π Lecture 32 β PHILLIPS CURVE: THE PHILLIPS RELATION
π Overview: This lecture explores the Phillips Curve, a fundamental macroeconomic model analyzing the interaction between inflation and unemployment. It progresses from the original relation through the expectations-augmented version, culminating in a dynamic model that uses second-order differential equations to derive time paths for expected inflation, actual inflation, and unemployment, revealing the long-run neutrality of monetary policy.
ποΈ Topics Covered
The lecture begins with the original and adapted Phillips relations, linking wage growth and inflation to unemployment. It then introduces the expectations-augmented Phillips relation incorporating inflation expectations and adaptive expectations formation. A feedback mechanism from inflation to unemployment via monetary policy is established. The core of the lecture derives the time path of expected inflation using a second-order differential equation, followed by numeric examples and derivations of time paths for actual inflation and unemployment, culminating in the concept of the natural rate and the long-run Phillips curve.
π Lecture Summary
TOPIC 140: PHILLIPS CURVE: THE PHILLIPS RELATION
The Phillips relation models the negative empirical relationship between money wage growth (ν) and the unemployment rate (νΌ). The Original Phillips Relation is ν = ν(νΌ) where νβ²(νΌ) < 0. An Adapted Phillips Relation substitutes the inflation rate (ν) for wage growth, based on the idea that higher wages increase production costs, leading to cost-push inflation: ν = ν β ν», where ν» is the growth rate of productivity. Assuming a linear form for the original relation (ν = ν β ν°νΌ), the adapted relation becomes ν = ν β ν°νΌ β ν», meaning both higher productivity and higher unemployment negatively affect the inflation rate.
π Definition β Original Phillips Relation: An inverse empirical relationship between the rate of growth of money wages and the rate of unemployment.
TOPIC 141: PHILLIPS CURVE: EXPECTATIONS AUGMENTED PHILLIPS RELATION
The expectations-augmented Phillips relation incorporates expected inflation (ν ) into wage demands: ν = ν(νΌ) + νν , where 0 < ν β€ 1. The adapted version becomes ν = ν β ν» β ν°νΌ + νν . Adaptive Expectations form the basis for how expectations evolve; here, the change in expected inflation is proportional to the discrepancy between actual and expected inflation: ν ν /ν ν = ν(ν β ν ), where 0 < ν β€ 1. If actual inflation exceeds expected (ν > ν ), expectations are revised upward; if ν = ν , no revision occurs; if ν < ν , downward revision is needed.
π Definition β Adaptive Expectations Hypthesis: The hypothesis that forecasts of future values are made using only past values of the variable, leading to a gradual adjustment based on recent forecast errors.
TOPIC 142: PHILLIPS CURVE: FEEDBACK FROM INFLATION TO UNEMPLOYMENT
A complete two-equation system models the interaction between inflation and unemployment. The first equation is the expectations-augmented Phillips relation (ν = ν β ν» β ν°νΌ + νν ). The second is the adaptive expectations hypothesis (ν ν /ν ν = ν(ν β ν )). The direction from unemployment to inflation (νΌ β ν) is the supply-side effect. The feedback effect from inflation to unemployment (νΌ β ν) operates through the demand side via government fiscal and monetary policy. Using a monetary policy channel, the rate of change of unemployment is inversely related to the growth rate of real money: ν νΌ/ν ν = βν(ν β ν), where ν is the growth rate of nominal money and ν > 0. Here, (ν β ν) is the rate of growth of real money.
π Definition β Feedback Effect (Monetary Policy): An inverse relationship between the rate of change of unemployment and the growth rate of real money, implying that higher real money growth reduces unemployment.
TOPIC 143: PHILLIPS CURVE: TIME PATH OF EXPECTED RATE OF INFLATION
The three-equation system (ν = ν β ν» β ν°νΌ + νν , ν ν /ν ν = ν(ν β ν ), and ν νΌ/ν ν = βν(ν β ν)) is reduced to a single second-order differential equation for the expected rate of inflation (ν ). The particular solution is ν _ν = ν/ν_ν, which simplifies to ν _ν = ν, meaning the equilibrium expected inflation equals the growth rate of nominal money. The nature of the roots for the complementary function depends on the determinant ν_ν^ν β νν_ν. Since ν_ν (ν°ν β ν(ν β 1)) and ν_ν (νν°ν) are both positive, but the relative magnitudes and signs of the parameters mean all three root cases (distinct real, repeated real, complex) are possible. Crucially, in all cases, the real part of the roots (or the roots themselves) are negative, making the time path dynamically stable. This means ν (ν) converges to ν over time.
π‘ Why this matters: The key result is that the long-run equilibrium expected inflation rate is determined only by the growth rate of nominal money, not by unemployment or other real factors.
TOPIC 144: PHILLIPS CURVE: TIME PATH OF EXPECTED RATE OF INFLATION - NUMERICAL
Using specific parameter values (ν° = 3, ν = 1, ν = 3/4, ν = 1/2), we find ν_ν = 3/2, ν_ν = 9/8, and ν = (9/8)ν. Since ν_ν^ν (9/4) is less than νν_ν (9/2), the roots are complex. The roots are ν_ν, ν_ν = -3/4 Β± (3/4)ν, giving a damping coefficient ν = -3/4 and a frequency ν = 3/4. The general solution for expected inflation is ν (ν) = ν^(-νν/ν) {ν¨_ν νν¨ν¬(νν/ν) + ν¨_ν ν¬ν’ν§(νν/ν)} + ν. This is a dynamically stable, fluctuating path converging to ν.
π Formula: General solution for a second-order differential equation with complex roots ν Β± νν: ν(ν) = ν^(νν) (νͺ_ν νν¨ν¬(νν) + νͺ_ν ν¬ν’ν§(νν)) + ν_ν.
TOPIC 145: PHILLIPS CURVE: TIME PATH OF INFLATION - NUMERICAL
The time path of actual inflation ν(ν) is derived from the adaptive expectations equation: ν = (ν ν /ν ν)(ν/ν) + ν . Using the derivative of ν (ν) and substituting into this formula, the general solution for actual inflation is ν(ν) = ν^(-νν/ν) {ν¨_ν νν¨ν¬(νν/ν) β ν¨_ν ν¬ν’ν§(νν/ν)} + ν. Like expected inflation, this is a dynamically stable, fluctuating path that converges to the equilibrium value of ν.
