized) π Formula: Constraints: 40x + 20y β₯ 900 (worker hours); y β€ (1/3)x (proportion); x β₯ 0; y β₯ 0 π Example: Corner points evaluated: (0,0) cost=$0; (19.3, 6.4) cost=$17,485.7; (22.5,0) cost=$18,000. The optimal solution is (x, y) = (19.3, 6.4)** with minimum cost of $17,485.7.
Integer Programming: Since workers cannot be fractions, whole numbers are tested. (20,5) gives cost=$17,600, which is the minimum integer solution, though higher than the fractional optimum.
TOPIC 201: NONLINEAR PROGRAMMING: UTILITY MAXIMIZATION
For nonlinear objective and constraint functions, Nonlinear Programming is used. It uses Kuhn-Tucker conditions (1951), which are similar to first-order conditions.
π Definition β Kuhn-Tucker Conditions for Maximization: For j variables: βZ/βxj β€ 0, xj β₯ 0, xj(βZ/βxj) = 0 (complementary slackness). For i multipliers: βZ/βΞ»i β₯ 0, Ξ»i β₯ 0, Ξ»i(βZ/βΞ»i) = 0.
Numerical Example: Maximize U = xy subject to x + y β€ 100, x β€ 40, and x,y β₯ 0.
π Definition β Lagrangian Function: Z(x,y,Ξ»β,Ξ»β) = xy + Ξ»β(100 - x - y) + Ξ»β(40 - x) π Example: Trial solution Case I with Ξ»β,Ξ»β > 0 gives x=40, y=60. Using complementary slackness: since x>0, βZ/βx=0 gives y-Ξ»β-Ξ»β=0; since y>0, βZ/βy=0 gives x-Ξ»β=0 β Ξ»β=40, Ξ»β=20. Verification: U = 40Γ60 = 2400; x+y=100β€100; x=40β€40. Solution: (x, y) = (40, 60)** with maximum utility 2,400.
TOPIC 202: NONLINEAR PROGRAMMING: COST MINIMIZATION
π Definition β Kuhn-Tucker Conditions for Minimization: For j variables: βZ/βxj β₯ 0, xj β₯ 0, xj(βZ/βxj) = 0. For i multipliers: βZ/βΞ»i β€ 0, Ξ»i β₯ 0, Ξ»i(βZ/βΞ»i) = 0.
Numerical Example: Minimize C = (xβ-4)Β² + (xβ-4)Β² subject to 2xβ+3xβ β₯ 6, -3xβ-2xβ β₯ -12, and xβ,xβ β₯ 0.
π Formula: Lagrangian Function: Z = (xβ-4)Β² + (xβ-4)Β² + Ξ»β(6-2xβ-3xβ) + Ξ»β(-12+3xβ+2xβ) π Example: Trial Solution Case I (Ξ»β,Ξ»β>0) gives xβ=-6/5, violating nonnegativity. Trial Solution Case II (xβ,xβ>0) with Ξ»β=0 gives xβ=28/13, xβ=36/13, Ξ»β=16/13. Verification: C = 68/13; 2(28/13)+3(36/13)=12.6β₯6; -3(28/13)-2(36/13)=-12β₯-12. Solution: (xβ, xβ) = (28/13, 36/13)** with minimum cost 68/13.
π‘ Why this matters: Kuhn-Tucker conditions provide the necessary conditions for optimality in constrained nonlinear optimization problems, extending the method of Lagrange multipliers to handle inequality constraints through complementary slackness.
β Key Takeaways
Linear programming solves allocation problems with linear objective and constraints through three steps: mathematizing, graphing, and selecting corner points of the feasible region. When fractional solutions don't make sense for whole units, integer programming tests nearby whole numbers. For nonlinear problems, Kuhn-Tucker conditions replace standard first-order conditions and require testing trial cases for variables and multipliers using complementary slackness. The key difference between maximization and minimization Kuhn-Tucker conditions is the direction of inequalities for partial derivatives of the Lagrangian function.
π§ Quick Revision Questions
- What are the three steps in solving a linear programming problem?
- In the glassware maximization problem, why was (25,20) chosen over (30,10) even though both satisfy constraints?
- What is the purpose of integer programming, and how does it relate to linear programming?
- How do Kuhn-Tucker conditions for maximization differ from those for minimization in terms of inequality direction?
- In the utility maximization example, what does complementary slackness mean when x>0?