TOPIC 146: PHILLIPS CURVE: TIME PATH OF UNEMPLOYMENT β NUMERICAL
The time path of unemployment νΌ(ν) is derived by rewriting the expectations-augmented Phillips relation for νΌ and substituting the solutions for ν (ν) and ν(ν). The general solution is νΌ(ν) = (ν/ν)ν^(-νν/ν) [(ν¨_ν β ν¨_ν)νν¨ν¬(νν/ν) + (ν¨_ν β ν¨_ν)ν¬ν’ν§(νν/ν)] + ν/νν. This is a dynamically stable, fluctuating path that converges to the equilibrium value of ν/νν, regardless of the rate of money growth (ν).
π Definition β Natural Rate of Unemployment (νΌ_ν΅νΉ): The equilibrium level of unemployment to which the economy tends in the long run, independent of the rate of inflation. In this example, it is ν/νν. π‘ Why this matters: This demonstrates the long-run vertical Phillips curve; in the long run, there is no trade-off between inflation and unemployment. Inflation is determined by money growth, while unemployment returns to its natural rate.
β Key Takeaways
The lecture evolves the Phillips Curve from a simple negative relationship between wage growth and unemployment to a dynamic model including inflation expectations and feedback effects. The crucial insight is that while expectations-augmented Phillips curve allows a short-run trade-off, the long-run equilibrium is characterized by the natural rate of unemployment, which is independent of the inflation rate. The time paths of all variables converge to their equilibrium values, demonstrating dynamic stability. The numerical example confirms that the equilibrium expected and actual inflation rates are equal to the rate of money growth (ν), while equilibrium unemployment is a constant natural rate (ν/νν). The model's solution uses second-order differential equations, showing that macroeconomic dynamics can exhibit cyclical convergence.
π§ Quick Revision Questions
- What is the key difference between the Original and the Expectations-Augmented Phillips Relation in terms of what determines wage growth?
- In the adaptive expectations framework, what condition causes expected inflation to be revised upward?
- How does the feedback effect from inflation to unemployment operate through monetary policy? Formulate the relationship.
- In the model, what does the equilibrium value of expected inflation (ν _ν) equal, and what does it imply about the long-run neutrality of money?
- What is the natural rate of unemployment, and how does the long-run Phillips curve relate to it?
π Lecture 33 β Higher Order Differential Equations
π Overview: This lecture extends the study of second-order differential equations to higher-order ones, focusing on finding particular integrals (ν¦_ν) when variable terms are present. It introduces the Method of Undetermined Coefficients and its modifications, then generalizes to nth-order equations and the complementary function, culminating in Routh's Theorem for assessing dynamic stability without calculating characteristic roots.
ποΈ Topics Covered
The lecture covers five main topics: differential equations with variable terms, a modification to the method when parts of the LHS are missing, finding particular integrals for higher-order equations, determining complementary functions for higher-order equations, and using Routh's Theorem to check convergence and stability of the time path without solving for roots.
π Lecture Summary
Topic 147: Differential Equation with Variable Term
This section introduces second-order differential equations with constant coefficients but a variable term f(t) instead of a constant b. The solution requires the Method of Undetermined Coefficients. This method works only if the variable term f(t) and its successive derivatives contain a finite and distinct set of expression types.
π Definition β Method of Undetermined Coefficients: A technique for finding the particular integral (ν¦_ν) of a differential equation by assuming a trial solution with unknown coefficients, then solving for those coefficients by equating terms of the same type on both sides of the equation.
π Key Concept: For a variable term f(t), the particular integral (ν¦_ν) changes, but the complementary function (ν¦_ν) remains unchanged, as it depends only on the roots of the characteristic equation rΒ² + aβr + aβ = 0.
π Example: Find the Particular Integral of: y''(t) + 5y'(t) + 3y(t) = 6tΒ² β t β 1. Here, aβ=5, aβ=3, and f(t)=6tΒ²βtβ1. Since the RHS is quadratic in t, we assume a trial general quadratic function: y(t) = BβtΒ² + Bβt + Bβ. Then y'(t) = 2Bβt + Bβ, and y''(t) = 2Bβ. Substituting into LHS: 2Bβ + 5(2Bβt + Bβ) + 3(BβtΒ² + Bβt + Bβ) = 6tΒ² β t β 1. Simplifying: (3Bβ)tΒ² + (10Bβ + 3Bβ)t + (2Bβ + 5Bβ + 3Bβ) = 6tΒ² + (β1)t + (β1). Comparing coefficients: 3Bβ=6 gives Bβ=2; 10(2)+3Bβ=β1 gives Bβ=β7; 2(2)+5(β7)+3Bβ=β1 gives Bβ=10. Therefore, the particular integral is: ν¦_ν = 2tΒ² β 7t + 10.
π‘ Why this matters: The Method of Undetermined Coefficients allows us to find a specific solution that satisfies the non-homogeneous equation.
Topic 148: Differential Equation with Variable Term β A Modification
When the LHS of a constant-coefficient variable-term differential equation is missing certain terms (e.g., the y(t) term is absent), the standard Method of Undetermined Coefficients fails because a direct polynomial trial solution leads to an illogical comparison. The solution requires a Modification: multiply the trial function by t.
π Example: Find the Particular Integral of: y''(t) + 5y'(t) = 6tΒ² β t β 1. Here, aβ=5, aβ=0, and f(t)=6tΒ²βtβ1. If we use the trial y(t)=BβtΒ²+Bβt+Bβ, substituting yields: LHS: (0)tΒ² + (10Bβ)t + (2Bβ+5Bβ) = 6tΒ² β t β 1. Comparing coefficients gives a contradiction: 0=6 for the tΒ² term. This is illogical, so the trial fails. Modification: Multiply the trial function by t: y(t) = t Γ (BβtΒ²+Bβt+Bβ) = BβtΒ³+BβtΒ²+Bβt. Now y'(t)=3BβtΒ²+2Bβt+Bβ, and y''(t)=6Bβt+2Bβ. Substituting into LHS: (6Bβt+2Bβ) + 5(3BβtΒ²+2Bβt+Bβ) = 15BβtΒ² + (6Bβ+10Bβ)t + (2Bβ+5Bβ). Equating to RHS: 15BβtΒ² + (6Bβ+10Bβ)t + (2Bβ+5Bβ) = 6tΒ² β t β 1. Comparing: 15Bβ=6 gives Bβ=2/5; 6(2/5)+10Bβ=β1 gives Bβ=β17/50; 2(β17/50)+5Bβ=β1 gives Bβ=β8/125. Therefore, the modified particular integral is: ν¦_ν = (2/5)tΒ² β (17/50)t β (8/125).
π‘ Why this matters: When the LHS lacks a y(t) term (aβ=0), the tΒ² term vanishes in the LHS. Multiplying by t adds one degree to every term, restoring the missing term type.
Topic 149: Higher-Order Differential Equation: Particular Integral
This section extends the concept of particular integrals from second-order to nth-order (higher-order) differential equations. The approach is analogous to a polynomial, where the order of derivative replaces the degree of polynomial. The solution for ν¦_ν depends on the values of the coefficients aβ, aβββ, etc.
For a general nth-order equation: aβyβ½βΏβΎ(t) + aβyβ½βΏβ»ΒΉβΎ(t) + ... + aβββy'(t) + aβy = b.
- If aβ β 0, then: ν¦_ν = b / aβ
- If aβ = 0 but aβββ β 0, then: ν¦_ν = (b / aβββ) Β· t
- If aβ = 0 and aβββ = 0, use trial solution y = ktΒ².
π Example 1: Find the Particular Solution of: yβ½β΄βΎ(t) + 6yβ΄(t) + 14yβ³(t) + 16y'(t) + 8y = 24. Here aβ=6, aβ=14, aβββ=aβ=16, aβ=aβ=8β 0. Since aββ 0, ν¦_ν = b/aβ = 24/8 = 3. This is a static equilibrium (constant value).
π Example 2: Find the Particular Solution of: yβ΄(t) + yβ³(t) + 3y'(t) = 1. Here aβ=1, aβββ=aβ=3, aβ=aβ=0. Since aβ=0 but aββββ 0, ν¦_ν = (b/aβββ)Β·t = (1/3)Β·t. This is a moving equilibrium (contains t).
π Example 3: Find the Particular Solution of: yβ΄(t) + 3yβ³(t) = 1/3. Here aβ=3, aβββ=aβ=0, aβ=aβ=0. Use trial y=ktΒ². Then y'(t)=2kt, yβ³(t)=2k, yβ΄(t)=0. Substituting: 0 + 3(2k) = 1/3 β 6k=1/3 β k=1/18. Therefore, ν¦_ν = (1/18)tΒ². This is also a moving equilibrium.
Topic 150: Higher-Order Differential Equation: Complementary Function
For higher-order differential equations, the complementary function (ν¦_ν) builds upon the methods for second-order equations. The characteristic equation for an nth-order equation extends naturally: for 4th order, it is rβ΄ + aβrΒ³ + aβrΒ² + aβr + aβ = 0.
π Example: Find the Complementary Function of: yβ½β΄βΎ(t) + 6yβ΄(t) + 14yβ³(t) + 16y'(t) + 8y = 24. The characteristic equation is rβ΄ + 6rΒ³ + 14rΒ² + 16r + 8 = 0. Factorizing gives: (r+2)(rβ2)(rΒ²+2r+2)=0. The roots are: rβ=β2, rβ=2 (distinct real roots), and rβ, rβ = [β2 Β± β(4β8)]/2 = β1 Β± i (complex conjugate roots). The complementary function is therefore hybrid: ν¦_ν = Aβeβ»Β²α΅ + AβeΒ²α΅ + eβ»α΅(Aβcos t + Aβsin t).
π‘ Why this matters: Complementary functions for higher-order equations can be a hybrid combination of distinct real roots, repeated real roots, and complex roots, all combined into one solution.
Topic 151: Convergence and the Routh Theorem
As the order of the differential equation increases, calculating roots becomes complex. Routh's Theorem provides a qualitative method to judge dynamic stability without solving for the roots. Its limitation is that it does not allow plotting the time path.
The theorem is based on the characteristic equation of the higher-order DE: aβrβΏ + aβrβΏβ»ΒΉ + ... + aβββr + aβ = 0. All roots will be negative (indicating dynamic stability) only if the first n of the following determinants are positive: |aβ|, |aβ aβ / aβ aβ|, |aβ aβ aβ / aβ aβ aβ / 0 aβ aβ|, |aβ aβ aβ aβ / aβ aβ aβ aβ / 0 aβ aβ aβ / 0 aβ aβ aβ|, ...
Pattern: 1) Number of determinants = order of DE. 2) A V-like shape forms along the principal diagonal. 3) Elements fill with ascending order vertically from bottom to top. 4) Zeros are used where elements are below aβ. Subscripts can exceed the order of the DE.
π Example 1: Apply Routh Theorem to: yβ½β΄βΎ(t) + 6yβ΄(t) + 14yβ³(t) + 16y'(t) + 8y = 24. Here aβ=1, aβ=6, aβ=14, aβ=16, aβ=8. Evaluate 4 determinants: 1st: |aβ| = |6| = 6 > 0. 2nd: |6 16 / 1 14| = (6)(14)β(1)(16) = 84β16 = 68 > 0. 3rd: |6 16 0 / 1 14 8 / 0 6 16| = 6(224β48) β 16(16β0) + 0 = 6(176) β 256 = 1056β256 = 800 > 0. 4th: |6 16 0 0 / 1 14 8 0 / 0 6 16 0 / 0 1 14 8| = 6(14168 + 800 + 0014 β 0160 β 8146 β 800) = 6400 > 0. Since all four determinants are positive, the time path is dynamically stable.
π Example 2: Apply Routh Theorem to: yβ΄(t) β 2yβ³(t) β y'(t) + 2y = 4. Here aβ=1, aβ=β2, aβ=β1, aβ=2. Evaluate 3 determinants: 1st: |aβ| = |β2| = β2 < 0. Since the first determinant is negative, the time path is dynamically unstable. (The 2nd and 3rd determinants also turn out to be non-positive, confirming this conclusion.)
π‘ Why this matters: Routh's Theorem allows for a quick check of stability without the onerous task of finding roots, especially useful for higher-order systems where root calculation is complex.
β Key Takeaways
The Method of Undetermined Coefficients provides a systematic way to find particular integrals for variable-term DEs, but when the LHS lacks the y(t) term, the trial function must be multiplied by t. For higher-order DEs, the particular integral follows simple rules based on which coefficients vanish (b/aβ, then bΒ·t/aβββ, then ktΒ²). The complementary function for higher-order equations can be a hybrid of distinct, repeated, and complex roots. Routh's Theorem offers a powerful shortcutβif any of its determinants for the characteristic equation's coefficients is negative, the system is dynamically unstable without needing to find the actual roots.
π§ Quick Revision Questions
- What is the precondition for applying the Method of Undetermined Coefficients to a variable-term differential equation?
- Why does the standard trial function fail when aβ=0, and how is this corrected?
- For a 5th-order differential equation with aβ β 0 and bβ 0, what is the particular integral?
- How is the complementary function for a higher-order DE constructed when the roots include a pair of complex conjugates and a repeated real root?
- According to Routh's Theorem, what must be true about the sequence of determinants for a time path to be dynamically stable?
π Lecture 34 β First Order Difference Equations
π Overview: This lecture introduces discrete time analysis, where time is treated as a sequence of distinct periods rather than a continuous flow. It explains first-order difference equations, their structure, and two solution methods: the iterative method and the general formula method. Understanding these equations is essential for modeling economic variables that change at specific intervals, such as seasonal prices or annual GDP.
ποΈ Topics Covered
The lecture begins by comparing continuous and discrete time analysis, introducing key terminology like lags and leads. It then defines the standard form of a first-order difference equation. Two solution methods are presented: the iterative (crude) method and the general (formula) method. The general method is further divided into two cases: when the coefficient a β -1 and when a = -1, each with its own formula for the particular integral and the complete solution.
π Lecture Summary
Topic 152: Discrete Time: First Order Difference Equation
This section introduces the concept of discrete time, where the variable ( t ) takes only integer values. This contrasts with continuous time analysis, which treats time as a continuous variable.
π Definition β Discrete Time: Time is treated as a non-continuous variable consisting of separate, distinct periods or intervals. Successive gaps between time periods are always equal ((\Delta t = 1)).
Key differences between continuous and discrete analysis:
- Nature of Values: Continuous allows decimals and integers; Discrete allows only integers.
- Analogy: Continuous is Point Analysis; Discrete is Period Analysis.
- Tool: Continuous uses derivatives; Discrete uses differences, lags, and leads.
- Equation Type: Continuous yields Differential Equations; Discrete yields Difference Equations.
π‘ Why this matters: A period does not have to be a calendar period; it is an analytical length of time before the variable ( y ) undergoes a change. This flexibility makes discrete analysis useful for modeling many economic phenomena.
The first difference of ( y ) is defined by the difference operator: π Formula: (\Delta y_t \equiv y_{t+1} - y_t)
Symbology:
- ( y_t ) = value of ( y ) in current time period ( t ).
- ( y_{t+1} ) = value of ( y ) in the next time period (a lead variable).
- ( y_{t-1} ) = value of ( y ) in the previous time period (a lag variable).
The standard form of a first-order difference equation is: π Formula: ( y_{t+1} + a y_t = c ) Here, ( a ) and ( c ) are constants.
Topic 153: Solving Difference Equation Using Iterative Method
The iterative method is a crude but direct way to find a solution by calculating ( y ) for successive periods starting from an initial condition. It involves repeatedly substituting the value from the previous step into the equation.
π Example 1: Solve ( y_{t+1} = y_t + 1 ) by iteration, given ( y_0 = 10 ).
- For ( t=0 ): ( y_1 = y_0 + 1 = 10 + 1 )
- For ( t=1 ): ( y_2 = y_1 + 1 = (10+1) + 1 = 10 + 2 )
- For ( t=2 ): ( y_3 = y_2 + 1 = (10+2) + 1 = 10 + 3 )
- Generalizing: ( y_t = y_0 + t = 10 + t )
π Example 2: Solve ( y_{t+1} = a y_t ) by iteration, given ( y_0 = \alpha ).
- For ( t=0 ): ( y_1 = a y_0 )
- For ( t=1 ): ( y_2 = a y_1 = a(a y_0) = a^2 y_0 )
- For ( t=2 ): ( y_3 = a y_2 = a (a^2 y_0) = a^3 y_0 )
- Generalizing: ( y_t = a^t y_0 = a^t \alpha )
Topic 154: Solving Difference Equation Using General Method: Aβ -1
This section develops a general formula to solve any first-order difference equation when ( a \neq -1 ). The general solution has two components: ( y_t = y_c + y_p )
- ( y_c ) = Complementary Function (the time-path deviation from equilibrium).
- ( y_p ) = Particular Integral (the intertemporal equilibrium value of ( y )).
Complementary Function: For the homogeneous reduced form ( y_{t+1} + a y_t = 0 ), the solution is found by assuming the form ( y_t = A b^t ). Substituting into the homogeneous equation gives ( b = -a ). π Formula: ( y_c = A (-a)^t ) The base (( -a )) in discrete time plays the same role in determining dynamic stability as the base (( e^r )) in continuous time.
Particular Integral: For ( a \neq -1 ), the simplest trial solution is a constant ( y_t = k ). Substituting into the standard form ( y_{t+1} + a y_t = c ) yields: π Formula: ( y_p = \frac{c}{1+a} )
General Solution: Combining these two parts gives: π Formula: ( y_t = A (-a)^t + \frac{c}{1+a} )
Definite Solution: To find a unique time path, an initial condition is used. Setting ( t=0 ) in the general solution, we solve for ( A ): π Formula: ( A = y_0 - \frac{c}{1+a} ) The definite solution is then: π Formula: ( y_t = \left( y_0 - \frac{c}{1+a} \right) (-a)^t + \frac{c}{1+a} ) This formula is valid only if ( a \neq -1 ).
Topic 155: Difference Equation Using General Method: Aβ -1 Numerical
This is a numerical application of the general formula from Topic 154.
π Example: Solve ( y_{t+1} - 5y_t = 1 ), given ( y_0 = \frac{7}{4} ).
- Identify parameters: The standard form is ( y_{t+1} + a y_t = c ). Comparing, we get ( a = -5 ) and ( c = 1 ). Note that ( a \neq -1 ).
- Calculate the Particular Integral: ( y_p = \frac{c}{1+a} = \frac{1}{1 + (-5)} = \frac{1}{-4} = -\frac{1}{4} ).
- Solve for the arbitrary constant ( A ): ( A = y_0 - \frac{c}{1+a} = \frac{7}{4} - \left( -\frac{1}{4} \right) = \frac{7}{4} + \frac{1}{4} = 2 ).
- Write the Definite Solution: ( y_t = A (-a)^t + y_p = 2 \cdot (-(-5))^t + \left( -\frac{1}{4} \right) = 2 (5)^t - \frac{1}{4} ).
The time path is ( y_t = 2(5)^t - \frac{1}{4} ), where the static equilibrium is ( y_p = -\frac{1}{4} ).
Topic 156: Solving Difference Equation Using General Method: A=-1
When ( a = -1 ), the previous formula for the Particular Integral becomes invalid because it involves division by zero (( 1 + a = 0 )). A different trial solution is required.
Complementary Function: The formula ( y_c = A (-a)^t ) remains valid. Since ( a = -1 ), we have ( -a = -(-1) = 1 ), so ( y_c = A (1)^t = A ).
Particular Integral: A constant trial solution fails. The next simplest trial solution is a linear function of time: ( y_t = k t ).
- Assume ( y_t = k t ). Its lead form is ( y_{t+1} = k (t+1) ).
- Substitute into the standard form: ( k(t+1) + a(k t) = c ).
- Simplify: ( k t + k + a k t = c \implies k(t + 1 + a t) = c ).
- Since ( a = -1 ), ( t + a t = t - t = 0 ). The equation simplifies to ( k (1) = c \implies k = c ). π Formula: ( y_p = c t )
General Solution: The complete solution for ( a = -1 ) is: π Formula: ( y_t = A (-a)^t + c t ) Since ( -a = 1 ), this is ( y_t = A + c t ).
Definite Solution: Using the initial condition at ( t=0 ): ( y_0 = A + c(0) \implies A = y_0 ) π Formula: ( y_t = y_0 (-a)^t + c t ), which simplifies to ( y_t = y_0 + c t ).
Topic 157: Difference Equation Using General Method: A=-1 Numerical
This is a numerical application of the formula for the ( a = -1 ) case.
π Example: Solve ( y_{t+1} - y_t = 2 ), given ( y_0 = 0 ).
- Identify parameters: The standard form is ( y_{t+1} + a y_t = c ). The equation is ( y_{t+1} + (-1) y_t = 2 ), so ( a = -1 ) and ( c = 2 ).
- Calculate the Particular Integral: ( y_p = c t = 2t ).
- Solve for the arbitrary constant ( A ): ( A = y_0 = 0 ).
- Write the Definite Solution: ( y_t = y_0 (-a)^t + c t = 0 \cdot (1)^t + 2t ). The time path is ( y_t = 2t ).
In this solution, the complementary function is zero (( y_c = 0 )), meaning the time path has no deviation from the moving equilibrium represented by ( y_p = 2t ).
β Key Takeaways
The most critical concepts from this lecture are the distinction between continuous and discrete time, and the two solution methods for first-order difference equations. The iterative method provides insight but is not practical for economic models. The general formula method is powerful: the standard form is ( y_{t+1} + a y_t = c ), and its solution depends critically on the value of ( a ). If ( a \neq -1 ), the particular integral is a constant equilibrium ( c/(1+a) ). If ( a = -1 ), the particular integral is a moving equilibrium ( ct ). The arbitrary constant ( A ) is always determined by the initial condition ( y_0 ), and the base (( -a )) in the complementary function dictates whether the time path converges to or diverges from equilibrium.
π§ Quick Revision Questions
- What is the standard form of a first-order difference equation?
- In the difference equation ( y_{t+1} + a y_t = c ), what trial solution is used for the particular integral if ( a \neq -1 )?
- What special modification to the particular integral is required when ( a = -1 )?
- Given the definite solution ( y_t = (y_0 - \frac{c}{1+a})(-a)^t + \frac{c}{1+a} ), what condition on the base (( -a )) ensures that the time path is dynamically stable (i.e., converges to equilibrium)?
- Solve the difference equation ( y_{t+1} - 3y_t = 4 ) with ( y_0 = 1 ) using the general method.
π Lecture 35 β Dynamic Stability of First Order Difference Equations
π Overview: This lecture examines the dynamic stability of first-order difference equations, focusing on how the complementary function determines whether the time-path is convergent or divergent. It introduces the seven cases (regions) of the parameter ( b ) that produce non-oscillatory and oscillatory time-paths, and explains the role of the coefficient ( A ) in scaling and mirroring these paths. Understanding these concepts is essential for analyzing economic models that evolve over discrete time periods.
ποΈ Topics Covered
The lecture covers dynamic stability of first-order difference equations beginning with non-oscillatory cases (Regions IβIV) and oscillatory cases (Regions VβVII). It then examines the role of the coefficient ( A ) in the complementary function, detailing its magnitude effect (scale) and sign effect (mirror). Finally, numerical examples demonstrate how to classify time-paths as dynamically stable or unstable based on the value of ( b ).
π Lecture Summary
TOPIC 158: DYNAMIC STABILITY OF FIRST ORDER DIFFERENCE EQUATION: NON-OSCILLATORY CASES
The dynamic stability of difference equations depends on the complementary function ( y_c ). Here, ( y_c = A(b)^t ). Setting ( A = 1 ) gives ( y_c = (b)^t ), so the expression ( (b)^t ) determines the dynamics of ( y_c ).
The base ( b ) is related to the coefficient ( a ) in the first-order difference equation ( y_t + a y_{t-1} = c ), where ( b = -a ).
The behavior of ( (b)^t ) is tabulated for positive, zero, and negative values of ( b ). When ( b = 2 > 0 ), the time-path shows exponential growth with positive signs, lying in the 1st quadrant (smooth graph). When ( b = 0 ), there is neither growth nor decay (horizontal straight line). When ( b = -2 < 0 ), the time-path shows exponential change with alternating positive/negative signs, lying in the 1st and 4th quadrants (ups and downs).
Terminology: A time-path is non-oscillatory when ( b \not< 0 ) (i.e., ( b \geq 0 )) and oscillatory when ( b < 0 ).
First-order difference equations have an additional feature of pattern (non-oscillatory or oscillatory) beyond the convergent or divergent nature seen in differential equations.
Using a number line to visualize regions of ( b ), seven values are identified: ( b > 1, b = 1, 0 < b < 1, b = 0 ) (non-oscillatory), and ( -1 < b < 0, b = -1, b < -1 ) (oscillatory). The first four (IβIV) are non-oscillatory.
Region β I (( b > 1, |b| > 1 )): Example ( y_c = (2)^t ). Values are 1, 2, 4, 8, 16... β all positive (non-oscillatory). Dynamic Stability: Time-path is divergent and hence dynamically unstable.
Region β II (( b = 1, |b| = 1 )): Example ( y_c = (1)^t ). Values are 1, 1, 1, 1, 1... β all positive (non-oscillatory). Dynamic Stability: Time-path is neither divergent nor convergent.
Region β III (( 0 < b < 1, |b| < 1 )): Example ( y_c = (1/2)^t ). Values are 1, 1/2, 1/4, 1/8, 1/16... β all positive (non-oscillatory). Dynamic Stability: Time-path is convergent and hence dynamically stable.
π Definition β Dynamic Stability: A time-path that converges to equilibrium over time. For first-order difference equations, this occurs when ( |b| < 1 ).
Region β IV (( b = 0, |b| = 0 )): Example ( y_c = (0)^t ). Values are 0, 0, 0, 0, 0... β all zero (non-oscillatory). Dynamic Stability: Time-path is neither divergent nor convergent. It is pre-converged at equilibrium; since ( y_c = 0 ), the time-path is already on equilibrium.
Summary: Cases IβIV have non-negative ( b ). A non-negative number raised to any power generates a positive number, giving a graph in the 1st quadrant only β hence non-oscillatory. Only when ( |b| < 1 ) is the time-path convergent.
Tabulation:
- Region I: ( b > 1 ), non-oscillatory, ( |b| > 1 ), non-convergent (divergent)
- Region II: ( b = 1 ), non-oscillatory, ( |b| = 1 ), non-convergent (neither)
- Region III: ( 0 < b < 1 ), non-oscillatory, ( |b| < 1 ), convergent
- Region IV: ( b = 0 ), non-oscillatory, ( |b| = 0 < 1 ), pre-converged (( y_c = 0 ), no deviation)
TOPIC 159: DYNAMIC STABILITY OF FIRST ORDER DIFFERENCE EQUATION: OSCILLATORY CASES
The remaining three cases (VβVII) are oscillatory because ( b < 0 ).
Region β V (( -1 < b < 0, |b| < 1 )): Example ( y_c = (-1/2)^t ). Values are 1, -1/2, 1/4, -1/8, 1/16... β alternating signs (oscillatory). Dynamic Stability: Time-path is convergent and hence dynamically stable. It moves above and below equilibrium with decreasing vertical distance.
Region β VI (( b = -1, |b| = 1 )): Example ( y_c = (-1)^t ). Values are 1, -1, 1, -1, 1... β alternating signs (oscillatory). Dynamic Stability: Time-path is neither divergent nor convergent. It moves above and below equilibrium with uniform vertical distance.
Region β VII (( b < -1, |b| > 1 )): Example ( y_c = (-2)^t ). Values are 1, -2, 4, -8, 16... β alternating signs (oscillatory). Dynamic Stability: Time-path is divergent and hence dynamically unstable. It moves above and below equilibrium with increasing vertical distance.
Summary: Cases VβVII have negative ( b ). A negative number raised to any power generates numbers with alternating signs, giving a graph in the 1st and 4th quadrants β hence oscillatory. Only when ( |b| < 1 ) is the time-path convergent.
Tabulation:
- Region V: ( -1 < b < 0 ), oscillatory, ( |b| < 1 ), convergent
- Region VI: ( b = -1 ), oscillatory, ( |b| = 1 ), non-convergent (neither)
- Region VII: ( b < -1 ), oscillatory, ( |b| > 1 ), non-convergent (divergent)
TOPIC 160: DYNAMIC STABILITY OF FIRST ORDER DIFFERENCE EQUATION: ROLE OF ( A )
Recall ( y_c = A(b)^t ). Among the seven cases, the convergent ones (Case III: non-oscillatory and Case V: oscillatory) are the desired cases. The coefficient ( A ) can affect ( y_c ) in two ways: magnitude and sign.
The dual effect of ( A ) on ( y_c ):
- Magnitude: ( |A| = 0 ) nullifies (yields ( y_c = 0 )); ( |A| < 1 ) minifies; ( |A| = 1 ) neutral; ( |A| > 1 ) amplifies.
- Sign: ( A > 0 ) retains the sign of ( (b)^t ); ( A < 0 ) reverses it.
1. Scale Effect: In Case III (convergent, non-oscillatory):
- Original: ( y_c = (1/2)^t ) starts from 1.
- Doubled: ( y_c = 2(1/2)^t ) starts from 2.
- Halved: ( y_c = (1/2)(1/2)^t ) starts from 0.5.
The magnitude of ( A ) (( \gtrless 1 )) scales the time-path up or down. This is called the Scale Effect of ( A ).
In Case V (convergent, oscillatory):
- Original: ( y_c = (-1/2)^t ) starts from 1.
- Doubled: ( y_c = 2(-1/2)^t ) starts from 2.
- Halved: ( y_c = (1/2)(-1/2)^t ) starts from 0.5.
Again, the magnitude of ( A ) scales the oscillatory time-path up or down.
2. Mirror Effect: In Case III (convergent, non-oscillatory):
- Original: ( y_c = (1/2)^t ) starts from +1.
- Reversed: ( y_c = -1(1/2)^t ) starts from -1.
The negative sign of ( A ) reverses the shape, creating a mirror image of the original time-path. This is called the Mirror Effect of ( A ).
In Case V (convergent, oscillatory):
- Original: ( y_c = (-1/2)^t ) starts from +1.
- Reversed: ( y_c = -1(-1/2)^t ) starts from -1.
Again, the negative sign of ( A ) creates a mirror image of the oscillatory time-path.
TOPIC 161: DYNAMIC STABILITY OF FIRST ORDER DIFFERENCE EQUATION: NUMERICALS
Example # 1: Check the dynamic stability of ( y_t = 2(-\frac{4}{5})^t + 9 ).
The formula is ( y_t = A(b)^t + y_p ), where ( y_p = \frac{c}{1+a} ).
Breaking up: ( y_c = 2(-\frac{4}{5})^t ), ( y_p = 9 ). So ( b = -\frac{4}{5} ).
Interpretation: Here, ( b = -\frac{4}{5} < 0 ) (oscillatory) and ( |b| = \frac{4}{5} < 1 ) (convergent). Therefore, the time-path is oscillatory in pattern and convergent in nature. The time-path is dynamically stable.
π Example: ( b = -\frac{4}{5} ). Since ( -\frac{4}{5} < 0 ), the path oscillates. Since ( |-\frac{4}{5}| = 0.8 < 1 ), the oscillations converge to equilibrium (the particular solution ( y_p = 9 )).
Note: This path differs from ( y_t = -2(\frac{4}{5})^t + 9 ), where the sign inside the parentheses is positive, giving a non-oscillatory but possibly convergent or divergent time-path.
Example # 2: Check the dynamic stability of ( y_t = 3(2)^t + 4 ).
The formula is ( y_t = A(b)^t + y_p ).
Breaking up: ( y_c = 3(2)^t ), ( y_p = 4 ). So ( b = 2 ).
Interpretation: Here, ( b = 2 \not< 0 ) (non-oscillatory) and ( |b| = 2 \not< 1 ) (non-convergent, i.e., divergent). Therefore, the time-path is non-oscillatory in pattern and non-convergent in nature. The time-path is dynamically unstable.
π Example: ( b = 2 ). Since ( 2 > 0 ), the path is smooth (non-oscillatory). Since ( |2| = 2 > 1 ), the path diverges away from equilibrium (( y_p = 4 )).
β Key Takeaways
- Dynamic stability of first-order difference equations depends entirely on the absolute value of ( b ) (where ( b = -a )): if ( |b| < 1 ), the time-path is convergent (dynamically stable); if ( |b| > 1 ), it is divergent (dynamically unstable); if ( |b| = 1 ), it is neither convergent nor divergent. 2. The sign of ( b ) determines the pattern: ( b \geq 0 ) yields non-oscillatory (smooth) time-paths while ( b < 0 ) yields oscillatory (alternating) time-paths. 3. The seven regions of ( b ) cover all possible time-path behaviors: non-oscillatory cases (IβIV: ( b > 1 ), ( b = 1 ), ( 0 < b < 1 ), ( b = 0 )) and oscillatory cases (VβVII: ( -1 < b < 0 ), ( b = -1 ), ( b < -1 )). 4. The coefficient ( A ) in the complementary function ( y_c = A(b)^t ) has a Scale Effect (magnitude ( |A| ) scales the path up or down) and a Mirror Effect (negative sign ( A < 0 ) reverses the path). 5. For any numerical example, always compute both ( b ) and ( |b| ) to classify the time-path as oscillatory/non-oscillatory and convergent/divergent, then state whether the system is dynamically stable (convergent) or unstable (divergent).
π§ Quick Revision Questions
- Under what condition on ( |b| ) is the time-path of a first-order difference equation dynamically stable (convergent)?
- What is the difference between a non-oscillatory and an oscillatory time-path in terms of the sign of ( b )?
- In Example #1, ( y_t = 2(-\frac{4}{5})^t + 9 ), what are the values of ( b ) and ( |b| ), and what do they imply about the time-path?
- Explain the Scale Effect and Mirror Effect of the coefficient ( A ) in the complementary function ( y_c = A(b)^t ).
- In Region VI (( b = -1 )), why is the time-path neither convergent nor divergent even though it oscillates?
π Lecture 36 β Income Dynamics of Difference Equations
π Overview: This lecture examines how national income evolves over time when consumption and investment decisions depend on past income levels. Using first-order difference equations, we model income dynamics under various economic scenarios, including lagged consumption, induced investment, capital market equilibrium, money market conditions, and balanced trade. Understanding these dynamics is crucial for analyzing economic stability and policy effects.
ποΈ Topics Covered
This lecture covers six topics: a lagged income determination model framework with consumption depending on last period's income, its numerical solution, income dynamics with induced investment alongside autonomous investment, income dynamics in the capital market where savings equal investment from previous period's income, income dynamics in the money market where money supply and demand depend on current and past income, and finally income dynamics for balanced foreign trade where imports and exports are functions of current and lagged income.
π Lecture Summary
Topic 162: Lagged Income Determination Model: Framework
Unlike the simple income determination model, we introduce lags in the consumption function, where current consumption depends on last period's income. The model is:
( C_t = C_0 + cY_{t-1} ) where ( c = MPC ) (0<c<1), and ( I_t = I_0 ) (autonomous investment).
Income identity: ( Y_t = C_t + I_t )
Substituting: ( Y_t = C_0 + cY_{t-1} + I_0 ), thus ( Y_t - cY_{t-1} = C_0 + I_0 )
This is a first-order difference equation (FODE) of the form ( Y_t + aY_{t-1} = c ), where ( a = -c ) and the constant is ( C_0 + I_0 ). Since ( a \neq -1 ) (as MPC<1), the definite time-path formula is used.
π Definition β First-Order Difference Equation: An equation expressing a variable's current value as a function of its immediately preceding value.
π Formula: ( Y_t = (Y_0 - \frac{C_0 + I_0}{1-c})(c)^t + \frac{C_0 + I_0}{1-c} )
This is the definite time-path. Dynamic stability depends on the exponential expression ( (c)^t ). Since ( |c| > 0 ) (non-zero) and ( c < 1 ), the time-path is non-oscillatory and convergent (dynamically stable).
π‘ Why this matters: The time-path converges to equilibrium because the MPC is less than 1, meaning each period's income moves closer to the steady state.
As ( t \to \infty ), the complementary function ( Y_c = (Y_0 - Y^*)(c)^t \to 0 ), so ( Y_t \to Y_p = \frac{C_0 + I_0}{1-c} ), the particular solution or equilibrium income.
Topic 163: Lagged Income Determination Model: Numerical
We solve the model numerically with specific values:
( C_t = 200 + 0.9Y_{t-1} ) (MPC = 0.9), ( I_t = 100 ), ( Y_0 = 4500 )
Substituting: ( Y_t = 200 + 0.9Y_{t-1} + 100 ) β ( Y_t - 0.9Y_{t-1} = 300 )
This is ( Y_t + (-0.9)Y_{t-1} = 300 ), so ( a = -0.9 \neq -1 ), ( c = 300 ).
π Formula: ( Y_t = (4500 - \frac{300}{1-0.9})(0.9)^t + \frac{300}{1-0.9} )
( Y_t = (4500 - 3000)(0.9)^t + 3000 = 1500(0.9)^t + 3000 )
Since ( |0.9| > 0 ) (non-zero) and ( 0.9 < 1 ), the time-path is non-oscillatory and convergent. As ( t \to \infty ), ( 1500(0.9)^t \to 0 ), so ( Y_t \to 3000 ) (equilibrium).
The complementary function ( Y_c = 1500(0.9)^t ) represents the deviation from equilibrium, which diminishes over time.
Topic 164: Income Dynamics with Induced Investment Using Difference Equations
We now introduce induced investment alongside autonomous investment.
( C_t = 200 + 0.4Y_{t-1} ) (MPC = 0.4), ( I_t = 100 + 0.2Y_{t-1} ) (autonomous + induced), ( Y_0 = 950 )
Substituting: ( Y_t = 200 + 0.4Y_{t-1} + 100 + 0.2Y_{t-1} ) β ( Y_t - 0.6Y_{t-1} = 300 )
So ( a = -0.6 \neq -1 ), ( c = 300 ).
π Formula: ( Y_t = (950 - \frac{300}{1-0.6})(0.6)^t + \frac{300}{1-0.6} = (950 - 750)(0.6)^t + 750 = 200(0.6)^t + 750 )
Since ( |0.6| > 0 ) and ( 0.6 < 1 ), the time-path is non-oscillatory and convergent to ( Y^* = 750 ).
π‘ Why this matters: The induced investment reduces the coefficient of the exponential term to 200 and the equilibrium to 750, showing how investment sensitivity affects dynamics.
Topic 165: Income Dynamics & Capital Market Using Difference Equations
In the capital market, savings are a proportion of current income, while investment depends on last period's income.
Savings: ( S_t = sY_t ) where ( s = MPS ) (marginal propensity to save, 0<s<1)
Investment: ( I_t = vY_{t-1} ) where ( v < 1 ) is the proportion of last year's income invested
Equilibrium: ( S_t = I_t ) β ( sY_t = vY_{t-1} ) β ( sY_t - vY_{t-1} = 0 )
Normalizing: ( Y_t - \frac{v}{s}Y_{t-1} = 0 ) β ( Y_t + (-\frac{v}{s})Y_{t-1} = 0 )
Here, ( a = -\frac{v}{s} \neq -1 ) (provided ( v \neq s )) and ( c = 0 ).
π Formula: ( Y_t = Y_0(\frac{v}{s})^t )
Dynamic stability depends on ( b = \frac{v}{s} ):
- Nature: If ( |b| < 1 ) (i.e., ( v < s )) β convergent; if ( |b| > 1 ) β non-convergent
- Pattern: Since ( b > 0 ) (positive), the path is non-oscillatory
Therefore, the time-path is convergent only if ( v < s ), and it moves in a non-oscillatory pattern.
Topic 166: Income Dynamics & Money Market Using Difference Equations
On the money market, the government sets money supply as a proportion of previous period's income, while money demand depends on current income.
( M_s = sY_{t-1} ) (proportion s of last year's income) ( M_D = a + vY_t ) (transaction demand, 0 < v < s < 1, a > 0)
Equilibrium: ( M_s = M_D ) β ( sY_{t-1} = a + vY_t ) Rearranging: ( vY_t - sY_{t-1} = -a ) β ( Y_t - \frac{s}{v}Y_{t-1} = -\frac{a}{v} )
So ( a = -\frac{s}{v} \neq -1 ) (since v < s), ( c = -\frac{a}{v} ).
π Formula: ( Y_t = (Y_0 + \frac{a}{v-s})(\frac{s}{v})^t - \frac{a}{v-s} )
Dynamic stability depends on ( b = \frac{s}{v} ):
- Since ( v < s ), ( |b| = \frac{s}{v} > 1 ) β non-convergent
- Since ( b > 0 ) β non-oscillatory
π‘ Why this matters: When money supply growth (s) exceeds money demand sensitivity (v), the system diverges, potentially leading to instability.
Topic 167: Income Dynamics & Balanced Trade Using Difference Equations
For balanced trade, imports depend on current income, while exports depend on both current and previous period's income.
( M_t = mY_t ) (0 < m < 1) ( X_t = aY_t + vY_{t-1} ) (0 < a, v < 1)
Balanced trade: ( M_t = X_t ) β ( mY_t = aY_t + vY_{t-1} ) ( (m-a)Y_t - vY_{t-1} = 0 )
Normalizing: ( Y_t - \frac{v}{m-a}Y_{t-1} = 0 ) β ( Y_t + (-\frac{v}{m-a})Y_{t-1} = 0 )
Here, ( a = -\frac{v}{m-a} \neq -1 ) (provided m-a β v), ( c = 0 ).
π Formula: ( Y_t = Y_0(\frac{v}{m-a})^t )
Dynamic stability depends on ( b = \frac{v}{m-a} ):
- Nature: If ( m-a > v ), then ( |b| < 1 ) β convergent
- Pattern: If convergent, ( m-a > 0 ) β ( b > 0 ) β non-oscillatory
Therefore, the time-path converges if ( m-a > v ), moving in a non-oscillatory pattern.
β Key Takeaways
The fundamental result is that all lagged income models reduce to first-order difference equations whose stability depends on the absolute value of the coefficient on the lagged term. When the coefficient's absolute value is less than one, the time-path converges to equilibrium; when greater than one, it diverges. All models in this lecture produce non-oscillatory patterns because the coefficients are positive. The particular solution (equilibrium income) is given by the constant term divided by one minus the coefficient on lagged income, generalizing the multiplier formula. Finally, the specific economic structureβwhether lags appear in consumption, investment, savings, money supply, or tradeβdetermines the exact coefficient values and thus the dynamic properties of national income.
π§ Quick Revision Questions
- What is the general solution for the lagged income determination model ( Y_t - cY_{t-1} = C_0 + I_t ), and under what condition is the time-path dynamically stable?
- In the capital market model ( S_t = sY_t ) and ( I_t = vY_{t-1} ), what condition ensures convergence of the income time-path?
- For the money market model with ( M_s = sY_{t-1} ) and ( M_D = a + vY_t ), why does the time-path diverge when ( s > v )?
- In the balanced trade model, if ( m = 0.3 ), ( a = 0.1 ), and ( v = 0.15 ), will the income time-path converge? Explain.
- Why do all the income dynamics models in this lecture produce non-oscillatory time-paths?